Neurocognition and neuroimaging differences across bipolar disorder with and without psychotic features and schizophrenia

The nosological boundary between bipolar disorder with psychotic features and schizophrenia remains unresolved, particularly in relation to Schneiderian first-rank symptoms, which may index shared core psychosis mechanisms across diagnostic categories. While classical categorical approaches assume discrete diagnostic entities, accumulating evidence supports a dimensional psychosis spectrum characterised by partially overlapping neurocognitive and neurobiological substrates. This study aimed to delineate neurocognitive and structural brain differences among psychotic bipolar disorder, non-psychotic bipolar disorder and schizophrenia, and to clarify their positioning within a dimensional framework of psychosis.

Ninety remitted patients (30 per group) meeting Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria were included in this cross-sectional study. All participants were clinically stable and free from acute symptom exacerbation. A comprehensive neurocognitive battery was administered to assess processing speed, attention, executive functions, working memory, verbal learning and memory, and social cognition. Structural magnetic resonance imaging data were acquired using high-resolution T1-weighted sequences and analysed via an automated volumetric segmentation pipeline to quantify global, lobar, subcortical, insular, opercular and cerebellar structures. Group differences were examined using analysis of variance with Tukey post hoc comparisons. Variables demonstrating significant group effects were subsequently entered into multinomial logistic regression models controlling for relevant clinical covariates, including age, illness duration and medication status.

Schizophrenia was associated with marked impairments in processing speed and verbal memory, as reflected by significantly prolonged trail making test part A completion times (p < 0.001) and reduced verbal learning and short-term recall performance (p < 0.001 for both), compared to both bipolar disorder groups. Structural analyses revealed regionally specific reductions in cerebellar vermis volumes (p = 0.040), posterior cerebellar lobules VIII to X (p = 0.049), occipital pole cortical thickness (p = 0.003) and anterior insula volumes (p = 0.005) in schizophrenia. In contrast, non-psychotic bipolar disorder showed relatively increased right parietal operculum volume (p = 0.004). Notably, no significant differences were observed in total grey or white matter volumes, indicating that group distinctions are driven by selective network-level alterations rather than global structural changes. Multinomial regression analyses demonstrated that a combined profile of processing speed deficits, verbal memory impairments and regional structural alterations provided robust discrimination between diagnostic groups, supporting their potential utility as transdiagnostic markers.

These findings support a partially overlapping yet hierarchically differentiated model of psychosis, in which schizophrenia is characterised by more extensive disruption of cerebellar–insular–cortical networks implicated in cognitive integration and salience processing. The convergence of neurocognitive deficits and region-specific structural alterations suggests that Schneiderian first-rank symptoms may arise from shared network dysfunctions, while their severity and distribution contribute to diagnostic differentiation. Importantly, the absence of global volumetric differences alongside selective regional abnormalities argues against a diffuse neurodegenerative model and instead supports a network-based dysconnectivity framework. Integrating neurocognitive and neuroimaging markers may enhance transdiagnostic stratification, improve diagnostic precision, and inform mechanism-based treatment targeting across the psychosis spectrum.

References

No references