Deep re-sequencing of 9 type 2 diabetes GWAS loci by comparison of extremes of dynamic indices of insulin secrection

Background and aims: The susceptibility genes for Type 2 Diabetes (T2D) identified so far are mainly involved in beta-cell function. To this point, Genome-Wide Association Studies (GWAS) have identified a large number of loci. Despite this large number, the major part of T2D inheritance is still uncovered. This missing heritability might be explained by multiple, low-frequency variants that are not captured by GWAS. A powerful approach to highlight causal variants is to deep re-sequence candidate genes. To enhance the probability to highlight causal or protective variants together with the appropriate statistical power, we aimed to identify by Next Generation Sequencing (NGS) lowfrequency variants in GWAS loci for T2D, searching for primary defects in beta-cell insulin secretion. For this, we applied a two stage study design: Stage 1, deep re-sequencing of coding and flanking regions of 9 candidate genes that reach GWAS significance (p<10-8), in individuals selected from the extremes of insulin secretion, adjusted for insulin resistance, i.e. Disposition Index (DI); Stage 2, confirmation of the association by genotyping the variants differently distributed between the two extremes of insulin secretion in larger and independent groups of Italian adults and children (N=3130). Materials and methods: In a large population, very well characterized from OGTT, measures of insulin secretion and resistance have been calculated, including insulinogenic index (IGI30), ISI (insulin-sensitivity index), and DI (IGI30xISI). NGS was performed on MiSeq system (Illumina) with TruSeq Custom Amplicon approach. Variants are investigated by proper bioinformatics tools. Discovered variants will be genotyped by Real-Time PCR or by suitable methods such as SNP array and genotyping by sequencing. Results: We sequenced 383 subjects from the discovery sample. Preliminary results show more than 1500 variants in this sample. Bioinformatics tools predict that at least 122 of them may have a functional effect on protein, being missense or nonsense. A small but relevant part of passing-filter variants seems to be newly discovered (no rs) or to presumptively affect protein function. We then searched for a different distribution of all the infrequent variants within the two extremes of DI. Variants in one of the genes, ADAMTS9, were significantly associated with the higher extreme (>80%) of DI distribution (OR= 1.30 p=0.03). A similar trend was observed for four of the 9 candidate genes. Multivariate analyses showed that carriers of one o more variants have an OR= 9.06 (1.73-47.42) p=0.009 and OR= 8.52 (1.32-55.01) p=0.024 respectively, to be in the >80% extreme of DI. Conclusion: The next stage involves the in-depth analysis of all the variants discovered, evaluating distribution, frequency, and possible function, together with the replication studies in large cohorts to confirm the association with altered insulin secretion. We expect that this study will deliver several results: from confirmatory gene association to newer T2D associated polymorphisms, to possible new insights into potential biological mechanisms influencing T2D pathogenesis. Supported by: fellowship granted by the Italian Diabete Ricerca Foundation and MSD Italy Disclosure: D. Bailetti: Grants; fellowship granted by the Italian Diabete Ricerca Foundation and Merck Sharp & Dohme Italy