Unraveling the genetics of AMD has been difficult. Although a few rare variants have been associated with AMD,
27 virtually all the genetic effect remains to be explained. Two complementary approaches can be used to tackle this problem. The genome screen approach has identified some common chromosomal regions,
13 and work is ongoing to identify these genes in this and other datasets.
17 59 Because of the underlying complexity, however, linkage analyses are unlikely to identify the locations of all relevant genes, and thus the examination of functional candidate genes is a useful complementary approach.
We examined 35 SNPs in the eight functional candidate genes. These SNPs were chosen initially for availability of assays, informativeness, and spacing across the gene. For this initial screen of candidates, we did not attempt to perform a comprehensive analysis of all polymorphisms in each gene. Our LD results confirm the emerging idea of haplotype blocks, with the SNPs in several genes (IL1A, A2M, LRP6, MGST1, and CKB) in strong LD across the entire gene. VEGF, VLDLR, and DCP1 each had two relatively independent blocks of SNPs. These data suggest that we effectively reduced the number of independent SNP results from 35 to 11.
Of the genes chosen for study here,
A2M did not demonstrate any nominally significant results.
IL1A showed nominally significant results only in the case-control analysis.
DCP1 showed nominally significant results only in the family-based association analysis. Two genes,
MGST1 and
CKB, generated modestly interesting LOD scores (>1.0), although none reached levels proposed as suggestive or significant.
60 The lack of consistently positive results for these genes strongly suggests that they do not play a significant role in the etiology of AMD. However, our dataset does not have sufficient power to detect more modest effects of these genes, and we cannot exclude the possibility that they may have a small effect.
Of more interest are the results for LRP6, VLDLR, and VEGF. LRP6 generated a modestly interesting HLOD score of 1.14 in the multiplex families, and independently generated the strongest association results in the case-control analysis (P = 0.004 in the more severe grade 5 clinical group). Although the linkage results for VLDLR were unimpressive (HLOD = 0.34), it generated a nominally significant result in the family-based association analysis (P = 0.03) and an independent significant result in the case-control dataset (P = 0.01). VEGF had a maximal LOD score of 1.32, the strongest family-based association result (P = 0.001, grade 3, 4, and 5) and a moderate case-control association result (P = 0.02, grade 5).
The nominally significant results must be interpreted cautiously, since we genotyped multiple SNPs and performed the analyses under two different clinical models. Given the LD results, we have effectively studied only 11 independent blocks of polymorphisms along with two clinical groups, resulting in 22 tests per dataset. The most conservative correction for such multiple comparisons (Bonferroni) would suggest an adjusted P-value of 0.002.
In this light, the
LRP6 case-control results are only on the border of significance. They are, however, independently supported by the linkage results, and thus this gene must remain of some interest. Functionally,
LRP6 is a low-density lipoprotein receptor involved in vasculature remodeling pathways.
48 Neither the family-based or case-control
VLDLR results survive this correction. However, the functional role of
VLDLR as a cell surface receptor for Reelin and the nominal results in two independent datasets suggest that this gene should be examined further. Finally, the family-based
VEGF result remains significant even with this conservative correction. Combined with the nominal result in the case-control dataset, the interesting genetic linkage results, and its functional role in vascular growth and regeneration,
51 this is perhaps our most strongly implicated gene in the etiology of AMD.
The authors thank all the participants and their relatives who generously participated in the study. The authors thank Melissa Allen for diligently genotyping the markers, and Ruth Domurath, Molly Klein, Jennifer Caldwell, and Katie Haynes for their tireless work in ascertaining data on many of the families used in this study. They also thank the following clinics and clinicians for referring individuals to the study: Southern Retina, L.L.C. (Charles Harris, MD, Savannah, GA); Vitreo-Retinal Surgeons (Michael E. Duan, MD, and Christopher J. Devine, MD, Cincinnati, OH); Georgia Retina, P.C. (Atlanta, GA); and The Retina Group of Washington (Washington, DC). The authors also thank Don Gass for his advice and mentorship.