http://jop.sagepub.com/Journal of Psychopharmacology

http://jop.sagepub.com/content/24/12/1747The online version of this article can be found at:

 DOI: 10.1177/0269881109353463

2010 24: 1747 originally published online 8 February 2010J PsychopharmacolDiana De Ronchi, Gloria Negri, Roberto Colombo, Luigi Janiri and Alessandro Serretti

Laura Mandelli, Marianna Mazza, Giovanni Martinotti, Daniela Tavian, Elisa Colombo, Sara Missaglia, Marco Di Nicola,bipolar depression: independent effect of brain-derived neurotrophic factor and harm avoidanceFurther evidence supporting the influence of brain-derived neurotrophic factor on the outcome of

  

Published by:

http://www.sagepublications.com

On behalf of: 

  British Association for Psychopharmacology

can be found at:Journal of PsychopharmacologyAdditional services and information for    

  http://jop.sagepub.com/cgi/alertsEmail Alerts:

 

http://jop.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://jop.sagepub.com/content/24/12/1747.refs.htmlCitations:  

What is This? 

- Feb 8, 2010 OnlineFirst Version of Record 

- Dec 7, 2010Version of Record >>

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

Original Paper

Further evidence supporting the influenceof brain-derived neurotrophic factor on theoutcome of bipolar depression: independenteffect of brain-derived neurotrophic factorand harm avoidance

Journal of Psychopharmacology

24(12) 1747–1754

! The Author(s) 2010

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/0269881109353463

jop.sagepub.com

Laura Mandelli1, Marianna Mazza2, Giovanni Martinotti2,Daniela Tavian3, Elisa Colombo3, Sara Missaglia3, Marco Di Nicola2,Diana De Ronchi1, Gloria Negri3, Roberto Colombo3,4, Luigi Janiri2

and Alessandro Serretti1

AbstractBrain-derived neurotrophic factor is a candidate gene for response to antidepressant treatment. However, response to pharmacological treatments is

moderated by both genetic and other factors within individuals. For example, there is evidence of an influence of the temperamental trait of harm

avoidance on the outcome of depressive disorders. In the present study we aimed to investigate the effect of the brain-derived neurotrophic factor gene

on medium-term outcome in a naturalistic sample of 86 depressed bipolar spectrum patients, taking into account harm avoidance.

Both single marker and haplotypes were significantly associated with severity of depression at month 6 after treatment initiation. The haplotype

comprising the A-C alleles was associated with a poorer outcome. Harm avoidance maintained a significant effect on depressive outcome in bipolar

disorder, independently from brain-derived neurotrophic factor genotypes. However, harm avoidance’s influence appeared to be more consistent in

patients carrying the protective G-T combination of alleles.

Our results indicate brain-derived neurotrophic factor as involved in the outcome of depression in bipolar disorder. Harm avoidance did not

interact with brain-derived neurotrophic factor genotypes, though its effect was still significant. Given that many factors may influence response

to pharmacological treatments, studies that consider personality and other individual characteristics are warranted also in pharmacogenetic

investigations.

Keywordsbipolar disorder, depression, BDNF gene, harm avoidance, outcome

Introduction

Much evidence exists of a genetic influence on the responseto pharmacological treatments. The past few decades have

witnessed much progress in the field of pharmacogenetics,and a number of interesting genes have been suggested asputative regulators of response to antidepressants (Drago

et al., 2009).Among these, recent attention has been focused on

brain-derived neurotrophic factor (BDNF), a neurotrophin

involved in the processes of differentiation and neuronal resil-ience. BDNF has been linked with severe psychiatric disor-ders, particularly schizophrenia (Ashe et al., 2001; Lu andMartinowich, 2008) and bipolar (BP) disorder (Fan and

Sklar, 2008; Serretti and Mandelli, 2008). Of interest, invitro and in vivo data provided direct evidence that BDNFis a key mediator of the therapeutic response to antidepres-

sants (see D’Sa and Duman, 2002). Chronic, but not acute,selective serotonin reuptake inhibitor treatment has beenassociated with an increased hippocampal expression of

BDNF in adult rats (Nibuya et al., 1995).

A first clinical study in humans reported reduced BDNFprotein levels in the brains of unmedicated depressed patients(Duman et al., 1999). It has been thus hypothesized thatdecreased levels of BDNF could participate in the hippocam-

pal atrophy observed in depressed patients. Indeed, BDNFhas also trophic effects on the serotoninergic neurons in thecentral nervous system (Mamounas et al., 2000). Further,

antidepressant treatment has been shown to restore serumlevels of BDNF in human patients with depression in several

1Department of Psychiatry, University of Bologna, Bologna, Italy.2Department of Psychiatry, Catholic University, Gemelli Hospital, Rome,

Italy.3Laboratory of Human Molecular Biology and Genetics, Catholic

University, Milan, Italy.4Institute of Biochemistry and Clinical Biochemistry, Catholic University,

Gemelli Hospital, Rome, Italy.

Corresponding author:Laura Mandelli, Institute of Psychiatry, University of Bologna, Viale C.

Pepoli 5, 40123, Bologna, Italy. Email: laura.mandelli@unibo.it

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

studies (Aydemir et al., 2005; Gervasoni et al., 2005; Gonulet al., 2005; Piccinni et al., 2008; Shimizu et al., 2003).

The serotonin transporter (SERT) is the major target

of antidepressants. Recently, evidence has been reportedthat constitutive reduction in BDNF levels modulates SERTreuptake capacity in adult hippocampus, while inducedhigher levels of BDNF in the hippocampus enhances SERT

function (Benmansour et al., 2008; Daws et al., 2007). Based onthis evidence, a number of studies, though still preliminary,investigated BDNF genetic variants on the acute response to

antidepressant treatment in human patients with depression.One of the most investigated genetic variations withinthe BDNF gene is rs6265 (196G/A), resulting in a Val66 to

Met (V66M) change in the 50-pro-region of the protein.This variation has been associated with poorer episodicmemory and abnormal hippocampal activation (Egan et al.,

2003). After a first negative report (Tsai et al., 2003), thers6265 A-allele has been associated with a better responseto acute treatment (6–8weeks) with antidepressants intwo independent Asian populations (Choi et al., 2006;

Yoshida et al., 2006), while negative findings werereported in Caucasians (Gratacos et al., 2008; Wilkie et al.,2007). However, Gratacos et al. found other polymorphisms

(haplotype rs12273363-rs908867-rs1491850) associated withremission after one adequate antidepressant treatment(at least 6weeks), and a recent meta-analysis confirmed

the association between the rs6265 polymorphism andresponse to antidepressants (Kato and Serretti, in press).However, controversial data have been subsequently obtainedin both Caucasians and Asiatics (Kang et al., in press, Licinio

et al., 2009).The role of BDNF in the treatment of BP depression, i.e.

not restricted to antidepressants, is less clear. There is evi-

dence of an involvement of BDNF in lithium prophylacticefficacy. Rybakowski et al. (2005) reported an associationbetween the rs6265 A-allele and a better response to lithium

prophylactic treatment for BP disorder, though no replicationcould be obtained on a Asiatic sample (Masui et al., 2006).Further, Rybakowski et al. found BDNF significantly inter-

acting with SERT gene (5-HTTLPR polymorphism) in mod-erating the efficacy of lithium treatment (Rybakowski et al.,2007b). Recently, changes in serum BDNF levels have beenalso reported after treatment with both mood stabilizers and

antipsychotics for an episode of acute mania (Tramontinaet al., 2009).

Inconsistencies among studies may be due to a number of

reasons, such as sample sizes, ethnicity, criteria for the selec-tion of patients, and other across-study methodological dis-parities. Further, the contribution of genes to the risk of

disorders and response to treatments is relatively small(Craddock and Jones, 1999) and individual and environmen-tal factors may interfere, influencing the observed phenotype.It is commonly acknowledged that complex human beha-

viours result from the interplay between the individual geneticprofile and other traits, both genetically and environmentallymediated. This has been recognized in many psychiatric dis-

orders as well (see Jaffee and Price, 2007). However, psycho-pharmacological studies rarely take into account otherclinical and individual features as predictors of response to

treatments (Serretti et al., 2008).

Features consistently associated with the risk for mooddisorder and response to treatment are ‘neurotic’ or ‘anxie-ty-related’ personality traits, also called neuroticism (Costa

and McCrae, 1989) or harm avoidance (HA) (Cloninger,1987). Studies have shown prevalent neurotic traits in patientswith history of depressive disorders and resistant to treat-ments (Pelissolo and Corruble, 2002). In a previous investi-

gation on a sample of BP spectrum patients, we found HAsignificantly influencing the outcome of a depressive episodeover a follow-up period of six months (unpublished data).

Further, we found HA to moderate the effect of a commonpolymorphism within SERT (5HTTLPR) on response totreatment (Mandelli et al., 2009).

In the present study, we aimed to investigate the effect ofBDNF (rs6265 and rs11030104 polymorphisms) in the samesample of BP spectrum patients: a naturalistic sample of

depressed BP patients followed up to six months. Wehypothesized a role for BDNF in the outcome of BP depres-sion after treatment with various agents (antidepressants,mood stabilizers and antipsychotics), according to previous

but sparse evidence. Moreover, we hypothesized a modulat-ing effect of personality traits, particularly HA, as previouslyobserved in our sample.

Methods

Sample and evaluations

Sample and methods of evaluation have been previouslydescribed (Mandelli et al., 2009). Briefly, patients were

included in the study if they met DSM-IV criteria for a BPspectrum disorder (Akiskal et al., 1977): Bipolar I (BP-I,n¼ 41, 47.7%), Bipolar II (BP-II, n¼ 20, 23.2%) or cyclothy-

mic disorder (CtD, n¼ 25, 29.1%), satisfying criteria for acurrent major depression (by the SCID-I interview (Firstet al., 1995)), and aged between 18 and 80 years. Exclusion

criteria were represented by mental retardation or docu-mented IQ< 70, cognitive impairment (evaluated by theMini Mental State Examination (Crum et al., 1993)), unstable

or severe medical condition that could impair evaluations,pregnancy or breastfeeding.

Finally, 86 unrelated Italian individuals, with Italian ante-cedents of at least three generations, were included in

the sample after they accepted the study procedures andsigned the informed consent as approved by the local ethicalcommittee. Patients were treated with conventional antide-

pressants and/or mood stabilizers at standard therapeuticdoses for at least three months. Concomitant treatmentwith antipsychotics and sedative/anxiolytic drugs did not

represent an exclusion criterion (Table 1). Treatment wasmaintained for six months when possible (drop out andswitches to mania listed in Table 1), whereas dosages ofdrugs could be adjusted on the basis of the clinician’s judg-

ment. At intake, patients were evaluated for Axis II person-ality disorders by the SCID-II (First et al., 1990) and for HAby the Italian version of the TCI-R (Martinotti et al., 2008).

Depressive symptoms’ severity was evaluated by theHamilton Depression Rating Scale (HDRS, or HAMD)(Hamilton, 1960) at intake and after one, three and six

months of treatment.

1748 Journal of Psychopharmacology 24(12)

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

Genetic analysis

Genomic DNA was extracted from a sample (200 ml) ofperipheral blood using High Pure PCR Template

Preparation Kit (Roche Diagnostic GmbH, Mannheim,Germany) according to the manufacturer’s instructions.Genotyping was performed by a duplex PCR followed by a

double digestion. A total of 50 ng of genomic DNA wasamplified in a total volume of 25ml containing GoTaq� 5XFlexi Buffer (1X; Fermentas Inc., Burlington, ON), MgCl2(2mM), dNTP (4mM), the oligonucleotide primers for thetwo BDNF fragments encompassing the rs6265 andrs11030104 loci (1 mM each; Table 1), and GoTaq� DNA

polymerase (1U; Promega Co., Madison, WI). Thermalcycling (PTC-200; MJ Research Inc., Waltham, MA) con-sisted of 5min of initial denaturation at 95�C followed by 6cycles of 94�C (30 s), 68�C (30 s) and 72�C (30 s), and 30 cycles

of 94�C (30 s), 62�C (30 s) and 72�C (30 s), with a final exten-sion step at 72�C (5min). Subsequently, 5ml of PCR productwas simultaneously digested with Eco72I and RsaI (2U each;

Fermentas). Digestion was performed in a 20ml reaction assay(Tango� buffer, Fermentas; 37�C; 4 h). For rs6265, two frag-ments (351 and 212 bp) were obtained in the case of the G

allele and just one (563 bp) in the case of the A allele, while forrs11030104 two fragments (398 and 115 bp) were obtained forthe C allele and a single fragment (513 bp) for the T allele.Digestion products were separated by electrophoresis on 3%

agarose gel containing ethidium bromide.

Statistical analysis

‘Haploview’ was employed to generate a map of linkage dis-equilibrium (LD) among markers and to calculate the

Hardy-Weinberg equilibrium (HWE). Univariate analyses

were performed by standard tests (t-Student, one-way analy-sis of variance – ANOVA – �2) when appropriate. Analyseswere systematically controlled for duration of current episode

prior intake, baseline severity and treatment by the one-wayor repeated-measures analysis of covariance (ANCOVA).Mean treatment dosages were recorded according the‘Antidepressant Treatment History Form’ (Sackeim, 2001).

The last observation carried forward (LOCF) method wasemployed for missing data for dropped-out patients, whileno LOCF was applied in the case of switch to mania (only

one subject). The ‘R’ environment (The R Project forStatistical Computing) was employed to analyse haplotypeassociation (haplo.stat package) with both discrete and con-

tinuous variables and controlling for covariates. Permutationanalysis (10,000) was performed in order to have a betterestimation of the significance of results. The effect of HA

on genetic association was analysed by the general linearmodel (GLM) and haplo.glm command in ‘R’ for haplotypes,which allows both the inclusion of variables as covariates andas factors to be tested for interactions.

With a conservative significance of 0.01, in our sample wehad a sufficient power (> 0.80) to detect only medium effectsizes of f¼ 0.37 in ANCOVA analyses, corresponding to a

difference of approximately 1.8 points on HDRS scoresbetween genotypes (Faul, 2007).

Results

Clinical and demographic features of the sample were pre-viously presented and analysed (Mandelli et al., unpublished

data; Mandelli et al., 2009). Briefly, the sample comprised 44females (51%) and 42 males (49%). Mean age at intake was46� 11 years (range: 22–76), mean age at onset was

30� 7 years (range: 18–51) and mean depressive severity,according to HDRS, was 18.4� 4.5 (range 12–36). A person-ality disorder was present in 36% of the individuals and half

of the subjects had a lifetime history of substance use disorder(SUD), though all subjects were detoxified and abstinent atintake. This high rate of lifetime SUD was due to our inten-

tional inclusion of this kind of patients.Treatments administered to the patients are listed in

Table 1. While neither SUD nor Axis II personality disordersinfluenced response to treatment for current episode, a result of

interest was the significant effect of the temperamental trait ofHA: subjects with high scores had a poorer outcome as com-pared with those characterized by lowHA scores. Other demo-

graphic and clinical features were not different from thosepreviously observed in the same sample (Mandelli et al.,2009) or in the larger one, which included non-clinically

depressed BP patients, apart from severity of depressive symp-toms (Mazza et al., 2009).

BDNF polymorphisms

Both polymorphisms were in HWE (rs6265 p¼ 0.9, rs11030104p¼ 0.41). Genotype frequencies were as follows: rs6265 GG

0.72, GA 0.26, AA 0.02; rs11030104 TT 0.65, TC 0.29, CC0.06, in line with those observed in Caucasian subjects(NCBI Entrez SNP database). Single nucleotide polymorphisms

(SNPs) were in reciprocal full LD (D’¼ 1, LOD¼ 45.6,

Table 1. Rates of drugs administered to patients, rates of drop out and

switches to mania

Type of treatment

% of

patients

Antidepressant only 26.4

Antidepressant plus a mood stabilizer 16.7

Antidepressant plus an antipsychotic 11.1

Antidepressant plus a mood stabilizer,

plus an antipsychotic

8.3

Mood stabilizer only 13.9

Mood stabilizer plus an antipsychotic 13.9

Antipsychotic only 9.7

Sedative/anxiolytic drugs 26.4

Drop out 30.2

Drop out at month 3 5.8

Drop out at month 6 24.4

Drop out for side effects 5.8

Drop out for lack of efficacy 5.8

Drop out for lack of compliance 18.6

Switched to mania 1.2

Mandelli et al. 1749

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

r2¼ 0.75). Haplotype frequencies were the following: G-T0.80, G-C 0.05, A-C 0.15.

BDNF and demographic/clinical variables

BDNF genotypes were not differently stratified for sex, age,marital status and educational level. BDNF genotypes were

not associated with family history for psychiatric disorders,BP sub-diagnosis, age at onset, lifetime SUD, Axis II person-ality disorder, comorbidity for anxiety disorders, HA scores,

number of previous illness episodes, duration of current epi-sode prior to intake, treatment administered and doses.Haplotype analyses did not reveal any association with clin-

ical and demographic variables. Similarly, BDNF genotypesand haplotypes were not associated with baseline total depres-sive scores (Table 2) (data not shown, all p values > 0.01).

However, as shown in Table 2, rs6265 and rs11030104genotypes were significantly associated with HDRS scoresafter six months of treatment. Associations were independentfrom baseline HDRS scores, treatment dosage, sex and age.

Further, the results were not influenced by HA scores,though, as expected, HA significantly influenced HDRSscores at month 6 (p¼ 0.006). HA was not influenced by

comorbidity for anxiety disorders, as its effect on outcomewas independent from anxiety disorders.

An effect of BDNF could be observed also on response (%

reduction from baseline) at month 3 (rs6265 co-dominantmodel: F¼ 4.3, df¼ 2, p¼ 0.016, A-dominant model:t¼ 2.72, df¼ 80 p¼ 0.008; rs11030104 co-dominant modelF¼ 3.9, df¼ 2, p¼ 0.024, C-dominant model: t¼ 2.6,

df¼ 80, p¼ 0.01), other than at month 6 (rs6265 co-dominantmodel: F¼ 6.0, df¼ 2, p¼ 0.0038, A-dominant model: t¼ 3.4,df¼ 80, p¼ 0.00098; rs11030104 co-dominant model: F¼ 7.3,

df¼ 2, p¼ 0.0012, C-dominant model: t¼ 3.8, df¼ 80,p¼ 0.00026). Again, the effects were neither influenced byHA scores, nor by treatment dosage, sex and age.

By repeated measures ANOVA, both polymorphisms weresignificantly associated with response to antidepressant treat-ment over the medium term (3–6months) (rs6265: F¼ 7.91

df¼ 3, 240, p¼ 0.0005, Figure 1; rs11030104: F¼ 8.32,df¼ 3, 240, p¼ 0.00003, Figure 2). Again, results were notinfluenced by HA scores.

The analysis of haplotypes confirmed a significant associ-

ation with severity at month 6 (Global stat¼ 10.86, df¼ 2,p¼ 0.004, sim p¼ 0.003). Carriers of the A-C haplotype (fre-quency 15%) were significantly more severe than other

patients at 6months after intake (Table 3). Including HA

Table 2. BDNF genotypes and severity of depressive symptoms at intake (baseline) and at each following evaluation (month 1, 3 and 6)

rs6265rs6265 A_dom

GG GA AA GG vs. GA/AA

N¼ 62 N¼ 22 N¼ 2 N¼ 62 vs N¼ 24HDRS scores Mean� SD Mean� SD Mean� SD F, p t, p

Baseline 18.8� 4.9 17.1� 2.5 21.5� 6.4 1.58, 0.21 1.18, 0.24

Month1 15.3� 3.7 14.9� 2.2 17.5� 4.9 0.59, 0.55 0.25, 0.80

Month 3 12.7� 4.9 14.5� 4.9 22.5� 2.1 4.67, 0.012 2.04, 0.04

Month 6 11.1� 4.7 15.1� 5.2 16.5� 6.4 6.19, 0.003 3.52, 0.0007

rs11030104 rs11030104 C_dom

TT TC CC TT vs. TC/CC

N¼ 56 N¼ 25 N¼ 5 N¼ 56 vs. N¼ 30

HDRS scores Mean� SD Mean� SD Mean� SD F, p t, p

Baseline 19.0� 4.9 17.3� 2.8 17.8� 5.9 1.30, 0.28 1.6, 0.11

Month1 15.4� 3.8 15.0� 2.1 15.2� 3.6 0.12, 0.89 0.48, 0.64

Month 3 12.8� 5.0 14.2� 4.7 16.5� 4.6 1.42, 0.25 1.45, 0.15

Month 6 11.0� 4.7 14.6� 5.0 14.7� 6.4 5.20, 0.007 3.25, 0.002

HDRS: Hamilton depressive rating scale; rs6265 A_dom: dominant model for allele A; rs11030104 C_dom: dominant model for allele C.

Vertical bars denote 0.95 confidence intervals

8

10

12

14

16

18

20

22

HD

RS

sco

res

GG GA and AA

Baseline Month 1 Month 3 Month 6

Figure 1. Effect of rs6265 genotype (A dominant model) on depressive

score over time.

1750 Journal of Psychopharmacology 24(12)

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

scores as a covariate or a factor (high/low HA) in GLM, bothDBNF haplotypes and HA showed a significant effect onHDRS scores at month 6, but no interaction could be

observed between the two factors (Table 4).Finally, although no significant interaction was observed

between BDNF and HA on course of BP depression, HAdid however have a significant influence on course of symp-

toms in subjects carriers of the ‘protective’ G-Thaplotype (F¼ 5.03, df¼ 3, 312, p¼ 0.002) (Figure 3, C).

A similar trend, though not significant, could be observedfor single markers (Figure 3, A, B).

Discussion

The present study confirms a significant influence of the BDNFgene on efficacy of treatment in bipolar depression. According

to preclinical studies, BDNF may be a key mediator of thetherapeutic response to drugs (D’Sa and Duman, 2002).BDNF protein levels have been associated with SERT reup-

take capacity (Daws et al., 2007), thus BDNF may indirectlyinterfere with the action of antidepressants. There isindeed evidence that BDNF variants may interfere with the

efficacy of fluoxetine in normalizing anxiety behavioursin mice (Chen et al., 2006). Further, levels of BDNF havebeen associated with abnormal hippocampal activation

(Egan et al., 2003) and hypothesized as a factor involved inthe hippocampal atrophy observed in depressed patients.Accordingly, reduced levels of BDNF have been observed inunmedicated depressed patients (Duman et al., 1999).

Previous studies reported BDNF involved in the responseto acute antidepressant treatment (Choi et al., 2006; Gratacoset al., 2008; Licinio et al., 2009; Yoshida et al., 2006), though

negative reports have been published as well (Kang et al., inpress; Tsai et al., 2003; Wilkie et al., 2007). BDNF has beenalso been found to be associated with efficacy of treatments

for acute mania (Tramontina et al., 2009) and lithium pro-phylactic efficacy in BP patients (Rybakowski et al., 2005).Studies have mainly focused the rs6265 polymorphism(Val66Met), which reported positive associations, though

not univocally. To our knowledge, this is the first study con-firming rs6265 as involved in the efficacy of different agents intreating BP depression.

Nevertheless, which allele has to be considered the riskvariant remains unclear. We found the rs6265 A-allele asassociated with a poorer outcome, while Choi et al. (2006)

reported it associated with a better response in a Korean pop-ulation. Yoshida et al. (2006) reported evidence for a heterosiseffect in Japanese. This discrepancy may be explained by the

genetic diversity between ethnicities. Indeed, in Caucasiansthe rs6265 A-allele has a frequency of about 0.17, while inAsian populations it is more frequent, ranging from 0.34 to0.63. Rs6265 may be further in different LD with other func-

tional variants (the flip-flop hypothesis of Lin et al., 2007).A similar case has already been reported: the short-allele of5HTTLPR (SERT gene) is more frequent in Asian popula-

tions and associated with a better response to antidepressants,in contrast to the observations in Caucasian populations,where the short allele is less frequent and associated with a

poor response (Kato and Serretti, in press). On the otherhand, Rybakowski et al. reported the same rs6265 A alleleassociated with a better efficacy of long-term efficacy oflithium as a prophylactic agent in a Caucasian population

(Rybakowski et al., 2005). Thus, further studies are neededto clarify the role of BDNF rs6265 polymorphism in the treat-ment of BP depression.

In our study, we also found a second SNP, rs11030104,associated with depressive outcome and a significant effect ofthe combination of alleles in the two markers. To our knowl-

edge, only Licinio et al. (Licinio et al., 2009) investigated

Vertical bars denote 0.95 confidence intervalsBaseline Month 1 Month 3 Month 6

8

10

12

14

16

18

20

22

HD

RS

sco

res

TT TC and CC

Figure 2. Effect of rs11030104 genotype (C dominant model) on

depressive scores over time.

Table 4. Analysis of haplotypes on severity at month 6 including harm

avoidance (HA) as a covariate or as an independent factor in general

linear model

Coefficient SE t-stat p-value

HA as a covariate

(intercept)* 5.42 1.53 3.53 0.0008

HA 3.27 1.06 3.09 0.003

G-C 2.21 1.59 1.39 0.17

A-C 3.92 0.99 3.97 0.0002

HA as an interaction factor

(intercept)* 4.70 1.83 2.56 0.013

G-C x HA �0.55 4.45 �0.12 0.90

A-C x HA �1.64 2.12 �0.77 0.44

*G-T haplotype as baseline category for the design matrix.

Table 3. Haplotype association with depressive severity at month 6 after

intake

Haplotype Frequency Stat-value p-value sim p-value

G-T 0.805 �2.89 0.004 0.003

G-C 0.043 0.028 0.98 0.98

A-C 0.152 3.26 0.001 0.0009

Mandelli et al. 1751

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

rs11030104 in response to antidepressant treatment, but nosignificant association was observed.

Other than investigating the influence of BDNF variants

on depressive outcome, a secondary aim of this study was totest a potential interaction with the trait of HA, which isstrongly associated with response to treatments (Pelissolo

and Corruble, 2002). We did not find a direct interactionbetween BDNF variants and HA levels. Multivariate analysesshould be taken with caution in this study, because of thelimited power to detect true positive associations and the

low number of subjects in groups of genotypes and levels ofHA. Nevertheless, the influence of HA remained statisticallysignificant, independently from the genetic profile of indivi-

duals. Of interest is the possible differential impact of HAlevels in subjects carrying the genetic variants associatedwith a better outcome, which was not evident in subjects

carrying the risk variants. However, larger studies arerequired to confirm this preliminary observation.

In a previous analysis performed on this sample of sub-jects, we found a significant effect of the 5-HTTLPR poly-

morphism on outcome of depression as well (Mandelli et al.,2009). There is evidence of a potential interaction between thetwo genes, SERT and BDNF, on the risk for depressive dis-

orders (for example, Kaufman et al., 2006), as well as onresponse to psychotropic treatments (Rybakowski et al.,2007a). Unfortunately, we could not test for the interaction

between BDNF variants and the 5-HTTLPR polymorphism,

because the small size of our sample strongly limited the pos-sibility to detect true associations.

The present study is different from those previously

reported for two main reasons. First, the length of follow-upwas considerably longer than most of previous studies, whichwere mainly limited to the acute treatment over 6–8 weeks.

Thus, our results are not completely in disagreement withnegative studies in Caucasians, since we did not observe asignificant effect of BDNF on acute treatment (8–12 weeks),but only at 6months after treatment initiation. Thus, it is

possible that BDNF variants influence the medium-term out-come of BP depression instead of acute response. Second, weemployed only few exclusion criteria while collecting subjects,

and the inclusion of comorbid disorders such as lifetimeSUD, personality disorders (36.0%) and anxiety disorders(11.6%) was allowed. We did not put a narrow diagnostic

criterion, including BP-I, BP-II and CtM patients, and neitherdid we include a cut-off of depressive severity for inclusion(for example HDRS > 15). Patients were included if theysatisfied DSM-IV criteria for a major depressive episode inde-

pendently from severity. Further, patients did not receive asingle standardized treatment as is often done in clinical trials,and only 55% of subjects were treated with antidepressants.

These choices were made in order to obtain a ‘naturalistic’sample of BP patients, representative of real clinical settings.Indeed, the inclusion of heterogeneous patients could allow

the testing of more clinical variables that could potentially

RS6255-RS11030104 haplotypes

G-T haplotype (freq. 0.80)

6

8

10

12

14

16

18

20

22

HD

RS

G-C, A-C haplotypes (freq. 0.20)

High HA Low HA

RS6265 and HA

GG

6

8

10

12

14

16

18

20

22

24

HD

RS

sco

res

GA, AA

High HA Low HA

RS11030104 and HA

TT

6

8

10

12

14

16

18

20

22

24

HD

RS

sco

res

TC, CCBaseline Month 1 Month 3 Month 6 Baseline Month 1 Month 3 Month 6

Baseline Month 1 Month 3 Month 6 Baseline Month 1 Month 3 Month 6

Baseline Month 1 Month 3 Month 6 Baseline Month 1 Month 3 Month 6

High HA Low HA

p=0.002

A B

C

Figure 3. Interaction between HA and BDNF genotypes and haplotypes.

1752 Journal of Psychopharmacology 24(12)

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

influence the individual response to treatments independently

or in interaction with genetic variations. On the other hand,unipolar major depressive patients were not included in oursample, introducing a further discrepancy with respect to the

large part of samples investigated so far for BDNF. Thus,although our sample may be to some extent representativeof a general population of BP patients, comparison with pre-vious investigations may be problematical. Moreover, the

naturalistic characterization of the sample may have intro-duced a number of confounding factors, although we system-atically controlled for some important features. On the other

hand, a number of variables potentially influencing the out-come of BP depression were not taken into account, such asdevelopmental history, social support and current stressors.

Moreover, a major limitation of the present study is repre-sented by the small sample size, which may have lead to falsepositives and reduced the reliability of multivariate analyses.

We applied a correction to the level of significance, but thisdoes not completely ensure the reliability of results. Thus,findings should be taken with caution and further studieson larger samples are required.

In conclusion, despite many critical points, results of ourinvestigation support the involvement of BDNF in the out-come of BP depression, in accordance with some previously

published reports, though in disagreement with others. Inparticular, BDNF seem to influence the medium-termresponse to treatments rather than the acute one. The influ-

ence of HA is independent of BDNF genotypes, though fur-ther research is required to confirm this finding.

References

Akiskal HS, Djenderedjian AM, Rosenthal RH, et al. (1977)

Cyclothymic disorder: validating criteria for inclusion in the bipo-

lar affective group. Am J Psychiatry 134: 1227–1233.

Ashe PC, Berry MD, Boulton AA (2001) Schizophrenia, a neurode-

generative disorder with neurodevelopmental antecedents.

Prog Neuropsychopharmacol Biol Psychiatry 25: 691–707.

Aydemir O, Deveci A, Taneli F (2005) The effect of chronic

antidepressant treatment on serum brain-derived neurotrophic

factor levels in depressed patients: a preliminary study. Prog

Neuropsychopharmacol Biol Psychiatry 29: 261–265.

Benmansour S, Deltheil T, Piotrowski J, et al. (2008) Influence

of brain-derived neurotrophic factor (BDNF) on serotonin

neurotransmission in the hippocampus of adult rodents. Eur J

Pharmacol 587: 90–98.

Chen ZY, Jing D, Bath KG, et al. (2006) Genetic variant BDNF

(Val66Met) polymorphism alters anxiety-related behavior.

Science 314: 140–143.

Choi MJ, Kang RH, Lim SW, et al. (2006) Brain-derived neuro-

trophic factor gene polymorphism (Val66Met) and citalopram

response in major depressive disorder. Brain Res 1118: 176–182.

Cloninger CR (1987) A systematic method for clinical description

and classification of personality variants. A proposal. Arch Gen

Psychiatry 44: 573–588.

Costa PT, McCrae RR (1989) NEO Personality Inventory.

Psichological Assessment Resources, Odessa.

Craddock N, Jones I (1999) Genetics of bipolar disorder. J Med

Genet 36: 585–594.

Crum RM, Anthony JC, Bassett SS, et al. (1993) Population-based

norms for the Mini-Mental State Examination by age and educa-

tional level. JAMA 269: 2386–2391.

Daws LC, Munn JL, Valdez MF, et al. (2007) Serotonin transporter

function, but not expression, is dependent on brain-derived

neurotrophic factor (BDNF): in vivo studies in BDNF-deficient

mice. J Neurochem 101: 641–651.

Drago A, De Ronchi D, Serretti A (2009) Pharmacogenetics of anti-

depressant response: an update. Hum Genomics 3: 257–274.

D’Sa C, Duman RS (2002) Antidepressants and neuroplasticity.

Bipolar Disord 4: 183–194.

Duman RS, Malberg J, Thome J (1999) Neural plasticity to stress and

antidepressant treatment. Biol Psychiatry 46: 1181–1191.

Egan MF, Kojima M, Callicott JH, et al. (2003) The BDNF val66met

polymorphism affects activity-dependent secretion of BDNF and

human memory and hippocampal function. Cell 112: 257–269.

Fan J, Sklar P (2008) Genetics of bipolar disorder: focus on BDNF

Val66Met polymorphism. Novartis Found Symp 289: 60–72;

discussion 72–73, 87–93.

Faul F (2007) G*Power. Universitat Kiel, Germany.

First MB, Spitzer RL, Gibbon M, et al. (1990) Structured Clinical

Interview for DSM-IV Axis II Personality Disorders (SCID-II).

New York: Biometrics Research Department, New York State

Psychiatric Institute.

First MB, Spitzer RL, Gibbon M et al. (1990) Structured Clinical

Interview for DSM-IV Axis I Disorders - Patient Edition (SCID

- I/P, Version 2.0). New York: Biometrics Research Department,

New York State Psychiatric Institute.

Gervasoni N, Aubry JM, Bondolfi G, et al. (2005) Partial normal-

ization of serum brain-derived neurotrophic factor in remitted

patients after a major depressive episode. Neuropsychobiology

51: 234–238.

Gonul AS, Akdeniz F, Taneli F, et al. (2005) Effect of treatment on

serum brain-derived neurotrophic factor levels in depressed

patients. Eur Arch Psychiatry Clin Neurosci 255: 381–386.

Gratacos M, Soria V, Urretavizcaya M, et al. (2008) A brain-derived

neurotrophic factor (BDNF) haplotype is associated with

antidepressant treatment outcome in mood disorders.

Pharmacogenomics J 8: 101–112.

Hamilton M (1960) A rating scale for depression. J Neurol Neurosurg

Psychiatry 23: 56–62.

Jaffee SR, Price TS (2007) Gene-environment correlations: a review

of the evidence and implications for prevention of mental illness.

Mol Psychiatry 12: 432–442.

Kang R, Chang H, Wong M, et al. (In press) Brain-derived neuro-

trophic factor gene polymorphisms and mirtazapine responses in

Koreans with major depression. J Psychopharmacol.

Kato M, Serretti A (In press) Review and meta-analysis of antide-

pressant pharmacogenetic findings in major depressive disorder.

Mol Psychiatry.

Kaufman J, Yang BZ, Douglas-Palumberi H, et al. (2006)

Brain-derived neurotrophic factor-5-HTTLPR gene interactions

and environmental modifiers of depression in children. Biol

Psychiatry 59: 673–680.

Licinio J, Dong C, Wong ML (2009) Novel sequence variations in

the brain-derived neurotrophic factor gene and association

with major depression and antidepressant treatment response.

Arch Gen Psychiatry 66: 488–497.

Lin PI, Vance JM, Pericak-Vance MA, et al. (2007) No gene is

an island: the flip-flop phenomenon. Am J Hum Genet 80:

531–538.

Lu B, Martinowich K (2008) Cell biology of BDNF and its relevance

to schizophrenia. Novartis Found Symp 289: 119–129; discussion

129–135, 193–195.

Mamounas LA, Altar CA, Blue ME, et al. (2000) BDNF promotes

the regenerative sprouting, but not survival, of injured serotoner-

gic axons in the adult rat brain. J Neurosci 20: 771–782.

Mandelli L, Mazza M, Di Nicola M, et al. (unpublished) The role of

substance abuse comorbidity and personality on the outcome of

depression in bipolar disorder.

Mandelli et al. 1753

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from

Mandelli L, Mazza M, Martinotti G, et al. (2009) Harm avoidance

moderates the influence of serotonin transporter gene variants

on treatment outcome in bipolar patients. J Affect Disord 119:

205–209.

Martinotti G, Mandelli L, Di Nicola M, et al. (2008) Psychometric

characteristic of the Italian version of the Temperament and

Character Inventory-Revised (TCI-R). Personality, psychopathol-

ogy and attachment styles. Compr Psychiatry 49: 514–522.

Masui T, Hashimoto R, Kusumi I, et al. (2006) Lithium response and

Val66Met polymorphism of the brain-derived neurotrophic factor

gene in Japanese patients with bipolar disorder. Psychiatr Genet

16: 49–50.

Mazza M, Mandelli L, Di Nicola M, et al. (2009) Clinical features,

response to treatment and functional outcome of bipolar disorder

patients with and without co-occurring substance use disorder:

1-year follow-up. J Affect Disord 115: 27–35.

NCBI Entrez SNP. Available at: http://www.ncbi.nlm.nih.gov/sites/

entrez?db¼snp

Nibuya M, Morinobu S, Duman RS (1995) Regulation of BDNF and

trkB mRNA in rat brain by chronic electroconvulsive seizure and

antidepressant drug treatments. J Neurosci 15: 7539–7547.

Pelissolo A, Corruble E (2002) Personality factors in depressive

disorders: contribution of the psychobiologic model developed

by Cloninger. Encephale 28: 363–373.

Piccinni A, Marazziti D, Catena M, et al. (2008) Plasma and serum

brain-derived neurotrophic factor (BDNF) in depressed patients

during 1 year of antidepressant treatments. J Affect Disord 105:

279–283.

R Project for Statistical Computing. Available at: http://www.

r-project.org/

Rybakowski JK, Suwalska A, Skibinska M, et al. (2007a) Response

to lithium prophylaxis: interaction between serotonin transporter

and BDNF genes. Am J Med Genet B Neuropsychiatr Genet 144:

820–823.

Rybakowski JK, Suwalska A, Skibinska M, et al. (2007b) Response

to lithium prophylaxis: interaction between serotonin transporter

and BDNF genes. Am J Med Genet B Neuropsychiatr Genet 144B:

820–823.

Rybakowski JK, Suwalska A, Skibinska M, et al. (2005) Prophylactic

lithium response and polymorphism of the brain-derived

neurotrophic factor gene. Pharmacopsychiatry 38: 166–170.

Sackeim HA (2001) The definition and meaning of treatment-resis-

tant depression. J Clin Psychiatry 62(Suppl 16): 10–17.

Serretti A, Kato M, Kennedy JL (2008) Pharmacogenetic studies

in depression: a proposal for methodologic guidelines.

Pharmacogenomics J 8: 90–100.

Serretti A, Mandelli L (2008) The genetics of bipolar disorder:

genome ‘hot regions,’ genes, new potential candidates and

future directions. Mol Psychiatry 13: 742–771.

Shimizu E, Hashimoto K, Okamura N, et al. (2003) Alterations of

serum levels of brain-derived neurotrophic factor (BDNF) in

depressed patients with or without antidepressants. Biol

Psychiatry 54: 70–75.

Tramontina JF, Andreazza AC, Kauer-Sant’anna M, et al. (2009)

Brain-derived neurotrophic factor serum levels before and after

treatment for acute mania. Neurosci Lett 452: 111–113.

Tsai SJ, Cheng CY, Yu YW, et al. (2003) Association study of a

brain-derived neurotrophic-factor genetic polymorphism and

major depressive disorders, symptomatology, and antidepressant

response. Am J Med Genet 123B: 19–22.

Wilkie MJ, Smith D, Reid IC, et al. (2007) A splice site poly-

morphism in the G-protein beta subunit influences antidepressant

efficacy in depression. Pharmacogenet Genomics 17: 207–215.

Yoshida K, Higuchi H, Kamata M, et al. (2007) The G196A

polymorphism of the brain-derived neurotrophic factor gene

and the antidepressant effect of milnacipran and fluvoxamine.

J Psychopharmacol 21: 650–656.

1754 Journal of Psychopharmacology 24(12)

at Scientific library of Moscow State University on December 1, 2013jop.sagepub.comDownloaded from