LANCET - 2012 - 24 - JUNIO 16

This Week in Medicine

Child survival June 14–15 saw the Governments of the USA, India, and Ethiopia convene a summit in Washington, DC, USA, where they pledged to help reduce preventable child deaths. The Child Survival Call to Action forum brought together politicians and academics, who together agreed to identify areas where child health can be improved. Needle distribution Kenya will begin distributing needles and syringes to more than 50 000 intravenous drug users next month. Intravenous drug use causes 4% of HIV infections nationally, and up to 17% of new infections in the Coast Province. In addition to needle distribution, the government will also encourage HIV testing, and provide antiretroviral and tuberculosis treatment. Elder abuse awareness June 15 was the seventh World Elder Abuse Awareness Day, organised by the US Department of Health and Human Services Administration on Aging. An event was held in the White House on June 14 to draw attention to the widespread problem of abuse, neglect, and exploitation of elderly people, and to its various forms. Aid for Yemen The US Government is to provide an extra US$6·5 million in humanitarian assistance to Yemen, officials aim to pledge $118 million in the 2012 fiscal year to improve the supply of clean water and food for internally displaced residents and refugees. Half of Yemen’s population has insufficient food, with 1 million children younger than 5 years being undernourished. Disney bans junk ads The Walt Disney Corporation has announced a plan to ban advertisements for food and drinks high in sugar, salt, and fat during programmes aimed at children younger than 12 years on its television, radio, and web content. The ban will not take effect until 2015, owing to long-term advertising contracts. Smarter health spending A report released on June 11 by the US Center for Global Development highlights the need for countries to establish health technology assessment agencies similar to the UK’s National Institute for Health and Clinical Excellence. A health spending system that aims to balance supply and demand, covering as many people as possible, should be implemented, the report states. IVF risks Pregnancies achieved via in-vitro fertilisation (IVF) carry an increased risk of some types of poor birth outcome, according to a report released by the UK’s Royal College of Obstetricians and Gynaecologists. A quarter of IVF pregnancies result in multiple births; there is also an increased risk of premature birth, low birthweight, pre-eclampsia, and cerebral palsy.
www.thelancet.com Vol 379 June 16, 2012

For the Child Survival Call to Action forum see http:// 5thbday.usaid.gov/pages/ ResponseSub/Event.aspx For more on World Elder Abuse Awareness Day see http://ncea. aoa.gov/ncearoot/Main_Site/ About/Initiatives/Take_A_ Stand.aspx For the Royal College of Obstetricians and Gynaecologists’ report on the risks of IVF see http://www.rcog. org.uk/news/rcog-release-ivflinked-poorer-birth-outcomesexperts-say/ For more on the USA’s humanitarian assistance to Yemen see http://www.usaid. gov/news-information/pressreleases/us-provides-additionalhumanitarian-assistance-yemen For Disney’s ban on junk food advertising see http:// thewaltdisneycompany. com/disney-news/pressreleases/2012/06/walt-disneycompany-sets-new-standardsfood-advertising-kids For the Center for Global Development report on health technology assessment see http://www.cgdev.org/content/ publications/detail/1426240/ For the EU Ombudsman’s recommendation on transparency at the EMA see http://www.ombudsman. europa.eu/en/cases/ draftrecommendation.faces/ en/11553/html.bookmark For MSF’s call to help refugees in South Sudan see http:// www.msf.org.uk/South_Sudan_ refugee_influx_20120606.news

Seeking refuge Médecins Sans Frontières has urged the UN refugee agency to find suitable safe ground for the 30 000 refugees who have trekked from Sudan’s Blue Nile state to South Sudan’s Upper Nile state to flee conflict. Diarrhoea and malnutrition are a frequent occurrence, meaning that there is a pressing need for new camps with adequate water supplies. Mental illness in China The Chinese Ministry of Health has released a draft guideline on mental health, which plans to establish a treatment network for serious mental illnesses in 95% of China’s counties and cities by 2015. China has an estimated 16 million people with serious mental illnesses such as schizophrenia, schizoaffective disorder, and bipolar mood disorder. Legionnaires’ disease An outbreak of Legionnaires’ disease in Edinburgh, UK, has resulted in 88 confirmed and suspected cases, and one death, as of June 11. 41 patients were being treated in hospital, including 15 in intensive care. Cooling towers in the southwest of the city are being investigated as a possible source of the outbreak, and have been chemically treated. Beer at the World Cup? Brazil’s longstanding ban on the sale of alcohol at football matches seems to have been relaxed for the World Cup in 2014, which Brazil is hosting. A recently passed bill on the event lacked any mention of a ban on alcohol sales. The World Cup’s governing body, FIFA, which is sponsored by Budweiser, had demanded that Brazil allow the sale of beer at World Cup matches.

Science Photo Library

Clear progress To help ensure the safety of paediatric medications, the EU Ombudsman has demanded that the European Medicines Agency increase the transparency of its decision-making procedures surrounding the testing of drugs in children. Better documentation of assessments, along with relevant guidelines for the pharmaceutical industry, were suggested.

Corbis

i

AFP/Getty Images

Editorial

Prediabetes and the potential to prevent diabetes
Around 366 million people worldwide have diabetes mellitus; it is estimated that this number will reach 552 million by 2030. Many more people are at risk of type 2 diabetes, and it is expected that ensuing microvascular and macrovascular disease will lead to a substantial burden of ill health, and demand for medical treatment, in both developed and developing countries. Today’s special issue of The Lancet is dedicated to research on this growing but still underappreciated health crisis, in collaboration with the American Diabetes Association (ADA). In this issue, Adam Tabák and colleagues review prediabetes, also known as intermediate hyperglycaemia, a state in which the risk of developing diabetes is increased. Prediabetes is defined by blood glucose concentrations higher than normal, but lower than established thresholds for diabetes itself. As defined by WHO, people with prediabetes have impaired fasting glucose (IFG), with a fasting plasma glucose (FPG) concentration between 6·1 and 7·0 mmol/L; ADA uses a lower cutoff value for IFG (FPG 5·6–6·9 mmol/L). Prediabetes is a high risk state not only for developing diabetes, but also the associated complications. Vascular complications, nephropathy, and neuropathies have been reported in people with prediabetes. Every year, in the region of 5–10% of people with prediabetes reach the clinical criteria for diabetes; according to an ADA expert panel, as many as 70% of individuals with prediabetes will progress to diabetes. It has been predicted that, globally, there will be about 470 million people with prediabetes by 2030. How should prediabetes be managed? Lifestyle changes to counteract obesity and physical inactivity are the first line of defence against progression to diabetes. Although controversial, there is also evidence for potential benefits of individual drugs. Metformin has been found to be beneficial in people with prediabetes with a higher baseline and higher FPG than in those with lower FPG concentrations. There is some evidence for use of pioglitazone in obese people with impaired glucose tolerance: incidence of type 2 diabetes was decreased in the ACT NOW study, although weight gain and oedema were also reported. The role of newer drugs, such as DPP-4 inhibitors and GLP-1 receptor agonists, in prediabetes has yet to be elucidated.
www.thelancet.com Vol 379 June 16, 2012

Also in today’s Lancet, Leigh Perreault and colleagues present research on individuals with impaired glucose tolerance who completed the US Diabetes Prevention Program (DPP) trial without developing diabetes, and were passively followed up within the Diabetes Prevention Program Outcome Study (DPPOS) for up to 6 years. Among 894 people who had at least one normal oral glucose tolerance test during DPP, there was a 56% lower relative risk of developing diabetes during passive follow-up than among 1096 people who never regressed to normoglycaemia during DPP. While confirmation of these findings is needed, it seems that even transient regression of impaired glucose tolerance to normoglycaemia can have lasting clinical effects. Regression to normal glucose tolerance could be a way to identify people at differential risk of progression to diabetes. Such stratification could help to target individuals for whom additional treatment might be needed to prevent diabetes or to impede disease progression. Andrea Tricco and colleagues report on quality improvement strategies for diabetes management in their systematic review, also in this issue. The active development of new drugs for type 2 diabetes needs to be complemented by intensified investment and research in interventions aimed at diabetes risk factors, particularly adiposity and physical inactivity. Public health strategies to deploy lifestyle interventions for diverse populations will also be important in coming decades. While we are learning more about the association between prediabetes and diabetes and the spectrum of disease, much remains to be elucidated. Should drug treatments be used early in people with prediabetes? Do early interventions intended to revert the prediabetic state to normal glucose tolerance prevent macrovascular and microvascular complications of disease? Diabetes is a chronic disease, and those affected are often exposed to long-term polypharmacy. Therefore, caution should be exercised about unnecessarily expanding the population receiving drugs. As shown in the DPPOS, the reduction in diabetes risk was achieved by reversing prediabetes, regardless of the method used. Above all, lifestyle changes early on in the predisease state should be emphasised. The Lancet

See Comment pages 2215, 2216, 2218, 2220, and 2222 See World Report page 2227 See Perspectives pages 2231, 2232, 2233, 2234, and 2235 See Articles pages 2243, 2252, 2262, and 2270 See Series pages 2279, 2291, and 2300

2213

Science Photo Library

Editorial

Gonorrhoea—old disease, new threat
Delegates at the 29th International AIDS Conference in Washington, DC, next month might consider how, in three decades, their field has changed. The progress from the first case reports, to fearful recognition of the extent of the HIV epidemic, to a point at which “an AIDS-free generation” is talked about represents a remarkable success story. Recognition of this achievement, however, must not turn to complacency; nor should it engender forgetfulness about other sexually transmitted infections (STIs). As medical science solves problems, it also creates new challenges. Such is the case with Neisseria gonorrhoeae. It is deeply concerning that in Australia, France, Japan, Norway, Sweden, and the UK, cases of gonorrhoea have been reported that are cephalosporin-resistant—and thus effectively untreatable. This resistance, accompanied by a high incidence of infections and possibly less noticeable symptoms, bodes ill for the future: fertility problems, ectopic pregnancies, stillbirth, spontaneous abortion, premature deliveries, and severe neonatal eye infections resulting in blindness might be its legacy. It is essential that steps are taken now in accordance with WHO’s Global Action Plan. Health-care workers must be vigilant and prescribe antibiotics appropriately. As with all cases of emerging antibiotic-resistant infections, research into new alternatives for treatment must be prioritised. And public health agencies should strengthen surveillance efforts, as well as targeting safer-sex educational programmes at key groups including young people, men who have sex with men, and sex workers. The growing threat of cephalosporinresistant gonorrhoea also indicates that it is time to break down the barriers between the efforts against HIV, and those against other STIs. As the at-risk groups and means of prevention are so similar, and as untreated gonococcal infection can significantly increase the risk of HIV infection and transmission, it would seem a logical step. Without new ways of thinking, the health of future generations will be blighted by a disease whose worst effects were—not so long ago—thought to have been consigned to history. The Lancet

Science Photo Library

Neisseria gonorrhoeae
See World Report page 2229 For the 29th International AIDS Conference see http://www. aids2012.org/ For WHO’s Global Action Plan to control the spread and impact of antimicrobial resistance in Neisseria gonorrhoeae see http:// whqlibdoc.who.int/ publications/2012/ 9789241503501_eng.pdf

Safeguarding Europe’s youngest citizens
Protecting the youngest and most vulnerable members of society from harm should concern all parents, health professionals, and governments in high-income and low-income nations. In Europe, unintentional injury is the leading cause of death and disability in children. Implementation of effective prevention strategies could save most of these lives as well as saving governments billions of Euros in treatment costs at a time of increasing economic woe. However, the latest set of Child Safety Report Cards for 31 countries in Europe produced by the European Child Safety Alliance show inconsistencies in policy adoption across the region. The report cards score countries on their level of adoption, implementation, and enforcement of more than 100 proven strategies and policies to prevent unintentional injury. Reports were produced for 18 participating countries in 2007, 26 in 2009, and 31 this year. Overall, there was a substantial improvement in country scores from 2007 to 2012. But the data show that every country could still do better. For example, only 13 countries have a national helmet law requiring use of a bicycle helmet while cycling. And only 15 countries have laws requiring child resistant packaging of medicines. Furthermore, no country has a law requiring the use of a rear facing child passenger restraint for children aged 0–4 years (although this is normal practice in Sweden where deaths in this age group have been reduced to almost zero). The report cards are an invaluable resource for policy makers in Europe. A detailed one exists for each participating country, highlighting performance gaps, action required, and inequities. No other WHO region has this level of regular assessment for child safety policies. However, many countries are still failing to measure the impact of safety measures that have been implemented—a crucial, necessary assessment to show the effectiveness of policies and which of them save the most lives. Additionally, to date, no European country has adopted all the recommended harm reduction measures. Taking every possible step available to protect children from preventable injuries should be a goal for all countries in Europe and beyond. The Lancet
www.thelancet.com Vol 379 June 16, 2012

Marc Romanelli/Getty Images

For more on the report cards see www.childsafetyeurope.org

2214

Comment

Is regression to normoglycaemia clinically important?
Impaired fasting glucose, impaired glucose tolerance, and type 2 diabetes are metabolic conditions characterised by raised blood glucose concentrations that put people at risk of various serious health outcomes. These concentrations generally rise with time so that patients with impaired fasting glucose or glucose tolerance progress to diabetes,1 and patients with diabetes need progressively more complex regimens to manage glucose concentrations. Several large randomised controlled trials have shown that diabetes can be prevented or delayed in people with impaired fasting glucose or impaired glucose tolerance by diet, physical activity, and various drugs. Less well understood is the clinical significance of regression of these conditions to normoglycaemia, which has been noted in several trials2–10 (table). Does regression predict a reduced risk of incident metabolic and vascular diseases, or does it simply reflect glucose fluctuations over time? In The Lancet, Leigh Perreault and colleagues11 describe epidemiological analyses of participants with impaired glucose tolerance who completed the Diabetes Prevention Program (DPP) trial without developing diabetes, and were passively followed up within the Diabetes Prevention Program Outcomes Study for up to 6 years. The investigators found that, among 894 people who had at least one normal oral glucose tolerance test during DPP, there was a 56% lower relative risk of developing diabetes during passive follow-up than among 1096 people who never regressed to normoglycaemia during DPP (hazard ratio [HR] 0·44, 95% CI 0·37–0·55). This relationship was independent of other predictors of incident diabetes and was unaffected by the treatment group assignment (people in the DPP trial having been assigned to intensive lifestyle intervention, metformin, or placebo groups). This reduction in the incidence of diabetes in people who regress to normoglycaemia might simply mean that people who spontaneously regress have a lower risk of developing diabetes than do those who do not regress. Alternatively, regression to normoglycaemia in response to either a lifestyle intervention, metformin, or other treatments might identify people who will ultimately derive the greatest benefit from these interventions. Whether a reduced incidence of diabetes after regression translates into a reduced incidence of diabetes-related health consequences such as blindness and vascular disease is unknown, and can be assessed by ongoing follow-up of this cohort, and by future clinical trials testing induction of normoglycaemia with any treatment versus maintaining glucose concentrations in the impaired fasting glucose or glucose tolerance range. Of interest was the counter-intuitive finding that people who had been assigned to the intensive lifestyle group were more likely to develop diabetes during passive follow-up than were people who had been allocated placebo (HR 1·31, 95% CI 1·03–1·68). However, because these analyses excluded people who had developed diabetes during the active treatment phase of DPP, they were not intention-to-treat comparisons. The retention of people who regressed to normoglycaemia despite being in the placebo group might have selected for people who were less likely to develop diabetes during subsequent follow-up than were people who regressed to normoglycaemia in response to intensive lifestyle therapy. Removal of the lifestyle intervention at the end of DPP would have removed its protective effect in people who were dependent on the lifestyle intervention. Moreover, adoption by the placebo group of some of the lifestyle changes that were proven to be effective in DPP would have accentuated this difference. Irrespective of these treatment group differences, the findings clearly suggest that transient regression of impaired glucose tolerance to normoglycaemia
N
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60828-9 See Online/Articles DOI:10.1016/S01406736(12)60525-X

Population Length Regression Active treatment (years) in controls (%) 3 2·5 NA 24·1% Lifestyle Metformin Lifestyle Metformin Lifestyle+metformin Lifestyle+pioglitazone Pioglitazone Rosiglitazone Ramipril Voglibose Acarbose

Risk of regression* (95% CI) 2·05 (1·66–2·53) 1·25 (0·99–1·58) 1·48 (0·99–2·22) 1·27 (0·85–1·89) 1·31 (0·87–1·95) 1·27 (0·98–1·65) 1·71 (1·33–2·19) 1·71 (1·57–1·87)‡ 1·16 (1·07–1·27)‡ 1·54 (1·36–1·75)‡ 1·14 (0·98–1·33)

DPP2† Indian DPP-13

2528 IFG+IGT 531 IGT

Indian DPP-24 ACT NOW5† DREAM6,7† CANOE8† STOP-NIDDM10

407 IGT 602 IGT 5269 IFG+/–IGT 207 IGT 1429 IFG+IGT

3 2·4 3 3·9 0·9 3·3

32·3% 28% 30·3% 38·2% 53·1% 51·5% 31%

Rosiglitazone+metformin 1·50 (1·21–1·86)

Kawamori et al9† 1780 IGT

NA=not available. *Unless otherwise noted, relative risk of regression to normoglycaemia was calculated from reported proportions of patients achieving normoglycaemia in treatment groups. †Normoglycaemia was defined on basis of oral glucose tolerance test; fasting plasma glucose cutoffs varied between studies. ‡Hazard ratio of regression to normoglycaemia was provided in publication.

Table: Risk of regression to normoglycaemia in people with impaired fasting glucose (IFG), impaired glucose tolerance (IGT), or both

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60828-9

1

Comment

that is either spontaneous or in response to treatment is of clinical relevance. On the assumption that this finding is confirmed, identification of regression to normoglycaemia could be an important way to stratify people into those at higher and lower risk of progression to diabetes. Such stratification could therefore identify individuals for whom additional treatment might be needed to prevent diabetes or to slow down disease progression. Finally, identification of the genetic, environmental, and behavioural predictors of regression might further elucidate pathophysiological mechanisms underlying diabetes development. Perreault and colleagues’ study focused on regression to normoglycaemia in people with impaired glucose tolerance. Other studies have described regression to normoglycaemia in people with diabetes. For example, short-term intensive treatment with insulin or oral hypoglycaemic agents has been shown to induce longlasting regression of recently diagnosed type 2 diabetes to normoglycaemia after discontinuation of therapy.12 Moreover, the American Diabetes Association now refers to regression as diabetes remission and has defined it as partial or complete normalisation of glycaemic parameters for at least 1 year without ongoing pharmacological or surgical treatments.13 In the same way as regression to normoglycaemia in people with impaired glucose tolerance predicts fewer cases of diabetes, regression to normoglycaemia in people with diabetes could predict fewer cases of eye, kidney, or cardiovascular disease and needs to be explored in future studies. Factors that predict regression to normoglycaemia, what makes this regression temporary or sustained, and whether regression reduces long-term outcomes are all questions that need further research. The results of such research might substantially change the therapeutic strategy from diabetes prevention and lifelong glucoselowering treatment to induction of regression and monitoring for relapse.

*Natalia Yakubovich, Hertzel C Gerstein
McMaster University and Hamilton Health Sciences Department of Medicine and Population Health Research Institute, Hamilton, ON L8S 4K1, Canada [email protected]
NY has accepted Sanofi grant support for travel expenses. HCG is on scientific advisory boards for Sanofi, GlaxoSmithKline, Novo Nordisk, Roche, Merck, Novartis, Janssen, Abbot, and AstraZeneca, is on trial steering committees for Bayer and Roche, has received a grant from Lilly, and has received speakers’ fees from Sanofi and Bayer. 1 Gerstein HC, Santaguida P, Raina P, et al. Annual incidence and relative risk of diabetes in people with various categories of dysglycemia: a systematic overview and meta-analysis of prospective studies. Diabetes Res Clin Pract 2007; 78: 305–12. Perreault L, Kahn SE, Christophi CA, et al. Regression from pre-diabetes to normal glucose regulation in the Diabetes Prevention Program. Diabetes Care 2009; 32: 1583–88. Snehalatha C, Mary S, Selvam S, et al. Changes in insulin secretion and insulin sensitivity in relation to the glycemic outcomes in subjects with impaired glucose tolerance in the Indian Diabetes Prevention Programme-1 (IDPP-1). Diabetes Care 2009; 32: 1796–801. Ramachandran A, Snehalatha C, Mary S, et al. Pioglitazone does not enhance the effectiveness of lifestyle modification in preventing conversion of impaired glucose tolerance to diabetes in Asian Indians: results of the Indian Diabetes Prevention Programme-2 (IDPP-2). Diabetologia 2009; 52: 1019–26. DeFronzo RA, Tripathy D, Schwenke DC, et al. Pioglitazone for diabetes prevention in impaired glucose tolerance. N Engl J Med 2011; 364: 1104–15. Gerstein HC, Yusuf S, Bosch J, et al. Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial. Lancet 2006; 368: 1096–105. Bosch J, Yusuf S, Gerstein HC, et al. Effect of ramipril on the incidence of diabetes. N Engl J Med 2006; 355: 1551–62. Zinman B, Harris SB, Neuman J, et al. Low-dose combination therapy with rosiglitazone and metformin to prevent type 2 diabetes mellitus (CANOE trial): a double-blind randomised controlled study. Lancet 2010; 376: 103–11. Kawamori R, Tajima N, Iwamoto Y, Kashiwagi A, Shimamoto K, Kaku K. Voglibose for prevention of type 2 diabetes mellitus: a randomised, double-blind trial in Japanese individuals with impaired glucose tolerance. Lancet 2009; 373: 1607–14. Chiasson JL, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M. Acarbose for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial. Lancet 2002; 359: 2072–77. Perreault L, Pan Q, Mather KJ, Watson KE, Hammam RF, Kahn SE, for the Diabetes Prevention Program Research Group. Effect of regression from prediabetes to normal glucose regulation on long-term reduction in diabetes risk: results from the Diabetes Prevention Program Outcomes Study. Lancet 2012; published online June 9. DOI:10.1016/S01406736(12)60525-X. Weng J, Li Y, Xu W, et al. Effect of intensive insulin therapy on β-cell function and glycaemic control in patients with newly diagnosed type 2 diabetes: a multicentre randomised parallel-group trial. Lancet 2008; 371: 1753–60. Buse JB, Caprio S, Cefalu WT, et al. How do we define cure of diabetes? Diabetes Care 2009; 32: 2133–35.

2

3

4

5 6

7 8

9

10

11

12

13

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60828-9

Comment

Hypoglycaemia: a therapeutic concern in type 2 diabetes
Although the Diabetes Control and Complications Trial showed clearly that hypoglycaemia is a limiting factor for control of type 1 diabetes,1 hypoglycaemia had been largely dismissed as a concern for therapy of type 2 diabetes. Studies have provided evidence that hypoglycaemia indeed limits the glycaemic management of patients with type 2 diabetes, is a substantial, independent cause of excess morbidity and mortality, and exacerbates the costs of type 2 diabetes to the patient and to employers and society as a whole.2–9 In ADOPT (A Diabetes Outcome Progression Trial),2 4360 newly diagnosed patients with type 2 diabetes and no excess cardiovascular risk factors were randomly assigned to initial therapy with a sulphonylurea (glyburide), metformin, or rosiglitazone. Over 4 years, 39% of patients assigned to sulphonylurea, compared with 9% of patients assigned rosiglitazone and 11% of those assigned metformin, had one or more episodes of hypoglycaemia, leading to a 15% greater discontinuation rate in the sulphonylurea group and potentially contributing to the lowest durability for sulphonylureas throughout the study period. More recently, the findings of excess mortality in intensively treated patients in the glucose control group of the ACCORD (Action to Control Cardiovascular Risk in Diabetes) trial3 have generated considerable interest. In this trial of intensive glucose control in patients with increased coronary heart disease risk, with antidiabetic therapies of the investigators’ choosing, most patients were given both sulphonylureas and basal insulin together. Although severe hypoglycaemia increased mortality by two times to four times in the intensive and conventional groups, respectively, hypoglycaemia was not conclusively identified as the cause for this excess mortality in the intensive group.4 A similar trial of intensive glycaemic control in type 2 diabetes with a sulphonylurea—gliclazide—without concomitant insulin in the intensive group also reported a link between hypoglycaemia and increased cardiovascular events and total mortality.5,6 Zoungas and colleagues6 have reported that, in the ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation) trial, minor hypoglycaemia occurred in 44·7% of 11 140 patients and severe hypoglycaemia occurred in 2·7% of intensively treated patients (glycated
www.thelancet.com Vol 379 June 16, 2012

haemoglobin A1c 6·5%) and 1·5% of conventionally treated patients (glycated haemoglobin A1c 7·5%). The hazard ratios for cardiovascular events and for total mortality arising from severe hypoglycaemia were 2·88 and 2·69, respectively. Because the interval between the reported hypoglycaemic event and the subsequent cardiovascular or fatal event in ADVANCE was variable and not immediate, hypoglycaemia might be a marker for patient-related disorders such as autonomic neuropathy or advanced age, which could predispose patients to an excess risk of coronary heart disease. Morbidity from hypoglycaemia has more to it than is shown by cardiovascular disease risk alone. Budnitz and colleagues7 reported a survey of emergency room visits with or without subsequent hospital admission for adverse drug reactions in 12 666 cases of older adults collected from a national adverse drug reaction database. They noted that antidiabetic drugs were the second most prevalent drug category; 94·6% of reactions were hypoglycaemic in nature. These findings extrapolated to 54 000 total emergency room visits per year caused by adverse reactions to antidiabetic therapies in the USA, of which nearly 23 000 led to a hospital stay. In this survey, insulin accounted for 13·9% of overall hospital admissions from adverse drug reactions, and 10·7% from oral antidiabetic agents. Insight into the costs of hypoglycaemia, even mild hypoglycaemia, can be obtained from Brod and colleagues’ internet survey of patients with diabetes.8,9 The 6756 respondents were from France, Germany, the UK, and the USA. Of the patients with type 2 diabetes, more than a quarter reported non-severe hypoglycaemia every week, if not more frequently. These episodes caused 8·3–15·9 h lost from work per episode. Further study investigated the effect of nocturnal hypoglycaemia.9 Of the patients with type 2 diabetes, 31% of non-severe hypoglycaemic episodes were noted to be nocturnal in occurrence. These nocturnal episodes produced a rate of absenteeism from work of greater than 28%. Thus, hypoglycaemia in type 2 diabetes results in substantial lost time from work, and productivity might suffer even if the worker is not absent. In an era of many therapeutic choices for management of type 2 diabetes, hypoglycaemia clearly no longer

Islet of Langerhans

2215

Science Photo Library

Comment

needs to be tolerated as an unavoidable result of the therapy selected. Incretin-based therapies (dipeptidyl peptidase-4 inhibitors and glucagon-like peptide-1 receptor agonists) do not increase hypoglycaemia or bodyweight and should be considered in those patients for whom metformin alone is inadequate, and in those at risk from hypoglycaemia; although less often used in practice, bile-acid-binding resins (colesevelam) and centrally acting agents (bromocriptine) could also be considered in this scenario.10 The same can be said for the newer insulin analogues rather than conventional human insulin. However, these newer agents have not been shown to reduce microvascular complications as have the older, established therapies such as metformin and sulphonylureas. Evidence has proven that the perception of hypoglycaemia having little effect in type 2 diabetes is unfortunate and incorrect. Hypoglycaemia itself produces or is associated with excess costs, lost productivity, and excess morbidity and mortality. These results can be minimised by appropriate therapeutic choices, which should be individually adapted to each patient’s needs and should improve patient adherence to drugs and outcomes. Although therapeutic choices vary in cost, increased acquisition costs can be offset by decreased emergency room and hospital costs. The goal for each patient should be to produce an optimum treatment response, which includes minimisation of hypoglycaemia.

Alan J Garber
Departments of Medicine, Molecular and Cellular Biology, Biochemistry, and Molecular Biology, Baylor College of Medicine, Houston, TX 77030, USA [email protected]
I have served as a consultant for Novo Nordisk, Daiichi-Sankyo, Merck, Takeda, Santarus, LipoScience, Boehringer Ingelheim, Sekris, Lexicon, and Halozyme, participated in speakers’ bureaux for Merck, Novo Nordisk, Santarus, and Daiichi-Sankyo, and provided expert testimony on behalf of Novo Nordisk. I am on the board of directors of the American Association of Clinical Endocrinologists. 1 The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993; 329: 977–86. Kahn SE, Haffner SM, Heise MA, et al. Glycemic durability of rosiglitazone, metformin, or glyburide monotherapy. N Engl J Med 2006; 355: 2427–43. The Action to Control Cardiovascular Risk in Diabetes Study Group. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008; 358: 2545–59. Bonds DE, Miller ME, Bergenstal R, et al. The association between severe symptomatic hypoglycaemia and mortality in type 2 diabetes: retrospective analysis from the ACCORD Study. BMJ 2010; 340: 4909–18. The ADVANCE Collaborative Group. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med 2008; 358: 2560–72. Zoungas S, Patel A, Chalmers J, et al. Severe hypoglycemia and risks of vascular events and death. N Engl J Med 2010; 363: 1410–18. Budnitz DS, Lovegrove MC, Shehab N, Richards CL. Emergency hospitalizations for adverse drug events in older Americans. N Engl J Med 2011; 365: 2002–12. Brod M, Christensen T, Thomsen TL, Bushnell DM. The impact of non-severe hypoglycemic events on work productivity and diabetes management. Value Health 2011; 14: 665–71. Brod M, Christensen T, Bushnell DM. Impact of nocturnal hypoglycemic events on diabetes management, sleep quality,and next-day function: results from a four-country survey. J Med Econ 2012; 15: 77–86. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach. Position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2012; 35: 1364–79.

2 3

4

5

6 7

8

9

10

Is regression to normoglycaemia clinically important?
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60828-9 See Articles page 2243

Impaired fasting glucose, impaired glucose tolerance, and type 2 diabetes are metabolic conditions characterised by raised blood glucose concentrations that put people at risk of various serious health outcomes. These concentrations generally rise with time so that patients with impaired fasting glucose or glucose tolerance progress to diabetes,1 and patients with diabetes need progressively more complex regimens to manage glucose concentrations. Several large randomised controlled trials have shown that diabetes can be prevented or delayed in people with impaired fasting glucose or impaired glucose tolerance by diet, physical activity, and various drugs. Less well understood is the clinical significance of regression of these conditions to normoglycaemia, which has been noted in several

trials2–10 (table). Does regression predict a reduced risk of incident metabolic and vascular diseases, or does it simply reflect glucose fluctuations over time? In The Lancet, Leigh Perreault and colleagues11 describe epidemiological analyses of participants with impaired glucose tolerance who completed the Diabetes Prevention Program (DPP) trial without developing diabetes, and were passively followed up within the Diabetes Prevention Program Outcomes Study for up to 6 years. The investigators found that, among 894 people who had at least one normal oral glucose tolerance test during DPP, there was a 56% lower relative risk of developing diabetes during passive follow-up than among 1096 people who never regressed to normoglycaemia during DPP (hazard ratio [HR] 0·44,
www.thelancet.com Vol 379 June 16, 2012

2216

Comment

Diabetes quality improvement: beyond glucose control
In The Lancet, Andrea Tricco and colleagues1 report their comprehensive systematic review and meta-analysis of quality improvement (QI) strategies in diabetes care; results from 94 randomised controlled trials were included, along with findings from 48 clusterrandomised trials. This systematic review updates and expands on previous reports2 by including not only outcomes for glycated haemoglobin (HbA1c), but also vascular risk management, monitoring of microvascular complications, and smoking cessation in patients with diabetes. The inclusion of cardiovascular outcomes and smoking cessation is important because of the growing recognition that blood glucose control alone is not adequate to prevent both the microvascular and macrovascular complications of diabetes.3–5 Because diabetes care is no longer glucose-centric, it is crucial that QI efforts represent a broader view. Evidence shows that people with diabetes are disproportionately affected by eye and renal disease, nontraumatic amputations, and cardiovascular disease,6 which result in significant health-care costs and morbidity. At present, only one person in eight with diabetes has their disease controlled to the respective goals for HbA1c, LDL cholesterol, and blood pressure,7 which are the key risk factors for diabetes complications. Strategies to improve HbA1c titres (both adequately and inadequately controlled) are clearly delineated by the evidence in the report—a mean reduction of 0·37% in HbA1c (95% CI 0·28–0·45) was achieved, for example, with a larger effect for baseline HbA1c above 8%.1 Yet the results were disappointing for cardiovascular outcomes, use of statins, hypertension control, and smoking cessation, with no significant differences noted. One partial explanation could be the few studies adopting a systems approach—ie, elements working together synergistically at the individual, health system, and community levels to improve outcomes—in their diabetes QI efforts. Electronic medical records for population-based care, promotion of self-management, and the use of standardised guidelines for both care and self-management education are central and crucial to improving diabetes care and QI efforts. Studies that intervened on the entire system were associated with the largest effects on outcomes, independent of HbA1c levels,1 making the case that QI strategies designed to optimise the entire system of care should be included in programmes to improve diabetes management. Although traditional diabetes care is based on an acute-care model that is fragmented and illness focused, diabetes clinical practice is beginning to change in a way that has important implications for care delivery and ultimately QI efforts. This movement towards a teambased model is based on outcomes like those shown in the accompanying meta-analysis1 and those reported elsewhere.8–10 Investigators showed that clinical (HbA1c, blood pressure, and non-HDL cholesterol), behavioural (self-monitoring of blood glucose), and psychosocial outcomes (empowerment and wellbeing) improved and were sustained after a multilevel randomised controlled trial by incorporating a diabetes nurse educator into primary-care practices within the context of the chroniccare model.8,9 Working closely with physicians, health professionals such as diabetes nurse educators, diabetes specialist nurses, and pharmacists not only have the knowledge and skills to manage drugs through the use of evidence-based algorithms to adjust diabetes, blood pressure, and cholesterol drugs,11 but also have the knowledge and skills to counsel patients effectively in the behavioural and psychosocial aspects of diabetes self management, including complex drug regimens, and can provide counselling and referral for smoking cessation.12 Full use of the skills of the entire team of health-care professionals facilitates improvements in both clinical care and self care—both of which are
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60637-0 See Online/Articles DOI:10.1016/S01406736(12)60480-2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60637-0

1

Science Photo Library

Comment

needed to improve and sustain outcomes throughout a lifetime of chronic disease. With the development and implementation of the patient-centred medical home, diabetes QI efforts that go beyond glycaemic control will probably receive even greater attention. Electronic medical records for population-based care, promotion of self-management, and use of standardised guidelines are central and crucial to improving diabetes care and QI efforts. Redesigning care to use the skills of all members of the health-care team might provide both the impetus and the ability for patients and practices to move beyond glycaemic control to create a comprehensive, patient-centred, and effective model of diabetes care. *Martha M Funnell, Gretchen A Piatt
Department of Medical Education and the Michigan Diabetes Research and Training Center, University of Michigan Medical School, Ann Arbor, MI 48109-5201, USA [email protected]
We declare that we have no conflicts of interest. 1 Tricco AC, Ivers NM, Grimshaw JM, et al. Effectiveness of quality improvement strategies on the management of diabetes: a systematic review and meta-analysis. Lancet 2012; published online June 9. DOI:10.1016/S0140-6736(12)60480-2.

2

3

4

5 6

7

8

9

10 11 12

Shojania KG, Ranji SR, McDonald KM, et al. Effects of quality improvement strategies for type 2 diabetes on glycemic control: a meta-regression analysis. JAMA 2006; 296: 427–40. Snow V, Weiss KB, Mottur-Pilson C. The evidence base for tight blood pressure control in the management of type 2 diabetes mellitus. Ann Intern Med 2003; 138: 587–92. Baigent C, Keech A, Kearney PM, et al. Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90 056 participants in 14 randomised trials of statins. Lancet 2005; 366: 1267–78. Haire-Joshu D, Glasgow RE, Tibbs TL. Smoking and diabetes (technical review). Diabetes Care 1999; 22: 1887–98. CDC. National diabetes fact sheet: general information and national estimates on diabetes in the United States. Atlanta, GA: Department of Health and Human Services, Centers for Disease Control and Prevention, 2009. Cheung BM, Ong KL, Cherny SS, Sham PC, Tso AW, Lam KS. Diabetes prevalence and therapeutic target achievement in the United States, 1999 to 2006. Am J Med 2009; 122: 443–53. Piatt G, Orchard T, Emerson S, et al. Translating the chronic care model into the community: results from a randomized controlled trial of a multifaceted diabetes care intervention. Diabetes Care 2006; 29: 811–17. Piatt G, Anderson R, Brooks M, et al. 3-year follow-up of clinical and behavioral improvements following a multifaceted diabetes care intervention: results of a randomized controlled trial. Diabetes Educ 2010; 36: 301–09. Coleman K, Austin B, Brach C, Wagner E. Evidence on the chronic care model in the new millennium. Health Affairs 2009; 28: 75–85. Davidson MB. Effect of nurse-directed diabetes care in a minority population. Diabetes Care 2003; 26: 2281–87. Anderson RM, Funnell MM, Aikens JE, et al. Evaluating the efficacy of an empowerment-based self-management consultant intervention: results of a two-year randomized controlled trial. Ther Patient Educ 2009; 1: 3–11.

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60637-0

Comment

Type 2 diabetes: which drug as add-on to metformin?
Type 2 diabetes is a progressive disease defined by hyperglycaemia and is associated with hypertension, dyslipidaemia, and an increased risk of cardiovascular events in most patients. The stepwise approach to management of type 2 diabetes starts with lifestyle interventions plus one or more antidiabetic drugs (figure).1–5 Most clinical guidelines indicate that metformin should be the first-line drug, but no consensus exists about additional use of a sulphonylurea, a dipeptidyl peptidase-4 inhibitor, pioglitazone, an α-glucosidase inhibitor, a GLP-1 receptor agonist, or insulin when glucose control deteriorates over time.1–5 Ideally, therapeutic decisions should be based on long-term clinical studies, but few studies have compared antidiabetic drugs for durability of effectiveness on glycaemic control, cost, quality of life, and effect on late diabetic complications, including cardiovascular disease.1–5 Instead, physicians choose the second drug on the basis of the effectiveness of reductions in glycated haemoglobin (HbA1c) concentration; safety; cost; expected properties in a particular patient, such as effect on weight; and preferences and tolerances of each patient.1–5 In The Lancet, Baptist Gallwitz and colleagues6 report the results of the European Exenatide (EUREXA) trial, which compared the GLP-1 receptor agonist exenatide twice daily with glimepiride, a sulphonylurea, once daily as add-on to metformin for up to 234 weeks, in overweight to obese patients with type 2 diabetes inadequately controlled by metformin alone. The study is so far the longest randomised trial to compare a GLP-1 receptor agonist with a commonly used treatment in type 2 diabetes. 203 (41%) of 490 patients had treatment failure in the exenatide group compared with 262 (54%) of 487 in the glimepiride group (hazard ratio 0·748, 95% CI 0·623–0·899). 221 (45%) patients in the exenatide group and 151 (31%) in the glimepiride group achieved an HbA1c concentration of less than 7% (p<0·0001). Patients in the exenatide group had a weight loss of 3·32 kg compared with those in the glimepiride group, who gained 1·15 kg (p<0·0001). Significantly fewer patients in the exenatide group than in the glimepiride group reported documented symptomatic (p<0·0001), nocturnal (p=0·007), and non-nocturnal (p<0·0001) hypoglycaemia. Discontinuation because of adverse events, which were mainly gastrointestinal, was significantly higher (p=0·0005) in the exenatide group than in the glimepiride group. One case of pancreatitis was reported in each study group. Overall, this 3-year follow-up study indicates that exenatide twice daily is more effective than glimepiride in terms of several clinical parameters related to treatment of patients with type 2 diabetes. Strengths of the study are the long-term follow-up and the comparison between the frequently used glimepiride and a GLP-1 receptor agonist. In the 26-week Liraglutide Effect and Action in Diabetes (LEAD)-2 study,7 liraglutide once daily induced similar glycaemic control, reduced bodyweight, and lowered the risk of hypoglycaemia compared with glimepiride when used as add-on to metformin. Physicians are divided in their opinions about the safety and effectiveness of sulphonylureas, which potentiate insulin secretion and lead to a rapid improvement in glycaemic control, but result in faster deterioration of glycaemic control than do metformin or a glitazone.1–5 The adverse effects of sulphonylureas are predictable, with most patients gaining weight, and risk of hypoglycaemia is higher than for metformin because the effect on insulin secretion is not glucose dependent.1–5 Studies linking sulphonylureas with cardiovascular adverse effects are based on databases
Diagnosis, lifestyle intervention, and metformin Metformin is widely accepted (if tolerated) as the first-line drug Published Online June 9, 2012 DOI:10.1016/S01406736(12)60769-7 See Online/Articles DOI:10.1016/S01406736(12)60479-6

HbA1c ≥6·5–7·0%

Add exenatide twice daily Pros • Effectiveness • No hypoglycaemia • Weight loss Cons • Gastrointestinal side-effects • No cardiovascular randomised controlled endpoint studies • High cost • Risk of pancreatitis?

Add sulphonylurea Pros • Effectiveness in short term • Cardiovascular randomised controlled endpoint studies • Low cost Cons • Low durability • Hypoglycaemia • Weight gain • Risk of cardiovascular disease?

Figure: Pros and cons of a sulphonylurea or GLP-receptor agonist as add-on to metformin for preventing deterioration in glycaemic control in patients with type 2 diabetes

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60769-7

1

Comment

rather than on randomised trials, and the UK Prospective Diabetes Study8 did not note an increased mortality risk for patients given sulphonylureas. As a drug class, GLP-1 receptor agonists improve glycaemia by stimulating insulin secretion and the inhibition of glucagon release, but only when glucose concentrations are raised, thus conferring a lower risk of hypoglycaemia than that noted with sulphonylureas.9–11 Moreover, exenatide twice daily reduces postprandial glucose excursions by delaying gastric emptying.10,11 GLP-1 receptor agonists induce weight loss in most patients, but are associated with gastrointestinal sideeffects9–11 and have been linked to pancreatitis, although with conflicting conclusions from the clinical controlled trials and use of different databases.12 These drugs display cardioprotection and reduce blood pressure and markers of inflammation, but increase heart rates.12 Analyses of phase 2 and phase 3 trials with exenatide twice daily versus placebo or insulin showed no evidence of cardiovascular harm with exenatide. Additionally, a retrospective analysis13,14 of cardiovascular events using the LifeLink database from 2005 to 2009 showed that patients given exenatide twice daily were significantly less likely to have a cardiovascular event (p=0·01) or cardiovascular-related hospital admission (p=0·02) than were those given other glucose-lowering drugs. After the lesson learned from rosiglitazone,15 the US Food and Drug Administration now requires the assessment of cardiovascular risks of new diabetic drugs both before and after approval, and results of cardiovascular outcome studies for the different GLP-1 receptor agonists are expected after 2015. Sten Madsbad
Department of Endocrinology, Hvidovre Hospital and University of Copenhagen, 2650 Hvidovre, Denmark [email protected]
I have been a consultant or adviser to Novartis Pharma, Novo Nordisk, Merck Sharp and Dohme, Sanofi-Aventis, AstraZeneca, Johnson and Johnson, Roche, Mannkind, Boehringer-Ingelheim, Zeeland, Lilly, and Intarcia Therapeutics, and have received fees for speaking from Novo Nordisk, Merck Sharp and Dohme, Johnson and Johnson, Roche, Schering-Plough, Sanofi-Aventis, Novartis Pharma, Lilly, Bristol-Myers Squibb, and AstraZeneca.

1

2

3

4

5

6

7

8

9 10

11

12

13

14

15

Bennett WL, Maruthur NM, Singh S, et al. Comparative effectiveness and safety of medications for type 2 diabetes: an update including new drugs and 2-drug combinations. Ann Intern Med 2011; 154: 602–13. Bennett WL, Odelola OA, Wilson LM, et al. Evaluation of guideline recommendations on oral medications for type 2 diabetes mellitus: a systematic review. Ann Intern Med 2012; 156: 27–36. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach. Position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2012; published online April 19. DOI:10.2337/dc12-0413. Qaseem A, Humphrey LL, Sweet DE, Starkey M, Shekelle P. Oral pharmacologic treatment of type 2 diabetes mellitus: a clinical practice guideline from the American College of Physicians. Ann Intern Med 2012; 156: 218–31. Schernthaner G, Barnett AH, Betteridge DJ, et al. Is the ADA/EASD algorithm for the management of type 2 diabetes (January 2009) based on evidence or opinion? A critical analysis. Diabetologia 2010; 53: 1258–69. Gallwitz B, Guzman J, Dotta F, et al. Exenatide twice daily versus glimepiride for prevention of glycaemic deterioration in patients with type 2 diabetes with metformin failure (EUREXA): an open-label, randomised controlled trial. Lancet 2012; published online June 9. DOI:10.1016/S01406736(12)60479-6. Nauck M, Frid A, Hermansen K, et al. Efficacy and safety comparison of liraglutide, glimepiride, and placebo, all in combination with metformin, in type 2 diabetes: the LEAD (liraglutide effect and action in diabetes)-2 study. Diabetes Care 2009; 32: 84–90. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000; 321: 405–12. Holst JJ. The physiology of glucagon-like peptide 1. Physiol Rev 2007; 87: 1409–39. Madsbad S, Krarup T, Deacon CF, Holst JJ. Glucagon-like peptide receptor agonists and dipeptidyl peptidase-4 inhibitors in the treatment of diabetes: a review of clinical trials. Curr Opin Clin Nutr Metab Care 2008; 11: 491–99. Madsbad S, Kielgast U, Asmar M, Deacon CF, Torekov SS, Holst JJ. An overview of once-weekly glucagon-like peptide-1 receptor agonists— available efficacy and safety data and perspectives for the future. Diabetes Obes Metab 2011; 13: 394–407. Drucker DJ, Sherman SI, Bergenstal RM, Buse JB. The safety of incretin-based therapies—review of the scientific evidence. J Clin Endocrinol Metab 2011; 96: 2027–31. Best JH, Hoogwerf BJ, Herman WH, et al. Risk of cardiovascular disease events in patients with type 2 diabetes prescribed the glucagon-like peptide 1 (GLP-1) receptor agonist exenatide twice daily or other glucose-lowering therapies: a retrospective analysis of the LifeLink database. Diabetes Care 2011; 34: 90–95. Ratner R, Han J, Nicewarner D, Yushmanova I, Hoogwerf BJ, Shen L. Cardiovascular safety of exenatide BID: an integrated analysis from controlled clinical trials in participants with type 2 diabetes. Cardiovasc Diabetol 2011; 10: 22. Goldfine AB. Assessing the cardiovascular safety of diabetes therapies. N Engl J Med 2008; 359: 1092–95.

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60769-7

Comment

Choosing a blood-glucose-lowering agent after metformin
In type 2 diabetes, early and effective treatment of hyperglycaemia is important in prevention of chronic complications.1 However, progressive decline of β-cell function leads to deterioration of glycaemic control, necessitating intensification of blood-glucose-lowering therapy during the course of the disease.2 Existing treatment algorithms advocate stepwise escalation of therapy, starting with metformin and subsequently adding other oral agents or insulin when glycated haemoglobin A1c (HbA1c) exceeds the treatment target.3,4 To date, however, there is insufficient evidence to guide clinicians in choice of the second agent after metformin.3–5 A recent analysis suggests that, after metformin treatment failure, less than half of patients with type 2 diabetes reach an HbA1c lower than 7%, irrespective of the second-line agent.6 Bayesian network meta-analysis showed that glucagon-like peptide-1 (GLP-1) receptor agonists and biphasic and basal insulin were most efficacious in lowering HbA1c after metformin failure.7 An even earlier start of insulin in type 2 diabetes has been advocated, because intensive insulin therapy immediately after diagnosis was shown to normalise blood glucose, preserve β-cell function, and induce disease remission.8 In clinical practice, however, insulin treatment is often postponed because of barriers perceived by patients and health-care providers to initiation of insulin.9 In these cases, various combinations of oral agents are prescribed, often at the cost of good glycaemic control. Since sulphonylurea use is associated with early treatment failure, weight gain, and hypoglycaemia risk, dipeptidyl peptidase-4 (DPP-4) inhibitors are increasingly used. Presently, whether these agents can sustain glycaemic control in the long term and improve outcomes in patients with type 2 diabetes is unknown.10,11 In The Lancet, Pablo Aschner and colleagues12 report a 26-week, multinational, open-label randomised trial comparing the HbA1c-lowering effect of insulin glargine versus sitagliptin, a DPP-4 inhibitor, added to ongoing metformin monotherapy, in patients with type 2 diabetes failing metformin treatment. The primary endpoint was difference in HbA1c. From a baseline HbA1c of 8·5%, insulin glargine resulted in a greater reduction than did sitagliptin (–1·72% vs –1·13%, mean adjusted difference –0·59%, 95% CI –0·77 to –0·42), with significantly more patients in the insulin glargine group reaching the prespecified HbA1c targets of both 7% and 6·5% than in the sitagliptin group. A small increase in bodyweight was noted with insulin glargine use (0·44 kg), whereas sitagliptin decreased bodyweight by a mean 1·08 kg (adjusted mean difference 1·51 kg, 95% CI 0·93–2·09). The number of symptomatic as well as severe hypoglycaemic episodes was greater with insulin glargine than with sitagliptin treatment. On the basis of these results, the authors concluded that, in patients with type 2 diabetes uncontrolled on metformin, insulin glargine might be more effective than sitagliptin in reducing HbA1c, with low rates of severe hypoglycaemia and mild weight gain. Although these results are not entirely unexpected, the first comparison of a DPP-4 inhibitor with basal insulin in metformin-treated patients with type 2 diabetes is undisputedly a strong point. This study is timely in the view of the gaps in the treatment algorithm and adds to the evidence base to guide individualisation of therapy in type 2 diabetes, as advocated in the most recent joint position statement of the European Association for the Study of Diabetes and American Diabetes Association.4 However, in addition to assessing surrogate outcome markers and satisfying potential commercial aims, longer-term overarching goals should drive the undertaking of elaborate and costly drug intervention trials, and their design should be adjusted accordingly.
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60780-6 See Online/Articles DOI:10.1016/S01406736(12)60439-5

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60780-6

1

Science Photo Library

Comment

Most patients and caregivers prefer to postpone insulin treatment as long as possible.9 Thus, the overarching goal could be to assess whether a DPP-4 inhibitor, which improves β-cell function in human beings, can provide sustained glycaemic control in a subgroup of patients not responding to metformin or a combination of metformin and sulphonylurea, thus postponing the need for insulin treatment for a prolonged time—eg, several years, which could be regarded as a relevant benefit. Assessment of durable effects of early initiation of insulin glargine, which defer use of prandial insulin, could be another overarching goal. These much hoped for durable effects, as well as the underlying mechanisms, can only be studied in sufficiently large and long-term trials with an appropriate design. Such an approach would provide a better rationale to evaluate a typical early-step oral drug and injectable final-step insulin therapy in a headto-head comparison. Therefore, the present study by Aschner and colleagues,12 with its very short duration and design, can only confirm the observations by others showing that addition of basal insulin, when dosed properly, to ongoing metformin monotherapy is more efficacious in lowering HbA1c than any currently available oral agent. To justify early initiation of basal insulin therapy, which is viewed by many as the last irreversible option in type 2 diabetes treatment, posing substantially greater burden on patients than oral therapies, arguments of benefit that go beyond glucose control need to be provided. Therefore, the use of composite endpoints, additionally encompassing favourable effects on bodyweight, hypoglycaemia risk, blood pressure, and lipids, all of which might improve patients’ cardiovascular risk, has been advocated to enable evaluation of an agent’s true benefit.4,13 At present, it is unclear whether initiation of basal insulin in the early stage of type 2 diabetes translates into longer term outcome benefits or whether its early use is ultimately offset by progressive weight gain and more hypoglycaemic events, both of which could result in increased cardiovascular risk, higher cost, and poor quality of life. The anticipated results of the Outcome Reduction with an Initial Glargine Intervention (ORIGIN) trial, comparing insulin glargine treatment in patients with dysglycaemia and early type 2 diabetes with usual

care for more than 7 years, will hopefully provide some answers to these questions.14 Michaela Diamant
Diabetes Centre, VU University Medical Centre, 1081 HV Amsterdam, The Netherlands [email protected]
I serve on advisory boards for Abbott, Eli Lilly, Merck Sharp & Dohme (MSD), Novo Nordisk, and Poxel Pharma; am a consultant for Sanofi; and a speaker for Eli Lilly, MSD, and Novo Nordisk. Through me the VU University Medical Center receives research grants from Amylin/Eli Lilly, MSD, Novo Nordisk, and Sanofi. I receive no personal payments in connection with any of these activities, payments being transferred to an institutional research foundation. 1 Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med 2008; 359: 1577–89. UKPDS Group. Glycaemic control with diet, sulfonylurea, metformin, or insulin in patients with type 2 diabetes mellitus. Progressive requirement for multiple therapies (UKPDS 49). JAMA 1999; 281: 2005–10. Nathan DM, Buse JB, Davidson MB, et al, American Diabetes Association; European Association for Study of Diabetes. Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2009; 32: 193–203. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach. Position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2012; published online April 19. DOI:10.2337/dc12-0413. Qaseem A, Humphrey LL, Sweet DE, Starkey M, Shekelle P, for the Clinical Guidelines Committee of the American College of Physicians. Oral pharmacological treatment of type 2 diabetes mellitus: a clinical practice guideline from the American College of Physicians. Ann Intern Med 2012; 156: 218–31. Esposito K, Bellastella G, Giugliano D. When metformin fails in type 2 diabetes mellitus. Arch Intern Med 2011; 171: 365–66. Liu SC, Tu YK, Chien MN, Chien KL. Effect of antidiabetic agents added to metformin on glycemic control, hypoglycemia and weight change in patients with type 2 diabetes: a network meta-analysis. Diabetes Obes Metab 2012; published online Apr 9. DOI:10.1111/j.1463-1326.2012.01606.x. Weng J, Li Y, Xu W, et al. Effect of intensive insulin therapy on β-cell function and glycaemic control in patients with newly diagnosed type 2 diabetes: a multicentre randomised parallel-group trial. Lancet 2008; 371: 1753–60. Kunt T, Snoek FJ. Barriers to insulin initiation and intensification and how to overcome them. Int J Clin Pract Suppl 2009; 63 (suppl 164): 6–10. van Genugten RE, van Raalte DH, Diamant M. Dipeptidyl peptidase-4 inhibitors and preservation of pancreatic islet-cell function: a critical appraisal of the evidence. Diabetes Obes Metab 2012; 14: 101–11. Fineman MS, Cirincione BB, Maggs D, Diamant M. GLP-1 based therapies: differential effects on fasting and postprandial glucose. Diabetes Obes Metab 2012; published online Jan 10. DOI:10.1111/j.1463-1326.2012.01560.x. Aschner PJ, Chan J, Owens DR, et al, on behalf of the EASIE investigators. Insulin glargine versus sitagliptin in insulin-naive patients with type 2 diabetes mellitus uncontrolled on metformin (EASIE): a multicentre, randomised, open-label trial. Lancet 2012; published online June 9. DOI:10.1016/S0140-6736(12)60439-5. Zinman B, Schmidt WE, Moses A, Lund N, Gough S. Achieving a clinically relevant composite outcome of an HbA1c of <7% without weight gain or hypoglycaemia in type 2 diabetes: a meta-analysis of the liraglutide clinical trial programme. Diabetes Obes Metab 2012; 14: 77–82. Origin Trial Investigators, Gerstein H, Yusuf S, Riddle MC, Ryden L, Bosch J. Rationale, design, and baseline characteristics for a large international trial of cardiovascular disease prevention in people with dysglycemia: the ORIGIN Trial (Outcome Reduction with an Initial Glargine Intervention). Am Heart J 2008; 155: 26–32.

2

3

4

5

6 7

8

9 10

11

12

13

14

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60780-6

Comment

As a drug class, GLP-1 receptor agonists improve glycaemia by stimulating insulin secretion and the inhibition of glucagon release, but only when glucose concentrations are raised, thus conferring a lower risk of hypoglycaemia than that noted with sulphonylureas.9–11 Moreover, exenatide twice daily reduces postprandial glucose excursions by delaying gastric emptying.10,11 GLP-1 receptor agonists induce weight loss in most patients, but are associated with gastrointestinal sideeffects9–11 and have been linked to pancreatitis, although with conflicting conclusions from the clinical controlled trials and use of different databases.12 These drugs display cardioprotection and reduce blood pressure and markers of inflammation, but increase heart rates.12 Analyses of phase 2 and phase 3 trials with exenatide twice daily versus placebo or insulin showed no evidence of cardiovascular harm with exenatide. Additionally, a retrospective analysis13,14 of cardiovascular events using the LifeLink database from 2005 to 2009 showed that patients given exenatide twice daily were significantly less likely to have a cardiovascular event (p=0·01) or cardiovascular-related hospital admission (p=0·02) than were those given other glucose-lowering drugs. After the lesson learned from rosiglitazone,15 the US Food and Drug Administration now requires the assessment of cardiovascular risks of new diabetic drugs both before and after approval, and results of cardiovascular outcome studies for the different GLP-1 receptor agonists are expected after 2015. Sten Madsbad
Department of Endocrinology, Hvidovre Hospital and University of Copenhagen, 2650 Hvidovre, Denmark [email protected]
I have been a consultant or adviser to Novartis Pharma, Novo Nordisk, Merck Sharp and Dohme, Sanofi-Aventis, AstraZeneca, Johnson and Johnson, Roche, Mannkind, Boehringer-Ingelheim, Zeeland, Lilly, and Intarcia Therapeutics, and

have received fees for speaking from Novo Nordisk, Merck Sharp and Dohme, Johnson and Johnson, Roche, Schering-Plough, Sanofi-Aventis, Novartis Pharma, Lilly, Bristol-Myers Squibb, and AstraZeneca. 1 Bennett WL, Maruthur NM, Singh S, et al. Comparative effectiveness and safety of medications for type 2 diabetes: an update including new drugs and 2-drug combinations. Ann Intern Med 2011; 154: 602–13. Bennett WL, Odelola OA, Wilson LM, et al. Evaluation of guideline recommendations on oral medications for type 2 diabetes mellitus: a systematic review. Ann Intern Med 2012; 156: 27–36. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach. Position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 2012; published online April 19. DOI:10.2337/dc12-0413. Qaseem A, Humphrey LL, Sweet DE, Starkey M, Shekelle P. Oral pharmacologic treatment of type 2 diabetes mellitus: a clinical practice guideline from the American College of Physicians. Ann Intern Med 2012; 156: 218–31. Schernthaner G, Barnett AH, Betteridge DJ, et al. Is the ADA/EASD algorithm for the management of type 2 diabetes (January 2009) based on evidence or opinion? A critical analysis. Diabetologia 2010; 53: 1258–69. Gallwitz B, Guzman J, Dotta F, et al. Exenatide twice daily versus glimepiride for prevention of glycaemic deterioration in patients with type 2 diabetes with metformin failure (EUREXA): an open-label, randomised controlled trial. Lancet 2012; published online June 9. DOI:10.1016/S0140-6736(12)60479-6. Nauck M, Frid A, Hermansen K, et al. Efficacy and safety comparison of liraglutide, glimepiride, and placebo, all in combination with metformin, in type 2 diabetes: the LEAD (liraglutide effect and action in diabetes)-2 study. Diabetes Care 2009; 32: 84–90. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000; 321: 405–12. Holst JJ. The physiology of glucagon-like peptide 1. Physiol Rev 2007; 87: 1409–39. Madsbad S, Krarup T, Deacon CF, Holst JJ. Glucagon-like peptide receptor agonists and dipeptidyl peptidase-4 inhibitors in the treatment of diabetes: a review of clinical trials. Curr Opin Clin Nutr Metab Care 2008; 11: 491–99. Madsbad S, Kielgast U, Asmar M, Deacon CF, Torekov SS, Holst JJ. An overview of once-weekly glucagon-like peptide-1 receptor agonists— available efficacy and safety data and perspectives for the future. Diabetes Obes Metab 2011; 13: 394–407. Drucker DJ, Sherman SI, Bergenstal RM, Buse JB. The safety of incretin-based therapies—review of the scientific evidence. J Clin Endocrinol Metab 2011; 96: 2027–31. Best JH, Hoogwerf BJ, Herman WH, et al. Risk of cardiovascular disease events in patients with type 2 diabetes prescribed the glucagon-like peptide 1 (GLP-1) receptor agonist exenatide twice daily or other glucose-lowering therapies: a retrospective analysis of the LifeLink database. Diabetes Care 2011; 34: 90–95. Ratner R, Han J, Nicewarner D, Yushmanova I, Hoogwerf BJ, Shen L. Cardiovascular safety of exenatide BID: an integrated analysis from controlled clinical trials in participants with type 2 diabetes. Cardiovasc Diabetol 2011; 10: 22. Goldfine AB. Assessing the cardiovascular safety of diabetes therapies. N Engl J Med 2008; 359: 1092–95.

2

3

4

5

6

7

8

9 10

11

12

13

14

15

An authority for crisis coordination and accountability
The demand for better coordination and control is heard during and after every major international disaster. We now have the potential framework to meet this demand and we should respond. The World Health Assembly altered WHO’s role in disasters after the outbreak of severe acute respiratory syndrome with the 2005 International Health Regulations
www.thelancet.com Vol 379 June 16, 2012

(IHR) Treaty.1 WHO changed from a mainly passive responder during short-term infectious disease crises to an unprecedented active authority with a mandate to address long-term prevention, preparedness, and response roles and responsibilities. This treaty obliges WHO to obtain expert advice on any declared public health emergency of international concern.

Published Online October 18, 2011 DOI:10.1016/S01406736(11)60979-3

2223

Comment

Additionally, National Focal Points should be identified to ensure a two-way channel of communication between WHO and its 194 member states, and countries are required to establish surveillance capacities and to share information relevant to public health risks. Finally, the treaty introduced a decision-instrument algorithm for the assessment and notification of events that might constitute a public health emergency of international concern. Although the IHR Treaty created political tensions because of the balance between sovereignty and trade concerns of individual nation states and the common good of the international community, it was eventually agreed on and adopted by all countries.1 The scope and timeliness of the implementation process proved to be a major accomplishment and complemented the existing role of WHO’s Health Action in Crises Cluster, which addressed increasingly wide and weighty international health emergencies. Undeniably, many nations still do not have the core capacities to detect, assess, and report risks, and the IHR Treaty is unable to enforce sanctions. Yet under the authority of the treaty, the response to H1N1 influenza was effective.1 Timely virus detection was achieved by the Global Influenza Surveillance Network, and resulted in effective partnering, interagency coordination, and rapid field deployment of experts and public health professionals. The development of first-candidate vaccine-seed strains and control reagents was achieved in a timely manner, as were early recommendations

Distribution of water after earthquake in Haiti, Jan 22, 2010

from a vulnerable-group analysis of surveillance data and the distribution of proper treatment courses to 72 countries. The IHR Review Committee admitted to some bureaucratic hiccups in May, 2011, but strongly supported an even larger increase in the number of fielded experts and in accelerated surveillance cooperation to help meet the 2012 core-capacity goals.1 Few could dispute that the treaty had functioned exactly as expected from a worldwide authority designed to mitigate the dire results of a threatening pandemic through international cooperation. The success of the IHR Treaty now opens the door of potential international cooperation wider and begs a larger question for the humanitarian community: can a similar model be introduced to guarantee the coordination of large-scale disasters and crises? Such a cooperative model is crucial to the provision of oversight, accountability, accreditation, and worldwide legitimacy that has been hitherto absent, which was painfully evident in the chaotic health response to the Haiti earthquake2 and to many previous major crises. Arguably, the treaty has to mature further and be broadened to encompass more than its exclusively epidemiological oversight to remain relevant: it should be incorporated into an agency with appropriate managerial skills and authorities to advance the positive aspects of the current humanitarian cluster model and universal standards of care.3 Historically, the first attempt at a broader mandated authority was addressed by the former UN Department of Humanitarian Affairs. This body was stripped of its short-lived operational responsibilities to avoid being seen as a competitor to UN field agencies and non-governmental organisations, and summarily disappeared in 1997 to become the Office for the Coordination of Humanitarian Affairs (OCHA). OCHA has many excellent disaster managers, but unfortunately has always been under-resourced, underfunded, and unable to compete with the dominant, often military-led resources and contractors who generally respond to health disasters. OCHA does not have the international power required of any authority like the IHR Treaty that is necessary for the reform of the UN Cluster System2 and in crises with major health effects.4 In view of the present movement toward a “blueprint for professionalizing humanitarian assistance”5 and
www.thelancet.com Vol 379 June 16, 2012

Reuters

2224

Comment

92% of humanitarians willing to be professionalised6 under a set of core competencies,7 the subject of an international authority should be urgently readdressed. Such an authority is needed to: guarantee a stable and strategic return to root-cause remediation and development; introduce universal standards at every level of response, prevention, and preparedness for the inevitable direct and indirect results that cause and compound catastrophic public health emergencies; and endorse a process for the accreditation and accountability of providers. Despite the enormous implications and challenges to general public health practice and policy worldwide, the need for this authority can no longer be rationally questioned or ethically denied. The encouragement of the IHR Treaty initiative and the demands of health generally should embolden that resolve. *Frederick M Burkle Jr, Anthony D Redmond, Dudley F McArdle
Harvard Humanitarian Initiative, Harvard School of Public Health, Cambridge, MA 02138, USA (FMB Jr); Woodrow Wilson International Centre for Scholars, Washington, DC, USA (FMB Jr); Manchester Medical School and Humanitarian and Conflict Response Institute, University of Manchester, Manchester, UK (ADR); and Monash University, Melbourne, VIC, Australia (DFM) [email protected]

DFM served as senior advisor to Assistant Director General, Health Action in Crises, WHO, from 2007 to 2009. The other authors declare that they have no conflicts of interest. 1 WHO. Implementation of the International Health Regulations (2005): report on the review committee on the functioning of the International Health Regulations (2005) in relation to pandemic (H1N1) 2009. May 5, 2011. http://apps.who.int/gb/ebwha/pdf_files/WHA64/A64_10-en.pdf (accessed June 13, 2011). Inter-Agency Standing Committee, Global Health Cluster. Concept paper: foreign medical teams. May 17, 2011. http://www.who.int/hac/global_ health_cluster/about/policy_strategy/fmt_concept_paper_16may11.pdf (accessed June 26, 2011). Institute of Medicine. Guidance for establishing crisis standards of care for use in disaster situations: a letter report. 2009. http://books.nap.edu/ openbook.php?record_id=12749&page=R1 (accessed June 13, 2011). Office for the Coordination of Humanitarian Affairs. OCHA in 2011: annual plan and budget. http://ochaonline.un.org/ocha2011/OCHA2011_ jpg2000_200dpi.pdf (accessed June 26, 2011). Walker P, Hein K, Russ C, Bertleff G, Caspersz D. A blueprint for professionalizing humanitarian assistance. Health Aff (Millwood) 2010; 29: 2223–30. Walker P, Russ C. Professionalising the humanitarian sector: a scoping study. April, 2010. http://www.elrha.org/uploads/Professionalising_the_ humanitarian_sector.pdf (accessed June 15, 2011). Hein K. The competency of competencies. Prehosp Disaster Med 2010; 25: 396–97.

2

3

4

5

6

7

www.thelancet.com Vol 379 June 16, 2012

2225

World Report

Europe to boost development of new antimicrobial drugs
A European public-private partnership has launched a €224 million research initiative to speed up the development of much-needed new antimicrobial drugs. Tony Kirby reports.
There are several reasons why development of new antimicrobial drugs has stalled in Europe (and in fact worldwide). Among them is the difficulty to develop antibiotics targeting pathogens endowed with antimicrobial resistance. In addition, the regulatory environment within Europe is such that new drugs need to show non-inferiority in clinical trials within a margin of 10% as compared with current gold standards; this is quite challenging since many participants in a clinical trial may not be infected with the bacteria being targeted by the new drug. As these drugs are taken for limited courses not for life, this also restricts possible profitability. These factors have coalesced to make development of new antimicrobials a largely unattractive prospect for drug companies. Antibiotic-resistant infections kill around 25 000 patients in Europe each year and represent a global cost in the European Union (EU) of around €1·5 billion (US$1·85 billion) per year. Yet as the pipeline for new antimicrobials has slowed to a crawl, so have levels of resistance to existing drugs increased. The problem of antimicrobial resistance is particularly acute for Gram-negative bacteria. In addition, meticillin-resistant Staphylococcus aureus (MRSA) remains a major threat worldwide as a hospitalacquired pathogen. Fortunately, the rates of MRSA are decreasing in hospitals in some EU Member States, as shown by the latest data from European Centre for Disease Control. But in the USA, community-acquired MRSA poses a substantial threat. Cases of community-acquired MRSA in Europe could also increase substantially in the near future. Add to this the recent emergence of the New Delhi metallo-β-lactamase (NDM-1) enzyme,
www.thelancet.com Vol 379 June 16, 2012

which can make many common Gramnegative bacteria resistant to all known antibiotics, and the necessity for rapid action to develop new drugs becomes startlingly clear.

“‘This is a major step forward, and we should applaud Europe for taking the lead’...”
The European Commission and the drug industry have joined forces to respond to this challenge with a major project under the Innovative Medicine’s Initiative (IMI)—a public– private partnership in which the European Union and the European Federation of Pharmaceutical Industries and Associations (EFPIA) each donate €1 billion ($1·23 billion) to stimulate innovation to confront Europe’s major health challenges. From this pot, IMI has initially dedicated €224 million ($275 million) to the new antimicrobials initiative, with a total of €600 million ($738 million) expected to be spent up to 2020. So far, the EFPIA companies taking part in the IMI antimicrobial resistance programme are GlaxoSmithKline (GSK), AstraZeneca, Johnson & Johnson, Sanofi, and Basilea. “This is a historic opportunity for Europe to overcome a public health problem which threatens millions of lives worldwide”, says Michel Goldman, Executive Director of IMI. “For researchers in universities, hospitals and small and mediumsized enterprises—that will all have the opportunity to apply for IMI funding— it is also a unique opportunity to speed up their research in the area of antimicrobial resistance, as the collaboration will give them access to the knowledge and expertise of the pharmaceutical industry.”

The initial projects will focus on building and training networks of researchers, facilitating and increasing the exchange of research data, and improving the efficiency of clinical trials on new antibiotics through better laboratory tests and better trial design, in which networks of clinical investigators will make use of novel diagnostic tests to assess the efficiency of drugs. The novel trial design will be applied in clinical trials testing the efficacy and safety of experimental antibiotics to fight particularly resistant bacteria. The first large-scale (phase 3) clinical trials will evaluate, in several hundred patients, the efficacy and safety of antibiotics designed to fight these resistant bacteria. For instance, trials will target the notorious MRSA, both the hospital-acquired and community-acquired versions. In parallel, new methods will be explored to improve antibiotic penetration into Gram-negative resistant bacteria, such as Acinetobacter baumannii. Antibiotic penetration is a key challenge in the development of drugs against these life-threatening infections. “Only two new classes of antibiotics have been developed and launched in the last 30 years. This reflects the difficulties we face—the science is very

See Editorial page 2214 For more on the crisis of no new antibiotics see Personal View Lancet Infect Dis 2012; 12: 249–53

2229

Bloomberg via Getty Images

World Report

complex and the risk and investment required is high yet antibiotic discovery has a higher attrition than other therapeutic areas”, says David Payne, Head of Antibiotic Discovery at GSK. “Also, carrying out studies in people for antibiotics where infections are often short lived can be challenging. Perhaps most importantly, the existing investment model for antibiotics is challenging and does not adequately reward innovation in antibiotic research.” “Discovering new, effective, and safe antibiotics is scientifically very difficult and costly. The IMI collaboration allows researchers, both in academia and industry, to build on each other’s experience and to take better advantage of each others’ specific knowledge and expertise”, says John Rex, Infection Clinical Vice President at AstraZeneca. “During the last 10–20 years there has been a steady decline in the number of large pharma companies actively investing in antibiotic research and AstraZeneca is among a small handful of key players that still remain.” “This is a major step forward, and we should applaud Europe for taking the lead”, says Laura Piddock, professor of Microbiology at the University of Birmingham, UK, and Director of Antibiotic Action, a global initiative of the British Society for Antimicrobial Chemotherapy to restimulate antimicrobial drug discovery, research, and development. But Piddock stresses there are many challenges to address to unlock what she calls the antimicrobial conundrum. “To make new antimicrobials last, we need to use them less, making them less profitable. We are also in desperate need of new bedside diagnostics that can help target treatment, which will in turn help make clinical trials smaller (and cheaper) whilst giving robust data for the efficacy and safety of the new drug.” Piddock adds: “We also have to address developing countries’ needs. If we end up with lots of expensive targeted antimicrobials, then, as with
2230

targeted cancer therapies, how can developing countries possibly afford them unless they are offered at a lower price? Maybe new ways of providing antibiotics are required.” WHO has called antimicrobial resistance one of the greatest threats facing the world, dedicating last year’s World Health Day to the topic. “Patients with impaired immune responses— such as the elderly, patients with HIV or cancer, transplant patients, and patients under mechanical ventilation—are especially prone to develop severe infections caused by antibiotic-resistant pathogens”, says Goldman. “These infections are responsible for life-threatening conditions such as intractable pneumonia, necrotising fasciitis, and septic shock, in addition to the well known challenges of MRSA and emerging challenges posed by NDM-1.” Europe is not the only region taking action. In the USA, the Generating Antibiotic Incentives Now (GAIN) Act is currently being debated by Congress. The act aims to extend the patents granted to drug companies that develop new drugs, and thus hopefully extend their profitability. “There are concerns in the international microbiology community that the act is short-term fix, rather than a long-term stimulus”, says Piddock. She points to the recent newsletter of the Alliance for Prudent Use of Antibiotics, which has reviewed the act as it currently stands. “The stated objectives of the GAIN Act include increased surveillance of resistant bacteria, more responsible use of existing antibiotics, and increased incentives to develop new antibiotics”, the newsletter quotes Kevin Outterson, Associate Professor of Law & Director of the Health Law Program, Boston University and Editor-in-Chief of the Journal of Law, Medicine and Ethics. He adds: “However, the current draft of the GAIN Act does not provide any binding requirements to implement antimicrobial stewardship, appropriate use, and conservation. It focuses exclusively on bringing new antibiotics

to market quickly, without any changes whatsoever to patterns of use in either human or animal populations.” It’s a scenario Outterson describes as “more brandy for the alcoholics”. “It has been recognised for some time that more concrete actions need to be materialised to make a meaningful change here, so the IMI initiative is incredibly important”, adds Payne. “The pharmaceutical industry has the will as well as the expertise, skills, and resources to be part of the solution in making new antibiotics available for the future, but there needs to be a fundamentally different approach with companies, public institutions, and academia, working together and sharing information to restimulate research so that when we do face a new bacterial challenge we are able to protect ourselves.” The initiative is not just about developing brand new antimicrobial molecules. Goldman is hopeful that this IMI initiative, by accessing data on molecules already held by pharmaceutical companies, will see new antimicrobial drugs coming to market within 5 years, possibly less. The initiative will facilitate phase 3 pivotal trials of these preexisting compounds, and thus greatly accelerate the commercialisation of innovative antibiotics and therefore access for patients. As far as drug discovery efforts are concerned, patents will remain owned by those who made the seminal discovery, whether they belong to industry or academia. “Importantly all results obtained during the IMI programme will be shared by the partners and further disseminated thereafter to the broad scientific community”, says Goldman. “As such, this initiative will do much more than provide funds for clinical research on antibiotics. It will set the stage for a new way to develop innovative medicines in Europe, based on open collaboration and partnership between health-care stakeholders.”

Tony Kirby
www.thelancet.com Vol 379 June 16, 2012

Comment

Offline: Why should we take women seriously?
A report from a UK research charity about how the public views science concludes: “This study showed that public understanding about medical research is weak.” The report continues: “It is important that the results of this study are considered within the context of the low level of understanding of the process of medical research, particularly where they are to be used to inform policy development.” Many scientists and doctors will have hoped that wider coverage of medical science in the media, better science education in schools, and greater access to scientific findings through the Internet would have improved the public’s understanding of research. But maybe these assumptions are unreasonable. The idea of a scientifically engaged society is still a distant dream. * One of the highlights of the UK medical research calendar is the Academy of Medical Sciences’ International Health Lecture. This year’s oration was delivered by Janet Hemingway, Director of the Liverpool School of Tropical Medicine. The Liverpool School sometimes (and unfairly) sits in the shadow of its London counterpart. Janet Hemingway showed why such a southern metropolitan bias would be a mistake. Her subject was the health impact of product development partnerships. Her special contribution is as Chief Executive Officer of one such partnership, the Innovative Vector Control Consortium. She showed how creating unusually collaborative public-private networks has created entirely new classes of insecticides and diagnostics. Her task is to promote “market rupture”—overcoming market failures. The prospects seem good. Within the next 5 years, perhaps as many as 11 new classes of insecticides will enter the market. * Although I am half-Norwegian and therefore have a fatal genetic conflict of interest, it would be a harsh observer who did not credit this small country with astonishing success in making maternal and child health a global political priority. When Hillary Clinton confirmed a visit to Norway to negotiate the future of the Arctic and its vast natural resources, Jonas Gahr Støre (Norway’s Foreign Minister) grasped an opportunity. He arranged a series of workshops to chart a new path in global health. In one
2226
McCoy Wynne Photography

of these workshops, held earlier this month, remarkable data were revealed. Most Norwegians (and the rest of the world) think that Norway is successful thanks to its oil. But that is only part of this country’s story. Hilde Singsaas, a State Secretary in the Ministry of Finance, presented data showing that the real secret to Norway’s success was a massive expansion of female participation in the workforce, from 44% in 1972 to 70% in 2011. Women’s contributions to Norway’s productive employment have been “a major factor explaining economic growth” and have been “equally important” to oil. This massive engagement of women in Norway’s economy itself depended on welfare reforms, child care provision, girls’ education, and parental leave. The conclusion for the rest of us is that taking women seriously in our societies is not only the right thing to do, but also brings huge and largely undocumented economic benefits. * Now is the summer for public health memoirs. Peter Piot’s No Time To Lose is to be published shortly. But first comes Desmond Avery’s biography of JW Lee, the former Director-General of WHO (Lee Jong-Wook: A Life in Health and Politics: Orient BlackSwan, 2012). “JW”, as he was known, died suddenly and unexpectedly in 2006, aged 61. The President of the Korea Foundation for International Healthcare, Han Kwang-su, calls him “the Vaccine Czar”, “the Schweitzer of Asia”, and “the Little Giant”. I remember JW differently—as a quiet and sometimes solitary man, and as someone who quickly saw opportunities to use WHO’s comparative advantage to maximum political value. When first elected DirectorGeneral, Lee attended a meeting in Bellagio, convened by Jennifer Bryce, at which the evidence to make child survival an international health priority was being assembled. Although not an expert in child public health, JW listened and decided to make child survival a key part of his future strategy. He did the same with “3 by 5”, noncommunicable diseases, and the social determinants of health. Lee’s tenure as DG was short, just 3 years. Yet he made his mark, and distinctively so. Desmond Avery’s exquisitely written book explains why. Richard Horton
[email protected]

AFP/Getty Images

AFP/Getty Images

Corbis

www.thelancet.com Vol 379 June 16, 2012

World Report

Diabetes saps health and wealth from China’s rise
Rapid economic change in China is propelling a wave of diabetes that health professionals and the public and are only beginning to wake up to. Ted Alcorn and Yadan Ouyang report.
Treatment for diabetes has changed dramatically over the past 40 years in China, and no one knows this better than Wang Wenying. Now retired from her government job but as active and gregarious as ever, the 76-year-old remembers the onset of her type 2 diabetes back in 1974, before she had been diagnosed. At that time she weighed 15 kg more than she does today, and she had begun suffering from fainting spells and often felt thirsty. “There’s nothing wrong with you: it’s a blessing that you eat a lot and are gaining weight”, she recalls the doctor saying. “Now I’ve realised that it’s a suffering, not a blessing”, she says. This is the paradox of diabetes in China: the epidemic is a direct byproduct of the country’s rapid increase in prosperity, and many of the factors contributing to it are luxuries the Chinese have worked a lifetime to achieve. Rising household incomes have allowed Chinese citizens to eat more and to shift to foods that are higher in fats and sugars. As rural residents relocate to cities, many adopt more sedentary lifestyles. And in embracing a marketbased economy, they have also taken on many daily stressors that come with it. Frank Hu, a professor of nutrition and epidemiology at Harvard University, MA, USA, says that all of these factors are at work in China. “It’s a perfect storm”, he says. At the time of Wang’s diagnosis in 1974, fewer than one in 150 of China’s citizens had developed the disease. But during the intervening decades the number of diabetics in the country has multiplied more than tenfold. Recent national surveys suggest that one in ten Chinese now have diabetes, and among those aged 60 years and older, the prevalence
www.thelancet.com Vol 379 June 16, 2012

is closer to one in five. When Wang joins a dozen friends for dinner in the evening, four of them bring their insulin-injection pens. Hu is especially alarmed to observe that the increasing prevalence of childhood obesity in China has outpaced the growth in western countries. “It’s possible that the prevalence of diabetes will be even greater for the next generation.”

“...China’s rate of hospital admissions for acute complications of diabetes is three times higher than that of the USA.”
The ramifications for China’s health system are enormous. Noncommunicable diseases account for two-thirds of China’s disease burden and the World Bank describes them as “China’s number one health threat”. Among them, diabetes is the most prevalent. Chronic conditions like diabetes are also disproportionately costly because they require daily treatment in perpetuity. The International Diabetes Federation estimated that in 2010, 13% of China’s health expenditures— US$25 billion—was attributable to diabetes. And if the country’s total health expenditures grow an additional 50% over the next 5 years as expected, the World Bank says it will “significantly undermine” China’s efforts to expand health insurance coverage, while increasing the odds of a future slowdown in economic growth. Of course, Chinese diabetics are getting something in return for all of those additional outlays. Back in the 1970s, even after her diabetes diagnosis, Wang could obtain little information about her condition

and resorted to fasting to control her symptoms. Each day after work, she mixed a few drops of urine with testing solution and boiled it over a kerosene lamp to check her bloodsugar levels. Today she has a detailed understanding of the importance of maintaining a stable blood glucose level, and technology is available to help her do so. She tests her blood daily with a glucose strip, purchases sugar-free foods, and takes oral drugs to keep her blood glucose in a healthy equilibrium. Where Wang and the Chinese Government see an expense, the private sector sees a business opportunity. In fact, Wang and her peers are now one of the most aggressively courted groups of diabetics in the world. Global Business Intelligence valued the Chinese market for diabetes treatment at $1·5 billion in 2011, and although this still represents a small fraction of global sales, with one in four diabetics on earth now in China, it won’t remain that way for long. Barclays Equity Research estimates that over the next 10 years, 29% of global growth in diabetes treatment will take place in China.

For more on non-communicable diseases in China see Editorial Lancet 2011; 378: 457

Rising household incomes have led Chinese citizens to buy fatty and sugary foods

2227

Ted Alcorn

World Report

Drug companies are expanding their facilities to meet the anticipated demand. In January, China’s Zhuhai Labs announced that it would build a new $151 million production facility, and in May, the French pharmaceutical giant Sanofi inaugurated a $90 million assembly plant in Beijing for the manufacture of pre-filled insulin-injection pens. Both are dwarfed by the latest investment of the market-leader, Danish firm Novo Nordisk, which sells 63% of the insulin consumed in China. The company is building a formulation and filling plant in Tianjin for $400 million. When completed, it will be the largest in the world, but Ron Christie, who heads commercial business in China for Novo Nordisk, cautions that even this facility may not satisfy Chinese demand. “If every diabetic patient was treated and treated properly, we’d need a couple of these factories. It’s a question of how fast the market evolves.” Aided by public policy and also by their own rising incomes, Chinese consumers are increasingly able to afford the products manufactured by these companies. Major reforms to the Chinese health-care system over the past 3 years have extended insurance coverage to nearly every citizen and increased the availability of basic drugs, including insulin, in rural areas. Because the reimbursement rates for diabetes treatment vary by province and by insurance scheme, coverage is still highly unequal across the population. But observers generally agree that affordability isn’t as big an obstacle to obtaining care as is unfamiliarity with the disease, amongst both practitioners and patients. Wang has become quite knowledgeable about diabetes in the 38 years she has managed her own condition, but most people lack her experience, even urbane residents of Beijing. She recalls how when a neighbour complained of a recurrent foot infection, she recognised it might be a symptom of diabetes
2228

and instructed the neighbour to get tested immediately. The neighbour was sceptical but a doctor later confirmed the diagnosis, before amputating two of her toes. 60% of Chinese diabetics are like Wang’s neighbour: they have developed the disease but don’t yet know it. By contrast, 20% of diabetics in the USA are undiagnosed. This is part of the reason why China’s rate of hospital admissions for acute complications of diabetes is three times higher than that of the USA. Delayed treatment invariably requires more invasive care, at greater cost, and with worse outcomes. Drug companies and the government thus share an interest in educating the public about diabetes, since it will expand the market for therapies while lowering overall health-care costs. In large and medium cities across China, they are collaborating to provide educational programmes for physicians and patients. But national objectives are thus far modest: the latest National Plan for Non-Communicable Disease Prevention and Control calls for providing 40% of the country’s diabetics with standard treatment by 2015 and achieving good blood glucose control in 60% of that group. Even if those goals are met, only one in four Chinese diabetics will be effectively controlling their disease. In addition to expanding care, China may have to reshape the way physicians and patients relate to one another if the epidemic is to be brought under control. Jason Mann, a physician who grew up in China and now heads China Healthcare at Barclays Equity Research, says that doctors in China typically examine and diagnose patients based on their chief complaints but rarely take a proactive approach to managing other observed risk factors. “In an environment where the patient has very limited experience with or awareness of diabetes, this sets you up for a silent epidemic.” In the next 5 years, he hopes that China’s

health-care system will begin to shift away from its current hospital-centric form towards a primary health care model in which patients develop a relationship with a single physician who helps them manage their health in the longer-term. A family-doctor programme launched in Jiangsu province in February indicates that the government may be exploring this direction. Better diagnosis and treatment will improve outcomes for the tens of millions of Chinese that have already developed diabetes, but staunching the torrent of new cases demands changes outside of the health sector, in areas that pit the government and commerce against one another. Taxes deterring the consumption of unhealthy foods or campaigns encouraging the population to get additional exercise would not be unprecedented in China, where the state often wields a heavy hand in the lives of the population. But the government has shown little appetite for experimenting with such tactics in the case of diabetes, perhaps because the factors contributing to the epidemic are so intertwined with economic growth, its most cherished objective. For now, prevention efforts are few and fragmentary. Last year, the government deepened its national healthy lifestyle campaign and pilot projects are underway in areas across the country to improve noncommunicable disease prevention by strengthening disease surveillance, educating the population about chronic disease, and standardising care. Chen Wei, newly elected Director-General of the Beijing Diabetes Prevention and Treatment Association, says that the efforts demonstrate vision but not action. “In my opinion, it hasn’t been carried out satisfactorily—that is to say, they have ideas but haven’t turned them into practice.”

Ted Alcorn, Yadan Ouyang
www.thelancet.com Vol 379 June 16, 2012

Perspectives

The Lancet Techology: June, 2012
Maps, apps—and evidence?
“GPs to ‘prescribe’ apps for patients” declared a press release from the UK Department of Health earlier this year. The announcement, made to promote the Maps and Apps initiative, included enthusiastic descriptions of apps that allow patients to access their medical records, or to manage type 2 diabetes by monitoring food intake. Health Secretary Andrew Lansley provided a mixed metaphor in support (“With more information at their fingertips, patients can truly be in the driving seat”), and UK Digital Champion Martha Lane-Fox added “Using apps that locate local health services or apps that help you to get fit can dramatically improve your daily life”. The management of chronic health conditions is an obvious area where apps might be beneficial, and type 2 diabetes is a case in point. Maintaining lifestyle changes over many years, as well as accurately recording data such as blood glucose measurements, presents a challenge to even the most motivated patient: I am reminded of Colin Dexter’s Inspector Morse, carefully forging his blood glucose readings on the record card in his doctor’s waiting room, making the figures realistic but not too alarming. That scene depicts a situation common to the old patriarchal form of medicine: the doctor as a sort of medical headmaster, sternly checking the patient’s end-of-term report. With an increasing emphasis on patient empowerment and self-management, however, the use of smartphones may provide a way in which individuals with type 2 diabetes can take greater responsibility for their own health. There are many apps available to help patients manage diabetes, as a search on the iTunes store will show. Diabetes UK Tracker, a free app produced by the charity Diabetes UK, is a good example. It has been available since September, 2011, and
www.thelancet.com Vol 379 June 16, 2012

there were over 9000 downloads by the end of the year. The app allows the patient to enter measurements such as blood glucose, caloric intake, and weight: these can be viewed on graphs, and shared with healthcare professionals. There is also an area to record medical information, and, interestingly, to note down personal feelings and reflections. I spoke to Pav Kalsi, a Clinical Advisor to Diabetes UK, who highlighted this as an important feature of the app. She told me that it was added in response to a consultation exercise with patients and clinicians. “It’s not just about blood glucose results and HbA1c results, it’s about how people feel, and perhaps how their mood may affect their glucose levels.” The ability to share these feelings and their results is also important to patients using the app. “People reported that it was encouraging— they’d post on Facebook or Twitter if they’d done well with their diabetes management, and that would be quite fulfilling. Or, if they weren’t doing so well, they’d get moral support from their peers via social media.” And what about what one might term the Inspector Morse problem? “People only see their health-care professional for a few hours out of a whole year, whereas they’re living with their condition the rest of the time”, said Pav. “Because it’s a self-management tool, no one has anything to hide from their health-care team.” I found Diabetes UK Tracker easy to use, and from Pav’s description it seems that it has been constructed with an admirable degree of liaison with the people who will be using it. My only caveat—which applies to all patient-oriented apps—is the issue of evidence for effectiveness. Looking again at the Department of Health press statement, there is no mention of actual evidence that smartphone apps improve

health. A search on Medline reveals why: very little is actually available. When any other prescribed health in ter vention—pharmacological, behavioural, or psychological— is subject to the scrutiny of the randomised controlled trial, why are apps exempted? One reason might be that people simply do not put apps in the same bracket as other health interventions: the sheer speed with which the app market has grown over the past few years has taken everyone, health-care workers included, by surprise. “The use of apps as a selfmanagement tool is a fairly new concept, so there is limited longterm evidence; however, Diabetes UK would welcome further research into this area”, said Pav at the end of our conversation. I agree: if doctors are going to be encouraged to prescribe apps, trial data are essential to reassure them that they are advising their patients correctly. Trials would also benefit app producers, helping them to develop more effective products. I admire the Department of Health for promoting innovation, but wonder if they should also think about supporting the development of the evidence base. The growth of smartphone use and social networking, a boon of industrial society, may prove valuable in combating one of the less desirable results of economic progress: the rise in obesity and type 2 diabetes. But we are only at the beginning of the journey. Perhaps gurus such as Eric Topol are right, and the so-called creative destruction of medicine via information and communication technology is at hand. It would be unwise, however, to discard the last century’s hard-won gains in evidence-based practice just yet.

For UK Department of Health press release see http:// mediacentre.dh.gov. uk/2012/02/22/gps-to%E2%80%98prescribe%E2% 80%99-apps-for-patients For Maps and Apps see http:// mapsandapps.dh.gov.uk For more on patient empowerment see Editorial Lancet 2012; 379: 1677 For Diabetes UK Tracker see http://www.diabetes.org.uk/ How_we_help/Diabetes-iPhoneTracker-app For more on Eric Topol see http://creativedestruction ofmedicine.com

Niall Boyce
[email protected] Twitter: @TheLancetTech

2231

Diabetes UK

Press Association

Corbis

Perspectives

Book Getting fatter and fatter
We need a public debate about what it means to keep markets in their place. And to have this debate, we have to think through the moral limits of markets. We need to recognise that there are some things that money can’t buy and other things that money can buy but shouldn’t. Michael Sandel, “Market and Morals”

false parochialism of a single company or a single country. Markets, thus constrained, tend to care little about such things.

Fat Fate and Disease: Why Exercise and Diet Are Not Enough Peter Gluckman, Mark Hanson. Oxford University Press, 2012. Pp 304. £16·99. ISBN 9780199644629

We live in a world where it is increasingly apparent that markets have all sorts of unwanted consequences, but they remain the bedrock of our civilisation. That we should relentlessly pursue economic growth is unquestioned, while the planet is drying up and we are becoming increasingly obese. And by a dominant political account the way to grow is to liberate the markets wherever we can. Thus the planet nears extinction more quickly and the prevalence of type 2 diabetes increases alarmingly across the globe. So is some kind of consensual good will required, as Michael Sandel suggested in his 2009 Reith Lecture? If we are to believe Peter Gluckman and Mark Hanson in Fat Fate and Disease, those who wish to avoid a worsening obesity epidemic should not oppose the food industry, but work with them. We should concentrate more on the epigenetic causes of rising obesity and less on current energy balances. Wise and pertinent it might be, but it seems to me yet another partial solution falsely constrained by our collective inability to protect the long-term health of populations in a market economy. It is clearly difficult to avoid that kind of deckchair-arranging polemic. Public health engages in this debate all the time, because it is necessarily concerned with externalities and the long-term health effects and societal costs of unhealthy exposure. The discussion has to do with these effects being properly set against shortterm gains—unconstrained by the

“It didn’t take the food industry long to discover that our primeval short-term needs responded well to the aggressive purveyance of junk food.”
In 2007, the UK’s Foresight Report on tackling obesity was clear: exercise and diet are central, but on their own wholly insufficient since both are subject to the influences of our environment— on our physiology and our inheritance during a life course. These are the determinants of our obesity epidemic and they are dominated by markets: marketing, pricing, and product reformulation, availability, and taste. J K Galbraith dealt with this 45 years ago when he argued that the local survival of commercial organisations depended on them creating demand, as much if not more than consumers do. It didn’t take the food industry long to discover that our primeval shortterm needs responded well to the aggressive purveyance of junk food. For them the long-term effects were irrelevant. The Foresight Report was not the least revolutionary, but it did suggest a blueprint for government policy: “healthy weight, healthy lives”. Again, largely dodging the main issue. Public health and epidemiology have reached a point where the causes of much ill health are clear. But unlike National Institute for Health and Clinical Excellence (NICE) clinical guidelines in the UK, which are mandatory for clinicians, those responsible ultimately for the health of populations seem largely to ignore population guidelines, often on the grounds that they “are not practical”. This tends to mean, in my experience,

that such guidelines would interfere too much with economic goals. Clinical decisions can be constrained to a single person, whereas public health decisions are broader. We now have the Responsibility Deal in the UK—an alliance of industry, government, and non-governmental organisations designed to temper the market by exacting health-related pledges voluntarily from (largely) the food industry. Strictly we have yet to measure its success, but until now the interest of shareholders apparently cannot countenance pledges that might compromise profit and growth. For success we need a more level playing field for industry to compete on with strong irresistible pressure in the direction of long-term health gain. We have yet to discover an acceptable way to make markets properly balance all the pay-offs caused by unhealthy production, one of which is to suffer loss at the point of production commensurate with the harm of causing bad health in the longer term. We are not even close. Good will and responsible citizenship are not, I suspect, going to solve this problem simply because profit and growth trump everything. Obviously consumers may choose more wisely—but experience suggests that this will have a slow, unequal, and barely perceptible effect. Fat Fate and Disease argues that our model of the determinants of the obesity epidemic is inadequate to deal with it. It is, and the analysis therein—just like all proposals that focus on individual responsibility in a strong and parochial market—can be only partial. Must we await the hard thinking of a contemporary Richard Titmuss or Galbraith? It is time to break out of this impasse, before most men aged 50 years have type 2 diabetes.

Klim McPherson
[email protected]

2232

www.thelancet.com Vol 379 June 16, 2012

Perspectives

Profile David Nathan: putting diabetes on trial
Cast an eye over the key physicians involved in most of the big US multicentre diabetes trials of the past 30 years—to say nothing of more general health studies with a diabetes component, such as Framingham—and one name keeps on appearing. David Nathan. Co-investigator, principal investigator, co-chairman, chairman…David Nathan. “His career has followed a consistent pattern of steadily working through aspects of the control of diabetes”, says diabetic physician Edwin Gale, emeritus professor at the University of Bristol. “He’s been involved not only with trials, their outcome and their analysis, but also with the development and evaluation of new ways of delivering insulin.” But how did Nathan, who is Director of the Diabetes Center and Clinical Research Center at Massachusetts General Hospital and a professor at Harvard Medical School, come to play such a role? The answer, it seems, is part chance, part perseverance, and part personality. He’s also smart—but, mindful of where he works, that’s a given. Nathan was initially drawn as much to literature as to medicine. “I still love writing”, he says, “but I decided to keep it as an avocation.” Or nearly so: “Writing an abstract, achieving brevity and accuracy while including as much content as possible, is as challenging as a haiku.” He finally opted for medicine because it represented an interesting form of applied biology. “I always intended to do research”, he adds. And this is where chance begins to figure: the choice of diabetes. The most exciting area at the time he was finishing medical school happened to be, in his view, endocrinology. “What made it more advanced than some other specialties at that time was radioimmunoassay. Because it was able to measure very small amounts of substances in the body it was a window into how we functioned. This captured my imagination.” Professor Joseph Avruch has known Nathan since 1978 when the latter joined the Massachusetts General Hospital as a clinical fellow. “David came into this area at a time when there was great change in the way that clinical care in diabetes was going to be provided”, says Avruch. “Partly this was because there had been great uncertainty about the benefit of tight glucose control.” Some people doubted its importance. “They thought that the complications had more to do with genetic susceptibility or other factors.” The study set up to resolve these doubts was the type 1 Diabetes Control and Complications Trial (DCCT). “It was an illustrious group and I was the kid when it started”, says Nathan. The methods used included HbA1C assay, which provides an index of long-term glucose control and has turned out to be the cornerstone of therapy as well as clinical research. The results couldn’t have been better. “More than 99% of the patients stayed with it for an average of six and
www.thelancet.com Vol 379 June 16, 2012

half years. I’d suggest that’s a world record. We were looking to reduce complications by about 30% and we ended up with a reduction that was more than double that”, says Nathan. The DCCT, with Nathan as chairman of its Publications Committee from 1985, became one of the most frequently cited studies in medicine. He himself went on to become co-chairman of the subsequent follow-up study, and also chaired the type 2 Diabetes Prevention Programme (DPP) that began in 1994. Nathan’s success, according to Gale, can be summarised in two qualities: perseverance and unobtrusive leadership. Judith Fradkin, Director of the diabetes section of the US National Institute of Diabetes and Digestive and Kidney Diseases, first met Nathan 20 years ago through the DCCT, and later worked with him on the DPP. “He has a very sharp mind in terms of recognising what’s important”, she says. It was through his influence that the DPP chose an endpoint (not some physiological measure but the prevention of diabetes) that could be translated into a simple, straightforward message when the trial was complete. “He’s also a very warm, witty, friendly person, and that’s important because everyone had their own ideas on how the trial should be designed. You need a leader who gives everyone a chance to express their opinion but can ultimately foster some agreement. He was extraordinarily effective at that”, says Fradkin. It was the same quality which Avruch had seen in Nathan’s chairmanship of the DCCT. “This warmth of personality allowed him to bring so many ‘diabetic geniuses’ to collaborate in the study”, he believes. “Focus and perseverance are essential to accomplish what he’s done—and he’s been at it for 35 years.” Nathan, in the view of Gale, also represents the best in American research. “The ability to tackle a big question, to organise and mobilise resources, and drive forward until you get answer. Only America can do that because only America has the resources. He’s a leading example of how successful that approach can be.” Glancing over Nathan’s CV you’re struck by the fact that he’s spent virtually all his career at Massachusetts General and Harvard. He travels a lot, and enjoys it. So was he never tempted to work elsewhere? Certainly not by the prospect of a prestige job that might take him away from research. “I’m a home body”, he says. “Scientifically I enjoy being in a stable environment. Being able to do mostly my own work is what I’ve aimed for. Massachusetts General and Harvard have allowed me to do this, and not saddled me with too much administration. They’ve left me alone!” Which, scientifically speaking, is pretty much his ideal.

Geoff Watts
2233

Perspectives

Profile Frances Ashcroft: the doyenne of diabetes research
In 1983 a young Frances Ashcroft sat alone in her new laboratory at Oxford University’s Department of Physiology, Anatomy, and Genetics, plagued by doubts. “I felt pretty useless as a scientist at the time, and I didn’t think I was terribly good in the lab”, she recalls. But she was, she says, “quite good at connecting ideas”; one instance of which was to apply the newly minted patch-clamp technique for studying the activity of ion channels to pancreatic β-cells, with the expectation that something interesting might turn up. It did. A year later she published a seminal paper in Nature that established the ATP-sensitive potassium channel (KATP channel) as the crucial link between blood glucose concentration and insulin secretion, which propelled her to the forefront of efforts to understand how cells regulate themselves in health and disease, and in diabetes in particular. Now a Royal Society Research Professor in that same Oxford lab, the decades since Ashcroft’s breakthrough discovery have brought awards by the bucket load, while her gift for explaining the intricacies of ion channels (and science in general) in a way that doesn’t make people want to flee in panic has seen her attain a degree of celebrity quite rare for a scientist. When The Lancet spoke to her in early June she was in the throes of preparing for an appearance at the Hay Festival to talk about her new book, The Spark of Life: Electricity in the Human Body. (Her first popular science work, Life at the Extremes—the Science of Survival, made it onto the best-seller lists and was translated into 13 languages.) And yet for all her success, some of the old doubts still remain. “The truth is I always feel pretty useless at everything”, she explains, “but the way I deal with it is to simply accept that fact and get on with it anyway”. And if things do go wrong, well, that’s just life she says. “Science is about learning to live with failure because most of the time things don’t work. You just have to try again— and again. But puzzling out how things work is also an enormously exciting and rewarding challenge and the moments of discovery, when everything suddenly falls into place, sustain you throughout the difficult times.” That desire to find out how things work has been Ashcroft’s driving force ever since she was a child growing up in rural Dorset. “I spent most of my life running around the fields and looking at flowers. And then later with my friends I would go bird watching or hunting for wild orchids. We didn’t have a camera so we used to paint them”, she laughs, “but it helped develop my observational skills”. At the same time, Ashcroft and her sister would hone their bodies and minds by pushing a
2234

pram laden with books the mile or so to where the village library van stopped every fortnight. The years spent in academia—first as an undergraduate studying Natural Sciences at Cambridge University and now as the head of a multidisciplinary team of researchers—haven’t quenched Ashcroft’s thirst for discovery; if anything the opposite is true. “If I was not able to do science any more, and I didn’t have such a wonderful team of brilliant young researchers in the lab, who are full of interesting ideas, I would be extremely miserable”, she says insistently. “Science is my life’s breath, I can’t survive without it.” Her enthusiasm is infectious, according to Jochen Roeper, Director of the Institute of Neurophysiology at the Goethe University Frankfurt in Germany and a former student of Ashcroft’s. “She knows how to bring out the best in young scientists—mostly by the enormous force of her love and enthusiasm for science”, says Roeper. Ashcroft’s way with people extends to her canny knack for forging lasting and extremely productive collaborations. Shortly after Ashcroft discovered the role of the KATP channel in insulin secretion, she learnt that Patrick Rorsman was trying to do the “same thing but in a different way in Germany”. A few months later there was a conference in Alicante, where the two potential rivals came face to face. Perhaps the warm climate helped break the ice, because the meeting proved to be the start of a long and fruitful partnership, with Rorsman eventually moving from Sweden with his whole family to be Professor of Diabetic Medicine at Oxford. “What can be nicer than having really good friends in your own field”, says Ashcroft cheerfully. Among a host of other collaborations, Ashcroft’s work with Andrew Hattersley’s team at the UK’s Peninsula Medical School in Exeter has been of special importance; first, they demonstrated that mutations in the genes that encode the KATP channel explain about half of cases of neonatal diabetes, and then they went on to show that these patients could be treated with high dose sulphonylurea tablets rather than insulin injections. It’s this close connection to the clinical consequences of her findings that really makes Ashcroft stand out, according to Christopher Miller, from Brandeis University in the USA, another collaborator and long-time fan. “She’s done translational research in the best sense of the word”, he says, “in showing how understanding the nuts and bolts of how a protein works can lead to dramatic, practical outcomes in the clinic”.

Robert Taylor

David Holmes
www.thelancet.com Vol 379 June 16, 2012

Perspectives

The art of medicine What goes around, comes around
It’s now several years since I could reasonably claim to be a card-carrying diabetologist, and most of my time in that field was in a previous millennium. The passage of time was brought home recently when the internet trawled up an image of an “antique” insulin pump, of a type that I remember all too clearly. This set me thinking about the clinical conundrum that kicked off my research career. It was 30 years ago, but the lessons we learned then— belatedly and with some difficulty—are still valid in this new, enlightened century. The conundrum was so-called brittle diabetes, defined pragmatically as metabolic instability sufficient to disrupt life, whatever the cause. The term dates from 1937, when the first long-acting insulins had recently been introduced and were failing to meet the expectation that they would bring diabetes magically under control. From this came the idea that there was a subset of insulin-dependent (type 1) diabetic patients whose blood glucose levels were intrinsically difficult to control. The notion of brittle diabetes resurfaced at intervals and then enjoyed a renaissance during the 1980s, when another revolution in insulin technology—the portable insulin pump—again shone the spotlight on a hard core of patients whose diabetes could not be tamed by state-of-the-art treatment. This was the cue for a handful of specialist centres on both sides of the Atlantic—Albuquerque in the USA, and Newcastle and Guy’s Hospital, London, in the UK—to declare an interest in “difficult” type 1 diabetes. Brittleness was now redefined as metabolic instability that even resisted pump therapy. Each centre built up its own population of brittle patients, with some referred from hundreds of miles away and a few shuttling between centres. These patients were easy to spot from their blood glucose profiles during “optimal” treatment with the insulin pump. Instead of the near-normal levels that were usually achieved, the brittle patients showed a chaotic sawtooth pattern that swung between hyperglycaemia and profound hypoglycaemia. Strikingly, many were taking high insulin dosages, sometimes hundreds of units each day. The record daily dosage was a terrifying 20 000 units, easily enough to clear the log-jam in consultant diabetologist posts that preoccupied trainees in Britain at the time. All our attempts to find an endocrine or other cause for this mysterious “insulin resistance” drew a blank. Curiously, almost all the patients with brittle diabetes were young women, which made us wonder whether this might be a genuine syndrome after all. Other diagnostic features of the “syndrome” were case notes that weighed several kilos, multiple hospital admissions totalling several weeks or months each year, and a trail of defeated diabetes
www.thelancet.com Vol 379 June 16, 2012

specialists. Some patients showed a baffling clinical sign: an enigmatic, Mona Lisa-like smile that could even appear in the emergency room, making an unsettling contrast with the fear in everyone else’s face. The significance of that smile completely passed me by. Brittle diabetes became a hot topic, with papers and editorials in top journals, and a posse of keen young things in the USA and UK locked in frantic competition to crack the mystery. Imaginative and variably plausible ideas were bounced around. Some researchers favoured hormonal disturbances (always worth a try) or exaggerated stress responses. Others wondered where all that insulin might go; perhaps it was being gobbled up by insulin-degrading enzymes at the injection site, or hoovered out of the bloodstream? Or maybe there was a defect in local blood flow where insulin was infused subcutaneously, which could impair the absorption of insulin into the circulation? There were plenty of leads to investigate, and it was a busy and exciting time. I still bear the scars of that research programme—and so do those of my friends who kindly volunteered to provide fat biopsies for various experiments. Throughout all this, the research teams were buoyed up by the amazing stoicism of the patients and the families’ confidence that we would find a cure. Collectively, we came up with some interesting suggestions, which the patients were always happy to try out. From Boston came the idea of injecting insulin mixed with enzyme inhibitors to protect it against being degraded by subcutaneous fat. At Guy’s, we tried bypassing the hazards of subcutaneous administration altogether, pumping the insulin instead into a muscle or even intravenously. A few patients struggled heroically for some weeks with continuous insulin

Mill Hill Infuser (1976), the first portable insulin infusion pump, made at the National Institute for Medical Research, based in Mill Hill, London, UK

2235

Science Museum/Science & Society Picture Library

Perspectives

Further reading Cartwright A, Wallymahed M, Macfarlane IA, Williams G, Gill GV. Life quality and experience among long-term survivors of brittle type 1 diabetes. Practical Diabetes 2011; 28: 332–35 Kent L, Gill GV, Williams G. Mortality and outcome of patients with brittle diabetes and recurrent ketoacidosis. Lancet 1994; 344: 778–81 Pickup JC, Williams G, Johns P, Keen H. Clinical features of brittle diabetic patients unresponsive to optimized subcutaneous insulin therapy (continuous subcutaneous insulin infusion). Diabetes Care 1983; 6: 279–84 Schade DS, Duckworth WC. In search of the subcutaneous insulin resistance syndrome. N Engl J Med 1986; 315: 147–53 Williams G, Pickup JC, Keen H. Massive insulin resistance apparently due to rapid clearance of circulating insulin. Am J Med 1987; 82: 1247–52

infusions into the central veins; some of these patients experienced recurrent bouts of septicaemia and central venous thrombosis. Luckily, these complications responded to treatment, although it was difficult to nail some of the bacteria responsible for septicaemic episodes, because for some reason they grew better at a decidedly cool 22°C, rather than body temperature. At best, these experimental therapies seemed to work for a few weeks, and then the patient would revert to her usual level of poor control. An increasingly gloomy picture began to form as the ideas were put properly to the test, and the experiments that had initially yielded such promising results were repeated by others. One by one, the grand hypotheses went through the cycle of hope, hype, reality check, and oblivion. Ultimately, none of them survived the impact of evidence. Interest gradually dried up, the keen young things published whatever they could and moved on to more productive pastures, and the patients drifted back home. The final nail in the coffin was a meticulous American study, showing that there was nothing exceptional about these patients—when they were closely observed. After being given standard insulin dosages under conditions where malefaction was impossible, their plasma insulin levels went up, and their glucose fell, exactly as in control diabetic patients whose metabolic control had always been stable. Conclusion: these patients were badly controlled because they deliberately sabotaged their own treatment. So much for the Americans. Over here in the UK, we’d occasionally suspected malefaction, but hadn’t wanted to believe anything that could undermine the grand hypotheses or the relationship with our patients. Confirmation emerged when we followed up our own patients a decade later. By then, they were older and wiser, and so were we. This time, the investigator was an empathic and agenda-free diabetes specialist nurse, not a publication-hungry male research registrar. Our patients told her things that we’d never thought of asking about. Those high insulin dosages? Well, they hadn’t been taking them. Some patients had learned to inject insulin straight through a skinfold and into the bedclothes, even when being watched, while others tipped insulin out of the pump and topped it up from the tap or a cistern in the ward lavatory. Water from the lavatory cistern mixed particularly badly with intravenous insulin infusions— hence the central venous thromboses and the episodes of septicaemia with those peculiar cold-growing bacteria. And the “unpredictable” episodes of hypoglycaemic coma? Easy—they’d hoarded insulin so that they could inject large doses at odd moments to knock their glucose down. Most revealingly, though, some of our patients also explained what drove them to do all this. To us, a life wrecked by uncontrollable diabetes, emergency admissions, and lengthy stays in hospital could only be a nightmare, but for many of our patients this was

preferable to whatever “normality” had to offer them. Trying to cope with diabetes had turned some of them into irretrievable failures at school, and a few were being sexually or physically abused at home. In hospital, they had company, attention, and safety. The Mona Lisa smile in the emergency room was no longer a mystery; the patient had managed to get herself back to where she most wanted to be. There were other motives too. Some had found that the rampant catabolism of uncontrolled diabetes was a neat way to stay thin; others seemed to get a kick out of defeating whatever clever solutions the keen young researchers had dreamed up. I still like to think that malefaction wasn’t the only explanation for brittle diabetes. Our record patient, for example, could not have faked an intravenous insulin dosage of 1000 units per hour, since she was deep in ketoacidotic coma at the time. But even this could have had a conventional explanation: the astronomically high circulating insulin levels could paralyse the mechanisms that pull insulin out of the circulation and deliver it to the tissues. We were never brave enough to do the control experiment of giving this massive amount to a “normal” diabetic patient in severe ketoacidosis. Sadly, we’ve missed the chance to revisit the truth with that patient. She died in her early thirties in hypoglycaemic coma. Similarly, our 30–year follow-up, completed last year, was restricted to fewer than a third of the original cohort, the others having died from metabolic catastrophe or diabetic complications. The survivors have been saved by lifetransforming events—a good job, a good partner, starting a family—rather than any medical breakthrough. Back in the 1980s, brittle diabetes began with unrealistic expectations of insulin pump therapy and somehow evolved into a folie à trois, a collusion between the patient, family, and doctor. Underpinning it are the sad truths that diabetes is a life sentence, and a disease that will never be perfectly treatable until we can mimic the knife-edge precision of the healthy pancreatic islet. Until then, we should not be surprised—and we must not be judgmental—if some of our patients are driven to manipulate their treatment in an attempt to tip the scales back in their favour. For the foreseeable future, diabetes will remain a brute to live with. Given human nature, it seems likely that “brittleness” (perhaps rebranded with a catchier, 21st-century name) will re-emerge when medical science next comes up with a therapeutic miracle for this difficult disease. When that happens, let’s hope that we won’t forget the lessons of the past. We won’t get the right answers if we ask the wrong questions.

Gareth Williams
University Department of Medicine, Southmead Hospital, Bristol BS10 5NB, UK [email protected]

2236

www.thelancet.com Vol 379 June 16, 2012

Correspondence

Neoadjuvant chemotherapy with HER2 inhibitors for breast cancer
José Baselga and colleagues (Feb 18, p 633)1 showed that neoadjuvant chemotherapy with trastuzumab and lapatinib for HER2-positive breast cancer increased pathological complete response from 29·5% in the trastuzumab-alone group and 24·7% in the lapatinib-alone group to 51·3%. Baselga and colleagues suggest that a high pathological complete response rate achieved by dual HER2 inhibition will lead to improved disease-free survival, because a correlation between increased pathological complete response and improved diseasefree survival has been shown in the TECHNO trial2 and the NOAH trial.3 In the TECHNO trial, failure to acquire a pathological complete response was indeed the only significant risk factor for relapse-related death (hazard ratio 2·5, 95% CI 1·2–5·1). However, although in the NOAH trial addition of trastuzumab to chemotherapy doubled the pathological complete response rate and significantly improved 3-year disease-free survival compared with chemotherapy alone, the relation between pathological complete response and disease-free survival was not shown. In 8·5 years’ follow-up of the National Surgical Adjuvant Breast and Bowel Project protocol B-27,4 addition of docetaxel to a neoadjuvant regimen did not improve diseasefree survival despite doubling the pathological complete response rate, while pathological complete response was a highly significant predictor of improved survival irrespective of the treatment group. In a more recent study5 that investigated the effect of the level of HER2 amplification on pathological complete response and survival after trastuzumab-based neoadjuvant therapy for breast cancer, high amplification was associated
www.thelancet.com Vol 379 June 16, 2012

I declare that I have no conflicts of interest.

Tetsuji Fujita
[email protected]
Department of Surgery, Jikei University School of Medicine, Tokyo 105-8461, Japan 1 Baselga J, Bradbury I, Eidtmann H, et al. Lapatinib with trastuzumab for HER2-positive early breast cancer (NeoALTTO): a randomised, open-label, multicentre, phase 3 trial. Lancet 2012; 379: 633–40. Untch M, Fasching PA, Konecny GE, et al. Pathologic complete response after neoadjuvant chemotherapy plus trastuzumab predicts favorable survival in human epidermal growth factor receptor 2-overexpressing breast cancer: results from the TECHNO trial of the AGO and GBG study groups. J Clin Oncol 2011; 29: 3351–57. Gianni L, Eiermann W, Semiglazov V, et al. Neoadjuvant chemotherapy with trastuzumab followed by adjuvant trastuzumab versus neoadjuvant chemotherapy alone, in patients with HER2-positive locally advanced breast cancer (the NOAH trial): a randomised controlled superiority trial with a parallel HER2-negative cohort. Lancet 2010; 375: 377–84. Rastogi P, Anderson SJ, Bear HD, et al. Preoperative chemotherapy: update of National Surgical Adjuvant and Bowel Project protocols B-18 and B-27. J Clin Oncol 2008; 26: 778–85. Guiu S, Gauthier M, Coudert B, et al. Pathological complete response and survival according to the level of HER-2 amplification after trastuzumabbased neoadjuvant therapy for breast cancer. Br J Cancer 2010; 103: 1335–42.

2

3

4

I declare that I have no conflicts of interest.

Katya Brede-Hekimian
[email protected]. de
Oncolgical Research Laboratory, Clinic for Internal Medicine II, University Hospital Jena, 07747 Jena, Germany 1 Baselga J, Bradbury I, Eidtmann H, et al. Lapatinib with trastuzumab for HER2-positive early breast cancer (NeoALTTO): a randomised, open-label, multicentre, phase 3 trial. Lancet 2012; 379: 633–40. Untch M, Fasching PA, Konecny GE, et al. Pathologic complete response after neoadjuvant chemotherapy plus trastuzumab predicts favorable survival in human epidermal growth factor receptor 2–overexpressing breast cancer: results from the TECHNO trial of the AGO and GBG study groups. J Clin Oncol 2011; 29: 3351–57. Perez EA, Suman VJ, Davidson NE, et al. Sequential versus concurrent trastuzumab in adjuvant chemotherapy for breast cancer. J Clin Oncol 2011; 29: 4491–97. Gajda M, Camara O, Oppel S, et al. Monitoring circulating epithelial tumor cells (CETC) during primary systemic chemotherapy including trastuzumab for early prediction of outcome in patients with Her2/neu-positive tumors. Ann Oncol 2008; 19: 2090–91.

5

Using pathological complete response to determine efficacy, José Baselga and colleagues1 report the results of neoadjuvant treatment of nonmetastatic HER2-positive breast cancer. Although the dual inhibition unquestionably resulted in higher rates of pathological complete response, the data with respect to larger tumours (>5 cm) for which neoadjuvant treatment was originally designed are confusing. Is pathological complete response adequate to compare treatment efficacy? This question also arises from the objective clinical tumour response

2

3

4

Submissions should be made via our electronic submission system at http://ees.elsevier.com/ thelancet/

2237

Science Photo Library

with better pathological complete response but not with improved disease-free survival. Thus, even if pathological complete response is a surrogate marker of outcome after neoadjuvant chemotherapy with a HER2 inhibitor, it is probably a weak one.

rate. And why do relapses occur in pathological complete response when the tumour is claimed to be eliminated2 even more frequently than in the adjuvant situation?3 Do we need to consider cells that can be shed into the circulation during tumour shrinkage4 and regrow into metastases? Neoadjuvant dual inhibition chemotherapy can undoubtedly result in an increase in breast-conserving surgery, but increasing proportions of mastectomies even when breastconserving therapy is recommended indicates that patients are concerned about possible residual tumour in their body. And what about patients who are unfortunate enough not to reach pathological complete response and in whom the relapse-free and overall survival results are even worse?2 Since we do not know which patients will achieve pathological complete response, they should be able to give fully informed consent to the planned treatment, whether it is neoadjuvant or adjuvant, especially if they have operable tumours.

Correspondence

Authors’ reply
In reply to Tetsuji Fujita, the incidence and prognostic effect of pathological complete response does vary between the different intrinsic subtypes of breast cancer.1 In HER2positive breast cancer, two large randomised studies2,3 have shown that pathological complete response correlates with disease-free survival in patients treated with anti-HER2 therapy. Additional information is expected from the NeoALTTO study once the disease-free survival data become available. On the other hand, the NASBP B27 study included a global breast cancer population, so its findings might not apply specifically to HER2-positive breast cancer. The mentioned correlation between levels of HER2 amplification and pathological complete response,4 although interesting, is from a small retrospective sample without a validated HER2 amplification cutoff. In the HERA trial5 of adjuvant trastuzumab, there was no difference in trastuzumab benefit according to either copy number or HER2 ratio in 3401 patients. In summary, we maintain our belief that pathological complete response is likely to be an early indicator of benefit of HER2targeted agents in neoadjuvant studies. Katya Brede-Hekimian addresses some issues similar to those of Fujita. Additionally, the reason why some tumours relapse after a pathological complete response is achieved might be explained on the basis of as yet unknown biological determinants of metastasis, the presence of a high systemic tumour burden before the start of therapy, tumour heterogeneity, and the existence of local host conditions that might favour the regrowth of tumours in certain metastatic niches. The observation that the higher pathological complete response rate seen in the combination group did not result in a higher rate of breastconserving surgery deserves careful
2238

analysis. At first glance, even in patients with a complete radiological response, the proportion of patients that underwent a mastectomy was high. We are currently analysing the reasons for this apparent discordancy and we do not believe that it is related to patients’ preferences alone. Finally, we agree that the patients who do not achieve a pathological complete response have a worse prognosis and that the potential implications of the lack of such a response have to be explained to all patients who receive neoadjuvant therapy, taking into consideration the tumour subtype, the administered therapy, the duration of therapy, and the planned postoperative (adjuvant) therapy. Additionally, the launch of clinical studies with novel antiHER2 treatments is currently being considered in patients with HER2positive tumours who do not achieve a pathological complete response with conventional anti-HER2 therapy.
JB has received honoraria from Roche. EdA has served on an advisory board and received a travel grant from GlaxoSmithKline, and has been a speaker for Roche. IB’s institution has received funding from GlaxoSmithKline and Roche. RG’s institution has received research funding from GlaxoSmithKline and Roche.

3

4

5

Untch M, Fasching PA, Konecny GE, et al. Pathologic complete response after neoadjuvant chemotherapy plus trastuzumab predicts favorable survival in human epidermal growth factor receptor 2–overexpressing breast cancer: results from the TECHNO trial of the AGO and GBG study groups. J Clin Oncol 2011; 29: 3351–57. Guiu S, Gauthier M, Coudert B, et al. Pathological complete response and survival according to the level of HER-2 amplification after trastuzumabbased neoadjuvant therapy for breast cancer. Br J Cancer 2010; 103: 1335–42. Dowsett M, Procter M, McCaskill-Stevens W, et al. Disease-free survival according to degree of HER2 amplification for patients treated with adjuvant chemotherapy with or without 1 year of trastuzumab: the HERA trial. J Clin Oncol 2009; 27: 2962–69.

The health of deaf people
Your March 17 Editorial (p 977)1 expressed concern about the quality of communication between deaf patients and various segments of the health-care system. The recognition that deafness is not just a pathological disability of hearing impairment but rather a handicap caused by difficulties in interpersonal communication2,3 is an important step towards improved health-care accessibility. We would like to share results from a study of deaf people in Israel.4 Similar difficulties to those mentioned by Alexander and colleagues2 were reported—eg, problems with booking appointments, poor communication, and scarcity of interpreters. Additionally, we found that deaf people had an increased risk of injury during emergency situations owing to untimely warning to seek shelter (usually provided by sound alarms), and difficulty in receiving information (provided mostly on radio or television, where subtitles are unmatched to the reading level of this population).4 In emergencies, system resources were stretched to the limit, leaving no slack for extra attention to special populations. The availability of translators was lower than usual, forcing use of hearing family members, compromising patients’ right to autonomy and privacy.2 In
www.thelancet.com Vol 379 June 16, 2012

*José Baselga, Evandro de Azambuja, Ian Bradbury, Richard Gelber
[email protected]
Division of Hematology/Oncology, Massachusetts General Hospital, Cancer Center, Boston, MA 02114, USA (JB); Breast European Adjuvant Study Team, Jules Bordet Institute, Brussels, Belgium (EdA); Frontier Science Scotland, Kincraig, UK (IB); Queen’s University, Belfast, UK (IB); Department of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston, MA, USA (RG); and Frontier Science and Technology Research Foundation, Boston, MA, USA (RG) 1 von Minckwitz G, Untch M, Blohmer J-U, et al. Definition and impact of pathologic complete response on prognosis after neoadjuvant chemotherapy in various intrinsic breast cancer subtypes. J Clin Oncol 2012; 30: 1796–804. Gianni L, Eiermann W, Semiglazov V, et al. Neoadjuvant chemotherapy with trastuzumab followed by adjuvant trastuzumab versus neoadjuvant chemotherapy alone, in patients with HER2-positive locally advanced breast cancer (the NOAH trial): a randomised controlled superiority trial with a parallel HER2-negative cohort. Lancet 2010; 375: 377–84.

2

Correspondence

Authors’ reply
In reply to Tetsuji Fujita, the incidence and prognostic effect of pathological complete response does vary between the different intrinsic subtypes of breast cancer.1 In HER2positive breast cancer, two large randomised studies2,3 have shown that pathological complete response correlates with disease-free survival in patients treated with anti-HER2 therapy. Additional information is expected from the NeoALTTO study once the disease-free survival data become available. On the other hand, the NASBP B27 study included a global breast cancer population, so its findings might not apply specifically to HER2-positive breast cancer. The mentioned correlation between levels of HER2 amplification and pathological complete response,4 although interesting, is from a small retrospective sample without a validated HER2 amplification cutoff. In the HERA trial5 of adjuvant trastuzumab, there was no difference in trastuzumab benefit according to either copy number or HER2 ratio in 3401 patients. In summary, we maintain our belief that pathological complete response is likely to be an early indicator of benefit of HER2targeted agents in neoadjuvant studies. Katya Brede-Hekimian addresses some issues similar to those of Fujita. Additionally, the reason why some tumours relapse after a pathological complete response is achieved might be explained on the basis of as yet unknown biological determinants of metastasis, the presence of a high systemic tumour burden before the start of therapy, tumour heterogeneity, and the existence of local host conditions that might favour the regrowth of tumours in certain metastatic niches. The observation that the higher pathological complete response rate seen in the combination group did not result in a higher rate of breastconserving surgery deserves careful
2238

analysis. At first glance, even in patients with a complete radiological response, the proportion of patients that underwent a mastectomy was high. We are currently analysing the reasons for this apparent discordancy and we do not believe that it is related to patients’ preferences alone. Finally, we agree that the patients who do not achieve a pathological complete response have a worse prognosis and that the potential implications of the lack of such a response have to be explained to all patients who receive neoadjuvant therapy, taking into consideration the tumour subtype, the administered therapy, the duration of therapy, and the planned postoperative (adjuvant) therapy. Additionally, the launch of clinical studies with novel antiHER2 treatments is currently being considered in patients with HER2positive tumours who do not achieve a pathological complete response with conventional anti-HER2 therapy.
JB has received honoraria from Roche. EdA has served on an advisory board and received a travel grant from GlaxoSmithKline, and has been a speaker for Roche. IB’s institution has received funding from GlaxoSmithKline and Roche. RG’s institution has received research funding from GlaxoSmithKline and Roche.

3

4

5

Untch M, Fasching PA, Konecny GE, et al. Pathologic complete response after neoadjuvant chemotherapy plus trastuzumab predicts favorable survival in human epidermal growth factor receptor 2–overexpressing breast cancer: results from the TECHNO trial of the AGO and GBG study groups. J Clin Oncol 2011; 29: 3351–57. Guiu S, Gauthier M, Coudert B, et al. Pathological complete response and survival according to the level of HER-2 amplification after trastuzumabbased neoadjuvant therapy for breast cancer. Br J Cancer 2010; 103: 1335–42. Dowsett M, Procter M, McCaskill-Stevens W, et al. Disease-free survival according to degree of HER2 amplification for patients treated with adjuvant chemotherapy with or without 1 year of trastuzumab: the HERA trial. J Clin Oncol 2009; 27: 2962–69.

The health of deaf people
Your March 17 Editorial (p 977)1 expressed concern about the quality of communication between deaf patients and various segments of the health-care system. The recognition that deafness is not just a pathological disability of hearing impairment but rather a handicap caused by difficulties in interpersonal communication2,3 is an important step towards improved health-care accessibility. We would like to share results from a study of deaf people in Israel.4 Similar difficulties to those mentioned by Alexander and colleagues2 were reported—eg, problems with booking appointments, poor communication, and scarcity of interpreters. Additionally, we found that deaf people had an increased risk of injury during emergency situations owing to untimely warning to seek shelter (usually provided by sound alarms), and difficulty in receiving information (provided mostly on radio or television, where subtitles are unmatched to the reading level of this population).4 In emergencies, system resources were stretched to the limit, leaving no slack for extra attention to special populations. The availability of translators was lower than usual, forcing use of hearing family members, compromising patients’ right to autonomy and privacy.2 In
www.thelancet.com Vol 379 June 16, 2012

*José Baselga, Evandro de Azambuja, Ian Bradbury, Richard Gelber
[email protected]
Division of Hematology/Oncology, Massachusetts General Hospital, Cancer Center, Boston, MA 02114, USA (JB); Breast European Adjuvant Study Team, Jules Bordet Institute, Brussels, Belgium (EdA); Frontier Science Scotland, Kincraig, UK (IB); Queen’s University, Belfast, UK (IB); Department of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston, MA, USA (RG); and Frontier Science and Technology Research Foundation, Boston, MA, USA (RG) 1 von Minckwitz G, Untch M, Blohmer J-U, et al. Definition and impact of pathologic complete response on prognosis after neoadjuvant chemotherapy in various intrinsic breast cancer subtypes. J Clin Oncol 2012; 30: 1796–804. Gianni L, Eiermann W, Semiglazov V, et al. Neoadjuvant chemotherapy with trastuzumab followed by adjuvant trastuzumab versus neoadjuvant chemotherapy alone, in patients with HER2-positive locally advanced breast cancer (the NOAH trial): a randomised controlled superiority trial with a parallel HER2-negative cohort. Lancet 2010; 375: 377–84.

2

Correspondence

an era of security risk during mass gatherings and occasional bombings of civilian populations, the system should be constantly ready. Finally, barriers to health-care provision are also obstacles to studying the health of deaf people. Recruitment of a researcher who is part of the deaf community enables a bottom-up approach towards problems and their solutions and is highly recommended, wherever possible.4
We declare that we have no conflicts of interest.

*Limor Aharonson-Daniel, Carolina Tannenbaum-Baruchi, Paula Feder-Bubis
[email protected]
Department of Emergency Medicine and Department of Health Systems Management, Faculty of Health Sciences, and PREPARED Center for Emergency Response Research, Ben-Gurion University of the Negev, PO Box 653, 84105 Beer-Sheva, Israel 1 The Lancet. The health of deaf people: communication breakdown. Lancet 2012; 379: 977. Alexander A, Ladd P, Powell S. Deafness might damage your health. Lancet 2012; 379: 979–81. Fellinger J, Holzinger D, Pollard R. Mental health of deaf people. Lancet 2012; 379: 1037–44. Tannenbaum-Baruchi C, Aharonson-Daniel L, Feder-Bubis P. The needs of deaf people during an emergency situation in Southern Israel. Presentated at the IPREDII conference, Tel-Aviv, Israel; January, 2012.

history-taking but it is often the focus of the assessment, providing essential clues about cognitive function. Deaf people display different cognitive competencies that arise through auditory deprivation and exposure to visuospatial, rather than spoken, language, but no neuropsychological tests are validated or normed for use with deaf signers. Even use of highly experienced interpreters is inappropriate, unreliable, and errorprone,5 since psychological validity is lost in translation. The development of specialist neurology services for deaf patients is vital to overcome these problems. A new monthly Cognitive Disorder Clinic for deaf patients at the UK’s National Hospital for Neurology and Neurosurgery already has a 6-month waiting list. We suggest that this is the tip of the iceberg.
We declare that we have no conflicts of interest.

discovery science, right across the UK. This research is crucial for subsequent translation into improvements in the prevention, diagnosis, and management of disease. Over the past few years, the MRC has maintained a basic medical science spend of more than £400 million per year. Furthermore, the MRC has protected 3-year project grants and early career fellowships, with 113 and 97, respectively, awarded in 2010–11. Indeed, 2011–12 has proved to be a record year for new investigator project grants, with 26 awards and a 28% success rate, based on our research boards’ enthusiasm for iterative interaction with colleagues starting out as university-funded investigators. Early-career basic scientists in the UK’s research-intensive medical schools should take heart—the MRC is still here to support you, because your research will help change lives.2
I am Chief Executive of the MRC. I am also head of the College of Medicine and Veterinary Medicine at the University of Edinburgh.

2 3 4

*Joanna Atkinson, Bencie Woll
[email protected]
Deafness, Cognition and Language Research Centre, University College London, London WC1H 0PD, UK 1 The Lancet. The health of deaf people: communication breakdown. Lancet 2012; 379: 977. Marshall J, Atkinson JA, Thacker A, Woll B. Is speech and language therapy meeting the needs of language minorities? The case of deaf people with neurological impairments. Int J Lang Comm Disord 2003; 38: 85–94. Atkinson J, Denmark T, Woll B, et al. Deaf with dementia: towards better recognition and services. J Dementia Care 2011; 19: 38–39. Parker J, Young A, Rogers K. “My Mum’s story”: a deaf daughter discusses her deaf mother’s experience of dementia. Dementia 2010; 9: 5–20. Hill-Briggs F, Dial JG, Morere DA, Joyce A. Neuropsychological assessment of persons with physical disability, visual impairment or blindness, and hearing impairment or deafness. Arch Clin Neuropsychol 2007; 22: 389–404.

John Savill
john.savill@headoffice.mrc.ac.uk
Medical Research Council, London WC2B 4AN, UK 1 2 The Lancet. Catastrophic neglect of the basic sciences in medicine. Lancet 2012; 379: 1273. Medical Research Council. Research changes lives: MRC strategic plan 2009–2014. http:// www.mrc.ac.uk/About/Strategy/ StrategicPlan2009-2014/index.htm (accessed May 23, 2012).

2

It was heartening to see The Lancet1 outline the barriers that deaf sign language users face in accessing general and mental health care. We would like to draw attention to a third overlooked arena: the lack of access to neurology services. Neurological disorders affect an estimated 6000–8000 deaf signers in the UK alone. These individuals experience serious under-representation in clinical referrals and inappropriate assessment, with adverse implications for accurate and timely diagnosis and treatment.2,3 Our current study of dementia in deaf people shows a pattern of late diagnosis. A typical experience is described by Parker and colleagues.4 Accurate diagnosis in the early stage of neurological disease is a particular challenge in deaf patients since not only is communication a barrier to
www.thelancet.com Vol 379 June 16, 2012

3

4

5

Catastrophic neglect of basic sciences in medicine
Your doom-laden Editorial (April 7, p 1273)1 glosses over the UK Medical Research Council’s (MRC’s) strong and continuing commitment to funding

Assuming that the ultimate goal of basic scientists and health professionals (and The Lancet) is to improve and sustain the wellbeing of humanity, I wish to place your Editorial1 on the neglect of the basic sciences in a wider context than that of medicine. From this perspective, what one observes is a neglect of basic services leading to catastrophic health consequences—eg, 8·5% of deaths in south and southeast Asia are due to diarrhoea, of which 88% results from poor sanitation and dirty water.2 This is despite the availability of cost-effective and simple technology (ie, pipe-borne water) for centuries.3 Inappropriate

For the Deaf People with Dementia Research Project see http://www.nursing.manchester. ac.uk/deafwithdementia/

2239

UIG via Getty Images

Correspondence

an era of security risk during mass gatherings and occasional bombings of civilian populations, the system should be constantly ready. Finally, barriers to health-care provision are also obstacles to studying the health of deaf people. Recruitment of a researcher who is part of the deaf community enables a bottom-up approach towards problems and their solutions and is highly recommended, wherever possible.4
We declare that we have no conflicts of interest.

*Limor Aharonson-Daniel, Carolina Tannenbaum-Baruchi, Paula Feder-Bubis
[email protected]
Department of Emergency Medicine and Department of Health Systems Management, Faculty of Health Sciences, and PREPARED Center for Emergency Response Research, Ben-Gurion University of the Negev, PO Box 653, 84105 Beer-Sheva, Israel 1 The Lancet. The health of deaf people: communication breakdown. Lancet 2012; 379: 977. Alexander A, Ladd P, Powell S. Deafness might damage your health. Lancet 2012; 379: 979–81. Fellinger J, Holzinger D, Pollard R. Mental health of deaf people. Lancet 2012; 379: 1037–44. Tannenbaum-Baruchi C, Aharonson-Daniel L, Feder-Bubis P. The needs of deaf people during an emergency situation in Southern Israel. Presentated at the IPREDII conference, Tel-Aviv, Israel; January, 2012.

history-taking but it is often the focus of the assessment, providing essential clues about cognitive function. Deaf people display different cognitive competencies that arise through auditory deprivation and exposure to visuospatial, rather than spoken, language, but no neuropsychological tests are validated or normed for use with deaf signers. Even use of highly experienced interpreters is inappropriate, unreliable, and errorprone,5 since psychological validity is lost in translation. The development of specialist neurology services for deaf patients is vital to overcome these problems. A new monthly Cognitive Disorder Clinic for deaf patients at the UK’s National Hospital for Neurology and Neurosurgery already has a 6-month waiting list. We suggest that this is the tip of the iceberg.
We declare that we have no conflicts of interest.

discovery science, right across the UK. This research is crucial for subsequent translation into improvements in the prevention, diagnosis, and management of disease. Over the past few years, the MRC has maintained a basic medical science spend of more than £400 million per year. Furthermore, the MRC has protected 3-year project grants and early career fellowships, with 113 and 97, respectively, awarded in 2010–11. Indeed, 2011–12 has proved to be a record year for new investigator project grants, with 26 awards and a 28% success rate, based on our research boards’ enthusiasm for iterative interaction with colleagues starting out as university-funded investigators. Early-career basic scientists in the UK’s research-intensive medical schools should take heart—the MRC is still here to support you, because your research will help change lives.2
I am Chief Executive of the MRC. I am also head of the College of Medicine and Veterinary Medicine at the University of Edinburgh.

2 3 4

*Joanna Atkinson, Bencie Woll
[email protected]
Deafness, Cognition and Language Research Centre, University College London, London WC1H 0PD, UK 1 The Lancet. The health of deaf people: communication breakdown. Lancet 2012; 379: 977. Marshall J, Atkinson JA, Thacker A, Woll B. Is speech and language therapy meeting the needs of language minorities? The case of deaf people with neurological impairments. Int J Lang Comm Disord 2003; 38: 85–94. Atkinson J, Denmark T, Woll B, et al. Deaf with dementia: towards better recognition and services. J Dementia Care 2011; 19: 38–39. Parker J, Young A, Rogers K. “My Mum’s story”: a deaf daughter discusses her deaf mother’s experience of dementia. Dementia 2010; 9: 5–20. Hill-Briggs F, Dial JG, Morere DA, Joyce A. Neuropsychological assessment of persons with physical disability, visual impairment or blindness, and hearing impairment or deafness. Arch Clin Neuropsychol 2007; 22: 389–404.

John Savill
john.savill@headoffice.mrc.ac.uk
Medical Research Council, London WC2B 4AN, UK 1 2 The Lancet. Catastrophic neglect of the basic sciences in medicine. Lancet 2012; 379: 1273. Medical Research Council. Research changes lives: MRC strategic plan 2009–2014. http:// www.mrc.ac.uk/About/Strategy/ StrategicPlan2009-2014/index.htm (accessed May 23, 2012).

2

It was heartening to see The Lancet1 outline the barriers that deaf sign language users face in accessing general and mental health care. We would like to draw attention to a third overlooked arena: the lack of access to neurology services. Neurological disorders affect an estimated 6000–8000 deaf signers in the UK alone. These individuals experience serious under-representation in clinical referrals and inappropriate assessment, with adverse implications for accurate and timely diagnosis and treatment.2,3 Our current study of dementia in deaf people shows a pattern of late diagnosis. A typical experience is described by Parker and colleagues.4 Accurate diagnosis in the early stage of neurological disease is a particular challenge in deaf patients since not only is communication a barrier to
www.thelancet.com Vol 379 June 16, 2012

3

4

5

Catastrophic neglect of basic sciences in medicine
Your doom-laden Editorial (April 7, p 1273)1 glosses over the UK Medical Research Council’s (MRC’s) strong and continuing commitment to funding

Assuming that the ultimate goal of basic scientists and health professionals (and The Lancet) is to improve and sustain the wellbeing of humanity, I wish to place your Editorial1 on the neglect of the basic sciences in a wider context than that of medicine. From this perspective, what one observes is a neglect of basic services leading to catastrophic health consequences—eg, 8·5% of deaths in south and southeast Asia are due to diarrhoea, of which 88% results from poor sanitation and dirty water.2 This is despite the availability of cost-effective and simple technology (ie, pipe-borne water) for centuries.3 Inappropriate

For the Deaf People with Dementia Research Project see http://www.nursing.manchester. ac.uk/deafwithdementia/

2239

UIG via Getty Images

Correspondence

attention to basic sciences could also unduly delay implementation of such simple interventions, resulting in unnecessary death and suffering.4 The conceit is not that “epidemiology is the basic science of clinical medicine”.1 The real conceit is to believe that health issues are mainly solved by basic scientists and health professionals. The truth, although unpalatable to some, is that health-related problems are mostly the result of the social and physical environment during our life course, and medical care accounts for only about 5 years of the 30 years gained in life expectancy during the 20th century.5
I declare that I have no conflicts of interest.

Saroj Jayasinghe
sarojoffi[email protected]
Department of Clinical Medicine, Faculty of Medicine, University of Colombo, Colombo, Sri Lanka 1 2 The Lancet. Catastrophic neglect of the basic sciences in medicine. Lancet 2012; 379: 1273. WHO, UNICEF. Diarrhoea: why children are still dying and what can be done. New York/ Geneva: WHO/UNICEF, 2010. http://www.who. int/maternal_child_adolescent/documents/ 9789241598415/en/index.html (accessed April 7, 2012). Jalan J, Ravallion M. Does piped water reduce diarrhea for children in rural India? J Econom 2003; 112: 153–73. Wickremasinghe AR, Peiris-John RJ, Wanigasuriya KP. Chronic kidney disease of unknown aetiology in the North Central Province of Sri Lanka: trying to unravel the mystery. Ceylon Med J 2011; 56: 144–45. Gold MR, Teutsch S. For a healthy nation: returns on investment in public health. Washington, DC: Department of Health and Human Services, 1994.

3

4

5

control achieved by insulin degludec compared with insulin glargine. Each focuses its main conclusion not on this primary outcome, but on one of several secondary measurements: nocturnal hypoglycaemia in the first paper and overall hypoglycaemia in the second. In both, the difference was of marginal significance and no mention is made of adjustment for multiple testing. These lower hypoglycaemia rates in unblinded studies should be considered, at best, hypothesis generating. At worst they are spurious. The papers focus on these secondary outcomes in such a way as to encourage clinicians to believe they warrant the selection of insulin degludec over alternatives. The Lancet’s reprints are a major source of revenue for the journal,3 and a major part of drug company marketing. These trials were written and analysed by NovoNordisk statisticians and NovoNordisk-funded professional writers. We applaud their skill, but regret the lack of editorial effort deployed to balance it. Commenting on dubious marketing strategies for diabetic drugs, Richard Horton recently tweeted, “We had at least 10 RCTs today where the sponsor had done absolutely everything, including writing the manuscript. What are authors for?” The same question might be asked of editors.
We declare that we have no conflicts of interest.

3

Lundh A, Barbateskovic M, Hróbjartsson A, Gøtzsche PC. Conflicts of interest at medical journals: the influence of industry-supported randomised trials on journal impact factors and revenue—cohort study. PLoS Med 2010; 7: e1000354.

Health problems in the temporary housing in Fukushima
We thank Justin McCurry for providing an excellent report (March 10, p 880)1 on the Great East Japan Earthquake 1 year on. We would like to add some comments based on our experience in Fukushima. Since November, 2011, we have been stationed near the temporary housing in Minamisoma City, in Hamadouri district, which is close to the evacuation zone of 20 km from the Fukushima nuclear power plant. Here we have been providing health checks and services such as vaccination programmes for more than 4000 evacuees. One of the most tragic aspects of the disaster is the break-up of communities and families because of the fear of radiation exposure. The population of 72 000 in Minamisoma City dropped to about 10 000 just after the nuclear disaster and only recovered up to about 43 000 in March, 2012. Of note, the proportion of those aged 65 years or older has increased from 25·9% to 32·2%. Many young families moved out of the city, which resulted in a sudden increase in the number of frail elderly people living alone in temporary housing. For example, we saw a 75-year-old man who was found in a hypovolaemic state in his small and chilly temporary accommodation; he died 2 weeks later. In Fukushima, 1323 of 10 664 evacuee households in January, 2012, were aged 65 years or older and lived alone, and officials reported that there were 22 cases of solitary death (15 of whom were aged 65 years or older) in temporary housing in three devastated prefectures; there were
www.thelancet.com Vol 379 June 16, 2012

*Druin Burch, Marion Mafham, John S Yudkin

What are editors for?
The two trials of insulin degludec (April 21, pp 1489 and 1498)1,2 share characteristics we think worthy of notice. Each deals with a common disease, yet recruits fewer than ten patients per centre—an inefficient way of gathering scientific data, although a good means to get a large number of units used to prescribing new drugs. Both trials produce the same finding: there does not seem to be a big difference in the glycaemic
2240

[email protected]
John Radcliffe Hospital, Oxford University Hospitals NHS Trust, Oxford OX3 9DU, UK (DB); Oxford University Hospitals NHS Trust, Oxford, UK (MM); and University College, London, UK (JSY) 1 Heller S, Buse J, Fisher M, et al. Insulin degludec, an ultra-longacting basal insulin, versus insulin glargine in basal-bolus treatment with mealtime insulin aspart in type 1 diabetes (BEGIN Basal-Bolus Type 1): a phase 3, randomised, open-label, treat-to-target noninferiority trial. Lancet 2012; 379: 1489–97. Garber AJ, King AB, Del Prato S, et al. Insulin degludec, an ultra-longacting basal insulin, versus insulin glargine in basal-bolus treatment with mealtime insulin aspart in type 2 diabetes (BEGIN Basal-Bolus Type 2): a phase 3, randomised, open-label, treat-to-target noninferiority trial. Lancet 2012; 379: 1498–507.

2

Correspondence

attention to basic sciences could also unduly delay implementation of such simple interventions, resulting in unnecessary death and suffering.4 The conceit is not that “epidemiology is the basic science of clinical medicine”.1 The real conceit is to believe that health issues are mainly solved by basic scientists and health professionals. The truth, although unpalatable to some, is that health-related problems are mostly the result of the social and physical environment during our life course, and medical care accounts for only about 5 years of the 30 years gained in life expectancy during the 20th century.5
I declare that I have no conflicts of interest.

Saroj Jayasinghe
sarojoffi[email protected]
Department of Clinical Medicine, Faculty of Medicine, University of Colombo, Colombo, Sri Lanka 1 2 The Lancet. Catastrophic neglect of the basic sciences in medicine. Lancet 2012; 379: 1273. WHO, UNICEF. Diarrhoea: why children are still dying and what can be done. New York/ Geneva: WHO/UNICEF, 2010. http://www.who. int/maternal_child_adolescent/documents/ 9789241598415/en/index.html (accessed April 7, 2012). Jalan J, Ravallion M. Does piped water reduce diarrhea for children in rural India? J Econom 2003; 112: 153–73. Wickremasinghe AR, Peiris-John RJ, Wanigasuriya KP. Chronic kidney disease of unknown aetiology in the North Central Province of Sri Lanka: trying to unravel the mystery. Ceylon Med J 2011; 56: 144–45. Gold MR, Teutsch S. For a healthy nation: returns on investment in public health. Washington, DC: Department of Health and Human Services, 1994.

3

4

5

control achieved by insulin degludec compared with insulin glargine. Each focuses its main conclusion not on this primary outcome, but on one of several secondary measurements: nocturnal hypoglycaemia in the first paper and overall hypoglycaemia in the second. In both, the difference was of marginal significance and no mention is made of adjustment for multiple testing. These lower hypoglycaemia rates in unblinded studies should be considered, at best, hypothesis generating. At worst they are spurious. The papers focus on these secondary outcomes in such a way as to encourage clinicians to believe they warrant the selection of insulin degludec over alternatives. The Lancet’s reprints are a major source of revenue for the journal,3 and a major part of drug company marketing. These trials were written and analysed by NovoNordisk statisticians and NovoNordisk-funded professional writers. We applaud their skill, but regret the lack of editorial effort deployed to balance it. Commenting on dubious marketing strategies for diabetic drugs, Richard Horton recently tweeted, “We had at least 10 RCTs today where the sponsor had done absolutely everything, including writing the manuscript. What are authors for?” The same question might be asked of editors.
We declare that we have no conflicts of interest.

3

Lundh A, Barbateskovic M, Hróbjartsson A, Gøtzsche PC. Conflicts of interest at medical journals: the influence of industry-supported randomised trials on journal impact factors and revenue—cohort study. PLoS Med 2010; 7: e1000354.

Health problems in the temporary housing in Fukushima
We thank Justin McCurry for providing an excellent report (March 10, p 880)1 on the Great East Japan Earthquake 1 year on. We would like to add some comments based on our experience in Fukushima. Since November, 2011, we have been stationed near the temporary housing in Minamisoma City, in Hamadouri district, which is close to the evacuation zone of 20 km from the Fukushima nuclear power plant. Here we have been providing health checks and services such as vaccination programmes for more than 4000 evacuees. One of the most tragic aspects of the disaster is the break-up of communities and families because of the fear of radiation exposure. The population of 72 000 in Minamisoma City dropped to about 10 000 just after the nuclear disaster and only recovered up to about 43 000 in March, 2012. Of note, the proportion of those aged 65 years or older has increased from 25·9% to 32·2%. Many young families moved out of the city, which resulted in a sudden increase in the number of frail elderly people living alone in temporary housing. For example, we saw a 75-year-old man who was found in a hypovolaemic state in his small and chilly temporary accommodation; he died 2 weeks later. In Fukushima, 1323 of 10 664 evacuee households in January, 2012, were aged 65 years or older and lived alone, and officials reported that there were 22 cases of solitary death (15 of whom were aged 65 years or older) in temporary housing in three devastated prefectures; there were
www.thelancet.com Vol 379 June 16, 2012

*Druin Burch, Marion Mafham, John S Yudkin

What are editors for?
The two trials of insulin degludec (April 21, pp 1489 and 1498)1,2 share characteristics we think worthy of notice. Each deals with a common disease, yet recruits fewer than ten patients per centre—an inefficient way of gathering scientific data, although a good means to get a large number of units used to prescribing new drugs. Both trials produce the same finding: there does not seem to be a big difference in the glycaemic
2240

[email protected]
John Radcliffe Hospital, Oxford University Hospitals NHS Trust, Oxford OX3 9DU, UK (DB); Oxford University Hospitals NHS Trust, Oxford, UK (MM); and University College, London, UK (JSY) 1 Heller S, Buse J, Fisher M, et al. Insulin degludec, an ultra-longacting basal insulin, versus insulin glargine in basal-bolus treatment with mealtime insulin aspart in type 1 diabetes (BEGIN Basal-Bolus Type 1): a phase 3, randomised, open-label, treat-to-target noninferiority trial. Lancet 2012; 379: 1489–97. Garber AJ, King AB, Del Prato S, et al. Insulin degludec, an ultra-longacting basal insulin, versus insulin glargine in basal-bolus treatment with mealtime insulin aspart in type 2 diabetes (BEGIN Basal-Bolus Type 2): a phase 3, randomised, open-label, treat-to-target noninferiority trial. Lancet 2012; 379: 1498–507.

2

Correspondence

at least 1465 deaths in evacuees by February, 2012.2,3 Fear of radiation exposure is also no exception among medical professionals and their families. In Minamisoma district, the number of staff doctors and nurses in December, 2011, decreased from 120 to 61, and from 1219 to 938, respectively. Because radiation doses measured so far are relatively small in Minamisoma City, more support is needed to increase the number of medical staff for this vulnerable area.
We declare that we have no conflicts of interest.

*Keitaro Harasawa, Tetsuya Tanimoto, Masahiro Kami, Tomoyoshi Oikawa, Yukio Kanazawa, Hideki Komatsu
[email protected]
Minamisoma Municipal General Hospital, Minamisoma, Fukushima 975-8686, Japan (KY, TO, YK); Jyoban Hospital, Tokiwakai Group, Fukushima, Japan (TT); Division of Social Communication for Advanced Clinical Research, Institute of Medical Science, University of Tokyo, Tokyo, Japan (TT, MK); and Kameda General Hospital, Chiba, Japan (HK) 1 2 McCurry J. Japan’s Tohoku earthquake: 1 year on. Lancet 2012; 379: 880–81. Anon. ‘Lonely death’ in evacuee community prompts renewed look at safety check system. Mainichi Daily News Feb 28, 2012. Anon. Urgent need for prevention of solitary deaths among evacuees. Mainich Shimbun March 11, 2012.

3

Syria: public health achievements and sanctions
The past year has witnessed major turbulence in Syria, with several thousand deaths and injuries in what is now incipient civil war. Sadly, interference from outside has meant that the “hire-a-gun crowd”—mainly unemployed young people from different regions of the Middle East— have been funded to travel and fight with the “free army”, many of whom are self-declared Islamic jihadists in one of the few secular countries in the region. Amid the political wrangling and rampage, we fear for the safety of civilians caught between both sides.
www.thelancet.com Vol 379 June 16, 2012

Few observers have even considered the health system, which we as public health doctors and teachers have helped to create painstakingly over the past two decades. Key achievements include: reorganisation of health services to increase local control as part of the process of decentralisation; systematisation of data and trends through a health metrics network and national accounts; significant declines in maternal and child mortality over the past three decades; progress in plans to address non-communicable diseases through partnerships with the private and not-for-profit sectors; and, since 2003, the INTRA Project in coordination with WHO, which aims to involve older people in redesigning health services at local level. For more than 1 year Syria has faced economic sanctions the direct and indirect effects of which are devastating. The value of the Syrian pound has collapsed, and the price of basic essentials such as milk, eggs, and rice has more than doubled. The cost of heating oil has increased more than threefold, with a direct effect on the ability of health services to function (eg, the vaccine chain). With thousands of job losses, the ability of people to afford basic essentials and medicines for vulnerable groups such as pregnant women, children, and older people has become nigh on impossible. Sanctions have also affected the ability to maintain clean water supplies, with increased concerns for waterborne infections and diarrhoeal diseases, especially in children. The plight of those fleeing from violence in refugee camps is dismal and a source of serious psychological distress, especially for children who do not understand what is happening. The health and human rights agenda is on the ascendance, with a collection of dedicated scholars and lawyers on board who are determined to defend the right to accessible, quality health care. The situation in Syria for most of the population is an urgent one. Should economic

and political sanctions be allowed to punish whole populations?
We declare that we have no conflicts of interest.

*Waleed Al Faisal, Yaser Al Saleh, Kasturi Sen
[email protected]
c/o The Lancet, 32 Jamestown Road, London NW1 7BY, UK

Health of Palestinian people in the ghettos: from Gaza to Shatila
Finally, I managed to get out of Gaza! For two successive years, I had been able to participate in scientific events outside the Strip only through video conferencing. This year, with a few others from Gaza, I passed through Egypt, successfully arriving in Lebanon to present a paper at the third Lancet Palestinian Health Alliance Conference at the American University of Beirut. Back in 2001, I had been admitted to the MPH programme at the Institute of Community and Public Health at Birzeit University in the West Bank; unfortunately, the sealing of Gaza prevented me from joining my classes there. At the conference, I met the professors who would have taught me public health. Among them was Rita Giacaman, founder of the Institute of Community and Public Health, with whom I had been in touch for more than 10 years, but only via email. In her welcome remarks, Rita coined the term ”PPES” —post-prison excitement syndrome—pointing to the irony of our inability, as Palestinians, to meet each other inside our country, and our huge excitement when we bump into each other abroad. The 2-day conference saw a smorgasbord of topics and researchers brought to the table. Entitled “Health of Palestinians inside and outside the occupied Palestinian territories”, the conference put Palestinian health back on its feet and could represent a paradigm shift in the way the health

Published Online June 1, 2012 DOI:10.1016/S01406736(12)60871-X

2241

Bassem Tellawi/AP/Press Association Images

Correspondence

at least 1465 deaths in evacuees by February, 2012.2,3 Fear of radiation exposure is also no exception among medical professionals and their families. In Minamisoma district, the number of staff doctors and nurses in December, 2011, decreased from 120 to 61, and from 1219 to 938, respectively. Because radiation doses measured so far are relatively small in Minamisoma City, more support is needed to increase the number of medical staff for this vulnerable area.
We declare that we have no conflicts of interest.

*Keitaro Harasawa, Tetsuya Tanimoto, Masahiro Kami, Tomoyoshi Oikawa, Yukio Kanazawa, Hideki Komatsu
[email protected]
Minamisoma Municipal General Hospital, Minamisoma, Fukushima 975-8686, Japan (KY, TO, YK); Jyoban Hospital, Tokiwakai Group, Fukushima, Japan (TT); Division of Social Communication for Advanced Clinical Research, Institute of Medical Science, University of Tokyo, Tokyo, Japan (TT, MK); and Kameda General Hospital, Chiba, Japan (HK) 1 2 McCurry J. Japan’s Tohoku earthquake: 1 year on. Lancet 2012; 379: 880–81. Anon. ‘Lonely death’ in evacuee community prompts renewed look at safety check system. Mainichi Daily News Feb 28, 2012. Anon. Urgent need for prevention of solitary deaths among evacuees. Mainich Shimbun March 11, 2012.

3

Syria: public health achievements and sanctions
The past year has witnessed major turbulence in Syria, with several thousand deaths and injuries in what is now incipient civil war. Sadly, interference from outside has meant that the “hire-a-gun crowd”—mainly unemployed young people from different regions of the Middle East— have been funded to travel and fight with the “free army”, many of whom are self-declared Islamic jihadists in one of the few secular countries in the region. Amid the political wrangling and rampage, we fear for the safety of civilians caught between both sides.
www.thelancet.com Vol 379 June 16, 2012

Few observers have even considered the health system, which we as public health doctors and teachers have helped to create painstakingly over the past two decades. Key achievements include: reorganisation of health services to increase local control as part of the process of decentralisation; systematisation of data and trends through a health metrics network and national accounts; significant declines in maternal and child mortality over the past three decades; progress in plans to address non-communicable diseases through partnerships with the private and not-for-profit sectors; and, since 2003, the INTRA Project in coordination with WHO, which aims to involve older people in redesigning health services at local level. For more than 1 year Syria has faced economic sanctions the direct and indirect effects of which are devastating. The value of the Syrian pound has collapsed, and the price of basic essentials such as milk, eggs, and rice has more than doubled. The cost of heating oil has increased more than threefold, with a direct effect on the ability of health services to function (eg, the vaccine chain). With thousands of job losses, the ability of people to afford basic essentials and medicines for vulnerable groups such as pregnant women, children, and older people has become nigh on impossible. Sanctions have also affected the ability to maintain clean water supplies, with increased concerns for waterborne infections and diarrhoeal diseases, especially in children. The plight of those fleeing from violence in refugee camps is dismal and a source of serious psychological distress, especially for children who do not understand what is happening. The health and human rights agenda is on the ascendance, with a collection of dedicated scholars and lawyers on board who are determined to defend the right to accessible, quality health care. The situation in Syria for most of the population is an urgent one. Should economic

and political sanctions be allowed to punish whole populations?
We declare that we have no conflicts of interest.

*Waleed Al Faisal, Yaser Al Saleh, Kasturi Sen
[email protected]
c/o The Lancet, 32 Jamestown Road, London NW1 7BY, UK

Health of Palestinian people in the ghettos: from Gaza to Shatila
Finally, I managed to get out of Gaza! For two successive years, I had been able to participate in scientific events outside the Strip only through video conferencing. This year, with a few others from Gaza, I passed through Egypt, successfully arriving in Lebanon to present a paper at the third Lancet Palestinian Health Alliance Conference at the American University of Beirut. Back in 2001, I had been admitted to the MPH programme at the Institute of Community and Public Health at Birzeit University in the West Bank; unfortunately, the sealing of Gaza prevented me from joining my classes there. At the conference, I met the professors who would have taught me public health. Among them was Rita Giacaman, founder of the Institute of Community and Public Health, with whom I had been in touch for more than 10 years, but only via email. In her welcome remarks, Rita coined the term ”PPES” —post-prison excitement syndrome—pointing to the irony of our inability, as Palestinians, to meet each other inside our country, and our huge excitement when we bump into each other abroad. The 2-day conference saw a smorgasbord of topics and researchers brought to the table. Entitled “Health of Palestinians inside and outside the occupied Palestinian territories”, the conference put Palestinian health back on its feet and could represent a paradigm shift in the way the health

Published Online June 1, 2012 DOI:10.1016/S01406736(12)60871-X

2241

Bassem Tellawi/AP/Press Association Images

Correspondence

of the Palestinian people is addressed. The shift is from the health of different populations in various isolated territories, which have traditionally been assessed and addressed separately, to the health of one dispossessed nation spanning the Levant, which should be tackled as a whole. For decades after the dispossession of Palestinians in 1948, their health was treated as the health of Arab refugees scattered in the Near East. The occupation of the West Bank and Gaza Strip in 1967 and the way the Palestinian socioeconomic sphere, including health, was dedeveloped by Israel, forced health activists on the ground to put health in the occupied Palestinian territories (OPT) on a separate itinerary, away from Palestinians elsewhere. This separation continued after the Oslo Accords and the establishment of the Palestinian National Authority (PNA) in 1994. Indeed, the creation of the PNA did away with Palestinian bodies that previously took care, albeit insufficiently, of the health of Palestinians outside Palestine. At the same time, the health system inside the OPT was addressed inappropriately: some dissimilarities between the West Bank and the Gaza Strip were ignored, while the illusion of state building under occupation remained the key agenda for two decades. A new paradigm in addressing Palestinian health affairs, as a health of one nation inside Palestine and across the borders, is not a knee-jerk reaction to the failed Oslo process, nor to the geopolitical break-up between the West Bank and the Gaza Strip. A new paradigm is dictated by the national aspirations of Palestinians and by the identical social determinants of health among them, analogous diseases and demographic patterns, and the commonality of the transborder Palestinian health-care system. Before I left Beirut, I phoned a former medical school classmate from the opulent hotel where I was
2242

staying in the shopping district of Hamra. I had not seen him for more than 20 years. Therefore, I could not turn down his invitation to go to his home in Shatila refugee camp. A world apart from glitzy Beirut, Shatila main street was accessed by car but then we had to continue on foot in the labyrinth of narrow alleys strung across with a chaotic web of electric cables and rusty water pipes. The misery of the camp was shocking. The high-rise buildings were packed tightly together, allowing only a small amount of light to filter through to the alleyways. “I do not know Beirut well”, he said when I asked him to get something from the city. “I live in Shatila and go daily to the hospital at Burj al-Barajna refugee camp. I come to Beirut only when somebody like you comes from Palestine or when my relatives who immigrated to Denmark visit us and ask for a tour inside the capital… Lebanon is not our place…we live in the camp and think about Palestine.” I realised that his world, like that of most of us Palestinians, is in the ghettos, where we have been forced to live for the past several decades.
I declare that I have no conflicts of interest.

AFP/Getty Images

Department of Error
Raviglione M, Marais B, Floyd K, et al. Scaling up interventions to achieve global tuberculosis control: progress and new developments. Lancet 2012; 379: 1902–13—In this Review (May 19), the fourth sentence of the section on tuberculosis in children (p 1906) should have read “WHO estimates that 500 000 children develop tuberculosis yearly, resulting in around 70 000 deaths.” Also, in the legend of figure 6, the last sentence should have said “Mycobacterium vaccae—originally developed for immunotherapy—has completed a phase 3 trial as a vaccine for HIV-infected Tanzanian adults.” These corrections have been made to the online version as of June 15. Avery AJ, Rodgers S, Cantrill JA, et al. A pharmacistled information technology intervention for medication errors (PINCER): a multicentre, cluster randomised, controlled trial and cost-effectiveness analysis. Lancet 2012; 379: 1310–19—In the Summary of this Article (April 7), Methods, line 4, the details of masking should have been: “The allocation was masked to researchers and statisticians involved in processing and analysing the data. The allocation was not masked to general practices, pharmacists, patients, or researchers who visited practices to extract data.” This correction has been made to the online version as of June 15, 2012.

Majdi Ashour
[email protected]
Apartment 2, Building 17, Elzhara Microcity, Gaza Strip, Occupied Palestinian Territory

www.thelancet.com Vol 379 June 16, 2012

Articles

Effect of regression from prediabetes to normal glucose regulation on long-term reduction in diabetes risk: results from the Diabetes Prevention Program Outcomes Study
Leigh Perreault, Qing Pan, Kieren J Mather, Karol E Watson, Richard F Hamman, Steven E Kahn, for the Diabetes Prevention Program Research Group

Summary
Background Our objective was to quantify and predict diabetes risk reduction during the Diabetes Prevention Program Outcomes Study (DPPOS) in participants who returned to normal glucose regulation at least once during the Diabetes Prevention Program (DPP) compared with those who consistently met criteria for prediabetes. Methods DPPOS is an ongoing observational study of participants from the DPP randomised trial. For this analysis, diabetes cumulative incidence in DPPOS was calculated for participants with normal glucose regulation or prediabetes status during DPP with and without stratification by previous randomised treatment group. Cox proportional hazards modelling and generalised linear mixed models were used to quantify the effect of previous (DPP) glycaemic status on risk of later (DPPOS) diabetes and normal glucose regulation status, respectively, per SD in change. Included in this analysis were 1990 participants of DPPOS who had been randomly assigned to treatment groups during DPP (736 intensive lifestyle intervention, 647 metformin, 607 placebo). These studies are registered at ClinicalTrials.gov, NCT00004992 (DPP) and NCT00038727 (DPPOS). Findings Diabetes risk during DPPOS was 56% lower for participants who had returned to normal glucose regulation versus those who consistently had prediabetes (hazard ratio [HR] 0·44, 95% CI 0·37–0·55, p<0·0001) and was unaffected by previous group assignment (interaction test for normal glucose regulation and lifestyle intervention, p=0·1722; normal glucose regulation and metformin, p=0·3304). Many, but not all, of the variables that increased diabetes risk were inversely associated with the chance of a participant reaching normal glucose regulation status in DPPOS. Specifically, previous achievement of normal glucose regulation (odds ratio [OR] 3·18, 95% CI 2·71–3·72, p<0·0001), increased β-cell function (OR 1·28; 95% CI 1·18–1·39, p<0·0001), and insulin sensitivity (OR 1·16, 95% CI 1·08–1·25, p<0·0001) were associated with normal glucose regulation in DPPOS, whereas the opposite was true for prediction of diabetes, with increased β-cell function (HR 0·80, 95% CI 0·71–0·89, p<0·0001) and insulin sensitivity (HR 0·83, 95% CI 0·74–0·94, p=0·0001) having a protective effect. Among participants who did not return to normal glucose regulation in DPP, those assigned to the intensive lifestyle intervention had a higher diabetes risk (HR 1·31, 95% CI 1·03–1·68, p=0·0304) and lower chance of normal glucose regulation (OR 0·59, 95% CI 0·42–0·82, p=0·0014) than did the placebo group in DPPOS. Interpretation We conclude that prediabetes is a high-risk state for diabetes, especially in patients who remain with prediabetes despite intensive lifestyle intervention. Reversion to normal glucose regulation, even if transient, is associated with a significantly reduced risk of future diabetes independent of previous treatment group. Funding US National Institutes of Health.
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60525-X See Online/Comment DOI:10.1016/S01406736(12)60828-9 University of Colorado Anschutz Medical Campus, Aurora, CO, USA (L Perreault MD); The Biostatistics Center, George Washington University, Diabetes Prevention Program Coordinating Center, Rockville, MD, USA (Q Pan); Indiana University School of Medicine, Indianapolis, IN, USA (K J Mather MD); University of California Los Angeles School of Medicine, Los Angeles, CA, USA (K E Watson MD); Colorado School of Public Health, Department of Epidemiology, Aurora, CO, USA (R F Hamman MD DrPh); and VA Puget Sound Health Care System and University of Washington, Seattle, WA, USA (S E Kahn MB ChB) Correspondence to: Diabetes Prevention Program Research Group, Diabetes Prevention Program Coordinating Center, The Biostatistics Center, George Washington University, Rockville, MD 20852, USA [email protected]

Introduction
Estimates from the US Centers for Disease Control and Prevention suggest that roughly 79 million Americans have impaired fasting glucose or impaired glucose tolerance (or both), collectively termed prediabetes.1 Screening for prediabetes is advocated because the disorder is a very high risk state for future type 2 diabetes and itself carries risk of diabetic complications.2–5 Clinical trials of diabetes prevention worldwide have largely enrolled participants with untreated prediabetes because every year roughly 11% of this group will acquire diabetes, when both fasting and postprandial plasma glucose concentrations are raised.6–12 Even when overt diabetes is delayed or prevented, both microvascular and macrovascular disease are more prevalent in those

with prediabetes compared with their normoglycaemic peers.2–5 Thus, there is reason to believe that true prevention of diabetes and its complications probably resides in the reversal of prediabetes and the restoration of normal glucose regulation. Several clinical trials have examined the feasibility and efficacy of lifestyle changes or drugs, or both, for prevention of diabetes in people with prediabetes.6–15 Together, these studies have shown reductions between 25% and 72% in incidence of diabetes during 2·4–6 year intervention periods, with most participants remaining with prediabetes. Less often discussed are the 20–50% of participants who not only did not progress to diabetes, but in fact reverted to normal glucose regulation (fasting glucose <5·6 mmol/L and 2-h
1

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

For the DPPOS protocol see http://www.bsc.gwu.edu/dpp

glucose <7·8 mmol/L).7–10 Three of these studies— STOP NIDDM,7 DREAM,9 and ACT NOW13—make particular mention of a subgroup who achieved normal glycaemic status, but offer no detail as to the demographic characteristics or predictive factors. We previously reported a post-hoc analysis from the Diabetes Prevention Program (DPP) examining predictors for the restoration of normal glucose regulation.16 This analysis revealed that improved β-cell function and young age, as well as weight loss and intensive lifestyle intervention—independent of one
158 183 patients made contact with DPP

another—were related to regression to normal glucose regulation. After completion of DPP, the Diabetes Prevention Program Outcomes Study (DPPOS) was initiated, which afforded a unique opportunity to establish the perseverance of predictors of normal glucose regulation—treatment-related or otherwise— and to quantify long-term diabetes risk reduction in participants who achieved normal glucose regulation. In pursuit of this objective, we postulated that participants reaching normal glucose regulation would have a significant and enduring decreased risk of incident diabetes compared with those with persistent prediabetes.

Methods
30 996 started OGTT 30 383 completed OGTT 17 893 shipped to laboratory

Trial design
DPPOS was an observational follow-up study to a randomised clinical trial undertaken at 27 centres involving 2761 participants who were at high risk of developing diabetes. The detailed methods have been reported17 and the protocol is available online. Institutional review boards at each centre approved the protocol, and all participants gave written informed consent before participation.

7525 eligible based on OGTT results

4720 started 3-week run-in

4078 completed 3-week run-in

Participants
3819 randomised

1082 placebo

1073 metformin

1079 lifestyle

585 troglitazone (enrolment and intervention stopped)

DPP* Year 1: n=1027 Year 2: n=1015 Year 3: n=975

DPP* Year 1: n=1017 Year 2: n=1006 Year 3: n=967

DPP* Year 1: n=1026 Year 2: n=1001 Year 3: n=972

Bridge period

Bridge period

Bridge period

1074 eligible for DPPOS†

1061 eligible for DPPOS†

1074 eligible for DPPOS†

DPPOS 932 enrolled Year 1: n=888 Year 2: n=872 Year 3: n=847 Year 4: n=830 Year 5: n=845 Year 6: n=756

DPPOS 924 enrolled Year 1: n=887 Year 2: n=856 Year 3: n=837 Year 4: n=827 Year 5: n=843 Year 6: n=745

DPPOS 910 enrolled Year 1: n=857 Year 2: n=825 Year 3: n=818 Year 4: n=814 Year 5: n=825 Year 6: n=736

Participants were followed up for a median 3·2 years during the intervention phase of DPP, which formally ended on July 31, 2001. All active DPP participants were eligible for entry into DPPOS. 2761 (85%) of the 3234 participants in DPP enrolled in DPPOS after a 13-month bridge period (the transition between DPP and DPPOS protocols) that followed the end of DPP (Aug 1, 2001, to Aug 31, 2002). Participants formerly randomly assigned to lifestyle intervention (n=909), metformin (n=921), and placebo (n=931) groups were followed up for a median 5·7 years during DPPOS (mean 5·4 years; range 0·01–5·98 years; Sept 1, 2002, to Oct 31, 2008). 1990 (72%) of the 2761 participants in DPPOS (736 lifestyle intervention, 647 metformin, 607 placebo) were classified as having persistent prediabetes or restoration of normal glucose regulation during the 5·7-year follow-up for DPPOS and were included in this analysis. Those who progressed to diabetes (173 lifestyle intervention, 274 metformin, 324 placebo) during DPPOS were excluded (figure 1).18

Procedures
DPPOS is a post-intervention ongoing observational study. Nevertheless, in view of the diabetes risk reduction noted from the interventions during DPP, all participants were offered group-implemented lifestyle sessions before the start of DPPOS, including those who had been randomly assigned to the intensive lifestyle group during DPP.19 The sessions were available quarterly for all participants (termed HELP), whereas lifestyle participants could attend up to four additional classes per year (termed BOOST). Open-label metformin

607 included in this analysis‡

647 included in this analysis‡

736 included in this analysis‡

Figure 1: Trial profile Figure is adapted from reference 17. Details of exclusions were previously published in reference 18. DPP=Diabetes Prevention Program. DPPOS=Diabetes Prevention Program Outcomes Study. *DPP enrolled participants during a 3-year period ending in June, 1999; therefore, participants had varying durations of DPP follow-up depending on their year of enrolment. †DPP participants surviving as of Sept 1, 2002, were eligible for DPPOS. ‡Participants with diabetes in DPPOS were excluded from this analysis (324 placebo, 274 metformin, 173 lifestyle intervention).

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

Intensive lifestyle intervention NGR (n=394) White African American Hispanic Asian Native American Women BMI (kg/m²) Waist (cm) Activity (met/h per week) Calorific intake (kcal per day) Fasting plasma glucose (mmol/L) 2-h plasma glucose (mmol/L) Insulin sensitivity (L/pmol) β-cell function (pmol/L) 212 (54%) 75 (19%) 59 (15%) 20 (5%) 28 (7%) 272 (69%) 32 (7) 100 (17)* 20 (18) 1748 (746) 5·61 (0·62)* 7·66 (2·17)* 9·22 (8·87)* 5·62 (4·29)* Prediabetes (n=342) 180 (53%) 64 (19%) 55 (16%) 23 (7%) 20 (6%) 222 (65%) 33 (6) 102 (15) 17 (17) 1626 (602) 5·99 (0·64) 8·99 (2·30) 7·63 (5·51) 4·1 (3·66)

Metformin NGR (n=274) 147 (54%) 60 (22%) 41 (15%) 8 (3%) 18 (7%) 173 (63%) 33 (7) 101 (15) 17 (19) 1916 (723)* 5·49 (0·58)* 7·60 (2·00)* 8·78 (7·78)* 5·76 (2·99)* Prediabetes (n=373) 223 (60%) 67 (18%) 55 (15%) 14 (4%) 14 (4%) 257 (69%) 33 (6) 102 (13) 19 (22) 1824 (759) 5·83 (0·61) 8·77 (2·08) 7·92 (4·86) 4·24 (3·04)

Placebo NGR (n=226) 133 (59%) 38 (17%) 27 (12%) 9 (4%) 19 (8%) 165 (73%) 33 (7) 100 (15) 17 (14) 1718 (725) 5·55 (0·50)* 7·60 (2·16)* 8·78 (7·33)* 5·63 (3·04)* Prediabetes (n=381) 208 (55%) 76 (20%) 64 (17%) 15 (4%) 18 (5%) 259 (68%) 33 (7) 103 (15) 17 (20) 1783 (803) 5·94 (0·62) 8·6 (2·02) 7·34 (5·03) 4·24 (2·57)

All participants Total NGR (n=894) 492 (55%) 173 (20%) 127 (14%) 37 (4%) 65 (7%) 610 (68%) 33 (7) 101 (16) 18 (16) 1766 (731) 5·55 (0·61)* 7·62 (2·11)* 8·93 (7·99)* 5·67 (3·44)* Total prediabetes (n=1096) 611 (56%) 207 (19%) 174 (16%) 52 (5%) 52 (5%) 738 (67%) 33 (6) 103 (14) 17 (20) 1716 (721) 5·92 (0·62) 8·79 (2·13) 7·63 (5·13) 4·19 (3·09)

Data are n (%) or mean (SD). The ethnic origin, sex, BMI, waist circumference, physical activity, calorific intake, glucose, and insulin measures were taken at the first DPPOS yearly visit. Asian refers to participants self-reporting as of Asian origin, including Chinese, Filipino, Korean, Vietnamese, Asian Indian, or other. Activity shown in met/h per week, where 1 met is the resting metabolic rate; higher mets show higher intensity activity. Insulin sensitivity was estimated by 1/fasting insulin. β-cell function was assessed with corrected insulin response. DPPOS=Diabetes Prevention Program Outcomes Study. NGR=normal glucose regulation. BMI=body-mass index. DPP=Diabetes Prevention Program. *p<0·05 NGR versus prediabetes, defined in and stratified by treatment group during DPP.

Table 1: Demographic and metabolic information at the beginning of DPPOS

was also continued in participants initially assigned to metformin (850 mg twice daily as tolerated) during DPPOS, unless discontinued because of development of diabetes needing management outside the protocol or for safety reasons. Outcome assessments during DPPOS continued on the same 6-month and 12-month schedule as in DPP. The primary outcome, as in DPP, was development of diabetes, defined as fasting plasma glucose 7·0 mmol/L or greater or 2-h glucose 11·0 mmol/L or greater, or both, after a 75 g oral glucose challenge (confirmed on repeat testing).20 For the present analysis, participants were classified as having normal glucose regulation if they achieved a fasting plasma glucose lower than 5·6 mmol/L and 2-h plasma glucose lower than 7·8 mmol/L at least once on yearly oral glucose tolerance test during DPP, and never met the criteria for diagnosis of diabetes. Participants were classified as having prediabetes20 if they consistently had fasting plasma glucose concentrations of 5·6–6·9 mmol/L or 2-h plasma glucose of 7·8–11·0 mmol/L, or both, on yearly oral glucose tolerance testing during DPP, and never met the criteria for the diagnosis of diabetes. Participants with a confirmed diagnosis of diabetes during DPP were excluded from this analysis, and are not part of the cohort described. Weight, blood pressure, plasma lipids, and medication history were obtained at a yearly examination using published methods.21 Measures of β-cell function (corrected insulin response = [100 × 30-min insulin] / [(30-min glucose) × (30-min glucose – 70)]; pmol/L) and insulin sensitivity (1 / [ fasting insulin]; L/pmol) were calculated with validated indices.22

0·40 0·35 Diabetes cumulative incidence rates 0·30 0·25 0·20 * 0·15 0·10 0·05 0 Reach NGR at least once in DPP Never reach NGR in DPP

0

1 Year 1 23 88

2 Year 2 24 46

3 Years in DPPOS follow-up Year 3 24 84

4 Year 4 24 53

5 Year 5 26 50

6 Year 6 23 35

NGR Prediabetes

n 894 1096

Figure 2: Diabetes cumulative incidence rates during DPPOS in participants who attained normal glucose regulation at least once during DPP compared with those who consistently had prediabetes Dashed lines show 95% CIs. DPP=Diabetes Prevention Program. DPPOS=Diabetes Prevention Program Outcomes Study. NGR=normal glucose regulation. *p<0·0001 between groups.

Statistical analysis
The cumulative incidence of diabetes during DPPOS, as well as probabilities of a participant reaching normal glucose regulation during DPPOS, were calculated and compared for those classified as having normal glucose regulation versus prediabetes during DPP. We evaluated predictors of diabetes risk using sequential Cox proportional hazards models since the groups were
3

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

0·45 0·40 Diabetes cumulative incidence rates 0·35 0·30

Lifestyle, prediabetes Metformin, prediabetes Placebo, prediabetes Lifestyle, NGR Metformin, NGR Placebo, NGR

Never reached NGR in DPP 0·25 0·20 0·15 0·10 Reached NGR at least once in DPP 0·05 0

period (DPP) and the observation period (DPPOS) in people who returned to normal glucose regulation versus those who remained with prediabetes, estimates of these metabolic measures are presented separately rather than together (as the disposition index). Of note, we did not adjust results for multiple comparisons because analyses were post-hoc and exploratory with the intention of generating hypotheses and discussion on this topic. These studies are registered at ClinicalTrials.gov, NCT00004992 (DPP) and NCT00038727 (DPPOS).

Role of the funding source
The sponsor of this study was represented on the steering committee and played a part in the study design, its execution, and the plans for publication. The funding agency was not represented in the writing groups, but could give input on proposed publications resulting from the study. All authors in the writing group had access to all data. The corresponding author had the final decision to submit for publication.

0 n

1 Year 1 14 5 4 36 22 30

2 Year 2 11 6 7 24 8 14

3 Years in DPPOS follow-up Year 3 8 13 3 21 32 31

4 Year 4 9 5 10 24 15 14

5 Year 5 12 6 8 8 23 19

6 Year 6 13 6 4 17 7 11

NGR Lifestyle Metformin Placebo Prediabetes Lifestyle Metformin Placebo

394 274 226 342 373 381

Results
The average number of lifestyle sessions attended during the bridge period between DPP and DPPOS was 2·87 (SD 4·73, 95% CI 0–14) in the intensive lifestyle intervention group, 5·35 (SD 5·95, 95% CI 0–16) in the metformin group, and 5·34 (SD 5·99, 95% CI 0–15) in the placebo group. Participants who had attained normal glucose regulation during DPP (n=894) attended more sessions (mean 4·61, SD 5·71) than did those who remained with prediabetes (n=1096) during DPP (mean 4·28, SD 5·63; p=0·0086 vs normal glucose regulation). The number of lifestyle sessions attended during the bridge period was lower in participants randomly assigned to the lifestyle intervention versus metformin or placebo groups during DPP (p<0·0001) and was slightly higher for those achieving normal glucose regulation (n=394, mean 3·24, SD 5·05) versus those who remained with prediabetes (n=342, mean 2·44, SD 4·30) within the lifestyle intervention (p=0·0229) and placebo (normal glucose regulation: n=226, mean 5·96, SD 5·85; prediabetes: n=381, mean 4·97, SD 6·05; p=0·0490) but not metformin groups (normal glucose regulation: n=274, mean 5·48, SD 6·07; prediabetes: n=373, mean 5·26, SD 5·87; p=0·6450). Baseline characteristics for the DPP and DPPOS cohorts have been previously published.17,23 Table 1 shows baseline demographic and metabolic information for DPPOS, stratified by prespecified glycaemic status (eg, prediabetes vs normal glucose regulation; all patients with diabetes were excluded). Briefly, ethnic origin, sex, BMI, and activity levels were similar at the outset of DPPOS for all treatment groups, whether they had attained normal glucose regulation during DPP or not. By definition, plasma fasting and 2-h glucose concentrations were higher for patients with prediabetes versus those with normal glucose regulation. Estimates of β-cell function

Figure 3: Diabetes cumulative incidence rates during DPPOS in participants who attained normal glucose regulation at least once during DPP compared with those who consistently had prediabetes, stratified by treatment group in DPP DPP=Diabetes Prevention Program. DPPOS=Diabetes Prevention Program Outcomes Study. NGR=normal glucose regulation.

defined during the randomised DPP period. Generalised mixed models were used to model probabilities of reaching normal glucose regulation during DPPOS because of the need for adjustment in group differences (percentage of weight change during DPP, insulin sensitivity, and β-cell function at the end of DPP; bodymass index [BMI] at the beginning of DPPOS) at the beginning of the non-randomised DPPOS follow-up. In the Cox proportional hazards model, hazard ratios (HRs) for continuous variables are presented as the ratio in hazards for one SD increase in the predictor; for categorical variables, HRs were calculated as the ratio between each category and the reference group. An HR greater than 1 shows increased risk of diabetes. However, in the generalised linear mixed models, an odds ratio (OR) greater than 1 shows increased chance of reaching normal glucose regulation. Interactions between normal glucose regulation status during DPP and treatment group were tested to establish whether the effect of previous normal glucose regulation status varied for participants receiving different treatments in DPP. We calculated p values for individual covariates using the Wald test and the likelihood ratio test for the overall model. Significance was set at an α level of 0·05. SAS (version 9.2) was used for all analyses. To show the natural history of insulin sensitivity and secretion during both the intervention
4

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

and insulin sensitivity were higher in normal glucose regulation versus prediabetes (difference 0·0011, p=0·0017 in lifestyle intervention; difference 0·00100, p=0·0399 in metformin; difference 0·0100, p=0·0007 in placebo). Waist circumference was also lower in normal glucose regulation versus prediabetes in the lifestyle intervention group (p<0·05). Weight loss during DPP was greater in participants who attained normal glucose regulation status (vs those who consistently had prediabetes) for each active treatment group during DPP (5·54% vs 2·37% weight loss in the lifestyle intervention group, p<0·0001; 3·71% vs 2·01% weight loss in metformin, p=0·0041; 1·29% vs 0·43% weight loss in placebo, p=0·1436). However, participants who had normal glucose regulation in DPP subsequently gained more weight during DPPOS (normal glucose regulation vs prediabetes: 4·92 vs 2·01%, p<0·0001 in the lifestyle intervention group; 4·79 vs 2·64%, p=0·0057 in metformin; 1·95 vs 0·94%, p=0·1731 in placebo), irrespective of previous treatment. Participants who achieved normal glucose regulation status at least once during DPP had a 56% reduced risk of progression to diabetes during DPPOS (HR 0·44, 95% CI 0·37–0·55, p<0·0001; figure 2). Diabetes risk reduction was strongly associated with the number of times normal glucose regulation was achieved. Specifically, diabetes risk was reduced 47% in DPPOS if normal glucose regulation was attained only once (HR 0·53, 95% CI 0·42–0·66, p<0·0001), 61% if it was reached twice (HR 0·39, 95% CI 0·28–0·56, p<0·0001), and 67% if it was reached three times (HR 0·33, 95% CI 0·19–0·58, p=0·0001) during DPP. Within treatment groups, normal glucose regulation status was attained once in 23% (170/736), 25% (161/647), and 23% (137/607), twice in 18% (130/736), 11% (71/647), and 9% (52/607), and three times in 9% (67/736), 4% (27/647), and 5% (28/607) of participants assigned to the lifestyle intervention, metformin, and placebo groups, respectively. Cox modelling also revealed that age younger than 45 years (HR 1·47, 95% CI 1·12–1·94, p=0·0060) and African American ethnic origin (HR 1·77, 95% CI 1·42–2·20, p<0·0001) were associated with increased diabetes risk. Hispanic (HR 1·17, 95% CI 0·90–1·51, p=0·2443), Asian (HR 1·22, 95% CI 0·78–1·90, p=0·3908), and Native American (HR 1·28, 95% CI 0·89–1·84, p=0·1849) participants did not have a significantly higher risk of diabetes than did those who are white. Paradoxically, increased weight loss during DPP adversely affected diabetes risk (HR 1·26, 95% CI 1·15–1·39, p<0·0001) in DPPOS independent of previous treatment (interaction test for normal glucose regulation and lifestyle intervention, p=0·8269; normal glucose regulation and metformin, p=0·3754), probably because of the high rate of weight regain in DPPOS with associated adverse effects on diabetes risk. High BMI at the beginning of DPPOS also related to diabetes risk during DPPOS follow-up (HR 1·14, 95% CI 1·05–1·25, p=0·0021), whereas higher β-cell function (HR 0·80,

A
6·0 NGR Prediabetes

5·5

5·0 Corrected insulin response (pmol/L)

4·5

4·0

3·5

3·0

2·5 0

e

1

2

3

4

4

3

1

5 SY PO DP
686 686

lin

PY

PY

PY

PY

SY

SY

SY

SY

se

DP

DP

DP

PO

PO

DP

PO

PO

Ba

DP

DP

DP

DP

NGR Prediabetes

871 1074

844 1032

842 1037

815 999

579 584

807 976

794 864

733 797

702 720

DP
585 563
5 SY
732 730

B
9·5

9·0

8·5 1/fasting insulin (L/pmol)

8·0

7·5

7·0

6·5

6·0 0

e

1

2

3

PY 4

PO SY 3

SY 4

1

PO
DP
628 566

PO SY 2

se lin

PY

PY

PY

SY

DP

DP

DP

PO

DP

PO

PO

Ba

DP

DP

NGR Prediabetes

902 1118

873 1070

861 1073

842 1042

597 622

837 1028

DP

818 909

770 844

DP

743 756

Figure 4: Insulin secretion (corrected insulin response; A) and insulin sensitivity (1/fasting insulin; B) in participants with prediabetes and normal glucose regulation at baseline and by year of DPP and DPPOS Data are mean; error bars show SD. Dotted line denotes the bridge period. Corrected insulin response=100×I30/ G30(G30–70). DPP=Diabetes Prevention Program. DPPOS=DPP Outcomes Study. NGR=normal glucose regulation.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

DP

PO

SY

6

SY

6

2

5

Articles

Odds ratio (95% CI) NGR vs prediabetes (placebo) NGR, lifestyle vs placebo NGR, metformin vs placebo Prediabetes, lifestyle vs placebo Prediabetes, metformin vs placebo Age <45 years Age 45–59 years Age ≥60 years Female African American Hispanic Asian Native American White BMI % weight change during DPP Insulin resistance β-cell function 3·01 (2·25–4·02) 0·80 (0·63–1·02) 0·89 (0·69–1·14) 0·59 (0·42–0·82) 0·98 (0·73–1·32) 0·95 (0·75–1·18) 1·00 (0·83–1·21) Reference 1·18 (1·00–1·38) 1·00 (0·81–1·24) 1·04 (0·84–1·30) 0·84 (0·57–1·24) 0·92 (0·67–1·26) Reference 0·79 (0·71–0·87) 0·78 (0·72–0·84) 1·16 (1·08–1·25) 1·28 (1·18–1·39)

p value <0·0001 0·0678 0·3481 0·0014 0·9028 0·6243 0·9968 ·· 0·0509 0·9697 0·7011 0·3794 0·5918 ·· <0·0001 <0·0001 <0·0001 <0·0001

remained with prediabetes despite intensive lifestyle intervention during DPP were less likely to subsequently achieve normal glucose regulation (vs placebo), consistent with their higher diabetes risk. Participants who remained with prediabetes despite metformin treatment during DPP had a similar likelihood of subsequently achieving normal glucose regulation as did those who remained with prediabetes with previous placebo treatment (figure 3). Generalised mixed models revealed no effect of any self-reported ethnic origin on regression to normal glucose regulation, but did show a positive effect of female sex on regression (table 2). By contrast, increased weight loss during DPP and high BMI at the beginning of DPPOS were associated with decreased regression to normal glucose regulation, reflecting the increased diabetes risk related to these factors.

Discussion
Although there is widespread consensus that diabetes prevention is crucially important,24 there is less agreement with respect to the particular intervention. Several studies have shown the efficacy of lifestyle modification for diabetes prevention.8,10,12,15,25 Long-term compliance with these lifestyle changes has proven difficult, however, and the benefits wane with weight regain.17 Various pharmacological agents also prevent diabetes, but the cost-effectiveness and risk–benefit ratio for most are unclear.6,7,9–11,13 Results from the present analysis would contend that the strategy is unimportant as long as the intervention is early (when someone has prediabetes) and can restore normal glucose regulation, even if transiently (panel). Further, maintenance of prediabetes despite the potent glucose-lowering effects of intensive lifestyle modification represents a high-risk state and might warrant additional preventive strategies. Although two-thirds of people with diabetes are overweight or obese,26 few of those who are obese will ever develop diabetes,27 whereas up to 70% of those with prediabetes might acquire the disease during their lifetime.10,12,28 Diagnosis of diabetes or prediabetes is made by defined glycaemic criteria, but can be less straightforward when blood glucose concentrations fluctuate above and below diagnostic thresholds. Spontaneous progression, regression, or interconversion of dysglycaemic states29 can contribute to the clinical conundrum. The patients in our study with normal glucose regulation could simply represent a group with either increased motivation for treatment (as evidenced by their long-term participation in this clinical trial) or a more modifiable glucose status, rather than true reversal of the diabetogenic process. However, we should point out that a single determination of impaired glucose tolerance has been shown to predict increased diabetes risk three-times over 5·8 years, even after the return to normal glucose regulation in a high-risk population.30 Our study is the first to show that the converse is true for the attainment of normal glucose regulation in people

All results based on a generalised linear mixed model for reaching NGR status or not at yearly DPPOS visits. Predictors for regression to NGR during DPPOS included: glycaemic status and treatment group during DPP; and demographic and concurrent weight and metabolic information during DPPOS. For categorical predictors, the ORs correspond to the comparison over the reference group; for continuous predictors, the ORs refer to per unit SD in the predictor. For BMI at the beginning of DPPOS, % weight change during DPP, 1/fasting insulin, and corrected insulin response, the SD between participants in DPPOS were 6·83 kg/m², 7·65%, 35 pmol/L, and 0·41 L/pmol, respectively. Insulin sensitivity was estimated by 1/fasting insulin. β-cell function was assessed with corrected insulin response. DPPOS=Diabetes Prevention Program Outcomes Study. OR=odds ratio. NGR=normal glucose regulation. BMI=body-mass index. DPP=Diabetes Prevention Program.

Table 2: Predictors of achieving normal glucose regulation during DPPOS

95% CI 0·71–0·89, p<0·0001) and insulin sensitivity (HR 0·83, 95% CI 0·74–0·94, p=0·0001) were protective. Previous randomisation group in DPP did not affect risk reduction in DPPOS in participants who attained normal glucose regulation (lifestyle intervention vs placebo, p=0·1722; metformin vs placebo, p=0·3304). Those who consistently had prediabetes during DPP, despite intensive lifestyle intervention, had an increased risk of diabetes during DPPOS (HR 1·31 vs placebo, 95% CI 1·03–1·68, p=0·0304; figure 3). Maintenance of prediabetes in participants previously randomly assigned to metformin did not carry the same risk (HR 0·93 vs placebo, 95% CI 0·72–1·20, p=0·5940). Normal glucose regulation status during DPP strongly predicted achievement of normal glucose regulation status during DPPOS (OR 3·18, 95% CI 2·71–3·72, p<0·0001) and was unaffected by previous treatment group (test for interaction between normal glucose regulation and lifestyle intervention, p=0·1363; normal glucose regulation and metformin, p=0·6066). This finding was partly accounted for by the maintenance of increased β-cell function and insulin sensitivity throughout DPP and DPPOS (figure 4). Participants who
6

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

with prediabetes. Specifically, determination of normal glucose regulation status at one or more yearly visits during DPP was associated with a 56% reduced risk of diabetes during subsequent follow-up in DPPOS. The magnitude of this risk reduction approximates that seen with intensive lifestyle intervention—the most potent intervention in DPP10—but with a greater enduring effect on long-term diabetes prevention.17 Further, we have shown that the risk reduction benefit was seen irrespective of previous treatment group (ie, no treatment interaction), suggesting that achievement of normal glucose regulation is more important than the method used to achieve it, with respect to lowering the risk of diabetes. Last, the demographic information of our multiethnic cohort closely resembles that of the US population with prediabetes, supporting the generalisability of our findings.1 Taken together, these facts serve as essential clinical information for people with prediabetes and their providers in the prevention of diabetes. We did not attempt to adjust models for attendance in the bridge period between DPP and DPPOS because groups were defined in DPP (before the bridge) and attendance during the bridge was generally low. We cannot exclude that attendance during the bridge affected normal glucose regulation status or diabetes development during DPP (as evidenced by the increased attendance in participants who had normal glucose regulation during DPP), yet attendance was higher for those who maintained normal glucose regulation with placebo versus the intensive lifestyle intervention during DPP, making its contribution suspect. DPP showed a clear reduction in diabetes incidence in participants randomly assigned to the lifestyle intervention or metformin during the intervention period.10 One might speculate that some of this benefit came from the potency of the intervention to restore normal glucose regulation. For example, the lifestyle intervention was about twice as effective as metformin for prevention of diabetes10 and was the only intervention associated with regression to normal glucose regulation.16 Implied are the many pleiotropic effects of lifestyle intervention (eg, reduction in lipids, blood pressure, and biomarkers) that go beyond the straightforward glucoselowering effect of metformin.31,32 Interestingly, the present analysis revealed that people with persistent prediabetes who had formerly been in the lifestyle intervention group had a higher, not lower, risk of diabetes in DPPOS. This seeming paradox can be reconciled by consideration of two potential explanations. First, the metformin group could have benefited from continued treatment in DPPOS, whereas lifestyle intervention adherence diminished—as evidenced by their weight regain and reduced attendance at the lifestyle sessions (ie, HELP and BOOST) during the bridge period. However, since weight regain was fairly small (1%) and metformin adherence also diminished during

Panel: Research in context Systematic review We searched from PubMed June 1, 1976, to Feb 29, 2012, for full reports of randomised clinical trials in people with prediabetes using the term “regression to normal glucose regulation” and “regression to normoglycemia”. Of the 154 references returned, four were clinical trials that examined the efficacy of an intervention to return people with prediabetes to normal glucose regulation. Two of the four trials were previous post-hoc analyses from the Diabetes Prevention Program (DPP). Of the two non-DPP clinical trials, one showed failure (of angiotensin-receptor blockade) and the other success (of lifestyle or glucose-lowering drugs, or both) in restoration of normoglycaemia in people with prediabetes, but neither these nor the DPP analyses quantified the risk reduction for future diabetes in people with normal glucose regulation. Interpretation This analysis draws attention to the significant long-term reduction in diabetes risk when someone with prediabetes returns to normoglycaemia, supporting a shift in the standard of care to early and aggressive glucose-lowering treatment in patients at highest risk.

the bridge period, these potential group differences in treatment efficacy were largely mitigated. A more likely explanation is that these findings reflect a survivor effect in the lifestyle intervention group, with participants who still had prediabetes being particularly susceptible to develop diabetes. Whether the susceptibility represents genetic predisposition, non-adherence to the lifestyle intervention, or other environmental factors is not known. Nonetheless, a combination of preventive strategies might be appropriate for patients whose dysglycaemia cannot be reversed with the potent glucose-lowering effect of lifestyle intervention. Notably, combination therapy (intensive lifestyle intervention plus low-dose metformin) was not more effective than were the individual treatments for prevention of diabetes in a group of high-risk Asian Indians.15 The possibility of increased potency of combination therapy in a multicultural cohort, such as DPPOS, cannot be excluded or confirmed without further study, but deserves consideration. Some analyses might have been underpowered for the endpoint (ie, ethnic origin and regression or progression). Last, erroneous misclassification of people as having normal glucose regulation versus prediabetes cannot be excluded. Thus, additional studies are needed to confirm our findings. Whether preventive strategies should target insulin resistance or decreasing β-cell function is a topic of much debate. Insulin sensitivity and secretion are integrally related and the deterioration of each is felt to be requisite in the development of type 2 diabetes.33 Our results suggest that both are related to restoration of normal glucose regulation. On closer examination, however,
7

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

See Online for appendix

insulin sensitivity in normal glucose regulation and prediabetes groups became more similar in DPPOS than did β-cell function. Although the continuous insulin response rose in both groups to compensate for the reduction in insulin sensitivity at the culmination of interventions invoked during DPP, the insulin secretory response remained substantially higher throughout DPPOS in participants who had attained normal glucose regulation during DPP. These data agree with others showing that β-cell failure ultimately dictates onset of diabetes.6,33,34 We should point out, however, that interventions that lower insulin resistance (ie, exercise) often improve β-cell function. For example, greater progression to diabetes in the younger participants of DPPOS than in older patients might reflect their lower success with the lifestyle intervention during DPP—an intervention shown to improve β-cell function.35 By contrast, greater regression to normal glucose regulation in women probably reflects their ability to maintain β-cell function. A previous analysis from DPP noted a higher index of insulin secretion, but not insulin sensitivity, in women versus men entering the trial.36 In sum, our data suggest that preservation of β-cell function is more closely related than is insulin sensitivity to the long-term prevention of diabetes and restoration of normal glucose regulation. Until therapeutic approaches can be confirmed as specifically preserving β-cell function, efforts should focus on intensive lifestyle intervention where its positive effects on the insulin secretory response and costeffectiveness have been shown in clinical trials.37,38 In conclusion, results from this analysis of DPPOS show a 56% reduction in diabetes incidence in highrisk individuals who reverted to normal glucose regulation, no matter how this reversion is achieved or however transiently. Surprisingly, diabetes risk was highest in participants who remained with prediabetes despite lifestyle intervention. This finding emphasises a particular susceptibility in this group, who might benefit from additional interventions. Together, these data serve as essential clinical information to support early and aggressive measures for long-term prevention of diabetes in people at risk.
Contributors LP contributed to the study design, analysis plan, interpretation, and writing and editing of the report. QP contributed to the analysis plan and interpretation and was responsible for analysis. KJM contributed to interpretation of results and review of the report. KEW contributed to the analysis plan, interpretation of results, and writing and editing of the report. RFH contributed to the study design, analysis plan, interpretation of the results, and review of the report. SEK contributed to the analysis plan, interpretation of the results, and writing and editing of the report. Conflicts of interest We declare that we have no conflicts of interest. Acknowledgments The research group thanks the participants of DPP and DPPOS for their commitment and dedication. During DPPOS, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the US National Institutes of Health (grant 5U01-DK048375-12) provided funding to the

clinical centres and the coordinating centre for the design and conduct of the study and collection, management, analysis, and interpretation of the data. The Southwestern American Indian Centers were supported directly by the NIDDK, including its Intramural Research Program, and the Indian Health Service. The General Clinical Research Center Program, National Center for Research Resources, and the Department of Veterans Affairs supported data collection at many of the clinical centres. Funding was also provided by the National Institute of Child Health and Human Development, the National Institute on Aging, the National Eye Institute, the National Heart Lung and Blood Institute, the Office of Research on Women’s Health, the National Center for Minority Health and Human Disease, the Centers for Disease Control and Prevention, and the American Diabetes Association. Bristol-Myers Squibb and Parke-Davis provided additional funding and material support during DPP, Lipha (Merck-Sante) provided drugs, and LifeScan Inc donated materials during DPP and DPPOS. The opinions expressed are those of the investigators and do not necessarily reflect the views of the funding agencies. A complete list of centres, investigators, and staff is shown in the appendix pp 1–3. This Article is dedicated to Christopher D Saudek, our friend and collaborator, who passed away on Oct 6, 2010. References 1 US Centers for Disease Control and Prevention. Diabetes fact sheet. Jan 26, 2011. 2 Diabetes Prevention Program Research Group. The prevalence of retinopathy in impaired glucose tolerance and recent-onset diabetes in the Diabetes Prevention Program. Diabet Med 2007; 24: 137–44. 3 Cheng YJ, Gregg EW, Geiss LS, et al. Association of A1C and fasting plasma glucose levels with diabetic retinopathy prevalence in the U.S. population: implications for diabetes diagnostic thresholds. Diabetes Care 2009; 32: 2027–32. 4 Ford ES, Zhao G, Li C. Pre-diabetes and the risk for cardiovascular disease: a systematic review of the evidence. J Am Coll Cardiol 2010; 55: 1310–17. 5 Ziegler D, Rathmann W, Dickhaus T, Meisinger C, Mielck A. Prevalence of polyneuropathy in pre-diabetes and diabetes is associated with abdominal obesity and macroangiopathy: the MONICA/KORA Augsburg Surveys S2 and S3. Diabetes Care 2008; 31: 464–69. 6 Buchanan TA, Xiang AH, Peters RK, et al. Preservation of pancreatic beta-cell function and prevention of type 2 diabetes by pharmacological treatment of insulin resistance in high-risk Hispanic women. Diabetes 2002; 51: 2796–803. 7 Chiasson JL, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M, for The STOP-NIDDM Trial Research Group. Acarbose for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial. Lancet 2002; 359: 2072–77. 8 Eriksson KF, Lindgarde F. Prevention of type 2 (non-insulin-dependent) diabetes mellitus by diet and physical exercise. The 6-year Malmo feasibility study. Diabetologia 1991; 34: 891–98. 9 The DREAM (Diabetes REduction Assessment with ramipril and rosiglitazone Medication) Trial Investigators. Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial. Lancet 2006; 368: 1096–105. 10 Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393–403. 11 Torgerson JS, Hauptman J, Boldrin MN, Sjostrom L. XENical in the prevention of diabetes in obese subjects (XENDOS) study: a randomized study of orlistat as an adjunct to lifestyle changes for the prevention of type 2 diabetes in obese patients. Diabetes Care 2004; 27: 155–61. 12 Tuomilehto J, Lindstrom J, Eriksson JG, et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001; 344: 1343–50. 13 DeFronzo RA, Tripathy D, Schwenke DC, et al. Pioglitazone for diabetes prevention in impaired glucose tolerance. N Engl J Med 2011; 364: 1104–15. 14 Pan XR, Li GW, Hu YH, et al. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care 1997; 20: 537–44.

8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

Articles

15

16

17

18

19

20

21

22

23

24

25

26

27

Ramachandran A, Snehalatha C, Mary S, Mukesh B, Bhaskar AD, Vijay V. The Indian Diabetes Prevention Programme shows that lifestyle modification and metformin prevent type 2 diabetes in Asian Indian subjects with impaired glucose tolerance (IDPP-1). Diabetologia 2006; 49: 289–97. Perreault L, Kahn SE, Christophi CA, Knowler WC, Hamman RF. Regression from pre-diabetes to normal glucose regulation in the Diabetes Prevention Program. Diabetes Care 2009; 32: 1583–88. Diabetes Prevention Program Research Group. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 2009; 374: 1677–86. The Diabetes Prevention Program Research Group. Strategies to identify adults at high risk for type 2 diabetes. Diabetes Care 2005; 28: 138–44. Venditti EM, Bray GA, Carrion-Petersen ML, et al. First versus repeat treatment with a lifestyle intervention program: attendance and weight loss outcomes. Int J Obes (Lond) 2008; 32: 1537–44. Tataranni PA, Ortega E. A burning question: does an adipokine-induced activation of the immune system mediate the effect of overnutrition on type 2 diabetes? Diabetes 2005; 54: 917–27. The Diabetes Prevention Program. Design and methods for a clinical trial in the prevention of type 2 diabetes. Diabetes Care 1999; 22: 623–34. Phillips DI, Clark PM, Hales CN, Osmond C. Understanding oral glucose tolerance: comparison of glucose or insulin measurements during the oral glucose tolerance test with specific measurements of insulin resistance and insulin secretion. Diabet Med 1994; 11: 286–92. The Diabetes Prevention Program: baseline characteristics of the randomized cohort. The Diabetes Prevention Program Research Group. Diabetes Care 2000; 23: 1619–29. Nathan DM, Davidson MB, DeFronzo RA, et al. Impaired fasting glucose and impaired glucose tolerance: implications for care. Diabetes Care 2007; 30: 753–59. Saito T, Watanabe M, Nishida J, et al. Lifestyle modification and prevention of type 2 diabetes in overweight Japanese with impaired fasting glucose levels: a randomized controlled trial. Arch Intern Med 2011; 171: 1352–60. Overweight, obesity, and health risk. National Task Force on the Prevention and Treatment of Obesity. Arch Intern Med 2000; 160: 898–904. Harris MI, Flegal KM, Cowie CC, et al. Prevalence of diabetes, impaired fasting glucose, and impaired glucose tolerance in U.S. adults. The Third National Health and Nutrition Examination Survey, 1988–1994. Diabetes Care 1998; 21: 518–24.

28

29

30

31

32

33

34

35

36

37

38

de Vegt F, Dekker JM, Jager A, et al. Relation of impaired fasting and postload glucose with incident type 2 diabetes in a Dutch population: The Hoorn Study. JAMA 2001; 285: 2109–13. Meigs JB, Muller DC, Nathan DM, Blake DR, Andres R. The natural history of progression from normal glucose tolerance to type 2 diabetes in the Baltimore Longitudinal Study of Aging. Diabetes 2003; 52: 1475–84. Saad MF, Knowler WC, Pettitt DJ, Nelson RG, Bennett PH. Transient impaired glucose tolerance in Pima Indians: is it important? BMJ 1988; 297: 1438–41. Mather KJ, Funahashi T, Matsuzawa Y, et al. Adiponectin, change in adiponectin, and progression to diabetes in the Diabetes Prevention Program. Diabetes 2008; 57: 980–86. Ratner R, Goldberg R, Haffner S, et al. Impact of intensive lifestyle and metformin therapy on cardiovascular disease risk factors in the diabetes prevention program. Diabetes Care 2005; 28: 888–94. Weyer C, Bogardus C, Mott DM, Pratley RE. The natural history of insulin secretory dysfunction and insulin resistance in the pathogenesis of type 2 diabetes mellitus. J Clin Invest 1999; 104: 787–94. Utzschneider KM, Prigeon RL, Faulenbach MV, et al. Oral disposition index predicts the development of future diabetes above and beyond fasting and 2-h glucose levels. Diabetes Care 2009; 32: 335–41. Crandall J, Schade D, Ma Y, et al. The influence of age on the effects of lifestyle modification and metformin in prevention of diabetes. J Gerontol A Biol Sci Med Sci 2006; 61: 1075–81. Perreault L, Ma Y, Dagogo-Jack S, et al. Sex differences in diabetes risk and the effect of intensive lifestyle modification in the Diabetes Prevention Program. Diabetes Care 2008; 31: 1416–21. Herman WH, Hoerger TJ, Brandle M, et al. The cost-effectiveness of lifestyle modification or metformin in preventing type 2 diabetes in adults with impaired glucose tolerance. Ann Intern Med 2005; 142: 323–32. Kitabchi AE, Temprosa M, Knowler WC, et al. Role of insulin secretion and sensitivity in the evolution of type 2 diabetes in the diabetes prevention program: effects of lifestyle intervention and metformin. Diabetes 2005; 54: 2404–14.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60525-X

9

Articles

Effectiveness of quality improvement strategies on the management of diabetes: a systematic review and meta-analysis
Andrea C Tricco, Noah M Ivers, Jeremy M Grimshaw, David Moher, Lucy Turner, James Galipeau, Ilana Halperin, Brigitte Vachon, Tim Ramsay, Braden Manns, Marcello Tonelli, Kaveh Shojania

Summary
Background The effectiveness of quality improvement (QI) strategies on diabetes care remains unclear. We aimed to assess the effects of QI strategies on glycated haemoglobin (HbA1c), vascular risk management, microvascular complication monitoring, and smoking cessation in patients with diabetes. Methods We identified studies through Medline, the Cochrane Effective Practice and Organisation of Care database (from inception to July 2010), and references of included randomised clinical trials. We included trials assessing 11 predefined QI strategies or financial incentives targeting health systems, health-care professionals, or patients to improve management of adult outpatients with diabetes. Two reviewers independently abstracted data and appraised risk of bias. Findings We reviewed 48 cluster randomised controlled trials, including 2538 clusters and 84 865 patients, and 94 patient randomised controlled trials, including 38 664 patients. In random effects meta-analysis, the QI strategies reduced HbA1c by a mean difference of 0·37% (95% CI 0·28–0·45; 120 trials), LDL cholesterol by 0·10 mmol/L (0·05–0.14; 47 trials), systolic blood pressure by 3·13 mm Hg (2·19–4·06, 65 trials), and diastolic blood pressure by 1·55 mm Hg (0·95–2·15, 61 trials) versus usual care. We noted larger effects when baseline concentrations were greater than 8·0% for HbA1c, 2·59 mmol/L for LDL cholesterol, and 80 mm Hg for diastolic and 140 mm Hg for systolic blood pressure. The effectiveness of QI strategies varied depending on baseline HbA1c control. QI strategies increased the likelihood that patients received aspirin (11 trials; relative risk [RR] 1·33, 95% CI 1·21–1·45), antihypertensive drugs (ten trials; RR 1·17, 1·01–1·37), and screening for retinopathy (23 trials; RR 1·22, 1·13–1·32), renal function (14 trials; RR 128, 1·13–1·44), and foot abnormalities (22 trials; RR 1·27, 1·16–1·39). However, statin use (ten trials; RR 1·12, 0·99–1·28), hypertension control (18 trials; RR 1·01, 0·96–1·07), and smoking cessation (13 trials; RR 1·13, 0·99–1·29) were not significantly increased. Interpretation Many trials of QI strategies showed improvements in diabetes care. Interventions targeting the system of chronic disease management along with patient-mediated QI strategies should be an important component of interventions aimed at improving diabetes management. Interventions solely targeting health-care professionals seem to be beneficial only if baseline HbA1c control is poor. Funding Ontario Ministry of Health and Long-term Care and the Alberta Heritage Foundation for Medical Research (now Alberta Innovates—Health Solutions).
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60480-2 See Online/Comment DOI:10.1016/S01406736(12)60637-0 Li Ka Shing Knowledge Institute of St Michael’s Hospital, Toronto, ON, Canada (A C Tricco PhD); Women’s College Hospital and Department of Family and Community Medicine (N M Ivers MD) and Division of Endocrinology, Department of Medicine (I Halperin MD), University of Toronto, Toronto, ON, Canada; Clinical Epidemiology Program, Ottawa Hospital Research Institute, Ottawa, ON, Canada (Prof J M Grimshaw PhD, D Moher PhD, L Turner MSc, J Galipeau PhD, B Vachon PhD, T Ramsay PhD); Department of Medicine (Prof J M Grimshaw), and Department of Epidemiology and Community Medicine (D Moher, T Ramsay), Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada; Occupational Therapy Program, School of Rehabilitation, Université de Montréal, Montreal, QC, Canada (B Vachon); Department of Community Health Sciences (B Manns MD) and Libin Cardiovascular Institute and Department of Medicine (B Manns, M Tonelli MD), University of Calgary, Calgary, AB, Canada; and Sunnybrook Health Sciences Centre and University of Toronto Centre for Patient Safety, Toronto, ON, Canada (K Shojania MD) Correspondence to: Dr Andrea C Tricco, Li Ka Shing Knowledge Institute of St Michael’s Hospital, 30 Bond Street, Toronto, ON M5B 1W8, Canada [email protected]

Introduction
Despite high-quality evidence showing improved clinical outcomes for patients with diabetes who receive various preventive and therapeutic interventions,1 many patients with diabetes do not receive them.2–5 The gap between ideal and actual care is not surprising in view of the complex nature of diabetes management, often needing coordinated services of primary-care physicians, allied health practitioners, and subspecialists. Moreover, it is a challenge to change patient behaviour and encourage healthy lifestyles.6 In view of the increasing prevalence of diabetes and the burgeoning cost of managing patients with this disease,7 improving the efficiency of diabetes care is an important goal. Although clinicians, managers, and policy makers expend significant time and resources attempting to

optimise care for patients with diabetes, the optimum approach to improving diabetes care (and outcomes) remains uncertain. A previous systematic review8 assessed the effect of quality improvement (QI) interventions to improve glycaemic control for patients with type 2 diabetes in 66 controlled studies published by April, 2006. Over a median follow-up of 13 months, the QI interventions significantly lowered glycated haemoglobin (HbA1c) by a mean 0·42% (95% CI 0·29–0·54). After adjustment for study size and baseline HbA1c, two of the 11 categories of QI strategies were associated with reductions in HbA1c of at least 0·50%: team changes (26 trials; 0·67%, 95% CI 0·43–0·91) and case management (26 trials; 0·52%, 0·31–0·73). Only these two strategies led to significant incremental reductions in HbA1c (ie, interventions that

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

1

Articles

included either of these two strategies achieved significantly greater improvements than strategies without them). Since the previous review8 noted a rapid growth of published work on this subject and did not assess the effect of QI strategies on outcomes other than HbA1c, we sought to update and expand the review by considering the effect of QI interventions on glycaemic control, vascular risk-factor management, monitoring of microvascular complications, and smoking cessation in patients with diabetes.

Methods
Study selection and search strategy
Our systematic review was based on a protocol with input from experts in diabetes care, methods, and statistics.9 We selected randomised clinical trials that assessed 11 predefined QI strategies or financial incentives targeting health-care professionals8 for the management of adult outpatients with diabetes (panel). The QI strategies targeted health systems (eg, team changes), professionals (eg, professional reminders), or patients (eg, promotion of self management). By use of a framework of outcomes (appendix), we required that studies reported at least one process of care measure (proportion of patients taking aspirin, statins, antihypertensive drugs, screened for retinopathy, screened for foot abnormalities, monitored for renal function) or intermediate outcome (HbA1c and LDL-cholesterol concentrations, diastolic and systolic blood pressure, proportion of patients with controlled hypertension, or who quit smoking). Consistent with the previous review, we excluded trials assessing the effect of QI strategies aimed solely at the patient (ie, with no associated health systems or professional change). We identified studies through Medline (July, 2003 [last date of the original search10], to July, 2010), the Cochrane Effective Practice and Organisation of Care (EPOC) database (July, 2003, to July, 2010), and the references of included trials. An experienced librarian developed the search strategy, which was peer reviewed independently by another information specialist.11 We restricted our final search strategy for Medline to reports in English (appendix); we adjusted it as necessary for searching the Cochrane EPOC database. To ensure reliability, we undertook a training exercise before the screening process with a random 5% sample of search results. Two reviewers subsequently screened the records from the updated search. Two reviewers obtained the full text of potentially relevant articles and screened them independently for inclusion. Discrepancies were resolved by discussion or involvement of a third reviewer. Two reviewers independently rescreened all full-text articles from the previous review for inclusion, since our inclusion criteria were slightly different from the original report.8 Since we relied on searches done in the previous review, we were unable to establish the

reason for exclusion for about 4% (220 of 5592) of the citations. We developed and modified a data abstraction form after a training exercise for reviewers. Data items were study details (eg, randomisation of clusters or patients , setting, duration of intervention, type of QI intervention), characteristics of participants (eg, mean age, proportion who were male), outcomes assessed, and study results (eg, mean HbA1c and SDs at baseline and the end of the intervention for the control and intervention groups). Two reviewers abstracted data independently from all of the included studies from the updated search, and those from the original review. Furthermore, two reviewers independently classified the QI strategies with our framework (panel). We contacted authors of the trials we included to obtain further information for data items that needed clarification. The Cochrane EPOC method was used to assess the risk of bias in individual studies.12,13 Discrepancies were resolved by discussion or the involvement of a third reviewer.

Statistical analysis
We used well established methods to adjust clusterrandomised controlled trials for meta-analysis with patient-randomised controlled trials.14,15 As in the previous review,8 many of the cluster trials we included did analyses at the patient level rather than the cluster level (ie, unit of analysis errors). In an attempt to avoid spurious estimates in patient-level outcomes, we calculated an effective sample size for each such trial by use of the intracluster correlation coefficient (ICC).16–18 We imputed unreported ICCs based on ICCs reported in other included trials for each outcome. To ensure that we maintained the independence of studies, we included a maximum of two groups in our analysis even if trials had more than two groups.19 For example, if a trial assessed team changes and education of patients versus education of patients alone versus usual care, we included only the team changes and education of patients versus usual care groups in our analysis. This restriction applied for only ten trials we included. We imputed unreported SDs by use of established methods.15,20 We assessed the effects of each QI strategy across the outcomes descriptively, assessing the data distributions, means, medians, and IQRs. We then used a random effects model to estimate the pooled risk ratio (RR, dichotomous data) or the mean difference (continuous data) across the included trials (Comprehensive Meta-analysis Version 2.2050).21 We assessed the consistency of results across the studies by use of forest plots and the statistical heterogeneity with the I² statistic.22 We did a post-hoc secondary analysis to explore whether the effectiveness of QI strategies varied in studies enrolling patients with diabetes who had poor baseline achievement of quality indicators (defined by baseline HbA1c, LDL cholesterol, and diastolic and systolic blood pressures).

See Online for appendix

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

Articles

Panel: Taxonomy of quality improvement strategies Quality improvement (QI) strategies targeting health systems Case management Any system for coordinating diagnosis, treatment, or routine management of patients (eg, arrangement for referrals, follow-up of test results) by a person or multidisciplinary team in collaboration with, or supplementary to, the primary-care clinician. For a randomised controlled trial to qualify, the case management had to happen more than once. Most of these studies had less involvement than in those with team changes (ie, case manager did not have to speak with primary-care physician). If the study called the intervention “case management” we classified it as such. Team changes Changes to the structure or organisation of the primary health-care team were defined as present if they met certain criteria: • Adding a team member or shared care—eg, routine visits with people other than the primary physician (including physician or nurse specialists in diabetic care, pharmacists, nutritionists, podiatrists). • Use of multidisciplinary teams—ie, active participation of professionals from more than one discipline (eg, medicine, nursing, pharmacy, nutrition) in the primary, routine management of patients. • Expansion or revision of professional roles (eg, nurse or pharmacist has a more active role in monitoring of the patient or adjusting drug regimens). To ensure that every study we classified as case management would not also qualify as a team change, we could classify a study that was already classified as case management also as a team change if at least two of the above conditions were met. Team changes involved more communication. If the study called the intervention “joint visits” or “shared care”, we classified it as a team change. To qualify, the intervention had to be done by a health-care professional and had to happen more than once. Electronic patient registry General electronic medical record system or electronic tracking system for patients with diabetes. We did not include websites unless patients were tracked over time. To qualify, it had to be a part of the clinical trial as an intervention (ie, not pre-existing infrastructure unless used more actively). Facilitated relay of information to clinicians Clinical information collected from patients and transmitted to clinicians by means other than the existing medical record. We excluded conventional means of correspondence between clinicians. For example, if the results of routine visits with a pharmacist were sent in a letter to the primary-care physician, the use of routine visits with a pharmacist would count as a “team” change, but the intervention would not also be counted as “facilitated relay”. However, if the pharmacist issued structured diaries for patients to record self-monitored glucose values, which were then taken to office visits to review with the primary physician, we would count the intervention as “facilitated relay”. Other examples include electronic or web-based methods through which patients provided self-care data and which clinicians reviewed, as well as point-of-care testing supplying clinicians with immediate HbA1c values. We included passports, referral systems, and dietary information (vs purely clinical information). In general, the patient should be facilitating the relay. To be included, the information must get to someone with prescribing or ordering ability. For example, if the nurse’s role was expanded to make drug changes, the patient had a passport, and the nurse could directly make a change, we would classify the intervention as case management and facilitated relay of clinical information (depending on the study and situation). If the nurse alerted the primary-care provider that the patient had run out of drugs, we did not deem this facilitated relay of information, because that is a normal part of a nurse’s role. Continuous QI Interventions explicitly identified as involving the techniques of continuous QI, total quality management, or plan-do-study-act, or any iterative process for assessing quality problems, developing solutions to those problems, testing their effects, and then reassessing the need for further action. QI strategies targeting health-care providers Audit and feedback Summary of clinical performance of health care delivered by an individual clinician or clinic over a specified period, which was then transmitted back to the clinician (eg, the percentage of a clinician’s patients who achieved a target HbA1c concentration or who underwent dilated-eye examinations with a specified frequency). This strategy was strictly based on clinical data and excluded clinical skills. It could include the number of patients with missing tests and dropouts. Clinician education Interventions designed to promote increased understanding of principles guiding clinical care or awareness of specific recommendations for a target disorder or population of patients. Subcategories of clinician education included conferences or workshops, distribution of educational materials (written, video, or other), and educational outreach visits (ie, academic detailing). We excluded teaching how to educate patients, counselling skills, motivational interviewing, self-directed learning, and skills related to the intervention (eg, teaching how to use the website for the randomised controlled trial). We included all health-care providers. If the education was part of the individual’s role (eg, teaching a case manager about diabetes) we did not categorise it as clinician education. (Continues on next page)

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

3

Articles

(Continued from previous page) Clinician reminders Paper-based or electronic systems intended to prompt a health professional to recall patient-specific information (eg, most recent HbA1c value) or to do a specific task (eg, foot examination). If the strategy was accompanied by a recommendation, we subclassified it as decision support (eg, giving targets to healthcare providers). An example is a yellow piece of paper clipped to the medical record with the patient’s information on it. This approach had to be systematic and part of the implementation of the intervention—we excluded ad-hoc clinician reminders. Financial incentives Interventions with positive or negative financial incentives directed at providers (eg, linked to adherence to some process of care or achievement of some target outcome). This strategy also includes positive or negative financial incentives directed at patients or system-wide changes in reimbursement (eg, capitation, prospective payment, or a shift from fee-for-service to salary pay structure). QI strategies targeting patients Education of patients Interventions designed to promote greater understanding of a target disorder or to teach specific prevention or treatment strategies, or specific in-person education (eg, individual or group sessions with diabetes nurse educator; distribution of

printed or electronic educational materials). Interventions with education of patients were included only if they also included at least one other strategy related to clinician or organisational change. We did not include occasions of optional education. Promotion of self-management Provision of equipment (eg, home glucose meters) or access to resources (eg, system for electronically transmitting home glucose measurements and receiving insulin dose changes based on those data) to promote self-management. Interventions promoting self-management were included only if they also included at least one other strategy related to clinician or organisational change. We also included established goals or a print off of a selfmanagement plan (ie, did not necessarily require equipment or resources). If the study called the intervention promotion of selfmanagement, personalised goal-setting, or action-planning, we included it here. We generally thought this a more active strategy than education of patients. Reminder systems Any effort (eg, postcards or telephone calls) to remind patients about upcoming appointments or important aspects of self care. Interventions with reminders were included only if they also included at least one other strategy related to clinician or organisational change. Examples included reminders to monitor glucose. If the intervention included case management, reminders to patients needed to be explicit and an extra task to the normal case management.

We decided a priori to do meta-regression with a linear fixed-effects model (Proc Mixed SAS Version 9.2) for studies reporting HbA1c. Our meta-regression adjusted for two study characteristics, median baseline HbA1c (<8·0% vs ≥8·0%) and median effective sample size (≤141 patients vs >141 patients). The sample size variable largely accounted for study design (ie, patient trials vs cluster trials), since cluster-randomised trials included many more patients than patient-randomised trials. We chose these characteristics a priori because they were methodologically relevant and significantly predicted HbA1c concentrations in univariate analysis. Because of the complexity of the combination of QI strategies we assessed in each trial and the restricted number of similar combinations across all trials, we assessed the QI strategies separately in our analysis (ie, the QI strategies were dichotomised). For example, if a trial compared five different QI strategies versus usual care and reported a reduction of 0·3% in HbA1c, we applied this result to each of the five QI strategies assessed for this trial. To assess the effects of individual QI strategies on the HbA1c results, we also did meta-regression analyses of trials with a given QI intervention versus trials without the particular QI intervention. For example, we included all trials in the model and then we assessed the effect of a given QI intervention (eg, team changes) on the HbA1c estimate by excluding the trials contributing data to team changes.
4

5592 titles and abstracts 3440 excluded 2064 not a randomised trial 1376 not an assessment of quality improvement interventions 2152 full-text articles reviewed 1990 excluded 617 not an assessment of quality improvement interventions 498 not a randomised trial 358 excluded topic 259 no component of clinician or organisational change 134 not diabetes care 109 did not report eligible outcomes or usable data 15 English translation unavailable

162 included randomised trials 48 cluster trials plus 6 companion reports 94 patient trials plus 14 companion reports

Figure 1: Study profile

Role of the funding source
The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

Articles

Results
Figure 1 shows the study profile. 48 cluster-randomised trials, including 2538 clusters and 84 865 patients, and 94 patient-randomised trials, including 38 664 patients, fulfilled our inclusion criteria. 20 companion reports provided supplementary information (appendix).

Many characteristics of studies and patients were similar for patient and cluster trials (table 1, appendix). However, the two types differed with respect to sample size, masking, and who gave the intervention. For patient-randomised trials, most included the QI strategies of patients’ education, promotion of selfmanagement, team changes, and case management (table 1). By contrast, cluster-randomised trials mostly
Patient trials (N=94; Cluster trials (N=48; 38 664 patients)* 84 865 patients) (Continued from previous column) Trials per QI strategies Audit and feedback 4 (4%) 49 (52%) 46 (49%) 19 (20%) 6 (6%) 9 (10%) 26 (28%) 41 (44%) 49 (52%) 15 (16%) 0 0 46 (49%) 9 (10%) 8 (9%) 7 (8%) 2 (2%) 2 (2%) 3 (3%) 2 (2%) 1 (1%) 1 (1%) 1 (1%) 2 (2%) 2 (2%) 1 (1%) 1 (1%) 1 (1%) 3 (3%) 1 (1%) 1 (1%) ·· ·· ·· ·· 11 (23%) 7 (15%) 8 (17%) 12 (25%) 20 (42%) 13 (27%) 8 (17%) 19 (40%) 14 (29%) 8 (17%) 4 (8%) 1 (2%) 22 (46%) 2 (4%) 6 (13%) ·· ·· 6 (13%) 3 (6%) 1 (2%) ·· 1 (2%) ·· ·· ·· ·· 2 (4%) 1 (2%) ·· 1 (2%) ·· 1 (2%) 1 (2%) 1 (2%) 1 (2%)

Patient trials (N=94; Cluster trials (N=48; 38 664 patients)* 84 865 patients) Duration of intervention (months) Longest duration of follow-up (months) Study outcomes Aspirin use Statin use Any hypertensive drug use Retinopathy screening Renal screening Foot screening HbA1c LDL cholesterol Diastolic blood pressure Systolic blood pressure Hypertension control Smoking cessation Number of clusters Number of patients Mean age (years) Percentage male Type of diabetes Type 1 Type 2 Types 1 and 2 Type unclear or not reported Primary-care physician Nurse Pharmacist Dietitian Psychiatrist Psychologist Ophthalmologist Specialist or endocrinologist Other Masking Intervention masked from patients Intervention masked from patients’ assessors Intervention masked from patients’ providers Number of QIs per trial 3 (3%) 11 (12%) NA 3 (2–4) 3 (6%) 12 (25%) 2 (4%) 3 (2–3) (Continues in next column) 8 (6%) 52 (55%) 26 (28%) 8 (9%) 1 (2%) 28 (58%) 8 (17%) 11 (23%) 4 (4%) 7 (8%) 8 (9%) 10 (11%) 4 (4%) 9 (10%) 84 (89%) 31 (33%) 39 (42%) 40 (44%) 10 (11%) 6 (6%) NA 127 (63–206) 56·4 (51·6–60·7) 50·6 (38·8–59·1) 7 (15%) 4 (8%) 3 (6%) 15 (31%) 11 (23%) 15 (31%) 32 (67%) 16 (33%) 23 (48%) 25 (52%) 8 (17%) 9 (19%) 29 (12–57) 684 (343–1549) 62·4 (58·1–65·1) 49 (44·7–52·7) 12 (6–12) 12 (6–13) 12 (11·8–18) 12 (12–21)

Case management Team changes Electronic patient registry Clinician education Clinician reminders Facilitated relay Patient education Promotion of selfmanagement Patient reminders Continuous quality improvement Financial incentives Country of publication USA Canada UK South Korea China Netherlands Australia Denmark Thailand

Administrators of interventions for patients 18 (19%) 53 (56%) 17 (18%) 19 (20%) 3 (3%) 1 (1%) 2 (2%) 18 (19%) 34 (36%)† 12 (25%) 14 (29%) 2 (4%) 3 (6%) 0 1 (21%) 0 3 (6%) 15 (31%)

Norway Finland France Germany Taiwan Israel Italy Spain Switzerland United Arab Emirates Belgium Ireland Mexico New Zealand

Data are median (IQR) or n (%). HbA1c=glycated haemoglobin. NA=not applicable. QI=quality improvement. *Includes three crossover trials and two quasi-randomised trials. †Includes investigators and community workers.

Table 1: Characteristics of studies and patients

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

5

Articles

included patients’ education, clinicians’ education, and promotion of self-management (table 1). Only 39% of the trials (55 of 142) adequately reported allocation sequence generation and 42% (60 of 142) adequately reported concealing the allocation sequence. The corresponding proportion for differences in baseline outcome measures was 10% (14 of 142), for differences in baseline characteristics (eg, demographics) was 16% (23 of 142), for incomplete outcome data was 17% (24 of 142), for potential knowledge of allocated interventions was 8% (11 of 142), for inadequate protection against contamination was 6% (nine of 142), and for potential for selective outcome reporting was 1% (one of 142; appendix). 120 trials reported a mean decrease in HbA1c concentration over a median follow-up of 12 months associated with QI interventions (table 2, figure 2). QI strategies were associated with lower LDL-cholesterol concentrations across 47 trials, lower systolic blood pressure across 65 trials, and lower diastolic blood pressure across 61 trials over a median follow-up of 12 months (table 2, figure 2). QI strategies were associated with an increase in use of aspirin over a median follow-up of 18 months and any antihypertensive drugs over a median follow-up of 13 months (table 2). There were no significant differences associated with QI strategies for use of statins over a median follow-up of 19 months and achievement of adequate control of hypertension over a median followup of 12 months (table 2). QI strategies were associated with increases in retinopathy screening, screening for renal involvement, and foot screening over a median follow-up of

12 months (table 2). QI strategies were not associated with a significant difference in smoking cessation rates over a median follow-up of 12 months (table 2). The six trials that included smoking cessation counselling as part of their QI strategy did not achieve greater cessation rates. In studies enrolling patients who had poor baseline achievement of quality indicators, QI strategies were associated with larger effects across HbA1c, systolic and diastolic blood pressure, and LDL cholesterol (table 3). The effectiveness of each QI strategy varied by baseline HbA1c concentration. Decreases in HbA1c of more than 0·5% were noted for four QI strategies (team changes, case management, patients’ education, and promotion of self-management) in trials enrolling patients with HbA1c greater than 8·0%, and one QI strategy (facilitated relay) in trials enrolling patients with HbA1c of 8·0% or less (table 4). After adjustment for median baseline HbA1c values and effective sample size, the QI strategies were associated with significantly lower HbA1c than usual care was (figure 3). All QI strategies were associated with significant changes in HbA1c, except for clinician education. In our planned analysis in which we sequentially omitted all trials with a given QI strategy from our metaregression model, HbA1c was further lowered when the QI strategy included team changes (0·33%), case management (0·21%), promotion of self-management (0·21%), clinician education (0·19%), patient education (0·16%), facilitated relay (0·12%), an electronic patient registry (0·08%), and patient reminders (0·02%).
Median baseline values (IQR) I² Pooled effect (95% CI)*

Studies Number of Median baseline (imputed SDs) patients compliance (IQR) Dichotomous outcomes Aspirin use Statin use Antihypertensive drug use Retinopathy screening Renal screening Foot screening Hypertension control Smoking cessation Hypoglycaemia Severe hypoglycaemia Hyperglycaemia Continuous outcomes HbA1c (%) LDL cholesterol (mmol/L) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) 120 (28) 47 (15) 65 (19) 61 (4) 22 811 11 676 14 791 12 808 NA NA NA NA 11 10 10 23 14 22 18 13 5 6 2 2258 1853 2264 10 455 7317 8144 3813 3231 987 1450 450 10·5% (0·2 to 25·8) 61·35% (55·0 to 74·0) 50·5% (21·3 to 67·8) 47·0% (39·0 to 65·0) 69·5% (44·5 to 76·0) 19·8% (16·3 to 31·8) NA NA NA

NA NA NA NA NA NA NA NA NA 8·19 (7·57 to 9·20) 2·93 (2·71 to 3·20) 139·75 (132·69 to 145·06) 80·00 (76·67 to 83·27)

38·5% 58·2% 91·4% 80·4% 91·6% 89·4% 67·5% 5·3% 0 66·8% 87·4% 73·5% 48·3% 60·3% 59·0%

1·33 (1·21 to 1·45 ) 1·12 (0·99 to 1·28) 1·17 (1·01 to 1·37) 1·22 (1·13 to 1·32) 1·28 (1·13 to 1·44) 1·27 (1·16 to 1·39) 1·01 (0·96 to 1·07) 1·13 (0·99 to 1·29) 0·99 (0·75 to 1·31) 1·0 (0·66 to 1·51) 0·74 (0·28 to 1·92) –0·37 (–0·45 to –0·28) –0·10 (–0·05 to –0·14) –3·13 (–4·06 to –2·19) –1·55 (–2·15 to –0·95)

32·76% (20·4 to 42·8) NA 84·53% (57·4 to 98·0) NA

Effective sample size was used for cluster trials. HbA1c=glycated haemoglobin. NA=not applicable. *Data are relative risk for dichotomous outcomes and mean difference for continuous outcomes.

Table 2: Meta-analysis results across all outcomes

6

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

Articles

However, none of these results were significantly different from the overall HbA1c effect. Five trials reported the proportion of hypoglycaemic events, six trials reported the proportion of severe hypoglycaemic events, and two trials reported the proportion of hyperglycaemic events in patients in the intervention and control groups. We did not identify any significant differences across all adverse events for patients in the intervention or control groups over a median follow-up of 12 months (table 2).

A

Number of trials

Mean difference (95% CI)

Post-intervention reduction in HbA1c (%)

Discussion
Our systematic review is an update of a previous review that assessed the effects of QI strategies on glycaemic control,8 includes more than twice as many trials, and reports the effects of QI strategies on other important aspects of diabetes management. By including outcomes that are deemed quality indicators in the management of diabetes, such as diastolic and systolic blood pressure, LDL cholesterol, medication use, and monitoring for diabetes complications, we were able to assess the effect of QI strategies on a broader range of diabetes care. On the basis of evidence from more than 140 trials, the QI strategies we assessed significantly improved HbA1c, LDL cholesterol, diastolic and systolic blood pressure, aspirin use, antihypertensive drug use, retinopathy screening, renal screening, and foot screening. We noted greater improvements in HbA1c control for QI strategies targeting health systems and patients. We also noted potentially clinically important but non-statistically significant improvements for statin use and smoking cessation. However, more evidence is needed to clarify these potential improvements, since only 13 trials (involving 3231 patients) were included in the smoking cessation meta-analysis and ten trials (involving 1853 patients) in the statin use meta-analysis. We noted no improvement in hypertension control. Since we did not include trials with interventions directed only towards the patient, the effectiveness of patients’ education, patients’ reminders, and promotion of self-management QI strategies should be interpreted as implemented in combination with QI strategies targeting health-care professionals. However, high-quality systematic reviews published in the past 5 years assessing the effects of patient-mediated interventions alone strongly support the benefits of these interventions.23,24 Across most outcomes of interest, most studies enrolled patients who were not achieving diabetes-relevant quality indicators (ie, HbA1c or blood pressure). For example, median HbA1c concentrations across all studies were 8·19%, the median proportion of patients on statins was 32·8%, and the median proportion of patients receiving
Figure 2: Findings from meta-analyses Findings of the meta-analyses for biological markers: glycated haemoglobin (HbA1c; A), LDL cholesterol (LDL; B), systolic blood pressure (C), and diastolic blood pressure (D). Quality improvement strategies with one trial are not based on meta-analysis (we present the individual trial result).

Promotion of self-management 60 Team changes 48 57 Case management Patient education 52 32 Facilitated relay Electronic patient register 27 Patient reminders 21 8 Audit and feedback Clinician education 15 18 Clinician reminders Financial incentives 1 Continuous quality improvements 2 120 All interventions

0·57 (0·31 to 0·83) 0·57 (0·42 to 0·71) 0·50 (0·36 to 0·65) 0·48 (0·34 to 0·61) 0·46 (0·33 to 0·60) 0·42 (0·24 to 0·61) 0·39 (0·12 to 0·65) 0·26 (0·08 to 0·44) 0·19 (0·03 to 0·35) 0·16 (0·02 to 0·31) 0·10 (–0·24 to 0·44) –0·23 (–0·41 to –0·05) 0·37 (0·28 to 0·45) –1·00 –0·50 0 0·50 1·00 Favours control Favours intervention

B
Promotion of self-management Team changes Facilitated relay Clinician reminders Patient education Case management Clinician education Electronic patient register Audit and feedback Patient reminders Continuous quality improvements All interventions

Number of trials 25 17 9 7 20 22 4 12 3 12 1 47

Mean difference (95% CI) 0·18 (0·10 to 0·26) 0·17 (0·07 to 0·27) 0·16 (0·06 to 0·25) 0·14 (0·04 to 0·25) 0·14 (0·04 to 0·23) 0·11 (0·02 to 0·21) 0·11 (–0·12 to 0·33) 0·09 (–0·01 to 0·18) 0·03 (–0·04 to 0·10) 0·01 (–0·04 to 0·07) –0·21 (–0·55 to 0·14) 0·10 (0·05 to 0·14)

Post-intervention reduction in LDL (mmol/L)

–0·50 –0·25 0 0·25 0·50 Favours control Favours intervention

C
Case management Team changes Facilitated relay Patient education Promotion of self-management Electronic patient register Clinician education Audit and feedback Financial incentives Patient reminders Continuous quality improvements Clinician reminders All interventions

Number of trials 25 27 12 28 28 14 18 8 1 12 1 12 65

Mean difference (95% CI) 4·62 (1·52 to 7·73) 4·32 (2·51 to 6·12) 4·31 (2·85 to 5·77) 4·02 (2·52 to 5·52) 3·69 (2·34 to 5·04) 3·35 (1·55 to 5·14) 2·56 (0·00 to 5·11) 2·52 (1·00 to 4·04) 2·00 (–2·73 to 6·73) 1·82 (0·29 to 3·36) 1·00 (–2·66 to 4·66) 0·65 (–1·14 to 2·44) 3·13 (2·19 to 4·06)

Post-intervention reduction in SBP (mm Hg)

–6·00 –3·00 0 3·00 6·00 Favours control Favours intervention

D
Patient education Promotion of self-management Team changes Clinician education Clinician reminders Case management Facilitated relay Electronic patient register Patient reminders Audit and feedback Financial incentives All interventions

Number of trials 29 28 25 15 11 25 12 11 11 7 1 61

Mean difference (95% CI) 2·25 (1·33 to 3·16) 1·89 (0·84 to 2·94) 1·75 (1·00 to 2·51) 1·13 (0·13 to 2·12) 1·11 (–0·02 to 2·24) 0·93 (0·16 to 1·71) 0·82 (0·04 to 1·59) 0·78 (–0·17 to 1·73) 0·76 (–0·24 to 1·76) 0·68 (–0·36 to 1·72) –1·00 (–4·15 to 2·15) 1·55 (0·95 to 2·15)

Post-intervention reduction in DBP (mm Hg)

–4·00 –2·00 0 2·00 4·00 Favours control Favours intervention

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

7

Articles

foot screening was 47·0% at baseline. Our secondary meta-analysis showed that the effectiveness of QI strategies varied depending on baseline glycaemic control. For example, we noted that team changes, case management, patients’ education, and promotion of selfmanagement were the most effective strategies in trials that enrolled patients with mean baseline HbA1c concentrations greater than 8·0%. We noted similar results in our meta-regression analysis when we sequentially omitted each QI strategy, with the most effective strategies being team changes, case management, and promotion of self-management. By contrast, the most effective strategies in trials that included patients with mean baseline HbA1c concentrations of 8·0% or less were facilitated relay, team changes, patients’ reminders, and electronic register of patients. Our findings suggest that QI strategies that intervened upon the entire system of chronic disease management were associated with the largest effects irrespective of baseline HbA1c. The effectiveness of interventions targeting health professionals and patients seemed to

Number of trials Glycated haemoglobin (%) <8·0 ≥8·0 LDL cholesterol (mg/dL) <2·59 ≥2·59 <140 ≥140 <80 >80 20 27 32 33 29 32 46 70

Mean difference (95% CI)

I²

–0·23 (0·34 to –0·13) –0·46 (–0·58 to –0·35) –0·05 (–0·09 to –0·01) –0·15 (–0·23 to –0·08) –2·92 (–4·13 to –1·70) –3·35 (–4·69 to –2·00) –1·13 (–1·98 to –0·29) –1·76 (–2·47 to –1·05)

69·8 72·5 57·5 53·0 72·0 38·3 53·3 71·0

Systolic blood pressure (mm Hg)

Diastolic blood pressure (mm Hg)

Table 3: Effects of quality improvement interventions by baseline levels

vary with baseline HbA1c. For example, clinicians’ education and audit and feedback led to an HbA1c reduction of 0·33% and 0·44%, respectively, when baseline HbA1c concentrations were greater than 8·0%, but no improvement when baseline HbA1c was less than 8·0%. Patients’ education seemed more effective than reminders when baseline HbA1c was greater than 8·0% but less effective when the HbA1c was less than 8·0%. These findings suggest that QI strategies that aim to optimise the systems of care should (whenever feasible) be included in programmes to improve diabetes management, irrespective of HbA1c. Interventions targeting patients might be beneficial irrespective of baseline HbA1c, whereas interventions targeting providers only seem beneficial when baseline HbA1c is greater than 8·0%. Our systematic review has some limitations. We were unable to include 15 trials published in languages other than English, although we did contact authors for English translations. Limitations of our data analysis include the complexity of the QI strategies, which were difficult to classify consistently, and we could not control for all potential confounding factors. We were unable to assess interactions in the meta-regression analysis (because too few trials per outcome were included), many of the analyses had substantial heterogeneity (which is to be expected in view of the large number of trials included and number of QI strategies assessed), and the definition of usual care was not consistent across the studies. Most trials reported HbA1c concentrations, with fewer reporting other key aspects of diabetes management, showing that glycaemic control remains (rightly or wrongly) the major focus for management of diabetes. As a result, we were unable to undertake meta-regression for other outcomes. We cannot tell whether the interventions that seemed more effective for HbA1c would have similar effects on other key endpoints, although our preliminary analysis suggests some consistency across outcomes. The data on adverse events should be interpreted with caution, since
Glycated haemoglobin ≤8·0% Rank Number Mean difference of trials (95% CI) 6 2 7 5 1 4 3 9 10 8 23 17 17 13 13 18 11 3 5 9 46 –0·29 (–0·47 to –0·12) –0·46 (–0·71 to –0·21) –0·25 (–0·44 to –0·07) –0·39 (–0·71 to –0·06) –0·54 (–0·79 to –0·30) –0·41 (–0·60 to –0·22) –0·42 (–0·70 to –0·15) –0·06 (–0·16 to 0·06) 0·03 (–0·18 to 0·25) –0·06 (–0·15 to 0·04) –0·23 (–0·34 to –0·13)

All studies Rank Number Mean difference of trials (95% CI) Promotion of self management Team changes Case management Patient education Facilitated relay Electronic patient register Patient reminders Audit and feedback Clinician education Clinician reminders All interventions 1 2 3 4 5 6 7 8 9 10 60 47 57 52 32 27 21 8 15 18 120 –0·57 (– 0·83 to –0·31) –0·57 (–0·71 to –0·42) –0·50 (–0·65 to –0·36) –0·48 (–0·61 to –0·34) –0·46 (–0·60 to –0·33) –0·42 (–0·61 to –0·24) –0·39 (–0·65 to –0·12) –0·26 (–0·44 to –0·08) –0·19 (–0·35 to 0·03) –0·16 (–0·31 to –0·02) –0·37 (–0·45 to –0·28)

Glycated haemoglobin >8·0% Rank Number Mean difference of trials (95% CI) 4 1 2 3 6 5 8 7 10 9 37 31 37 39 19 9 10 5 10 9 70 –0·56 (–0·70 to –0·42) –0·62 (–0·79 to –0·46) –0·61 (–0·80 to –0·42) –0·59 (–0·74 to –0·43) –0·42 (–0·56 to –0·29) –0·47 (–0·79 to –0·14) –0·39 (–0·77 to –0·00) –0·40 (–0·77 to –0·03) –0·33 (–0·57 to –0·10) –0·35 (–0·56 to –0·13) –0·46 (–0·58 to –0·35)

Table 4: Ranking of quality improvement strategies across glycated haemoglobin primary and secondary meta-analyses

8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

Articles

few trials reported safety data and those trials might differ systematically from trials that did not report safety data with respect to adverse events. Furthermore, trials reporting this data followed up patients for only a short duration (longest duration of follow-up was 12 months) and included a small number of patients. Most studies had short intervention durations and brief follow-up, meaning that we were unable to assess longer-term outcomes, such as mortality. Our findings suggest that key aspects and intermediate outcomes of diabetes care can be improved and that a larger effect is evident when baseline achievement of quality indicators is poor. If implemented widely, the population benefits of the observed effects are potentially important. For example, data from the United Kingdom Prospective Diabetes Study (UKPDS) suggested that a 1% reduction in mean HbA1c results in 21% fewer deaths, 14% fewer myocardial infarctions, and a 37% decrease in microvascular complications at the population level.25 We recorded a 0·33% reduction in mean HbA1c, which, if the QI strategies are employed, might translate to 7% fewer deaths, 5% fewer myocardial infarctions, and 12% fewer microvascular complications at the population level. It is plausible that further population-level improvements in these outcomes could be achieved through improved vascular risk-factor management (eg, better blood pressure and lipid control). Larger population benefits would probably accrue in populations with poor-quality indicators. The large I² values from our meta-analysis results suggest that some of the QI interventions might be more effective than others. Detailed descriptions of the QI strategies were lacking in many of the trial reports and optimum combinations of the QI strategies, as well as ways of implementing and delivering the QI interventions, remain unclear. As such, although our report provides information on the relative effectiveness of the different QI strategies, how best to deliver the most effective QI strategies remains uncertain. This information is crucial, since it will allow policy makers to tailor the choice of intervention to the desired outcome, available resources, and local health-care context. Since the strategies seem more effective in patients not achieving quality indicators, careful selection of patients who will benefit most from these QI strategies needs consideration by decision makers. Moreover, since several strategies were marginally beneficial relative to other strategies, and the resource intensity of the different strategies varied significantly (probably being highest for case management and team changes), further exploration of the relative costeffectiveness of these QI strategies is needed. Decision makers might also consider how they value the expected benefits before widely implementing such QI strategies. Future assessments should explicitly build on the present evidence base by targeting a broad range of important diabetes process and outcome measures and

Number of trials Team changes Facilitated relay Promotion of self-management Case management Patient education Electronic patient register Clinician reminders Patient reminders Audit and feedback Clinician education All interventions 47 31 57 52 52 28 16 20 9 12 117

Mean difference (95% CI) 0·52 (0·00 to 1·04) 0·49 (0·02 to 0·96) 0·45 (0·04 to 0·87) 0·41 (0·00 to 0·82) 0·40 (0·00 to 0·80) 0·39 (0·00 to 0·78) 0·35 (0·00 to 0·70) 0·31 (0·00 to 0·62) 0·22 (0·00 to 0·44) 0·16 (0·01 to 0·33) 0·33 (0·01 to 0·65)

Post-intervention reduction in HbA1c (%)

–0·50 0 0·50 1·00 Favours control Favours intervention

Figure 3: Glycated haemoglobin meta-regression results We derived estimates from a meta-regression model, adjusting for median baseline glycated haemoglobin values (<8·0% vs ≥8·0%) and the median number of patients included in the randomised clinical trials (≤141 patients vs >141 patients).

carefully assessing the role of context. The QI strategy should be carefully tailored (eg, intervention mapping26) and the interventions should be thoroughly described.27 Stakeholders should prioritise testing different QI strategies head-to-head in adequately powered, multigroup trials and assess explicitly postulated mechanisms of action of the interventions (ie, process assessments) to inform generalisation to different settings. Further research is needed to identify which interventions and combination of QI strategies will optimally improve important outcomes in patients with diabetes at an acceptable cost to aid health-system planning.
Contributors JMG and KS conceived the systematic review. JMG, KS, ACT, and DM designed the systematic review. ACT, NMI, LT, JG, IH, and BV selected studies for inclusion and abstracted data. LT and TR analysed the data. ACT wrote the first draft, which was revised by all authors. All authors approved the final draft. Conflicts of interest We declare that we have no conflicts of interest. Acknowledgments We acknowledge the generous funding from the Ontario Ministry of Health and Long-term Care and the Alberta Heritage Foundation for Medical Research Interdisciplinary Team Grants program (now Alberta Innovates—Health Solutions). NI is supported by a Canadian Institutes for Health Research (CIHR) Fellowship Award; JMG and MT are supported by Canada Research Chairs. DM is supported by a University of Ottawa Research Chair. BM and MT are also supported by the Alberta Heritage Foundation for Medical Research Population Health Scholar awards. We thank the Advisory Board, Joan Canavan, Michael Hillmer, Erin Keely, and Baiju Shah for their guidance on the systematic review protocol. We thank Michelle Fiander for undertaking the searches of published work, Raymond Daniel for obtaining the articles, Steve Doucette for his statistical analysis expertise, and Alain Mayhew for guiding the review conceptualisation. We thank Natasha Wiebe, Natasha Krahn, Avtar Lal, and Mohammad Karkhaneh for their assistance with abstracting some of the included studies and appraising their study quality. References 1 American Diabetes Association. Standards of medical care in diabetes—2010. Diabetes Care 2010; 33 (suppl 1): S11–61. 2 Braga M, Casanova A, Teoh H, et al. Treatment gaps in the management of cardiovascular risk factors in patients with type 2 diabetes in Canada. Can J Cardiol 2010; 26: 297–302.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

9

Articles

3

4

5

6 7

8

9

10

11

12

13

14 15

Saaddine JB, Engelgau MM, Beckles GL, Gregg EW, Thompson TJ, Narayan KM. A diabetes report card for the United States: quality of care in the 1990s. Ann Intern Med 2002; 136: 565–74. McFarlane SI, Jacober SJ, Winer N, et al. Control of cardiovascular risk factors in patients with diabetes and hypertension at urban academic medical centers. Diabetes Care 2002; 25: 718–23. Narayan KM, Benjamin E, Gregg EW, Norris SL, Engelgau MM. Diabetes translation research: where are we and where do we want to be? Ann Intern Med 2004; 140: 958–63. Kogan AJ. Overcoming obstacles to effective care of type 2 diabetes. Am J Manag Care 2009; 15 (9 suppl): S255–62. Simpson SH, Corabian P, Jacobs P, Johnson JA. The cost of major comorbidity in people with diabetes mellitus. CMAJ 2003; 168: 1661–67. Shojania KG, Ranji SR, McDonald KM, et al. Effects of quality improvement strategies for type 2 diabetes on glycemic control: a meta-regression analysis. JAMA 2006; 296: 427–40. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ 2009; 339: b2535. Shojania KG, Ranji SR, Shaw LK, et al. Diabetes Mellitus Care— vol 2. In: Shojania KG, McDonald KM, Wachter RM, Owens DK. Closing the quality gap: a critical analysis of quality improvement strategies—technical review 9. Rockville, MD: Agency for Healthcare Research and Quality, 2004. Sampson M, McGowan J, Cogo E, Grimshaw J, Moher D, Lefebvre C. An evidence-based practice guideline for the peer review of electronic search strategies. J Clin Epidemiol 2009; 62: 944–52. Armstrong R, Waters E, Doyle J, eds. Chapter 21: reviews in public health and health promotion. In: Higgins JPT, Green S, eds. Cochrane handbook for systematic reviews of interventions version 5.1.0. Oxford: The Cochrane Collaboration, 2011. Cochrane Effective Practice and Organisation of Care Group. Risk of bias—EPOC specific. http://epoc.cochrane.org/epoc-authorresources (accessed July 1, 2009). Rao JN, Scott AJ. A simple method for the analysis of clustered binary data. Biometrics 1992; 48: 577–85. Higgins JPT, Green S, eds. Cochrane handbook for systematic reviews of interventions version 5.1.0. Oxford: The Cochrane Collaboration, 2011.

16 17 18

19

20

21 22 23

24

25

26

27

Kerry SM, Bland JM. The intracluster correlation coefficient in cluster randomisation. BMJ 1998; 316: 1455. Donner A, Klar N. Issues in the meta-analysis of cluster randomized trials. Stat Med 2002; 21: 2971–80. Campbell M, Grimshaw J, Steen N. Sample size calculations for cluster randomised trials. Changing Professional Practice in Europe Group (EU BIOMED II Concerted Action). J Health Serv Res Policy 2000; 5: 12–16. Cook C. Julia H Littell, Jacqueline Corcoran, Vijayan Pillai, Systematic reviews and meta-analysis (2008) Oxford University Press, New York 202 ISBN: 978-0-19-532654-3. Child Youth Serv Rev 2009; 31: 495–96. Furukawa TA, Barbui C, Cipriani A, Brambilla P, Watanabe N. Imputing missing standard deviations in meta-analyses can provide accurate results. J Clin Epidemiol 2006; 59: 7–10. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–88. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002; 21: 1539–58. Allemann S, Houriet C, Diem P, Stettler C. Self-monitoring of blood glucose in non-insulin treated patients with type 2 diabetes: a systematic review and meta-analysis. Curr Med Res Opin 2009; 25: 2903–13. Duke SA, Colagiuri S, Colagiuri R. Individual patient education for people with type 2 diabetes mellitus. Cochrane Database Syst Rev 2009; 1: CD005268. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000; 321: 405–12. van Bokhoven MA, Kok G, van der Weijden T. Designing a quality improvement intervention: a systematic approach. Qual Saf Health Care 2003; 12: 215–20. Michie S, Fixsen D, Grimshaw JM, Eccles MP. Specifying and reporting complex behaviour change interventions: the need for a scientific method. Implement Sci 2009; 4: 40.

10

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60480-2

Articles

Insulin glargine versus sitagliptin in insulin-naive patients with type 2 diabetes mellitus uncontrolled on metformin (EASIE): a multicentre, randomised open-label trial
Pablo Aschner*, Juliana Chan*, David R Owens, Sylvie Picard, Edward Wang, Marie-Paule Dain, Valérie Pilorget, Akram Echtay, Vivian Fonseca, on behalf of the EASIE investigators

Summary
Background In people with type 2 diabetes, a dipeptidyl peptidase-4 (DPP-4) inhibitor is one choice as second-line treatment after metformin, with basal insulin recommended as an alternative. We aimed to compare the efficacy, tolerability, and safety of insulin glargine and sitagliptin, a DPP-4 inhibitor, in patients whose disease was uncontrolled with metformin. Methods In this comparative, parallel, randomised, open-label trial, metformin-treated people aged 35–70 years with glycated haemoglobin A1c (HbA1c) of 7–11%, diagnosis of type 2 diabetes for at least 6 months, and body-mass index of 25–45 kg/m² were recruited from 17 countries. Participants were randomly assigned (1:1) to 24-week treatment with insulin glargine (titrated from an initial subcutaneous dose of 0·2 units per kg bodyweight to attain fasting plasma glucose of 4·0–5·5 mmol/L) or sitagliptin (oral dose of 100 mg daily). Randomisation (via a central interactive voice response system) was by random sequence generation and was stratified by centre. Patients and investigators were not masked to treatment assignment. The primary outcome was change in HbA1c from baseline to study end. Efficacy analysis included all randomly assigned participants who had received at least one dose of study drug and had at least one on-treatment assessment of any primary or secondary efficacy variable. This trial is registered at ClinicalTrials.gov, NCT00751114. Findings 732 people were screened and 515 were randomly assigned to insulin glargine (n=250) or sitagliptin (n=265). At study end, adjusted mean reduction in HbA1c was greater for patients on insulin glargine (n=227; –1·72%, SE 0·06) than for those on sitagliptin (n=253; –1·13%, SE 0·06) with a mean difference of –0·59% (95% CI –0·77 to –0·42, p<0·0001). The estimated rate of all symptomatic hypoglycaemic episodes was greater with insulin glargine than with sitagliptin (4·21 [SE 0·54] vs 0·50 [SE 0·09] events per patient-year; p<0·0001). Severe hypoglycaemia occurred in only three (1%) patients on insulin glargine and one (<1%) on sitagliptin. 15 (6%) of patients on insulin glargine versus eight (3%) on sitagliptin had at least one serious treatment-emergent adverse event. Interpretation Our results support the option of addition of basal insulin in patients with type 2 diabetes inadequately controlled by metformin. Long-term benefits might be expected from the achievement of optimum glycaemic control early in the course of the disease. Funding Sanofi.
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60439-5 See Online/Comment DOI:10.1016/S01406736(12)60780-6 *Equal contributions Pontificia Universidad Javeriana, Hospital Universitario San Ignacio, Bogotá, Colombia (Prof P Aschner MD); The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong Special Administrative Region, China (Prof J Chan FRCP); Centre for Endocrine and Diabetes Science, Cardiff University, Cardiff, UK (Prof D R Owens MD); Point Medical Rond Point de la Nation, Dijon, France (S Picard MD); Sanofi, Bridgewater, NJ, USA (E Wang PhD); Sanofi, Paris, France (M-P Dain MD, V Pilorget MD); Rafic Hariri University Hospital, Beirut, Lebanon (A Echtay MD); and Tulane University Medical Center, New Orleans, LA, USA (Prof V Fonseca MD) Correspondence to: Prof Pablo Aschner, Pontificia Universidad Javeriana, Hospital Universitario San Ignacio, Bogotá, Colombia [email protected]

Introduction
In type 2 diabetes, large-scale randomised clinical trials have shown that early achievement of glycaemic control with glycated haemoglobin A1c (HbA1c) less than 7% resulted in long-term benefits in reduction of microvascular complications and might reduce macrovascular problems.1–3 With the exception of patients with substantial renal impairment or gastrointestinal intolerance, metformin is the mainstay of treatment in type 2 diabetes. However, most people eventually need additional treatment to achieve their glycaemic goal.4–6 In view of the low associated risk of hypoglycaemia and neutral effect on bodyweight, dipeptidyl peptidase-4 (DPP-4) inhibitors are increasingly added to metformin as an alternative second agent to sulphonylureas.7,8 Alternatively, randomised clinical trials9,10 and meta-analysis11 have suggested that the

early addition of basal insulin to metformin treatment can lower HbA1c effectively with good tolerability. To date, no studies have compared the use of DPP-4 inhibitors versus basal insulin in people with type 2 diabetes who have not responded to metformin monotherapy. In the EASIE (Evaluation of insulin glargine versus Sitagliptin in Insulin-naive patients) trial, we aimed to compare the efficacy, safety, and tolerability of basal insulin (insulin glargine) versus a DPP-4 inhibitor (sitagliptin) in such a population during a 24-week period.

Methods
Study design
EASIE was a multicentre, 6-month, comparative, twoarm, parallel, randomised, open-label trial undertaken in 17 countries (appendix p 1) from Nov 12, 2008, to

See Online for appendix

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

1

Articles

Baseline week 0

Treatment Insulin glargine plus metformin

Follow-up

Insulin-naive patients with type 2 diabetes mellitus HbA1c ≥7% and <11% n=515

Intermediate clinical visits at weeks 2, 6, 12, 16 Randomisation Intermediate telephone calls visits at weeks 1, 3, 4, 5, 8, 10, 20 Sitagliptin plus metformin 24 weeks 1–7 days

Figure 1: Study design HbA1c=glycated haemoglobin A1c.

732 people screened

217 excluded 201 inclusion or exclusion criteria not met 16 other

had HbA1c of 7% or greater and less than 11% and bodymass index (BMI) between 25 kg/m² and 45 kg/m² (inclusive). Furthermore, they had to be willing to take structured self-monitored blood glucose measurements and complete a monitoring diary. People were excluded if they had been treated with oral glucose-lowering drugs other than metformin within the past 3 months, had received combination treatment with metformin plus a sulphonylurea in the past year, or had previous treatment with glucagon-like peptide-1 agonists or DPP-4 inhibitors. Other exclusion criteria were fasting plasma glucose of 15·4 mmol/L or more, impaired renal function (serum creatinine ≥133 μmol/L in men or ≥124 μmol/L in women) or hepatic function (greater than three times the upper limit of the normal range for alanine aminotransferase or aspartate aminotransferase), or any disorder (present or expected) that the investigator felt would compromise the patient’s safety or restrict the patient’s successful participation in the study.

Randomisation and masking
Eligible people were allocated a four-digit randomisation number through a central interactive voice response system in France. The randomisation list, generated centrally by a clinical research organisation (Cardinal Systems, Paris, France), linked sequential numbers to the treatment allocated at random. Stratification by centre was done to ensure a balanced number of participants in each treatment group at each centre (1:1 ratio). Participants were randomly assigned to treatment groups in the order in which they qualified for inclusion in the study. Participants and investigators were not masked to group assignment.

515 randomised

250 glargine

265 sitagliptin

13 not treated 8 withdrew consent 4 poor compliance 1 other

1 not treated (withdrew consent)

Safety population

237 glargine

264 sitagliptin

10 no postbaseline assessment for any efficacy variable

11 no postbaseline assessment for any efficacy variable

Procedures
Clinic visits were scheduled for screening, randomisation (week 0), and weeks 2, 6, 12, 16, and 24 with telephone visits at weeks 1, 3, 4, 5, 8, 10, and 20. After week 24 and within a week, a follow-up phone call was done. All participants received a glucose meter (calibrated to read plasma glucose values) to record self-monitored glucose values. Individuals randomly assigned to insulin glargine implemented insulin titration to attain self-monitored fasting plasma glucose concentrations between 4·0 mmol/L and 5·5 mmol/L (inclusive). The initial subcutaneous dose was 0·2 units per kg of bodyweight injected at dinner or bedtime using a prefilled SoloSTAR pen (sanofi-aventis, Frankfurt, Germany). The dose was either decreased by two units if fasting plasma glucose concentration was less than 4·0 mmol/L with or without symptomatic hypoglycaemia, increased by two units if the concentration was 5·6–7·7 mmol/L, and increased by four units if the concentration was greater than 7·7 mmol/L. Participants monitored fasting plasma glucose daily and generally used the middle of the past three values to undertake the titration twice a week. Selfmonitored glucose values and insulin doses were reviewed by an international titration committee on an ongoing

Efficacy population 227 glargine

253 sitagliptin

Figure 2: Study profile

July 28, 2011. It included an initial 2-week screening period followed by 6 months of treatment (insulin glargine or sitagliptin) and, finally, a 1–7 day follow-up to record any new adverse event or episode of symptomatic hypoglycaemia that occurred during the 24 h after the last study dose (figure 1). The study was undertaken in accordance with the Declaration of Helsinki and the Guidelines for Good Clinical Practice. Every centre obtained local research ethics committee approval after approval from a multicentre research ethics committee. All participants gave full informed written consent before entry into the study.

Patients
People aged 35–70 years (inclusive) and diagnosed with type 2 diabetes for at least 6 months were eligible if they
2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

Articles

basis via a website and the study investigators were contacted by email if titration was inadequate. Minor departures from the algorithm were allowed, although the final decision rested with the investigator. Participants in the sitagliptin group received a fixed oral dose of 100 mg once daily taken in the morning either with or without food and no changes in dose were allowed during the trial. HbA1c was recorded at baseline (week 0), week 12, and week 24, and seven-point plasma glucose profiles were recorded twice during the week before clinic visits at weeks 0, 12, and 24. Seven-point plasma glucose was measured immediately before and 2 h after breakfast, lunch, and dinner and at bedtime. Self-monitored fasting plasma glucose measurements were recorded over 6 consecutive days before the visits at weeks 0, 6, 12, 16, and 24 with bodyweight recorded at the same clinic visits. For the insulin glargine group, the last insulin dose given before the visits on weeks 2, 6, 12, 16, and 24 was recorded. HbA1c and other laboratory blood tests were analysed by a central laboratory. The primary objective was to show the superiority of insulin glargine over sitagliptin in reduction of HbA1c from baseline to the end of the 6-month treatment period. The primary efficacy variable was change in HbA1c from baseline to study endpoint. We also assessed several secondary efficacy variables: HbA1c at baseline, week 12, week 24, and study end; proportion of participants achieving HbA1c less than 7% or less than 6·5% for the same timepoints; self-monitored fasting plasma glucose measured on 6 consecutive days at or before baseline (week 0) and weeks 6, 12, 16, and 24; seven-point plasma glucose profiles over 2 days in the week before baseline and weeks 12 and 24; insulin doses at weeks 0, 2, 6, 12, 16, and 24 and study end; and lipid profile at baseline and study end. Safety variables were adverse events reported by the patient or noted by the investigator, standard haematology and blood chemistry tests, bodyweight, vital signs, and hypoglycaemia. Symptomatic hypoglycaemia was defined as an event with typical symptoms (eg, sweating, palpitation, feeling of hunger) with or without confirmation by a plasma glucose less than 4·0 mmol/L. Severe symptomatic hypoglycaemia was defined as episodes necessitating assistance from another person and associated with a measured plasma glucose lower than 2·0 mmol/L or with prompt recovery after oral carbohydrate, intravenous glucose, or glucagon administration.

assessments under treatment; in particular, this method allowed evaluation of the change in HbA1c from baseline. All participants who were randomly assigned to treatment groups and who were treated were included in the safety population for analysis.
Glargine (n=227) Sitagliptin (n=253) Total (n=480) Age (years) Women Bodyweight (kg) Body-mass index (kg/m²) Duration of diabetes (years) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Heart rate (beats per min) Duration of OADs (years) Antidiabetic treatments in previous 3 months α-glucosidase inhibitors Metformin Fast-acting insulin or insulin analogues Dose of metformin (mg) Any late diabetes complication Myocardial infarction Angina pectoris Coronary artery disease Heart failure Stroke Transient ischaemic attack Peripheral vascular disease Diabetic neuropathy Diabetic nephropathy Diabetic retinopathy HbA1c (%) FPG (mmol/L) SMFPG (mmol/L) Total cholesterol (mmol/L) HDL cholesterol (mmol/L) LDL cholesterol (mmol/L) Triglycerides (mmol/L) Aspartate aminotransferase (UI/L) Alanine aminotransferase (UI/L) Serum creatinine (μmol/L) Concomitant treatments (other than OADs) Any β blockers Calcium-channel blockers Diuretic agents Agents acting on renin-angiotensin system Lipid-modifying agents Antithrombotic agents 198 (87%) 52 (23%) 33 (15%) 41 (18%) 131 (58%) 111 (49%) 75 (33%) 217 (86%) 47 (19%) 39 (15%) 35 (14%) 134 (53%) 114 (45%) 90 (36%) 415 (86%) 99 (21%) 72 (15%) 76 (16%) 265 (55%) 225 (47%) 165 (34%) 1 (<1%) 227 (100%) 1 (<1%) 1852 (535) 65 (29%) 11 (5%) 10 (4%) 26 (11%) 4 (2%) 5 (2%) 2 (1%) 4 (2%) 24 (11%) 11 (5%) 12 (5%) 8·5% (1·0) 9·1 (2·2) 9·1 (2·1) 4·8 (1·1) 1·2 (0·4) 2·9 (0·9) 2·2 (1·6) 24·3 (10·8) 31·4 (17·4) 69·5 (17·1) 0 253 (100%) 0 1835 (486) 67 (26%) 16 (6%) 11 (4%) 20 (8%) 0 3 (1%) 2 (1%) 8 (3%) 28 (11%) 7 (3%) 9 (4%) 8·5% (1·1) 9·5 (2·3) 9·3 (2·1) 4·8 (1·0) 1·2 (0·3) 3·0 (0·9) 2·1 (1·3) 23·8 (12·5) 29·7 (17·0) 70·0 (16·6) 1 (<1%) 480 (100%) 1 (<1%) 1843 (509) 132 (28%) 27 (6%) 21 (4%) 46 (10%) 4 (1%) 8 (2%) 4 (1%) 12 (3%) 52 (11%) 18 (4%) 21 (4%) 8·5% (1·1) 9·3 (2·3) 9·2 (2·1) 4·8 (1·0) 1·2 (0·3) 2·9 (0·9) 2·1 (1·4) 24·1 (11·7) 30·5 (17·2) 69·8 (16·8) 53·9 (8·9) 113 (50%) 83·4 (18·2) 31·1 (4·9) 3·9 (1·9–8·2) 129·8 (13·3) 79·5 (8·7) 75·6 (8·7) 2·5 (1·0–5·3) 53·3 (8·7) 121 (48%) 84·2 (18·3) 31·3 (4·9) 4·8 (1·9–8·2) 131·7 (15·1) 80·0 (8·3) 76·3 (9·3) 3·1 (1·1–6·1) 53·6 (8·8) 234 (49%) 83·8 (18·2) 31·1 (4·9) 4·5 (1·9–8·2) 130·8 (14·3) 79·7 (8·5) 76·0 (9·0) 2·9 (1·0–6·0)

Statistical analysis
SAS version 9.2 was used for all analyses. Data are expressed as mean (SD), median (Q1–Q3), estimates (SE), or differences with 95% CI. Efficacy analysis included all participants randomly assigned to treatment groups who had received at least one dose of study drug and had at least one on-treatment assessment of any primary or secondary efficacy variable, irrespective of compliance with the study protocol and procedures. This group was used to focus on patients who had available

Data are n (%), mean (SD), or median (Q1–Q3). OADs=oral antidiabetes drugs. HbA1c=glycated haemoglobin A1c. FPG=fasting plasma glucose. SMFPG=self-monitored fasting plasma glucose. *All participants randomly assigned to treatment groups who had received at least one dose of study drug and had at least one on-treatment assessment of any primary or secondary efficacy variables.

Table 1: Baseline clinical characteristics of patients included in efficacy analysis*

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

3

Articles

A
10 9 HbA1c (%) 8 7 6 5 0 12 Weeks 24 Endpoint Glargine Sitagliptin

B
100 HbA1c <7% or <6·5% (% of participants) HbA1c <7%, glargine HbA1c <6·5%, glargine HbA1c <7%, sitagliptin HbA1c <6·5%, sitagliptin

80

For the primary endpoint, we undertook ANCOVA with the change from baseline in HbA1c as the dependent variable, treatment group as fixed effect, and baseline HbA1c value as covariate. The corresponding 95% CIs were calculated for the adjusted difference (insulin glargine–sitagliptin) in means. On the assumption of a difference in the HbA1c change of 0·4% in favour of insulin glargine, a SD of 1·3%, a two-tailed α risk of 5%, 446 evaluable people were needed (223 per group) to ensure a statistical power of 90%. ANCOVA was used to describe the change from baseline in HbA1c, fasting plasma glucose, and seven-point plasma glucose profile. For categorical variables, Pearson χ² or Fisher’s exact test were used. The rate of hypoglycaemia per patient-year was analysed with a generalised linear model based on a Poisson, negative binomial, zero-inflated Poisson, or zeroinflated negative binomial distribution. The best model was fitted according to likelihood ratio test and Vuong test. This trial is registered at ClinicalTrials.gov, NCT00751114.

60

Role of the funding source
40

20

0 0 12 Weeks 24 Endpoint

C
12 11 10 SMFPG (mmol/L) 9 8 7 6 5 4 0 6 12 Weeks 18 24 Endpoint Glargine Sitagliptin

The funding source participated in initial discussions about trial design, participated in the respective study steering committees, and undertook the data analysis and preparation of study reports. Company representatives are named as authors and contributed to development of the report as described in the authors’ contributions. The authors had access to all data and participated in the analysis and interpretation of data. The authors vouch for the completeness and veracity of the data and analyses. PA, JC, DRO, EW, M-PD, VP, and VF served on the steering committee and had full access to the data. All authors jointly made the decision to submit for publication.

Results
732 people were initially screened and 515 were randomly assigned to insulin glargine (n=250) or sitagliptin (n=265; figure 2). Most of the 217 people excluded after screening did not meet inclusion criteria; the most common cause was an HbA1c value out of range (n=146, 67%). 13 people randomly assigned to the insulin glargine group (mean HbA1c 8·4% [68 mmol/mol]) and one to the sitagliptin group (HbA1c 7·5% [58 mmol/mol]) were never treated, mostly because of withdrawal of consent, and ten (mean HbA1c 8·6% [70 mmol/mol]) and 11 (HbA1c 8·5% [69 mmol/mol]) people in the respective safety populations had no postbaseline assessment for any efficacy variables and were excluded from the efficacy analysis. The groups had similar baseline characteristics (table 1).
Figure 3: HbA1c (A), HbA1c less than 7% or 6·5% (B), self-monitored fasting plasma glucose (C) and seven-point self-monitored blood glucose profiles during a 24-h period (D) in a 24-week study comparing glargine versus sitagliptin in patients with type 2 diabetes who did not respond to metformin monotherapy HbA1c=glycated haemoglobin A1c. SMFPG=self-monitored fasting plasma glucose. *p<0·0001. †p=0·0012. ‡p=0·0008 versus sitagliptin (endpoint).

D
16 14 Plasma glucose (mmol/L) 12 10 8 6 * 4 2 0 Before After Before After * * * * Glargine, baseline Glargine, endpoint Before After † ‡ Sitagliptin, baseline Sitagliptin, endpoint Bedtime

Breakfast

Lunch

Dinner

4

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

Articles

Insulin glargine (n=237) Patients with ≥1 event All symptomatic hypoglycaemia Symptomatic hypoglycaemia with plasma glucose ≤3·9 mmol/L Nocturnal symptomatic hypoglycaemia Nocturnal symptomatic hypoglycaemia with plasma glucose ≤3·9 mmol/L Severe symptomatic hypoglycaemia Severe nocturnal symptomatic hypoglycaemia Symptomatic hypoglycaemia with plasma glucose ≤3·1 mmol/L Nocturnal symptomatic hypoglycaemia with plasma glucose ≤3·1 mmol/L 108 (46%) 86 (36%) 41 (17%) 36 (15%) 3 (1%) 1 (<1%) 56 (24%) 20 (8%) Events per patient-year 4·21 (0·54) 3·16 (0·47) 0·92 (0·18) 0·76 (0·16) 0·03 (0·02) 0·01 (0·01) 0·90 (0·13) 0·29 (0·08)

Sitagliptin (n=264) Patients with ≥1 event 35 (13%) 28 (11%) 8 (3%) 8 (3%) 1 (<1%) 1 (<1%) 12 (5%) 2 (1%) Events per patient-year 0·50 (0·09) 0·38 (0·08) 0·07 (0·03) 0·07 (0·03) 0·01 (0·01) 0·01 (0·01) 0·11 (0·03) 0·02 (0·01)

Insulin glargine/sitagliptin for events per patient-year Ratio (95% CI) 8·45 (5·55–12·87) 8·24 (5·07–13·40) 12·41 (5·43–28·35) 10·34 (4·46–23·98) 3·40 (0·35–32·72)* 1·13 (0·07–18·14)* 8·42 (4·40–16·11) 17·64 (3·87–80·34) p value <0·0001 <0·0001 <0·0001 <0·0001 0·29* 0·93* <0·0001 0·0002

Data are n (%), estimated event rate (SE), estimated rate ratio (95% CI), or p value. The safety population consisted of all participants randomly assigned to treatment groups and treated. Estimated rate ratios and p values were derived from a binomial negative model with the exception of those denoted by *, which were from a Poisson model.

Table 2: Rate of symptomatic, severe, and nocturnal hypoglycaemia during 24-week treatment with glargine or sitagliptin in the safety population

HbA1c was reduced to a greater extent with insulin glargine than with sitagliptin throughout the study (figure 3), with the reduction at study end greater for patients on insulin glargine (adjusted mean –1·72% [SE 0·06] or –18·8 [0·7] mmol/mol) than for those on sitagliptin (–1·13% [0·06] or –12·4 [0·7] mmol/mol). The adjusted mean difference in HbA1c (insulin glargine– sitagliptin) at study end was –0·59% (95% CI –0·77 to –0·42) or –6·4 mmol/mol (–8·4 to –4·6) in favour of insulin glargine (p<0·0001). Throughout the study, more participants on insulin glargine than on sitagliptin achieved HbA1c less than 7% or less than 6·5% (figure 3). At study end, 152 (68%) of 224 participants on insulin glargine had HbA1c less than 7% compared with 104 (42%) of 248 on sitagliptin (p<0·0001); 90 (40%) on insulin glargine had HbA1c less than 6·5% compared with 42 (17%) on sitagliptin (p<0·0001). Participants in the insulin glargine group had greater reduction in selfmonitored fasting plasma glucose and seven-point plasma glucose profile than did those in the sitagliptin group (figure 3). The adjusted mean difference in selfmonitored fasting plasma glucose (insulin glargine– sitagliptin) was –2·3 mmol/L (95% CI –2·6 to –2·0) lower with insulin glargine than with sitagliptin (p<0·0001). The adjusted mean difference (insulin glargine–sitagliptin) also favoured insulin glargine at each timepoint of the seven-point plasma glucose profile (after dinner, p=0·0012; at bedtime, p=0·0008; all others, p<0·0001). Changes in vital signs and lipid profiles were similar for the two treatments (appendix p 2). Dose of insulin glargine increased throughout the study, mainly during the first 12 weeks. At baseline, the mean daily dose was 0·19 (SD 0·3) units per kg (15·8 [4·2] units), which increased to 0·27 (0·08) units per kg at 2 weeks, 0·38 (0·16) units per kg at 6 weeks, 0·45 (0·2) units per kg at 12 weeks, 0·48 (0·23) units per kg at 16 weeks, and

0·5 (0·26) units per kg at 24 weeks. At study endpoint, the dose was 0·49 (SD 0·36) units per kg or 41·4 (25·8) units daily. Bodyweight increased in the insulin glargine group and decreased in the sitagliptin group. The adjusted mean change in bodyweight from baseline to endpoint was 0·44 (SE 0·22) kg in the insulin glargine group and –1·08 (0·2) kg in the sitagliptin group with an adjusted mean difference of 1·51 kg (95% CI 0·93–2·09; p<0·0001). More participants in the insulin glargine group had symptomatic hypoglycaemia than in the sitagliptin group; severe symptomatic and severe nocturnal hypoglycaemia were rare events in either group (table 2). Treatment-emergent adverse events were reported by 108 (46%) of 237 participants in the insulin glargine and 143 (54%) of 264 participants in the sitagliptin group. The most frequently reported treatment-emergent events (reported in >3% of participants in at least one of the groups) were influenza (eight [3%] patients on insulin glargine and 15 [6%] on sitagliptin), nasopharyngitis (eight [3%] and 15 [6%]), headache (15 [6%] and 14 [5%]), dizziness (eight [3%] and eight [3%]), diarrhoea (five [2%] and ten [4%]), and nausea (four [2%] and 12 [5%]). Serious treatment-emergent adverse events were reported by 15 (6%) patients on insulin glargine and eight (3%) on sitagliptin (table 3). Most types of event were reported by only one individual, except for hypoglycaemia (three patients, including one with unconsciousness) and unstable angina (two events in the insulin glargine group). Two (1%) of 237 people in the insulin glargine group and four (2%) of 264 in the sitagliptin group withdrew from the study because of a treatmentemergent adverse event. Withdrawal in the insulin glargine group was due to malaise, injection-site inflammation, and nausea, whereas cellulitis, acute myocardial infarction, overdose, and pregnancy were causes of withdrawal in the sitagliptin group.
5

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

Articles

Insulin glargine (n=237) Any class Infections and infestations Anal abscess Cellulitis General disorders and administration site Non-cardiac chest pain Nervous system disorders Carotid artery occlusion Epilepsy Loss of consciousness Nerve compression Muscular and connective tissue disorders Pain in leg Gastrointestinal disorders Impaired gastric emptying Diverticulum intestinal Vascular disorders Hypertension Orthostatic hypotension Peripheral arterial occlusive disease Metabolism and nutrition disorders Hypoglycaemia Hypoglycaemic unconsciousness Cardiac disorders Angina unstable Angina pectoris Acute myocardial infarction Injury, poisoning, and procedural complications Vascular pseudoaneurysm Renal and urinary disorders Calculus ureteric Renal colic Blood and lymphatic system disorders Haemorrhagic anaemia Neoplasms (benign, malignant, and unspecified, including cysts and polyps) Kaposi’s sarcoma Prostate cancer 15 (6%) 1 (<1%) 1 (<1%) 0 1 (<1%) 1 (<1%) 3 (1%) 1 (<1%) 1 (<1%) 1 (<1%) 1 (<1%) 0 0 1 (<1%) 1 (<1%) 0 2 (1%) 0 1 (<1%) 1 (<1%) 3 (1%) 2 (1%) 1 (<1%) 3 (1%) 2 (1%) 1 (<1%) 0 1 (<1%) 1 (<1%) 2 (1%) 1 (<1%) 1 (<1%) 1 (<1%) 1 (<1%) 0 0 0

Sitagliptin (n=264) 8 (3%) 1 (<1%) 0 1 (<1%) 0 0 0 0 0 0 0 1 (<1%) 1 (<1%) 1 (<1%) 0 1 (<1%) 1 (<1%) 1 (<1%) 0 0 0 0 0 2 (1%) 0 1 (<1%) 1 (<1%) 0 0 0 0 0 0 0 2 (1%) 1 (<1%) 1 (<1%)

Data are number (%) of patients with at least one adverse event. The safety population consisted of all participants randomly assigned to treatment groups and treated. MedDRA version 14.0 was used for assessment.

Table 3: Serious treatment-emergent adverse events in the safety population

Discussion
In this international, multicentre trial, 24-week treatment with either insulin glargine or sitagliptin was well tolerated with low discontinuation rates in individuals with type 2 diabetes inadequately controlled on metformin (panel). Treatment with insulin glargine lowered HbA1c by 0·59% more than did sitagliptin, and was 1·6-times more likely than was sitagliptin to achieve
6

HbA1c less than 7% and 2·5-times more likely to achieve HbA1c less than 6·5% and with concurrently lower fasting and postprandial plasma glucose concentrations. Most guidelines agree on a glycaemic target of HbA1c 7% or less, although some now recommend a tighter glycaemic goal of 6·5% in specific groups of patients with a low cardiovascular risk in view of the adverse results of the ACCORD study in high-risk individuals.4–6 Recent metaanalyses, however, including the ACCORD trial,2,12,13 showed that more intensive control reduces non-fatal coronary heart disease events and has a neutral effect on mortality. Additionally, the UK Prospective Diabetes Study14,15 showed that a low HbA1c reduced the risk of microvascular complications. Data are insufficient to prove or refute a relative risk reduction for cardiovascular mortality, non-fatal myocardial infarction, composite microvascular complications, or retinopathy with intensive glycaemic control.16 The prevalence of diabetesrelated complications in our study was similar to that in other similar trials or even in trials recruiting recently diagnosed patients. In addition to a greater likelihood of reaching HbA1c less than 7% at study end, significantly more participants on insulin glargine than on sitagliptin achieved the goal by week 12. By week 24, there was a further increase in the number of participants in the insulin glargine group who reached the goal, whereas the number in the sitagliptin group increased only slightly. Although the duration of the study was fairly short, the results show that if people learn to titrate basal insulin and do not have episodes of severe hypoglycaemia, more of them might continue over time to reach their glycaemic goal with the combination of insulin glargine plus metformin. The results of this study are in general agreement with previous results for both insulin glargine and sitagliptin. In previous clinical trials, insulin glargine added to metformin treatment reduced HbA1c by 1·7% or more,9–11 an effect size similar to our findings. On the other hand, sitagliptin added to metformin reduced HbA1c by an average of 0·7% in earlier trials,7,8,17 but in subgroups of participants with baseline HbA1c of 8–9% the reduction in HbA1c was nearly identical to the present study.8 Weight loss with sitagliptin was greater than expected since DPP-4 inhibitors are usually weight-neutral,17 but a similar weight loss has been reported in patients treated with the combination of sitagliptin and metformin.18 The prevalence of diabetes-related complications in our study was similar to that in other similar trials and in trials recruiting recently diagnosed patients. During the seven-point plasma glucose monitoring, insulin glargine was more effective than was sitagliptin in reducing both fasting and postprandial plasma glucose. This finding could be accounted for by the effect of basal insulin on reduction of hepatic glucose production.19 Amelioration of glucotoxicity that is often accompanied by improved endogenous insulin secretion might also play a part, but was not addressed in this

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

Articles

study. Intensive insulin treatment has been shown to improve insulin secretion.20–22 The estimated rate of hypoglycaemia per patient-year was eight-times higher with insulin glargine than with sitagliptin. However, not all episodes were confirmed by blood glucose testing and 13% of patients treated with sitagliptin (known to have low risk [<3%] of hypoglycaemia17) also reported hypoglycaemia. The risk of severe hypoglycaemia, although three-times higher with insulin glargine, was not significantly different from that with sitagliptin, and the risk of severe nocturnal hypoglycaemia was the same for the two groups. Lessons from recent megatrials have shown that intensive lowering of blood glucose should be accompanied by structured self-monitoring of blood glucose, especially in high-risk people with long disease duration, in an attempt to avoid hypoglycaemia-related adverse effects such as cardiovascular disease and related death.23 In view of the possible long-term benefits of intensive lowering of blood glucose and the superior efficacy of insulin to optimise glycaemic control, strong arguments could be made to use insulin early when the dose is expected to be fairly low, with a reduced risk of hypoglycaemia.24 As shown in the present study, glycaemic control can be achieved with an average supplementary dose of 0·5 units per kg of insulin glargine and a weight gain of less than 0·5 kg. Upcoming results from the ORIGIN (Outcome Reduction with an Initial Glargine Intervention) trial25 will further address this question and other aspects related to the benefit of early intensive glucose lowering with basal insulin. Despite the large sample size and multinational nature of the study population, the present study is not without limitations. We recognise that the open-label nature of the study and the absence of placebo groups preclude determination of absolute HbA1c changes. We also acknowledge that initiation of insulin treatment is inevitably accompanied by increased frequency of monitoring and contacts with care teams, which might contribute to metabolic improvement. However, many randomised clinical trials have shown the benefits of early attainment of goals with intensive insulin treatment.20 Our objective was to establish the relative effect of basal insulin glargine versus the DPP-4 inhibitor sitagliptin under conditions that were as close to real life as can be simulated in a randomised clinical trial. The results extend the earlier findings to show the feasibility of application of this knowledge in day-to-day practice. Although the increased reduction in HbA1c in the insulin glargine group might be partly attributable to the increased frequency of blood-glucose monitoring needed for insulin titration and thus a different level of interaction with the carers, otherwise, both groups had a similar number of clinic visits, telephone reminders, and seven-point plasma glucose monitoring. We also recognise the short-term nature of the study and realise that long-term studies will be needed to reaffirm our

Panel: Research in context Systematic review We searched PubMed Clinical Queries up to Feb 28, 2012, with the search terms “type 2 diabetes”, “insulin therapy”, and “dipeptidyl peptidase-4 inhibitor”. We reviewed randomised clinical trials and meta-analyses published in English. Both treatments effectively improve glycaemic control with clinically acceptable safety. However, we were unable to find a comparative effectiveness trial comparing the two strategies head to head. Interpretation This randomised trial is the first to compare two therapeutic approaches often considered as second-line add-on treatment to metformin, under conditions that are as close to real life as is possible in a randomised trial. In this study, insulin glargine when added to metformin reduced glycated haemoglobin A1c more than did sitagliptin added to metformin, with an effect size similar to previous clinical trials.9–11 Hypoglycaemia overall, however, was less frequent with sitagliptin. There was a small weight loss with sitagliptin, by contrast with minimum weight gain with insulin glargine. The results of this comparative effectiveness trial might help physicians to choose between these two drugs for patients whose diabetes is uncontrolled on metformin and provide clinical experience to guide the design of future studies needed to assess the long-term efficacy of these two therapeutic strategies.

findings and to provide greater understanding about the differences between these two therapeutic options. Acceptance of insulin treatment is another consideration despite the likely benefits. In conclusion, we have shown in this 24-week trial the feasibility of initiating basal insulin therapy early in individuals who have not achieved the desired glycaemic target with metformin monotherapy. Early introduction of basal insulin glargine was associated with lower HbA1c and fasting and postprandial blood glucose concentrations with a higher rate of attainment of HbA1c goals compared with sitagliptin. Minimum weight gain was noted with insulin glargine and a small weight loss with sitagliptin. The results of this study support the option of introduction of basal insulin in patients with type 2 diabetes inadequately controlled by metformin, with the potential for long-term benefits arising from the achievement of optimum glycaemic control early in the course of the disease.
Contributors PA, JC, DRO, EW, M-PD, VP, and VF served on the steering committee. SP served on the international titration committee. VP undertook statistical analyses for the report. All authors contributed to the interpretation of the data and provided comments on the report at various stages in its development. EASIE Study Group Austria T Hermann, A Luger, G Schernthaner, G Fließer-Görzer, B Paulweber, S Pusnarig, J Föchterle, K Elcic-Mihaljevic, B Hölzl; Brazil J Gross, F Eliaschewitz, M Hissa, R Réa, R Kupfer; Colombia A Almanzar, P Aschner, A Orduz, E Maria; Greece V Klisiaris, N Tentolouris, A Melidonis, S Pappas, G Dimitriadis; Hong Kong R Osaki, C C Tsang, S C Siu, M W Tsang, V Yeung; India S Pendsey, S Trivedi, S Murthy; Israel T Zornitzki, J Wainstein, O Shemuel, I Harman Boehm; Lebanon S Azar, C Saab, A Echtay; Mexico L-G Mancillas-Adame, C-A Dominguez-Reyes, A Romero-Zazueta, N-A Caracas-Portilla, F de M C Mercado-Palacios; Netherlands J van de Walle, M van Vollevelde, S Jenniskens, B Sombekke, W van Kempen, L de Schipper, W Everts, A Kooy, I Agous; Portugal L Guerra, J Guimaraes, R Silva; South Korea

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

7

Articles

D Kim, W Lee, S Baik, H Jang, M Lee, K Won; Spain C Zapon Llópis, D Bellido Guerrero, M Rivas Fernández, F Botella Romero, F Tinahones Madueño, A Robles Iniesta, J A Paniagua González, L de Teresa Parreño, L Flores Meneses; Turkey H Sargin, B Efe, S Gonen, I Sahin, R Sari, S Shelbaya; UK D R Owens, J Litchfield, A Ellery, N Armstrong; USA M Wong, D Robbins, N Patel, L Fogelfeld, B Mirada-Palma, J McGill, L Chaykin, S Garg, C Lovell, R Lorraine, P Snell, S Rizvi, H Maheshwari, S Nakhle, S Ong, J Mitchell, K Roberts. Conflicts of interest PA has served on advisory boards for AstraZeneca, Eli Lilly & Co, GlaxoSmithKline, Janssen, Merck, Sharpe & Dohme, Novartis, and Sanofi and on speakers’ bureaus for AstraZeneca, Eli Lilly & Co, Merck, Sharpe & Dohme, Novartis, and Sanofi. JC has served on advisory boards for Amylin, AstraZeneca, Bayer Healthcare, Eli Lilly & Co, GlaxoSmithKline, Merck-Serono, Merck, Sharpe & Dohme, Pfizer, and Sanofi and on speakers’ bureaus for AstraZeneca, Bayer Healthcare, Eli Lilly & Co, GlaxoSmithKline, Merck-Serono, Merck, Sharpe & Dohme, Pfizer, Sanofi, and Takeda; she has received research support from Amylin, AstraZeneca, Bayer Healthcare, Eli Lilly & Co, GlaxoSmithKline, Merck-Serono, Merck, Sharpe & Dohme, Sanofi, and Takeda. DRO has served on advisory boards and speakers’ bureaus for Roche and Sanofi and has received research support from Boehringer Ingelheim, Roche, and Sanofi. SP has served on advisory boards, as a board member, and as a consultant for Medtronics, Novo Nordisk, and Sanofi and on speakers’ bureaus for Eli Lilly & Co, Lifescan, Medtronic, Merck-Serono, Novartis, Pierre Fabre, Novo Nordisk, Sanofi, and Solvay. EW, M-PD, and VP are employees of Sanofi. AE has served on an advisory board for Merck, Sharpe & Dohme and on speakers’ bureaus for AstraZeneca, Eli Lilly & Co, Merck & Co, Merck, Sharpe & Dohme, Novartis, Novo Nordisk, and Sanofi. VF has served as a consultant and on speakers’ bureaus for AstraZeneca, Daiichi Sankyo, Eli Lilly & Co, GlaxoSmithKline, Novo Nordisk, Pamlabs, Sanofi, Takeda, and Xoma. Acknowledgments The study was funded by Sanofi. Editorial support was provided by Tom Claus of PPSI (a PAREXEL Company; Hackensack, NJ, USA) and funded by Sanofi. References 1 Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med 2008; 359: 1577–89. 2 Ray KK, Seshasai SRK, Wijesuriya S, et al. Effect of intensive control of glucose on cardiovascular outcomes and death in patients with diabetes mellitus: a meta-analysis of randomised controlled trials. Lancet 2009; 373: 1765–72. 3 Turnbull FM, Abraira C, Anderson RJ, et al. Intensive glucose control and macrovascular outcomes in type 2 diabetes. Diabetologia 2009; 52: 2288–98. 4 IDF Clinical Guidelines Task Force. Guideline for management of postmeal glucose. Brussels: International Diabetes Federation, 2007. http://www.idf.org/webdata/docs/Guideline_PMG_final.pdf (accessed Jan 10, 2012). 5 Nathan DM, Buse JB, Davidson MB, et al. Medical management of hyperglycaemia in type 2 diabetes mellitus: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia 2009; 52: 17–30. 6 Rodbard HW, Blonde L, Braithwaite SS, et al. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the management of diabetes mellitus. Endocr Pract 2007; 13 (suppl 1): 1–68. 7 Charbonnel B, Karasik A, Liu J, Wu M, Meininger G. Efficacy and safety of the dipeptidyl peptidase-4 inhibitor sitagliptin added to ongoing metformin therapy in patients with type 2 diabetes inadequately controlled with metformin alone. Diabetes Care 2006; 29: 2638–43. 8 Nauck MA, Meininger G, Sheng D, Terranella L, Stein PP. Efficacy and safety of the dipeptidyl peptidase-4 inhibitor, sitagliptin, compared with the sulfonylurea, glipizide, in patients with type 2 diabetes inadequately controlled on metformin alone: a randomized, double-blind, non-inferiority trial. Diabetes Obes Metab 2007; 9: 194–205.

9

10

11

12

13

14

15

16

17 18

19

20

21

22

23 24

25

Riddle MC, Rosenstock J, Gerich J. The treat-to-target trial: randomized addition of glargine or human NPH insulin to oral therapy of type 2 diabetic patients. Diabetes Care 2003; 26: 3080–86. Rosenstock J, Davies M, Home PD, Larsen J, Koenen C, Schernthaner G. A randomised, 52-week, treat-to-target trial comparing insulin detemir with insulin glargine when administered as add-on to glucose-lowering drugs in insulin-naive people with type 2 diabetes. Diabetologia 2008; 51: 408–16. Fonseca V, Gill J, Zhou R, Leahy J. An analysis of early insulin glargine added to metformin with or without sulfonylurea: impact on glycaemic control and hypoglycaemia. Diabetes Obes Metab 2011; 13: 814–22. Boussageon R, Bejan-Angoulvant T, Saadatian-Elahi M, et al. Effect of intensive glucose lowering treatment on all cause mortality, cardiovascular death, and microvascular events in type 2 diabetes: meta-analysis of randomised controlled trials. BMJ 2011; 343: d4169. Mannucci E, Monami M, Lamanna C, Gori F, Marchionni N. Prevention of cardiovascular disease through glycemic control in type 2 diabetes: a meta-analysis of randomized clinical trials. Nutr Metab Cardiovasc Dis 2009; 19: 604–12. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998; 352: 837–53. UK Prospective Diabetes Study (UKPDS) Group. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). Lancet 1998; 352: 854–65. Hemmingsen B, Lund SS, Gluud C, et al. Intensive glycaemic control for patients with type 2 diabetes: systematic review with meta-analysis and trial sequential analysis of randomised clinical trials. BMJ 2011; 343: d6898. Ahren B. Use of DPP-4 inhibitors in type 2 diabetes: focus on sitagliptin. Diabetes Metab Syndr Obes 2010; 3: 31–41. Wainstein J, Katz L, Engel SS, et al. Initial therapy with the fixed-dose combination of sitagliptin and metformin results in greater improvement in glycaemic control compared with pioglitazone monotherapy in patients with type 2 diabetes. Diabetes Obes Metab 2011; 14: 409–18. Triplitt C, Glass L, Miyazaki Y, et al. Comparison of glargine insulin versus rosiglitazone addition in poorly controlled type 2 diabetic patients on metformin plus sulfonylurea. Diabetes Care 2006; 29: 2371–77. Weng J, Li Y, Xu W, et al. Effect of intensive insulin therapy on β-cell function and glycaemic control in patients with newly diagnosed type 2 diabetes: a multicentre randomised parallel-group trial. Lancet 2008; 371: 1753–60. Noh YH, Lee SM, Kim EJ, et al. Improvement of cardiovascular risk factors in patients with type 2 diabetes after long-term continuous subcutaneous insulin infusion. Diabetes Metab Res Rev 2008; 24: 384–91. Chen HS, Wu TE, Jap TS, Hsiao LC, Lee SH, Lin HD. Beneficial effects of insulin on glycemic control and beta-cell function in newly diagnosed type 2 diabetes with severe hyperglycemia after short-term intensive insulin therapy. Diabetes Care 2008; 31: 1927–32. Desouza CV, Bolli GB, Fonseca V. Hypoglycemia, diabetes, and cardiovascular events. Diabetes Care 2010; 33: 1389–94. Pozzilli P, Leslie RD, Chan J, et al. The A1C and ABCD of glycaemia management in type 2 diabetes: a physician’s personalized approach. Diabetes Metab Res Rev 2010; 26: 239–44. Gerstein H, Yusuf S, Riddle MC, Ryden L, Bosch J. Rationale, design, and baseline characteristics for a large international trial of cardiovascular disease prevention in people with dysglycemia: the ORIGIN Trial (Outcome Reduction with an Initial Glargine Intervention). Am Heart J 2008; 155: 26–32.

8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60439-5

Articles

Exenatide twice daily versus glimepiride for prevention of glycaemic deterioration in patients with type 2 diabetes with metformin failure (EUREXA): an open-label, randomised controlled trial
Baptist Gallwitz, Juan Guzman, Francesco Dotta, Bruno Guerci, Rafael Simó, Bruce R Basson, Andreas Festa, Jacek Kiljański, Hélène Sapin, Michael Trautmann, Guntram Schernthaner

Summary
Background Glycaemic control deteriorates progressively over time in patients with type 2 diabetes. Options for treatment escalation remain controversial after failure of first-line treatment with metformin. We compared add-on exenatide with glimepiride for durability of glycaemic control in patients with type 2 diabetes inadequately controlled by metformin alone. Methods We did an open-label, randomised controlled trial at 128 centres in 14 countries between Sept 5, 2006, and March 29, 2011. Patients aged 18–85 years with type 2 diabetes inadequately treated by metformin were randomly assigned via a computer-generated randomisation sequence to receive exenatide twice daily or glimepiride once daily as add-on to metformin. Randomisation was stratified by predetermined categories of glycated haemoglobin (HbA1C) concentration. The primary outcome was time to inadequate glycaemic control and need for alternative treatment, defined as an HbA1c concentration of more than 9% after the first 3 months of treatment, or more than 7% at two consecutive visits after the first 6 months. Analysis was by intention to treat. This trial is registered with EudraCT, number 2005-005448-21, and ClinicalTrials.gov, number NCT00359762. Findings We randomly assigned 515 patients to the exenatide group and 514 to the glimepiride group, of whom 490 versus 487 were the intention-to-treat population. 203 (41%) patients had treatment failure in the exenatide group compared with 262 (54%) in the glimepiride group (risk difference 12·4 [95% CI 6·2–18·6], hazard ratio 0·748 [0·623–0·899]; p=0·002). 218 (44%) of 490 patients in the exenatide group, and 150 (31%) of 487 in the glimepiride group achieved an HbA1c concentration of less than 7% (p<0·0001), and 140 (29%) versus 87 (18%) achieved concentrations of 6·5% and less (p=0·0001). We noted a significantly greater decrease in bodyweight in patients given exenatide than in those given glimepiride (p<0·0001). Five patients in each treatment group died from causes unrelated to treatment. Significantly fewer patients in the exenatide group than in the glimepiride group reported documented symptomatic (p<0·0001), nocturnal (p=0·007), and non-nocturnal (p<0·0001) hypoglycaemia. Discontinuation because of adverse events (mainly gastrointestinal) was significantly higher (p=0·0005) in the exenatide group than in the glimepiride group in the first 6 months of treatment, but not thereafter. Interpretation These findings provide evidence for the benefits of exenatide versus glimepiride for control of glycaemic deterioration in patients with type-2 diabetes inadequately controlled by metformin alone. Funding Eli Lilly and Company; Amylin Pharmaceuticals.
Published Online June 9, 2012 DOI:10.1016/S01406736(12)60479-6 See Online/Comment DOI:10.1016/S01406736(12)60769-7 Department of Medicine IV, Eberhard-Karls-University Tübingen, Tübingen, Germany, (Prof B Gallwitz MD); Celaya Centre for Specialist Medicine, Guanajuato, Mexico (J Guzman MD); Department of Internal Medicine, Endocrine and Metabolic Sciences, Policlinico Le Scotte, Sienna, Italy (F Dotta MD); Hospital Brabois and Centres d’Investigation Clinique Inserm, CHU de Nancy, Vandoeuvre-Lès-Nancy, France (B Guerci MD); Vall d’Hebron Research Institute and Centro de Investigación Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas, Carlos III Health Institute, Barcelona, Spain (R Simó MD); Eli Lilly and Company, St Cyr au Mont D’Or, France (B R Basson MSc); Eli Lilly and Company, Vienna, Austria (A Festa MD); Eli Lilly Polska, Warsaw, Poland (J Kiljański MD); Eli Lilly and Company, Suresnes, France (H Sapin MSc); Lilly Deutschland, Hamburg, Germany (M Trautmann MD); and Department of Medicine I, Rudolfstiftung Hospital, Vienna, Austria (Prof G Schernthaner MD) Correspondence to: Prof Baptist Gallwitz, Medizinische Klink IV, Universitätklinikum Tübingen, 72076 Tübingen, Germany [email protected]

Introduction
Metformin is widely used as a first-line glucose-lowering drug;1,2 however, selection of the most appropriate treatment after metformin failure is poorly established. Sulphonylureas are commonly chosen as add-on treatment because of their rapid effect and low cost.1,3,4 Although these drugs can improve the short-term function of β cells, glycaemic control subsequently deteriorates; furthermore, because effects are not glucosedependent, risk of hypoglycaemia might be increased, which can restrict doses used in clinical practice.5 Glucagon-like peptide (GLP)-1 receptor agonists have become established as treatments for type 2 diabetes.6,7

They improve glycaemic control, with glucosedependent stimulation of insulin secretion and no increased risk of hypoglycaemia, and have been associated with weight loss and improvements in biomarkers of cardiovascular risk.8-11 These drugs have shown protective action in β cells, and findings from clinical trials have noted improved β-cell function,12–14 thus raising expectations that GLP-1 receptor agonists might delay disease progression.15,16 We aimed to assess durability of glycaemic control achieved with GLP-1 receptor agonist exenatide twice a day and sulphonylurea glimepiride in patients with type 2 diabetes inadequately controlled by metformin alone.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

1

Articles

Methods
1404 patients screened 375 excluded 13 for exclusion of 1 site 1 adverse event 312 did not meet entry criteria 45 patient decision 3 physician decision 1 lost to follow-up 1029 randomised

Study design and participants
We undertook this open-label, randomised controlled European Exenatide (EUREXA) trial at 128 centres in 14 countries (Austria, Czech Republic, Finland, France, Germany, Hungary, Ireland, Israel, Italy, Mexico, Poland, Spain, Switzerland, and the UK) between Sept 5, 2006, and March 29, 2011. The rationale and baseline characteristics of the EUREXA phase 3, multinational study have been described previously.17 Eligible participants had type 2 diabetes; were overweight to obese (body-mass index [BMI] ≥25 kg/m² to <40 kg/m²); aged 18–85 years; had been on stable, maximum tolerated doses of metformin; and had developed suboptimum glycaemic control, defined by a glycated haemoglobin (HbA1c) concentration of 6·5% and more or 9·0% and less. Exclusion criteria were contraindications for metformin or glimepiride, according to the product-specific label; active or untreated malignancy or remission for less than 5 years; evidence of renal or liver disease or dysfunction; haemoglobinopathy or clinically significant chronic anaemia; active proliferative retinopathy or macular oedema; or severe gastrointestinal disease. Excluded drugs were those affecting gastrointestinal motility, chronic systemic glucocorticoids, prescription drugs to promote weight loss in the past 3 months, and treatment for more than 2 weeks in the past 3 months with insulin, thiazolidinediones, α-glucosidase inhibitors, sulphonylureas, or meglitinides. The study protocol was approved by appropriate institutional review boards, in accordance with country-specific regulations. We did the study in compliance with Good Clinical Practice and the Declaration of Helsinki, and obtained signed informed consent from all patients.

515 allocated to exenatide

514 allocated to glimepiride

203 met primary endpoint

262 met primary endpoint

174 discontinued 49 had an adverse event* 4 patients died 4 did not meet entry criteria 8 lack of efficacy 5 lost to follow-up 70 patient decision 23 physician decision 11 protocol violation

128 discontinued 17 adverse event* 2 patients died 8 did not meet entry criteria 11 lack of efficacy 5 lost to follow-up 50 patient decision 17 physician decision 18 protocol violation

138 completed study with no treatment failure

124 completed study with no treatment failure

Figure 1: Trial profile The intention-to-treat population consisted of 490 patients randomised to exenatide (five did not receive the study drug and 20 did not have at least one baseline or post-baseline HbA1c measurement) and 487 randomised to glimepiride (six did not receive the study drug and 21 did not have at least one baseline or post-baseline HbA1c measurement). *p=0·001 for difference between groups.

Exenatide (n=490) Age (years) Age ≥65 years Sex Male Female Race White Hispanic African or Asian Bodyweight (kg) Body-mass index (kg/m²) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Heart rate (beats per min) Diabetes duration (years) Metformin dose (mg per day) HbA1c concentration (%) Fasting plasma glucose (mmol/L) Taking antihypertensive drugs 450 (92%) 36 (7%) 4 (<1%) 92·8 (16·7) 32·6 (4·2) 132·8 (15·7) 80·4 (9·4) 74·1 (9·3) 5·8 (4·8) 1956 (596) 7·5% (0·7) 8·9 (2·3) 340 (69%) 272 (56%) 218 (44%) 56 (10·0) 102 (21%)

Glimepiride (n=487) 56 (9·1) 98 (20%)

Randomisation and masking
252 (52%) 235 (48%) 444 (91%) 35 (7%) 8 (2%) 91·1 (14·8) 32·3 (3·9) 133·4 (15·1) 79·8 (9·9) 74·0 (10·1) 5·5 (4·3) 1989 (634) 7·4% (0·7) 8·6 (1·9) 367 (75%)

We used a computer-generated randomisation sequence to randomly assign patients, in a 1:1 ratio, to receive either exenatide or glimepiride. Randomisation was stratified by HbA1c categories of 7·3% and less, more than 7·3% to 8·2% and less, and more than 8·2%. Before database lock the study team were masked to group assignment and statistical anlyses were planned with no knowledge of groups.

Procedures
Exenatide was injected subcutaneously within 60 min before breakfast and evening meals, starting at 5 μg twice daily for 4 weeks, followed by 10 μg twice daily for the remaining study period. If patients had daily episodes of nausea for more than 1 week, the 10 μg dose was reduced to 5 μg twice daily and could be increased again after nausea subsided. The recommended starting dose for patients in the glimepiride group was 1 mg per day, given once daily immediately before breakfast. Attending physicians established the glimepiride dose as per their

Data are mean (SD) or n (%). Data are for the intention-to-treat population. HbA1c=glycated haemaglobin A1c.

Table 1: Baseline demographic and clinical characteristics

2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

Articles

Proportion of patients not meeting criteria for treatment failure

usual practice, and investigators were instructed to adjust the dose every 4 weeks, according to tolerability, up to the maximum tolerated dose in accordance with the countryspecific summary of product characteristics. Concomitant metformin was continued throughout the study for all patients, in the same form and at the same dose as used at study entry.

A
1·0 0·9 0·8 0·7 0·6 0·5 0·4 0·3 0·2 0·1 0 0 Number at risk Exenatide 490 Glimepiride 487 6 437 466 12 310 336 18 24 220 236 30 36 168 170 42 48 32 41 Exenatide twice daily Glimepiride

Study outcomes
The primary outcome was time to inadequate glycaemic control, defined as an HbA1c concentration of more than 9% after the first 3 months of treatment, or more than 7% at two consecutive visits 3 months apart after the first 6 months. We defined treatment failure in line with recommendations of diabetes associations and the known timecourse of changes in HbA1c concentration, and allowed quick identification of patients with poor glycaemic control who needed alternative treatment.1 Because the primary outcome was a time-to-event measure, we regarded a study period of 2–3 years as appropriate. Patients who had treatment failure were discontinued, but could enrol in an extension study to examine further treatment options; findings from this study will be described elsewhere. Secondary outcomes were markers of β-cell function, bodyweight, hypoglycaemia, and surrogate markers of cardiovascular risk (blood pressure and heart rate). Laboratory measurements were done at a central laboratory (Interlab GmbH, Munich, Germany). Plasma glucose was measured with an automated hexokinase method (Cobas Gluco-quant, Roche Diagnostics GmbH, Mannheim, Germany), HbA1c with automated highperformance liquid chromatography (Tosoh Bioscience Inc, San Francisco, CA, USA), and insulin with a two-site chemiluminescent immunometric assay (Immulite 2000, Siemens Diagnostics, Tarrytown, NY, USA). All patients underwent oral glucose-tolerance tests, starting in a fasted state and before the morning doses of metformin and study drug. Homoeostatic model assessment (HOMA)-B and HOMA-IR were calculated with standard formulas and programs. We established insulinogenic index from changes in glucose and insulin at 30 min and adjusted the index for HOMA-IR for the disposition index.15 Additionally, patients self-monitored blood glucose before and 2 h after meals for 2 consecutive days before study visits. We classified hypoglycaemic episodes as recommended by the American Diabetes Association Workgroup on Hypoglycemia.18 Blood pressure and heart rate were measured at all study visits. We recorded and classified adverse events according to the Medical Dictionary for Regulatory Activities.

B
1·0 0·9 Proportion of patients not meeting criteria for treatment failure 0·8 0·7 0·6 0·5 0·4 0·3 0·2 0·1 0 0 Number at risk HbA1c ≤7·3% Exenatide 235 Glimepiride 257 HbA1c >7·3%–≤8·2% Exenatide 189 Glimepiride 166 HbA1c >8·2% Exenatide 66 Glimepiride 64 6 12 18 24 30 Time (months) 130 163 75 63 15 10

HbA1c ≤7·3% glimepiride HbA1c >7·3%–≤8·2% glimepiride HbA1c >8·2% glimepiride HbA1c ≤7·3% exenatide HbA1c >7·3%–≤8·2% exenatide HbA1c >8·2% exenatide

36

42

48

204 244 177 161 56 61

170 212 111 103 29 21

98 130 61 36 9 4

18 33 9 6 5 2

Figure 2: Time-to-event curves for patients meeting criteria for treatment failure (A) and for those meeting treatment failure criteria according to baseline HbA1c categories (B) Event rates in figure 2A are Kaplan-Meier estimates. The sharp drop at 9 months corresponds to when patients could first meet the criterion of HbA1c concentration >7·0% at two consecutive visits after the first 6 months of treatment. HbA1c=glycated haemaglobin A1c.

Statistical analysis
We declared non-inferiority of exenatide to glimepiride if the 97·5% CI for the hazard ratio (HR), with a Cox proportional hazards model with baseline HbA1c as

covariate, excluded 1·25, thus rejecting the hypothesis that risk of treatment failure with exenatide was more than 25% greater than that with glimepiride. If non-inferiority was shown, we tested superiority with 95% CI (excluding 1).19 Kaplan-Meier curves were calculated for patients with inadequate HbA1c criteria. We calculated sample size on the basis of the non-inferiority test of exenatide versus glimepiride, an expected mean baseline HbA1c concentration of 8·2%, a 1 year patient accrual, maximum follow-up of 3 years, drop-out rate of 15% per year (for reasons other than treatment failure), and a 58%

For the HOMA calculator see http://www.dtu.ox.ac.uk/ homacalculator/index.php

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

3

Articles

7·6 7·5 7·4 HbA1c concentration (%) 7·3 7·2 7·1 7·0 6·9 6·8 6·7 6·6 0 0 Number at risk Exenatide 486 Glimepiride 480

Exenatide twice daily Glimepiride p=0·035 p=0·008

number 2005-005448-21, and ClinicalTrials.gov, number NCT00359762.

Role of the funding source
The sponsor took part in study design, data collection, data analysis, data interpretation, and writing of the report. All authors had full access to the data and responsibility for the content of the report. BGa and GS had final responsibility for the decision to submit for publication.

p=0·141

Results
Figure 1 shows the trial profile. We randomly assigned 1029 of 1404 screened patients to receive either exenatide or glimepiride as add-on treatment to metformin. The intention-to-treat population consisted of 490 patients in the exenatide group and 487 in the glimepiride group; conclusions from the as-treated population were not different from those from the intention-to-treat analysis and are therefore not presented. Of patients who met the primary endpoint, five of those in the exenatide group and seven in the glimepiride group had an HbA1c concentration of more than 9%, and 198 versus 255 had concentrations of more than 7% at two visits. One patient in the glimepiride group had an HbA1c concentration of more than 9% after the first 6 months of treatment and 113 and 164 patients in the exenatide and glimepiride groups, respectively, had concentrations of more than 7% at two visits after the first 9 months of treatment. The most common reason for study discontinuation was patient decision (figure 1). Table 1 shows baseline demographic and diabetes characteristics of the intention-to-treat population. The mean HbA1c concentration of enrolled patients was lower than originally assumed, and history of type 2 diabetes was fairly short. Consistent with inclusion criteria, patients were taking metformin at close to the recommended maximum dose, with a median dose of 2000 mg per day (IQR 1700–2550). Average treatment time was about 2 years (exenatide group, mean 101·9 weeks [SD 73·8]; glimepiride group, 113·1 weeks [70·9]). Mean exenatide dose was 17·35 (4·07) μg per day and mean glimepiride dose was 2·01 (1·02) mg per day. Treatment failure diverged for each group with time (figure 2). 203 (41%) of 490 patients in the exenatide group had treatment failure compared with 262 (54%) of 487 in the glimepiride group (risk difference 12·4%, 95% CI 6·2–18·6), despite more patients from the exenatide group discontinuing treatment (figure 1). The HR for inadequate glycaemic control with exenatide compared with glimepiride was 0·748. The upper one-sided CI of the Cox proportional hazard analysis was 0·899, which was less than the predefined non-inferiority value of 1·25. According to the two-sided 95% CI, exenatide was more effective than glimepiride as add-on treatment for patients with metformin failure (95% CI 0·623–0·899; p=0·002). Median time to inadequate HbA1c control was 180 weeks (IQR 52·3 [upper values not reached because of

3

6 440 471

12 331 371

18 24 Time (months) 230 264

30

36 182 197

42

Figure 3: Changes in HbA1c concentration during treatment with exenatide or glimepiride Values show least-squares means with 95% CI from mixed model repeated measures model analysis, including terms for baseline HbA1c, visit, and treatment by visit interaction, with an unstructured covariance matrix. We included only visits with >25% of originally enrolled patients remaining. p values for treatment difference are shown are at years 1, 2, and 3. HbA1c=glycated haemaglobin A1c.

event rate in each group after 1 year. With these assumptions, 527 patients per study group would provide about a 90% power to conclude non-inferiority of exenatide. Analyses were by intention to treat with the caveat that only randomly assigned patients receiving at least one dose of study treatment, and with baseline and at least one post-baseline HbA1c measurement were included. We analysed the as-treated population according to treatment actually received and included only patients with at least 6 months’ follow-up for HbA1c. For all measures except primary endpoint, tests were two-sided (α=0·05). We did sensitivity analysis to examine the effect of discontinuations as a possible competing risk before the primary endpoint was met. Furthermore, we did a posthoc analysis to examine proportions of patients meeting each of the two definitions of inadequate HbA1c control. We used a mixed model repeated measures analysis for continuous variables, with terms for visit, treatment, and interaction, and included the baseline value as a covariate. We included only visits with more than 25% of originally enrolled patients and made no imputations for missing data. Least-squares means with 95% CI were derived from the model for 1, 2, and 3 years (visits eight, 12, and 16). Analyses of covariance (ANCOVA), including terms for treatment, baseline HbA1c stratum, and baseline values were done for changes from baseline to treatment failure or other endpoint. For secondary outcomes not identified at each study visit, we used last observation carried forward to account for missing values. We based safety analyses on all patients who received study drug. Percentages of patients with adverse events after treatment, and those who had hypoglycaemia, were compared between treatment groups with Pearson’s χ² test. This trial is registered with EudraCT,
4

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

Articles

Exenatide (n=490)

Glimepiride (n=487)

Exenatide LS mean (95% CI; n)

Glimepiride LS mean (95% CI; n)

Difference LS mean (95% CI)

p value

Ratio of change in glucose and insulin at 30 min Baseline 1 year 2 years 3 years HOMA-IR Baseline 1 year 2 years 3 years Disposition index Baseline 1 year 2 years 3 years Baseline 1 year 2 years 3 years 3·84 (2·16–6·62) 7·89 (4·11–14·83) 7·90 (4·25–15·43) 8·75 (4·55–18·37) 0·12 (0·07–0·20) 0·14 (0·08–0·22) 0·15 (0·08–0·24) 0·14 (0·06–0·23) 3·90 (2·15–6·46) 5·24 (3·26–8·42) 5·39 (3·00–9·41) 5·31 (3·21–9·30) 0·12 (0·07–0·20) 0·15 (0·10–0·25) 0·16 (0·10–0·27) 0·14 (0·08–0·28) ·· 12·43 (10·78–14·07; 279) 13·79 (6·20–21·38; 182) 12·56 (10·25–14·88; 130) ·· 0·19 (0·16–0·21; 292) 0·21 (0·18–0·24; 200) 0·22 (0·18–0·27; 152) ·· 7·37 (5·83–8·91; 320) 7·89 (5·78–10·01; 156) ·· 0·20 (0·19–0·22; 338) 0·24 (0·21–0·26; 228) 0·23 (0·18–0·27; 170) ·· 5·06 (2·81–7·31) 4·67 (1·53–7·81) ·· –0·02 (–0·05 to 0·01) –0·03 (–0·07 to 0·01) –0·00 (–0·07 to 0·06) ·· <0·0001 0·047 0·004 ·· 0·163 0·134 0·904 4·87 (3·06–7·48) 3·13 (1·83–5·10) 2·84 (1·65–5·08) 2·54 (1·35–4·75) 4·66 (2·97–7·22) 4·16 (2·67–6·78) 4·24 (2·43–7·05) 3·76 (2·12–6·42) ·· 4·50 (4·00–5·00; 299) 4·15 (3·66–4·65; 197) 3·51 (2·98–4·03; 149) ·· 5·30 (4·83–5·77; 336) 5·05 (4·59–5·51; 230) 4·83 (4·34–5·33; 166) ·· –0·80 (–1·48 to –0·12) –0·89 (–1·57 to –0·21) –1·33 (–2·05 to –0·60) ·· 0·022 0·010 0·0003 17·9 (10·4–30·1) 25·1 (14·0–43·6) 22·5 (15·0–38·8) 23·4 (13·0–41·3) 16·9 (10·4–29·8) 22·7 (14·1–33·3) 12·4 (13·0–33·9) 19·8 (11·6–31·4) ·· 35·3 (29·9–40·7; 279) 31·2 (16·0–46·4; 182) 25·8 (19·3–32·3; 130) ·· 27·4 (22·4–32·4; 320) 16·4 (2·4–30·3; 216) 26·4 (20·4–32·3; 156) ·· 7·9 (0·5–15·2) 14·9 (–5·8 to 35·5) –0·6 (–9·4 to 8·3) ·· 0·036 0·157 0·900

3·36 (–3·61 to 10·33; 216) 10·43 (0·13–20·73)

Proinsulin to insulin ratio

Data are unadjusted median (IQR), unless otherwise indicated. LS=least-squares. HOMA=homoeostatic model assessment.

Table 2: Variables from oral glucose tolerance tests at baseline and after treatment with exenatide or glimepiride

inadequate control in less than 75% of patients in each group]) with exenatide versus 142·1 weeks (52·3) with glimepiride (p=0·032). Risk of treatment failure was significantly affected by baseline HbA1c concentration (HR 2·417, 95% CI 2·127–2·745; p<0·0001). Risk of treatment failure was greatest for patients with high baseline HbA1c, and the reduction in risk with exenatide compared with glimepiride was greater for those with higher baseline HbA1c concentrations (figure 2). We noted no significant interactions of treatment with country, age or sex (data not shown). Mean HbA1c concentration fell from baseline to treatment failure or other endpoint in the exenatide group from 7·45% (SD 0·69) to 7·08% (0·89), and in the glimepiride group from 7·42% (0·71) to 7·22% (0·79). Least-squares mean change in HbA1c from baseline to treatment failure differed significantly (p=0·002) in patients in the exenatide group (–0·36%; 95% CI –0·43 to –0·30) compared with those in the glimepiride group (–0·21%; –0·28 to –0·14). Figure 3 shows mean HbA1c concentration over time for each treatment group. We noted an overall treatment effect in favour of the exenatide group, and the difference between groups in least-squares mean HbA1c concentration was significant at years 2 (p=0·008) and 3 (p=0·035); however, the difference in HbA1c change from baseline was significant at years 1 (p=0·043), 2 (p=0·001), and 3 (p=0·013). Significantly more patients attained an HbA1c concentration of less than 7% in the exenatide group than in the glimepiride group

Exenatide (n=511) Serious adverse events* Treatment-emergent adverse events† Nausea Nasopharyngitis Diarrhoea Headache Influenza Back pain Vomiting Bronchitis Arthralgia Pharyngitis Dyspepsia 147 (29%) 96 (19%) 62 (12%) 56 (11%) 55 (11%) 52 (10%) 44 (9%) 34 (7%) 21 (4%) 26 (5%) 26 (5%) 73 (14%)

Glimepiride (n=508) 68 (13%) 11 (2%) 93 (18%) 33 (7%) 48 (9%) 35 (7%) 54 (11%) 12 (2%) 31 (6%) 42 (8%) 21 (4%) 21 (4%)

Data are n (%). Hypoglycaemia was reported separately and not included in adverse events. *Most frequent (>0·5% of patients) adverse events in the exenatide group were cases of fall (n=3), breast cancer (3), and nephrolithiasis (3); and in the glimepiride group were osteoarthritis (7), coronary artery disease (4), meniscus lesion (4), goitre (3), and tendon rupture (3). †Reported by >5% of either group.

Table 3: Treatment-emergent adverse events

(218 [45%] of 490 vs 150 [31%] of 487; p<0·0001), and a target HbA1c concentration of 6·5% and less (140 [29%] vs 87 [18%]; p=0·0001). Fasting plasma glucose concentration was significantly lower in the exenatide group after years 1 (p=0·048), 2 (p=0·004), and 3 (p<0·0001) of treatment (appendix).

See Online for appendix

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

5

Articles

1·0 0·9 0·8 Proportion of patients with no hypoglycaemic episodes 0·7 0·6 0·5 0·4 0·3 0·2 0·1 0 0 Number at risk Exenatide 511 Glimepiride 508 6 446 284 12 297 153 18 24 Time (months) 185 81 30

Exenatide twice daily Glimepiride

36 100 35

42

Figure 4: Incidence of hypoglycaemia Kaplan-Meier survival curves for patients reporting any hypoglycaemic episodes; median time to first hypoglycaemic episode was 42·3 months (IQR 7·5 [upper values not reached because <75% of patients had a hypoglycaemic episode]) in the exenatide group vs 5·0 months [IQR 0·8–27·8]) in the glimepiride group. p<0·0001 for treatment difference.

Exenatide (n=511) Documented symptomatic* Documented symptomatic† At least one hypoglycaemic episode reported Nocturnal hypoglycaemia Non-nocturnal hypoglycaemia Severe hypoglycaemia 102 (20%) 34 (7%) 186 (36%) 53 (10%) 178 (35%) 1 (<1%)

Glimepiride p value (n=508) 240 (47%) 63 (12%) 338 (67%) 82 (16%) 333 (66%) 0 (0%) <0·0001 0·002 <0·0001 0·007 <0·0001 0·319

Data are n (%). *Blood glucose <3·9 mmol/L. †Blood glucose <2·8 mmol/l.

Table 4: Patients reporting hypoglycaemia at any time during the study

Plasma glucose concentration at 0·5 h of the oral glucose tolerance test fell in both treatment groups, with no statistical difference between groups at any time (data not shown). The decrease from baseline to endpoint in plasma glucose at 2 h of the tolerance test was greater in patients in the exenatide group (p<0·0001) than in those in the glimepiride group, and mean value was significantly lower at 1, 2, and 3 years of treatment (p<0·0001 at all times). Plasma insulin at fasting and 0·5 h of the oral glucose tolerance test did not differ between treatments at any timepoint. Insulin at 2·0 h in the test was significantly greater in patients in the exenatide group at years 2 (p=0·008) and 3 (p=0·022). Mean insulinogenic index—ie, changes in glucose and insulin at 30 min—differed between groups only at 1 year (table 2). The decrease from baseline to treatment failure with HOMA-IR was significantly greater in the exenatide group than in the glimepiride group (least-squares mean difference between groups –0·99, 95% CI –1·86 to –0·11; p=0·027), and mean HOMA-IR was significantly lower
6

in the exenatide group than the glimepiride group at all 3 years (table 2). The increase in disposition index was significantly greater in patients in the exenatide group than in those in the glimepiride group (least-squares mean difference 6·16, 0·40–11·91; p=0·036), and mean disposition index was significantly higher in the exenatide group at all 3 years (table 2). We noted no significant differences between treatments in proinsulin to insulin ratio (table 2), or in HOMA-B mean values or changes from baseline (data not shown). Self-monitored excursions of blood glucose after meals were significantly lower in the exenatide group than in the glimepiride group after breakfast (least-squares mean excursion from ANCOVA: exenatide 0·30 mmol/L [95% CI 0·13–0·47] vs glimepiride 0·90 mmol/L [0·73–1·07]; p<0·0001), lunch (1·09 mmol/L [0·91–1·27] vs 1·70 mmol/L [1·52–1·89; p<0·0001), and dinner (0·63 mmol/L [0·44–0·82] vs 1·60 mmol/L [1·41–1·79]; p<0·0001). Five patients in each treatment group died; death was given as the reason for discontinuation for six patients, with other reasons given for discontinuation for four patients (figure 1). All deaths were from causes regarded by investigators as unrelated to study treatment. Significantly more patients discontinued in the exenatide group than in the glimepiride group because of adverse events (49 vs 17; p=0·001). However, discontinuations due to adverse events were only significantly different between treatments in the first 6 months of study (32 patients in the exenatide and six in the glimepiride group; p=0·0005), and not thereafter. Consistent with the known tolerability profile of exenatide, adverse events leading to discontinuations in the exenatide group were mainly gastrointestinal, and included nausea (22 [4%] patients in the exenatide group vs 0 in the glimepiride group) and diarrhoea (13 [3%] vs 0). Table 3 summarises the most frequent adverse events occurring after treatment. One patient in each study group had pancreatitis and one in the glimepiride group had thyroid cancer. Systolic blood pressure decreased in patients in the exenatide group (change to endpoint –1·9 mm Hg; p=0·006), but not in the glimepiride group (1·1 mm Hg; p=0·096), resulting in a significant difference between groups from year 1 (–3·1 mm Hg, 95% CI –5·0 to –1·2; p=0·001) to year 3 (–5·2 mm Hg, –7·6 to –2·8; p<0·0001). Heart rate increased at endpoint in patients given exenatide (1·2 beats per min [bpm]; p=0·024), but not in those given glimepiride (0·6 bpm; p=0·282), with no difference between groups at any time. Bodyweight fell from baseline to endpoint in the exenatide group (–3·32 kg [SD 5·45]) and rose in the glimepiride group (1·15 kg [4·18]); difference in change from baseline between groups was significant after 4 weeks and at each time thereafter (p<0·0001). Consequently, BMI was significantly lower in the exenatide group than the glimepiride group from 1 month (least-squares mean difference –0·39 kg/m²

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

Articles

[95% CI –0·46 to –0·32]; p<0·0001) to 3 years (–1·88 kg/m² [–2·20 to –1·57; p<0·0001). Proportion of patients reporting hypoglycaemia was lower in patients in the exenatide group than in those in the glimepiride group (p<0·0001; figure 4). Occurrences of symptomatic documented hypoglycaemia and nocturnal and nonnocturnal hypoglycaemia were significantly lower for patients in the exenatide group than for those in the glimepiride group (table 4). Although one patient reported severe hypoglycaemia within the first month after randomisation to exenatide, blood glucose was not measured for confirmation. For hypoglycaemia of any type during the study, the least-squares mean rate estimated from a negative binomial model was 1·52 (95% CI 1·26–1·82) episodes per year in the exenatide group compared with 5·32 (4·47–6·34) episodes per year in the glimepiride group, with an exenatide to glimepiride ratio for rate of hypoglycaemia of 0·29 (95% CI 0·22–0·37; p<0·0001). In patients on glimepiride, the dose at which any type of hypoglycaemia was reported was 1 mg per day for 49·3% of episodes, 2 mg per day for 37%, 3 mg per day for 5·6%, 4 mg per day for 6·5%, 5 mg per day for 0·3%, and 6 mg per day for 0·7%.

Panel: Research in context Systematic review Common practice for patients with type 2 diabetes inadequately controlled by metformin has been to add a sulphonylurea to the treatment regimen.1,4 GLP-1 receptor agonists can be used as add-on to metformin as an alternative treatment option.6–8 We searched PubMed from 1970 to 2012 for “randomised clinical trials”, “GLP-1 receptor agonists” and “sulphonylureas”, with no restriction on language. This search identified 22 studies, of which six compared a GLP-1 receptor agonist with a sulphonylurea; three11,14,20 with treatments as add-on to metformin for up to 1 year, and three as monotherapies for up to 2 years.21–23 None of the reported studies continued beyond 2 years. Interpretation This is the longest randomised controlled study of a GLP-1 receptor agonist reported so far. With comparative treatment for up to 4·5 years, our findings show that glycaemic control in terms of HbA1c concentration was maintained for longer and in a higher proportion of patients given exenatide than for those given glimepiride as add-on to metformin. Furthermore, those in the exenatide group had maintained weight loss and reduced rates of hypoglycaemia. Exenatide twice a day is therefore a more effective treatment option than is glimepiride for patients with type 2 diabetes with metformin failure.

Discussion
Our findings show that exenatide twice daily as add-on to metformin reduced worsening of glycaemic control and rate of hypoglycaemia compared with add-on glimepiride in patients with type 2 diabetes inadequately controlled by metformin alone. Furthermore, exenatide was more effective than glimepiride for fasting glucose, glucose excursions after meals, and HbA1c concentration. Overall, safety and tolerability of both drugs was consistent with the known safety profiles. The most frequently reported adverse events with exenatide were gastrointestinal; these events resulted in more frequent study discontinuations at the start of the study, but not after the first 6 months of treatment. Randomised controlled head-to-head studies are important to guide clinical decisions, especially for new treatments different from the standard of care, which, according to our findings, would be a sulphonylurea. Up to now, EUREXA is the longest study undertaken with a GLP-1 receptor agonist, and could contribute substantially to decisions in clinical practice. We chose treatment failure needing alternative treatment as the primary endpoint to assess the clinical effect of two different second-line treatments on disease progression; therefore, our analyses represent real-life medical practice with early and late treatment failure (panel). This endpoint is similar to that of the ADOPT study5 in previously untreated patients. ADOPT showed a reduction in treatment failure for rosiglitazone compared with metformin or glyburide as monotherapy, with the effect subsequently shown to correspond with improved β-cell function and insulin sensitivity.24 Although our

findings did not show a significant difference in HOMA-B, HOMA-IR and disposition index were significantly improved in patients in the exenatide group, which might be related to the improvements in bodyweight and glycaemia. When our study started, treatment options in the event of metformin failure were scarce.1 Add-on options were insulin25 or a thiazolidinedione;26 however, these drugs are associated with increased risks of hypoglycaemia, weight gain, oedema, congestive heart failure, and bone fractures.27,28 Dipeptidyl peptidase-4 inhibitors are now also used as add-on drugs,7,9,29,30 but were not available at the start of this study, and their effectiveness might not be higher than that of sulphonylureas.31 Improvements in fasting glucose, glucose excursions after meals, and HbA1c concentration in patients in the exenatide group were associated with initially enhanced insulinogenic index, increased disposition index, and decreased HOMA-IR. Improvements from baseline in such factors were similar to those reported in previous studies with exenatide,12,13 and our findings were consistent with the decreased risk of hypoglycaemia and decreased bodyweight associated with exenatide treatment. Improved glycaemic control and β-cell function with reduced hypoglycaemia noted with exenatide use are achieved through glucose-dependent stimulation of insulin secretion, by contrast with glimepiride, which increases insulin secretion via
7

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

Articles

non-glucose dependent pathways. The effectiveness of both treatments could have been affected by the lower than anticipated baseline concentration of HbA1c. However, exenatide was better than glimepiride for prevention of inadequate glycaemic control in patients with raised HbA1c at baseline, and inclusion of patients with decreased control would not change the benefits of exenatide. The population enrolled was not ethnically diverse, which limits applicability of our study findings mainly to white patients. Furthermore, the glimepiride dose used was fairly low, despite titration to the maximum tolerated dose recommended by the protocol. However, the highest dose given to individual patients was identified by the attending physicians, and high incidence of hypoglycaemia could have prevented investigators from increasing the sulphonylurea dose. For use of an increased dose of glimepiride, findings from two previous studies did not show improved effectiveness. In the GUIDE study,32 although glimepiride was titrated from 1 mg to 6 mg daily, HbA1c reduction was not improved. When glimepiride was compared with vildagliptin,33 the increased glimepiride dosage of 4·5 mg per day decreased HbA1c by only 0·53%, from 7·3% at baseline; these values are almost identical to our findings at 12 months with a glimepiride dosage of 2 mg, when HbA1c was decreased by 0·5%, from 7·4% at baseline. In conclusion, our findings provide evidence for a beneficial effect of exenatide twice daily versus usual care with glimepiride, for deterioration of glycaemia in patients with type 2 diabetes.
Contributors BGa and GS designed, submitted, and drafted the report. All authors analysed and interpreted data, and revised the text. Conflicts of interest BGa has been a consultant for, and received honoraria from, AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Eli Lilly, Novartis, Novo Nordisk, Merck, Roche, Sanofi-Aventis, and Takeda. BGu has been a consultant for GlaxoSmithKline, Eli Lilly, Merck, AstraZeneca, Bristol-Myers Squibb, Pfizer, Novo Nordisk, Novartis, Abbott, Lifescan, Medtronic, and Menarini. RS has been a consultant for, and received honoraria from, Novo Nordisk, Eli Lilly, and Abbott. GS has been a consultant for, and received honoraria, from Amgen, AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Eli Lilly, Merck, Novartis, Poxel, Roche, Sanofi-Aventis, Novo Nordisk, Servier, and Takeda. BRB, AF, JK, MT and HS are employees of Eli Lilly and Company, and BRB, AF, JK, and MT hold stocks in Eli Lilly and Company. Acknowledgments We thank the EUREXA investigators and clinical teams; the study participants; Ludger Rose (Münster, Germany) for his helpful advice; Christof Kazda (Lilly Research Center) for his involvement in study design and implementation; and Peter Bates (Cambridge Medical Writing Services, UK) for writing support and assistance with manuscript preparation. References 1 Nathan DM, Buse JB, Davidson MB, et al. Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2009; 32: 193–203.

2

3

4

5

6

7

8 9

10

11

12

13

14

15

16 17

18

19

20

21

22

Holman RR, Paul SK, Bethel MA Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med 2008; 359: 1577–89. International Diabetes Federation. Global guideline for type 2 diabetes. 2005. http://www.idf.org/guidelines/type-2-diabetes (accessed Jan 19, 2012). International Diabetes Federation. Treatment algorithm for people with type 2 diabetes. 2011. http://www.idf.org/treatment-algorithmpeople-type-2-diabetes (accessed Jan 19, 2012). Kahn SE, Haffner SM, Heise MA, et al. Glycemic durability of rosiglitazone, metformin, or glyburide monotherapy. N Engl J Med 2006; 355: 2427–43. Drucker DJ, Sherman IS, Gorelick FS, Bergenstal RM, Sherwin RS, Buse JB. Incretin-based therapies for the treatment of type 2 diabetes: evaluation of the risks and benefits. Diabetes Care 2010; 33: 428–33. Tahrani AA, Bailey CJ, Del Prato S, Barnett AH. Management of type 2 diabetes: new and future developments in treatment. Lancet 2011; 378: 182–97. Gallwitz B. Glucagon-like peptide-1 analogues for type 2 diabetes mellitus: current and emerging agents. Drugs 2011; 71: 1675–88. Bergenstal RM, Wysham C, MacConell L, et al. Efficacy and safety of exenatide once weekly versus sitagliptin or pioglitazone as an adjunct to metformin for treatment of type 2 diabetes (DURATION-2): a randomised trial. Lancet 2010; 376: 431–39. Drucker DJ, Buse JB, Taylor K, et al. Exenatide once weekly versus twice daily for the treatment of type 2 diabetes: a randomised, open-label, non-inferiority study. Lancet 2008; 372: 1240–50. Nauck M, Frid A, Hermansen K, et al. Efficacy and safety comparison of liraglutide, glimepiride, and placebo, all in combination with metformin, in type 2 diabetes: the LEAD (liraglutide effect and action in diabetes)-2 study. Diabetes Care 2009; 32: 84–90. DeFronzo RA, Triplitt C, Qu Y, Lewis MS, Maggs D, Glass LC. Effects of exenatide plus rosiglitazone on beta-cell function and insulin sensitivity in subjects with type 2 diabetes on metformin. Diabetes Care 2010; 33: 951–57. Bunck MC, Corner A, Eliasson B, et al. Effects of exenatide on measures of β-cell function after 3 years in metformin-treated patients with type 2 diabetes. Diabetes Care 2011; 34: 2041–47. Buse JB, Rosenstock J, Sesti G, et al. A study of two glucagon-like peptide-1 receptor agonists for the treatment of type 2 diabetes: liraglutide once-daily compared with exenatide twice daily in a randomised, 26-week, open-label trial (LEAD-6). Lancet 2009; 374: 39–47. DeFronzo RA, Abdul-Ghani MA. Preservation of β-cell function: the key to diabetes prevention. J Clin Endocrinol Metab 2011; 96: 2354–66. Visbøll T. The effects of glucagon-like peptide-1 on the beta cell. Diabetes Obes Metab 2009; 11 (suppl 3): 11–18. Kazda C, Gallwitz B, Simo R, et al. The European Exenatide study of long-term exenatide vs. glimepiride for type 2 diabetes: rationale and patient characteristics. Diabetes Obes Metab 2009; 11: 1131–37. American Diabetes Association Workgroup on Hypoglycemia. Defining and reporting hypoglycemia in diabetes. Diabetes Care 2005; 28: 1245–49. European Agency for Evaluation of Medicinal Products. Committee for Proprietary Medicinal Product (CPMP). Points to consider on switching between superiority and non-inferiority. July 27, 2000. http://www.emea.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2009/09/WC500003658.pdf (accessed Jan 19, 2012). Buse JB, Rosenstock J, Sesti G, et al. Liraglutide once a day versus exenatide twice a day for type 2 diabetes: a 26-week randomised, parallel-group, multinational, open-label trial (LEAD-6). Lancet 2009; 374: 39–47. Garber A, Henry RR, Ratner R, et al. Liraglutide, a once-daily human glucagon-like peptide 1 analogue, provides sustained improvememnts in galycaemic control and weight for 2 years as monotherapy compared with glimepiride in patients with type 2 diabetes. Diabetes Obes Metab 2011; 13: 348–56. Seino Y, Rasmussen ME, Nishida T, Kaku K. Efficacy and safety of the once-daily human GLP-1 analogue, liraglutide, vs glibenclamide monotherapy in Japanese patients with type 2 diabetes. Curr Med Res Opin 2010; 26: 1013–22.

8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

Articles

23

24

25

26

27

28

Derosa G, Maffioli P, Salvadeo SA, et al. Exenatide versus glibenclamide in paitents with diabetes. Diabetes Technol Ther 2010; 26: 233–40. Kahn SE, Lachlin JM, Zinman B, et al. Effects of rosiglitazone, glyburide, and metformin on beta-cell function and insulin sensitivity in ADOPT. Diabetes 2011; 60: 1552–60. Malone JK, Kerr LF, Campaigne BN, Sachson RA, Holcombe JH, for the Lispro Mixture-Glargine Study Group. Combined therapy with insulin lispro Mix 75/25 plus metformin or insulin glargine plus metformin: a 16-week, randomized, open-label, crossover study in patients with type 2 diabetes beginning insulin therapy. Clin Ther 2004; 26: 2034–44. Derosa G, Tineli C, Maffioli P. Effects of pioglitazone and rosiglitazone combined with metformin on body weight in people with diabetes. Diabetes Obes Metab 2009; 11: 1091–99. Bergenstal RM, Bailey CJ, Kendall DM. Type 2 diabetes: assessing the relative risks and benefits of glucose-lowering medications. Am J Med 2010; 123: 374.e9–18. Aguilar RB. Evaluating treatment algorithms for the management of patients with type 2 diabetes mellitus: a perspective on the definition of treatment success. Clin Ther 2011; 33: 408–24.

29

30

31

32

33

DeFronzo RA, Hissa MN, Garber AJ, et al. The efficacy and safety of saxagliptin when added to metformin therapy in patients with inadequately controlled type 2 diabetes with metformin alone. Diabetes Care 2009; 32: 1649–55. Pratley RE, Nauck M, Bailey T, et al. Liraglutide versus sitagliptin for patients with type 2 diabetes who did not have adequate glycaemic control with metformin: a 26-week, randomised, parallel-group, open-label trial. Lancet 2010; 375: 1447–56. Matthew DR, Dejager S, Ahren B, et al. Vildagliptin add-on to metformin produces similar efficacy and reduced hypoglycaemic risk compared with glimepiride, with no weight gain: results from a 2-year study. Diabetes Obes Metab 2010; 12: 780–89. Schernthaner G, Grimaldi A, Di Mario U, et al. GUIDE study: double-blind comparison of once-daily gliclazide MR and glimepiride in type 2 diabetic patients. Eur J Clin Invest 2004; 34: 535–42. Ferrannini E, Fonseca V, Zinman B, et al. Fifty-two week efficacy and safety of vildagliptin vs glimepiride in patients with type 2 diabetes mellitus inadequately controlled on metformin therapy. Diabetes Obes Metab 2009; 11: 157–66.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60479-6

9

Series

Diabetes 2 Diabetes and cognitive dysfunction
Rory J McCrimmon, Christopher M Ryan, Brian M Frier

Cognitive dysfunction in type 1 and type 2 diabetes share many similarities, but important differences do exist. A primary distinguishing feature of type 2 diabetes is that people with this disorder often (but not invariably) do poorly on measures of learning and memory, whereas deficits in these domains are rarely seen in people with type 1 diabetes. Chronic hyperglycaemia and microvascular disease contribute to cognitive dysfunction in both type 1 and type 2 diabetes, and both disorders are associated with mental and motor slowing and decrements of similar magnitude on measures of attention and executive functioning. Additionally, both types are characterised by neural slowing, increased cortical atrophy, microstructural abnormalities in white matter tracts, and similar, but not identical, changes in concentrations of brain neurometabolites. Disconcertingly, the rapid rise in obesity and type 2 diabetes in all age groups might result in a substantial increase in prevalence of diabetes-related cognitive dysfunction.

Published Online June 9, 2012 DOI:10.1016/S01406736(12)60360-2 This is the second in a Series of three papers about diabetes Medical Research Institute, Ninewells Hospital and Medical School, University of Dundee, Dundee, UK (R J McCrimmon MD), Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA, USA (Prof C M Ryan PhD); and Department of Diabetes, Royal Infirmary of Edinburgh, Edinburgh, UK (Prof B M Frier MD) Correspondence to: Dr Rory J McCrimmon, Medical Research Institute, Ninewells Hospital and Medical School, Dundee DD1 9SY, UK [email protected]

Introduction
Around 171 million people worldwide have diabetes mellitus, and this number is estimated to double by 2030. The two primary forms of diabetes are type 1 diabetes—an autoimmune disorder characterised by an absolute or near total loss of insulin secretion—and type 2 diabetes, which is characterised by reduced insulin sensitivity and relative insulin deficiency. In both forms, chronic hyperglycaemia can lead to microvascular and macrovascular complications. Although most focus has been on end-organ disease affecting the kidney, eyes, and peripheral nervous system, the brain is also affected. Diabetes, its complications, and its treatment can induce transient or permanent cognitive abnormalities, which result from acute and chronic disturbances of blood glucose homoeostasis. Here, we focus on diabetes-related cognitive dysfunction rather than on other related effects of diabetes on the brain, such as glucose-sensing, the regulation of wholebody glucose homoeostasis, or stroke, which have been reviewed elsewhere.1,2 We consider the lessons learned from cognitive, neurophysiological, neuroanatomical, and neurochemical assessments of brain function in diabetes and compare results of studies of type 1 and type 2 diabetes. The increasing prevalence of diabetes and the decreasing age of diabetes diagnosis suggest that diabetesrelated cognitive dysfunction will probably increase and have a substantial effect on society. Efforts to understand the pathophysiological changes that underpin the development and progression of diabetes-related cognitive dysfunction are of vital importance to develop treatments to reverse or prevent these cognitive complications.

nephropathy.3 Intensive insulin therapy, designed to achieve near-normal glucose control, minimises the development and severity of these complications, but at the expense of increased severe hypoglycaemia (blood glucose concentrations so low that external assistance is needed for recovery).4 Severe hypoglycaemia can be partly attributed to limitations of insulin regimens and profound defects in the physiological, symptomatic, and behavioural responses to hypoglycaemia that individuals with type 1 diabetes develop.2 Recurrent hypoglycaemia causes cerebral adaptation, with resetting of glycaemic thresholds for symptom generation and counter-regulation—a syndrome known as impaired hypoglycaemia awareness. Thus, type 1 diabetes treatment is often beset by fluctuations in blood glucose concentrations that, along with associated hormonal and metabolic disturbances, might contribute to the development of diabetes-related cognitive dysfunction (panel 1).

Cognitive dysfunction
Type 1 diabetes has a specific effect on a subset of cognitive domains in adults, including intelligence, attention,

Search strategy and selection criteria A professional clinical librarian searched PubMed with the following MeSH headings, combined with a Boolean “or” operator: “diabetes mellitus”, “diabetes mellitus, type 2”, “diabetes mellitus, type 1”, “diabetes complications”, “hyperglycemia”, “blood glucose”, “hypoglycemia”, or “hemoglobin A, glycosylated” (set 1) and “brain”, “cognition”, “cognition disorders”, “dementia”, “memory”, “memory disorders”, “neuropsychological tests”, “executive function”, or “psychomotor performance” (set 2). These two sets were combined with a Boolean “and” operator and the search was restricted to human studies published in English between 1995 and 2012. Additional searches of authors were done and reference lists were reviewed after assessment of relevant studies identified from the PubMed retrieval. The final search took place on Jan 12, 2012.

Type 1 diabetes
Glucose control
Type 1 diabetes develops most often in childhood or adolescence and always needs insulin replacement therapy. Chronic hyperglycaemia associated with inadequate insulin replacement heightens the risk of microvascular complications such as retinopathy, neuropathy, and

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

1

Series

Panel 1: Factors contributing to the development of cognitive dysfunction in diabetes Individuals with type 1 and type 2 diabetes can develop several microvascular and macrovascular complications that can contribute to cognitive dysfunction, which might be exacerbated by a genetic predisposition to neuroinflammatory brain disease. Metabolic factors • Chronic hyperglycaemia • Acute hypoglycaemia • Recurrent hypoglycaemia • Protein glycation • Changes in fuel metabolism and transport Vascular disease • Microvascular disease • Macrovascular disease • Endothelial dysfunction • Inflammation • Changes in blood–brain barrier permeability • Rheological factors • Dyslipidaemia Endocrine factors • Reduced insulin sensitivity • Hyperinsulinaemia • Hypothalamic–pituitary–adrenal axis dysregulation • Increased antidiuretic hormone • Hyperleptinaemia CNS factors • Genetic predisposition • Amyloid disposition • Oxidative stress • Changes in neuronal calcium homoeostasis • Depression

brains might be more susceptible to the effects of diabetes than adult’s brains, although this might be because glycaemic control is more difficult to achieve in this population. Individuals who develop type 1 diabetes early in life (younger than 7 years) are at a higher risk of developing more severe cognitive deficits than are those who develop diabetes at an older age.9 One study in adolescents10 reported that 24% of those with early-onset diabetes (diagnosed before age 6 years) showed clinically significant impairments in a wide range of cognitive domains compared with only 6% of the later-onset patients, and 6% of people without diabetes.

Neurophysiological and cerebrovascular changes
Neurophysiological changes underpinning cognitive dysfunction in type 1 diabetes are largely unknown, mainly because of the difficulties surrounding the study of specific brain regions in man. Nevertheless, studies of electroencephalograms in adults11 and adolescents12 with diabetes have noted significant reductions in fast brainwave (α, β, and γ) activity, particularly in temporooccipital regions, compared with people without diabetes, whereas slow wave (δ and θ) activity was increased in several frontal areas. Reduced γ-band activity is associated with cognitive decline, whereas activity in the δ and θ areas is associated with subcortical lesions and metabolic encephalopathy. Magnetic encephalography detected abnormalities in functional magnetic fields and in the neural connectivity of the brain at the scalp in people with type 1 diabetes.13 These abnormalities were present irrespective of microvascular disease status, but were greater when accompanied by retinopathy than when not. Magnetic encephalography has revealed similar changes in people with other brain diseases. Others have reported an association between neurophysiological abnormalities and peripheral neuropathy in people with diabetes, and suggested that this association represents a central neuropathy, although direct evidence is scarce.14 Measurement of cerebral blood flow with single-photon emission tomography in children15 and adults with type 1 diabetes shows significant regional variation in cerebral perfusion (either increased or decreased compared with controls without diabetes) in many brain regions, but most noticeably in the cerebellum, frontal brain, and frontotemporal brain.16,17 Changes in perfusion correlated most with poor glycaemic control and the presence of microvascular complications such as retinopathy. Although many cognitive studies have shown that changes in perfusion are related to poor glycaemic control or cognitive test results,18 a strong relation has not been reported, and no evidence exists to suggest that perfusion abnormalities play an important part in the development of cognitive dysfunction in type 1 diabetes.

psychomotor speed, cognitive flexibility, and visual perception (table).5 The magnitude of cognitive dysfunction reported in most trials is moderate, with effect sizes (a measure of the strength of the difference between people with diabetes and healthy controls) ranging from 0·3 to 0·8 SD units. Most cognitive tests examine the participant’s ability to respond rapidly, and mental slowing is thought to be the fundamental cognitive deficit associated with type 1 diabetes.6 Toddlers7 and children8 with type 1 diabetes also show the same pattern of cognitive dysfunction as that shown in adults. By contrast, learning and memory, the cognitive domains thought to be most susceptible to early brain disease, seem to be unaffected even when patients have had a long history of poor glycaemic control. Thus, in type 1 diabetes, cognitive dysfunction emerges early in the disease course (within 2 years of diagnosis) and tends to have very circumscribed effects, particularly on intelligence and psychomotor speed. Age is also an important variable, and children’s
2

Imaging studies
Neuroimaging has consistently shown slight structural changes in the brains of people with type 1 diabetes,

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Series

particularly in cortical grey matter.19 In the largest study of its type so far, brain density in 82 adults aged 25–40 years with type 1 diabetes and 36 healthy participants was assessed with voxel-based morphometry—a well established, semi-automated quantitative MRI technique.20 Those with type 1 diabetes showed reductions of 4–5% in grey matter density in several brain regions compared with controls, which correlated with lifetime glycated haemoglobin A1c (HbA1c) values, but the reduction in density was unrelated to cognitive test scores or history of recurrent hypoglycaemia.20 That study excluded patients with proliferative retinopathy, but others21 have shown that patients with retinopathy had substantial reductions in grey matter density in frontal gyri, occipital lobe, and cerebellum. Studies using diffusion tensor imaging (DTI) have shown white matter microstructural abnormalities in middle-aged adults with long-standing type 1 diabetes. DTI measures the direction of water diffusion in tissues and is an index of the integrity of highly organised and constrained tissues such as white matter. DTI revealed abnormalities in the integrity of several white matter tracts, particularly the posterior corona radiata and optic radiations, and these abnormalities were correlated with increased duration of diabetes, raised HbA1c concentrations, and poor results in cognitive tasks that tested visuospatial analysis and hand–eye coordination.22 Subsequent regional connectivity analyses showed that these white matter abnormalities were linked directly to a reduction in grey matter cortical thickness, suggesting that long-standing diabetes leads to concurrent microstructural changes to both grey and white matter, mainly in posterior cerebral regions.23 In view of the possibility that these abnormalities underlie the mental slowing that is characteristic of many people with diabetes, DTI techniques clearly have a place in future research studies.

Number of studies Overall cognition Intelligence Crystallised Fluid Language Attention Visual Sustained Learning and memory Working memory Verbal learning Verbal delayed memory Visual learning Visual delayed memory Psychomotor speed Cognitive flexibility Visual perception 8 5 3 5 4 8 9 5 5 3 5 4 4 16

n

Effect size (SD units) 0·40 0·80 0·50 0·05 0·40 0·30 0·10 0·20 0·30 0·10 0·10 0·60 0·50 0·40

p value

660 276 168 144 195 217 244 204 157 187 157 368 364 202

<0·001 <0·01 <0·01 NS <0·001 <0·01 NS NS NS NS NS <0·05 <0·001 <0·001

Data are from a meta-analysis of 33 case-control studies of individuals aged 18–50 years.5 The standardised effect sizes (Cohen’s d) show differences between people with diabetes and those without. NS=not significant.

Table: Characteristics of diabetes-related cognitive dysfunction in adults with type 1 diabetes compared with healthy controls

concentrations of glucose in the brain in patients with type 1 diabetes with retinopathy.25

Biomedical risk factors for cognitive dysfunction
Reports from the early 1990s suggested that cognitive dysfunction in type 1 diabetes might be more pronounced in individuals exposed to repeated severe hypoglycaemia, a finding consistent with anecdotal reports of severe hypoglycaemia inducing cortical changes in several brain regions such as the frontal and temporal cortex, basal ganglia, and hippocampus.26,27 However, longitudinal epidemiological studies have tended to implicate chronic hyperglycaemia and microvascular disease in the pathogenesis of diabetes-related cognitive dysfunction. The Diabetes Control and Complications Trial/ Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) cognitive follow-up study28 of 1144 participants showed no evidence for cognitive dysfunction in eight cognitive domains (assessed with a comprehensive battery of cognitive tests) in patients who had experienced one or more episodes of severe hypoglycaemia (defined as blood glucose concentration lower than 2·8 mmol/L accompanied by seizure or coma) during an 18·5-year follow-up. Similarly, data from a systematic meta-analysis did not show any relation between recurrent hypoglycaemia and performance in cognitive tests.5 By contrast, many studies using various measures of brain integrity have shown that microvascular complications are associated with
3

Imaging brain metabolism
Several investigators have examined the effect of type 1 diabetes on whole-brain concentrations of metabolites and neurotransmitters. Unless done under controlled conditions, these studies can be difficult to interpret because the ratio of interstitial brain glucose to blood glucose (about 1:5) remains constant across a range of glucose concentrations, so measurements need to be made under similar glucose concentrations. However, correlations between lifetime glycaemic control and increased frontal brain glucose and neurotransmitter concentrations measured with proton magnetic resonance spectroscopy in adults with type 1 diabetes do suggest an underlying pathophysiological change.24 Moreover, adults with type 1 diabetes had compromised memory and executive function and reduced psychomotor speed compared with controls. Studies using magnetic resonance spectroscopy have also shown that poor glycaemic control is associated with biomarkers of gliosis and altered neuronal integrity, and with high

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Series

increased risk of cognitive dysfunction.29 The DCCT/ EDIC cognitive study identified five variables that independently predicted reduction in psychomotor speed throughout the 18·5-year follow-up: old age, low number of years in education, high lifetime HbA1c concentrations, and two clinically significant microvascular complications (proliferative diabetic retinopathy and renal complications).18 Additionally, increased carotid intima-media thickness (a marker of early macrovascular disease) was marginally associated with decreased performance in cognitive tests. In this study, retinopathy was most strongly associated with cognitive dysfunction, a finding also reported in type 2 diabetes. A limitation of the DCCT/EDIC cognitive study is that it has examined a young population (aged <50 years) of low-risk adults with type 1 diabetes that are in relatively good control in whom cognitive sequelae of repeated hypoglycaemia might not yet have emerged. However, findings from trials such as DCCT/EDIC suggest that, at least within this younger diabetic population with reasonable glycaemic control, any cognitive decline throughout follow-up was small and, as such, likely to have progressed very slowly. Overall, a growing body of research suggests that moderate-to-severe hypoglycaemia does not seem to result in substantial cognitive dysfunction, although that might not be true for some high-risk groups such as children diagnosed within the first few years of life.30,31

Type 2 diabetes
Cerebrovascular disease
Type 2 diabetes is a heterogeneous metabolic disorder, characterised by reduced insulin sensitivity and relative insulin deficiency. Coexisting disorders, including obesity, hypertension, and dyslipidaemia, contribute to the severity of type 2 diabetes. By contrast with type 1 diabetes, macrovascular disease causes about 80% of mortality in people with type 2 diabetes. Interventions to reduce blood glucose in people with type 2 diabetes significantly lower the risk of microvascular, and possibly macrovascular, disease.32 The brain is a target end-organ in type 2 diabetes and prediabetes, but the cause of diabetes-related cognitive dysfunction is difficult to establish because of the prevalence of several comorbidities, each of which might affect cognitive function (panel 1). Perhaps the most important of these comorbidities is cerebrovascular disease. Diabetes is associated with a 1·5–2·0-fold increased risk of stroke,33 with a stroke relative-risk increase of 1·15 (95% CI 1·08–1·23) for every 1% rise in HbA1c.34 Cerebrovascular disease, when present, probably contributes substantially to cognitive dysfunction in type 2 diabetes.

Cognitive dysfunction
Although changes in psychomotor speed and other cognitive modalities have been reported, learning and memory deficits are the cognitive abnormalities that most clearly differentiate adults with type 2 diabetes
4

from those with type 1. Effect sizes are small, ranging between 0·25 and 0·5 SD units.35,36 Additionally, the highly variable extent of cognitive dysfunction seen in adults with type 2 diabetes might be caused by the presence of several comorbid disorders, such as hypertension and obesity. Indeed, a study of elderly people with type 2 diabetes showed that their learning and memory skills were significantly disrupted, but when the analysis was adjusted for the presence of hypertension this difference was not significant.37 In cross-sectional studies of older adults (aged 60–85 years) with type 2 diabetes, extent of cognitive dysfunction was associated most strongly with long disease duration and poor metabolic control.37,38 Surprisingly, diabetes might not increase the rate of cognitive decline over time. When people aged 56–80 years with diabetes were followed up for 4 years, their overall cognitive decline, measured by a comprehensive battery of cognitive tests, did not differ significantly from agematched participants without diabetes (figure 1),39 although people with diabetes did more poorly overall.39,40 People with type 2 diabetes aged 85 years followed up for 5 years had the same pattern of results (ie, similar rate of decline to controls, but poorer overall performance at each timepoint of assessment than non-diabetic controls).41 Although these follow-up periods were quite short, the findings suggest that cognitive dysfunction in type 2 diabetes might have a demarcated onset—similar to a traumatic brain injury, which impairs cognitive performance absolutely, but does not otherwise affect the rate of change over time. Whether dysfunction takes place during a critical period, perhaps soon after or even before the diagnosis of diabetes or the emergence of microvascular complications,42 is unknown. This hypothesis suggests that therapeutic intervention and reversal of diabetes-related cognitive dysfunction is possible, provided the main underlying pathophysiological changes can be identified. By contrast, large-scale longitudinal studies and systematic reviews suggest that diabetes increases the risk of developing dementia.38,43 Diabetes is associated with a 50–100% increased risk of Alzheimer’s disease and a 100–150% increased risk of vascular dementia.40 Several endocrine, metabolic, and vascular abnormalities have been linked to diabetes and dementia, including ischaemic cerebrovascular disease, hyperglycaemiaassociated neurotoxicity (glucose toxicity), changes in insulin and amyloid metabolism, increased oxidative stress, and increased release of inflammatory factors such as C-reactive protein, interleukin 6, and tumour necrosis factor α (panel 1).41,42 However, the causal pathway that underlies the statistical associations between type 2 diabetes and dementia is unknown. Comparison of the results of neurocognitive assessments of individuals with type 1 diabetes with those of people with type 2 diabetes is useful. Brand and colleagues44 compared age, sex, and estimated-IQ-matched patients

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Series

with type 1 and 2 diabetes and reported that patients with type 2 diabetes had significantly greater cortical atrophy and more lesions in deep white matter. The type 2 diabetes group had a much shorter duration of disease than did the type 1 group (7 years vs 34 years), better glycaemic control, and lower rates of clinically significant microvascular disease (laser-treated retinopathy, 8% vs 38%), but had higher rates of macrovascular disease and more atherosclerosis risk factors (eg, hypercholesterolaemia, hypertriglyceridaemia, hypertension, and high body-mass index). These findings implicate reduced insulin sensitivity and macrovascular disease in the pathogenesis of cortical atrophy and cognitive dysfunction in type 2 diabetes, by contrast with type 1 diabetes in which substantial disruptions in metabolic state and microvascular disease probably have a primary role.

0·5 Information-processing speed (Z score)

0·3

0·1

–0·1

–0·3

–0·5 Baseline 4-year follow-up Baseline 4-year follow-up

Neurophysiological and cerebrovascular changes
Adults with type 2 diabetes have evidence of neural slowing on recordings of sensory-evoked potentials. Neural slowing can occur soon after diagnosis and is indicative of metabolic status or perhaps subclinical disease45 because it is significantly worse in patients with overt peripheral neuropathy.46 The slowing of evoked potentials seems to proceed at the same rate as that in individuals without diabetes, but overall, adults with type 2 diabetes have 4–11% slower evoked potentials than do those without diabetes,47–49 again supporting the notion of a crucial period for development of this complication. Furthermore, morbidly obese adolescents with diabetes performed more poorly in tests of several cognitive domains than did obese adolescents without diabetes, despite an average time of less than 2 years from diagnosis and no evidence of microvascular disease.50 Structural brain abnormalities were also present. Changes in cerebral blood flow, measured with a range of methods, are well documented in older adults (mean age 62 years) with type 2 diabetes. Advanced techniques, such as continuous arterial spin labelling MRI, have shown that patients with diabetes have significantly lower cerebral blood flow than do healthy controls, and have evidence of cortical and subcortical atrophy.51 People with type 1 and type 2 diabetes show similar neural slowing very early in the disease, and in both types this slowing is exacerbated by the presence of microvascular complications, suggesting that overall metabolic state and underlying vascular disease both contribute to neural activity. However, as in type 1 diabetes, the relation between these measures and cognition is poorly understood and, in many studies, neurophysiological function, cerebral blood flow, and cognitive function are not closely correlated.52

Figure 1: Cognitive decline in people with type 2 diabetes compared with non-diabetic controls Data are for people with type 2 diabetes (n=68) and matched non-diabetic control participants (n=38), followed up for 4 years.39 A repeated-measures ANOVA (mean standardised domain score; error bars show SE of measurement) showed no significant decrease in information-processing speed over time for both groups (p=0·28), a significant main effect of group (p=0·02), and a non-significant time×group interaction (p=0·23), whereas for attention and executive functioning, both groups’ performance decreased significantly over time (p<0·001), main effect of group was significant (p=0·04), but time×group interaction was not (p=0·37). Adapted from van den Berg and colleagues,39 by permission of Springer on behalf of the authors.

Imaging studies
Pronounced structural changes in the brain have been noted in people with type 2 diabetes. Cerebral atrophy, white matter lesions, and infarctions are frequently

reported,53,54 and correlate with the presence of microvascular and macrovascular complications.55 The rate of change in cerebral atrophy can be quite slow—eg, a 0·11% increase in ventricular volume over 4 years54—and is also similar to age-related changes, although the number of people examined in such studies is small and the follow-up times might have been too short to discern subtle differences in rate of increase in brain atrophy. Lacunar infarcts are also reported more frequently in people with type 2 diabetes than in those without diabetes.56 Lacunar infarcts result from occlusion of penetrating arteries that supply arterial blood to deep brain structures. A meta-analysis of the few studies of cognitive function in patients with diabetes reported a significant association between diabetes and lacunar infarcts (odds ratio 1·3, 95% CI 1·1–1·6).56 Lacunar infarcts are often silent (ie, the individual is unaware that they have had a stroke), but contribute to changes in mood, personality, and cognitive dysfunction. Another consistent neuroimaging finding is the presence of hippocampal atrophy in patients with type 2 diabetes. Hippocampal atrophy is evident early in the course of the disease, and has even been documented in elderly people with prediabetes.57 As much as a 10–15% loss in hippocampal volume has been reported in otherwise healthy middle-aged and elderly people with diabetes, despite them having similar frontal and temporal brain volumes to those without diabetes.36 Hippocampal atrophy correlated with impairments in immediate memory, and was best predicted by HbA1c concentration, but not by hypertension.36 The hippocampus is susceptible to acute metabolic changes such as hypoglycaemia,58 suggesting that it might be particularly susceptible to diabetes-related metabolic and vascular change.
5

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Attention and executive functioning (Z score)

Controls Type 2 diabetes

Series

0·4 0·2 0 –0·2 –0·4 –0·6 –0·8 –1·0

Men Women

these studies and those of patients with type 1 diabetes suggest that the neurometabolic changes occurring in the brain vary appreciably between these two disorders, with the greatest changes seen in adults with type 1 diabetes (who tend to have had the disorder for a much longer time). However, few studies have measured brain metabolites in these two populations, therefore clear conclusions are difficult to draw. More research is needed using magnetic resonance spectroscopy techniques to directly compare people with type 1 and type 2 diabetes to carefully ascertain the nature and extent of biomedical complications.
Diabetic retinopathy Moderate to severe diabetic retinopathy

g score

No diabetic retinopathy

Biomedical risk factors for cognitive dysfunction
Chronic hyperglycaemia and long duration of diabetes are both associated with increased development of cognitive dysfunction, as is the presence of vascular risk factors (eg, hypertension, hypercholesterolaemia, and obesity) and microvascular and macrovascular complications.29,62 The Edinburgh type 2 diabetes study63—a cross-sectional survey of more than 1000 elderly people aged 60–75 years with type 2 diabetes—showed that age-adjusted and sexadjusted general cognitive ability was significantly lower in people with moderate-to-severe diabetic retinopathy (mean –0·44, 95% CI –0·73 to –0·16) than in those without retinopathy (0·05, –0·03 to 0·12; p=0·003; figure 2). Moreover, in men, but not women, diabetic retinopathy was associated with estimated lifetime cognitive decline.63 Reduced insulin sensitivity, which is common in people with type 2 diabetes, might also affect cognitive processes via several mechanisms,64,65 as might hypoglycaemia.66

Figure 2: Retinopathy and cognitive function in elderly people with type 2 diabetes Data are estimated means of general cognitive ability (g) scores of 1046 men and women aged 60–75 years with type 2 diabetes who underwent standard seven-field binocular digital retinal photography and a battery of seven cognitive function tests. The g score was generated by principal components analysis. In the total study population, mean g score was significantly lower for people with moderate-to-severe diabetic retinopathy (–0·44, 95% CI –0·73 to –0·16) than for those without retinopathy (0·05, –0·03 to 0·12; p=0·003). Error bars show ±2 SE. Adapted from Ding and colleagues,63 by permission of the American Diabetes Association.

Hippocampal atrophy is one of the neuroanatomical features that differs between people with types 1 and 2 diabetes. Both have reduced grey matter density and white matter lesions, although cortical atrophy is generally more pronounced in type 2 diabetes (possibly because this population is older on average). Why the hippocampus is more affected in type 2 diabetes is unclear, particularly because this region is susceptible to acute metabolic change, which is a more prominent feature of type 1 diabetes. These findings suggest that age, associated comorbidities, and macrovascular disease or insulin resistance might be important risk factors for hippocampal atrophy.

Therapeutic interventions for cognitive dysfunction in diabetes
Long-term prospective trials such as DCCT/EDIC suggest that cognitive decline in people with type 1 diabetes is likely to be slowly progressive and mild, at least in those with good glycaemic control. The correlation between lifetime HbA1c concentrations, retinopathy, and cognitive decline in DCCT/EDIC18 suggests that intensive insulin therapy to improve overall glycaemic control is a prudent approach. DCCT/EDIC also showed a significant reduction in overall vascular events with improved glycaemic control67 and a reduction in progression of carotid intima-media thickness, suggesting that this approach would be beneficial because it would reduce progression of cerebrovascular risk. Benefits of aggressive glucose management in type 2 diabetes are less clear. The Memory in Diabetes (MIND) substudy68 of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial showed no benefit of intensive glucose-lowering therapy on cognitive function or total brain volume in a large population of people with type 2 diabetes during 40-month follow-up. Similarly, ACCORD,69 ADVANCE (Action in Diabetes, and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation),70

Imaging brain metabolism
Adults with type 2 diabetes assessed with proton magnetic resonance spectroscopy had concentrations of specific brain metabolites that were different from those reported in older adolescents and adults with type 1 diabetes.24,59 Myo-inositol is located in astrocytes and its concentration changes in brain disease. It is increased in the frontal white matter of elderly people with type 2 diabetes.60,61 Concentrations of myo-inositol correlate with the presence of macrovascular disease and complications, but not with HbA1c, suggesting that frontal gliosis can arise secondary to cerebrovascular changes. People older than 60 years with type 2 diabetes do not show abnormalities in other neurotransmitters and metabolites, whereas younger adults with type 1 diabetes show abnormalities in concentrations of myo-inositol, choline, and N-acetylaspartate (a marker of neuronal damage).59 Taken together, the differences between
6

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Series

and the Veterans’ Affairs diabetes trial71 reported that intensive glucose control had no significant effect on macrovascular disease in people with type 2 diabetes. By contrast, PROactive (PROspective pioglitAzone Clinical Trial In macroVascular Events), which examined the effect of pioglitazone in a large cohort of people with type 2 diabetes, reported that, in a prespecified analysis, an HbA1c concentration difference of 0·5% was associated with a reduction in both fatal and non-fatal stroke (hazard ratio [HR] 0·53, 95% CI 0·34–0·85).72 Particularly in type 2 diabetes, but also in type 1, treatment of hypertension and dyslipidaemia are important. The links between hypertension, ischaemic stroke, and cognitive dysfunction in type 2 diabetes suggest that modification of these risk factors will also ameliorate cognitive decline. Societies such as the European Stroke Organisation recommend a target blood pressure of 130/80 mm Hg in people with diabetes undergoing therapy that includes an angiotensinconverting enzyme inhibitor or angiotensin-receptor blocker.73 Statin use also reduces vascular risk. In a 5-year follow-up study of 1674 Mexican-Americans older than 60 years (about a third of whom had diabetes), those taking statins were about half as likely as were non-statin users to develop dementia or cognitive impairment (HR 0·52, 95% CI 0·34–0·80).74

Panel 2: Screening and assessment of cognitive dysfunction in diabetes Discussion of the patient’s cognitive complaints and level of functioning Assessments should focus on perception of deficits relative to previous performance. Collation of information about their concurrent affective state (eg, depressed, anxious, stressed), metabolic status at the time of cognitive dysfunction (eg, hypoglycaemic or recently hypoglycaemic), and their overall level of metabolic control is important. Identification of other potential biomedical causes of declining cognitive function Common disorders include hypertension, hypercholesterolaemia, obesity, and poor sleep quality and each risk factor should be managed according to national recommendations.75–77 Counselling of the patient about possible causes of cognitive dysfunction Several factors, such as poor glycaemic control, especially when associated with microvascular disease, might contribute to the development of cognitive dysfunction. If a connection between improved control and improved cognitive function is made, it might motivate patients to work more actively to improve glycaemia. Mood can also interfere with cognitive function.78 Referral for assessment by clinical neuropsychologist if necessary If the patient reports having substantial difficulties with daily activities or undertaking tasks necessary for disease management, referral should be considered. The neuropsychologist can systematically document cognitive strengths and weaknesses, establish extent of impairment, and work with the health-care team to identify factors other than diabetes that might be disrupting cognitive functioning. Consideration of cognitive remediation services No study has examined whether cognitive rehabilitation is effective in reversal of diabetes-associated cognitive dysfunction, but studies of other disorders have suggested that these approaches might have some value when applied to patients with well documented cognitive dysfunction.79,80 Assistance of patients with accommodation to their new, normal level of functioning If dysphoria or anxiety accompanies cognitive impairment, psychotropic drugs should be considered.81

Recommendations for clinical practice
Cognitive dysfunction associated with diabetes is mild in most instances and rarely meets criteria for clinically significant impairment, but can occur in children as well as adults and is irrespective of diabetes type. Preliminary evidence suggests that cognitive changes begin early in the disease course and can worsen over time.40 Reduction in mental efficiency might be sufficient to disrupt performance in the classroom, workplace, and home. If a patient reports that their performance in school or at work is worsening, or their ability to undertake activities of daily living is deteriorating, including diabetes selfmanagement behaviours, or if they ask about the effects of diabetes on functioning, we recommend the approach to screening and assessment outlined in panel 2.

Conclusions and future perspectives
In general, cognitive characteristics of people with type 2 diabetes are similar to those seen in people with type 1 diabetes. Both groups show evidence of mental and motor slowing (a nearly ubiquitous finding) and similar performance decrements on measures of executive functioning such as planning, attention, working memory, and problem solving (effect size is about 0·3–0·4 SD units). People with type 2 diabetes perform worse than healthy controls on learning and memory tests, unlike those with type 1 diabetes, who rarely have deficits in these domains; however, people with type 1 diabetes and those with type 2 show evidence of neural slowing, changes in cerebral perfusion, increased

cortical atrophy, and microstructural abnormalities in white matter tracts. Hippocampal atrophy seems to be a more prominent feature of type 2 diabetes than of type 1. Several biomedical risk factors might contribute to cognitive dysfunction in diabetes. In type 1 diabetes, evidence suggests that chronic exposure to high glucose concentrations and the presence of microvascular disease, in particular retinopathy, are major contributors to the development of the disorder. This finding is especially interesting and might be indicative of the structural homology between the retinal and cerebral blood supply. If this link is substantiated, digitised fundal photography might provide a non-invasive assessment of cerebral microcirculation. In type 2 diabetes, insulin resistance, dyslipidaemia, hypertension, and cerebrovascular disease seem to be of great importance to the development of cognitive dysfunction and should be addressed in the management of this disorder. The evidence suggesting that intensive glycaemic control improves cognitive outcomes is weak, although pioglitazone and metformin might prove effective through reduction of macrovascular risk. These
7

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Series

questions will only be answered by large, long-term intervention trials in which detailed cognitive assessment is combined with neuroimaging.
Contributors RJM, CMR, and BMF contributed equally to the review of published work and to the writing and editing of this report. Conflicts of interest We declare that we have no conflicts of interest. Acknowledgments We thank Michele Klein-Fedyshin, Falk Medical Library, University of Pittsburgh School of Medicine (Pittsburgh, PA, USA), for her support and advice. References 1 Quinn TJ, Dawson J, Walters MR. Sugar and stroke: cerebrovascular disease and blood glucose control. Cardiovasc Ther 2011; 29: e31–42. 2 McCrimmon RJ, Sherwin RS. Hypoglycemia in type 1 diabetes. Diabetes 2010; 59: 2333–39. 3 The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 1993; 329: 977–86. 4 The Diabetes Control and Complications Trial Research Group. Hypoglycemia in the diabetes control and complications trial. Diabetes 1997; 46: 271–86. 5 Brands AMA, Biessels G-J, De Haan EHF, Kappelle LJ, Kessels RPC. The effects of type 1 diabetes on cognitive performance: a meta-analysis. Diabetes Care 2005; 28: 726–35. 6 Ryan CM. Diabetes, aging, and cognitive decline. Neurobiol Aging 2005; 26 (suppl 1): 21–25. 7 Patiño-Fernández AM, Delamater AM, Applegate EB, et al. Neurocognitive functioning in preschool-age children with type 1 diabetes mellitus. Pediatr Diabetes 2010; 11: 424–30. 8 Northam EA, Anderson PJ, Jacobs R, Hughes M, Warne GL, Werther GA. Neuropsychological profiles of children with type 1 diabetes 6 years after disease onset. Diabetes Care 2001; 24: 1541–46. 9 Ryan CM. Diabetes and brain damage: more (or less) than meets the eye? Diabetologia 2006; 49: 2229–33. 10 Ryan C, Vega A, Drash A. Cognitive deficits in adolescents who developed diabetes early in life. Pediatrics 1985; 75: 921–27. 11 Brismar T, Hyllienmark L, Ekberg K, Johansson B-L. Loss of temporal lobe beta power in young adults with type 1 diabetes mellitus. Neuroreport 2002; 13: 2469–73. 12 Hyllienmark L, Maltez J, Dandenell A, Ludviggson J, Brismar T. EEG abnormalities with and without relation to severe hypoglycaemia in adolescents with type 1 diabetes. Diabetologia 2005; 48: 412–19. 13 van Duinkerken E, Klein M, Schoonenboom NS, et al. Functional brain connectivity and neurocognitive functioning in patients with long-standing type 1 diabetes with and without microvascular complications: a magnetoencephalography study. Diabetes 2009; 58: 2335–43. 14 Dejgaard A, Gade A, Larsson H, Balle V, Parving A, Parving H. Evidence for diabetic encephalopathy. Diabetic Med 1991; 8: 162–67. 15 Salem MA, Matta LF, Tantawy AA, Hussein M, Gad GI. Single photon emission tomography (SPECT) study of regional cerebral blood flow in normoalbuminuric children and adolescents with type 1 diabetes. Pediatr Diabetes 2002; 3: 155–62. 16 Jiménez-Bonilla JF, Quirce R, Hernández A, et al. Assessment of cerebral perfusion and cerebrovascular reserve in insulin-dependent diabetic patients without central neurological symptoms by means of 99mTc-HMPAO SPET with acetazolamide. Eur J Nucl Med 2001; 28: 1647–55. 17 Quirce R, Carril JM, Jiménez-Bonilla JF, et al. Semi-quantitative assessment of cerebral blood flow with 99mTc-HMPAO SPET in type 1 diabetic patients with no clinical history of cerebrovascular disease. Eur J Nucl Med 1997; 24: 1507–13. 18 Jacobson AM, Ryan CM, Cleary PA, et al. Biomedical risk factors for decreased cognitive functioning in type 1 diabetes: an 18 year follow-up of the Diabetes Control and Complications Trial (DCCT) cohort. Diabetologia 2011; 54: 245–55.

19 20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

Jongen C, Biessels GJ. Structural brain imaging in diabetes: a methodological perspective. Eur J Pharmacol 2008; 585: 208–18. Musen G, Lyoo IK, Sparks CR, et al. Effects of type 1 diabetes on gray matter density as measured by voxel-based morphometry. Diabetes 2006; 55: 326–33. Wessels AM, Simsek S, Remijnse PL, et al. Voxel-based morphometry demonstrates reduced gray matter density on brain MRI in patients with diabetic retinopathy. Diabetologia 2006; 49: 2474–80. Kodl CT, Franc DT, Rao JP, et al. Diffusion tensor imaging identifies deficits in white matter microstructure in subjects with type 1 diabetes that correlate with reduced neurocognitive function. Diabetes 2008; 57: 3083–89. Franc DT, Kodl CT, Mueller BA, Muetzel RL, Lim KO, Seaquist ER. High connectivity between reduced cortical thickness and disrupted white matter tracts in long-standing type 1 diabetes. Diabetes 2011; 60: 315–19. Lyoo IK, Yoon SJ, Musen G, et al. Altered prefrontal glutamateglutamine-gamma-aminobutyric acid levels and relation to low cognitive performance and depressive symptoms in type 1 diabetes mellitus. Arch Gen Psychiatry 2009; 66: 878–87. Mäkimattila S, Malmberg-Cèder K, Häkkinen A-M, et al. Brain metabolic alterations in patients with type 1 diabetes-hyperglycemiainduced injury. J Cereb Blood Flow Metab 2004; 24: 1393–99. Gold AE, Deary IJ, Frier BM. Recurrent severe hypoglycaemia and cognitive function in type 1 diabetes. Diabetic Med 1993; 10: 503–08. Deary I, Crawford J, Hepburn DA, Langan SJ, Blackmore LM, Frier BM. Severe hypoglycemia and intelligence in adult patients with insulin-treated diabetes. Diabetes 1993; 42: 341–44. Jacobson AM, Musen G, Ryan CM, et al, for the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Study Research Group. Long-term effects of diabetes and its treatment on cognitive function. N Engl J Med 2007; 356: 1842–52. Wessels AM, Scheltens P, Barkhof F, Heine RJ. Hyperglycaemia as a determinant of cognitive decline in patients with type 1 diabetes. Eur J Pharmacol 2008; 585: 88–96. Aye T, Reiss AL, Kesler S, et al. The feasibility of detecting neuropsychologic and neuroanatomic effects of type 1 diabetes in young children. Diabetes Care 2011; 34: 1458–62. Asvold BO, Sand T, Hestad K, Bjorgaas MR. Cognitive function in type 1 diabetic adults with early exposure to severe hypoglycemia: a 16-year follow-up study. Diabetes Care 2010; 33: 1945–47. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998; 352: 837–53. Folsom AR, Rasmussen ML, Chambless LE, et al, for the Atherosclerosis Risk in Communities (ARIC) study investigators. Prospective associations of fasting insulin, body fat distribution, and diabetes with risk of ischemic stroke. Diabetes Care 1999; 22: 1077–83. Selvin E, Bolen S, Yeh HC, et al. Cardiovascular outcomes in trials of oral diabetes medications: a systematic review. Arch Intern Med 2008; 168: 2070–80. Strachan MWJ, Deary IJ, Ewing FME, Frier BM. Is type 2 (non-insulin dependent) diabetes mellitus associated with an increased risk of cognitive dysfunction? Diabetes Care 1997; 20: 438–45. Gold SM, Dziobek I, Sweat V, et al. Hippocampal damage and memory impairments as possible early brain complications of type 2 diabetes. Diabetologia 2007; 50: 711–19. Van Harten B, Oosterman J, Muslimovic D, Potter van Loon B-J, Scheltens P, Weinstein HC. Cognitive impairments and MRI correlates in the elderly patients with type 2 diabetes mellitus. Age Ageing 2007; 36: 164–70. Kloppenborg PR, Van den Berg E, Kappelle LJ, Biessels GJ. Diabetes and other vascular risk factors for dementia: what factor matters most? A systematic review. Eur J Pharmacol 2008; 585: 97–108. van den Berg E, Reijmer YD, de Bresser J, et al. A 4 year follow-up study of cognitive functioning in patients with type 2 diabetes mellitus. Diabetologia 2010; 53: 58–65. Biessels GJ, Deary IJ, Ryan CM. Cognition and diabetes: a lifespan perspective. Lancet Neurol 2008; 7: 184–90.

8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

Series

41 42 43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58 59

60

61

Haan MN. Therapy insight: type 2 diabetes mellitus and the risk of late-onset Alzheimer’s disease. Nat Rev Neurol 2006; 2: 159–66. Whitmer RA. Type 2 diabetes and risk of cognitive impairment and dementia. Curr Neurol Neurosci Rep 2007; 7: 373–80. Biessels GJ, Staekenborg S, Brunner E, Scheltens P. Risk of dementia in diabetes mellitus: a systematic review. Lancet Neurol 2006; 5: 64–74. Brands AMA, Biessels GJ, Kappelle LJ, et al. Cognitive functioning and brain MRI in patients with type 1 and type 2 diabetes mellitus: a comparative study. Dement Geriatr Cogn Disord 2007; 23: 343–50. Pozzessere G, Rizzo PA, Valle E, et al. Early detection of neurological involvement in IDDM and NIDDM: multimodal evoked potentials versus metabolic control. Diabetes Care 1988; 11: 473–80. Várkonyi TT, Petõ T, Dégi R, et al. Impairment of visual evoked potentials: an early manifestation of diabetic neuropathy? Diabetes Care 2002; 25: 1161–62. Hissa MN, D’Almeida JA, Cremasco F, de Bruin VM. Event related P300 potentials in NIDDM patients without cognitive impairment and its relationship with previous hypoglycemic episodes. Neuro Endocrinol Lett 2002; 23: 226–30. Kurita A, Katayama K, Mochio S. Neurophysiological evidence for altered higher brain functions in NIDDM. Diabetes Care 1996; 19: 361–64. Dey J, Misra A, Desai NG, Mahapatra AK, Padma MV. Cerebral function in a relatively young subset of NIDDM patients. Diabetologia 1995; 38: 251. Yau PL, Javier DC, Ryan CM, et al. Preliminary evidence for brain complications in obese adolescents with type 2 diabetes mellitus. Diabetologia 2010; 53: 2298–306. Last D, Alsop DC, Abduljalil AM, et al. Global and regional effects of type 2 diabetes on brain tissue volumes and cerebral vasoreactivity. Diabetes Care 2007; 30: 1193–99. Tiehuis AM, Vincken KL, van den Berg E, et al. Cerebral perfusion in relation to cognitive function and type 2 diabetes. Diabetologia 2008; 51: 1321–26. Kumar A, Haroon E, Darwin C, et al. Gray matter prefrontal changes in type 2 diabetes detected using MRI. J Magn Reson Imaging 2008; 27: 14–19. de Bresser J, Tiehuis AM, van den Berg E, et al. Progression of cerebral atrophy and white matter hyperintensities in patients with type 2 diabetes. Diabetes Care 2010; 33: 1309–14. Manschot SM, Biessels GJ, de Valk HW, et al. Metabolic and vascular determinants of impaired cognitive performance and abnormalities on brain magnetic resonance imaging in patients with type 2 diabetes. Diabetologia 2007; 50: 2388–97. van Harten B, de Leeuw FE, Weinstein HC, Scheltens P, Biessels GJ. Brain imaging in patients with diabetes: a systematic review. Diabetes Care 2006; 29: 2539–48. Convit A, Wolf OT, Tarshish C, De Leon MJ. Reduced glucose tolerance is associated with poor memory performance and hippocampal atrophy among normal elderly. Proc Natl Acad Sci USA 2003; 100: 2019–22. Auer RN. Hypoglycemic brain damage. Metab Brain Dis 2004; 19: 169–75. Northam EA, Rankins D, Lin A, et al. Central nervous system function in youth with type 1 diabetes 12 years after disease onset. Diabetes Care 2009; 32: 445–50. Ajilore O, Haroon E, Kumaran S, et al. Measurement of brain metabolites in patients with type 2 diabetes and major depression using proton magnetic resonance spectroscopy. Neuropsychopharmacology 2007; 32: 1224–31. Geissler A, Fründ R, Schölmerich J, Feuerbach S, Zietz B. Alterations of cerebral metabolism in patients with diabetes mellitus studied by proton magnetic resonance spectroscopy. Exp Clin Endocrinol Diabetes 2003; 111: 421–27.

62

63

64

65 66

67

68

69

70

71

72

73

74

75 76

77

78

79

80

81

Reijmer YD, van den Berg E, Ruis C, Jaap Kappelle L, Biessels GJ. Cognitive dysfunction in patients with type 2 diabetes. Diabetes Metab Res Rev 2010; 27: 195–202. Ding J, Strachan MWJ, Reynolds RM, et al. Diabetic retinopathy and cognitive decline in older people with type 2 diabetes: the Edinburgh type 2 diabetes study. Diabetes 2010; 59: 2883–89. Hallschmid M, Schultes B. Central nervous insulin resistance: a promising target in the treatment of metabolic and cognitive disorders? Diabetologia 2009; 52: 2264–69. Cardoso S, Correia S, Santos RX, et al. Insulin is a two-edged knife on the brain. J Alzheimers Dis 2009; 18: 483–507. Zammitt NN, Frier BM. Hypoglycemia in type 2 diabetes: pathophysiology, frequency, and effects of different treatment modalities. Diabetes Care 2005; 28: 2948–61. Nathan DM, Cleary PA, Backlund JY, et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med 2005; 353: 2643–53. Launer LJ, Miller ME, Williamson JD, et al. Effects of intensive glucose lowering on brain structure and function in people with type 2 diabetes (ACCORD MIND): a randomised open-label substudy. Lancet Neurol 2011; 10: 969–77. Gerstein HC, Miller ME, Byington RP, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008; 358: 2545–59. Duckworth W, Abraira C, Moritz T, et al. Glucose control and vascular complications in veterans with type 2 diabetes. N Engl J Med 2009; 360: 129–39. Patel A, MacMahon S, Chalmers J, et al. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med 2008; 358: 2560–72. Wilcox R, Bousser MG, Betteridge DJ, et al. Effects of pioglitazone in patients with type 2 diabetes with or without previous stroke: results from PROactive (PROspective pioglitAzone Clinical Trial In macroVascular Events 04). Stroke 2007; 38: 865–73. European Stroke Organisation. Guidelines for stroke management. http://www.eso-stroke.org/recommendations.php?cid=9&sid=1 (accessed March 13, 2012). Cramer C, Haan MN, Galea S, Langa KM, Kalbfleisch JD. Use of statins and incidence of dementia and cognitive impairment without dementia in a cohort study. Neurology 2008; 71: 344–50. Waldstein SR. The relation of hypertension to cognitive function. Curr Dir Psychol Sci 2003; 12: 9–12. Waters F, Bucks RS. Neuropsychological effects of sleep loss: implications for neuropsychologists. J Int Neuropsychol Soc 2011; 17: 571–86. Gunstad J, Lhotsky A, Wendell CR, Ferrucci L, Zonderman AB. Longitudinal examination of obesity and cognitive function: results from the Baltimore Longitudinal Study of Aging. Neuroepidemiology 2010; 34: 222–29. Elderkin-Thompson V, Kumar A, Bilker W, et al. Neuropsychological deficits among patients with late-onset minor and major depression. Arch Clin Neuropsychol 2003; 18: 529–49. Rohling ML, Faust ME, Beverly B, Demakis G. Effectiveness of cognitive rehabilitation following acquired brain injury: a meta-analytic re-examination of Cicerone et al.’s (2000, 2005) systematic reviews. Neuropsychology 2009; 23: 20–39. Cicerone KD, Dahlberg C, Malec JF, et al. Evidence-based cognitive rehabilitation: updated review of the literature from 1998 through 2002. Arch Phys Med Rehabil 2005; 86: 1681–92. Petrak F, Herpertz S. Treatment of depression in diabetes: an update. Curr Opin Psychiatry 2009; 22: 211–17.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60360-2

9

Series

Diabetes 1 Prediabetes: a high-risk state for diabetes development
Adam G Tabák, Christian Herder, Wolfgang Rathmann, Eric J Brunner, Mika Kivimäki

Prediabetes (intermediate hyperglycaemia) is a high-risk state for diabetes that is defined by glycaemic variables that are higher than normal, but lower than diabetes thresholds. 5–10% of people per year with prediabetes will progress to diabetes, with the same proportion converting back to normoglycaemia. Prevalence of prediabetes is increasing worldwide and experts have projected that more than 470 million people will have prediabetes by 2030. Prediabetes is associated with the simultaneous presence of insulin resistance and β-cell dysfunction—abnormalities that start before glucose changes are detectable. Observational evidence shows associations between prediabetes and early forms of nephropathy, chronic kidney disease, small fibre neuropathy, diabetic retinopathy, and increased risk of macrovascular disease. Multifactorial risk scores using non-invasive measures and blood-based metabolic traits, in addition to glycaemic values, could optimise estimation of diabetes risk. For prediabetic individuals, lifestyle modification is the cornerstone of diabetes prevention, with evidence of a 40–70% relative-risk reduction. Accumulating data also show potential benefits from pharmacotherapy.

Published Online June 9, 2012 DOI:10.1016/S01406736(12)60283-9 This is the first in a Series of three papers about diabetes Department of Epidemiology and Public Health, University College London, London, UK (A G Tabák MD, E J Brunner PhD, Prof M Kivimäki PhD); 1st Department of Internal Medicine, Faculty of Medicine, Semmelweis University, Budapest, Hungary (A G Tabák); and Institute of Clinical Diabetology (C Herder PhD), and Institute for Biometry and Epidemiology (W Rathmann MD), German Diabetes Centre, Leibniz Centre for Diabetes Research at Heinrich Heine University of Düsseldorf, Düsseldorf, Germany Correspondence to: Dr Adam G Tabák, Department of Epidemiology and Public Health, University College London, London WC1E 6BT, UK [email protected]

Introduction
Prediabetes, typically defined as blood glucose concentrations higher than normal, but lower than diabetes thresholds, is a high-risk state for diabetes development. Diagnostic criteria for prediabetes have changed over time and vary depending on the institution of origin (table 1). According to WHO, people are at high risk of developing diabetes if they have one of two distinct states: impaired fasting glucose (IFG), defined as a fasting plasma glucose (FPG) concentration of ≥6·1 and <7·0 mmol/L, without impaired glucose tolerance (IGT); and IGT, defined as an FPG concentration of <7·0 mmol/L and a 2 h postload plasma glucose concentration of ≥7·8 and <11·1 mmol/L, measured during a 75 g oral glucose tolerance test (OGTT).1 The American Diabetes Association (ADA) applies the same thresholds for IGT, but uses a lower cutoff value for IFG (FPG 5·6–6·9 mmol/L), and has introduced glycated haemoglobin A1c (HbA1c) 5·7–6·4% as a new category for high diabetes risk.2 The term prediabetes has been criticised because many people with prediabetes do not progress to diabetes, and it might imply that no intervention is necessary because no disease is present. Furthermore, diabetes risk does not necessarily differ between people with prediabetes and those with a combination of other diabetes risk factors. Indeed, WHO use the term intermediate hyperglycaemia and an International Expert Committee convened by the ADA prefers the “high-risk state of developing diabetes” to prediabetes.1,3 For brevity, we use the term prediabetes in this Series paper to refer to IFG, IGT, and high-risk HbA1c concentrations. Reproducibility of thresholds used to define prediabetes (around 50%) is lower than that for diabetes diagnostic criteria (>70%),4 and each of the alternative definitions (based on IFG, IGT, or HbA1c) produce overlapping groups with distinct and shared abnormalities. People with IFG can have different pathophysiological abnormalities

from those with IGT—eg, in white people, overlap in abnormalities between those with IFG and those with IGT can be as low as 25%5—and those with both IFG and IGT tend to have more advanced disturbance of glycaemic homoeostasis.5 Individual risk factors for diabetes (eg, history of gestational diabetes or a first-degree relative with diabetes) or a combination of risk factors (eg, metabolic syndrome) can also be used to define populations at risk

Search strategy and selection criteria We searched PubMed for work published up to and including January, 2012, with the terms “prediabetes”, “impaired glucose tolerance”, or “impaired fasting glucose”. For the epidemiology section, we also searched with the terms “incidence” or “prevalence”; for the complications section, “nephropathy”, “albuminuria”, “microalbuminuria”, “chronic kidney disease”, “neuropathy”, “autonomic”, “heart rate variability”, “orthostatic”, “idiopathic neuropathy”, “erectile dysfunction”, or “Valsalva”; for the pathophysiology section, “pathophysiology”, “clamp”, “intravenous glucose tolerance test”, “insulin secretion”, or “insulin sensitivity”; and for the treatment section, “diabetes prevention”, “lifestyle intervention”, “metformin”, “troglitazone”, “rosiglitazone”, “pioglitazone”, “acarbose”, “voglibose”, “exenatide”, “liraglutide”, “nateglinide”, “ramipril”, “valsartan”, “orlistat”, “bariatric surgery”, or “fibrate”. We mainly selected publications from the past 5 years, but did not exclude widely referenced and highly regarded older publications. We also searched the reference lists of articles identified by this search strategy and selected those we judged relevant. We have cited several review articles and book chapters to provide readers with more details and references. We modified our reference list on the basis of comments from peer reviewers and restricted it to 120 references.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

1

Series

Venous plasma glucose WHO, 1965 WHO, 1980 WHO, 1985 WHO, 1999 and 2006 (most recent) ADA, 1997 ADA, 2003 ADA, 2010 (most recent) 7·1–8·2 mmol/L (postload) <8·0 mmol/L (fasting), ≥8·0 mmol/L and <11·0 mmol/L (2 h postload) <7·8 mmol/L (fasting), ≥7·8 mmol/L and <11·1 mmol/L (2 h postload) IGT: <7·0 mmol/L (fasting), ≥7·8 mmol/L and <11·1 mmol/L (2 h postload); IFG: ≥6·1 mmol/L and <7·0 mmol/L (fasting), <7·8 mmol/L (2 h postload, if measured*) IGT: <7·0 mmol/L (fasting), ≥7·8 mmol/L and <11·1 mmol/L (2 h postload); IFG: 6·1–6·9 mmol/L (fasting) IGT: <7·0 mmol/L (fasting), 7·8–11·0 mmol/L (2 h postload, if measured); IFG: 5·6–6·9 mmol/L (fasting)† IGT: <7·0 mmol/L (fasting), 7·8–11·0 mmol/L (2 h postload); IFG: 5·6–6·9 mmol/L (fasting)†; HbA1c: 5·7–6·4%

79 million adults with prediabetes.9 Prevalences of IFG and IGT vary between ethnic groups and both disorders are more common in people older than 40 years.10 Additionally, IFG is more prevalent in men than in women, although the reasons for this difference are poorly understood.10 Figure 1 shows worldwide projections of IGT prevalence for 2030, according to the International Diabetes Federation.11 The number of adults with IGT is expected to increase worldwide, reaching 472 million by 2030. The greatest absolute rises are expected in southeast Asia and the western Pacific region.11

Progression from prediabetes to diabetes
Around 5–10% of people with prediabetes become diabetic every year, although the conversion rate varies with population characteristics and prediabetes definitions.12,13 In a meta-analysis of prospective studies published between 1979 and 2004, annualised incidence rates of progression to diabetes in patients with isolated IGT (4–6%) or isolated IFG (6–9%) were lower than in those with both IFG and IGT (15–19%).14 In subsequent major studies, progression estimates have been similar— annualised incidence was 11% in the Diabetes Prevention Program (DPP) outcomes study,15 6% in participants with IFG in the US Multi-Ethnic Study of Atherosclerosis,16 and 9% in participants with IFG and 7% in those with HbA1c 5·7–6·4% in a Japanese population-based study.17 Data suggest that risk of diabetes development on the basis of FPG and 2 h postload glucose is broadly similar to that defined by HbA1c concentration.14,18 According to an ADA expert panel, up to 70% of individuals with prediabetes will eventually develop diabetes. In a Chinese diabetes prevention trial,19 the 20 year cumulative incidence of diabetes in controls with IGT defined with repeated OGTTs was even higher (>90%) than that predicted by previous studies. By comparison, women with gestational diabetes have a 20–60% risk of developing diabetes 5–10 years after pregnancy.20–22 This large heterogeneity in estimates is probably caused by variation in criteria used to define gestational diabetes and type 2 diabetes in these studies. In a meta-analysis of 20 studies,23 13% of mothers with gestational diabetes developed diabetes after pregnancy compared with 1% of mothers without gestational diabetes.

One abnormal test result defines prediabetes; no repeat testing is required. IGT=impaired glucose tolerance. IFG=impaired fasting glucose. ADA=American Diabetes Association. HbA1c=glycated haemoglobin A1c. *Measurement is recommended to exclude diabetes or IGT. †2 h postload glucose measurement not recommended.

Table 1: Diagnostic criteria for prediabetes

for diabetes, but their predictive value is poorer than that of a prediabetes classification. Additionally, risk scores for incident diabetes based on a combination of noninvasive or blood-based risk factors are under development to identify individuals at high risk of developing diabetes.6 In this Series paper we provide an updated review of the evidence of vascular complications and underlying pathophysiology of prediabetes, and discuss the clinical implications.

Epidemiology and temporal trends
Glycaemic concentrations are rapidly rising in people living in developed and developing countries.7 Pooled data from 2·7 million adults participating in health surveys and epidemiological studies suggest that age-standardised mean FPG was 5·5 mmol/L in men and 5·4 mmol/L in women in 2008, a rise of 0·1 mmol/L since 1980. People living in Oceania had the highest mean FPG of any region (6·1 mmol/L for men and women), but mean FPG was also high in those from some other regions (south and central Asia, Latin America, the Caribbean, north Africa, and the Middle East).7 Increases in glycaemia have resulted in a rise in prediabetes prevalence, although in some populations IGT has not risen despite increasing diabetes incidence, probably because increases in obesity have affected FPG more than 2 h glucose, and because of improved detection of diabetes.8 The population-based US National Health and Nutrition Examination Survey (NHANES) suggests that 35% of US adults older than 20 years and 50% of those older than 65 years had prediabetes in 2005–08, defined by FPG or HbA1c concentrations.9 Application of these percentages to the entire US population in 2010 yielded an estimated
2

Reversion to normoglycaemia
Several trials have reported reductions in the risk of diabetes development in prediabetic individuals after lifestyle and drug-based interventions.15,24–28 Prediabetes can convert back to normoglycaemia. In a populationbased observational study of the natural history of diabetes in England, 55−80% of participants with IFG at baseline had normal FPG at 10 year follow-up.12 Other studies have reported lower conversion rates29—eg, 19% in controls in the DPP outcomes study.15

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

Risk prediction
As with prediabetic status, diabetes risk models provide methods for identification of individuals at risk of diabetes on the basis of indices available to family doctors. However, no diabetes prediction model has been universally accepted, and given that ethnic origin is strongly related to diabetes risk, recalibration of prediction algorithms might be necessary when models are applied to different populations.30 Table 2 presents a selection of diabetes risk models used in Australia, Europe, and the USA. These models consist of many of the same risk factors, but they weigh these components differently. In clinical practice, a two-stage process could be efficient—diabetes prediction models with non-invasive variables such as age, sex, body-mass index (BMI), blood pressure, diabetes family history, and lifestyle information allow a first assessment of diabetes risk with little effort and cost. Laboratory measures, particularly glucose values, can improve results of non-invasive models. Thus, for patients with an increased risk at first assessment, models consisting of routinely obtained blood measures can be applied for more precise risk estimation. Classification of people as either healthy or prediabetic (those with IFG or IGT or both) neglects the fact that diabetes risk substantially increases for those with FPG values at the higher end of the normal range.31 Thus, in diabetes risk prediction, glycaemic measures (fasting or 2 h glucose, or HbA1c) might be more accurate if treated as continuous rather than categorical variables.32,33 Furthermore, incorporation of postload glucose into a model that already includes FPG can improve prediction. The KORA (Cooperative Health Research in the Region of Augsburg) study32 and the Framingham Offspring Study33 reported the usefulness of straightforward clinical and laboratory measurements to derive diabetes prediction models suitable for general practice.32,33 Derivation of both models suggested that some information about metabolic traits (eg, glucose, uric acid, and lipids) beyond personal diabetes risk factors is important to adequately establish future risk of type 2 diabetes. Most attempts to substantially improve diabetes prediction with measurements from genetics and transcriptomics have not been successful, and whether serial measurements might decrease variations in non-genetic biomarkers, resulting in a more precise estimation of their concentrations, is not known.34–36
Prevalence of IGT (millions)

180 160 140 120 100 80 60 40 20 0

2010 (actual) 2030 (projected)

Africa

Middle East and North Africa

Europe

North America and Caribbean

South and Central America

Southeast Asia

Western Pacific region

Figure 1: Actual and projected prevalence of impaired glucose tolerance (IGT) by region in adults aged 20–79 years in 2010 and 2030 Data are from the International Diabetes Federation Diabetes Atlas,11 which provides a breakdown of the countries included in each of the geographical regions.

is a continuous process.35,36,40,41 We described trajectories of fasting and postload glucose, and trajectories of insulin sensitivity and insulin secretion (β-cell function) measured by homoeostatic model assessment, preceding development of type 2 diabetes in the British Whitehall II study (figure 2).36 In people who developed diabetes, increased glucose values were seen as early as 13 years before diagnosis, although glucose values seemed to be tightly regulated within the normal range until 2–6 years before diagnosis, when an abrupt increase was found. Other studies have confirmed this pattern of glycaemic changes.35,40,41 Figure 2 shows that insulin sensitivity was already reduced 13 years before onset of diabetes, with a steeper fall seen 5 years before diagnosis. Insulin secretion (β-cell function) was steady throughout the 13 year observation period and showed a substantial compensatory increase 3–4 years before diagnosis before decreasing steeply.36 These results support the notion that insulin resistance starts years before diabetes development and that decreased β-cell function is already present in the prediabetic stage.37,42

Multistage model of diabetes development
Weir coined a multistage model of diabetes development43 that corresponds to these findings. The first stage is defined by a long period of insulin resistance accompanied by a compensatory increased rate of insulin secretion44 and increased β-cell mass.39 The second stage is the stable adaptation period when β cells are no longer fully compensating for increased insulin resistance; thus, fasting and postload glucose values are not completely maintained. This period probably starts when fasting and postload glucose levels are still within the normal range36,39,43 and is usually accompanied by a decrease in acute insulin secretion at FPG concentrations of around 5·6 mmol/L.39 Much of
3

Pathophysiology of prediabetes
Trajectories of glycaemic changes in prediabetes
In healthy people, blood glucose is strictly regulated. FPG is maintained at 3·9–5·6 mmol/L,37 and postmeal increases rarely exceed 3 mmol/L.38 During development of type 2 diabetes, homoeostasis of fasting and postload glucose becomes abnormal.39 As proven by studies with repeat measures of glucose concentrations, insulin sensitivity, and insulin secretion, development of diabetes from normal glucose tolerance

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

San Antonio Year Country Age Sex Ethnicity BMI Waist circumference Height Family history of diabetes Systolic blood pressure HDL cholesterol Triglycerides Uric acid Antihypertensive treatment Hypertension Cardiovascular disease Use of corticosteroids Diet Physical inactivity Smoking Deprivation index Fasting glucose HbA1c 2002 USA P P P P × × P P P × × × × × × × × × × P ×

FINDRISK 2003 Finland P P × P P × × × × × × P × × × P P × × × ×

ARIC 2005 USA P P P × P P P P P P × × × × × × P × × P ×

Framingham Cambridge Offspring Risk Score 2007 USA P P × P × × P P P P × × × × × × × × × × × 2008 UK P P × P × × P × × × × P × × P × × P × × ×

QDScore 2009 UK P P P × × × × × × × × P P P P × × P P × ×

AUSDRISK 2010 Australia P P P P P × P × × × × P × × × × P P × P ×

KORA 2010 Germany P P × P × × P × × × P × P × × × × P × P P

FINDRISK=Finnish Diabetes Risk Study. ARIC=Atherosclerosis Risk in Communities study. AUSDRISK=Australian Type 2 Diabetes Risk Assessment Tool based on demographic, lifestyle, and simple anthropometric measures. KORA=Cooperative Health Research in the Region of Augsburg study. BMI=body-mass index. HbA1c=glycated haemoglobin A1c.

Table 2: Examples of externally validated diabetes risk models6

the first and second stages therefore occur before the prediabetic phase. During the unstable early decompensation period—the third stage of diabetes development—β cells become unable to compensate for insulin resistance and consequently glucose concentrations start to increase rapidly,39,43 as was seen in Whitehall II and other longitudinal studies.36,41 This period probably extends from prediabetes to manifest diabetes. The subsequent two stages of diabetes development (stable decompensation and severe decompensation) relate to manifest diabetes and thus are beyond the scope of this review.43

Glucose dysregulation
FPG values are determined by endogenous glucose production (EGP), which depends mostly on the liver. EGP and fasting insulin are used as markers of hepatic insulin resistance and show a strong relation with fasting glycaemia.38,39,45 During absorption of a glucosecontaining meal, changes in glucose concentrations are caused by intestinal absorption, suppression of EGP, and total body glucose uptake.38,39 EGP is greatly suppressed in people with normal glucose tolerance after glucose ingestion, whereas this suppression is less pronounced
4

in prediabetes and diabetes.38,39 In type 2 diabetes, total body glucose disposal is decreased, and 85–90% of this impairment is related to muscle insulin resistance.46 If insulin secretion was able to compensate for insulin resistance perfectly, no observable changes in glucose concentration would occur. This factor means that, by definition, β-cell dysfunction is already present in the prediabetic phase. However, β-cell function cannot be characterised solely on the basis of insulin secretion without consideration of underlying insulin resistance. β cells respond to an increase in glucose concentration with a rise in insulin secretion that is dependent on whole body insulin sensitivity. Accordingly, the relation between insulin secretion and insulin sensitivity is hyperbolic, and the ratio of incremental insulin to incremental glucose divided by insulin resistance is described by a constant known as the disposition index.39,47 This index, therefore, is a measure of insulin secretion after the underlying degree of insulin resistance (higher for healthy people and lower for prediabetic and diabetic individuals) has been accounted for. Studies using different measures of β-cell function have reported severely abnormal (up to 80% decreased) insulin secretion in prediabetic people.37,42,48 These findings are supported by autopsies

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

reporting a 50% decrease in β-cell volume in those with IFG.49

12

2 h postload glucose (mmol/L) Fasting glucose (mmol/L)

Insulin sensitivity (HOMA2-%S) β-cell function (HOMA2-%B)

200 180 160

IFG versus IGT
Patients with isolated IFG differ from those with isolated IGT in their fasting and 2 h postload glucose values and by the shape of their glucose concentration curves during OGTT. Both groups present with insulin resistance, but the site of their insulin resistance is different. High hepatic insulin resistance is a typical finding in patients with IFG, with almost normal values in skeletal muscle.38,39,45 In patients with IGT, the main site of insulin resistance is muscle, with only small changes in liver insulin sensitivity.37,38 This notion is supported by the finding that total body glucose disposal can gradually worsen from normal glucose tolerance to IFG to IGT and then to type 2 diabetes.48 β-cell dysfunction is present both in people with isolated IFG and those with isolated IGT. Individuals with IFG have severely impaired early insulin responses during OGTT, but their insulin secretion improves during the second phase of the test. By contrast, people with IGT present with impaired earlyphase and late-phase insulin secretion.38,39,50 These findings suggest distinct pathophysiological mechanisms of isolated IFG and isolated IGT, although the clinical relevance of these results needs further clarification.

10

Glucose (mmol/L)

8 140 6 120 4 100 2 80 60 –12 –10 –8 –6 –4 Time until diagnosis of diabetes (years) –2 0

HOMA2 (%S or %B)

0 –14

Figure 2: Fasting and 2 h postload glucose, insulin sensitivity, and β-cell function trajectories before the diagnosis of diabetes Data are from the British Whitehall II study of 505 diabetes cases.36 Time 0 is diagnosis of diabetes. The updated homoeostasis model assessment (HOMA2) was used to calculate insulin sensitivity (HOMA2-%S) and β-cell function (HOMA2-%B) trajectories. Adapted from figures 1 and 2 of Tabák and colleagues,36 by permission of Elsevier.

Nephropathy and kidney disease in prediabetes
People with prediabetes can have concomitant damage to end organs such as eyes, kidneys, blood vessels, and heart, which is traditionally thought to be a complication of diabetes. Here, we briefly review evidence for complications that are particularly relevant to prediabetes: nephropathies and chronic kidney disease; neuropathies; diabetic retinopathy; and macrovascular diseases. Prediabetes has been linked to increased risk of early forms of nephropathy and chronic kidney disease, defined by methods such as urinary albumin excretion rate and estimated glomerular filtration rate.51–55 NHANES, 1999–2006, showed that the prevalences of microalbuminuria and macroalbuminuria increase as glycaemia worsens—ie, from normoglycaemia (6% prevalence of microalbuminuria and 0·6% prevalence of macroalbuminuria), to IFG (10% and 1·1%), undiagnosed diabetes (29% and 3·3%), or diagnosed diabetes (29% and 7·7%).54 Of note, microalbuminuria can be indicative of hypertension and is therefore an imprecise marker of diabetes-related early nephropathy.54 Other data for increased albuminuria and glomerular filtration rate—an early marker of kidney involvement in hyperglycaemia— also support the notion that some nephropathic changes might be present in the prediabetic stage before onset of diabetes.51,53,56–58 By contrast, evidence of a cross-sectional association between prediabetes and decreased estimated glomerular filtration rate—a late marker of chronic kidney disease—is mixed, consisting of studies with both positive54 and null findings.55,57 Longitudinal studies

suggest that prediabetes is a risk factor for chronic kidney disease, but whether this prospective association is attributable to the effects of prediabetes itself, increased incidence of diabetes, or common causes contributing to both hyperglycaemia and kidney pathology is unclear.59,60

Neuropathies in prediabetes
The strongest supportive evidence is for the association between prediabetes and autonomic neuropathy in particular, although the method used to measure autonomic neuropathy seems to be crucial. Prediabetes is associated with decreased heart-rate variability61 (a marker of parasympathetic function),62–65 decreased postural changes in heart rate,62 increased prevalence of erectile dysfunction in men,66 and a worse profile in tests of sympathetic and parasympathetic function.67 No consistent evidence is available to suggest that prediabetes is associated with orthostatic blood pressure63 (a late marker of diabetic neuropathy61), decreased expiratory-to-inspiratory ratio, or changes in heart rate during breathing.63 Studies of prediabetes and sensorimotor neuropathy68–70 suggest that small demyelinated fibres might be implicated in IGT and early diabetic neuropathy.61 Distal intraepidermal nerve fibre density, quantitative sudomotor testing, total sweat volume and arm-to-foot sweat responses, deep tendon reflexes, and temperature sensation are sensitive markers of sensorimotor neuropathy,71,72 whereas tests such as the Michigan neuropathy screening instrument, calibrated tuning fork, and classical nerve conduction tests, and vibration and temperature perception thresholds, might not detect neuropathy in prediabetic people.
5

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

Finally, evidence is accumulating for increased prevalence of idiopathic polyneuropathy (eg, idiopathic sensory or painful neuropathy,73–78 and sensory or small-fibre-only neuropathy)73,75,78 in individuals with prediabetes, with IGT more strongly related to painful than non-painful neuropathy.73,75,78

account. Individual-level evidence from prospective studies suggests that fasting hyperglycaemia (figure 3), postload glucose, and HbA1c are all robust predictors of vascular mortality86,88,89 and, according to multivariable adjusted analyses, these associations are independent of vascular risk factors such as obesity, blood pressure, triglyceride, and lipoproteins.84,85,87

Diabetic retinopathy
Prediabetes might be associated with an increased risk of diabetic retinopathy, although findings vary depending on how diabetic retinopathy is detected.51,79–83 In a study of more than 5000 Pima Indians, retinopathy ascertained by direct ophthalmoscopy was associated with prediabetic status.51 Measures of retinal vascular changes, such as lower arteriole-to-venule ratio and increased retinal arteriole or venular calibre, have also been related to prediabetes or increased risk of diabetes, although evidence is not entirely consistent.81–83

Treatment
Lifestyle intervention
Prediabetes should be treated to prevent progression to diabetes, mitigate some of the potential results of progression to diabetes, and prevent the potential effects of prediabetes itself. Most studies in this research specialty have focused on diabetes incidence in prediabetic individuals, and support the notion that lifestyle change should be the cornerstone for diabetes prevention. The primary aim of lifestyle interventions is to prevent or delay development of type 2 diabetes and its complications13,45 by targeting obesity and physical inactivity, the two most important modifiable risk factors of diabetes development.3,25 The Finnish Diabetes Prevention Study and the US DPP (the largest so far) with a 3 year follow-up reported a 58% risk reduction after interventions aimed at weight loss, dietary change, and increased physical activity.25,28 In the first trial, benefits were dependent on the number of goals achieved by the participant (weight reduction >5%, fat intake <30%, saturated fat intake <10%, fibre intake >15 g/1000 kcal, exercise >4 h/week),28 whereas in the DPP the most important determinant of risk reduction was weight loss (every 1 kg decrease reduced risk by 16%).90 The beneficial effect of lifestyle interventions has also been confirmed in Asian populations.26,91 Successful lifestyle interventions seem to improve insulin sensitivity and β-cell function.92,93

Macrovascular disease
Prediabetes is linked with increased risks of major manifestations of vascular disease, but whether raised disease risks depend on development of clinical diabetes is unclear.84,85 Cross-sectional studies provide evidence in favour of vascular risk effects of mild or moderate hyperglycaemia because an excess prevalence of coronary disease is reported in people with fasting or postload hyperglycaemia lower than the diabetic threshold.86,87 Compared with coronary disease, less certainty exists with respect to cerebrovascular disease and aortic aneurysm.87 Diabetes is a known risk factor for ischaemic and haemorrhagic stroke, but whether risk increases before development of diabetes remains to be established.84 The dose–response effect of fasting hyperglycaemia for vascular mortality might be weaker than the effect of postload glucose. The DECODE (Diabetes Epidemiology: Collaborative analysis Of Diagnostic criteria in Europe) pooling study of European cohorts showed that IGT was associated with increased risk of coronary death and total cardiovascular death, independent of the concentration of FPG, although the converse was not the case.88 Irrespective of whether basal or challenged blood glucose concentration is more important for atherogenesis, average glucose concentrations, indexed by HbA1c concentration, predict incident coronary disease at least as well as fasting and postload glucose, although prospective studies of HbA1c are fairly rare.89 The epidemiological relation between prediabetes and macrovascular disease can be confounded by clustering of vascular risk factors within individuals. Blood glucose in the prediabetic range is correlated with many risk factors, including general and central obesity, blood pressure, and triglyceride and lipoprotein concentrations.84 As a result, the strength of the glycaemia effect depends on the extent to which related vascular risk factors are taken into
6

Pharmacological intervention based on antidiabetic drugs
Evidence of potential benefits from pharmacotherapy is accumulating. The biguanides are a class of drug that include metformin, used for decades to treat diabetes. Metformin has beneficial effects on BMI and lipid concentrations and has been proven to be safe by trial evidence showing no serious adverse effects (only minor gastrointestinal side-effects were detected).94 It reduces fasting glucose mainly through its effect on hepatic glucose output.95 According to trial evidence in people with IGT, metformin lowers risk of type 2 diabetes by 45%.96 Its effect was similar to lifestyle intervention in the Indian DPP-1 study,26 although in the US DPP it was less effective than lifestyle.25 The beneficial effect of metformin was greater in prediabetic people with a higher baseline BMI and higher FPG than in their leaner counterparts with lower FPG concentrations.25 Gastrointestinal side-effects of the drug were mostly mild to moderate, so the intervention seemed to be safe.25,26

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

Thiazolidinediones, such as troglitazone, rosiglitazone, and pioglitazone, act through the peroxisome proliferatoractivated receptor-γ by increasing hepatic and peripheral insulin sensitivity and preserving insulin secretion.45,95 Rosiglitazone was effective in a 3 year randomised trial that showed a 60% reduction in incident diabetes risk, but it was also associated with a statistically and clinically significant weight increase (roughly 2 kg) compared with placebo and increased risk of heart failure (0·1% vs 0·5% in controls).24,97 Pioglitazone showed effectiveness in the ACT NOW study29 in obese people with IGT. The risk of diabetes decreased by more than 70% and the drug was associated with improved diastolic blood pressure, improved HDL cholesterol, and a reduced rate of carotid intima-media thickening. However, weight gain was about 3 kg greater with pioglitazone than with placebo, and oedema was more frequently reported (13% vs 6%).29 A possible link between pioglitazone and bladder cancer has been suggested and therefore individuals with a history of bladder cancer or unexplained haematuria should probably not receive this drug.98,99 In the Indian DPP-2 study,100 no difference in the rate of diabetes development was noted between lifestyle intervention alone and lifestyle intervention plus pioglitazone during a 3 year trial. Two thiazolidinedione drugs were withdrawn from the European market: troglitazone for probable serious hepatotoxicity, and rosiglitazone because of possible increases in cardiovascular risk.45,95 In the CANOE (CAnadian Normoglycemia Outcomes Evaluation) trial,101 low doses of rosiglitazone (2 mg twice a day) in combination with metformin were tested against placebo to examine whether lower doses would cause reduced side-effects. The risk of incident diabetes was reduced by 66% in the active treatment group with no significant difference in weight gain compared with controls. However, more people complained of diarrhoea in the active treatment group (16% vs 6%). α-glucosidase inhibitors reduce the rate of polysaccharide digestion from the proximal small intestine. They mainly lower postprandial glucose without causing hypoglycaemia. Since their effect on HbA1c is smaller than that of other oral antidiabetic agents, they are seldom used in the treatment of type 2 diabetes.95 However, two large trials102,103 support their effectiveness in the prevention of diabetes and, importantly, one of them (STOP-NIDDM [Study to Prevent Non-InsulinDependent Diabetes Mellitus])103 shows evidence of decreased cardiovascular disease and hypertension risk in treated IGT patients. In this trial, a 25% relative-risk reduction for diabetes was reported in people with IGT who were randomly assigned to either acarbose (100 mg three times a day) or identical placebo during 3·3 years of follow-up,103,104 but almost a third of the acarbose group could not complete the trial because of gastrointestinal side-effects such as flatulence and diarrhoea.103 A recent study investigating voglibose,102 another α-glucosidase inhibitor, reported a 40% reduction in incident diabetes

2·5

History of diabetes at baseline: Yes No

2·0

Hazard ratio

1·5

1·0 0·9

0 0 3 4 5 6 7 8 9 10 Mean fasting glucose (mmol/L)

Figure 3: Hazard ratios for vascular death according to baseline concentrations of fasting glucose Data are from 50 studies and include 16 211 vascular deaths. Glucose concentrations for participants without a known history of diabetes at baseline were classified into several groups (<4·0 mmol/L; 4·0 mmol/L to <4·5 mmol/L; 4·5 mmol/L to <5·0 mmol/L; 5·0 mmol/L to <5·5 mmol/L; 5·5 mmol/L to <6·0 mmol/L; 6·0 mmol/L to <6·5 mmol/L; 6·5 mmol/L to <7·0 mmol/L; 7·0 mmol/L to <7·5 mmol/L, and ≥7.5 mmol/L). Hazard ratios are plotted against mean fasting glucose concentration in each group (reference category, 5·0 to <5·5 mmol/L). Error bars show 95% CIs. Reproduced from Seshasai and colleagues85 of the Emerging Risk Factor Collaboration, by permission of Massachusetts Medical Society.

risk during 48 weeks of follow-up in high-risk Japanese individuals with IGT. Although gastrointestinal sideeffects were similar to those reported in previous trials, more people completed that study. The glucagon-like peptide-1 analogues exenatide and liraglutide both produced sustained weight loss in obese patients and were associated with increased reversion from prediabetes to normoglycaemia during 1–2 years of follow-up. The most frequent side-effects were nausea and vomiting.105–107 A multicentre multinational study investigated the effect of nateglinide (a short-acting insulin secretagogue) in more than 9000 people with IGT and reported no effect on the rate of diabetes or cardiovascular outcomes during 6·5 year follow-up.108

Pharmacological interventions based on non-antidiabetic drugs
The anti-obesity drug orlistat is a gastrointestinal lipase inhibitor. In a post-hoc analysis of obese people,109 orlistat was associated with greater weight loss than was placebo (6·7 kg vs 3·8 kg) and significantly reduced the conversion rate from IGT to diabetes (7·6% vs 3·0%) in a 1·5 year follow-up. This finding is consistent with the
7

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

4 year XENDOS (XENical in the prevention of Diabetes in Obese Subjects) trial27 that reported a 37% reduction in relative risk of diabetes in obese people given orlistat, although in that study only 52% of people on the drug and 34% of placebo recipients completed treatment. An explanatory analysis suggested that the preventive effect was mainly confined to participants with IGT. At least one randomised 6 month trial in people with prediabetes and hypertriglyceridaemia has reported higher rates of regression to normoglycaemia in patients given fenofibrate (>50%) than in those given placebo (30%). Lipotoxicity is thought to be an important factor in development of diabetes, so these findings might have important clinical implications.110 Whether or not inhibitors of the renin–angiotensin– aldosterone system have an effect on the development of diabetes is the subject of debate. Secondary analyses of hypertension trials have suggested that people with high cardiovascular risk who receive angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers have lower incidences of diabetes than do those receiving different active treatments or placebo.111 However, these findings might be biased because the comparator active treatment groups had different proportions of other antihypertensive treatments known to increase diabetes risk (eg, β blockers and thiazide diuretics).112,113 Furthermore, in the DREAM (Diabetes REduction Assessment with ramipril and rosiglitazone Medication) trial,113 ramipril, another renin–angiotensin–aldosterone system blocker, did not significantly reduce incidence of new-onset diabetes. In view of the evidence, the effect of these drugs is much smaller than that of antidiabetic drugs, and they are not recommended for treatment of prediabetes.

The DPP outcomes study120 found that reversion from prediabetes to normoglycaemia during the randomised phase of the study, even if transient, was associated with a 56% reduced risk of future diabetes, independent of whether the reversion occurred spontaneously or during lifestyle or metformin therapy during the 5·7 year followup (hazard ratio=0·44, 95% CI 0·37–0·55; p<0·0001). In the 20 year follow-up of the DaQing Study, the evidence of intervention effects on macrovascular complications is not consistent. In a recent metaanalysis of trials in prediabetic people,116 lifestyle and drug-based interventions had no significant effect on the risk of all-cause mortality or cardiovascular death during mean follow-up of 3·8 years, except for a borderline significant reduction in stroke risk. Allcause mortality was lower in the diet and exercise intervention group than in the control group during a 12 year follow-up in the Malmö Preventive Project,121 but this study was not randomised.

Clinical and public health implications
By defining people as prediabetic (also known as intermediate hyperglycaemia or high risk for diabetes), a heterogeneous patient population is identified, characterised by the simultaneous presence of insulin resistance and β-cell dysfunction. Multifactorial diabetes risk scores are promising approaches to further improve identification of individuals at high risk of diabetes development, although whether risk scores will help prevent diabetes more than the classic definition of prediabetes is unknown. Prediabetes is not only related to an increased risk of diabetes and its complications, but also might cause damage to kidney and nerves, according to accumulating evidence. Identification and treatment of prediabetic individuals is therefore crucial. Recent evidence suggests that prevention of progression to diabetes is possible, although evidence of reduced cardiovascular disease risk is scarce. On the basis of randomised trials that show the effectiveness of lifestyle intervention and several antidiabetic drugs in the prevention of diabetes, lifestyle intervention aimed at achieving more than 7% weight reduction and 150 min per week of moderate intensity physical activity is recommended for all people with prediabetes. In view of long-standing safety information about metformin, this drug could be given to people who are unable to comply with lifestyle advice. For other potential drugs, further long-term studies are needed on safety and vascular outcomes before lifelong treatment can be safely recommended. Economic considerations are important for policy makers, public health agencies, insurers, and health-care providers and consumers, but few studies have assessed different prediabetes screening and treatment strategies in terms of cost-effectiveness and health benefits. Diabetes is projected to be one of the five leading causes of death in high-income countries by 2030 and one of the

Other treatments that reduce diabetes risk
In morbidly obese people, bariatric surgery is associated with sustained weight loss; a substantial reduction in 2 year and 10 year incidence of type 2 diabetes;114 and, in individuals with blood glucose greater than 4·5 mmol/L, a reduced risk of cardiovascular disease.115 Corresponding benefits have not been reported for other weight-loss interventions.

Long-term effects of lifestyle and antidiabetic drug interventions
Several trials support a long-term reduction in diabetes risk or a delay in onset of the disease as a result of lifestyle and drug-based interventions.15,19,116–118 In the 20 year follow-up of the DaQing Diabetes Prevention Study,119 for example, those receiving a lifestyle intervention had a 43% reduced risk of diabetes, translating to a mean 3·6 year delay in development of diabetes. In the same study, lifestyle intervention was also associated with an almost 50% reduction in relative risk of incident severe retinopathy, whereas rates of other microvascular complications, such as nephropathy and neuropathy, were similar to those seen in controls.119
8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

ten leading causes of death worldwide, which emphasises the public health importance of reducing diabetes risk at the population level. Strategies targeting interventions aimed at the entire population to reduce key diabetes risk factors, such as adiposity and physical inactivity, are important. However, these need to be complemented with diabetes prevention strategies specifically aimed at prediabetic and other high-risk individuals.
Contributors All authors contributed to the search of published work and wrote parts of the paper. AGT and MK wrote the first draft and all authors contributed to the final version. Conflicts of interest We declare that we have no conflicts of interest. Acknowledgments MK is supported by the UK Medical Research Council, the US National Institutes of Health (R01HL036310; R01AG034454), and the Academy of Finland. EB is supported by the British Heart Foundation and Stroke Association. References 1 WHO, International Diabetes Foundation. Definition and diagnosis of diabetes mellitus and intermediate hyperglycaemia: report of a WHO/IDF consultation. Geneva: World Health Organization, 2006. 2 American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2011; 34 (suppl 1): S62–69. 3 International Expert Committee. International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care 2009; 32: 1327–34. 4 Balion CM, Raina PS, Gerstein HC, et al. Reproducibility of impaired glucose tolerance (IGT) and impaired fasting glucose (IFG) classification: a systematic review. Clin Chem Lab Med 2007; 45: 1180–85. 5 DECODE Study Group. Age- and sex-specific prevalences of diabetes and impaired glucose regulation in 13 European cohorts. Diabetes Care 2003; 26: 61–69. 6 Buijsse B, Simmons RK, Griffin SJ, Schulze MB. Risk assessment tools for identifying individuals at risk of developing type 2 diabetes. Epidemiol Rev 2011; 33: 46–62. 7 Danaei G, Finucane MM, Lu Y, et al. National, regional, and global trends in fasting plasma glucose and diabetes prevalence since 1980: systematic analysis of health examination surveys and epidemiological studies with 370 country-years and 2·7 million participants. Lancet 2011; 378: 31–40. 8 Katikireddi SV, Morling JR, Bhopal R. Is there a divergence in time trends in the prevalence of impaired glucose tolerance and diabetes? A systematic review in south Asian populations. Int J Epidemiol 2011; 40: 1542–53. 9 Centers for Disease Control and Prevention. National Diabetes Fact Sheet: national estimates and general information on diabetes and prediabetes in the United States, 2011. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, 2011. 10 Cowie CC, Rust KF, Ford ES, et al. Full accounting of diabetes and pre-diabetes in the U.S. population in 1988–1994 and 2005–2006. Diabetes Care 2009; 32: 287–94. 11 International Diabetes Federation. IDF Diabetes Atlas, 5th edn. Brussels: International Diabetes Federation, 2011. 12 Forouhi NG, Luan J, Hennings S, Wareham NJ. Incidence of type 2 diabetes in England and its association with baseline impaired fasting glucose: the Ely study 1990–2000. Diabet Med 2007; 24: 200–07. 13 Nathan DM, Davidson MB, DeFronzo RA, et al, for the American Diabetes Association. Impaired fasting glucose and impaired glucose tolerance: implications for care. Diabetes Care 2007; 30: 753–59. 14 Gerstein HC, Santaguida P, Raina P, et al. Annual incidence and relative risk of diabetes in people with various categories of dysglycemia: a systematic overview and meta-analysis of prospective studies. Diabetes Res Clin Pract 2007; 78: 305–12.

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30 31

32

33

34

Diabetes Prevention Program Research Group. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 2009; 374: 1677–86. Yeboah J, Bertoni AG, Herrington DM, Post WS, Burke GL. Impaired fasting glucose and the risk of incident diabetes mellitus and cardiovascular events in an adult population: MESA (Multi-Ethnic Study of Atherosclerosis). J Am Coll Cardiol 2011; 58: 140–46. Heianza Y, Hara S, Arase Y, et al. HbA1c 5·7–6·4% and impaired fasting plasma glucose for diagnosis of prediabetes and risk of progression to diabetes in Japan (TOPICS 3): a longitudinal cohort study. Lancet 2011; 378: 147–55. Zhang X, Gregg EW, Williamson DF, et al. A1C level and future risk of diabetes: a systematic review. Diabetes Care 2010; 33: 1665–73. Li G, Zhang P, Wang J, et al. The long-term effect of lifestyle interventions to prevent diabetes in the China Da Qing Diabetes Prevention Study: a 20-year follow-up study. Lancet 2008; 371: 1783–89. Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: a systematic review. Diabetes Care 2002; 25: 1862–68. Feig DS, Zinman B, Wang X, Hux JE. Risk of development of diabetes mellitus after diagnosis of gestational diabetes. CMAJ 2008; 179: 229–34. Lauenborg J, Hansen T, Jensen DM, et al. Increasing incidence of diabetes after gestational diabetes: a long-term follow-up in a Danish population. Diabetes Care 2004; 27: 1194–99. Bellamy L, Casas J-P, Hingorani AD, Williams D. Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet 2009; 373: 1773–79. The DREAM (Diabetes REduction Assessment with ramipril and rosiglitazone Medication) Trial Investigators. Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial. Lancet 2006; 368: 1096–105. Knowler WC, Barrett-Connor E, Fowler SE, et al, for the Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393–403. Ramachandran A, Snehalatha C, Mary S, et al. The Indian Diabetes Prevention Programme shows that lifestyle modification and metformin prevent type 2 diabetes in Asian Indian subjects with impaired glucose tolerance (IDPP-1). Diabetologia 2006; 49: 289–97. Torgerson JS, Hauptman J, Boldrin MN, Sjöström L. XENical in the prevention of diabetes in obese subjects (XENDOS) study: a randomized study of orlistat as an adjunct to lifestyle changes for the prevention of type 2 diabetes in obese patients. Diabetes Care 2004; 27: 155–61. Tuomilehto J, Lindström J, Eriksson JG, et al, for the Finnish Diabetes Prevention Study Group. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001; 344: 1343–50. DeFronzo RA, Tripathy D, Schwenke DC, et al, for the ACT NOW Study. Pioglitazone for diabetes prevention in impaired glucose tolerance. N Engl J Med 2011; 364: 1104–15. Noble D, Mathur R, Dent T, Meads C, Greenhalgh T. Risk models and scores for type 2 diabetes: systematic review. BMJ 2011; 343: d7163. Tirosh A, Shai I, Tekes-Manova D, et al, for the Israeli Diabetes Research Group. Normal fasting plasma glucose levels and type 2 diabetes in young men. N Engl J Med 2005; 353: 1454–62. Rathmann W, Kowall B, Heier M, et al. Prediction models for incident type 2 diabetes mellitusin the older population: KORA S4/F4 cohort study. Diabet Med 2010; 27: 1116–23. Wilson PW, Meigs JB, Sullivan L, Fox CS, Nathan DM, D’Agostino RB Sr. Prediction of incident diabetes mellitus in middle-aged adults: the Framingham Offspring Study. Arch Intern Med 2007; 167: 1068–74. Carstensen M, Herder C, Kivimäki M, et al. Accelerated increase in serum interleukin-1 receptor antagonist starts 6 years before diagnosis of type 2 diabetes: Whitehall II prospective cohort study. Diabetes 2010; 59: 1222–27.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

9

Series

35

36

37

38 39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

Sattar N, McConnachie A, Ford I, et al. Serial metabolic measurements and conversion to type 2 diabetes in the west of Scotland coronary prevention study: specific elevations in alanine aminotransferase and triglycerides suggest hepatic fat accumulation as a potential contributing factor. Diabetes 2007; 56: 984–91. Tabák AG, Jokela M, Akbaraly TN, Brunner EJ, Kivimäki M, Witte DR. Trajectories of glycaemia, insulin sensitivity, and insulin secretion before diagnosis of type 2 diabetes: an analysis from the Whitehall II study. Lancet 2009; 373: 2215–21. Abdul-Ghani MA, Tripathy D, DeFronzo RA. Contributions of beta-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care 2006; 29: 1130–39. Ferrannini E, Gastaldelli A, Iozzo P. Pathophysiology of prediabetes. Med Clin North Am 2011; 95: 327–39, vii–viii. DeFronzo RA. Banting Lecture. From the triumvirate to the ominous octet: a new paradigm for the treatment of type 2 diabetes mellitus. Diabetes 2009; 58: 773–95. Ferrannini E, Nannipieri M, Williams K, Gonzales C, Haffner SM, Stern MP. Mode of onset of type 2 diabetes from normal or impaired glucose tolerance. Diabetes 2004; 53: 160–65. Mason CC, Hanson RL, Knowler WC. Progression to type 2 diabetes characterized by moderate then rapid glucose increases. Diabetes 2007; 56: 2054–61. Gastaldelli A, Ferrannini E, Miyazaki Y, Matsuda M, DeFronzo RA, for the San Antonio metabolism study. Beta-cell dysfunction and glucose intolerance: results from the San Antonio metabolism (SAM) study. Diabetologia 2004; 47: 31–39. Weir GC, Bonner-Weir S. Five stages of evolving beta-cell dysfunction during progression to diabetes. Diabetes 2004; 53 (suppl 3): S16–21. Polonsky KS, Given BD, Hirsch L, et al. Quantitative study of insulin secretion and clearance in normal and obese subjects. J Clin Invest 1988; 81: 435–41. DeFronzo RA, Abdul-Ghani MA. Preservation of beta-cell function: the key to diabetes prevention. J Clin Endocrinol Metab 2011; 96: 2354–66. Pendergrass M, Bertoldo A, Bonadonna R, et al. Muscle glucose transport and phosphorylation in type 2 diabetic, obese nondiabetic, and genetically predisposed individuals. Am J Physiol Endocrinol Metab 2007; 292: E92–100. Kahn SE. The relative contributions of insulin resistance and beta-cell dysfunction to the pathophysiology of type 2 diabetes. Diabetologia 2003; 46: 3–19. Ferrannini E, Balkau B, Coppack SW, et al, for the RISC Investigators. Insulin resistance, insulin response, and obesity as indicators of metabolic risk. J Clin Endocrinol Metab 2007; 92: 2885–92. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC. Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes 2003; 52: 102–10. Kanat M, Mari A, Norton L, et al. Distinct beta-cell defects in impaired fasting glucose and impaired glucose tolerance. Diabetes 2012; 61: 447–53. Gabir MM, Hanson RL, Dabelea D, et al. Plasma glucose and prediction of microvascular disease and mortality: evaluation of 1997 American Diabetes Association and 1999 World Health Organization criteria for diagnosis of diabetes. Diabetes Care 2000; 23: 1113–18. Hoehner CM, Greenlund KJ, Rith-Najarian S, Casper ML, McClellan WM. Association of the insulin resistance syndrome and microalbuminuria among nondiabetic native Americans. The Inter-Tribal Heart Project. J Am Soc Nephrol 2002; 13: 1626–34. Metcalf PA, Baker JR, Scragg RK, Dryson E, Scott AJ, Wild CJ. Microalbuminuria in a middle-aged workforce. Effect of hyperglycemia and ethnicity. Diabetes Care 1993; 16: 1485–93. Plantinga LC, Crews DC, Coresh J, et al, for the CDC CKD Surveillance Team. Prevalence of chronic kidney disease in US adults with undiagnosed diabetes or prediabetes. Clin J Am Soc Nephrol 2010; 5: 673–82. Xu M, Li XY, Wang JG, et al. Retinol-binding protein 4 is associated with impaired glucose regulation and microalbuminuria in a Chinese population. Diabetologia 2009; 52: 1511–19.

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70 71 72

73

74

75

76

77

Fujita H, Narita T, Ito S. Abnormality in urinary protein excretion in Japanese men with impaired glucose tolerance. Diabetes Care 1999; 22: 823–26. Hermans MM, Henry R, Dekker JM, et al. Estimated glomerular filtration rate and urinary albumin excretion are independently associated with greater arterial stiffness: the Hoorn Study. J Am Soc Nephrol 2007; 18: 1942–52. Melsom T, Mathisen UD, Ingebretsen OC, et al. Impaired fasting glucose is associated with renal hyperfiltration in the general population. Diabetes Care 2011; 34: 1546–51. Fox CS, Larson MG, Leip EP, Meigs JB, Wilson PW, Levy D. Glycemic status and development of kidney disease: the Framingham Heart Study. Diabetes Care 2005; 28: 2436–40. Thomas G, Sehgal AR, Kashyap SR, Srinivas TR, Kirwan JP, Navaneethan SD. Metabolic syndrome and kidney disease: a systematic review and meta-analysis. Clin J Am Soc Nephrol 2011; 6: 2364–73. Tesfaye S, Boulton AJ, Dyck PJ, et al, for the Toronto Diabetic Neuropathy Expert Group. Diabetic neuropathies: update on definitions, diagnostic criteria, estimation of severity, and treatments. Diabetes Care 2010; 33: 2285–93. Wu JS, Yang YC, Lin TS, et al. Epidemiological evidence of altered cardiac autonomic function in subjects with impaired glucose tolerance but not isolated impaired fasting glucose. J Clin Endocrinol Metab 2007; 92: 3885–89. Gerritsen J, Dekker JM, TenVoorde BJ, et al. Glucose tolerance and other determinants of cardiovascular autonomic function: the Hoorn Study. Diabetologia 2000; 43: 561–70. Singh JP, Larson MG, O’Donnell CJ, et al. Association of hyperglycemia with reduced heart rate variability (The Framingham Heart Study). Am J Cardiol 2000; 86: 309–12. Schroeder EB, Chambless LE, Liao D, et al, for the Atherosclerosis Risk in Communities (ARIC) study. Diabetes, glucose, insulin, and heart rate variability: the Atherosclerosis Risk in Communities (ARIC) study. Diabetes Care 2005; 28: 668–74. Grover SA, Lowensteyn I, Kaouache M, et al. The prevalence of erectile dysfunction in the primary care setting: importance of risk factors for diabetes and vascular disease. Arch Intern Med 2006; 166: 213–19. Putz Z, Tabák AG, Tóth N, et al. Noninvasive evaluation of neural impairment in subjects with impaired glucose tolerance. Diabetes Care 2009; 32: 181–83. Ziegler D, Rathmann W, Dickhaus T, Meisinger C, Mielck A, for the KORA Study Group. Prevalence of polyneuropathy in pre-diabetes and diabetes is associated with abdominal obesity and macroangiopathy: the MONICA/KORA Augsburg Surveys S2 and S3. Diabetes Care 2008; 31: 464–69. Ylitalo KR, Sowers M, Heeringa S. Peripheral vascular disease and peripheral neuropathy in individuals with cardiometabolic clustering and obesity: National Health and Nutrition Examination Survey 2001–2004. Diabetes Care 2011; 34: 1642–47. Bruce SG, Young TK. Prevalence and risk factors for neuropathy in a Canadian First Nation community. Diabetes Care 2008; 31: 1837–41. Smith AG, Russell J, Feldman EL, et al. Lifestyle intervention for pre-diabetic neuropathy. Diabetes Care 2006; 29: 1294–99. Grandinetti A, Chow DC, Sletten DM, et al. Impaired glucose tolerance is associated with postganglionic sudomotor impairment. Clin Auton Res 2007; 17: 231–33. Hoffman-Snyder C, Smith BE, Ross MA, Hernandez J, Bosch EP. Value of the oral glucose tolerance test in the evaluation of chronic idiopathic axonal polyneuropathy. Arch Neurol 2006; 63: 1075–79. Smith AG, Singleton JR. The diagnostic yield of a standardized approach to idiopathic sensory-predominant neuropathy. Arch Intern Med 2004; 164: 1021–25. Singleton JR, Smith AG, Bromberg MB. Increased prevalence of impaired glucose tolerance in patients with painful sensory neuropathy. Diabetes Care 2001; 24: 1448–53. Nebuchennykh M, Løseth S, Jorde R, Mellgren SI. Idiopathic polyneuropathy and impaired glucose metabolism in a Norwegian patient series. Eur J Neurol 2008; 15: 810–16. Smith AG, Ramachandran P, Tripp S, Singleton JR. Epidermal nerve innervation in impaired glucose tolerance and diabetes-associated neuropathy. Neurology 2001; 57: 1701–04.

10

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Series

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

Sumner CJ, Sheth S, Griffin JW, Cornblath DR, Polydefkis M. The spectrum of neuropathy in diabetes and impaired glucose tolerance. Neurology 2003; 60: 108–11. Algvere P, Efendić S, Luft R, Wajngot A. Retinal microangiopathy and pigment epithelial lesions in subjects with normal, borderline, and decreased oral glucose tolerance. Br J Ophthalmol 1985; 69: 416–19. Tapp RJ, Tikellis G, Wong TY, et al. Longitudinal association of glucose metabolism with retinopathy: results from the Australian Diabetes Obesity and Lifestyle (AusDiab) study. Diabetes Care 2008; 31: 1349–54. Nguyen TT, Wang JJ, Wong TY. Retinal vascular changes in pre-diabetes and prehypertension: new findings and their research and clinical implications. Diabetes Care 2007; 30: 2 708–15. Wong TY, Klein R, Sharrett AR, et al, for the ARIC Investigators. Retinal arteriolar narrowing and risk of diabetes mellitus in middle-aged persons. JAMA 2002; 287: 2528–33. Nguyen TT, Wang JJ, Islam FM, et al. Retinal arteriolar narrowing predicts incidence of diabetes: the Australian Diabetes, Obesity and Lifestyle (AusDiab) Study. Diabetes 2008; 57: 536–39. The Emerging Risk Factors Collaboration. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet 2010; 375: 2215–22. Seshasai SR, Kaptoge S, Thompson A, et al, for the Emerging Risk Factors Collaboration. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med 2011; 364: 829–41. Barr EL, Zimmet PZ, Welborn TA, et al. Risk of cardiovascular and all-cause mortality in individuals with diabetes mellitus, impaired fasting glucose, and impaired glucose tolerance: the Australian Diabetes, Obesity, and Lifestyle Study (AusDiab). Circulation 2007; 116: 151–57. Brunner EJ, Shipley MJ, Witte DR, Fuller JH, Marmot MG. Relation between blood glucose and coronary mortality over 33 years in the Whitehall Study. Diabetes Care 2006; 29: 26–31. DECODE Study Group, the European Diabetes Epidemiology Group. Glucose tolerance and cardiovascular mortality: comparison of fasting and 2-hour diagnostic criteria. Arch Intern Med 2001; 161: 397–405. Sarwar N, Aspelund T, Eiriksdottir G, et al. Markers of dysglycaemia and risk of coronary heart disease in people without diabetes: Reykjavik prospective study and systematic review. PLoS Med 2010; 7: e1000278. Hamman RF, Wing RR, Edelstein SL, et al. Effect of weight loss with lifestyle intervention on risk of diabetes. Diabetes Care 2006; 29: 2102–07. Saito T, Watanabe M, Nishida J, et al, for the Zensharen Study for Prevention of Lifestyle Diseases Group. Lifestyle modification and prevention of type 2 diabetes in overweight Japanese with impaired fasting glucose levels: a randomized controlled trial. Arch Intern Med 2011; 171: 1352–60. Snehalatha C, Mary S, Selvam S, et al. Changes in insulin secretion and insulin sensitivity in relation to the glycemic outcomes in subjects with impaired glucose tolerance in the Indian Diabetes Prevention Programme-1 (IDPP-1). Diabetes Care 2009; 32: 1796–801. Kitabchi AE, Temprosa M, Knowler WC, et al, for the Diabetes Prevention Program Research Group. Role of insulin secretion and sensitivity in the evolution of type 2 diabetes in the diabetes prevention program: effects of lifestyle intervention and metformin. Diabetes 2005; 54: 2404–14. Salpeter SR, Buckley NS, Kahn JA, et al. Meta-analysis: metformin treatment in persons at risk for diabetes mellitus. Am J Med 2008; 121: 149–57. Nathan DM, Buse JB, Davidson MB, et al, for the American Diabetes Association and the European Association for the Study of Diabetes. Medical management of hyperglycaemia in type 2 diabetes mellitus: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia 2009; 52: 17–30. Lilly M, Godwin M. Treating prediabetes with metformin: systematic review and meta-analysis. Can Fam Physician 2009; 55: 363–69.

97

98

99 100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

Dagenais GR, Gerstein HC, Holman R, et al, for the DREAM Trial Investigators. Effects of ramipril and rosiglitazone on cardiovascular and renal outcomes in people with impaired glucose tolerance or impaired fasting glucose: results of the Diabetes REduction Assessment with ramipril and rosiglitazone Medication (DREAM) trial. Diabetes Care 2008; 31: 1007–14. Piccinni C, Motola D, Marchesini G, Poluzzi E. Assessing the association of pioglitazone use and bladder cancer through drug adverse event reporting. Diabetes Care 2011; 34: 1369–71. Tseng CH. Pioglitazone and bladder cancer: a population-based study of Taiwanese. Diabetes Care 2012; 35: 278–80. Ramachandran A, Snehalatha C, Mary S, et al. Pioglitazone does not enhance the effectiveness of lifestyle modification in preventing conversion of impaired glucose tolerance to diabetes in Asian Indians: results of the Indian Diabetes Prevention Programme-2 (IDPP-2). Diabetologia 2009; 52: 1019–26. Zinman B, Harris SB, Neuman J, et al. Low-dose combination therapy with rosiglitazone and metformin to prevent type 2 diabetes mellitus (CANOE trial): a double-blind randomised controlled study. Lancet 2010; 376: 103–11. Kawamori R, Tajima N, Iwamoto Y, Kashiwagi A, Shimamoto K, Kaku K, for the Voglibose Ph-3 Study Group. Voglibose for prevention of type 2 diabetes mellitus: a randomised, double-blind trial in Japanese individuals with impaired glucose tolerance. Lancet 2009; 373: 1607–14. Chiasson J-L, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M, for the STOP-NIDDM Trial Research Group. Acarbose for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial. Lancet 2002; 359: 2072–77. Chiasson JL, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M, for the STOP-NIDDM Trial Research Group. Acarbose treatment and the risk of cardiovascular disease and hypertension in patients with impaired glucose tolerance: the STOP-NIDDM trial. JAMA 2003; 290: 486–94. Astrup A, Carraro R, Finer N, et al. Safety, tolerability and sustained weight loss over 2 years with the once-daily human GLP-1 analog, liraglutide. Int J Obes (Lond) 2011; published online Aug 16. DOI:10.1038/ijo.2011.158. Rosenstock J, Klaff LJ, Schwartz S, et al. Effects of exenatide and lifestyle modification on body weight and glucose tolerance in obese subjects with and without pre-diabetes. Diabetes Care 2010; 33: 1173–75. Astrup A, Rössner S, Van Gaal L, et al, for the NN8022-1807 Study Group. Effects of liraglutide in the treatment of obesity: a randomised, double-blind, placebo-controlled study. Lancet 2009; 374: 1606–16. Holman RR, Haffner SM, McMurray JJ, et al, for the NAVIGATOR Study Group. Effect of nateglinide on the incidence of diabetes and cardiovascular events. N Engl J Med 2010; 362: 1463–76. Heymsfield SB, Segal KR, Hauptman J, et al. Effects of weight loss with orlistat on glucose tolerance and progression to type 2 diabetes in obese adults. Arch Intern Med 2000; 160: 1321–26. Wan Q, Wang F, Wang F, et al. Regression to normoglycaemia by fenofibrate in pre-diabetic subjects complicated with hypertriglyceridaemia: a prospective randomized controlled trial. Diabet Med 2010; 27: 1312–17. Al-Mallah M, Khawaja O, Sinno M, Alzohaili O, Samra AB. Do angiotensin converting enzyme inhibitors or angiotensin receptor blockers prevent diabetes mellitus? A meta-analysis. Cardiol J 2010; 17: 448–56. McMurray JJ, Holman RR, Haffner SM, et al, for the NAVIGATOR Study Group. Effect of valsartan on the incidence of diabetes and cardiovascular events. N Engl J Med 2010; 362: 1477–90. Bosch J, Yusuf S, Gerstein HC, et al, for the DREAM Trial Investigators. Effect of ramipril on the incidence of diabetes. N Engl J Med 2006; 355: 1551–62. Sjöström L, Lindroos AK, Peltonen M, et al, for the Swedish Obese Subjects Study Scientific Group. Lifestyle, diabetes, and cardiovascular risk factors 10 years after bariatric surgery. N Engl J Med 2004; 351: 2683–93. Sjöström L, Peltonen M, Jacobson P, et al. Bariatric surgery and long-term cardiovascular events. JAMA 2012; 307: 56–65.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

11

Series

116 Hopper I, Billah B, Skiba M, Krum H. Prevention of diabetes and reduction in major cardiovascular events in studies of subjects with prediabetes: meta-analysis of randomised controlled clinical trials. Eur J Cardiovasc Prev Rehabil 2011; 18: 813–23. 117 Lindström J, Ilanne-Parikka P, Peltonen M, et al, for the Finnish Diabetes Prevention Study Group. Sustained reduction in the incidence of type 2 diabetes by lifestyle intervention: follow-up of the Finnish Diabetes Prevention Study. Lancet 2006; 368: 1673–79. 118 Gerstein HC, Mohan V, Avezum A, et al, for the DREAM On (Diabetes Reduction Assessment with Ramipril and Rosiglitazone Medication Ongoing Follow-up) Investigators. Long-term effect of rosiglitazone and/or ramipril on the incidence of diabetes. Diabetologia 2011; 54: 487–95. 119 Gong Q, Gregg EW, Wang J, et al. Long-term effects of a randomised trial of a 6-year lifestyle intervention in impaired glucose tolerance on diabetes-related microvascular complications: the China Da Qing Diabetes Prevention Outcome Study. Diabetologia 2011; 54: 300–07.

120 Perreault L, Pan Q, Mather KJ, Watson KE, Hammam RF, Kahn SE, for the Diabetes Prevention Program Research Group. Effect of regression from prediabetes to normal glucose regulation on long-term reduction in diabetes risk: results from the Diabetes Prevention Program Outcomes Study. Lancet 2012; published online June 9. DOI:10.1016/S0140-6736(12)60525-X. 121 Eriksson KF, Lindgarde F. No excess 12-year mortality in men with impaired glucose tolerance who participated in the Malmo Preventive Trial with diet and exercise. Diabetologia 1998; 41: 1010–16.

12

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60283-9

Case Report

Older Still...
Phillip Ferdinand, Timothy Warrener, Lauren Mitchell, Rasheed Zahir
Lancet 2012; 379: 2312 Department of Acute Medicine, Wolverhampton New Cross Hospital, Wolverhampton, UK (P Ferdinand MRCP, T Warrener MBChB, L Mitchell MBChB, R Zahir FRCP) Correspondence to: Dr Phillip Ferdinand, Department of Acute Medicine, Wolverhampton New Cross Hospital, Wolverhampton, West Midlands, WV10 0QP, UK [email protected]

In April, 2011, a 52-year-old man presented to us with a 7-day history of fleeting joint pain and stiffness of his hips, right elbow, left shoulder, and lumbar region. The symptoms were worst in the evening and accompanied by myalgia in his quadriceps and gastrocnemius muscles. He had been unwell for 3 weeks with a swinging fever (>39°C), sore throat, cervical lymphadenopathy, and an intermittent pink rash across his chest, arms, and thighs. He had had no previous similar episodes and no recent diarrhoea, genitourinary symptoms, or red eyes. He denied any new sexual partners or travel to the tropics. On examination, he was febrile (temperature 39·2°C). He had swelling and reduced movement of his right elbow and left shoulder and discomfort on mobilisation of his lumbar spine. A striking pink rash covered his chest, arms, and thighs. There was no palpable lymphadenopathy or abdominal mass. Initial investigations showed a white cell count of 28×10⁹/L (neutrophils 26·9×10⁹/L), C-reactive protein 224 mg/L, ESR 91 mm/h, and normal renal and hepatic function. There was a delayed rise in alkaline phosphatase (360 IU/L) and alanine aminotransferase (48 IU/L). His serum ferritin was very high at 17 000 pmol/L. Chest radiograph was unremarkable. Serial blood and urine cultures grew no organisms, and broad-spectrum antibiotics were given. We detected no anti-nuclear antibodies, rheumatoid factor, antibody to cyclic citrullinated peptide, or extractable nuclear antigens. Epstein-Barr virus, cytomegalovirus, and parvovirus serology and screening for infectious mononucleosis and malaria were also negative. Thoracic, abdominal, and pelvic CT showed no fluid collections, abscesses, or evidence of malignancy. Transthoracic echocardiography did not reveal any valvular vegetations, and MRI of his
Panel: Yamaguchi criteria for diagnosis of AOSD2 Major criteria • Fever ≥39°C for 1 week or longer • Arthralgia for 2 weeks or longer • Typical rash • Leucocytosis (>10×10⁹/L) with >80% granulocytes Minor criteria • Lymphadenopathy and/or splenomegaly • Liver dysfunction (raised serum aminotransferases) • Sore throat • Negative rheumatoid factor and antinuclear antibodies
Exclusions: infection, malignancy, and rheumatic diseases. Classification requires at least five criteria, of which two must be major.

spine showed no discitis. After exclusion of competing diagnoses, and completion of the diagnostic Yamaguchi criteria (panel), a diagnosis of adult-onset Still’s disease (AOSD) was made. Our patient was unresponsive to high-dose NSAIDs, but after treatment with prednisolone 60 mg daily for 8 weeks, his arthritic symptoms improved and the biochemical markers had normalised. He tolerated a graded reduction in prednisolone to 20 mg daily, with normal ESR and concentrations of CRP and ferritin. At last follow-up in November, 2011, he was well and had begun a phased return to work. With an incidence of 0·16 per 100 000,1 AOSD is a multi-system inflammatory disorder usually presenting in the second or third decade of life. It is diagnosed by use of the sensitive (96·2%) and specific (92·1%) Yamaguchi criteria.2 Typical features are a high swinging fever, polyarthritis, and an evanescent salmon-pink rash that is more prominent during episodes of fever. Although not part of the diagnostic criteria, ferritin concentrations are very high in many patients; together with its glycosylated form, ferritin might be important in future diagnostic criteria.3 The disease course is evenly split between monocyclic, polycyclic with remitting periods, and chronic forms.4 In mild disease, high-dose NSAIDs can be useful, but corticosteroids, combined with or substituted by methotrexate, are required in most cases. Other therapies including anakinra and tocilzumab have also been successful.5 Arthralgia, fever, and rashes are common symptoms; they have broad differential diagnoses, most of which are self-limiting or secondary to easily identifiable, common causes. Our case shows that persistence of these symptoms requires thorough investigation to facilitate diagnosis of disorders such as AOSD, which has a defined treatment that reverses otherwise disabling symptoms.
Contributors PF, TW, LM, and RZ all looked after the patient and wrote the report. Written consent to publication was obtained. References 1 Magadur-Joly G, Billaud E, Barrier J, et al. Epidemiology of adult Still’s disease: estimate of the incidence by a retrospective study in west France. Ann Rheum Dis 1995; 54: 587–90. 2 Yamaguchi M, Ohta A, Tsunematsu T, et al. Preliminary criteria for classification of adult Still’s disease. J Rheumatol 1992; 19: 424–30. 3 Fautrel B, Le M, Saint-Marcoux B, et al. Diagnostic value of ferritin and glycosylated ferritin in adult onset Still’s disease. J Rheumatol 2001; 28: 322–29. 4 Catagay Y, Gul A, Catagay A, et al. Adult-onset Still’s disease. Int J Clin Pract 2009; 63: 1050–55. 5 Fautrel B. Adult-onset Still’s disease. Best Prac Res Clin Rheumatol 2008; 22: 773–92.

2312

www.thelancet.com Vol 379 June 16, 2012

Series

Diabetes 3 Bariatric surgery for type 2 diabetes
John B Dixon, Carel W le Roux, Francesco Rubino, Paul Zimmet

Bariatric surgery provides substantial, sustained weight loss and major improvements in glycaemic control in severely obese individuals with type 2 diabetes. However, uptake of surgery in eligible patients is poor, and the barriers are difficult to surmount. We examine the indications for and efficacy and safety of conventional bariatric surgical procedures and their effect on glycaemic control in type 2 diabetes. How surgical gastrointestinal interventions achieve these changes is of great research interest, and is evolving rapidly. Old classifications about restriction and malabsorption are inadequate, and we explore understanding of putative mechanisms. Some bariatric procedures improve glycaemic control in people with diabetes beyond that expected for weight loss, and understanding this additional effect could provide insights into the pathogenesis of type 2 diabetes and assist in the development of new procedures, devices, and drugs both for obese and non-obese patients.

Published Online June 9, 2012 DOI:10.1016/S01406736(12)60401-2 This is the third in a Series of three papers about diabetes Baker IDI Heart and Diabetes Institute, Melbourne, VIC, Australia (Prof J B Dixon PhD, Prof P Zimmet FRACP); Imperial Weight Centre, Imperial College London, London, UK (C W le Roux MRCP); and Weill Cornell Medical College, Cornell University, New York, NY, USA (F Rubino MD) Correspondence to: Prof John B Dixon, Baker IDI Heart and Diabetes Institute, 75 Commercial Road, Melbourne, VIC 3004, Australia [email protected]

Introduction
The worldwide prevalence of type 2 diabetes is rising in association with an increasing frequency of overweight and obesity driven by several interactions with an obesogenic environment.1 The perceived two main factors, namely sedentary behaviour and ready access to appealing, energy-dense foods, are not the complete story. Many aspects of early-life development including in-utero factors could set the scene for lifelong difficulty with maintenance of a healthy weight.2,3 In 2011, an estimated 366 million people had diabetes, and this number is predicted to rise to 522 million by 2030.4 Prevention and treatment of the combination of obesity and type 2 diabetes—often called diabesity5,6—are important public health priorities, and substantial societal change at many levels is needed. More than 60% of patients with type 2 diabetes are obese7 and frequency increases disproportionately in those who are severely obese;7,8 thus weight loss is an attractive, but challenging, therapeutic option. Type 2 diabetes is both heterogeneous and progressive9 and to expect it to be controlled by any one therapeutic intervention, or even a combination of two or more interventions, is unrealistic.10 Traditionally used drugs such as sulphonylureas and metformin, and newer ones such as the glitazones, dipeptidyl peptidase 4 inhibitors and glucagon-like peptide 1 agonists, all have a role but do not provide adequate control in many cases.9,11 Stepwise algorithms of treatment can lead to treatment failure,12 resulting in a need to intensify treatment that could involve use of drugs that induce weight gain. Treatments that avoid this issue in severely obese patients with type 2 diabetes deserve careful consideration.13 Bariatric surgery, a form of gastrointestinal surgery that is designed to achieve and sustain substantial weight loss, effectively prevents and treats type 2 diabetes.14 Improvement or remission of diabetes after gastrectomy was initially reported more than 50 years ago.15 In 1995, Pories and colleagues16 described sustained changes in glycaemic control for up to 14 years after

gastric bypass surgery in morbidly obese patients with diabetes. Work in rodents, however, suggested that specific gastrointestinal operations can have direct, weight-independent effects on diabetes,17 thus providing an additional scientific rationale for bariatric surgery as a reasonable approach to treat the disease. The implementation of laparoscopic, minimally invasive techniques and the pronounced reduction in morbidity and mortality generated interest in surgery, leading to a Diabetes Surgery Summit of experts in Rome in 2007,18 the inclusion by the American Diabetes Association of bariatric surgery as a treatment option for diabetes in 2009,19 and an International Diabetes Federation position statement in 2011.20 Zimmet and colleagues21 pointed out that although type 2 diabetes is usually treated by physicians, surgeons can now provide successful outcomes in obese patients with type 2 diabetes. Bariatric surgery provides additional benefits through improvements in other obesity-related comorbidities—eg, dyslipidaemia and obstructive sleep apnoea.22 Additionally,
Search strategy and selection criteria We reviewed studies about conventional bariatric surgery procedures and their effect on glycaemic control in type 2 diabetes with the following inclusion criteria: studies needed to include non-surgically treated controls, outcomes had to be based on biochemical indicators of glycaemic control or mortality, and patients needed to be followed up for at least 12 months. We searched the Cochrane Library, Medline, and Embase for reports published in English between Jan 1, 1990, and Oct 31, 2011, with the search terms “bariatric”, “gastric bypass”, “gastric band”, “gastric sleeve”, “sleeve gastrectomy”, “biliopancreatic diversion”, and “duodenal switch”, in combination with the term “diabetes”. We largely selected publications from the past 5 years, but did not exclude commonly referenced and highly regarded older publications. Short-term studies and those looking at new procedures were excluded.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

1

Series

A

B

C

D

survival—specifically, reduced mortality from cardiovascular disease, cancer in women, and type 2 diabetes itself.24–26 Available analyses suggest that bariatric surgery is cost effective and, in some circumstances, reduces health-care costs.27,28 Despite these findings, surgery is rarely used; fewer than 1% of patients eligible for surgery are treated each year.29 Reasons include stigmatisation and discrimination against obese people and methods to treat obesity, professional boundaries (ie, thinking of diabetes as a medical rather than surgical disorder), little awareness of surgical options in patients and physicians, barriers to access to surgical care, cost, and concerns about effectiveness and risks.30–32 The implications of bariatric surgery as a valuable treatment option for type 2 diabetes are far reaching. The clinical effectiveness of bariatric–metabolic surgery in control of diabetes, metabolic disease, and obesity suggests the gastrointestinal tract as a target for future interventional diabetology with surgical procedures, devices, and drugs that could change the way obesity and diabetes are considered.

Conventional surgical procedures
Four bariatric procedures are used conventionally, and two dominate worldwide—the Roux-en-Y gastric bypass (figure 1A) and the laparoscopic adjustable gastric band (figure 1B). Surveys33,34 by the International Federation for the Surgery of Obesity and Metabolic Disorders showed regional and temporal trends in the choice of procedure. In the USA in 2003 and 2008, respectively, Roux-en-Y gastric bypasses accounted for 65% and 49% of procedures, and laparoscopic adjustable gastric bands 24% and 42%. By contrast, in Europe for the same years, Roux-en-Y gastric bypasses accounted for 11% and 39% of procedures, and laparoscopic adjustable gastric band surgery 64% and 43%. The biliopancreatic diversion (figure 1C) and its duodenal switch variant (figure 1D), which are truly malabsorptive procedures, have a long-established history, but use has fallen during the past decade, and they are rarely used (<2% of procedures).33,34 Sleeve gastrectomy (figure 1E) is a more recent procedure that was originally used as the first of two stages in high-risk patients undergoing biliopancreatic diversion–duodenal switch. Sleeve gastrectomy is gaining some acceptance as a standalone operation and accounts for an estimated 5–15% of all procedures; this figure is increasing.29,33,35 Few data with more than 3 years’ follow-up are available for sleeve gastrectomy. Table 1 shows the safety and effectiveness of these bariatric surgical techniques. Choice of bariatric surgical procedure depends on many factors including regional expertise and experience in the different techniques, aftercare, and the balance of effectiveness; safety; complexity; and reversibility. Additionally, patients’ factors such as general health, susceptibility to perioperative morbidity and mortality, and obesity-associated comorbidities can

E

Figure 1: Conventional surgical procedures (A) Roux-en-Y gastric bypass. (B) Laparoscopic adjustable gastric band. (C) Biliopancreatic diversion. (D) Biliopancreatic diversion with duodenal switch. (E) Sleeve gastrectomy.

health-related quality of life improves, symptoms of depression are reduced, and other psychosocial benefits are noted.23 Several studies have shown improvements in
2

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

Roux-en-Y gastric bypass Mean weight loss (%) 25–35% Excess weight* loss at 3–5 years22,35,36 (%) 60% (75% with banded bypass)

Laparoscopic adjustable gastric band 20–30% 50% 0·05–0·1% 1% 4·6% Gradual, usually maximum at 2–3 years Yes Yes

Biliopancreatic diversion with or without duodenal switch 30–40% 75% 0·75–1·0% No data available 25·6% Rapid, maximum at 1–2 years

Sleeve gastrectomy

20–30% 50–60% (few reports at ≥3 years) 0·4%† No data available 10·8% Rapid, maximum at 1–2 years

30 day postoperative 0·3–0·5% mortality35,37–39 Major 30 day morbidity37,40 Morbidity at 1 year40 Pattern of weight loss22,35,36 Long-term‡ data available24,41,42 Laparoscopic 4·8%, open 7·8% 14·9% Rapid, maximum at 1–2 years; weight regain at 3–5 years Yes

Yes No High: deficiencies in iron, vitamin B12, folate, calcium, vitamin D, copper, zinc, and fat-soluble vitamins Lifelong assessment and nutritional support Abdominal pain, staple-line leak, stomach ulcer, intestinal obstruction, gallstones, nutritional deficiency, weight regain, malabsorption, hypoalbuminaemia, excessive fat malabsorption, progressive liver damage, renal calculae

No No Moderate: deficiencies in iron, vitamin B12, folate, calcium, vitamin D, copper, and zinc Lifelong assessment and nutritional support Staple-line leak, gastro-oesophageal reflux disease, dilation of the gastric remnant, nutritional deficiency, weight regain

Evidence of improved Yes survival24,25,43 Nutritional concerns

Moderate: deficiencies in Low: deficiencies in iron, vitamin B12, folate, iron, vitamin B12, and folate calcium, vitamin D, copper, and zinc Lifelong assessment and Lifelong (high in the nutritional support first 12 months) band adjustments Abdominal pain, staple-line leak, stomach ulcer, intestinal obstruction, gallstones, nutritional deficiency, weight regain Gastric pouch dilation, erosion of band into the stomach, leaks to the gastric-band system, weight regain

Follow-up requirements Key complications

*Excess weight is defined as the weight of an individual in excess of their weight at a body-mass index of 25 kg/m2. †30-day postoperative mortality for sleeve gastrectomy is based on fewer than 1000 cases.25 ‡10 years or longer.

Table 1: Summary of conventional bariatric procedures

affect risk-to-benefit assessments. The patient’s choice after he or she has been fully informed about the procedures available is also crucial. Morbidity and mortality associated with conventional bariatric procedures are generally low and similar to those of gallbladder surgery (table 1).37,38 Long-term complications are associated with the nature of the surgery and changes to gastrointestinal physiology. Complications necessitating surgical revision are not unusual for any procedure, and correction of abnormalities or conversion to alternative procedures carries additional risks.44,45 Bariatric surgery takes time to learn and apply safely, so outcomes vary. The favourable results reported in large centres might be poorer in small centres where doctors have less experience of the procedures. Longterm monitoring and aftercare, which are recommended in all guidelines, are needed to achieve optimum results and allow early detection of complications.44

Indications
A range of national and international guidelines and position statements outline the indications for bariatric surgery in obese people with type 2 diabetes (table 2). All

include similar stipulations, including failed previous weight-loss attempts, no specific contraindication to surgery, and the patient’s commitment to long-term follow-up and aftercare. The guidelines have generally followed the 1991 National Institutes of Health guidelines,51 and have focused on body-mass index (BMI) cutoffs and the presence of comorbidities. Concern is increasing that BMI should not be a dominant feature of surgical indication because it does not adequately show individual risk and benefit.52 Interest is increasing in the use of bariatric surgery in people who are class 1 obese (ie, people with a BMI 30–35 kg/m²), and a strong evidence base is needed. The International Diabetes Federation position statement20 describes suitable criteria modifications for ethnic origin (particularly for Asian people) and people with poorly controlled diabetes, especially when associated with failed medical weight management measures and difficulty in controlling associated comorbidities such as hypertension and dyslipidaemia. Bariatric surgery has typically been judged an elective procedure, with corresponding indications defining patients who might be eligible rather than those in whom it could be the most appropriate treatment for
3

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

International Diabetes Federation,20 2011

American Diabetes Association,46 2011

Scottish Intercollegiate Bariatric Scientific Guidelines Network,47 Collaborative Group48 2010 (2007)

National Institute for Health and Clinical Excellence,49 2006

US Centers for Medicare and Medicaid Services (National Coverage Determination),50 2006 ..

Recommended BMI >40 kg/m² or BMI >35 kg/m² Surgery not prioritised for any groups to be prioritised when diabetes and other comorbidities not controlled by for surgery optimum medical treatment Eligible BMI with type 2 diabetes* BMI >35 kg/m² or BMI >30 kg/m² when diabetes and other comorbidities not controlled by optimum medical treatment Adjustment of BMI for patients of Asian ethnic origin advised

··

··

BMI >50 kg/m²

BMI >35 kg/m² if patient has type 2 BMI >35 kg/m² with one BMI >35 kg/m² with one BMI >35 kg/m2 with one weight-loss-responsive weight-loss-responsive diabetes or other comorbidity not serious weight-lossresponsive comorbidity comorbidity controlled by lifestyle and comorbidity pharmacological treatment Little evidence for BMI between 30 and 35 kg/m²—suggest use is restricted to research protocols ·· ·· Weight loss before surgery does not change eligibility

BMI >35 kg/m² with one obesity-related comorbidity and unsuccessful medical treatment for obesity Conventional surgery with the exception of sleeve gastrectomy

Comment

BMI=body-mass index. *In all guidelines, type 2 diabetes is deemed a weight-loss responsive comorbidity.

Table 2: Guidelines and position statements for eligibility for bariatric surgery in adults with type 2 diabetes

their disorder and for whom access to surgery should be prioritised as best care. Despite data for the benefits of bariatric surgery in obese people with type 2 diabetes,53 uptake among eligible patients is very low—roughly 0·5–2·0% per year. In the UK, less than 0·5% of eligible people each year receive this treatment.29 Because of the progressive nature of type 2 diabetes and its complications and comorbidities, bariatric surgery should be seen as a potential treatment to be recommended and prioritised in obese patients with the disorder, rather than as a last resort.

Benefits
When people who are clinically severely obese undergo bariatric surgery, overall mortality is reduced compared with that of community control populations receiving usual care. Specific reductions in cardiovascular disease, cancer in women, and diabetes-related mortality are the most substantial benefits reported.24–26,54 The authors of a Cochrane systematic review,55 which included patients with and without diabetes, concluded that bariatric surgery is more effective than conventional treatment in achievement and sustainment of weight loss in people who are obese. Improvements in health-related quality of life and obesity-related comorbidities are also reported, including type 2 diabetes, dyslipidaemia, and sleep apnoea. Another systematic review,56 which had less rigorous inclusion criteria than did the Cochrane review, included 103 largely observational studies reporting on the remission of clinical or biochemical manifestations of diabetes in relation to bariatric surgery. Remission was reported in 78% of patients and 62% remained in remission more than 2 years after surgery. Rates of remission were associated with the extent of weight loss, the period the patient had diabetes, and the type of surgery. However, this review had pronounced
4

limitations because remission was largely based on clinical reports rather than glycated haemoglobin A1c (HbA1c) or other biochemical outcomes, and the followup of most cohorts was poorly described. Our search of published work identified one randomised controlled trial, one large, prospective case-control study, three smaller case-control studies, and three retrospective case-control studies (table 3). Studies were heterogeneous, with different measures of primary outcome and definitions of remission. All studies showed improved outcomes in type 2 diabetes in the surgically treated groups. The large Swedish Obese Subjects casecontrol study14 provided the only high-quality long-term data and clearly showed sustained weight loss and remission of type 2 diabetes in severely obese patients who elected to have bariatric surgery when compared with well matched controls at 2 and 10 years’ follow-up. Of the patients with type 2 diabetes at baseline, 72% were in remission at 2 years and 36% were still in remission at 10 years. The sole prospective randomised controlled trial63 that investigated bariatric surgery specifically as a treatment for type 2 diabetes compared laparoscopic adjustable gastric band surgery as part of a comprehensive management programme with conventional diabetes treatment with a focus on weight loss by diet and exercise. After 2 years, remission of diabetes was significantly more common in patients who underwent surgery than in those who received standard treatment (73% vs 13%), and improvements in the biochemical components of metabolic syndrome were greater. The investigators of three studies24,25,61 reported a mortality advantage in surgically treated groups compared with matched community controls. Three other studies14,57,58 examined incident diabetes in surgical and control patients without diabetes at baseline, and all three showed a decreased frequency of diabetes in the

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

Study type

Type 2 diabetes

BMI Procedure Follow-up Surgery Control (kg/m2) (n) (n)

Primary outcome

Definition of primary outcome

Proportion reaching primary endpoint (surgery) 73%

Proportion reaching primary endpoint (control) 13%

Dixon et al,57 2008 Randomised controlled trial Sjöström et al,14 2004 Sjöström et al,14 2004 Sjöström et al,14 2004 Sjöström et al,14 2004 Adams et al,57 2010 Adams et al,58 2010 Pontiroli et al,58 2005 Pontiroli et al,58 2005

Yes (diagnosis <2 years)

30–40

LAGB

2 years

30

30

Remission

HbA1c <6·2%, FPG <7·0 mmol/L, and no hypoglycaemic drugs FPG <7 mmol/L and no hypoglycaemic drugs FPG ≥7 mmol/L or on hypoglycaemic drugs FPG <7 mmol/L and no hypoglycaemic drugs FPG ≥7 mmol/L or on hypoglycaemic drugs FPG <126 mg/dL, no hypoglycaemic drugs FPG ≥126 mg/dL, or on hypoglycaemic drugs New diagnosis of type 2 diabetes HbA1c, FPG, and medications HbA1c <6·5% Significant difference between groups Significant difference between groups

Large prospective Yes case-control Large prospective No case-control Large prospective Yes case-control Large prospective No case-control Prospective case-control Prospective case-control Prospective case-control Prospective case-control Yes No No Yes Yes Yes Yes

>35 >35 >35 >35 >35 >35 >35 >35 >35 >35 25–35

Mixed Mixed Mixed Mixed RYGB RYGB LAGB LAGB RYGB BPD BPD

2 years 2 years 10 years 10 years 2 years 2 years 4 years 4 years 1 year 10 years 1 year

342 1489 118 517 61 252 56 17 20 22 30

248 1402 84 539 50*; 64†

Remission Incident diabetes Remission Incident diabetes Remission

72% 1% 36% 8% 79% 0% 0% 45% 70% 4·9%‡ 6·5%‡

21% 8% 13% 24% 0%; 5% 5·6%; 9·5% 17% 4% 33% 7·8%‡ 7·7%‡

177*; 210† Incident diabetes 29 20 18 28 38 Incident diabetes Remission Remission HbA1c HbA1c

Hofso et al,59 2010 Prospective case-control Iaconelli et al,42 2011 Scopinaro et al,60 2011 MacDonald et al,61 1997 Leslie et al,62 2011 Prospective case-control Prospective case-matched historic controls Retrospective case-control Retrospective case-control

Yes Yes

>35 >35

RYGB RYGB

·· 2 years

154 152

78 115

Mortality/year Mortality All three ADA composite goals HbA1c <7%, LDL cholesterol <100 mg/dL, systolic blood pressure <130 mm Hg

1% 38·2%

4·5% 17·4%

BMI=body-mass index. LAGB=laparoscopic adjustable gastric band. HbA1c=glycated haemoglobin A1c. FPG=fasting plasma glucose. RYGB=Roux-en-Y gastric bypass. BPD=biliopancreatic diversion. ADA=American Diabetes Association. *Comparator group sought surgery but did not have surgery. †Comparator group did not seek surgery. ‡Data are mean achieved HbA1c concentrations.

Table 3: Studies comparing bariatric surgery with conventional medical treatment for type 2 diabetes

groups who had surgery. Few high-quality studies have compared the safety and efficacy of the conventional surgical techniques as treatments for type 2 diabetes. A broad systematic review by Buchwald and colleagues56 included three of the four commonly used procedures. The percentage of excess bodyweight lost at 2 years or more was 63% for Roux-en-Y gastric bypass, 49% for laparoscopic adjustable gastric band, and 73% for biliopancreatic diversion with or without duodenal switch, and remission of type 2 diabetes occurred in 71%, 58%, and 95% of patients, respectively. In a separate review64 of sleeve gastrectomy in patients with type 2 diabetes, a loss of 47% of excess bodyweight with a 66% remission rate was reported at a mean of 13 months’ follow-up. These results are closely similar to those of a detailed review65 of the lapararoscopic adjustable gastric band procedure in patients with type 2 diabetes. However, the absence of high-quality studies, inconsistent reporting of diabetes outcomes, and high attrition rates

are important limitations for all observational case series reported. We also identified four direct prospective comparisons of two or more of the conventional procedures in patients with type 2 diabetes and at least 12 months’ follow-up (table 4). Much variability in achievement of remission is clear; use of the American Diabetes Association consensus group’s strict definition of remission70 generated lower remission rates after all surgical procedures than did other definitions of remission. The rates therefore depend on the definition of remission and the final anatomical arrangement of the procedure. Irrespectively, all procedures allow exceptional and closely similar glycaemic control when measured on the basis of HbA1c concentrations.68 Perhaps the notion of cure or remission is premature and instead work should focus on targets for control of glycaemia, blood pressure, and lipids, and hard endpoints, which are more clinically relevant and transparent at this time. Combination of treatments is
5

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

Study type

Length Intervention 1 Intervention 2 Body-mass Type 2 diabetes of study (n) (n) index (kg/m2) 25–35 Yes 1 year RYGB (30)

Outcome definition

Achieved Achieved outcome with outcome with intervention 1 intervention 2 47%

Lee et al,66 2011

Randomised controlled trial

Sleeve Remission—ie, HbA1c 93% gastrectomy (30) <6·5%, fasting plasma glucose <7·0 mmol/L, and no oral hypoglycaemics RYGB* (45) Control—ie, normal 58% glycaemia and off oral hypoglycaemics Remission—ie, HbA1c 72% <6·0%, normal fasting glucose, and off oral hypoglycaemics 87%

Pinheiro Randomised >50 et al,67 2008* controlled trial Pournaraset et al,68 2010 Prospective comparison >35

Yes

Mean 4 years 3 years

RYGB* (55)

93%

Yes

RYGB (8)

LAGB (11)

17%

Vidal et al,69 2008

Prospective comparison

>35

Yes

1 year

RYGB (52)

Sleeve Remission— gastrectomy (39) ie, normal HbA1c concentrations and fasting plasma glucose <7 mmol/L

87%

RYGB=Roux-en-Y gastric bypass. HbA1c=glycated haemoglobin A1c. LAGB=laparoscopic adjustable gastric band. *Two variants of RYGB were compared: a biliary limb of 50 cm and Roux limb of 150 cm with a biliary limb of 100 cm and Roux limb of 250 cm.

Table 4: Studies that have directly compared two bariatric surgical procedures for remission of type 2 diabetes

not new in the management of type 2 diabetes, and after non-diversionary procedures—such as laparoscopic adjustable gastric band and sleeve gastrectomy—patients are more likely to use oral hypoglycaemic drugs than they are after diversionary procedures; poor control of blood pressure and LDL cholesterol after bariatric surgery might be related to inappropriate cessation of drug treatment.71,72 The Swedish Obese Subjects study14 described well the small long-term effect of bariatric surgery on blood pressure and LDL cholesterol concentrations.

Mechanism of action
Interest has increased rapidly in the mechanism of bariatric surgery that causes weight loss and improvements in glycaemic control in people with type 2 diabetes. These mechanisms do not support the old theory that either restriction or malabsorption is a key putative factor for the most commonly used procedures.73 Bariatric surgery is an excellent model to study integrated physiology of energy balance and weight-related metabolic disorders, including type 2 diabetes. Investigators have tried to show that bariatric surgery has effects beyond weight loss, but the beneficial physiological responses that occur after substantial weight loss have been difficult to control for. Care should be taken not to interpret reduced inflammatory, glucotoxic, and lipotoxic effects as specific mechanisms of bariatric surgery without consideration of the effects of the rapid and sustained weight loss. Usually, non-surgical weight loss leads to substantial physiological responses to reverse the process and promote weight (fat) gain.74–76 These mechanisms lead to increased hunger and reduced energy expenditure, underlying the difficulty in
6

sustainment of even slight weight loss by lifestyle and pharmacological interventions. The most striking effect of bariatric surgery is its ability to prevent such physiological responses to non-surgical weight loss. Bariatric operations result in a characteristically attenuated appetite despite reduced calorie intake in people, and some procedures increase energy expenditure in animals after weight loss. How surgery changes homoeostatic mechanisms is poorly understood. Furthermore, bariatric surgery induces rapid improvement of hyperglycaemia, reduction in hepatic insulin resistance, and improved insulin secretion, and changes food preferences.68,77,78 Although weight loss has a role in improvement of diabetes, other mechanisms also seem to be implicated and are under investigation. The non-weight-loss effect of bariatric surgery has caused interest in surgery as a treatment for people with type 2 diabetes and BMI in the overweight range.79 However, Scopinaro and colleagues60 showed that caution is needed in patients with less fat. Biliopancreatic diversion was less effective in achievement of glycaemic control in patients with diabetes with a BMI of 25–30 kg/m² than in those with a BMI of 30–35 kg/m². The role of insulin resistance could be less important than that of β-cell dysfunction and could be less responsive to gastrointestinal interventions.60

Changes in insulin resistance
Weight loss reduces insulin resistance, and bariatric surgery is the most successful way to induce and maintain weight loss.24 The procedures associated with the most weight loss have the most pronounced effects on insulin resistance.56,80,81 Reduction in peripheral insulin resistance occurs only once weight loss has been established, but

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

hepatic insulin resistance can change earlier.68,82 The acute calorie restriction immediately after bariatric surgery and before substantial weight loss improves insulin sensitivity.82,83 Thus, to establish the relative contribution of calorie restriction and factors related to surgically induced changes in gastrointestinal structure and function remains difficult. The usual return of compensatory hunger after a period of calorie restriction and weight loss does not happen with bariatric surgery. Dixon and colleagues’ double-blind crossover study84 of weight-stable patients after laparascopic adjustable gastric band surgery showed that an active band provided reduced hunger after a fast, greater early satiation after a small meal, and prolonged satiety after meals. The effect, possibly due to gentle intraluminal pressure on vagal afferent mechanoreceptors at the gastric band,85 could be the main mechanism84 that allows patients to reduce meal size without a compensatory increase in meal frequency and maintain a substantially lower energy intake than before surgery.86 Patients who have had a gastric bypass increase meal frequency, but the calorie total remains low because of reductions in meal size and increased fullness. This behaviour could be an important feature in the sustained weight loss after gastric bypass.87,88 Contrary to initial suggestions that a small gastric pouch after gastric bypass causes restriction, manometry studies have noted low pressure in the pouch, but pressure in the alimentary limb might contribute to the reduction in meal size.89 Furthermore, the exaggerated and early responses of satiety gut hormones such as peptide YY, oxyntomodulin, and glucagon-like peptide 1 are associated with rapid fullness after small meals.90–94 The relation seems causative, because blockade of the gut hormone responses specifically or together with suppression of other hormones results in increased food intake.90–95 Maintenance of weight loss is probably more important than initial weight loss for long-term glycaemic control. Glycaemic control deteriorates in bariatric surgery patients who are unable to maintain weight loss.96,97 This deterioration could be partly explained by the increased energy expenditure after gastric bypass and specifically increased diet-induced thermogenesis, which has been shown in rodents.98,99 However, caution is needed in translation of these findings to people because most human studies suggest a reduction in energy expenditure.100

secretion is pronounced, but does not seem complete.77,104 Gluconeogenesis in the gut can contribute to the early and exaggerated insulin secretion that follows food intake,105 although this finding and many others that have been reported in rodents need to be confirmed in human beings. After gastric bypass, glucose sensing can change, with rapid absorption of refined carbohydrates from the alimentary limb, which is associated with unpleasant side-effects of dumping syndrome almost immediately.106 Initially, a conditioned taste aversion was thought to develop in patients who like to eat sweet foods, and gastric bypass was deemed the treatment of choice in such people.107 However, gastric banding seems to be as suitable a procedure as gastric bypass for these patients.108 The incretin response after bariatric surgery has attracted much interest, especially since therapeutic analogues of glucagon-like peptide 1 have become available. Glucagon-like peptide 1 has an exaggerated response after Roux-en-Y gastric bypass and biliopancreatic diverion–duodenal switch, with few side-effects such as nausea. However, it might not be the only cause of increased insulin secretion.109 Other incretin hormones such as glucose-dependent insulinotropic polypeptide have also been studied, but results are controversial, possibly because of differences between procedures and difficulties with the existing assays.110,111 Investigators who have shown reductions in glucose-dependent insulinotropic polypeptide have speculated that the hormone could bring about the lipolysis noted after surgery and might not contribute to increased insulin secretion associated with bariatric surgery.111

Changes in food preference
A shift from high-glycaemic-index, high-fat foods to lowglycaemic-index, low-fat foods is beneficial for people with type 2 diabetes.112 All bariatric surgery procedures change food choices—eg, gastric banding usually limits consumption of breads and pasta,113 biliopancreatic diversion–duodenal switch reduces fat intake, and gastric bypass reduces intake of sweet and fatty foods and possibly increases vegetable consumption.100 The mechanisms causing these changes implicate the sensory, reward, and physiological domains of altered taste,114,115 but the causative anatomical and physiological factors need to be established. The frameworks that have been developed to elucidate the substantial improvements in glycaemic control after bariatric surgery might be insufficient to account for the changes in biomarkers that have been described after these procedures. Introduction of different undigested nutrients from mixed meals distally in the gut could have pronounced effects. Concentrations of plasma bile acids increase after gastric bypass,116 and could have direct effects on the farnesoid X receptor (FXR) and TGR5 (a G-protein-coupled receptor for bile acids) receptors and affect the intraluminal environment and the gut microbiota.117 The major changes to the
7

Changes in insulin secretion
The substantial reductions in lipotoxic and glucotoxic effects and decreased inflammation achieved quickly after bariatric surgery result in better β-cell function.101,102 The commonly used preoperative very low calorie diet could contribute to the improved glycaemic control noted immediately before surgery.82 Each procedure results in different insulin secretion curves, which might account for the differing frequencies of postprandial hypoglycaemia after bariatric surgery.103 Improvement in β-cell

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

A

B

C

Figure 2: Novel bariatric procedures (A) Duodenal–jejunal bypass. (B) Duodenal–jejunal bypass with sleeve gastrectomy. (C) Ileal interposition.

proportions of the gut microbiota that have been reported are largely unexplained, and any downstream effects on the metabolome could be far reaching. Branched-chain aminoacids also seem to be changed.109 Whether any of these downstream effects directly or indirectly contribute to the reduction in inflammation remains unclear, but taking all the physiological changes into account results in the most pronounced improvement in and maintenance of glycaemic control noted to date.

led to interest in the gastrointestinal tract as a rational target for the development of new treatments for obesity and diabetes. Weight-independent gastrointestinal effects also raise the possibility that interventional treatments for type 2 diabetes with little or no weight loss could be developed. Glucagon-like peptide 1 agonists are an excellent example of a gastrointestinal-based incretin treatment used in type 2 diabetes.118 Several gastrointestinal anatomical arrangements that have been associated with conventional bariatric surgery are being used to develop new procedures to treat diabetes specifically. These arrangements include removal of the gastric fundus to reduce ghrelin secretion, bypass of the duodenum and upper jejenum to provide duodenal exclusion and deliver chyme prematurely to the distal jejenum (perhaps activating foregut and hindgut mechanisms), and ileal interposition to engage hindgut mechanisms. These procedures have been used alone and in combination. Duodenal–jejunal bypass (figure 2A) is one of the first surgical interventions in people to match the impressive results of duodenal exclusion reported in a non-obese diabetic rat model.17 Human studies have varied in the extent of improvement in glycaemic control, but the need for drugs for glycaemic control has been reduced and slight weight loss has been reported in patients who have undergone the procedure.119,120 Duodenal–jejunal bypass is increasingly being done in association with a conventional sleeve gastrectomy (figure 2B) to improve weight loss, glycaemic control, and gastric emptying, and reduce the risk of stomal ulcers. This more complex procedure produces a similar anatomical arrangement as a biliopancreatic diversion–duodenal switch, but with a much longer common limb. It is an alternative to Roux-en-Y gastric bypass in high-risk Asian populations because the residual stomach is available for endoscopic screening for gastric cancer.121 Another new procedure is ileal interposition, where a length of the ileum, along with its mesentery, is moved proximally and positioned in the jejunum (figure 2C). Gastrointestinal contents therefore present to this segment of ileum prematurely. Ileal interposition alone or in combination with sleeve gastrectomy has been done in patients with type 2 diabetes who were not severely obese. Preliminary results show short-term and midterm improvements in glycaemic control and substantial weight loss, but the complexity of the surgery and longterm effectiveness and safety have caused concern.122,123

Gastrointestinal devices for obesity and diabetes
Several devices are under investigation for changes to energy balance and non-weight-loss effects on glucose tolerance. These techniques can be divided broadly into three groups by mode of action and method of placement: endoscopically placed devices that change gastric volume, shape, or transit to induce early satiation and prolong satiety; endoscopically placed endoluminal liners to

Medical, surgical, and device-based interventions in the gut
The sustained weight loss of bariatric surgery and emerging evidence that surgical manipulations of gastrointestinal anatomy can bring about weightindependent improvements in glycaemic control have
8

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

mimic the upper gastrointestinal mucosal exclusion in Roux-en-Y gastric bypass or duodenal–jejunal bypass; and laparoscopically placed electrophysiological devices with several possible mechanisms. Few devices are commercially available and only the intragastric balloon has been used extensively. A small range of intragastric balloons are available (outside the USA—none is approved by the US Food and Drug Administration). Most published work pertains to the Bioenterics Intragastric Balloon (Allergan, Irvine, CA, USA), which is also called Orbera. A systematic review124 of the balloon examined 18 studies, of varying quality, including 4877 patients, and reported a mean weight loss of 17 kg and weight-loss-related changes in comorbidity and quality of life. Reporting of complications in studies was variable, and 2–7% of balloons were removed early for intolerance. Serious complications—such as gastric perforation and intestinal obstruction—were uncommon. Balloons are placed for 6–12 months and are reported to induce loss of 10–15% of bodyweight during this period. Evidence from animals suggests that exclusion of chyme from the proximal small intestine mucosa and secretions directly and positively affects glucose homoeostasis, and this finding has prompted the development of devices aiming to mimic the foregut exclusion of bypass procedures.125 One duodenal–jejunal bypass liner is approved for use in Europe and is designed to stay in place for 12 months. EndoBarrier Gastrointestinal Liner (GI Dynamics, Lexington, MA, USA) is a 60 cm impermeable fluoropolymer sleeve that is placed endoscopically and anchored in the proximal duodenum to prevent contact of chyme with the proximal intestine—similar to duodenal–jejunal bypass surgery. Weight loss and improvements in glycaemic control in diabetes are reported but few data are available. Endoluminal devices are designed for temporary placement and removed after 6–12 months, by contrast with bariatric surgery, which is thought of as definitive treatment. Thus, the effectiveness of such devices is restricted in the management of severe obesity and diabetes, which are chronic disorders. Electrophysiological devices designed to change brain–gut signalling through mechanisms including gastric stimulation, pacing, neuromodulation, and vagal blocking are under investigation. Although they are placed laparoscopically and thought of as long-term treatment, these devices are judged less invasive than most conventional bariatric surgical procedures. Early results were promising,126 but data from randomised controlled trials were disappointing.127

therapy complements, but does not replace lifestyle, behavioural, and medical treatments. However, surgery should not be thought of as a last resort, but rather as a timely and appropriate intervention when glycaemic control is suboptimum and weight management is an issue, especially when other obesity-related comorbidities are poorly controlled. After many decades in obscurity, bariatric surgery has emerged as an impressive treatment for type 2 diabetes and provides an opportunity to explore the mechanisms implicated in provision of sustained changes in energy balance and glycaemic control. A better understanding of the role of the gastrointestinal tract in obesity and type 2 diabetes could provide a range of new therapeutic targets for surgery, devices, and drugs. Bariatric surgery is associated with important clinical and health-service challenges. Robust criteria for selection of patients and prioritisation for surgery should be established, and evidence-based matching of the patient’s profile to the procedure should be provided. Development of clinical pathways for integration of bariatric procedures into management of diabetes and obesity is needed, as are clinical studies to establish the mechanisms of surgical success and failure, the best regimens for diabetes management after bariatric surgery, and durable methods of nutritional monitoring and prevention of long-term complications of surgery. Furthermore, the durability of surgery and effects on progressive loss of β-cell function and microvascular complications and the long-term effect on macrovascular (cardiovascular) complications and overall mortality should be assessed. Finally, randomised controlled trials are needed to assess different bariatric procedures for the treatment of diabetes compared with each other and emerging nonsurgical treatments.
Contributors JBD did extensive searches and led critical analysis of published work. He was central to design and planning, collation of all authors’ contributions, writing of all sections, and editing of the report for submission. ClWR did the review of published work, wrote the section about mechanisms of action, and provided critical review; editing; and approval of the entire report. FR did the review of published work, wrote the section about novel procedures and devices, and provided critical review; editing; and approval of the entire report. PZ contributed to the initial design and planning, critical review of all sections (including writing and editing), and approval of the article. Conflicts of interest JBD has received research funding and support from Allergan and Scientific Intake. He is on the Medical Advisory Board for Nestlé Australia and has received consultancy fees from Allergan and Metagenics. FR has received research grants from Covidien and Roche. He is a member of the scientific advisory board of NGM Biopharmaceuticals. PZ has received a consultancy fee from Covidien in the past. He has received speakers’ fees from Metacure, Novo Nordisk, GlaxoSmithKline, Abbott, Eli Lilly, ResMed, and Novartis. CWlR declares that he has no conflicts of interest. Acknowledgments JBD receives a research fellowship from the National Health and Medical Research Council of Australia. We thank Toni McGee for her assistance with the collation, review, and submission of the article.

Conclusion
In view of the worldwide diabetes crisis,128 health-care professionals need an effective range of treatment options for the management of type 2 diabetes. Bariatric surgery is an additional option in obese patients. Surgical

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

9

Series

References 1 Unwin N, Gan D, Whiting D. The IDF diabetes atlas: providing evidence, raising awareness and promoting action. Diabetes Res Clin Pract 2010; 87: 2–3. 2 McAllister EJ, Dhurandhar NV, Keith SW, et al. Ten putative contributors to the obesity epidemic. Crit Rev Food Sci Nutr 2009; 49: 868–913. 3 Gluckman PD, Hanson MA. Developmental and epigenetic pathways to obesity: an evolutionary-developmental perspective. Int J Obes (Lond) 2008; 32 (suppl 7): S62–71. 4 Whiting DR, Guariguata L, Weil C, Shaw J. IDF diabetes atlas: global estimates of the prevalence of diabetes for 2011 and 2030. Diabetes Res Clin Pract 2011; 94: 311–21. 5 From the NIH: successful diet and exercise therapy is conducted in Vermont for “diabesity”. JAMA 1980; 243: 519–20. 6 Astrup A, Finer N. Redefining type 2 diabetes: ‘diabesity’ or ‘obesity dependent diabetes mellitus’? Obes Rev 2000; 1: 57–59. 7 Kramer H, Cao G, Dugas L, Luke A, Cooper R, Durazo-Arvizu R. Increasing BMI and waist circumference and prevalence of obesity among adults with type 2 diabetes: the National Health and Nutrition Examination Surveys. J Diabetes Complications 2010; 24: 368–74. 8 Gregg EW, Cheng YJ, Narayan KM, Thompson TJ, Williamson DF. The relative contributions of different levels of overweight and obesity to the increased prevalence of diabetes in the United States: 1976–2004. Prev Med 2007; 45: 348–52. 9 Nathan DM, Buse JB, Davidson MB, et al. Medical management of hyperglycaemia in type 2 diabetes mellitus: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia 2009; 52: 17–30. 10 Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res Clin Pract 2010; 87: 4–14. 11 UKPDS16. UK prospective diabetes study 16. Overview of 6 years’ therapy of type II diabetes: a progressive disease. UK Prospective Diabetes Study Group. Diabetes 1995; 44: 1249–58. 12 Liebl A, Mata M, Eschwege E. Evaluation of risk factors for development of complications in type II diabetes in Europe. Diabetologia 2002; 45: S23–28. 13 Pinkney JH, Sjöström CD, Gale EAM. Should surgeons treat diabetes in severely obese people? Lancet 2001; 357: 1357–59. 14 Sjöström L, Lindroos AK, Peltonen M, et al. Lifestyle, diabetes, and cardiovascular risk factors 10 years after bariatric surgery. N Engl J Med 2004; 351: 2683–93. 15 Friedman MN, Sancetta AJ, Magovern GJ. The amelioration of diabetes mellitus following subtotal gastrectomy. Surg Gynecol Obstet 1955; 100: 201–04. 16 Pories WJ, Swanson MS, MacDonald KG, et al. Who would have thought it? An operation proves to be the most effective therapy for adult-onset diabetes mellitus. Ann Surg 1995; 222: 339–52. 17 Rubino F, Marescaux J. Effect of duodenal-jejunal exclusion in a non-obese animal model of type 2 diabetes: a new perspective for an old disease. Ann Surg 2004; 239: 1–11. 18 Rubino F, Kaplan LM, Schauer PR, Cummings DE. The Diabetes Surgery Summit consensus conference: recommendations for the evaluation and use of gastrointestinal surgery to treat type 2 diabetes mellitus. Ann Surg 2010; 251: 399–405. 19 American Diabetes Association. Standards of medical care in diabetes—2009. Diabetes Care 2009; 32 (suppl 1): S13–61. 20 Dixon JB, Zimmet P, Alberti KG, Rubino F. Bariatric surgery: an IDF statement for obese type 2 diabetes. Diabet Med 2011; 28: 628–42. 21 Zimmet P, Alberti KGMM, Rubino F, Dixon JB. IDF’s view of bariatric surgery in type 2 diabetes. Lancet 2011; 378: 108–10. 22 Buchwald H, Avidor Y, Braunwald E, et al. Bariatric surgery: a systematic review and meta-analysis. JAMA 2004; 292: 1724–37. 23 Dixon JB, Dixon ME, O’Brien PE. Depression in association with severe obesity: changes with weight loss. Arch Intern Med 2003; 163: 2058–65. 24 Sjöström L, Narbro K, Sjöström CD, et al. Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med 2007; 357: 741–52. 25 Adams TD, Gress RE, Smith SC, et al. Long-term mortality after gastric bypass surgery. N Engl J Med 2007; 357: 753–61.

26 27

28

29

30

31

32

33 34 35

36

37

38

39

40

41

42

43 44

45 46 47

48 49

Adams TD, Hunt SC. Cancer and obesity: effect of bariatric surgery. World J Surg 2009; 33: 2028–33. Picot J, Jones J, Colquitt JL, et al. The clinical effectiveness and cost-effectiveness of bariatric (weight loss) surgery for obesity: a systematic review and economic evaluation. Health Technol Assess 2009; 13: 1–190, 215–357, iii–iv. Finkelstein EA, Allaire BT, Burgess SM, Hale BC. Financial implications of coverage for laparoscopic adjustable gastric banding. Surg Obes Relat Dis 2011; 7: 295–303. Welbourn R, Fiennes A, Kinsman R, Walton P. First registry report to March 2010: Henly-on-Thames: the United Kingdom National Bariatric Surgery Registry; 2011. Oxfordshire UK: Dendritic Clinical Systems, 2011. Hayden MJ, Dixon JB, Piterman L, O’Brien PE. Physician attitudes, beliefs and barriers towards the management and treatment of adult obesity: a literature review. AJPH 2008; 14: 9–18. Padwal RS, Damjanovic S, Schulze KM, Lewanczuk RZ, Lau DC, Sharma AM. Canadian physicians’ use of antiobesity drugs and their referral patterns to weight management programs or providers: the SOCCER Study. J Obes 2011; 2011: 686521. Afonso BB, Rosenthal R, Li KM, Zapatier J, Szomstein S. Perceived barriers to bariatric surgery among morbidly obese patients. Surg Obes Relat Dis 2010; 6: 16–21. Buchwald H, Oien DM. Metabolic/bariatric surgery worldwide 2008. Obes Surg 2009; 19: 1605–11. Buchwald H, Williams SE. Bariatric surgery worldwide 2003. Obes Surg 2004; 14: 1157–64. Brethauer SA, Hammel JP, Schauer PR. Systematic review of sleeve gastrectomy as staging and primary bariatric procedure. Surg Obes Relat Dis 2009; 5: 469–75. O’Brien PE, McPhail T, Chaston TB, Dixon JB. Systematic review of medium-term weight loss after bariatric operations. Obes Surg 2006; 16: 1032–40. Flum DR, Belle SH, King WC, et al. Perioperative safety in the longitudinal assessment of bariatric surgery. N Engl J Med 2009; 361: 445–54. Buchwald H, Estok R, Fahrbach K, Banel D, Sledge I. Trends in mortality in bariatric surgery: a systematic review and meta-analysis. Surgery 2007; 142: 621–32. Chapman A, Kiroff G, Game P, et al. Laparoscopic adjustable gastric banding in the treatment of obesity: a systematic review. Surgery 2004; 135: 326–51. DeMaria EJ, Pate V, Warthen M, Winegar DA. Baseline data from American Society for Metabolic and Bariatric Surgery–designated Bariatric Surgery Centers of Excellence using the bariatric outcomes longitudinal database. Surg Obes Relat Dis 2010; 6: 347–55. Favretti F, Segato G, Ashton D, et al. Laparoscopic adjustable gastric banding in 1,791 consecutive obese patients: 12-year results. Obes Surg 2007; 17: 168–75. Iaconelli A, Panunzi S, De Gaetano A, et al. Effects of bilio-pancreatic diversion on diabetic complications: a 10-year follow-up. Diabetes Care 2011; 34: 561–67. Peeters A, O’Brien PE, Laurie C, et al. Substantial intentional weight loss and mortality in the severely obese. Ann Surg 2007; 246: 1028–33. Mechanick JI, Kushner RF, Sugerman HJ, et al. American Association of Clinical Endocrinologists, the Obesity Society, and American Society for Metabolic and Bariatric Surgery medical guidelines for clinical practice for the perioperative nutritional, metabolic, and nonsurgical support of the bariatric surgery patient. Endocr Pract 2008; 14 (suppl 1): 1–83. Deitel M, Gagner M, Dixon JB, Himpens J, Madan AK, eds. Handbook of obesity surgery. Toronto: FD-Communications Inc, 2010. American Diabetes Association. Standards of medical care in diabetes—2011. Diabetes Care 2011; 34 (suppl 1): S11–61. Logue J, Thompson L, Romanes F, Wilson DC, Thompson J, Sattar N. Management of obesity: summary of SIGN guideline. BMJ 2010; 340: c154. Fried M, Hainer V, Basdevant A, et al. Interdisciplinary guidelines on surgery of severe obesity. Int J Obes (London) 2007; 31: 569–77. NICE. Obesity: guidance on the prevention, identification, assessment and management of overweight and obesity in adults and children. London: National Institute for Health and Clinical Excellence, 2006.

10

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Series

50

51

52

53

54

55 56

57

58

59

60

61

62

63

64

65

66

67

68

69

70 71

Centers for Medicare and Medicaid Services. Bariatric surgery CMS national coverage determination. http://www.cms.gov/ medicare–coverage–database/details/ncd–details.aspx?NCDId= 57&ncdver=3&NCAId=160&ver=30&NcaName=Bariatric+ Surgery+for+the+Treatment+of+Morbid+Obesity+(1st+Recon)& bc=BEAAAAAAEAAA& (accessed Nov 13, 2011). NIH. Gastrointestinal surgery for severe obesity: National Institutes of Health Consensus Development Conference Statement. Am J Clin Nutr 1992; 55 (suppl): 615S–19S. Pories WJ, Dohm LG, Mansfield CJ. Beyond the BMI: the search for better guidelines for bariatric surgery. Obesity (Silver Spring) 2010; 18: 865–71. Vetter ML, Cardillo S, Rickels MR, Iqbal N. Narrative review: effect of bariatric surgery on type 2 diabetes mellitus. Ann Intern Med 2009; 150: 94–103. Sjöström L, Gummesson A, Sjöström CD, et al. Effects of bariatric surgery on cancer incidence in obese patients in Sweden (Swedish Obese Subjects Study): a prospective, controlled intervention trial. Lancet Oncol 2009; 10: 653–62. Colquitt JL, Picot J, Loveman E, Clegg AJ. Surgery for obesity. Cochrane Database Syst Rev 2009; 2: CD003641. Buchwald H, Estok R, Fahrbach K, et al. Weight and type 2 diabetes after bariatric surgery: systematic review and meta-analysis. Am J Med 2009; 122: 248–56 e5. Adams TD, Pendleton RC, Strong MB, et al. Health outcomes of gastric bypass patients compared to nonsurgical, nonintervened severely obese. Obesity (Silver Spring) 2010; 18: 121–30. Pontiroli AE, Folli F, Paganelli M, et al. Laparoscopic gastric banding prevents type 2 diabetes and arterial hypertension and induces their remission in morbid obesity: a 4-year case-controlled study. Diabetes Care 2005; 28: 2703–09. Hofso D, Nordstrand N, Johnson LK, et al. Obesity-related cardiovascular risk factors after weight loss: a clinical trial comparing gastric bypass surgery and intensive lifestyle intervention. Eur J Endocrinol 2010; 163: 735–45. Scopinaro N, Adami GF, Papadia FS, et al. Effects of biliopanceratic diversion on type 2 diabetes in patients with BMI 25 to 35. Ann Surg 2011; 253: 699–703. MacDonald KG Jr, Long SD, Swanson MS, et al. The gastric bypass operation reduces the progression and mortality of non-insulin-dependent diabetes mellitus. J Gastrointest Surg 1997; 1: 213–20. Leslie DB, Dorman RB, Serrot FJ, et al. Efficacy of the Roux-en-Y gastric bypass compared to medically managed controls in meeting the American Diabetes Association composite end point goals for management of type 2 diabetes mellitus. Obes Surg 2012; 22: 367–74. Dixon JB, O’Brien PE, Playfair J, et al. Adjustable gastric banding and conventional therapy for type 2 diabetes: a randomized controlled trial. JAMA 2008; 299: 316–23. Gill RS, Birch DW, Shi X, Sharma AM, Karmali S. Sleeve gastrectomy and type 2 diabetes mellitus: a systematic review. Surg Obes Relat Dis 2010; 6: 707–13. Dixon JB, Murphy DK, Segel JE, Finkelstein EA. Impact of laparoscopic adjustable gastric banding on type 2 diabetes. Obes Rev 2012; 13: 57–67. Lee WJ, Chong K, Ser KH, et al. Gastric bypass vs sleeve gastrectomy for type 2 diabetes mellitus: a randomized controlled trial. Arch Surg 2011; 146: 143–48. Pinheiro JS, Schiavon CA, Pereira PB, Correa JL, Noujaim P, Cohen R. Long-long limb Roux-en-Y gastric bypass is more efficacious in treatment of type 2 diabetes and lipid disorders in super-obese patients. Surg Obes Relat Dis 2008; 4: 521–27. Pournaras DJ, Osborne A, Hawkins SC, et al. Remission of type 2 diabetes after gastric bypass and banding: mechanisms and 2 year outcomes. Ann Surg 2010; 252: 966–71. Vidal J, Ibarzabal A, Romero F, et al. Type 2 diabetes mellitus and the metabolic syndrome following sleeve gastrectomy in severely obese subjects. Obes Surg 2008; 18: 1077–82. Buse JB, Caprio S, Cefalu WT, et al. How do we define cure of diabetes? Diabetes Care 2009; 32: 2133–35. Serrot FJ, Dorman RB, Miller CJ, et al. Comparative effectiveness of bariatric surgery and nonsurgical therapy in adults with type 2 diabetes mellitus and body mass index <35 kg/m2. Surgery 2011; 150: 684–91.

72

73 74 75

76

77

78

79

80 81

82

83

84

85

86

87

88

89

90

91

92

93

94

Pournaras DJ, Aasheim ET, Sovik TT, et al. Effect of the definition of type II diabetes remission in the evaluation of bariatric surgery for metabolic disorders. Br J Surg 2012; 99: 100–03. O’Brien PE. Bariatric surgery: mechanisms, indications and outcomes. J Gastroenterol Hepatol 2010; 25: 1358–65. Rosenbaum M, Kissileff HR, Mayer LE, Hirsch J, Leibel RL. Energy intake in weight-reduced humans. Brain Res 2010; 1350: 95–102. Rosenbaum M, Hirsch J, Gallagher DA, Leibel RL. Long-term persistence of adaptive thermogenesis in subjects who have maintained a reduced body weight. Am J Clin Nutr 2008; 88: 906–12. Sumithran P, Prendergast LA, Delbridge E, et al. Long-term persistence of hormonal adaptations to weight loss. N Engl J Med 2011; 365: 1597–604. Nannipieri M, Mari A, Anselmino M, et al. The role of beta-cell function and insulin sensitivity in the remission of type 2 diabetes after gastric bypass surgery. J Clin Endocrinol Metab 2011; 96: e1372–79. le Roux CW, Bueter M, Theis N, et al. Gastric bypass reduces fat intake and preference. Am J Physiol Regul Integr Comp Physiol 2011; 301: R1057–66. Reis CE, Alvarez-Leite JI, Bressan J, Alfenas RC. Role of bariatric-metabolic surgery in the treatment of obese type 2 diabetes with body-mass index <35 kg/m2: a literature review. Diabetes Technol Ther 2011; 14: 41–47. Mingrone G, Castagneto M. Bariatric surgery: unstressing or boosting the beta-cell? Diabetes Obes Metab 2009; 11 (suppl 4): 130–42. Mari A, Manco M, Guidone C, et al. Restoration of normal glucose tolerance in severely obese patients after bilio-pancreatic diversion: role of insulin sensitivity and beta cell function. Diabetologia 2006; 49: 2136–43. Lim EL, Hollingsworth KG, Aribisala BS, Chen MJ, Mathers JC, Taylor R. Reversal of type 2 diabetes: normalisation of beta cell function in association with decreased pancreas and liver triacylglycerol. Diabetologia 2011; 54: 2506–14. Isbell JM, Tamboli RA, Hansen EN, et al. The importance of caloric restriction in the early improvements in insulin sensitivity after Roux-en-Y gastric bypass surgery. Diabetes Care 2010; 33: 1438–42. Dixon AF, Dixon JB, O’Brien PE. Laparoscopic adjustable gastric banding induces prolonged satiety: a randomized blind crossover study. J Clin Endocrinol Metab 2005; 90: 813–19. Burton PR, Brown W, Laurie C, et al. The effect of laparoscopic adjustable gastric bands on esophageal motility and the gastroesophageal junction: analysis using high-resolution video manometry. Obes Surg 2009; 19: 905–14. Colles SL, Dixon JB, O’Brien PE. Hunger control and regular physical activity facilitate weight loss after laparoscopic adjustable gastric banding. Obes Surg 2008; 18: 833–40. Laurenius A, Larsson I, Bueter M, et al. Changes in eating behaviour and meal pattern following Roux-en-Y gastric bypass. Int J Obes (Lond) 2012; 36: 348–55. Borg CM, le Roux CW, Ghatei MA, Bloom SR, Patel AG, Aylwin SJ. Progressive rise in gut hormone levels after Roux-en-Y gastric bypass suggests gut adaptation and explains altered satiety. Br J Surg 2006; 93: 210–15. Bjorklund P, Laurenius A, Een E, Olbers T, Lonroth H, Fandriks L. Is the Roux limb a determinant for meal size after gastric bypass surgery? Obes Surg 2010; 20: 1408–14. le Roux CW, Aylwin SJ, Batterham RL, et al. Gut hormone profiles following bariatric surgery favor an anorectic state, facilitate weight loss, and improve metabolic parameters. Ann Surg 2006; 243: 108–14. le Roux CW, Welbourn R, Werling M, et al. Gut hormones as mediators of appetite and weight loss after Roux-en-Y gastric bypass. Ann Surg 2007; 246: 780–85. Korner J, Inabnet W, Febres G, et al. Prospective study of gut hormone and metabolic changes after adjustable gastric banding and Roux-en-Y gastric bypass. Int J Obes (Lond) 2009; 33: 786–95. Van der Schueren BJ, Homel P, Alam M, et al. Magnitude and variability of the glucagon-like peptide-1 response in patients with type 2 diabetes up to 2 years following gastric bypass surgery. Diabetes Care 2012; 35: 42–46. Laferrere B, Swerdlow N, Bawa B, et al. Rise of oxyntomodulin in response to oral glucose after gastric bypass surgery in patients with type 2 diabetes. J Clin Endocrinol Metab 2010; 95: 4072–76.

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

11

Series

95

96

97

98 99

100

101

102 103

104

105

106 107

108

109

110

111

Fenske WK, Bueter M, Miras AD, Ghatei MA, Bloom SR, le Roux CW. Exogenous peptide YY3-36 and Exendin-4 further decrease food intake, whereas octreotide increases food intake in rats after Roux-en-Y gastric bypass. Int J Obes (Lond) 2012; 36: 379–84. DiGiorgi M, Rosen DJ, Choi JJ, et al. Re-emergence of diabetes after gastric bypass in patients with mid- to long-term follow-up. Surg Obes Relat Dis 2010; 6: 249–53. Chikunguwo SM, Wolfe LG, Dodson P, et al. Analysis of factors associated with durable remission of diabetes after Roux-en-Y gastric bypass. Surg Obes Relat Dis 2010; 6: 254–59. Bueter M, Lowenstein C, Olbers T, et al. Gastric bypass increases energy expenditure in rats. Gastroenterology 2010; 138: 1845–53. Stylopoulos N, Hoppin AG, Kaplan LM. Roux-en-Y gastric bypass enhances energy expenditure and extends lifespan in diet-induced obese rats. Obesity (Silver Spring) 2009; 17: 1839–47. Olbers T, Bjorkman S, Lindroos A, et al. Body composition, dietary intake, and energy expenditure after laparoscopic Roux-en-Y gastric bypass and laparoscopic vertical banded gastroplasty: a randomized clinical trial. Ann Surg 2006; 244: 715–22. Weir GC, Marselli L, Marchetti P, Katsuta H, Jung MH, Bonner-Weir S. Towards better understanding of the contributions of overwork and glucotoxicity to the beta-cell inadequacy of type 2 diabetes. Diabetes Obes Metab 2009; 11 (suppl 4): 82–90. Wajchenberg BL. Beta-cell failure in diabetes and preservation by clinical treatment. Endocr Rev 2007; 28: 187–218. Cummings DE. Endocrine mechanisms mediating remission of diabetes after gastric bypass surgery. Int J Obes (Lond) 2009; 33 (suppl 1): S33–40. Camastra S, Gastaldelli A, Mari A, et al. Early and longer term effects of gastric bypass surgery on tissue-specific insulin sensitivity and beta cell function in morbidly obese patients with and without type 2 diabetes. Diabetologia 2011; 54: 2093–102. Troy S, Soty M, Ribeiro L, et al. Intestinal gluconeogenesis is a key factor for early metabolic changes after gastric bypass but not after gastric lap-band in mice. Cell Metab 2008; 8: 201–11. Ritz P, Hanaire H. Post-bypass hypoglycaemia: a review of current findings. Diabetes Metab 2011; 37: 274–81. Sugerman HJ, Starkey JV, Birkenhauer R. A randomized prospective trial of gastric bypass versus vertical banded gastroplasty for morbid obesity and their effects on sweets versus non-sweets eaters. Ann Surg 1987; 205: 613–24. Hudson SM, Dixon JB, O’Brien PE. Sweet eating is not a predictor of outcome after lap-band placement. Can we finally bury the myth? Obes Surg 2002; 12: 789–94. Laferrere B, Reilly D, Arias S, et al. Differential metabolic impact of gastric bypass surgery versus dietary intervention in obese diabetic subjects despite identical weight loss. Sci Transl Med 2011; 3: 80re2. Laferrere B, Teixeira J, McGinty J, et al. Effect of weight loss by gastric bypass surgery versus hypocaloric diet on glucose and incretin levels in patients with type 2 diabetes. J Clin Endocrinol Metab 2008; 93: 2479–85. Korner J, Bessler M, Inabnet W, Taveras C, Holst JJ. Exaggerated glucagon-like peptide-1 and blunted glucose-dependent insulinotropic peptide secretion are associated with Roux-en-Y gastric bypass but not adjustable gastric banding. Surg Obes Relat Dis 2007; 3: 597–601.

112 Thomas DE, Elliott EJ. The use of low-glycaemic index diets in diabetes control. Br J Nutr 2010; 104: 797–802. 113 Favretti F, O’Brien PE, Dixon JB. Patient management after lap-band placement. Am J Surg 2002; 184: S38–41. 114 Bueter M, Miras AD, Chichger H, et al. Alterations of sucrose preference after Roux-en-Y gastric bypass. Physiol Behav 2011; 104: 709–21. 115 Spector AC. Linking gustatory neurobiology to behavior in vertebrates. Neurosci Biobehav Rev 2000; 24: 391–416. 116 Patti ME, Houten SM, Bianco AC, et al. Serum bile acids are higher in humans with prior gastric bypass: potential contribution to improved glucose and lipid metabolism. Obesity (Silver Spring) 2009; 17: 1671–77. 117 Li JV, Ashrafian H, Bueter M, et al. Metabolic surgery profoundly influences gut microbial-host metabolic cross-talk. Gut 2011; 60: 1214–23. 118 Amori RE, Lau J, Pittas AG. Efficacy and safety of incretin therapy in type 2 diabetes: systematic review and meta-analysis. JAMA 2007; 298: 194–206. 119 Lee HC, Kim MK, Kwon HS, Kim E, Song KH. Early changes in incretin secretion after laparoscopic duodenal-jejunal bypass surgery in type 2 diabetic patients. Obes Surg 2010; 20: 1530–35. 120 Ferzli GS, Dominique E, Ciaglia M, Bluth MH, Gonzalez A, Fingerhut A. Clinical improvement after duodenojejunal bypass for nonobese type 2 diabetes despite minimal improvement in glycemic homeostasis. World J Surg 2009; 33: 972–79. 121 Kasama K, Tagaya N, Kanahira E, et al. Has laparoscopic bariatric surgery been accepted in Japan? The experience of a single surgeon. Obes Surg 2008; 18: 1473–78. 122 DePaula AL, Stival A, Halpern A, Vencio S. Thirty-day morbidity and mortality of the laparoscopic ileal interposition associated with sleeve gastrectomy for the treatment of type 2 diabetic patients with BMI <35: an analysis of 454 consecutive patients. World J Surg 2011; 35: 102–08. 123 Depaula AL, Stival AR, Halpern A, Vencio S. Surgical treatment of morbid obesity: mid-term outcomes of the laparoscopic ileal interposition associated to a sleeve gastrectomy in 120 patients. Obes Surg 2011; 21: 668–75. 124 Dumonceau JM. Evidence-based review of the Bioenterics intragastric balloon for weight loss. Obes Surg 2008; 18: 1611–17. 125 Rubino F, Forgione A, Cummings DE, et al. The mechanism of diabetes control after gastrointestinal bypass surgery reveals a role of the proximal small intestine in the pathophysiology of type 2 diabetes. Ann Surg 2006; 244: 741–49. 126 Cigaina V. Long-term follow-up of gastric stimulation for obesity: the Mestre 8-year experience. Obes Surg 2004; 14 (suppl 1): S14–22. 127 Shikora SA, Bergenstal R, Bessler M, et al. Implantable gastric stimulation for the treatment of clinically severe obesity: results of the SHAPE trial. Surg Obes Relat Dis 2009; 5: 31–37. 128 UN News Centre. UN launches global campaign to curb death toll from non-communicable diseases. http://www.un.org/apps/news/ story.asp?NewsID=39600&Cr=non–communicable+diseases&Cr1= &Kw1=summit&Kw2=chronic&Kw3= (accessed Jan 1, 2012).

12

www.thelancet.com Published online June 9, 2012 DOI:10.1016/S0140-6736(12)60401-2

Clinical Picture

Multiple endocrine carcinomas of the ileum
Lilian Schwarz, Michel Scotté

A 79-year-old man presented with overt gastrointestinal bleeding. Gastroscopy and colonoscopy did not show the origin of the bleeding. Video capsule endoscopy showed several ileal submucosal tumours with a diameter of 1·0 cm to 1∙5 cm (figure A). One of these lesions was the source of the bleeding (figure A, arrow). A CT scan with enteroclysis (fluoroscopic imaging of the small intestine) showed multiple nodular lesions in the distal ileum, clustered like a series of beads (figure B). A 130 cm ileal resection was done. Pathological examination showed 50 tumours (figure C), which were well-differentiated and had been producing serotoninA

chromogranin-synaptophysin. Plasma-chromogranin A and urinary 5-hydroxyindoleacetic acid serum concentrations were assessed, and somatostatin receptor scintigraphy was done, all of which were normal. Postoperative recovery was uneventful, with no recurrent bleeding. Endocrine tumours of the small bowel are multiple in less than 40% of cases. In most patients, no more than two or three tumours are found. After surgical resection of small bowel endocrine tumours, reported 5-year overall survival rates range from 50% to 65%, which are worse for jejuno-ileal locations and for multiple tumours.

Lancet 2012; 379: e55 Published Online March 9, 2012 DOI:10.1016/S01406736(11)61611-5 Department of Digestive Surgery, Rouen University Hospital, Rouen, France (L Schwarz, Prof M Scotté MD) Correspondence to: Prof Michel Scotté, Department of Digestive Surgery, Charles Nicolle Hospital, 1 rue de Germont, 76031 Rouen, France [email protected]

B

C

Figure: Multiple endocrine carcinomas of the ileum (A) Video capsule endoscopy showing a bleeding tumour (arrow); (B) CT showing multiple nodular lesions (arrows); (C) resected ileum and tumours (arrow).

www.thelancet.com Vol 379 June 16, 2012

e55