Association Between the Cholesteryl Ester Transfer Protein Gene and Polypoidal Choroidal Vasculopathy

Abstract

Purpose.: To determine whether genetic variants in the lipid-associated genes are related to the risk of developing polypoidal choroidal vasculopathy (PCV) in a Japanese population.

Methods.: Five hundred eighty-one patients with PCV and 793 controls were enrolled in the study. Association analysis of allele and genotype frequencies was performed for the following single-nucleotide polymorphisms (SNPs) that are associated with high-density lipoprotein cholesterol levels in blood: rs493258 at the hepatic lipase gene (LIPC), rs3764261 at the cholesteryl ester transfer protein gene (CETP), and rs12678919 at the lipoprotein lipase gene (LPL). A further model adjusting for age-related maculopathy susceptibility 2 (ARMS2) A69S, complement factor H (CFH) I62V, age, sex, and smoking status was used to confirm the independent association of these SNPs with other covariates.

Results.: CETP rs3764261 was significantly associated with the development of PCV; the frequency of the minor allele A was higher in the PCV cases (24.0%) than in the control subjects (18.5%) (P = 0.0025; odds ratio [OR], 1.41; 95% confidence interval, 1.13–1.75). Furthermore, we found an independent association of CETP variants with age, sex, smoking status, and genetic background of ARMS2 A69S, CFH I62V, LIPC rs493258, and LPL rs12678919 (P = 0.0013; OR, 1.50). LIPC rs493258 and LPL rs12678919 did not show significant associations with the development of PCV (P > 0.05).

Conclusion.: CETP variants are associated a risk of developing PCV among the Japanese population.

Introduction

Polypoidal choroidal vasculopathy (PCV) is characterized by aneurysmal dilations with interconnecting vessels that are best demonstrated by indocyanine green angiography. ^1–3 Clinically, PCV is classified into a specific subtype of age-related macular degeneration (AMD), and the incidence of PCV in Asian populations has been reported to be higher than that in Caucasians. ^4–6 Controversies exist about the pathogenesis of PCV; whether this condition represents inner choroidal vascular abnormalities or a particular variety of choroidal neovascularization (CNV) remains undetermined. However, because there are apparent differences in the demographic risk profile, clinical course, and visual prognosis, PCV is thought to be a distinct clinical entity. ⁷ For example, the response to treatment, particularly in photodynamic therapy for PCV, is completely different from that for typical AMD and CNV. ^8,9  

Cholesterol and lipids are reported to accumulate underneath the retinal pigment epithelium (RPE) with age. When sufficient debris, including lipids, accumulates and forms a mound between the RPE cell and its basement membrane, it can be seen clinically as drusen. Because many population-based studies have shown the association between drusen and the progression of AMD, drusen is thought to be one of the determinants of both early and late AMD. In fact, an association between high-density lipoprotein (HDL) cholesterol level and the development of AMD has been reported in several studies. ^10–12  

Previous studies ^13–15 showed that the prevalence of drusen under RPE was reported to be lower in PCV than in AMD. Therefore, the absence of drusen was thought to be one of the criteria necessary to diagnose PCV. ^6,15,16 However, the results of a clinical study ¹⁶ suggested that drusen is frequently seen in PCV eyes, and several studies ^6,17,18 reported that drusen were observed in 20% to 27% of unaffected, fellow eyes in patients with unilateral PCV. Therefore, whether drusen has a functional role in the development of PCV remains controversial. 

While previous investigations showed a lower prevalence of drusen among patients with PCV, lipid deposits that distribute from the inner retina to the outer retina are known to be the paramount features of PCV (Figure). Some recent investigations, including a study ¹⁹ in a large cohort of Caucasians, showed significant associations between the lipid-associated genes and the development of AMD. These discoveries of genetic variants in the lipid pathway provided new insight into the pathogenesis of AMD. However, there are limited reports evaluating the association between the lipid-associated genes and the development of PCV. Although several genes are thought to be involved in regulating susceptibility to the development of PCV, ^20–23 almost all are identical to those involved in the development of AMD, including the age-related maculopathy susceptibility 2 and high-temperature requirement factor A1 genes (ARMS2/HTRA1) locus ^24,25 and the complement factor H gene (CFH). ^26–29 Considering that several studies ^13–15 reported a difference in the clinical features of drusen between AMD and PCV, there could be different roles of the lipid-associated genes in these subtypes. Thus, we aimed in this study to determine whether genetic variants in the lipid-associated genes, including variants affecting HDL cholesterol levels, are related to the risk of developing PCV in a Japanese population. 

Figure

 

A 64-year-old woman with a typical case of PCV in the right eye. (A) Fundoscopic examination shows massive subretinal hemorrhage, lipid deposits, and reddish orange nodules. (B) Indocyanine green angiography demonstrates a small branching vascular network terminating in polypoidal lesions (white arrow). The speckle noise–reduced spectral-domain optical coherence tomography image of a horizontal section corresponding to the arrow indicated in fluorescein angiography (C) shows hyperreflective foci, indicating lipids ([D], arrowhead), in the outer retina beside the polyp ([D], white arrow).

Methods

All procedures in this study adhered to the tenets of the Declaration of Helsinki, and the ethics committee of each institution involved approved the study protocols. All patients were fully informed about the purpose and procedures of this study, with each patient providing written consent. 

Five hundred eighty-one patients with PCV were recruited from the departments of ophthalmology at Kyoto University Hospital, Fukushima Medical University Hospital, and Kobe City Medical Center General Hospital. The diagnosis of PCV was based on indocyanine green angiography, which showed a branching vascular network terminating in polypoidal swelling (Figure), and was confirmed by three retina specialists (KY, AT, AO); a fourth specialist (NY) was consulted when the diagnosis could not be agreed on by the initial three reviewers. Patients who had both typical CNV and polypoidal lesions were excluded from this study. The control group consisted of 793 unrelated individuals 60 years or older recruited in the Nagahama Prospective Genome Cohort for Comprehensive Human Bioscience (the Nagahama Study). ³⁰ Fundoscopic photographs of both eyes confirmed the absence of any signs of AMD (large drusen or pigment change) using the Age-Related Eye Disease Study ³¹ severity scale, with grading by two independent ophthalmologists (IN, YAK), followed by grading by a senior reviewer (KY). 

We targeted three single-nucleotide polymorphisms (SNPs) of three genes reported to be associated with HDL cholesterol levels in blood, including rs493258 at the hepatic lipase gene (LIPC), rs3764261 at the cholesteryl ester transfer protein gene (CETP), and rs12678919 at the lipoprotein lipase gene (LPL). ³² Genomic DNA was prepared from peripheral blood using a DNA extraction kit (QuickGene-610L; Fujifilm, Minato, Tokyo, Japan). All case samples were genotyped using the Taqman SNP assay with an ABI PRISM 7700 system (Applied Biosystems, Foster City, CA). Controls were genotyped using Human610-Quad BeadChips and HumanOmni2.5 BeadChips (Illumina, Inc., San Diego, CA). ARMS2 A69S (rs10490924) and CFH I62V (rs800292) were also genotyped in the same manner. Fasting serum samples from the control subjects were analyzed for HDL cholesterol level, measured using a direct assay system with the selective inhibitory method on an automatic analyzer (LABOSPECT 008; Hitachi, Ltd., Tokyo, Japan). We did not have HDL cholesterol data for the case samples. 

Information on smoking status was obtained via a self-reported questionnaire with three categories of never smoker, former smoker, and current smoker. The never smokers were those who had smoked fewer than 100 cigarettes in the past, current smokers were those who had smoked in the past year, and former smokers were those who had quit smoking more than 1 year earlier. 

Deviations in genotype distributions from the Hardy-Weinberg equilibrium (HWE) of the controls were assessed with the HWE exact test. Statistical differences in the observed allelic distribution were identified using logistic regression analyses with age and sex adjustments, under the assumption of an additive genetic effect where the genotypes of each SNP are coded numerically as 0, 1, and 2 for the number of minor alleles carried. A linear regression analysis was performed to assess the association between HDL cholesterol level and genotype. R software (http://www.r-project.org/ in the public domain) was used for statistical analyses. P < 0.05 was considered statistically significant. 

Results

Demographics of the study population are given in Table 1. Genotype and allele frequencies of the three SNPs were analyzed in 581 patients with PCV and compared with those of 793 age-matched individuals without any signs of AMD or PCV. The genotyping of all evaluated SNPs had a success rate exceeding 99.4%. 

Table 1

 

Characteristics of the Study Population

Variable	Cases, n = 581	Controls, n = 793	P Value
Age, y
Mean ± SD	72.59 ± 8.13	65.99 ± 4.33	<0.0001
Range	48–92	60–75
Sex, n (%)
M	420 (72.3)	326 (41.1)	<0.0001
F	161 (27.7)	467 (58.9)
Smoking status, n (%)
Never	200 (38.5)	509 (64.3)	<0.0001
Former	195 (37.6)	176 (22.3)
Current	124 (23.9)	106 (13.4)

Table 2 gives details of genotype and allele frequencies and summary statistics. The distributions of the genotypes for all evaluated SNPs were in HWE (P > 0.05). We found that CETP rs3764261 was significantly associated with the development of PCV; the frequency of the minor allele A in the patients with PCV (24.0%) was higher than that in the controls (18.5%) (P = 0.0025; odds ratio [OR], 1.41; 95% confidence interval [CI], 1.13–1.75). This significant association remained even after a correction for multiple testing (P = 0.0075). LIPC rs493258 and LPL rs12678919 did not show significant associations with the development of PCV (P > 0.05). 

Table 2

 

Distribution of Genotypes and Results of the Association Tests

Gene	SNP	Allele		Cases, n = 581				Controls, n = 793				Association Results*
		1	2	11	12	22	MAF	11	12	22	MAF	P Value	OR (95% CI)
LIPC	rs493258	G	A	32	185	354	0.22	37	259	497	0.21	0.706	1.04 (0.84–1.30)
CETP	rs3764261	C	A	332	210	33	0.24	528	237	28	0.19	0.0025	1.41 (1.13–1.75)
LPL	rs12678919	A	G	439	135	3	0.12	602	179	12	0.13	0.883	1.02 (0.77–1.35)

 MAF, minor allele frequency.

*  Adjusted for age and sex.

Next, we conducted a logistic regression analysis that included the effects of the most robust Japanese variants associated with AMD and PCV, ARMS2 A69S (rs10490924) and CFH I62V (rs800292), as well as age, sex, smoking status, LIPC rs493258, and LPL rs12678919 in the regression model. Table 3 gives the results of the logistic regression analysis. CETP rs3764261 remained significant for the development of PCV even after including the effects of these covariates (P = 0.0013; OR, 1.50; 95% CI, 1.17–1.92). 

Table 3

 

Logistic Regression Analysis, Including Major Factors Associated With PCV

Variable	P Value*	OR (95% CI)
Age	<0.0001	1.18 (1.16–1.21)
F:M sex	<0.0001	3.16 (2.20–4.52)
ARMS2 rs10490924 (G/T)	<0.0001	2.27 (1.86–2.77)
CFH rs800292 (A/G)	<0.0001	1.77 (1.43–2.19)
LIPC rs493258 (G/A)	0.689	1.05 (0.82–1.35)
CETP rs3764261 (C/A)	0.0013	1.50 (1.17–1.92)
LPL rs12678919 (A/G)	0.948	0.99 (0.72–1.35)
Smoking (never, former, or current)	0.0107	1.35 (1.07–1.69)

*  A logistic regression model was used for covariate adjustment.

Finally, we investigated the role of CETP rs3764261 in blood HDL cholesterol level using fasting serum samples from 793 control subjects. The mean ± SD HDL cholesterol level of the control samples was 61.3 ± 16.1 mg/dL. In this analysis, we found that the A allele of rs3764261 was associated with the following increases in HDL cholesterol: 59.3 mg/dL for the CC genotype, 64.8 mg/dL for the CA genotype, and 67.2 mg/dL for the AA genotype (P < 0.0001). 

Discussion

Plasma CETP was first described as a high-molecular-weight protein stimulating the transfer of cholesteryl ester between lipoproteins in plasma of hypercholesterolemic rabbits. ³³ Other studies demonstrated various roles of CETP in the lipid pathway: CETP facilitates the transfer of triglycerides and phospholipids ³⁴ ; it is an important component of reverse cholesterol transport, which is chiefly characterized by the transport of cholesterol from peripheral tissues to the liver; and it regulates the concentration of HDL cholesterol. ^35,36  

After the discovery of the association between HDL cholesterol level and cardiovascular diseases, ³⁷ studies ^38,39 evaluated the functional role of the lipid-associated genes that can affect the HDL cholesterol level. Among those genes, the A allele of CETP rs3764261 was associated with an increase in HDL cholesterol by 5.6 mg/dL among the Japanese population. ⁴⁰ Herein, we confirmed the role of rs3764261 in increased HDL cholesterol levels among 793 healthy Japanese individuals. 

In the present study comparing the allelic distributions of CETP variants in a sample of 581 patients with PCV and 793 control subjects, the A allele of CETP rs3764261 was significantly associated with a risk of developing PCV (OR, 1.41; 95% CI, 1.13–1.75), which indicates a higher level of HDL cholesterol in patients with PCV. In addition, the association of CETP variants remained significant even when we adjusted for the effects of other established risk factors for developing AMD and PCV (age, sex, smoking status, and genetic background of ARMS2 A69S, CFH I62V, LIPC rs493258, and LPL rs12678919). Although the effect of CETP variants (OR, 1.50) was not as large as the effects of the major genes associated with AMD and PCV (ORs, 2.27 for ARMS2 and 1.77 for CFH) in this regression analysis, we were able to confirm that CETP variants have a significant role in the development of PCV. Our findings for CETP rs3764261 were similar to the associations already documented in AMD among Caucasians, ^41,42 which suggests that a higher HDL cholesterol level may be a risk factor in both PCV and Caucasian AMD. The hypothesis that a higher level of HDL cholesterol is associated with the development of PCV might appear contradictory to the fact that a lower level of HDL cholesterol is associated with an increased risk of cardiovascular disease. However, despite the well-known antiatherogenic properties of HDL cholesterol, some studies ^10,11,43 found elevated levels of HDL cholesterol in Caucasian patients with AMD. 

Recently, Zhang et al. ⁴⁴ reported an investigation of lipid-associated SNPs for PCV and neovascular AMD in a Chinese population. In that article, they showed a significant association of CETP with PCV, while no association was found with neovascular AMD. Thus, they concluded that the HDL cholesterol pathway in the pathogenesis of PCV likely differs from that of neovascular AMD. However, the sample size evaluated in the their article was small (204 controls, 250 patients with PCV, and 157 patients with neovascular AMD), which suggests that the negative result of the association between CETP and neovascular AMD could have been due to insufficient power to detect the association. To confirm whether the observed association of CETP with PCV exists for neovascular AMD as well, we performed an additional analysis using another Japanese cohort of neovascular AMD cases (n = 452). In this evaluation, we found a significant association between CETP and neovascular AMD (P = 0.0246; OR, 1.35). 

Adenosine triphosphate–binding cassette, subfamily A member 1 (ABCA1) is also known to be associated with the lipid pathway. Because ABCA1 has been reported to be another susceptible gene for the development of AMD in Caucasians, ¹⁹ we also evaluated whether ABCA1 rs1883025 has a significant role in the development of PCV but found no significant association with PCV (P > 0.05). In previous genome-wide association analyses for HDL cholesterol, the strongest and most consistently associated SNPs have been reported in the CETP locus. ^45,46 Study ³² findings also suggest that LIPC rs493258 and LPL rs12678919 are associated with HDL cholesterol level in Caucasians, so the lack of association in the present study could be due to insufficient statistical power or racial/ethnic differences. Further study that includes a larger number of participants is needed to clarify the association between genetic variants of HDL cholesterol–associated genes and the development of PCV. 

In the present study, there was a large sex difference between the PCV cases and the general population controls. It remains unknown why there is such a high prevalence of PCV among men. In a previous meta-analysis by Kawasaki et al., ⁴⁷ the prevalence of late AMD among Asian women was reported to be much lower than that among Asian men. In contrast, a male predominance was reported in PCV. ⁴ Considering the high prevalence of PCV among Asian populations, these results suggest that men are more likely to develop PCV. In our study, genetic factors had an enormous influence on whether participants developed PCV (Table 3). However, sex had the largest effect among all covariates on the development of PCV (OR, 3.16). A previous genetic study ²³ among Japanese may provide insight into this question because the results suggested that differences in sex would affect phenotypic differences in AMD. Another limitation of the present study was the age difference between cases and controls. Although we enrolled only controls who were 60 years or older, the average age of the control cohort was still younger than that of the case cohort, which means that some of the young controls may develop PCV in the future. To exclude a potential confounder of genetic background with age, a logistic regression analysis adjusting for age and sex was performed in the present study. However, given that the prevalence of late AMD among the Japanese population is reported to be 0.5%, ⁴⁸ the magnitude of statistical bias of the association analysis is negligible. In addition, considering that case-control association analyses among such subjects are less likely to be statistically significant, our positive results should be acceptable. 

Overall, this study provides the first evidence to date that CETP variants have a significant role in the risk of developing PCV among the Japanese population. Our study also indicates the same role of HDL cholesterol in both PCV and Caucasian AMD, although the role of fatty acids in Japanese AMD is reported to be different from that in Caucasian AMD. ⁴⁹ Further studies are needed to increase the understanding of the genetic backgrounds of PCV, as well as the molecular pathogenesis, particularly the role of lipids. 

Acknowledgments

The authors thank the participants of the Nagahama Study, the Nagahama City Office, and the nonprofit organization Zero-ji Club for Health Promotion. 

Supported by grants-in-aid from the following organizations: the Ministry of Education, Culture, Sports, Science, and Technology of Japan (2006–2012); Grants 19390442, 22791706, and 22791653 from the Japan Society for the Promotion of Science; the Japanese National Society for the Prevention of Blindness; and Takeda Science Foundation (2008–2012). 

Disclosure: I. Nakata, None; K. Yamashiro, None; T. Kawaguchi, None; N. Gotoh, None; H. Nakanishi, None; Y. Akagi-Kurashige, None; M. Miyake, None; A. Tsujikawa, None; A. Oishi, None; M. Saito, None; T. Iida, None; R. Yamada, None; F. Matsuda, None; N. Yoshimura, None 

Footnotes

 See the Appendix for the members of the Nagahama Study Group.

Appendix

The following investigators were core members of the Nagahama Study Group: Takeo Nakayama (Department of Health Informatics, Kyoto University School of Public Health, Kyoto, Japan), Akihiro Sekine (Department of Genome Informatics, Kyoto University School of Public Health, Kyoto, Japan), Shinji Kosugi (Department of Medical Ethics, Kyoto University School of Public Health, Kyoto, Japan), and Yasuharu Tabara (Center for Genomic Medicine, Graduate School of Medicine, Kyoto University, Kyoto, Japan).