Genes To Mental Health (G2MH): A Framework to Map the Combined Effects of Rare and Common Variants on Dimensions of Cognition and Psychopathology

Abstract

Rare genomic disorders (RGDs) confer elevated risk for neurodevelopmental psychiatric disorders. In this era of intense genomics discoveries, the landscape of RGDs is rapidly evolving. However, there has not been comparable progress to date in scalable, harmonized phenotyping methods. As a result, beyond associations with categorical diagnoses, the effects on dimensional traits remain unclear for many RGDs. The nature and specificity of RGD effects on cognitive and behavioral traits is an area of intense investigation: RGDs are frequently associated with more than one psychiatric condition, and those studied to date affect, to varying degrees, a broad range of developmental and cognitive functions. Although many RGDs have large effects, phenotypic expression is typically influenced by additional genomic and environmental factors. There is emerging evidence that using polygenic risk scores in individuals with RGDs offers opportunities to refine prediction, thus allowing for the identification of those at greatest risk of psychiatric illness. However, translation into the clinic is hindered by roadblocks, which include limited genetic testing in clinical psychiatry, and the lack of guidelines for following individuals with RGDs, who are at high risk of developing psychiatric symptoms. The Genes to Mental Health Network (G2MH) is a newly funded National Institute of Mental Health initiative that will collect, share, and analyze large-scale data sets combining genomics and dimensional measures of psychopathology spanning diverse populations and geography. The authors present here the most recent understanding of the effects of RGDs on dimensional behavioral traits and risk for psychiatric conditions and discuss strategies that will be pursued within the G2MH network, as well as how expected results can be translated into clinical practice to improve patient outcomes.

Keywords: Autism Spectrum Disorder; Diagnosis and Classification; Genetics/Genomics; Intellectual Disabilities; Neurodevelopmental Disorders; Schizophrenia Spectrum and Other Psychotic Disorders.

Figure 1

Effects of CNVs on cognitive and behavioral dimensions (A) RGDs and dimensional phenotyping, current knowledge, and hypotheses. NPD CNVs affect multiple cognitive and behavioral dimensions and increase the risk for ASD, SZ, and ID, albeit with different effect sizes. This scenario incorporates key features of the genomic architecture of psychiatric disorders including polygenicity, phenotypic variability -also referred to as pleiotropy- and genetic overlap between conditions, cognitive and behavioral dimensions. In this scenario, combinations in different proportions of common dimensions lead to different clinical manifestations classified as psychiatric diagnoses. The 4 dimensions described in figure 1A may approximately align with RDoC dimensions: Cognitive ability → Cognitive systems; Disorganized thought/perceptual abnormalities → perception (a (sub)construct of the cognitive systems domain); Social responsiveness → Systems for social processes; Anxiety-mood → Negative valence. (B) Effects of 1q21.1, 16p11.2 and 22q11.2 deletions and duplications on neurocognitive and behavioral functioning. Measures are standardized to control mean and standard deviation. For visualization purposes, scores are converted to absolute, positive values to highlight impairments in CNV groups compared to controls (Z=0 indicated by vertical dashed line). Behavioral measures = scales from Child Behavior Checklist (CBCL). Social responsiveness was assessed by the Social Responsiveness Scale (SRS). Sample sizes: 1q21.1 Del n =11, 1q21.1 Dup n = 12 (REF); 16p11.2 Del n = 137, 16p11.2 Dup n = 127 (11); 22q11.2 Del n = 99, 22q11.2 Dup n = 34 (21); combined controls n = 214.

Figure 2

Risk prediction in rare CNV carriers with and without polygenic risk (PRS) information. (A) Schematic showing baseline risk for the general population (dark green) and a large effect size RGD (dark red). Risk for individuals of both groups with a top decile PRS score (light green and light red). Although the PRS has the same small effect size in both groups, it results in a larger increase in the penetrance of a diagnosis in the RGD group. (B) Comparing risk conferred by CNVs for carriers without PRS information (baseline risk, table 1) to those with top decile PRS values. For schizophrenia PRS, OR= 1.5 (23, 59). For Autism spectrum disorder PRS, OR = 1.91 (96). Effect size of top decile PRS cognitive ability compared to 50th percentile = 0.45 z-score (23, 97). Risk in RGD carriers with top decile PRS values is computed based on an additive model. Y-axis for IQ: IQ values. Y-axis for ID, SZ, and ASD: penetrance (from 0 to 100%) of a diagnosis in CNV carriers.

Figure 3

Heritability / SNP heritability for cognitive and behavioral dimensions relevant to RGDs Comparison between SNP-based (GWAS) heritabilities and twin-based heritabilities for phenotypes of interest. Heritability estimates and 95% CI are shown. Heritability estimates were derived from studies listed in supplemental table 1. Heritability estimates can differ depending on age (e.g., head circumference) and sex (e.g., hallucinations).