The Ala12 variant of PPARG has been associated with protection from Type 2 diabetes [1]. It has also been associated with increased insulin sensitivity and lower body mass, and seems to modulate response to environmental cues including diet and exercise [2, 3].

Although not changing the coding sequence of PPARG, the C1431T polymorphism has been associated with increased body weight and is in tight allelic disequilibrium with the Ala12 variant [3]. Allelic disequilibrium can result in cross contamination of association studies where only one of the variants is considered. In this report we consider both variants in haplotypic context and dissect their specific associations with susceptibility to Type 2 diabetes.

Subjects and methods

Tayside diabetic population

In Tayside, Scotland detailed information on all individuals with diabetes mellitus is recorded on an electronic clinical information system which is known as DARTS (Diabetes Audit and Research in Tayside Scotland) and has been described elsewhere [4]. At the time of this study, about 2500 of the approximately 11,500 people registered on DARTS were enrolled in the genetic study. Of these, 1997 had Type 2 diabetes. This free-living clinical population was therefore selected solely on the basis of having Type 2 diabetes and attending diabetes clinics in Tayside. All subjects were white.

Non-diabetic children population (Tayside)

Healthy school children (4–9 years old) recruited for a gene–nutrition interaction study in Tayside, Scotland were used as a control population (n=2444). This population was largely white (>97%), and being selected at random from local primary schools, was derived from the same population as the Type 2 diabetic patients.

Middle-aged control population

This consisted of 1061 white middle-aged adults from three Scottish cities, recruited without regard to diabetic status. The first group numbered 192 adults randomly selected from the Glasgow area [3]. The second group consisted of 489 adults from Tayside, recruited as control subjects for studies of breast cancer, colon cancer and systemic sclerosis [3]. The third population of 380 women was randomly selected from the Aberdeen Prospective Osteoporosis Screening Study (APOSS) [5]. These women had been genotyped for PPARG polymorphisms as part of a prospective study on the genetic and environmental determinants of bone mass, bone loss and osteoporotic fracture. Whereas the APOSS population comprised women only, the allele frequencies between men and women were identical in the other study populations. The allele frequencies in each individual control population reflected the final combined dataset. The individual case-control comparisons for each study will be presented elsewhere. Full ethics approval and informed consent were obtained for each of the studies referred to above.

Genotype determination

Mouthwash samples were collected from the children and blood samples were taken from the adults for DNA analysis. DNA was stored at −20°C on 96-well plates. Genotyping for PPARG Pro12Ala and C1431T polymorphisms was done on all populations using Taqman (Applied Biosystems, Foster City, Calif., USA.) allelic discrimination assays as described [3].

Statistical analyses

All statistical analyses were carried out using SPSS for Windows version 11 and Instat for Macintosh version 3 (Graphpad, San Diego, Calif., USA). Allele frequencies were compared between populations using Pearson’s chi square. Allelic associations were analysed as described [3]. Differences in estimated haplotype frequency between populations or between phenotype were compared by log likelihood ratio using chi square. Fisher exact tests were used to determine odds ratios. A p value of less than 0.05 was considered significant.

Results and discussion

The Ala12 variant was present at a frequency of 0.111 in the 1997 individuals genotyped from the Tayside Type 2 diabetic cohort (Table 1). This is consistent with a summary frequency of 0.118 obtained from 3444 white Type 2 diabetic subjects compiled from nine studies [6, 7]. Initial analysis showed that, similarly to previous reports, the Ala12 allele was under-represented in the Type 2 diabetic cohort compared with adult population controls (Table 1) (OR=0.74, 95% CI: 0.62 to 0.88, p=0.0006). Indeed, in the adult control populations the allele frequency was highly concordant with a literature-based sample of 4065 white control subjects (0.143 vs 0.139 from the literature). The association with Type 2 diabetes was consistent with a co-dominant model, as a trend between magnitude of protection and Ala12 allele dosage was observed (chi squared test for trend = 12.8, p=0.0003). However, the Ala12 frequency was very similar in the Type 2 diabetic population and the children’s cohort (OR=0.92, 95% CI: 0.79 to 1.05). The latter was even closer to the literature-based estimate of the Ala12 allele frequency in Type 2 diabetic patients (children: 0.121, diabetic: 0.118). Thus, the Ala12 variant was under-represented in the children, when compared with the adult control populations (OR=0.81, 95% CI: 0.68 to 0.95, p=0.012).

Table 1 PPARG genotype frequencies in the three groups, with allele frequencies and 95% confidence intervals
Full size table

The T1431 variant was also under-represented in the Type 2 diabetic population compared to the adult population. However, this association was weaker than that observed with the Ala12 variant. The T1431 variant was in allelic association with the Ala12 variant, with about 70% of individuals with Ala12 also carrying T1431 (Table 2). This raised the possibility that the protection from diabetes observed with the single polymorphisms could result from their association with each other.

Table 2 PPARG haplotype frequencies in the three cohorts
Full size table

We therefore examined the distribution of Ala12 variant with and without the T1431 variant in the two cohorts. This was done by statistically estimating the frequency of specific haplotypes from the combined observed genotypes (Table 2). When the Ala12 variant was examined with regard to the chromosome not bearing the T1431 variant (Pro-C vs Ala-C), it was more substantially under-represented in the Type 2 diabetic population (OR=0.66, p=0.003) than in the adult control populations. However, when the Ala12 variant was examined in chromosomes bearing the T1431 variant (70% of all Ala12 carriers; Pro-T vs Ala-T), no association was seen (OR=0.99, p=0.94).

Whereas these data further support the hypothesis that Ala12 variant confers protection against Type 2 diabetes, the protective signal in our data was totally derived from a specific minority haplotype. Moreover, the protective effect attributable to the Ala12 variant was absent in cases where the Ala12 variant was on the same chromosome as the T1431 variant. This would suggest that only a minority subgroup of Ala12 carrier individuals are protected from Type 2 diabetes. These data also support our previous studies, where we observed that the association of the Ala12 allele with reduced body weight is modulated by its chromosomal context relative to the silent T1431 allele. This modulation could be due to an independent association of the T1431 variant with increased body weight [3]. As body weight is a major determinant of Type 2 diabetes susceptibility, the observed effects of these variants on body weight could well be related to differential susceptibility to Type 2 diabetes.

The use of children from the same region as the population with diabetes to estimate the local birth frequency of these variants showed that the Ala12 frequency had mainly changed in the relatively healthy middle-aged population-based controls, rather than in the diseased cohort. This is consistent with the emerging view that variation at the PPARG locus is likely to have pleiotropic effects, modulating susceptibility to hypertension and myocardial infarction [8, 9, 10]. The Ala12 variant could therefore provide protection not only from developing Type 2 diabetes, but also from more general metabolic morbidity. This could result in the selective recruitment of elderly Ala12 carriers in control populations.

In conclusion, we have confirmed that the Ala12 variant of PPARG modulates susceptibility to Type 2 diabetes in a large population-based cohort of Type 2 diabetics and have shown that this could be modulated by additional common variation at the PPARG locus.