Our understanding about the genetic causes and therefore pathological pathways involved in the development of AMD has increased substantially in the last few years. The p.Tyr402His polymorphisms of the CFH gene, the rs11200638 polymorphism of the serine peptidase HTRA1 and the p.Ala69Ser polymorphism of the LOC387715 gene are unequivocally associated to AMD in different populations with robust risks. Our results further support these observations and demonstrate robust and highly significant adjusted risk conferred by these polymorphisms as shown by the analysis of the pooled AMD population. After dividing the patient group to exudative (wet) and non-exudative (dry) subgroups, other associations and an interesting genetic interplay was revealed. The rs2230199 polymorphism of the C3 gene contributed significant disease risk to non-exudative AMD in the simultaneous absence of homozygous CFH and HTRA1 polymorphisms. Alternative complement activation is the major biological process linked to AMD development so far. C3 plays a central role in the alternative pathway and its role in AMD pathogenesis is supported by epidemiological and experimental observations. The C3 polymorphism had no significant effect on the development of wet AMD in our study population.
Epidemiological reports on the p.Arg102Gly C3 polymorphism are usually limited to advanced cases including neovascular AMD and geographic atrophy. This approach, however, falls short of detecting any difference between the etiology of dry and wet AMD. Despite the fact that CFH and C3 are members of the same biological pathway, no evidence of a genetic interplay between these two has been demonstrated so far 
. Carriers of two risk alleles of the rs2230199 polymorphism of the C3 gene are at moderate risk to develop AMD compared to wild type subjects (OR: 1.88, 95%CI: 1.59–2.23) according to a recent meta-analysis 
. Besides that, it is well known that, drusen, the hallmark of dry AMD contains complement factors 3a and 5a among other inflammatory proteins, which therefore could play a role in drusen formation and might even have an effect on neovascularisation 
. Experimental evidence showed that suppression of the complement cascade in the retina delayed and reversed the onset of AMD in a monkey model of dry AMD 
. The effect of CFH and C3 polymorphisms are mutually exclusive on the risk of dry AMD in our study population. On the other hand, CFH has a robust effect on neovascular AMD independent of the C3 status. Since the two proteins are located in the same biological pathway, our results can be explained with the stronger effect overriding the other. Interestingly, the interaction between C3 and HTRA1 is very similar, which indicates that C3 and HTRA1 are also competing in dry AMD development. This finding underscores previous reports on the role of HTRA1 in complement regulation 
. Taken together our findings indicate a central role of C3 in dry AMD formation, however, CFH and HTRA1 genotype have an influence on the effect of the C3 p.Arg102Gly polymorphism most likely explained by their complement regulatory activity. CFI has been associated to AMD development with conflicting results. Fagerness et al. reported an OR of 0.7 for the lower risk C allele (p
) of the rs10033900 polymorphism within the CFI gene in a large population of European descent 
. Chen et al found a statistically significant association of the rs2285714 risk allele and AMD with a moderate OR of 1.31 (95%CI: 1.18–1.45, p
. Seddon et al. demonstrated that the protective effect of the hepatic lipase (LIPC) rs10468017 polymorphism on AMD pathogenesis was stronger in the presence of double wild type rs10033900 compared to the homozygous polymorphism 
. The findings of the original article by Fagerness were reflected in three case-controls studies. A report on a Japanese population demonstrated an OR of 0.28 (95%CI: 0.11–0.69, p
0.0035) for the rare homozygous CC genotype 
. The work of Ennis et al. supports the involvement of the CFI gene in AMD development by demonstrating four SNP's in the CFI gene associating to AMD, however the effect of the rs10033900 polymorphism, which had the strongest association signal in the original work of Fagerness, was not statistically significant in their cohort (p
0.135). Interestingly, the haplotype including rs10033900 polymorphism and another non-significant polymorphism together conferred statistically significant disease risk (OR
. A very well powered case-control study based on independent samples from England and Scotland failed to detect any statistically significant effect of the rs10033900 polymorphism on AMD (OR
0.95, 95%CI: 0.83–1.09, p
. Similarly to that, in the present study we could not provide any additional support for the association of the rs10033900 polymorphism and AMD. Although our sample size is not as large as some of the cited works, it is large enough to detect strong or even weak but significant associations with other well established risk factors of AMD as demonstrated by our findings on the CFH, LOC387715, HTRA1 and C3 polymorphisms. Our results are in line with studies showing no effect of the rs10033900 polymorphism. Although one cannot exclude such an association based on a single negative result, there are a growing number of publications falling short of detecting any effect of the polymorphism which clearly indicates that the association is not unequivocal and calls for further epidemiological and functional analysis to clarify the link between CFI and AMD.
No evidence of association with AMD could be demonstrated in the case of ApoE alleles. We neither could confirm previous reports on the protective effect of the E4 allele, nor could we show any risk related to the E2 allele. Two recent very-well powered epidemiological studies 
concluded on the disease causing effect of the E2 allele and the protective role of the E4 allele. Albeit these convincing studies, our results still do not fit in the line. Interestingly, we could not observe any statistically significant disease association with any of the ApoE alleles, moreover the frequency of the E2 risk allele was higher in controls than in patients, while the protective E4 allele showed similar frequencies in the wet AMD and the control group. These data indicate that ApoE is not a key player in AMD pathogenesis in the Hungarian population. This can either be explained by a geographical difference, or by other different environmental factors, probably the different smoking habits of the Hungarian population.
Candidate gene approach is widely used to identify biological pathways and genes involved in pathological processes. Based on their biological functions, we directly tested for disease association of common polymorphisms in the FXIII, Gas6 and in one of the Gas6 receptor genes, MerTK. Blood coagulation factor XIII is a plasma transglutaminase that cross links adjacent fibrin chains in the final step of the coagulation cascade. In addition, FXIII is known to participate in wound healing, tissue remodeling and embryo implantation at least partially through its proangiogenic effect 
. The cardioprotective p.Val34Leu polymorphism of the blood coagulation factor XIII has been associated with accelerated thrombin activation, recurrent subconjunctival haemorrhage, intracerebral haemorrhage, and a decreased immune reaction in humans 
. Most importantly, since angiogenesis plays a crucial role in neovascular AMD, we found it reasonable to investigate if the polymorphism had any effect on AMD formation. However, we failed to detect any association of AMD and the polymorphism in our present study, indicating a neutral polymorphism in relation to both dry and wet AMD development.
Gas6, a Protein S structural homologue, has important functions in the regulation of angiogenesis, cell migration and proliferation. Its common polymorphism Gas6 c.834+7G>A is associated with decreased risk of cardiovascular diseases. Its receptors (TAM) play a crucial role in phagocytosis of apoptotic cells in the immune, nervous, and reproductive systems. These functions make Gas6 and its receptors a potential candidate involved in both dry and wet AMD development. In the pooled early and late AMD population the Gas6 polymorphism showed no statistically significant association to AMD. When analyzing dry and wet subgroups separately, no association with the dry AMD could be demonstrated; however a protective effect was detected in the wet AMD subgroup (p
0.04). As formation of new vessels is a hallmark of wet AMD, it is likely that it is this function of Gas6 that is important in the context of the disease. Gas6 is able to inhibit VEGF-A signaling through Axl activation of SHP-2 phosphatases 
. Recently, a similar inhibitory effect has been shown in metastasis induced angiogenesis 
. This finding suggests a role of Gas6 in controlling pathological angiogenesis, and that c.834+7G>A polymorphism of the GAS6 gene, which has been linked to disease repeatedly 
, could have a direct effect on the function of the gene through an unknown mechanism. Future epidemiological and functional studies are required to further evaluate the effect of the GAS6 polymorphism on AMD development. It is to be noted, that no formal correction for multiple testing was applied in the multiple logistic regression analysis, however, this does not invalidate the conclusion that the suggested effect of the GAS6 polymorphism is worthy of further investigations.
In our case-control study we confirmed previous findings on the association of AMD and genetic polymorphisms in the CFH, LOC387715 and HTRA1 genes, however could not demonstrate any association with the CFI gene. More interestingly, we found a genetic interplay of CFH, HTRA1 and C3 genes, showing that the C3 polymorphism is a major contributor of dry AMD in the absence of the other polymorphisms, while has no effect on wet AMD development in our population. This result indicates that C3 plays a critical role in dry AMD rather than wet AMD pathogenesis, and its effect can be overdriven by other known genetic risk factors in the CFH and HTRA1 genes. We detected a protective effect of a common Gas6 c.834+7G>A polymorphism on wet AMD formation. Our results suggest a possible new player and novel genetic interactions in AMD pathogenesis, however our data should be confirmed in other populations, especially in the case of the Gas6 polymorphism.