in both studies (In Loh's Taiwanese and our Irish samples: case = 0.29, control = 0.32). However, this marker exhibited genotypic difference (P = 0.01) between their male AD cases and controls, but did not show such difference (P = 0.42) in our study. For GAD2, marker rs2839670 which is located near 5′ of the gene was genotyped in all three studies with varying MAF distribution across populations (Among AD cases, MAF were 0.12, 0.18, 0.28 for Russian, Irish, Taiwanese samples; among controls, MAF were 0.19, 0.20, 0.23 for the three populations). This marker showed significant genotypic association in the Russian sample (P = 0.01) but not in the Irish and Taiwanese samples. With the sample size in the current study, in most of the situations we shall have power of 0.8 to detect an effect size (such as odds ratio) greater than 1.3. Therefore, the negative association findings with AD diagnosis are unlikely due to inadequate power if loci have notable effects on AD. Overall, although we did not observe direct effects of single markers in the GAD genes on the risk to develop AD, they may have impact via their influences on initial sensitivity to ethanol and the
