population from our Yale-UPenn samples (Fig 4). Rs2066702, also at ADH1B, has not previously been associated with MaxDrinks, likely because of its low frequency in EAs and Asians, who are most heavily represented in studies of ADH1B variants and alcohol phenotypes. This SNP has a higher minor allele frequency in populations of African descent (0.19 in the present study). The p-value was 10-10 in the AA population in the combined Yale-UPenn and SAGE samples. Rs2066702 (Arg370Cys) is located in exon 9 and encodes the β3 subunit of ADH1B. The substitution of the cysteine residue, which results in more rapid alcohol metabolism than the common arginine residue, was associated with lower MaxDrinks. This finding is consistent with the substitution causing higher levels of acetaldehyde, which are aversive. In addition to the two known functional SNPs in ADH1B, rs20047889 was also associated with Maxdrinks, independent of the effect of rs2066702 in the AA samples. These findings, along with our previous findings of significant association of ALDH2 with MaxDrinks in a Chinese population (Quillen et al., 2014), provide additional strong evidence of a convergent biological pathway that involves alcohol metabolism in different populations.
