Table 1.

SNPs showing the strongest association with AMD

SNP	Allele		Risk allele frequency	lod, 1 df	P	Penetrances			λsib	RR1	RR2	Logistic regression, including covariates for the following
	Risk (−)	Nonrisk (+)				+/+	+/−	−/−				rs10490924	rs11200638
rs10490924	T	G	0.284	28.12	5.3 × 10−30	0.092	0.245	0.652	1.275	2.656	7.054	NA	6.27 × 10−8
rs3750847	T	C	0.247	18.95	9.4 × 10−21	0.114	0.264	0.61	1.191	2.313	5.347	0.067	0.003
rs3793917	G	C	0.243	17.75	1.6 × 10−19	0.116	0.265	0.603	1.182	2.275	5.174	0.051	0.002
rs3750848	G	T	0.246	17.45	3.1 × 10−19	0.117	0.263	0.594	1.178	2.253	5.080	0.103	0.008
rs11200638	A	G	0.248	17.36	3.8 × 10−19	0.118	0.262	0.577	1.167	2.207	4.872	0.241	NA
rs932275	A	G	0.235	14.75	1.7 × 10−16	0.12	0.269	0.601	1.172	2.238	5.007	0.140	0.050
rs2672587	G	C	0.254	14.19	6.3 × 10−16	0.123	0.256	0.534	1.142	2.086	4.351	0.137	0.124
rs2253755	G	A	0.312	8.54	3.6 × 10−10	0.132	0.23	0.398	1.078	1.736	3.014	0.027	0.017
rs2248799	T	C	0.494	7.97	1.4 × 10−9	0.112	0.188	0.316	1.066	1.680	2.820	0.307	0.421
rs10510110	C	T	0.51	5.75	2.7 × 10−7	0.116	0.187	0.302	1.055	1.612	2.599	0.347	0.531

For each SNP, the risk allele (−) is defined as the allele with increased frequency in affected individuals. Evidence for association, as evaluated by the LAMP program (26), is summarized through the risk allele frequency in the population (estimated using a parametric model that, in effect, weights cases and controls according to the estimated disease prevalence); lod (logarithm of odds) score (log₁₀ likelihood ratio statistic comparing model with and without association); P value; and a series of estimated penetrances for nonrisk homozygotes (+/+), heterozygotes (+/−), and risk allele homozygotes (−/−); genotype relative risks RR1 and RR2 (which are computed by comparing estimated penetrances in heterozygotes and risk-allele homozygotes, respectively, and those for nonrisk homozygotes); and sibling recurrence risks λ _sib. The λ _sib measure characterizes the overall contribution of a locus to disease susceptibility. It quantifies the increase in risk to siblings of affected individuals attributable to a specific locus (48). For example, λ_sib of 1.27 signifies that the SNP could account for 27% in risk of AMD for relatives of affected individuals. Association analysis using a simple χ ² statistic produced similar results. The last two columns summarize P value results of logistic regression analysis, including either rs10490924 or rs11200638 as covariates. As suggested in ref. 40, missing genotypes were imputed prior to the sequential analyses reported in the last two columns.