Keywords

8.1 Introduction

Age-related macular degeneration (AMD) is a leading cause of blindness and central vision impairment and blindness in elderly patients [1]. There are two types of AMD, dry and wet (exudative or neovascular) AMD (nAMD), which is characterized by atrophy of retinal pigment epithelium (RPE) and choroidal neovascularization respectively. The clinical manifestations of nAMD include choroidal neovascularization, subretinal fluid, hemorrhage, exudation, and fibrosis. Polypoidal choroidal vasculopathy (PCV) is characterized by the branching vascular network of the choroid and polyp-like aneurysmal dilations of its terminals [2]. Clinically, PCV is manifested as serosanguineous detachments of the pigmented epithelium and exudative changes that can recur in several episodes.

It is still controversy whether PCV presents a subtype of nAMD or a distinct disease. PCV and nAMD have some shared characters but there are also some different features, including risk factors, clinical manifestations, natural course, and response to treatment.

Both PCV and nAMD are commonly seen in elderly patients. However, PCV presents a younger age than nAMD [3, 4]. Although both PCV and nAMD occur in any race, PCV is known to be more prevalent in pigmented ethnicity while nAMD has a high prevalence in European than in Asian [5]. Smoking is a proven risk factors for both PCV and nAMD, while female gender is a protective factor for both conditions [3, 6]. Diabetes was found to be more prevalent in nAMD than in PCV patients [4].

Clinically, Both PCV and nAMD present as exudation or hemorrhage at the macular region. But there are also different characters between them. nAMD is predominantly located at the fovea or parafoveal region, while PCV may involve perifoveal, peripapillary, or even peripheral retina. The histological feature of PCV is majorly polypoidal enlargement of the terminal of the choroidal vessel. nAMD is characterized by choroidal neovascularization above or underneath the RPE. The choroidal thickness of nAMD is usually thin but that of PCV is usually thick. The natural history of PCV is multiple, recurrent episodes while nAMD is a progressive disease. Although both disorders can be treated using photodynamic therapy or anti-vascular endothelial growth factor (VEGF) antibody, nAMD responses better to anti-VEGF therapy and PCV responses better to photodynamic therapy [7].

Genetic studies of AMD have identified susceptibility single-nucleotide polymorphisms (SNPs) in multiple genes, including rs1061170 in complement factor H (CFH), rs10490924 in age-related maculopathy susceptibility 2 (ARMS2), and rs11200638 in high-temperature requirement factor H (HTRA1) [8, 9]. In 2016, the International AMD Genomics Consortium reported 34 loci associated with AMD [10]. Due to the similarities between nAMD and PCV, major gene SNPs for nAMD have also been evaluated in PCV. The CFH SNP rs1061170 was not found to be associated with PCV [11], while an adjacent SNP rs800292 was significantly associated [11,12,13]. Both rs10490924 and rs11200638 at the ARMS2-HTRA1 locus were associated with PCV [11, 12, 14, 15]. In 2012, we published a meta-analysis investigating genetic associations of PCV with SNPs in the ARMS2, HTRA1, CFH, and complement component 2 (C2) genes. The results also showed that one SNP, rs10490924, in ARMS2 showed a significant difference between PCV and AMD [16]. In 2015, we reported the updated meta-analysis of the association of genetic variants with PCV, which found 31 polymorphisms in 10 genes/loci (including ARMS2, HTRA1, CFH, C2, CFB, RDBP, SKIV2L, CETP, 8p21, and 4q12) were significantly associated with PCV. Twelve polymorphisms at the ARMS2-HTRA1 locus showed significant differences between PCV and nAMD. There are many new articles investigating these topics since the publication of the latest meta-analysis. In this chapter, we further updated our meta-analysis comparing the genetic association profiles between PCV and nAMD.

8.2 Methods of Meta-Analysis

A systematic literature search was performed using EMBASE, PubMed, Web of Science, and Chinese Biomedical Literature Database. The search used the terms (polypoidal choroidal vasculopathy or PCV) and (gene or genetic or polymorphism or variant or SNP or DNA). We retrieved all related records published from February 1, 2015, and September 27, 2018, and then added the articles published before Feb 2015 that were included in our previous meta-analysis. The reference lists of all eligible studies, reviews, and meta-analyses were also screened to prevent that any relevant studies were omitted.

The retried records were reviewed by two independent reviewers (L.M. and X. L.) and any inconsistency was resolved by discussion with another reviewer (H.C.). The following criteria were used when assessing the records [1]. case-control studies, cohort studies, or population-based studies that evaluated the difference of gene variants between PCV and nAMD; and [2] allele or genotype counts and/or frequencies being presented or able to be calculated from the data in the study. For those reports published by the same study group on the same gene markers, only the latest study was included. Case reports, animal studies, reviews, conference abstracts, comments, articles without sufficient data, or published in language other than English were excluded.

The data from included studies were extracted by the two independent reviewers (L.M. and Z.L.) and any inconsistency was resolved by discussion with another reviewer (H.C.). If there were several cohorts in the same article, they were treated as independent study. The following information from each record was extracted: first author, year of publication, the ethnicity of study subjects, study design, genotyping method, and sample size, demographics, allele, and genotype distribution in PCV and nAMD.

The distribution of genetic variants between PCV and nAMD from all included studies were pooled. Three genetic models were used, including allelic, dominant, and recessive models. The effect size was assessed using a summary odds ratio (OR) and its 95% confidence intervals (CIs) of each SNP. The software, Review Manager software (RevMan, version 5.3.5, The Cochrane Collaboration, Copenhagen, Denmark) was used for statistical analysis. The I2 statistic was adopted to assess the heterogeneity among the studies. The I2 values correspond with no (<25%), low (25%–50%), moderate (50%–75%), and high heterogeneity (≥75%). If the I2 value was ≥50%, the fixed effects model was used in the meta-analysis, otherwise, the random effects model was used. A summary P value <0.05 was considered statistically significant. We performed a sensitivity analysis by omitting one study at a time and calculating the pooled ORs for the remaining studies. Funnel plots were constructed to assess potential publication bias.

8.3 Results of Updated Meta-Analysis

Our literature search yielded a total of 1315 reports published between February 1, 2015, and September 27, 2018, from EMBASE, PubMed, Web of Science and Chinese Biomedical Literature Database. Out of these, 502 articles were excluded due to duplicates. After assessing the titles and abstracts, a further 606 reports with unrelated topics were omitted. For the remaining 107 studies, the full-texts were retrieved and reviewed. Another 89 reports were excluded, among which 62 studies were on AMD but not PCV, 2 were reviews, 23 were non-genetic studies, and 1 was a case report. Finally, 18 articles were eligible for the meta-analysis. A further 66 studies published before 2015 that were used in our previous meta-analysis were added. However, 19 of these studies were excluded because they only studied in PCV patients. Thus, a total of 65 studies were included in the meta-analysis. Figure 8.1 shows the flowchart of literature inclusion and exclusion with the specification of reasons and Table 8.1 shows the characters of the included studies.

Fig. 8.1
figure 1

Flow chart of literature screen

Full size image
Table 8.1 Characteristics of the included studies in the meta-analysis
Full size table

In these 65 studies , both PCV and nAMD were assessed for associations with a total of 57 SNPs in 20 genes or loci (i.e., ARMS2, HTRA1, CFH, VEGF-A, C2, CFB, SKIV2L, CETP, 8p21, 4q12, ELN, LIPC, LPL, FGD6, ABCA1, ABCG1, PGF, TLR3, LOXL1, and PEDF; Table 8.1). In total, 11 SNPs at the ARMS2-HTRA1 locus and 1 in FGD6 showed significant differences between PCV and nAMD (Tables 8.2 and 8.3). There was no significant difference between PCV and nAMD in the remaining 45 SNPs (Table 8.4).

Table 8.2 Polymorphisms with significant differences between PCV and neovascular AMD
Full size table
Table 8.3 Gene SNPs with significant differences between PCV and nAMD in dominant, recessive, homozygous and heterozygous models
Full size table
Table 8.4 Gene variants not significantly different between PCV and neovascular AMD
Full size table

There are 12 studies tested the most-investigated SNP, ARMS2 rs10490924, involving 2361 PCV and 2138 nAMD patients (Table 8.2) [3, 12, 15, 22, 26, 27, 35, 36, 38, 43, 44, 78]. The frequency of the T allele was significantly lower in PCV than in nAMD (summary OR 0.69; 95% CI 0.63–0.75; P = 5.50 × 10−16; Table 8.2 and Fig. 8.2). The association was also statistically significant in both dominant and recessive models (OR = 0.64, P = 8.80 × 10−8 and OR = 0.62, P = 1.47 × 10−13 respectively; Table 8.3 and Fig. 8.2). The results of the sensitivity analysis found that the association remains significant after omitting any single included cohorts (data not shown). And there was no asymmetry on the funnel plots (Fig. 8.5). There are 8 other SNPs in ARMS2, namely rs3750848, rs36212731, rs36212732, rs36212733, rs3750846, rs10664316, c.372_815del443ins54 and rs2672587, were evaluated in 2 to 3 cohorts, and also showed significant differences between PCV and nAMD (ORs values between 0.48 and 0.71, P values between 7.19 × 10−9 and 0.05; Table 8.2).

Fig. 8.2
figure 2

Forest plots of meta-analysis comparing the SNP rs10490924 at ARMS2 between PCV and nAMD. (a) allele frequencies; (b) dominant model; (c) recessive model

Full size image

There are seven studies tested the HTRA1 SNP rs11200638 in 1362 PCV and 1364 nAMD patients [3, 14, 18, 43, 44, 57, 73]. The A allele frequency was lower in PCV compared to nAMD, with a summary OR of 0.75 (95% CI, 0.67–0.84; P = 2.14 × 10−5; Table 8.2 and Fig. 8.3). The association was also statistically significant in both dominant and recessive models (OR = 0.67, P = 0.006 and OR = 0.70, P = 9.87 × 10−6 respectively, Table 8.3 and Fig. 8.3). The results of the sensitivity analysis found that the association remain significant after omitting any single included cohorts (data not shown). And there was no asymmetry on the funnel plots (Fig. 8.5). Another HTRA1 SNP, rs2672587, was also evaluated in two cohorts, and showed significant differences between PCV and nAMD (G allele; OR, 1.41; 95% CI, 1.07–1.85; P = 0.01; Table 8.2).

Fig. 8.3
figure 3

Forest plots of meta-analysis comparing the SNP rs11200638 at HTRA1 between PCV and nAMD. (a) allele frequencies; (b) dominant model; (c) recessive model

Full size image

The SNP rs77466370 in FGD6 was studied in 3318 PCV and 2457 nAMD patients from five cohorts. The summary OR for the T allele was 1.86 (95% CI, 1.48–2.35; P = 1.29 × 10−7; Table 8.2 and Fig. 8.4). The association was statistically significant in the dominant model but no in the recessive model (OR = 1.89, P = 1.52 × 10−7 and OR = 2.19, P = 0.27 respectively; Table 8.3 and Fig. 8.4). The results of the sensitivity analysis found that the association remain significant after omitting any single included cohorts (data not shown). And there was no asymmetry on the funnel plots (Fig. 8.5) .

Fig. 8.4
figure 4

Forest plots of meta-analysis comparing the SNP rs77466370 at FGD6 between PCV and nAMD. (a) allele frequencies; (b) dominant model; (c) recessive model

Full size image
Fig. 8.5
figure 5

Funnel plots of the meta-analysis comparing rs10490924 (a–c), rs11200638 (d–f) and rs77466370 (g–i) between PCV and nAMD. (a, d, g) Allele frequencies; (b, e, h) dominant model; C, F, I. recessive model

Full size image

8.4 Discussion

Genetic variants are important risk factors for both nAMD and PCV. This updated systematic review and meta-analysis compared the distribution of genetic variants between nAMD and PCV. The results showed that 57 SNPs in 20 genes had been investigated in both PCV and nAMD in the same cohorts. The pooled outcomes showed 11 SNPs at the ARMS2-HTRA1 locus and 1 SNP in FGD6 had significant differences between PCV and nAMD. The results are robust because the sensitivity test found consistency when omitting any included studies. There was no publication bias found on the funnel plots. There was no significant difference between PCV and nAMD in the remaining 45 SNPs in CFH, VEGF, C2, CFB, SKIV2L, CETP, 8p21, 4q12, ELN, LIPC, LPL, FGD6, ABCA1, ABCG1, PGF, TLR3, LOXL1, and PEDF.

The similarity and difference of PCV and nAMD attracted the interest in investigating the genetic susceptibility between them. There was a large sample size study investigated the genetic variants of 34 AMD loci for PCV and nAMD in East Asians [75]. The results showed that PCV and tAMD were highly correlated (rg = 0.69, P = 4.68 × 10−3) in genetic variants. Weaker association for PCV compared to nAMD was found at ARMS2-HTRA1 and KMT2E-SRPK2. The different association of ARMS2-HTRA1 variants between PCV and nAMD was confirmed in this meta-analysis. But KMT2E-SRPK2 was investigated in only one study and therefore no meta-analysis was performed. In 2016, an article using exome sequencing identified a rare variant, rs77466370, in FGD6 was significantly associated with PCV (OR = 2.12) but not with CNV (OR = 1.13) [70]. Our meta-analysis confirmed that most genetic polymorphisms were distributed similarly between nAMD and PCV, But also some polymorphisms had a statistically significant difference between PCV and nAMD. These results suggest that PCV and nAMD have shared the majority of genetic background, while the differences of ARMS2-HTRA1 locus and FGD6 variants may be correlated with the differences in the pathologic and clinical manifestations of PCV and nAMD. The molecular mechanisms underlying their differences in pathogenesis remain to be further investigated.

ARMS2-HTRA1 locus located at chromosome 10q26. It was one of the most strong associated locus with AMD [79, 80]. There are many SNPs in this locus and they are in strong linkage disequilibrium. ARMS2 was expressed in the mitochondria of the outer segment of photoreceptors [81]. The function of ARMS2 was suggested to be associated with loss of function of RPE [81]. HTRA1 can inhibit transforming growth factor-βin chronic inflammation [82]. In the HTRA1 transgenic mice model, retinal pigment epithelium atrophy, photoreceptor degeneration, and grape-cluster structure in choroidal vasculature were reported, which is similar to the PCV phenotype [83]. However, our meta-analysis found that the effect size of ARMS2-HTRA1 locus was weaker in PCV compared to nAMD. Further functional studies are needed to elucidate the role of ARMS2-HTRA1 locus in PCV/nAMD.

FGD6 located at chromosome 12q22. FGD6 expresses in all human tissue but has a higher level of expression in retina and choroid, especially in retinal microvascular endothelial cells. Rs77466370, c.986A > G (p.Lys329Arg), is a rare variant with the minor allele frequency of 0.02–0.03 in normal subjects. FGD6-Arg329 has a different pattern of intracellular localization from FGD6-Lys329. In vitro, FGD6 could promote endothelial cells tube formation, furthermore, FGD6-Arg329 promoted more abnormal vessel development in the mouse retina than FGD6-Lys329 [70]. These functional studies support the role of FGD6 in the pathogenesis of PCV.

8.5 Summary

In summary, we pooled the results 57 SNPs in 20 genes that had been investigated in both PCV and nAMD in the same studies. Among them, 11 SNPs at the ARMS2-HTRA1 locus and rs77466370 in FGD6 showed significant differences between PCV and nAMD, but the other SNPs had similar distribution between PCV and nAMD. Our results suggest that PCV and nAMD have shared the majority of genetic components, but the variants distributed differently between these two conditions may explain the pathogenic and clinical differences of PCV and nAMD.