gender differences in prepulse inhibition (ppi) in bipolar disorder: men have reduced ppi, women...
TRANSCRIPT
Gender differences in prepulse inhibition (PPI)in bipolar disorder: men have reduced PPI,women have increased PPI
Andrea Gogos1,2,3, Maarten van den Buuse1,2 and Susan Rossell3,4
1 Behavioural Neuroscience Laboratory, Mental Health Research Institute of Victoria, Parkville, VIC, Australia2 Centre for Neuroscience, University of Melbourne, Parkville, VIC, Australia3 Department of Cognitive Neuropsychiatry, Mental Health Research Institute of Victoria, Parkville, VIC, Australia4 Cognitive Neuropsychiatry Laboratory, Monash-Alfred Psychiatry Research Centre, School of Psychology, Psychiatry and
Psychological Medicine, Monash University, The Alfred, Prahran, VIC, Australia
Abstract
Prepulse inhibition (PPI) is a measure of sensorimotor gating or information processing. Few studies have
examined PPI in bipolar disorder (BD) ; two studies reported a PPI disruption and two reported no
change. There are gender differences in PPI and within the clinical profile of BD, which may explain some
of these discrepancies. Thus, the effect of gender on PPI in BDwas the focus of the current study. Euthymic
BD patients (14 male/15 female) were compared to age- and IQ-matched healthy control participants
(16 male/16 female). Assessment of PPI included 21 pulse-alone trials (115 dB) and a total of 42 prepulse-
pulse trials (seven of each prepulse : 74, 78, 86 dB) at two stimulus onset asynchrony levels (SOA: 60,
120 ms). There was a grouprSOA and a grouprgender interaction, reflecting that men with BD showed
reduced PPI compared to control males at the 60-ms SOA (3% in BD vs. 26% in controls), but not the
120-ms SOA. In contrast, women with BD had significantly increased PPI compared to female controls at
the 120-ms SOA (49% in BD vs. 29% in controls), but not the 60-ms SOA. Compared to control participants
BD patients showed changes in PPI, which are gender-dependent ; male BD participants had reduced PPI,
whereas female BD participants had increased PPI. This gender difference highlights the need to consider
men and women with BD as two distinct groups, at least in PPI studies.
Received 9 November 2008 ; Reviewed 12 February 2009 ; Revised 19 March 2009 ; Accepted 24 April 2009 ;
First published online 3 June 2009
Key words : Bipolar disorder (BD), gender differences, PPI, schizophrenia, startle.
Introduction
Bipolar disorder (BD) is a serious mental illness where
patients may experience both manic episodes and
syndromal major depression. BD can be characterized
by an abrupt onset and a fluctuating course in which
the patient returns to a relatively normal state between
episodes (Berns &Nemeroff, 2003). Gender differences
are present in the clinical profile of BD, e.g. men have
an earlier onset of BD, as determined by the first manic
episode (Kennedy et al. 2005). Compared to men,
women have higher incidence rates throughout adult-
hood rather than early life and are at a greater risk
for depressive episodes, episodes of mixed mania
and rapid-cycling (Curtis, 2005; Kennedy et al. 2005 ;
Leibenluft, 1996). These gender differences may be
attributed to the effect of the female sex hormone,
oestrogen, in the brain. Oestrogen has been implicated
in mood, higher cognitive function and sensorimotor
gating (Bethea et al. 1998 ; Gogos et al. 2006 ; McEwen,
2001 ; Rubinow et al. 1998).
Prepulse inhibition (PPI) is a measure of sensori-
motor gating or in other words, information pro-
cessing. Sensorimotor gating is a normal protective
mechanism in the brain that filters irrelevant sensory,
motor or cognitive information, allowing for coherent
thought (Braff et al. 2001). Deficits in gating sensori-
motor information may result in an information over-
load or a misinterpretation of stimuli and may also
cause certain symptoms of schizophrenia, such as
thought disorder (Braff et al. 2001 ; Perry & Braff, 1994).
Address for correspondence : A. Gogos, Ph.D., Mental Health
Research Institute of Victoria, National Neuroscience Facility,
161 Barry St, Carlton South, VIC 3053, Australia.
Tel. : +613 8344 1851 Fax : +613 9387 5061
Email : [email protected]
International Journal of Neuropsychopharmacology (2009), 12, 1249–1259. Copyright f CINP 2009doi:10.1017/S1461145709000480
ARTICLE
CINP
In humans, PPI of the acoustic startle is measured via
the eye-blink startle response (Braff et al. 2001). There
is an extensive literature on how PPI is affected in
schizophrenia patients (for review see Braff et al. 2001).
PPI is also deficient in other disorders including
obsessive–compulsive disorder (Hoenig et al. 2005)
and Huntington’s disease (Swerdlow et al. 1995).
These disorders have a common underlying deficit, i.e.
PPI deficits are the result of abnormalities within a
specific brain circuit rather than a specific psycho-
pathology (Swerdlow et al. 2001).
A limited number of studies have recently ex-
amined PPI in BD and found that remitted or acutely
manic BD patients, as well as their unaffected siblings,
show disruptions of PPI compared to healthy control
subjects (Giakoumaki et al. 2007 ; Perry et al. 2001).
However, other studies have shown no disruption
of PPI in BD adults (Barrett et al. 2005; Carroll et al.
2007) and children (Rich et al. 2005). These inconsistent
findings do not appear to be due to current symptom
severity, as two studies tested euthymic participants
(Barrett et al. 2005 ; Giakoumaki et al. 2007) and two
studies tested symptomatic participants [acute mania
and psychosis (Perry et al. 2001), manic or mixed epi-
sode (Carroll et al. 2007)]. It is possible that a history
of psychotic episodes may explain PPI deficits in BD,
as psychosis can occur in BD and schizophrenia.
Alternatively, gender may play a role in mediating PPI
deficits.
Gender differences in PPI have been described
in healthy participants, where men show greater PPI
than women (Swerdlow et al. 1993). Further, women
have greater PPI during the early follicular (low oes-
trogen) phase of the menstrual cycle, compared to
the luteal phase (Jovanovic et al. 2004 ; Swerdlow et al.
1997). The previous PPI literature for BD does not state
which stage of the menstrual cycle the women were in
when tested for PPI. In the schizophrenia literature
some studies have reported PPI gender differences,
with disruptions in male, but not female, clinical par-
ticipants (Kumari et al. 2004), alternatively disruptions
in female schizophrenia patients (Braff et al. 2005) and
lastly female patients have a greater PPI disruption
compared to male patients (Swerdlow et al. 2006a),
overall suggesting variable findings.
The present study aimed to examine gender differ-
ences in PPI in BD. We tested PPI in men and women
with BD compared to gender-, age- and IQ-matched
healthy controls. We recruited euthymic BD patients
to reduce the potential confounds present when test-
ing acutely unwell patients. Additionally, this recruit-
ment strategy enabled us to obtain patient groups that
were matched for symptom severity, a potential cause
of a gender difference. We hypothesized that there
would be a gender difference in PPI when comparing
BD patients and controls.
Methods
Participants
The present study tested 66 participants : 31 BD
patients and 35 healthy controls. All participants were
aged between 21 and 61 yr, were fluent in English,
and were recruited from the local community via
advertisements, outpatient facilities (Northern Hospital
and St Vincent’s Health) and word-of-mouth. The
procedures described in this study were approved
by the Mental Health Research and Ethics Committee
(Melbourne Health, Victoria, Australia) and partici-
pants gave their written, informed consent prior to
testing.
Participants from either group were excluded if
they reported a history of brain trauma, neurological
illness (n=1), concurrent substance abuse (n=1), co-
morbidity with Axis I disorders, had undergone elec-
troconvulsive therapy in the previous year, or had an
estimated full-scale verbal IQ<80 (n=1) as indicated
by the National Adult Reading Test (NART; Nelson,
1982). Handedness was assessed with the Edinburgh
Handedness Inventory (Oldfield, 1971). Control par-
ticipants were free from psychiatric illness [self-
reported and screened for current symptoms using
the Brief Psychiatric Rating Scale (BPRS; Ventura et al.
1993)] and were excluded if they had a first-degree
relative with schizophrenia or BD (n=1). Finally, one
male control showed inadequate startle responses
(defined in the PPI section below) and was excluded
from the final analysis. Therefore, the final analysis
included 61 participants (14 male/15 female BD par-
ticipants, 16 male/16 female controls) (Table 1).
BD patients were remitted outpatients, self-
reported euthymic and most were on medication. BD
diagnosis was confirmed according to DSM-IV criteria
during a clinical interview with a trained researcher
(independent and naive to the study hypothesis)
using the combined Structured Clinical Interview for
DSM-IV and Positive and Negative Syndrome Scale
(PANSS; Table 1) (First et al. 1997). As assessed using
the Bech–Rafaelsen Mania Rating Scale (MRS; Bech
et al. 1979), 27 patients were currently euthymic (score
0–10) and two patients had mild mania (score 11–15).
As assessed using the Hamilton Depression Rating
Scale (HAMD; Hamilton, 1960), 18 patients were cur-
rently euthymic (score 0–7), eight patients had mild
depression (score 8–13), two patients had moderate
1250 A. Gogos et al.
depression (score 14–18) and one female patient had
very severe depression (o23, scored 25). Of the 29 BD
patients included in the study, three were unmedi-
cated and 26 were prescribed psychotropic medication
[median number of medications 2 (S.D.=0.9) ; Table 2].
Study design
All participants were tested once, with females tested
during low levels of circulating oestrogen and pro-
gesterone in order to eliminate the confounding
variable of cycling hormone levels. Therefore, women
with a natural menstrual cycle were tested during the
early follicular phase (days 1–8 of cycle ; n=9 BD,
n=10 control), women prescribed the oral contra-
ceptive pill were tested during administration of
the inactive pills (n=2 BD, n=2 control), and post-
menopausal women (lack of period for>2 years) were
tested at any time (n=4 BD, n=4 control). Although
all women were tested during low levels of circulating
hormones, it is unclear whether the varying hormonal
conditions (natural cycle vs. contraceptive pill vs. post-
menopause) may be associated with endocrine effects
on PPI. However, there was a similar distribution of
control and BD women in each of the three hormonal
conditions.
Measures
PPI involved measuring the eye-blink component of
the acoustic startle response using the EMG Startle
System (San Diego Instruments, USA), which was
regularly calibrated (Quest SoundPro dB Meter, Quest
Technologies, USA). Electromyogram (EMG) activity
of the orbicularis oculi was recorded using two small
Ag/AgCl electrodes (electrolyte gel-filled) placed
under the right eye ; the reference electrode was at-
tached to the right mastoid (average impedance
<16 kV). The EMG system delivered acoustic stimuli
binaurally to the participant via headphones (Maico,
San Diego Instruments). Before being tested for PPI, all
participants refrained from smoking for >20 min and
Table 1. Demographic information
Bipolar disorder Control
Male (n=14) Female (n=15) Male (n=16) Female (n=16)
Age (yr) 45 (10.9) 41 (10.9) 40 (11.7) 41 (12.0)
Predicted IQ 105 (12.2) 108 (11.1) 111 (7.0) 112 (4.9)
Smoking status (Y/N) 6/7* 3/11 0/16 1/15
Handedness (R/L) 13/1 15/0 11/5 15/1
BPRS – – 26 (2.1) 27 (2.8)
Age at onset (yr) 28 (11.1) 20 (8.0)# – –
Illness duration (yr) 17 (8.8) 21 (12.1) – –
MRS 2 (3.6) 3 (3.8) – –
HAMD 6 (5.6) 8 (6.9) – –
PANSS
Total 44 (9.3) 44 (10.0) – –
Positive 11 (3.7) 11 (3.9) – –
Negative 10 (2.8) 9 (1.9) – –
General 23 (5.3) 24 (5.2) – –
Sodium valproate (Y/N) 8/6 (57%) 6/9 (40%) – –
Antipsychotics (Y/N) 10/4 (71%) 5/10 (33%)# – –
Lithium (Y/N) 3/11 (21%) 4/11 (27%) – –
Antidepressants (Y/N) 3/11 (21%) 8/7 (53%) – –
BPRS, Brief Psychiatric Rating Scale ; HAMD,Hamilton Depression Rating Scale ; MRS, Mania Rating Scale ; PANSS, Positive and
Negative Syndrome Scale.
Data represent mean (S.D.) except for smoking status and medication frequencies.
Age at onset was defined as the first appearance of symptoms. Predicted IQ was determined using the National Adult Reading
Test (NART). Smoking status (Yes/No) represents the number of participants that smoke regularly (i.e. every day), status was
not recorded for one male and one female participant with bipolar disorder. The medication frequencies (Yes/No) indicate the
number and percentage of participants within that group that were prescribed that medication.
* p<0.05, compared to controls ; # p<0.05, compared to males in the same group.
Gender differences in PPI in bipolar disorder 1251
underwent a hearing test indicating that they could
hear o40 dB. EMG activity was amplified (full scale
0.25), band-pass filtered (10–1000 Hz) and recorded
for 250 ms following stimulus onset (sample interval
1 ms). The startle amplitude output was in mV, where
1 mV=0.1 mV at the electrode.
The PPI task took 20 min to complete, during which
time participants were seated comfortably, instructed
to minimize movement, relax, ignore the sounds and
keep their eyes open during the task. The task began
with 3 min of 70-dB white noise (acclimation period),
which continued throughout the session. The partici-
pant was then presented with 67 acoustic stimuli in a
pseudo-random order at variable intervals (12–24 s,
average 16.5 s). There were 21 pulse-alone trials, i.e.
a startling pulse of 115 dB, presented for 40 ms.
The pulse-alone trials were grouped in three blocks :
(1) the first seven trials of the session, (2) seven trials
Table 2. Medications
Sodium
valproate
Antipsychotics
(CPZeq) Lithium Antidepressants Anticonvulsants Anxiolytics
Male
1 – – – – – –
2 – 150 Zu – – – –
3 1700 – – – – –
4 2000 285 Ri – – – –
5 – 670 Qu – – – Diazepam 10
6 – – 1250 – – –
7 1000 402 Qu – – – Oxazepam 30
8 1500 536 Qu – Citalopram 20 – –
9 1500 – – Citalopram 10 – –
10 1500 385 Ri & CPZ 1350 – – –
11 – 202.5 Ol 1250 – – –
12 500 67.5 Ol – – – –
13 1500 100 Ar – – Lamotrigine 200 –
14 – 405 Ol – Moclobemide 600 – –
Mean 1400 320 1283 – – –
(S.D.) (457) (195) (58)
Female
1 – – – – – –
2 – – – – – –
3 – 200 Am – Venlafaxine 450 Lamotrigine 400 –
4 – – 1500 – – –
5 1500 270 Ol – – – –
6 – 400 Ar 900 – – –
7 – – 500 – Carbamazepine 200 –
8 – – – Escitalopram 80 Lamotrigine 250 –
9 1500 402 Qu – Citalopram 20 – –
10 1400 – – – – –
11 1000 – – Venlafaxine 450 – –
12 – 16.75 Qu 900 Fluoxetine 40 – –
13 – – – Fluvoxamine 100 Carbamazepine 800 –
14 1000 – – – – –
15 1000 – – Escitalopram 10,
Mirtazapine 20
– –
Mean 1233 258 950 – – –
(S.D.) (258) (160) (412)
Am, Amisulpride ; Ar, aripiprazole ; CPZ, chlorpromazine ; Ol, olanzapine ; Qu, quetiapine ; Ri, risperidone ; Zu, zuclopenthixol.
Data represent dose (mg/day) of the different medications prescribed to the participants with bipolar disorder, except for the
antipsychotics which are expressed in chlorpromazine equivalents (CPZeq).
1252 A. Gogos et al.
interspersed with the prepulse-pulse trials, (3) the last
seven trials. There were 42 prepulse-pulse trials con-
sisting of a 20-ms prepulse of 74, 78 and 86 dB (4, 8 and
16 dB above background) followed 60 or 120 ms later
by the startling pulse of 115 dB. The startling pulse
and the prepulse-pulse trials involved bursts of white
noise with near instantaneous rise-fall times. The
mean of the seven trials per stimulus type was used in
data analysis. There were also four trials where no
sound was presented (control trials).
All EMG recordings were screened for spontaneous
eye-blink activity and trials were excluded if excessive
EMG activity occurred during 20-ms post-stimulus
(baseline shifted >90 mV). Nine percent of trials were
excluded for control participants and 10% of trials for
BD patients. Raw EMG data were smoothed: rolling
average 3, filter order 4, pass frequency 350, response
criterion 2. The magnitude of the startle response was
defined as the peak EMG response within 20–100 ms
post-stimulus. One participant was classified as a
‘non-responder’ and excluded from any further data
analysis as the average startle amplitude (of the three
blocks) was <50 mV when accounting for the control
trials. The following measurements of the startle re-
sponse were conducted: PPI (where %PPI is) :
amplitude of block 2
pulse-alone trials
� �x
amplitude of
prepulse-pulse trials
� �
amplitude of block 2 pulse-alone trialsr100,
startle reactivity (amplitude of block 1 pulse-alone
trials), startle amplitude (amplitude of blocks 1–3),
habituation (decrease in the pulse-alone amplitudes
across blocks 1–3).
Data analysis
Group and gender differences in terms of demo-
graphic and clinical factors were examined using
Pearson’s x2, one-way ANOVA, independent-sample
t test or Mann–Whitney U test, as required using
SPSS software (SPSS Inc., USA). A p value of <0.05
was considered significant, where applicable the
Greenhouse–Geisser corrected p value was instead
reported.
According to the Kolmogorov–Smirnov statistic,
PPI data were normally distributed and therefore
analysed with a repeated-measures ANOVA. For PPI
data, the main effects and interactions were analysed
for group (control, BD), gender (male, female), stimu-
lus onset asynchrony (SOA; 60 or 120 ms) and pre-
pulse intensity (PP4, PP8, PP16). Startle data (i.e. blocks
1–3) were highly positively skewed (Kolmogorov–
Smirnov test, p<0.01 for each block) and had a
significant heterogeneity of variance (Levene’s test,
p<0.05). After log-transformation (Csomor et al. 2008),
startle amplitudes were normally distributed and
there was homogeneity of variance. For graphical
purposes the non-transformed startle data was pre-
sented. The %PPI was based on non-transformed
startle data. For startle amplitude and habituation,
the main effects and interactions were analysed for
group, gender and block (1–3). The between-subjects
variables were gender and group, while the within-
subjects, repeated–measures variables refer to the PPI
protocol (i.e. SOA, prepulse intensity, block). When
the three participants with moderate to severe de-
pression were removed, the PPI findings did not
change direction, therefore they were retained in the
analysis. Further, when the analysis was repeated with
handedness as an additional between-subjects vari-
able, handedness did not interact with PPI, nor did the
PPI findings change when the seven left-handed par-
ticipants were removed.
Using Pearson’s r correlation, the relationship be-
tween average PPI scores (60 and 120 ms; i.e. the
average of the three prepulse intensities for 60 and
120 ms) and/or startle amplitude, and each demo-
graphic and clinical variable, was examined across the
four groups individually. These variables included
BPRS, age, predicted IQ, age of onset, illness duration,
PANSS (total and subscales), MRS and HAMD. Due
to the large number of comparisons, the Bonferroni
correction was applied, therefore p<0.002 was con-
sidered significant. Correlations with medication
and PPI were only completed with the BD partici-
pants that were actively taking the medication of
interest (Table 2). Pearson’s r correlations were per-
formed on the mean dosage of chlorpromazine equi-
valents (CPZeq), sodium valproate and lithium, and
Spearman’s rho correlations were performed on the
medication frequencies for antipsychotics, antide-
pressants, sodium valproate and lithium. Similarly to
Barrett et al. (2005), we performed exploratory com-
parisons between patients being prescribed a certain
medication or not, and whether this changed PPI
levels. Based on previous literature suggesting that
smoking can influence PPI (Della Casa et al. 1998 ;
Postma et al. 2006), the effect of smoking status on PPI
was examined using Spearman’s rho correlation and
also by using smoking status as a covariate.
Results
Demographic variables (Tables 1 and 2)
Using one-way ANOVA, the four groups were not
significantly different in terms of age and predicted
Gender differences in PPI in bipolar disorder 1253
full-scale IQ. There was a difference in the number
of smokers between the four groups [x2(3)=12.6,
p<0.01], which was significant when comparing male
BD and control participants [x2(1)=9.3, p<0.01], but
not female participants. Forty-six percent of male and
21% of female BD participants and 6% of female and
0% of male control participants were smokers. The
BPRS scores of the male and female control partici-
pants were not significantly different.
There were no significant gender differences in BD
patients in terms of illness duration, positive, negative,
general or total PANSS, MRS, HAMD, or mean dose of
medication (antipsychotic, sodium valproate, lithium).
There was a difference in medication frequency with
more male patients prescribed antipsychotics [x2(1)=4.2, p=0.04], but not for lithium, sodium valproate or
antidepressants. There was a significant difference in
terms of age of illness onset (U14,14=x2.1, p=0.04),
reflecting an earlier onset in women with BD (Table 1).
PPI
The 2 groupr2 genderr2 SOAr3 prepulse intensity
ANOVA revealed no main effect of group or gender.
There was a significant main effect of SOA (F1,57=71.4,
p<0.001) and prepulse intensity (F2,114=88.0, p<0.001), reflecting the expected increase in PPI (i.e.
greater reduction in startle) with increasing prepulse
intensity and increasing SOA (Fig. 1). There was an
SOArprepulse intensity interaction (F2,114=7.6, p<0.01), reflecting the increase in PPI, particularly at the
lower prepulse intensities, with increasing SOA. There
was an SOArprepulse intensityrgender interaction
(F2,114=5.3, p<0.01). Importantly, there was a signifi-
cant grouprSOA interaction (F1,57=8.8, p<0.01) and a
grouprgender interaction (F1,57=7.5, p<0.01), sug-
gesting that the groups exhibit different PPI depend-
ing on the SOA or gender, respectively.
In exploring the grouprgender interaction, a 2
groupr2 SOAr3 prepulse intensity ANOVA for
male participants demonstrated a trend for a group
effect (F1,28=3.7, p=0.06) and grouprSOA interaction
(F1,28=3.4, p=0.08), reflecting the tendency for BD
participants to have reduced PPI at the 60-ms SOA.
The ANOVA for female participants revealed a sig-
nificant grouprSOA interaction (F1,29=5.8, p=0.02)
and a trend for a group effect (F1,29=3.9, p=0.06),
reflecting the increased PPI in BD participants at the
120-ms SOA. Further, when comparing gender within
a group, there was a trend for a gender effect for con-
trols (F1,30=3.8, p=0.06) and BD (F1,27=3.6, p=0.07),
reflecting that female controls tended to have reduced
PPI compared to male controls, but male BD partici-
pants tended to have reduced PPI compared to female
BD participants, respectively (Fig. 2).
In exploring the grouprSOA interaction, a 2
groupr2 genderr3 prepulse intensity ANOVA at the
60-ms SOA demonstrated significant interactions of
grouprgender (F1,57=4.7, p=0.03) and prepulse in-
tensityrgender (F2,114=4.1, p=0.03). Examination of
the data (Fig. 2) suggests that male BD participants
had reduced PPI compared to male controls, while fe-
male control and BD participants showed similar PPI.
This observation was confirmed using post-hoc one-
way ANOVAs (male BD vs. male controls : F1,28=6.4,
80
40
0
–40
%P
PI
60 ms 120 ms
80
40
0
–40
60 ms 120 ms
Control: Males BD: Males
80
40
0
%P
PI
60 ms 120 ms
80
40
0
60 ms 120 ms
Control: Females BD: Females
SOA SOA
Fig. 1. Mean¡S.E.M. % prepulse inhibition (PPI) of healthy
control males (n=16) and females (n=16), and bipolar
disorder (BD) males (n=14) and females (n=15). %PPI is
shown with three levels of prepulse (PP) intensity [PP4 (%),
PP8 ( ) and PP16 (&) dB above 70 dB background] with
two levels of stimulus onset asynchrony (SOA) (60 and
120 ms).
60
40
20
0
Ave
rag
e %
PP
I
60 ms 120 ms
Males
*
60
40
20
060 ms 120 ms
Females
SOA SOA
*
Fig. 2. Mean¡S.E.M. average % prepulse inhibition (PPI) of
male and female participants that are healthy controls (%) or
participants with bipolar disorder (BD, &). Average %PPI is
the average of the three levels of prepulse intensity, and is
shown at two stimulus onset asynchrony (SOA) levels (60 and
120 ms) (* p<0.05 compared to controls).
1254 A. Gogos et al.
p=0.02 ; female BD vs. female controls : p=0.7).
ANOVA for the 120-ms SOA also demonstrated a
significant grouprgender interaction (F1,57=6.7, p=0.01). One-way ANOVA comparing male control and
BD participants revealed no significant main effect of
group (p=0.4). In contrast, when comparing female
participants, BD participants showed significantly
greater PPI than controls (F1,29=11.4, p<0.01).
Startle amplitude
Startle reactivity (Fig. 3), as measured by the magni-
tude of block 1 pulse-alone trials, was different across
the groups depending on gender (grouprgender
interaction : F1,57=4.3, p=0.04). When comparing male
participants, there was a significant group effect
(F1,28=8.0, p<0.01), reflecting the reduced startle re-
activity in men with BD compared to controls ; there
was no significant difference when comparing female
participants.
When analysing all three blocks of startle amplitude
data, there was a significant main effect of block
(F2,114=109.2, p<0.001), reflecting the reduction in
startle amplitude after repeated administration of the
startling pulse (i.e. habituation; Fig. 3). There was no
blockrgroup interaction, reflecting the similar habi-
tuation across groups. There was a grouprgender
interaction (F1,57=4.5, p=0.04) and a trend for a group
effect (F1,57=3.1, p=0.08). Of interest, when examining
block 2 (i.e. the startle pulses used to calculate PPI)
there was a grouprgender interaction (F1,57=5.6,
p=0.02), reflecting the reduced startle amplitude in
men with BD compared to male controls (F1,28=9.8,
p<0.01).
Given that the significant difference in startle
amplitude between male BD patients and controls
may have influenced PPI, we repeated the analysis on
a sample of participants with matched startle ampli-
tude: male BD patients (n=9, mean=781, S.E.M.=121)
and male controls (n=9, mean=778, S.E.M.=99). PPI
analysis revealed a significant main effect of group
(F1,16=5.4, p=0.03), which remained significant at the
60-ms SOA (F1,16=5.9, p=0.03), but not at 120 ms.
Clinical and demographic correlates
There was a trend for a positive correlation between
average PPI at the 60-ms and the 120-ms SOA in male
(r=0.69, p=0.003) and female controls (r=0.62, p=0.01), but not in BD. There was no correlation between
average PPI and startle data (either average startle,
block 2 original or log-transformed) in any of the four
groups, or between average startle and any demo-
graphic or clinical variable. In all participants there
were no significant correlations between PPI (at either
SOA) and demographic variables. In male BD partici-
pants there was a trend for a correlation between
60-ms average PPI and age of onset (r=0.67, p=0.009).
Importantly, there were no significant correlations
between PPI and symptoms (PANSS, HAMD, MRS) or
any medication (dosages : CPZeq, sodium valproate,
lithium; frequencies : antipsychotics, sodium valpro-
ate, lithium, antidepressants). When all BD patients
were combined, there was no difference in the effect of
each medication on PPI, for either medication type.
To examine the effects of smoking on PPI, smoking
status was added as a covariate to the PPI analyses.
There was no significant effect of smoking, nor did
it change the direction of our findings. The analyses
were repeated comparing men and women in both
groups for non-smokers only and the findings were
similar to the whole-group results. Using Spearman’s
rho correlation, we found that in female BD partici-
pants there was a negative correlation between 60-ms
average PPI and cigarette smoking (r=x0.58, p=0.03) ; however, this does not explain the PPI effect that
was observed in these participants at the 120-ms SOA.
3000
2000
1000
0
Am
plit
ud
e
3000
2000
1000
01 2 3
Am
plit
ud
e
ConM ConFBDM BDF
Startle reactivity
Startle habituation
Group
*
Block
Con MBD MCon FBD F
Fig. 3.Mean¡S.E.M. startle responses (mV) in the four groups :
healthy control males (Con M, n=16) and females (Con F,
n=16), and bipolar disorder (BD) males (BD M, n=14) and
females (BD F, n=15). Startle reactivity refers to the startle
response to the first block of seven 115 dB pulse-alone trials
(top panel). Startle habituation refers to the decrease in the
startle response to the pulse-alone trials across blocks 1, 2
and 3.
Gender differences in PPI in bipolar disorder 1255
Discussion
The present study examined gender differences in
PPI in BD. The main finding was that men and women
with BD both show changes in PPI compared to
healthy controls ; however, in the opposite direction.
Specifically, compared to controls, men with BD
showed reduced PPI, while women with BD had in-
creased PPI.
Our finding of disrupted PPI in men with BD was
consistent with previous literature showing a PPI dis-
ruption in BD (Giakoumaki et al. 2007 ; Perry et al.
2001). In both of these previous studies, the PPI dis-
ruption was greater at the 60-ms than the 120-ms SOA,
similar to the present results. The present finding also
parallels that of patients with schizophrenia, where
men with schizophrenia have reduced PPI compared
to male controls (Braff et al. 2001; Kumari et al. 2004).
A PPI disruption has been suggested to be related to
some symptoms of mental illness, such as psychosis,
mania or thought disturbance (Perry & Braff, 1994;
Perry et al. 2001). However, in the present study and
the Giakoumaki et al. (2007) study, PPI was not cor-
related with psychopathology, although this was not
surprising as we recruited patients with low levels
of psychopathology. Not all the previous literature
supports the present findings ; Carroll et al. (2007) and
Barrett et al. (2005) found that there was no change in
PPI in BD compared to controls. This may be ex-
plained by a number of methodological and sample
differences. For example, the Carroll et al. (2007) study
used a low-intensity startling pulse (95 dB) and a long
duration prepulse (50 ms) which may have caused a
floor effect in their study; they also tested sympto-
matic patients. The Barrett et al. (2005) study measured
PPI using the left eye, while the majority of PPI re-
search tests the right eye. Recently, it has been re-
ported that PPI can be affected by startle magnitude,
with a reduction in startle amplitude resulting in
an increase in PPI (Csomor et al. 2008). However, this
does not appear to be a confounding factor in the
present study as male BD patients showed reduced
startle amplitude and reduced PPI. Further, when
male participants were matched for startle, there was
still a significant disruption of PPI in male BD patients
compared to male controls.
Strikingly, the opposite effect was observed for fe-
male participants, where women with BD showed in-
creased PPI compared to female controls. This finding
was not supported by the previous PPI literature in
BD, where the majority of participants have been men
(Perry et al. 2001) or there was an equal mix of men
and women (Barrett et al. 2005 ; Carroll et al. 2007;
Giakoumaki et al. 2007). It is possible that in some
studies the decrease in PPI in men with BD may have
predominated, while in the other studies an increase
in PPI in women was counterbalanced by a decrease in
PPI in men with the net effect being no change in PPI.
However, the Giakoumaki et al. (2007) study used
gender as a factor in their statistical analyses, but
found no significant effects of gender. The previous
studies of PPI in BD did not report the stage of the
menstrual cycle the women were in when tested for
PPI. Given that PPI can vary in women across the
menstrual cycle (Jovanovic et al. 2004 ; Swerdlow et al.
1997), testing women at various stages may also
lead to a net effect of no change in PPI. In the present
study, both women with BD and control women were
tested at a stage with low circulating oestrogen and
progesterone. Kumari et al. (2004) speculated that
the gender difference observed in their PPI study of
schizophrenia patients may have been because the
women were less symptomatic than the men. In the
present study, there was no difference between
the men and women on any of the symptom ratings.
However, it is possible, that a difference in the fre-
quency or history of psychotic episodes may influence
PPI. The two studies showing a disruption of PPI in
BD tested 100% acutely psychotic patients (Perry et al.
2001) or 50% of patients with a history of psychotic
symptoms (Giakoumaki et al. 2007). In contrast, the
two studies showing no change in PPI in BD were
Carroll et al. (2007), who found no relationship be-
tween psychosis and PPI, and Barrett et al. (2005),
who did not report on psychotic features. The present
study found no correlation between the Positive
PANSS subscale and PPI in BD patients. Future stu-
dies should, however, specifically examine the pres-
ence of psychotic features (past and present) and their
influence on PPI.
As was also speculated by Kumari et al. (2004),
a simple explanation for the increased PPI in female
BD patients was that the control women showed rela-
tively low levels of PPI compared to other studies.
However, within the present study the four groups
had similar PPI levels, i.e. within a close range.
Further, the increase observed in female BD patients
was even greater than the male participants (average
PPI at 120-ms SOA: female BD 49%, male control 42%,
male BD 35%), where males generally have greater
PPI than females (Swerdlow et al. 1993). Alternatively,
medication may have had a greater effect in women
with BD. Medication can influence PPI ; some studies,
but not all, have found that antipsychotics can nor-
malize PPI in schizophrenia patients (Braff et al. 2001).
In the present study, it is unlikely that antipsychotic
1256 A. Gogos et al.
medication is responsible for the PPI increase in fe-
male BD patients, as only a small number of women
were prescribed an antipsychotic (5/15 patients).
Further, the majority of our male BD patients were
prescribed antipsychotics (10/14 patients) and they
still showed a disruption of PPI. Barrett et al. (2005)
suggested that BD patients prescribed sodium val-
proate (n=6) had reduced PPI compared to those who
were not taking sodium valproate (n=17). Our results
did not confirm this finding, where we had a fairly
equal distribution of patients prescribed vs. not pre-
scribed sodium valproate (Table 1). In our sample,
most patients were on multiple and various medica-
tions (Table 2), thus making it very difficult to extract
the effect of medication on PPI. Overall, the present
study found no correlation between medication and
PPI. Future research in drug-naive BD patients should
investigate the role of medication on PPI. Another
possible limitation of the present results is that we
did not match participants for smoking status, which
may influence PPI (Della Casa et al. 1998 ; Postma et al.
2006). Although there were more male smokers, fur-
ther analysis revealed no significant effect of smoking
status on PPI.
An increase in PPI in humans has been seen in a
number of studies, e.g. after treatment with nicotine
(Kumari et al. 1997), the indirect dopamine receptor
agonist, amantadine (Swerdlow et al. 2002), the
NMDA receptor antagonist, ketamine (Abel et al.
2003 ; Duncan et al. 2001 ; Heekeren et al. 2007), the
serotonin (5-HT) releaser, MDMA (Liechti et al.
2001 ; Vollenweider et al. 1999), the 5-HT2A receptor
agonist, psilocybin (Gouzoulis-Mayfrank et al. 1998 ;
Vollenweider et al. 2007) and by atypical antipsy-
chotics including quetiapine (Swerdlow et al. 2006b)
and clozapine (Vollenweider et al. 2006). Of these
studies, few included female participants and only one
was gender-balanced (Vollenweider et al. 2007). In the
latter study, psilocybin treatment increased PPI at a
long SOA (120 ms), but a PPI reduction occurred at a
short SOA (30 ms). Thus, it is possible that the 5-HT2A
receptor may be mediating the PPI increase in BD
women observed in the present study. There is some
evidence for a role of 5-HT2A receptors in BD (Lopez-
Figueroa et al. 2004 ; Pandey et al. 2003; Ranade et al.
2003). Other 5-HT receptor subtypes or other neuro-
transmitters may also be involved and therefore the
mechanism by which a PPI increase occurred remains
unknown and warrants further investigation.
In the present study, the PPI disruption in men with
BD occurred only at the 60-ms SOA, while the PPI in-
crease in women with BD occurred at the 120-ms SOA.
Using different SOAs can result in different effects on
PPI, with the optimal SOA (maximum inhibition) typ-
ically used in human PPI experiments being 120 ms
(Braff et al. 2001). While PPI is generally thought of as
an automatic process, at an SOA greater than 120 ms
attentional influences may play a role (Braff et al. 2001 ;
Filion et al. 1993). Future studies examining gender
differences in BD are needed whereby attention to
prepulses is controlled. Further investigation is also
required in order to clarify the different roles of vary-
ing SOAs on PPI.
In conclusion, the present results showed reduced
PPI in men and increased PPI in women with BD,
compared to controls. Although we have proposed a
number of mechanisms underlying these opposite
effects on PPI, further studies are required to clarify
which of these, if any, is responsible. The fact that men
and women with BD showed opposite effects in PPI
has implications for the interpretation of past and
future BD studies. In particular, combining men and
women in one group might be masking two opposite
effects and this highlights the importance of acknowl-
edging gender differences in PPI studies.
Acknowledgements
The authors gratefully acknowledge the financial
support of the J. & P. Clemenger Trust and the
National Health and Medical Research Council of
Australia (A. Gogos, Peter Doherty Fellowship, ID
435690). This research was also supported by Oper-
ational Infrastructure Support (OIS) from the Victorian
State Government. We acknowledge the assistance of
Alison O’Regan in administering the SCID-PANSS
and Nicole Joshua for assistance with recruitment.
Statement of Interest
None.
References
Abel KM, Allin MP, Hemsley DR, Geyer MA (2003). Low
dose ketamine increases prepulse inhibition in healthy
men. Neuropharmacology 44, 729–737.
Barrett SL, Kelly C, Watson DR, Bell R, et al. (2005). Normal
levels of prepulse inhibition in the euthymic phase of
bipolar disorder. Psychological Medicine 35, 1737–1746.
Bech P, Bolwig TG, Kramp P, Rafaelsen OJ (1979). The
Bech-Rafaelsen Mania Scale and the Hamilton Depression
Scale. Acta Psychiatrica Scandinavica 59, 420–430.
Berns GS, Nemeroff CB (2003). The neurobiology of bipolar
disorder. American Journal of Medical Genetics 123, 76–84.
Bethea CL, Pecins-Thompson M, Schutzer WE, Gundlah C,
et al. (1998). Ovarian steroids and serotonin neural
function. Molecular Neurobiology 18, 87–123.
Gender differences in PPI in bipolar disorder 1257
Braff DL, Geyer MA, Swerdlow NR (2001). Human studies
of prepulse inhibition of startle : Normal subjects, patient
groups, and pharmacological studies. Psychopharmacology
156, 234–258.
Braff DL, Light GA, Ellwanger J, Sprock J, et al. (2005).
Female schizophrenia patients have prepulse inhibition
deficits. Biological Psychiatry 57, 817–820.
Carroll CA, Vohs JL, O’Donnell BF, Shekhar A, et al. (2007).
Sensorimotor gating in manic and mixed episode bipolar
disorder. Bipolar Disorders 9, 221–229.
Csomor PA, Yee BK, Vollenweider FX, Feldon J, et al.
(2008). On the influence of baseline startle reactivity on
the indexation of prepulse inhibition. Behavioral
Neuroscience 122, 885–900.
Curtis V (2005). Women are not the same as men: specific
clinical issues for female patients with bipolar disorder.
Bipolar Disorders 7 (Suppl. 1), 16–24.
Della Casa V, Hofer I, Weiner I, Feldon J (1998). The effects
of smoking on acoustic prepulse inhibition in healthy men
and women. Psychopharmacology (Berlin) 137, 362–368.
Duncan EJ, Madonick SH, Parwani A, Angrist B, et al.
(2001). Clinical and sensorimotor gating effects of
ketamine in normals. Neuropsychopharmacology 25, 72–83.
Filion DL, Dawson ME, Schell AM (1993). Modification of
the acoustic startle-reflex eyeblink : a tool for investigating
early and late attentional processes. Biological Psychology
35, 185–200.
First MB, Spitzer RL, Gibbon M, Williams JBW (1997).
User’s Guide for the Structured Clinical Interview for DSM-IV
Axis I Disorders, Clinician Version (SCID-I). Washington DC:
American Psychiatric Press.
Giakoumaki SG, Roussos P, Rogdaki M, Karli C, et al.
(2007). Evidence of disrupted prepulse inhibition in
unaffected siblings of bipolar disorder patients. Biological
Psychiatry 62, 1418–1422.
Gogos A, Nathan PJ, Guille V, Croft RJ, et al. (2006).
Estrogen prevents 5-HT1A receptor-induced disruptions
of prepulse inhibition in healthy women.
Neuropsychopharmacology 31, 885–889.
Gouzoulis-Mayfrank E, Heekeren K, Thelen B,
Lindenblatt H, et al. (1998). Effects of the hallucinogen
psilocybin on habituation and prepulse inhibition of the
startle reflex in humans. Behavioral Pharmacology 9,
561–566.
Hamilton M (1960). A rating scale for depression. Journal of
Neurology, Neurosurgery & Psychiatry 23, 56–62.
Heekeren K, Neukirch A, Daumann J, Stoll M, et al. (2007).
Prepulse inhibition of the startle reflex and its attentional
modulation in the human S-ketamine and N,N-
dimethyltryptamine (DMT) models of psychosis. Journal of
Psychopharmacology 21, 312–320.
Hoenig K, Hochrein A, Quednow BB, Maier W, et al. (2005).
Impaired prepulse inhibition of acoustic startle in
obsessive-compulsive disorder. Biological Psychiatry 57,
1153–1158.
Jovanovic T, Szilagyi S, Chakravorty S, Fiallos AM, et al.
(2004). Menstrual cycle phase effects on prepulse inhibition
of acoustic startle. Psychophysiology 41, 401–406.
Kennedy N, Boydell J, Kalidindi S, Fearon P, et al. (2005).
Gender differences in incidence and age at onset of
mania and bipolar disorder over a 35-year period in
Camberwell, England. American Journal of Psychiatry 162,
257–262.
Kumari V, Aasen I, Sharma T (2004). Sex differences in
prepulse inhibition deficits in chronic schizophrenia.
Schizophrenia Research 69, 219–235.
Kumari V, Cotter PA, Checkley SA, Gray JA (1997). Effect
of acute subcutaneous nicotine on prepulse inhibition of
the acoustic startle reflex in healthy male non-smokers.
Psychopharmacology (Berlin) 132, 389–395.
Leibenluft E (1996). Women with bipolar illness : clinical and
research issues. American Journal of Psychiatry 153, 163–173.
Liechti ME, Geyer MA, Hell D, Vollenweider FX (2001).
Effects of MDMA (ecstasy) on prepulse inhibition and
habituation of startle in humans after pretreatment with
citalopram, haloperidol, or ketanserin.
Neuropsychopharmacology 24, 240–252.
Lopez-Figueroa AL, Norton CS, Lopez-Figueroa MO,
Armellini-Dodel D, et al. (2004). Serotonin 5-HT1A,
5-HT1B, and 5-HT2A receptor mRNA expression in
subjects with major depression, bipolar disorder, and
schizophrenia. Biological Psychiatry 55, 225–233.
McEwen BS (2001). Invited review: estrogens effects on the
brain : multiple sites and molecular mechanisms. Journal
of Applied Physiology 91, 2785–2801.
Nelson HE (1982). The National Adult Reading Test (NART).
Windsor : NFER.
Oldfield RC (1971). The assessment and analysis of
handedness : the Edinburgh inventory. Neuropsychologia
9, 97–113.
Pandey GN, Pandey SC, Ren X, Dwivedi Y, et al. (2003).
Serotonin receptors in platelets of bipolar and
schizoaffective patients : effect of lithium treatment.
Psychopharmacology (Berlin) 170, 115–123.
Perry W, Braff DL (1994). Information-processing deficits
and thought disorder in schizophrenia. American Journal
of Psychiatry 151, 363–367.
Perry W, Minassian A, Feifel D, Braff DL (2001).
Sensorimotor gating deficits in bipolar disorder patients
with acute psychotic mania. Biological Psychiatry 50,
418–424.
Postma P, Gray JA, Sharma T, Geyer M, et al. (2006). A
behavioural and functional neuroimaging investigation
into the effects of nicotine on sensorimotor gating in
healthy subjects and persons with schizophrenia.
Psychopharmacology (Berlin) 184, 589–599.
Ranade SS, Mansour H, Wood J, Chowdari KV, et al. (2003).
Linkage and association between serotonin-2A receptor
gene polymorphisms and bipolar I disorder. American
Journal of Medical Genetics 121B, 28–34.
Rich BA, Vinton D, Grillon C, Bhangoo RK, et al. (2005).
An investigation of prepulse inhibition in pediatric
bipolar disorder. Bipolar Disorders 7, 198–203.
Rubinow DR, Schmidt PJ, Roca CA (1998).
Estrogen–serotonin interactions : implications for affective
regulation. Biological Psychiatry 44, 839–850.
1258 A. Gogos et al.
Swerdlow NR, Auerbach P, Monroe SM, Hartston H, et al.
(1993). Men are more inhibited than women by weak
prepulses. Biological Psychiatry 34, 253–260.
Swerdlow NR, Geyer MA, Braff DL (2001). Neural circuit
regulation of prepulse inhibition of startle in the rat :
current knowledge and future challenges.
Psychopharmacology 156, 194–215.
SwerdlowNR, Hartman PL, Auerbach PP (1997). Changes in
sensorimotor inhibition across the menstrual cycle :
implications for neuropsychiatric disorders. Biological
Psychiatry 41, 452–460.
Swerdlow NR, Light GA, Cadenhead KS, Sprock J, et al.
(2006a). Startle gating deficits in a large cohort of patients
with schizophrenia : relationship to medications,
symptoms, neurocognition, and level of function.
Archives of General Psychiatry 63, 1325–1335.
Swerdlow NR, Paulsen J, Braff DL, Butters N, et al.
(1995). Impaired prepulse inhibition of acoustic and
tactile startle response in patients with Huntington’s
disease. Journal of Neurology, Neurosurgery & Psychiatry
58, 192–200.
Swerdlow NR, Stephany N, Shoemaker JM, Ross L, et al.
(2002). Effects of amantadine and bromocriptine on
startle and sensorimotor gating : parametric studies and
cross-species comparisons. Psychopharmacology (Berlin) 164,
82–92.
Swerdlow NR, Talledo J, Sutherland AN, Nagy D, et al.
(2006b). Antipsychotic effects on prepulse inhibition in
normal ‘low gating’ humans and rats.
Neuropsychopharmacology 31, 2011–2021.
Ventura J, Green MF, Shaner A, Liberman RP (1993).
Training and quality assurance with the Brief Psychiatric
Rating Scale : ‘The drift busters’. International Journal of
Methods in Psychiatry Research 3, 221–244.
Vollenweider FX, Barro M, Csomor PA, Feldon J (2006).
Clozapine enhances prepulse inhibition in healthy
humans with low but not with high prepulse inhibition
levels. Biological Psychiatry 60, 597–603.
Vollenweider FX, Csomor PA, Knappe B, Geyer MA, et al.
(2007). The effects of the preferential 5-HT2A agonist
psilocybin on prepulse inhibition of startle in healthy
human volunteers depend on interstimulus interval.
Neuropsychopharmacology 32, 1876–1887.
Vollenweider FX, Remensberger S, Hell D, Geyer MA
(1999). Opposite effects of 3,4-
methylenedioxymethamphetamine (MDMA) on
sensorimotor gating in rats versus healthy humans.
Psychopharmacology (Berlin) 143, 365–372.
Gender differences in PPI in bipolar disorder 1259