Abstract
These findings suggest two distinctive psychosis-related cognitive biotypes with differing functional brain patterns shared with their relatives. Patient stratification based on these biotypes instead of traditional diagnosis may help to optimize future research and clinical trials addressing cognitive dysfunction in psychotic disorders.
Introduction
Cognitive dysfunction is a core feature of psychotic disorders like schizophrenia and bipolar disorder, yet it remains notoriously difficult to treat. A major roadblock is biological heterogeneity: two patients with the same clinical diagnosis (e.g., Schizophrenia) may have entirely different underlying brain pathologies causing their cognitive struggles.
Traditional diagnostic tools like the DSM rely on observable symptoms rather than neurobiology. This study seeks to move beyond symptom-based grouping by identifying "biotypes"—subgroups defined by specific patterns of functional brain network connectivity (FNC) that correlate directly with cognitive performance. By mapping functional brain connectivity to cognitive performance, researchers identified two distinct "cognitive biotypes" that cut across traditional DSM diagnoses and are frequently shared with first-degree relatives.
Methodology and Data
The study analyzed a massive dataset from the Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) Consortium 1 and 2, totaling 2,270 individuals (1,179 patients, 465 first-degree relatives, and 626 healthy controls).
The researchers employed a multi-stage analytic pipeline to identify these biotypes:
- Neuromark Framework: A standardized framework used a canonical template from over 100,000 individuals to identify 105 multi-scale intrinsic connectivity networks (ICNs).
- MOO-ICAR: Subject-specific brain networks were estimated using Multivariate-Objective Optimization Independent Component Analysis with Reference.
- PCA-CCA: To bridge brain function and behavior, Principal Component Analysis and Canonical Correlation Analysis identified specific FNC features (Cognitive FNC Features in Psychosis, or CFPs) most strongly associated with cognitive scores.
- K-means Clustering: Unsupervised clustering grouped patients based on these CFPs into two distinct biotypes.
To ensure robustness, the sample was split into a Discovery set (80%) and a Replication set (20%), with the clustering patterns successfully validated across both.
Key Insights: Transdiagnostic and Familial Biotypes
The identified biotypes were found to be transdiagnostic, meaning a patient with Bipolar Disorder could share the same brain connectivity signature as someone with Schizophrenia. This suggests that for cognitive issues, brain function is a more reliable classifier than clinical symptoms.
Furthermore, these patterns appear to have a strong familial component. Approximately 70.12% of relatives belonged to the same biotype as their affected family member. This suggests these connectivity patterns are likely rooted in shared genetic or early environmental factors, rather than just being a byproduct of the illness or medication.
Characterizing the Two Biotypes
The two biotypes represented distinct pathologies rather than just varying degrees of severity:
- Biotype 1: Characterized by widespread hypoconnectivity, particularly in cerebellar-subcortical and somatomotor-visual networks. These patients exhibited significantly worse cognitive performance and a higher prevalence of childhood learning difficulties.
- Biotype 2: Exhibited hyperconnectivity in somatomotor-subcortical networks but hypoconnectivity in high-level cognitive processing networks. These patients had relatively preserved cognitive performance.
Statistical analysis showed that these biotypes were more consistent across relatives (70%) compared to traditional DSM diagnoses (where only ~55% of relatives matched the patient's diagnosis cluster).
Limitations and Future Directions
While promising, the study has limitations. Approximately 60% of the sample was White/Caucasian, which may limit generalizability. Additionally, the use of "hard" K-means clustering forces patients into discrete buckets, whereas some may exist on a spectrum. Future work should also adjust for specific drug use or cardiometabolic conditions that impact cognition.
Ultimately, these biotypes offer a more precise neurobiological framework. Stratifying patients by biotype in future clinical trials could lead to more targeted cognitive remediation or pharmacological treatments, addressing the specific biological causes of impairment in each group.
Summary updated