The association between polygenic risk scores and transdiagnostic vulnerability traits in a population-based youth sample

Evidence points towards a shared genetic structure between AN and OCD and SCZ. Whether this genetic correlation manifests itself with shared brain and behavioural characteristics between these disorders remains to be confirmed. This knowledge can in turn provide insight on developmental pathways relevant to both shared and unique psychiatric risk. PRS is often studied in relation to vulnerability traits with the underlining assumption that those with a higher genetic liability are more likely to express a psychiatric trait than those with lower genetic risk. The goal of the study was to identify the relationship PRS for AN, binge eating, OCD and SCZ and vulnerability traits (brain/cognitive characteristics) associated with these disorders using clinical, neurocognitive, and neuroimaging data from the population-based Philadelphia Neurodevelopmental Cohort (PNC). The study aimed to gain a better understanding of the complex mechanisms underlining EDs and associated comorbidities. We analysed publicly available genetic, clinical, and neurocognitive data from the PNC (N=4729), a population-based sample of United States youth, 8 to 22 years old. A separate analysis was run on 626 individuals that additionally had structural MRI data collected. Participants were genome-wide genotyped on the Illumina genotyping platform. Psychiatric phenotypes related to EDs, OCD, and SCZ were assessed via a computerized structured interview (Kiddie-SADS Family Study Interview). Neurocognitive phenotypes such as executive functioning, episodic memory, complex cognition, and social cognition performance were assessed via an adapted computerized neurocognitive battery (CNB). Total intracranial volume, grey matter volume, white matter volume, gyrification, and thickness were extracted from structural MRI data. Regression models were constructed to test the relation between PRSs and the clinical, neurocognitive, and cortical dependent variables. A separate analysis was run to test the relation between trait expression and cortical parameters. The AN-PRS was significantly associated with body image distortion (OR=1.210, 95% CI [1.080, 1.355], pFDR=0.012)), and a trend towards significance was present in association with compulsive behaviour (puncorrected= 0.034). A trend was also seen between SCZ-PRS and binge eating (OR=1.116, puncorrected=0.027) and perfectionism (OR=1.094, puncorrected=0.021) and between OCD-PRS and psychosis (OR = 0.075; 95% CI = 0.001, 0.150; puncorrected = 0.046). Moreover, body image distortion was significantly related to a reduction in gross grey matter volume (β= -30.091, t= -2.857, pFDR=0.053, Mean difference = 38). Subjects at high genetic risk for AN, OCD, and SCZ do not show abnormal brain characteristics, but in contrast subjects displaying body image distortion show reduced gross grey matter volume. These findings point to brain abnormalities as state markers (effect of illness) rather than trait markers. Additionally, our results confirm previous findings that AN-PRS significantly correlates with AN traits in the general population. Our results also show that vulnerability traits can transcend clinical labels and add to our knowledge of psychiatric comorbidity. To check for the stability of our findings, independent of the cohort analysed and PRS version, we aim to replicate these findings in another non-clinical sample of youth from the United States using the latest generated PRS for AN, BE and SCZ.