Impact of Hardy-Weinberg equilibrium deviation on allele-based risk effect of genetic association studies and meta-analysis

Abstract

Deviations from Hardy-Weinberg equilibrium (HWE) in control subjects may bias the estimates of genetic effects in genetic association studies (GAS) and meta-analysis. A large empirical evaluation was carried out to evaluate the impact of HWE deviation and explore the effect of variance adjustment for the allele-based odds ratio in 833 individual GAS and 72 meta-analyses. In individual GAS, the variance adjustment for any deviation from HWE resulted in stronger associations, and 10 GAS (1%) became significant (P < 0.05). One hundred sixteen GAS (14%) showed significant deviation from HWE (P (HWE) < 0.05); however, only 37 GAS (4%) had more than 90% power to detect significant deviation from HWE at the 5% level. In meta-analyses, adjustment for any deviation from HWE improved the significance in 53 meta-analyses (74%). Then, a formal statistical significance (P < 0.05) was revealed for one previously negative meta-analyses whereas one meta-analysis lost its significance. Between-study heterogeneity was enhanced in 50 meta-analyses (69%). None of the meta-analyses lost the significance of heterogeneity (P (Q) < 0.10) whereas in one meta-analysis, the non significant heterogeneity became significant. Sensitivity analysis for studies not conforming to HWE (P (HWE) < 0.05) was applied to 45 meta-analyses (69%). Then, the significance of association was increased in 26 the meta-analyses (58%) and one meta-analysis became significant (P < 0.05) whereas seven meta-analyses (16%) were no longer significant. Adjustment for HWE deviation could be an effective strategy for dealing with HWE violations in GAS and meta-analyses.