Effects of Clozapine on Auditory Event-Related
Potentials in Schizophrenia
Daniel Umbricht, Daniel Javitt, Gerald Novak, John Bates, Simcha Pollack,
Jeffrey Lieberman, and John Kane
Background: Schizophrenia is associated with cognitive
deficits that are an intrinsic component of the disorder.
Clozapine is an atypical antipsychotic that is superior to
typical agents in the treatment of positive symptoms. The
degree to which clozapine ameliorates cognitive deficits,
however, is still controversial. Mismatch negativity
(MMN), N200 (N2), and P300 (P3) are cognitive eventrelated potentials (ERPs) that index preattentive (MMN)
and attention-dependent information processing (N2, P3)
and provide a measure of cognitive deficits associated
with schizophrenia. In schizophrenic patients deficient
generation of MMN, N2, and P3 has been observed,
suggesting impairments of discrete stages of information
processing.
Methods: This study investigates the effects of clozapine
treatment on MMN, N2, and P3 generation. Patients were
recruited from a haloperidol-controlled, double-blind
treatment study of clozapine in chronic schizophrenia.
ERPs were obtained at the beginning of the study and after
9 weeks (4 patients) and 16 weeks (13 patients) of
treatment.
Results: Clozapine treatment was associated with a significant increase of P3 amplitude, which was not observed
in the haloperidol group; however, clozapine treatment
did not affect deficits in MMN and N2.
Conclusions: These findings suggest that clozapine—in
contrast to conventional antipsychotics—improves electrophysiological measures of attention-dependent information processing, but does not ameliorate preattentive
deficits. Biol Psychiatry 1998;44:716 –725 © 1998 Society of Biological Psychiatry
Key Words: Event-related potentials, mismatch negativity, clozapine, atypical antipsychotics, schizophrenia,
treatment resistance
From the Research Department, Psychiatric University Hospital, Zurich, Switzerland (DU); Research Department, Hillside Hospital, Glen Oaks, New York
(DU, DJ, GN, JB, SP, JK); Nathan Kline Institute for Psychiatric Research,
Orangeburg, New York (DJ); St. John’s University, Jamaica, New York (SP);
and Department of Psychiatry, University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina (JL).
Address reprint requests to Daniel Umbricht, MD, Research Department, Psychiatric University Hospital, PO Box 68, Lenggstr.31, 8029 Zurich, Switzerland.
Received January 21, 1997; revised September 18, 1997; revised November 5,
1997; accepted November 12, 1997.
© 1998 Society of Biological Psychiatry
Introduction
C
ognitive deficits are a major cause of disability in
schizophrenia and an important hurdle in the rehabilitation of schizophrenic patients (Gold and Harvey 1993;
Green 1996). There are no proven treatments available for
this important aspect of schizophrenia. Clozapine, an
atypical antipsychotic, has been shown to be superior to
conventional agents in the treatment of positive symptoms
(Kane et al 1988, 1996; Breier et al 1994), yet it has not
been proven to have clinically significant effects on
cognitive deficits (Goldberg et al 1993; Hagger et al 1993;
Williams et al 1993; Buchanan et al 1994; Zahn et al 1994;
Lee et al 1994; Goldman et al 1996; Hoff et al 1996). This
study investigates the effects of clozapine on cognitive
deficits by utilizing the technique of auditory event-related
potentials (ERPs).
ERPs provide an objective index of cognitive dysfunction, and a reliable method for assessing effects of medication on underlying brain activity. Most ERP studies in
schizophrenia have focused on abnormalities in the generation of the P300 (P3) amplitude—a positive potential
that occurs with an approximate latency of 300 ms after
the presentation of a novel, behaviorally relevant target
stimulus embedded among irrelevant stimuli (McCarley et
al 1991, 1993; Pritchard 1986). Its amplitude depends,
among other factors, on overall probability of the deviant
stimulus and is believed to reflect allocation of attention
and activation of immediate memory (Johnson 1986;
Polich and Kok 1995). In recent years several studies have
also demonstrated abnormalities in the generation of
several potentials preceding P3 that index discrete stages
of information processing. They include mismatch negativity (MMN) (Shelley et al 1991; Oades et al 1993; Catts
et al 1995; Javitt et al 1993, 1995) and N200 (N2)
(O’Donnell et al 1993; Egan et al 1994). MMN is an early
cognitive potential, occurring with an approximate latency
of 100 –200 ms after the presentation of a stimulus that
deviates in any physical dimension (pitch, duration, location) from preceding stimuli. MMN is assumed to be the
manifestation of a comparison process that operates independently of the specific task and the subject’s attention
0006-3223/98/$19.00
PII S0006-3223(97)00524-6
Clozapine and ERPs in Schizophrenia
and motivation, but requires the presence of a sensory
memory trace of the standard stimulus with which the
deviant stimulus is compared (Näätänen 1990; Novak et al
1990; Ritter et al 1992). In contrast, N2, occurring
frontocentrally with an approximate latency of 200 ms,
is—like P3—an attention-dependent potential that is
thought to reflect stimulus categorization (Ritter et al
1979, 1983; Sams et al 1985). MMN, N2, and P3 thus
provide a sequence of potentials that index different
stages—preattentive and attention-dependent—in the processing of behaviorally relevant target stimuli (Novak et al
1990). Studies in schizophrenic patients have provided
some evidence that abnormalities in the early stages of
auditory information processing as evidenced in abnormal
MMN contribute to subsequent neurophysiological dysfunction, manifested in deficient N2 and P3 generation
(Javitt et al 1995).
Treatment with typical antipsychotics followed by
symptomatic remission is not associated with a normalization of deficient auditory ERPs in schizophrenic patients (Blackwood et al 1987; Duncan et al 1987; Eikmeier
et al 1992; Ford et al 1994a). Despite the fact that
clozapine has been in use for over two decades, its effects
on ERP abnormalities in schizophrenia have not been
investigated under controlled conditions. The goal of this
study was to determine whether clozapine would differ
from conventional antipsychotics, i.e., haloperidol, and
ameliorate deficits in preattentive and attention-dependent
information processing as evidenced in abnormalities of
MMN, N2, and P3 in schizophrenia. This question was
investigated in the context of a double-blind treatment
study, in which patients were switched from treatment
with conventional antipsychotics to treatment with either
clozapine or haloperidol and, additionally, with patients
whose treatment was changed to clozapine by their clinicians.
Methods and Materials
Study Design
ERPs were recorded in a group of normal controls (n 5 13), in
patients who had consented to participate in a double-blind
treatment study comparing clozapine to haloperidol (n 5 14),
and in patients who were assigned to open clozapine treatment by
their clinician (n 5 3). In patients ERPs were recorded on two
occasions: at baseline, and after 16 weeks (13 patients) and 9
weeks (4 patients), respectively. At baseline patients were treated
with conventional antipsychotics and other psychotropics (see
below) prescribed by patients’ clinicians. After the baseline
assessments all psychotropics (with the exception of lorazepam
on p.r.n. basis) were tapered off, and study medication or
open-label clozapine was started. Thus, the design of the study
assessed the effects of switching form treatment with conventional antipsychotics to treatment with clozapine or haloperidol.
BIOL PSYCHIATRY
1998;44:716 –725
717
In the normal control group, ERPs were obtained once.
To be eligible for the double-blind treatment study and for
open-label clozapine treatment, patients had to be partially
treatment refractory to treatment with conventional antipsychotics as evidenced in a rating of 4 (5 moderate) on any of the four
positive symptom items (conceptual disorganization, suspiciousness, hallucinatory behavior, unusual thought content) of the
Brief Psychiatric Rating Scale (BPRS, anchored version;
Woerner et al 1988). Patients had to meet the diagnostic criteria
of schizophrenia or schizoaffective disorder according to DSMIII-R established by a structured clinical interview (SCID;
Spitzer et al 1990) (n 5 15) or by a clinical interview and chart
review (n 5 2) performed by the first author (D.U.). All patients
provided written informed consent to participate in the ERP
study.
Patients
A total of 17 patients were studied. Nine patients were diagnosed
with chronic schizophrenia, undifferentiated type, 6 patients with
chronic schizophrenia, paranoid type, and 2 patients with chronic
schizophrenia, disorganized type. Patient characteristics are described in Table 1. At baseline patients received the following
medications in addition to neuroleptic treatment (in parentheses
are the numbers of patients who went on to receive clozapine or
haloperidol plus 4 mg of benztropine): anticholinergic medication 10 (6/4), chlomipramine 3 (1/2), lithium carbonate 3 (3/0),
and benzodiazepines 5 (4/1). The clozapine group comprised 11
patients; 6 patients were assigned to haloperidol. At follow-up
the mean doses of clozapine and haloperidol were 540 6 217 mg
and 20 6 8 mg, respectively. Patients in the two treatment groups
did not differ with regard to age, duration of illness, prior months
on neuroleptics, mean neuroleptic dose in chlorpromazine (CPZ)
equivalents, or total scores of BPRS, and SANS (Schedule for
the Assessment of Negative Symptoms; Andreasen 1982), and
CGI (Clinical Global Impression; Guy 1976) ratings at baseline
(Table 1).
Controls
Thirteen normal controls were recruited from staff at the Long
Island Jewish Medical Center (Table 1). Absence of any past or
present psychopathology was ensured by a clinical interview.
Normal controls were younger than the patients, but this difference did not reach statistical significance (t 5 2.02, df 5 28, p .
.5).
Clinical Ratings and Assessments
Symptomatology was assessed with the help of the BPRS,
SANS, and CGI. Behavioral ratings were performed within 5
days of the ERP recording session. The measures of psychopathology used for analysis were the BPRS total score, the BPRS
psychosis factor score, the SANS total score, and the CGI
severity of illness rating.
Information regarding past history was obtained through
patient interview and chart review. Writing preference was used
to assess handedness.
718
D. Umbricht et al
BIOL PSYCHIATRY
1998;44:716 –725
Table 1. Characteristics of Normal Controls and Patients
N (m/f)
Age (y)
Handedness (r/l)
Duration of Illness (y)
Age at first psychotic
hospitalization
Number of previous
hospitalizations
Neuroleptic Dose at Baseline
(in mg CPZ equivalents)
BPRS Total Score
BPRS Psychosis Factor
Score
SANS Total Score
CGI
Patients
Normal
Controls
Total Sample
Clozapine Group
Haloperidol Group
13 (8/5)
31.2 6 8.9
13/0
—
—
17 (15/2)
37.2 6 7.4
16/1
17.4 6 8.9
23.7 6 7.5
11 (11/0)
36.8 6 8.6
10/1
16.7 6 8.2
24.2 6 7.3
6 (4/2)
38.0 6 7.6
6/0
18.7 6 10.8
22.8 6 8.6
—
6.7 6 6.1
6.12 6 3.8
7.5 6 8.7
—
1026 6 761
855 6 638
1342 6 924
—
—
41.1 6 10.8
14.2 6 4.1
41.9 6 12.6
14.1 6 4.7
39.7 6 7.1
14.3 6 2.9
—
—
31.9 6 7.8
4.6 6 0.9
31.3 6 7.1
4.6 6 1.1
33.0 6 9.5
4.7 6 0.5
ERP Recording and Analysis
ERPs were recorded during an active and a “no response”
auditory “oddball” paradigm. Standard stimuli were 1000-Hz
tones, deviants were tones of 1200 Hz. Both standard and deviant
stimuli were of 50-ms duration with 5-ms rise/fall time, delivered
binaurally via headphones. Intensity of all stimuli was 85 dB
SPL. Interstimulus interval varied randomly between 700 and
800 ms. Deviants were randomly presented with an overall
probability of .14, with the exception that deviants were always
followed by one or more standards. In the active paradigm
subjects were asked to press a button after the presentation of a
deviant tone. Two blocks with 80 deviants each were presented.
In the “no response” paradigm subjects were watching a silent
movie and were told to ignore the tones. Four blocks with 110
deviants each were presented. The active paradigm always
preceded the “no response” condition.
ERPs were recorded using a 16-electrode montage [F3, Fz, F4,
left mastoid (LM), T3, C3, Cz, C4, T4, right mastoid (RM), T5,
P3, Pz, P4, T6, and one electrode attached above the left outer
canthus for monitoring of blinks and eye movements] referenced
to a nose electrode. Electroencephalograms (EEGs) were recorded with a Nihon Koden EEG amplifier (band-pass 0.1–70
Hz). EEGs were digitized with a sampling rate of 256 Hz and
saved with markers indexing the various stimuli. All data
processing was performed off-line using NEUROSCAN software. Averages of 1024 ms with a 100-msec prestimulus baseline
were constructed after automatic rejection of sweeps with potentials exceeding 6100 mV in any of the channels. Average waves
were obtained separately for standard and deviant stimuli. In the
active paradigm only sweeps associated with correctly detected
deviant stimuli were included in the averaging. A button press
within 200 –1000 ms poststimulus was considered a correct
response. Average waves were baseline corrected and digitally
filtered with a low-pass filter of 30 Hz (6 dB down). Since N1
and MMN are known to invert between Fz and mastoid leads,
MMN and N1 averages were rereferenced to a mathematically
computed average mastoid derivation for the purpose of peak
measurements. For the determination of MMN and N2, difference waves (ERPs to deviants minus ERPs to standards) were
used. N1 was determined from waveforms to standard stimuli in
the “no response” paradigm and defined as peak negativity at Fz
within the 50 –150-ms latency range. MMN was defined as the
peak negativity at Fz within the 100 –225-ms latency range in the
“no response” condition difference wave. N2 was defined as
peak negativity at Cz within the 150 –300-ms latency range in the
active condition difference wave. P3 was defined as the peak
positivity at Pz within a latency range of 225–550 ms in the ERPs
to deviant tones. For the measurement of MMN, the required
minimum of sweeps surviving artifact rejection was set at 100;
and for the measurement of N2 and P3, the required minimum of
sweeps surviving artifact rejection was set at 25. Automatic
latency and peak measurements were done by the NEUROSCAN
software.
The number of sweeps surviving artifact rejection in the
“attend” paradigm and usable for analyses of N2 and P3 waves
was too low in 2 patients at baseline and 2 patients at follow-up.
ERPs for the “attend” paradigm were therefore available for
analyses in 13 patients (9 in clozapine group; 4 in haloperidol
group). In the control group ERPs in the attend paradigm were
available in 10 subjects.
Statistical Test
The main analyses concerned baseline comparisons between the
patients and the normal controls and the effects of the two
treatments on ERP parameters and psychopathology. Baseline
comparisons between the patients and the normal controls and
between the two treatment groups, respectively, were performed
with the help of t tests for the four defined ERP components (N1,
MMN, N2, P3) and for the behavioral measures. Topographical
analyses were performed for MMN, N2, and P3 with the help of
repeated-measures analysis of variance (ANOVA) with group as
between-subject factor and electrode site as repeated measure
(within-subject factor). Drug effects on peak amplitude and
Clozapine and ERPs in Schizophrenia
BIOL PSYCHIATRY
1998;44:716 –725
719
Table 2. Amplitudes and Latencies of Mismatch Negativity (MMN), N2 and P3 of the Normal Controls and Patients at Baseline
and Follow-up
Patients
Normal
Controls
N1
(Fz)
MMN
(Fz)
N2
(Cz)
P3*
Fz
Cz
Pz
Amplitude
Latency
Amplitude
Latency
Amplitude
Latency
Amplitude
Amplitude
Amplitude
Latency
21.9 6 1.7
90 6 21
24.1 6 1.6
150 6 22
25.3 6 4.4
211 6 64
6.4 6 4.0
7.9 6 4.6
10.0 6 4.0
394 6 62
Total Sample
(n 5 17)
21.1 6 1.7
88 6 27
22.9 6 1.4
148 6 30
22.9 6 3.0
179 6 60
2.8 6 3.4
5.5 6 4.7
8.4 6 5.8
395 6 93
Clozapine Group (n 5 11)
Haloperidol Group (n 5 6)
Baseline
Follow-up
Baseline
Follow-up
21.5 6 1.9
85 6 31
22.7 6 1.2
141 6 33
23.5 6 3.2
189 6 59
1.9 6 3.1
4.2 6 4.3
7.4 6 6.1
400 6 93
20.9 6 1.5
102 6 21
22.4 6 1.6
159 6 38
24.1 6 4.8
200 6 83
4.1 6 3.3
7.6 6 4.5
9.4 6 4.8
464 6 65
20.4 6 1.0
94 6 22
23.1 6 1.8
161 6 18
21.6 6 2.5
158 6 63
4.7 6 3.6
8.3 6 4.7
10.6 6 5.3
383 6 105
20.6 6 1.5
97 6 22
22.7 6 1.3
148 6 23
24.8 6 7.1
155 6 40
4.3 6 2.8
7.4 6 4.2
7.8 6 5.34
445 6 78
Amplitudes in mVolts, latencies in ms; * 5 Normal Controls N 5 10, Clozapine Group N 5 9, Haloperidol Group N 5 4.
latency of the four defined ERP variables (MMN, N1, N2, P3)
and on behavioral measures were evaluated using repeatedmeasures ANOVAs with drug treatment as between-subject
factor and session as repeated measure. Because of the limited
number of patients relative to electrode positions, multivariate
analysis of treatment effects on ERP topography could not be
employed. Instead, univariate repeated-measures ANOVA with
Greenhouse–Geisser correction with the treatment group as
between-subject factor and session and electrode site as withinsubject factors was used. Alpha values of .05 were considered
significant. Values in text are mean 6 standard deviation.
Results
Schizophrenics versus Controls
TASK PERFORMANCE. Behaviorally, reaction time,
hit rate, and false alarm rate differed significantly between
patients and controls, with normal controls showing
shorter reaction times, higher hit rates, and lower false
alarm rates (reaction time: 512 6 87 ms versus 430 6 90
ms; t 5 2.2, df 5 21, p , .05; hit rate: 80.4 6 21.3%
versus 95.7 6 4.4%; t 5 22.5, df 5 13.36, p , .05; false
alarm rate: 1.2 6 1.4% versus 0.07 6 0.2%, t 5 2.9, df 5
12.4, p , .05).
N1 AND MMN. At baseline patients showed smaller
peak amplitudes of N1, but this difference was not
statistically significant (t 5 1.31, df 5 28, p 5 ns; Table
2). MMN amplitude at baseline was largest at Fz and
inverted between Fz and the mastoids in both the patient
and the normal control group (Figure 1). In the patient
group MMN amplitude at Fz was significantly smaller in
the patient group than in the normal control group (t 5
2.37, df 5 28, p , .05; Table 2, Figures 1 and 2).
Topographical comparison using a repeated-measures
ANOVA confirmed a significant effect of group [F(1,28)
5 5.97, p , .05], with patients showing smaller MMN,
but no group 3 electrode interaction [F(14,16) 5 1.82,
p 5 ns]. Peak latencies of N1 and MMN were not
significantly different between patients and normal controls (see Table 2).
N2. For all subject groups, N2 was the largest at Cz.
The mean amplitude of N2 at Cz was smaller in the
patients than the normal controls, but this difference did
not reach statistical significance (t 5 1.57, df 5 21, p 5
ns; Table 2). Topographical analyses did not show significant group effects or group 3 electrode interactions [main
effect: F(1,21) 5 0.99, p 5 ns; group 3 electrode
interaction F(3,49) 5 1.89, p 5 ns]. Peak latencies of N2
were not significantly different between patients and
normal controls (see Table 2).
P3. The mean amplitude of P3 at Pz was smaller in the
patients than the normal controls, but these differences did
not reach statistical significance (t 5 20.74, df 5 21, p 5
ns; Table 2). Topographical analyses did not show significant group effects or group 3 electrode interactions [main
effect: F(1,21) 5 1.38, p 5 ns; group 3 electrode
interaction: F(14,8) 5 0.97, p 5 ns]. Peak latencies P3
were not significantly different between patients and
normal controls (see Table 2).
Effects of Treatment on Symptomatology and ERP
Parameters
SYMPTOMATOLOGY AND TASK PERFORMANCE. In
the clozapine group the BPRS total score decreased by
6.3 6 9.5 points and the BPRS psychosis factor score by
3.9 6 3.8 points, whereas in the haloperidol group the
mean BPRS total score increased by 5.2 6 7.7 points and
the BPRS psychosis factor score by 0.5 6 3.9 points.
These differences were statistically significant [BPRS
total score: repeated-measures ANOVA, treatment
720
BIOL PSYCHIATRY
1998;44:716 –725
D. Umbricht et al
rate was not significantly affected by clozapine or haloperidol treatment [Hit rate: repeated-measures ANOVA
treatment group 3 session interaction: F(1,13) 5 0.24,
p 5 ns; False alarm rate: repeated-measures ANOVA
treatment group 3 session interaction: F(1,13) 5 0.51,
p 5 ns].
N1 AND MMN. Peak amplitudes and latencies of N1
and MMN did not differ between the two treatment groups
at baseline (Table 2 and Figures 1 and 3). Repeatedmeasures (session) ANOVAs did not reveal any significant differences in the effects of treatment with clozapine
or haloperidol on peak amplitude and latency of N1 at Fz
[amplitude: treatment 3 session interaction: F(1,15) 5
2.36, p 5 ns; latency: treatment group 3 session interaction: F(1,15) 5 1.10, p 5 ns] and MMN at Fz [amplitude:
treatment group 3 session interaction: F(1,15) 5 0.08,
p 5 ns; latency: treatment group 3 session interaction:
F(1,15) 5 2.80, p 5 ns].
Topographical analysis of the drug effects of MMN
amplitudes did not show any significant treatment
group 3 session [F(1,15) 5 1.51, p 5 ns] and treatment
group 3 session 3 electrode interactions [F(14,2) 5 1.77,
p 5 ns].
Figure 1. Mean MMN amplitudes at baseline and follow-up of
clozapine group, haloperidol group, and normal controls. Open
squares, clozapine group at baseline; filled squares, clozapine
group at follow-up; open circles, haloperidol group at baseline;
filled circles, haloperidol group at follow-up; triangles, normal
controls.
group 3 session interaction: F(1,15) 5 6.43, p 5 .02;
BPRS psychosis factor score: repeated-measures
ANOVA, treatment group 3 session interaction: F(1,15)
5 5.14, p , .05]. In the clozapine group mean BPRS total
and mean BPRS psychosis factor scores were significantly
lower at follow-up than at baseline (BPRS total score:
paired t test: t 5 2.2, df 5 10, p 5 .05; BPRS psychosis
factor score: paired t test: t 5 3.4, df 5 10, p , .01).
Clinical Global Impression Change scores were significantly greater for the clozapine than the haloperidol group
at follow-up (3.1 6 0.8 vs. 4.0 6 0.9; t 5 22.1, df 5 15,
p 5 .05). No significant changes in SANS total scores
were observed in any treatment group.
Task performance as measured by hit and false alarm
N2. Peak amplitudes and latencies of N2 did not differ
between the two treatment groups at baseline (Table 2 and
Figures 1 and 3). Repeated-measures (session) ANOVA
did not reveal any significant differences in the effects of
treatment with clozapine or haloperidol on peak amplitude
and latency of N2 at Cz [amplitude: treatment group 3
session interaction: F(1,11) 5 0.44, p 5 ns; latency:
treatment group 3 session interaction: F(1,11) 5 0.54,
p 5 ns; Table 2 and Figures 1 and 3]. Topographical
analyses of the drug effects on N2 amplitudes did not
show any significant treatment group 3 session [F(1,11)
5 0.25, p 5 ns] and treatment group 3 session 3
electrode interactions [F(2.13) 5 0.50, p 5 ns].
P3. Peak amplitudes and latencies of P3 did not differ
significantly between the two treatment groups at baseline
(Table 2 and Figures 1 and 3). Clozapine treatment was
associated with a significant increase of P3 amplitude at
Pz compared to the haloperidol group [repeated-measures
(session) ANOVA drug 3 session interaction: F(1,11) 5
4.3, p 5 .03].
The topographical analysis of P3 revealed a significant
treatment group 3 session interaction [F(1,11) 5 11.31,
p , .01], but no treatment group 3 session 3 electrode
interaction [F(2.72) 5 0.82, p 5 ns], confirming an
overall enhancing effect of clozapine on P3 amplitude.
Neither clozapine nor haloperidol treatment was associated with an effect on P3 latency at Pz [repeated-measures
Clozapine and ERPs in Schizophrenia
BIOL PSYCHIATRY
1998;44:716 –725
721
Figure 2. Grand averages of MMN (difference wave) at baseline and follow-up of clozapine group, haloperidol group, and normal
controls at midline electrodes. Solid lines, clozapine group; dashed lines, haloperidol group.
(session) ANOVA drug 3 session interaction: F(1,11) 5
0.47, p 5 ns].
The increase of P3 amplitude at Pz in the clozapine
group was not significantly correlated with change of
BPRS psychosis score (Figure 4).
Discussion
Despite its clinical use for over 30 years, the effects of
clozapine on brain function as measured by ERPs have not
been studied under controlled conditions. This study is the
first to attempt this, investigating the effects of clozapine
on ERP measures of neurocognitive deficits in a doubleblind, haloperidol-controlled study in partially treatmentrefractory schizophrenic subjects. In the present study
clozapine significantly improved electrophysiological
measures of attention-dependent information processing
(P3), but failed to alter deficits in preattentive information
processing as indexed by MMN. Clozapine treatment led
to an increase in P3 amplitudes, but did not reverse the
deficient MMN generation in schizophrenic patients. In
addition, clozapine treatment was associated with a significant reduction of positive symptoms and clearly
showed its superiority over haloperidol even in this small
patient sample.
The observed effect of clozapine on auditory P3 generation differs from those of typical antipsychotics, since it
has been well documented that treatment with standard
antipsychotics does not normalize P3 deficits in schizophrenic patients (Blackwood et al 1987; Eikmeier et al
1992; Ford et al 1994a). The results of this study are in
agreement with a brief report by Schall et al (1995) on a
study comparing the effects of clozapine on auditory ERPs
to those of haloperidol in an uncontrolled, open study.
Schall et al also observed a significant increase of P3
amplitude in clozapine-treated patients, but failed to find
an effect of clozapine on MMN. Taken together, the
results of this study and those reported by Schall et al
(1995) support the concept that clozapine has actions in
schizophrenia that are qualitatively different from those of
typical antipsychotics. There are different possible explanations for the observed effect of clozapine on P3 generation. P3 depends, among other factors, on attentional
functions. First, the increase of P3 by clozapine might thus
be the “top-down” effect of clozapine-induced improvements in attention. In agreement with such a view are the
reports of a small, but significant positive effect of
long-term treatment clozapine on the performance of the
Digit Symbol Test—a measure of attention (Hagger et al
1993; Lee et al 1994); however, other studies using this
722
BIOL PSYCHIATRY
1998;44:716 –725
D. Umbricht et al
Figure 3. P3 amplitudes (grand averages) at baseline and follow-up of clozapine group, haloperidol group, and normal controls at
midline electrodes. Solid lines, clozapine group; dashed lines, haloperidol group.
specific test did not detect a significant improvement
during clozapine treatment (Goldberg et al 1993). Second,
since P3 also reflects the processing of relevant stimuli in
specific brain areas, the observed increase of P3 amplitude
might be the direct manifestation of an improved functioning of these brain areas, secondarily resulting in the
improvement of certain cognitive functions. A third possibility is given by results of a study by Ford et al (1994b),
who conducted a single-trial analysis of P3 generation in
schizophrenic patients and found that schizophrenic patients showed P3 waves to fewer trials to which they had
correctly responded and smaller P3 amplitude on each
trial. The authors suggested that patients show intermittent
waning of attentional engagement, which might result in a
disengagement of P3 generation from the motor response
process. It is possible that the observed effect of clozapine
could be the result of patients remaining more engaged
during the task, thus simply generating more P3 waves
without any changes of P3 amplitude. Future studies will
have to answer these questions.
The observation that clozapine treatment not only led to
a significant reduction of positive symptoms, but also
improved auditory P3 as a measure of attention-dependent
information processing is only partly consistent with the
results seen in studies that investigated the effect of
clozapine on a wide range of neuropsychological deficits
(Goldberg et al 1993; Hagger et al 1993; Williams et al
1993; Buchanan et al 1994; Zahn et al 1994; Lee et al
1994; Goldman et al 1996; Hoff et al 1996). Goldberg et
al (1993), investigating the effects of clozapine in an
open-label treatment study, did not find an ameliorative
effect of clozapine on deficits in attention, memory, and
higher-level problem solving in 13 patients on long-term
clozapine treatment. Zahn et al (1994) also failed to show
any beneficial effects of clozapine—in comparison with
fluphenazine treatment and placebo— on measures of
sustained and selective attention in a reaction time paradigm. In both studies, however, patients showed significant reductions of positive symptoms despite the absence
of neuropsychological improvement. In other studies
(Haggre et al 1993; Lee et al 1994; Buchanan et al 1994;
Hoff et al 1996) clozapine was found to improve performance on tests of retrieval from reference memory,
measures of attention, verbal and category fluency and
perceptual discrimination, whereas on other tests such as
the Wisconsin Card Sorting Test no improvement was
seen; however, the observed improvements were modest
and much smaller than the effects of clozapine on positive
symptoms. In the context of the findings in this study, it is
of interest that some of the studies—as mentioned
Clozapine and ERPs in Schizophrenia
Figure 4. Mean P3 amplitudes at baseline and follow-up of
clozapine group, haloperidol group, and normal controls. Open
squares, clozapine group at baseline; filled squares, clozapine
group at follow-up; open circles, haloperidol group at baseline;
filled circles, haloperidol group at follow-up; triangles, normal
controls.
above—found improvement on measures of attention
(Hagger et al 1993; Lee et al 1994) and on tests of the
ability to concentrate (Hoff et al 1996). In summary, these
studies of the effects of clozapine on neurocognition have
not yielded consistent results. The results of this study thus
suggest that normalization of certain brain activity may
occur during clozapine therapy, but be poorly detected by
routine neuropsychological test batteries and difficult to
define with traditional instruments. It is of interest that in
this study symptomatic improvement did not correlate
with ERP measures of attention. This is in agreement with
the results of all neuropsychological studies, even those
that reported significant effects on some aspects of cognition. None of them found a correlation of changes of
neuropsychological test performance with symptomatic
BIOL PSYCHIATRY
1998;44:716 –725
723
improvement. The finding of this study thus supports the
notion that neuropsychological deficits and positive symptomatology are independent areas of schizophrenic pathology.
In contrast to its effect on P3, clozapine treatment did
not alter deficient MMN generation, indicating that it does
not affect deficits at more basic levels of information
processing in schizophrenic patients. The question to what
extent the failure of clozapine to affect these abnormalities
relates to its lacking effect on deficit symptoms (Breier et
al 1994; Kane et al 1996), and to the negative findings in
the neuropsychological studies cited above, remains open
and an issue for further research.
This study is limited by several factors. The number of
subjects studied was small, reducing the power to detect
small differences and changes. In addition, at baseline
patients could not be studied in a medication-free condition. Thus, the observed increase of P3 amplitude in the
clozapine group could be the result of the taper of
conventional antipsychotics; however, Ford et al (1994a)
observed a slight decrease of P3 amplitudes after a 1 week
washout of conventional antipsychotics, suggesting that
P3 deficits, if anything, become more pronounced in a
medication-free condition. A third limitation is the lack of
follow-up data in the normal control group, which would
have allowed the evaluation of effects of retesting on
ERPs in the analyses.
In conclusion, the results of our study indicate that
clozapine, in contrast to conventional antipsychotics, may
improve certain deficits in attention-dependent functions
as manifested in reduced P3 amplitude, but lacks a distinct
and unique effect on ERP indices of preattentive information processing in schizophrenia. Given the small number
of subjects in our study, these findings require replication
in a larger sample.
This study was supported by a grant from the National Alliance for
Research on Schizophrenia and Depression (NARSAD) to Dr. Umbricht,
grant R29 MH49334 (Dr. Javitt), grant MH46633 (Dr. Kane), and Mental
Health Clinical Research Center Grant for the Study of Schizophrenia at
Hillside Hospital MH41960.
References
Andreasen NC (1982): Negative symptoms in schizophrenia:
Definition and reliability. Arch Gen Psychiatry 39:784 –788.
Blackwood DHR, Whalley LJ, Christie JE, Blackburn IM, St.
Clair DM, McInnes A (1987): Changes in auditory P3
event-related potential in schizophrenia and depression. Br J
Psychiatry 150:154 –160.
Breier A, Buchanan RW, Kirkpatrick B, Davis OR, Irish D,
Summerfeldt A, et al (1994): Effects of clozapine on positive
and negative symptoms in outpatient with schizophrenia.
Am J Psychiatry 151:20 –26.
724
BIOL PSYCHIATRY
1998;44:716 –725
Buchanan RW, Holstein C, Breier A (1994): The comparative
efficacy and long-term effect of clozapine treatment on
neuropsychological test performance. Biol Psychiatry 36:
717–725.
Catts SV, Shelley A, Ward PB, Liebert B, McConaghy N,
Andrews S, et al (1995): Brain potential evidence for an
auditory sensory memory deficit in schizophrenia. Am J
Psychiatry 152:213–219.
Duncan CC, Morihisa J, Fawcett RW, Kirch DG (1987): P300 in
schizophrenia: State or trait marker? Psychopharmacol Bull
23:497–501.
Egan MF, Duncan CC, Suddath RL, Kirch DG, Mirsky AF,
Wyatt RJ (1994): Event-related potential abnormalities correlate with structural brain alterations and clinical features in
patients with chronic schizophrenia. Schizophr Res 11:259 –
271.
Eikmeier G, Lodeman E, Zerbin D, Gastpar M (1992): P300,
clinical symptoms, and neuropsychological parameters in
acute and remitted schizophrenia: A preliminary report. Biol
Psychiatry 31:1065–1069.
Ford JM, White PM, Csernansky JG, Faustman WO, Roth WT,
Pfefferbaum A (1994a): ERPs in schizophrenia: Effects of
antipsychotic medication. Biol Psychiatry 36:153–170.
Ford JM, White PM, Lim KO, Pfefferbaum A (1994b): Schizophrenics have fewer and smaller P300s: A single-trial analysis. Biol Psychiatry 35:96 –103.
Gold JM, Harvey PD (1993): Cognitive deficits in schizophrenia.
Psychiatr Clin Am 16:295–312.
Goldberg TE, Greenberg RD, Griffin SJ, Gold JM, Kleinman JE,
Pickar D, et al (1993): The effect of clozapine on cognition
and psychiatric symptoms in patients with schizophrenia. Br J
Psychiatry 162:43– 48.
Goldman RS, Bates JA, Bilder RM, Kane J, Schooler NR,
Marder S (1996): The effects of clozapine and haloperidol on
memory functioning in treatment refractory schizophrenics.
Schizophr Res 18:221.
Green MF (1996): What are the functional consequences of
neurocognitive deficits in schizophrenia? Am J Psychiatry
153:321–330.
Clinical Global Impression Scale. In: Guy W, editor. EDCEU
Assessment Manual. Rockville, Md: US Department of
Health, Education and Welfare, 1976, pp 217–222.
Hagger C, Buckley P, Kenny JT, Friedman L, Uboby D, Meltzer
HY (1993): Improvement in cognitive functions and psychiatric symptoms in treatment-refractory schizophrenic patients
receiving clozapine. Biol Psychiatry 34:702–712.
Hoff AL, Faustman WO, Wienecke M, Espinoza S, Costa M,
Wolkowitz O, et al (1996): The effects of clozapine on
symptom reduction, neurocognitive function, and clinical
management in treatment-refractory state hospital schizophrenic inpatients. Neuropsychopharmacology 15:361–369.
Javitt DC, Doneshka P, Zylberman I, Ritter W, Vaughan HG
(1993): Impairment of early cortical processing in schizophrenia: An event-related potential confirmation study. Biol
Psychiatry 33:513–519.
Javitt DC, Doneshka P, Grochowski S, Ritter W (1995): Impaired mismatch negativity generation reflects widespread
dysfunction of working memory in schizophrenia. Arch Gen
Psychiatry 52:550 –558.
D. Umbricht et al
Johnson R Jr (1986): A triarchic model of P300 amplitude.
Psychophysiology 23:367–384.
Kane JM, Honigfeld G, Singer J, Meltzer HY, the Clozaril
Collaborative Study Group (1988): Clozapine for the treatment-resistant schizophrenic: A double-blind comparison
with chlorpromazine. Arch Gen Psychiatry 45:789 – 896.
Kane MM, Schooler NR, Marder S, Wirshing W, Ames D,
Umbricht D, et al (1996): Efficacy of clozapine versus haloperidol in a long-term clinical trial. Schizophr Res 18:127.
Lee MA, Thompson PA, Meltzer HY (1994): Effects of clozapine on cognitive function in schizophrenia. J Clin Psychiatry
55(suppl B):82– 87.
McCarley RW, Faux SF, Shenton ME, Nestor PG, Adams J
(1991): Event-related potentials in schizophrenia: Their biological and clinical correlates and a new model of schizophrenic pathophysiology. Schizophr Res 4:209 –231.
McCarley RW, Shenton ME, O’Donnell BF, Faux SF, Kikinis R,
Nestor PG, et al (1993): Auditory P300 abnormalities and left
posterior superior temporal gyrus volume reduction in schizophrenia. Arch Gen Psychiatry 50:190 –197.
Näätänen R (1990): The role of attention in auditory information
processing as revealed by event-related potentials and other
brain measures of cognitive function. Behav Brain Sci 13:
201–288.
Novak GP, Ritter W, Vaughan HG Jr, Wiznitzer ML (1990):
Differentiation of negative event-related potentials in an
auditory discrimination task. Electroencephalogr Clin Neurophysiol 75:255–275.
Oades RD, Zerbin D, Eggers C (1993): Stimulus-Vergleichsprozesse bei psychotischen Jugendlichen mit paranoiden and
nicht paranoiden Symptomen: “Mismatch Negativity” deutet
auf differenzierte Beeinträchtigung hin. In: Bauman P, editor.
Biologische Psychiatrie der Gegenwart. Vienna: SpringerVerlag, pp 69 –74.
O’Donnell BF, Shenton ME, McCarley RW, Faux SF, Smith RS,
Salisbury DF, et al (1993): The auditory N2 component in
schizophrenia: Relationship to MRI temporal lobe gray matter
and to other ERP abnormalities. Biol Psychiatry 34:26 –40.
Polich J, Kok A (1995): Cognitive and biological determinants of
P300: An integrative review. Biol Psychol 41:103–146.
Pritchard WS (1986): Cognitive event-related potential correlates
of schizophrenia. Psychol Bull 100:43– 66.
Ritter W, Simson R, Vaughan HG Jr, Friedman D (1979): A
brain event related to the making of a sensory discrimination.
Science 203:1358 –1361.
Ritter W, Simson R, Vaughan HG Jr (1983): Event-related
potential correlates of two stages of information processing in
physical and semantic discrimination tasks. Psychophysiology 20:168 –179.
Ritter W, Paavilainen P, Lavikainen J, Reinikainen K, Alho K,
Sams M, et al (1992): Event-related potentials to repetition
and change of auditory stimuli. Electroencephalogr Clin
Neurophysiol 83:306 –321.
Sams M, Paavilainen P, Alho K, Naatanen R (1985): Auditory
frequency discrimination and event-related potentials. Electroencephalogr Clin Neurophysiol 62:437– 448.
Schall U, Catts SV, Chaturvedi S, Redenbach J, Karayanidis F,
Clozapine and ERPs in Schizophrenia
Ward PB (1995): The effect of clozapine therapy on psychometric and event-related potential (ERP) measures of cognitive dysfunction in schizophrenia. Schizophr Res 15:164.
Shelley AM, Ward PB, Catts SV, Michie PT, Andrews S,
McConaghy N (1991): Mismatch negativity: An index of a
preattentive processing deficit in schizophrenia. Biol Psychiatry 30:1059 –1062.
Spitzer RI, Williams JBW, Gibbon M, First MB (1990): Structured Clinical Interview for DSM-III-R–Patient Version.
Washington, DC: American Psychiatric Press.
BIOL PSYCHIATRY
1998;44:716 –725
725
Williams R, Baillie P, Dickson RA, Dalby JT (1993): Cognitive
and behavioral efficacy of clozapine in clinical trials. Can
J Psychiatry 38:522.
Woerner MG, Mannuzza S, Kane JM (1988): Anchoring the
Brief Psychiatric Rating Scale: An aid to improve reliability.
Psychopharmacol Bull 24:112–117.
Zahn TP, Pickar D, Haier RJ (1994): Effects of clozapine,
fluphenazine, and placebo on reaction time measures of
attention and sensory dominance in schizophrenia. Schizophr
Res 13:133–144.