Neurobiology-based cognitive biotypes using multi-scale intrinsic connectivity networks in psychotic disorders

Pablo Andrés-Camazón, Covadonga M. Díaz‐Caneja, Ram Ballem, Jiayu Chen, Vince D. Calhoun, Armin IrajiSchizophrenia 2025 · DOI

Abstract

Understanding neurobiology and developing effective interventions for cognitive dysfunction in psychotic disorders remain elusive. Insufficient knowledge about the biological heterogeneity of cognitive dysfunction hinders progress. We aimed to identify subgroups of patients with psychosis and distinct patterns of functional brain alterations related to cognition (cognitive biotypes). We analyzed B-SNIP consortium data (2,270 participants including participants with psychotic disorders, relatives, and controls). We used reference-informed independent component analysis with the standardized and fully automated framework NeuroMark and the 100k multi-scale intrinsic connectivity networks (ICN) template to obtain subject-specific ICNs and whole-brain functional network connectivity (FNC). FNC features associated with cognitive performance were identified using multivariate joint analysis. K-means clustering identified patient subgroups based on these features. Two biotypes with different functional brain alteration patterns were identified. Relative to controls, biotype 1 exhibited hypoconnectivity in cerebellar-subcortical and somatomotor-visual networks and worse cognitive performance. Biotype 2 exhibited hyperconnectivity in somatomotor-subcortical networks, hypoconnectivity in somatomotor-high cognitive processing networks, and better-preserved cognitive performance. Demographic, clinical, cognitive, and FNC characteristics of biotypes were consistent in discovery and replication sets and in relatives. 76.56% of relatives were assigned to a psychosis biotype; of those, 70.12% were to the same biotype as their affected family members. These findings suggest two distinctive psychosis-related cognitive biotypes with differing functional brain patterns shared with their relatives. Instead of traditional diagnosis, patient stratification based on these biotypes may help optimize future research and identify biological targets for the treatment of cognitive dysfunction in psychosis.

Introduction

Cognitive dysfunction is a core feature of psychotic disorders, such as schizophrenia and bipolar disorder, and serves as a primary predictor of long-term functional outcomes. Despite its clinical significance, effective treatments remain elusive, largely due to the biological heterogeneity of psychosis. Patients sharing a single DSM diagnosis often exhibit vastly different underlying neurobiological abnormalities. Traditional research paradigms comparing broad patient groups against healthy controls frequently mask this diversity. This study aims to shift the focus from symptom-based labels toward "biotypes"—subgroups defined by objective neurobiological markers—to identify specific biological targets for cognitive interventions.

Methodology

The researchers utilized data from the Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) consortium, encompassing 2,270 individuals, including patients with schizophrenia (SZ), schizoaffective disorder (SAD), and bipolar disorder (BDP), along with their first-degree relatives and healthy controls.

To analyze this large dataset, the team employed the NeuroMark framework, an automated Independent Component Analysis (ICA) tool, to extract 105 intrinsic connectivity networks (ICNs) from resting-state fMRI data. By using a template derived from over 100,000 individuals, the study examined brain connectivity across multiple spatial scales. A multivariate technique combining Principal Component Analysis (PCA) and Canonical Correlation Analysis (CCA) was then used to isolate Functional Network Connectivity (FNC) patterns most strongly correlated with cognitive performance, measured via the BACS battery.

Figure 1: Overview of the data processing pipeline (A) and the discovery/replication study design (B).

Results

Using k-means clustering on the identified cognitive-related brain features, the researchers discovered two stable patient biotypes that transcended traditional diagnostic categories.

  • Biotype 1 (Severely Impaired): Characterized by significantly poorer performance across all cognitive domains. Neurobiologically, this group showed widespread hypoconnectivity between the cerebellum, subcortical regions, and somatomotor-visual networks. This biotype was more frequently associated with schizophrenia diagnoses and lower socioeconomic status.
  • Biotype 2 (Better Preserved): Exhibited relatively preserved cognitive function. Its brain signature included a mix of hyperconnectivity in somatomotor-subcortical circuits—potentially a compensatory mechanism—and hypoconnectivity in high-level cognitive processing networks.
Figure 2: Identification of cognitive biotypes through canonical correlation analysis (C) and k-means clustering (D) of functional network connectivity features.

Crucially, these biotypes were consistent across both discovery and replication sets. Furthermore, 70.12% of healthy relatives assigned to a psychosis biotype shared the same biotype as their affected family members, suggesting these connectivity patterns are heritable intermediate phenotypes.

Discussion

These findings demonstrate that neurobiological signatures of cognitive impairment in psychosis do not align with traditional DSM boundaries. Instead, they reveal two distinct pathological mechanisms: one defined by widespread hypoconnectivity and severe impairment, and another showing mixed connectivity patterns with better-preserved function.

While these results provide a robust framework for patient stratification, several limitations remain. The sample was primarily Caucasian and based in the US, necessitating validation in more diverse global populations. Additionally, as a cross-sectional study, it remains unclear how these biotypes evolve over time or respond to treatment. Future research should focus on determining whether these biotypes can predict differential responses to pharmacological or cognitive remediation therapies, moving psychiatry closer to personalized medicine.

Summary updated

Abstract

Understanding neurobiology and developing effective interventions for cognitive dysfunction in psychotic disorders remain elusive. Insufficient knowledge about the biological heterogeneity of cognitive dysfunction hinders progress. We aimed to identify subgroups of patients with psychosis and distinct patterns of functional brain alterations related to cognition (cognitive biotypes). We analyzed B-SNIP consortium data (2 270 participants including participants with psychotic disorders, relatives, and controls, 55% females). We used reference-informed independent component analysis with the standardized and fully automated framework NeuroMark and the 100k multi-scale intrinsic connectivity networks (ICN) template to obtain subject-specific ICNs and whole-brain functional network connectivity (FNC). FNC features associated with cognitive performance were identified using multivariate joint analysis. K-means clustering identified patient subgroups based on these features. Two biotypes with different functional brain alteration patterns were identified. Relative to controls, biotype 1 exhibited hypoconnectivity in cerebellar-subcortical and somatomotor-visual networks and worse cognitive performance. Biotype 2 exhibited hyperconnectivity in somatomotor-subcortical networks, hypoconnectivity in somatomotor-high cognitive processing networks, and better-preserved cognitive performance. Demographic, clinical, cognitive, and FNC characteristics of biotypes were consistent in discovery and replication sets and in relatives. 76.56% of relatives were assigned to a psychosis biotype, of those, 70.12% were to the same biotype as their affected family members. These findings suggest two distinctive psychosis-related cognitive biotypes with differing functional brain patterns shared with their relatives. Instead of traditional diagnosis, patient stratification based on these biotypes may help optimize future research and identify biological targets for the treatment of cognitive dysfunction in psychosis.
BibTeX
@inproceedings{andrescamazon2025neurobiology,
  author = {Pablo Andrés-Camazón and Covadonga M. Díaz‐Caneja and Ram Ballem and Jiayu Chen and Vince D. Calhoun and Armin Iraji},
  title = {{Neurobiology-based cognitive biotypes using multi-scale intrinsic connectivity networks in psychotic disorders}},
  booktitle = {Schizophrenia},
  year = {2025},
  doi = {10.1038/s41537-025-00593-2},
}
Figure