Two hundred nine patients presenting for treatment of an ongoing depressive episode, who met the Diagnostic and Statistical Manual of the American Psychiatric Association (DSM-IV) criteria for MDD, were enrolled in the study. They received treatment in the Department of Psychiatry, Sungkyunkwan University, Seoul, South Korea. A diagnosis of MDD was based on clinical assessment by a board certified psychiatrist, followed by the Structured Clinical Interview for DSM-IV Axis I Disorders-Clinical Version (SCID-IV),¹⁰ which was administered by a trained research assistant. In accordance with our routine clinical procedures, these interviews were conducted with the patient and at least one family member, who provided collateral information. The clinical observations, medical records, past histories, and the DSM-IV criteria were also assessed independently by a second board-certified psychiatrist (DKK). At entry, all patients received an additional semi-structured diagnostic interview, the Samsung Psychiatric Evaluation Schedule (SPES).¹¹ The SPES provides information about psychiatric symptoms, comorbid psychiatric diagnoses, and psychosocial variables (age, sex, age of onset, duration of current episode, episode number, family psychiatric history with a focus on mood disorders), and initial severity rating with the Hamilton Depression Rating Scale (HAM-D).¹² Each patient's medical records were subjected to a retrospective and detailed review to extract demographic data, age of onset, and family history. These reviews were performed without knowledge of the BDNF genotypes. Inclusion criteria were 18 years of age or older, the existence of a unipolar major depressive episode as verified by DSM-III/IV criteria, elapsed time of at least 2 years since first episode onset, and consent for inclusion.

Exclusion criteria in the presenting episode were pregnancy, significant medical conditions, abnormal laboratory baseline values, a history of alcohol or drug dependence, seizure, neurological illness, or concomitant Axis I psychiatric disorder.

One hundred one healthy volunteers with no history of psychiatric illness were recruited by advertisement. Healthy control participants were recruited on the basis of age and sex matching to patients. They were screened by the Minnesota Multiphasic Personality Inventory (MMPI), on which they had validity and clinical scale scores ranging from 35 to 65.¹³ All patients and control subjects are ethnic Koreans.

The protocol was approved by the Institutional review board of Samsung Medical Center, Seoul, South Korea. Signed informed consent was obtained from all participants.

In keeping with DSM-IV-TR, we defined a chronic major depressive episode as one that is continuously present for at least 2 years.¹⁴ The recurrence rate may increase with each subsequent episode.¹⁵ In addition, with each subsequent episode of depression, the length of time to recurrence is shortened. At a certain point, the progressively more rapid time to recurrence in sequential episodes appears to level off, occurring within the third or fourth episode.¹⁵ Thus, the recurrent group (with three or more previous episodes) has a significantly higher relapse rate after future recoveries.¹⁵ Based on this evidence, recurrent tendencies in this study were defined as three or more lifetime episodes including the current episode.

Venous blood was drawn into Vacutainer tubes containing EDTA as an anticoagulant. Samples were then prepared into plasma and buffy coat by centrifugation at 3500 rpm for 10 min. Genomic DNA (gDNA, chromosomal DNA) was purified from lymphocytes in the buffy layer using Wizard gDNA purification kit (Promega Inc, Madison, WI, USA).

To examine the BDNF Val66Met polymorphism, an RFLP (Restriction Fragment Length Polymorphism) method was applied.¹⁶ In brief, the polymorphic region was amplified with primers flanking the exon 2 region of the BDNF gene. These primers were 5'-ACTCTGGAGAGCGTGAAT-3' and 5'-ATACTGTCACACACGCTC-3'. The Polymerase Chain Reaction (PCR) reaction mixture consisted of 20 ng gDNA, 0.25 mM dNTP mix, 1x buffer containing 1 mM MgCl₂, 2.5 mM each of forward and reverse primer, and 0.2 uL of i-MaxII polymerase (iNtRon Biotech, local company in R.Korea). The reaction was performed under the following conditions, one cycle of pre-denaturation at 95℃ for 5 min, 35 cycles of denaturation at 95℃ for 30 sec, annealing at 57℃ for 45 sec, and elongation at 72℃ for 45 sec and one cycle of post-elongation at 72℃ for 10 min. Then, 10 uL amplified 308 bp product was fragmented by incubation with 15 units of the restriction enzyme, NlaIII (NEB Inc., Beverly, MA, USA) at 37℃ overnight. The restriction fragments were separated on a 3% agarose gel containing ethidium bromide through electrophoresis at 100 V for 1 hour and visualized on a UV transluminator. The cutting band pattern was 168 bp+75 bp+65 bp for the homozygous Methionine (Met)-BDNF allele, 243 bp+65 bp for the homozygous Valine (Val)-BDNF allele, and 168 bp+75 bp+65 bp+243 bp for the heterozygous Val/Met genotype. Figure 1 shows the cutting band pattern.

For continuous variables, we presented median and interquartile range as descriptive statistics, and the Mann-Whitney test was used for two group comparisons because the variables were not normally distributed. Normality was examined by the Kolmogorov Smirnov test and the Shapiro Wilk test. For categorical variables, frequencies and proportions were presented and the Chi-square test was performed to compare the distributions of the proportions between groups. To investigate gene-dose effect of Met-BDNF allele, the Cochran-Armitage test was applied. Bonferroni's correction was used to control inflated type I error rate due to multiple testing. Deviations from Hardy-Weinberg equilibrium in the patients and control samples for the BDNF Val66Met polymorphisms were investigated. We assumed that the maximum Met/Met homozygote proportion in patients with duration <24 month and with ≥24 month was 0.1 and 0.5, respectively. Power calculation indicated power >80% at the 0.05 level of significance. Results were considered as statistically significant at two-sided p-value <0.05. All statistical analyses were performed using PASW 17.0.

Descriptive statistics for clinical and demographic characteristics are shown in Table 1. Among 209 patients, 47 (22.5%) were male, 162 (77.5%) were female, and the age range was 22-85 years (median 62, interquartile range 56-67). Healthy control participants comprised 21 (21%) males and 80 (79%) females with an age range of 46-77 years (median 61, interquartile range 58-64). There was no significant difference between the MDD and healthy control group in age (p=0.453) and sex (p=0.735). There was no significant difference in sex, age, age of onset, initial HAM-D scores, or first degree family history between patients with and without chronicity. The age (p=0.383) and sex (p=0.144) distributions between patients with and without recurrent tendencies were not different. However, patients with recurrent tendencies exhibited earlier ages of onset (46.5 vs. 56, p<0.001 by Mann-Whitney test), higher initial HAM-D score (21 vs. 19, p=0.033 by Mann-Whitney test) and higher rate of first degree family history (36% vs. 12%, p<0.001 by chi-square test). Patients with a positive first degree family history had a 67% rate of recurrent illness (31 of 46), whereas patients with a negative first degree family history had a 34% rate of recurrent illness (55 of 163) (odds ratio 4.06 (2.02-8.14); p<0.0001 by chi-square test).

The detailed distributions of the gene variants of the BDNF Val66Met polymorphism, consisting of count and frequency (%) of genotypes (Met/Met, Met/Val and Val/Val), and of alleles (Met, Val) are shown in Table 2. Among 209 MDD patients, 62 patients (29.7%) had the Val/Val genotype group, 102 (48.8%) had the Val/Met genotype group and 45 patients (21.5%) had the Met/Met genotype. Genotype distribution of 101 healthy control subjects was Val/Val (22.8%), Val/Met (61.4%) and Met/Met (15.8%). The distributions of Val66Met genotypes did not differ significantly between patients and healthy controls (p=0.228 by chi-square test). The Val and Met-BDNF allele distributions also did not differ significantly between the healthy control group and patient group (p=1.000 by chi-square test).

Patients with Val/Val genotype showed the lowest chronicity rate (12.9%) and Met/Met genotype group had the highest chronicity rate (33.3%). Moreover, there was a gene-dosage effect for Met-BDNF allele on chronicity showing linear relationship between the number of Met-BDNF alleles and chronic depression (p=0.024 by Cochran-Armitage test). There was a statistically significant difference in chronicity rates (27.1% vs .16.8%, p<0.022 by chi-square test) associated with the Met and Val alleles, respectively (Table 2). Compared with the Val/Val genotype, the relative risk of chronicity was 1.67 for the Val/Met genotype, and 2.58 for the Met/Met genotype.

There were no significant associations of BDNF genotype or allele with recurrent course of illness (Table 2).

Among the BDNF gene polymorphisms, the Met allele showed a significant dosage association with chronicity of depressive episodes. Compared with the Val/Val genotype, the relative risk of chronicity was 1.67 for the Val/Met genotype, and 2.58 for the Met/Met genotype. Knowledge of a patient's Val66Met BDNF genotype, therefore, may signal the need for early intensive treatment aimed at preventing chronic depressions.