Background: Relationships between brain-derived neurotrophic factor (BDNF), interleukin (IL)-6, and salivary cortisol and both depression severity and treatment response were assessed in patients enrolled in a double-blind, placebo-controlled trial of desvenlafaxine 50. mg/d for MDD. Methods: Outpatients with MDD were randomly assigned to 12weeks of double-blind treatment with desvenlafaxine 50mg/d or placebo (2:1). Baseline severity was assessed using the 17-item Hamilton Rating Scale for Depression (HAM-D17); treatment response at week 12 was based on HAM-D17 total score and response and remission status. Saliva (cortisol) and blood (BDNF, IL-6) samples for biomarker assay were collected at baseline and week 12. Spearman correlations were calculated between the biomarkers at baseline, and between biomarkers and HAM-D17 total score at baseline. Logistic regression analyses were used to assess whether baseline biomarker levels predicted treatment response at week 12, with and without adjustment for baseline HAM-D17 score, treatment, and geographic region. Similarly, an analysis of covariance was used to assess whether baseline disease severity predicted biomarker change at week 12. Results: A total of 427 patients who received ≥1 dose of study drug and had baseline and ≥1 on-therapy primary efficacy evaluations were included in the analysis. At baseline, there was a statistically significant although weak correlation between levels of IL-6 and BDNF (Spearman correlation coefficient [rs]=0.120; P=0.014), but no significant correlation between baseline biomarker levels and baseline HAM-D17 total score (absolute value of all rs, ≤0.061). Desvenlafaxine 50mg/d treatment significantly reduced HAM-D17 total score from baseline at week 12 compared with placebo (P=0.006), but the three potential biomarkers did not predict treatment effects. No significant correlations were observed between the change from baseline in any biomarker level and change in HAM-D17 total score at week 12, either overall, or in desvenlafaxine or placebo groups (absolute value of all rs, 0.003-0.196). Baseline levels of BDNF, IL-6, and salivary cortisol did not significantly predict response to treatment at week 12. Although median increase in BDNF was not significantly different between desvenlafaxine (13.7%) and placebo (5.7%) groups, the increase was significantly greater (33.4% vs 4.3%; P=0.003) in patients with more severe depression at baseline (HAM-D17>22) vs those with less severe depression (HAM-D17≤22). No similar findings were observed for IL-6 or salivary cortisol. Discussion: Weak or no relationships were observed at baseline between the potential biomarkers or between biomarkers and disease severity. While baseline biomarker level did not predict treatment response, improvement in BDNF was significantly greater among patients who were more severely depressed at baseline.© 2013 Elsevier Inc.