Serum proteomic profiling of major depressive disorder

Abstract

Much has still to be learned about the molecular mechanisms of depression. This study aims to gain insight into contributing mechanisms by identifying serum proteins related to major depressive disorder (MDD) in a large psychiatric cohort study. Our sample consisted of 1589 participants of the Netherlands Study of Depression and Anxiety, comprising 687 individuals with current MDD (cMDD), 482 individuals with remitted MDD (rMDD) and 420 controls. We studied the relationship between MDD status and the levels of 171 serum proteins detected on a multi-analyte profiling platform using adjusted linear regression models. Pooled analyses of two independent validation cohorts (totaling 78 MDD cases and 156 controls) was carried out to validate our top markers. Twenty-eight analytes differed significantly between cMDD cases and controls (P < 0.05), whereas 10 partly overlapping markers differed significantly between rMDD cases and controls. Antidepressant medication use and comorbid anxiety status did not substantially impact on these findings. Sixteen of the cMDD-related markers had been assayed in the pooled validation cohorts, of which seven were associated with MDD. The analytes prominently associated with cMDD related to diverse cell communication and signal transduction processes (pancreatic polypeptide, macrophage migration inhibitory factor, ENRAGE, interleukin-1 receptor antagonist and tenascin-C), immune response (growth-regulated alpha protein) and protein metabolism (von Willebrand factor). Several proteins were implicated in depression. Changes were more prominent in cMDD, suggesting that molecular alterations in serum are associated with acute depression symptomatology. These findings may help to establish serum-based biomarkers of depression and could improve our understanding of its pathophysiology.

Figure 1

Regression coefficients and 95% confidence intervals of significant log10-transformed biomarkers in individuals with cMDD and rMDD compared with controls (reference group, total n=1589). Results were from linear regression analyses, all analyses were conducted separately for each log10-transformed biomarker. The figure only shows biomarkers that were statistically different in cMDD or rMDD compared with controls (P<0.05). Results are sorted by P-value. All models were adjusted for sex and age. Other relevant covariates were selected by stepwise regression. The following additional covariates were considered: ethnicity, research center, plate, smoking, alcohol intake, BMI, physical activity, corticosteroid use, anti-inflammatory drug use, sex hormone use, diabetes treatment and cardiovascular medication use. *Analytes that significantly differ between rMDD and cMDD (P<0.05). cMDD, current major depressive disorder; rMDD, remitted major depressive disorder.