Introduction
Glaucoma is a leading cause of irreversible vision loss worldwide [
1]. Adult-onset primary open-angle glaucoma (POAG), the major type of glaucoma, has an inheritable element but also results from a complex interplay among various environmental, genetic, and lifestyle factors [
2]. Recent studies on the heritability of POAG suggest that genetic factors make a significant contribution to POAG, with high heritability [
3,
4]. Similarly, a systemic meta-analysis reported high heritability in the individual endophenotypes responsible for POAG [
5].
Genetic factors are mediated by intermediate phenotypes; these phenotype attributes known to contribute to risk for POAG can be classified as either ocular or systemic [
6]. Well-recognized systemic risk factors include diabetes, hypertension, low diastolic blood pressure, migraine, and thyroid disease [
7]. Among ocular risk factors, the most important is elevated intraocular pressure (IOP) [
5]. However, given the fact that a significant proportion of patients diagnosed under the broad classification of open-angle glaucoma actually display IOP values within the normal range, it is supposed that individual biomechanical architecture around the optic nerve head could mediate susceptibility to glaucomatous damage. In particular, myopia, which involves elongation of the eye and resultant stretching of the peripapillary sclera and optic nerve, shows substantial causal association with POAG [
8]. Corneal hysteresis (CH), a measure of viscoelastic damping of cornea, represents an alteration of ocular biomechanics. Low level of CH is associated with the development and progression of glaucoma [
9‐
11]. The close association of myopia and CH with POAG suggests a role of altered ocular biomechanics in glaucoma pathogenesis [
12].
Several large-scale genome-wide association studies (GWAS) on POAG have been conducted to date, identifying hundreds of associated loci [
13]. Polygenic risk scores (PRS) are widely used to synthesize disease-associated single-nucleotide polymorphisms (SNPs) into a single score and offer a means of risk stratification for common chronic diseases, such as POAG [
14,
15]. The polygenic nature of POAG suggests that a synthesizing measure such as a PRS could be useful for determining individual disease risk [
15]. Previous research has shown that polygenic variants for POAG are associated with more than 2.5-fold increased odds of incident glaucoma, which is comparable to known monogenic risk variants, and they are ~ 15 times more prevalent than a monogenic variant in the general population [
16]. A PRS for POAG has likewise demonstrated utility in risk stratification, with significant association with disease onset, disease severity and treatment intensity [
14].
Considering that glaucoma is phenotypically linked with many systemic and ocular conditions, it is crucial to explore the associations between genetic predisposition to glaucoma and these phenotypes. A phenome-wide association study (PheWAS) is a hypothesis-free approach that tests a set of selected genetic variants or a PRS for association with hundreds of phenotypes without prior assumptions. As such, a PheWAS can be useful for investigating the genetic basis of glaucoma, where the pathogenesis and underlying genetic factors are not well understood.
In this study, we used genomic and phenotypic data from the UK Biobank and Penn Medicine Biobank to conduct a PRS-based PheWAS for POAG, exploring a wide range of disease phenotypes. We then assessed the distribution of patient characteristics across PRS-based glaucoma genetic risk groups and the associations of POAG PRS with ocular factors, focusing specifically on IOP, myopia, and CH. Finally, we investigated the impact of POAG PRS on risk of developing incident glaucoma and explored the combined impact of genetic factors and ocular factors on the risk of incident glaucoma.
Discussion
Our study aimed to identify a broad spectrum of phenotypes associated with POAG genetic risk and to investigate how this genetic risk interacts with ocular risk factors in the development of POAG. Notably, we observed that individuals in the top 1% PRS for POAG exhibited a significantly higher risk of new glaucoma events, with a HR of 8.8. Furthermore, our analysis of the joint associations of polygenic risk and ocular risk factors revealed that individuals with low baseline IOP and very high polygenic risk had a similar hazard ratio for developing glaucoma compared to those with high baseline IOP and low polygenic risk (absolute risk, 3.62% vs. 3.20%).
Increased IOP is considered to be the most important evidence-based risk factor for the development and progression of glaucoma, with a considerable number of SNPs in the POAG PRS implicated in increasing IOP [
6,
16]. Nevertheless, noteworthy is the observation that disease susceptibility was substantially increased even among individuals with a high genetic susceptibility but low IOP. This underscores the strong influence of polygenic risk on the development of glaucoma. In this regard, in addition to regular IOP checks, the assessment of POAG genetic risk could serve as a robust tool for the early identification of individuals at high risk for the disease.
Genetic predisposition to POAG has been previously associated with the severity and predictive aspects of the disease. Individuals in the top 5% PRS reportedly had 2.8-fold increased odds of glaucoma [
16]. In the cohort study of PROGRESSA, subjects with early POAG with high polygenic risk had faster structural and functional progression [
34]. In the glaucoma population, subjects with high PRS had increased risk of younger age at diagnosis and higher requirement for incisional surgery [
14]. In addition to the findings of the previous studies, we also demonstrated the robust synergistic effect between genetic and ocular factors in predicting the future glaucoma development, showing that the participants with high genetic burden exhibited 28.9-fold, 11.8-fold, and 9.2-fold increased risk of glaucoma incidence, when they accompanied high IOP, moderate to high myopia, and low CH, respectively. Craig et al. [
14] previously reported the improved predictive capacity through the integration of polygenic risk and IOP and vertical CD ratio when estimating glaucoma odds ratios. However, to our knowledge, a large and well-phenotyped prospective cohort study assessing the comprehensive joint association of ocular factors and genetic risk has not been reported. The manifestation of genetic risk for glaucoma substantially increased risk of glaucoma incidence by joint associations of genetic and ocular factors, emphasizing the significance of risk stratification based on genetic and ocular risk factors for early detection and prevention of the disease.
In this study, we conducted a PRS-based PheWAS analysis to gain better insight and identify associations between genetic predisposition to POAG and disease phenotypes available in the UK Biobank and Penn Medicine Biobank cohorts. Our findings consistently revealed that the genetic liability for POAG to have a predominant impact on the development of ocular conditions, with little correlation to systemic disease phenotypes in both the discovery and replication sets.
In contrast to the results of a PRS-based PheWAS, our investigation using PheWAS with POAG diagnosis as the independent variable revealed significant associations with various systemic conditions, which were previously identified as systemic risk factors for glaucoma in epidemiologic studies [
7,
35‐
39]. Despite the distinct disease associations of POAG, we found no positive associations between genetic liability for POAG and systemic disease phenotypes. In accordance with our study, a previous work has reported limited genetic correlation of diabetes with glaucoma-related endophenotypes and POAG [
40]. Kolli et al. [
41] recently reported that the association between glaucoma and cardiometabolic disease differed by genetic risk for glaucoma, reporting that glaucoma patients with low genetic risk tended to have a higher prevalence of cardiometabolic disease, while the opposite trend was observed for those with high genetic risk. The differential PheWAS results of POAG and POAG polygenic risk in our study suggests that systemic conditions may have a substantial impact on the development of glaucoma, potentially by sharing confounding risk factors with POAG or serving as prominent non-genetic risk factors.
In our study, both POAG PRS and POAG diagnosis were significantly associated with overrepresentation of the ocular disease phenotypes such as cataract, uveitis, retinal detachment, and myopia with the same directional effect. POAG is reportedly significantly associated with cataracts and myopia [
42,
43]. Elevated IOP and the use of glaucoma medication increase the risk of 5-year incident cataracts [
42]; additionally, there is a higher prevalence of retinal detachment among patients with glaucoma, particularly in cases where glaucoma is advanced or the IOP is poorly controlled [
44]. Meanwhile, in consistent with results from a recent large multi-ethnic meta-analysis of GWASs, we found diseases such as celiac disease or unspecified diffuse connective tissue disease to have negative associations with POAG PRS, suggesting that these conditions may have genetic factors that are protective against the development of glaucoma or vice versa [
45]. Predominant association of genetic predisposition to POAG with ocular disease phenotypes in the PheWAS suggest that common genetic variants associated with POAG primarily drive ocular conditions, rather than involving disease pathways associated with metabolic traits.
In line with the finding from the results of PheWAS, our prospective study confirmed that the initial presence of ocular factors, specifically elevated IOP, myopia, and low CH, was significant predictors for the development of incident glaucoma. Furthermore, these ocular factors were significantly associated with the genetic risk of glaucoma. Choquet et al. [
46] reported that myopia and glaucoma demonstrate a shared genetic architecture. Recently, low CH was reportedly associated with POAG PRS and greater prevalence of disc hemorrhage [
47]. These significant associations of POAG PRS with myopia and CH implicates that the genetic pathways responsible for POAG have clinically significant implications for the pathogenic role of the ocular biomechanics. In addition to ocular hypertension, these altered ocular biomechanics could collectively increase ocular susceptibility to glaucomatous damage in the general population. Further prospective studies are necessary to clarify the association between myopia, CH, and the onset of new glaucoma.
In this study, we developed and assessed PRS-PheWAS for individuals of European ancestry. Nonetheless, we acknowledge the importance of developing PRSs for groups of non-European ancestry, given the potential challenges associated with the transferability of PRSs across different ancestries and ethnic groups. Recently, Verma et al. [
48] reported a multi-cohort GWAS in Africans including the Penn Medicine Biobank and showed that, for individuals of African descent, PRS for POAG based on African GWAS outperformed PRS from a much larger GWAS derived from European populations. Additionally, the study by Gharahkhani et al. [
49] conducted a large multi-ethnic meta-analysis of GWAS, confirming that several genetic variants share their effects on POAG across ethnicities, including those of European descent. However, it also identified ancestry-specific genetic factors (e.g., rs16944405 in IQGAP1), indicating that polygenic diseases with numerous ancestry-specific genetic variants, like POAG, might exhibit lower transferability across ethnicities in PRS analyses. Therefore, it is important to perform GWAS and validate PRS and PRS-PheWAS results in multi-ethnic populations in further studies.
In this study, to maintain consistency, the PRS-CS approach was used in the PRS-PheWAS and analyses of joint associations with ocular factors. This is one of the widely used Bayesian polygenic modeling approaches and minimizes parameter selection bias through the auto option using the linkage disequilibrium reference panel. However, it is also crucial to acknowledge that several other approaches (PRS-CT [clumping and thresholding], PRSice-2, lassosum, LDpred2, SBLUP, SbayesR, DBSLMM, etc.) [
50‐
55] exist, and different methods can influence the results. For transparency and reproducibility in genetic research, we used the PRS model (PRS-CS) that is publicly available.
This study has several strengths. First, to better identify the multiple disease phenotypes associated with POAG PRS, we used a hypothesis-free approach, which overcomes an incomplete understanding of the disease mechanism. Additionally, a notable strength of this study is the large sample size, which allowed us to investigate whether genetic markers associated with POAG may be related to ocular factors in the general population. Furthermore, our findings were overall replicated in an independent data set. The two cohort datasets used in this study might have distinct demographic and clinical characteristics. The UK Biobank is a prospective national cohort study based on healthy participants, whereas the Penn Medicine Biobank is an academic research cohort derived from a regional university hospital with diverse ancestries. Despite these potential differences, our replicated results might enhance the generalizability of our study, expanding its relevance to both the healthy population and those associated with a hospital setting. Next, we used a longitudinal study design to assess the impact of POAG PRS on glaucoma incidence. Finally, this study defined the very high-risk group as those with a top 1% PRS, as opposed to using deciles in previous studies [
16,
41,
47]. This difference in risk stratification could potentially account for the elevated HR observed in this study.
However, the study also has limitations. First, the assessment of glaucoma progression was constrained by the limited parameters available within the datasets. Second, although we used POAG GWAS summary statistics from a large-scale meta-analysis, our analyses were limited to individuals of European ancestry. Therefore, further studies with other cohorts are warranted to verify the generalizability of our findings. Third, drawing upon the definition used in previous reports [
28,
29], we defined POAG on the basis of ICD codes or self-reports. This method of definition could have resulted in overestimation or underestimation of disease. Fourth, our study is less likely to uncover POAG risk factors that have no or only a weak genetic link to the disease. The stringent statistical criteria applied in our PheWAS may have excluded associations that could be potentially noteworthy. Fifth, in our study, genotype data of the two biobanks were imputed using different imputation reference panels. To mitigate potential biases arising from these limitations, we focused our analysis on individuals of European descent and used the HapMap3 variants using the PRS-CS approach for PRS-PheWAS. Nevertheless, the accuracy and consistency of PRS could be further improved by using the same reference panel across discovery and replication datasets.
In conclusion, our study demonstrated the potential of PRS-based PheWAS in revealing associations between genetic risk factors for glaucoma and various ocular conditions. The findings emphasized the importance of considering genetic susceptibility in understanding glaucoma and highlighted shared genetic bases between glaucoma and other ocular conditions. We could find that the overall genetic liability to POAG primarily contributes to ocular conditions and has only limited associations with systemic disease phenotypes. Genetic predisposition to POAG accompanies lower CH, and refractive error, as well as ocular hypertension in the study population. The genetic risk for POAG demonstrated a robust association with the occurrence of incident glaucoma and the participants with high genetic burden exhibited a substantially increased risk of glaucoma incidence when they accompanied high IOP, moderate to high myopia, and low CH. The novel joint effect between genetic burden and ocular risk factors for POAG implicates that the interplay between genetic and ocular factors may substantially contribute to the increased risk of glaucoma, highlighting the necessity for rigorous glaucoma monitoring in populations with a high genetic burden.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.