The contribution of copy number variants to psychiatric symptoms and cognitive ability

Abstract

Copy number variants (CNVs) are deletions and duplications of DNA sequence. The most frequently studied CNVs, which are described in this review, are recurrent CNVs that occur in the same locations on the genome. These CNVs have been strongly implicated in neurodevelopmental disorders, namely autism spectrum disorder (ASD), intellectual disability (ID), and developmental delay (DD), but also in schizophrenia. More recent work has also shown that CNVs increase risk for other psychiatric disorders, namely, depression, bipolar disorder, and post-traumatic stress disorder. Many of the same CNVs are implicated across all of these disorders, and these neuropsychiatric CNVs are also associated with cognitive ability in the general population, as well as with structural and functional brain alterations. Neuropsychiatric CNVs also show incomplete penetrance, such that carriers do not always develop any psychiatric disorder, and may show only mild symptoms, if any. Variable expressivity, whereby the same CNVs are associated with many different phenotypes of varied severity, also points to highly complex mechanisms underlying disease risk in CNV carriers. Comprehensive and longitudinal phenotyping studies of individual CNVs have provided initial insights into these mechanisms. However, more work is needed to estimate and predict the effect of non-recurrent, ultra-rare CNVs, which also contribute to psychiatric and cognitive outcomes. Moreover, delineating the broader phenotypic landscape of neuropsychiatric CNVs in both clinical and general population cohorts may also offer important mechanistic insights.

Fig. 1. Types of copy number variants (CNVs).

CNVs are variation in the number of copies of segments of DNA and can be classified primarily as deletion (segment B deleted), duplication (segment B duplicated), inversion (segments B and C inverted), and insertion (segment E inserted).

Fig. 2. Summary of known associations between CNVs and psychiatric phenotypes.

Red = deletion associated with phenotype; blue = duplication associated with phenotype; black = both deletion and duplication associated with phenotype. Total % and Control % = prevalence estimates from Coe et al. (2014) for patients and controls combined (Total %) and controls only (Control %). NDD = neurodevelopmental disorders (intellectual disability, developmental delay and/or autism spectrum disorder)(Coe et al. 2014); ASD = autism spectrum disorder (Iakoucheva et al. 2019); ADHD = attention deficit hyperactivity disorder (Gudmundsson et al. 2019); SCZ = schizophrenia (Marshall et al. 2017); BD = bipolar disorder (Green et al. 2016); PTSD = post-traumatic stress disorder (Maihofer et al., 2022); OCD = obsessive compulsive disorder (McGrath et al. 2014); AD = Alzheimer’s disease (Chapman et al. 2013); Neurotic = neuroticism (Auwerx et al. 2021); Cog = cognition (Kendall et al. 2019); TBV = total brain volume (Modenato et al. 2021). Ns represent number of patients (NDD; ASD; ADHD; SCZ; Depression; BD; PTSD; OCD; AD) or participants (Neuroticism; Cognition).

Fig. 3. Prevalence and effect size estimates for select CNVs.

Prevalence estimates in % units; Effect sizes in standardized units, with values of 0.2, 0.5, and 0.8 indicating small, medium, and large effects. UKBB = UK Biobank (prevalence estimated excluding participants with primary or secondary International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) psychiatric diagnoses); NDD = neurodevelopmental disorders (intellectual disability, developmental delay and/or autism spectrum disorder)(Coe et al. 2014); ASD = autism spectrum disorder (Sanders et al. 2015; Pinto et al. 2014); ADHD = attention deficit hyperactivity disorder (Gudmundsson et al. 2019); SCZ = schizophrenia (Marshall et al. 2017); Depression (Kendall et al. 2019); Bipolar = bipolar disorder (Green et al. 2016); Cognition (Kendall et al. 2019). *p_FDR<.05.

Fig. 4. Detectable mean trait difference by number of CNV carriers.

Figure illustrates the differences in trait means detectable at 80% power at a p-value threshold of 8 × 10⁻⁵ by number of CNV carriers. For the rarest CNVs (10–15 carriers), studies are powered to detect only large effects of 1–1.5 standard deviation (SD) units (without any further stratification), but for the majority of CNVs (20–100 carriers) studies are well powered to detect differences of medium effect size in the range of 0.5–1 SD, and for the most common CNVs (100+ carriers), studies are well powered to detect differences of small effect size on the order of 0.25 SD.