Role of ARMS2/HTRA1 risk alleles in the pathogenesis of neovascular age-related macular degeneration

Abstract

Age-related macular degeneration (AMD) is one of the leading causes of severe irreversible blindness worldwide in the elderly population. AMD is a multifactorial disease mainly caused by advanced age, environmental factors, and genetic variations. Genome-wide association studies (GWAS) have strongly supported the link between ARMS2/HTRA1 locus on chromosome 10q26 and AMD development, encompassing multiple variants, rs10490924 (c.205G > T, p.A69S in ARMS2), insertion/deletion (del443/ins54 in ARMS2), and rs11200638 (in HTRA1 promoter region). In this comprehensive review, we provide an overview of the role played by ARMS2/HTRA1 risk alleles in neovascular AMD pathogenesis, covering GWAS, in vitro studies, and animal models, shedding light on their underlying molecular genetic mechanisms. Further extensive research is also imperative, including confirmation of these findings, identifying novel treatment targets, and advancing primary and secondary prevention strategies for AMD.

Introduction

Age-related macular degeneration (AMD) is the leading cause of irreversible vision loss among individuals aged 55 years and older in the developed world.[1] It accounts for 6%–9% of legal blindness cases worldwide and is projected to affect 288 million individuals by 2024.[2] The clinical classification systems for early-stage AMD typically categorize it based on drusen size, with medium-size drusen classified as early AMD and large drusen as intermediate AMD. The risk of progressing to late-stage AMD, such as neovascular AMD (nAMD) and/or any geographic atrophy (GA), is highest when large drusen are present along with pigmentary changes. GA, also known as dry AMD, is characterized by confluent atrophy of photoreceptors, retinal pigment epithelial (RPE), and choriocapillaris, leading to scotoma.[3] In AMD at any stage, new vessels may invade the outer retina, subretinal space, or sub-RPE space, giving rise to macular neovascularization (MNV), the hallmark lesion of nAMD.[4] The exudative stage of nAMD (also called exudative or wet AMD) becomes evident when these new vessels leak or rupture, causing fluid accumulation and/or hemorrhages and vision distortion and deterioration. Without treatment, exudative MNV typically results in extensive fibrosis with severe central vision loss.

In 2014, a meta-analysis of pooled global population-based studies involving individuals aged 45 to 85 years estimated the prevalence of AMD as 8.69%.[2] Notably, this prevalence varies among racial and ethnic groups, with 12.3% in people of European descent, 10.4% in those of Latin American heritage, 7.5% in individuals of African descent, and 7.4% in people of Asian ancestry. These differences are predicted to arise from environmental or genetic factors. A reanalysis of the Multi-Ethnic Study of Atherosclerosis data, considering common factors such as smoking, body mass index, inflammatory factors, diabetes, and alcohol, failed to explain these variations.[5]

To gain a better genetic understanding across populations, Fritsche et al. compared the minor allele frequency (MAF) of the known AMD risk variants among ancestry groups in the 1000 Genomes Project.[6] They found significantly higher MAFs for the ARMS2/HTRA1 locus in Asian populations (40.2%) compared to Western populations (19.9%), although the CFH variant rs10737680 showed similar MAFs (European: 59.1% and East Asian: 57.9%).[6] In addition, in people of European descent, the ratio of nAMD to GA is approximately 1:1 across all age groups. In contrast, a meta-analysis of data from ten different Asian countries found a 3:1 ratio of nAMD to GA.[7] These findings suggest that the ARMS2/HTRA1 risk variant may predispose individuals toward nAMD, whereas CFH variants might be stronger risk factors for GA. However, the underlying biological mechanism in Asian populations remains elusive, despite the greater significance of the genetic risk in ARMS2/HTRA1 locus. Understanding the molecular mechanism of this genetic risk is crucial for elucidating the etiology of AMD. In this review, we explore the literature on several key aspects: (1) the role of the ARMS2/HTRA1 locus in Asian populations, (2) animal models related to ARMS2, (3) animal models related to HTRA1, (4) enhancers binding to ARMS2 and its effect on HTRA1 transcription, and (5) HTRA1 expression in human samples.

Genetic Dissection of ARMS2/HTRA1

The initial evidence of genetic involvement in AMD susceptibility came from familial aggregation and twin studies.[8] The heritability of late AMD is estimated to be as high as 71%, greater than that of most complex age-related diseases.[1] In 2005, several independent genome-wide association studies (GWAS) identified two major loci associated with an increased risk of AMD in White populations: CFH (rs1061170, p.Y402H) on chromosome 1q31.3[9,10,11,12] and ARMS2/HTRA1 on chromosome 10q26.[13,14] However, ethnic differences have been observed in the prevalence of nAMD and GA.[15] In Asian populations, nAMD is significantly more common than GA,[7] which differs from White populations where the prevalence is relatively equal.[15]

To gain a better genetic understanding of the Asian population, in 2006 and 2007, our group validated the association between the CFH loci and nAMD[16] as well as ARMS2/HTRA1 loci and nAMD[17] in the Japanese population using temperature gradient capillary electrophoresis and direct sequencing, respectively. Interestingly, we found that the frequencies of each haplotype in the CFH gene were significantly different from those in published White populations. Moreover, the allelic association test for rs11200638 (in the HTRA1 promoter region) yielded P < 10⁻¹¹. Single-nucleotide polymorphism (SNP) rs11200638 conferred disease risk in an autosomal recessive fashion, with an odds ratio of 10.1 (95% confidence interval: 4.36, 23.06), significantly higher than that reported in previous papers. In 2009, our group conducted the first GWAS of Japanese patients with nAMD, classified as 5b by Seddon et al.,[18] using the Affymetrix GeneChip Human Mapping 500K Array and TaqMan assay.[19] Japanese nAMD patients showed a stronger association with ARMS2/HTRA1 locus than CFH locus, in contrast to individuals with late AMD in White populations. In addition, lower associations of rs1061170 and other variations in the CFH gene were reported in Japanese,[20] Chinese,[21,22,23] and Korean populations,[24] suggesting that, in Asian populations, the ARMS2/HTRA1 locus plays a more significant role compared to the CFH locus.

To date, numerous genetic association studies have been conducted. In the current largest GWAS of AMD, carried out by the International AMD Genomics Consortium, 34 independent loci were identified to be associated with late AMD.[25] These 52 variants have been estimated to explain almost 50% of the variability of AMD in Europeans, with the ARMS2/HTRA1 locus alone accounting for approximately 30% of this genetic variance. Notably, the most robust genetic signal for AMD, with odds ratios of 2.81, was discovered in the ARMS2/HTRA1 locus.[25] Consequently, the ARMS2/HTRA1 locus stands out as one of the most strongly associated loci for AMD, suggesting that either ARMS2, HTRA1, or both may be the causal genes for AMD risk.

It is worth mentioning that the ARMS2 and HTRA1 genes are separated by approximately 4 Kbp adjacent to each other on chromosome 10q26 in the same linkage disequilibrium (LD) block.[26] Although several sophisticated fine-mapping approaches, based on Bayesian models, are currently employed to identify the most likely causal variants, prioritizing a single-credible SNP over highly correlated SNPs with absolute LD remains a challenge.[27] Thus, GWAS has not been able to definitively determine which variant (s) in ARMS2/HTRA1 locus are responsible for AMD risk. Furthermore, GWAS do not specify whether one or both genes are affected by the causal variant. Consequently, the functional implications of these two genes will be explored in the next section using animal models.

ARMS2 Animal Models

ARMS2 is a primate-specific gene that encodes an 11.4-kDa secreted protein.[28,29] Its localization and function have been disputed since its discovery.[28,29] Initially, it was believed to confer AMD risk, potentially through mitochondrial-related pathways.[30] However, later attempts to replicate its mitochondrial localization were unsuccessful.[31] In 2010, Kortvely et al. discovered that the endogenous ARMS2 protein is primarily localized to the choroid pillars in human eyes.[32] This localization suggests that it is a component of the extracellular matrix, corresponding to the principal sites of drusen formation. Given the pivotal role of the extracellular matrix in AMD’s progression, which often involves abnormal deposition of extracellular debris in the macula, it was hypothesized that ARMS2 protein may be necessary for proper matrix function.

To determine the function of the ARMS2 protein in the choroid and retina, in 2014, our group generated ARMS2 transgenic (Tg) mice using the CMV early enhancer/chicken beta actin (CAG), resulting in the ubiquitous expression of human ARMS2 protein in mice.[33] Furthermore, numerous studies have consistently demonstrated that a conserved primate-specific missense polymorphism, rs10490924 (p.A69S), affects the ARMS2 coding sequence and contributes to over 50% of AMD susceptibility.[13] The A69S alteration was suggested to potentially impact its function within the mitochondria or trigger RPE apoptosis through some toxic effects of ARMS2 (A69S).[30] Consequently, we also created ARMS2 (p.A69S) Tg mice using the same promoter. Nevertheless, histopathological examination and immunohistochemistry analysis of aged ARMS2 Tg and ARMS2 (p.A69S) Tg mice revealed no retinal or choroid abnormalities compared to age-matched controls.

Moreover, smoking stands out as the most potent modifiable risk factor for AMD development.[26] Given the intricate nature of AMD pathogenesis, our research group also combined direct inhalation of cigarette smoke with advanced age in these Tg mouse models, aiming to establish a link between the ARMS2 protein and AMD risk.[33] Similarly, none of the aged ARMS2 and ARMS2 (p.A69S) Tg mice exposed to mainstream cigarette smoke displayed any abnormal retinal changes or subretinal deposits. In 2015, Liu et al. developed an alternative Cre-conditional human ARMS2 Tg mouse model using the cytomegalovirus (CMV) promoter. Their findings align with our results, supporting the conclusion that ubiquitous overexpression of human ARMS2 protein does not lead to a typical AMD phenotype in aged mice.[34]

Furthermore, a common protective haplotype arises due to a premature stop codon (rs2736911, p.R38X) in the ARMS2 gene, leading to the termination of ARMS2 protein synthesis. This means that homozygous carriers lack the ARMS2 protein. However, this variant is not associated with AMD.[35] A recently published paper also mentioned that ARMS2 expression cannot be detected in human ocular samples by RNA sequencing, real-time quantitative polymerase chain reaction (RT-qPCR), or Western blotting.[36] Therefore, the function of ARMS2 risk variants might not be strictly associated with ARMS2 protein, including its transcription and presumed function.

Notably, ARMS2 is adjacent to the HTRA1 gene and is in LD with the HTRA1 gene.[26] Thus, its function might be closely associated with the HTRA1 protein rather than the ARMS2 protein. In these three published Tg mouse models, the ARMS2 transgene was inserted randomly in the mouse genome, and its interactions with HTRA1 could not be examined. The appropriate method for investigating the function of the ARMS2 gene in mice could entail creating a knock-in model to introduce either the human ARMS2 gene or its mutant into the corresponding HtrA1 promoter region.

HTRA1 Mouse Models

The serine protease HTRA1, among the most extensively studied members within the mammalian HTRA family (HTRA1-4), is a 50-kDa secreted protein.[26] This protein consists of an N-terminal signal peptide, followed by an insulin/insulin-like growth factor-binding protein domain, a Kazal inhibitor domain (Kaz), a trypsin-like serine protease domain, and one postsynaptic density protein 95, Drosophila disc large tumor suppressor and zonula occludens-1 domain.[37] Convincing evidence has supported the pivotal role of HTRA1 in several important biological processes and intricate cellular singling pathways, including cleaving extracellular matrix proteins, regulating signaling pathways (transforming growth factor-beta [TGF-β], IGF, and TSP1-CD47), and controlling mammalian target of rapamycin signaling by tuberous sclerosis 2 degradation. Dysregulation of HTRA1 has been implicated in the pathogenesis of several diseases, including cancer, arthritis, musculoskeletal disorder, cerebral small vessel disease, and AMD.[26,38,39]

Htra1 knockout mouse model

In 2005, the first study of mouse Htra1 knockout (KO) model was published, in which Htra1 exon 1 was replaced with β-galactosidase cDNA. This study aimed to investigate the role of HTRA1 in the pathogenesis of arthritis.[40] To assess the impact of HTRA1 deficiency in the retina, our research group crossed this Htra1 KO mouse model with C57BL/6J mice line. Subsequently, we conducted fundus imaging, hematoxylin and eosin (H and E) staining, and electroretinogram measurements, including rod and mixed responses and oscillatory potentials. Interestingly, no significant physiological or histological effects of Htra1 deficiency were observed in the mouse retina at 12 months. These findings were presented at the 2008 Association for Research in Vision and Ophthalmology annual meeting.[41]

Consistent with our results, a study conducted by Jones et al. in 2011 also examined Htra1 KO mice and found that these mice did not exhibit any features of polypoidal choroidal vasculopathy (PCV) or choroidal neovascularization (CNV), as determined by H and E staining, fluorescein angiography, indocyanine green angiography (ICGA), or spectral-domain optical coherence tomography (OCT).[42,43] Furthermore, the following year, Zhang et al. observed reduced vascular development and heightened small of mothers against decapentaplegic (SMAD) signaling in the retinas of Htra1 KO mice, suggesting a critical role for HTRA1 in regulating angiogenesis through the TGF-β signaling pathway.[43]

HTRA1 transgenic models

To further investigate the impact of HTRA1 overexpression in the retina, we developed Htra1 Tg mice using the CAG promoter for ubiquitous mouse Htra1 overexpression.[33] The Htra1 levels were 3.0-fold higher in the retinas of 12-month-old Htra1 Tg mice compared to age-matched controls. We observed fluorescein diapedesis in the fundi of Htra1 Tg mice, indicating the breakdown of the blood–retinal barrier and exacerbation of retinal vascular permeability. In addition, low fluorescence lesions were detected in the choroids of Htra1 Tg mice through ICGA, suggesting the presence of abnormally branched vessel networks. Furthermore, OCT revealed the presence of radial capillary branching that extended from the choroid, traversing the RPE and reaching into the retina. The formation of radial CNV was further explored through H and E staining and immunostaining with anti-CD-31, an endothelial cell marker. Moreover, Htra1 Tg mice exhibited ruptures and deficiencies in Bruch’s membranes (BMs), as revealed by Elastica van Gieson staining. Notably, 18.2% of Htra1 Tg mice developed CNV, whereas none of the age-matched WT mice exhibited this condition. Recently, Ahamed et al. conducted a more in-depth analysis of this same Htra1 Tg mouse model and identified signs of subclinical inflammation, which could have led to increased RPE cell activation and angiogenic potential.[44]

Simultaneously, Htra1 Tg mice were exposed to mainstream cigarette smoke, as described in ARMS2 Tg mice.[33] Surprisingly, 20% of them developed CNV by the age of 12 months. Although no CNV was observed in the control group exposed to air, equivalent damage was also observed in WT mice exposed to mainstream cigarette smoke. This suggests that exposure to mainstream cigarette smoke triggered CNV, and the formation of CNV was independent of the level of Htra1 expression. Due to dense deposits in the subretinal space of aged Htra1 Tg mice, electron microscopy was also performed, revealing abnormal deposits or vacuolization in the subretinal space of Htra1 Tg mice exposed to mainstream cigarette smoke.[33]

Several other research groups have also studied the effects of overexpressing human HTRA1 or murine Htra1 on ocular phenotypes using mouse or zebrafish models. They employed different promoters to explore the association between HTRA1 overexpression and AMD risk. In 2011, an investigation of human HTRA1 overexpression in mouse RPE (Tg 73), driven by the CMV-IE enhancer/promoter and VMD2 promoter, revealed key features of PCV in 59% of the 114 HTRA1 Tg 73 mice, beginning at 3 weeks, and developed occult CNV in 50% of PCV⁺ mice older than 11 months.[42] Simultaneously, another group compared RPE and BM defects in their Htra1 Tg mouse line, which overexpresses murine Htra1 specifically in RPE using the Rpe65 promoter.[45] They found altered elastogenesis in the BM through fibulin-5 cleavage. In 2014, human HTRA1 Tg mice (Tg44), driven solely by the VMD2 promoter (expressing HTRA1 in RPE), exhibited a more consistent range of human PCV-like features; 94.7% of Tg 44 mice developed PCV-like features at 4–5 weeks old.[46] To investigate whether the proteolytic activity of HTRA1 was crucial for PCV development, they generated two other Tg lines (Tg33 and Tg 26) expressing the protease-inactive mutant human HTRA1 (S328A) in mouse RPE using the human VMD2 promoter.[47] Neither line exhibited any PCV phenotype. Similarly, intravitreal delivery of an HTRA1 inhibitor targeting the catalytic site of HTRA1 effectively prevented PCV initiation.[47] Furthermore, Tg zebrafish that overexpressed human HTRA1 in rod photoreceptors exhibited photoreceptor death and lipofuscin accumulation, similar to early AMD findings.[48] In summary, all of these studies involving HTRA1/Htra1 animal models suggest that HTRA1 plays a critical role in AMD pathogenesis.

The Impact of ARMS2/HTRA1 Variants on HTRA1 Transcription

In ARMS2/HTRA1 locus, seven SNPs have been identified within a 17-kilobase region on chromosome 10q26.[49] These SNPs, including rs10490924, rs3750846, insertion/deletion (del443ins54), rs11200638, rs1049331, and rs2293870, have been widely studied due to their strong associations with the risk of AMD. Notably, the most prominently associated haplotype within the ARMS2/HTRA1 locus comprises three variants: rs10490924 (c. 205G > T, resulting in p.A69S in ARMS2), the in/del mutation (consisting of a 443-base pair deletion and an adjacent 54-base pair insertion in the 3’-untranslated region of ARMS2 and its flanking region), and rs11200638 (located within the HTRA1 promoter region).[26] It is worth highlighting that these three highly associated variants are found within the same LD block. Consequently, despite large-scale GWAS, it has proven challenging to definitively pinpoint which specific variant within this locus is responsible for the observed AMD association. As a result, researchers have turned their attention to elucidating the molecular mechanisms underlying how these genetic variants may influence the regulation of HTRA1 expression. Over the past decade, multiple research groups have reported conflicting findings concerning the functional roles of HTRA1 rs11200638 and ARMS2 in/del mutations in relation to the transcriptional regulation of HTRA1.

Age-related Macular Degeneration Risk Variant rs11200638 and HTRA1

In 2006, the homozygous risk genotype AA at rs1200638 was initially found to be strongly associated with higher levels of HTRA1 mRNA and protein in blood lymphocytes and RPE cells from four AMD patients compared to three controls without the risk variant.[50] This correlation was subsequently confirmed in archived eye tissues and fresh placenta tissues through RT-qPCR and immunohistochemistry.[51,52,53,54] It was hypothesized that this correlation could be attributed to the disruption of a CpG island, which removed inhibitory effects on the methylated CpG island, leading to altered transcription of HTRA1 or increased responsiveness of the HTRA1 gene to microenvironmental factors.[55,56] However, subsequent studies have yielded conflicting results, with several investigations suggesting that the SNP rs11200638 is not related to HTRA1 transcription in the retina or lymphocytes.[30,52,57]

In vitro studies have also produced varying results. Dewan et al. observed that the risk homozygote (AA) genotype led to a persistent trend of higher luciferase expression than the GG genotype in ARPE-19 and HeLa S3 cells.[58] However, the following year, Kanda et al. found no significant difference in luciferase activity associated with rs1200638 in HEK 293, ARPE-19, and Y79 cells.[30] Recently, He et al. discovered that the risk SNP rs1200638 significantly increased the responsiveness of the HTRA1 promoter to oxidative stress and inflammatory responses, resulting in elevated HTRA1 expression in ARPE-19 cells.[59] These discordant results regarding the effect of rs1200638 on HTRA1 transcription may be attributed to cell type-dependent effects in vitro. In a recent study, Chang et al. isolated the effect of rs1200638 using CRISPR manipulation of induced pluripotent stem cell (iPSC)-RPE cells and found that the risk SNP rs1200638 significantly increased HTRA1 transcription and expression, as confirmed by RT-qPCR and Western blot (WB) analysis, respectively.[60]

Age-related Macular Degeneration Risk Variant ARMS2 in/del

Researchers have also explored the impact of the in/del mutation on HTRA1 expression. In 2010, Yang et al. observed that the haplotype tagged by the in/del mutation led to an upregulation of HTRA1 transcription in human and mouse RPE cells using luciferase constructs.[54] However, in 2011, Friedrich et al. failed to detect this upregulation in ARPE-19 and rMC-1 cells.[52] To validate the regulatory effect, in 2015, our group performed transcription regulator activity assays using luciferase constructs containing the −4320 to +1 base pair region of the non-in/del regulatory region and the −3936 to +1 base pair region of the in/del regulatory region.[61] The transcription was significantly upregulated with in/del variant in both 661W and RGC5 cells but not in ARPE-19 cells, indicating cell type-dependent effects on HTRA1 transcriptional regulation. In 2018, Oura et al. replicated this upregulation in Y79 human retinoblastoma cells.[48]

To identify the activator that binds to the in/del region, we performed gel electrophoresis mobility shift assays and liquid chromatography-mass spectrometry using nuclear extracts from 661W cells. Among the listed in/del interacting proteins, general transcription factor IIi (Gtf2i) was identified with most number of peptide hits and was the sole transcription factor protein. In addition, Gtf2i commonly targets the c-fos cis-element-binding nucleotide sequence, which was found within the in/del sequence.[62] We further validated that Gtf2i bound to the in/del region can enhance HTRA1 expression through WB and enzyme-linked immunosorbent assay in transfected 661W cell lines.[62] This discovery provides new insights into the molecular mechanism of HTRA1 transcription regulated by the highly AMD-associated in/del haplotype.

HTRA1 Expression in Human Samples

Induced pluripotent stem cells derived from age-related macular degeneration patients

Since 2007, human iPSCs have rapidly become a valuable investigative tool for gaining insights into the pathology and cellular biology of retinal diseases.[63] In 2015, to measure the HTRA1 expression levels in AMD patients, our group generated human iPSCs, revealing a significant increase of HTRA1 mRNA in both heterozygous and homozygous in/del-AMD iPSCs compared to nonrisk controls.[61] Three years later, Lin et al. observed a similar elevation of HTRA1 expression in iPSC-RPE cells from heterozygous in/del-AMD compared to controls.[64] Recently, Chang et al. also replicated the increase of HTRA1 expression in iPSC-RPE cells derived from AMD patients carrying the ARMS2/HTRA1 risk variants. Furthermore, they isolated the effects of the individual SNV using CRISPR and found that both ARMS2 rs10490924 and HTRA1 rs11200638 independently elevated HTRA1 expression.[60]

HTRA1 in serum/plasma or aqueous humor samples

Since HTRA1 is known as a widely secreted serine protease, we investigated the relationship between blood circulation HTRA1 levels in plasma/serum and AMD. We compared 478 AMD patients to 481 controls from diverse geographical regions, including Japan, India, Australia, and the USA. Our findings revealed a statistically significant increase of HTRA1 in serum or plasma in AMD cases compared with controls.[62] Notably, we observed a consistent rise in HTRA1 levels with age in controls but not in AMD cases.[62] Recently, this trend was further verified in human retina samples.[26] In addition, we identified that in/del variant increased the secretion of HTRA1 in both plasma and serum samples.[62] Our results align with the findings of Qureshi and Ambreen, who conducted a similar investigation using 190 serum samples from Pakistan and detected significantly elevated serum HTRA1 levels in AMD patients.[65] Furthermore, Tosi et al. measured HTRA1 concentrations in the aqueous humor of nAMD patients and found that it was significantly higher than that in controls.[66] These collective observations strengthen the association between circulating HTRA1 levels and AMD, emphasizing the potential utility of HTRA1 as a biomarker for AMD.

HTRA1 in retina tissue

To improve the accuracy of HTRA1 expression levels in association with AMD risk, several other research groups have investigated HTRA1 expression in human retinal tissues. These reports presented varying conclusions regarding HTRA1 expression in human retina tissues, which can be categorized as follows: (1) elevated HTRA1 levels in AMD retina (positive),[50,51,53,67] (2) reduced HTRA1 levels in AMD retina (negative),[30,36] and (3) no significant association.[68] These divergent findings could be attributed to the limited sample sizes used in these studies. Currently, the largest published sample size for AMD research consists of only 18 cases. Given that AMD is a multifactorial disease influenced by factors such as advanced age, environmental variables, genetic variations, and other yet-to-be-discovered factors, the current number of AMD retinal samples in published studies may not be sufficient to establish the definite relationship between HTRA1 expression level and AMD risk. Therefore, future research should be undertaken to thoroughly investigate HTRA1 expression in the retinas of AMD patients with age-matched controls.

Conclusion

AMD is a complex multifactorial disease with contributions from various risk factors. Human genetics research has illuminated the pivotal pathways involved in AMD pathogenesis. Notably, numerous recent genetic studies have underscored the importance of the ARMS2/HTRA1 locus as one of the strongest genetic factors contributing to AMD onset. The compelling association between AMD and HTRA1 expression upregulated by ARMS2/HTRA1 risk variants has progressed to developing HTRA1 inhibitors as potential drug targets. To date, clinical trials targeting HTRA1 in AMD have shown limited success, possibly due to the intricate interplay of multiple risk factors affecting AMD. Therefore, a comprehensive approach, encompassing diverse perspectives, integrating various technologies, and fostering multidisciplinary collaboration among experts, is essential for a thorough understanding of AMD mechanisms.

Ethical statement

The Institutional Review Board of NHO Tokyo Medical Center (No. R21-120) granted approval for this study.

Declaration of patient consent

The authors certify that they have obtained all appropriate patient consent forms. In the form, the patients have given their consent for their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal identity, but anonymity cannot be guaranteed.

Data availability statement

The datasets generated during and/or analyzed during the current study are publicly available.

Conflicts of interest

The authors declare that there are no conflicts of interests of this paper.

Funding Statement

Nil