Journal of Psychiatric Research 101 (2018) 34–41

Contents lists available at ScienceDirect

Journal of Psychiatric Research

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

Running for extinction? Aerobic exercise as an augmentation of exposure T

therapy in panic disorder with agoraphobia
Sophie Bischoffa,∗, Gesine Wiederb, Franziska Einsleb, Moritz B. Petzolda, Christiane Janßenc,
Jennifer L.M. Mumma, Hans-Ulrich Wittchenb,d, Thomas Fydrichc, Jens Plaga,1, Andreas Ströhlea,1
a
Charité– Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of Health, Department of Psychiatry
and Psychotherapy, Campus Mitte, Charitéplatz 1, 10117, Berlin, Germany
b
Technische Universität Dresden, Institute of Clinical Psychology and Psychotherapy, Chemnitzer Str. 46, 01187, Dresden, Germany
c
Department of Psychology, Humboldt Universität Berlin, Unter den Linden 6, 10099, Berlin, Germany
d
Department of Psychiatry and Psychotherapy, Ludwig-Maximilians-Universität, Nußbaumstraße 7, 80336, Munich, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: Exposure-based Cognitive Behavioral Therapy (eb-CBT) represents the most evidence-based psychotherapeutic
Panic approach in anxiety disorders. However, its efficacy may be limited by a delay in onset of action and a sub-
Agoraphobia stantial number of patients does not respond sufficiently to treatment. In this context, aerobic exercise was found
Exercise to be effective in reducing clinical anxiety as well as to improve (elements of) disorder-specific CBT in some
Exposure
mental disorders. We therefore investigated the effect of aerobic exercise supplementary to an eb-CBT in panic
CBTClinicalTrials.gov Identifier:
NCT01928810
disorder and agoraphobia (PD/AG). 77 patients with PD/AG performed a 30 min treadmill task with moderate or
low intensity (70% or 30% of the maximal oxygen uptake [VO2max]) prior to five exposure sessions within a
standardized seven-week CBT. At baseline, after completing the treatment period (post) and six month after post
(follow-up), several measures of (un)specific psychopathology (Hamilton Anxiety Rating Scale [Ham-A],
Mobility Inventory [MI], Panic and Agoraphobia Scale [PAS], Agoraphobic Cognitions Questionnaire [ACQ],
Body Sensations Questionnaire [BSQ]) were established to assess for clinical changes. All patients experienced a
significant improvement of symptoms from baseline to post (for all measures p < .001) but repeated-measures
analyses of variance found a trend towards a significant time × group interaction in the Ham-A in favor for the
moderate intense exercise group (f[1, 74] = 4.15, p = .045, α = .025). This trend, however, disappeared at
follow-up since the low-intense exercise group further improved significantly in Ham-A after post. Our findings
therefore might point to an accelerating effect of moderate-intense exercise within an exposure-based CBT for
AG/PD.

1. Introduction and a significant number of patients do not respond to treatment

Psychotherapy and second-generation antidepressants currently re-
present the first-line treatment options in anxiety disorders (e.g.
Bandelow et al., 2012). In this field, exposure-based cognitive beha-
vioral therapy (CBT) has already been established for years in the daily
routine and has to be regarded as the best evidence-based psy-
chotherapeutic procedure for these conditions (Kaczkurkin and Foa,
2015). Although the efficacy of disorder-specific CBT was proven in
numerous randomized-controlled trials, a few clinical aspects limit
enthusiasm to some extent. Available data point to only moderate ef-
fects on symptoms of social anxiety disorder (SAD), generalized anxiety
disorder (GAD) or panic disorder (PD) with/without agoraphobia (AG)
(Hofmann and Smits, 2008; Taylor et al., 2012). Moreover, due to the
protracted changes in cognition and behavior, CBT may show a general
delay in onset of action of up to several weeks or months.
For these reasons, current research focused on strategies eligible to
facilitate the efficacy of exposure-based CBT in anxiety disorders.
Augmentation of exposure via pharmacological interventions (e.g. by
using the NMDA-partial agonist d-cycloserine) therefore has gained
substantial interest in this context and several modifications of psy-
chological techniques have been discussed in order to optimize fear
extinction (Mataix-Cols et al., 2017; Pittig et al., 2016).
In addition, growing evidence also points to a substantial impact of
physical activity in the treatment of clinical anxiety. Randomized-

∗
Corresponding author.
E-mail address: sophie.bischoff@charite.de (S. Bischoff).
1
Shared senior authorship.

https://doi.org/10.1016/j.jpsychires.2018.03.001
Received 11 December 2017; Received in revised form 11 February 2018; Accepted 4 March 2018
0022-3956/ © 2018 Elsevier Ltd. All rights reserved.
S. Bischoff et al.
controlled trials found an overall moderate effect of several-week
aerobic exercise programs on disorder-specific symptomatology in PD,
GAD or SAD (Stubbs et al., 2017) and single bouts of exercise demon-
strated acute anxiolytic activity in specific phobia or PD (Lindenberger
et al., 2017; Ströhle et al., 2009). Several biological and psychological
aspects such as reduction of anxiety sensitivity, changes in serotonergic
and endophinergic systems, improvement of self-efficacy or exercise-
induced exposure to fear-related bodily sensations were suspected to be
mediating mechanisms of these findings (Asmundson et al., 2013).
Furthermore, available data suggest that physical activity also may
enhance exposure training by facilitating the extinction of fear. As re-
ported by several studies, exercise was found to elevate levels of brain-
derived neurotrophic factor (BDNF; e.g. Neeper et al., 1996; Van
Kummer and Cohen, 2015) that in turn plays a prominent role in ex-
tinction learning (Andero and Ressler, 2012). In accordance with this
hypothesis, one preclinical trial found wheel-running to be significantly
associated with increased extinction learning in mice whereby the
duration of training and the amount of extinction learning were cor-
related positively to each other (Siette et al., 2014). Despite these
promising findings, clinical trials focusing on exercise-induced aug-
mentation of extinction learning in distinct mental disorders are rare.
To date, only one study investigated the immediate impact of exercise
on in-vivo disorder-specific exposure in Posttraumatic Stress Disorder
(PTSD). Within a small sample of only nine patients, a single bout of
70% VO2max exercise directly prior to sessions of prolonged exposure
(PE) has shown to be significantly more effective than PE alone (Powers
et al., 2012).
The aim of the present study therefore was to investigate the ad-
ditional effect of moderate-intense aerobic exercise (70% VO2max) on
exposure in patients with PD/AG who underwent a manualized ex-
posure-based CBT. In the light of former findings in this field, we hy-
pothesized that this type of aerobic exercise would be associated with a
stronger amplification and acceleration of psychotherapy than a less
intense exercise condition.
2. Material and methods
2.1. Design
The present study was conducted as a multi-center trial at three
centers in Germany: Department of Psychiatry and Psychotherapy,
Campus Charité Mitte, Charité – Universitätsmedizin Berlin;
Department of Psychology, Humboldt Universität Berlin and Institute of
Clinical Psychology and Psychotherapy, Technische Universität
Dresden. It was approved by the local ethics committee (registration
number: EA1/223/10) and registered (ClinicalTrials.gov Identifier:
NCT01928810). Patients were recruited over a period of 1.5 years. A
total of 77 patients with PD/AG were subsequently allocated to one of
the two treatment conditions by applying a single randomization. The
randomization was reciprocally accomplished by selected staff mem-
bers of the different study centers via the assignment of a random
number to one of the two treatment conditions. There were more
random numbers than necessary and therefore it was impossible to
predict the allocation of patients. Considering the nature of the trial,
patients could not be blinded to intervention but were blind to hy-
potheses. Therapists and assessors were blind to condition.
2.2. Sample size calculation
According to a study using a pharmacological approach to augment
exposure-based CBT in patients with PD/AG we expected an interaction
effect of η2p = .18 for the primary outcome measures (Siegmund et al.,
2011). With an alpha-level of 0.05 and power of 0.95 (one tailed),
power analysis using G*Power (Faul et al., 2009) resulted in a sample
size of 64 patients. The addition of further 14 patients for an expected
dropout rate of 20% led to a required sample size of n = 78.
35
Journal of Psychiatric Research 101 (2018) 34–41
2.3. Participants
Participants were recruited via specialized outpatient units at the
participating centers. Diagnosis of PD/AG was made according to the
Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (DSM-IV)
(American Psychiatric Association, 2000) and was verified by applying
the Composite International Diagnostic Interview (CIDI) for DSM-IV
(Essau and Wittchen, 1993). The following criteria were obligate to be
included in the study: age 16–65 years, Hamilton Anxiety Scale (Ham-
A; Hamilton, 1959) ≥ 18, Clinical Global Impression (CGI; Guy,
1976) ≥ 4 and being able to attend all therapy sessions. Exclusion
criteria were changes in pharmacological treatment within the last four
weeks, ongoing psychotherapy, suicidality, diagnosis of psychotic or
bipolar disorder, borderline personality disorder, and active alcohol
dependence. Other mental comorbidities were acceptable if PD/AG
represented the main diagnosis. Medical contraindications for ex-
posure-based CBT or exercise (e.g. severe cardiovascular, musculoske-
letal, pulmonary or neurological disorders) were assessed via recording
a detailed medical history by the study physician (JP). Moreover, car-
diopulmonary fitness of participants was ensured by a spiroergometry
that was performed prior to the start of the trial (see also “exercise
interventions”). For explorative analyses the amount of physical ac-
tivity in minutes per week (defined as activity that made subjects sweat
or getting out of breath) was obtained via two questions derived from
the “German Health Update 2009” survey (Lampert et al., 2012).
2.4. Psychotherapy
2.4.1. CBT
All patients received a manualized exposure-based CBT (Gloster
et al., 2011; Lang et al., 2012) consisting of twelve sessions over a time
period of seven weeks (two sessions à 100 min per week). The treatment
included empirically verified components such as psychoeducation,
cognitive reframing and interoceptive exposure. In addition, therapist-
guided in-vivo exposure were conducted during session 6, 7, 8, 10 and
11. The first three exposure situations were standardized (bus, shopping
center, forest) and the last two sessions were chosen individually by
visiting the patients' most relevant agoraphobic context.
2.4.2. Therapists
The therapist team consisted of 31 psychologists (four of them were
male). Eight of them were certified psychotherapists while the others
were in training for psychotherapy. They all underwent a two-day
training consisting of detailed instructions for applying the manual
(Lang et al., 2012) and including several practical exercises. A video-
tape of each therapist performing a central therapy exercise (thought
experiment) was rated for adherence as well as competence and served
as basis for becoming licensed as a study therapist. This procedure was
based on the procedure within the “MAC” study published by Gloster
and colleagues (Gloster et al., 2009). In order to control for treatment
integrity, all therapy sessions (except for exposure) were videotaped
and analyzed by two trained master students of psychology. Therapists
showed good adherence and competence (adherence: M = 6.00,
SD = 1.59, competence: M = 5.49, SD = 1.42) on the therapist ad-
herence and competence rating scale for PD/AG (Gloster et al., 2008).
There were no significant differences between both study groups (see
below) concerning adherence (f[1,40] = 0.60, p = .45) and compe-
tence (f[1,40] = 0.07, p = .79).
2.5. Exercise interventions
2.5.1. Determination of optimum training level
In order to determine the optimum training level as well as to de-
finitively exclude any somatic contraindications to physical strain,
every patient underwent a modified Bruce treadmill spiroergometry
(Knubben et al., 2007) prior to the start of psychotherapy. Patients were
S. Bischoff et al. Journal of Psychiatric Research 101 (2018) 34–41

Fig. 1. Participants flow.

36
S. Bischoff et al.
instructed to run until they subjectively felt exhausted while the
workload increased every three minutes (Borg, 1974). The individual
maximal oxygen uptake was charged and normalized by body weight in
ml/kg/min (VO2max) according to the guidelines for exercise testing
and prescription by the American College of Sports Medicine (ACSM,
2013).
2.5.2. Moderate and low intense aerobic exercise
Prior to each of the five exposure sessions patients performed a
30 min treadmill task of either 70% VO2max (moderate intense exercise;
MI-E) or 30% VO2max (low intense exercise; LI-E). Speed and slope of
the treadmill could be adjusted independently in order to ensure the
required heart rate. Interventions were monitored by a staff member
and were identical in the two conditions regarding the content, dura-
tion, structure and the process of the treadmill intervention.
2.6. Assessment
Assessment of symptoms took place before (baseline), immediately
after (post) as well as 6 months after the therapy (follow-up) and in-
cluded a structured interview and several self-reports. To ensure the
correct administration of instruments all assessors were trained by
certified experts.
2.6.1. Outcome variables
The scores of the Hamilton Anxiety Rating Scale (Ham-A; Hamilton,
1959) and the Mobility Inventory (MI; Chambless et al., 1985) served as
primary outcome variables. The Ham-A is a structured interview to
investigate the severity of anxiety disorders with the subscales psychic
anxiety (mental agitation and psychological distress) and somatic an-
xiety (physical complaints related to anxiety). The MI is a self-report
questionnaire to assess the amount of agoraphobic avoidance (alone
and accompanied) for 27 situations. As secondary outcome parameters,
the scores of the self-rated version of the Panic and Agoraphobia Scale
(PAS; Bandelow, 1995), the Body Sensations Questionnaire (BSQ; Ehlers
et al., 1993), the Agoraphobic Cognitions Questionnaire (ACQ; Ehlers
et al., 1993) and the EuroQol (EQ-5D; Rabin and de Charro, 2001) were
chosen to further assess for severity of PD/AG, fear from bodily sen-
sations and health-related quality of life, respectively. Since HAMA, MI,
PAS, BSQ and ACQ provide a positive correlation between score and
symptom severity, greater values in the EQ-5D correspond to a better
quality of life.
2.7. Statistical procedures
Repeated-measures analyses of variance were conducted with group
as between-subject factor and time as within-subject factor. To in-
vestigate the amplification of CBT by MI-E, data from baseline to post
were investigated. Time stability of this effect was analyzed with data
from post to follow-up. Sphericity was analyzed using the Maulchy test.
In case of violation of sphericity (p ≤ .05) the degrees of freedom were
corrected by Greenhouse-Geisser. In the following, p-values are pre-
sented as two-tailed tests. As there were two primary outcome mea-
sures, we used Bonfferoni correction to adjust for multiple testing.
Therefore alpha level for repeated-measures analyses of variance of
these outcomes was set to ≤ .025. Effect sizes are reported as partial
eta-squared (η2p) for analyses of variance and for t-tests. Additionally
Cohen's d is reported as an effect size for the change from baseline to
post, from baseline to follow-up and from post to follow-up for each
group separately. All procedures were executed by IBM SPSS Statistics
22.
For intention-to-treat analysis missing data (due to drop out or in-
complete questionnaires) were replaced by the expectation-maximiza-
tion algorithm (EM) on scale basis via SPSS Statistics 22. EM is one of
the best procedures to estimate parameters and is based on a maximum-
likelihood distribution (Dempster et al., 1977). Imputation was carried
37
Journal of Psychiatric Research 101 (2018) 34–41
out separately for MI-E and LI-E to preserve interaction effects. Baseline
values were always included in imputation models. In addition, all
analyses were conducted with treatment completers (n = 51). The
analyses showed similar results as the ones with EM replacement and
led to the same conclusions (results available on request).
3. Results
3.1. Sample characteristics
149 out of the total of 388 screened patients provided informed
consent. 77 of them were eligible for inclusion in the study and were
subsequently randomized to one of the two study arms (see Fig. 1).
During course of therapy 33.8% (26 of 77) of patients dropped out due
to diverse reasons. The most common dropout reasons were organiza-
tional problems like difficulties to meet the strict time schedule. Two
participants decided to quit treatment because they were dissatisfied
with the exercise intervention. Most of the drop-outs occurred between
session 4 and post assessment (n = 19) which also presented the most
demanding phase due to exposure and exercise interventions.
22 of the patients were treated with antidepressants (mostly SSRIs
or SNRIs, n = 16) at the beginning of the study. Before the treatment
started nine patients received beta blockers and four took benzodiaze-
pine on demand. On-demand medication was strictly prevented during
treatment.
Patients in the MI-E and LI-E group did not differ significantly in
age, gender, dropout rates, duration of agoraphobia, the number of 12-
month diagnoses, the use of anxiolytic medication as well as socio-
demographic characteristics (see Table 1). Moreover, there were no
significant group differences regarding baseline severity of illness (see
Table 2).
3.2. Effect of treatment conditions
3.2.1. Baseline to post
Unadjusted repeated-measures analyses of variance with data from
baseline to post revealed a trend (after Bonferroni correction) towards
significance on the interaction effect for the Ham-A (f[1, 75] = 4.04,
p = .048) in favor for the MI-E group. After adjusting for the baseline
difference concerning the amount of PA this effect remained a trend
towards significance (f[1, 74] = 4.15, p = .045; see Table 3 for ad-
justed results): The Ham-A reduction of 11.4 points from baseline to
post in the MI-E group was non-significantly greater than the reduction
of 8.0 points in the LI-E group (η2p = .053; see Fig. 2). No interaction
effects were found for data from baseline to post for the PAS, MI, ACQ,
BSQ and EQ-5D in both unadjusted and adjusted analyses (see Table 3).
The change in physical activity (minutes per week) from baseline to
post also did not differ significantly between the groups.
Overall, the exposure based CBT led to highly significant symptom
improvements according to all used psychometric instruments in both
groups.
3.2.2. Post to follow-up
Unadjusted repeated-measures analyses of variance with data from
post to follow-up found no significant interaction effects. However, the
LI-E group reduced its Ham-A score from post to follow-up significantly
by 2.1 points (t[37] = 2.10, p = .043, d = 0.28) while patients per-
formed MI-E showed a non-significant reduction of 0.5 points (t
[38] = 0.64, p = .524, d = 0.07) (see Fig. 2 and Table 2). Both groups
gained a further significant symptom improvement from post to follow-
up in the majority of instruments (see Tables 2 and 4).
4. Discussion
For the first time, this study investigated the effects of MI-E on
disorder-specific exposure therapy as a part of a standardized CBT in
S. Bischoff et al. Journal of Psychiatric Research 101 (2018) 34–41

Table 1
Patient characteristics by treatment group and t-tests or chi-square tests.

Total sample MI-E LI-E t/χ2 df P

N 77 39 38
Age (M ± SD, range) 36.7 ± 10.0 37.2 ± 10.0 36.2 ± 10.1 0.45 75 .656
22–64 23–63 22–64
Gender
Male 28 15 13
Female 49 24 25 0.15 1 .698
Dropout (from BL to FU)
Yes 26 15 11
No 51 24 27 0.78 1 .377
Social class
Lowest 5 1 4 4.72 4 .318
Lower middle 24 15 9
Middle 37 18 19
Upper middle 10 4 6
Upper 0 0 0
Marital status
Married 22 11 11 2.97 4 .5639
Separated 3 2 1
Divorced 7 5 2
Widowed 1 1 0
Never married 44 20 24
Employment
Yes 47 23 24 1.04 2 .594
No 29 15 14
Housewife/Househusband 1 1 0
Number of 12-month diagnoses (M ± SD) 4.6 ± 2.0 4.8 ± 2.0 4.4 ± 2.1 0.97 75 .335
Duration of agoraphobia, years (M ± SD) 11.8 ± 10.2 13.7 ± 11.8 9.8 ± 7.8 1.53 59 .132
Antidepressants
Yes 22 12 10 0.19 1 .665
No 55 27 28

VO2max (M ± SD, range) 32.4 ± 8.1 32.1 ± 8.3 32.8 ± 8.0 −0.34 72 .736
11–52 11–52 21–52
PA, minutes per week (M ± SD, range) 87 ± 100 68 ± 70 108 ± 120 −1.76 75 .083
0–455 0–260 0–455

Ham-A (M ± SD) 22.2 ± 7.8 23.4 ± 8.2 20.9 ± 7.2 1.44 75 .155
PAS (M ± SD) 25.0 ± 9.0 25.8 ± 8.2 24.2 ± 9.8 0.78 75 .439
MI (M ± SD) 2.5 ± 0.8 2.7 ± 0.9 2.4 ± 0.7 1.34 75 .185
ACQ (M ± SD) 2.2 ± 0.6 2.3 ± 0.6 2.1 ± 0.6 1.38 75 .173
BSQ (M ± SD) 46.6 ± 10.7 47.4 ± 11.1 45.7 ± 10.4 0.71 75 .477
EQ-5D (M ± SD) 0.6 ± 0.2 0.6 ± 0.2 0.6 ± 0.1 −0.99 75 .326

MI-E, moderate intense exercise group; LI-E, low intense exercise group; M, mean; SD, standard deviation; BL, Baseline; FU, Follow-up; VO2max, maximal oxygen uptake; Ham-A,
Hamilton Anxiety Scale; PAS, Panic and Agoraphobia Scale; MI, Mobility Inventory; ACQ, Agoraphobic Cognitions Questionnaire; BSQ, Body Sensations Questionnaire; EQ-5D, EuroQol:
health-related quality of life.

patients with AG/PD. Since exposure-based CBT led to an overall im-
provement of symptoms from baseline to post in our sample, patients
conducted MI-E showed a trend to stronger reduction of the Ham-A
score at the end of therapy than those who performed a less intense
exercise protocol. While the MI-E group maintained its Ham-A im-
provement over time, the LI-E group then caught up by showing a
further significant decrease in Ham-A score from post to follow-up. As a
result, both groups reached similar improvements at the end of the
observation period.
To date, one further trial investigated MI-E with respect to in-vivo
exposure training in a distinct mental condition. 30-minutes exercise in
an intensity of 70% VO2max prior was performed to each of 12 sessions
of prolonged exposure (PE) in nine patients suffered from posttraumatic
stress disorder (PTSD). Subsequently, MI-E plus PE led to a significantly
greater reduction of several core symptoms of PTSD (as measured by
the “PTSD Symptom Scale”) as well as to a significantly stronger rise of
BDNF after the end of the treatment period than PE alone (Powers et al.,
2012). Obviously, this finding is in contrast to our results since MI-E in
the present trial did not affect any of the disorder-specific measures
(e.g. PAS, MI) and we only found a non-significant improvement of
symptomatic clinical anxiety assessed by the Ham-A. These divergent
findings, however, might be explained in some extend by methodolo-
gical differences between the two trials. In the study of Powers and
colleagues, the exercise task was performed more than twice as often as
we did it in the present study. Since a relationship between the duration
of exercise and the degree of extinction learning was already described
(Siette et al., 2014), the lack of significant efficacy of MI-E on disorder-
specific symptoms in our sample may be due to the smaller number of
exercise sessions. Furthermore, for reasons of feasibility (e.g. different
reachability of individual exposure settings) there has been a time gap
of about 33 min on average between exercise intervention and start of
exposure in the present trial. In addition to the relatively small number
of exercise sessions, an enlarged exercise-exposure interval therefore
may also account for lowering the effect of exercise on exposure. A
biological base of these assumptions might be provided by BDNF and its
role for learning mechanisms. Since a beneficial effect of BDNF on ex-
tinction learning was already described in preclinical trials (Andero and
Ressler, 2012), a couple of evidence also points to the influenceability
of BDNF by physical exercise (Huang et al., 2014). Powers et al. also
investigated BDNF as a potential mediator of the effect of exercise on PE
efficacy in their sample of PTSD patients. As a result, patients per-
formed a combined approach of exercise and PE showed both, a sig-
nificant reduced symptomatology and an increased BDNF level com-
pared to those who received PE alone (Powers et al., 2012). However,
former data suggest that the impact of aerobic exercise on BDNF ap-
pears to be dependent on the type of exercise. Regular aerobic exercise

38
S. Bischoff et al. Journal of Psychiatric Research 101 (2018) 34–41

Table 2
Mean scores for outcome measures at different assessment points and effect sizes compared to baseline and post.

Baseline Post Follow-up Effect size baseline – post (d) Effect size baseline – follow-up (d) Effect size post – follow-up (d)

Ham-A
MI-E group
M ± SD 23.4 ± 8.2 12.0 ± 7.6 11.5 ± 7.6 1.39 1.45 0.07
LI-E group
M ± SD 20.90 ± 7.1 13.0 ± 7.5 10.9 ± 7.1 1.11 1.41 0.28
PAS
MI-E group
M ± SD 25.8 ± 8.2 14.1 ± 8.5 10.1 ± 7.6 1.43 1.91 0.47
LI-E group
M ± SD 24.2 ± 9.8 12.1 ± 7.7 10.0 ± 6.5 1.23 1.45 0.27
MI
MI-E group
M ± SD 2.7 ± 0.9 1.8 ± 0.8 1.5 ± 0.4 1.01 1.36 0.35
LI-E group
M ± SD 2.4 ± 0.7 1.5 ± 0.4 1.3 ± 0.3 1.29 1.67 0.37
ACQ
MI-E group
M ± SD 2.3 ± 0.6 1.6 ± 0.4 1.5 ± 0.5 1.17 1.33 0.25
LI-E group
M ± SD 2.1 ± 0.6 1.6 ± 0.4 1.4 ± 0.4 0.83 1.17 0.50
BSQ
MI-E group
M ± SD 47.4 ± 11.1 34.1 ± 10.3 31.4 ± 10.7 1.20 1.44 0.26
LI-E group
M ± SD 45.7 ± 10.4 32.7 ± 8.0 29.1 ± 8.4 1.25 1.60 0.45
EQ-5D
MI-E group
M ± SD 0.6 ± 0.2 0.7 ± 0.2 0.8 ± 0.1 −0.50 −1.00 −0.50
LI-E group
M ± SD 0.6 ± 0.1 0.7 ± 0.2 0.8 ± 0.2 −1.00 −2.00 −0.50
Physical activity minutes/week
MI-E group
M ± SD 68 ± 70 85 ± 91 78 ± 71 −0.24 −0.14 0.08
LI-E group
M ± SD 108 ± 121 122 ± 105 119 ± 86 −0.12 −0.09 0.03

MI-E, moderate intense exercise; LI-E, low intense exercise; Ham-A, Hamilton Anxiety Scale; PAS, Panic and Agoraphobia Scale; MI, Mobility Inventory; ACQ, Agoraphobic Cognitions
Questionnaire; BSQ, Body Sensations Questionnaire; EQ-5D, EuroQol: health-related quality of life.

(of up to several weeks) was mostly found to be associated with an
increase of resting BDNF levels even after the exercise was dis-
continued. In contrast, the overall majority of clinical trials that in-
vestigated the effect of acute exercise did not observe such a sustained
effect. Although BDNF also significantly increased while exercise was
performed, BDNF levels quickly returned to baseline during the ensuing
resting period (Huang et al., 2014). In the present study, an augmen-
tative effect of MI-E induced elevated BDNF on agoraphobic exposure
therefore might impeded by the small number of MI-E units as well as
the time gap between termination of MI-E and the start of exposure
sessions. With respect to optimum standardization we further per-
formed an active control condition (LI-E) while Powers and colleagues
conducted PE with or without MI-E. Therefore, it seems to be possible
that even LI-E might have beneficial effects for our patients with AG/PD
and overlapping psychological mechanisms (e.g. increase of self-effi-
cacy increase, reduction of anxiety sensitivity or exposure to fear-re-
lated bodily sensations, e.g. Asmundson et al., 2013) might have re-
duced the differences between our groups. Finally, the well proven and
comprehensive CBT program in our study (that includes not only ex-
posure but also psychoeducation and cognitive reframing) has shown to
be very effective in reducing agoraphobic fear (Cohens d = −1.3 for
PAS). Therefore, there might be a ceiling effect without any space for
significantly improvement by MI-E.
Considering these aspects, however, the trend to greater baseline-to-
post reduction of the Ham-A score in the present trial suggests that MI-E
also might hold some potential for augmentation of exposure therapy in
patients with AG/PD. In our group of patients, the augmentative effect
seems to occur as an acceleration of treatment since the between-group
difference disappeared at follow-up. Unfortunately, Powers and
coworkers missed to carry out a follow-up measurement and therefore
it remains unclear whether the beneficial effect of MI-E on PTSD re-
mained stable or not. In fact, our finding agree in some extend with
studies focusing on a pharmacological augmentation of disorder-spe-
cific exposure training with d-cycloserine (DCS). DCS is a partial N-
methyl-D-aspartate agonist that has shown to facilitate extinction
learning in animals and patients suffering from several anxiety dis-
orders, obsessive compulsive disorder and posttraumatic stress disorder
(e.g. Mataix-Cols et al., 2017). In several trials addressing social anxiety
disorder as well as PD with or without AG, however, a benefit of DCS
administration prior to exposure sessions was only found at early as-
sessments (Hofmann et al., 2013; Otto et al., 2016) or speeded up the
therapy outcome particularly in severely affected patients (Siegmund
et al., 2011). Comparable to DCS, MI-E therefore might contribute to
the efficacy of exposure training in PD/AG by accelerating the response
to treatment.
However, there are is also one trial that failed to detect any effect of
moderate to high-intense exercise on exposure therapy in subjects with
elevated fear (Jacquart et al., 2017). Within a randomized-controlled
design, exercise or a period of rest were administered prior to a unit of
virtual reality exposure therapy (VRET) in 59 patients with fear of
heights. Exercise contains a single bout of 30 min treadmill training
within 80% of the maximum heart (HRmax). As measured by the
“Acrophobia Questionnaire”, patients in both groups significantly im-
proved up to 14 days after exposure, respectively, but there was no
additional (non-)significant effect in the exercise group at any mea-
surement (Jacquart et al., 2017). Beside several strengths of this trial
(e.g. the lack of an active control condition in order to avoid con-
founding influences), some limitations have to be taken into account

39
S. Bischoff et al. Journal of Psychiatric Research 101 (2018) 34–41

Table 3 Table 4
Repeated-measures ANOVA with data from baseline to post assessment. Repeated-measures ANOVA with data from post to follow-up assessment.

Measure Effect df f p η2p Measure Effect df f p η2p

Primary Outcomes Primary Outcomes

Ham-A a time 1/74 62.67 < .001 .459 Ham-A a time 1/74 2.85 .096 .037
group 1/74 0.16 .687 .004 group 1/74 0.66 .798 .001
time x 1/74 4.15 .045 .053 time x group 1/74 1.40 .241 .019
a
group MI time 1/74 20.48 < .001 .217
a
MI time 1/74 159.21 < .001 .683 group 1/74 3.42 .068 .044
group 1/74 2.09 .152 .027 time x group 1/74 0.55 .461 .007
time x 1/74 0.60 .807 .001 Secondary Outcomes
group PAS a time 1/74 13.76 < .001 .157
Secondary Outcomes group 1/74 0.07 .791 .001
PAS a time 1/74 41.66 < .001 .360 time x group 1/74 0.86 . 356 .012
a
group 1/74 0.61 .438 .008 ACQ time 1/74 2.41 .125 .032
time x 1/74 < 0.01 .999 < .001 group 1/74 0.16 .693 .002
group time x group 1/74 0.70 .407 .009
a a
ACQ time 1/74 74.45 < .001 .502 BSQ time 1/74 9.62 .003 .115
group 1/74 0.67 .415 .009 group 1/74 0.76 .386 .010
time x 1/74 1.70 .197 .022 time x group 1/74 0.29 .594 .004
a
group EQ-5D time 1/74 1.08 .302 .014
a
BSQ time 1/74 84.67 < .001 .534 group 1/74 0.01 .923 < .001
group 1/74 0.43 .513 .006 time x group 1/74 1.04 . 311 .014
time x 1/74 0.01 .943 < .001
Physical activity minutes/week time 1/74 0.63 .802 .001
group a
a group 1/74 3.82 .054 .049
EQ-5D time 1/74 32.64 < .001 .306
time x group 1/74 0.31 .861 < .001
group 1/74 0.06 .807 .001
time x 1/74 1.33 .252 .018 a
ANOVA adjusted for baseline amount of physical activity; Ham-A, Hamilton Anxiety
group
Scale; MI, Mobility Inventory; PAS, Panic and Agoraphobia Scale; ACQ, Agoraphobic
Physical activity minutes/ time 1/75 0.97 .329 .013 Cognitions Questionnaire; BSQ, Body Sensations Questionnaire; EQ-5D, EuroQol: health-
week group 1/75 5.19 .026 .065 related quality of life.
time x 1/75 0.63 .802 .001
group
exposure was conducted. As already discussed with regards to our trial,
a
ANOVA adjusted for baseline amount of physical activity; Ham-A, Hamilton Anxiety these aspects might have negatively impacted the efficacy of exercise on
Scale; MI, Mobility Inventory; PAS, Panic and Agoraphobia Scale; ACQ, Agoraphobic exposure. Moreover, more than half of the patients included by Jac-
Cognitions Questionnaire; BSQ, Body Sensations Questionnaire; EQ-5D, EuroQol: health- quart and colleagues demonstrated sub-categorical symptom severity
related quality of life. and did not meet the criteria for acrophobia. As discussed by the au-
thors themselves, this aspect also might have attenuated the additional
anxiolytic effect of exercise in this sample, since significant symptom
severity has thought to be critical for the efficacy of other augmentative
strategies such as DCS (Guastella et al., 2007). In contrast to the present
trial as well as to the PTSD trial, Jacquart and colleagues further per-
formed a VRET instead of a real-life in-vivo exposure. VRET has also
shown to be an effective treatment in anxiety disorders (e.g. Botella
et al., 2017), but several additional components have been identified to
be critical for its efficacy in a therapeutical context (e.g. the “presence”
within the virtual reality or the sensory “perception” of presented
material; Diemer et al., 2015). Until now, there are no studies available
that compare the impact of exercise on in-vivo exposure and VRET in
mental disorders. Therefore, VRET-specific characteristics might also be
responsible for the absence of detectable effects of MI-E in some extend.
However, there are some limitations with regards to our trial. As
already noted above, a significant exercise-exposure latency, the use of
an active control group and a small number of augmented exposure
sessions may have undermined significant effects of MI-E in our sample.
We also had to deal with a considerable amount of dropouts (33.8%)
during active treatment and the follow-up period. Actually, only two
patients discontinued for reasons related to exercise. Moreover, all
patients received social interaction during CBT and exercise. Since the
psychotherapy as well as the exercise program were strictly standar-
dized for both groups, however, this aspect should not significantly
impact our findings. The strengths of the present study are the use of an
evidence-based and well-proved CBT manual as well as the assessment
Fig. 2. Hamilton Anxiety Scale (Ham-A) Score at baseline post and follow-up. Error bars
of therapists' qualities in order to gain ideal therapeutic results. In ad-
indicate the standard deviation (SD).
dition, the quantification of individual optimum training level by con-
ducting a spiroergometry in our trial should be further a significant
that might contribute to the negative result. First, there also was a advantage of this trial since it should have been contribute to a best
significant time gap of about 20 min between the termination of ex- possible personalized application of MI-E and MI-L.
ercise and the start of VRET and only one session of combined exercise/

40
S. Bischoff et al.
5. Conclusion
Although we only found a trend for a stronger reduction of anxiety
at post, MI-E might provide some potential for accelerating the effects
of exposure training in patients with PD/AG as seen in trials using a
pharmacological augmentation approach. The absence of a significant
effect on disorder-(un)specific symptomatology, however, mainly might
be due to methodological aspects already discussed above. Future trials
therefore should address these concerns particularly by choosing a non-
active control group, shortening the exercise-exposure gap as well as
providing a larger number of exercise-exposure sessions. Moreover, the
“pure” effect of MI-E on exposure in PD/AG might be elucidated more
clearly if exposure is carried out as a stand-alone therapy rather than as
an integral part of CBT.
References
ACSM, 2013. ACSM's Guidelines for Exercise Testing and Prescription, ninth ed.
Lippincott, Williams & Wilkins, Philadelphia.
American Psychiatric Association, 2000. Diagnostic and Statistical Manual of Mental
Disorders, fourth ed. APA, Washington DC text rev.
Andero, R., Ressler, K.J., 2012. Fear extinction and BDNF: translating animal models of
PTSD to the clinic. Gene Brain Behav. 11, 503–512.
Asmundson, G.J., Fetzner, M.G., Deboer, L.B., Powers, M.B., Otto, M.W., Smits, J.A.,
2013. Let's get physical: a contemporary review of the anxiolytic effects of exercise
for anxiety and its disorders. Depress. Anxiety 30, 362–373.
Bandelow, B., 1995. Assessing the efficacy of treatments for panic disorder and agor-
aphobia. II. The Panic and Agoraphobia Scale. Int. Clin. Psychopharmacol. 10, 73–81.
Bandelow, B., Sher, L., Bunevicius, R., Hollander, E., Kasper, S., Zohar, J., Möller, H.J.,
WFSBP Task Force on Mental Disorders in Primary Care; WFSBP Task Force on
Anxiety Disorders, OCD and PTSD, 2012. Guidelines for the pharmacological treat-
ment of anxiety disorders, obsessive-compulsive disorder and posttraumatic stress
disorder in primary care. Int. J. Psychiatr. Clin. Pract. 16, 77–84.
Botella, C., Fernández-Álvarez, J., Guillén, V., García-Palacios, A., Baños, R., 2017. Recent
Progress in Virtual Reality Exposure Therapy for Phobias: A Systematic Review. Curr.
Psychiatry Rep. 19, 42.
Borg, G.A., 1974. Perceived exertion. Exerc. Sport Sci. Rev. 2, 131–153.
Chambless, D.L., Caputo, G.C., Jasin, S.E., Gracely, E.J., Williams, C., 1985. The mobility
inventory for agoraphobia. Behav. Res. Ther. 23, 35–44.
Dempster, A.P., Laird, N.M., Rubin, D.B., 1977. Maximum-Likelihood from incomplete
data via the EM algorithm. J. Royal Stat. Soc. Series B Stat. Methodol. 39, 1–38.
Diemer, J., Alpers, G.W., Peperkorn, H.M., Shiban, Y., Mühlberger, A., 2015. The impact
of perception and presence on emotional reactions: a review of research in virtual
reality. Front. Psychol. 30, 26.
Ehlers, A., Margraf, J., Chambless, D., 1993. Fragebogen zu körperbezogenen Ängsten,
Kognitionen und Vermeidung. AKV (mit BSQ und ACQ). Beltz Test, Weinheim.
Essau, C.A., Wittchen, H.U., 1993. An overview of the composite international Diagnostic
interview (CIDI). Int. J. Methods Psychiatr. Res. 1993, 79–85.
Faul, F., Erdfelder, E., Buchner, A., Lang, A.G., 2009. Statistical power analyses using
G*Power 3.1: tests for correlation and regression analyses. Behav. Res. Meth. 41,
1149–1160.
Gloster, A.T., Einsle, F., Lang, T., Hauke, T., Wittchen, H.U., 2008. The Therapist
Adherence and Competency Rating Scales for Panic Disorder with Agoraphobia.
Unpublished manuscript.
Gloster, A.T., Wittchen, H.U., Einsle, F., Hofler, M., Lang, T., Helbig-Lang, S., Fydrich, T.,
Fehm, L., Hamm, A.O., Richter, J., Alpers, G.W., Gerlach, A.L., Ströhle, A., Kircher,
T., Deckert, J., Zwanzger, P., Arolt, V., 2009. Mechanism of action in CBT (MAC):
methods of a multi-center randomized controlled trial in 369 patients with panic
disorder and agoraphobia. Eur. Arch. Psychiatr. Clin. Neurosci. 259 (2), S155–S166.
Gloster, A.T., Wittchen, H.U., Einsle, F., Lang, T., Helbig-Lang, S., Fydrich, T., Fehm, L.,
Hamm, A.O., Richter, J., Alpers, G.W., Gerlach, A.L., Ströhle, A., Kircher, T., Deckert,
J., Zwanzger, P., Hofler, M., Arolt, V., 2011. Psychological treatment for panic dis-
order with agoraphobia: a randomized controlled trial to examine the role of thera-
pist-guided exposure in situ in CBT. J. Consult. Clin. Psychol. 79, 406–420.
Guastella, A.J., Dadds, M.R., Lovibond, P.F., Mitchell, P., Richardson, R., 2007. A ran-
domized controlled trial of the effect of D-cycloserine on exposure therapy for spider
fear. J. Psychiatr. Res. 41, 466–471.
Guy, W., 1976. Clinical global impressions. In: Guy, W. (Ed.), ECDEU Assessment Manual
for Psychopharmacology (Revised). National Institute of Mental Health, Rockville,
pp. 217–221.
Hamilton, M., 1959. The assessment of anxiety states by rating. Br. J. Med. Psychol. 32,
41
Journal of Psychiatric Research 101 (2018) 34–41
50–55.
Hofmann, S.G., Smits, J.A., 2008. Cognitive- therapy for adult anxiety disorders: a meta-
analysis of randomized placebo-controlled trials. J. Clin. Psychiatr. 69, 621–632.
Hofmann, S.G., Smits, J.A., Rosenfield, D., Simon, N., Otto, M.W., Meuret, A.E., Marques,
L., Fang, A., Tart, C., Pollack, M.H., 2013. D-Cycloserine as an augmentation strategy
with cognitive-behavioral therapy for social anxiety disorder. Am. J. Psychiatr. 170,
751–758.
Huang, T., Larsen, K.T., Ried-Larsen, M., Møller, N.C., Andersen, L.B., 2014. The effects of
physical activity and exercise on brain-derived neurotrophic factor in healthy hu-
mans: a review. Scand. J. Med. Sci. Sports 24, 1–10.
Jacquart, J., Roquet, R.F., Papini, S., Powers, M.B., Rosenfield, D., Smits, J.A.J., Monfils,
M.H., 2017. Effects of acute exercise on fear extinction in rats and exposure therapy
in humans: null findings from five experiments. J. Anxiety Disord. 50, 76–86.
Kaczkurkin, A.N., Foa, E.B., 2015. Cognitive-behavioral therapy for anxiety disorders: an
update on the empirical evidence. Dialogues Clin. Neurosci. 17, 337–346.
Knubben, K., Reischies, F.M., Adli, M., Schlattmann, P., Bauer, M., Dimeo, F., 2007. A
randomised, controlled study on the effects of a short-term endurance training pro-
gramme in patients with major depression. Br. J. Sports Med. 41, 29–33.
Lampert, T., Mensink, G.B., Muters, S., 2012. Physical and sporting activity among adults
in Germany. Results from the “German Health Update 2009” survey.
Bundesgesundheitsblatt - Gesundheitsforsch. - Gesundheitsschutz 55, 102–110.
Lang, T., Helbig-Lang, S., Westphal, D., Gloster, A.T., Wittchen, H.U., 2012.
Expositionsbasierte Therapie der Panikstörung mit Agoraphobie: Ein
Behandlungsmanual. Hogrefe Verlag, Göttingen.
Lindenberger, B.L., Plag, J., Schumacher, S., Gaudlitz, K., Bischoff, S., Bobbert, T., Dimeo,
F., Petzold, M.B., Kirschbaum, C., Dudás, Z., Ströhle, A., 2017. Clinical and neuro-
biological effects of aerobic exercise in dental phobia: a randomized controlled trial.
Depress. Anxiety 2017. http://dx.doi.org/10.1002/da.22659.
Mataix-Cols, D., Fernández de la Cruz, L., Monzani, B., Rosenfield, D., Andersson, E.,
Pérez-Vigil, A., Frumento, P., de Kleine, R.A., Difede, J., Dunlop, B.W., Farrell, L.J.,
Geller, D., Gerardi, M., Guastella, A.J., Hofmann, S.G., Hendriks, G.J., Kushner, M.G.,
Lee, F.S., Lenze, E.J., Levinson, C.A., McConnell, H., Otto, M.W., Plag, J., Pollack,
M.H., Ressler, K.J., Rodebaugh, T.L., Rothbaum, B.O., Scheeringa, M.S., Siewert-
Siegmund, A., Smits, J.A.J., Storch, E.A., Ströhle, A., Tart, C.D., Tolin, D.F., van
Minnen, A., Waters, A.M., Weems, C.F., Wilhelm, S., Wyka, K., Davis, M., Rück, C.,
and the DCS Anxiety Consortium,Altemus, M., Anderson, P., Cukor, J., Finck, C.,
Geffken, G.R., Golfels, F., Goodman, W.K., Gutner, C., Heyman, I., Jovanovic, T.,
Lewin, A.B., McNamara, J.P., Murphy, T.K., Norrholm, S., Thuras, P., 2017. D-cy-
closerine augmentation of exposure-based cognitive behavior therapy for anxiety,
obsessive-compulsive, and posttraumatic stress disorders: a systematic review and
meta-analysis of individual participant data. JAMA Psychiatry 74, 501–510.
Neeper, S.A., Gomez-Pinilla, F., Choi, J., Cotman, C.W., 1996. Physical activity increases
mRNA for brain-derived neurotrophic factor and nerve growth factor in rat brain.
Brain Res. 726, 49–56.
Otto, M.W., Pollack, M.H., Dowd, S.M., Hofmann, S.G., Pearlson, G., Szuhany, K.L.,
Gueorguieva, R., Krystal, J.H., Simon, N.M., Tolin, D.F., 2016. Randomized trial of d-
cycloserine enhancement of cognitive-behavioral therapy for panic disorder. Depress.
Anxiety 33, 737–745.
Pittig, A., van den Berg, L., Vervliet, B., 2016. The key role of extinction learning in
anxiety disorders: behavioral strategies to enhance exposure-based treatments. Curr.
Opin. Psychiatr. 29, 39–47.
Powers, M.B., Medina, J.L., Burns, S., Kauffman, B.Y., Monfils, M., Asmundson, G.J.,
Diamond, A., McIntyre, C., Smits, J.A., 2012. Exercise augmentation of exposure
therapy for PTSD: rationale and pilot efficacy data. Cognit. Behav. Ther. 44, 314–327.
Rabin, R., de Charro, F., 2001. EQ-5D: a measure of health status from the EuroQol group.
Ann. Med. 33, 337–343.
Siegmund, A., Golfels, F., Finck, C., Halisch, A., Rath, D., Plag, J., Ströhle, A., 2011. D-
cycloserine does not improve but might slightly speed up the outcome of in-vivo
exposure therapy in patients with severe agoraphobia and panic disorder in a ran-
domized double blind clinical trial. J. Psychiatr. Res. 45, 1042–1047.
Siette, J., Reichelt, A.C., Westbrook, R.F., 2014. A bout of voluntary running enhances
context conditioned fear, its extinction, and its reconsolidation. Learn. Mem. 21,
73–81.
Ströhle, A., Graetz, B., Scheel, M., Wittmann, A., Feller, C., Heinz, A., Dimeo, F., 2009.
The acute antipanic and anxiolytic activity of aerobic exercise in patients with panic
disorder and healthy control subjects. J. Psychiatr. Res. 43, 1013–1017.
Stubbs, B., Vancampfort, D., Rosenbaum, S., Firth, J., Cosco, T., Veronese, N., Salum,
G.A., Schuch, F.B., 2017. An examination of the anxiolytic effects of exercise for
people with anxiety and stress-related disorders: a meta-analysis. Psychiatr. Res. 249,
102–108.
Taylor, S., Abramowitz, J.S., McKay, D., 2012. Non-adherence and non-response in the
treatment of anxiety disorders. J. Anxiety Disord. 26, 583–589.
Van Kummer, B.H., Cohen, R.W., 2015. Exercise-induced neuroprotection in the spastic
han wistar rat: the possible role of brain-derived neurotrophic factor. BioMed Res.
Int. 2015, 834543.