Performance of Polygenic Risk Scores for Primary Open-Angle Glaucoma in Populations of African Descent

Jennifer M Chang-Wolf; Tyler G Kinzy; Sjoerd J Driessen; Lauren A Cruz; Sudha K Iyengar; Neal S Peachey; Tin Aung; Chiea Chuen Khor; Susan E Williams; Michele Ramsay; Olusola Olawoye; Adeyinka Ashaye; Caroline C W Klaver; Michael A Hauser; Alberta A H J Thiadens; Jessica N Cooke Bailey; Pieter W M Bonnemaijer

doi:10.1001/jamaophthalmol.2024.4784

. 2024 Nov 14;143(1):7–14. doi: 10.1001/jamaophthalmol.2024.4784

Performance of Polygenic Risk Scores for Primary Open-Angle Glaucoma in Populations of African Descent

Jennifer M Chang-Wolf ^1,^2,³, Tyler G Kinzy ^4,^5,^6,^7,⁸, Sjoerd J Driessen ^1,⁹, Lauren A Cruz ^4,⁵, Sudha K Iyengar ^4,^5,⁶, Neal S Peachey ^6,^10,¹¹, Tin Aung ¹², Chiea Chuen Khor ¹³, Susan E Williams ¹⁴, Michele Ramsay ¹⁵, Olusola Olawoye ¹⁶, Adeyinka Ashaye ¹⁶, Caroline C W Klaver ^1,^9,^17,¹⁸, Michael A Hauser ^2,^3,¹², Alberta A H J Thiadens ^1,⁹, Jessica N Cooke Bailey ^4,^5,^6,^7,⁸, Pieter W M Bonnemaijer ^1,^9,^19,^✉, for the for the for the Genetics in Glaucoma Patients of African Descent (GIGA) Study Group; Genetics of Glaucoma in People of African Descent (GGLAD) Study Group; Million Veteran Program (MVP)

¹Department of Epidemiology, Erasmus Medical Center, Rotterdam, the Netherlands

²Department of Ophthalmology, Duke University, Durham, North Carolina

³Department of Medicine, Duke University Medical Center, Durham, North Carolina

⁴Cleveland Institute for Computational Biology, Case Western Reserve University, Cleveland, Ohio

⁵Department of Population & Quantitative Health Sciences, Case Western Reserve University, Cleveland, Ohio

⁶Research Service, VA Northeast Ohio Healthcare System, Cleveland, Ohio

⁷Department of Pharmacology & Toxicology, Brody School of Medicine, East Carolina University, Greenville, North Carolina

⁸Center for Health Disparities, Brody School of Medicine, East Carolina University, Greenville, North Carolina

⁹Department of Ophthalmology, Erasmus Medical Center, Rotterdam, the Netherlands

¹⁰Cole Eye Institute, Cleveland Clinic Foundation, Cleveland, Ohio

¹¹Department of Ophthalmology, Cleveland Clinic Lerner College of Medicine of Case Western Reserve University, Cleveland, Ohio

¹²Singapore Eye Research Institute, Singapore

¹³Genome Institute of Singapore, Agency for Science, Technology and Research, Singapore

¹⁴Division of Ophthalmology, Department of Neurosciences, University of the Witwatersrand, Johannesburg, South Africa

¹⁵Sydney Brenner Institute for Molecular Bioscience, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa

¹⁶Department of Ophthalmology, College of Medicine, University of Ibadan, Ibadan, Nigeria

¹⁷Department of Ophthalmology, Radboud University Medical Centre, Nijmegen, Gelderland, the Netherlands

¹⁸Institute of Molecular and Clinical Ophthalmology, Basel, Switzerland

¹⁹The Rotterdam Eye Hospital, Rotterdam, the Netherlands

Group Information: The members of the Genetics in Glaucoma Patients of African Descent (GIGA) Study Group, Genetics of Glaucoma in People of African Descent (GGLAD) Study Group, Million Veteran Program (MVP), and Cooperative Studies Program (CSP) appear in Supplement 2.

Accepted for Publication: August 18, 2024.

Published Online: November 14, 2024. doi:10.1001/jamaophthalmol.2024.4784

^✉

Corresponding Author: Pieter Bonnemaijer, MD, PhD, Department of Epidemiology, Erasmus Medical Center, NA2808, Dr. Molewaterplein 40, 3015 GD Rotterdam, the Netherlands (p.bonnemaijer@erasmusmc.nl ).

Author Contributions: Drs Bonnemaijer and Cooke Bailey had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Dr Chang-Wolf and Mr Kinzy are considered co–first authors of this work. Drs Cooke Bailey and Bonnemaijer are considered co–senior authors of this work.

Concept and design: Chang-Wolf, Kinzy, Driessen, Aung, Olawoye, Klaver, Thiadens, Cooke Bailey, Bonnemaijer.

Acquisition, analysis, or interpretation of data: Chang-Wolf, Kinzy, Driessen, Cruz, Iyengar, Peachey, Aung, Khor, Williams, Ramsay, Ashaye, Klaver, Hauser, Thiadens, Cooke Bailey, Bonnemaijer.

Drafting of the manuscript: Chang-Wolf, Kinzy, Aung, Thiadens, Cooke Bailey, Bonnemaijer.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Chang-Wolf, Kinzy, Driessen, Khor, Ashaye, Thiadens, Bonnemaijer.

Obtained funding: Iyengar, Peachey, Olawoye, Klaver, Thiadens, Cooke Bailey, Bonnemaijer.

Administrative, technical, or material support: Chang-Wolf, Kinzy, Iyengar, Peachey, Aung, Olawoye, Hauser, Thiadens.

Supervision: Driessen, Klaver, Hauser, Thiadens, Cooke Bailey, Bonnemaijer.

Conflict of Interest Disclosures: Mr Kinzy reported receiving grants from National Institutes of Health and the Department of Veterans Affairs during the conduct of the study. Ms Cruz reported receiving grants from the National Institutes of Health during the conduct of the study. Dr Olawoye reported receiving grants from National Institute of Health H3 Africa during the conduct of the study. Dr Cooke Bailey reported receiving grants from National Institutes of Health and from Department of Veterans Affairs during the conduct of the study. Dr Bonnemaijer reported receiving grants from Brightfocus, Uitzicht, and Rottersamse Blindenbelangen during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was supported in part by the BrightFocus Foundation, the Combined Ophthalmic Research Rotterdam Foundation, the Rotterdamse Stichting Blindenbelangen, the Landelijke Stichting voor Blinden en Slechtzienden, the Stichting Beheer Het Schild, Algemene Nederlandse Vereniging ter Voorkoming van Blindheid, and het Glaucoomfonds through Uitzicht; grant R01EY033829 from the National Institutes of Health (Dr Cooke Bailey, Mr Kinzy, and Dr Hauser); and data from the Million Veteran Program (MVP) were supported by the MVP Office of Research and Development, Veterans Health Administration, and by award I01 BX004557 (Dr Cooke Bailey, Mr Kinzy, Dr Iyengar, and Dr Peachey). Ms Cruz was supported by National Institutes of Health grants T32 HL007567 and T32 EY007157.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: This publication does not represent the views of the Department of Veteran Affairs or the US government.

Data Sharing Statement: See Supplement 3.

^✉

Corresponding author.

PMCID: PMC11565374 PMID: 39541127

This cross-sectional study investigates how published polygenic risk scores for primary open-angle glaucoma (POAG) perform in groups with African ancestry compared with European ancestry.

Key Points

Question

How do published polygenic risk scores (PRSs) for primary open-angle glaucoma (POAG) perform in African-ancestry compared with European-ancestry groups?

Findings

In this cross-sectional study comparing 3 PRSs and including 11 673 cases and 66 432 controls across 7 ancestral groups, modest odds ratio increases were detected for the highest POAG risk quintile in Tanzanians, South Africans, Ghanaians, African Americans, and Europeans. However, overall predictive performance remains limited for African-ancestry compared with European-ancestry groups in terms of area under the curve and coefficient of determination.

Meaning

Results suggest that despite some improvements in POAG risk stratification among African-ancestry groups, this study underscores the need for further research to develop more accurate risk prediction models tailored to diverse populations for equitable health care outcomes.

Abstract

Importance

Primary open-angle glaucoma (POAG) polygenic risk scores (PRSs) continue to be evaluated in primarily European-ancestry populations despite higher prevalence and worse outcomes in African-ancestry populations.

Objective

To evaluate how established POAG PRSs perform in African-ancestry samples from the Genetics in Glaucoma Patients of African Descent (GIGA), Genetics of Glaucoma in Individuals of African Descent (GGLAD), and Million Veteran Program (MVP) datasets and compare these with European-ancestry samples.

Design, Setting, and Participants

This was a multicenter, cross-sectional study of POAG cases and controls from Tanzania, South Africa, Nigeria, Ghana, and the US. Included were individuals of African descent from South Africa and Tanzania from the GIGA dataset; individuals of African descent from Ghana, Nigeria, and the US from the GGLAD dataset; and individuals of African or European descent from the US in the MVP dataset. Data were analyzed from January 2022 to July 2023.

Exposures

Three PRSs derived from large meta-analyses of European and Asian populations, namely Gharahkhani et al (Gharahkhani PRS), Han et al (Han PRS), and Craig et al (Craig PRS).

Main Outcomes and Measures

Odds ratios (ORs) for POAG risk stratification comparing the highest and lowest quintiles; area under the receiver operating characteristic curve (AUROC), and liability coefficient of determination (R²) for the addition of PRS to a baseline of age, sex, and first 5 principal components.

Results

A total of 11 673 cases and 66 432 controls were included in this study across 7 ancestral groups. Mean (SD) age of the total participants was 76.9 (8.7) years, with 74 304 males (95.1%). The following were included in each dataset: GIGA (663 cases, 476 controls), GGLAD (1471 cases, 1482 controls), and MVP (9559 cases, 64 474 controls). Increases in ORs were found for the highest POAG risk quintile ranging from an OR of 1.68 (95% CI, 1.17-2.43) in Ghanaians to 7.05 (95% CI, 2.73-19.6) in the South African multiple ancestry group (which derives from at least 5 distinct ancestral groups: Khoisan, Bantus, Europeans, Indians, and Southeast Asians) with the Gharahkhani PRS. The Han PRS showed OR increases for the highest POAG risk quintile ranging from 2.27 (95% CI, 1.49-3.47) in African American individuals in the GGLAD dataset to 7.24 (95% CI, 6.47-8.12) in Europeans. The Craig PRS predicted OR increases in the highest quintile for all groups ranging from 1.51 (95% CI, 1.05-2.18) in Ghanaians to 6.31 (95% CI, 5.67-7.04) in Europeans. However, AUROC and R² increases above baseline were lower for all African-ancestry compared with European-ancestry groups in the 3 tested PRSs.

Conclusions and Relevance

In this cross-sectional study, despite some improvements in OR-based risk stratification using the Gharahkhani PRSs, Han PRSs, and Craig PRSs, consistently lower improvements in AUROC and R² for African-ancestry compared with European-ancestry groups highlight the need for risk prediction models tailored to diverse populations.

Introduction

Disproportionately affecting those of African ancestry, primary open-angle glaucoma (POAG) is a leading cause of irreversible blindness worldwide and is projected to become more prevalent in the coming decades.^1,2,3 POAG affected an estimated 76 million people globally in 2020 and is projected to affect nearly 120 million by 2040, making it a significant cause of irreversible blindness.^4,5,6,7 People of African descent face a higher risk, with up to 7 times the likelihood of developing POAG and experiencing blindness 4 times more frequently than those of European descent.^8,9 Risk stratification based on clinical and demographic data has long been used to aid patient care, which enables the identification of patients with high-risk POAG for more frequent follow-up or more aggressive clinical management.^10,11,12,13 Adding genetic data in the form of polygenic risk scores (PRSs) further improves risk stratification.¹⁴

Despite the disproportionate impact on individuals of African ancestry, genetic studies on POAG have focused on European and Asian populations, leaving a gap in understanding the genetic factors in African ancestral groups.^15,16,17,18 Recent genomic research advancements allow genetic data to enhance disease risk prediction. PRSs integrate variants from genome-wide association studies (GWAS) to predict disease risk. However, the construction of these PRSs currently relies heavily on data derived from European populations, raising concerns about their applicability and transferability to other populations with higher incidences of the target diseases.^{16,19,20,21,22}

PRS approaches have performed well in POAG, predicting advanced glaucoma risk and influencing treatment decisions.^23,24,25 Importantly, these PRSs were constructed from and applied to population samples that were primarily European and Asian and are thus potentially more reflective of the genetic architecture of those groups. In this study, we evaluated the performance of these tools for their ability to stratify POAG risk in individuals of African ancestry from South Africa, Tanzania, Ghana, Nigeria, and the US.

Methods

The Genetics in Glaucoma Patients of African Descent (GIGA) study, the Genetics of Glaucoma in People of African Descent (GGLAD) study, and the Million Veteran Program (MVP) adhered to the tenets of the Declaration of Helsinki, received ethical approval from the local institutional review boards, and obtained informed consent from all participants with no stipend or other incentive to participate. Only complete cases with age, sex, genotype, and phenotype data were included. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.²⁶

Study Population

GIGA is a multicenter POAG case-control study of individuals from the Groote Schuur Hospital (Cape Town, South Africa), Muhimbili National Hospital, and CCBRT Disability Hospital (Dar es Salaam, Tanzania).²⁷ The GIGA study included Tanzanians, Black South African people, and a group of individuals of multiple ancestries, which derives from at least 5 distinct ancestral groups: Khoisan, Bantus, Europeans, Indians, and Southeast Asians and for whom we use the term South African multiple ancestry.²⁸ Participants from the GGLAD discovery dataset included individuals of African descent recruited from Ghana (Lions International Eye Centre and Korle-Bu Teaching Hospital), Nigeria (University of Ibadan, University of Nigeria, and Nigerian Navy Reference Hospital), and North Carolina in the US (Duke University, Durham, North Carolina).^29,30,31 The MVP is a national research program sponsored by the Department of Veterans Affairs (VA) Office of Research and Development that includes health, genetic, and lifestyle information from 1 million US military Veterans who consented to participate in an observational cohort study and mega biobank. The genetic ancestry of over 450 000 MVP individuals with available genome-wide data was determined using the harmonized ancestry and race/ethnicity algorithm.^32,33

Genotyped data from the GIGA and GGLAD study groups were imputed to 1000 Genomes (1000G) phase 3, version 5.³⁴ MVP samples were imputed on either the 1000G phase 3, version 5, reference panel or the African Genome Resource (not publicly available), which comprises all 1000G samples with an additional 2000 samples from Uganda and 100 samples from each of 5 populations in Ethiopia, Egypt, Namibia (Nama/Khoesan), and South Africa (Zulu).^32,35

POAG Definition

GIGA POAG cases met International Society for Geographical and Epidemiological Ophthalmology classification categories 1 or 2, defined by a definite visual field defect and vertical cup disc ratio (VCDR) of greater than or equal to 0.7 or greater than 0.8 without visual field testing.³⁶ Cases had an open angle on gonioscopy and onset age older than 35 years. Controls underwent similar assessments and were characterized by no glaucoma signs or family history, intraocular pressure (IOP) less than or equal to 21 mm Hg, VCDR less than 0.5, VCDR intereye asymmetry less than 0.2, and age older than 55 years.^27,30,31,37 The GGLAD POAG definition included the presence of glaucomatous optic neuropathy characterized by loss of neuroretinal rim, VCDR greater than 0.7, intereye asymmetry greater than 0.2, and glaucoma-associated notching. Control samples were recruited from hospitals or the general population, with hospital-based controls being older than 40 years, free from (family history of) glaucoma or major eye diseases, IOP less than 21 mm Hg, open angles, healthy optic nerves, and normal visual fields. MVP cases and controls were defined by a computable phenotyping algorithm based on International Classification of Diseases–Clinical Modification codes, Current Procedural Terminology codes, and prescriptions in the VA computerized patient record system.^38,39

Statistical Analysis

We tested 3 PRSs using weighted additive models where the imputed dosage of each risk allele (j) was multiplied by its calculated β (log odds ratio [OR]) and summed using PLINK2 to produce a score for each individual (i)⁴⁰:

score_i = Σ^k_j_{= 1} β_jM_ij

Gharahkhani PRS (k = 111): Based on 127 POAG-associated variants identified in the largest to date cross-ancestry single-trait POAG GWAS meta-analysis.⁴¹ To reduce potential bias due to inclusion of overlapping African ancestral data from the GIGA and GGLAD studies, we applied a fixed-effects inverse variance–weighted meta-analysis to 239 168 controls and 23 612 POAG cases of European or Asian descent. Without the African datasets, 115 single-nucleotide variants (SNVs) maintained genome-wide significance. The X chromosome was excluded because it was unavailable in some cohorts, leaving 111 autosomal SNVs.
Given potential differences in linkage disequilibrium (LD) patterns, we developed a modified LD proxy PRS (k = 111) to incorporate SNV proxies while accounting for between-population LD variation. We analyzed SNV transferability via the local approach and evaluated a 500-kb window around each variant for SNVs with LD r² greater than 0.2 (as defined by the 1000 Genomes African subset).²⁷ We examined these for evidence of higher association in the African-only GWAS.⁴¹ Of these, 63 were nominally associated (P < .05), and the lead SNV was replaced with the local replacement SNV.
Han PRS (k = 223): Based on multitrait GWAS meta-analysis of POAG, VCDR, and IOP that identified 263 novel POAG loci, of which we included 223 SNVs from the European discovery set that replicated in an independent 23andMe cohort in which the P value remained <.001 after Bonferroni correction to avoid inflation due to overlap between Asian and African samples.²³
Craig PRS (k = 2673): Based on a POAG PRS derived from 67 040 UK Biobank samples by combining multiple endophenotypes (eg, VCDR and IOP) in multivariate analysis and validating the 2673 SNVs with P ≤ .001 in independent European datasets.²⁵

To investigate score value distributions across study groups, we plotted PRS densities standardized with a mean of 0 and an SD of 1 (eFigure 1 in Supplement 1). To account for potential nonlinear association and facilitate comparisons between different sample sizes, we constructed PRS quintiles. To adjust for confounding factors, we included age, sex, and the first 5 principal components (PCs) as covariates. For each quintile, we used a logistic regression model to estimate ORs of POAG risk and 95% CIs.

Aside from quintile stratification, we implemented logistic regression models of baseline alone (ie, age, sex, and 5 PCs) and with the addition of continuous PRS values. We examined the area under the receiver operating characteristic curve (AUROC) and used the DeLong test to compare area under the curve increases from the addition of PRS with the baseline models. We additionally evaluated the added value of continuous PRSs to baseline by determining increases in R² coefficients of determination. We computed R² on the liability scale to account for varying proportions of phenotypic variance explained by each PRS.^42,43 Case/control ratios in this study ranged from 0.09 in the European American group to 0.70 in the South African Black group. We fixed disease prevalence at 2.4% for comparison, although prevalence may vary across ancestral groups.⁴⁴ Data were analyzed from January 2022 to July 2023, using R statistical software, version 4.2.1 (R Foundation for Statistical Computing).

Results

Study Demographics for POAG Cases and Controls

A total of 11 673 cases and 66 432 controls were included in this study across 7 ancestral groups: Tanzanian (366 cases, 329 controls), South African multiple ancestry (179 cases, 96 controls), South African Black (118 cases, 51 controls), Nigerian (125 cases, 185 controls), Ghanaian (640 cases, 690 controls), African American (4627 cases, 6272 controls), and European American (5618 cases, 58 809 controls). The African American group comprised 2 subgroups: the GGLAD dataset (706 cases, 607 controls) and MVP dataset (3921 cases, 5665 controls). The following were included in each dataset: GIGA (663 cases, 476 controls), GGLAD (1471 cases, 1482 controls), and MVP (9559 cases, 64474 controls). Mean (SD) age of the total participants was 76.9 (8.7) years, with 3801 females (4.9%) and 74 304 males (95.1%). Study sample characteristics are presented in the Table. Sex proportions differed between cases and controls in the Tanzanian (female sex: 125 of 366 cases [34.2%], 172 of 329 controls [52.3%]; P = 2.1 × 10⁻⁶), South African Black (female sex: 62 of 118 cases [52.5%], 40 of 51 controls [78.4%]; P = 2.8 × 10⁻³), Nigerian (female sex: 48 of 125 cases [38.4%], 104 of 185 controls [56.2%]; P = 3.1 × 10⁻³), African American MVP (female sex: 216 of 3921 cases [5.5%], 171 of 5665 controls [3.0%]; P = 1.9 × 10⁻⁹), and European American (female sex: 211 of 5618 cases [8.7%], 1314 of 58 809 controls [2.2%]; P = 5.91 × 10⁻¹³) groups based on χ² testing. The mean (SD) age differed between cases and controls in the South African Black (62.9 [10.7] years; P = 2 × 10⁻⁴), Tanzanian (64.8 [9.8] years; P = 3.4 × 10⁻⁶), Ghanaian (59.6 [11.1] years; P = 4.5 × 10⁻²³), and African American GGLAD (61.3 [12.7] years; P = 8.1 × 10⁻⁹⁴) as well as African American MVP (74.2 [7.9] years; P = 2.2 × 10⁻¹⁶) groups based on Welch t test.

Table. Demographic Summary^a.

Variable	GIGA			GGLAD			MVP
Variable	Tanzanian	South African multiple ancestry^b	South African Black	Nigerian	Ghanaian	African American	African American	European American
No. (%)
Total	695	275	169	310	1330	1313	9586	64 427
Cases	366 (52.7)	179 (65.1)	118 (69.8)	125 (40.3)	640 (48.1)	706 (53.8)	3921 (40.9)	5618 (8.7)
Controls	329 (47.3)	96 (34.9)	51 (30.2)	185 (59.7)	690 (51.9)	607 (46.2)	5665 (59.1)	58 809 (91.3)
Female sex, No. (%)
Cases	125 (34.2)	93 (52.0)	62 (52.5)	48 (38.4)	279 (43.6)	336 (47.6)	216 (5.5)	211 (3.8)
Controls	172 (52.3)	54 (56.3)	40 (78.4)	104 (56.2)	297 (43.0)	279 (46.0)	171 (3.0)	1314 (2.2)
Male sex, No. (%)
Cases	241 (65.8)	86 (48.0)	56 (47.5)	77 (61.6)	361 (56.4)	370 (52.4)	3705 (94.5)	5407 (96.2)
Controls	157 (47.7)	42 (43.8)	11 (21.6)	81 (43.8)	393 (56.9)	328 (54.0)	5494 (97.0)	57 495 (97.8)
P value (χ² test)	2.1 × 10⁻⁶	.58	2.8 × 10⁻³	3.1 × 10⁻³	.88	.54	1.9 × 10⁻⁹	5.91 × 10⁻¹³
Age, mean (SD), y
Total	64.8 (9.8)	69.7 (10.0)	62.9 (10.7)	63.5 (11.5)	59.6 (11.1)	61.3 (12.7)	74.2 (7.9)	78.2 (7.6)
Cases	63.2 (11.0)	70.5 (11.0)	61.3 (11.5)	63.7 (11.5)	62.7 (12.4)	67.4 (12.3)	71.2 (9.5)	78.2 (7.6)
Controls	66.6 (8.1)	68.3 (7.7)	66.8 (7.2)	63.4 (11.5)	56.7 (9.0)	54.2 (9.0)	76.2 (6.7)	78.2 (7.4)
P value (Welch t test)	3.4 × 10⁻⁶	.06	2 × 10⁻⁴	.82	4.5 × 10⁻²³	8.1 × 10⁻⁹⁴	2.2 × 10⁻¹⁶	.47

Open in a new tab

Abbreviations: GGLAD, Genetics of Glaucoma in Individuals of African Descent; GIGA, Genetics in Glaucoma Patients of African Descent; MVP, Million Veteran Program.

^{^a}

The table presents information for the included population groups across different demographic variables.

^{^b}

The South African multiple ancestry group derives from at least 5 distinct ancestral groups: Khoisan, Bantus, Europeans, Indians, and Southeast Asians.

POAG Risk Stratification by PRS

We evaluated the performance of the 3 PRS in 8 study groups by comparing ORs categorized into quintiles, as illustrated in Figure 1 and detailed in eTable 1 in Supplement 1. The distribution of case proportions by quintile is visually depicted in eFigure 2 in Supplement 1. Figure 2 shows the increase in AUROC compared with the baseline model; detailed AUROC values are provided in eTable 2 in Supplement 1. Additionally, we examined R² values on the liability scale to further characterize the predictive accuracy of the evaluated scores (eTable 3 in Supplement 1). Here, we present the results of these measures separately for each score.

Figure 2. — The improvement of the AUROC of a model with the genetic risk score/PRS compared with a baseline model of sex, age, and 5 principal components is shown. The data are organized by population groups, including Tanzanian, South African multiple ancestry (which derives from at least 5 distinct ancestral groups: Khoisan, Bantus, Europeans, Indians, and Southeast Asians), South African Black, Nigerian, Ghanaian, African American (Genetics of Glaucoma in People of African Descent [GGLAD]), African American (Million Veteran Program [MVP]), and European. The heights of the bars reflect the AUROC increase values, and each bar is color-coded according to baseline vs values after addition of the Gharahkhani PRS (designated G), Han PRS (designated H), and Craig PRS (designated C) as indicated on the x-axis.

Gharahkhani PRS

Using the Gharahkhani PRS consisting of 111 SNVs, we observed an increase in the OR for POAG risk within the highest PRS quintile, compared with the lowest quintile across all groups except Nigerian (Figure 1). Relative to the European American group in this study (OR, 5.02; 95% CI, 4.54-5.57), the ORs within the highest quintile were increased but had wider CIs in the South African groups. Specifically, increases in ORs were found for the highest POAG risk quintile ranging from an OR of 1.68 (95% CI, 1.17-2.43) in Ghanaians to 7.05 (95% CI, 2.73-19.6) in the South African multiple ancestry group with the Gharahkhani PRS. The OR was 5.89 (95% CI, 1.72-24.30) for the highest quintile of the South African Black group compared with the lowest quintile and 7.05 (95% CI, 2.73-19.60) in the South African multiple ancestry group (Figure 1 and eTable 1 in Supplement 1). The performance of the GRS compared with a baseline model of age, sex, and the first 5 PCs was assessed by the AUROC. We observed a gain in the AUROC for the Tanzanian, South African multiple ancestry, African American (GGLAD and MVP), and European American groups (Figure 2 and eTable 2 in Supplement 1). However, the increase in AUROC was relatively modest in the Tanzanian (0.03 AUROC improvement from 0.66 baseline, P = .03) and African American (GGLAD: 0.02 AUROC improvement from 0.81 baseline, P < .001; MVP: 0.01 AUROC improvement from 0.69 baseline, P < .001) groups compared with the European American (0.06 AUROC improvement from 0.52 baseline, P < .001) group (eTable 2 in Supplement 1). We also evaluated the estimated proportion of POAG variance explained by the score. When comparing the genetic prediction performance of the Gharahkhani PRS from the African ancestral groups with the European American group, we observed lower phenotypic variance explained across all the African groups. Adding PRS to the baseline model, an improvement of the liability-R² value was only detected in the South African multiple ancestry group (0.03 R² increment over 0.01 baseline) and European American group (0.06 R² increment over 0.002 baseline) (eTable 3 in Supplement 1). We attempted to consider differences in LD using the LD proxy score, however, this approach yielded unremarkable results across all groups when compared with the original Gharahkhani PRS (eTable 4 in Supplement 1).

Han PRS

The Han PRS, consisting of 223 SNVs, showed increases in ORs for the highest genetic risk quintile compared with the lowest risk quintile in the Tanzanian, Ghanaian, GGLAD and MVP African American, and European American groups (Figure 1 and eTable 1 in Supplement 1). In the European American group, the Han PRS demonstrated the most substantial OR estimate for the highest quintile compared with the lowest among all tested scores (7.24; 95% CI, 6.47-8.12, P < .001) (eTable 1 in Supplement 1). In contrast, when comparing the highest with the lowest quintiles, the OR estimates were lower in the Han PRS than in the Gharahkhani PRS in the Tanzanian, South African, Ghanaian, and GGLAD African American groups. Consequently, the Han PRS only exhibited superior high-risk identification in the MVP African American and European American groups. The prediction performance of the Han PRS demonstrated an improvement in the AUROC compared with the baseline model in the Tanzanian (0.03 AUROC improvement from 0.66 baseline, P = .02), African American (GGLAD: 0.01 AUROC improvement from 0.82 baseline, P = .005; MVP: 0.02 AUROC improvement from 0.71 baseline, P < .001), and European American (0.16 AUROC improvement from 0.52 baseline, P < .001) groups (the H blue bars) (Figure 2 and eTable 2 in Supplement 1). Regarding R² by the Han PRS, increases for African ancestry groups were modest relative to the European American group. We only observed an improvement exceeding the contribution of the baseline alone in the European American group (0.08 R² increment over 0.002 baseline) (eTable 3 in Supplement 1). The Han PRS outperformed all other scores regarding AUROC and R² increases within the MVP African American and European American groups.

Craig PRS

Of the 3 scores, only the Craig PRS produced OR increases for the highest quintile in all tested groups ranging from 1.51 (95% CI, 1.05-2.18) in Ghanaians to 6.31 (95% CI, 5.67-7.04) in Europeans (Figure 1 and eTable 1 in Supplement 1). Prediction of POAG risk using the Craig PRS score increased the AUROC in the same groups as the Gharahkhani PRS (the G blue bars): Tanzanian (0.04 AUROC improvement from 0.66 baseline, P = .007), South African multiple ancestry (0.08 AUROC improvement from 0.59 baseline, P = .03), African American (GGLAD: 0.01 AUROC improvement from 0.81 baseline, P = .002; MVP: 0.01 AUROC improvement from 0.69 baseline, P < .001), and European American (0.16 AUROC improvement from 0.52 baseline, P < .001) (Figure 2 and eTable 2 in Supplement 1). Increases in R² were smaller for all African groups than for European Americans. The Craig PRS showed a slight improvement in R² for the South African multiple ancestry group compared with the baseline (0.03 R² increment over 0.01 baseline) (eTable 3 in Supplement 1). Conversely, the increase in R² within the European American group was more considerable (0.07 R² increment over 0.002 baseline) (eTable 3 in Supplement 1), mirroring the performance observed with the Han PRS.

Discussion

This study evaluates 3 PRSs for POAG, derived from a meta-analysis of European and Asian population GWAS data, for distinguishing between cases and controls in a European ancestry group and 7 African ancestry groups. All 3 PRSs performed best for classifying POAG cases in the European American group, emphasizing the importance of constructing genetic risk scores using reference data from the same ancestral population as the study group. The improvement in AUROC remained modest for most African ancestral groups whereas performance was best in those groups with the highest proportion of European ancestry, the African American and South African multiple ancestry groups (Figure 2). Performance of risk scores is limited to stratifying individuals into case and control groups when comparing the lowest to the highest quintiles of genetic risk as defined by each risk score. This distinction does little to help guide clinical management.^45,46

Strengths and Limitations

Our study has several strengths. We leverage ancestrally diverse data from 2 of the largest currently available datasets containing the genetic information of native African populations for glaucoma. However, even within African groups with low P values and relatively small CIs suggesting adequate power to detect differences, the OR gain was substantially higher in European groups. This suggests that the marked increase in ORs across quintiles of PRS in Europeans may not be solely attributable to higher statistical power but may also be influenced by ancestry-related factors. Others have reported this disparity in other diseases, and it has been suggested that this performance could be improved using models with more genetic variants.^15,21,47,48 Our data do not support this—the Gharahkhani PRS performed as well as or better than the 2 larger models. The most important contributing factor to poor overall performance is the lack of population-specific genetic risk data for glaucoma. It is well recognized that individuals of African ancestry are underrepresented in GWAS studies, resulting in a small number of associated genetic variants with potentially imprecise estimates of effect sizes.^49,50 At the same time, African continental and diaspora populations worldwide have very high levels of POAG.⁵¹ Extensive GWAS studies that include diverse African ancestry populations hold great promise for advancing our understanding of POAG risk in African individuals to enhance clinical management of this understudied and underserved patient population.⁵²

We also recognize several limitations in this study. Despite using 2 of the most extensive available datasets, our African datasets had low statistical power, especially the South African group. This potentially limits the generalizability of our findings and emphasizes the need to support the implementation of more diverse genetic studies that include African populations. Furthermore, the GIGA and GGLAD studies draw cases and controls from hospital settings, whereas the MVP draws cases and controls from a defined cohort of VA medical center users. This difference in sampling methodology may impact the genetic risk profiles and the demographic composition, potentially accounting for the disparities observed in the baseline AUROC statistics. Moreover, the POAG definition in the MVP populations differed from that in the GIGA and GGLAD studies.

Conclusions

In this cross-sectional study, results suggest that adding genetic variants to the risk model moderately improved PRS performance for Africans, whereas more substantial improvements were seen in risk identification for Europeans. This emphasizes that simply including more variants in the PRS does not address the model performance disparity across ancestral populations and may reinforce it. Genetically diverse discovery datasets are necessary for building population-specific PRSs.

Supplement 1.

eFigure 1. Density Distributions

eTable 1. Highest Quintile Odds Ratios

eFigure 2. Case Proportions by Quintile

eTable 2. Area Under the Receiver Operating Characteristic Curve

eTable 3. R² Values

eTable 4. Linkage Disequilibrium Proxy Polygenic Risk Score Results

jamaophthalmol-e244784-s001.pdf^{(643KB, pdf)}

Supplement 2.

Nonauthor Collaborators. The Genetics in Glaucoma Patients of African Descent (GIGA) Study Group, Genetics of Glaucoma in People of African Descent (GGLAD) Study Group, Million Veteran Program (MVP), and Cooperative Studies Program (CSP).

jamaophthalmol-e244784-s002.pdf^{(202KB, pdf)}

Supplement 3.

Data Sharing Statement.

jamaophthalmol-e244784-s003.pdf^{(17.4KB, pdf)}

References

1.Quigley HA, Broman AT. The number of people with glaucoma worldwide in 2010 and 2020. Br J Ophthalmol. 2006;90(3):262-267. doi: 10.1136/bjo.2005.081224 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Jonas JB, Aung T, Bourne RR, Bron AM, Ritch R, Panda-Jonas S. Glaucoma. Lancet. 2017;390(10108):2183-2193. doi: 10.1016/S0140-6736(17)31469-1 [DOI] [PubMed] [Google Scholar]
3.Varma R, Vajaranant TS, Burkemper B, et al. Visual impairment and blindness in adults in the US: demographic and geographic variations from 2015 to 2050. JAMA Ophthalmol. 2016;134(7):802-809. doi: 10.1001/jamaophthalmol.2016.1284 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Hattenhauer MG, Johnson DH, Ing HH, et al. The probability of blindness from open-angle glaucoma. Ophthalmology. 1998;105(11):2099-2104. doi: 10.1016/S0161-6420(98)91133-2 [DOI] [PubMed] [Google Scholar]
5.Kwon YH, Fingert JH, Kuehn MH, Alward WLM. Primary open-angle glaucoma. N Engl J Med. 2009;360(11):1113-1124. doi: 10.1056/NEJMra0804630 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Allison K, Patel D, Alabi O. Epidemiology of glaucoma: the past, present, and predictions for the future. Cureus. 2020;12(11):e11686. doi: 10.7759/cureus.11686 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Tham YC, Li X, Wong TY, Quigley HA, Aung T, Cheng CY. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology. 2014;121(11):2081-2090. doi: 10.1016/j.ophtha.2014.05.013 [DOI] [PubMed] [Google Scholar]
8.Bonnemaijer PWM, Lo Faro V, Sanyiwa AJ, et al. ; GIGA study group . Differences in clinical presentation of primary open-angle glaucoma between African and European populations. Acta Ophthalmol. 2021;99(7):e1118-e1126. doi: 10.1111/aos.14772 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Kyari F, Abdull MM, Bastawrous A, Gilbert CE, Faal H. Epidemiology of glaucoma in Sub-Saharan Africa: prevalence, incidence and risk factors. Middle East Afr J Ophthalmol. 2013;20(2):111-125. doi: 10.4103/0974-9233.110605 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Weinreb RN, Aung T, Medeiros FA. The pathophysiology and treatment of glaucoma: a review. JAMA. 2014;311(18):1901-1911. doi: 10.1001/jama.2014.3192 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Gordon MO, Beiser JA, Brandt JD, et al. The Ocular Hypertension Treatment Study: baseline factors that predict the onset of primary open-angle glaucoma. Arch Ophthalmol. 2002;120(6):714-720. doi: 10.1001/archopht.120.6.714 [DOI] [PubMed] [Google Scholar]
12.Chauhan BC, Mikelberg FS, Balaszi AG, LeBlanc RP, Lesk MR, Trope GE; Canadian Glaucoma Study Group . Canadian Glaucoma Study: 2. risk factors for the progression of open-angle glaucoma. Arch Ophthalmol. 2008;126(8):1030-1036. doi: 10.1001/archopht.126.8.1030 [DOI] [PubMed] [Google Scholar]
13.Leske MC, Heijl A, Hyman L, Bengtsson B. Early Manifest Glaucoma Trial: design and baseline data. Ophthalmology. 1999;106(11):2144-2153. doi: 10.1016/S0161-6420(99)90497-9 [DOI] [PubMed] [Google Scholar]
14.Wiggs JL, Pasquale LR. Genetics of glaucoma. Hum Mol Genet. 2017;26(R1):R21-R27. doi: 10.1093/hmg/ddx184 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Cooke Bailey JN, Funk KL, Cruz LA, et al. Diversity in polygenic risk of primary open-angle glaucoma. Genes (Basel). 2022;14(1):111. doi: 10.3390/genes14010111 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Duncan L, Shen H, Gelaye B, et al. Analysis of polygenic risk score usage and performance in diverse human populations. Nat Commun. 2019;10(1):3328. doi: 10.1038/s41467-019-11112-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Fatumo S, Sathan D, Samtal C, et al. Polygenic risk scores for disease risk prediction in Africa: current challenges and future directions. Genome Med. 2023;15(1):87. doi: 10.1186/s13073-023-01245-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Fritsche LG, Ma Y, Zhang D, et al. On cross-ancestry cancer polygenic risk scores. PLoS Genet. 2021;17(9):e1009670. doi: 10.1371/journal.pgen.1009670 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Martin AR, Kanai M, Kamatani Y, Okada Y, Neale BM, Daly MJ. Clinical use of current polygenic risk scores may exacerbate health disparities. Nat Genet. 2019;51(4):584-591. doi: 10.1038/s41588-019-0379-x [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Majara L, Kalungi A, Koen N, et al. Low and differential polygenic score generalizability among African populations due largely to genetic diversity. HGG Adv. 2023;4(2):100184. doi: 10.1016/j.xhgg.2023.100184 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Kamiza AB, Toure SM, Vujkovic M, et al. Transferability of genetic risk scores in African populations. Nat Med. 2022;28(6):1163-1166. doi: 10.1038/s41591-022-01835-x [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Liu C, Zeinomar N, Chung WK, et al. Generalizability of polygenic risk scores for breast cancer among women with European, African, and Latinx Ancestry. JAMA Netw Open. 2021;4(8):e2119084. doi: 10.1001/jamanetworkopen.2021.19084 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Han X, Gharahkhani P, Hamel AR, et al. ; 23andMe Research Team; International Glaucoma Genetics Consortium . Large-scale multitrait genome-wide association analyses identify hundreds of glaucoma risk loci. Nat Genet. 2023;55(7):1116-1125. doi: 10.1038/s41588-023-01428-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Marshall HN, Mullany S, Han X, et al. High polygenic risk is associated with earlier initiation and escalation of treatment in early primary open-angle glaucoma. Ophthalmology. 2023;130(8):830-836. doi: 10.1016/j.ophtha.2023.03.028 [DOI] [PubMed] [Google Scholar]
25.Craig JE, Han X, Qassim A, et al. ; NEIGHBORHOOD consortium; UK Biobank Eye and Vision Consortium . Multitrait analysis of glaucoma identifies new risk loci and enables polygenic prediction of disease susceptibility and progression. Nat Genet. 2020;52(2):160-166. doi: 10.1038/s41588-019-0556-y [DOI] [PMC free article] [PubMed] [Google Scholar]
26.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344-349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
27.Bonnemaijer PWM, Iglesias AI, Nadkarni GN, et al. ; GIGA Study Group; Eyes of Africa Genetics Consortium; NEIGHBORHOOD Consortium . Genome-wide association study of primary open-angle glaucoma in continental and admixed African populations. Hum Genet. 2018;137(10):847-862. doi: 10.1007/s00439-018-1943-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Patterson N, Petersen DC, van der Ross RE, et al. Genetic structure of a unique admixed population: implications for medical research. Hum Mol Genet. 2010;19(3):411-419. doi: 10.1093/hmg/ddp505 [DOI] [PubMed] [Google Scholar]
29.Li Z, Allingham RR, Nakano M, et al. ; ICAARE-Glaucoma Consortium; NEIGHBORHOOD Consortium . A common variant near TGFBR3 is associated with primary open angle glaucoma. Hum Mol Genet. 2015;24(13):3880-3892. doi: 10.1093/hmg/ddv128 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Bailey JN, Loomis SJ, Kang JH, et al. ; ANZRAG Consortium . Genome-wide association analysis identifies TXNRD2, ATXN2, and FOXC1 as susceptibility loci for primary open-angle glaucoma. Nat Genet. 2016;48(2):189-194. doi: 10.1038/ng.3482 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Gharahkhani P, Burdon KP, Fogarty R, et al. ; Wellcome Trust Case Control Consortium 2, NEIGHBORHOOD consortium . Common variants near ABCA1, AFAP1, and GMDS confer risk of primary open-angle glaucoma. Nat Genet. 2014;46(10):1120-1125. doi: 10.1038/ng.3079 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Hunter-Zinck H, Shi Y, Li M, et al. ; VA Million Veteran Program . Genotyping array design and data quality control in the Million Veteran Program. Am J Hum Genet. 2020;106(4):535-548. doi: 10.1016/j.ajhg.2020.03.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Fang H, Hui Q, Lynch J, et al. ; VA Million Veteran Program . Harmonizing genetic ancestry and self-identified race/ethnicity in genome-wide association studies. Am J Hum Genet. 2019;105(4):763-772. doi: 10.1016/j.ajhg.2019.08.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Auton A, Brooks LD, Durbin RM, et al. ; The 1000 Genomes Project Consortium . A global reference for human genetic variation. Nature. 2015;526(7571):68-74. doi: 10.1038/nature15393 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.O’Connell J, Yun T, Moreno M, et al. ; 23andMe Research Team . A population-specific reference panel for improved genotype imputation in African Americans. Commun Biol. 2021;4(1):1269. doi: 10.1038/s42003-021-02777-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Foster PJ, Buhrmann R, Quigley HA, Johnson GJ. The definition and classification of glaucoma in prevalence surveys. Br J Ophthalmol. 2002;86(2):238-242. doi: 10.1136/bjo.86.2.238 [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Hauser MA, Allingham RR, Aung T, et al. ; Genetics of Glaucoma in People of African Descent (GGLAD) Consortium . Association of genetic variants with primary open-angle glaucoma among individuals with African Ancestry. JAMA. 2019;322(17):1682-1691. doi: 10.1001/jama.2019.16161 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Fihn SD, Francis J, Clancy C, et al. Insights from advanced analytics at the Veterans Health Administration. Health Aff (Millwood). 2014;33(7):1203-1211. doi: 10.1377/hlthaff.2014.0054 [DOI] [PubMed] [Google Scholar]
39.Nealon CL, Halladay CW, Kinzy TG, et al. ; VA Million Veteran Program . Development and evaluation of a rules-based algorithm for primary open-angle glaucoma in the VA Million Veteran Program. Ophthalmic Epidemiol. 2022;29(6):640-648. doi: 10.1080/09286586.2021.1992784 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience. 2015;4(1):7. doi: 10.1186/s13742-015-0047-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Gharahkhani P, Jorgenson E, Hysi P, et al. ; NEIGHBORHOOD consortium; ANZRAG consortium; Biobank Japan project; FinnGen study; UK Biobank Eye and Vision Consortium; GIGA study group; 23 and Me Research Team . Genome-wide meta-analysis identifies 127 open-angle glaucoma loci with consistent effect across ancestries. Nat Commun. 2021;12(1):1258. doi: 10.1038/s41467-020-20851-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Nagelkerke NJD. A note on a general definition of the coefficient of determination. Biometrika. 1991;78(3):691-692. doi: 10.1093/biomet/78.3.691 [DOI] [Google Scholar]
43.Lee SH, Goddard ME, Wray NR, Visscher PM. A better coefficient of determination for genetic profile analysis. Genet Epidemiol. 2012;36(3):214-224. doi: 10.1002/gepi.21614 [DOI] [PubMed] [Google Scholar]
44.Zhang N, Wang J, Li Y, Jiang B. Prevalence of primary open angle glaucoma in the last 20 years: a meta-analysis and systematic review. Sci Rep. 2021;11(1):13762. doi: 10.1038/s41598-021-92971-w [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Hingorani AD, Gratton J, Finan C, et al. Performance of polygenic risk scores in screening, prediction, and risk stratification: secondary analysis of data in the Polygenic Score Catalog. BMJ Med. 2023;2(1):e000554. doi: 10.1136/bmjmed-2023-000554 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Sud A, Horton RH, Hingorani AD, et al. Realistic expectations are key to realising the benefits of polygenic scores. BMJ. 2023;380:e073149. doi: 10.1136/bmj-2022-073149 [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Hoffmann TJ, Tang H, Thornton TA, et al. Genome-wide association and admixture analysis of glaucoma in the Women’s Health Initiative. Hum Mol Genet. 2014;23(24):6634-6643. doi: 10.1093/hmg/ddu364 [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Kachuri L, Chatterjee N, Hirbo J, et al. ; Polygenic Risk Methods in Diverse Populations (PRIMED) Consortium Methods Working Group . Principles and methods for transferring polygenic risk scores across global populations. Nat Rev Genet. 2024;25(1):8-25. doi: 10.1038/s41576-023-00637-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Rotimi CN, Bentley AR, Doumatey AP, Chen G, Shriner D, Adeyemo A. The genomic landscape of African populations in health and disease. Hum Mol Genet. 2017;26(R2):R225-R236. doi: 10.1093/hmg/ddx253 [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Sirugo G, Williams SM, Tishkoff SA. The missing diversity in human genetic studies. Cell. 2019;177(1):26-31. doi: 10.1016/j.cell.2019.02.048 [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Budenz DL, Barton K, Whiteside-de Vos J, et al. ; Tema Eye Survey Study Group . Prevalence of glaucoma in an urban West African population: the Tema Eye Survey. JAMA Ophthalmol. 2013;131(5):651-658. doi: 10.1001/jamaophthalmol.2013.1686 [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Verma SS, Gudiseva HV, Chavali VRM, et al. ; Regeneron Genetics Center; VA Million Veteran Program . A multicohort genome-wide association study in African ancestry individuals reveals risk loci for primary open-angle glaucoma. Cell. 2024;187(2):464-480.e10. doi: 10.1016/j.cell.2023.12.006 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eFigure 1. Density Distributions

eTable 1. Highest Quintile Odds Ratios

eFigure 2. Case Proportions by Quintile

eTable 2. Area Under the Receiver Operating Characteristic Curve

eTable 3. R² Values

eTable 4. Linkage Disequilibrium Proxy Polygenic Risk Score Results

jamaophthalmol-e244784-s001.pdf^{(643KB, pdf)}

Supplement 2.

jamaophthalmol-e244784-s002.pdf^{(202KB, pdf)}

Supplement 3.

Data Sharing Statement.

jamaophthalmol-e244784-s003.pdf^{(17.4KB, pdf)}

[eoi240072r1] 1.Quigley HA, Broman AT. The number of people with glaucoma worldwide in 2010 and 2020. Br J Ophthalmol. 2006;90(3):262-267. doi: 10.1136/bjo.2005.081224 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r2] 2.Jonas JB, Aung T, Bourne RR, Bron AM, Ritch R, Panda-Jonas S. Glaucoma. Lancet. 2017;390(10108):2183-2193. doi: 10.1016/S0140-6736(17)31469-1 [DOI] [PubMed] [Google Scholar]

[eoi240072r3] 3.Varma R, Vajaranant TS, Burkemper B, et al. Visual impairment and blindness in adults in the US: demographic and geographic variations from 2015 to 2050. JAMA Ophthalmol. 2016;134(7):802-809. doi: 10.1001/jamaophthalmol.2016.1284 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r4] 4.Hattenhauer MG, Johnson DH, Ing HH, et al. The probability of blindness from open-angle glaucoma. Ophthalmology. 1998;105(11):2099-2104. doi: 10.1016/S0161-6420(98)91133-2 [DOI] [PubMed] [Google Scholar]

[eoi240072r5] 5.Kwon YH, Fingert JH, Kuehn MH, Alward WLM. Primary open-angle glaucoma. N Engl J Med. 2009;360(11):1113-1124. doi: 10.1056/NEJMra0804630 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r6] 6.Allison K, Patel D, Alabi O. Epidemiology of glaucoma: the past, present, and predictions for the future. Cureus. 2020;12(11):e11686. doi: 10.7759/cureus.11686 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r7] 7.Tham YC, Li X, Wong TY, Quigley HA, Aung T, Cheng CY. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology. 2014;121(11):2081-2090. doi: 10.1016/j.ophtha.2014.05.013 [DOI] [PubMed] [Google Scholar]

[eoi240072r8] 8.Bonnemaijer PWM, Lo Faro V, Sanyiwa AJ, et al. ; GIGA study group . Differences in clinical presentation of primary open-angle glaucoma between African and European populations. Acta Ophthalmol. 2021;99(7):e1118-e1126. doi: 10.1111/aos.14772 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r9] 9.Kyari F, Abdull MM, Bastawrous A, Gilbert CE, Faal H. Epidemiology of glaucoma in Sub-Saharan Africa: prevalence, incidence and risk factors. Middle East Afr J Ophthalmol. 2013;20(2):111-125. doi: 10.4103/0974-9233.110605 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r10] 10.Weinreb RN, Aung T, Medeiros FA. The pathophysiology and treatment of glaucoma: a review. JAMA. 2014;311(18):1901-1911. doi: 10.1001/jama.2014.3192 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r11] 11.Gordon MO, Beiser JA, Brandt JD, et al. The Ocular Hypertension Treatment Study: baseline factors that predict the onset of primary open-angle glaucoma. Arch Ophthalmol. 2002;120(6):714-720. doi: 10.1001/archopht.120.6.714 [DOI] [PubMed] [Google Scholar]

[eoi240072r12] 12.Chauhan BC, Mikelberg FS, Balaszi AG, LeBlanc RP, Lesk MR, Trope GE; Canadian Glaucoma Study Group . Canadian Glaucoma Study: 2. risk factors for the progression of open-angle glaucoma. Arch Ophthalmol. 2008;126(8):1030-1036. doi: 10.1001/archopht.126.8.1030 [DOI] [PubMed] [Google Scholar]

[eoi240072r13] 13.Leske MC, Heijl A, Hyman L, Bengtsson B. Early Manifest Glaucoma Trial: design and baseline data. Ophthalmology. 1999;106(11):2144-2153. doi: 10.1016/S0161-6420(99)90497-9 [DOI] [PubMed] [Google Scholar]

[eoi240072r14] 14.Wiggs JL, Pasquale LR. Genetics of glaucoma. Hum Mol Genet. 2017;26(R1):R21-R27. doi: 10.1093/hmg/ddx184 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r15] 15.Cooke Bailey JN, Funk KL, Cruz LA, et al. Diversity in polygenic risk of primary open-angle glaucoma. Genes (Basel). 2022;14(1):111. doi: 10.3390/genes14010111 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r16] 16.Duncan L, Shen H, Gelaye B, et al. Analysis of polygenic risk score usage and performance in diverse human populations. Nat Commun. 2019;10(1):3328. doi: 10.1038/s41467-019-11112-0 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r17] 17.Fatumo S, Sathan D, Samtal C, et al. Polygenic risk scores for disease risk prediction in Africa: current challenges and future directions. Genome Med. 2023;15(1):87. doi: 10.1186/s13073-023-01245-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r18] 18.Fritsche LG, Ma Y, Zhang D, et al. On cross-ancestry cancer polygenic risk scores. PLoS Genet. 2021;17(9):e1009670. doi: 10.1371/journal.pgen.1009670 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r19] 19.Martin AR, Kanai M, Kamatani Y, Okada Y, Neale BM, Daly MJ. Clinical use of current polygenic risk scores may exacerbate health disparities. Nat Genet. 2019;51(4):584-591. doi: 10.1038/s41588-019-0379-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r20] 20.Majara L, Kalungi A, Koen N, et al. Low and differential polygenic score generalizability among African populations due largely to genetic diversity. HGG Adv. 2023;4(2):100184. doi: 10.1016/j.xhgg.2023.100184 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r21] 21.Kamiza AB, Toure SM, Vujkovic M, et al. Transferability of genetic risk scores in African populations. Nat Med. 2022;28(6):1163-1166. doi: 10.1038/s41591-022-01835-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r22] 22.Liu C, Zeinomar N, Chung WK, et al. Generalizability of polygenic risk scores for breast cancer among women with European, African, and Latinx Ancestry. JAMA Netw Open. 2021;4(8):e2119084. doi: 10.1001/jamanetworkopen.2021.19084 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r23] 23.Han X, Gharahkhani P, Hamel AR, et al. ; 23andMe Research Team; International Glaucoma Genetics Consortium . Large-scale multitrait genome-wide association analyses identify hundreds of glaucoma risk loci. Nat Genet. 2023;55(7):1116-1125. doi: 10.1038/s41588-023-01428-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r24] 24.Marshall HN, Mullany S, Han X, et al. High polygenic risk is associated with earlier initiation and escalation of treatment in early primary open-angle glaucoma. Ophthalmology. 2023;130(8):830-836. doi: 10.1016/j.ophtha.2023.03.028 [DOI] [PubMed] [Google Scholar]

[eoi240072r25] 25.Craig JE, Han X, Qassim A, et al. ; NEIGHBORHOOD consortium; UK Biobank Eye and Vision Consortium . Multitrait analysis of glaucoma identifies new risk loci and enables polygenic prediction of disease susceptibility and progression. Nat Genet. 2020;52(2):160-166. doi: 10.1038/s41588-019-0556-y [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r26] 26.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344-349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]

[eoi240072r27] 27.Bonnemaijer PWM, Iglesias AI, Nadkarni GN, et al. ; GIGA Study Group; Eyes of Africa Genetics Consortium; NEIGHBORHOOD Consortium . Genome-wide association study of primary open-angle glaucoma in continental and admixed African populations. Hum Genet. 2018;137(10):847-862. doi: 10.1007/s00439-018-1943-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r28] 28.Patterson N, Petersen DC, van der Ross RE, et al. Genetic structure of a unique admixed population: implications for medical research. Hum Mol Genet. 2010;19(3):411-419. doi: 10.1093/hmg/ddp505 [DOI] [PubMed] [Google Scholar]

[eoi240072r29] 29.Li Z, Allingham RR, Nakano M, et al. ; ICAARE-Glaucoma Consortium; NEIGHBORHOOD Consortium . A common variant near TGFBR3 is associated with primary open angle glaucoma. Hum Mol Genet. 2015;24(13):3880-3892. doi: 10.1093/hmg/ddv128 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r30] 30.Bailey JN, Loomis SJ, Kang JH, et al. ; ANZRAG Consortium . Genome-wide association analysis identifies TXNRD2, ATXN2, and FOXC1 as susceptibility loci for primary open-angle glaucoma. Nat Genet. 2016;48(2):189-194. doi: 10.1038/ng.3482 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r31] 31.Gharahkhani P, Burdon KP, Fogarty R, et al. ; Wellcome Trust Case Control Consortium 2, NEIGHBORHOOD consortium . Common variants near ABCA1, AFAP1, and GMDS confer risk of primary open-angle glaucoma. Nat Genet. 2014;46(10):1120-1125. doi: 10.1038/ng.3079 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r32] 32.Hunter-Zinck H, Shi Y, Li M, et al. ; VA Million Veteran Program . Genotyping array design and data quality control in the Million Veteran Program. Am J Hum Genet. 2020;106(4):535-548. doi: 10.1016/j.ajhg.2020.03.004 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r33] 33.Fang H, Hui Q, Lynch J, et al. ; VA Million Veteran Program . Harmonizing genetic ancestry and self-identified race/ethnicity in genome-wide association studies. Am J Hum Genet. 2019;105(4):763-772. doi: 10.1016/j.ajhg.2019.08.012 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r34] 34.Auton A, Brooks LD, Durbin RM, et al. ; The 1000 Genomes Project Consortium . A global reference for human genetic variation. Nature. 2015;526(7571):68-74. doi: 10.1038/nature15393 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r35] 35.O’Connell J, Yun T, Moreno M, et al. ; 23andMe Research Team . A population-specific reference panel for improved genotype imputation in African Americans. Commun Biol. 2021;4(1):1269. doi: 10.1038/s42003-021-02777-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r36] 36.Foster PJ, Buhrmann R, Quigley HA, Johnson GJ. The definition and classification of glaucoma in prevalence surveys. Br J Ophthalmol. 2002;86(2):238-242. doi: 10.1136/bjo.86.2.238 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r37] 37.Hauser MA, Allingham RR, Aung T, et al. ; Genetics of Glaucoma in People of African Descent (GGLAD) Consortium . Association of genetic variants with primary open-angle glaucoma among individuals with African Ancestry. JAMA. 2019;322(17):1682-1691. doi: 10.1001/jama.2019.16161 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r38] 38.Fihn SD, Francis J, Clancy C, et al. Insights from advanced analytics at the Veterans Health Administration. Health Aff (Millwood). 2014;33(7):1203-1211. doi: 10.1377/hlthaff.2014.0054 [DOI] [PubMed] [Google Scholar]

[eoi240072r39] 39.Nealon CL, Halladay CW, Kinzy TG, et al. ; VA Million Veteran Program . Development and evaluation of a rules-based algorithm for primary open-angle glaucoma in the VA Million Veteran Program. Ophthalmic Epidemiol. 2022;29(6):640-648. doi: 10.1080/09286586.2021.1992784 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r40] 40.Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience. 2015;4(1):7. doi: 10.1186/s13742-015-0047-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r41] 41.Gharahkhani P, Jorgenson E, Hysi P, et al. ; NEIGHBORHOOD consortium; ANZRAG consortium; Biobank Japan project; FinnGen study; UK Biobank Eye and Vision Consortium; GIGA study group; 23 and Me Research Team . Genome-wide meta-analysis identifies 127 open-angle glaucoma loci with consistent effect across ancestries. Nat Commun. 2021;12(1):1258. doi: 10.1038/s41467-020-20851-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r42] 42.Nagelkerke NJD. A note on a general definition of the coefficient of determination. Biometrika. 1991;78(3):691-692. doi: 10.1093/biomet/78.3.691 [DOI] [Google Scholar]

[eoi240072r43] 43.Lee SH, Goddard ME, Wray NR, Visscher PM. A better coefficient of determination for genetic profile analysis. Genet Epidemiol. 2012;36(3):214-224. doi: 10.1002/gepi.21614 [DOI] [PubMed] [Google Scholar]

[eoi240072r44] 44.Zhang N, Wang J, Li Y, Jiang B. Prevalence of primary open angle glaucoma in the last 20 years: a meta-analysis and systematic review. Sci Rep. 2021;11(1):13762. doi: 10.1038/s41598-021-92971-w [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r45] 45.Hingorani AD, Gratton J, Finan C, et al. Performance of polygenic risk scores in screening, prediction, and risk stratification: secondary analysis of data in the Polygenic Score Catalog. BMJ Med. 2023;2(1):e000554. doi: 10.1136/bmjmed-2023-000554 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r46] 46.Sud A, Horton RH, Hingorani AD, et al. Realistic expectations are key to realising the benefits of polygenic scores. BMJ. 2023;380:e073149. doi: 10.1136/bmj-2022-073149 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r47] 47.Hoffmann TJ, Tang H, Thornton TA, et al. Genome-wide association and admixture analysis of glaucoma in the Women’s Health Initiative. Hum Mol Genet. 2014;23(24):6634-6643. doi: 10.1093/hmg/ddu364 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r48] 48.Kachuri L, Chatterjee N, Hirbo J, et al. ; Polygenic Risk Methods in Diverse Populations (PRIMED) Consortium Methods Working Group . Principles and methods for transferring polygenic risk scores across global populations. Nat Rev Genet. 2024;25(1):8-25. doi: 10.1038/s41576-023-00637-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r49] 49.Rotimi CN, Bentley AR, Doumatey AP, Chen G, Shriner D, Adeyemo A. The genomic landscape of African populations in health and disease. Hum Mol Genet. 2017;26(R2):R225-R236. doi: 10.1093/hmg/ddx253 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r50] 50.Sirugo G, Williams SM, Tishkoff SA. The missing diversity in human genetic studies. Cell. 2019;177(1):26-31. doi: 10.1016/j.cell.2019.02.048 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r51] 51.Budenz DL, Barton K, Whiteside-de Vos J, et al. ; Tema Eye Survey Study Group . Prevalence of glaucoma in an urban West African population: the Tema Eye Survey. JAMA Ophthalmol. 2013;131(5):651-658. doi: 10.1001/jamaophthalmol.2013.1686 [DOI] [PMC free article] [PubMed] [Google Scholar]

[eoi240072r52] 52.Verma SS, Gudiseva HV, Chavali VRM, et al. ; Regeneron Genetics Center; VA Million Veteran Program . A multicohort genome-wide association study in African ancestry individuals reveals risk loci for primary open-angle glaucoma. Cell. 2024;187(2):464-480.e10. doi: 10.1016/j.cell.2023.12.006 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Performance of Polygenic Risk Scores for Primary Open-Angle Glaucoma in Populations of African Descent

Jennifer M Chang-Wolf, MD

Tyler G Kinzy, MS

Sjoerd J Driessen, MD

Lauren A Cruz, MPH

Sudha K Iyengar, PhD

Neal S Peachey, PhD

Tin Aung, MBBS, MMed, PhD

Chiea Chuen Khor, MD, PhD

Susan E Williams, MD, PhD

Michele Ramsay, BSc, MSc, PhD

Olusola Olawoye, MBBS, MSc, PhD

Adeyinka Ashaye, MD

Caroline C W Klaver, MD, PhD

Michael A Hauser, PhD

Alberta A H J Thiadens, MD, PhD

Jessica N Cooke Bailey, PhD, MA

Pieter W M Bonnemaijer, MD, PhD

Key Points

Question

Findings

Meaning

Abstract

Importance

Objective

Design, Setting, and Participants

Exposures

Main Outcomes and Measures

Results

Conclusions and Relevance

Introduction

Methods

Study Population

POAG Definition

Statistical Analysis

Results

Study Demographics for POAG Cases and Controls

Table. Demographic Summarya.

POAG Risk Stratification by PRS

Figure 1. Odds Ratios by Quintile.

Figure 2. Improvement in Area Under the Receiver Operating Characteristic (AUROC) Curve Adding a Polygenic Risk Score (PRS) to the Baseline Model by Population and PRS.

Gharahkhani PRS

Han PRS

Craig PRS

Discussion

Strengths and Limitations

Conclusions

References

Associated Data

Supplementary Materials

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Table. Demographic Summary^a.