Type 2 diabetes TCF7L2 risk alleles reduce beta-cell function and increase birth weight. R.M. Freathy¹, M.N. Weedon¹, M.I. McCarthy², M. Walker³, G.A. Hitman⁴, Y. Ben-Shlomo⁵, G. Davey Smith⁵, S.M. Ring⁵, A.T. Hattersley¹, T.M. Frayling¹. 1) Peninsula Medical School, Exeter, UK; 2) Oxford, UK; 3) Newcastle, UK; 4) London, UK; 5) Bristol, UK.
   The transcription factor TCF7L2 is the most important type 2 diabetes gene found to date. Heterozygotes and minor allele homozygotes for rs12255372, which constitute 41% and 9% of the population, have odds ratios for type 2 diabetes of 1.30 and 1.66 respectively, and this association has been replicated in multiple studies. The mechanism by which variation in TCF7L2 predisposes to diabetes is unknown. Identifying intermediate traits in the non-diabetic population will help define the mechanism of action but is notoriously difficult. We studied the role of common variation in the TCF7L2 gene in diabetes-related intermediate traits in the general population. We assessed the role of TCF7L2 in beta-cell function using 1053 non-diabetic subjects from two studies (Barry Caerphilly Growth Study, n=631, mean age 25; Warren 2 Extension, n=422, mean age 39) with oral glucose tolerance test data available, and in birth weight using 7211 children and 6573 mothers from the ALSPAC study. All subjects were UK whites. Each copy of the type 2 diabetes predisposing allele of rs12255372 (frequency 0.29) decreased beta-cell function (early insulin response) by 0.15 (0.06-0.24) SD (P=0.001). Each copy of the diabetes risk allele in the fetus was associated with a 21g (3g-40g) increase in birth weight (ANOVA P=0.004), while each copy of the diabetes risk allele in the mother was associated with a 32g (12g-52g) increase in offspring birth weight (ANOVA P=0.0006). There was no association with fasting glucose, insulin or BMI (all P>0.05). Our results suggest that prior to the development of type 2 diabetes, common variation in TCF7L2 is associated with reduced beta-cell function and increased birth weight. Our results indicate that with large sample sizes it is possible to identify subtle effects of established disease risk alleles in the general population.