Discovery of 318 novel loci for type-2 diabetes and related micro- and macrovascular outcomes among 1.4 million participants in a multi-ethnic meta-analysis.
Marijana VujkovićiD, Jacob M. KeatoniD, Julie Ann LynchiD, Donald R. MilleriD, Jin J. ZhouiD, Catherine TcheandjieuiD, Jennifer E. HuffmaniD, Themistocles L. AssimesiD, Kimberly LorenziD, Xiang ZhuiD, Austin T. HilliardiD, Renae JudyiD, Jie HuangiD, Kyung Min LeeiD, Derek M. KlariniD, Saiju PyarajaniD, John DaneshiD, Olle MelanderiD, Asif Rasheed, Nadeem Hayat MallickiD, Shahid Hameed, Irshad Hussain Qureshi, Muhammad AfzaliD, Uzma Malik, Anjum JalaliD, , Saib S. Gappy
Abstract
We investigated type 2 diabetes (T2D) genetic susceptibility in a multi-ethnic meta-analysis of 228,499 cases and 1,178,783 controls in the Million Veteran Program (MVP) and other biobanks. We identified 558 autosomal and 10 X-chromosome T2D-associated variants, of which 286 autosomal and 7 X-chromosome variants were previously unreported. Ancestry-specific analyses identified 25 additional novel T2D-susceptibility variants. Transcriptome-wide association analysis detected 3,568 T2D-associations with T2D-colocalized genetically predicted gene expression of 804 genes in 52 tissues, of which 687 are novel. Fifty-four of these genes are known to interact with FDA-approved drugs and chemical compounds. T2D polygenic risk score was strongly associated with increased the risk of T2D-related retinopathy, and additionally showed evidence for association with chronic kidney disease (CKD), neuropathy, and peripheral artery disease (PAD). We investigated the genetic etiology of T2D-related vascular outcomes in the MVP and observed statistical SNP-T2D interactions at 13 variants, including 3 for coronary heart disease, 1 for PAD, 2 for stroke, 4 for retinopathy, 2 for CKD, and 1 for neuropathy. Our findings may identify potential novel therapeutic targets for T2D and genomic pathways that link T2D and its vascular outcomes.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 777
- FWCI
- 15.15
- Field percentile
- top 1%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · max effort · 8 Nov 2026 · from the full text of version 1 (25 Nov 2019) · CC BY 4.0
Multi-ethnic type 2 diabetes (T2D) meta-analysis of 228,499 cases and 1,178,783 controls finds 318 previously unreported loci, 804 expression-colocalized genes, and 13 SNP-by-T2D vascular-outcome interactions.
Most T2D genome-wide association studies (GWAS) have sampled European and Asian ancestry groups, and the genetics of diabetic vascular sequelae remain unclear because few large, multi-ethnic, deeply phenotyped cohorts link genotypes to clinical records. Using electronic health record-linked Million Veteran Program (MVP) data, this work adds ancestry-diverse discovery, polygenic and SNP-by-T2D interaction tests for vascular outcomes, and expression-based gene prioritization.
Key results
ShownFixed-effects meta-analysis across five ancestry groups identified 558 autosomal and 10 X-chromosome sentinel T2D variants; adding ancestry-restricted scans, 318 variants were previously unreported, and new loci showed smaller average per-allele effects than established ones.
558 autosomal sentinels split into 286 new and 272 known; 7 of 10 X-chromosome signals new; 25 more new variants from ancestry-specific analyses (318 overall); mean per-allele effect 0.032 ± 0.012 for new vs 0.054 ± 0.045 for established SNPs; chip heritability explained 19% of T2D risk. · Abstract; Single-variant autosomal analyses; Chromosome-X analyses; Discussion
SuggestedAn African American meta-analysis of over 56K participants found 21 genome-wide significant loci, including 3 new autosomal signals that appeared ancestry-specific (risk alleles more common in African than European reference populations), plus 1 ancestry-specific X-chromosome locus.
21 loci, 16 in strong linkage disequilibrium (LD) with established variants; 44 (7.9%) loci had significantly different European vs African American effects, with SLC30A8, PTPRQ, GRB10 and COLB loci stronger in African Americans; among prior African-ancestry loci only rs3842770 (INS-IGF2) replicated, and rs73284431 did not despite minor allele frequency 9% vs 10%. · Single-variant autosomal analyses; Chromosome-X analyses; Effect heterogeneity between Europeans and…; Discussion
ShownS-PrediXcan across 52 tissues with COLOC colocalization prioritized 804 genes whose genetically predicted expression associates with T2D, most not previously tied to diabetes; 54 S-PrediXcan genes are targets of approved drugs or compounds without T2D indications or diabetic adverse-event reports.
4,468 significant gene-tissue pairs; 3,568 pairs with colocalization posterior P4 > 0.5 spanning 804 genes, 687 absent from GWAS catalog entries for diabetes or its complications; 873 genes missed by nearest-gene annotation; 54 of 991 S-PrediXcan genes link to 283 FDA-approved drugs or compounds, and DGIdb 3.0 gave 322 gene-drug pairs predicted to lower glucose. · Transcriptome-wide association analyses; Assessment of gene–drug relationships; S-PrediXcan and colocalization analyses
ShownAmong European-ancestry MVP participants, a T2D polygenic score stratified T2D risk and, within T2D cases, associated most strongly with retinopathy and also with neuropathy, chronic kidney disease (CKD) and peripheral artery disease (PAD), but not coronary heart disease (CHD) or ischemic stroke.
Genome-wide polygenic risk score (gPRS) weights from DIAMANTE Europeans via PRSice (pruning r^2 0.8, P threshold 0.05); top vs bottom decile T2D odds ratio 5.21 (95% CI 4.94-5.49), cross-sectional; complication P-values used continuous gPRS within T2D cases, with decile odds ratios given only in tables and figures. · Polygenic risk scores and T2D-related micro- and…; Polygenic risk scores and risk of T2D and…; Abstract; Discussion
SuggestedGenome-wide SNP-by-T2D interaction scans of micro- and macrovascular outcomes in European-ancestry MVP participants found 13 variants whose outcome associations differ by diabetes status, including 9p21 and SORT1 for CHD, UMOD for CKD, and TCF7L2 and GJA8 for retinopathy.
Interaction P < 5×10^-8 plus P < 0.001 in at least one stratum: 3 CHD, 2 CKD, 1 PAD, 1 neuropathy, 4 retinopathy and 2 stroke variants; rs1831733 (9p21), rs602633 (SORT1) and rs34857077 (UMOD) were also genome-wide significant within a stratum; outcome case counts ranged from 11,796 (stroke) to 67,403 (CKD). · T2D-related micro- and macrovascular outcomes; Analysis of T2D-related outcomes; Abstract; Discussion
Methods, data and reproducibility
Ancestry-stratified GWAS from MVP and other cohorts were pooled by inverse-variance fixed-effects meta-analysis (METAL). Follow-up comprised conditional, coding-variant, phenome-wide (PheWAS), S-PrediXcan/COLOC, drug-gene, LD score regression and DEPICT analyses, plus polygenic-score and SNP-by-T2D interaction models of vascular outcomes in MVP.
- Multi-ethnic T2D GWAS meta-analysis spanning European, African American, East Asian, South Asian and Hispanic participants from MVP and non-overlapping cohorts · human · n = 228,499 cases; 1,178,783 controls
- Million Veteran Program genotyped veterans (custom Affymetrix Axiom array) with linked health records; 72.1% European, 19.5% African American, 7.5% Hispanic, 0.9% Asian · human · n = 273,409 participants (91.6% male)
- External summary statistics from DIAMANTE, Biobank Japan, Pakistan Genomic Resource, Penn Medicine Biobank, MedStar/PennCath and the Malmö Diet and Cancer Study · human · n = 917,392 European; 2,647 African American; 213,834 Asian
- MVP ancestry-stratified subsets used as LD reference for GCTA-COJO, variance-explained models and PheWAS of loss-of-function variants · human · n = 197,066 European (69,869 T2D cases); 53,445 African American
- MVP European-ancestry participants used for SNP-by-T2D interaction tests on CKD, CHD, PAD, acute ischemic stroke, retinopathy and neuropathy · human · n = 11,796 (stroke) to 67,403 (CKD) cases per outcome
- Expression quantitative trait locus (eQTL) references for 52 tissues: 48 GTEx tissues, kidney glomerulus and tubulus, pancreatic islet alpha and beta cells · human (GTEx) · n = 52 tissues; islet panels of 30 alpha and 19 beta cells
- LD Hub collection of precomputed GWAS summary statistics on other complex traits and diseases, used for genetic correlation with European T2D results · human · n = 774 traits
- 1000 Genomes phase 3 (version 5) imputation reference, whose individuals were also co-analyzed for ancestry principal components and LD covariance · human (1000 Genomes) · n = 2,504 individuals added to MVP for principal component analysis
- Methods & tools
- PLINK2a logistic-regression association testing, METAL inverse-variance fixed-effects meta-analysis with Cochran's Q heterogeneity, EAGLE v2 prephasing with Minimac3 imputation (MiniMac4 named in code statement), Affymetrix Power Tools v1.18, KING relatedness, FlashPCA principal components, HARE ancestry and race/ethnicity harmonization, XWAS chromosome-X quality-control pipeline, GCTA-COJO conditional analysis, Ensembl Variant Effect Predictor and PheWAS R package (R v3.2.0), S-PrediXcan with COLOC colocalization, SUGEN v8.8 interaction models (robust variance; relative excess risk due to interaction), PRSice prune-and-threshold polygenic scores; LD Hub v1.9.3 LD score regression, DEPICT enrichment; DGIdb 3.0 drug-gene and SIDER side-effect databases
- Data & accessions
- http://ldsc.broadinstitute.org/
- Reagents & environment
- 1000 Genomes phase 3 v5 imputation panel (Haplotype Reference Consortium panel for DIAMANTE); Affymetrix Power Tools v1.18, KING, EAGLE v2, Minimac3, FlashPCA, PLINK2a, METAL, GCTA-COJO, S-PrediXcan, SUGEN v8.8, PRSice, LD Hub v1.9.3, DGIdb 3.0, R v3.2.0.
- Assay conditions & compute
- Genotyping on a custom Affymetrix Axiom MVP 1.0 biobank array (723,305 SNPs), called in two batches; no computational hardware, runtime or memory reported.
- Reproducibility gaps
- Summary statistics are promised via dbGaP but accession codes were pending at posting; no individual-level MVP data release or study-specific analysis code is described, only pointers to public tool websites.
Limitations (6)
- No significant predicted-expression associations emerged in pancreatic alpha or beta cells, which the authors attribute to small islet reference samples (30 alpha, 19 beta cells) and highly variable islet transcriptomes. · stated by the authors
- Proximity of a coding variant to a tagging SNP is acknowledged as insufficient evidence of causality, and sequencing is described as necessary to pinpoint causal variants and ultra-rare allele contributions. · stated by the authors
- Apart from skeletal muscle, tissues most enriched for T2D expression associations (e.g., spinal cord, basal ganglia) have no known role in T2D etiology, so the authors hypothesize eQTL effects are broadly shared rather than tissue-specific. · stated by the authors
- The MVP discovery cohort is 91.6% male and over-represents men compared with the general population, though it is described as representative of US race and ethnicity. · stated by the authors
- Reporting is internally inconsistent: genetic-correlation significance is P < 5.0×10^-8 in Results but Bonferroni P < 6.5×10^-5 in Methods, the Figure 2 legend calls CKD macrovascular and stroke microvascular, and complication-model covariates differ between that legend and Methods. · noted in this summary
- Vascular-outcome interaction and polygenic analyses were confined to European-ancestry MVP participants with mostly code-based phenotypes and lacked independent replication; several interaction hits reached only nominal (P < 0.001) significance within a T2D stratum. · noted in this summary
Who should read:Read this if you study T2D genetics beyond European ancestry, need expression-prioritized genes or drug-repurposing leads for diabetes, or model genetic risk of retinopathy, kidney disease and other diabetic vascular complications.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Multi-ancestry genome-wide study in >2.5 million individuals reveals heterogeneity in mechanistic pathways of type 2 diabetes and complications
Suzuki Ken, Konstantinos Hatzikotoulas, … H. Janaka de Silva·Yearbook of pediatric endocrinologyPeer-reviewed·Sep 8, 2023
- Novel susceptibility loci and genetic regulation mechanisms for type 2 diabetes
Angli Xue, Yang Wu, … Jian Yang·Nature CommunicationsPeer-reviewed·Jul 23, 2018
- Multi-omics characterization of type 2 diabetes associated genetic variation.
Ravi Mandla, Kimberly Lorenz, … Josep Maria Mercader·medRxiv·Jul 15, 2024
- Time-to-Event Genome-Wide Association Study for Incident Cardiovascular Disease in People with Type 2 Diabetes Mellitus
Soo Heon Kwak, Ryan B. Hernandez-Cancela, … James B. Meigs·Diabetes CarePeer-reviewed·Apr 23, 2024
- Type 2 diabetes genetics in 125,000 admixed adults from Mexico City
Jason Torres, Jaime Berumen-Campos, … Roberto Tapia‐Conyer·medRxiv·Jun 25, 2026
- Integrative cross tissue analysis of gene expression identifies novel type 2 diabetes genes
Jason Torres, Alvaro Barbeira, … Hae Kyung Im·bioRxiv·Feb 27, 2017