CFH and ARMS2 Variations in Age-Related Macular Degeneration, Polypoidal Choroidal Vasculopathy, and Retinal Angiomatous Proliferation

Abstract

Purpose.: To seek an association in Japanese individuals between the CFH polymorphisms Y402H and I62V and the ARMS2 polymorphism A69S and age-related macular degeneration (AMD) or its three subtypes: typical (t)AMD, polypoidal choroidal vasculopathy (PCV), and retinal angiomatous proliferation (RAP).

Methods.: The three polymorphisms were genotyped in a case–control study of 1351 control subjects and 962 patients with AMD.

Results.: The three polymorphisms correlated with AMD (Y402H, P = 1.54 × 10⁻⁶; I62V, P =1.94 × 10⁻²⁹; and A69S, P = 9.56 × 10⁻⁴³). The I62V and A69S polymorphisms were associated with all three subtypes: tAMD (P = 3.74 × 10⁻¹⁸ and 1.37 × 10⁻³⁵, respectively), PCV (P = 3.18 × 10⁻¹⁹ and 3.96 × 10⁻¹⁸, respectively), and RAP (P = 0.034 and 2.49 × 10⁻¹⁸, respectively). Y402H was associated with tAMD (P = 3.00 × 10⁻⁵) and with PCV (P = 9.73 × 10⁻⁵), but no association was found with RAP, possibly because of the small sample size and the rare minor allele. The risk allele contribution of A69S was stronger for RAP than for tAMD or PCV and was stronger for tAMD than for PCV.

Conclusions.: CFH Y402H is associated with AMD, tAMD, and PCV, whereas I62V is associated with all three subtypes. ARMS2 A69S has a strong association with all three subtypes, with the association being strongest for RAP and weakest for PCV. PCV and RAP may thus be subtypes of AMD that are genetically distinct from tAMD.

Age-related macular degeneration (AMD) is the leading cause of severe impairment of visual function in people over 50 years of age who reside in industrialized countries. Maruko et al. ¹ surveyed the distribution of subtypes of exudative AMD in the Japanese and showed that 54.7% of patients had polypoidal choroidal vasculopathy (PCV), 35.3% had typical (t)AMD, and 4.5% had retinal angiomatous proliferation (RAP). PCV and RAP is differentiated from tAMD by the presentation of choroidal neovascularization (CNV), and PCV is characterized by orange subretinal polypoidal dilations arising from the choroidal vascular network. ^{2 –5} Intraretinal neovascularization occurs early in the course of RAP and extends downward into the subretinal space, where it communicates with the CNV. ⁶ Reactions to treatment are also different, depending on the subtype of AMD. AMD has been treated recently with anti-vascular endothelial growth factor (VEGF) therapy and/or photodynamic therapy (PDT), and although PDT tends to be of benefit in PCV, ^{7 –9} anti-VEGF therapy does not appear to diminish the polypoidal lesions. ^{10 –12} Although there have been numerous reports of a variety of treatments for RAP, most reports have shown that these lesions are very difficult to treat effectively. ^{13 –20}  

Evidence has suggested an association of AMD with polymorphisms in the complement factor H (CFH) gene ^{21 –26} and age-related maculopathy susceptibility 2 (ARMS2) gene. ^{24,25,27 –29} The differences in genotype association to AMD have been carefully investigated among various ethnic groups and by subtypes of AMD. 

Among the various polymorphisms in the CFH gene, the Y402H and I62V variants have been most intensively evaluated. An association of the Y402H variant with AMD has been reported in the United States, the United Kingdom, Iceland, The Netherlands, France, Germany, Russia, China, India, and Australia, ^{21 –23,26,28,30 –37} although this association has not been replicated in several studies conducted in east Asian countries, including Japan, China, and Korea. ^{38 –45} In contrast, the I62V variant has been shown to be associated with AMD in both Caucasian and Asian subjects. ^{26,42 –45} As for an association with PCV, the I62V polymorphism has shown an association, whereas the Y402H polymorphism has not. ^46,47 Since the C allele of the Y402H variant, which has been shown to be a risk allele for AMD in many nationalities, is very rare in Asians and in PCV, it is important to use a large cohort to look for its association. As for RAP, Wegscheider et al. ⁴⁸ reported an association of Y402H polymorphism with RAP in Caucasians. To date, however, the association of the I62V polymorphism has not been investigated in patients with RAP. 

The A69S variant in the ARMS2 gene is another polymorphism that is associated with AMD, ^{27 –29,36,49 –53} an association that has been reported to be present in both Caucasian and Asian populations. The A69S variant is associated also with PCV. ^46,49,54 However, whether the A69S polymorphism is associated with RAP has not been investigated. 

In the study described herein, we used a large cohort to examine the association of the Y402H and I62V polymorphisms in the CFH gene and of the A69S polymorphism in the ARMS2 gene in Japanese patients with neovascular AMD and then performed a subtype analysis in which we evaluated these associations with three subtypes of AMD: tAMD, PCV, and RAP. Furthermore, we compared the strength of the risk allele contribution among the three subtypes. 

Patients and Methods

This study was performed in accordance with the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board/Ethics Committee of Kyoto Graduate School of Medicine, Kyoto, Japan. Written informed consent was obtained from each patient. 

Nine hundred sixty-two patients with neovascular AMD were recruited from the Department of Ophthalmology at Kyoto University Hospital, Fukushima Medical University Hospital, and the Kobe City Medical Center General Hospital. Of these, 408 had tAMD, 518 had PCV, and 36 had RAP. As a control, blood samples of 1351 randomly selected healthy individuals who had been recruited by the Aichi Cancer Center Research Institute for an epidemiologic study were included. Some of these samples were also used as control samples in a genome-wide association analysis for pathologic myopia. ⁵⁵ All subjects in the present study were unrelated and were of Japanese descent. 

In the case group, all patients had a complete ophthalmic examination, including fundus photography, fluorescein and indocyanine green angiography, and optical coherence tomography. The diagnosis of PCV was based on the indocyanine green angiography showing of a branching vascular network terminating in polypoidal swelling. The diagnosis of RAP was based on the criteria of Yannuzzi et al., ⁶ via fundus biomicroscopy, fluorescein angiography, indocyanine green angiography, and optical coherence tomography. Typical AMD involved classic CNV, occult CNV, or both. All diagnoses were made by three retina specialists (KY, AT, and AO); a fourth specialist (NY) was called on when the subtype classification could not be decided on by the initial three reviewers. We included patients older than 50 years in the case group of this study because several studies on AMD have included patients 50 to 54 years of age. ^{28,40,42,46,53,56}  

Genomic DNAs were prepared from peripheral blood with a DNA extraction kit (QuickGene-610L; Fujifilm, Minato, Tokyo, Japan). CFH Y402H rs1061170, I62V rs800292, and ARMS2 A69S rs10490924 were genotyped (TaqMan SNP assay; Prism 7700 system; Applied Biosystems, Inc. [ABI], Foster City, CA). Deviations in genotype distributions from the Hardy-Weinberg equilibrium (HWE) were assessed with the HWE exact test. To evaluate the genotype effect, taking into account the trend of the genotype effect strength, we used the χ² test for trend or its exact counterpart for comparing two groups, and we used the Kruskal-Wallis test for comparing three groups. The ratio of the sexes was compared with the χ² test. Mean ages were compared by using analysis of variance (ANOVA) with post hoc comparisons tested by the Scheffé procedure. Statistical significance was set at P < 0.05. 

Results

Demographics of the study population are shown in Table 1. The mean age was significantly different between the controls and cases. Of the 2313 individuals studied (controls plus neovascular AMD patients), the Y402H SNP in CFH was successfully genotyped in 2287 (98.9%). The I62V SNP (single-nucleotide polymorphism) in CFH and the A69S SNP in ARMS2 were both successfully genotyped in 2285 (98.8%). The genotype distribution of Y402H and I62V was in HWE in the control subjects, in all the neovascular AMD patients, and in the three subtypes studied: tAMD, PCV, and RAP (P > 0.12). The A69S genotype distribution in the control subjects was also in HWE (P = 0.82), whereas it was not in the neovascular AMD patients, regardless of subtype. CFH Y402H and I62V showed a strong linkage disequilibrium (D′ = 0.907). 

Table 1.

 

Characteristics of the Study Population

	Control	tAMD	PCV	RAP	P
n	1351	408	518	36
Mean age ± SD, y	51.2 ± 16.3*	77.7 ± 8.4*	75.1 ± 8.5*	83.8 ± 7.0*	<0.0001†
Sex (male/female)	722/629	293/115	381/137	19/17	<0.0001‡

*  One-way ANOVA revealed significant difference in the age among control, tAMD, PCV and RAP (P < 0.0001). Scheffé post hoc test was used to compare the age of control vs. tAMD, control vs. PCV, control vs. RAP, tAMD vs. PCV, tAMD vs. RAP, and PCV vs. RAP. The age of tAMD and RAP was not significantly different (P = 0.082). The age was significantly different between control and tAMD, control and PCV, control and RAP, tAMD and PCV, and PCV and RAP (P < 0.0001).

†  One-way ANOVA.

‡  χ² test.

The frequency of the C allele in Y402H was 6.6% in the control subjects and 10.6% in the AMD patients (Table 2). The frequency of homozygosity for the at-risk genotype (CC) of Y402H was 0.6% in the control subjects and 1.3% in the AMD patients. The χ² test for trend revealed that the C allele contributed significantly to neovascular AMD (P = 1.54 × 10⁻⁶), and subtype analysis showed a significant contribution of the C allele to tAMD (P = 3.00 × 10⁻⁵) and to PCV (P = 9.73 × 10⁻⁵). In the present study, however, we detected no C allele contribution to RAP (P = 0.847). When the genotype distribution of the Y402H polymorphism was analyzed with the Kruskal-Wallis test, there was no significant difference between tAMD, PCV, and RAP (P = 0.655). 

Table 2.

 

Distribution of Genotype of CFH Y402H in the Control Subjects and in the Patients with AMD, tAMD, PCV, or RAP

Group	Genotype, n (%)			P with χ2 Test for Trend or Its Exact Counterpart
	CC	CT	TT	vs. control	vs. PCV	vs. RAP
Control	8 (0.6)	160 (11.9)	1174 (87.5)
AMD	12 (1.3)	176 (18.6)	758 (80.1)	1.54 × 10−6
tAMD	7 (1.7)	75 (18.7)	319 (79.6)	3.00 × 10−5	0.632	0.317
PCV	5 (1.0)	96 (18.8)	409 (80.2)	9.73 × 10−5		0.341
RAP	0 (0)	5 (14.3)	30 (85.7)	0.847

We compared our C allele frequency finding with previously reported C allele frequencies in Asians. Table 3 shows the reported C allele frequencies in both Asian control subjects and patients with AMD. When pooled, the C allele frequency was 5.2% in the control subjects and 8.7% in the AMD patients, and the nominal P value of these pooled samples was 6.93 × 10⁻⁶. Meta-analysis with the Mantel-Haenszel method revealed a significant contribution of the C allele to AMD (P = 2.15 × 10⁻⁶). 

Table 3.

 

Previously Reported C Allele Frequency of CFH Y402H Polymorphism in the Asian Control Subjects and the Asian Patients with AMD

	AMD, n (%)		Control, n (%)		P
	C	T	C	T
Lau et al.34	37 (11.3)	289 (88.7)	13 (2.8)	451 (91.2)	<0.00001
Ng et al.45	19 (5.8)	307 (94.2)	9 (2.9)	301 (97.1)	0.07
Chen et al.44	19 (5.8)	307 (94.2)	19 (3.9)	469 (96.1)	0.38
Gotoh et al.38	22 (8.2)	246 (91.8)	8 (4.1)	188 (95.9)	0.42
Mori et al.42	33 (8.8)	343 (91.2)	15 (5.3)	263 (94.7)	0.10
Kim et al.43	24 (10.5)	204 (89.5)	24 (6.5)	350 (93.5)	0.20
Fuse et al.41	7* (4.4)	153 (95.6)	28 (7.3)	356 (92.7)	0.25
Uka et al.39	19 (14.2)	115 (85.8)	24 (11.2)	190 (88.8)	0.52
Total	173† (8.7)	1811† (91.3)	140 (5.2)	2568 (94.8)	6.93 × 10−6
Meta-analysis‡					6.78 × 10−8

*  Dry AMD.

†  Samples from patients with dry AMD are omitted.

‡  Samples from Fuse et al.⁴¹ are omitted.

The frequency of the minor allele of the CFH I62V polymorphism was 41.7% in the control subjects and 25.2% in the AMD patients (Table 4). When the genotype was analyzed, the G allele contributed significantly to neovascular AMD (P = 1.94 × 10⁻²⁹), tAMD (P = 3.74 × 10⁻¹⁸), PCV (P = 3.18 × 10⁻¹⁹), and RAP (P = 0.034). When the genotype distribution of the I62V polymorphism was analyzed with the Kruskal-Wallis test, there was no significant difference between tAMD, PCV, and RAP (P = 0.853). 

Table 4.

 

Distribution of the Genotype of CFH I62V in the Control Subjects and in the Patients with AMD, tAMD, PCV, or RAP

	Genotype, n (%)			P with χ2 Test for Trend or Its Exact Counterpart
	GG	GA	AA	vs. Control	vs. PCV	vs. RAP
Control	456 (34.1)	649 (48.5)	233 (17.4)
AMD	538 (56.8)	341 (36.0)	68 (7.2)	1.94 × 10−29
tAMD	228 (57.0)	148 (37.0)	24 (6.0)	3.74 × 10−18	0.651	0.386
PCV	290 (56.8)	182 (35.6)	39 (7.6)	3.18 × 10−19		0.500
RAP	20 (55.6)	11 (30.6)	5 (13.8)	0.034

The A69S polymorphism in the ARMS2 gene contributed to neovascular AMD more than did the polymorphisms in the CFH gene. The T allele contributed significantly to neovascular AMD (P = 9.56 × 10⁻⁴³; Table 5); furthermore, subtype analysis showed a contribution of the T allele to all subtypes of neovascular AMD: tAMD (P = 3.11 × 10⁻³⁸), PCV (P = 4.69 × 10⁻¹⁹), and RAP (P = 9.15 × 10⁻¹⁹). Because the Kruskal-Wallis test showed a significant difference in genotype contribution among these three subtypes of AMD (P = 1.26 × 10⁻⁹), we compared the contribution of each of these three genotypes. This comparison revealed that the contribution of the T allele was strongest in RAP and weakest in PCV. When examined with the χ² test for trend, the T allele's contribution was significantly greater in RAP than in tAMD (P = 4.38 × 10⁻⁵), and was significantly greater in RAP than in PCV (P = 6.26 × 10⁻⁸). Furthermore, the contribution of the T allele was significantly greater in tAMD than in PCV (P = 1.38 × 10⁻⁴). 

Table 5.

 

Distribution of the Genotype of ARMS2 A69S in the Control Subjects and the Patients with AMD, tAMD, PCV, or RAP

	Genotype, n (%)			P with χ2 Test for Trend or Its Exact Counterpart
	TT	TG	GG	vs. Control	vs. PCV	vs. RAP
Control	196 (14.6)	638 (47.8)	502 (37.6)
AMD	385 (40.5)	374 (39.4)	191 (20.1)	9.56 × 10−43
tAMD	183 (45.2)	155 (38.3)	67 (16.5)	1.37 × 10−35	1.38 × 10−4	4.38 × 10−5
PCV	171 (33.6)	216 (42.4)	122 (24.0)	3.96 × 10−18		6.26 × 10−8
RAP	31 (86.1)	3 (8.3)	2 (5.6)	2.49 × 10−18

Since it is well known that patients with PCV are relatively young and RAP patients relatively old and that RAP is a predominantly female condition, we performed logistic regression analysis to adjust for age and sex differences among the three subtypes of AMD. Again, the T allele's contribution was strongest in RAP and weakest in PCV and was significantly greater in RAP than in tAMD (P = 1.40 × 10⁻³). When the genotypes of TT and GG were compared, the odds ratio was 2.09 (95% confidence interval [CI], 1.00–4.37). The T allele's contribution was significantly greater in RAP than in PCV (P = 2.27 × 10⁻⁵) with an odds ratio of 3.22 (95% CI, 1.53–6.77) between the genotype of TT and GG. The contribution of the T allele was significantly greater in tAMD than in PCV (P = 5.38 × 10⁻⁴), with an odds ratio of 1.37 (95% CI, 1.13–1.64) between the genotype of TT and that of GG. 

Discussion

In the study reported herein, the CFH Y402H and I62V polymorphisms and the ARMS2 A69S polymorphism were associated with neovascular AMD in a Japanese cohort. Subtype analysis showed the association of I62V and A69S to the three subtypes of AMD: tAMD, PCV, and RAP. Furthermore, our findings suggest that the A69S association is strongest for RAP and weakest for PCV. As for Y402H, we detected an association with tAMD and PCV, whereas an association with RAP could not be evaluated precisely because of the infrequent finding of the C allele and the sparsity of RAP in our Japanese cohort. 

In contrast to reports on Caucasians, many reports on Asians have shown that the Y402H association with AMD cannot be replicated (Table 3). ^{38 –45} Furthermore, PCV has been shown to not to be associated with the Y402H polymorphism in the Japanese or Chinese. ^46,47 In the present study, however, the Y402H polymorphism appeared to be associated with AMD, tAMD, and PCV. Because the frequency of the C allele was similar between our cohort and the pooled samples in previous reports, this discrepancy may well be due to the relatively small size of the sample used in the other reports. 

As for the I62V polymorphism, our findings are consistent with those of previous reports. The I62V polymorphism is associated with neovascular AMD, tAMD, and PCV. Considering the strong linkage disequilibrium between Y402H and I62V, it seems natural that both Y402H and I62V have a significant association with neovascular AMD, tAMD, and PCV. An evaluation of the CFH gene polymorphism I62V would be more useful in Asians than an evaluation of Y402H. The frequency of the risk-homozygous genotype of Y402H in Asians has been shown to be 0% to 3.0% in control subjects, 0.6% to 3.4% in AMD, and 0% to 1.0% in PCV. ^{38 –45,47,57} These small amounts of the risk allele could lead to false-negative results if a small cohort were examined. In contrast to Y402H, the distribution of the I62V polymorphism was well balanced, and so it would be a useful marker in both Asians and Caucasians, because the association of I62V with AMD has been reported to be strong in Caucasians. ²⁶  

In the present study, we did not detect associations between RAP and the CFH Y402H polymorphism. However, considering the infrequent finding of the C allele in Y402H in Asians and the sparsity of RAP in Asians, our results may be falsely negative. The power of this study to detect a significant difference in the allele frequency of Y402H between RAP and the control was calculated as 4.4%, and the age difference between cases and controls could lower the power even more. Since the prevalence of neovascular AMD in the Japanese is reported to be 0.5%, ⁵⁸ approximately 0.5% of the control subjects in the present study would have AMD if they were the same age as those in the case group. 

Our findings on ARMS2 A69S suggest an association of the A69S polymorphism with RAP, but, because A69S in RAP has not yet been investigated in depth, further study is warranted. 

Because the distribution of the ARMS2 A69S polymorphism had been reported to be similar between AMD and PCV, ^49,51,57,59 we hypothesized in a previous report that patients with AMD and PCV may have the same genetic background. ⁵⁷ However, the present study revealed a significant difference in the distribution of the A69S polymorphism among the three subtypes of AMD, and so the possibility exists that PCV and RAP are genetically distinct from tAMD. Although in the present study we detected no difference in the distribution of the CFH polymorphism among the three subtypes of AMD, Wegscheider et al. ⁴⁸ showed that the association of the Y402H polymorphism is weaker for RAP than for tAMD if classic CNV is present. Taken together with our findings, their evidence shows that ARMS2 would be more important in RAP and less important in PCV, whereas CFH would affect the occurrence of the three subtypes of AMD to the same extent or might be less important for RAP than for tAMD. 

In regard to the location of neovascular lesions in AMD, PCV is characterized by abnormal vessels beneath the retinal pigment epithelium (RPE), tAMD is characterized by subretinal CNV located both beneath and above the RPE, and RAP is characterized by intraretinal neovascularization above the RPE. As for the location of gene expression, CFH would play a major role beneath the RPE, whereas ARMS2 would play a major role above the RPE, because CFH has been shown to be expressed primarily in the RPE, drusen, and choroidal capillaries, ²¹ and ARMS2 has been shown to be expressed in the ellipsoid region of the photoreceptor cells. ⁵² Since it seems that the location of characteristic neovascularization corresponds to the location of susceptible gene expression in each subtype of AMD, further studies on the functions of ARMS2 and CFH during the course of AMD may elucidate the different or common pathway that leads to the three distinct subtypes of AMD. 

In the present study, the frequency of the ARMS2 A69S polymorphism in AMD patients showed a departure from the HWE, and although this deviation could suggest selection bias, deviation from the HWE in affected individuals could also be regarded as indicative of the presence of susceptibility loci. ^60,61 Indeed, deviation from the HWE has been observed in several studies of the ARMS2 gene polymorphism. ^28,32,53  

In summary, our findings indicate a strong association of the ARMS2 A69S variant with three subtypes of AMD (tAMD, PCV, and RAP) in Japanese subjects. Among these three subtypes, the association is strongest for RAP and weakest for PCV. Furthermore, we showed an association of the CFH Y402H variants for AMD, tAMD, and PCV in the Japanese by using a large cohort. Further genetic study, however, may well deepen our understanding of these three subtypes of AMD. 

Footnotes

 Supported in part by Grants-in-aid for Scientific Research 21249084 and 200791294 from the Japanese Society for the Promotion of Science, Tokyo, and the Japanese National Society for the Prevention of Blindness, Tokyo, Japan.

Footnotes

 Disclosure: H. Hayashi, None; K. Yamashiro, None; N. Gotoh, None; H. Nakanishi, None; I. Nakata, None; A. Tsujikawa, None; A. Otani, None; M. Saito, None; T. Iida, None; K. Matsuo, None; K. Tajima, None; R. Yamada, None; N. Yoshimura, None

The authors thank Hiroshi Tamura, Sotaro Ooto, Kuniharu Saito, Akio Oishi, and Yasuo Kurimoto for their assistance in the recruitment of patients.