The only significant PRS measure in the random forest model for classifying the memory and control groups was derived from the MVP study of DSM-5 AUD, suggesting the importance of AUD-PRS, rather than consumption-related PRS, in predicting neurocognitive outcomes such as alcohol-induced memory problems. This may partly be because individuals from both the memory and control groups had a lifetime diagnosis of DSM-IV alcohol dependence. While the DSM-IV alcohol dependence PRS derived from the PGC was not found to be significant, it is possible that this could be because of its relatively smaller GWAS sample size, compared to that of the MVP dataset, as well as the presence of fewer participants of non-European ancestry in the discovery GWAS (see Table 4) and/or the more inclusive diagnosis of DSM-5 AUD compared to DSM-IV AD. Nevertheless, the finding that AUD-PRS significantly contributed to the classification suggests that alcohol-induced memory issues, at least in part, are associated with genomic liability. In general, family studies, twin studies, and GWAS have all demonstrated the heritability of AUD [153,154,155] and the utility of PRS to identify
