Journal of Psychopharmacology
27(2) 135 –145
© The Author(s) 2013
Reprints and permission:
Generalised Anxiety Disorder (GAD) is common, with a US survey
using Diagnostic and Statistical Manual of Mental Disorders
Fourth Edition (DSM-IV) criteria showing a lifetime prevalence as
high as 6% (Kessler et al., 2005). A recent Chinese survey also
using these criteria found a prevalence of 4.1%, with 72% of these
patients also meeting criteria for major depression (Ying et al.,
2010). Treatment can be complex, with many patients failing to
respond to first-line treatments, requiring combination therapies or
switching between treatments (Baldwin et al., 2005). Current treat-
ments can have significant disadvantages, such as slow onset of
action, abuse potential and unpleasant side effects (Davidson et al.,
2010). There is a need for new evidence-based effective treatments,
with a good side effect profile, to improve patient productivity and
quality of life and reduce the burden on health care resources.
The CO2 model of anxiety and panic may be useful in the
development of such treatments. It has been shown to produce
robust subjective and objective effects, some of which are similar
to symptoms of GAD or panic, which may provide a useful tool
to model anxiety in healthy volunteers for the purposes of anxiety
research and new drug development (Bailey, 2003, 2005, 2007,
2009; Papadopoulos et al., 2010). These CO2-induced effects
can be attenuated by selective serotonin reuptake inhibitors
(SSRIs), benzodiazepines and cognitive behaviourtherapy in
panic disorder patients (e.g. Bertani, et al., 1997, 2001; Gorman
et al., 1997; Nardi et al., 2000; Perna et al., 1997; Polset al., 1996;
Sanderson et al., 1994; Schruers and Griez, 2004). Bailey and
Nutt (2008) have postulated that a mechanism by which CO2
inhalation can elicit anxiety responses is by reducing the amount
of available gamma-amino butyric acid (GABA),both centrally
and peripherally. It has been shown that GABA dysfunction may
Evaluation of the effects of venlafaxine
and pregabalin on the carbon dioxide
inhalation models of Generalised
Anxiety Disorder and panic
Alison Diaper1, Victoria Osman-Hicks1, Ann S Rich1, Kevin Craig2,
Colin T Dourish2, Gerard R Dawson2, David J Nutt3 and
Jayne E Bailey4
Previous studies have shown that subjective and objective symptoms of anxiety induced by 7.5% CO2 inhalation can be attenuated by anxiolytics such
as lorazepam and, to a lesser extent, paroxetine. Venlafaxine and pregabalin, two other licensed treatments for Generalised Anxiety Disorder, were
used to further investigate the 7.5% and 35% CO2 models of anxiety in healthy volunteers. Fifty-four participants were randomised to receive either
placebo, venlafaxine or pregabalin. Study treatments were dosed incrementally over a three week period, to reach daily doses of 150mg venlafaxine
and 200mg pregabalin by the CO2 challenge test day. Participants inhaled air 7.5% CO2 for 20 minutes (single-blind presentation), and a non-blinded
single vital capacity of 35% CO2. Subjective ratings were recorded before and after each inhalation. Both 7.5% and 35% CO2 inhalations produced
the expected effects of increased ratings of symptoms of panic and anxiety, with increased blood pressure and heart rate. No significant treatment
effects were found, although there were trends towards a reduction in feeling tense and nervous by both drugs compared with placebo during the
7.5% CO2 challenge, and a reduction in alertness generally in the venlafaxine group compared with the pregabalin group. In contrast with the clear
anxiolytic effects of benzodiazepines reported in several previous CO2 studies, these findings suggest that the anxiogenic effects of CO2 challenges are
not significantly influenced by these serotonergic and GABAergic anxiolytics. This may be due to a lack of sensitivity of the CO2 challenges in healthy
volunteers to these drug types.
Anxiety, carbon dioxide, experimental human model, panic, pregabalin, venlafaxine
1 Psychopharmacology Unit, University of Bristol, Bristol, UK
2 P1vital Ltd, Department of Psychiatry, University of Oxford, Warneford
Hospital, Oxford, UK
3 Neuropsychopharmacology Unit, Division of Experimental Medicine,
Imperial College London, London, UK
4 Severnside Alliance for Translational Research, University of Bristol,
School of Medical Sciences, Bristol, UK
Alison Diaper, Psychopharmacology Unit, University of Bristol, Dorothy
Hodgkin Building, Whitson Street, Bristol, BS1 3NY, UK.
443742 JOP27210.1177/0269881112443742Diaper et al.Journal of Psychopharmacology
Journal of Psychopharmacology 27(2)
play a role in various anxiety disorders. GABAA receptors are
targets for barbiturates, neurosteroids and alcohol, as well as ben-
zodiazepines. In GAD, there is evidence of reduced benzodiaze-
pine receptor function in the GABA system (Tiihonen et al.,
1997), and drugs acting on the GABA system have been shown to
be effective in treating GAD (Pohl et al., 2005).
The effect and efficacy of psychotropic medication has been
examined by administering the 7.5% and 35% CO2 challenges in
healthy volunteers and patient groups with varying results. It has
been hypothesised that different anxiolytic and antidepressant
medication might have drug specific effects on CO2 inhalations.
In this study, we present data regarding the effects of two anxio-
lytics, to investigate the suitability of these CO2 challenges as
models of anxiety and panic, and perhaps in turn elucidate further
the mechanisms of these CO2 challenges. The first anxiolytic is
pregabalin, which binds to the alpha-2-delta subunit of a voltage-
gated calcium channel and inhibits neurotransmitter release (Gee
et al., 1996), and the second is venlafaxine, a serotonin noradrena-
line reuptake inhibitor (SNRI) that binds with high affinity to
monoamine transporter sites and thereby increases the synaptic
levels of serotonin and noradrenaline (and to a lesser extent dopa-
mine) by preventing their reuptake (Holliday and Benfield, 1995).
Pregabalin is prescribed for epilepsy and neuropathic pain as
well as GAD, and venlafaxine is licensed for depression and several
anxiety disorders including GAD and panic disorder. Both venlafax-
ine and pregabalin have been shown to be effective treatments in
GAD, and venlafaxine for panic disorder and social anxiety disorder
(Bandelow et al., 2007; Kim et al., 2006; Liebowitz et al., 2005,
2009; Pohl et al., 2005; Pollack et al., 2007). A study by Montgomery
et al. (2006) compared pregabalin and venlafaxine treatment for six
weeks in a study of GAD patients and found similar efficacy for both
drugs, with pregabalin achieving effectiveness after one week, a
week before venlafaxine. Kelsey (2000) has also shown that venla-
faxine requires administration for a couple of weeks before it is more
effective than placebo, and Lydiard et al. (2010) found that reduc-
tions in HAM-A scores are observed following 300–600 mg/day
pregabalin after one week of dosing. This suggests that venlafaxine
and pregabalin, unlike benzodiazepine agonists such as lorazepam,
are not immediately active and that dosing needs to be continuous
for a period of time before they demonstrate clinical efficacy.
The study protocol was approved by the Cambridgeshire 2
Research Ethics Committee and local NHS Trusts, and performed
in accordance with ICH Good Clinical Practice. All participants
gave written informed consent after receiving a complete descrip-
tion of the study prior to their participation.
A power calculation of peak 7.5% CO2 effects (absolute values)
based on the mean Visual Analogue Scale ratings of anxiety of
31.2 (placebo) and 17.9 (lorazepam, 2mg), with a common stand-
ard deviation of 12 (Bailey et al., 2007), suggested that to detect a
significant difference at the p≤0.05 level, 13 participants were
required for this study. This number was increased to 18 in each
arm to allow for variation in parameters between experiments and
to balance the experimental design. As there were three arms, 54
participants were required in total.
Participants were recruited using existing databases at the
Psychopharmacology Unit, University of Bristol, and via adver-
tisements on the University careers website and on campus.
At screening, all participants passed strict inclusion and exclu-
sion criteria before participation. Prior to inclusion, participants
were given a physical examination by a study physician, including
an electrocardiogram (ECG) and vital signs, and medical histories
were taken. Participants were interviewed by a study physician to
ensure no history of significant mental disorder or first degree rela-
tives with severe anxiety or panic disorder. Other exclusion criteria
were: current or history of drug or alcohol abuse or dependence,
smoking more than five cigarettes per day, current or history of
cardiovascular, respiratory or renal disease, hypertension, migraine
or epilepsy, and participation in another CO2 study within the last
six months. An alcohol breath test was performed, and a urine sam-
ple was tested to screen for drugs of abuse, pregnancy in females
and for urinalysis. A blood sample was taken to screen for general
health problems. Medication was not allowed for two weeks prior
to and during testing, unless deemed by the investigator not to
interfere with the study or compromise safety. No alcohol or illicit
drugs were allowed for the duration of the whole study. Participants’
family doctors were informed of their participation and partici-
pants were compensated £400 for their time.
Study design and procedure
This was a randomised, double-blind, placebo-controlled, parallel
design study. Baseline measures of trait anxiety and personality
were taken using the Anxiety Sensitivity Index (ASI) (Reiss and
McNally, 1985), Spielberger State Trait Inventory –Trait (STAI)
(Spielberger, 1983), the Eysenck Personality Questionnaire –
Revised (EPQ-R) (Eysenck et al., 1985) and Cloninger’s
Temperament and Character Inventory-125 (TCI) (Cloninger et al.,
1994). Participants were randomised to receive venlafaxine, pre-
gabalin or placebo over a 26 day period.
After randomisation, participants were asked to keep a diary of
adverse events and dose times, and attended the research facility
(the Psychopharmacology Unit clinical testing rooms at the
Bristol Royal Infirmary) weekly for a brief health and compliance
check (breath alcohol, urine drugs of abuse screen, temperature
and blood pressure). After three weeks of dose-increasing treat-
ment, participants underwent the CO2 challenge procedure. This
consisted of the inhalation of air for 20 minutes, 10 minutes’ rest,
the inhalation of 7.5% CO2 for 20 minutes, 10 minutes’ rest, inha-
lation of 35% CO2 for one breath (vital capacity inhalation) and
30 minutes’ rest. During the challenge, subjective and physiologi-
cal measures were collected. For five days after the CO2 test day,
dose levels were decreased and participants attended a follow-up
health check once withdrawn from the drug.
The study treatments were venlafaxine, pregabalin and placebo.
The active treatments were over-encapsulated to give the appear-
ance of the red placebo capsule. Treatments were dose-increased
Diaper et al. 137
to avoid adverse events over a three week period, to reach doses
of 150mg venlafaxine and 200mg pregabalin by day 21 (CO2 test
day). These doses were chosen to be within therapeutic range, but
it was considered more important that the drugs had minimal side
effects in this healthy volunteer sample, rather than being exactly
equipotent. Studies of GAD patients by Davidson et al. (1999) and
Pohl et al. (2005) have shown that these doses reduce HAM-A
scores compared with placebo. British National Formulary rec-
ommended doses for anxiety range between 75mg and 225mg of
venlafaxine and 150mg and 600mg daily of pregabalin.
The dose of venlafaxine was 75mg on days 0–2, 112.5mg on
days 3–6, 150mg on days 7–21, 75mg on days 22–24 and 37.5mg
on days 25–26. The dose of pregabalin was 100mg on days 0–6,
200mg on days 7–21, 100mg on days 22–24 and 50mg on days
25–26. The last administration of the maximum dose of each drug
was on the day of the CO2 challenge. Doses were taken as two
capsules daily. The study drugs were provided in blister packs and
participants were instructed to take one capsule every morning
and one capsule every evening with food. Capsules were taken in
the order set out on the blister pack (day 1 morning, day 1 even-
ing, day 2 morning, day 2 evening, etc.). The drugs were labelled
according to the point in the study when they were taken.
The first doses of study medication, and the doses on the CO2
test day (day 21), were administered on-site under direct supervi-
sion of two researchers. No other doses were witnessed, as partici-
pants were given the study medication to take home and
self-administer, noting the times of doses and any side effects in a
diary. Participants were required to bring their diaries and full or
empty drug packaging with them at each visit for a compliance
check that all medication had been taken, or to note when partici-
pants had forgotten to take their medication. If participants forgot
to take more than two capsules in one week, they would have been
withdrawn from the study for non-compliance.
Delivery of gas
The delivery of gas followed procedures previously reported by
Bailey et al. (2009) and Papadopoulos et al. (2010). Participants
were seated comfortably during inhalations and at least two inves-
tigators attended the sessions, with one investigator remaining in
sight of the participant throughout the procedure.
For the 20 minute inhalations, the gas mixtures used were an
inhalation of CO2 7.5%/O2 21%/N 71.5% and an inhalation of
medical air, delivered via a nasal–oral exercise facemask (Hans
Rudolf, Kansas) attached to a 500 L Douglas bag via tubing. Gas
flow was monitored to allow for a reservoir of gas in the bag at all
times. Gas cylinders were kept out of view in a separate room to
blind the participant as to the order of gas presentation, which was
single blind. Air was always presented first but participants were
informed that presentation was random to avoid expectation
effects and differing levels of anticipatory anxiety.
The gas mixture used for the vital capacity single breath inha-
lation was CO2 35%/O2 65%. Full instructions were given before
participants underwent this non-blinded inhalation. First, partici-
pants were asked to insert a mouthpiece attached to a small
Douglas bag. A nose clip was used to ensure participants only
breathed the contents of the bag. After expiring fully, a full vital
capacity inhalation was taken and held for a count of four seconds.
The equipment was then removed and participants were asked to
Participants scored Visual Analogue Scales (VASs) verbally on a
scale of 0 (not at all) to 100 (the most ever) using the adjectives:
alert, fearful, relaxed, anxious, happy, feel like leaving the room,
stressed, tense, nervous, irritable and worried.
The Panic Symptom Inventory (PSI) was used to rate panic
anxiety and the associated symptoms of autonomic arousal, with
the option of rating 0 = not at all, 1 = slight, 2 = moderate, 3=
severe, 4 = very severe. The PSI was adapted from Clark and
Hemsley (1982) and lists 34 items. It has been used in studies of
panic provocation (Bell et al., 2002; Nutt et al., 1990) and previ-
ous CO2 studies (Argyropoulos et al., 2002; Bailey et al., 2005,
2007). The Generalised Anxiety Disorder Criteria Inventory
(GAD-C) was used to measure participants’ current anxiety state.
PSI, GAD-C and VAS ratings were performed 30 minutes
(baseline) and 15 minutes before inhalations and at the end of the
session (30 minutes after the last (35% CO2) inhalation).
Immediately after each inhalation, participants were asked to rate
how they felt when the effects of the gas were at their greatest, this
rating being ‘peak’.
The Spielberger State Trait Anxiety Inventory – State (SSAI)
(Spielberger, 1983) was used to measure state anxiety and was
administered 10 minutes prior to each inhalation and at the end of
the testing session.
Continuous measurements of blood pressure and heart rate were
obtained during all inhalations using the Finapres (Ohmeda,
Englewood, CO). For full details see Coupland et al. (1995), but
briefly, the participants wore a finger cuff with a photosensitive
cell connected via a servo-controlled pump, which inflated the cuff
to maintain a constant pressure on the finger. To avoid artefact, the
participant was instructed to stay still and not to move their hand
or cross their legs. Data were captured onto a DOS-based program
and later analysed using software designed for the purpose.
Comparisons of differences between baseline (at the beginning of
the test session before sight of the inhalation equipment) and peak
gas effects (peak air minus baseline versus peak 7.5% CO2 minus
baseline versus peak 35% CO2 minus baseline) were made for
subjective variables. Also, direct comparisons between peak 7.5%
and peak 35% CO2 subjective ratings were made. Blood pressure
and heart rate were averaged over the air and 7.5% CO2 inhala-
tions and compared. No baseline cardiovascular measures were
taken using the Finapres. For the 35% CO2 inhalation, blood pres-
sure and heart rate 30 seconds before inhalation, at the point of
inhalation and 30 seconds after inhalation were compared.
Physiological measures during the 35% CO2 inhalation for two
participants contained too many artefacts to be included in the
analysis, so here 17 participants were included in the analysis for
the venlafaxine and pregabalin groups. Subjective and objective
variables were analysed using mixed model and repeated meas-
ures ANOVAs respectively for Drug, Gas and Drug*Gas interac-
tions using the Greenhouse-Geisser correction. Although not all
VAS variables were normally distributed, it is assumed that the
test is robust enough to withstand such violations. Post hoc
Journal of Psychopharmacology 27(2)
analyses were simple main effects and pairwise comparisons with
Bonferroni correction unless otherwise stated. All data were ana-
lysed using SPSS Version 16.0 for Windows.
We were contacted by 282 volunteers, of which 115 underwent
an initial brief telephone questionnaire. Seventy-nine of these
volunteers were screened, and 60 were suitable for study inclu-
sion and randomised. Six participants withdrew from the study
because of adverse events; five were in the venlafaxine group and
one was in the placebo group. These participants were replaced to
make a total of 54 completing participants, 29 male and 25
female, aged between 20 and 43 years of age (mean age 23.1,
standard deviation (SD) 4.68). Each treatment group consisted of
Chi-square tests showed that there was no significant differ-
ence between numbers of males and females in each treatment
group, and no significant difference between numbers of each eth-
nic origin in each treatment group. One way ANOVAs showed
there was no significant difference between the treatment groups
with respect to age, usual caffeine and alcohol intake, usual
amounts of smoking, body mass index and scores on the ASI
(Reiss and McNally, 1985), STAI (Spielberger, 1983) and all sub-
scales of the EPQ-R (Eysenck et al., 1985) and TCI-125 (Cloninger
et al., 1994, see Table 1). There was also no significant difference
between compliance (the numbers of capsules missed) by partici-
pants in each treatment group (an average throughout the study of
0.9 capsules in the placebo group, 0.8 in the venlafaxine group
and 0.7 in the pregabalin group). Capsules were missed because
participants forgot to take them.
A two-way mixed model ANOVA of the SSAI showed anxiety
levels returned to baseline (pre-air) before each inhalation and at
the end of the test session, with the exception of prior to the 35%
CO2 inhalation, where anxiety levels were significantly higher than
at all other time points (F(3,54)= 22.19, p<0.001; Bonferroni pair-
wise comparisons: pre-air versus pre-7.5% CO2, p=0.135; pre-
7.5% versus pre-35% CO2, p<0.001; pre-air versus pre-35% CO2,
p<0.001). This is likely to be anticipatory anxiety as the partici-
pants knew they were about to receive an inhalation of a high CO2
concentration. There was no significant difference in SSAI
between drug groups at this time point. Scores on average were 31
at baseline (SD 6.9), 32 before 7.5% CO2 (SD 7.1), 36 before 35%
CO2 (SD 9.1) and 31 at the end of the session (SD 8.3). For com-
parison, average state anxiety score for males and females aged 19
to 39 years is 36 (SD 10.6) (Spielberger, 1983).
Accounting for baseline and compared with air, the inhalation
of 7.5% CO2 significantly increased the total PSI scores
(F(1,51)=62.4, p<0.001). Compared with baseline, the inhalation
of 35% CO2 also significantly increased the total PSI scores
(F(1,51)=149.5, p<0.001). There was a trend towards a Drug*Gas
interaction (F(2,51)=2.4, p=0.097), whereby scores in the prega-
balin group were lower than the venlafaxine group at peak 35%
CO2 effects. Accounting for baseline and compared with air, 7.5%
CO2 inhalation significantly increased GAD Criteria Inventory
(GAD-C) scores (F(1,51)=40.7, p<0.001), and compared with
baseline, inhalation of 35% CO2 also significantly increased
GAD-C scores (F(1,51)=105.6, p<0.001), but there was no sig-
nificant difference between scores for the two CO2 inhalations
overall. However, by treatment group, GAD-C scores on venla-
faxine were significantly higher at peak 35% compared with peak
7.5% (t(17)=2.14, p<0.05). There were no significant differences
between treatment (see Figures 1 and 2).
Table 1. Participant demographics and characteristics at baseline.
MeasurePlaceboVenlafaxinePregabalinMatched population norm
9 M, 9 F
12 M, 6 F
8 M, 10 F
ASI: Anxiety Sensitivity Index; STAI: Spielberger State–Trait Anxiety Inventory; EPQ-R: Eysenck Personality Questionnaire – Revised; TCI-125: Cloninger’s Temperament and
Character Inventory (125 items). Normal values taken from Peterson and Reiss (1992; ASI); Spielberger (1983; STAI); Eysenck and Eysenck (1991; EPQ-R) and Cloninger
et al. (1991; TCI-125). Table shows mean scores with standard deviations in parentheses.
Diaper et al. 139
Dividing the PSI into somatic (26 symptoms, e.g. sweating,
heart racing) and psychic (nine symptoms, e.g. going mad, appre-
hension) symptoms, and taking account of the difference in the
number of symptoms queried, paired t-tests showed no significant
differences between the CO2 inhalations and air on venlafaxine or
pregabalin. However, on placebo, there was a non-significant
trend towards more somatic reports of anxiety than psychic during
the 7.5% CO2 inhalation (t(17)=1.786, p=0.092), but not the 35%
CO2 inhalation (see Figure 3).
Visual Analogue Scales
The 7.5% CO2 inhalation (accounting for baseline scores and
compared with air inhalation) and the 35% CO2 inhalation (com-
pared with baseline) significantly increased ratings at peak of feel-
ing fearful, anxious, like leaving the room, stressed, tense,
nervous, irritable and worried, and decreased ratings of feeling
relaxed and happy (see Table 2).
There were no significant main effects of drug for any VAS
measure, probably because the variances were large. However,
there were significant Gas*Drug interactions for the ratings of
feeling tense (F(2,51)=3.9, p<0.05) and nervous (F(2,51)=3.4,
p<0.05) peak 7.5% CO2 compared with peak air (accounting for
baseline). One way ANOVAs showed trends towards feeling less
tense during this inhalation (F(2,51)=2.9, p=0.062) on pregabalin
compared to placebo (Tukey’s, p=0.065), and less nervous
(F(2,51)=2.8, p=0.072) on pregabalin and venlafaxine compared
with placebo (Tukey’s post hoc analysis was not significant;
p=0.120 and p=0.108 respectively). Taking account of baseline,
direct comparison of the 7.5% CO2 and 35% CO2 inhalations by
paired t-tests revealed that participants felt more like leaving the
room (t(53)=3.13, p<0.01) and more irritable (t(53)=2.67, p<0.05)
during the 7.5% CO2 inhalation compared with the 35% CO2
inhalation, indicating it was a less pleasant experience. All means
and standard deviations are shown in Table 3.
Alertness (F(2,51)=4.1, p<0.05) was increased, compared with
air, during the 7.5% CO2 inhalation but not the 35% CO2 inhala-
tion. During the 7.5% CO2 inhalation, alertness was significantly
reduced by venlafaxine compared with pregabalin (pairwise com-
parisons, p<0.05) and there was a trend towards reduced alertness
by venlafaxine compared to placebo (pairwise comparisons,
p=0.076). There was a trend towards reduced alertness generally
(F(2,54)=2.86, p=0.067) in the venlafaxine group compared with
the pregabalin group (p=0.094) (see Figures 4 and 5).
Physiological measures taken during weekly health checks and at
the beginning of the CO2 challenge day showed blood pressure
and heart rate were significantly higher in the venlafaxine group
compared with pregabalin and placebo (data not shown).
Compared with air, the 7.5% CO2 inhalation significantly
increased systolic blood pressure (SBP) (F(1,54)=67.89, p<0.001),
diastolic blood pressure (DBP) (F(1,54)=40.70, p<0.001) and
heart rate (HR) (F(1,54)=30.73, p<0.001). Significant effects of
drug were also found for SBP (F(2,54)=13.82, p<0.001), DBP
(F(2,54)=11.43, p<0.001) and HR (F(2,54)=6.07, p<0.01) in that
SBP, DBP and HR were significantly higher in the venlafaxine
group compared with the pregabalin and placebo groups (all
p values <0.05). There were no Drug*Gas interactions.
PEAK Air PEAK 7.5% PEAK 35% End
PSI total score
Figure 1. Graph shows the total Panic Symptom Inventory (PSI) score
over time in the three drug conditions.
PEAK Air PEAK 7.5% PEAK 35%End
GAD-C total score
Figure 2. Graph shows the total Generalised Anxiety Disorder Criteria
Inventory (GAD-C) scores over time in the three drug conditions.
PSI Score adjusted for number of symptoms
Figure 3. Graph shows the scores for Panic Symptom Inventory
(PSI) Somatic and Psychic symptoms at the peak effects of the three
Journal of Psychopharmacology 27(2)
Table 2. Effect of CO2 inhalations on Visual Analogue Scale (VAS) ratings.
Peak 7.5% CO2 - baseline vs.
peak air - baseline
Peak 35% CO2 - baseline vs.
peak air - baseline
F p valueEffect
F p value Effect
Table 3. Table showing the effect of CO2 inhalations on Visual Analogue Scale (VAS) ratings by drug treatment.
Peak air - baselinePeak 7.5% CO2 - baselinePeak 35% CO2 - baseline
Placebo Venlafaxine PregabalinPlaceboVenlafaxine Pregabalin PlaceboVenlafaxine Pregabalin
-18.3 (15.72) -27.2 (19.87)
-12.2 (14.47) -13.3 (13.50)
0.0 (22.82) -19.4 (25.37) -4.7* (27.36)
17.8 (22.57)13.3 (17.24)
-7.8 (24.98) -17.2 (35.74)
14.4 (20.07) 0.8 (8.09)
-2.5 (22.51) -10.8 (21.44) -37.8 (21.71) -27.5 (24.51) -32.8 (28.66) -39.2 (21.64) -41.1 (30.71) -36.7 (31.62)
-3.3 (18.15)0.8 (11.79)22.8 (24.57)9.7 (23.54)
-32.8 (22.83) -25.3 (15.48) -30.0 (21.49) -27.2 (23.21) -33.9 (30.08) -28.6 (24.72)
-0.3 (14.09)18.6 (23.31)29.2 (30.01)
-0.3 (11.56)18.9 (25.64)17.8 (22.31)
-0.8 (7.91)30.3 (22.52)18.3 (15.81)
0.3 (11.31)1.4 (9.20)21.4 (22.87)9.7 (10.91)
-0.3 (11.69)15.0 (19.10)12.2 (21.09)
-1.9 (11.00)15.8 (22.83)7.2 (12.63)
14.7 (18.75)25.6 (20.93)
16.1 (22.66)15.0 (20.22)13.1 (27.34)15.3 (17.19)
Table shows mean scores of the differences between inhalation and baseline with standard deviations in parentheses.
*Pregabalin significantly different from venlafaxine (p<0.05).
PEAK Air PEAK 7.5% PEAK 35%End
Figure 4. Graph shows ratings of Visual Analogue Scales (VAS) Alertness
over time in the three drug conditions.
PEAK Air PEAK 7.5% PEAK 35%End
Figure 5. Graph shows ratings of VAS Anxious over time in the three
drug conditions. Although non-significant, this pattern is also seen in
all negative effect VASs, the reverse seen in positive effect VASs.
Diaper et al. 141
Analysis of the 35% CO2 inhalation showed significant Time
(F(2,52)=13.13, p<0.001) and Drug (F(2,52)=3.47, p<0.05)
effects of SBP, and Time (F(2,52)=9.60, p<0.001) and Drug
(F(2,52)=8.85, p<0.01) effects of HR. SBP and HR were signifi-
cantly higher after inhalation compared with before inhalation and
the point of inhalation (p values <0.01) and significantly higher on
venlafaxine compared with pregabalin (p values <0.05) and also
placebo for HR (p values <0.05).
There were also significant Time (F(2,52)=21.48, p<0.001)
and Drug (F(2,52)=3.32, p<0.05) effects of DBP, in that DBP
was significantly higher after 35% CO2 inhalation compared
with before inhalation and the point of inhalation, and DBP at
the point of inhalation was significantly higher than before
inhalation (p values <0.05). DBP was significantly higher on
venlafaxine compared with pregabalin (p<0.05). There was a
significant Time*Drug interaction (F(4,52)=3.93, p<0.01).
A one-way ANOVA showed a significant difference in DBP
between the drug groups 30 seconds after inhalation
(F(2,52)=5.81, p<0.01). Tukey HSD analysis showed DBP for
the venlafaxine group was significantly higher at this time point
than for the pregabalin and placebo groups (p values <0.05).
This indicates that the increase in DBP associated with the 35%
CO2 inhalation was significantly greater in the venlafaxine
group. See Figure 6(a) and (b).
Air inhalation7.5% CO2 inhalation 30 beats after 35% CO2
Blood pressure (mmHg)
Placebo - SBP
Venlafaxine - SBP
Pregabalin - SBP
Placebo - DBP
Venlafaxine - DBP
Pregabalin - DBP
Air inhalation7.5% CO2 inhalation 30 beats after 35% CO2
Heart rate (beats per minute)
Placebo - HR
Venlafaxine - HR
Pregabalin - HR
Figure 6. Graph shows that systolic (SBP) and diastolic (DBP) blood pressure (a) and heart rate (HR) (b) were significantly elevated in the
venlafaxine group compared with the pregabalin and placebo groups.
Journal of Psychopharmacology 27(2)
Adverse events (AEs) and concomitant
In total, 317 adverse events (AEs) were reported after randomisa-
tion by 48 of the 54 participants. Six participants reported no AEs
at any point during the study. Four of these participants were tak-
ing pregabalin, one was taking placebo and one was taking
There were 79 AEs reported in the placebo group (86% mild
and 14% moderate, as judged by the volunteer according to pre-
defined criteria), 143 reported in the venlafaxine group (81% mild
and 18% moderate) and 95 reported in the pregabalin group (89%
mild and 10% moderate). Pregabalin was better tolerated than
venlafaxine with respect to number and severity of AEs, although
the doses used may not have been comparable. Headache, insom-
nia and somnolence were most commonly reported on placebo;
fatigue, insomnia and headache on venlafaxine, and headache and
dizziness on pregabalin. All AEs were followed up, no adverse
sequelae were reported, and there were no serious AEs.
Thirty-four participants took 57 concomitant medications dur-
ing the study. Thirty-one medications were taken to alleviate
adverse events (e.g. paracetamol, mouth ulcer treatment), two
were local anaesthetic for dental work, nine were for pre-existing
conditions (e.g. vitamin supplements) and 15 were taken for pro-
phylactic reasons (e.g. multivitamins, contraceptives). None were
judged to interfere with the study measures.
This is the first study to investigate the effects of venlafaxine and
pregabalin using an experimental model of GAD and panic anxi-
ety symptoms in healthy volunteers. This was a study of 54
healthy volunteers monitored and assessed over a 25 day period.
The inhalation of 7.5% CO2 for 20 minutes and the inhalation of
35% CO2 as a single breath, vital capacity inhalation robustly
increased subjective measures of panic and anxiety. Both CO2
inhalations increased ratings of feeling fearful, anxious, like leav-
ing the room, stressed, tense, nervous and worried, and decreased
ratings of feeling relaxed and happy. The 7.5% CO2 inhalation
increased ratings of alertness and irritability compared with air.
The 35% CO2 inhalation increased anticipatory state anxiety com-
pared with the other inhalations. It is thought this was due to the
anticipation of the higher concentration of CO2. However, as the
presentation of air and 7.5% CO2 inhalations was not counterbal-
anced, a hangover of the 7.5% CO2 inhalation effects cannot be
ruled out. A limitation of this study is a lack of baseline measures
taken before each inhalation; however, another study using the
same procedure performed by our group found Visual Analogue
Scales of anxiety returned to baseline levels within 10 minutes
after the 7.5% CO2 inhalation (Bailey et al., 2009). Physiological
analysis of the CO2 inhalations showed both concentrations increased
systolic and diastolic blood pressure, and heart rate. These find-
ings replicate our previous CO2 inhalation studies (Bailey, 2003,
2005, 2007, 2009; Papadopoulos et al., 2010; Seddon et al., 2010).
No significant effects of drug treatment on CO2 responses
were found. There were trends towards reduced alertness gener-
ally in the venlafaxine group compared with the pregabalin group,
and PSI scores of panic were lower in the pregabalin group at the
peak of the 35% CO2 rating compared with the venlafaxine group.
Ratings of feeling tense and nervous were non-significantly lower
in the venlafaxine and pregabalin groups compared with placebo at
the peak of the 7.5% CO2 inhalation. Systolic and diastolic blood
pressure and heart rate were significantly higher in the venlafaxine
group compared with both the pregabalin and placebo group at
weekly health checks, and on the CO2 challenge days regardless of
inhalation, which is probably due to the noradrenergic effects of
venlafaxine. In a study by Hood et al. (2010) of a clonidine chal-
lenge in 10 untreated GAD patients, seven venlafaxine treated
GAD patients and seven controls showed no effect of venlafaxine
treatment on clonidine sensitivity. The authors concluded that the
noradrenergic effects of venlafaxine were not pivotal in reducing
anxiety, which may explain why, in our study, there was no signifi-
cant change in psychological response to venlafaxine in line with
the observed change in physiological response.
Based on previous studies, Bailey and Nutt (2008) propose
that CO2 inhalation can elicit anxiety responses by reducing the
amount of available GABA, and that this may occur alongside
noradrenergic and serotonergic activation. A single dose of loraz-
epam (a benzodiazepine agonist) significantly reduced CO2-
induced feelings of fear, feeling like leaving the room, tension and
worry (Bailey et al., 2007). Twenty-one days’ dosing of paroxe-
tine (a SSRI), however, only significantly reduced CO2-induced
feelings of nervousness (Bailey et al., 2007). Reasons suggested
for this were the small number of participants, and perhaps insen-
sitive scales. Because of this, the current study included 18 par-
ticipants in each group, and also used a scale based on the DSM-IV
and the International Classification of Diseases version 10 (ICD-10)
criteria for GAD-C, but still drug attenuation of the 7.5% and 35%
CO2 challenge did not reach significance. Another reason for a
lack of significant effect may have been inter-individual variabil-
ity. Although balanced for demographic characteristics, other fac-
tors, such as timing of cigarettes smoked, may affect response to
the CO2 challenge. Attwood et al. (2009) found non-abstinent
smokers reported reduced anxiety in the 7.5% CO2 challenge
compared with abstinent smokers, with non-smokers positioned
in between. All smokers were restricted from smoking for 3.5
hours before inhalations, but only 1.5 hours of this time was under
observation in the research facility. Also, the anxiety sensitivity
and state anxiety of participants on test days were slightly lower
than normal values would suggest, which may have biased
responses. Notably, participant scores for extraversion were
higher and scores for neuroticism and harm avoidance were lower
than normal values, which is to be expected in individuals who
volunteer for medical trials and in line with our previous studies
(Bailey and Nutt, 2001, 2005). Compliance was thought to be
good, but most doses were not witnessed. However, physiological
measures taken during weekly health checks showed blood pres-
sure and heart rate were significantly higher in the venlafaxine
group compared with pregabalin and placebo (data not shown),
indicating general compliance.
A study by Papadopoulos et al. (2010) showed no significant
subjective effects of acute doses of the anxiolytic drugs proprano-
lol, hydroxyzine or flupentixol on the 7.5% CO2 challenge when
compared with placebo. The authors suggest this may have been
due to propranolol being more effective for somatic symptoms
rather than for psychological symptoms (File and Lister, 1985)
and the doses of hydroxyzine (an antihistamine; H1 receptor
agonist) and flupentixol (an antipsychotic; dopamine D1 and D2
receptor antagonist) being sub-clinical (to minimise side effects).
However, a recent study by Bailey et al. (2011) found seven days’
Diaper et al. 143
dosing of a corticotropin-releasing factor antagonist (R317573)
significantly reduced subjective panic and anxiety measures dur-
ing the 7.5% CO2 inhalation compared with placebo, but not quite
to the extent of a single dose of lorazepam. Comparing different
anxiolytics with fast and slower onset of action is difficult, as
there may not be an equivalence of doses.
Rather than a difference in effect between acute or chronic
dosing, an alternative postulation would be that a significant influ-
ence on the CO2 challenge may only be elicited by benzodiaz-
epines or similarly potent anxiolytics acting on the GABAA
receptor. Indeed, several patient studies (but not all) have shown
that a single dose of alprazolam (Pols et al., 1996; Sanderson
et al., 1994) and clonazepam (Nardi et al., 2000; Valença et al.,
2000) can attenuate the anxiety response to 35% CO2. Valença
et al. (2002) have also demonstrated this effect after six weeks
of treatment with clonazepam. A study by Zwanzger et al. (2003)
showed that a single dose of alprazolam can attenuate CCK-4 pro-
voked panic in healthy volunteers. A recent study by Bailey et al.
(2009) of the acute effects of zolpidem (5 mg), alprazolam (1 mg)
and placebo showed alprazolam (a benzodiazepine) significantly
reduced ratings of fear, panic and generalised anxiety compared
with placebo during the 7.5% CO2 challenge, and reduced panic
and generalised anxiety scores during the 35% CO2 challenge.
Zolpidem (an agonist with preference for alpha-1 subtype binding
in the GABAA receptor) significantly reduced feeling like leaving
the room, and feeling worried and stressed compared with placebo
during the 7.5% CO2 challenge.
Although venlafaxine and pregabalin at 150mg and 200mg
respectively have been shown to have anxiolytic properties in
GAD patients (Davidson et al., 1999; Pohl et al., 2005), it appears
that they may not provide a sufficiently potent anxiolytic effect for
the CO2 challenge model, or the CO2 models were not mechanisti-
cally similar enough to GAD. Either way, 21 days’ tapering up to
these clinically effective doses was not shown to consistently
attenuate the anxiogenic effects of hypercapnic gas, although
trends towards attenuation were noted. It is possible that the doses
of the drugs used in this study were not high enough to elicit sig-
nificant effects on the CO2 challenges. This is unlikely with respect
to venlafaxine as it is licensed for anxiety at the dose range of
75mg–225mg daily, but may be the case for pregabalin, given that
up to 600mg daily is permitted. We did not progress to this dose on
account of tolerability concerns, although no participants withdrew
from the pregabalin group and this appears to have been ground-
less. A limitation to this study was the lack of use of a benzodiaz-
epine for direct comparison, which may have resolved some issues
such as equivalent dosing.
One consideration is that of differences between the modes of
action of the anxiolytics that the CO2 challenge may be exposing.
Benzodiazepines have been shown in most (but not all) GAD
patient studies to be more efficacious at alleviating somatic anxi-
ety rather than psychic anxiety (e.g. Rickels et al., 1988, 1993). In
contrast, pregabalin (after 4–6 weeks (Lydiard et al., 2010)) and
venlafaxine (after eight weeks (Stahl et al., 2007); eight weeks
and six months (Meoni et al., 2004)) have been shown to be more
efficacious at alleviating psychic than somatic anxiety in patients
with GAD. The PSI includes both somatic and psychic questions,
and analysis of these subdivisions showed that those on placebo
reported a trend towards a larger increase in somatic symptoms
compared with psychic symptoms when inhaling 7.5% CO2. This
may be an important distinction when developing novel anxiolyt-
ics, as somatic symptoms can often be the primary complaint of
anxious patients (Meoni et al., 2004). The patients’ perception of
the improvement in the somatic symptoms of anxiety can provide
a considerable contribution to their perception of treatment effi-
cacy, so much so that Meoni et al. (2004) suggest the improve-
ment of somatic anxiety should be regarded as an essential
requirement of any anxiolytic medication.
To conclude, this study showed that venlafaxine and pregaba-
lin had no significant effect on either CO2 challenge model in
healthy volunteers. This may have been due to some classes of
anxiolytics, depending on their mode of action or speed of onset,
mediating these effects more than others. Further research is
required into the emerging distinctions between anxiolytics found
when administered for CO2-induced anxiety, or the use of other
techniques, such as cognitive tasks, to separate the expressions of
anxiety elicited during the CO2 challenges.
Thanks to the staff, students and study medics of the Psychopharmacology
Unit, University of Bristol, for their kind assistance during the course of
the study, and to the study participants.
This work was supported by the P1vital CNS Experimental Medicine
Consortium (members AstraZeneca, GlaxoSmithKline, Lundbeck, Organon
(a subsidiary of Merck) and Pfizer), reference P1V-ANX-CT01-07.
Conflict of interest
AD, VO-H, ASR and JEB have no conflicts of interest to declare. GRD,
CTD and KC are directors and shareholders in P1vital Limited. DJN has
provided consultancy services to Pfizer, GSK, Novartis, Organon,
Cypress, Lilly, Janssen, Lundbeck, BMS, Astra Zeneca, Servier, Hythiam
and Sepracor, received honoraria from Wyeth, Reckitt-Benkiser and
Cephalon, received grants or clinical trials payments from MSD, GSK,
Novartis, Servier, Janssen, Lundbeck, Pfizer, Wyeth and Organon and
holds shares in GSK.
Argyropoulos SV, Bailey JE, Hood SO, et al. (2002) Inhalation of
35% CO2 results in activation of the HPA axis in healthy volunteers.
Psychoneuroendocrinology 27: 715–729.
Attwood AS, Saeed F, Bailey J, et al. Effects of carbon dioxide challenge
on anxiety in cigarette smokers. Paper presented at the Society for
Research on Nicotine and Tobacco, Dublin, April 2009.
Bailey JE, Argyropoulos SV, Lightman SL and Nutt DJ (2003) Does
the brain noradrenaline network mediate the effects of the CO2
challenge? J Psychopharmacol 17(3): 252–259.
Bailey JE and Nutt DJ (2001) How normal is normal? Personality char-
acteristics of volunteers participating in psychopharmacological
research studies. J Psychopharmacol 15: A63, K1.
Bailey JE and Nutt DJ (2005) Are personality factors related to the CO2
response in healthy volunteers? J Psychopharmacol 19: A18.
Bailey JE and Nutt DJ (2008) GABA-A receptors and the response to CO2
inhalation – a translational trans-species model of anxiety? Pharma-
col Biochem Behav 90: 51–57.
Bailey JE, Argyropoulos SV, Kendrick A, et al. (2005) Behavioral and
cardiovascular effects of 7.5% CO2 in human volunteers. Depress
Anxiety 21: 18–25.
Bailey JE, Kendrick A, Diaper A, et al. (2007) A validation of the 7.5%
CO2 model of GAD using paroxetine and lorazepam in healthy volun-
teers. J Psychopharmacol 21: 42–49.
Bailey JE, Papadopoulos A, Seddon K, et al. (2009) A comparison of
the effects of a subtype selective and non-selective benzodiazepine
Journal of Psychopharmacology 27(2)
receptor agonist in two CO2 models of experimental human anxiety. J
Psychopharmacol 23: 117–122.
Bailey JE, Papadopoulos A, Diaper A, et al. (2011) Preliminary evidence
of anxiolytic efficacy of the CRF1 receptor antagonist R317573 in the
7.5% CO2 proof-of-concept experimental model of human anxiety.
J Psychopharmacol 25(9): 1199–1206.
Baldwin DS, Anderson IM, Nutt DJ, et al. (2005) Evidence-based guide-
lines for the pharmacological treatment of anxiety disorders: recom-
mendations from the British Association for Psychopharmacology.
J Psychopharmacol 19: 567–596.
Bandelow B, Wedekind D and Leon T. (2007) Pregabalin for the treat-
ment of generalized anxiety disorder: a novel pharmacologic inter-
vention. Exp Rev Neurother 7: 769–781.
Bell C, Forshall S, Adrover M, et al. (2002) Does 5-HT restrain panic?
A tryptophan depletion study in panic disorder patients recovered on
paroxetine. J Psychopharmacol 16: 5–14.
Bertani A, Perna G, Arancio C, et al. (1997) Pharmacologic effect of imip-
ramine, paroxetine, and sertraline on 35% carbon dioxide hypersen-
sitivity in panic patients: a doubleblind, random, placebo-controlled
study. J Clin Pharmacol 17: 97–101.
Bertani A, Caldirola D, Bussi R, et al. (2001) The 35% CO2 hyperreactiv-
ity and clinical symptomology in patients with panic disorder after 1
week of treatment with citalopram: an open study. J Clin Psychophar-
macol 21: 262–267.
Clark OM and Hemsley DR (1982) The effects of hyperventilation; indi-
vidual variability and its relation to personality. J Behav Ther Exp
Psychiatry 13: 41–47.
Cloninger CR, Przybeck TR and Svrakic DM. (1991) The tridimensional
personality questionnaire: U.S. normative data. Psychol Rep 69:
Cloninger CR, Pryzbeck TR, Svrakic DM and Wetzel RD (1994) The
Temperament and Character Inventory (TCI): a guide to its develop-
ment and use. St. Louis, MO : Center for Psychobiology of Personal-
ity, Washington University.
Coupland NJ, Bailey JE, Wilson SJ, et al. (1995) The effects of clonidine
on cardiovascular responses to standing in healthy volunteers. Clin
Autonom Res 5: 171–177.
Davidson JRT, DuPont RL, Hedges D, et al. (1999) Efficacy, safety,
and tolerability of venlafaxine extended release and buspirone in
outpatients with generalized anxiety disorder. J Clin Psychiatry 60:
Davidson JRT, Feltner DE and Dugar A (2010) Management of General-
ized Anxiety Disorder in primary care: identifying the challenges and
unmet needs. J Clin Psychiatry 12: e1–e13.
Eysenck HJ and Eysenck SBG (1991) Manual of the Eysenck Personality
Scales (EPS Adult). London: Hodder and Stoughton.
Eysenck SBG, Eysenck HJ and Barrett P (1985) A revised version of the
psychoticism scale. Person Indiv Diff 6: 21–29.
File SE and Lister RG (1985) A comparison of the effects of lorazepam
with those of propranolol on experimentally-induced anxiety and per-
formance. Br J Clin Pharmacol 19: 445–451.
Gee NS, Brown JP, Dissanayake VU, et al. (1996) The novel anticonvul-
sant drug, gabapentin (neurontin), binds to the alpha2delta subunit of
a calcium channel. J Biol Chem 271: 5768–5776.
Gorman JM, Browne ST, Papp LA, et al. (1997) Effect of antipanic treat-
ment on response to carbon dioxide. Biol Psychiatry 42: 982–991.
Holliday SM and Benfield P (1995) Venlafaxine. A review of its
pharmacology and therapeutic potential in depression. Drugs 49:
Hood SD, Melichar JK, Taylor LG, et al. (2010) Noradrenergic function
in generalized anxiety disorder: impact of treatment with venlafax-
ine on the physiological and psychological responses to clonidine
challenge. J Psychopharmacol 25: 78–86.
Kelsey JE (2000) Efficacy, safety, and tolerability of venlafaxine XR in
generalized anxiety disorder. Depress Anxiety 12: 81–84.
Kessler RC, Berglund P, Demler O, et al. (2005) Lifetime prevalence
and age-of-onset distributions of DSM-IV disorders in the National
Comorbidity Survey Replication. Arch Gen Psychiatry 62: 593–602.
Kim TS, Pae CU, Yoon SJ, et al. (2006) Comparison of venlafaxine
extended release versus paroxetine for treatment of patients with gen-
eralized anxiety disorder. Psychiatry Clin Neurosci 60: 347–351.
Liebowitz MR, Mangano RM, Bradwejn J, et al.; SAD Study Group
(2005) A randomized controlled trial of venlafaxine extended release
in generalized social anxiety disorder. J Clin Psychiatry 66: 238–247.
Liebowitz MR, Asnis G Mangano R, et al. (2009) A double-blind, placebo-
controlled, parallel-group, flexible-dose study of venlafaxine extended
release capsules in adult outpatients with panic disorder. J Clin Psy-
chiatry 70: 550–561.
Lydiard RB, Rickels K, Herman B, et al. (2010) Comparative efficacy of
pregabalin and benzodiazepines in treating the psychic and somatic
symptoms of generalized anxiety disorder. Int J Neuropsychopharma-
col 13: 229–241.
Meoni P, Hackett D and Lader M (2004) Pooled analysis of venlafaxine
XR efficacy on somatic and psychic symptoms of anxiety in patients
with generalized anxiety disorder. Depress Anxiety 19: 127–132.
Montgomery SA, Tobias K, Zornberg GL, et al. (2006) Efficacy and
safety of pregabalin in the treatment of generalized anxiety disorder:
a 6-week, multicenter, randomized, double-blind, placebo-controlled
comparison of pregabalin and venlafaxine. J Clin Psychiatry 67:
Nardi AE, Valenca AM, Nascimento I, et al. (2000) Double-blind acute
clonazepam vs. placebo in carbon dioxide-induced panic attacks.
Psychiatry Res 94, 179–184.
Nutt DJ, Glue P, Lawson C, et al. (1990) Flumazenil provocation of panic
attacks: evidence for altered benzodiazepine receptor sensitivity in
panic disorder. Arch Gen Psychiatry 47, 917–925.
Papadopoulos A, Rich A, Nutt DJ, et al. (2010) The effects of single dose
anxiolytic medication on the CO2 models of anxiety: differentiation of
subjective and objective measures, J Psychopharmacol 24: 649–656.
Perna G, Bertani A, Gabriele A, et al. (1997) Modification of 35% carbon
dioxide hypersensitivity across one week of treatment with clomip-
ramine and fluvoxamine: a double-blind, randomised, placebo-con-
trolled study. J Clin Psychopharmacol 17: 173–178.
Peterson RA and Reiss RJ (1992) Anxiety Sensitivity Index Manual
(2nd ed.). Worthington, OH: International Diagnostic Systems.
Pohl RB, Feltner DE, Fieve RR, et al. (2005) Efficacy of pregabalin in
the treatment of generalized anxiety disorder: double-blind, placebo-
controlled comparison of BID versus TID dosing. J Clin Psychophar-
macol 25: 151–158.
Pollack MH, Lepola U, Koponen H, et al. (2007) A double-blind study of
the efficacy of venlafaxine extended-release, paroxetine, and placebo
in the treatment of panic disorder. Depress Anxiety 24: 1–14.
Pols HJ, Hauzer RC, Meijer JA, et al. (1996) Fluvoxamine attenu-
ates panic induced by 35% CO2 challenge. J Clin Psychiatry 57:
Reiss S and McNally RJ (1985) The expectancy model of fear. In: Reiss S
and Bootzin RR (eds) Theoretical Issues in Behavior Therapy.
New York: Academic Press. pp. 107–121.
Rickels K, Schweizer E, Csanalosi I, et al. (1988) Long-term treatment
of anxiety and risk of withdrawal. Prospective comparison of cloraz-
epate and buspirone. Arch Gen Psychiatry 45: 444–450.
Rickels K, Downing R, Schweizer E, et al. (1993) Antidepressants for
the treatment of generalized anxiety disorder. A placebo-controlled
comparison of imipramine, trazodone, and diazepam. Arch Gen Psy-
chiatry 50: 884–895.
Sanderson WC, Wetzler S and Asnis GM (1994) Alprazolam blockade of
CO2-provoked panic in patients with panic disorder. Am J Psychiatry
Schruers K and Griez E (2004) The effects of tianeptine or paroxetine on 35%
CO2 provoked panic in panic disorder. J Psychopharmacol 18: 553–558.
Diaper et al. 145 Download full-text
Seddon KM, Morris K, Bailey J, et al. (2010) Effects of 7.5% CO2
challenge in generalized anxiety disorder. J Psychopharmacol 25:
Spielberger CD (1983) Manual for the State-Trait Anxiety Inventory
(Form Y). California, USA: Consulting Psychologists Press, Inc.
Stahl SM, Ahmed S and Haudiquet V (2007) Analysis of the rate of
improvement of specific psychic and somatic symptoms of general
anxiety disorder during long-term treatment with venlafaxine ER.
CNS Spectr 12: 703–711.
Tiihonen J, Kuikka J, Rasanen P, et al. (1997) Cerebral benzodiazepine
receptor binding and distribution in generalized anxiety disorder:
A fractal analysis. Mol Psychiatry 2: 463–471.
Valença A, Nardi AE, Nascimento I, et al. (2000) Acute clonazepam dose
in carbon dioxide induced panic attacks. CINP Meeting, abstract no.
Valença AM, Nardi AE, Nascimento I, et al. (2002) Carbon dioxide test as
an additional clinical measure of treatment response in panic disorder.
Arq Neuropsiquiatr 60-B: 358–361.
Ying DG, Jiang S, Yang H, et al. (2010) Frequency of generalized anxi-
ety disorder in Chinese primary care. Postgrad Med 122: 32–38.
Zwanzger P, Eser D, Aicher S, et al. (2003) Effects of alprazolam
on cholecystokinin-tetrapeptide-induced panic and hypothalamic–
pituitary–adrenal-axis activity: a placebo-controlled study. Neuropsy-
chopharmacology 28: 979–984.