Overlap of Genetic Loci for Central Serous Chorioretinopathy With Age-Related Macular Degeneration

Abstract

Importance: Central serous chorioretinopathy (CSC) is a serous maculopathy of unknown etiology. Two of 3 previously reported CSC genetic risk loci are also associated with AMD. Improved understanding of CSC genetics may broaden our understanding of this genetic overlap and unveil mechanisms in both diseases.

Objective: To identify novel genetic risk factors for CSC and compare genetic risk factors for CSC and AMD.

Design, setting, and participants: Using International Classification of Diseases, Ninth (ICD-9) and Tenth (ICD-10) Revision code-based inclusion and exclusion criteria, patients with CSC and controls were identified in both the FinnGen study and the Estonian Biobank (EstBB). Also included in a meta-analysis were previously reported patients with chronic CSC and controls. Data were analyzed from March 1 to September 31, 2022.

Main outcomes and measures: Genome-wide association studies (GWASs) were performed in the biobank-based cohorts followed by a meta-analysis of all cohorts. The expression of genes prioritized by the polygenic priority score and nearest-gene methods were assessed in cultured choroidal endothelial cells and public ocular single-cell RNA sequencing data sets. The predictive utility of polygenic scores (PGSs) for CSC and AMD were evaluated in the FinnGen study.

Results: A total of 1176 patients with CSC and 526 787 controls (312 162 female [59.3%]) were included in this analysis: 552 patients with CSC and 343 461 controls were identified in the FinnGen study, 103 patients with CSC and 178 573 controls were identified in the EstBB, and 521 patients with chronic CSC and 3577 controls were included in a meta-analysis. Two previously reported CSC risk loci were replicated (near CFH and GATA5) and 3 novel loci were identified (near CD34/46, NOTCH4, and PREX1). The CFH and NOTCH4 loci were associated with AMD but in the opposite direction. Prioritized genes showed increased expression in cultured choroidal endothelial cells compared with other genes in the loci (median [IQR] of log 2 [counts per million], 7.3 [0.6] vs 4.7 [3.7]; P = .004) and were differentially expressed in choroidal vascular endothelial cells in single-cell RNA sequencing data (mean [SD] fold change, 2.05 [0.38] compared with other cell types; P < 7.1 × 10-20). A PGS for AMD was predictive of reduced CSC risk (odds ratio, 0.76; 95% CI, 0.70-0.83 per +1 SD in AMD-PGS; P = 7.4 × 10-10). This association may have been mediated by loci containing complement genes.

Conclusions and relevance: In this 3-cohort genetic association study, 5 genetic risk loci for CSC were identified, highlighting a likely role for genes involved in choroidal vascular function and complement regulation. Results suggest that polygenic AMD risk was associated with reduced risk of CSC and that this genetic overlap was largely due to loci containing complement genes.

Figure 1.. Significant Loci in the Genome-Wide Association Study (GWAS) Meta-analysis

A, Manhattan plot for the central serous chorioretinopathy (CSC) genome-wide association study meta-analysis. A total of 1176 patients with CSC and 526 787 controls from the FinnGen study, Estonian Biobank, and a previously reported European clinical chronic CSC (cCSC) cohort were included in the meta-analysis. Each data point corresponds to a single genetic variant. A Bonferroni-corrected 2-sided genome-wide P value threshold of 5 × 10⁻⁸ (dashed line) was used to establish significance accounting for multiple comparisons. Each locus is annotated with the name(s) of the nearest protein coding gene(s). B, Expected and observed P value distributions are represented in a quantile-quantile plot. C, Cohort-specific effect estimates are shown for the meta-analysis and separately for the FinnGen study (552 patients with CSC and 343 461 controls), Estonian Biobank (103 patients with CSC and 178 573 controls), and a previously reported European CCSC cohort (521 patients with CSC and 3577 controls). Squares and lines correspond to odds ratios and 95% CIs, respectively.

Figure 2.. Expression of Prioritized and Nonprioritized Genes in Human Primary Cultured Choroidal Endothelial Cells

Basal gene expression levels were quantitated in cultured choroidal endothelial cells from 10 human cadaveric donors (5 males and 5 females) using RNA sequencing. Expression is shown separately for 6 genes in the association loci that were prioritized by the polygenic priority score or nearest-gene method (CFH, CD34, CD46, NOTCH4, PREX1, and LAMA5), 54 nonprioritized genes in the association loci, and all other 13 128 quantifiable genes. Genes were designated as belonging to the association loci if they were located within ±500 kilobases from the genome-wide association meta-analysis lead variants. The 25%, 50%, and 75% percentiles and 1.5 × IQRs for each group are represented by box and whisker plots. P values were calculated with the Wilcoxon rank sum test. Exact expression levels for genes in the association loci are additionally presented in eTable 14 in Supplement 2. No measurable expression was observed for the nearest protein-coding gene GATA5.

Figure 3.. Comparison of Effect Estimates Between Central Serous Chorioretinopathy (CSC) and Age-Related Macular Degeneration (AMD)

Effect estimates (natural logarithms of odds ratios [ORs]) for CSC (y-axis) and AMD (x-axis) are contrasted for 5 lead variants from the genome-wide association study (GWAS) meta-analysis of CSC (A), and 32 lead variants that were reported in a previous GWAS of AMD (B). Pearson correlation coefficients and best-fit lines based on linear regression are shown.

Figure 4.. Predictive Utility of Polygenic Scores (PGS) for Central Serous Chorioretinopathy (CSC) and Age-Related Macular Degeneration (AMD)

PGSs were used to predict prevalent CSC (552 patients with CSC and 343 461 controls) and AMD (8913 patients with AMD and 348 936 controls) in the FinnGen study. PGSs were constructed based on previously reported genome-wide association studies of chronic CSC (cCSC) and AMD. Additional versions of the PGS for AMD were calculated after removing variants in the CFH locus and all 6 loci previously annotated as complement associated (CFH, CFI, C9, C2-CFB-SKIV, C3, and TMEM97-VTN loci). In each logistic regression model, a PGS, age at end of follow-up or death, age at end of follow-up or death squared, sex, first 5 genomic principal components, genotyping batch, and genotyping array were used as covariates. OR indicates odds ratio.