Publication Date



JAMA Psychiatry





First Page


Last Page



Importance Most disorders caused by copy number variants (CNVs) display significant clinical variability, often referred to as incomplete penetrance and variable expressivity. Genetic and environmental sources of this variability are not well understood.

Objectives To investigate the contributors to phenotypic variability in probands with CNVs involving the same genomic region; to measure the effect size for de novo mutation events; and to explore the contribution of familial background to resulting cognitive, behavioral, and motor performance outcomes in probands with de novo CNVs.

Design, Setting, and Participants Family-based study design with a volunteer sample of 56 individuals with de novo 16p11.2 deletions and their noncarrier parents and siblings from the Simons Variation in Individuals Project.

Main Outcomes and Measures We used linear mixed-model analysis to measure effect size and intraclass correlation to determine the influence of family background for a de novo CNV on quantitative traits representing the following 3 neurodevelopmental domains: cognitive ability (Full-Scale IQ), social behavior (Social Responsiveness Scale), and neuromotor performance (Purdue Pegboard Test). We included an anthropometric trait, body mass index, for comparison.

Results A significant deleterious effect of the 16p11.2 deletion was demonstrated across all domains. Relative to the biparental mean, the effect sizes were −1.7 SD for cognitive ability, 2.2 SD for social behavior, and −1.3 SD for neuromotor performance (P < .001). Despite large deleterious effects, significant positive correlations between parents and probands were preserved for the Full-Scale IQ (0.42 [P = .03]), the verbal IQ (0.53 [P = .004]), and the Social Responsiveness Scale (0.52 [P = .009]) scores. We also observed a 1-SD increase in the body mass index of probands compared with siblings, with an intraclass correlation of 0.40 (P = .07).

Conclusions and Relevance Analysis of families with de novo CNVs provides the least confounded estimate of the effect size of the 16p11.2 deletion on heritable, quantitative traits and demonstrates a 1- to 2-SD effect across all neurodevelopmental dimensions. Significant parent-proband correlations indicate that family background contributes to the phenotypic variability seen in this and perhaps other CNV disorders and may have implications for counseling families regarding their children’s developmental and psychiatric prognoses. Use of biparental mean scores rather than general population mean scores may be more relevant to examine the effect of a mutation or any other cause of trait variation on a neurodevelopmental outcome and possibly on systems of diagnosis and trait ascertainment for developmental disorders.