Abstract
Purpose:
Age-related macular degeneration (AMD) is the leading cause of legal blindness in industrialized countries. Recently, the International AMD Genomics Consortium (IAMDGC) conducted a large genome-wide association study for AMD that included more than 50,000 individuals. A total of 34 independent loci were found to be associated with advanced AMD with genome wide significance. The next step required resolution of the associated signals and identification of the true causal variants at each locus. The current study is well powered and thus is suited to statistically refine regions of strong LD such as in the ARMS2/HTRA1 region, where the causal gene and the causal functional variant are still elusive as previous association data failed to distinguish between the two genes.
Methods:
Step-wise conditional logistic regression analysis was used to identify independent signals with genome-wide significance at the 34 AMD risk loci. Several algorithms were applied to determine the number of distinct haplotypes underlying each independent signal. Finally, two-SNP haplotypes and conditional logistic regression analyses were used for further fine mapping.
Results:
In total, we found 52 association signals at the 34 AMD loci with further evidence that the association of at least seven of these signals can be fully explained by two rather than one associated haplotype. By further fine mapping of two-SNP haplotypes, we excluded associated variants from being disease associated on the basis of recombination events in our large data set. This approach points to variants in ARMS2 to account for the AMD association signal at the ARMS2/HTRA1 locus.
Conclusions:
Our dataset was used to refine the mapping of AMD-associated variants. This has led to the identification of 14 novel disease associated haplotypes and the differentiation of associated variants through recombination events between the ARMS2 and HTRA1 genes. This approach is being extended to other AMD loci.