Abstract Copy number variants (CNVs) can alter disease susceptibility by gene deletion, duplication, and other mechanisms, and have been implicated in neuropsychiatric diseases. However, their rarity or de novo nature impedes linkage analysis. Therefore, we identified recurrent CNVs (rCNVs) in Native Americans with low genetic admixture and high prevalence of alcohol use disorder (AUD) and other psychiatric disorders. Large (> 200 kb) rCNVs were abundant in Plains Indians (PI) and Southwest American Indians (SWI), almost all carrying at least one rCNV, with some CNVs found in both geographically and linguistically distinct tribes. In patients carrying rCNVs, gene deletions led to haploinsufficiency, and duplications led to increased gene dosage. Haplotype analysis revealed a common chromosome 6p21.33 recurrent CNV (rCNV) that persisted in Native Americans for at least 750 generations, leading to haploinsufficiency of at least two genes. Gene-based CNV burden and CNV count did not predict AUD or other psychiatric disorders. However, an rCNV, found in PI and duplicating three genes within the 22q11.2 velocardiofacial syndrome region, showed nominally elevated odds ratios in generalized linear mixed models accounting for kinship as a random effect and age and sex as fixed covariates. For AUD, the odds ratio was 3.19 (95% CI 1.12–9.07, p = 0.03), and for psychiatric diagnosis, the odds ratio was 4.80 (95% CI 1.37–16.82, p = 0.014). These associations, calculated with adjustment for relatedness, did not remain statistically significant after correction for multiple testing, illustrating the challenge of linking CNVs to behavior even if they are recurrent and potentially of large effect. The effects of most CNVs are undetectable via genome-wide association studies with single SNPs, and in cosmopolitan populations, most CNVs are nonrecurrent, sometimes affecting similar genomic regions but differing in their properties, and origins. Recurrent CNVs having ancient origins are prevalent in Native American populations providing an opportunity to examine their relationships to disease risk. However, we observed that neither gene-based CNV burden nor individual rCNVs did not predict AUD or other psychiatric outcomes, after adjustment for the numbers of rCNVs tested.
Wakil et al. (Thu,) studied this question.