Psychoneuroendocrinology (2014) 43, 81—89

Available online at www.sciencedirect.com

ScienceDirect

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n

BDNF in late-life depression: Effect of SSRI

usage and interaction with childhood abuse
Annemarie van der Meij a,b, Hannie C. Comijs c,
Annemieke Dols c, Joost G.E. Janzing b,
Richard C. Oude Voshaar b,d,*

a
Pro Persona, Department of Psychiatry, Nijmegen, The Netherlands
b
Department of Psychiatry, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
c
GGZ Ingeest, Department of Psychiatry & EMGO Institute of Health and Care Research, VU Medical Centre,
Amsterdam, The Netherlands
d
University Centre of Psychiatry, University Medical Centre Groningen, University of Groningen, Groningen,
The Netherlands

Received 16 October 2013; received in revised form 11 January 2014; accepted 3 February 2014

KEYWORDS Summary Brain-Derived Neurotrophic Factor (BDNF) serum levels are abnormally low in
Aged; depressed patients as compared to healthy controls and normalize with SSRI treatment. The
Aged, 80 years and over; aim of this study is to examine serum BDNF levels in late-life depression, stratified for SSRI usage,
Brain-Derived and to explore the relation between BDNF levels and specific depression characteristics as well as
Neurotrophic Factor; between BDNF levels and early and recent life stressors in late-life depression. We assessed serum
Depression, major BDNF levels in 259 depressed patients not using an SSRI, 99 depressed patients using an SSRI and
depressive disorder; 119 non-depressed controls (age range 60—93 years). Depressive disorders were diagnosed with
Child abuse; the Composite International Diagnostic Interview (CIDI, version 2.1). Serum BDNF levels were
Serotonin uptake significantly higher in depressed patients who used an SSRI compared to depressed patients not
inhibitors using SSRIs and compared to non-depressed controls, when adjusted for age, sex, life style
characteristics, cognitive functioning and somatic comorbidity. Recent life-events, assessed with
the List of Threatening Events-Questionnaire, were significantly associated with lower BDNF
levels in non-depressed subjects only. Although a summary score of early traumatization (before
the age of 16 years) was not associated with serum BDNF levels in any of the three groups, we
found an interaction between a history of severe physical abuse and SSRI usage in the depressed
group. Interestingly, higher serum levels of BDNF in depressed patients using SSRIs were only
found in those patients without a history of severe childhood abuse and not in those with a history
of severe childhood abuse.
# 2014 Elsevier Ltd. All rights reserved.

* Corresponding author at: University Medical Center Groningen, University Center for Psychiatry (CC44), PO Box 30.001, 9700 RB Groningen,
The Netherlands. Tel.: +31 50 3612367; fax: +31 50 3611699.
E-mail address: r.c.oude.voshaar@umcg.nl (R.C. Oude Voshaar).

http://dx.doi.org/10.1016/j.psyneuen.2014.02.001
0306-4530/# 2014 Elsevier Ltd. All rights reserved.
82
1. Introduction
Brain-Derived Neurotrophic Factor (BDNF) is a critical reg-
ulator of the formation, plasticity and integrity of neurons
within the limbic system (Duman and Monteggia, 2006).
Moreover, BDNF plays an important role in the molecular
and behavioural responses to both chronic and acute stress
(Martinowich and Lu, 2008; Elzinga et al., 2011). In this sense
it is not surprising that the past decade witnessed a growing
interest in BDNF in relation to several psychiatric disorders.
To date, three meta-analyses have shown that peripheral
BDNF levels are abnormally low in depressed patients as
compared to healthy controls (Sen et al., 2008; Brunoni
et al., 2008; Molendijk et al., 2013) and that BDNF levels
normalize in the course of anti-depressant drug treatment
(Sen et al., 2008). Our group recently confirmed the effect of
depressive disorder and anti-depressant treatment on BDNF
levels in a much larger sample, albeit with a much smaller
effect-size (Molendijk et al., 2011). Interestingly, only the
use of SSRIs, in contrast to other types of antidepressants,
was associated with normalized BDNF levels during a
depressed state. Although some studies have included broad
age-ranges, age-stratified results or specific studies in later
life samples remain scarce (Bus et al., 2012). Therefore, it
remains unknown whether serum BDNF levels are also abnor-
mal in late-life depression.
Quite recently, a meta-analysis of the BDNF Val66Met
polymorphism showed an increased odds on the risk for
late-life depression in Met-allele carriers compared to Val/
Val homozygotes (Pei et al., 2012). Although this result may
point to the relevance of BDNF in late-life depression, studies
on peripheral BDNF levels in later life show conflicting results
with both positive (Palhagen et al., 2010; Zhou et al., 2011;
Diniz et al., 2010; Chu et al., 2012) and negative results
(Ziegenhorn et al., 2007; Kobayakawa et al., 2011). These
inconsistencies may be explained by small sample sizes (Diniz
et al., 2010; Chu et al., 2012) or inclusion of specific subtypes
of late-life depression like post-stroke depression (Zhou
et al., 2011), depression in Parkinson’s disease (Palhagen
et al., 2010), or depression in lung-cancer patients (Kobaya-
kawa et al., 2011).
Furthermore, the central role of psychological stress
within the neurotrophic hypothesis of depression has not
received any attention in late-life depression. Animal
research showed consistently decreased cerebral expression
of BDNF in reaction to early life stressors and subsequently
neuronal atrophy and degeneration in the hippocampus and
the cortex (Murakami et al., 2005; Song et al., 2006; Roceri
et al., 2004; Smith et al., 1995). In humans, child abuse is a
significant etiological factor in the development and persis-
tence of MDD across the life cycle (Comijs et al., 2013),
whereas more recent life stressors, like financial problems or
a divorce, often precipitate or exacerbate depressive symp-
toms (Kendler et al., 1999). Both, early and recent life
stressors may thus be hypothesized to impact on late-life
depression by the neurotrophic system.
We hypothesized that, as in younger adults, BDNF serum
levels are lower in depressed elderly, stratified by SSRI use,
compared to non-depressed controls. Furthermore, we
explored a possible relationship between BDNF levels and
specific depressive symptom profiles or characteristics and
A. van der Meij et al.
between BDNF levels and early and recent life-events stra-
tified by depression status and SSRI usage as the largest
effects may be hypothesized in depressed patients not using
SSRIs (which has large effects on peripheral BDNF levels) and
non-depressed persons (which may be less vulnerable for
stressors).
2. Methods
2.1. Sample
For the present study, we used the baseline assessment of the
Netherlands Study of Depression in Older people (NESDO)
(Comijs et al., 2011). NESDO is an on-going cohort study
designed to examine the (determinants of the) course and
consequences of depressive disorders in 378 depressed and
132 non-depressed persons aged 60—93 years. Recruitment of
depressed older persons took place in five regions in the
Netherlands from both mental health care institutes and
general practitioners in order to include persons with late-
life depression of various severity, representing the clinical
population in the Netherlands. Persons with a primary diag-
nosis of dementia, a Mini Mental State Examination-score
(MMSE) under 18 (out of 30 points), and insufficient command
of the Dutch language were excluded. Non-depressed con-
trols were recruited from general practitioners. Inclusion
criteria for non-depressed controls were: no lifetime diag-
nosis of depression, dementia or other serious psychiatric
disorders, and good command of the Dutch language.
Data collection of the baseline assessment started in 2007
and was finished in September 2010. The baseline assessment
included written questionnaires, interviews and physical
assessments. Interviews were audio taped to control the
quality of the data. The ethical review boards of the parti-
cipating institutes have approved this study. All participants
gave informed consent after oral and written information
about the study.
2.2. Measures
2.2.1. Brain-Derived Neurotrophic Factor (BDNF)
Fifty millilitre of blood was withdrawn into vacuum tubes
between 0730 h and 0930 h after an overnight fast. We put no
restrictions on patients with respect to their drug use prior to
blood withdrawal. Following blood collection, serum was
separated and stored at 80 8C until it was assayed. BDNF
protein levels were measured using the Emax Immuno Assay
system from Promega according to the manufacturer’s pro-
tocol (Madison, WI, USA, see www.promega.com), in 1
laboratory (Maastricht University) by 1 technician who was
blind to diagnoses. Undiluted serum was acid treated since
this reliably increased the detectable BDNF in a dilution-
dependent way. Greiner Bio-One high affinity 96-well plates
were used. Serum samples were diluted 100 times and the
absorbency was read in duplicate using a Biorad Benchmark
microplate reader at 450 nm. The assay sensitivity threshold
was ascertained at 1.56 ng/ml reflecting the minimum level
of BDNF in the serum that could be reliably determined. The
intra- and inter-assay coefficients of variation were found to
be within 3% and 9% respectively. In the present sample,
BDNF in late-life depression
serum BDNF levels were normally distributed (skew-
ness = 0.71; kurtosis = 0.61).
2.2.2. Depression diagnoses
Past 6-month diagnoses of depression and dysthymia accord-
ing to DSM-IV-R criteria (APA, 2000) were assessed with the
Composite International Diagnostic Interview (CIDI; WHO
version 2.1; 12 month version). The CIDI is a structured
clinical interview that is designed for use in research settings
and has high validity for depressive and anxiety disorders. As
in NESDA (The Netherlands Study of Depression in Adults;
Penninx et al., 2008), we added questions to determine the
research DSM-IV diagnosis of current minor depression
(Comijs et al., 2011).
2.2.3. Depression characteristics
Within the depressed sample, 339 (95.5%) met criteria for a
major depressive disorder (MDD) in the past 6 months, 17
(3.8%) for a minor depression and 93 (26.2%) for dysthymia.
Due to double diagnoses, numbers do not add up to 100%.
Based on data from the CIDI-interview, we assessed age of
onset of the depression (age of the participant at the time of
the first depressive episode) and recurrence (presence of
depressive episode prior to the current episode).
Severity of depression was measured by the 30-item self-
rating Inventory of Depressive Symptomatology (IDS), which
has adequate psychometric properties (Rush et al., 1996). To
examine symptom profiles of late-life depression, three
subscales of the IDS were used: a mood, a motivation and
a somatic subscale (Hegeman et al., 2012). These three
homogenous subscales yielded a good fit with exploratory
and confirmatory factor analysis in the NESDO study (Hege-
man et al., 2012).
Comorbid anxiety was also taken into account. The CIDI
12-month anxiety diagnosis was used to determine the pre-
sence of an anxiety disorder in the past year i.e. generalized
anxiety disorders, panic disorder, agoraphobia or social pho-
bia.
Medication use was assessed on drug container inspection
of all drugs used in the past month and classified according to
the World Health Organization Anatomical Therapeutic Che-
mical classification (ATC) (WHO, 2010). Medication was only
considered when taken on a regular basis (at least 50% of the
time). Antidepressant medication was classified in Selective
Serotonin Reuptake Inhibitors (SSRIs) (N06AB), tricyclic anti-
depressants (TCAs) (N06AA) and other antidepressants
(N06AX16, N06AX21, N06AX03, N06AX05, and N06AX11).
2.2.4. Early life stressors (childhood trauma)
Childhood trauma, including emotional neglect as well as
psychological, physical and sexual abuse, was assessed using
a structured inventory previously used in the Mental Health
Survey and Incidence Study (de Graaf et al., 2004) and the
Netherlands Study of Depression and Anxiety (Penninx et al.,
2008). In this inventory participants were asked whether they
had experienced any kind of emotional neglect, psychologi-
cal, physical and/or sexual abuse before the age of 16.
Emotional neglect included the lack of parental attention
or support and ignorance of one’s problems and experiences.
Psychological abuse included verbal abuse, punishment with-
out reason, subordination to siblings and being blackmailed.
83
Physical abuse included being kicked, hit with or without an
object and any other physical harm. Sexual abuse was defined
as being sexually touched against your will, or being forced to
touch someone sexually. After an affirmative answer, a ques-
tion was asked about the frequency of these events, which
was recorded as: never, once, sometimes, regularly, often or
very often. A childhood abuse index was constructed by
recoding the frequency scores in (0) never, (1) once or some-
times, and (2) regularly, often or very often. These scores
were summed up, resulting in a childhood abuse index that
ranges from 0 to 8, with higher scores indicating a higher
frequency of childhood abuse (see Comijs et al., 2013). As
childhood abuse in depression might be prone for recall bias,
we also looked at the presence or absence of physical and/or
sexual abuse. Memories on physical and sexual abuse have
been proven to be relatively stable and only minimally
affected by the presence of depression (see Spinhoven
et al., 2012).
2.2.5. Recent life-stress
The occurrence of 12 recent stressful life events was assessed
using the List of Threatening Events Questionnaire (LTE-Q;
Brugha et al., 1985; Brugha and Cragg, 1990). These events
reflect the presence of life stressors during the past year,
such as serious illness and injury, death of close friend or
relative, unemployment, major financial loss, and loss of
important relationships. The LTE-Q has good test—retest
reliability, high agreement between participant and infor-
mant ratings, and good agreement with interview-based
ratings (Brugha and Cragg, 1990). For the present study,
we dichotomized the number of life-events in the past year
as none versus one or more life-events.
2.3. Covariates
All characteristics that have been reported previously as
potentially associated with serum BDNF levels were consid-
ered as covariates in the present study. We included age,
gender, and educational level (years of education) as poten-
tially important socio-demographic characteristics. Fasten
blood withdrawal (yes/no) and duration of serum storage
(over three years or not) were controlled for since BDNF
levels vary according to variation on these variables (Bus
et al., 2011). Additionally, we controlled for global cognitive
functioning (MMSE score), number of chronic somatic dis-
eases, season of blood withdrawal according to the equinox
(dummies for autumn, winter and spring with summer as the
reference category) and life-style since these characteristics
are associated with BDNF and mood (Bus et al., 2011, 2012;
Molendijk et al., 2012).
As lifestyle characteristics we included smoking, use of
alcohol, body mass index (BMI) and physical activity. Smoking
was defined as currently smoking (yes/no). Based on the first
two questions of the Alcohol Use Disorder Identification Test
(AUDIT; Saunders et al., 1993), we classified alcohol con-
sumption into three categories, i.e. no drinking, moderate
alcohol use and problematic alcohol use. Problematic alcohol
use was defined as taking 5—10 units on a typical drinking day
irrespective of the frequency of drinking or 3 or 4 units on a
typical drinking day at least 4 or more days a week. Moderate
alcohol use was defined as any alcohol use not being
84
problematic use. Physical activity in the past week was
measured with the short-form (8-items) of the International
physical Activities Questionnaire (IPAQ; Craig et al., 2003).
Psychometric properties of the long and short version of the
IPAQ are acceptable.
Global cognitive functioning was measured by Mini Mental
State Examination (MMSE; Folstein et al., 1975). The MMSE
score ranges from 0 to 30, with higher scores indicating
better cognitive functioning.
The number of chronic diseases was assessed with pre-
viously used self-report questions about the presence of the
following chronic diseases or disease events: cardiac disease
(including myocardial infarction), peripheral atherosclerosis,
stroke, diabetes mellitus, COPD (asthma, chronic bronchitis
or pulmonary emphysema), arthritis (rheumatoid arthritis or
osteoarthritis) and cancer. The accuracy of self-reports of
these diseases was shown to be adequate and independent of
cognitive impairment compared to data obtained from gen-
eral practitioners (Kriegsman et al., 1996).
2.4. Statistical analysis
The three groups of interest, i.e. depressed persons with and
without SSRI usage and their non-depressed counterparts,
were compared by ANOVA in case of continuous variables and
by chi-square tests in case of categorical variables. Subse-
quently, ANCOVA was applied to adjusted for confounders. All
covariates described above were considered as confounding
variables; only those variables that contributed significantly
to the multivariate model (using a backward elimination
strategy) were included. LSD post hoc tests for pair-wise
comparisons were conducted to examine main effects in
more depth.
Within the depressed subgroup, multiple linear regression
analyses with BDNF serum level as the dependent variable
were conducted to examine associations between different
characteristics of depression and serum BDNF in levels. All
linear regression models were adjusted for confounders as
described above.
Finally, multiple linear regression analyses, stratified for
the three groups of interest, were conducted to explore the
potential impact of recent life-events and early life stressors
on BDNF serum levels.
p-Values < .05 will be considered significant. For signifi-
cant differences, Cohen’s d effect-sizes will be presented. All
analyses were carried out using the Statistical Package for
the Social Sciences (SPSS) version 20.0 (Inc. Chicago).
3. Results
3.1. Population characteristics
From the 510 participants included in NESDO, we had BDNF
levels of 478 persons (93.7%), i.e. 120 non-depressed com-
parisons subjects and 358 depressed patients. One non-
depressed person using an SSRI (for generalized anxiety
disorder) was excluded, as this would confound the control
group. Excluded subjects (n = 33) did not differ from included
subjects with respect to age (mean (SD) age was 70.8 (7.8)
versus 70.6 (7.3) years; t = 0.17, df = 508, p = .859), gender
(24 (72.7%) versus 307 (64.4%) females; x2 = 0.95, df = 1,
A. van der Meij et al.
p = .330), educational level (mean (SD) number of 11.3
(3.6) versus 10.9 (3.6) years; t = 0.59, df = 508, p = .554),
and severity of depressive symptoms based on the IDS sum
score (mean (SD) 24.3 (14.1) versus 24.4 (15.3) points;
t = 0.05, df = 500, p = .961). Table 1 presents the demo-
graphic, clinical and sampling characteristics of the study
population by depression status and SSRI drug use (n = 477).
3.2. BDNF levels in depressed persons with and
without SSRI usage and controls
As shown in Table 1, serum BDNF levels differed significantly
across the three groups. An ANCOVA model adjusted for con-
founders (see methods) showed a main effect of diagnostic
status on serum levels of BDNF (F = 5.11; df = 2.447; p = .006).
Pair wise comparisons (see Fig. 1) indicated that serum BDNF
levels were higher in depressed patients who used an SSRI
compared to depressed patients not using SSRIs ( p = .006,
Cohen’s d = .31) as well as compared to controls ( p = .003,
Cohen’s d = .40). BDNF levels did not differ between controls
and depressed patients using no SSRIs ( p = .452).
To confirm the hypothesized SSRI-specific effect, we also
checked BDNF-levels by type of antidepressant within the
depressed group. For analysis, we excluded those patients
using antidepressants from two classes simultaneously
(n = 22). The fully adjusted ANCOVA model was significant
(F = 3.37; df = 3.310; p = .019). LSD post hoc analyses showed
that the SSRI-users (n = 82) had significantly higher serum
BDNF levels compared to depressed patients not using anti-
depressant drugs (n = 90; p = .042, Cohen’s d = .29), com-
pared to TCA users (n = 66; p = .013, Cohen’s d = .38) and
compared to users of other antidepressants (n = 85; p = .004,
Cohen’s d = .42) (see also Fig. 2).
3.3. Association of BDNF levels with depression
characteristics
Based on significantly higher BDNF levels in depressed
patients using SSRIs, the association between BDNF serum
levels and characteristics of depression were analyzed sepa-
rately for depressed older persons with and without SSRI
usage. As shown in Table 2, none of the characteristics in
either group achieved significance.
3.4. Association of BDNF levels with early and
recent life-events
Using multiple linear regression analyses, we also examined
whether early and recent life-events were associated with
serum BDNF levels. As shown in Table 3, the presence of
recent life-events was significantly associated with lower
BDNF levels in non-depressed subjects ( p = .035, Cohen’s
d = 0.47). The childhood trauma index, however, was not
associated with BDNF serum levels in any of the three groups.
Nonetheless, the presence of physical and/or sexual abuse
approached statistical significance among depressed patients
using SSRIs ( p = .064).
Although the effect of sexual abuse was not significant,
the opposite effects of sexual abuse in depressed persons
with and without SSRI usage triggered us for more detailed
BDNF in late-life depression 85

Table 1 Characteristics of study sample by depression status and SSRI use (n = 477).

Variable Non-depressed Depressed participants Statistics

group (n = 119)
No SSRI use SSRI use
(n = 259) (n = 99)
Socio-demographics
Age (years) Mean (SD) 70.0 (7.1) 70.9 (7.6) 70.3 (6.7) F = 0.71; df = 2474; p = .494
Female sex n (%) 70 (58.8) 167 (64.5) 70 (70.7) x2 = 3.3; df = 2; p = .189
Educational level (years) Mean (SD) 12.6 (3.4) 10.5 (3.4) 10.0 (3.4) F = 19.1; df = 2474; p < .001
Life-style characteristics
Smoker n (%) 10 (8.4) 70 (27.1) 26 (26.5) x2 = 17.7; df = 2; p < .001
Alcohol use (AUDIT score) Mean (SD) 3.6 (2.6) 2.7 (3.6) 2.2 (3.2) F = 5.32; df = 2465; p = .005
Physical activity: x2 = 11.0; df = 4; p = .026
Low n (%) 20 (17.2) 77 (30.6) 34 (35.8)
Moderate n (%) 50 (43.1) 93 (36.9) 36 (37.9)
High n (%) 46 (39.7) 82 (32.5) 25 (26.3)
Body mass index Mean (SD) 27.0 (4.1) 26.4 (4.7) 26.0 (4.5) F = 1.4; df = 2474; p = .251
Clinical characteristics
MMSE score Mean (SD) 28.4 (1.6) 27.9 (1.9) 27.3 (2.1) F = 8.6; df = 2473; p < .001
No. of chronic diseases Mean (SD) 1.5 (1.1) 2.1 (1.5) 2.1 (1.5) F = 9.5; df = 2473; p < .001
IDS sum score Mean (SD) 7.5 (6.3) 29.1 (13.5) 32.3 (12.1) F = 159.7; df = 2466; p < .001
Stressors
Childhood trauma index Mean (SD) 0.2 (0.4) 1.1 (1.2) 1.0 (1.2) F = 28.9; df = 2471; p < .001
Life-events, past year n (%) 42 (35.3) 78 (30.2) 29 (29.6) x2 = 1.2; df = 2; p = .563
Sampling characteristics
Not fasting blood withdrawal n (%) 3 (2.3) 8 (3.0) 4 (3.9) x2 = 0.50; df = 2; p = .780
> 3 year blood storage n (%) 9 (6.8) 103 (38.1) 29 (28.2) x2 = 43.2; df = 2; p < .001
Season of blood withdrawal x2 = 8.17; df = 6; p = .226
Summer n (%) 49 (37.1) 72 (26.7) 28 (27.5)
Autumn n (%) 17 (12.9) 56 (20.7) 20 (19.6)
Winter n (%) 21 (915.9) 54 (20.0) 23 (22.5)
Spring n (%) 45 (34.1) 88 (32.6) 31 (40.4)
Serum BDNF level
BDNF (ng/ml) Mean (SD) 6.82 (4.19) 7.37 (4.15) 8.71 (4.41) F = 5.75; df = 2474; p = .003
Abbreviations: SSRI, Selective Serotonin Reuptake Inhibitor; SD, standard deviation; AUDIT, Alcohol Use Disorder Identification Test; MMSE,
Mini Mental State Examination; No., number; IDS, Inventory of Depressive Symptoms.

analyses. A post hoc, ANCOVA in the depressed sample
showed a significant interaction term between SSRI usage
(yes/no) and physical and/or sexual abuse before the age of
16 (yes/no) (F = 3.70; df = 1328; p = .055). As shown in Fig. 3,
SSRI usage is not associated with BDNF serum levels in
depressed older persons with a history of sexual abuse,
whereas in case of no-sexual abuse BDNF levels are signifi-
cantly higher in case of SSRI usage. The mean difference in
BDNF levels in sexually abused patients with and without SSRI
usage is smaller than 0.01 [95% CI: 1.79 to 1.79] ng/ml
( p = .911), whereas the mean difference between patients
with and without SSRI usage who are not abuse is 2.09 [95%
CI: 3.31 to 0.87] ng/ml ( p = .001).
4. Discussion
Contrary to findings in younger depressed patients, we did
not find decreased BDNF serum levels in depressed older
persons. SSRI usage, however, was associated with increased
serum BDNF levels, in contrast to other types of antidepres-
sants. No depression characteristic was associated with
serum BDNF levels, neither in depressed patients using SSRIs
nor in depressed patients not using SSRIs.
Interestingly, post hoc analyses suggest that increase in
BDNF serum levels due to SSRI usage does not occur in
depressed patients with a history of early physical and/or
sexual abuse. Although this has not been reported before in
human studies, these results are in line with preclinical
studies. The induction of cerebral BDNF expression due to
an enriched environments as well as SSRI usage is lower in
early-traumatized rats compared to control rats (for review
see Chourbaji et al., 2011). Comparable to our study, the
SSRI-associated increase of peripheral BDNF levels in rats has
been shown to be conditional on early maternal separation
(Carboni et al., 2010). Finally, we also found one human study
that showed that the SSRI-related increase in BDNF is con-
ditional upon underlying factors. In patients with Parkinson’s
disease and major depressive disorder BDNF levels were
86 A. van der Meij et al.

Figure 1 Adjusted mean (standard error) serum BDNF levels in

non-depressed controls, depressed patients using an SSRI and Figure 2 Adjusted mean (standard error) serum BDNF level in
depressed patients not using an SSRI. depressed patients splitted by antidepressant drug class.

lower than in patients with solely major depressive disorder
after treatment with citalopram (Palhagen et al., 2010).
Thus, the BDNF response on citalopram in late-life depression
seems also conditional on the presence of comorbid Parkin-
son’s disease (Palhagen et al., 2010).
We did neither detect an association between BDNF serum
levels and depression in our entire sample nor between
depressive symptom severity and BDNF level in the depressed
subgroup. The first studies on peripheral BDNF levels in
depression have reported large effect-sizes when compared
to non-depressed persons (Sen et al., 2008; Brunoni et al.,
2008). More recent studies have reported much lower effect-
sizes (Molendijk et al., 2011) as well as negative findings like
ours in both younger and older samples (for review, see
Molendijk et al., 2013). In line with our study, the largest
study on serum BDNF levels to date, also found no association
between depressive symptom severity and BDNF level in the
depressed subgroup (Molendijk et al., 2011). Therefore, it
seems unlikely that these negative findings can solely be
contributed to age-effects only.
In our sample we found that SSRI usage was significantly
associated with increased serum BDNF. Although knowledge
on the peripheral physiology of BDNF is still limited, BDNF in
serum is most likely derived from platelets (storage), vas-
cular endothelial cells (production) as well as from the brain
(Pan et al., 1998; Karege et al., 2005; Serra-Millas et al.,
2011). BDNF crosses the blood-brain barrier (Pan et al., 1998)
and in animal studies cerebral BDNF expression is highly
correlated with serum BDNF levels (Karege et al., 2005).
The SSRI-specific findings might thus be explained by

Table 2 Associations of serum BDNF with depression characteristics in depressed patients with and without SSRI usage (fully
adjusteda).

Characteristics No SSRI usage (n = 259) SSRI usage (n = 99)

B (SE) b p-Value B (SE) b p-Value
Depression severity scales
IDS sum score 0.03 (0.02) .09 193 0.02 (0.04) .05 .670
IDS mood subscale 0.04 (0.05) .05 423 0.06 (0.10) .06 .543
IDS motivation subscale 0.12 (0.09) .09 189 0.03 (0.16) .02 .843
IDS somatic subscale 0.08 (0.07) .07 .259 0.13 (0.11) .13 .231
Clinical characteristics
Age of onset (continuous) 0.00 (0.01) .01 .825 0.01 (0.02) .02 .837
Age of onset (60 years) 0.20 (0.56) .02 .719 0.33 (0.99) .03 .741
Recurrent depression 0.35 (0.55) .04 .520 0.88 (0.92) .10 .341
Comorbid dysthymia (past year) 0.30 (0.60) .03 .623 0.60 (1.00) .06 .547
Comorbid anxiety disorder (past year) 0.56 (0.55) .06 .316 0.67 (0.92) .08 .468
Abbreviations: BDNF, Brain-Derived Neurotrophic Factor; SSRI, Selective Serotonin Reuptake Inhibitor; SE, standard error; IDS, Inventory of
Depressive Symptoms; MDD, major depressive disorder.
a
Adjusted for MMSE score, physical activity (low, moderate, high), fasten blood withdrawal (yes/no), and season of blood withdrawal
(summer, autumn, winter, spring).
BDNF in late-life depression 87

Table 3 Association of early and recent life-stressors on showing upregulation of serum BDNF levels in humans after
BDNF.a antidepressant treatment, were all based on treatment with
SSRI specifically or mixed antidepressants including SSRIs
B (SE) b p-Value
(Sen et al., 2008). Based on the observation that SSRI usage
Recent life-events and not other types of antidepressants were associated with
Non-depressed group 1.74 (0.81) .20 .035 increased serum BDNF levels in currently depressed patients,
Depressed, no SSRI 0.42 (0.60) .05 .482 but not remitted depressed patients, led to the hypothesis
Depressed, SSRI 0.39 (1.02) .04 .705 that increases in BDNF serum levels do not parallel clinical
effectiveness but may have direct effects on the BDNF
Childhood trauma index
system. This hypothesis is in line with preclinical studies that
Non-depressed group 0.22 (0.91) .02 .808
have shown that SSRIs induce a state of increased neuroplas-
Depressed, no SSRI 0.18 (0.22) .05 .417
ticity due to upregulation of BDNF expression in cortical
Depressed, SSRI 0.407 (0.39) .11 .299
neurons (Maya Vetencourt et al., 2008).
Physical and/or sexual abuse We did not find an effect of recent life events in older
Non-depressed group 0.43 (1.61) .03 .789 depressed patients and only a trend for a history of sexual
Depressed, no SSRI 0.71 (0.58) .08 .219 abuse. However, we did find that the presence of recent life-
Depressed, SSRI 1.82 (0.97) .19 .064 events was significantly associated with lower BDNF levels in
non-depressed subjects. Earlier studies in depressed patients
Abbreviations: BDNF, Brain-Derived Neurotrophic Factor; SSRI,
Selective Serotonin Reuptake Inhibitor; SE, standard error. showed that levels of BDNF were decreased in case of exposure
a
Adjusted for MMSE score, physical activity (low, moderate, to recent life events and also in case of a history of childhood
high), duration of storage (>3 years), fasten blood withdrawal abuse (Elzinga et al., 2011). This concurred with the finding that
(yes/no), season of blood withdrawal (summer, autumn, winter, traumatic events are associated with lower BDNF levels in
spring). bipolar patients (Kauer-Sant’Anna et al., 2007) as well as with
a twin study reporting lower BDNF levels in females with a
genetic vulnerability for depression as well as for those sub-
increased availability of extra-synaptic levels of serotonin, as jected to stressful life events (Trajkovska et al., 2008). A
serotonin stimulates BDNF expression (Mattson et al., 2004; possible explanation for our findings might thus be that occur-
Martinowich and Lu, 2008), as well as by effecting platelets rence of late-life depression is less dependent on early and
to release BDNF that may mirror BDNF release by neurongs recent psychosocial stressors. Nonetheless, two alternative
(Serra-Millas et al., 2011). SSRI-specific findings have also explanations can be put forward explaining these negative
been reported before in a much larger, but younger sample results. First, effects may have been obscured in our study
including both currently depressed patients as well as non- due to lack of statistical power as previous human studies in this
depressed persons with a history of depression with and field have reported very small effect-sizes (Elzinga et al., 2011).
without SSRI usage (Molendijk et al., 2011). In this latter Secondly, the aetiology of depression in later life is much more
study, highest BDNF levels were found in patients using SSRIs heterogeneous due to prodromal, but not yet detected, neu-
and St. John’s wort, know to have large effects of extra- rodegenerative diseases, which may also have masked these
synaptic serotonin levels, and lowest in patients using mir- effects. Interestingly, however, despite an overall effect of
tazapine, know to have little impact on the availability of early and recent life events on serum BDNF levels, our results
serotonin. Moreover, the studies included in a meta-analysis suggest that sexual abuse may result in a lower responsiveness
of the neurotrophic system, as serum BDNF levels did not differ
by SSRI usage in sexually abused older patients.

4.1. Methodological considerations

The effect of SSRI usage on BDNF serum levels in depressed

Figure 3 Adjusted mean (standard error) serum BDNF level in
depressed patients stratified by sexual abuse (yes/no) and SSRI
usage (yes/no).
older persons, however, should be interpreted cautiously, as
our subjects were not randomly assigned to the various drugs
(or no drug) conditions, and thus they might be confounded by
indication. Furthermore, data on previous treatment with
antidepressants and on duration of treatment were not avail-
able. Moreover, the significant findings with respect to physical
and sexual abuse need to be replicated as these analyses were
post hoc and may have been underpowered. Additional weak-
nesses of our study are recognized as well. First, we relied on
data that were collected in a single wave, precluding any form
of causality. Information about history of child abuse and life
events was obtained retrospectively. It is hard to tell whether
this has led to under reporting events due to memory problem
and unwillingness to report embarrassing events or to disclose
painful memories, or to over reporting as a consequence of
negative mood. Second, we measured serum levels of BDNF
88
and assume that these measurements mirror the amount of
BDNF activity in the brain as BDNF passes the blood-brain
barrier (Pan et al., 1998). This assumption is validated on
preclinical work that showed that cortical and peripheral
levels of BDNF are highly correlated (Karege et al., 2002;
Sartorius et al., 2009), but remains complicated, because in
addition to neurons, several other tissues serve as sources of
BDNF in serum (e.g. Ernfors et al., 1990; Donovan et al., 2000;
Linker et al., 2010). Especially in older persons confounding by
somatic comorbidity or underlying prodromal neurodegenera-
tive diseases can hardly be eliminated. Especially somatic
morbidity associated with a pro-inflammatory profile may
account for negative effects in late-life depression as these
conditions are associated with increased prevalence rates of
depression (Vogelzangs et al., 2012). Nonetheless, some
strengths of our study seem evident as well and these include,
notably, the use of multivariate techniques and a large sample
of depressed older persons.
Although it remains unclear whether late-life depression
is associated with decreased serum BDNF levels, it seems
likely that SSRI still led to an up regulation of serum BDNF
levels in later life. Whether this effect is indeed dampened by
a history of sexual abuse, should be replicated in experi-
mental studies.
Sponsor’s role
The authors maintained full independence in the conduct of
this work. The sponsors had no role in design, methods,
subject recruitment, data collections, analysis or prepara-
tion of the manuscript.
Role of funding source
The infrastructure for NESDO is funded through the Fonds
NutsOhra, Stichting tot Steun VCVGZ, NARSAD The Brain and
Behaviour Research Fund, and the participating universities
and mental health care organizations (VU University Medical
Center, Leiden University Medical Center, University Medical
Center Groningen, Radboud University Nijmegen Medical
Center, and GGZ inGeest, GGNet, GGZ Nijmegen, GGZ Riv-
ierduinen, Lentis, and Parnassia). R.C. Oude Voshaar is
research fellow of the Dutch Scientific Organisation (ZonMW)
(research grant 90700231).
Conflict of interest statement
None of the authors have any financial or any other kind of
personal conflicts with this paper.
Acknowledgments
We thank Jos Prickaerts for performing the BDNF levels in his
lab.
References
American Psychiatric Association, 2000. Diagnostic and Statistical
Manual of Mental Disorders, 4th ed., text rev., http://
dx.doi.org/10.1176/appi.books.9780890423349.
A. van der Meij et al.
Brugha, T., Bebbington, P., Tennant, C., et al., 1985. The List of
Threatening Experiences: a subset of 12 life event categories with
considerable long-term contextual threat. Psychol. Med. 15,
189—194.
Brugha, T.S., Cragg, D., 1990. The List of Threatening Experiences:
the reliability and validity of a brief life events questionnaire.
Acta Psychiatr. Scand. 82, 77—81.
Brunoni, A.R., Lopes, M., Fregni, F., 2008. A systematic review and
meta-analysis of clinical studies on major depression and BDNF
levels: implications for the role of neuroplasticity in depression.
Int. J. Neuropsychopharmacol. 11, 1169—1180.
Bus, B.A., Molendijk, M.L., Penninx, B.J., et al., 2011. Determinants
of serum brain-derived neurotrophic factor. Psychoneuroendocri-
nology 36, 228—239.
Bus, B.A., Tendolkar, I., Franke, B., et al., 2012. Serum brain-derived
neurotrophic factor: determinants and relationship with depres-
sive symptoms in a community population of middle-aged and
elderly people. World J. Biol. Psychiatry 13, 39—47.
Carboni, L., Becchi, S., Piubelli, C., et al., 2010. Early-life stress and
antidepressants modulate peripheral biomarkers in a gene-envir-
onment rat model of depression. Prog. Neuropsychopharmacol.
Biol. Psychiatry 34, 1037—1048.
Chourbaji, S., Brandwein, C., Gass, P., 2011. Altering BDNF expres-
sion by genetics and/or environment: impact for emotional and
depression-like behaviour in laboratory mice. Neurosci. Biobehav.
Rev. 35, 599—611.
Chu, C.L., Liang, C.K., Chou, M.Y., et al., 2012. Decreased plasma
brain-derived neurotrophic factor levels in institutionalized
elderly with depressive disorder. J. Am. Med. Direct. Assoc.
13, 434—437.
Comijs, H.C., van Exel, E., van der Mast, R.C., et al., 2013. Childhood
abuse in late-life depression. J. Affect. Disord. 147, 241—246.
Comijs, H.C., van Marwijk, H.W., van der Mast, R.C., et al., 2011.
The Netherlands study of depression in older persons (NESDO); a
prospective cohort study. BMC Res. Notes 4, 524.
Craig, C.L., Marshall, A.L., Sjostrom, M., et al., 2003. International
physical activity questionnaire: 12-country reliability and valid-
ity. Med. Sci. Sports Exerc. 35, 1381—1395.
Diniz, B.S., Teixeira, A.L., Talib, L.L., et al., 2010. Serum brain-
derived neurotrophic factor level is reduced in antidepressant-
free patients with late-life depression. World J. Biol. Psychiatry
11, 550—555.
Donovan, M.J., Lin, M.I., Wiegn, P., et al., 2000. Brain derived neuro-
trophic factor is an endothelial cell survival factor required for
intramyocardial vessel stabilization. Development 127, 4531—4540.
Duman, R.S., Monteggia, L.M., 2006. A neurotrophic model for stress-
related mood disorders. Biol. Psychiatry 59, 1116—1127.
Elzinga, B.M., Molendijk, M.L., Oude Voshaar, R.C., et al., 2011. The
impact of childhood abuse and recent stress on serum brain-
derived neurotrophic factor and the moderating role of BDNF
Val66Met. Psychopharmacology (Berl.) 214, 319—328.
Ernfors, P., Wetmore, C., Olson, L., et al., 1990. Identification of
cells in rat brain and peripheral tissues expressing mRNA for
members of the nerve growth factor family. Neuron 5, 511—526.
Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. Mini-mental state.
A practical method for grading the cognitive state of patients for
the clinician. J. Psychiatr. Res. 12, 189—198.
de Graaf, R., Ten Have, M., van Dorsselaer, S., et al., 2004. Negative
and positive participant responses to the Composite International
Diagnostic Interview Results of the Netherlands Mental Health
Survey and Incidence Study. Soc. Psychiatry Psychiatr. Epidemiol.
39, 521—527.
Hegeman, J.M., Wardenaar, K.J., Comijs, H.C., et al., 2012. The
subscale structure of the Inventory of Depressive Symptomatol-
ogy Self Report (IDS-SR) in older persons. J. Psychiatr. Res. 46,
1383—1388.
Karege, F., Bondolfi, G., Gervasoni, N., et al., 2005. Low Brain-
Derived Neurotrophic Factor (BDNF) levels in serum of depressed
BDNF in late-life depression
patients probably results from lowered platelet BDNF release
unrelated to platelet reactivity. Biol. Psychiatry 57, 1068—1072.
Karege, F., Perret, G., Bondolfi, G., et al., 2002. Decreased serum
brain-derived neurotrophic factor levels in major depressed
patients. Psychiatry Res. 109, 143—148.
Kauer-Sant’Anna, M., Tramontina, J., Andreazza, A.C., et al., 2007.
Traumatic life events in bipolar disorder: impact on BDNF levels
and psychopathology. Bipolar. Disord. 9 (Suppl. 1) 128—135.
Kendler, K.S., Karkowski, L.M., Prescott, C.A., 1999. Causal relation-
ship between stressful life events and the onset of major depres-
sion. Am. J. Psychiatry 156, 837—841.
Kobayakawa, M., Inagaki, M., Fujimori, M., et al., 2011. Serum brain-
derived neurotrophic factor and antidepressant-naive major
depression after lung cancer diagnosis. Jpn. J. Clin. Oncol. 41,
1233—1237.
Kriegsman, D.M., Penninx, B.W., van Eijk, J.T., et al., 1996. Self-
reports and general practitioner information on the presence of
chronic diseases in community dwelling elderly. A study on the
accuracy of patients’ self-reports and on determinants of inac-
curacy. J. Clin. Epidemiol. 49, 1407—1417.
Linker, R.A., Lee, D.H., Demir, S., et al., 2010. Functional role of
brain-derived neurotrophic factor in neuroprotective autoimmu-
nity: therapeutic implications in a model of multiple sclerosis.
Brain J. Neurol. 133, 2248—2263.
Martinowich, K., Lu, B., 2008. Interaction between BDNF and seroto-
nin: role in mood disorders. Neuropsychopharmacology 33, 73—83.
Mattson, M.P., Maudsley, S., Martin, B., 2004. BDNF and 5-HT: a
dynamic duo in age related neuronal plasticity and neurodegen-
erative disorders. Trends Neurosci. 27, 589—594.
Maya Vetencourt, J.F., Sale, A., Viegi, A., et al., 2008. The anti-
depressant fluoxetine restores plasticity in the adult visual cor-
tex. Science 320, 385—388.
Molendijk, M.L., Spinhoven, P., Polak, M., et al., 2013. Serum BDNF
concentrations as peripheral manifestations of depression. Evi-
dence from a systematic review and meta-analyses on 179 asso-
ciations (N = 9484). Mol. Psychiatry, http://dx.doi.org/10.1038/
mp.2013.105.
Molendijk, M.L., Bus, B.A., Spinhoven, P., et al., 2011. Serum levels
of brain-derived neurotrophic factor in major depressive disor-
der: state-trait issues, clinical features and pharmacological
treatment. Mol. Psychiatry 16, 1088—1095.
Molendijk, M.L., Haffmans, J.P., Bus, B.A., et al., 2012. Serum BDNF
concentrations show strong seasonal variation and correlation
with the amount of ambient sunlight. PLoS ONE 7, e48046.
Murakami, S., Imbe, H., Morikawa, Y., et al., 2005. Chronic stress, as
well as acute stress, reduces BDNF mRNA expression in the rat
hippocampus but less robustly. Neurosci. Res. 53, 129—139.
Palhagen, S., Qi, H., Martensson, B., et al., 2010. Monoamines, BDNF,
IL-6 and corticosterone in CSF in patients with Parkinson’s disease
and major depression. J. Neurol. 257, 524—532.
Pan, W., Banks, W.A., Fasold, M.B., et al., 1998. Transport of brain-
derived neurotrophic factor across the blood—brain barrier. Neu-
ropharmacology 37, 1553—1561.
89
Pei, Y., Smith, A.K., Wang, Y., et al., 2012. The brain-derived
neurotrophic-factor (BDNF) val66met polymorphism is associated
with geriatric depression: a meta-analysis. Am. J. Med. Genet. B.
Neuropsychiatr. Genet. 159B, 560—566.
Penninx, B.W., Beekman, A.T., Smit, J.H., et al., 2008. The Nether-
lands Study of Depression and Anxiety (NESDA): rationale, objec-
tives and methods. Int. J. Methods Psychiatr. Res. 17, 121—140.
Roceri, M., Cirulli, F., Pessina, C., et al., 2004. Postnatal repeated
maternal deprivation produces age-dependent changes of brain-
derived neurotrophic factor expression in selected rat brain
regions. Biol. Psychiatry 55, 708—714.
Rush, A.J., Gullion, C.M., Basco, M.R., et al., 1996. The Inventory of
Depressive Symptomatology (IDS): psychometric properties. Psy-
chol. Med. 26, 477—486.
Sartorius, A., Hellweg, R., Litzke, J., et al., 2009. Correlations and
discrepancies between serum and brain tissue levels of neuro-
trophins after electroconvulsive treatment in rats. Pharmacop-
sychiatry 42, 270—276.
Saunders, J.B., Aasland, O.G., Babor, T.F., et al., 1993. Development
of the alcohol use disorders identification test (AUDIT): WHO
collaborative project on early detection of persons with harmful
alcohol consumption-II. Addiction 88, 791—804.
Sen, S., Duman, R., Sanacora, G., 2008. Serum brain-derived neuro-
trophic factor, depression, and antidepressant medications:
meta-analyses and implications. Biol. Psychiatry 64, 527—532.
Serra-Millas, M., López-Vı́lchez, I., Navarro, V., et al., 2011. Changes
in plasma and platelet BDNF levels induced by S-citalopram in
major depression. Psychopharmacology 216, 1—8.
Smith, M.A., Makino, S., Kvetnansky, R., et al., 1995. Effects of
stress on neurotrophic factor expression in the rat brain. Ann. N.
Y. Acad. Sci. 771, 234—239.
Song, L., Che, W., Min-Wei, W., et al., 2006. Impairment of the
spatial learning and memory induced by learned helplessness and
chronic mild stress. Pharmacol. Biochem. Behav. 83, 186—193.
Spinhoven, P., Bamelis, L., Haringsma, R., 2012. Consistency of
reporting sexual and physical abuse during psychological treat-
ment of personality disorder: an explorative study. J. Behav. Ther.
Exp. Psychiatry 43 (Suppl. 1) S-43—S-50.
Trajkovska, V., Vinberg, M., Aznar, S., et al., 2008. Whole blood BDNF
levels in healthy twins discordant for affective disorder: associa-
tion to life events and neuroticism. J. Affect. Disord. 108, 165—
169.
Vogelzangs, N., Duivis, H.E., Beekman, A.T., et al., 2012. Association
of depressive disorders, depression characteristics and antide-
pressant medication with inflammation. Transl. Psychiatry 2, e79.
Zhou, Z., Lu, T., Xu, G., et al., 2011. Decreased serum brain-derived
neurotrophic factor (BDNF) is associated with post-stroke depres-
sion but not with BDNF gene Val66Met polymorphism. Clin. Chem.
Lab. Med. 49, 185—189.
Ziegenhorn, A.A., Schulte-Hebruggen, O., Danker-Hopfe, H., et al.,
2007. Serum neurotrophins–—a study on the time course and
influencing factors in a large old sample. Neurobiol. Aging 28,
1436—1445.