Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
[Preprint]. 2025 Jul 23:2025.07.21.25331778.
doi: 10.1101/2025.07.21.25331778.

Multi-ancestry polygenic risk scores for the prediction of type 2 diabetes and complications in diverse ancestries

Alicia Huerta-Chagoya  1   2   3   4 Joohyun Kim  5 Ravi Mandla  1   2   3 Yingchang Lu  5 Ken Suzuki  6   7   8 Lauren E Petty  5 Hong Kiat Ng  9 Jaewon Choi  10 Simon Lee  11 Madhusmita Rout  12 Kuang Lin  13 Linda S Adair  14 Adebowale Adeyemo  15 Habibul Ahsan  16 Masato Akiyama  17   18 Ping An  19 Sonia S Anand  20   21   22 Diane M Becker  23 Alain G Bertoni  24 Zheng Bian  25 Lawrence F Bielak  26 John Blangero  27 Michael Boehnke  28 Erwin P Bottinger  29   30   11 Donald W Bowden  31   32   33 Fiona Bragg  13   34 Jennifer A Brody  35 Thomas A Buchanan  36 Brian E Cade  37   38 Jin-Fang Chai  39 John C Chambers  9 Giriraj R Chandak  40   41 Li-Ching Chang  42 Kyong-Mi Chang  43   44 Miao-Li Chee  45 Chien-Hsiun Chen  42 Yuan-Tsong Chen  42 Zhengming Chen  13 Yii-Der I Chen  46 Ji Chen  47 Guanjie Chen  15 Shyh-Huei Chen  48 Wei-Min Chen  49 Ching-Yu Cheng  50   51   45   52 Yoon Shin Cho  53 Hyeok Sun Choi  53 Lee-Ming Chuang  54   55 Miguel Cruz  56 Mary Cushman  57 Swapan K Das  58 Ralph A DeFronzo  59 H Janaka deSilva  60 Latchezar Dimitrov  32 Ayo P Doumatey  15 Shufa Du  14 Qing Duan  61 Ravindranath Duggirala  27 Leslie S Emery  62 James C Engert  63 Daniel S Evans  64 Michele K Evans  65 Sarah Finer  66 Jose C Florez  1   4   67 James S Floyd  35 Myriam Fornage  68 Elizabeth G Frankel  5 Barry I Freedman  69 Lourdes García-García  70 Pauline Genter  71 Hertzel C Gerstein  20   21   22 Mark O Goodarzi  72 Penny Gordon-Larsen  14 Mariaelisa Graff  73 Myron Gross  74 Yu Guo  25 Xiuqing Guo  46 Yang Hai  46 Craig L Hanis  75 MGeoffrey Hayes  76   77   78 Momoko Horikoshi  79 Annie-Green Howard  80 Sarah Hsu  1   3 Willa Hsueh  81 Wei Huang  82 Mengna Huang  83   84 Yi-Jen Hung  85   86 Mi Yeong Hwang  87 Chii-Min Hwu  88   89 Sahoko Ichihara  90 Michiya Igase  91 Eli Ipp  71 Mohammad T Islam  92 Masato Isono  93 Hye-Mi Jang  87 Farzana Jasmine  16 Jost B Jonas  94   95   96   97   98 Yoonjung Y Joo  99   78 Edmond Kabagambe  100 Takashi Kadowaki  7 Yoichiro Kamatani  101   102 Fouad R Kandeel  103 Sharon L R Kardia  26 Elizabeth W Karlson  104 Anuradhani Kasturiratne  105 Norihiro Kato  93 Tomohiro Katsuya  106   107 Varinderpal Kaur  1   2   3 Takahisa Kawaguchi  108 Jacob M Keaton  32   100 Abel N Kho  109   110 Chiea-Chuen Khor  111   45   112 Muhammad Kibriya  16 Bong-Jo Kim  87 Woon-Puay Koh  113   39 Katsuhiko Kohara  114 Jaspal S Kooner  115 Charles Kooperberg  116 Raymond J Kreienkamp  1   4   2   117 Amel Lamri  20   21 Leslie A Lange  118 Nanette R Lee  119 Myung-Shik Lee  120   121 Jung-Jin Lee  122 Donna M Lehman  59 Liming Li  123 Yun Li  124 Victor Jy Lim  39 Jianjun Liu  111   125 Yongmei Liu  126   24 Simin Liu  127   128   129   130 Jirong Long  100 Tin Louie  62 Xi Luo  131 Jun Lv  123 Julie A Lynch  132   133 Shiro Maeda  79   134   135 Anubha Mahajan  136 Nisa M Maruthur  137 Fumihiko Matsuda  108 Mark I McCarthy  136   138   139 Roberta McKean-Cowdin  140 James B Meigs  1   4   141 Iona Y Millwood  13 Karen L Mohlke  61 Ayesha A Motala  142 Girish N Nadkarni  11   29 Jerry L Nadler  143 Masahiro Nakatochi  144 Mike A Nalls  145   146   147 Uma Nayak  148 Aude Nicolas  145 Kari E North  73 Darryl Nousome  140 Yukinori Okada  149   150   151 Ian Pan  84 James S Pankow  152 Guillaume Paré  153   21   22 Jaehyun Park  5 Kyong Soo Park  154   155 Esteban J Parra  156 Sanjay R Patel  157 Mark A Pereira  152 Patricia A Peyser  26 Fraser J Pirie  142 Michael Preuss  11 Michael A Province  19 Bruce M Psaty  158 Leslie J Raffel  159 Laura M Raffield  61 Laura J Rasmussen-Torvik  160 Susan Redline  161   37   38 Alexander P Reiner  162   116 Stephen S Rich  163 Rebecca Rohde  73 Kathryn Roll  46 Rashedeh Roshani  5 Charles N Rotimi  15 Charumathi Sabanayagam  45   50 Danish Saleheen  43 Kevin Sandow  46 Claudia Schurmann  29   30   11 Mohammad Shahriar  16 Douglas M Shaw  5 Wayne H-H Sheu  86   88   89   164   165 Jinxiu Shi  82 Xiao-Ou Shu  100 Megan M Shuey  5 Moneeza K Siddiqui  66 Jennifer A Smith  26   166 Tamar Sofer  167   168   4 Cassandra N Spracklen  61   169 Adrienne M Stilp  62 Meng Sun  170 Yasuharu Tabara  108 E-Shyong Tai  125   39   112 Salman M Tajuddin  65 Atsushi Takahashi  101   171 Fumihiko Takeuchi  93 Jingyi Tan  46 Kent D Taylor  46 Katherine Taylor  1 Farook Thameem  172 Lin Tong  16 Fuu-Jen Tsai  173 Philip S Tsao  174   175 Miriam S Udler  1   4   2   3 Adan Valladares-Salgado  56 David A van Heel  176 Rob M vanDam  39   125 Rohit Varma  177 Maheak Vora  1   2 Niels Wacher-Rodarte  178 Ya-Xing Wang  179 Ellie Wheeler  180   47 Eric A Whitsel  73   181 Ananda R Wickremasinghe  105 Genevieve L Wojcik  182 Tien Y Wong  45   50   51 Jer-Yuarn Wu  42 Yong-Bing Xiang  183   184 Anny H Xiang  185 Chittaranjan S Yajnik  186 Ken Yamamoto  187 Toshimasa Yamauchi  7 Lisa R Yanek  23 Jie Yao  46 Mitsuhiro Yokota  187 Canqing Yu  123 Jian-Min Yuan  188   189 Salim Yusuf  190   21   22 Eleftheria Zeggini  191   192 Liang Zhang  45 Weihua Zhang  193 Wei Zheng  194 Alan B Zonderman  65 ENSA Genomics ConsortiumGenes & Health Research TeamVA Million Veteran ProgramCarlos A Aguilar-Salinas  195   196 Clicerio González-Villalpando  197 Christopher A Haiman  198 Young Jin Kim  87 Soo Heon Kwak  154 Aaron Leong  1   2   199   3   141   200 Ruth J F Loos  11   201 Andres Moreno-Estrada  202 Andrew P Morris  8   203 Lorena Orozco  204 Jerome I Rotter  46 Dharambir Sanghera  12 Teresa Tusie-Luna  205 Benjamin F Voight  206   207   208   209 Marijana Vujkovic  210   206 Robin G Walters  13 Tian Ge  211 Alisa K Manning  212   1   4 Marie Loh  9   213   214 Jennifer E Below  5 Xueling Sim  39 Josep M Mercader  1   4   2   3   215 Maggie C Y Ng  5 D-PRISM Consortium
Affiliations

Multi-ancestry polygenic risk scores for the prediction of type 2 diabetes and complications in diverse ancestries

Alicia Huerta-Chagoya et al. medRxiv. .

Abstract

Background: Polygenic risk scores (PRSs) improve type 2 diabetes (T2D) prediction beyond clinical risk factors but perform poorly in non-European populations, where T2D burden is often higher, undermining their global clinical utility.

Methods: We conducted the largest global effort to date to harmonize T2D genome-wide association study (GWAS) meta-analyses across five ancestries-European (EUR), African/African American (AFR), Admixed American (AMR), South Asian (SAS), and East Asian (EAS)-including 360,000 T2D cases and 1·8 million controls (41% non-EUR). We constructed ancestry-specific and multi-ancestry PRSs in training datasets including 11,000 T2D cases and 32,000 controls, and validated their performance in independent datasets including 39,000 T2D cases and 126,000 controls of diverse ancestries. In the All of Us Research Program, we compared these PRSs to those from the Polygenic Score Catalog and assessed their ability to predict diabetes micro- and macrovascular complications.

Findings: Ancestry-specific PRSs showed limited prediction power for T2D in AFR, AMR, and SAS compared to EUR and EAS. In contrast, multi-ancestry PRSs, built using GWAS data from five ancestries, substantially improved T2D prediction across all ancestries. Compared to those in the interquartile range, individuals at the 97·5th percentile of their PRSs had a 6-fold increased T2D risk in AMR, EAS, and EUR, and ≥3-fold in AFR and SAS. These PRSs were also associated with the development of microvascular complications and outperformed all previously reported PRSs for all ancestries.

Interpretation: We developed and extensively validated the most up-to-date T2D PRSs across diverse ancestry groups. These PRSs are publicly available to support further evaluation of their clinical utility in diverse ancestries.

PubMed Disclaimer

Conflict of interest statement

H.C.G. holds the McMaster-Sanofi Population Health Institute Chair in Diabetes Research and Care. He reports research grants from Eli Lilly, Novo Nordisk, and Hanmi Pharmaceutical; grants to support continuing education programs from Eli Lilly, Abbott, Sanofi, Novo Nordisk, and Boehringer Ingelheim; honoraria for speaking from AstraZeneca, Eli Lilly, Zuellig, and Jiangsu Hanson; and consulting fees from Abbott, Bayer, Biolinq, Eli Lilly, Novo Nordisk, Pfizer, Shionogi, and Zealand. M.S.U. has consulting activity and research funded in collaboration with Novo Nordisk. A.K.M. has research funded in collaboration with Novo Nordisk. M.S.U. has research funded in collaboration with Novo Nordisk and is an unpaid research collaborator with AstraZeneca. J.M.M. has research funded in collaboration with Novo Nordisk. S.R.P. has had research funded by Philips Respironics and consulting fees from Apnimed, Bayer, Philips Respironics, Mineralys, and SleepRes. M.A.N. ‘s participation in this project was part of a competitive contract awarded to DataTecnica LLC by the National Institutes of Health to support open science research. He also currently owns stock in Character Bio and Neuron23 Inc.

Figures

Fig.1 |
Fig.1 |. Overall analysis approach.
a, Overview of the 125 T2D GWAS datasets, variant sets, and pairwise LD information used to generate five ancestry-specific T2D GWAS meta-analyses and 20 LD reference panels for PRSs training. b, Independent, ancestry-specific cohorts used to train the PRS models and select the optimal continuous shrinkage prior from five phi values (i.e., 0·01, 0·001, 1×10−4, 1×10−5, 1×10−6) based on predictive performance. Single-ancestry PRSs using GWAS summary statistics and LD panels matched to the validation ancestry or using data from the EAS or EUR ancestries. Multi-ancestry PRSs jointly modeled GWAS summary statistics and LD panels from all five ancestry groups. c, Set of 23 ancestry-specific and independent cohorts used to validate the 18 best-performing PRS. d, Evaluation of PRS predictive performance, including: i) incremental AUC, calculated as the difference between the AUC of the full model (PRS + covariates) and the model without the PRS, ii) proportion of variation in T2D status explained by the PRS, estimated using Nagelkerke’s r2, iii) odds ratio per standard deviation (OR per SD) of the PRS distribution or odds ratio (OR) comparing PRS distribution extremes relative to the interquartile range.
Fig. 2 |
Fig. 2 |. Performance of the T2D PRSs in the validation cohorts across ancestry groups.
ae: Incremental AUC (iAUC) of the T2D PRS in the validation cohorts across ancestry groups: a, AFR, b, AMR, c, EAS, d, EUR, e, SAS. For each ancestry, the best-performing single-ancestry and multi-ancestry PRSs were evaluated. Each bar represents a single cohort. Bar colors represent the ancestry group: purple for AFR, yellow for AMR, green for EAS, red for EUR, and blue for SAS. Line colors represent the ancestry of the T2D GWAS summary statistics and LD panels used to train the PRS, using the same color codes for single-ancestry PRSs, and black for multi-ancestry PRSs *De Long test p<0·05. fj: Odds ratio from the meta-analysis of validation cohorts across ancestry groups: f, AFR, g, AMR, h, EAS, i, EUR, j, SAS. Points represent the odds ratio per standard deviation (OR per SD) of the PRS distribution or the odds ratio (OR) comparing different PRS distribution extremes relative to the interquartile range. Error bars show the 95% confidence intervals (95% CI). Point colors represent the ancestry of the T2D GWAS summary statistics and LD panels used to train the PRS. *De Long p<0.05.
Fig. 3 |
Fig. 3 |. Performance of D-PRISM multi-ancestry PRSs compared to published T2D PRSs from the PGS Catalog and other sources in the All of Us cohort.
ae: Incremental AUC (iAUC) across ancestry groups: a, AFR, b, AMR, c, EAS, d, EUR, e, SAS. Black bars highlight this study’s multi-ancestry PRSs. *De Long p<0.05; **Bonferroni-corrected De Long p<9×10−4.
Fig. 4 |
Fig. 4 |. Association of D-PRISM multi-ancestry PRS with common complications of diabetes in the All of Us cohort.
Odds ratio of this study’s multi-ancestry PRS for the following outcomes: CAD (cardiovascular disease), IS (ischemic stroke), DN (diabetic nephropathy), ESDN (end-stage diabetic nephropathy), DR (diabetic retinopathy), and PDR (proliferative diabetic retinopathy). Points represent the odds ratios per standard deviation (OR per SD) and are colored according to the genetic ancestry of the individuals tested: purple for AFR, yellow for AMR, and red for EUR. Error bars show the 95% confidence intervals (95% CI).

References

    1. Magliano Dianna J., Boyko Edward J., IDF Diabetes Atlas 10th edition scientific committee. IDF DIABETES ATLAS. 2021.
    1. Mercader JM, Ng MCY, Manning AK, Rich SS. Predicting diabetes risk in diverse populations: what next? Lancet Diabetes Endocrinol 2021; 9: 808–10. - PMC - PubMed
    1. Suzuki K, Hatzikotoulas K, Southam L, et al. Genetic drivers of heterogeneity in type 2 diabetes pathophysiology. Nature 2024; 627: 347–57. - PMC - PubMed
    1. Udler MS, McCarthy MI, Florez JC, Mahajan A. Genetic Risk Scores for Diabetes Diagnosis and Precision Medicine. Endocr Rev 2019; 40: 1500–20. - PMC - PubMed
    1. Diabetes Prevention Program Research Group, Knowler WC, Fowler SE, et al. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 2009; 374: 1677–86. - PMC - PubMed

Publication types

LinkOut - more resources