The prevalence of type 2 diabetes mellitus in adult Indians living in cities is estimated to be between 12 and 16%, and India has the greatest number of diabetic subjects in any single country [1]. It is well established that Asian Indians are thin (low BMI) but have higher adiposity (percentage body fat), are more centrally obese [2] and more insulin-resistant than Europeans [3]. These factors are thought to increase their susceptibility to diabetes mellitus [4]. It has been reported that the genetic basis of some diseases in Indians may be different from that reported in Europeans [5], but only a few such studies have been performed.

There has been some progress in defining the genetic susceptibility to type 2 diabetes mellitus in European populations. The Pro12Ala polymorphism in PPARG (also known as PPARγ) and the E23K variant in KCNJ11 predispose to type 2 diabetes with allelic odds ratios (OR) in the range of 1.15–1.3 [6, 7]. There are a limited number of studies on the genetic basis of type 2 diabetes mellitus in the Indian population, mostly on Southern Indians [8, 9]. Recently, variants in TCF7L2 gene have been strongly associated with increased risk of type 2 diabetes mellitus in Icelandic individuals and replicated in Europid subjects from USA and Denmark, with a combined OR of 1.56 (p = 4.7 × 10−18) [10]. Similar observations have been made in large case–control and familial-association studies from the UK [11]. However, all these studies are from European populations. We report for the first time the association of the variants in the TCF7L2 gene with type 2 diabetes mellitus in a South Asian Indian population.

Subjects and methods

We studied 955 type 2 diabetes mellitus patients and 399 ethnically matched control subjects (Indo-Europeans) from Pune, Maharashtra, Western India. The patients were from the Diabetology Research Centre, King Edward Memorial Hospital and the Research Centre, Pune, and were part of the Wellcome Genetic collection (WellGen) of young type 2 diabetes mellitus patients (diagnosed before age 45 years). The centre receives patients from Pune and adjoining areas for routine treatment, while some are referred for specific problems (secondary and tertiary referrals). We recruited consecutive young type 2 diabetic patients attending the outpatient department in the study. Diagnosis of type 2 diabetes mellitus was based on clinical criteria. Those clinically judged to be insulin-dependent (history of ketoacidosis, to be unresponsive to oral hypoglycaemic agents, to be on continuous insulin treatment since diagnosis) or to have exocrine pancreatic disease (fibrocalculous pancreatic diabetes) or to fulfil clinical criteria for monogenic forms of diabetes were excluded. We studied only one member from a family where more than one attended the clinic. Sixty-five per cent of the patients had osmotic symptoms or weight changes at diagnosis. Subsequent clinical course and response to treatment with oral hypoglycaemic agents supports the diagnosis of type 2 diabetes. The control subjects were parents of children studied in the Pune Maternal Nutrition Study (PMNS) in six villages near Pune [12], who were normal glucose tolerant on a 75 g OGTT. The clinical characteristics of the case and the control subjects are presented in Table 1. The study was approved by the Institutional Ethics Committee following the Indian Council of Medical Research guidelines for research on human subjects. All subjects gave written informed consent.

Table 1 Clinical characteristics of the study population

We genotyped the SNPs rs7903146 and rs12255372, which showed the strongest association in the study by Grant et al. [10], and rs4506565, as it was the best correlated proxy of these SNPs in HapMapII. SNP genotyping was carried out by direct sequencing (using ABI3730 Genetic Analyzer; Applied BioSystems, Foster City, CA, USA) of the purified PCR products, amplified using primers flanking the three variants (available on request). Fifteen per cent of the randomly selected samples were re-genotyped and the discrepancy rate on duplicate genotyping was 1 / 752 (0.13%).

Allele and genotype frequencies were calculated and consistency of genotype frequencies at each SNP with Hardy–Weinberg equilibrium was tested on a contingency table of observed and expected genotype frequencies using the Markov simulation-based goodness of fit test [13]. The allele and genotype frequencies between the diabetic and the control subjects were compared using standard contingency table analysis. Continuous trait data are shown as mean ± SD unless otherwise specified. Variables with skewed distributions were log transformed to satisfy assumptions of normality and back-transformed values are shown. We used regression analysis to examine the effects of TCF7L2 genotypes on quantitative traits. Homeostasis model assessment of insulin resistance (HOMA-R) was calculated using the equation (fasting plasma insulin × fasting plasma glucose) / 22.5, and all analyses were carried out with Stata (version 7; Stata, College Station, TX, USA).

Results and discussion

The allele and genotype frequencies for all the variants in the patients and the control subjects are shown in Table 2. The risk allele frequency at rs12255372 in control subjects was lower in Indian subjects than Icelandic subjects (22 vs 29%, p = 0.001), but that at rs7903146 was very similar to the cohorts reported by Grant et al. [10]. The genotype distribution at all the SNPs did not show any deviation from the Hardy–Weinberg equilibrium (p > 0.05 in controls). All three SNPs showed strong linkage disequilibrium with D′ = 0.86–0.93 and r 2 = 0.71–0.88 (Electronic supplementary material [ESM] Table 1).

Table 2 Allelic and genotypic frequencies and estimates of relative risks for the TCF7L2 variants in type 2 diabetes patients and control subjects

We found a strong association of TCF7L2 variants with type 2 diabetes mellitus in this Indian population, replicating observations in European populations [10, 11]. All the SNPs showed similar association with the rare allele (rs12255372 [OR = 1.50, 95% CI = 1.24–1.82, p = 4.0 × 10−5], rs7903146 [OR = 1.46, 95% CI = 1.22–1.75, p = 3.0 × 10−5] and rs4506565 [OR = 1.48, 95% CI = 1.24–1.77, p = 2.0 × 10−5]). For all variants, the risk of type 2 diabetes mellitus in homozygotes was higher than for the heterozygous carriers (Table 2). This was most marked for rs12255372, with heterozygotes OR = 1.43 (95% CI = 1.11–1.83, p = 5.3 × 10−3) and homozygotes OR = 2.28 (95% CI = 1.40–3.72, p = 6.9 × 10−4). These results support the multiplicative mode of inheritance proposed by Grant et al. [10].

As recent studies have suggested that the higher prevalence of type 2 diabetes mellitus in the South Asian Indians could be partially due to central obesity and altered distribution of fat and muscle mass [2], we investigated whether TCF7L2 genotypes were associated with altered BMI and WHR. However, no significant association with BMI or WHR was observed in the patients or control subjects. Similarly, the TCF7L2 genotypes were not associated with age at diagnosis, sex or family history of diabetes mellitus (p > 0.05) (data not shown). However, in the non-diabetic subjects, we found that possessing the at-risk allele at rs12255372 predicted higher fasting and 2-h plasma glucose concentrations and higher HOMA-R, suggesting both a defect in insulin secretion from the beta cells and an increase in insulin resistance (Table 3).

Table 3 Correlation of TCF7L2 rs12255372 genotype with measures of quantitative traits in non-diabetic control subjects

Our results confirm that genetic variation in TCF7L2 is as strongly associated with type 2 diabetes mellitus in India, as previously described in European populations [10, 11]. Variation in TCF7L2 is the most significant genetic factor for diabetes mellitus described in the Indian population to date [8, 9]. The similar strength of association for TCF7L2 in this Indian study contrasts with the results observed for the PPARG Pro12Ala SNP. The 12Ala allele is consistently associated with protection of type 2 diabetes in European populations, but not in South Asian Indians (n = 697) in India and in Dallas, TX, USA [9]. However, in view of the genetic diversity of Indians, both these results need to be replicated in other groups of patients. Our results show that the strongest risk variant found for type 2 diabetes mellitus to date in European populations is a risk allele of similar effect size in Indians. This is consistent with genetic factors playing an important role in risk for type 2 diabetes mellitus, even in populations where environmental factors may have resulted in a dramatic recent increase in prevalence.

A potential limitation of our study is the relatively small number of patients and control subjects; thus, estimates of the association have relatively large confidence risks, especially for homozygous subjects. However, they are larger than the replication cohorts used in the initial study by Grant et al. [10]. Despite testing our control subjects with OGTT, there may be some potential diabetic patients in the control group, since they were younger and thinner than the patients and might develop diabetes in later life. However, the fact that some control subjects are at risk of diabetes would result in a slight reduction in the ORs, so our results may represent an underestimate of the strength of the association. Since we do not have data on the 30-min insulin levels in the control subjects, the dynamic measurements of insulin secretion, such as the insulinogenic and insulin disposition indices, are not available.

It is not yet clear how TCF7L2 contributes to the pathogenesis of type 2 diabetes mellitus [10]. Association of the at-risk alleles in non-diabetic control subjects with higher glycaemia and higher HOMA-R suggests defects in both insulin secretion and insulin sensitivity mechanisms for its possible effects. A causal variant or functional defect in this gene is yet to be identified, so further study of the gene is required.

To conclude, we have replicated the strong association of variants in TCF7L2 gene and shown it to be a susceptibility gene for type 2 diabetes mellitus in the South Asian Indians. In view of a similar strength of association in Indians as in European populations, TCF7L2 is an important susceptibility marker for risk of type 2 diabetes mellitus in different ethnic groups.