Trans-ethnic Meta-analysis and Functional Annotation Illuminates the Genetic Architecture of Fasting Glucose and Insulin

Abstract

Knowledge of the genetic basis of the type 2 diabetes (T2D)-related quantitative traits fasting glucose (FG) and insulin (FI) in African ancestry (AA) individuals has been limited. In non-diabetic subjects of AA (n = 20,209) and European ancestry (EA; n = 57,292), we performed trans-ethnic (AA+EA) fine-mapping of 54 established EA FG or FI loci with detailed functional annotation, assessed their relevance in AA individuals, and sought previously undescribed loci through trans-ethnic (AA+EA) meta-analysis. We narrowed credible sets of variants driving association signals for 22/54 EA-associated loci; 18/22 credible sets overlapped with active islet-specific enhancers or transcription factor (TF) binding sites, and 21/22 contained at least one TF motif. Of the 54 EA-associated loci, 23 were shared between EA and AA. Replication with an additional 10,096 AA individuals identified two previously undescribed FI loci, chrX FAM133A (rs213676) and chr5 PELO (rs6450057). Trans-ethnic analyses with regulatory annotation illuminate the genetic architecture of glycemic traits and suggest gene regulation as a target to advance precision medicine for T2D. Our approach to utilize state-of-the-art functional annotation and implement trans-ethnic association analysis for discovery and fine-mapping offers a framework for further follow-up and characterization of GWAS signals of complex trait loci.

A Type 2 Diabetes–Associated Functional Regulatory Variant in a Pancreatic Islet Enhancer at the ADCY5 Locus

“…Our data are consistent with the fact that rs11708067 was the strongest candidate for an effect on regulatory function based on its location in a putative islet enhancer element. rs11708067 also showed the highest posterior probability (highest log-Bayes factor 12.39) of driving the association signal from a credible set analysis (11). Considering that some GWAS loci have multiple functional regulatory variants responsible for the association (43), other variants at this locus may also contribute to allelic differences in transcriptional activity.…”

“…rs11708067-A was associated ( P = 1.7 × 10 −14 ) with higher fasting glucose levels in individuals of European ancestry (10) and higher fasting glucose levels ( P = 6.3 × 10 −6 ) in individuals of African ancestry (11). An additional variant, rs2877716, in strong LD ( r 2 = 0.837, 1000 Genomes phase 3 European [EUR]) (7) with rs11708067, was associated ( P = 4.2 × 10 −16 ) with 2-h glucose level (12).…”

“…The type 2 diabetes–, fasting glucose–, and 2-h glucose–associated variant, rs11708067, was also associated ( P = 2.0 × 10 −3 ) with impaired proinsulin-to-insulin conversion as individuals homozygous for the type 2 diabetes risk A allele showed higher proinsulin levels and a higher proinsulin/insulin ratio after an oral glucose tolerance test (15). Transethnic fine-mapping studies (11) and credible set analyses suggest rs11708067 is a functional variant at this locus.…”

Dissecting the genetics of complex traits using summary association statistics

“…In addition to increasing fine-mapping accuracy, these studies have also provided insights into polygenic architectures (see below) by identifying tissue-specific functional annotations that are enriched for causal disease signals. This can also be achieved by conducting fine-mapping without integrating functional annotation data (typically under a single causal variant assumption) and then overlapping the resulting credible sets with functional annotation data to assess enrichment 62–64 . Future integrative methods could increase fine-mapping resolution by integrating probabilistic functional annotations (e.g., ChIP-seq peak intensity) or modeling the strength of association between SNPs and chromatin marks in population-based studies 65,66 .…”

A Decade of Genetic and Metabolomic Contributions to Type 2 Diabetes Risk Prediction

“…Third, the majority of T2D variants have been identified in European populations or trans-ancestry meta-analyses that are still predominantly European [17••]. While many loci associated with T2D and related glycemic traits are transferrable across different ancestries [17••, 18, 19, 43–45], sample sizes of these non-European GWASs are often smaller than European GWASs, and so discovery of variants in these populations has lagged behind those in European populations [19, 27•, 29, 43, 46–48]. A GRS composed of T2D-related variants that are common in people of European descent, but of lower frequency in other ancestries, may not evenly predict incident T2D in diverse populations even if genetic effects were consistent across ancestries.…”