The role of genetically driven OPRM1 expression also appeared complex in this study. In our S-PrediXcan analysis, we observed statistically significant, predicted differential expression of OPRM1 for OA in cerebellum, the brain tissue with the highest level of OPRM1 expression in GTEx. Moreover, one of the two cis-eQTL variants in the version 8 GTEx model for OPRM1 expression (rs478498) is in high LD with our top association variant (rs9478500; r2 = 0.56, D’ = 0.98), suggesting that the intron 1 locus may have its effect on OA through OPRM1 expression. However, the colocalization analysis was more equivocal. The hypotheses with the greatest posterior probabilities were that both OPRM1 expression and OA risk are associated with this locus, but with different causal variants (H3, posterior probability = 0.46) and with a single causal variant (H4, posterior probability = 0.38). Given the generally low level of OPRM1 expression across bulk brain tissues, larger sample sizes and single nuclei experiments will be needed to further distinguish which of these hypotheses is most likely. Ultimately, model organism or organoid experiments are likely to be necessary to fully test gene expression as a potential mechanism for the association of this locus with OA.
