Common and rare variants in patients with early onset drusen maculopathy

Abstract Early onset drusen maculopathy (EODM) can lead to advanced macular degeneration at a young age, affecting quality of life. However, the genetic causes of EODM are not well studied. We performed whole genome sequencing in 49 EODM patients. Common genetic variants were analysed by calculating genetic risk scores based on 52 age‐related macular generation (AMD)‐associated variants, and we analysed rare variants in candidate genes to identify potential deleterious variants that might contribute to EODM development. We demonstrate that the 52 AMD‐associated variants contributed to EODM, especially variants located in the complement pathway. Furthermore, we identified 26 rare genetic variants predicted to be pathogenic based on in silico prediction tools or based on reported pathogenicity in literature. These variants are located predominantly in the complement and lipid metabolism pathways. Last, evaluation of 18 genes causing inherited retinal dystrophies that can mimic AMD characteristics, revealed 11 potential deleterious variants in eight EODM patients. However, phenotypic characteristics did not point towards a retinal dystrophy in these patients. In conclusion, this study reports new insights into rare variants that are potentially involved in EODM development, and which are relevant for future studies unravelling the aetiology of EODM.


| INTRODUCTION
Age-related macular degeneration (AMD) is one of the leading causes of severe visual impairment among elderly globally. 1 Genetic predisposition in combination with ageing and several environmental factors underlie the disease aetiology. In a recent genome-wide association study (GWAS), including~16 000 AMD cases and~1 7 000 control individuals, 52 genetic variants are found to be independently associated with AMD, including 45 common and seven rare variants. 2 The complement pathway is considered the most important pathway involved in AMD, as 19/52 risk increasing and protective variants are located in or near genes of the complement pathway. 2 In addition, a collective enrichment of rare variants in complement genes has been identified in AMD case-control and family studies, 3,4 however, due to their low allele frequency, only for a limited number of individual rare variants a robust association could be established. 2,5,6 Although AMD mainly affects the elderly, some individuals develop AMD characteristics, including drusen, which are deposits that accumulate underneath the retinal pigment epithelium (RPE), RPE alterations, geographic atrophy (GA) and choroidal neovascularization (CNV) at a much younger age. Consequently, these patients with early onset drusen maculopathy (EODM) often develop the vision-threatening end stages of the disease, including GA and CNV, earlier in life. Its early onset suggests a prominent role for genetic factors rather than environmental factors in EODM development. In particular rare, highly penetrant variants may contribute to an early onset of AMD characteristics. [7][8][9][10] Recently, we evaluated a large cohort of 89 EODM patients, and identified rare variants in complement factor H (CFH) in~30% of EODM patients. 11 We hypothesised that in EODM patients without a potential causative rare variant in CFH, rare variants in other genes might contribute to EODM. In addition, common AMD-associated variants may also contribute to EODM, as we hypothesise that EODM is an early manifestation of AMD, caused by a combination of common variants and highly penetrant rare variants. Caution is warranted when studying EODM, as phenotypically some inherited retinal dystrophies (IRDs) resemble EODM, for example, Stargardt disease (characterised by yellow flecks, which are different from drusen, and GA) and Sorsby fundus dystrophy (SFD, characterised by yellow, subretinal, drusen-like deposits and CNV).
The aim of this study is to determine the contribution of common AMD-associated variants, and to identify potential disease-causing rare variants in patients with EODM.

| Study population
We collected a cohort of 89 unrelated patients with EODM from the European Genetic Database (EUGENDA), as described previously, 11 and one EODM patient from the Institut de la Màcula, Barcelona. EODM was defined as any sign of age-related maculopathy diagnosed ≤55 years of age, or severe signs of age-related maculopathy diagnosed between 56 and 65 years of age. In this current study, we selected 49 patients from the EODM cohort in whom no rare, potential causative variants in the complement genes CFH, CFI, C3, C9 and CFB were identified. 11 All except one of the EODM patients were from European ancestry. Colour fundus photographs of EODM patients were graded by experienced graders from EyeNED Reading Center under supervision of a senior specialist according to the international grading system based on the Wisconsin Age-Related Maculopathy Grading System, and reclassified into the Rotterdam Classification, as described previously. [12][13][14] In addition, a case-control cohort from the EUGENDA database, including 925 control individuals without AMD ≥65 years of age, 577 early/intermediate AMD patients, and 1155 advanced AMD patients, was used as reference for the genetic risk score (GRS) distribution. Grading of the reference cohort was performed according to the CIRCL grading protocol, as described previously. 15

| Genetic risk scores
The 52 AMD-associated variants, as reported previously, 2 were extracted from the WGS dataset (Table S1). Subsequently, an overall GRS (including all 52 variants), a complement GRS (including 19/52 complement-related variants), and a lipid GRS (including 7/52 lipidrelated variants) were calculated based on the formula:

| Filtering of the rare variants
We performed multiple filtering steps to identify potential diseasecausing variants in EODM patients. As a first step, WGS data were Note: General characteristics of the 49 patients with early onset drusen maculopathy, included in this study. Disease stage is based on the Rotterdam Classification, as described previously. 14 Abbreviations: AMD, age-related macular degeneration; ARM, age-related maculopathy; EODM, early onset drusen maculopathy.

| Evaluation of 18 IRD genes
Overlap in phenotypic characteristics between IRD and AMD has previously been reported. 21 To investigate whether EODM may be a phenocopy from Mendelian macular dystrophies, we screened 18 IRD genes to rule out potential misdiagnoses. In total, 44 rare nonsense, frameshift, protein-altering or splice-site variants were identified in nine different IRD genes (Table S5). None of the EODM patients carried biallelic variants in the six genes described to cause AR IRDs, and none of the copy number variants or structural variants were identified in the EODM cohort. We evaluated the available imaging in detail for 11 variants identified in the ABCA4, ABCC6, FSCN2, PRDM13, TIMP3 and IMPG1 genes, as they were predicted to be pathogenic based on all three in silico prediction tools, or they were reported to be (likely) pathogenic in ClinVar or in literature (Table S6). Retinal images of the patients carrying potential pathogenic variants showed drusen in combination with other fundus features, such as RPE alterations, GA and CNV, which are characteristic for AMD, and not for IRD ( Figure S1). All rare variants identified in candidate genes in EODM patients are listed in Table S7. Potential deleterious rare variants in non-candidate genes are depicted in Table S8.

| DISCUSSION
The present study comprehensively analysed the role of common and rare variants in EODM. We showed that the 52 AMD- Regarding the contribution of common variants to EODM, we observed that especially complement-related variants are involved in EODM, and complement-related variants seem to contribute most in advanced disease stages, although this latter warrants further research in larger EODM cohorts. Our findings are in line with previous literature, in which a higher complement GRS was observed in advanced AMD stages, 27 and higher systemic complement activation levels in advanced AMD stages were reported. 28 Lipid-related variants did not contribute to EODM in this current study, although we did ciencies (e.g. C9 deficiency in the Japanese population). 37,38 Higher complement activation levels are reported in AMD patients compared to control individuals. 28 Therefore, we expect that nonsense mutations in C5-C9 are protective for AMD, and presumably also for EOMD, considering the same underlying disease mechanisms.
Nevertheless, it is important to note that EODM is likely due to a combination of risk increasing and risk decreasing genetic variants. To our knowledge, this is the first study using WGS in a large cohort of EODM patients, and allowed for the evaluation of both common and rare variants, and revealed insights into biological disease pathways that may contribute to EODM. This study also has its limitations. The limited sample size and the design of the study did not allow to detect significant associations of rare variants with EODM. However, we identified several rare variants in EODM patients that are potentially involved in EODM. These findings should be replicated in a larger cohort of EODM patients, to elucidate their contribution to EODM. Furthermore, many variants were not reported previously, were of uncertain clinical significance, or were predicted to be (likely) pathogenic based on in silico prediction tools. Testing the effect of these variants in AMD/EODM model systems (e.g. mouse or iPSc-based models), or by measuring protein or metabolite levels in patients with specific rare variants, will improve the interpretation of these variants and may result in new insights in the EODM pathogenesis. In the current study family members of EODM patients were unfortunately not available, therefore, we recommend ascertaining family members for segregation analysis in a follow-up study. We consider EODM as an early manifestation of AMD, which is a multifactorial disease. In some EODM cases a high GRS based on the AMD-associated variants may already explain the disease. In other cases where the GRS is low and where we did not find potential deleterious rare variants in the studied candidate genes, the cause may lie in other genomic regions or in genes that were not evaluated in this study. In addition, we focused solely on genetic factors, and did not analyse non-genetic risk factors such as age, smoking and diet, as we hypothesised that EODM is more genetically driven compared to AMD due to the early age of onset.

| CONCLUSION
In this current study we showed that common AMD-associated variants contribute to EODM development, and we identified rare, potential deleterious variants that might play a role in EODM pathogenesis.
Together with the identification of rare CFH variants in a large proportion of EODM patients, 11 we consider EODM as an early manifestation of AMD, in which both common and rare variants contribute to its pathogenesis. Roche.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article.