VEGFA Haplotype and VEGF-A and VEGF-R2 Protein Associations with Exudative Age-Related Macular Degeneration

Our study aimed to reveal the associations between VEGFA SNPs (rs1570360, rs699947, rs3025033, and rs2146323), their haplotypes, VEGF-A and VEGF-R2 serum concentrations, and early and exudative AMD. A total of 339 subjects with early AMD and 419 with exudative AMD groups, and 374 healthy subjects, were genotyped for four VEGFA SNPs (rs1570360, rs699947, rs3025033, and rs2146323). VEGF-A and VEGFR-2 serum concentrations were measured in exudative AMD and controls. The results revealed that rs3025033 G allele was significantly associated with lower odds of exudative AMD under the dominant model (OR = 0.67; 95% CI: 0.49–0.80; p = 0.0088) and additive (OR = 0.7; 95% CI: 0.54–0.90; p = 0.0058) models after Bonferroni correction. In the female group, rs3025033 AG genotype was associated with exudative AMD under the codominant model (OR = 0.57; 95% CI: 0.37–0.87; p = 0.009) and G allele under the dominant (OR = 0.55; 95% CI: 0.37–0.82; p = 0.0032) and additive models (OR = 0.60; 95% CI: 0.42–0.84; p = 0.0028). Haplotype analysis revealed that individuals carrying rs1570360, rs699947, rs3025033, and rs2146323 haplotype A-A-G-A had decreased risk of exudative AMD (OR = 0.46, 95% CI: 0.23–0.90; p = 0.023). The VEGF-A and VEGF-R2 serum concentrations did not differ between study groups; we found that patients with exudative AMD carrying at least one C allele at rs699947 have statistically significantly higher VEGF-A serum concentrations compared to AA genotype carriers (485.95 (945.93) vs. 194.97 (-), respectively, p = 0.046). In conclusion, we found that VEGFA rs3025033 and haplotype rs1570360A-rs699947A-rs3025033G- rs2146323A play a protective role for exudative AMD in the Caucasian population. Furthermore, rs699947 is associated with elevated VEGF-A serum concentrations in exudative AMD.


Introduction
Age-related macular degeneration (AMD) is widely described as a multifactorial, progressive, neurodegenerative disease of the macula, causing loss of visual functions leading to blindness. The macula is a part of the retina where the photoreceptors, responsible for visual acuity and colour perception, are concentrated [1]. Most eye function impairments are associated with age-related alterations and, together with the other risk factors such as environmental and genetic factors, lead to severe eye conditions [2]. Exponential population ageing is a recent worldwide problem affecting human health, including eye diseases. It is known that AMD affects about 10% of people older than 65 years and more than 25% of people older than 75 years in developed countries [2]. The previous study's projection shows that the number of people with early AMD will increase up to 21.5 million and late AMD up to 4.8 million in Europe [3], while worldwide the numbers are expected to increase dramatically, even up to 288 million in 2040 [3].

AMD Groups
All the AMD patients underwent optical coherence tomography (OCT), and optical coherence tomography angiography (OCT-A) was performed in patients suspected of having late AMD after the OCT examination. Exudative AMD was diagnosed in one eye of the study subjects.
Age-Related Eye Disease Study (AREDS) classification was used for AMD diagnosis and has been described previously [25].
Early AMD was defined as a combination of multiple small drusen and several intermediate drusen (63-124 µm in diameter) or retinal pigment epithelial abnormalities.
Intermediate AMD was characterized by the presence of extensive intermediate drusen and at least one large (giant) druse (≥125 µm in diameter) or geographic atrophy (GA) not involving the center of the fovea.
Advanced AMD was characterized by GA involving the fovea and/or any of the features of neovascular AMD.

The Control Group
Subjects without ophthalmologic pathologies and patients after senile cataract surgeries (without any other ocular comorbidities) were included in the control group. The following exclusion criteria were also described in previous publications [26]: (1) unrelated eye disorders, e.g., high refractive error, cloudy cornea, lens opacity (nuclear, cortical, or posterior subcapsular cataract) except minor opacities, keratitis, acute or chronic uveitis, glaucoma, or diseases of the optic nerve; (2) systemic illnesses, e.g., diabetes mellitus, malignant tumors, systemic connective tissue disorders, chronic infectious diseases, hypertension, coronary artery disease, stroke or conditions following organ or tissue transplantation; (3) ungraded colour fundus photographs resulting from obscuring the ocular optic system or because of fundus photograph quality.

DNA Extraction
DNA extraction was carried out at the Laboratory of Ophthalmology, Neuroscience Institute, LUHS. The DNA was extracted from 200 µL peripheral venous blood samples utilizing silica-based membrane technology, using a genomic DNA extraction kit (GeneJET Genomic DNA Purification Kit, Thermo Fisher Scientific, Vilnius, Lithuania), based on the manufacturer's recommendations.

SNP Selection
In the present study, two tag SNPs (intronic variants rs3025033A/G and rs2146323A/C) covering two haploblocks ( Figure 1) (rs833068 have already been analyzed in our previous research study) [26] were selected from the CEU population using the public HapMap database. The pairwise option of the online Tag SNP tool was used with the following settings: r 2 = 0.8 set and the minimum number of SNPs tagged by each tag SNP was 2. Two other VEGFA promoter polymorphisms, −2578C/A (rs699947) and −1154G/A (rs1570360), were selected additionally based on the previous inconsistent results [27] and potential multiple SNP associations with the AMD [19].

Genotyping
The genotyping of four VEGFA gene polymorphisms, rs1570360, rs699947, rs3025033, and rs2146323, was carried out at the Laboratory of Ophthalmology, Neuroscience Institute, LUHS. The identification of all single-nucleotide polymorphisms was performed on a "StepOnePlus" real-time PCR quantification system (Thermo Fisher Scientific, Singapore) using predesigned TaqMan ® Genotyping assays (Thermo Fisher Scientific, Pleasanton, CA, USA) according to the manufacturer's recommendations. Genotyping results were obtained using Genotyping program on the StepOne software.

Quality Control of Genotyping
A total of 5% randomly chosen samples were repetitively genotyped for all four SNPs to confirm the same rate of genotypes from initial and repetitive genotyping.

Statistical Analysis
Statistical analysis was performed using the SPSS/W 27.0 software (Statistical Package for the Social Sciences for Windows, Inc., Chicago, IL, USA). Continuous data (age, protein serum level data distributions were evaluated for normality by the Kolmogorov-Smirnov test. Continuous variables presented as median with interquartile range (IQR) based on data distribution. The Mann-Whitney test was used to compare two groups for non-normally distributed data.
Categorical data (gender, genotype, and allele distributions) are presented as absolute numbers with percentages in brackets and compared between the early, exudative, AMD, and control groups using the chi-square (χ 2 ) test. Hardy-Weinberg equilibrium was evaluated to compare the observed and expected frequencies of VEGFA rs1570360, rs699947, rs3025033, and rs2146323 using χ 2 test in the control group as well.
The impact of SNPs on early and exudative AMD was evaluated using binomial logistic regression analysis. Results are presented as odds ratios (OR) with 95% confidence interval (CI) and adjusted by covariate effect for age in the exudative AMD groups. Logistic regression analysis results were expressed as genetic models (codominant: heterozygotes vs. major allele homozygotes and minor allele homozygotes vs. major allele homozygotes; dominant: minor allele homozygotes and heterozygotes vs. major allele homozygotes; recessive: minor allele homozygotes vs. major allele homozygotes and heterozygotes; overdominant: heterozygotes vs. major allele homozygotes and minor allele homozygotes); the additive model was used to evaluate the impact of each minor allele on AMD: major allele homozygotes vs. heterozygotes vs. minor allele homozygotes. The best genetic model selection was based on the Akaike information criterion (AIC); therefore, the best genetic models had the lowest AIC values. We introduced an adjusted significance threshold for multiple comparisons alpha = 0.0125 (0.05/4, since we analyzed four SNPs in the VEGF-A gene) [28].
Haplotype analysis was performed in the early AMD and control groups, and exudative AMD and control groups separately, using online SNPStats software (https://www. snpstats.net/snpstats/ (accessed on 15 December 2021)) [29]. Linkage disequilibrium (LD) analysis was assessed by D' and r 2 measures. The associations between the haplotypes and different AMD forms were calculated by logistic regression and presented as ORs and 95% CI and values adjusted for age in exudative AMD analysis. Haplotypes with less than 1% frequencies were pooled into one group and described as "rare". A two-sided test with a value less than 0.05 was considered statistically significant. Graphs were performed using GraphPad Prism version 9.0.0 for Mac, GraphPad Software, San Diego, CA, USA, www.graphpad.com (accessed on 21 February 2022).
Allele frequency analysis showed that G allele at rs3025033 was statistically significantly less frequent in the exudative AMD group than controls (16.9% vs. 22.3%, respectively, p = 0.007) ( Table 2).
While the logistic regression analysis showed significant associations between rs2146323 and exudative AMD in females, the results did not survive Bonferroni correction (Supplementary Material, Table S2). No associations were found between rs2146323 and early AMD in females (Supplementary Material S1, Tables S1 and S2) nor males (Supplementary Material, Tables S3 and S4).
We also compared the genotype and allele frequencies between females with early AMD and exudative AMD and males with early AMD and exudative AMD. We found that allele G at rs3025033 was less frequent in females with exudative AMD than in females with early AMD (14.4% vs. 20%, respectively p = 0.019). Still, the result did not survive Bonferroni correction (Supplementary Material, Table S1).

VEGF-A and VEGF-R2/KDR Serum Concentration Analysis
Serum protein concentrations were measured in 20 patients with exudative AMD before treatment and 21 control group samples. The Control group for VEGF-A and VEGF-R2/KDR serum concentration measurement consisted of subjects considering the age and gender distributions based on the exudative AMD group.
We compared the VEGF-A serum concentrations between exudative AMD and control groups but did not find a significant difference (422.674 (677.02) vs. 615.489 (425.49), respectively, p = 0.424) (Figure 2). We also compared the VEGF-R2/KDR serum concentrations between exudative AMD and control groups, but there was also no statistical difference (12,

VEGF-A and VEGF-R2/KDR Concentrations by VEGFA Genotypes
We also performed VEGF-A and VEGF-R2/KDR serum concentration and VEGFA rs1570360, rs699947, rs302503, and rs2146323 genotype association analysis and found that patients with exudative AMD carrying at least one C allele at rs699947 have statistically significantly higher VEGF-A serum concentrations compared to wild-type allele A homozygous genotypes carriers (485.95 (945.93) vs. 194.97 (-), respectively, p = 0.046) ( Table 7). However, there were only two subjects with the AA genotype.

Discussion
This study analyzed four SNPs in the VEGFA gene and their associations with early and exudative AMD. Study results revealed that the G allele at rs3025033 was significantly associated with lower odds of exudative AMD. We also found that these associations with exudative AMD remained only in females but not in males, suggesting the potential gender role in AMD development. Furthermore, the differences of rs3025033 allele G frequencies were observed between females with exudative AMD and females with early AMD, but the results did not survive Bonferroni correction. Moreover, associations between rs2146323 and exudative AMD in females were found, but the results did also not survive Bonferroni correction.
A haplotype of VEGFA SNPs analysis revealed that individuals carrying rs1570360, rs699947, rs3025033, and rs2146323 haplotype A-A-G-A had decreased risk of exudative AMD, showing a protective role of this haplotype. No studies have previously included all these SNPs in AMD analysis. Only Mori et al. (2010) analyzed VEGFA −116A, rs1570360 associations with AMD, but did not reveal significant results [27]. Another three studies examined rs1570360 associations with exudative AMD treatment efficacy but not with the AMD occurrence [30][31][32].
This SNP was also studied in patients with AMD, but statistically significant differences in the genotypic distribution of VEGFA rs699947 were not found, and any significant associations were revealed [20,27,37,38]. Scientists also tried to find the differences between the AMD subtypes comparing neovascular and atrophic AMD groups but did not reveal significant results [39].
Previous results were found in much smaller study groups. Still, our study confirmed those findings in bigger samples groups, including 339 subjects in early AMD and 419 in the exudative AMD groups, and 374 subjects in the control group.
Other research studies have analyzed rs699947 associations with anti-VEGFA treatment [30,31,[40][41][42] or photodynamic therapy response [32,43,44] but conflicting results suggest that other risk factors [45] or SNP combinations may be associated with the AMD treatment as well [46,47]. Our analysis of SNPs and VEGF-A and VEGF-R2 serum concentrations revealed that carriers of at least one C allele at rs699947 have statistically significantly higher VEGF-A serum concentrations compared to wild-type allele A homozygous genotype carriers (485.95 (945.93) vs. 194.97 (-), respectively, p = 0.046). Bulgu et al. (2014) included in their study an intronic VEGFA variant rs3025033 and genotyped it for 82 AMD patients and 80 controls. Unfortunately, about 98% of genotypes were determined as AA, so no further statistical analysis was performed [48]. Li et al. (2021) analyzed the rs3025033 effect on VEGF165b protein production but did not reveal significant results; on the other hand, they showed that this SNP promoted cell proliferation in human retinal vascular endothelial cells (hRVECs) [49].
Another study performed by Immonen et al. (2010) analyzed the associations between rs3025033 and photodynamic therapy response, but no associations were revealed [44].
In contrast, the frequencies of the VEGFA +5092, rs2146323 were significantly different in photodynamic therapy non-responders and responders [44]. However, similar frequencies of this SNP were found between the exudative AMD patients and controls [37] and any AMD patients and controls [48]. Furthermore, it was shown that rs2146323 C allele was protective against dry AMD, and the allele A was associated with the disease. Moreover, there was a significant difference in the genotype frequencies of this SNP between the wet type of AMD and dry type AMD [48]. Our study results showed associations between rs2146323 and exudative AMD only in females, but these results did not survive when we applied strict Bonferroni correction. Furthermore, we did not find any associations between this SNP and VEGF-A or VEGF-R2 serum concentrations.
Other widely studied SNPs show significant associations with the AMD and highlight the possible VEGFA rs1413711 and rs833061 polymorphisms contributions to AMD susceptibility [50]. Bulgu et al. (2014) found that the presence of ancestral allele (G) in rs1413711 was protective for all AMD patients, and the AA genotype was a risk factor for AMD and even a highly increased risk factor for dry AMD [48]. Opposite results were shown in another study: SNP +674, rs1413711 CC genotype was significantly associated with a higher risk of exudative AMD [19]. The conflicting results of these studies may be explained by differences in the size of study groups and populations (Turkish and Caucasian of Northern European origin). Habibi et al. (2014) showed that VEGFA +405, rs2010963 CC and VEGFA +936, rs3025039 TT genotype frequencies were higher in Tunisian AMD patients than in controls [51].
These results confirmed the previous associations between VEGFA +936, rs3025039, and wet AMD in the Japanese population [52]. Deeper analysis showed that genotype TT for rs3025039 was associated with elevated VEGF-A protein serum levels [20].
Moreover, an extensive analysis of VEGFA promoter and gene polymorphisms showed that SNPs +674, +4618, +5092, +9162, and +9512 their haplotypes, CTCCT and TCACC, were associated with a 15-fold increased risk of exudative AMD and the promoter SNPs −460T, −417T, −172C, −165C, −160C, −152G, −141A, −116A, +405C haplotype was associated with about an 18-fold greater risk [19]. Two of these polymorphisms were included in our study and, together with another two SNPs, revealed a protective haplotype for exudative AMD. It is important to elucidate the SNP combinations and their role in AMD to understand the pathogenesis and possible treatment strategies better.
VEGF-A serum levels, as well as VEGFA SNPs, are widely studied in AMD patients. While we did not find any statistical differences in VEGF-A serum levels between the exudative AMD patients and controls, we confirmed results from several other studies which included total AMD patients or only exudative AMD patients, consisting of 27 to 71 samples per group in different populations [53][54][55][56][57][58][59][60]. Further studies revealed significantly elevated VEGF-A levels in exudative AMD patients compared to controls [61,62]. Significantly elevated VEGF-A levels in total AMD patients compared to controls were also found in a few studies [20,51,63].
VEGF-R2 protein associations are not so widely analyzed as VEGF-A, but conflicting results were also found. Örnek et al. (2016) found that decreased VEGFR-2 serum levels were associated with both dry and wet type AMD. Furthermore, negative correlations between VEGFR-2 with foveal retinal thickness in AMD patients and a significant positive correlation with subfoveal choroidal thickness revealed the possible VEGF-R2 role in AMD development [64]. Another two studies showed opposite associations and found elevated VEGFR2 protein levels in patients with exudative AMD group than in the controls [65,66], suggesting the role of VEGFR-2 in the pathogenesis of AMD.

Conclusions
In conclusion, we found that rs3025033 polymorphism of VEGFA and the haplotype rs1570360A-rs699947A-rs3025033G-rs2146323A can play a protective role for exudative AMD in our population. Furthermore, the C allele at rs699947 is associated with elevated VEGF-A serum concentrations in exudative AMD patients. Since the genetic variations were associated with exudative AMD, our findings need to be replicated in additional studies. While the main strength of our study was a large number of early and exudative AMD patients, further studies are still needed to investigate the pharmacologic role of these angiogenesis-related markers in AMD therapy.