Psychiatry Research 178 (2010) 342–347

Contents lists available at ScienceDirect

Psychiatry Research
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p s yc h r e s

Associations between personality traits and CCK-4-induced panic attacks in

healthy volunteers
Innar Tõru a,⁎,1, Anu Aluoja a,1, Ülle Võhma b, Mait Raag c, Veiko Vasar a, Eduard Maron a,b,e, Jakov Shlik d
a
Department of Psychiatry, University of Tartu, Tartu, Estonia
b
North Estonia Medical Centre Foundation, Psychiatry Clinic, Tallinn, Estonia
c
Department of Public Health, University of Tartu; Tartu, Estonia
d
Department of Psychiatry, University of Ottawa, Ottawa, Ontario, Canada
e
Department of Neuropsychopharmacology and Molecular Imaging, Imperial College London, London, UK

a r t i c l e i n f o a b s t r a c t

Article history: In this study we examined how personality disposition may affect the response to cholecystokinin
Received 24 September 2009 tetrapeptide (CCK-4; 50 μg) challenge in healthy volunteers (n = 105). Personality traits were assessed with
Received in revised form 30 March 2010 the Swedish universities Scales of Personality (SSP). Statistical methods employed were correlation analysis
Accepted 1 April 2010
and logistic regression. The results showed that the occurrence of CCK-4-induced panic attacks was best
predicted by baseline diastolic blood pressure, preceding anxiety and SSP-defined traits of lack of
Keywords:
Cholecystokinin
assertiveness, detachment, embitterment and verbal aggression. Significant interactions were noted between
Anxiety the abovementioned variables, modifying their individual effects. For different subsets of CCK-4-induced
symptoms, the traits of physical aggression, irritability, somatic anxiety and stress susceptibility also
appeared related to panic manifestations. These findings suggest that some personality traits and their
interactions may influence vulnerability to CCK-4-induced panic attacks in healthy volunteers.
© 2010 Elsevier Ireland Ltd. All rights reserved.

1. Introduction response through mainly biological mechanisms, the somewhat

In the past decades several neurobiological and psychological
factors have been identified to play a crucial role in the pathophys-
iology of panic disorder (PD). As the experimental induction of panic
attacks offers a unique opportunity to study the core symptoms of PD
under controlled conditions, several panic challenge procedures have
been established to investigate the mechanisms of panic attacks and
anxiolytic treatments. Panic induction with cholecystokinin-tetra-
peptide (CCK-4) or pentagastrin (CCK-5) has been established as a
valid experimental model of human panic attacks. CCK-4 and CCK-5
are synthetic analogues of the endogenous neuropeptide cholecysto-
kinin (CCK) acting as agonists of the central subtype of CCK receptors
to induce panic attacks in patients with PD and, to a lesser extent, in
healthy subjects (Bradwejn et al., 1991; Abelson and Nesse, 1994).
The CCK-induced panic symptoms resemble spontaneous panic
attacks in PD patients (Bradwejn et al., 1991; Abelson and Nesse,
1994), suggesting the suitability of such experimental challenge
setting to clinical investigation of panic phenomena. The mechanisms
underlying panic induction by CCK-4 are not yet fully understood.
Although previous studies have accounted for CCK-4-induced panic
⁎ Corresponding author. University of Tartu, Department of Psychiatry, Raja 31,
50417 Tartu, Estonia. Tel.: +372 7 318812; fax: +372 7 318801.
E-mail address: Innar.Toru@kliinikum.ee (I. Tõru).
1
Joint first authors.
contradictory data suggest that cognitive appraisal, fear of anxiety
sensations, some personality traits and baseline anxiety may have at
least some role (Koszycki et al., 1993, 1996; Flint et al., 1998; Aluoja
et al., 1997). Koszycki et al. (1996) found that in patients with PD
(n = 29), anxiety sensitivity, a trait characterized by the propensity to
appraise symptoms of anxiety as threatening, correlated significantly
with cognitive, but not with somatic or affective, response to CCK-4.
Radu et al. (2003) demonstrated that in healthy volunteers (n = 20),
the Anxiety Sensitivity Index scores correlated with ratings of anxiety
and discomfort following the administration of pentagastrin. Howev-
er, several studies have observed that anxiety sensitivity did not
predict panic response to CCK-4 in healthy subjects (Koszycki et al.,
1993 (n = 36); Flint et al., 1998 (n = 80)). Furthermore, Van Megen
et al. (1994) found no association between the fear of anxiety-related
symptoms and response to pentagastrin in patients with PD (n = 30).
The associations between baseline state or trait anxiety and response
to CCK-4 also vary from study to study. While Radu et al. (2003)
showed that in healthy volunteers, baseline state anxiety correlated
with anxiety/discomfort reaction after the administration of penta-
gastrin, Aluoja et al. (1997) in their study on healthy volunteers
(n = 14) found that state/trait anxiety predicted the reaction to
placebo, but not to CCK-4. Also, Eser et al. (2008), studying the impact
of state and trait anxiety on CCK-4-induced panic in healthy
volunteers (n = 33), found no significant differences between the
groups of panickers and non-panickers.

0165-1781/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.psychres.2010.04.003
 I. Tõru et al. / Psychiatry Research 178 (2010) 342–347
Among basic personality traits, introversion and neuroticism may
have some relationships with CCK-4-induced panic. Koszycki et al.
(1996) found significant correlations between the scores of Minne-
sota Multiphasic Personality Inventory (MMPI) Social Introversion
scale and somatic, cognitive, and affective response to CCK-4 in
patients with PD. Koszycki and Bradwejn (1997) also found an inverse
relationship between the scores of extraversion on the Eysenck
Personality Questionnaire and baseline nervousness in healthy males
(n = 40). Extraversion scores correlated positively with the onset of
CCK-4-induced symptoms and some physiological indices. The same
study showed that neuroticism correlated positively with pre-
challenge self-rated nervousness, the number of CCK-4-induced
symptoms, and post-challenge self-rated anxiety, nervousness and
fearfulness. Predisposing influence of negative affect on panic reaction
in response to a biological challenge was further supported by Radu
et al. (2003), showing that Karolinska Scales of Personality subscales
of Muscular Tension, Indirect Aggression, Verbal Aggression and
Suspicion were associated with the anxiety reaction to pentagastrin.
This brings attention to the possible role of aggression-related traits in
panic reaction to CCK-4. However, the predictive value of personality
traits in CCK-4 challenge studies has been equivocal. Eser et al. (2007)
studying healthy males (n = 85) found no correlation between CCK-4-
induced panic symptom severity and any of the revised Minnesota
Multiphasic Personality Inventory (MMPI-2) subscales. Furthermore,
the MMPI-2 clinical scales did not correlate with panic status
according to either panic criterion posed by the authors. These
discrepancies warrant further exploration of the role of personality in
laboratory panic, especially in larger samples, which could yield more
reliable results compared with the relatively small samples of
previous studies.
We aimed to test the hypothesis that susceptibility to CCK-4-
induced panic attacks in healthy subjects is influenced by anxiety-
related personality traits. Based on the findings outlined above, we
hypothesized that (a) higher pre-challenge anxiety will be associated
with a stronger panic reaction to CCK-4 challenge; (b) some SSP
anxiety-related traits will predict the emergence and magnitude of
the panic reaction—specifically, somatic trait anxiety will predict
somatic symptoms and psychic trait anxiety will predict cognitive
symptoms of panic; and (c) SSP detachment and trait aggression
scores will predict CCK-4-induced panic reaction.
2. Methods
2.1. Subjects
A total of 105 healthy volunteers (mean age 22.4, range 18–49, 55% females) were
recruited by flyer advertisements. All subjects gave written informed consent after
learning about the study aims and procedures. The study protocol and informed
consent form were approved by the Human Studies Ethics Committee of the University
of Tartu, Estonia. The inclusion criteria were age between 18 and 50 years, no personal
or family (first degree relatives) psychiatric history, negative urine drug screen and
good physical health. Pregnancy was excluded with a urine pregnancy test.
2.2. Assessments
The diagnostic assessment was done using the Estonian translation of the Mini
International Neuropsychiatric Interview (M.I.N.I. 5.0.0; Sheehan et al., 1998). Baseline
anxiety and depression were measured by the clinician-rated Hamilton Anxiety Scale
(HAS; Hamilton, 1969) and two subscales of the self-rated Emotional State
Questionnaire (EST-Q; Aluoja et al., 1999) measuring the symptoms of depression
and anxiety during past four weeks. Personality traits were assessed with the Swedish
universities Scales of Personality (SSP; Gustavsson et al., 2000; Aluoja et al., 2009), a self-
rated questionnaire based on the Karolinska Scales of Personality (Schalling, 1978). The
SSP comprises 91 items grouped into 13 scales: Somatic Trait Anxiety (STA), Psychic
Trait Anxiety (PsTA), Stress Susceptibility (SS), Lack of Assertiveness (LA), Impulsive-
ness (I), Adventure Seeking (AS), Detachment (D), Social Desirability (SD), Embitter-
ment (E), Trait Irritability (TI), Mistrust (M), Verbal Trait Aggression (VTA), and Physical
Trait Aggression (PhTA). Each scale is formed by 7 items rated on a scale of 1 (does not
apply at all) to 4 (applies completely). Factor analysis suggested that SSP scales measure
three broad constructs: neuroticism, extraversion and aggressiveness (Gustavsson et al.,
2000). The Estonian translation of SSP has previously been validated showing a factor
343
structure similar to the original version (Aluoja et al., 2009). The subjective effects of
CCK-4 challenge were assessed by Visual Analogue Scales (VAS), based on scales of Bond
et al. (1974), consisting of 100 mm lines for the dimensions of anxiety (VAS-A) and
discomfort (VAS-D). CCK-4-induced symptoms were rated on the Panic Symptom Scale
(PSS; Bradwejn et al., 1991) assessing the intensity of 18 symptoms derived from the
DSM-III-R/IV criteria for panic attack from 0 (not present) to 4 (extremely severe). The
main outcome variable was occurrence of a panic attack. The a priori criteria to define
the panic attack were a sudden onset of at least 4 PSS symptoms of at least moderate
intensity (score ≥2) and anxiety/apprehension/fear PSS item scored as at least 3 (severe
or extreme). Other measures derived from the PSS were the number of symptoms
(number of items scored ≥1), sum intensity score (the sum of all individual item
ratings), and subscale scores for somatic and cognitive symptoms. Arterial blood
pressure (BP) and heart rate (HR) were measured with an automatic sphygmoma-
nometer (Dinamap Pro 100, Criticon, Tampa, FL, USA). Maximum changes in the studied
parameters were calculated by subtracting baseline scores from the highest values
obtained within 90 s after injection.
2.3. Procedures
The subjects arrived at the research unit on the challenge day at about 10:00 am.
Upon arrival, they completed the EST-Q and HAS. Then, an intravenous cannula was
inserted into an antecubital vein and saline infusion was started to keep the cannula
open. The subjects stayed in the same room resting. At 11 am, after the registration of
baseline values of BP, HR and VAS-A, a bolus injection of 50 µg of CCK-4 (Clinalfa, Merck
Biosciences AG, Switzerland) diluted in 2.5 ml of normal saline solution was given
through the cannula (during ca 3 s). The subjects were asked to describe any symptoms
they experienced after the injection. The BP, HR and VAS-A were registered every 30 s
for 2 min, then at 5 and 15 min. After the CCK-4-induced symptoms had abated, the
subjects were assessed on PSS and rated VAS-A and VAS-D for the peak symptoms. After
completion of the last measurements the cannula was removed and the subject was
allowed to leave when comfortable with an option of phone contact over the next 24 h.
2.4. Data analysis
For comparison of the groups with and without panic reaction we used t-tests for
normally distributed data and Mann–Whitney U tests for the rest of the variables.
Correlation analysis was done to investigate the relationships between the continuous
or nearly continuous variables. A logistic model was developed to determine the
variables possibly predicting the occurrence of panic attack. All possible main-effect
models that did not contain interaction terms with up to 10 parameters (out of 23
possible) were created. After excluding the models with the parameters resulting in
multicollinearity, three models with the least Akaike's information criterion (AIC) were
chosen for detailed investigation. Adding quadratic terms did not decrease AIC. The
interaction terms which improved AIC were added to the models. For easier
interpretation of interactions, the models were fitted to standardised data. Finally, to
find out if the model with better AIC was also significantly better by likelihood, the log-
likelihood chi-squared tests were performed. Correlation analysis was conducted to
examine how personality traits were associated with the number of panic symptoms,
self-assessed discomfort and anxiety. In addition to investigating predictive values of
personality traits, regression models were developed for the PSS total, somatic and
cognitive scores, number of panic symptoms, and VAS-A and VAS-D by stepwise
regression towards minimizing AIC. For PSS symptoms, Poisson regression was used; in
other cases, response variables (or suitably transformed response variables) followed
normal distribution and ordinary linear regression models were developed. The data
were controlled for multiple testing where appropriate and power analyses were
performed ad hoc. The R and Stata statistical software were used for the data analysis.
3. Results
Forty seven subjects (45%) met the study definition for a CCK-4-
induced panic attack. Panic rate was not different between male and
females. The comparison of mean values and predictor variables
between panic and non-panic groups is presented in Table 1. Among
all variables only age and the scores on Verbal Trait Aggression
subscale were significantly different between the groups. After Holm–
Bonferroni correction for multiple testing none of the comparisons
retained significant difference. None of the pairwise comparison tests
achieved 80% power (min 5%, max 62%). The sample sizes required to
achieve 80% power with the observed effect sizes assuming similar
panic attack rates are shown in Table 1.
The comparison of logistic regression models indicated that the
best set (by AIC) of variables predicting the occurrence of PA included
baseline diastolic blood pressure (RRD-0; lower values result in higher
chance of PA), baseline anxiety (EST-Q-A; higher scores give higher
chance of PA), SSP Lack of Assertiveness (LA; higher scores result in
higher chance of PA), SSP Verbal Trait Aggression (VTA; higher scores
344 I. Tõru et al. / Psychiatry Research 178 (2010) 342–347

Table 1
Means, standard deviations and tests for difference between CCK-4-induced panic and non-panic groups.

Group Types of achieved Sample size

needed to achieve
Panickers Non-panickers

Mean S.D. Median Mean S.D. Median P-value Analysis Power 80% power

Age 23.17 6.04 21 21.47 4.31 20 0.021 (a) 0.36 311

HAS 2.4 2.4 2 3.17 4.04 2 0.425 (a) 0.16 567
RRS-0 122.9 9.87 122.5 123.6 10.61 125 0.708 (b) 0.06 6756
RRD-0 73.79 7.49 73.5 71.4 7.68 70 0.112 (b) 0.36 322
fr-0 76.57 19.62 72 77.19 22.54 74 0.872 (a) 0.05 36′419
VAS-A-0 8.19 8.25 5 10.66 10.4 9 0.295 (a) 0.25 451
EST-Q-D 2.93 3.04 2 3.7 3.36 3 0.219 (a) 0.22 547
EST-Q-A 2.48 2.23 2 3.77 3.36 3 0.068 (a) 0.61 152
STA 1.58 0.36 1.57 1.65 0.39 1.57 0.43 (a) 0.15 889
PsTA 1.88 0.48 1.86 1.94 0.46 2 0.561 (b) 0.1 1950
SS 2.03 0.46 2 2.17 0.53 2.14 0.183 (a) 0.29 391
LA 2.18 0.43 2.14 2.29 0.43 2.29 0.208 (b) 0.24 487
I 2.43 0.41 2.43 2.46 0.47 2.43 0.861 (a) 0.06 6829
AS 2.84 0.44 2.86 2.85 0.49 3 0.877 (b) 0.05 67′173
D 2.09 0.39 2.14 2.09 0.4 2 0.995 (b) 0.05 –
SD 3.03 0.31 3 2.97 0.32 3 0.309 (b) 0.16 849
E 1.82 0.4 1.79 1.84 0.37 1.86 0.438 (a) 0.06 12′073
TI 2.18 0.48 2.14 2.25 0.51 2.29 0.479 (b) 0.11 1557
M 1.93 0.47 1.86 2.04 0.37 2.14 0.056 (a) 0.25 487
VTA 2.43 0.49 2.43 2.65 0.49 2.71 0.024 (b) 0.62 159
PhTA 2.09 0.5 2 2.24 0.57 2.29 0.155 (b) 0.29 398

PA—panic attack, (a) Mann–Whitney U test, (b) t-test, HAS—Hamilton's anxiety scale, RRS-0—baseline systolic blood pressure, RRD-0—baseline diastolic blood pressure, fr-0—
baseline heart frequency, VAS-A-0—baseline anxiety on VAS, EST-Q-D—EST-Q depression score, EST-Q-A—EST-Q anxiety score, SSP scales: STA—somatic trait anxiety score, PsTA—
psychic trait anxiety, SS—stress susceptibility, LA—lack of assertiveness, I—impulsiveness, AS—adventure seeking, D—detachment, SD—social desirability, E—embitterment, TI—trait
irritability, M—mistrust, VTA—verbal trait aggression, PhTA—physical trait aggression. Data in bold: significant on significance level P b .05.

result in higher chance of PA), SSP Detachment (D; lower scores
predict higher chance of PA), and SSP Embitterment (E; lower scores
result in higher chance of PA) (Table 2). The best predictive model did
not include age, although Mann–Whitney U-test showed that it was
significantly different between the groups of panickers and non-
panickers. Notably, significant interactions influencing the effect of
the individual factors emerged. For instance, a higher RRD-0 was
associated with lower chance of PA, but this effect was weaker when
Verbal Trait Aggression scores were lower. Similarly, higher scores on
Lack of Assertiveness predicted greater risk of PA, but lower scores on
Detachment as well as higher scores on Embitterment decreased the
effect of Lack of Assertiveness. Furthermore, higher scores on
Embitterment generally lowered the risk of PA, but a high score on
Lack of Assertiveness decreased this effect. For the logit model for PA,
the sample size required to achieve 80% power in testing the
significance should have been 247 for RRD-0, 87 for EST-Q-A, 90 for
LA, 331 for D, 164 for E, 95 for VTA, 166 for RRD-0 VTA, 114 for LA D,
and 244 for LA E.
Correlations between personality traits and the number of panic
symptoms, and self-assessed discomfort and anxiety are presented in
Table 3. Regression models developed for the PSS total, somatic and
cognitive scores, number of panic symptoms, and VAS-A and VAS-D
are presented in Table 4. We observed that higher HAS scores
Table 2
The best predictive logistic regression model for CCK-4-induced PA.
Model for logit(PA) Df Deviance, Gen. R2 AIC LR-X2 P were more likely to react with panic symptoms to provocation with
−0.36 − 0.36·RRD-0 + 0.62· 9 117.2 0.305 137.2 27.19 0.0013
EST-Q-A + 0.82·LA − 0.46·
D − 0.62·E + 0.57·VTA − 0.58·
RRD-0·VTA
+ 0.43·LA·D-0.56·LA·E
Df—degrees of freedom of the model; Gen. R2—Nagelkerke's generalised R2, AIC—Akaike's
information criterion, LR-X2—model likelihood chi-squared statistic, P—model's P-value.
RRD-0—standardised baseline diastolic blood pressure (ratio of centred baseline diastolic
blood pressure, EST-Q-A—EST-Q anxiety score, LA—SSP lack of assertiveness, E—SSP
embitterment score, D—SSP detachment score, VTA—SSP verbal trait aggression score.
predicted higher PSS cognitive subset scores, total PSS scores, and the
number of PSS symptoms. Higher EST-Q anxiety scores predicted
higher PSS total and somatic scores. Higher Physical Trait Aggression
was linked to higher PSS total, somatic and cognitive scores, and the
number of PSS symptoms. Higher Stress Susceptibility was associated
with higher PSS cognitive scores. Higher Lack of Assertiveness
predicted higher PSS somatic scores, number of PSS symptoms, and
maximum post-challenge discomfort. Higher Embitterment predicted
lower PSS somatic scores. Higher Trait Irritability was linked to higher
maximum post-challenge discomfort.
4. Discussion
The 45% panic rate in our study was similar to the earlier findings
in healthy volunteers (e.g. 47% in the study of Bradwejn et al., 1991).
In our sample, the panickers were significantly younger than non-
panickers. Although the age range in our study was narrower, this
observation is in parallel with the data of Flint et al. (1998) showing
that older subjects (age 65 and older vs. 20–35 years) had less
pronounced panic response to CCK-4. The occurrence of CCK-4-
induced panic attack was best predicted by the EST-Q assessed
anxiety, SSP subscales of Lack of Assertiveness, Detachment, Embit-
terment and Verbal Trait Aggression, and baseline diastolic blood
pressure.
The effect of baseline anxiety on susceptibility to panic challenge
has been a subject of some contradictions. Studies in patients with PD
have shown that the subjects with a higher level of baseline anxiety
sodium lactate and hyperventilation (Aronson et al., 1989; Spinhoven
et al., 1993). In healthy volunteers, Radu et al. (2003) have found that
the pre-challenge HAS scores correlated positively with the ratings of
anxiety and discomfort following administration of pentagastrin
(0.05 µg and 0.2 µg/kg). Eser et al. (2007) have found that the
panickers reported higher baseline panic scores on two panic rating
scales used, although the difference reached a statistical significance
only with one of the two applied panic definitions. On the other hand,
Aluoja et al. (1997) have shown that in healthy volunteers baseline
 I. Tõru et al. / Psychiatry Research 178 (2010) 342–347 345

Table 3
Pearson correlations between variables measured before injection of CCK and PSS-scores and VAS-based discomfort and anxiety scores.

PSS-TOT PSS-COG PSS-SOM PSS-NrS VAS-D VAS-A-max VAS-A-dif

Age −0.045 −0.139 −0.004 −0.023 −0.023 −0.155 −0.129

HAM-anx 0.309** 0.205* 0.305** 0.279** 0.013 0.020 −0.070
RRS-0 0.013 0.042 0.001 −0.015 0.009 −0.080 −0.118
RRD-0 −0.173 −0.137 −0.163 −0.141 0.034 −0.115 −0.077
fr-0 −0.062 0.006 −0.079 −0.100 −0.065 −0.023 −0.034
VAS-A-0 0.077 0.157 0.037 0.030 0.171 0.182 −0.248*
EST-Q-D 0.238* 0.137 0.243* 0.220* 0.144 0.095 −0.003
EST-Q-A 0.324*** 0.249* 0.307** 0.241* 0.123 0.129 0.026
STA 0.276** 0.233* 0.255** 0.245* 0.182 0.231* 0.121
PsTA 0.172 0.079 0.183 0.182 0.230* 0.136 0.004
SS 0.229* 0.226* 0.200* 0.241* 0.244* 0.233* 0.108
LA 0.099 −0.041 0.136 0.151 0.245* 0.089 0.005
I 0.051 0.070 0.037 −0.011 0.108 0.053 −0.021
AS 0.062 0.150 0.022 0.021 0.159 0.182 0.114
D 0.082 −0.081 0.130 0.149 0.077 −0.042 −0.079
SD −0.156 −0.071 −0.166 −0.216* −0.148 −0.053 −0.012
E 0.085 0.115 0.063 0.065 0.193* 0.112 −0.010
TI 0.227* 0.235 0.193* 0.179 0.223* 0.093 0.016
M 0.118 0.155 0.089 0.146 0.166 0.165 0.060
VTA 0.208* 0.136 0.207* 0.187 0.157 0.119 0.095
PhTA 0.265** 0.297** 0.217* 0.212* 0.086 0.038 −0.013

*P b 0.05, **P b 0.01, ***P b 0.001. HAS—Hamilton's anxiety score, RRS-0 baseline systolic blood pressure, RRD-0—baseline diastolic blood pressure, fr-0—baseline heart frequency, VAS-A-0—
baseline anxiety on VAS, EST-Q-D—EST-Q depression score, EST-Q-A—EST-Q anxiety score, STA—SSP somatic trait anxiety score, PsTA—SSP psychic trait anxiety score, SS—SSP stress
susceptibility score, LA—SSP lack of assertiveness score, I—SSP impulsiveness score, AS—SSP adventure seeking score, D—SSP detachment score, SD—SSP social desirability score, E—SSP
embitterment score, TI—SSP trait irritability score, M—SSP mistrust score, VTA—SSP verbal trait aggression score, PhTA—SSP physic trait aggression score. PSS-TOT—sum of all PSS item
scores, PSS-COG—sum of cognitive PSS item scores, PSS-SOM—sum of somatic PSS item scores, PSS-NrS—Number of PSS symptoms, VAS-D—maximum discomfort score on VAS, VAS-A-
max—maximum anxiety score on VAS, VAS-A-dif—difference between baseline and maximum anxiety scores on VAS. Data in bold: significant on significance level P b .05.

anxiety, as well as anxiety sensitivity, did predict reactions to placebo,
but not to CCK-4. Also, Koszycki et al. (1998) observed no association
between pre-challenge VAS measurements of subjective arousal and
anxiety and susceptibility to CCK-4-induced panic in healthy controls.
In the present study, higher levels of anxiety as assessed by the HAS
and EST-Q predicted the occurrence of panic attack, as well as higher
PSS total scores and somatic intensity scores. Higher PSS cognitive
subscores were predicted by higher scores on the HAS, but not by
higher EST-Q anxiety, probably reflecting contextual differences
between these two scales. Thus, our data support the assumption
that baseline anxiety has at least some role in determining the
reaction to a panic challenge agent.
The predictive effect of RRD-0 in our study is difficult to explain. In
other studies, the data concerning relationship between RRD-0 and
CCK-4-induced symptoms, as well as the effect of CCK-4 on RRD, have
been equivocal. Eser et al. (2007) reported no difference between
panickers and non-panickers in RRD-0 and no effect of CCK-4 on RRD.
Bradwejn et al. (1992), in a dose-ranging study, observed no influence
of RRD-0 on post-challenge variables but found a consistent post-
challenge increase in RRD with a trend for dose–response effect.
Koszycki et al. (1998) reported no differences between panickers and
non-panickers in RRD-0 but noted differences between the groups in
RRD response to CCK-4. Koszycki et al. (1998) have suggested that an
augmented diastolic blood pressure response may be explained by the
augmented parasympathetic activation by CCK-4 or a blunted
neuronal influence of vagal afferents, both of which might be linked
to some personality characteristics. Pertinently, Koszycki and Brad-
wejn (1997) have shown that in healthy volunteers, extraversion
scores correlated negatively with CCK-4-induced increase in RRD. In
the present study, the mean RRD-0 did not differ between the groups
of panickers and non-panickers, yet lower RRD-0 predicted the
occurrence of panic attack. The contradiction, however, might
dissipate if the modulating influence of other factors was considered.
In particular, the predictive effect of RRD-0 was lessened by the lower
Verbal Trait Aggression score. In addition, many other moderating
factors not assessed in the present study may exist.
We identified four personality traits as potential predictors of CCK-4-
induced panic attack: higher Lack of Assertiveness and Verbal Trait

Table 4
Predictive regression models for number of CCK-4-induced panic symptoms, somatic reactions, and self-assessed discomfort and anxiety.

Model Gen. R2 Model characteristics AIC

PSS-TOT = 8.40 + 0.57·HAS + 0.57·EST-Q-A + 3.73·PhTA 0.184 SE: 8.126 on 101 df; F(3,101): 7.564, P b 0.001 743.86
(PSS-COG)0.5 = 0.65 + 0.30·SS + 0.29·PhTA 0.151 SE: 0.5827 on 102 df; F(2,102): 7.473, P b 0.001 189.53
PSS-SOM = 3.64 + 0.44·HAS + 0.56·EST-Q-A + 3.39·LA − 0.201 SE: 6.593 on 99 df; F(5,99): 4.971, P b 0.001 701.86
3.11·E + 3.41·PhTA
ln(PSS-NrS) = 1.50 + 0.02·HAS + 0.14·LA + 0.16·PhTA 0.138 Null deviance: 113.970 on 104 df, Residual deviance: 527.76
98.452 on 101 df
(VAS-D)1.8 = −98.6 + 526.7·LA + 464.5·TI 0.111 SE: 927.9 on 102 df, F(2,102): 6.35, P = 0.003 1737.85
(VAS-A-max)1.7 = 650.0 + 418.8·STA 0.061 SE: 618.1 on 103 df, F(1,103): 6.641, P = 0.011 1651.55
(VAS-A-dif)1.5 = 322.2 − 8.3·VAS-A-0 + 129.2·STA 0.071 SE: 224.2 on 102 df, F(2,102): 6.638, P = 0.002 1439.57

Gen. R2—Nagelkerke's generalised R2; AIC—Akaike's information criterion; SE—residual standard error; PSS-TOT—sum of all PSS item scores; PSS-COG—sum of cognitive PSS item
scores; PSS-SOM—sum of somatic PSS item scores; PSS-NrS—number of PSS symptoms; VAS-D—maximum discomfort score on VAS; VAS-A-max—maximum anxiety score on VAS;
VAS-A-dif—difference between baseline and maximum anxiety scores on VAS; HAS—Hamilton Anxiety Scale score; EST-Q-A—EST-Q anxiety score; PhTA—SSP physical trait
aggression; SS—SSP stress susceptibility; LA—SSP lack of assertiveness; E—SSP embitterment; TI—SSP trait irritability; STA—SSP somatic trait anxiety; VAS-A-0—baseline anxiety on
VAS.
346 I. Tõru et al. / Psychiatry Research 178 (2010) 342–347
Aggression as well as lower Detachment and Embitterment. We did not
confirm our assumption that higher Detachment is associated with
panic reaction. Nevertheless, we found an association of panic attack
with Lack of Assertiveness. Higher Lack of Assertiveness was also
associated with a greater number of panic symptoms, higher scores of
PSS somatic symptoms, and higher maximum discomfort during the
provocation. Lack of Assertiveness in SSP is a trait resembling social
anxiety. Katzman et al (2004) have shown a trend for CCK-4-induced
panic to occur more frequently in patients with Social Phobia (SP) as
compared to healthy controls. Patients with SP also reported a greater
number and intensity of panic symptoms as compared to the patients
with Obsessive–Compulsive Disorder and control subjects in both CCK-
4 and placebo challenge conditions. This suggests a higher non-specific
responsiveness in SP. Socially anxious subjects may have additional
sensitivity to the interpersonal aspects of the CCK-4 provocation
procedure. The panic-predicting role of low detachment is also in
agreement with social sensitivity hypothesis. It has been argued that
besides mental and physical catastrophes panic subjects fear social
humiliation while experiencing arousal symptoms (Austin and
Richards, 2001). Individuals who isolate less from social environment
could be more sensitive to the reactions of others and therefore respond
with heightened anxiety to the challenge. At the same time low
assertiveness may inhibit direct communication of discomfort and lead
to higher anxiousness and panic symptoms. In the presence of higher
detachment, the communication-blocking role of low assertiveness may
be enhanced, as indicated by the interaction between SSP Detachment
and the Lack of Assertiveness subscales. Detachment has been
conceptualized as part of a broader extraversion–introversion construct
(Gustavsson et al., 2000). Other studies have shown associations
between higher introversion and aspects of CCK-4-induced panic in
healthy volunteers (Koszycki and Bradwejn, 1997) and patients with PD
(Koszycki et al 1996). This is in contrast to our findings of higher
detachment predicting a lower chance of panic attack. Notably, the
constructs in these two studies were measured by different instruments
and it is not known how much the MMPI Social Introversion and SSP
Detachment may overlap. The inverse relationship between embitter-
ment and CCK-4-induced panic was a counterintuitive finding. SSP
Embitterment, which is associated with neuroticism and aggression,
indicates general dissatisfaction, feelings of injustice, and resentment
and, therefore, may be expected to heighten propensity to panic. At the
same time embitterment can be conceptualized as a habitual way of
stress management; blaming others and expecting the worst, which can
reduce the reactivity to sudden uncontrollable acute stress encountered
in panic provocation situation.
In accordance with our hypothesis we found that personality traits
reflecting aggressive tendencies were associated with CCK-4-induced
panic. Verbal trait aggression predicted the occurrence of panic attack,
whereas physical trait aggression predicted all aspects of panic
symptoms. Previously it has also been indicated that the tendency to
be verbally aggressive was associated with panic response to
pentagastrin administration (Radu et al., 2003). The association
between aggressiveness and provoked panic response is further
supported by George et al. (2000), who showed that persons
exhibiting high rate of physical aggression in domestic relationships
react with more panic symptoms to sodium lactate provocation. The
exact nature of these associations is unclear. A Radu et al. (2003)
finding that personality emerges as a significant predictor of panic
reaction only with a lower dose of pentagastrin suggests rather non-
specific stress reactivity. It is possible that certain personality traits
heighten an individual's reactivity to different kinds of stressors,
including panic-provoking agents. On the other hand, the anxiety and
aggression are known to share underlying neurobiology, including
aspects of CCK neurotransmission and pharmacotherapy approaches
(Siegel et al., 2007). Of note, our preliminary study in patients with
PD, also using SSP, indicated an increased aggressiveness in the
patients as compared to healthy controls, which was more pro-
nounced in cases of PD with agoraphobia and affective co-morbidity
(Võhma et al., 2010).
Our hypothesis that some SSP anxiety-related traits will predict
the emergence and magnitude of the panic reaction was only partially
supported. Although anxiety-related traits did not predict the
occurrence of panic attack, we still found that stress susceptibility
had significant associations with cognitive panic symptoms. The SSP
Stress Susceptibility denotes sensitivity to situational pressure and
easy fatigability. According to Abelson et al. (2007), heightened
reactivity to novel and uncontrollable situations may be a vulnera-
bility factor for PD and can account for reactivity in panic challenge
experiments. Somatic trait anxiety predicted increase of subjective
anxiety during the CCK provocation. Stress susceptibility and somatic
trait anxiety are part of the broader neuroticism construct; thus, our
study supports the findings of Koszycki and Bradwejn (1997) on the
converging relationship of neuroticism and number of CCK-4 induced
symptoms and self-rated anxiety. Our results did not confirm the
specific hypotheses that somatic trait anxiety will predict somatic
symptoms and that psychic trait anxiety will predict cognitive
symptoms of panic. The interaction of personality traits in predicting
occurrence of panic showed a complex nature of associations between
personality characteristics and panic reaction. Most studies so far have
looked at the associations of single personality traits or dimensions
with panic reaction without considering the influence of combina-
tions of traits. Our results suggest that the effect of a certain
personality trait on panic reaction can be enhanced or decreased by
the presence of other traits.
When interpreting the results of the present study in the context
of earlier and future studies, one needs to bear in mind that different
personality scales assessing slightly or even robustly diverse para-
meters have been used across the various studies, making direct
comparisons difficult and possibly explaining some inconsistencies.
Also, variations in challenge settings used in different laboratories
may have played a role in amplifying or attenuating the impact of
certain personality traits. Some other limitations of our study have to
be considered. As we did not have a placebo control for CCK-4, it
remains uncertain whether the results are specific to CCK-4 or can be
attributed to non-specific factors of challenge procedure. Our model
predicted the occurrence of PA with moderate accuracy. To increase
the model's precision, additional potentially relevant factors may
need to be included. For example, the effect of menstrual cycle phase
in females suggested by Le Melledo et al. (1999) was not taken into
account in our protocol. Also, we did not assess all the personality
characteristics that have been shown to influence CCK-4-induced
panic reaction in earlier studies. The effect of different personality
traits on panic reaction seems to be of a small statistical magnitude
and to involve complex interactions between traits; therefore, and
based on power calculations, it is obvious that for reliable evaluation
of the associations between personality traits and experimentally
induced anxiety much larger sample sizes are necessary (see Table 1).
In conclusion, challenge studies with CCK-4 seem productive in
uncovering the role of individual psychological characteristics in
vulnerability to panic attacks. Our results suggest that neuroticism and
aggressiveness predispose to a greater panic reaction in laboratory
challenge in healthy volunteers, but the relevance of these findings to
clinical aspects of PD is not clear. The influence of personality traits on
susceptibility to panic attacks requires more extensive research using
different challenge and treatment approaches in the individuals
affected by PD and those at risk to develop PD.
Acknowledgements
This study was supported by Estonian Science Foundation grant 7034 (EM) and by
target grants SF0182590As03 (VV) and SF0180125s08 (VV) from the Ministry of Education
of Estonia. The authors thank study subjects for their participation, the nurses Katri Sööt,
Jane Puusepp, Birgit Aumeste, Merle Taevik and Ketlin Veeväli for their dedicated assistance
and Emilie Chan, MSc, for careful proofreading.
 I. Tõru et al. / Psychiatry Research 178 (2010) 342–347
References
Abelson, J.L., Nesse, R.M., 1994. Pentagastrin infusion in patients with panic
disorder. I. Symptoms and cardiovascular responses. Biological Psychiatry 36,
73–83.
Abelson, J.L., Khan, S., Liberzon, I., Young, E.A., 2007. HPA axis activity in patients with
panic disorder: review and synthesis of four studies. Depression and Anxiety 24,
66–76.
Aluoja, A., Shlik, J., Vasar, V., Kingissepp, P.H., Jagomagi, K., Vasar, E., Bradwejn, J., 1997.
Emotional and cognitive factors associated with response to CCK-4 in healthy
volunteers. Psychiatry Research 66, 59–67.
Aluoja, A., Shlik, J., Vasar, V., Luuk, K., Leinsalu, M., 1999. Development and
psychometric properties of the Emotional State Questionnaire, a self-report
questionnaire for depression and anxiety. Nordic Journal of Psychiatry 53, 443–449.
Aluoja, A., Voogne, H., Maron, E., Gustavsson, J.P., Võhma, Ü., Shlik, J., 2009. Personality
traits measured by the Swedish universities Scales of Personality: factor structure
and position within the five-factor model in an Estonian sample. Nordic Journal of
Psychiatry 63, 231–236.
Aronson, T.A., Whitaker-Azmitia, P., Caraseti, I., 1989. Differential reactivity to lactate
infusions: the relative role of biological, psychological and conditioning variables.
Biological Psychiatry 25, 469–481.
Austin, D.W., Richards, J.C., 2001. The catastrophic misinterpretation model of panic
disorder. Behaviour Research and Therapy 39, 1277–1291.
Bond, A.J., James, D.C., Lader, M.H., 1974. Physiological and psychological measures in
anxious patients. Psychological Medicine 4, 364–373.
Bradwejn, J., Koszycki, D., Shriqui, C., 1991. Enhanced sensitivity to cholecystokinin
tetrapeptide in panic disorder. Clinical and behavioral findings. Archives of General
Psychiatry 48, 603–610.
Bradwejn, J., Koszycki, D., Annable, L., Couėtoux du Tetre, A., Reines, S., Karkanias, C.A., 1992.
Dose-ranging study of the behavioral and cardiovascular effects of CCK-tetrapeptide in
panic disorder. Biological Psychiatry 32, 903–912.
Eser, D., Schüle, C., Baghai, T., Floesser, A., Krebs-Brown, A., Enunwa, M., de la Motte, S.,
Engel, R., Kucher, K., Rupprecht, R., 2007. Evaluation of the CCK-4 model as a
challenge paradigm in a population of healthy volunteer within a proof-of-concept
study. Psychopharmacology 192, 479–487.
Eser, D., Wenninger, S., Baghai, T., Schüle, C., Rupprecht, R., 2008. Impact of state and
trait anxiety on the panic response to CCK-4. Journal of Neural Transmission 115
(6), 917–920.
Flint, A.J., Koszycki, D., Vaccarino, F.J., Cadieux, A., Boulenger, J.P., Bradwejn, J., 1998.
Effect of aging on cholecystokinin-induced panic. American Journal of Psychiatry
155, 283–285.
George, D.T., Hibbeln, J.R., Ragan, P.W., Umhau, J.C., Phillips, M.J., Doty, L., Hommer, D.,
Rawlings, R.R., 2000. Lactate-induced rage and panic in a select group of subjects
who perpetrate acts of domestic violence. Biological Psychiatry 47, 804–812.
347
Gustavsson, J.P., Bergman, H., Edman, G., Ekselius, L., von Knorring, L., Linder, J., 2000.
Swedish universities Scales of Personality (SSP): construction, internal consistency
and normative data. Acta Psychiatrica Scandinavica 102, 217–225.
Hamilton, M., 1969. Diagnosis and rating of anxiety. British Journal of Psychiatry 76–79
Special publication.
Katzman, M.A., Koszycki, D., Bradwejn, J., 2004. Effects of CCK-tetrapeptide in patients
with social phobia and obsessive–compulsive disorder. Depression and Anxiety 20,
51–58.
Koszycki, D., Bradwejn, J., 1997. Personality correlates of response to CCK-4 in healthy
males. Biological Psychiatry 42 (1S–297S), 18S.
Koszycki, D., Cox, B., Bradwejn, J., 1993. Anxiety sensitivity and response to
cholecystokinin-tetrapeptide in healthy volunteers. American Journal of Psychiatry
150, 1881–1883.
Koszycki, D., Bradwejn, J., Zacharko, R.M., 1996. Influence of personality on behavioral
response to cholecystokinin-tetrapeptide in patients with panic disorder. Psychi-
atry Research 62, 131–138.
Koszycki, D., Zacharko, R.M., Le Melledo, J.M., Bradwejn, J., 1998. Behavioral, cardiaovas-
cular, and neuroendocrine profiles following CCK-4 challenge in healthy volunteers: a
comparison of panickers and nonpanickers. Depression and Anxiety 8, 1–7.
Le Melledo, J.M., Merani, S., Koszycki, D., Palmour, R., Gutkowska, Y., Bellavance, F.,
Steinberg, S.I., Bichet, D.G., Bradwejn, J., 1999. Influences of the menstrual cycle and
premenstrual dysphoric disorder on the response to CCK-4. Neuropsychopharma-
cology 20, 81–91.
Radu, D., Ahlin, A., Svanborg, P., Lindefors, N., 2003. Pentagastrin test for anxiety—
psychophysiology and personality. Psychopharmacology 166, 139–145.
Schalling, D., 1978. Psychopathy-related personality variables and the psychophysiol-
ogy of socialization. In: Hare, R.D., Schalling, D. (Eds.), Psychopathic Behaviour.
Approaches to Research. John Wiley and Sons, Chichester, pp. 85–106.
Sheehan, D.V., Lecrubier, Y., Sheehan, K.H., Amorim, P., Janavs, J., Weiller, E., Hergueta,
T., Baker, R., Dunbar, G.C., 1998. The Mini-International Neuropsychiatric Interview
(MINI): the development and validation of a structured diagnostic psychiatric
interview for DSM-IV and ICD-10. Journal of Clinical Psychiatry 59, 22–33.
Siegel, A., Bhatt, S., Bhatt, R., Zalcman, S.S., 2007. The neurobiological bases for
development of pharmacological treatments of aggressive disorders. Current
Neuropharmacology 5, 135–147.
Spinhoven, P., Onstein, E.J., Sterk, P.J., Le Haen-Versteijnen, D., 1993. Discordance
between symptom and physiological criteria for the hyperventilation syndrome.
Journal of Psychosomatic Research 37, 281–289.
Van Megen, H.J.G.M., Westenberg, H.G.M., den Boer, J.A., Haigh, J.R., Traub, M., 1994.
Pentagastrin induced panic attacks: enhanced sensitivity in panic disorder patients.
Psychopharmacology 114, 449–455.
Võhma, Ü., Aluoja, A., Vasar, V., Shlik, J., Maron, E., 2010. Evaluation of personality traits
in panic disorder using Swedish universities Scales of Personality. Journal of
Anxiety Disorders 24, 141–146.