Symptom networks and working memory in schizophrenia: a multi-methodological cross-sectional study from phenotype to endophenotype.

BACKGROUND

A notable deficit in working memory (WM) is well established in schizophrenia. Nevertheless, the intricate relationship between various symptoms and WM impairment is still not fully understood. We use three distinct methodologies-symptom network analysis (SNA), Connectome-Based Predictive Modeling (CPM), and brain gene annotation enrichment analysis-to explore the connectome patterns that link WM deficits and symptoms, and their related gene expression.

METHODS

255 patients with schizophrenia were recruited as two distinct samples. SNA was used to pinpoint the core psychiatric symptoms influenced by WM performance. CPM identified the subnetwork of the functional connectome that was recruited under the 2-back load of the N-back WM task, and predicted the severity of the SNA-based key symptoms. Gene annotation enrichment analysis explored the likely molecular biological processes underlying the symptom-predictive functional WM network.

RESULTS

SNA revealed that disorganized attention (G11 of PANSS) is most closely linked to WM performance in schizophrenia. The WM-based connectome significantly predicted disorganized attention ( = 0.278, = 0.001, permutation- = 0.046), and this model was validated in the second dataset ( = 0.274, = 0.014). The predictive network primarily involved the frontoparietal and frontolimbic networks. Gene enrichment analysis revealed a preferential role for cytoplasmic protein binding, indicating a potential molecular basis for the WM-related, symptom-predictive functional connectivity.

CONCLUSIONS

Impaired WM performance in schizophrenia relates to frontoparietal and frontolimbic connectivity and preferentially influences the severity of disorganized attention, a clinically observable phenomenon. The potential role of cytoplasmic protein binding in WM deficits and attentional disorganization in schizophrenia warrants further investigation.