Diabetes mellitus is a complex metabolic disorder associated with an increased risk of microvascular and macrovascular disease; its main clinical characteristic is hyperglycaemia. The last century has been characterised by remarkable advances in our understanding of the mechanisms leading to hyperglycaemia. The central role of insulin in glucose metabolism regulation was clearly demonstrated during the early 1920s, when Banting, Best, Collip and Macleod successfully reduced blood glucose levels and glycosuria in a patient treated with a substance purified from bovine pancreata. Later, during the mid-1930s, clinical observations suggested a possible distinction between ‘insulin-sensitive’ and ‘insulin-insensitive’ diabetes. Only during the 1950s, when a reliable measure of circulating insulin was available, was it possible to translate these clinical observations into pathophysiological and biochemical differences, and the terms ‘insulin-dependent’ (indicating undetectable insulin levels) and ‘non-insulin-dependent’ (normal or high insulin levels) started to emerge. The next 30 years were characterised by pivotal progress in the field of immunology that were instrumental in demonstrating an immune-mediated loss of insulin-secreting β-cells in subjects with ‘insulin-dependent’ diabetes. At the same time, new experimental techniques allowing measurement of insulin ‘impedance’ showed a reduced peripheral effect of insulin in subjects with ‘non-insulin-dependent’ diabetes (insulin resistance). The difference between the two types of diabetes emerging from decades of observations and experiments was further formally recognised in 1979, when the definitions ‘type I’ and ‘type II’ diabetes were introduced to replace the former ‘insulin-dependent’ and ‘non-insulin-dependent’ terms. In the following years, many studies elucidated the natural history and temporal contribution of insulin resistance and β-cell insulin secretion in ‘type II’ diabetes. Furthermore, a central role for insulin resistance in the development of a cluster of cardiometabolic alterations (dyslipidaemia, inflammation, high blood pressure) was suggested. Possibly as a consequence of the secular changes in diabetes risk factors, in the last 10 years the limitation of a simple distinction between ‘type I’ and ‘type II’ diabetes has been increasingly recognised, with subjects showing the coexistence of insulin resistance and immune activation against β-cells. With the advancement of our cellular and molecular understanding of diabetes, a more pathophysiological classification that overcomes the historical and simple ‘glucocentric’ view could result in a better patient phenotyping and therapeutic approach.
- DIABETES & ENDOCRINOLOGY
- HISTORY (see Medical History)
Statistics from Altmetric.com
Background: the burden of diabetes mellitus
Diabetes mellitus is a complex metabolic disorder whose main clinical and diagnostic feature is hyperglycaemia.1 Diabetes has reached epidemic proportions, affecting around 387 million people worldwide. Over the next 20 years, its prevalence is predicted to double, and more than half-a billion people will be affected.2 Estimated regional diabetes prevalence ranges from 5.1% in Africa to 11.4% in North America and the Caribbean, with more than 75% of subjects living in low- and middle-income countries.2 Moreover, the increase in prevalence is estimated to be greater in developing areas,3 as many countries adopt Western lifestyle habits (sedentary behaviour, lack of physical activity and energy-dense diet) which are well-recognised risk factors for type 2 diabetes mellitus (T2DM).4
Although much progress has been made in the identification of risk factors associated with diabetes, in particular T2DM, its health and socioeconomic impact is increasing, mainly because of its associated complications.5 Diabetes (particularly T2DM) approximately doubles the risk of a wide range of cardiovascular diseases, including coronary heart disease and stroke.6 Moreover, T2DM is also associated with a wide range of non-vascular diseases, including cancer, mental and nervous system disorders, infections and liver disease.7 Similarly, type 1 diabetes mellitus (T1DM; ∼10% of all diabetes cases) is associated with an increased risk of both vascular and non-vascular complications.8 ,9 Therefore, a better understanding of the mechanisms that result in hyperglycaemia could help to identify potential therapeutic targets to curb diabetes and its related complications.
The aim of this review is to summarise historical developments in our understanding of the pathophysiology of T1DM and T2DM from epidemiological, clinical and biological studies, describing the extraordinary progress made in the last century.
Pathophysiology of diabetes mellitus
Initial discoveries in the pathophysiology of diabetes mellitus are intrinsically linked to polyuria, historically considered to be its main (and diagnostic) characteristic. The term ‘diabetes’ is derived from the ancient Greek word ‘diabainen’, meaning ‘go through’, to indicate the excessive passing of urine through the kidney.10 It was not until the 1600s, however, that Willis added the term ‘mellitus’ (‘sweet’) to distinguish this condition from an excessive production of non-sweet urine (diabetes ‘insipidus’).10 Almost 200 years later (1776), Dobson demonstrated that the sweet taste of urine was due to an excess of sugar in the urine and blood.11
Another 100 years were necessary to clarify the pathogenesis of diabetes mellitus. In 1889, Minkowski and von Mering found that pancreatectomised dogs developed symptoms of diabetes, thus linking diabetes for the first time to a specific organ.12 In 1910, Sharpey-Schafer suggested that people with diabetes were deficient in a substance produced in the pancreatic islets (discovered in 1869 by Langerhans) and called it ‘insulin’; therefore, a link between the pancreas, insulin and diabetes was starting to emerge and form the basis of the modern understanding of the disease. It was only in 1921, however, that a more precise picture emerged: Banting, Best and Macleod showed that diabetes in pancreatectomised dogs could be reversed after the intravenous administration of the ‘islet’ extraction from normal canine pancreata.13 Subsequently, Banting, Best and Collip purified this substance from bovine pancreata, and the first patient was successfully treated in 1922, resulting in a reduction in blood glucose and glycosuria.
The possibility that diabetes may exist in different forms was hypothesised during the 1920s and 1930s. In 1926, MacLean suggested a distinction between ‘hepatic glycosuria’ (uncommon in the young and characterised by mild glycosuria and low concentration of urinary ketone bodies) and ‘true diabetes’ (a condition affecting young subjects, with significant ketonuria and fatal unless treated with insulin).14 Ten years later, Himsworth, summarising his previous research, distinguished between ‘insulin-sensitive’ and ‘insulin-insensitive’ diabetes mellitus,15 with the latter more insidious condition characterised by less severe hyperglycaemia. During the 1950s, a reliable measurement of circulating insulin with a radioimmunoassay technique allowed a clear distinction between ‘insulin-dependent’ and ‘non-insulin-dependent’ diabetes mellitus,16 and the paradigm of two pathophysiologically distinct diseases became more and more evident in the following years.
Type 1 diabetes mellitus
It is now well-recognised that T1DM is an autoimmune disorder characterised by the destruction of insulin-producing pancreatic β-cells.17 Like many other immune-mediated diseases, T1DM shows heterogeneity in terms of age of onset, severity of autoimmune response, and efficacy of therapy. A common distinction is made between type A (accounting for up to 90% of overall cases), with a detectable serological autoimmune response, and type B (or idiopathic), where apparently no humoral autoimmunity is present.18 This distinction has, however, not been widely adopted.
Historically, the definition of T1DM as a nosological entity dates back to 1979,19 when the National Diabetes Data Group divided diabetes into two major subtypes (table 1). By the 1960s and 1970s, however, a comprehensive picture had already emerged, following the discovery of inflammation within the pancreatic Langerhans islets (insulitis), the identification of an association with human leucocyte antigen (HLA) genes,20 the detection of autoantibodies against islets cells, and a better method for measuring insulin.16 HLA genes are critical in regulating the immune response, as they encode cell surface proteins involved in the antigen presentation and self-tolerance. Genetically determined variations in the amino acid sequence of these proteins can therefore alter the repertoire of presented peptides and result in the loss of self-tolerance.21 These observations, along with the contemporary knowledge of an association between HLA and other autoimmune disorders and evidence of the efficacy of immunosuppressive therapies on T1DM disease progression,22 ,23 strongly supported the idea that ‘insulin-dependent’ diabetes was an immune-mediated disease involving the pancreatic islets of Langerhans.
Progress in understanding the pathophysiology of T1DM cannot be separated from advances in the field of immunology. Like the vast majority of autoimmune disorders, the primary cause of T1DM is still unknown. T1DM is characterised by a selective, specific involvement of β-cells without apparent pathological alterations of other Langerhans cells, such as α- (secreting glucagon), δ- (somatostatin) and PP- (pancreatic polypeptide) cells.24 In recently diagnosed T1DM, generally more than two-thirds of pancreata have no evidence of insulin at all, around one-fifth display insulin to some extent, and some have no detectable alterations.17 The reason for this heterogeneity remains unknown, possibly being related to a true difference (different pathophysiological entities) or representative of different stages (or severity) of the same disease.
Both humoral and cellular immunity is involved in T1DM pathogenesis. T lymphocytes (which mature in the thymus and play a central role in cell-mediated immunity) are predominant in islet lesions, with lower concentrations of other immunological cells, such as macrophages, B lymphocytes and plasma cells.24 The presence of humoral immunity, on the other hand, was recognised over 40 years ago, when autoantibodies against pancreatic islets were detected in subjects with T1DM.25 From the 1980s, targets of autoantibodies have been discovered, and several autoantigens are currently widely used in clinical practice, such as insulin, proinsulin, glutamic acid decarboxylase (GAD65), glucose 6-phosphatase catalytic subunit-related protein (G6PC2, also known as IGRP), islet cell antibody (ICA) and zinc transporter 8 (ZnT8A).26 In genetically predisposed individuals (eg, first-degree relatives), autoantibodies can also be detected months to years before the clinical diagnosis is made.27
A landmark description of the natural history of T1DM was published in 1986 by Eisenbarth, after reported discoveries in the 1970s and 1980s (figure 1).28 Although subsequent findings have helped to clarify in more detail the role of immunity in T1DM pathogenesis,17 ,29 ,30 these earlier concepts remain applicable. In predisposed individuals, early-life environmental triggering factors (infections, nutrition, chemicals) are able to ‘activate’ self-targeting immune cascades. The initial event, however, is still unclear. Of note, the presence of detectable autoantibodies is not sufficient for the clinical development of T1DM, as some subjects with serological positivity will never develop T1DM.18 These observations underline the importance of β-cell function and β-cell turnover (mass) in the pathogenesis of insulin deficiency.31 In the initial phases, the progressive destruction of β-cells and serological positivity are not associated with changes in blood glucose, as a ‘functional’ pancreatic reserve is sufficient to maintain euglycaemia (figure 1). In the following stages, there is further β-cell destruction, with a consequent loss of insulin production and a parallel increase in blood glucose concentration. When the majority of β-cells are destroyed, overt diabetes develops. Tight glycaemic control after diagnosis is of paramount importance, as ‘near-to-normal’ glucose has been shown not only to reduce the risk of diabetic complications,32 but also to preserve any remaining β-cell mass/function.33
Type 2 diabetes mellitus
While in recent years many major risk factors for the emergent T2DM epidemic have been identified, the mechanisms linking them to the clinical manifestations of T2DM and its complications are intensively investigated (figure 2). The availability of radioimmunoassays in the 1950s helped differentiate ‘insulin-dependent’ from ‘non-insulin-dependent’ diabetes, and such differences were formally recognised in the 1979 classification of diabetes by the National Diabetes Data Group in type I (former ‘insulin-dependent’) and type II (‘non-insulin-dependent’) diabetes mellitus.19 It was not until the 1980s and 1990s, with the standardisation and evolution of techniques to measure glucose disposal by insulin,34 ,35 that researchers performed independent experiments confirming the presence of insulin resistance in ‘maturity-onset’ diabetes.35 ,36
Several pivotal pathophysiological studies underpin our understanding of insulin resistance and secretion in the course of disease onset and progression.37–46 Subjects at risk of T2DM (obese subjects and first-degree relatives) display an initial state of insulin resistance compensated by β-cell hypersecretion of insulin (hyperinsulinaemia). Such pancreatic ‘functional’ reserve, however, is eventually unable to cope with the required insulin secretion. Compared with lean euglycaemic subjects, obese euglycaemic people have ∼30% reduced insulin sensitivity46 and therefore show increased insulin secretion to maintain normal glucose tolerance (euglycaemic hyperinsulinaemia). Over time, obese euglycaemic subjects experience a further reduction in insulin sensitivity that is no longer associated with compensatory hyperinsulinaemia, resulting in an increased blood glucose concentration (hyperglycaemic hyperinsulinaemia). By the time diabetes is diagnosed, β-cells are no longer able to secrete enough insulin, with consequent manifestation of overt hyperglycaemia (hyperglycaemic hypoinsulinaemia).46 Although the relative contribution of β-cell dysfunction and insulin resistance can vary, it is generally accepted that ‘abnormal’ insulin sensitivity precedes the clinical diagnosis of diabetes by up to 15 years.47 The results of these pathophysiological studies illustrate the failure of compensatory hyperinsulinaemia as the ‘hallmark’ of frank hyperglycaemia. Therefore, along with mechanistic studies investigating mechanisms forming the basis of insulin resistance, more recent research has also focused on the pathways leading to β-cell ‘failure’.
Liver and muscles have long been recognised as major contributors of systemic insulin resistance.37 To ensure constant availability of a carbohydrate energy source during fasting, the liver produces glucose from non-glucose substrates (gluconeogenesis).48 Several studies have shown increased gluconeogenesis in subjects with T2DM, which occurs despite a state of hyperinsulinaemia, suggesting hepatic insulin resistance as a main determinant of fasting hyperglycaemia.49 ,50 The reasons behind reduced hepatic insulin sensitivity are poorly defined, but accumulation of fat within the liver (steatosis) is considered a major determinant.51 Interestingly, liver steatosis precedes overt T2DM and is commonly associated with obesity,52 particularly visceral (android or abdominal) obesity.53 It is now well accepted that a continuous positive energy balance due to excess calories and a lack of physical activity leads initially to fat accumulation in the subcutaneous tissue. When this storage capacity is exceeded, fat is diverted to ‘ectopic’ compartments such as the liver, pancreas, muscles, perivascular, pericardium and omentum (‘spill-over’ or ‘adipose tissue overflow’ hypothesis).53 Hepatic and muscle fat accumulation results in impaired insulin-mediated glucose uptake due to intracellular impairment of insulin signalling.37 Of note, muscle insulin resistance has been shown in lean T2DM,35 suggesting that mechanisms independent of body fat may also be involved in its pathogenesis. Fat accumulation within the pancreatic islets, on the other hand, determines β-cell dysfunction and increases in plasma glucose which, in turn, reduce insulin response to ingested glucose (‘twin cycle hypothesis’52).
As highlighted, defects in pancreatic β-cell function are essential for the manifestation of hyperglycaemia.54 Research performed in the last few years has attempted to clarify the mechanisms of β-cell failure. In genetically predisposed individuals,55 ,56 the increased demand of insulin synthesis and secretion eventually results in β-cell dysfunction. Yet, among the proposed potential mechanisms causing β-cell dysfunction (including direct effects of glucose and free fatty acids), their comparative and chronological role is unknown. It has been suggested that the ‘stressed’ β-cell may stimulate local inflammation and modify the balance between α- and β-cell mass and function within the Langerhans islets. Of note, insulin exerts negative paracrine (a signal inducing changes in nearby cells) action on α-cells, thus limiting the secretion of glucagon;57 therefore, lack of insulin results in higher levels of glucagon, which further increase blood glucose concentration via hepatic gluconeogenesis.
Among the regulators of β-cells, a central role for gut hormones has also been discovered over the last two to three decades. Insulin secretion is greater after glucose ingestion than after an intravenous glucose infusion with superimposable glucose excursion (isoglycaemic infusion). This observation suggests the existence of factors that stimulate insulin secretion after glucose ingestion. These factors have been shown to be gut ‘messengers’ (termed incretins) able to stimulate insulin secretion.58 Two incretins, glucagon-like peptide-1 (GLP-1, secreted by L cells located predominantly in the ileum and colon) and gastric inhibitory polypeptide (GIP, secreted by enteroendocrine K cells concentrated in the duodenum and proximal jejunum, also interpreted as glucose-dependent insulinotropic polypeptide), have been identified as the main hormones responsible for this phenomenon. These polypeptides are secreted after food ingestion and are able to increase insulin (GLP-1 and GIP) and reduce glucagon (GLP-1) secretion.58 While a major secretory defect in GIP secretion does not seem to exist in T2DM, significantly decreased secretion of GLP-1 has been consistently found in T2DM,59 insulin resistance60 and obesity.61 Therefore, after food ingestion, subjects with T2DM have a blunted GLP-1 response, resulting in lower postprandial GLP-1 and insulin concentrations and relative hyperglucagonaemia. Further to this, T2DM subjects also show reduced responsiveness to both GIP and GLP-1, a condition that can be improved by restoring euglycaemia,62 suggesting that the loss of incretin is secondary to the development of hyperglycaemia and not a direct cause of it.
The role of the kidney in blood glucose regulation has long been known.14 About 180 g of glucose is filtered by the glomerulus every day: about 90% is reabsorbed in the proximal tubule through the sodium–glucose cotransporter 2 (SGLT2) membrane transporter and 10% in the descending tubule (SGLT1).63 Reabsorption of filtered glucose increases linearly until the maximal reabsorptive capacity (Tm) is exceeded, which is ∼11.0 mmol/L (200 mg/dL) in healthy adults.64 Notably, Tm values seem to be higher in T2DM subjects, further exacerbating hyperglycaemia in a vicious cycle.65 The mechanism possibly associated with the change in Tm could be the upregulation of SGLT2 secondary to hyperglycaemia.66
Maturity-onset diabetes of the young, latent autoimmune diabetes of adults, and ‘double diabetes’
With a better understanding of diabetes pathophysiology, it has also become clearer that some forms of diabetes do not entirely fall into ‘type 1’ or ‘type 2’ categories. The term ‘maturity-onset diabetes of the young (MODY)’ describes single-gene disorders causing type 2 diabetes-like conditions in younger age groups.67 These monogenic forms, possibly representing 2% of all diabetic cases in the UK,68 display an autosomal dominant pattern of inheritance of diabetes and are usually diagnosed around childhood or adolescence. While there is variability in the natural history, the progression from normal to ‘mild’ hyperglycaemia and frank diabetes is generally slow.67 Initial genetic studies on MODY started in the mid 1980s, and several genetic defects have been discovered since then. The two most common conditions are MODY2 (mutation of a glucokinase (GCK) gene) and MODY3 (hepatocyte nuclear factor (HNF)), with the latter characterised by a remarkable response to sulfonylureas (drugs stimulating insulin secretion). Of note, recognised forms of MODY are characterised by ineffective insulin production and/or release by pancreatic β-cells.69
Latent autoimmune diabetes of adults (LADA), also known as slowly progressing insulin-dependent diabetes,70 defines a form of diabetes characterised by three features: adult age at diagnosis, presence of diabetes-associated autoantibodies, and no need for insulin therapy at diagnosis.71 From a pathophysiological perspective, this form of diabetes can indeed be considered a slowly progressive T1DM. Moreover, given the increased prevalence of obesity and insulin resistance, cases of combined type 1 diabetes and insulin resistance are increasing, giving rise to the definitions of ‘type 1.5’72 or ‘double diabetes’.73 These patients often represent a diagnostic challenge, particularly if some β-cell insulin is initially preserved.
Diabetes mellitus can also be present in a range of rare genetic (eg, Alström syndrome, Friedrich's ataxia, Huntington's disease, congenital lipodystrophy, mitochondrial DNA) and chromosomal (eg, Down's syndrome, Turner's syndrome, Klinefelter's syndrome) disorders involving multiple organs.74
Current research: role of brown adipose tissue and gut
A great deal of research into the pathophysiology of diabetes and its complications now centres on the biological role of fat tissue. Compared with white adipose tissue (the main role of which is energy storage), brown adipose tissue (BAT) is a biologically distinct form involved in adaptive thermogenesis to maintain normal body temperature.75 Animal studies have demonstrated the importance of BAT in regulating energy and glucose homeostasis. BAT activation increases energy expenditure and is associated with peripheral insulin resistance and glucose levels.76 ,77 Although further research on BAT is essential, preliminary data suggest BAT to be an attractive therapeutic target.
The role of the gut in the pathophysiology of diabetes can be viewed from two different perspectives. Beyond malabsorption and/or anatomical restriction, studies have suggested that several mechanisms may be involved in weight loss and diabetes control after bariatric surgery.78 Indeed, complex changes in multiple gut hormones have been shown after bariatric procedures and have been proposed as adjunctive (weight-loss-independent) mechanisms for short- and long-term positive metabolic effects.79 A better understanding of these physiological adaptations would help to clarify the role of gut hormones in the pathogenesis of diabetes, as well as offer novel therapeutic approaches to obesity and insulin resistance.80 Further to this, in the last few years, a potential role of the gut microbiome (1014 bacteria) in energy balance has been demonstrated.81 The gut microbiome differs between obese and lean subjects,82 and a faecal microbiome transplantation from lean donors to insulin-resistant subjects results in beneficial metabolic effects.83 These observations suggest a causal role of gut bacteria (and possibly their products, such as short-chain fatty acids) in metabolism, and research is ongoing to disentangle the inter-relationship between bariatric procedures and changes in gut hormones and microbiome.84
Over the last 50 years, many pathophysiological studies have contributed to a better understanding of diabetes. Available evidence clearly demonstrates that diabetes mellitus is a spectrum of disorders characterised by variable degrees of insulin resistance and β-cell dysfunction. Factors modifying either or both lead eventually to hyperglycaemia and appear to have independent actions. It is therefore possible that, on a background of monogenic β-cell dysfunction (ie, MODY) or obesity-related insulin resistance, a patient may also develop an autoimmune response against β-cells, which is particularly relevant in the context of the current diagnosis and treatment of the disease based on polarised definitions of type 1 and type 2.
Historical definitions of diabetes may become less applicable in the future as classification of diabetes based on a greater pathophysiological understanding emerges.85 Although much discussed, a ‘personalised’ treatment of hyperglycaemia has not yet been widely implemented in clinical practice. It would be reasonable to identify and treat patients on the basis of the degree of β-cell dysfunction and insulin resistance, as a simple ‘glucocentric’ therapeutic approach could potentially result in no benefit (and possibly harm) in some patients.86 Moreover, a clearer definition of diabetes phenotypes would further help us to understand the comparative contribution of hyperglycaemia and insulin resistance in the development of diabetes-related vascular and non-vascular complications and help us to design future randomised clinical trials in pathophysiologically similar subgroups of patients with T2DM.
Lastly, it is worth noting that, from an historical perspective, a detailed picture of diabetes pathophysiology has emerged only after decades of research. Similarly, the role of ‘emerging’ organs (ie, kidney, adipose tissue, gut) can be clarified only in independent, replicated studies, with the main goal being to establish their temporal involvement in the natural history of diabetes.
Self assessment questions
Please answer true (T) or false (F) to the below
Type 1 diabetes mellitus is
A single-gene disease
A multifactorial disease
An autoimmune disease
Type 2 diabetes mellitus is associated with an increased risk of
Only vascular diseases
Only non-vascular diseases
Both vascular and non-vascular diseases (eg, cancer)
Insulin secretion is regulated by
Insulin resistance is present
Only in type 1 diabetes mellitus
Only in type 2 diabetes mellitus
Only in subjects with hyperglycaemia
HLA genes are associated with
Type 1 diabetes mellitus
Type 2 diabetes mellitus
Epidemiological, human and molecular studies in the last 50 years have substantially clarified the pathophysiology of diabetes mellitus.
The current classification of diabetes reflects only in part recent progress in understanding its pathophysiology.
A more ‘pathophysiological’ classification could potentially result in a better patient phenotyping and therapeutic approach.
Current research questions
Can a better pathophysiological definition of type 2 diabetes phenotypes result in a personalised therapy and better outcomes?
What is the temporal involvement of the kidney and gut in the pathophysiology of type 2 diabetes?
Are the natural history and the risk of type 2 diabetes complications related to its different phenotypes?
Banting FG, Best CH, Collip JB, et al. Pancreatic extracts in the treatment of diabetes mellitus: preliminary report. 1922. CMAJ 1991;145:1281–6.
Himsworth HP. Diabetes mellitus: its differentiation into insulin-sensitive and insulin-insensitive types. Lancet 1936;227:127–30.
Yalow RS, Berson SA. Immunoassay of endogenous plasma insulin in man. J Clin Invest 1960;39:1157–75.
National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979;28:1039–57.
Tabak AG, Jokela M, Akbaraly TN, et al. 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.
A (F); B (T); C (T).
A (F); B (F); C (T).
A (T); B (T); C (T).
A (F); B (F); C (F).
A (F); B (T); C (F).
We acknowledge the support of the following institutes in this work: the National Institute for Health Research; Collaboration for Leadership in Applied Health Research and Care–East Midlands (NIHR CLARHC East Midlands); the National Institute for Health Research (NIHR) Diet, Lifestyle & Physical Activity Biomedical Research Unit based at University Hospitals of Leicester and Loughborough University. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.
Contributors All authors planned the outline, contributed to figures and tables, and wrote and approved the manuscript. FZ carried out the literature search.
Funding This study is funded by the National Institute for Health Research.
Competing interests FZ is a clinical research fellow funded with an educational grant from Sanofi-Aventis to the University of Leicester. MJD has acted as a consultant, advisory board member and speaker for Novo Nordisk, Sanofi-Aventis, Lilly, Merck Sharp & Dohme, Boehringer Ingelheim, AstraZeneca and Janssen and as a speaker for Mitsubishi Tanabe Pharma Corporation. She has received grants in support of investigator and investigator-initiated trials from Novo Nordisk, Sanofi-Aventis and Lilly.
Provenance and peer review Commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.