Abstract
Symptom diversity in psychoses complicates the search for biological markers. Using Positive and Negative Symptom Scale data from 362 recent-onset patients in the multicenter PRONIA study, we identified four subgroups with dominant symptom patterns: motor/cognitive, positive, social withdrawal, and affective. Subgroups were compared against each other and to 338 healthy controls across neurocognition, brain imaging, and genetic risk. Patient subgroups shared a profile of impaired processing speed and altered functional connectivity, with increased coupling in sensorimotor networks and reduced connectivity within and between default-mode, salience, and control networks. Variations in modality-specific neurobiological underpinnings differentiated subgroups, with fronto-temporal gray-matter loss characterizing the motor/cognitive and positive subgroups, and elevated genetic risk best separating the positive subgroup from the rest. The motor/cognitive group showed the most severe alterations across modalities, reaching the highest multimodal separability from controls (Balanced Accuracy=82.1%, sensitivity=74.9%, specificity=89.3%). Our findings support a framework of shared biological dysfunction with modality-selective vulnerabilities shaping symptom heterogeneity in early-stage psychotic disorders.