Cerebellar dysconnectivity in schizophrenia spectrum: task-based functional connectivity analysis and cognitive stratification

Introduction Schizophrenia is conceptualized as a disorder of brain network dysconnectivity, yet relationships between neural alterations, cognitive deficits, and genetic risk remain unclear. Methods We examined 86 participants: schizophrenia patients (SCZ), unaffected siblings (SCZ-SIB), healthy controls (CON), and control siblings (CON-SIB). We used a multiscale graph-theoretic analysis of task-based fMRI during N-back working memory and unsupervised clinical-cognitive clustering. Results We found that reduced cerebellum-sensorimotor (CER-SM) and cerebellum-cingulo-opercular (CER-CO) connectivity during the 1-back condition robustly discriminated SCZ from CON (AUC = 0.89). Critically, these dysconnectivity patterns were linked to clinical state, present in SCZ vs. SCZ-SIB but absent in SCZ-SIB vs. CON-SIB, suggesting illness expression rather than familial risk. Unsupervised clustering revealed three data-driven subtypes with distinct cognitive- symptomatic profiles: subtype 1 with relative preservation of verbal abilities (predominantly controls), subtype 2 with marked fluid cognitive impairment (enriched in SCZ), and subtype 3 with intermediate performance with working memory sparing (mixed composition). Cerebellar-cortical hypoconnectivity showed graded alignment across these profiles. Discussion These findings demonstrate that cerebellar dysconnectivity is most detectable under moderate cognitive load, tracks with clinical state, and covaries with transdiagnostic cognitive profiles, advancing circuit-based understanding of schizophrenia heterogeneity.