Article Text

Download PDFPDF
Genetic associations of central serous chorioretinopathy: a systematic review and meta-analysis
  1. Zhen Ji Chen1,
  2. Shi Yao Lu1,
  3. Shi Song Rong2,
  4. Mary Ho1,3,
  5. Danny Siu-Chun Ng1,
  6. Haoyu Chen4,
  7. Bo Gong5,
  8. Jason C Yam1,
  9. Alvin L Young1,3,
  10. Marten Brelen1,3,
  11. Clement C Tham1,
  12. Chi Pui Pang1,4,
  13. Li Jia Chen1,3
  1. 1 Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, China
  2. 2 Department of Ophthalmology, Harvard Medical School, Boston, Massachusetts, USA
  3. 3 Department of Ophthalmology and Visual Sciences, Prince of Wales Hospital, Hong Kong, China
  4. 4 Joint Shantou International Eye Center, Shantou University, Shantou, China
  5. 5 Sichuan Key Laboratory for Disease Gene Study, Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital, Chengdu, China
  1. Correspondence to Dr Li Jia Chen, Department of Ophthalmology and Visual Science, The Chinese University of Hong Kong, Hong Kong, China; lijia_chen{at}cuhk.edu.hk

Abstract

Aims To identify single-nucleotide polymorphisms (SNPs) associated with central serous chorioretinopathy (CSCR) by a systematic review and meta-analysis, and to compare the association profiles between CSCR, neovascular age-related macular degeneration (nAMD) and polypoidal choroidal vasculopathy (PCV).

Methods We searched the EMBASE, PubMed and Web of Science for genetic studies of CSCR from the starting dates of the databases to 12 September 2020. We then performed meta-analyses on all SNPs reported by more than two studies and calculated the pooled OR and 95% CIs. We also conducted sensitivity analysis and adopted the funnel plot to assess potential publication bias.

Results Totally 415 publications were reviewed, among them 10 were eligible for meta-analysis. We found 10 SNPs that have been reported at least twice. Meta-analysis and sensitivity analysis confirmed significant associations between CSCR and six SNPs in three genes, namely age-related maculopathy susceptibility 2 (ARMS2) (rs10490924, OR=1.37; p=0.00064), complement factor H (CFH) (rs800292, OR=1.44; p=7.80×10−5; rs1061170, OR=1.34; p=0.0028; rs1329428, OR=1.40; p=0.012; and rs2284664, OR=1.36; p=0.0089) and tumour necrosis factor receptor superfamily, member 10a (TNFRSF10A) (rs13278062, OR=1.34; p=1.44×10−15). Among them, only TNFRSF10A rs13278062 showed the same trend of effect on CSCR, nAMD and PCV, while the SNPs in ARMS2 and CFH showed opposite trends in the SNP associations.

Conclusions This study confirmed the associations of ARMS2, CFH and TNFRSF10A with CSCR, and revealed that ARMS2, CFH and TNFRSF10A may affect different phenotypic expressions of CSCR, nAMD and PCV.

  • choroid
  • genetics
  • retina
  • macula
  • neovascularisation

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information. All the data included in our study are from published studies which can be searched in PubMed, Embase and Web of science.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Central serous chorioretinopathy (CSCR) is a group of chorioretinal diseases characterised by serous detachment of the neurosensory retina with dysfunction of the retinal pigment epithelium (RPE) and choroidal vascular hyperpermeability. It can cause central visual blurring and metamorphopsia.1 CSCR is among the 10 most common diseases of the posterior segment of the eye and a frequent cause of visual impairment during working age.2 It is more common in males than females, affecting about 10 in 100 000 males and 2 in 100 000 females aged between 20 and 50 years.3–5 CSCR has variable presentations. Some present as a self-limiting unilateral disease, but a substantial proportion of cases present with bilateral occurrence, chronic persistent subretinal fluid and high recurrence rate. According to disease duration, CSCR can be classified into two forms: acute CSCR and chronic CSCR.5 Acute CSCR often recovers spontaneously with no significant visual deficit, but it may recur and progress to chronic CSCR in some patients.4 Chronic CSCR manifests as persistent subretinal fluid more than 3–6 months in duration, and such persistent fluid leads to loss of RPE cells and hence impaired visual function.6

CSCR is a multifactorial disease resulted from the interactions of inherent, environmental, genetic and epigenetic factors. Known risk factors of CSCR include male gender, type A personality, pregnancy, obstructive sleep apnoea, corticosteroid use, Helicobacter pylori infection and family history.7 The involvement of genetic factors in CSCR has been demonstrated by the identification of disease-associated genes for CSCR. In 2014, the CFH gene, an associated gene of neovascular age-related macular degeneration (nAMD) and polypoidal choroidal vasculopathy (PCV), was first reported to be associated with CSCR in a Japanese cohort.8 Thereafter, studies on specific genes or pathways also identified a number of candidate genes for CSCR, such as age-related maculopathy susceptibility 2 (ARMS2), tumour necrosis factor receptor superfamily, member 10a (TNFRSF10A), nuclear receptor subfamily 3 group C member 2 (NR3C2) and Cadherin 5 (CDH5).9–19 In 2018, the first genome-wide association study (GWAS) for CSCR reported the association of CFH with chronic CSCR among Europeans.20 Another CSCR GWAS in Japanese identified a new associated single-nucleotide polymorphism (SNP) rs11865049 in SLC7A5.21 Later in 2019, a GWAS in three Japanese cohorts and one European cohort found a novel locus, GATA5, for CSCR.22

Apart from these studies, other candidate gene analyses also reported the association of specific gene variants with CSCR in different populations. However, variation in the association results existed across the studies. For example, the alleles of three SNPs in CFH, namely rs1065489-T, rs1329428-C and rs3753394-T, were reported as risk alleles for CSCR among Greeks13 but as protective alleles among Dutchman, German and Japanese.8 9 12 20 23 This could be due to genuine ethnic specificity or individual cohort effects. So far, however, a thorough summary of the genetic associations of CSCR is lacking. Therefore, we performed a systematic review and meta-analysis on all CSCR genetic association reports, with a view to confirm the genetic associations and highlight potential biomarkers and pathological pathways of CSCR. Moreover, certain genes, such as ARMS2, CFH and TNFRSF10A, have also been associated with nAMD and PCV,24 suggesting the three maculopathies may share some common disease pathways. Therefore, we compared the genetic association profiles among CSCR, nAMD and PCV.

Methods

Literature search

We conducted a literature search in EMBASE, PubMed and Web of Science. All related records from the start dates of respective databases to 12 September 2020 were retrieved using Boolean logic and controlled vocabularies, that is, Medical Subject Heading terms. The searching strategy was (‘central serous chorioretinopathy’ OR ‘central serous retinopathy’) AND (‘genetics’ OR ‘single nucleotide polymorphism’ OR ‘SNP’ OR ‘genetic association’ OR ‘genotype’ OR ‘allele’ OR ‘gene’ OR ‘genetic variation’ OR ‘variant’ OR ‘polymorphism’ OR ‘single nucleotide polymorphisms’ OR ‘SNPs’ OR ‘genetic associations’ OR ‘genotypes’ OR ‘alleles’ OR ‘genes’ OR ‘genetic variations’ OR ‘variants’ OR ‘polymorphisms’).

Selection criteria

In this meta-analysis, we included articles that fulfilled the following criteria: (1) original genetic association studies of CSCR; (2) studies comparing the allelic or genotypic frequencies between patients with CSCR and normal controls; and (3) studies with sufficient data for estimation of effect size, that is, OR with 95% CIs or allele frequency with sample size. We excluded reviews, abstracts of conferences, commentaries and republished data. Moreover, all SNPs included in the meta-analysis had been reported in at least two different cohorts. Language of the articles was not limited.

Data extraction

Two reviewers (ZC and SL) performed the literature search and data extraction independently. The following information and data were extracted: studies (combined with first author’s name and year of publication), study cohorts, genotyping methods, sample sizes of cases and controls, genes, SNPs, allele counts and minor allele frequencies in cases and controls and/or ORs with 95% CIs, if available. Disagreements were resolved through discussion with another two reviewers (SSR and LJC).

Statistical analysis

We generated summary outcomes (ORs and 95% CIs) using inverse-variance-weighted meta-analysis. The I 2 statistics was used to evaluate individual heterogeneity. The fixed-effect model was deployed in case of an I 2 value <25%, which suggested low heterogeneity among studies. If heterogeneity existed among studies with an I 2 value of ≥25%, we measured the outcomes using the random-effect model (DerSimoniane-Laird estimator). A summary p value of <0.05 was considered statistically significant. Moreover, to find out whether there might be specific studies that affected the overall results, we conducted a sensitivity analysis by excluding individual studies one at a time and recalculated the summary ORs and 95% CIs after removing the study. We also estimated potential publication bias using funnel plots. The R software (V.3.6.1; http://cran.r-project.org/) and R studio (V.1.2.5019, https://rstudio.com/) were adopted for statistical analyses.25

Results

Inclusion of studies

A total of 415 publications were screened for eligible studies (figure 1). After removing 143 duplicated studies and 165 irrelevant records or conference abstracts, we further excluded 86 studies which did not meet our eligibility criteria. Among them, 10 were review articles, 7 were cell biology studies and 69 were clinical studies on CSCR without genetics data. We then reviewed the full texts of 21 studies which investigated the genetic associations of CSCR. Eleven of them were further removed, including four studies that involved only CSCR cases but no controls,14 15 26 27 a study on CSCR pedigrees to explore rare variants,16 one study on copy number variants (CNVs),28 and five reports on specific variations which appeared only once in the database.17 19 21 29 30 Eventually, 10 studies were included in the meta-analysis (table 1). Among them, six studies had defined the subtypes of CSCR,9–12 20 23 while four used patients with mixed CSCR.8 13 18 22 Only one study conducted subgroup analysis based on acute and chronic CSCR.12 Therefore, we analysed the data of mixed CSCR subtypes.

Figure 1

Flowchart of the study inclusion process. CSCR, central serous chorioretinopathy.

Table 1

Characteristics of eligible studies for the meta-analysis

SNPs associated with CSCR

We identified six SNPs in three genes that showed significant associations with CSCR in the meta-analysis, including rs10490924 in ARMS2, rs800292, rs1061170, rs1329428 and rs2284664 in CFH, and rs13278062 in TNFRSF10A (table 2 and online supplemental figure 1). ARMS2 rs10490924 had a pooled OR of 1.37 (G allele; 95% CI: 1.14 to 1.64; p=0.00064; I 2=22%). TNFRSF10A rs13278062 had a pooled OR of 1.34 (T allele; 95% CI: 1.25 to 1.45; p=1.44×10−15; I 2=0). In CFH, four SNPs were associated with an increased risk of CSCR, including rs800292 (A allele; OR=1.44; 95% CI: 1.20 to 1.73; p=7.80×10−5; I 2=68%), rs1061170 (T allele; OR=1.34; 95% CI: 1.10 to 1.62; p=0.0028; I 2=33%), rs1329428 (T allele; OR=1.40; 95% CI: 1.08 to 1.82; p=0.012; I 2=86%) and rs2284664 (T allele; OR=1.36; 95% CI: 1.08 to 1.70; p=0.0089; I 2=61%). The association of the other four SNPs with CSCR were not statistically significant, that is, CDH5 rs7499886, CFH rs1065489 and rs3753394 and NR3C2 rs2070951 (p>0.05, table 2).

Table 2

Allelic associations of SNPs with CSCR

Sensitivity analysis and potential publication bias

We performed sensitivity analysis to evaluate the effect of individual studies by omitting one study at a time (online supplemental figure 2). The allelic association results remained stable for four SNPs, that is, CFH rs800292, rs1329428 and rs2284664, and TNFRSF10A rs13278062. The association of three SNPs (CFH rs1065489 and rs3753394, and NR3C2 rs2070951) with CSCR were changed in the sensitivity analysis after removing Moschos et al’s study13 in Greeks (for CFH rs1065489 and rs3753394) or Hosoda et al’s study23 in East Asians (for NR3C2 rs2070951) (table 3 and online supplemental figures 2 and 3). Moreover, the shapes of the funnel plots of the three SNPs indicated potential publication bias (online supplemental figure 4).

Table 3

Sensitivity analysis of allelic associations of SNPs with CSCR

Discussion

In this systematic review and meta-analysis, we summarised the associations of all eligible SNPs with CSCR, and confirmed significant association of six SNPs in three genes, including rs10490924 in ARMS2, rs800292, rs1061170, rs1329428 and rs2284664 in CFH and rs13278062 in TNFRSF10A. Notably, high interstudy heterogeneities were found in most of the SNPs, which might be caused by cohort-specific effects or epigenetics effects. Differences in linkage disequilibrium (LD) structures across different populations may also lead to various genetic association results. So far, however, only a few studies had included haplotype association of CSCR and the SNPs in the haplotypes were different across the studies.8–10 Therefore, further haplotype-based association analyses in CSCR should be warranted.

Genetic similarities and differences between CSCR and nAMD/PCV by reported SNPs

The six CSCR-associated SNPs in ARMS2 (rs10490924), CFH (rs800292, rs1061170, rs1329428 and rs2284664) and TNFRSF10A (rs13278062) have also been associated with nAMD and PCV. We therefore compared the association profiles of these SNPs among CSCR, nAMD and PCV (table 4). Notably, only the T allele of TNFRSF10A rs13278062 conferred the same trend of effect. The ORs of rs13278062-T were comparable in CSCR (OR=1.34), nAMD (OR=1.26) and PCV (OR=1.28). In contrast, the SNPs in ARMS2 (rs10490924) and CFH (rs800292, rs1061170, rs1329428 and rs2284664) conferred opposite effects for CSCR as compared with nAMD and PCV (table 4).24 31–33 Although the risk alleles of all these SNPs were the same between nAMD and PCV, it had been shown that ARMS2 rs10490924-T had a significantly smaller OR in PCV than in nAMD.24 34 35 Other genes, such as SKIV2L and FGD6, also showed different effects between nAMD and PCV.36 37 Thus, even though CSCR, nAMD and PCV share some genetic components, different association patterns exist. Recently, a review summarised and compared the genetic characteristics of AMD and pachychoroid diseases among Japanese, and demonstrated the A allele of CFH rs800292 as a risk to CSCR but protective for AMD.38 Currently, how such opposite allelic effects affect the phenotypic expressions of CSCR and AMD/PCV remained unclear. It could be resulted from epigenetic effects, which are yet to be investigated. Also, the reported alleles can be proxies in LD with the causal alleles, which may have the same trends of effect among the diseases. Further deep sequencing of these genes with comprehensive LD analysis may help to elucidate the genetic architectures of the three maculopathies.

Table 4

Comparison of genetic associations of CSCR with nAMD and PCV.

Complement pathway in CSCR

This meta-analysis confirmed the association of CFH SNPs with CSCR. Complement factor H (CFH) is a negative regulator for immune response. Genetic variants at CFH affect the expression of CFH and subsequent complement activation at the human macular RPE–choroid interface.39 Activation of complement can induce vascular leakage.40 Leakage of sub-RPE choroidal vasculature is a feature in CSCR.41 Thus, the CFH variants and involved complement pathway may confer risk to CSCR via influencing choroidal vasculature leakage.

One of the CSCR-associated SNP, CFH rs800292-A, was reported by a recent GWAS of choroidal thickness, and was proposed to increase the risk of CSCR via a thicker choroid,23 which is a feature in most patients with CSCR.42 In contrast, rs800292-G was associated with thinner choroidal thickness,23 which increased the risk of some nAMD and PCV.24 31 Both thicker and thinner choroid have been revealed in different stages and subtypes of patients with nAMD and PCV.43–46 So far, there is no sufficient evidence to demonstrate the effects of thick or thin choroid on the development of CSCR, nAMD and PCV. Further genetic studies in choroidal parameters may provide some insights for dissecting the genetic associations of the diseases.

The CFH rs1061170 (Y402H) affects choroidal blood flow and ocular perfusion pressure.47 Young healthy subjects homozygous for the C allele, which is a risk allele for nAMD and PCV but protective for CSCR (table 4), had increased choroidal blood flow.47 Thus, CFH likely affects the pathogeneses via alterations of choroidal blood flow regulation. Of note, in aged CFH-H/H (where H is encoded by the C allele) transgenic mice fed with high-fat diet, the CFH Y402H variant induced AMD-like phenotypes consisting of increased RPE stress and increased basal laminar deposits.48 These features were not observed in aged CFH-Y/0 mice or in younger CFH mice of both genotypes fed with either diet, suggesting there were genotype-dependent changes in plasma and eyecup lipoproteins in aged CFH mice after high-fat diet, but not complement activation.48

These findings suggested that CFH may be involved in the pathogenesis of CSCR, nAMD and PCV via multiple processes, such as inflammation and complement activation, choroidal thickness variation, choroidal blood flow alteration and lipoprotein balance. In addition, the gene effects can be affected by factors such as ageing and diet, suggesting epigenetic effects. Further functional characterisation of CFH variants and epigenetics studies are warranted.

Gene involved in the regulation of RPE and choroidal thickness

This meta-analysis revealed a significant association between ARMS2 rs10490924 and CSCR in Europeans. Further confirmation of this association in other populations should be warranted. SNP rs10490924 is a major genetic risk factor for nAMD and PCV, while the effect is stronger for nAMD.24 Again, the allelic effects are opposite between CSCR and nAMD/PCV. The ARMS2 gene encodes a 107 amino acid protein, age-related maculopathy susceptibility protein 2 (ARMS2), which was reported to express in the mitochondria of the outer segment of photoreceptors and may lead to RPE dysfunction.49 50 Moreover, SNP rs10490924 was associated with subfoveal choroidal thickness on treatment-naive patients with PCV in a Japanese study.51 These indicated that ARMS2 may affect the pathogeneses by interrupting RPE function and choroidal thickness. However, how rs10490924 can affect the differential development of CSCR, nAMD and PCV is yet to be investigated.

Inflammation and apoptosis in CSCR

This meta-analysis also revealed a significant association between TNFRSF10A rs13278062 and CSCR. It is the only SNP that confers the same trend of risk effect on CSCR, nAMD and PCV. Notably, the risk effect of the T allele was the largest on CSCR (OR=1.34), followed by PCV (OR=1.28) and nAMD (OR=1.26) (table 4). But such ORs were generated from different study cohorts, therefore the comparison should be cautionary. TNFRSF10A encodes the tumor necrosis factor (TNF) -related apoptosis-inducing ligand receptor 1 (TRAILR1), a member of TNF-receptor family, expressing in both normal tissues and tumour cells.52 This receptor is also known as death receptor 4 and transmits the death signals to induce cell apoptosis.53 The proapoptotic death-signalling receptor for TNF-related apoptosis-inducing ligand (TRAIL) in a mouse model suppressed inflammation and tumourigenesis, indicating a TRAIL receptor (TRAILR) may regulate the inflammation and tumour suppressor in vivo.54 TRAILR1 was found in normal RPE cells and overexpressed in RPE cells of proliferative vitreoretinopathy patients.55 These suggested that TNFRSF10A may contribute to the pathogenesis of CSCR, nAMD and PCV through the function of TRAILR1 in RPE cells.

Other potential candidate genes for CSCR

Among the eligible studies in our meta-analysis, one study revealed a significant association between the NR3C2 rs2070951 and chronic CSCR, while another three studies showed a lack of significant association.10 12 18 23 In sensitivity analysis, we found that NR3C2 rs2070951 had a significant association with CSCR in European population (OR=1.23; p=0.00065, I 2=0). The excluded study was conducted in a Japanese cohort, in which the results indicated no association between rs2070951 and CSCR.23 The findings suggested that NR3C2 is likely a candidate gene for CSCR specifically for populations of European origin, which need further confirmation in other populations. NR3C2 encodes a mineralocorticoid receptor (MR), which is involved in signalling pathway affecting choroidal vasodilation and linking to chorioretinopathy in rat and human.56 A MR–aldosterone system is localised in the retina. It can be regulated by MR antagonism and angiotensin II type 1 receptor blockade. MR antagonism also improves pathological angiogenesis in oxygen-induced retinopathy and may involve in inflammatory pathways.57 Moreover, MR is expressed in the neuroretina and controls hydration of the healthy retina by regulating the ion and water channels.56 Thus MR likely involves in the pathogenesis of retinopathy, although further functional exploration is warranted.

Two eligible studies included in our meta-analysis involved the assessment of SNPs in CDH5 (Cadherin 5). One reported a significant association between CSCR and CDH5 SNPs (rs7499886, p=0.0001, OR=1.30; rs1130844, p=0.002, OR=0.78; and rs1073584, p=0.004, OR=0.77),11 while the other one reported negative results (rs7499886, p=0.96, OR=1.00; rs3837775, p=0.86, OR=0.98).11 23 CDH5 rs7499886 was not associated with CSCR in our meta-analysis (OR=1.17; p=0.081; table 2). Though it appeared significant after sensitivity test, there were only two cohorts from one study and hence it was excluded from further meta-analysis (online supplemental figure 3). CDH5 encodes a classical cadherin protein, which is a main constituent of endothelial adherens junctions. Cdh5 regulates vascular morphological development by interacting with F-Actin during the elongation of the endothelial cell. It may affect vascular changes and leakage in the choroid.58 Furthermore, Cdh5 is likely involved in corticosteroid response pathway in human and animal eyes.11 Thus, CDH5 may play a role in interplay with MR. Further validation of its SNPs in CSCR pathogenesis should be warranted.

We found other possible candidate gene variants for CSCR which, however, were not eligible for meta-analysis due to single report (online supplemental table 1). SNPs in ADAMTS9 9 and CDH5 11 were reported in candidate gene studies. SNPs in VIPR2,23 EMX2OS,21 SLC7A5 21 and near GATA5 22 were discovered by GWAS. Moreover, rare variants based on family studies and CNVs have been reported for CSCR. For example, a SNP rs61758735 in PTPRB (c.4145C>T; p.T1382I) segregated in CSCR families.16 Higher copy number of the C4B gene was found in patients with CSCR than in healthy controls.28 Some of these genes are involved in pathways related to complement (eg, CFH); immune system (eg, C4B) and choroidal thickness (eg, VIPR2). Replication of these SNPs (online supplemental table 1) will provide further evidence for evaluating these genes in CSCR.

Limitations

This study highlighted some limitations in the current literature of CSCR genetics. First, some of the SNPs were only reported by two studies, thus the meta-analysis results can be affected by publication bias, which however, cannot be assessed by sensitivity analysis and funnel plot. Likely due to the limited numbers of reports, some SNPs showed marginal p values, such as CFH rs1329428 (p=0.012) and CDH5 rs7499886 (p=0.081). Second, subgroup analysis by disease subtypes (acute vs chronic CSCR) and ethnic groups was not feasible due to the insufficient information provided. Finally, only a few downstream studies have been conducted on CSCR. Functional mechanisms, microbiome, epigenetics and genetic-environmental interactions are future research directions that can lead to better understanding of the roles of the genes in CSCR.

In summary, this meta-analysis confirmed significant associations of six SNPs in three genes with CSCR, including ARMS2 rs10490924, CFH rs800292, rs1061170, rs1329428 and rs2284664, and TNFRSF10A rs13278062, and revealed a number of candidate gene variants awaiting validation. Moreover, we noted that TNFRSF10A rs13278062 conferred similar risk effects for CSCR, nAMD and PCV, whereas the SNPs in ARMS2 and CFH conferred opposite effects on CSCR compared with nAMD and PCV. Thus, further differentiating the genetic architectures of CSCR, nAMD and PCV, detailed genotype–phenotype correlation analysis, and unravelling the underlying functional implications should be warranted to shed new light on the phenotypic expressions of the three maculopathies.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information. All the data included in our study are from published studies which can be searched in PubMed, Embase and Web of science.

Ethics statements

Patient consent for publication

References

Footnotes

  • ZJC and SYL contributed equally.

  • Contributors ZC and SL performed the literature searching, data analysis, data interpretation and manuscript drafting. SSR and MH conducted literature checking, data quality control, data interpretation and revised the manuscript. DS-CN, HC, BG, JCSY, ALY and MB interpreted data and revised the manuscript critically for important intellectual content. LJC, CCT and C-PP proposed the idea, designed the study and manuscript structure and revised it. All authors agreed to be accountable for all aspects of the work. All authors approved the submitted version.

  • Funding This work was supported in part by a research grant from the Department of Science and Technology of Sichuan Province (2020ZYD035; BG); a Direct Grant from The Chinese University of Hong Kong (4054560; LJC), and the Endowment Fund for Lim Por-Yen Eye Genetics Research Centre, Hong Kong.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Linked Articles

  • Highlights from this issue
    Frank Larkin