IRS1 G972R polymorphism and type 2 diabetes. a paradigm for the difficult ascertainment of the contribution to disease susceptibility of ‘low-frequency–low-risk’ variants

Aims/hypothesis: The aim of the study was to determine the association between IRS1 G972R polymorphism and type 2 diabetes; published data concerning this association have been conflicting. To obtain further insight into this topic, we performed a meta-analysis of all available case-control studies. Methods: We performed a meta-analysis of 32 studies (12,076 cases and 11,285 controls). Results: The relatively infrequent R972 variant was not significantly associated with type 2 diabetes (OR 1.09, 95% CI 0.96-1.23, p∈=∈0.184 under a dominant model). Some evidence of heterogeneity was observed across studies (p∈=∈0.1). In the 14 studies (9,713 individuals) in which the mean age at type 2 diabetes diagnosis was available, this variable explained 52% of the heterogeneity (p∈=∈0.03). When these studies were subdivided into tertiles of mean age at diagnosis, the OR for diabetes was 1.48 (95% CI 1.17-1.87), 1.22 (95% CI 0.97-1.53) and 0.88 (95% CI 0.68-1.13) in the youngest, intermediate and oldest tertile, respectively (p∈=∈0.0022 for trend of ORs). Conclusions/interpretation: Our findings illustrate the difficulties of ascertaining the contribution of 'low-frequency-low-risk' variants to type 2 diabetes susceptibility. In the specific context of the R972 variant, ~200,000 study individuals would be needed to have 80% power to identify a 9% increase in diabetes risk at a genome-wide significance level. Under these circumstances, a strategy aimed at improving outcome definition and decreasing its heterogeneity may critically enhance our ability to detect genetic effects, thereby decreasing the required sample size. Our data suggest that focusing on early-onset diabetes, which is characterised by a stronger genetic background, may be part of such a strategy