only rs10014685 was present in independent cohorts, but it was not significant in either (deCODE p = 0.89; UHS p = 0.36). Examining a recently reported GWAS of prescription opioid misuse (POU)49, we see a genetic correlation between OA and POU (rg = 0.74, p = 2.24 × 10–12) and extend their association of rs640561 to OA (rs640561-T, beta = -0.061, p = 0.009). Finally, we examined our gene-based GWAS results for evidence supporting previously reported genes and found no support for GRM822 (p = 0.655), CNIH324 (p = 0.174), CCDC4250 (p = 0.307) or SPDYE450 (p = 0.856). However, we found nominal support for BEND4 (p = 0.0023) association with OA, which was reported in the PGC-SUD GWAS for the opioid use phenotype (exposed vs. unexposed controls)26 and PTPRF for POU (p = 0.026)49.
