Personality classification enhances blood metabolome analysis and biotyping for major depressive disorders: two-species investigation.

BACKGROUND

The relationship between depression and personality has long been suggested, however, biomarker investigations for depression have mostly overlooked this connection.

METHODS

We collected personality traits from 100 drug-free patients with major depressive disorders (MDD) and 100 healthy controls based on the Five-Factor Model (FFM) such as Neuroticism (N) and Extraversion (E), and also obtained 63 plasma metabolites profiles by LCMS-based metabolome analysis.

RESULTS

Partitional clustering analysis using the NEO-FFI data classified all subjects into three major clusters. Eighty-six subjects belonging to Cluster 1 (C1: less personality-biased group) constituted half of MDD patients and half of healthy controls. C2 constituted 50 subjects mainly MDD patients (N  + E ), and C3 constituted 64 subjects mainly healthy subjects (N  + E ). Using metabolome information, the machine learning model was optimized to discriminate MDD patients from healthy controls among all subjects and C1, respectively. The performance of the model for all subjects was moderate (AUC = 0. 715), while the performance was extremely improved when limited to C1 (AUC = 0. 907). Tryptophan-pathway plasma metabolites including tryptophan, serotonin and kynurenine were significantly lower in MDD patients especially among C1. We also validated metabolomic findings using a social-defeat mice model of stress-induced depression.

LIMITATIONS

A case-control study design and sample size is not large.

CONCLUSIONS

Our results suggest that personality classification enhances blood biomarker analysis for MDD patients and further translational investigations should be conducted to clarify the biological relationship between personality traits, stress and depression.