Journal of Affective Disorders 165 (2014) 24–30

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Research report

The effect of exercise on hippocampal volume and neurotrophines

in patients with major depression—A randomized clinical trial
Jesper Krogh a,n, Egill Rostrup b, Carsten Thomsen c, Betina Elfving d, Poul Videbech e,
Merete Nordentoft a
a
Mental Health Centre Copenhagen, Mental Health Services in the Capital Region of Denmark, University of Copenhagen, Copenhagen, Denmark
b
Functional Imaging Unit, Department of Diagnostics, Copenhagen University Hospital, Glostrup Hospital, Denmark
c
Department of Radiology, Rigshospitalet, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark
d
Translational Neuropsychiatry Unit, Department of Clinical Medicine, Aarhus University, Aarhus, Denmark
e
Department for Depression and Anxiety Q, Aarhus University Hospital, Risskov, Denmark

art ic l e i nf o a b s t r a c t

Article history: Background: The hippocampal volume is reduced in patients with major depression. Exercise leads to an
Received 4 January 2014 increased hippocampal volume in schizophrenia and in healthy old adults. The effect of exercise on
Received in revised form hippocampal volume is potentially mediated by brain derived neurotrophic factor (BDNF), vascular
15 April 2014
endothelial growth factor (VEGF), and insulin like growth factor 1 (IGF-1). The aim of this trial was to
Accepted 15 April 2014
Available online 23 April 2014
assess the effect of an aerobic exercise intervention on hippocampal volume and serum BDNF, VEGF, and
IGF-1 in patients with major depression.
Keywords: Methods: Patients were randomized to an aerobic exercise intervention (n ¼ 41) or a control condi-
Hippocampus tion (n ¼38). Both interventions consisted of three supervised sessions per week during a three months
Exercise
period.
Depression
Results: Post-intervention the increase in maximal oxygen uptake was 3.90 ml/kg/min (SD 5.1) in the
Neurogenesis
Memory aerobic exercise group and 0.95 ml/kg/min (SD 6.2) in the control group (p¼0.03). The hippocampal
BDNF volume, BDNF, VEGF, or IGF-1 did not differ between the two groups. Post-hoc we found a positive
association between change in hippocampal volume and verbal memory (Rho ¼0.27; p¼ 0.05) and
change in hippocampal volume and depressive symptoms (Rho ¼ 0.30; p¼0.03).
Limitations: Participation was low in both groups corresponding to an average participation of one
session per week.
Conclusion: Despite a significant increase in maximal oxygen uptake, a pragmatic exercise intervention
did not increase hippocampal volume or resting levels of neurotrophines in out-patients with mild to
moderate major depression. Trial identifier: ClinicalTrials.gov (NCT00695552)
& 2014 Elsevier B.V. All rights reserved.

1. Introduction the observed reduction in hippocampal volume in patients with

The hippocampal region is thought to play a pivotal role in major
depression. This temporal structure is involved in cognitive skills
such as memory, learning, regulation of the hypothalamic–pituitary–
adrenal (HPA) axis, and mood (Eisch and Petrik, 2012). Meta-analysis
suggests that the bilateral hippocampal volume is reduced 0.3–0.4
standard deviations in patients with depression compared to healthy
controls (Videbech and Ravnkilde, 2004). It has been speculated that
n increase in memory and learning is mediated through an up-
Correspondence to: Mental Health Centre Copenhagen, Bispebjerg Bakke 23,
Opg. 13a, DK-2400, Denmark. Tel.: þ45 2553 54 85.
E-mail address: Jesper.krogh@regionh.dk (J. Krogh).
depression is due to decreased neurogenesis.
Randomised clinical trials suggest that exercise leads to an
increased hippocampal volume in healthy old adults (Erickson
et al., 2011) while the evidence for schizophrenia is conflicting
(Pajonk et al., 2010; Scheewe et al., 2013). Furthermore, in two of
these trials the increase in hippocampal volume was positively
associated to increased maximal oxygen uptake, brain derived
neurotrophic factor (BDNF), spatial and verbal memory (Erickson
et al. 2011; Pajonk et al., 2010).
Studies on rodents have found that the exercise related
regulation of BDNF, which has the ability to stimulate synaptic-
plasticity (Tapia-Arancibia et al., 2004). The exercise induced cell

http://dx.doi.org/10.1016/j.jad.2014.04.041
0165-0327/& 2014 Elsevier B.V. All rights reserved.
 J. Krogh et al. / Journal of Affective Disorders 165 (2014) 24–30
proliferation is inhibited by peripheral blockade of either insulin
growth factor 1 (IGF-1) or vascular endothelial growth factor
(VEGF) (Fabel et al., 2003; Trejo et al., 2008), which indicate that
the exercise induced neurogenesis is dependent on not only BDNF
but peripheral IGF-I and VEGF. Furthermore, it has been shown
that antidepressant treatment up-regulates BDNF in humans
(Shimizu et al., 2003) and a relationship between genetic variants
of BDNF locus and depression has been implied (Schumacher et al.,
2005). On this background, it could be argued that low BDNF levels
contribute to the pathophysiology of depression, and that exercise
has a beneficial effect on mood and cognitive function in patients
with depression through an up-regulation of BDNF, VEGF, and
IGF-1.
The aim of the present study was to determine if a pragmatic
exercise intervention compared to an exercise control condition
would increase hippocampal volume and increase serum levels of
neurotrophines (BDNF, VEGF, and IGF-1) in patients with major
depression.
2. Methods
2.1. Design
The study was designed as a randomised, parallel-group,
superiority trial with blinded outcome assessment. The trial proto-
col was approved by the local ethics committee (H-A-2008-046),
the Danish Data Protection Agency (J.nr. 2008-41-2354), and
registered at ClinicalTrials.gov (NCT00695552). Verbal and written
informed consent was obtained from all participants involved in
the trial.
2.2. Study population
All patients were recruited for participation from the DEMO-II
trial, a randomized clinical trial evaluating the antidepressant
effect of aerobic exercise (Krogh et al., 2012). The initial 90
consecutively enroled participants were asked to participate in
the current sub-study. Eligible participants were men and women
between 18 and 60 years of age, referred from a clinical setting by
a physician or a psychologist, a diagnose of major depression
(DSM-IV) based on the Danish version of the Mini International
Neuropsychiatric Interview (Bech et al., 1999). The participants all
scored above 12 on the HAM-D17 and were able to comprehend
the informed consent statement. Exclusion criteria were current
drug abuse, any antidepressant medication within the last two
months, current psychotherapeutic treatment, contraindications
to physical exercise, regular recreational exercise over 1 h per
week, suicidal behaviour according to the 17-item Hamilton
Depression Rating Scale (HAM-D17, item 3 over 2), pregnancy, or
current/previous psychotic or manic symptoms.
2.3. Randomisation
Participants were randomized with a 1:1 ratio to either aerobic
exercise or an attention control group. Randomisation was strati-
fied according to severity of depression (high or low depression
score; high 417 HAM-D17) and blood pressure (high or low blood
pressure; high 4 140/95 mmHg). The randomisation was centra-
lised and carried out by the Copenhagen Trial Unit (CTU) using
a computerised randomisation sequence with alternating block
sizes (alternating 8, 10, and 12) unknown to the investigators. Prior
to the first training session the trial physiotherapist did contact the
CTU by phone for participant allocation.
25
2.4. Blinding
Neither participants nor the physiotherapist conducting the
intervention were blinded to the allocation. The outcome assessors
(the study nurse, the laboratory and the MRI technicians) were all
blinded to participant allocation. Prior to the follow-up interview,
participants were instructed not to reveal their allocation to the
outcome assessors. The statistical analysis and preparation of the
first draft was carried out blinded to group assignments.
2.5. Physical examination
For the physical examination, the participants were requested to
report at the research department between 8:00 and 10:00 a.m. The
participants were instructed not to take any food or liquids except for
water beginning from midnight prior to the examination and abstain
from strenuous physical activity prior to the examination. Height and
weight was measured using an electronic weight (Soehnle Medicals,
Type 7700, Backnang, Germany). Blood pressure was measured after
5 min of rest in a sitting position. The mean of three consecutive
measurements is reported. To estimate the cardiovascular fitness of
the participants a bicycle cardiopulmonary exercise test (Ergomedic
839 e, Monark, Vansbro, Sweden) was used based on Andersen's
cycle exercise protocol (Andersen, 1997).
2.6. Intervention
After baseline assessment the patients were randomised to
either aerobic exercise or a control group as described in detail
elsewhere(Krogh et al., 2012). In short, both groups were sched-
uled to participate in 45 min supervised sessions three times per
week in a 3 months programme. In the aerobic training group the
participants exercised on stationary bikes at approximately 80% of
their maximal heart rate, while the training in the exercise control
group was predominantly stretching exercises and low impact
exercise such as throwing and catching balls.
2.7. Outcomes
2.7.1. Hippocampal volume
The primary outcome for this sub-study was the left and right
hippocampal volumes measured post-intervention. Hippocampal
volume was estimated using magnetic resonance imaging (MRI).
MRI volumetric images were acquired on a 3.0T Siemens system
using a T1-weighted 3D gradient MPR sequence with a resolution
of 1  1  1 mm with the following parameters: TR ¼2250 ms;
TE 3 ms; FoV read ¼256 mm; FoV phase ¼100.0%; and flip angle
91. All images were performed in the coronal plane at an angle
perpendicular to the hippocampal long axis. All tracings were
manually performed by the same rater (JK) who was instructed by
a specialist in neuroimaging (ER). After an initial training period
the intra-rater reliability was obtained by estimating the left
and right hippocampus of the same 10 patients with at least one
day between the first and second estimations. The intraclass-
correlation-coefficient of the first and second assessments was
0.98 (95% CI. 0.91–0.99) and 0.95 (95% CI. 0.82–0.99) for the right
and left hippocampal regions. In absolute numbers the median
difference between first and second assessments was  12.5 mm3
(IQR  121 to 96) and 129 mm3 (IQR 25–178) for right and left
hippocampal volumes.
Hippocampus was defined as the cornu ammonis, dentate
gyrus, and subiculum. Regions of interest were traced in the
coronal plane. Sagittal and axial planes were used continuously
to assist tracing in the coronal plane. All estimations of volumes
were acquired by tracing regions of interest on each coronal image
and finally all slices were added in order to determine the volume.
26 J. Krogh et al. / Journal of Affective Disorders 165 (2014) 24–30
For tracing and volume estimation an in house programme was
used (RIP, written for Matlab). Border definitions were based on
common segmentation protocols (Konrad et al., 2009). The ante-
rior border was located in the sagittal plane. Differentiation of
amygdala and hippocampus was based on the alveus. The superior
border was identified using alveus as an internal landmark, or the
temporal horn of the lateral ventricle as an external landmark. The
inferior border was defined as the white matter of the parahippo-
campal gyrus. The inferomedial border was drawn from the tip of
white matter from the parahippocampal gyrus medially to cisterna
ambiens. The lateral border was defined by CSF of the lateral
ventricle as an external landmark. The posterior border was
located where an ovoid grey matter appear inferiomedially to
the trigone of the lateral ventricle. Head of the hippocampus was
anteriorly identified as described above and posteriorly as the last
coronal image where the uncal apex was visible (Malykhin et al.,
2007). The total volume of the brain tissue and grey matter
was estimated with SIENAX (Smith et al., 2002), part of the FSL
software package (2013) (FMRIB Software Library).
2.7.2. Neurotrophic factors
Previously, we have shown the BDNF levels in plasma samples,
but not in serum or whole blood samples, to be affected by the
preanalytical conditions (Elfving et al., 2010). BDNF, VEGF, and IGF-
1 were all estimated using serum samples collected from an
antecubital vein after five minutes of rest and stored at 801
until analysis. Quantification of serum BDNF, VEGF, and IGF-1
levels was performed with the enzyme-linked immunosorbent
assay (ELISA) kits. The same batch number was used for the entire
experiments of BDNF, VEGF, and IGF-1. The standard curves and
the samples were run in duplicate, blinded for case and control.
Serum BDNF levels were measured using the Quantikine Human
BDNF Immunoassay from R&D Systems (USA). The determination
was processed according to the manufacturer's specifications and
the absorbance was immediately measured at 450 nm with wave-
length correction set to 540 nm (EL 800 Universal Microplate reader,
Bio-Tek instruments, INC). For the BDNF measurements the samples
were diluted 1:20 to be within the range of the standard curve.
The standard curves ranged from 62.5 to 2000 pg/ml BDNF. Three
internal BDNF controls (high: 1959–3315 pg/ml, medium: 1099–
1525 pg/ml, low: 344–506 pg/ml), commercial available from R&D
Systems (USA), were included on each plate.
The serum VEGF levels were measured using the Quantikine
Human VEGF Immunoassay from R&D Systems (USA). The deter-
mination was processed according to the manufacturer's specifica-
tions and the absorbance was immediately measured at 450 nm
with wavelength correction set to 540 nm (EL 800 Universal
Microplate reader, Bio-Tek instruments, INC). The standard curves
ranged from 31.2 to 1000 pg/ml VEGF and the samples were
run undiluted to be within the range of the standard curve.
Three internal VEGF controls (high: 682–1354 pg/ml, medium:
350–686 pg/ml, low: 115–259 pg/ml), commercially available from
R&D Systems (USA), were included on each plate.
The serum IGF-1 levels were measured using RayBios Human
IGF-1 ELISA kit from RayBiotech, Inc (USA). The determination was
processed according to the manufacturer's specifications and the
absorbance was immediately measured at 450 nm (EL 800 Uni-
versal Microplate reader, Bio-Tek instruments, INC). The standard
curves ranged from 0.25 to 60 ng/ml IGF-1 and the samples were
run undiluted to be within the range of the standard curve.
2.7.3. Assessment of mood and cognitive function
Depression severity was assessed using the Hamilton Depression
Rating Scale with 17-items (Hamilton, 1960). Assessment of verbal
memory was assessed by the Buschke's Selective Reminding test
(Buschke and Fuld, 1974). A list of 10 different unrelated words was
read aloud to the participant. The participant is then asked to recall the
list. The interviewer repeats the words that the patient misses and the
participant is asked to try again until all ten words can be said, or until
ten attempts. The score is the total number of blanks or mistakes; thus,
a high score indicates poorer performance. Visuospatial memory was
assessed by the Rey's Complex Figure Test (Meyers et al., 1996). In this
test the participant was shown a geometrically complex figure on a
sheet of paper and asked to copy it to another sheet of paper. When
this is done the drawings and the original are put away and after three
minutes the participant is asked to draw as much of the figure they
can recall. The score is calculated based on the tree-minute recall
drawing. A high score reflects better performance.
2.8. Statistics
The statistical analysis was based on the intention-to-treat princi-
ple including all randomised patients regardless of subsequent with-
drawal or deviation from the protocol. Assessment of variables
measured at baseline and post-intervention was reported using mean
and standard deviation. Post-intervention values were compared using
student's t-test for unpaired data. Missing values were imputed using
the multiple imputation technique available in SPSS, with 100 impu-
tations and 20 iterations having allocation and baseline values as the
predicting variables. Per protocol analysis was undertaken including
participants who attended Z50% of the scheduled sessions.
All significance tests were 2-tailed and p-values r0.05 were
considered significant. Statistical analysis was conducted using
SPSS version 19.0 (SPSS, Inc.: Chicago, USA).
3. Results
The initial 90 patients included in the DEMO-II trial were
considered for enrolment in the current sub-study, as displayed in
Fig. 1. The inclusion of patients occurred between September 2008
and April 2010. Eleven patients were excluded from MRI due to either
severe claustrophobia (n¼10) or metallic implants (n¼1), leaving 79
participants to constitute the current sample. Forty-one were allo-
cated to the aerobic exercise group and 38 were allocated to the
control group. The mean age in the included group was 41.3 (SD 12.1)
years and consisted of 53/79 (67.1%) female participants. The mean
HAM-D17 was 19.0 (SD 4.3), 32/79 (40.5%) had recurrent depression,
and 43/79 (54.4%) generalised anxiety disorder comorbid with
their depression. The aerobic exercise group and control group
were comparable with respect to baseline demographic and clinical
characteristics as displayed in Table 1.
3.1. Compliance
The mean attendance was 12.9 (SD 9.6) sessions in the aerobic
exercise group compared to 12.4 (SD 8.6) sessions in the control
group (p¼0.54) of a planned total of 36 sessions. This corresponds to
an average participation of one session per week. The per-protocol
population (Z18 sessions) consisted of 14 participants from the
aerobic exercise group with a mean number of attended sessions of
24.1 (SD 4.7) and 13 participants from the control group with a mean
number of attended sessions of 22.4 (SD 2.8) (p¼0.10).
3.2. Follow-up
Post-intervention, 33/41 (80.5%) in the aerobic exercise group
participated in a MRI scan and 22/38 (57.9%) in the control group
(p¼ 0.03) suggesting that follow-up was skewed. The pre-inter-
vention right hippocampal volume was 182 mm3 lower (95% CI
0.1–364; p¼0.05), diastolic blood pressure was 5.3 mmHg higher
 J. Krogh et al. / Journal of Affective Disorders 165 (2014) 24–30 27

Fig. 1. Flow-diagram of participant flow.

Table 1 3.4. Hippocampal volume

Clinical and demographical characteristics.
Post-intervention, there were no differences in right or left
Aerobic exercise Exercise control P
N ¼ 41 N¼ 38 value hippocampal volume as illustrated in Fig. 2. Neither did the
intervention result in any changes of total hippocampal volume
Age, years 38.9 (11.7) 43.8 (12.2) 0.07 or grey matter volume as displayed in Table 2.
Female, n (%) 30 (73.2) 23 (60.5) 0.34
Post-hoc we adjusted for age and gender as main effects. This
Smoking, daily, n (%) 13 (31.7) 15 (39.5) 0.54
did not change the results. Neither did inclusion of age as an
Psychometrics interaction term (age  allocation) influence post-intervention
HAM-D17 19.0 (3.9) 18.9 (4.6) 0.92
hippocampal volumes.
HAM-A14 18.1 (5.0) 16.3 (6.0) 0.17
Generalised anxiety, n (%) 23 (56.1) 20 (52.6) 0.82
Recurrent depression, n (%) 17 (41.4) 20 (52.6) 0.37 3.5. Neurotrophines and cognitive skills
Age at first depression 33.9 (11.6) 38.8 (13.1) 0.09

Somatic Post-intervention, we found no differences in serum BDNF,

Systolic blood pressure, 121.8 (15.5) 124.3 (17.7) 0.52 VEGF, and IGF-1 levels or any of the assessed cognitive skills
mmHg between the two groups.
Diastolic blood pressure, 80.1 (9.2) 82.6 (13.0) 0.34
mmHg
Weight, kg (SD) 76.0 (22.5) 75.6 (17.5) 0.96 3.6. Per-protocol analysis
Body mass index (SD) 25.8 (6.4) 25.2 (5.1) 0.70
Cholesterol (SD) 5.2 (1.1) 5.3 (1.4) 0.72 The per-protocol population (Z18 sessions) consisted of 14
Maximal oxygen uptake 26.5 (7.2) 27.2 (5.9) 0.65
participants from the aerobic exercise group with a mean atten-
Mean and standard deviation are reported unless otherwise noted. dance of 24.1 (SD 4.7) sessions per week and 13 participants from
Abbreviations: HAM-D17—Hamilton Depression Scale, 17 items; HAM-A14—Hamil- the control group with a mean attendance of 22.4 (SD 2.8) sessions
ton Anxiety Scale, 14 items. per week (p ¼0.10). Also in this restricted sample we did not find
any differences in right hippocampal volume (p ¼0.31), left hippo-
(95% CI 0.1–10.7; p¼0.05), and number of attended sessions was on campal volume (p¼ 0.35), total hippocampal volume (p¼ 0.85), or
average 10.5 higher (95% CI 7.5–13.5; po0.001) in participants grey matter volume (p ¼0.37).
attending the post-intervention MRI compared to non-attenders.
Otherwise the attenders and non-attenders to post-intervention 3.7. Post-hoc analysis
MRI were not significantly different on any other reported variable.
As illustrated in Fig. 3, change in maximal oxygen uptake was
3.3. Maximal oxygen uptake not associated to change in total hippocampal volume in this study
(p ¼0.85). A positive association was found between an increase in
Post-intervention the mean maximal oxygen uptake increased right hippocampal volume and a decrease in HAM-D17 score and
with 3.90 ml/kg/min (SD 5.1) in the aerobic exercise group and an increase in left hippocampal volume and an increase in verbal
0.95 ml/kg/min (SD 6.2) in the control group (p ¼0.03). performance of the Buschke selective reminding test as illustrated
28 J. Krogh et al. / Journal of Affective Disorders 165 (2014) 24–30

cm3 ITT analysis PP analysis

Right Left Right Left
P = 0.91 P = 0.80 P = 0.17 P = 0.50 P = 0.31 P = 0.35

Fig. 2. Right and left hippocampal volumes in response to an exercise intervention. Data are presented as mean (SD). Intention-to-treat analysis (ITT; n¼ 41þ 38) was
conducted using multiple imputation adjusting for baseline right hippocampal volume and diastolic blood pressure. Per protocol analysis (PP; n¼ 14þ 13) was conducted and
post-intervention data is shown here.

Table 2
Hippocampus volume, neurotrophic factors and cognitive skill before and after an exercise intervention.

Aerobic exercise Exercise control P valuea

N ¼ 41 N ¼ 38 Between-group

Pre-intervention Post-intervention Pre-intervention Post-intervention Pre Post

Brain volumes
Total hippocampal, cm3 6.353 (0.7) 6.325 (0.7) 6.421 (0.6) 6.380.0 (0.7) 0.54 0.88
Right hippocampal 3.151 (0.4) 3.117 (0.4) 3.136 (0.4) 3.110 (0.4) 0.91 0.80
Anterior part 1.537 (0.3) 1.546 (0.3) 1.599 (0.3) 1.622 (0.3) 0.43 0.35
Left hippocampal 3.202 (0.3) 3.201 (0.3) 3.284 (0.3) 3.267 (0.4) 0.17 0.50
Anterior part 1.644 (0.2) 1.605 (0.2) 1.699 (0.2) 1.651 (0.2) 0.38 0.50
Grey matter 624.337 (60.8) 621.455 (62.0) 605.591 (101.0) 605.960 (103.0) 0.32 0.41
Total brain 1166.187 (108.0) 1163.100 (108.0) 1167.996 (92.7) 1165.000 (96.0) 0.97 0.64

Neurotrophic factors
BDNF, pg/ml 25347.0 (6224) 26005.5 (7184) 25488.0 (7165) 25043.8 (7428) 0.93 0.47
VEGF, pg/ml 309.2 (175.7) 303.8 (182.5) 332.1 (211.3) 312.6 (244.7) 0.62 0.86
IGF-1, ng/ml 86.6 (110) 81.7 (114.6) 88.6 (104) 67.4 (110.3) 0.94b 0.54

Cognitive function
DART 34.0 (9.9) – 34.6 (7.6) – 0.79 –
Buschke, total 16.4 (12.0) 11.4 (10.9) 17.9 (11.8) 12.9 (12.3) 0.57 0.60
Rey recall 20.9 (6.7) 25.5 (6.4) 21.3 (7.2) 23.4 (7.4) 0.79 0.28

Mean and standard deviation are reported.

Missing data post-intervention was handled by multiple imputations.
a
P values are based on independent t-test. All comparison of brain volumes are adjusted for whole brain volume.
b
Confirmed in non-parametric analysis.

in Fig. 4. An association between increase in right hippocampal

volume and improved performance of the Rey memory test
(rho¼ 0.23; p ¼0.10) was borderline significant. No other associa-
tion between change in maximal oxygen uptake, hippocampal
volume, and neurotrophines was observed.
At follow-up, 4/33 (12.1%) patients in the aerobic exercise group
and 4/22 (18.2%) in the control group had started antidepressant
medication (p¼ 0.70). Adjusting for this in the analysis did not
change the results.

4. Discussion

The primary aim of the current study was to assess hippocam-

Fig. 3. Association between change in total hippocampal volume and change in
maximal oxygen uptake. Spearmann's rho and p-value is based on all randomized
participants. Missing data post-intervention were handled by multiple imputations.
pal volume in response to aerobic exercise. We did not find that a
three month aerobic exercise intervention increased the hippo-
campal volume in a group of mild to moderate depressed out-
patients. Furthermore, the aerobic exercise intervention did not
increase serum levels of BDNF, VEGF or IGF-1. We did, however,
find an association between an increase in hippocampal volume,
improved depression, and verbal memory independent of exercise.
 J. Krogh et al. / Journal of Affective Disorders 165 (2014) 24–30
Fig. 4. The association between change in right hippocampal volume and HAM-D17 and change in left hippocampal volume and verbal memory. Spearmann's rho and p-
value is based on all randomized participants. Missing data post-intervention were handled by multiple imputations. 1Busckhes Selective Reminding Test.
To our best knowledge, this is the first study to assess hippo-
campal volume in response to aerobic exercise in patients with
major depression. Several randomized clinical trials have pre-
viously shown an increase in hippocampal volume in response
to aerobic exercise. An increase of 12% was found in a small study
of patients with schizophrenia in response to a three month
exercise intervention (Pajonk et al., 2010) and an increase of 2%
was found in healthy older adults in response to a 6 months
exercise programme (Erickson et al., 2011). These findings are
supported by non-controlled prospective studies in healthy
humans demonstrating an association between increase in hippo-
campal volume and increase in aerobic fitness (Parker et al., 2011;
Pereira et al., 2007). Also, a substantial volume of animal studies
support the association between exercise and hippocampal
volume (Voss et al., 2013b). However, a recent semi-large rando-
mized clinical trial of patients with schizophrenia was not able to
reproduce the previously reported findings in subjects with
schizophrenia (Scheewe et al., 2013). Interestingly, the increase
in hippocampal volume in trials with positive findings has been
linked to increase in maximal oxygen uptake. In the current study
the increase in maximal oxygen uptake in the aerobic exercise
group was 14.7% compared to 11% in the study by Pajonk et al.
(2010) and 7.8% in the study by Erickson et al. (2011) illustrating
that increase in maximal oxygen uptake is not sufficient for
hippocampal growth and a frequent stimulation of exercise
mediated factors on the hippocampus is potentially necessary.
Therefore, one potential explanation is the low participation rate
in the current study compared to the study by Pajonk et al. (2010)
who included 16 patients that participated in 75% of the sessions
or more. In our per protocol analysis (n ¼27), including partici-
pants that participated in more than 50% of the sessions we did
not find any suggestions of hippocampal growth. It is possible that
the effect of exercise on hippocampal volume was less than the
uncertainty of our measurement and changes of 2% would be
smaller than our method would detect. Including patients for this
sub-study ended after approaching patient number 90. This was
based partly on a rough estimate on expected effect observed in
the study by Pajonk et al. (2010), in which exercise programme
was very similar to current study, and partly due to the availability
and costs of MRI. Assuming a two-sided type 1 error of 5% the
current study had a 92% chance (power) of finding a 10% difference
post-intervention in left or right hippocampal volume, however,
the current setting allow only a 10% chance (power) of finding a 2%
29
difference post-intervention as observed in the study by Erickson
et al. (2011). Therefore, we cannot exclude smaller increases in
hippocampal volume in response to an exercise intervention.
However, this potential effect on hippocampal volume did not
produce clinically relevant results. A neurotrophin like BNDF
increases in response to an acute bout of exercise and it is possible
that the continuous stimulation of BDNF and other neurotrophines
to the brain is responsible for hippocampal growth. To our knowl-
edge, only three randomised clinical trials on exercise and hippo-
campal growth have been published to date (Erickson et al., 2011;
Pajonk et al., 2010;Scheewe et al., 2013) and none of these on
patients with depression. Both schizophrenia and increasing age
are associated with decreased hippocampal volume. Recent meta-
analysis of hippocampal volume in patients with depression
suggests that hippocampal volume is reduced only in a sub-
group of patients minority of patients (McKinnon et al., 2009)
and it could be argued that a pragmatic exercise intervention is
unlikely to increase the volume of a non-affected brain structure.
The evidence of hippocampal growth in response to exercise in
humans is therefore still uncertain and with the current study no
data from human subjects supports that exercise in patients with
depression will induce hippocampal growth.
The increase in hippocampal volume is thought to be mediated
partly by neurotrophines such as BDNF, VEGF, and IGF-1. BDNF is
responsible for neural survival, growth, and synaptic plasticity and
BDNF mRNA is upregulated in the dentate gyrus of the hippocam-
pal region in response to exercise (Vaynman et al., 2004). The
current study did not find that aerobic exercise training increases
serum levels of BDNF, VEGF, or IGF-1, which is in line with a
similar study of healthy older adults (Voss et al., 2013a). In rodents,
peripheral blockade of VEGF and IGF-1 has been shown to inhibit
the exercise induced hippocampal growth. While BDNF increases
in response to acute aerobic exercise, the effect of chronic aerobic
exercise training is less clear (Huang et al., 2014). Of particular
interest is that Erickson et al. (2011) found that the exercise
mediated increase in hippocampal volume was positively asso-
ciated to increase in BDNF in healthy old participants, which was
not confirmed in the present study. Neither did the study by
Erickson et al., 2011 find an effect of exercise training on VEGF or
IGF-1. IGF-1 should ideally be analysed in the context of the IGF-1
carrier protein IGFB-3, but that analysis was not undertaken in the
current study. While it is possible that BDNF in some populations
will increase in response to aerobic training there is at present no
30 J. Krogh et al. / Journal of Affective Disorders 165 (2014) 24–30
evidence in favour of this in psychiatric populations. Neither is
there any support of aerobic exercise training will increase or
decrease VEGF or IGF-1 in this patient group.
We conducted post-hoc analysis to examine a number of
correlations and these results should purely be seen as hypothesis
generating. As illustrated, we found an increase in verbal memory
performance (Buschke Selective Reminding Test). This association
between verbal memory and left hippocampal volume is sup-
ported by a previous study finding an association between verbal
memory and hippocampal microstructure but these findings were
not specifically related to left or right hippocampal region (van
Norden et al., 2012). The association between verbal memory
performance and left hippocampal volume should also be inter-
preted in the context of speech production predominantly being
situated in the left frontal–temporal region. The association
between hippocampal growth and reduction in depressive symp-
toms supports a role for hippocampus in depression and as
previously shown the volume reduction in depression is poten-
tially state dependent and will after recovery (Ahdidan et al., 2011)
regain the pre-morbid volume.
In summary, an exercise intervention for outpatients with mild
to moderate major depression does not increase hippocampal
volume and does not increase serum levels of BDNF, VEGF, or
IGF-1. Furthermore, we found no association between change in
hippocampal volume and change in maximal oxygen uptake
Role of the funding source
None of the funders had any role in trial design, data collection,
manuscript preparation or analysis of data. Jesper Krogh had full
access to all of the data in the study and takes full responsibility
for the integrity of the data and the accuracy of the data analysis.
Conflict of interests
None of the authors report any conflicts of interest.
Acknowledgement
We thank all the participants, without their interest in our research this study
would not have been possible. We are also grateful for the assistance of Eivy Olsen
for assessing the participants' cognitive skills and Poul-Henrik Frandsen for
assistance with MRI, and Pia Hogh Plougmann for analysis of neutrophines. This
study was funded by Trygfonden, Nordea-Danmark fonden, Helsefonden, and Aase
and Ejnar Danielsensfond.
References
Ahdidan, J., Hviid, L., Chakravarty, M., Ravnkilde, B., Rosenberg, R., Rodell, A.,
Stodkilde-Jorgensen, H., Videbech, P., 2011. Longitudinal MR study of brain
structure and hippocampus volume in major depressive disorder. Acta Psy-
chiatr. Scand. 123 (3), 211–219.
Andersen, L.B., 1997. A maximal cycle exercise protocol to predict maximal oxygen
uptake. Scand. J. Med. Sci. Sports 5, 143–146.
Bech,P., Bech-Andersen,G., Schütze,T., 1999. Mini Internationalt Neuropsykiatrisk
Interview. Dansk Version 5.0.0. www. ccmh. dk.
Buschke, H., Fuld, P.A., 1974. Evaluating storage, retention, and retrieval in
disordered memory and learning. Neurology 24, 1019–1025.
Eisch, A.J., Petrik, D., 2012. Depression and hippocampal neurogenesis: a road to
remission? Science 338, 72–75.
Elfving, B., Plougmann, P., Wegener, G., 2010. Detection of brain-derived neuro-
trophic factor (BDNF) in rat blood and brain preparations using ELISA: pitfalls
and solutions. J. Neurosci. Methods 187, 73–77.
Erickson, K., Voss, M., Prakash, R., Basak, C., Szabo, A., Chaddock, L., Kim, J., Heo, S.,
Alves, H., White, S., Wojcicki, T, Mailey, E, Vieira, V., Martin, S., Pence, B.,
Woods, J., McAuley, E, Kramer, A., 2011. Exercise training increases size of
hippocampus and improves memory. Proc. Natl. Acad. Sci. USA 108, 3017–3022.
FMRIB Software Library. www.fmrib.ox.ac.uk/fsl.
Fabel, K., Fabel, K., Tab, B., Kaufer, D., Baiker, A., Simmons, N., Kuo, C., Palmer, T.,
2003. VEGF is necessary for exercise-induced adult hippocampal neurogenesis.
Eur. J. Neurosci. 18, 2803–2812.
Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry
23, 56–62.
Huang, T., Larsen, K.T., Ried-Larsen, M., Moller, N.C., Andersen, L.B., 2014. The
effects of physical activity and exercise on brain-derived neurotrophic factor in
healthy humans: a review. Scand. J. Med. Sci. 24 (1), 1–10.
Konrad, C., Ukas, T., Nebel, C., Arolt, V., Toga, A.W., Narr, K.L., 2009. Defining the
human hippocampus in cerebral magnetic resonance images – an overview of
current segmentation protocols. NeuroImage 47, 1185–1195.
Krogh, J., Videbech, P., Thomsen, C., Gluud, C., Nordentoft, M., 2012. DEMO-II trial.
aerobic exercise versus stretching exercise in patients with major depression –
a randomised clinical trial. PLoS One 7 (10), e48316.
Malykhin, N.V., Bouchard, T.P., Ogilve, C.J., Coupland, N.J., Seres, P., Camicioli, R.,
2007. Three-dimensional volumetric analysis and reconstruction of amygdala,
hippocampal head, body and tail. Psychiatr. Res. 155 (2), 155–165.
McKinnon, M.C., Yucel, K., Nazarov, A., MacQueen, G.M., 2009. A meta-analysis
examining clinical predictors of hippocampal volume in patients with major
depressive disorder. J. Psychiatry Neurosci. 34 (1), 41–54.
Meyers, J.E., Bayless, J.D., Meyers, K.R., 1996. Rey complex figure: memory error
patterns and functional abilities. Appl. Neuropsychol. 3, 89–92.
van Norden, A.G., de Laat, K.F., Fick, I., van Uden, I.W., van Oudheusden, L.J., Gons, R.A.,
Norris, D.G., Zwiers, M.P., Kessels, R.P., de Leeuw, F.E., 2012. Diffusion tensor
imaging of the hippocampus and verbal memory performance: the RUN DMC
study. Hum. Brain Mapp. 33 (3), 542–551.
Pajonk, F., Wobrock, T., Gruber, O., Scherk, H., Berner, D., Kaizl, I., Kierer, A., Muller, S.,
Oest, M., Meyer, T., Backens, M., Schneider-Axmann, T., Thornton, A., Honer, W.,
Falkai, P., 2010. Hippocampal plasticity in response to exercise in schizophrenia.
Arch. Gen. Psychiatry 67, 133–143.
Parker, B.A., Thompson, P.D., Jordan, K.C., Grimaldi, A.S., Assaf, M., Jagannathan, K.,
Pearlson, G.D., 2011. Effect of exercise training on hippocampal volume in
humans: a pilot study. Res. Q. Exerc. Sport 82 (3), 585–591.
Pereira, A., Huddleston, D., Brickman, A., Sosunov, S., Hen, R., McKhann, G., Sloan, R.,
Gage, F., Brown, T., Small, S., 2007. An in vivo correlate of exercise-induced
neurogenesis in the adult dentate gyrus. Proc. Natl. Acad. Sci. USA 104,
5638–5643.
Scheewe, T.W., van Haren, N.E., Sarkisyan, G., Schnack, H.G., Brouwer, R.M., de, G.M.,
Hulshoff Pol, H.E., Backx, F.J., Kahn, R.S., Cahn, W., 2013. Exercise therapy,
cardiorespiratory fitness and their effect on brain volumes: a randomised
controlled trial in patients with schizophrenia and healthy controls. Eur.
Neuropsychopharmacol. 23 (7), 675–685.
Schumacher, J., Jamra, R., Becker, T., Ohlraun, S., Klopp, N., Binder, E.B., et al., 2005.
Evidence for a relationship between genetic variants at the brain-derived
neurotrophic factor (BDNF) locus and Major Depression. Biol. Psychiatry 58,
307–314.
Shimizu, E., Hashimoto, K., Okamura, N., Koike, K., Komatsu, N., Kumakiri, C., 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.
Smith, S.M., Zhang, Y., Jenkinson, M., Chen, J., Mathews, P., Frederico, A., De Stefano,
N., 2002. Accurate, robust and automated longitudinal and cross-sectional brain
change analysis. NeuroImage 17 (1), 479–489.
Tapia-Arancibia, L., Rage, F., Givalois, L., Arancibia, S., 2004. Physiology of BDNF:
focus on hypothalamic function. Front. Neuroendocrinol. 25, 77–107.
Trejo, J.L., Llorens-Martin, M.V., Torres-Aleman, I., 2008. The effects of exercise on
spatial learning and anxiety-like behavior are mediated by an IGF-I-dependent
mechanism related to hippocampal neurogenesis. Mol. Cell. Neurosci. 37 (2),
402–411.
Vaynman, S., Ying, Z., Gomez-Pinilla, F., 2004. Hippocampal BDNF mediates the
efficacy of exercise on synaptic plasticity and cognition. Eur. J. Neurosci. 20,
2580–2590.
Videbech, P., Ravnkilde, B., 2004. Hippocampal volume and depression: a meta-
analysis of MRI studies. Am. J. Psychiatry 161 (11), 1957–1966.
Voss, M.W., Erickson, K.I., Prakash, R.S., Chaddock, L., Kim, J.S., Alves, H., Szabo, A.,
Phillips, S.M., Wojcicki, T.R., Mailey, E.L., Olson, E.A., Gothe, N., Vieira-Potter, V.J.,
Martin, S.A., Pence, B.D., Cook, M.D., Woods, J.A., McAuley, E., Kramer, A.F.,
2013a. Neurobiological markers of exercise-related brain plasticity in older
adults. Brain Behav. Immun. 28, 90–99.
Voss, M.W., Vivar, C., Kramer, A.F., van, P.H., 2013b. Bridging animal and human
models of exercise-induced brain plasticity. Trends Cogn. Sci. 17 (10), 525–544.