For survival analysis, we excluded cases with AAO below 60 and cases with prevalent stroke. For CSF analysis, individuals under age 45 years were removed because prior studies have demonstrated that the relationship between CSF Aβ42 levels and age appears to differ in individuals below 45 years vs. those above 45 years69. Of the remaining individuals in both analyses, we excluded individuals who had > 5% missing genotype rates, who showed a discrepancy between reported sex and sex estimated on the basis of genetic data, or who showed evidence of non-European ancestry based on principal component analysis using PLINK1.970. We identified unanticipated duplicates and cryptic relatedness using pair-wise genome-wide estimates of proportion identity by descent (IBD) using PLINK. When duplicate samples or a pair of samples with cryptic relatedness was identified, the sample with the lower genotyping call rate was removed. We excluded potentially related individuals so that all remaining individuals have kinship coefficient below 0.05. Finally, we excluded individuals with missing disease status, age or gender information.
