BIOL PSYCHIATRY 1991;30:753-769
753
Magnetic Resonance Imaging in Schizophrenia: Altered Brain Morphology Associated with P300 Abnormalities and Eye Tracking Dysfunction Douglas H. R. Blackwood, Allan H. Young, Judith K. McQueen, Mike J. Martin, Hilary M. Roxborough, Walter J. Muir, David M. St. Clair, and David M. Kean
This study was designed to investigate whether auditory P300 event-related potential and smooth pursuit eye-movement abnormalities in schizophrenia are associated with brain structural changes measured using magnetic resonance imaging (MRI). Serial coronal MRI scans obtained from 31 schizophrenic subjects and 33 volunteer controls were analysed by a rater who had no knowledge of the subjects' diagnoses. The brain areas measured bilaterally were the temporal lobe, hippocampus, amygdala, parahippocampal gyrus, head of caudate, cingulate cortex, frontal cortex, and the lateral ventricles. The area of the third ventricle, the thickn?ss of the corpus callosum, and the intracranial area were also measured. Auditory P300 and eye tracking performance were recorded on all subjects. There was a significant increase in the latency and a reduction in amplitude of the P300 in the schizophrenic group. Only in the schizophrenic group was P300 latency correlated negatively with the area of the right and left cingulate cortex and positively with the difference in size between the right and left amygdala. In the subgroup of schizophrenic subjects whose P300 latency was greater than 2 standard deviations above the control mean, the arc~ of the left cingulate cortex was significantly smaller than in controls, and the absolute right-left difference in the area of the amygdala was significantly increased. Eye tracking dysfunction in schizophrenia was not related to changes in the amygdala or cingulate cortex but was sign~cantly correlated wuh eniargemem ~/ the lateral ventricles. Schizophrenic subjects with poor eye tracking had significantly larger lateral ventricles than controls. Eye tracking dysfunction, but not P300 abnormality, was correlated with the severity of both positive and negative symptom of schizophrenia. These findings demonstrate that psychophysiological abnormalities are associated with altered brain structure in schizophrenia. From the University Department of Psychiatry (DHRB, AHY, HMR, WJM, DMS) Royal Edinburgh Hospital, Edinburgh, Department of Radiology (DMK), Royal Infirmary, Edinburgh, and the MRC Brain Metabolism Unit (JKM, MIM), Edinburgh, Scotland. Address reprint requests to Dr. Douglas H.R. Blackwood, UniversityDepartmentof Psychiatry, Royal Edinburgh Hospital, Edinburgh, EHIO 5HF. Received May 21, 1990; revised November 30, 1990.
© 1991 Societyof Biological Psychiatry
0006-3223/91/$03.50
754
BIOL PSYCHIATRY 1991 ;30:753-769
D.H.R. Blackwood et al
Introduction Changes in the amplitude and latency of event-related potentials following auditory, visual, and somatosensory stimuli are among the most consistently reported abaormalities in schizophrenia (Shagass et al 1977, 1978). In particular, abnormalities occur in the response of schizophrenics to tasks requiting selective attention and the evaluation of signals. One such response is the P300 wave form generated during a two-tone discrimination task in which the subject is asked to respond to an infrequent stimulus, a highpitched tone, randon~ly inte~,-ningled with a series of low-pitched tones. Several studies have shown tlr~t ,~:hizephrenics have a reduced P300 amplitude following auditory stimuli (Roth and Cannon 1972; Levit et al 1973; Verleger and Cohen 1978; Roth Horvath et al 1980; Roth Pfefferbaum et al 1980; Roth et al 1981; Baribeau-Braun et al 1983; Morstyn et al 1983; Pfefferbaum et al 1984; Barrett et al 1986; Blackwood et al 1987; Romani et al I o~S,. - , . Faux et al 1988; Pfeffe~aum et all io8q; . . . .Ebme~ar . . . .at .n!. lOgO) . P300 latency has al~;o bezn found to be prolonged in a high proportion of schizophrenic patients (Pfefferb mm et al 1984; Blackwood et al 1987; Romani et ai 1987; Ebmeier et al 1989). A disorder of eye movement measured as increased saccadic intrusions and saccadic tracking while the subject focuses on a moving visual target has also been established as a potential vulnerability trait for schizophrenia (Holzman et N !974). Eye tracking abnormalities are stable with respect to neuroleptic medication and clinical state (lacono et al 1981; Levy et al 1983) and the disorder is found with increased frequency among parents and siblings of schizophrenic subjects (Holzman et al 1974, 1984) leading to the proposal that eye movement disorder is one expression of a "schizophrenia" gene transmitted in some families in an autosomal dominant fashion (Matthysse et al 1986, Holzman et al 1988). Although these psychophysiological abnormalities in schizophrenia have been well replicated, there have been few studies that aim to relate the physiological findings to underlying anatomical changes. Bartfai et al (1985) reported on association between poor eye tracking and lateral ventricular enlargement. Romani et al (1987) performed computed tomography (CT) scans and recorded auditory P300 on 20 schizophrenic subjects. Although tiae vehicular size v,as eaiaEgcd ~aid P3G0 ~ho~ed prolonged latency and reduced amplitude in the schizophrenic subjects there was no consistent association between the P300 abnormalities and ventricular enl -gement. A bilateral decrease in P300 amplitude maximal over the left temporal area in schizophrenia has been reported (Morstyn et al 1983; Faux et al 1988). McCarley et al (1989) compared these evoked-potential abnormalities with changes found on computed tomography scans in nine schizophrenic and nine control subjects and found the P300 amplitude reduction over the left temporal region was significantly correlated with left sylvian-fissure enlargement and with positive symptoms of schizophrenia. Recently, however, Pfefferbaum et al (1989) have found no topographic differences in P300 amplitude between control and schizophrenic subjects. In the present study we aimed to test the hypothesis that changes in auditory P300 and smooth pursuit eye tracking in schizophrenia are associated with ventricular enlargement or abnormalities in temporal lobe structures measured by Magnetic Resonance Imaging (MRI). There is some evidence to suggest that temporal lobe structures may be implicated in the generation or at least the modulation of the P300 wave form, and likewise temporal lobe pathology has been implicated in schizophrenia (Stevens 1973). There is a well-established association between structural lesions involving the temporal lobe (especially on the left) and schizophrenia-like illness (Davison and Bagley 1969).
P300, Eye Tracking, and MR! in Schizophrenia
BI(~L PSYCHIATRY
1991,30:753--769
755
Postmortem studies have found a reduction in volume of the amygdala, hippocarnpal formation, and pallidum internum in the left hemisphere of sch~zepl-aenic subjects (Bogerts et al !985) and cytoarchitectural abnormalities in the entorhinal cortex, particularly the parahippocampal gyms (Jakob and Beckmann 1986). Brown et al (1986), found ventricular enlargement in the temporal horns and a thinning of the cortex of the l : Z ~ p pocampal gyms. A reduced number of hippocampal pyramidal ceils and neurones i~ the entorhinal region of schizophrenic brains has been reported (Falkai and klogerts 1986; Falkai et al 1988). Other studies have silo~n disarray of hippocampal pyranudal-cell orientation (Scheibel and Kovelman 1981; Kovelman and Scheibel 1984) but these findings are controversial, and have not been fully replicated (Altshuler et al 1987; Christison et al 1989). Jeste and Lohr (1989) have reported significantly lower pyramidal-cell density in schizophrenic patients than in normal controls and suggested that hippocampal pathology in schizophrenia may be quite localized. A reduced density of neurons in the cingulate cortex of postmortem schizophrenic brains was reported by Benes et al (1986) and Benes and Bird (1987) and Benes et al (i987) reported increased numbers of vertical axons, believed to be associated afferents, in the cingulate area. The brain stmctures responsible for generafng the scalp-recorded P300 response during an odd-ball listening task, remain unknowra. Lovrich et al (1988) reported that topographic analysis was consistent with cortical generators of P300 in the supratemporal and superior temporal gyms during semantic processing tasks. There is evidence from patients undergoing investigation of temporal lobe seizures that during a two-tone discrimination task~ endogenous potentials are generated within medial temporal lobe structures and patho!ogy in the region of the amygdala and hippocampus is accompanied by changes in these potentials (Halgren et al 1980; Squires et al 1983; Meador et al 1987; Smith et al 1986; Stapleton and Halgren, 1987). However, the result of other studies suggest that medial temporal structures are probably not the generators of the scalp-recorded P300. Stapleton et al (1987) examined the topography of endogenous potentials including the P30C in 11 subjects who had undergone temporal lobectomy for intractable seizures and found no differences between postoperative patients and unoperated centrol subjects. This would support the view that while medial temporal-lobe pathology may modify the P300 response in some patients, the scalp-recorded P300 is not a reliable indicator of the integrity of medial temporal structures. Smith et al (1990) concluded from a study of 10 epileptic patients undergoing stereo¢lct:uoc.t:cpnttlU~ttpny t t l a t a major generator ,_,, ,,,,,,,,,j *,-ot,-,,,o.,--,- . . . . . . . . . . trecorded P300 is the inferior parietal cortex, while other regions including frontal cortex and hippocampus also contribute but to a lesser extent. This project was designed to compare clinical, neurophysiological, and MRI data in schizophrenic and control subjects. The clinical and MRI findings are reported by Young et al (1991) and the relation between the neurophysiological and neuropsychological impairments in schizophrenia will be described by Roxborough et al (in preparation).
Methods
Subjects Thirty-one patients consented to take part in the study. All satisfied Research Diagnostic Criteria (RDC) (Spitzer et al 1978) for schizophrenia based on individual interview, (using
756
BIOL PSYCHIATRY 1991:30:753-769
D.H.R. Blackwood et al
the Schedule for Affective Disorders and Schizophrenia), review of the case notes, and
discussion with the clinician in charge of the case. All patients were receiving therapeutic doses of antipsychotic medication, and all were unequivocally right-handed as measured by a Handedness Scale (Annett 1970). None of the participants had a history of head injury, substance dependence, or a known neurological condition. Patients were rated for clinical symptoms using the Brief Psychiatric Rating Scale, BPRS (Overall and Gotham 1962), and the Hamilton Rating Scale for Depression (Hamilton 1960). The Withdrawal
Retardation subscores of the BPRS (Emotional Wi~drawal, Motor Rehuxlation, and Blunted Affect) were taken to indicate negative symptoms and the Thinking Disturbance subscores on the BPRS (Conceptual Disorganization, Hallucinatory Behavior, Unusual Thought Content) were used to rate positive symptoms. Premorbid intelligence was estimated using the National Adult Reading Test (NART) (Nelson 1982) and tL~. score on this test was c.~~r,-~-~'-d to an equivalent intelligence quotient (IQ) in the range 8 7 128. A control group ~f 33 healthy volunteers who were matched for age and sex with the patient group was recruited from hospital staff and students and members of the local community. All controls were unequivocally right-handed and none had a personal or family history of psychiatric iimess, or a history of head injury, substance dependence, or a known neurological conditio~.
MRI Scans MRI scans were performed using a 0.08 Tesla resistive magnetic resonance imaging system. In a preliminary study a pulse sequence was chosen that produced the clearest images of medial temporal lobe structures. Matrix size was 256, slice thickness was 8 mm and six averages were performed. In the same preliminary study, the settings that gave the clearest images of the ventficular system were identified. For the low-resolution scan the matrix size was 128, slice thickness was 12 mm and one average was done. Ten coronal slices beginning at the anterior pole of the temporal lobe were obtained first at the low-resolution settings than the high-resolution settings. The MRI images were fed to a Quantimet 800 Image Analysis System and areas for measurement were outlined on the monitor screen by a rater (MM) who had no knowledge of the diagnostic status of subjects. To ensure consistency five repeated measurements of each area were made and ,,,,. ,,,,.~,., ~,,,,,.,,. ,,,,. ,.,,,.,,,,.,,..,, o v on c"" measurements ..4, ,ho !.,,o..ol ,,°,,,,~,.loo from the low-resolution scan was 6.9%. The coefficient of variation for measurements on the high-resolution scan was less than 8% except for the right amygdala, which was 10.8 and the left amygdala, which was 10.2%. The measurements were given in mm and sq mm of the actual photographic image, and approximate to one-quarter and onesixteenth real size, respectively. The brain areas studied shown on Figure I were identified using an atlas of the central nervous system (Nienwerhuys et al 1979). The intracranial area (ICA) was measured on the low-resolution scan at the level of C (Figure l) and was the area delineated by the inner table of the skull and a straight line exFapolated across the lower margin, occupied by the brain stem. All measurements were converted into a ratio relative to ICA for that subject. The ventricular brain ratio (VBR) was taken from the low-resolution scan section in which the lateral ventricles had the greatest area, the right and left sides were added together to give a total area, which was then expressed as a ratio of the ICA to give the VBR. All other measurements were carded out on the high-resolution scan.
P300, Eye Tracking, and MRI in Schizophrenia
a
BIOL PSYCHIATRY 1991;30:753-'/69
757
b
Figure 1. Illustrations of coronal sections: (a) Anterior to optic chiasm. (b) Level of third ventricle and amygdala. (c) Level of cerebral peduncles and anterior hippocampus. (d) Level of fourth ventricle. Measured areas are stippled. AMY = amygdala; CCx = cingulate cortex; CNC = head of caudate nucleus; FCx = frontal cortex; HIP = hippocampus; LV = lateral ventricle; PHG = parahippocampal gyrus; TL = temporal lobe; 3V = third ventricle, 4V = fourth ventricle.
Auditory Event-Related Potential The method has been previously described (Blackwood et al 1987). In this study signal generation and data acquisition were performed ,.~ing software developed by the Department of Medical Physics, University of Edinburgh, on a BBC microcomputer. Subjects reclined in a chair in a sound-proof room and performed a two-tone auditory discrimination task. A silver/silver chloride disc electrode was secured at the vertex (cz) position and an indifferent ear-clip electrode was attached to the iefi ear lobe. The ear~ electrode was attached to the right ear. Eye movements were recorded from electrodes placed at the outer canthus and supra orbitally to the right eye. Electrode impedances were less than two kilohms in all cases. Tone pips were delivered binaurally through headphones. Subjects were asked to count silently "infrequent" tones of 1500 Hz randomly presented in a series of 1900-Hz tones that were not counted. The ratio of high to low
758
BIOL PSYCHIATRY 1991 ;30:753-769
D.H.R. Blackwood et al
pitch tones was 1:9. The stimulus rate was 1.1 per see, intensity was 75 dB binaurally. Tone duration was 20 msecs with a rise/fall time of 9.9 reset. The EEG was amplified 10,000 times with a band pass of I to 30 Hz ( - 3 dB attenuation points of 12 dB/octave roll off). The responses to frequent and rare tones were averaged separately with a sampling rate of 1000 Hz. Trials were excluded when the voltage exceeded 45 microvolts. The recording of eye movements confirmed th,tt there were no time-locked artifacts contributing to the scalp-recorded potentials. Five hundred trials were averaged with a sweep time of 750 msecs including a 75 msec prestimulus baseline. Two separate recordings were carded out at each testing and the baseline was estimated from the average response to frequent tones over the 75 msecs before stimulus. The P300 response following the infrequent tones was a positive wave between 260 and 500 msecs. The amplitude of P300 was measured from baseline to peak on the pen-recorded trace. Other components of the wave form were: N1 (negativity 70-120 msec) P2 (Positivity 140-230 msec) both in the averaged response to "frequent" tones. N2 immediately preceded P300 in the response to the infrequent tones. Figure 2 shows representative tracings of auditory event-related potentials recorded from a control subject and from a schizophrenic subject whosc trace showed a delay in P300 latency.
Smooth Pursuit Eye Movement The method followed that of Holzman et al (1974). The electrooculogram (EOG) was recorded in the horizontal plane via silver/silver chloride electrodes attached 1 cm lateral to the outer canthus of each eye and a reference earth electrode attached to the fight earlobe. Impedance was less than 2 kilohms, the EOG was amplified x 10,000, the amplifier band width was 0.016 to 30 Hz and the signal was recorded for later analysis on an FM instrumentation taperecorder. The target, a round spot subtending 0.5 ° of visual arc displayed on a monitor screen, moved sinusoidally at a frequency of 0.4 Hz subtending a maximum angular amplitude of 30 °. The recorded signal was later digitized at a rate of 50 samples a second to a total of 1024 samples. The reconstructed signal was displayed on a monitor and 10 sec of blink-free signal selected for fourier transform. Spectral power density in the range 0.3 to 0.5 Hz (Signal) was compared to power density in the range 0.8 to 8.0 Hz (Noise). The natural logarithm of the signal-to-noise ratio was used as the measure of smooth pursuit eye movement, a low ratio indicating increased saccadic movements and hence greater abnormality.
Results Subjects The 33 control subjects (25 men and 8 women) had a mean age of 28.3 years (range 2152 years) and mean IQ (calculated from NART score) of 121 (range 110-127). The 31 schizophrenic subjects (24 men and 7 women) had a mean age of 29.3 years (range 2050) and mean IQ of 1 l0 (range 91-125). The mean IQ was significantly higher in controls than in schizophrenics (Student's two-tailed t test: t = 5.0, df = 62, p < 0.001). However, there was no significant correlation (at the 5% level using Pearson's correlation) between brain regions and NART score in schizophrenics or controls.
BIOL PSYCHIATI,,,,V
P300, Eye Tracking, and MRI in Schizophrenia
759
1991 ;30:753-769
NormalControl # bJ
/" "t7 WJ
#' l
'
l
spv
!
~
/
Schizophrenic
\_.,.'---"..
Figure 2. Representative recording of auditory event-related potentials recorded during a two-tone discrimination task. The dotted line is the averaged response to frequent low pitched tones and the continuous line which includes the P300, is the averaged response to the infrequent "attended" high-pitch tones. In schizophrenia the P300 response has reduced amplitude and delayed latency.
XL •
0
$ STIMULUS
I
I
300
750
Latency (m secsl
Event-Related Potentials and Eye Tracking Table 1 shows that the schizophrenic group differed from controls in the latency of N2 and P300, and in the amplitude of P2 and P300 event-related potentials. This group of schizophrenics, however, did not differ from controls in eye tracldng performance, although in a larger group of schizophrenic subjects a significant impairment of eye tracking was measured (Blackwood et al in press). The event-related potential results in this smaller group of patients and controls are similar to our larger study in which P300 latency was 301 _+ 22.6 msec and P300 amplitude was 9.8 --- 2.9 microvolts in 212 controls. In 94 schizophrenics P300 latency was 338 +- 35.3 mSec and amplitude 6.4 +__ 3.2 microvolts. Eye tracking (signal/noise ratio) was 4.95 --. 0.92 in 135 controls and 4.24 --- 1.16 in 57 schizophrenics. In the present study the 31 schizophrenic patients were divided into
760
D.H.R. Blackwood et al
BIOL PSYCHIATRY 1991:30:753-769
Table 1. Differences in the Schizophrenic Group in the Latency of N2 and P300, and the Amplitude of P2 and P300 Event-related Potentials from that of the Control Group Latency (msec) mean -+ SD (n) Control NI P2 N2 P300
99 175 217 300
.-4=_19.1 _+ 23.5 +- 27.7 _ 24.5
Schizophrenia (29) (29) (27) (29)
96 164 238 343
"*- 12.5 _ 18.0 _+ 29.9 _ 40.0
(30) (31) (31) (31)
- 1.8 -4- 1.5 _ 3.5 ___ 5.6 _ !.!6
(30) (31) (31) (31)
Amplitude (micro v) mean +_ SD (n) Nl P2 N2 P300 Eye tracking In signal/noise
4.2 4.3 3.8 10.5 4.93
± -+ -4±
4.6 2.1 3.3 3.3 1.12
(29) (29) (24) (29) (28)
2.6 2.9 3.1 8.0 4.45
(31)
Two-Tailed t-tests N21atency P30Olatency P2amplitude P3OOamplitude
t t t t
= = = =
2.8 5.1 3.0 2.1
df = df = df= df =
56 58 58 58
p = 0.007 p 3 4 6 msecs, which is 2 standard deviations above the control mean) were compared with controls. The mean area (ratio) of the left cingulate cortex was 0.89 +_ 0.2 (n = 13) in schizophrenics with P300 abnormality and 1.06 _ 0.27 (n = 25) in controls and this difference was significant (t = 2.2, p = 0.03). The right cingulate cortex was also smaller in this group of schizophrenics (0.89 _+ 0.2) compared to controls but did not reach significance. In schizophrenics with normal P300 latency, the area of the cingulate cortex on the left (0.97 _+ 0.18) and on the right (1.03 _+ 0.2) did not differ from controls. P300 latency was significantly positively correlated with the absolute right-left dif•,-~°'-~"o,..,,~,.in ~mygdala size in the schizophrenic group (r = 0.47, p = 0.03, n = 21) but not in the control group (r = 0.09, p = 0.7, n = 20). The 12 schizophrenics whose P300 latency was abnormally delayed also showed a greater absolute right-left difference in the amygdala compared to the 20 controls. This difference in size of the right and left amygdala was significant (t = 2.24, df = 30, p = 0.03). There were no significant differences between these two groups in the right-
P300, Eye Tracking, and MRI in Schizophrenia
BIOL PSYCHIATRY
763
1991;30:753-769
Table 3. Correlations between P300 Latency and Eye Tracking (Ln Signal:Noise Ratio) with Clinical Variables in 31 Schizophrenic Subjects (Pearson Correlation Coefficient = r) P300 latency
P300 amplitude
Eye tracking (Ln Signal:Noise}
Parameter
r
p
r
p
r
p
Total BPRS Thinking disturbance (positive symptoms) Withdrawal/retardation (negative symptoms) Anxiety/depression
0.32 0.35
NS NS
- 0.01 - 0. i 2
NS NS
- 0.42 - 0.48
0.02 0.006
0.22
NS
0.13
NS
- 0.37
0.04
0.12
NS
- 0.13
NS
0.07
NS
left difference score of other structures. In the 12 schizophrenic subjects who had abnormal P300 latency, the right amygdala was larger than the left in seven and smaller than the left in five.
Relation Between Eye Tracking and MRI Measurements The correlation between eye movement disorder (Ln signal:noise ratio) and ventricular brain ratio was significant in the schizophrenic group (r = - 0 . 4 , p < 0 . 0 3 , n = 31) but not in the control group (r = 0.2, p < 0.03, n = 28) where r = Pearson correlation coefficient. The schizophrenic group was subdivided into a group of 14 whose Ln signal:noise ratio was considered abnormal ( 3.1) using Student's t-test (two tailed). The groups differed significantly in total BPRS score (t = 2.7, df = 29, p < 0.01) and in thinking disturbance (t = 3.6, df = 29, p < 0.001). There was no significant difference between groups in age, NART, withdrawal/retardation score, or anxiety depression score. Similarly 17 schizophrenic patients with abnormal P300 latency (>346 msec) were compared with 14 patients with normal P300 (
753
Magnetic Resonance Imaging in Schizophrenia: Altered Brain Morphology Associated with P300 Abnormalities and Eye Tracking Dysfunction Douglas H. R. Blackwood, Allan H. Young, Judith K. McQueen, Mike J. Martin, Hilary M. Roxborough, Walter J. Muir, David M. St. Clair, and David M. Kean
This study was designed to investigate whether auditory P300 event-related potential and smooth pursuit eye-movement abnormalities in schizophrenia are associated with brain structural changes measured using magnetic resonance imaging (MRI). Serial coronal MRI scans obtained from 31 schizophrenic subjects and 33 volunteer controls were analysed by a rater who had no knowledge of the subjects' diagnoses. The brain areas measured bilaterally were the temporal lobe, hippocampus, amygdala, parahippocampal gyrus, head of caudate, cingulate cortex, frontal cortex, and the lateral ventricles. The area of the third ventricle, the thickn?ss of the corpus callosum, and the intracranial area were also measured. Auditory P300 and eye tracking performance were recorded on all subjects. There was a significant increase in the latency and a reduction in amplitude of the P300 in the schizophrenic group. Only in the schizophrenic group was P300 latency correlated negatively with the area of the right and left cingulate cortex and positively with the difference in size between the right and left amygdala. In the subgroup of schizophrenic subjects whose P300 latency was greater than 2 standard deviations above the control mean, the arc~ of the left cingulate cortex was significantly smaller than in controls, and the absolute right-left difference in the area of the amygdala was significantly increased. Eye tracking dysfunction in schizophrenia was not related to changes in the amygdala or cingulate cortex but was sign~cantly correlated wuh eniargemem ~/ the lateral ventricles. Schizophrenic subjects with poor eye tracking had significantly larger lateral ventricles than controls. Eye tracking dysfunction, but not P300 abnormality, was correlated with the severity of both positive and negative symptom of schizophrenia. These findings demonstrate that psychophysiological abnormalities are associated with altered brain structure in schizophrenia. From the University Department of Psychiatry (DHRB, AHY, HMR, WJM, DMS) Royal Edinburgh Hospital, Edinburgh, Department of Radiology (DMK), Royal Infirmary, Edinburgh, and the MRC Brain Metabolism Unit (JKM, MIM), Edinburgh, Scotland. Address reprint requests to Dr. Douglas H.R. Blackwood, UniversityDepartmentof Psychiatry, Royal Edinburgh Hospital, Edinburgh, EHIO 5HF. Received May 21, 1990; revised November 30, 1990.
© 1991 Societyof Biological Psychiatry
0006-3223/91/$03.50
754
BIOL PSYCHIATRY 1991 ;30:753-769
D.H.R. Blackwood et al
Introduction Changes in the amplitude and latency of event-related potentials following auditory, visual, and somatosensory stimuli are among the most consistently reported abaormalities in schizophrenia (Shagass et al 1977, 1978). In particular, abnormalities occur in the response of schizophrenics to tasks requiting selective attention and the evaluation of signals. One such response is the P300 wave form generated during a two-tone discrimination task in which the subject is asked to respond to an infrequent stimulus, a highpitched tone, randon~ly inte~,-ningled with a series of low-pitched tones. Several studies have shown tlr~t ,~:hizephrenics have a reduced P300 amplitude following auditory stimuli (Roth and Cannon 1972; Levit et al 1973; Verleger and Cohen 1978; Roth Horvath et al 1980; Roth Pfefferbaum et al 1980; Roth et al 1981; Baribeau-Braun et al 1983; Morstyn et al 1983; Pfefferbaum et al 1984; Barrett et al 1986; Blackwood et al 1987; Romani et al I o~S,. - , . Faux et al 1988; Pfeffe~aum et all io8q; . . . .Ebme~ar . . . .at .n!. lOgO) . P300 latency has al~;o bezn found to be prolonged in a high proportion of schizophrenic patients (Pfefferb mm et al 1984; Blackwood et al 1987; Romani et ai 1987; Ebmeier et al 1989). A disorder of eye movement measured as increased saccadic intrusions and saccadic tracking while the subject focuses on a moving visual target has also been established as a potential vulnerability trait for schizophrenia (Holzman et N !974). Eye tracking abnormalities are stable with respect to neuroleptic medication and clinical state (lacono et al 1981; Levy et al 1983) and the disorder is found with increased frequency among parents and siblings of schizophrenic subjects (Holzman et al 1974, 1984) leading to the proposal that eye movement disorder is one expression of a "schizophrenia" gene transmitted in some families in an autosomal dominant fashion (Matthysse et al 1986, Holzman et al 1988). Although these psychophysiological abnormalities in schizophrenia have been well replicated, there have been few studies that aim to relate the physiological findings to underlying anatomical changes. Bartfai et al (1985) reported on association between poor eye tracking and lateral ventricular enlargement. Romani et al (1987) performed computed tomography (CT) scans and recorded auditory P300 on 20 schizophrenic subjects. Although tiae vehicular size v,as eaiaEgcd ~aid P3G0 ~ho~ed prolonged latency and reduced amplitude in the schizophrenic subjects there was no consistent association between the P300 abnormalities and ventricular enl -gement. A bilateral decrease in P300 amplitude maximal over the left temporal area in schizophrenia has been reported (Morstyn et al 1983; Faux et al 1988). McCarley et al (1989) compared these evoked-potential abnormalities with changes found on computed tomography scans in nine schizophrenic and nine control subjects and found the P300 amplitude reduction over the left temporal region was significantly correlated with left sylvian-fissure enlargement and with positive symptoms of schizophrenia. Recently, however, Pfefferbaum et al (1989) have found no topographic differences in P300 amplitude between control and schizophrenic subjects. In the present study we aimed to test the hypothesis that changes in auditory P300 and smooth pursuit eye tracking in schizophrenia are associated with ventricular enlargement or abnormalities in temporal lobe structures measured by Magnetic Resonance Imaging (MRI). There is some evidence to suggest that temporal lobe structures may be implicated in the generation or at least the modulation of the P300 wave form, and likewise temporal lobe pathology has been implicated in schizophrenia (Stevens 1973). There is a well-established association between structural lesions involving the temporal lobe (especially on the left) and schizophrenia-like illness (Davison and Bagley 1969).
P300, Eye Tracking, and MR! in Schizophrenia
BI(~L PSYCHIATRY
1991,30:753--769
755
Postmortem studies have found a reduction in volume of the amygdala, hippocarnpal formation, and pallidum internum in the left hemisphere of sch~zepl-aenic subjects (Bogerts et al !985) and cytoarchitectural abnormalities in the entorhinal cortex, particularly the parahippocampal gyms (Jakob and Beckmann 1986). Brown et al (1986), found ventricular enlargement in the temporal horns and a thinning of the cortex of the l : Z ~ p pocampal gyms. A reduced number of hippocampal pyramidal ceils and neurones i~ the entorhinal region of schizophrenic brains has been reported (Falkai and klogerts 1986; Falkai et al 1988). Other studies have silo~n disarray of hippocampal pyranudal-cell orientation (Scheibel and Kovelman 1981; Kovelman and Scheibel 1984) but these findings are controversial, and have not been fully replicated (Altshuler et al 1987; Christison et al 1989). Jeste and Lohr (1989) have reported significantly lower pyramidal-cell density in schizophrenic patients than in normal controls and suggested that hippocampal pathology in schizophrenia may be quite localized. A reduced density of neurons in the cingulate cortex of postmortem schizophrenic brains was reported by Benes et al (1986) and Benes and Bird (1987) and Benes et al (i987) reported increased numbers of vertical axons, believed to be associated afferents, in the cingulate area. The brain stmctures responsible for generafng the scalp-recorded P300 response during an odd-ball listening task, remain unknowra. Lovrich et al (1988) reported that topographic analysis was consistent with cortical generators of P300 in the supratemporal and superior temporal gyms during semantic processing tasks. There is evidence from patients undergoing investigation of temporal lobe seizures that during a two-tone discrimination task~ endogenous potentials are generated within medial temporal lobe structures and patho!ogy in the region of the amygdala and hippocampus is accompanied by changes in these potentials (Halgren et al 1980; Squires et al 1983; Meador et al 1987; Smith et al 1986; Stapleton and Halgren, 1987). However, the result of other studies suggest that medial temporal structures are probably not the generators of the scalp-recorded P300. Stapleton et al (1987) examined the topography of endogenous potentials including the P30C in 11 subjects who had undergone temporal lobectomy for intractable seizures and found no differences between postoperative patients and unoperated centrol subjects. This would support the view that while medial temporal-lobe pathology may modify the P300 response in some patients, the scalp-recorded P300 is not a reliable indicator of the integrity of medial temporal structures. Smith et al (1990) concluded from a study of 10 epileptic patients undergoing stereo¢lct:uoc.t:cpnttlU~ttpny t t l a t a major generator ,_,, ,,,,,,,,,j *,-ot,-,,,o.,--,- . . . . . . . . . . trecorded P300 is the inferior parietal cortex, while other regions including frontal cortex and hippocampus also contribute but to a lesser extent. This project was designed to compare clinical, neurophysiological, and MRI data in schizophrenic and control subjects. The clinical and MRI findings are reported by Young et al (1991) and the relation between the neurophysiological and neuropsychological impairments in schizophrenia will be described by Roxborough et al (in preparation).
Methods
Subjects Thirty-one patients consented to take part in the study. All satisfied Research Diagnostic Criteria (RDC) (Spitzer et al 1978) for schizophrenia based on individual interview, (using
756
BIOL PSYCHIATRY 1991:30:753-769
D.H.R. Blackwood et al
the Schedule for Affective Disorders and Schizophrenia), review of the case notes, and
discussion with the clinician in charge of the case. All patients were receiving therapeutic doses of antipsychotic medication, and all were unequivocally right-handed as measured by a Handedness Scale (Annett 1970). None of the participants had a history of head injury, substance dependence, or a known neurological condition. Patients were rated for clinical symptoms using the Brief Psychiatric Rating Scale, BPRS (Overall and Gotham 1962), and the Hamilton Rating Scale for Depression (Hamilton 1960). The Withdrawal
Retardation subscores of the BPRS (Emotional Wi~drawal, Motor Rehuxlation, and Blunted Affect) were taken to indicate negative symptoms and the Thinking Disturbance subscores on the BPRS (Conceptual Disorganization, Hallucinatory Behavior, Unusual Thought Content) were used to rate positive symptoms. Premorbid intelligence was estimated using the National Adult Reading Test (NART) (Nelson 1982) and tL~. score on this test was c.~~r,-~-~'-d to an equivalent intelligence quotient (IQ) in the range 8 7 128. A control group ~f 33 healthy volunteers who were matched for age and sex with the patient group was recruited from hospital staff and students and members of the local community. All controls were unequivocally right-handed and none had a personal or family history of psychiatric iimess, or a history of head injury, substance dependence, or a known neurological conditio~.
MRI Scans MRI scans were performed using a 0.08 Tesla resistive magnetic resonance imaging system. In a preliminary study a pulse sequence was chosen that produced the clearest images of medial temporal lobe structures. Matrix size was 256, slice thickness was 8 mm and six averages were performed. In the same preliminary study, the settings that gave the clearest images of the ventficular system were identified. For the low-resolution scan the matrix size was 128, slice thickness was 12 mm and one average was done. Ten coronal slices beginning at the anterior pole of the temporal lobe were obtained first at the low-resolution settings than the high-resolution settings. The MRI images were fed to a Quantimet 800 Image Analysis System and areas for measurement were outlined on the monitor screen by a rater (MM) who had no knowledge of the diagnostic status of subjects. To ensure consistency five repeated measurements of each area were made and ,,,,. ,,,,.~,., ~,,,,,.,,. ,,,,. ,.,,,.,,,,.,,..,, o v on c"" measurements ..4, ,ho !.,,o..ol ,,°,,,,~,.loo from the low-resolution scan was 6.9%. The coefficient of variation for measurements on the high-resolution scan was less than 8% except for the right amygdala, which was 10.8 and the left amygdala, which was 10.2%. The measurements were given in mm and sq mm of the actual photographic image, and approximate to one-quarter and onesixteenth real size, respectively. The brain areas studied shown on Figure I were identified using an atlas of the central nervous system (Nienwerhuys et al 1979). The intracranial area (ICA) was measured on the low-resolution scan at the level of C (Figure l) and was the area delineated by the inner table of the skull and a straight line exFapolated across the lower margin, occupied by the brain stem. All measurements were converted into a ratio relative to ICA for that subject. The ventricular brain ratio (VBR) was taken from the low-resolution scan section in which the lateral ventricles had the greatest area, the right and left sides were added together to give a total area, which was then expressed as a ratio of the ICA to give the VBR. All other measurements were carded out on the high-resolution scan.
P300, Eye Tracking, and MRI in Schizophrenia
a
BIOL PSYCHIATRY 1991;30:753-'/69
757
b
Figure 1. Illustrations of coronal sections: (a) Anterior to optic chiasm. (b) Level of third ventricle and amygdala. (c) Level of cerebral peduncles and anterior hippocampus. (d) Level of fourth ventricle. Measured areas are stippled. AMY = amygdala; CCx = cingulate cortex; CNC = head of caudate nucleus; FCx = frontal cortex; HIP = hippocampus; LV = lateral ventricle; PHG = parahippocampal gyrus; TL = temporal lobe; 3V = third ventricle, 4V = fourth ventricle.
Auditory Event-Related Potential The method has been previously described (Blackwood et al 1987). In this study signal generation and data acquisition were performed ,.~ing software developed by the Department of Medical Physics, University of Edinburgh, on a BBC microcomputer. Subjects reclined in a chair in a sound-proof room and performed a two-tone auditory discrimination task. A silver/silver chloride disc electrode was secured at the vertex (cz) position and an indifferent ear-clip electrode was attached to the iefi ear lobe. The ear~ electrode was attached to the right ear. Eye movements were recorded from electrodes placed at the outer canthus and supra orbitally to the right eye. Electrode impedances were less than two kilohms in all cases. Tone pips were delivered binaurally through headphones. Subjects were asked to count silently "infrequent" tones of 1500 Hz randomly presented in a series of 1900-Hz tones that were not counted. The ratio of high to low
758
BIOL PSYCHIATRY 1991 ;30:753-769
D.H.R. Blackwood et al
pitch tones was 1:9. The stimulus rate was 1.1 per see, intensity was 75 dB binaurally. Tone duration was 20 msecs with a rise/fall time of 9.9 reset. The EEG was amplified 10,000 times with a band pass of I to 30 Hz ( - 3 dB attenuation points of 12 dB/octave roll off). The responses to frequent and rare tones were averaged separately with a sampling rate of 1000 Hz. Trials were excluded when the voltage exceeded 45 microvolts. The recording of eye movements confirmed th,tt there were no time-locked artifacts contributing to the scalp-recorded potentials. Five hundred trials were averaged with a sweep time of 750 msecs including a 75 msec prestimulus baseline. Two separate recordings were carded out at each testing and the baseline was estimated from the average response to frequent tones over the 75 msecs before stimulus. The P300 response following the infrequent tones was a positive wave between 260 and 500 msecs. The amplitude of P300 was measured from baseline to peak on the pen-recorded trace. Other components of the wave form were: N1 (negativity 70-120 msec) P2 (Positivity 140-230 msec) both in the averaged response to "frequent" tones. N2 immediately preceded P300 in the response to the infrequent tones. Figure 2 shows representative tracings of auditory event-related potentials recorded from a control subject and from a schizophrenic subject whosc trace showed a delay in P300 latency.
Smooth Pursuit Eye Movement The method followed that of Holzman et al (1974). The electrooculogram (EOG) was recorded in the horizontal plane via silver/silver chloride electrodes attached 1 cm lateral to the outer canthus of each eye and a reference earth electrode attached to the fight earlobe. Impedance was less than 2 kilohms, the EOG was amplified x 10,000, the amplifier band width was 0.016 to 30 Hz and the signal was recorded for later analysis on an FM instrumentation taperecorder. The target, a round spot subtending 0.5 ° of visual arc displayed on a monitor screen, moved sinusoidally at a frequency of 0.4 Hz subtending a maximum angular amplitude of 30 °. The recorded signal was later digitized at a rate of 50 samples a second to a total of 1024 samples. The reconstructed signal was displayed on a monitor and 10 sec of blink-free signal selected for fourier transform. Spectral power density in the range 0.3 to 0.5 Hz (Signal) was compared to power density in the range 0.8 to 8.0 Hz (Noise). The natural logarithm of the signal-to-noise ratio was used as the measure of smooth pursuit eye movement, a low ratio indicating increased saccadic movements and hence greater abnormality.
Results Subjects The 33 control subjects (25 men and 8 women) had a mean age of 28.3 years (range 2152 years) and mean IQ (calculated from NART score) of 121 (range 110-127). The 31 schizophrenic subjects (24 men and 7 women) had a mean age of 29.3 years (range 2050) and mean IQ of 1 l0 (range 91-125). The mean IQ was significantly higher in controls than in schizophrenics (Student's two-tailed t test: t = 5.0, df = 62, p < 0.001). However, there was no significant correlation (at the 5% level using Pearson's correlation) between brain regions and NART score in schizophrenics or controls.
BIOL PSYCHIATI,,,,V
P300, Eye Tracking, and MRI in Schizophrenia
759
1991 ;30:753-769
NormalControl # bJ
/" "t7 WJ
#' l
'
l
spv
!
~
/
Schizophrenic
\_.,.'---"..
Figure 2. Representative recording of auditory event-related potentials recorded during a two-tone discrimination task. The dotted line is the averaged response to frequent low pitched tones and the continuous line which includes the P300, is the averaged response to the infrequent "attended" high-pitch tones. In schizophrenia the P300 response has reduced amplitude and delayed latency.
XL •
0
$ STIMULUS
I
I
300
750
Latency (m secsl
Event-Related Potentials and Eye Tracking Table 1 shows that the schizophrenic group differed from controls in the latency of N2 and P300, and in the amplitude of P2 and P300 event-related potentials. This group of schizophrenics, however, did not differ from controls in eye tracldng performance, although in a larger group of schizophrenic subjects a significant impairment of eye tracking was measured (Blackwood et al in press). The event-related potential results in this smaller group of patients and controls are similar to our larger study in which P300 latency was 301 _+ 22.6 msec and P300 amplitude was 9.8 --- 2.9 microvolts in 212 controls. In 94 schizophrenics P300 latency was 338 +- 35.3 mSec and amplitude 6.4 +__ 3.2 microvolts. Eye tracking (signal/noise ratio) was 4.95 --. 0.92 in 135 controls and 4.24 --- 1.16 in 57 schizophrenics. In the present study the 31 schizophrenic patients were divided into
760
D.H.R. Blackwood et al
BIOL PSYCHIATRY 1991:30:753-769
Table 1. Differences in the Schizophrenic Group in the Latency of N2 and P300, and the Amplitude of P2 and P300 Event-related Potentials from that of the Control Group Latency (msec) mean -+ SD (n) Control NI P2 N2 P300
99 175 217 300
.-4=_19.1 _+ 23.5 +- 27.7 _ 24.5
Schizophrenia (29) (29) (27) (29)
96 164 238 343
"*- 12.5 _ 18.0 _+ 29.9 _ 40.0
(30) (31) (31) (31)
- 1.8 -4- 1.5 _ 3.5 ___ 5.6 _ !.!6
(30) (31) (31) (31)
Amplitude (micro v) mean +_ SD (n) Nl P2 N2 P300 Eye tracking In signal/noise
4.2 4.3 3.8 10.5 4.93
± -+ -4±
4.6 2.1 3.3 3.3 1.12
(29) (29) (24) (29) (28)
2.6 2.9 3.1 8.0 4.45
(31)
Two-Tailed t-tests N21atency P30Olatency P2amplitude P3OOamplitude
t t t t
= = = =
2.8 5.1 3.0 2.1
df = df = df= df =
56 58 58 58
p = 0.007 p 3 4 6 msecs, which is 2 standard deviations above the control mean) were compared with controls. The mean area (ratio) of the left cingulate cortex was 0.89 +_ 0.2 (n = 13) in schizophrenics with P300 abnormality and 1.06 _ 0.27 (n = 25) in controls and this difference was significant (t = 2.2, p = 0.03). The right cingulate cortex was also smaller in this group of schizophrenics (0.89 _+ 0.2) compared to controls but did not reach significance. In schizophrenics with normal P300 latency, the area of the cingulate cortex on the left (0.97 _+ 0.18) and on the right (1.03 _+ 0.2) did not differ from controls. P300 latency was significantly positively correlated with the absolute right-left dif•,-~°'-~"o,..,,~,.in ~mygdala size in the schizophrenic group (r = 0.47, p = 0.03, n = 21) but not in the control group (r = 0.09, p = 0.7, n = 20). The 12 schizophrenics whose P300 latency was abnormally delayed also showed a greater absolute right-left difference in the amygdala compared to the 20 controls. This difference in size of the right and left amygdala was significant (t = 2.24, df = 30, p = 0.03). There were no significant differences between these two groups in the right-
P300, Eye Tracking, and MRI in Schizophrenia
BIOL PSYCHIATRY
763
1991;30:753-769
Table 3. Correlations between P300 Latency and Eye Tracking (Ln Signal:Noise Ratio) with Clinical Variables in 31 Schizophrenic Subjects (Pearson Correlation Coefficient = r) P300 latency
P300 amplitude
Eye tracking (Ln Signal:Noise}
Parameter
r
p
r
p
r
p
Total BPRS Thinking disturbance (positive symptoms) Withdrawal/retardation (negative symptoms) Anxiety/depression
0.32 0.35
NS NS
- 0.01 - 0. i 2
NS NS
- 0.42 - 0.48
0.02 0.006
0.22
NS
0.13
NS
- 0.37
0.04
0.12
NS
- 0.13
NS
0.07
NS
left difference score of other structures. In the 12 schizophrenic subjects who had abnormal P300 latency, the right amygdala was larger than the left in seven and smaller than the left in five.
Relation Between Eye Tracking and MRI Measurements The correlation between eye movement disorder (Ln signal:noise ratio) and ventricular brain ratio was significant in the schizophrenic group (r = - 0 . 4 , p < 0 . 0 3 , n = 31) but not in the control group (r = 0.2, p < 0.03, n = 28) where r = Pearson correlation coefficient. The schizophrenic group was subdivided into a group of 14 whose Ln signal:noise ratio was considered abnormal ( 3.1) using Student's t-test (two tailed). The groups differed significantly in total BPRS score (t = 2.7, df = 29, p < 0.01) and in thinking disturbance (t = 3.6, df = 29, p < 0.001). There was no significant difference between groups in age, NART, withdrawal/retardation score, or anxiety depression score. Similarly 17 schizophrenic patients with abnormal P300 latency (>346 msec) were compared with 14 patients with normal P300 (
