Sixty-six Korean patients with MDD were recruited from the clinical trials program of the Samsung Medical Center Geropsychiatry and Affective Disorder Clinics (Seoul, Korea). Patients fulfilled the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR), criteria for major depressive episode.²¹²² Diagnoses were confirmed by a board certified psychiatrist on the basis of an initial clinical interview and a structured research interview.²³²⁴ A minimum baseline 17-item Hamilton Rating Scale for Depression (HAM-D) score of 15 was required.²⁵ Exclusion criteria were pregnancy, significant medical conditions, abnormal laboratory baseline values, unstable psychiatric features (e.g, suicide attempt), histories of alcohol or drug dependence, seizure, neurological illnesses, including significant cognitive impairment, or concomitant Axis I psychiatric disorders (schizophrenia, bipolar affective disorder, primary diagnosis of adjustment disorder, or posttraumatic stress disorder). No patient met the DSM-IV criteria for the specifier ‘Severe with Psychotic Features’, and none had received psychotropic medication with the current episode. In addition, no patients had received psychotropic medication within 4 weeks. Twenty-two healthy volunteers without histories of psychiatric illness were recruited by advertisement. A psychiatrist screened the control group with an interview and then biochemical tests were administered. Volunteers with a family history of mood disorders or evidence of inflammatory illness (flu, allergic disease, or dermatitis) or immunopathies were not included. The protocol was approved by the ethics review board of Samsung Medical Center, Seoul, Korea.

Patients received an antidepressant monotherapy for 6 weeks. SSRIs [escitalopram (n=31), sertraline (n=2), paroxetine (n=11), fluoxetine (n=1)] and mirtazapine (n=21) were chosen by clinician based on the anticipated adverse effects and the symptomatic characteristics of patients. Dose titration was completed within two weeks to check compliance. The HAM-D scores²⁶ were obtained by a single trained rater every two weeks. A response was defined as a decrease in HAM-D score of 50% or more at week 6.²⁷ The rater and laboratory workers were blinded to the purpose of this study. In addition, cytokine data were not disclosed to the rater and the HAM-D scores were not exposed to the laboratory workers.

Through a literature survey based on their likely importance for the psychiatric disorder, we initially selected 19 cytokines (exotaxin, sCD40L, IL-1a, IL-1b, IL-2, IL-4, IL-7, IL-8, IL-10, IL-12, FGF-2, MCP-1, TNF-alpha, alpha2-macroglobulin, Apoprotein AI, Apoprotein E, C3, INF-gamma and MIP-1alpha). We evaluated the analytical performance of the commercially available assay kits, Human Cytokine/Chemokine Magnetic Bead Panel 1 and Human Neurodegenerative Disease Panel 1 (Millipore, Billerica, MA, USA). Seven cytokines (eotaxin, sCD40L, IL-8, MCP-1, TNF-alpha, INF-gamma, and MIP-1alpha) were finally selected in the consideration of precision, calibration range and the limit of quantification of the assay method. These inflammatory markers have been shown to be relevant in psychiatric diseases.^714152829 Venous blood was drawn at baseline and at week 6. After serum isolation by centrifugation, the samples were stored at -80℃. Inflammatory markers were measured using 25 µL of serum with a Luminex 200 analyzer (Luminex Corporation, Austin, TX, USA) with the Human Cytokine/Chemokine Magnetic Bead Panel 1.

Markers for which more than 10% of the samples had values below limit of quantitation (LOQ) at baseline were not included in the analysis.³⁰ Two cytokines were excluded from the analysis because they frequently had values below the LOQ (53.4% for INF-gamma, and 67.0% for MIP-1alpha). Eotaxin (none), sCD40L (none), IL-8 (1.1%), MCP-1 (none), TNF-alpha (8.0%) were included in the analysis. For each marker retained for analysis, samples with undetectable values were given the value zero.

Continuous variables are presented as median and interquartile range. The Wilcoxon rank-sum test was employed to compare these variables between healthy control subjects and patients or the responsive and non-responsive groups. Categorical variables were presented as frequencies and proportions, and tested with Fisher's exact test. Median regression was employed to compare non-normally distributed continuous variables adjusting for other variables. The Wilcoxon signed-rank test was used to compare the cytokine values of pre- and post-antidepressant treatment. Relationships between two continuous variables were tested by Spearman's correlation analysis. p-values were corrected by Bonferroni's correction for the multiple association tests of cytokines and noted as corrected p. All differences were considered to be significant at p<0.05. Statistical analysis was undertaken using STATA SE 10.0 for Windows (College Station, TX, USA).

Clinical and demographic characteristics are shown in Table 1. There were no notable differences between the control subjects and patients with respect to age and sex. The rate of response to antidepressants was 40 of 66 patients (60.6%). A family history of depression was present in 15.2% of the patients. The median baseline HAM-D score was 19 (interquartile range: 17-21), indicating moderate to severe depression. Age, sex, episode number, current episode duration, age at onset, and baseline HAM-D score were not associated with antidepressant responsiveness. In addition, choice of drug had no effect on the rate of response (p>0.05). The median HAM-D score gradually decreased from 19 to 9 (week 2: 15,^{12131415161718} week 4: 12,^{91011121314151617} week 6: 9^{67891011121314}).

Baseline plasma concentrations of cytokines in the participants are shown in Table 2 and Figure 1. The median value of sCD40L of healthy controls was higher than that of depressive patients (44950 pg/mL, 2880-66700 pg/mL vs 19700, 5990-40800 pg/mL, respectively; corrected p=0.01). This difference was maintained after adjusted for age and sex [adjusted p=0.001, 95% confidence interval (CI)=14036.24-43266.99]. IL-8 (corrected p=0.01) and MCP-1 (corrected p=0.002) were significantly associated with a diagnosis of major depressive disorder. The levels of IL-8 and MCP-1 were higher in healthy controls when compared to depressive patients. These associations were robust after adjusted for age and sex (IL-8: adjusted p=0.004, 95% CI=3.09-10.73, MCP-1: adjusted p=0.03, 95% CI=44.71-254.38). The levels of these cytokines did not differ with sex, old age (>60), late onset (>60) and recurrence (number of episode ≥2).

We found no association between baseline cytokine concentration and antidepressant response (Table 2). In addition, changes of cytokine concentrations after 6 weeks of antidepressant treatment were not associated with responsiveness.

We found the level of MCP-1 significantly increased after antidepressant treatment (corrected p=0.01) (Table 3). This association was observed in both the response (corrected p=0.04) and non-response (corrected p=0.04) groups. Moreover, the significance of difference in MCP-1 concentration between the healthy controls and depressive patients disappeared after antidepressant treatment (p=0.06, 95% CI=-2.34-168.80).

There was no significant correlation between HAM-D score and cytokine concentration at baseline and at week 6. The HAM-D score changes did not correlate with cytokine level changes after antidepressant treatment.