Recent studies have suggested that CNVs play important roles in the development of inheritable diseases 
. Although less abundant than SNP, CNV seems to account for more nucleotide variation due to its sheer size 
. By spanning thousands of bases, CNVs often encompass functional DNA sequences. A recent comparison of the relative impact of SNP and CNV on gene expression identified that a substantial proportion (~18%) of gene expression variability might be attributable to known CNVs 
In this study, we found three out of four candidate regions showed CNVs in AMD patients or control subjects. CNV frequencies in some regions were found to differ from those of our previous screening study. For instance, we previously found loss of CFH in 5 out of 30 individuals. In this study, however, none of the AMD patients and control subjects demonstrated CN loss at this site. Ethnic differences may be responsible for this discrepancy. In case of CFH loss, every CNV loss was discovered in the Chinese and Japanese, yet not in Koreans. The detection of CNV could be platform-dependent as well. A previous study discovered CNVs using array CGH, whereas this study used TaqMan probes based on real-time PCR. Furthermore, we used the CopyCaller program and maximum-likelihood algorithm to determine CN or predicted CN, and set the most frequent CN at 2. However, if there is a region with more than 2 copies in the actual human genome, its CN determination would have a number of false results. Regions with high frequencies of CNV, such as VLDLR in our study, could be suspected as multi-copy loci.
Considering factors mentioned above, the issue of false CNV may be the first hurdle to overcome for this kind of association study. In addition to the duplication of experiments, we repeated experiments for samples with CN gain and loss to avoid false positive CNV calls, and confirmed whether they were consistent with previous experiments. CN gain or loss of VLDLR, however, was excluded from these re-experiments, since this region appeared to be highly CNV-prone in both patient and control groups. CN gains and losses of other three regions shown in were all confirmed by two separate experiments. Besides, all the plates were designed as almost an equal number of patient and control samples since false calls might occur randomly in each plate.
PCN would be the most apparent and decisive value showing the CNV status of each individual if RCV is converted well into PCN without error. However, we found several cases showing borderline RCVs, which make it difficult to determine whether they have two copies or not. To overcome these kinds of false negative or positive calls, we also compared the mean RCV between patient and control groups and showed the possible association of candidate genes with the AMD development.
Vascular endothelial growth factor (VEGF) is a major molecular mediator of neovascularization. Intraocular VEGF expression was found to be increased in neovascular AMD patients, which led to the development of VEGF inhibition therapies with anti-VEGF antibodies for neovascular AMD 
. However, studies analyzing SNP did not show consistent results 
. In this study, CN loss in VEGFA
was discovered in 3 control subjects, which had not been reported in the Asian or Caucasian populations before. Although our study analyzed a relatively large number of AMD patients and controls, further investigations should be conducted in other ethnic groups to confirm the possible effects of CNV on VEGFA
gene in AMD development. In addition, because the CNV in VEGFA
gene appeared to be rare in the Korean population, larger sample sizes are required for further studies.
gene encodes a member of a family of serine proteases expressed in both mouse and human retinas, and its expression in human fibroblasts increases with aging 
. Over-expression of ARMS2/HTRA1
alters the integrity of the Bruch's membrane, favoring the invasion of choroid capillaries across the extracellular matrix, as occurs in wet AMD 
. In this study, on the contrary, the mean RCV at ARMS2/HTRA1
was lower in AMD patients compared to control subjects. Since there was no CN gain or loss predicted, studies to elucidate the genuine CNV status of ARMS2/HTRA1
in AMD patients, and its possible effect on ARMS2/HTRA1
expression would be needed.
Genes encoding complement factors have been identified as AMD susceptibility loci with convincing statistical evidence. They include complement factor H gene (CFH
) on chromosome 1q32 
, complement component 3 gene on 19p13 
, and 2 neighboring genes on 6p21 (complement factor B and complement component 2) 
. This study discovered CN loss of CFH
gene in one AMD patient but it could not find any association of CFH
variation with AMD. Other complement factor genes and complement factor related regions may find significant genetic variations in complement factor related genes in AMD patients.
Very low density lipoprotein receptor (VLDLR) is involved in lipid transportation and chronic inflammation through the Wnt pathway 
gene knockout (VLDLR
−/−) mice were shown to develop sub-retinal neovascularization 
and the features of neovascularization resemble those of retinal angiomatous proliferation in human — a subtype of AMD 
. In an association study, variations (rs10967213, rs2290465) of VLDLR
showed associations with AMD prevalence 
. In this study, the CN gain or loss of VLDLR
was very common in control subjects and in AMD patients. It may be due to its regional complexity in the human genome, and therefore this region should be interpreted with caution.
In this study, we selected target CNVs out of Asian-specific CNVs discovered in our previous study. Consequently, several genes which have been thought to play important roles in AMD pathogenesis and widely studied in other ethnic groups, such as C2/BF, TIMP, CFHR1 and CFHR3, were not included in our target regions.
In conclusion, the outcome of this study suggests that some CNVs might be associated with AMD development, although the actual action of these CNVs in vivo remains unelucidated. Further CNV-oriented research is required to improve our understanding of the mechanism that underlies AMD.