Psychosis and Affective Research Domains and Intermediate Phenotypes (PARDIP)

The B-SNIP research consortium previously obtained dense phenotypes across the psychosis spectrum in an effort to observe features both (i) distinctive to and (ii) shared between DSM-type categorical diagnoses. Despite the broad range of extra-clinical phenotypes, we had limited success finding clinical SZ-BD diagnosis-specific features; instead, most phenotypes were distributed continuously across DSM diagnoses. To describe more biologically homogeneous groups, therefore, we combined all psychosis probands and implemented a multi-stage analysis procedure, beginning with identification of psychosis biomarkers (variables with the largest effect sizes for differentiating psychosis and healthy groups) including cognitive, electrophysiological, and oculo-motor measures ('classical' endophenotypes). We then estimated the number of subgroups that efficiently optimized variance among the biomarkers (n=3) and differentiated the individual psychosis cases into these subgroups. Subsequently, the subgroups were tested for biological uniqueness using meaningful external validators (structural and functional brain imaging, social functioning, and familial data). Given the neurobiological distinctiveness of these subgroups, we called them psychosis Biotypes. DSM diagnoses were distributed across all Biotypes. Compared to DSM diagnoses, Biotype membership enhanced group separations on biomarkers. These results indicate that groups of psychosis cases can be generated with homogeneous phenotypic characteristics independent of DSM diagnoses. The proposed study aims to further develop Biotype definitions and demonstrate that psychosis Biotypes constructed from a dense biomarker panel (i) are replicable, (ii) neurobiologically distinctive, and (iii) have unique genetic characteristics.