Proteomic signatures of bipolar disorders

Introduction

Bipolar disorders (BD) are characterized by alternating episodes of mania or hypomania and depression. Patients are functionally impaired and they have a high prevalence of cardiovascular and metabolic comorbidities. These disorders are often associated with early life adversity. Although the pathophysiology is not yet fully understood, current evidence points to immunometabolic, inflammatory, and mitochondrial abnormalities. To date, most studies have focused on isolated biomarkers, which limits the interpretation of systemic biological processes and understanding of heterogeneity. In this context, high-dimensional proteomics offers a promising approach for identifying biological signatures and molecular subgroups in BD, paving the way for precision psychiatry.

Objectives

The primary objective was to identify proteomic differences between patients with BD and healthy controls and to explore biological heterogeneity within the BD group using unsupervised methods. The secondary objective was to characterize the identified subgroups using circulating inflammatory and endothelial markers, clinical variables, and childhood adversity.

Methods

Plasma samples from 248 patients with BD and 76 healthy controls were analyzed using the Olink Reveal platform, which quantifies 1,034 proteins. For the case-control analysis, linear models adjusted for age, sex, and BMI were applied, with imputation of missing data and sensitivity analysis performed on complete cases. Genetic Set Enrichment Analysis (GSEA) was performed using the KEGG and Reactome databases. Subsequently, within the BD group, unsupervised clustering was performed based on 100 proteins with the highest variability. The resulting clusters were compared using differential proteomic analysis, biological enrichment, and an independent panel of circulating inflammatory biomarkers (CRP, SAA, sICAM-1, and sVCAM-1). Finally, associations with clinical variables and childhood trauma were explored.

Results

For the first objective based on the comparison between patients and controls, 527 differentially expressed proteins were identified in the imputed analysis and 136 in the complete-case analysis, with high agreement between the two models. The differentially expressed proteins were primarily associated with inflammation, metabolism, oxidative stress, vascular biology, and intracellular processes, while pathway enrichment was limited, with enrichment observed only in B-cell signaling and pathways related to chronic myeloid leukemia in BD. For the second objective, the unsupervised analysis of BD identified two robust proteomic subgroups. One of these clusters showed a significant increase in protein expression, with enrichment in pathways linked to intracellular signaling, FTL3 signaling, cellular stress response, senescence, and transcriptional regulation. Among circulating markers, only SAA consistently differentiated the two clusters and was associated with the overall proteomic burden. Clinical variables did not show relevant differences between the subgroups. In contrast, childhood physical abuse was higher in the cluster with greater proteomic expression.

Conclusions

BD exhibits significant biological heterogeneity that is not adequately captured by either broad diagnostic comparisons or conventional clinical variables. Proteomic stratification allowed for the identification of biologically distinct subgroups, one of which was associated with a stronger inflammatory signal and greater exposure to childhood physical abuse. These findings support the value of proteomics for the biological characterization of BD and for future stratification strategies in precision psychiatry.

References

No references