Association of Genetic Variants on 8p21 and 4q12 with Age-Related Macular Degeneration in Asian Populations

Abstract

Purpose.: To evaluate the association of genetic variants at chromosomes 8p21 and 4q12 with the risk of developing AMD and its two main subtypes, choroidal neovascular membrane (CNV) and polypoidal choroidal vasculopathy (PCV), in Asian populations.

Methods.: The study population comprised 2360 patients with neovascular AMD (1013 typical AMD-CNV and 1282 PCV), and 3598 controls from four independent cohorts, two of Japanese (n = 4859) and two of Chinese (n = 1099) ethnicity. We performed a meta-analysis in case-control studies of two reported single nucleotide polymorphisms (SNPs) (rs13278062 at TNFRSF10A-LOC389641 on 8p21 and rs1713985 at REST-C4orf14-POLR2B-IGFBP7 on 4q12) by using logistic regression analysis adjusted for age and sex. Subgroup analysis by CNV and PCV subtypes were performed to evaluate the significance of these two variants.

Results.: The reported association between rs13278062 at 8p21 and neovascular AMD was replicated in this population (P = 1.12 × 10⁻⁴, odds ratio [OR] = 0.79, 95% confidence interval [CI] = 0.70–0.89). However, there was no association of rs1713985 at 4q12 with neovascular AMD, or its two subtypes, typical AMD-CNV and PCV (all P > 0.05). The study sample size had a statistical power of greater than 99% to detect an association of a risk allele with AMD with an OR of 1.30, as reported in the original study of rs1713985 and AMD.

Conclusions.: The present results did not replicate the reported association between rs1713985 at 4q12 and neovascular AMD. However, we confirmed the association between rs13278062 at 8p21 and neovascular AMD in Asian populations.

Introduction

AMD is the leading cause of visual impairment in the elderly and the most common cause of blindness in developed countries. ¹ Several genes have been reported to be associated with this disease, ^2–10 and these genetic studies have helped to reveal the mechanisms underlying the development of AMD, specifically suggesting an inflammation-based model of AMD pathogenesis. ^11–13  

There is increasing recognition that the prevalence of AMD in Asian populations is as high as in Caucasian populations. ¹⁴ However, it has been suggested that the frequency of AMD subtypes may differ between the Asians and Caucasians. ^15,16 In Asian populations, neovascular AMD (“wet” AMD) is the major type of late AMD, and is associated with choroidal neovascular membrane (CNV), collectively referred to as AMD-CNV in this study. However, recent studies revealed that almost all reported genetic risk factors for developing AMD in Asians are identical to those for developing Caucasian AMD. ^17–20  

Polypoidal choroidal vasculopathy (PCV) is the other subtype of neovascular AMD, and it is usually diagnosed clinically by indocyanine green angiography. ²¹ The prevalence of PCV in Asian populations with neovascular AMD has been reported to be higher than in Caucasians. ^22–24 Previous studies have revealed several genes that are linked to the development of PCV. ^17–20,25 However, because PCV has many similarities with AMD, including genetic factors, demography, pathology, and clinical manifestation, it remains controversial whether PCV is a subtype of AMD or a distinct clinical entity. ^16,22,24,26  

Recently, Arakawa et al. ²⁷ suggested that two new loci (rs13278062 at TNFRSF10A-LOC389641 on chromosome 8p21 and rs1713985 at REST-C4orf14-POLR2B-IGFBP7 on chromosome 4q12) conferred risk for the development of AMD-CNV in case-control studies involving a Japanese population. They noted that TNFRSF10A on 8p21 and all four genes on 4q12 were expressed in human adult retinal pigment epithelium (RPE) and rod photoreceptors in mice, which suggests that these loci have functional roles in the development of AMD. To accurately evaluate the gene–disease association, it is important to replicate the positive association reported in previous studies using the same or different ethnic cohorts. In addition, it is important to evaluate the significance of these two variants for typical AMD-CNV and PCV in a larger number of cases, because the original study by Arakawa et al. ²⁷ involved subgroup analysis of AMD with relatively small numbers of patients (298 typical AMD-CNV and 480 PCV). 

The aim of this study was to investigate whether the two suggested loci play a significant role in the development of AMD in Asians and its subtypes, typical AMD-CNV and PCV, by assessing 6000 participants from Japanese and Chinese populations. 

Methods

We conducted a case-control study with meta-analysis for 2360 patients with neovascular AMD (1013 typical AMD-CNV and 1282 PCV) and 3598 controls from four cohorts. 

All procedures in this study adhered to the tenets of the Declaration of Helsinki. This study was approved by the ethics committee of each institute involved (Kyoto University Graduate School and Faculty of Medicine, ethics committee, the ethical committee of Fukushima Medical University, the ethical committee of Kobe City Medical Center General Hospital, the ethical committee at Aichi Cancer Center, the Ethics and Gene Analysis Committee in the Faculty of Medicine, University of Yamanashi, the Singapore Eye Research Institute (SERI) institutional review board, and the Ethics Committee on Human Research, the Chinese University of Hong Kong). All the patients were fully informed about the purpose and procedures of this study, and written consent was obtained from each. 

Patients and Controls

Kyoto Cohort.

The Kyoto cohort consisted of 1364 neovascular AMD cases and 3057 general healthy population controls. Neovascular AMD patients were recruited from the Department of Ophthalmology at Kyoto University Hospital, Fukushima Medical University Hospital, and Kobe City Medical Center General Hospital. The diagnoses of AMD were based on the International Classification System for age-related maculopathy. ²⁸ Of these patients, 720 were diagnosed as PCV. As proposed by the Japanese Study Group of PCV, ²⁹ the diagnoses were based on indocyanine green angiography, which showed a branching vascular network terminating in polypoidal swelling. Typical AMD-CNV showed classic CNV, occult CNV, or mixed CNV with clear images of vascular CNV networks or diffuse staining of the CNV membrane without polypoidal lesions in indocyanine green angiography. 

Patients displaying any of the following characteristics were excluded from the study: (1) high myopia (spherical equivalent < −6.00 diopters [D]), (2) geographic atrophy or drusen only, (3) patients with one eye affected by typical choroidal neovascularization and the other with polypoidal lesions, or (4) an old lesion without a clear diagnosis. The healthy Japanese individuals were obtained from the three studies, which have been reported elsewhere: the Pharma SNP Consortium (PSC), ³⁰ the Aichi Cancer Center Research Institute (with patients confirmed not to have cancer according to the cancer registry, medical record, and self-reporting) ³¹ and the Japanese Single Nucleotide polymorphism (JSNP) database. ³² We recruited them without ophthalmic data, and they served as general population controls. All subjects in this cohort were unrelated and of Japanese ethnicity. 

Yamanashi Cohort.

The Yamanashi cohort consisted of 323 neovascular AMD cases and 115 controls. All participants were recruited from the Macular Clinic, Department of Ophthalmology, University of Yamanashi Hospital. Of the patients, 211 patients were diagnosed with PCV. All diagnoses were made as described for the Kyoto cohort. All control subjects were confirmed not to have any signs of AMD by funduscopic examination. All subjects in this cohort were unrelated and of Japanese ethnicity. 

Singapore Cohort.

The Singapore cohort consisted of 240 neovascular AMD cases and 151 controls. The AMD cases were recruited from a tertiary eye hospital, the Singapore National Eye Center, between September 2007 and April 2008. ³³ Controls comprised participants from the Singapore Chinese Eye Study ¹⁵ without any sign of AMD. ³⁴ Of the AMD patients, 118 were diagnosed with PCV. All diagnoses were based on criteria as described for the Kyoto cohort. All subjects in this cohort were unrelated and of Chinese ethnicity. 

Hong Kong Cohort.

The Hong Kong cohort consisted of 433 neovascular AMD cases and 275 controls. All participants were recruited from two tertiary ophthalmic centers in Hong Kong, the Hong Kong Eye Hospital and the Prince of Wales Hospital. Of these patients, 233 were diagnosed with PCV. All diagnoses were made as described for the Kyoto cohort. All control subjects were confirmed not to have any signs of AMD by funduscopic examination. All subjects in this cohort were unrelated and of Chinese ethnicity. 

Genotyping

We targeted rs13278062 at TNFRSF10A-LOC389641 on chromosome 8p21 and rs1713985 at REST-C4orf14-POLR2B-IGFBP7 on chromosome 4q12, which have been described to have a positive association with development of AMD in a prior study. ²⁷ In the Kyoto cohort, all case samples and PSC samples were genotyped using TaqMan SNP assays with an ABI PRISM 7700 system (Applied Biosystems, Foster City, CA). Controls from the Aichi Cancer Center Research Institute were genotyped using Illumina Human610-Quad BeadChips (Illumina Inc., San Diego, CA). In the Yamanashi cohort, all samples were genotyped using TaqMan genotyping assays with a 7300/7500 Real-Time PCR System (Applied Biosystems). In the Singapore cohort, all samples were genotyped using Illumina Human610-Quad BeadChips (Illumina Inc.). In the Hong Kong cohort, the SNPs were genotyped using TaqMan genotyping assays on an ABI Prism 7000 Sequence Detection System, according to the manufacturer's instructions (Applied Biosystems). 

Statistical Analyses

Deviations in genotype distributions from the Hardy-Weinberg equilibrium (HWE) were assessed with the HWE exact test. Statistical analyses for differences in the observed genotypic distributions were performed by the χ² test for trend; logistic regression analyses were performed for age and sex adjustments. Meta-analyses were conducted using a weighted-inverse variance approach in METAL (http://www.sph.umich.edu/csg/abecasis/metal/; provided in the public domain by Center for Statistical Genetics, The University of Michigan, Ann Arbor, MI). ³⁵ The P values of heterogeneity across the population were determined using a Cochran's Q test. The coordinates presented were from NCBI Built 36.1 (http://www.ncbi.nlm.nih.gov/; provided in the public domain by National Center for Biotechnology Information, Bethesda, MD). The statistical power calculation was performed using QUANTO version 1.2. ³⁶ P value correction was not performed because the SNPs were analyzed independently. P values of less than 0.05 were considered statistically significant. 

Results

Demographics of the four Japanese and Chinese cohorts are shown in Table 1. Overall, genotype and allele frequencies of the two reported SNPs (rs13278062 at TNFRSF10A-LOC389641 and rs1713985 at REST-C4orf14-POLR2B-IGFBP7) were analyzed in the 2360 patients with neovascular AMD, and compared with those of the 3598 controls. The genotyping of evaluated SNPs was more than 99.3% successful, and the distributions of the genotypes for all study groups were in HWE (P > 0.05). Details of allele frequencies and summary statistics for rs13278062 are shown in Table 2. The minor allele frequency (MAF) of rs13278062 was lower in Chinese populations (0.25–0.32) than in Japanese populations (0.30–0.39). After age and sex adjustments, meta-analyses of the four cohorts revealed a significant association of rs13278062 with developing neovascular AMD (P = 1.12 × 10⁻⁴, odds ratio [OR] = 0.79, 95% confidence interval [CI] = 0.70–0.89). 

Table 1. 

 

Characteristics of the Study Population

	Japanese				Chinese
	Kyoto		Yamanashi		Singapore		Hong Kong
	Cases	Controls	Cases	Controls	Cases	Controls	Cases	Controls
n	1364	3057	323	115	240	151	433	275
Age, y	74.3	45.3	74.0	72.5	70.8	65.7	71.6	74.3
SD	8.7	15.4	8.5	9.4	8.7	5.0	9.1	7.6
Range	42 to 96	20 to 79	47 to 93	45 to 91	44 to 93	60 to 84	43 to 94	60 to 94
Male, %	71.6	50.2	74.1	59.8	65.0	53.6	62.8	44.0
PCV, %	52.8	-	65.3	-	49.2	-	53.8	-

Table 2. 

 

Distribution of Genotypes and the Results of the Association Tests for TNFRSF10-LOC389641 rs13278062

	n	Genotype								Association Analysis
		Cases				Controls				Nominal		Adjusted*
		GG	GT	TT	MAF	GG	GT	TT	MAF	P Value	OR (95% CI)	P Value	OR (95% CI)
Japanese
Kyoto	4421	506	642	205	0.39	1327	1354	348	0.34	5.22 × 10−6	0.80 (0.73–0.88)	0.040	0.84 (0.71–0.99)
Yamanashi	438	114	163	46	0.39	60	41	14	0.30	0.011	0.66 (0.48–0.91)	0.012	0.65 (0.47–0.91)
Chinese
Singapore	391	111	103	26	0.32	85	56	10	0.25	0.033	0.71 (0.51–0.97)	0.051	0.72 (0.51–1.00)
Hong Kong	708	206	186	41	0.31	143	113	19	0.27	0.161	0.84 (0.67–1.07)	0.110	0.82 (0.64–1.04)
Meta-analysis
Total	5958									1.98 × 10−8	0.79 (0.73–0.86)	1.12 × 10−4	0.79 (0.70–0.89)

*  Adjusted for age and sex.

Table 3 shows details of allele frequencies and summary statistics for rs1713985. The MAF of rs1713985 was similar between Japanese and Chinese populations. We found no significant association between rs1713985 and neovascular AMD (P = 0.785, OR = 1.02, 95% CI = 0.90–1.15). Had there been a true association between rs1713985 and development of AMD at the level reported in the original study (OR = 1.30), ²⁷ our sample size had more than 99.9% power to detect it (unmatched case-control design, log-additive genotype model, 0.67% for the prevalence of exposure in general population, ³⁷ and 0.22–0.31 for allele frequency). We found no evidence of heterogeneity in these meta-analyses for rs13278062 and rs1713985 (heterogeneity P > 0.05). 

Table 3. 

 

Distribution of Genotypes and the Results of the Association Tests for REST-C4orf14-POLR2B-IGFBP7 rs1713985

	n	Genotype								Association Analysis
		Cases				Controls				Nominal		Adjusted*
		GG	GT	TT	MAF	GG	GT	TT	MAF	P Value	OR (95% CI)	P Value	OR (95% CI)
Japanese
Kyoto	4421	109	557	689	0.29	274	1204	1547	0.29	0.730	0.98 (0.89–1.09)	0.359	0.92 (0.77–1.10)
Yamanashi	438	22	147	154	0.30	7	39	69	0.23	0.058	1.40 (0.99–1.99)	0.058	1.39 (0.99–2.05)
Chinese
Singapore	391	20	102	118	0.30	4	59	88	0.22	0.023	1.47 (1.05–2.06)	0.131	1.32 (0.92–1.89)
Hong Kong	708	39	183	210	0.30	25	121	129	0.31	0.725	0.96 (0.76–1.21)	0.530	0.93 (0.73–1.18)
Meta-analysis
Total	5958									0.250	1.12 (0.92–1.37)	0.785	1.02 (0.90–1.15)

*  Adjusted for age and sex.

Next, we evaluated whether these two SNPs were associated with developing typical AMD-CNV or PCV. In this subgroup analysis (Table 4), 1013 typical AMD-CNV and 1282 PCV patients were evaluated. In the meta-analysis, TNFRSF10A-LOC389641 rs13278062 was found to be significantly associated with development of both typical AMD-CNV (P = 8.21 × 10⁻³, OR = 0.81, 95% CI = 0.70–0.95) and PCV (P = 3.79 × 10⁻⁵, OR = 0.74, 95% CI = 0.65–0.86) after age and sex adjustments, although rs1713985 did not show significant association with either typical AMD-CNV or PCV (P > 0.05). In the evaluation of rs1713985 in the development of PCV, we found significant evidence of heterogeneity (P = 0.02). 

Table 4. 

 

Subgroup Analysis of the Two SNPs for Neovascular AMD

	No. of Cases	rs1713985					rs13278062
		MAF	Nominal		Adjusted*		MAF	Nominal		Adjusted*
			P	OR (95% CI)	P	OR (95% CI)		P	OR (95% CI)	P	OR (95% CI)
tAMD
Kyoto	579	0.29	0.844	0.99 (0.86–1.13)	0.687	0.95 (0.76–1.20)	0.39	2.89 × 10−4	0.78 (0.69–0.89)	0.196	0.87 (0.70–1.08)
Yamanashi	112	0.27	0.303	1.25 (0.82–1.91)	0.340	1.24 (0.81–1.93)	0.42	0.010	0.60 (0.41–0.89)	0.016	0.61 (0.41–0.91)
Singapore	122	0.31	0.024	1.56 (1.06–2.29)	0.104	1.46 (0.92–2.32)	0.30	0.179	0.77 (0.53–1.13)	0.337	0.80 (0.52–1.25)
Hong Kong	200	0.30	0.634	0.93 (0.71–1.24)	0.860	0.97 (0.73–1.30)	0.30	0.347	0.87 (0.66–1.16)	0.330	0.86 (0.65–1.16)
Meta-analysis	1013		0.378	1.09 (0.90–1.34)	0.598	1.04 (0.89–1.22)		7.22 × 10−6	0.78 (0.70–0.87)	8.21 × 10−3	0.81 (0.70–0.95)
PCV
Kyoto	720	0.29	0.882	0.99 (0.87–1.12)	0.242	0.89 (0.73–1.08)	0.39	1.69 × 10−4	0.79 (0.71–0.90)	0.044	0.82 (0.68–0.99)
Yamanashi	211	0.31	0.035	1.49 (1.03–2.15)	0.040	1.95 (1.19–3.19)	0.38	0.032	0.69 (0.49–0.97)	0.029	0.67 (0.48–0.96)
Singapore	118	0.28	0.099	1.39 (0.94–2.06)	0.256	1.28 (0.84–1.96)	0.34	0.020	0.64 (0.44–0.94)	0.017	0.63 (0.43–0.92)
Hong Kong	233	0.31	0.890	0.98 (0.75–1.28)	0.830	0.97 (0.72–1.30)	0.32	0.154	0.82 (0.63–1.08)	0.030	0.71 (0.53–0.97)
Meta-analysis	1282		0.245	1.12 (0.92–1.36)	0.820	1.02 (0.88–1.18)		8.41 × 10−7	0.78 (0.70–0.86)	3.79 × 10−5	0.74 (0.65–0.86)

 tAMD, typical age-related macular degeneration with choroidal neovascular membrane.

*  Adjusted for age and sex.

Discussion

A recent article by Arakawa et al. ²⁷ suggested that two loci (rs13278062 at TNFRSF10A-LOC389641 on chromosome 8p21 and rs1713985 at REST-C4orf14-POLR2B-IGFBP7 on chromosome 4q12) conferred risk for the development of neovascular AMD in case-control studies of a Japanese population. In the present study, we evaluated these variants by using a larger number of cases (n = 2360) than in the previous study (n = 1536). However, our current study, including four studies of several Asian cohorts analyzed both independently and collectively, shows no significant association between rs1713985 at REST-C4orf14-POLR2B-IGFBP7 and neovascular AMD (P = 0.785, OR = 1.02, 95% CI = 0.90–1.15). Although one of the cohorts (Yamanashi) suggested marginal significance (P = 0.058), this result is likely due to the limited sample size of the control group in the Yamanashi cohort, and, thus, is theoretically more prone to sampling error. In fact, the MAF of rs1713985 in the pooled controls (Japanese: 0.287, Chinese: 0.279) were almost the same as in the previous study that used 18,894 individuals with various diseases as controls (0.282–0.286). ²⁷ Considering similarity in the MAF of the controls, the discrepancy of the association for rs1713985 may be due to the population substructure of the cases in the previous study. 

On the other hand, the current study confirmed the significance of rs13278062 at TNFRSF10A-LOC389641 on chromosome 8p21 for development of AMD in Asians (P = 1.12 × 10⁻⁴, OR = 0.79, 95% CI = 0.70–0.89). Although the calculated risk for developing neovascular AMD was smaller than that reported in the previous study (OR = 0.73), ²⁷ data from all four studies analyzed here suggested that the T allele of rs13278062 conferred risk for AMD (Fig. 1). 

Figure 1. 

 

Forest plot of the combined analysis of six case-control cohorts for rs13278062 at 8p21. Horizontal lines represent the study-specific odds ratio and 95% CI, and various-sized squares correspond to the sample size of cases. Data from all studies suggested that the T allele of rs13278062 conferred risk for AMD. We found no evidence of heterogeneity in this evaluation (P = 0.53).

In the subgroup analysis targeting the development of typical AMD-CNV and PCV, an association similar to that of all neovascular AMDs was obtained; rs1713985 did not show a significant association for either typical AMD-CNV or PCV, while rs13278062 showed a significant association for both typical AMD-CNV and PCV. In particular, rs13278062 showed almost the same susceptibility for developing typical AMD-CNV and PCV (OR = 0.82 and 0.75, respectively), which suggests that there is no difference between typical AMD and PCV with regard to the role of rs13278062 in disease development. These results are consistent with previous studies that have investigated the significance of genetic backgrounds for both typical AMD-CNV and PCV. ^17–20  

Currently, several genes have been reported to be associated with developing AMD in Asians, including the following major AMD-associated loci in Caucasians: (1) age-related maculopathy susceptibility 2 and high-temperature requirement factor A1 genes (ARMS2/HTRA1) locus, ^2,3 (2) complement factor H gene (CFH), ^4–7 and (3) complement component 2 and factor B genes (C2/CFB) locus. ⁸ Considering the reported OR of each major locus was approximately 2.0, ^{2–8,17–20} the susceptibility of rs13278062 at TNFRSF10A-LOC389641 for developing AMD would be smaller than those of the other major loci. In fact, the population attributable risk (PAR) ³⁸ for rs13278062 was 5.9% in our cohorts. However, since rs13278062 has been reported to associate with the transcriptional activity of TNF-related apoptosis inducing ligand (TRAIL) receptors (TRAILR1), ³⁹ which are known to be involved in apoptosis and inflammation, ⁴⁰ rs13278062 might have a functional role in developing AMD. 

In the present study, we focused on just two SNPs that were identified as the “peak” associations in the two novel regions reported by Arakawa et al. ²⁷ The failure to replicate the 4q12 association might be due to the limited attempt to replicate only one SNP. However, in the original study, ²⁷ they found that rs1713985 represents a linkage disequilibrium block that spans from 57.421 to 57.611 million bases (Mb) on 4q12. In addition, they showed a successful replication for the observed association in the discovery phase using the additional cohort of 708 AMD cases. Thus, if there were true associations between variants at 4q12 and developing AMD, the association of rs1713985 should be replicated in our cohorts that include the same population. 

In conclusion, this study provides an Asian population-wide replication study for associations of rs13278062 at TNFRSF10A-LOC389641 and rs1713985 at REST-C4orf14-POLR2B-IGFBP7 with the development of neovascular AMD; we confirmed the significance of rs13278062 for Asian neovascular AMD but found no association for rs1713985. Thus, the results suggest that only one locus (TNFRSF10A-LOC389641) of the two suggested loci confers increased risk of developing AMD in Asian populations. Because most of the reported genetic risk factors for developing AMD in Asians are similar to those for developing Caucasian AMD, a replication study using a Caucasian cohort would be needed to confirm the significance of TNFRSF10A-LOC389641 on chromosome 8p21. 

Footnotes

 Supported by grants in part by grants-in-aid for scientific research (19390442, 22791706, 23791972, and 22791653) from the Japan Society for the Promotion of Science, Tokyo, Japan; by the Japanese National Society for the Prevention of Blindness; by funding from the American Health Assistance Foundation (M2011068); by a Endowment Fund for Lim Por-Yen Eye Genetics Research Centre and the General Research Fund from the Research Grants Council (473410), Hong Kong; and by grants from the SingHealth, Singapore (SHF/FG381P/2007), the National Medical Research Council, Singapore (STaR/0003/2008), and the Biomedical Research Council, Singapore (BMRC 09/1/35/19/616 and 08/1/35/19/550).

Footnotes

 Disclosure: I. Nakata, None; K. Yamashiro, None; Y. Akagi-Kurashige, None; M. Miyake, None; K. Kumagai, None; A. Tsujikawa, None; K. Liu, None; L.J. Chen, None; D.T.L. Liu, None; T.Y.Y. Lai, None; Y. Sakurada, None; S. Yoneyama, None; C.-Y. Cheng, None; P. Cackett, None; I.Y. Yeo, None; W.T. Tay, None; B.K. Cornes, None; E.N. Vithana, None; T. Aung, None; K. Matsuo, None; F. Matsuda, None; T.-Y. Wong, None; H. Iijima, None; C.P. Pang, None; N. Yoshimura, None