Comparison of effect sizes for early AMD from this study versus published effect estimates for late AMD.

aSuperscript shows reference for the largest study reporting genome-wide association of the relevant SNP with late AMD, from which the “Late AMD” effect estimates were derived:1Chen et al, 2010 11.2Yu et al, 2011 15.3Klein et al, 2005 12.4Kopplin et al, 2010 13.5Arakawa et al, 2011 10.6Neale et al, 2010 14.bNCBI Human Genome Build 36.3 coordinates;cEffective allele;dFrequency of the effective allele;eSummary meta-analysis regression coefficient, indicating the overall, estimated change in log(odds) associated with each additional copy of the effective allele;fEstimated odds ratio and 95% confidence interval for each additional copy of the effective allele, based on fixed-effects meta-analysis of European-ancestry cohorts;gP-value associated with the estimated OR;hNR: not reported;iP-value from test of heterogeneity of regression coefficients between early and advanced AMD. The threshold for study-wise significance was 0.0036, after accounting for multiple tests. Significant results are shown in bold. Heterogeneity could not be assessed for SNPs with no published confidence interval for the late AMD effect estimate;jRatio of regression coefficient for advanced vs early AMD, formulated as Betaadv/Betaearly.Notes: This study did not have data and could not assess association for additional published SNPs rs4711751 in VEGFA and rs11200638 in HTRA1.