Deciphering the Heterogeneity of Schizophrenia: A Multimodal and Multivariate Neuroimaging Framework for Unveiling...

Schizophrenia manifests large heterogeneities in either symptoms or brain abnormalities. However, the neurobiological basis of symptomatic diversity remains poorly understood. We hypothesized that schizophrenia's diverse symptoms arise from the interplay of structural and functional alterations across multiple brain regions, rather than isolated abnormalities in a single area.

A total of 495 schizophrenia patients and 507 healthy controls from 8 sites were recruited. Five symptomatic dimensions of schizophrenia patients were derived from the Positive and Negative Syndrome Scale. Multivariate canonical correlation analysis was introduced to identify symptom-related multimodal magnetic resonance imaging composite indicators (MRICIs) derived from gray matter volume, functional connectivity strength, and white matter fractional anisotropy. The intergroup differences in MRICIs were compared, and the paired-wise correlations between symptom dimensions and MRICIs were resolved. Finally, K-means clustering was used to identify the underlying biological subtypes of schizophrenia based on MRICIs.

Canonical correlation analysis identified 15 MRICIs in schizophrenia that were specifically contributed by the neuroimaging measures of multiple regions, respectively. These MRICIs can effectively characterize the complexity of symptoms, showing correlations within and across symptom dimensions, and were consistent across both first-episode and chronic patients. Additionally, some of these indicators could moderately differentiate schizophrenia patients from healthy controls. K-means clustering identified 2 schizophrenia subtypes with distinct MRICI profiles and symptom severity.

Symptom-guided multimodal and multivariate MRICIs could decode the symptom heterogeneity of schizophrenia patients and might be considered as potential biomarkers for schizophrenia.