Introduction

South Asians (people living in, or with ancestry from, the Indian subcontinent), represent nearly 2 billion people globally and have a high prevalence of type 2 diabetes, even at lower BMIs [1]. This increased diabetes risk among South Asians was first reported in the diaspora but has gathered greater attention as diabetes sweeps across South Asian countries, alongside economic development and dramatic changes in lifestyles [2]. Recent environmental changes, epidemiological transition and lifestyle changes (e.g. changes to diet and physical activity) are important in explaining the rise of type 2 diabetes in South Asians. However, given the high diabetes risk at low BMI in South Asians, the unique biological reasons driving the heightened risk of type 2 diabetes in South Asians compared with other ethnicities also need to be identified [3].

The often-cited biological reason for increased diabetes risk in South Asians is the propensity for insulin resistance, driven by higher visceral adiposity [1, 4, 5]. Much of this evidence placing obesity-driven insulin resistance as central to the pathophysiology of type 2 diabetes has come from studies in high-income well-nourished countries with an established obesity epidemic or from smaller replicative studies in South Asia. Furthermore, studies using objective measures of visceral adiposity have raised doubts about its unique or dominant role in the pathophysiology of type 2 diabetes in South Asians [6]. Thus, we may be overlooking potential differences in the pathophysiology and risk factors in the South Asian or other similar populations, especially in those who develop type 2 diabetes in the absence of obesity.

In this review, we will summarise recent longitudinal epidemiological data on diabetes incidence in South Asians. We acknowledge that diabetes risk in South Asians is a complex product of multiple factors, including genes, lifestyle, culture, socioeconomic factors and migration [1]. However, in this article, we focus selectively but critically on novel clues from studies of type 2 diabetes pathophysiology in South Asians placed in the context of intriguing historical knowledge from evolutionary biology studies in this population. We will emphasise three previously understudied areas investigating the high risk of type 2 diabetes in South Asians. We hypothesise that the high type 2 diabetes susceptibility in the South Asian population is evolutionarily set through dual parallel and/or interacting mechanisms: (1) reduced beta cell function and (2) impaired insulin action owing to low lean mass, which is further accentuated by (3) ectopic fat deposition in the liver and muscle (Fig. 1).

Fig. 1
figure 1

Proposed pathophysiological pathways for type 2 diabetes. Both impaired insulin secretion and impaired insulin action (or insulin resistance) have evolutionary roots among South Asians, with low beta cell function and low lean muscle mass. Adverse lifestyle behaviours further exacerbate insulin action with ectopic fat deposition in liver and muscle. This figure is available as a downloadable slide

Full size image

Emerging epidemiological data on diabetes incidence in South Asian populations

A few published studies on South Asians in the USA and Europe have indicated a higher incidence of type 2 diabetes in this population than in other ethnic groups [7, 8]. A recent study from Chennai, India, noted an alarmingly high incidence of prediabetes and diabetes (29.5 and 22.2 per 1000 person-years, respectively), and a possibly rapid conversion from prediabetes to diabetes [9]. Preliminary data from the largest population-based cohort study in South Asians with follow-up data for multiple time points, the Center for cArdiometabolic Risk Reduction in South Asia (CARRS) [10], indicates that among urban people aged 20–44 years, the age-standardised diabetes incidence (cases/1000 person-years) in South Asians was 14.2 (95% CI 12.2, 16.2) in men and 14.8 (95% CI 13.0, 16.5) in women. Furthermore, a recent analysis of diabetes incidence in South Asians compared with Pima Indians, a population reputedly with the world’s highest diabetes incidence revealed some important findings [11]. Prospective analyses were done using data from two population-based cohorts of adults aged 20–44 years: the Pima Indian Study (Pima Indians, n = 1852), and CARRS (South Asians, n = 6676). Diabetes was defined as fasting glucose ≥7.0 mmol/l (126 mg/dl), HbA1c ≥ 6.5% (48 mmol/mol) or glucose-lowering medication use. At baseline, South Asians were less obese than Pima Indians (BMI: 24.4 vs 33.8 kg/m2; waist circumference: 82.5 vs 107 cm). Age-standardised diabetes incidence (cases/1000 person-years) was much lower in South Asians than in Pima Indians (men: 14.2 [95% CI 12.2, 16.2] vs. 37.3 [95% CI 31.8, 42.8; women: 14.8 [13.0, 16.5] vs 46.1 [41.2, 51.1]). However, among 20–44 year olds with a BMI <25 kg/m2, the incidence of diabetes in South Asian men was about fivefold that in South Asian men. Among those with a BMI ≥30 kg/m2, diabetes incidence in South Asian men was nearly as high as in the Pima men [11].

Prospective cohort analyses in adults aged ≥45 years compared the incidence of diabetes in South Asians, using data from the CARRS cohort (n = 3136), with that in white and black people in the US Atherosclerosis Risk in Communities Study (ARIC; black, n = 3059; white, n = 9924) [12]. Among those aged ≥45 years, the age-standardised diabetes incidence in South Asian men and women was 26.0 (95% CI 22.2, 29.8) and 31.9 (95% CI 27.5, 36.2), respectively. In those with a BMI <25 kg/m2, the diabetes incidence in South Asian men and women was nearly 3 times and 5.3 times that in the corresponding sex in the white subgroup, respectively. Among those with a BMI ≥30 kg/m2, the diabetes incidence in South Asian men and women was 2.6 and 2.1 times that in US white men and women, respectively [12]. Together, these data on incidence indicate that the South Asian population may have a higher risk of incident diabetes than other ethnic groups. This difference is particularly evident, even at low BMIs, suggesting that factors other than obesity and insulin resistance may be important to explore [11, 12].

Pathophysiology

Contextual knowledge from evolutionary biology

South Asians, on average, have a low lean mass [13], which may confer specific susceptibility for diabetes. This may have historical roots. In a recent study of skeletons spanning the past 11,000 years, Pomeroy et al. investigated the ancient origins of low lean mass in the South Asian population [14]. They found that, compared with a worldwide sample, the skeletons of South Asians indicated low lean mass. The stature of South Asians decreased sharply when agriculture was adapted, while stature-adjusted lean mass increased to a minor extent over time. Pomeroy et al. concluded that the South Asian population has been characterised by low lean mass for thousands of years, and this phenomenon is unlikely to change in the short term. Historical data suggest that the South Asian population may have had good nutrition status in the Mesolithic period, as indicated by their relative tallness during that time [15]. However, since then, the population may have become increasingly undernourished for generations, and this may have become more pronounced during the period of British colonisation of the subcontinent [16]. These long-term intergenerational influences, especially those adversely affecting maternal and/or early childhood during the first 3 years of life, may have had an impact on the metabolic capacity of people in these populations [17], e.g. by leading to shorter stature, lower lean muscle mass, poorer organ development and increased diabetes susceptibility [18,19,20]. A number of studies also support the role of transgenerational metabolic pathways (e.g. poor insulin secretion, ectopic fat deposition and low lean mass) linking early malnourishment with diabetes risk [21, 22].

Role of poor insulin secretion very early in the natural history of diabetes

There are reasons to consider South Asians as a population variant with higher frequencies of phenotypes of type 2 diabetes, wherein the primary problem may be poor insulin secretion (or impaired beta cell reserve) to start with [23,24,25]. A study comparing non-diabetic adult South Asians with Chinese Americans, Hispanics, African Americans and whites in the USA found that, in every age group, South Asians had the lowest insulin secretion, followed by Chinese Americans [3]. Studies of adults in India, youth-onset diabetes populations in India, and adult South Asians in the USA all point to the potential role of deficient insulin secretion in the early natural history of type 2 diabetes in South Asians [26,27,28]. Furthermore, in the Whitehall II cohort study in the UK, South Asians had a lower beta cell reserve relative to Europids [29], and a study from the Netherlands suggested that family members of South Asians with type 2 diabetes may have poorer beta cell adaptation than family members of Dutch individuals, of European descent, with type 2 diabetes [30]. A comparison of Indians living in urban Chennai with Pima Indians in Arizona revealed that both populations have a roughly equal diabetes prevalence, despite the Chennai Indians being an average of 7 kg/m2 lighter and having a waist circumference 25 cm smaller [31]. In addition, in all age groups, across weight and glucose categories (including those with normal BMI and normal glucose levels), the Pima Indians were two to four times more severely insulin resistant than the Chennai Indians and the Chennai Indians secreted a half to a third the amount of insulin compared with the Pima Indians. Notably, compared with Pima Indians, South Asians have relatively poorer insulin secretion at baseline and at 30 and 120 min after an oral glucose tolerance test. A limitation of these studies is that surrogate measures of insulin resistance and insulin secretion were used (e.g. HOMA-IR and HOMA-B) rather than the gold standard measurements. One small study using hyperinsulinaemic–euglycaemic clamps found that South Asians, especially underweight or normal-weight individuals with metabolic abnormalities, did not have reduced insulin action, implying that reduced insulin secretion may need to be investigated in lean individuals at risk of developing type 2 diabetes [19].

Compared with other ethnic groups, South Asians have a higher prevalence of isolated impaired fasting glucose (iIFG), which is primarily the result of hepatic insulin resistance, with early phase, stationary impairment in beta cell function [32]. While the rate of conversion of prediabetes to diabetes in South Asians appears high (up to 14% per year), regardless of the prediabetes type, it is unclear whether those with iIFG have lower rates of conversion than those with isolated impaired glucose tolerance (iIGT), as few data on an adequate sample size exist [9, 33]. In a lifestyle intervention followed by addition of metformin (when appropriate) among South Asians with prediabetes in Chennai, there was an overall relative risk reduction of 32% of progression to diabetes [33]. However, in the group with iIFG, the relative risk reduction was a non-significant 12%. South Asians may also respond relatively less well to pharmacological agents, such as glitazones, that target insulin sensitivity (beta cell response to blood glucose) and/or insulin resistance (glucose uptake) [34]. Interventions that preserve or help recover beta cell function during the early normoglycaemic periods in South Asians need be researched, and possible strategies may include high-intensity interval training or specific pharmacological approaches [35,36,37].

Role of abdominal visceral fat vs hepatic fat

Greater body adiposity has been a theorised culprit for insulin resistance and type 2 diabetes among South Asians. Early studies from the UK reported that South Asians had a higher waist circumference compared with Europeans, supporting a role for central adiposity underpinning insulin resistance and its metabolic consequences. However, recent studies using computed tomography (CT) scans for ectopic fat measurements, including abdominal visceral fat and intrahepatic fat, have shown less consistent results [38,39,40,41]. A recent meta-analysis of published and unpublished studies with direct measures of visceral fat by CT or magnetic resonance imaging found no difference in abdominal visceral fat area among South Asian men and women compared with their white counterparts. However, there were consistently higher amounts of hepatic fat among South Asian compared with white individuals (Iliodromiti S, Queen Mary University, London, UK, personal communication).

This distinction between visceral fat and intrahepatic fat stores is important because of the different metabolic consequences at each site, with the liver being the prime organ for postprandial glucose metabolism and glycogen synthesis. An intriguing possibility to explain why South Asians have greater hepatic fat content than other groups is that the relatively higher saturated fat content of traditional South Asian diets may be involved [42]. In a short-term dietary trial of overfeeding saturated fat or polyunsaturated fat in normal-weight Swedish adults, while weight gain was similar in both the high-saturated-fat and high-polyunsaturated-fat groups, relatively higher amounts of hepatic fat than visceral fat were gained by the saturated fat group [43].

Few studies have measured both abdominal visceral fat and hepatic fat to determine their relative contributions to incident diabetes. In the Mediators of Atherosclerosis in South Asians Living in America (MASALA) study, abdominal visceral fat area and hepatic attenuation were independently associated with glycaemic progression (from normal, to impaired fasting glucose, to impaired glucose tolerance, to diabetes) [8]. Hepatic fat was more closely linked to progression from prediabetes to diabetes, indicating its role once dysglycaemia has already begun [9]. A recent Mendelian randomisation analysis of the UK Biobank data found that genetically predicted higher circulating liver enzymes, surrogate markers for non-alcoholic fatty liver disease, were associated with increased risk of type 2 diabetes [44]. Moreover, genetic variants associated with insulin resistance were also associated with these liver enzymes, indicating a causal relationship between insulin resistance resulting in fatty liver disease, which in turn increases type 2 diabetes risk [44].

Weight loss is a proven treatment for fatty liver, even among normal-weight Asians [45], and modest weight loss can decrease abdominal fat stores in general [46]. Several lifestyle modifications for non-alcoholic fatty liver disease (NAFLD) have been tested in short-term randomised trials in primarily non-Asian populations, showing that both hypoenergetic low-carbohydrate diets and low-fat diets, as well as isoenergetic (isocaloric) higher protein diets and diets high in fibre, can decrease intrahepatic lipid content [47]. In addition, aerobic, resistance and high-intensity interval training forms of exercise have each been shown to have similar effects on reducing hepatic fat, even independently of weight loss [47, 48]. Culturally appropriate dietary and exercise interventions, with and without weight loss, focusing on hepatic steatosis are, therefore, needed for South Asians.

Role of skeletal muscle

Low lean muscle mass is another important contributing factor that may promote susceptibility to diabetes among South Asians. Several studies have shown that South Asians have lower lean muscle mass than other race/ethnic groups [40, 41, 49]. As mentioned above, South Asians have exhibited low lean mass consistently over time, most likely representing long-term adaptation to climate or other more neutral variation [14]. Genetic studies of several populations from the South Asian subcontinent have found increased selection of a gene encoding myostatin [50], a protein that inhibits skeletal muscle growth in utero through poor placental glucose uptake. Lower lean muscle mass affects peripheral glucose uptake and clearance, since the skeletal muscle is responsible for metabolising 80% of circulating glucose [51]. Long-standing peripheral insulin resistance may further lead to earlier beta cell exhaustion [52], and may be especially important when insulin secretory reserves are already in short supply.

Another form of ectopic fat deposition, accumulation of lipids within the skeletal muscle from ineffective lipid metabolism as a result of dysfunctional mitochondrial oxidation, may also play a role in insulin resistance among South Asians. In a UK study comparing South Asians and Europids, South Asians had 30% higher intramuscular triacylglycerols than BMI-matched Europids [53], and in a US study, South Asians had higher intermuscular fat on CT scans compared with other US race/ethnic groups [41]. While a build-up of intramyocellular lipids is not directly associated with insulin resistance [54], an accumulation of diacylglycerols and ceramides have molecular effects that may promote insulin resistance [55].

The role of the mitochondria in skeletal muscle in promoting insulin resistance remains heavily debated. The few studies investigating skeletal muscle mitochondrial function in South Asians have had conflicting results. In a small US study reported in 2008, among Asian Indians muscle oxidative capacity was similar for those with and without diabetes, and Asian Indians had a higher oxidative capacity than the white participants though the Asian Indians had more severe insulin resistance [56]. In an exercise study of 20 South Asian and 20 BMI-matched Europid men, the South Asian men had significantly more severe insulin resistance, had lower cardiorespiratory fitness and lower skeletal muscle fat oxidation with exercise, but this was not explained by genes known to be involved in fat oxidation and lipid metabolism [57]. Newer evidence refutes an impairment of mitochondrial capacity and posits a mitochondrial bioenergetics problem, with an oversupply of fuel that leads to mitochondrial reactive oxidation species production that impairs insulin signalling [58].

Lifestyle modification may directly affect lean muscle mass and mitochondrial function. In the Swedish overfeeding study mentioned above, the group assigned to polyunsaturated fat had a greater increase in lean muscle mass and no significant increase in other ectopic fat stores [43]. Dietary long-chain saturated fatty acids impaired mitochondrial function while n-3 polyunsaturated fatty acids improved mitochondrial function [55]. Exercise training, both endurance and resistance exercise, can stimulate mitochondrial biogenesis and improve mitochondrial function and efficiency [59]. Few dietary or exercise interventions have been conducted among South Asians to determine the most beneficial diet or the most effective dose, type, intensity or duration of exercise to improve beta cell function, reduce ectopic fat stores or increase lean muscle mass [60] and thereby reduce diabetes risk in this population.

Conclusion

While data on type 2 diabetes prevalence have suggested that South Asians are at high risk, even at low BMI, recent data on diabetes incidence reviewed above further support this. Furthermore, these data on diabetes incidence also suggest that South Asians have a higher diabetes risk than other ethnic populations, and normal-weight South Asians have higher risk than normal-weight people of other ethnic groups, including white, black and Pima Indians. We postulate that the evolutionary history of the South Asians has endowed the population with significant susceptibility to type 2 diabetes through a dual mechanism (Fig. 1). Our hypothesis is that this susceptibility is primarily driven by poor metabolic capacity (i.e. reduced beta cell mass and/or function, which impairs insulin secretion) coupled with low lean muscle mass, which may be responsible for reduced insulin action. This is accentuated by a propensity to ectopic fat deposition in the liver and muscle, which further impairs insulin action. These unique evolutionary and anthropomorphic features of South Asians, relative to other ethnic groups, will need to be considered to formulate an innovative research agenda for solving the problem. Well-supported programmes of interdisciplinary research, spanning basic/translational sciences, longitudinal epidemiology and intervention studies are needed, with robust measures of insulin secretion, lean mass, hepatic fat and glucose patterns, in South Asian populations in host countries and in the diaspora. There is also an urgent need to unravel the basic biological reasons (e.g. poor insulin secretion, low lean mass, hepatic steatosis, myosteatosis, mitochondrial metabolism) that put South Asians at such high risk for diabetes, through high-quality basic science research. Simultaneously, research into potential clinical and public health interventions to address these specific susceptibilities in South Asians, using robust measurements and appropriate endpoints, are needed. Among them are behavioural and pharmacological strategies to improve beta cell preservation and/or approaches to improve lean mass and reduce ectopic fat deposition.