Identifying genetic risk factors for type 2 diabetes and related traits (e.g. lipids, obesity and blood pressure) in the Mexican population

    Type 2 diabetes (T2D) is present in epidemic proportions in Native American and Mexican-American populations. In the US, the prevalence of T2D in Native Americans and Latinos is higher than in non-Hispanic whites. The prevalence of diabetes-related complications is also very high in these populations. Similarly, in Canada, the T2D rates in Aboriginal communities is 3 to 5 times higher than in the general population (

    In collaboration with Dr. Miguel Cruz, from the Medical Center "Siglo XXI" in Mexico City (Mexico), we carried out a Genome-Wide Association (GWA) study of T2D in a sample from Mexico using high density microarrays (Affymetrix 5.0). We then joined efforts with the research groups of Craig Hanis (University of Texas at Houston) and Nancy Cox (University of Chicago), in order to perform a meta-analysis including the Mexico City study, and a similar study in Starr County, Texas. We observed highly suggestive evidence of association with T2D in 10 genomic regions, including two regions identified in previous GWA studies: HNF1A (first identified in Europeans) and KCNQ1 (first identified in Japanese). In a meta-analysis including the two Hispanic samples and data from the large European DIAGRAM+ study, two regions (CDKN2A/CDKN2B and HNF1A) reached genome-wide significance (p<5x10-8), and three additional regions showed highly suggestive results (p<10-5) (IGF2BP2, HNF1A, and the previously undescribed C14orf70). An analysis of expression data showed that among the top signals observed in the Mexico City and Starr County studies there is a significant enrichment of genetic markers that predict gene transcript levels in muscle and adipose tissue. The results of this study were published as back-to-back papers in the journal Diabetologia (see publication list).

   We also collaborated with researchers throughout the world in a meta-analysis including many T2D GWA studies from different population groups, including East Asians, Europeans, Hispanics and South Asians. The results of this meta-analysis were published in the prestigious journal Nature Genetics (see list of publications). Seven new T2D susceptibility loci were identified in this collaborative effort. Additionally, this study showed that there was a remarkable concordance between populations in the direction of effects for loci with nominal evidence of association (p<0.001), but no such concordance was observed for loci showing no significance. This indicates that there are many common causal variants with relatively small effects on T2D susceptibility that are shared by different populations and have not been identified due to limited sample sizes, but are amenable for discovery in future meta-analytic efforts with larger samples. Finally, combining GWA studies from different populations led to improvements in the resolution of fine-mapping of common variant association signals at several loci. This is due to the increased sample sizes and importantly, the differences in patterns of Linkage Disequilibrium (LD) between populations. We are currently expanding these analyses using datasets imputed with the latest version of the 1,000 Genomes Project reference panels (Phase 3).

   We have data for many other traits of medical importance, including lipids (e.g. LDL-cholesterol, HDL-cholesterol, Total-cholesterol, Triglycerides), anthropometric traits (e.g. Body Mass Index, Waist-to-Hip Ratio) and blood pressure. Additionally, we expanded our original study to increase the number of samples, using the Latino Axiom microarray, which provides better genome coverage than previous arrays. Very recently, in collaboration with many research groups from the US, we carried out one of the largest meta-analyses of lipid traits in Hispanic populations, including 4,383 individuals in the discovery sample and 7,876 in the replication sample (Below et al. 2016. Sci. Rep. 5:19429). In this study, we described genome-wide signals in many regions previously described in European populations. Overall, as for T2D, we saw a substantial concordance of direction of effects, and little heterogeneity of effect sizes, when we compared the signals reported in European studies with our Hispanic dataset. However, for some regions (ABCA1 and LIPC for HDL-cholesterol, and NCAN /MAU2 for triglycerides), the signals that we identified in our study seem to be independent from those reported in European samples. A meta-analysis of the European GLGC and our Hispanic datasets identified five novel regions reaching genome-wide significance. Finally, the top meta-analysis signals were found to be enriched for SNPs associated with gene expression in a tissue-specific fashion, suggesting an enrichment of tissue-specific function in lipid-associated loci.

Manhattan plots showing the results of the genome-wide analyses for lipid traits (Below et al. 2016. Sci. Rep. 5:19429)