Genome-wide Survival Analysis for Macular Neovascularization Development in Central Serous Chorioretinopathy Revealed Shared Genetic Susceptibility with Polypoidal Choroidal Vasculopathy
概要
Genome-wide Survival Analysis for Macular
Neovascularization Development in Central
Serous Chorioretinopathy Revealed Shared
Genetic Susceptibility with Polypoidal
Choroidal Vasculopathy
Yuki Mori, MD,1,2 Masahiro Miyake, MD, PhD,1,2 Yoshikatsu Hosoda, MD, PhD,3 Akiko Miki, MD, PhD,4
Ayako Takahashi, MD, PhD,1 Yuki Muraoka, MD, PhD,1 Manabu Miyata, MD, PhD,1 Takehiro Sato, PhD,5
Hiroshi Tamura, MD, PhD,1 Sotaro Ooto, MD, PhD,1 Ryo Yamada, MD, PhD,2 Kenji Yamashiro, MD, PhD,6
Makoto Nakamura, MD, PhD,4 Atsushi Tajima, PhD,5 Masao Nagasaki, PhD,2 Shigeru Honda, MD, PhD,7
Akitaka Tsujikawa, MD, PhD1
Purpose: To identify susceptibility genes for macular neovascularization (MNV) development in central serous chorioretinopathy (CSC).
Design: Genome-wide survival analysis using a longitudinal cohort study.
Participants: We included 402 and 137 patients with CSC but without MNV at their first visit from the Kyoto
CSC Cohort and Kobe CSC dataset, respectively. All patients underwent detailed ophthalmic examinations,
including multimodal imaging, such as fundus autofluorescence, spectral-domain OCT, and fluorescein angiography/indocyanine green angiography or OCT angiography.
Methods: We conducted a genome-wide survival analysis using the Kyoto CSC Cohort. We applied the Cox
proportional hazard model to adjust for age, sex, and the first principal component. Single nucleotide polymorphisms (SNPs) with P values < 1.0 10-5 were carried forward to the replication in the Kobe CSC dataset.
Moreover, we evaluated the contribution of previously reported age-related macular degeneration (AMD) susceptibility loci. We used FUMA and ToppFun for the functional enrichment analysis.
Main Outcome Measures: The association between SNPs and MNV development in patients with CSC.
Results: Rs370974631 near ARMS2 displayed a genome-wide significant association in the meta-analysis of
discovery and replication result (hazard ratio [HR]meta, 3.63; Pmeta ¼ 5.76 10-9). Among previously reported AMD
susceptibility loci, we additionally identified CFH rs800292 (HR, 0.39, P ¼ 2.55 10-4), COL4A3 rs4276018 (HR,
0.26, P ¼ 1.56 10-3), and B3GALTL rs9564692 (HR, 0.56, P ¼ 8.30 10-3) as susceptibility loci for MNV
development in CSC. The functional enrichment analysis revealed significant enrichment of 8 pathways
(GO:0051561, GO:0036444, GO:0008282, GO:1990246, GO:0015272, GO:0030955, GO:0031420, and
GO:0005242) related to ion transport.
Conclusions: ARMS2, CFH, COL4A3, and B3GALTL were identified as susceptibility genes for MNV
development in CSC. These 4 genes are known as susceptibility genes for AMD, whereas COL4A3 and B3GALTL
were previously reported to be polypoidal choroidal vasculopathy (PCV)-specific susceptibility genes. Our findings revealed the shared genetic susceptibility between PCV and MNV secondary to
CSC. Ophthalmology 2022;129:1034-1042 ª 2022 by the American Academy of Ophthalmology
Supplemental material available at www.aaojournal.org.
Central serous chorioretinopathy (CSC) is characterized by
serous retinal detachment around the fovea and is one of the
common causes of vision loss, principally in young to
middle-aged men.1 Central serous chorioretinopathy
consists of 2 clinically distinct entities, namely, acute CSC
and chronic CSC.2 Acute CSC resolves spontaneously
within 3 months such that its visual prognosis is generally
good. However, a recent long-term observational study by
1034
ª 2022 by the American Academy of Ophthalmology
Published by Elsevier Inc.
Mrejen et al3 reported that chronic CSC lasting > 6 months
resulted in legal blindness in 12.8% of the patients with an
average 11.3 years of follow-up. In this report, 24% of the
patients developed macular neovascularization (MNV),
which was strongly associated with visual acuity at their last
visit. Therefore, MNV is considered a major cause of poor
visual prognosis in patients with CSC. The prevalence of
MNV secondary to CSC ranges from 2% to 28.7%.4-7
https://doi.org/10.1016/j.ophtha.2022.04.018
ISSN 0161-6420/22
Mori et al
Susceptibility Genes for MNV in CSC
Researchers have reported on several susceptibility genes
for CSC. CFH was the first reported CSC susceptibility gene,
which is also an established susceptibility gene for agerelated macular degeneration (AMD); however, the effect
direction is opposite between CSC and AMD.8 We have also
identified additional susceptibility genes for CSC through 2
genome-wide association studies (GWASs): (1) an intermediate phenotype 2-staged GWAS of 6110 healthy Japanese
individuals, which led to the identification of VIPR2,9 and (2)
a 2-staged GWAS of 1546 CSC samples and 13 029 controls
from Asian and White participants that identified
TNFRSF10A and GATA5.10 Despite reports on these
susceptibility genes for CSC development, researchers have
not yet explored a detailed phenotype-genotype correlation.
Among a variety of clinical phenotypes for CSC, MNV
development is most important. Macular neovascularization
secondary to CSC is paid more attention after being located
within the novel disease concept, pachychoroid spectrum
diseases, and named pachychoroid neovasculopathy
(PNV).11 Despite speculating that the mechanical damage of
choriocapillaris and Sattler’s layer due to dilated choroidal
vessels may cause MNV,12 its molecular biological
susceptibility is unknown. We aimed to conduct a genomewide survival analysis for MNV development in patients
with CSC, followed by replication in an independent dataset.
Methods
This study followed the tenets of the Declaration of Helsinki and
was approved by the ethics committee of the Kyoto University
Graduate School of Medicine and Kobe University Graduate
School of Medicine. Written informed consent was obtained from
all participants.
Discovery Dataset
We used the Kyoto CSC Cohort10 for the discovery stage. It consists
of 610 unrelated patients with CSC who visited the Macular Service
of the Kyoto University Hospital (Kyoto, Japan) and were recruited
for genetic studies on disease development and clinical course with
written informed consent. All patients underwent detailed
ophthalmic examinations, including a measurement of the bestcorrected visual acuity (BCVA), intraocular pressure, slit-lamp
biomicroscopy, color fundus photography, infrared reflectance,
fundus autofluorescence, fluorescein angiography (FA), indocyanine
green angiography (ICGA), and spectral-domain OCT, with or
without OCT angiography. Central serous chorioretinopathy was
diagnosed as subretinal fluid and choroidal thickening in the macula
on OCT with leakage on FA regardless of choroidal hyperpermeability on ICGA. Two retina specialists (M.M. and Y.H.)
independently diagnosed the patients. The discrepancies were
resolved by a face-to-face discussion. We excluded patients with
MNV or characteristics suggestive of AMD, such as confluent soft
drusen, pseudodrusen, or choroidal thinning. Moreover, we excluded
those without follow-up visits and any visits after 2013.
Replication Dataset
Central serous chorioretinopathy was defined as eyes with central
serous retinal detachment without subretinal hemorrhage or suspected MNV in ICGA or OCT. The exclusion criteria were as
follows: (1) a history of corticosteroid treatment; (2) central
choroidal thickness < 250 mm; (3) age > 80 years; and (4) a
history of retinal vessel occlusion or uveitis.
Genotyping
The Kyoto CSC Cohort was genotyped using a series of BeadChip
DNA arrays (Illumina), namely, OmniExpress (N ¼ 250) and
Asian Screening Array (N ¼ 360). We performed genotype
imputation using the Michigan imputation server (https://imputationserver.sph.umich.edu/index.html#!pages/home) and used the
East Asian population of the 1000 Genomes dataset (phase III,
version 5 release) as a reference for each dataset. In each dataset,
single nucleotide polymorphisms (SNPs) with a call rate < 90% or
a minor allele frequency (MAF) < 1% were excluded before the
genotype imputation. After the imputation, we also excluded the
following imputed SNPs from the subsequent association analysis:
those with R2 < 0.9, a call rate < 90%, MAF < 1%, or significant
deviation (P < 1.0 10-5) from the HardyeWeinberg equilibrium.
The SNPs included in both datasets after imputation and quality
control were used for the analysis.
The Kobe CSC dataset was genotyped using the Human
OmniExpress BeadChips (Illumina).13 We performed genomic
imputation using the BEAGLE 4.1 with the 1000 Genomes
dataset (phase III, version 5 release) as reference panels. Imputed
SNPs with R2 < 0.7 were excluded. In the current study, the
coordinates and variant identifiers were based on GRCh37/hg19.
Genome-wide Survival Analysis
We reviewed the medical records to obtain phenotypic information
of all participants. The duration from CSC diagnosis to the event
(MNV development) or censoring (final visit in August 2019 or
lost to follow-up) was recorded daily. The age at the first visit was
used for statistical analysis.
We conducted a 2-staged genome-wide survival analysis. In the
discovery stage, using the Kyoto CSC Cohort, we conducted a
survival analysis on the MNV-free period for each SNP with an
MAF 10% using the Cox proportional hazard regression model
to adjust for the age at first visit, sex, and the first principal
component. We carried SNPs with a suggestive association (P <
1.0 10-5) forward to the replication stage. In the replication stage,
using the Kobe CSC dataset, we performed a similar analysis
adjusted for the age at first visit and sex. Eventually, both Cox
regression analysis results were meta-analyzed using the METAL
software14 (http://www.sph.umich.edu/csg/abecasis/Metal/), with
the inverse variance-based method and the fixed effect model.
Association of AMD Susceptibility Genes
We investigated the contribution of AMD susceptibility genes to
MNV development in CSC. ...