Table 4 shows the relationship between AMD phenotypes and genotypes for
ABCA1 and
LIPC based on multivariate logistic models controlling for all genetic, demographic, and behavioral risk factors. The TT genotype of
LIPC (rs10468017) was protective against intermediate drusen (OR [95% CI] = 0.52 [0.30–0.89];
P = 0.018), large drusen (OR [95% CI] = 0.55 [0.31–0.95];
P = 0.033), and advanced AMD (OR [95% CI] = 0.37 [0.19–0.71],
P = 2.8 × 10
−3). The CT genotype of
LIPC (rs10468017) was not significant, suggesting that the genetic effect of this locus best fits a recessive genetic model. Using the collapsed genotypes of CC and CT as the reference genotype, we found that the TT homozygous genotype of
LIPC was consistently associated with a reduced risk of intermediate, large drusen, and advanced AMD. Both homozygous and heterozygous genotypes with the T allele of
ABCA1 (rs1883025) were protective against intermediate drusen (TT:
P = 0.011, OR [95% CI] = 0.48 [0.27–0.85]; CT:
P = 0.037, OR [95% CI] = 0.70 [0.50–0.98]), large drusen (TT:
P = 3.1 × 10
−3, OR [95% CI] = 0.41 [0.23–0.74]; CT:
P = 5.8 × 10
−3, OR [95% CI] = 0.62 [0.44–0.87]), and advanced AMD (TT:
P = 3.5 × 10
−3, OR [95% CI] = 0.35 [0.17–0.71]; CT,
P = 6.8 × 10
−3, OR [95% CI] = 0.59 [0.40–0.87]). The genetic effect of the T allele of
ABCA1 (rs1883025) seems to fit an additive genetic model. Although the
P values for both the
LIPC and the
ABCA1 genes were higher in the GA group than in the NV group, possibly because of the smaller sample size in the GA group, the estimates of the effects of these genes on GA and NV were similar.