An unique feature of our study was the extensive sensitivity analyses that provided more insight into the reported genetic and epidemiological correlations between MDD, loneliness, and CAD. Genetic correlations alone can be difficult to interpret because they may be influenced by phenotypic hitchhiking. For example, if MDD is comorbid with heart disease, then a GWAS of MDD based on prevalent cases will also ascertain patients with heart disease, which could then induce a genetic correlation between MDD and an independent GWAS of heart disease. Any GWAS that ascertains prevalent cases will be similarly enriched for comorbidities. In the absence of large GWAS meta-analyses of incident cases, we tackled this problem by controlling for multiple clinical risk factors, known CAD risk variants, and genetic correlations, resulting in strong evidence that genes conferring risk for MDD and loneliness exert pleiotropic effects on CAD.
