Psychotic disorders remain among the most severe and debilitating forms of psychopathology, exerting a dramatic cost on patients, families and society as a whole. Recently there has been considerable hope that understanding the clinical and neurodevelopmental mechanisms preceding the emergence of psychosis, will allow earlier and more effective treatment strategies. However, progress has been hindered by difficulties in investigating stages preceding onset of psychosis. Moreover, researchers have struggled with the massively multifactorial nature of the disease mechanisms at stake. In order to better understand the pathophysiology of psychosis longitudinal designs and dedicated computational approaches and will be required.

The present dissertation will focus on the study of 22q11.2 Deletion Syndrome (22q11DS), a genetic disorder that contributes a dramatically increased risk of developing psychosis. Longitudinal follow-up of youth with 22q11DS offers a unique opportunity to characterize early stages of vulnerability to psychosis. The overall objective of this dissertation will be the development and implementation of computational tools of dynamic network analysis to characterize clinical pathways and neurodevelopmental mechanisms of vulnerability to psychopathology in 22q11DS.

The first study employs structural covariance (SC) to describe alterations of large-scale cortical networks in 22q11DS. We show that alterations of SC networks related to the presence of psychotic symptoms in 22q11DS, with a particularly prominent role anterior cingulate dysconnectivity. In the second study we implement a novel dynamic network analysis technique to describe trajectories of SC maturation in 22q11DS. We show that late adolescence is a critical period for the disruption of SC architecture in 22q11DS, that might contribute vulnerability to psychosis. In the third study, we employ t dynamic network analysis and graph signal processing to analyse longitudinal clinical data. We show that this approach can provide an intuitive and quantitative characterization of relevant clinical pathways of casual interaction between symptoms. Specifically, we demonstrate the central role of reduced tolerance to stress in 22q11DS. Moreover, we show that a network approach to psychopathology carries potential for predicting clinical evolution of individual subjects. In study 4 we further explored the role exposure to environmental stress in contributing to psychopathology in 22q11DS. We propose that the syndrome could represent an ideal model to characterize mechanisms of gene-environment interaction. In study 5 we explored neurobiological mechanisms contributing to increased vulnerability to stress in 22q11DS, focusing on the candidate role of dysregulation of the Hypothalamus-Pituitary-Adrenal axis. We show that an atypical longitudinal trajectory of pituitary volume decline, suggestive of extinction of HPA-axis functionality, is associate with reduced tolerance to environmental stress and contributes pleiotropic vulnerability to psychopathology. HPA-axis dys-maturation also affected trajectories of hippocampal and cortical maturation in 22q11DS. In conclusion, our results suggest that a dynamic network approach to developmental psychopathology has the potential of improving our understanding of both neurodevelopmental mechanisms and clinical pathways of vulnerability to mental health disturbances.