S24. PATHWAYS LINKING ENVIRONMENTAL RISKS, POLYGENIC RISK SCORE TO SCHIZOPHRENIA AND PSYCHOTIC EXPERIENCES IN A BRAZILIAN COHORT OF CHILDREN AND ADOLESCENTS

Abstract

BackgroundPolygenic risk score to schizophrenia (PRS-SZ) provides a liability measure summarizing each genetic risk variant and the polyenviromic risk score (PERS) proposes the same regarding exposure factors to psychosis, yet few studies addressed how both scopes interplay, especially in early developmental stages. Psychotic experiences (PE) rest on the lower range of psychosis spectrum, representing an important asset to study psychotic disorders, ie. schizophrenia. However, investigators failed to find significant associations between PRS-SZ and PE in children. We hypothesize that unspecific psychopathology – also previously linked to PE – can mediate the effects of higher risk load for psychosis during neurodevelopment. Thus, our aim is to test a moderated mediation model in which PERS and general psychopathology in youths can lead to PE, prospectively, through SZ genetic liability.MethodsWe analyzed data from the Brazilian High-Risk Cohort for Psychiatric Disorders, a youth community sample with 2 time-points: baseline (w0) and 3year follow-up (w1), from São Paulo and Porto Alegre, both urban centers. PRS-SZ was calculated using summary statistics from the PGC and corrected for the 10 principal components of the GWAS. PE were assessed at w0 and w1 with the Community Assessment Psychotic Experiences – CAPE and trained psychologists rated the reliability of students’ answers. The Development and Well-Being Assessment – DAWBA, a structured interview with a transdiagnostic approach, was used to extract a general factor for psychopathology (P-factor) on w0. Latent variables for PE and P-factor were generated through confirmatory factor analysis yielding good model fits. We calculated PERS on w0, as validated, with birth season, urbanicity, cannabis use, paternal age, obstetric/perinatal complications and physical/sexual abuse, neglect or parental loss/separation. Last, we built a moderated mediation diagram based on model 15 of Haye’s PROCESS builder on SPSS: (X) PERS > (M) P-factor > (Y) PE w1, with (V) PRS-SZ as a moderator for PERS > PE and P-factor > PE. Age, sex, site and PE w0 were covariates.Results2,511 students (6–14 y/o, mean=10.2 ± 1.9, 53% male) completed the w0 assessment and 2,010 the follow-up (mean=13.5 y/o ± 1.9). In our moderated mediation model, P-factor emerged as a full mediator between PERS and PE w1 (B=.324, BootLL–UL CI=.138 to .553). We found PRS-SZ provided a significant moderation effect on the P-factor > PE relation (M*V=.053, R2-chng=.003, p=.037), with the moderator effects of the focal predictor rising considerably according to values of PRS-SZ: p16 (B=.047, p=.192), p50 (B=.099, p=.000) and p84 (B=.153, p=.000). PRS-SZ did not moderate PERS > PE separately (X*V=.016, R2-chng=.001, p=.974). However, conditional indirect coefficients for the complete model were also significant for higher PRS-SZ levels: p16 (B=.143, BootLL–UL CI=-.072 to .389), p50 (B=.304, BootLL–UL CI=.126 to .529) and p84 (B=.470, BootLL–UL CI=.197 to .814).DiscussionOur findings suggest environmental risk factors and intermediate phenotypes – namely unspecific non-psychotic psychopathology – can play crucial and intertwined roles in children and adolescents with higher genetic liability to SZ. Moreover, the moderation effects of PRS-SZ imply the existence of thresholds for those relations. The non-clinical nature and age of our sample could explain the low effect sizes. Next steps would include additional phenotypic tracks, such as cognition and social functionality – both previously connected to PRS-SZ as well. We hope our results can help disentangle the genetic and environmental trajectories bonding SZ proneness and PE, and possibly contribute to risk assessment in youths, especially among vulnerable populations.