Multi-level signatures of antisociality, grey and white matter volume and the social exposome in early psychosis and depression

Brisbane Convention & Exhibition Centre 

Antisocial behavior is a multifaceted phenomenon influenced by the interplay of brain structure and social environmental factors across diagnostic boundaries. Investigating these interactions is crucial for understanding their role in clinical trajectories, necessitating advanced mathematical modeling. Therefore, we present a novel multivariate machine learning approach to elucidate the clinical and neuroanatomical complexity of antisocial behavior in a transdiagnostic adolescent and young adult cohort.
We used data from the prospective, multicentric European Personalized Prognostic Tools for Early Psychosis Management (PRONIA) project, including individuals with recent-onset depression or psychosis and clinical high-risk states for psychosis. To detect parsimonious associations between antisocial behavior (PANSS items P4, P7, G4, G8, G14), social environmental factorsexposome (CTQ, EDS, BS, LEE, MSPSS, RSA) and brain structurevolume (GMV, WMV, CSF), the multi-block sparse partial least squares algorithm was employed within a nested cross-validation framework.
The analysis yielded two significant signatures. The first (P = 0.003, Frobenius norm = 2.32) captured an ageing-related GMV pruning pattern of GMV reduction in frontal and parietal regions. The second (P = 0.01, Frobenius norm = 2.22) linked higher levels of hostility (P7) and excitement (P14) and experience of past and present discrimination and childhood trauma (sexual and physical abuse) to GMV reductions in predominantly frontal areas and enlargement of the third ventricle. This finding was particularly pronounced in younger individuals with recent-onset psychosis.
This study presents the first application of the multi-block sparse partial least squares approach to integrate multiple clinical and neurobiological domains to study antisocial behavior in early-stage mental disorders. The multi-level signature linking antisocial traits, frontal GMV reductions, and social adversities in younger individuals with recent-onset psychosis highlights the complex multi-factor etiology of antisocial behavior in first-episode psychosis and emphasizes the need for early assessment, intervention and long-term risk management specifically identified clinical subpopulations.