This study is the largest independent AD GWAS to date, and, by including samples through collaboration and public resources, we augmented the overall available sample from 8186 subjects to an overall sample of 16 087. Other strengths include the following: (1) two analytical approaches were used, including an ordinal trait approach (to increase power and correct for other substance dependence) and case–control using exposed controls only (less power but increased signal-to-noise ratio); (2) multiple replication samples were used; and (3) functional evidence was presented for the novel chromosome 2 risk locus. There are also several limitations. We had limited phenotypic data available on the German (European-ancestry) replication sample, and limited follow-up SNP genotyping was possible. Also our replication data sets include proportionally fewer AA subjects than the GWAS sample, limiting our ability to identify AA-specific risk loci. Although our sample and the SAGE sample were evaluated in similar ways, they differ in that the majority of our sample was recruited for studies of drug, rather than alcohol, dependence. We adjusted for these comorbidities (in both samples) analytically, however, the heterogeneity may have decreased statistical power.
