400
BIOL PSYCHIATRY !990;27:400--410
Visual Evoked Potential Correlates of Positive/Negative Symptoms in Schizophrenia Steven B. Schwarzkopf, J. Steven Lamberti, Maureen Jiminez, Catherine F. Kane, Michael Henricks, and Henry A. Nasrallah
Previous studies of schizophrenic patients have found evoked potential (EP) correlates of clinical symptomatology, including EP differences between subtypes of schizophrenia. In the current study, 14 medicated male schizophrenics underwent flash visual evoked potentials (VEP) and were clinically rated for positive and negative symptoms. We tested the hypothesis that positive symptoms would be associated with VEP latency reduction and negative symptoms with latency prolongation. Patients were divided into predominantly positive symptom and predominantly negative symptom groups using a combination of positive and negative symptom ratings. Patients with predominantly positive symptoms exhibited reduced latencies when compared with predominantly negative ~m~tom patients. Similarly, significant negative correlations between positive symptom ratings and P200 latency variables were found. Correlations between negative symptom measures and P200 latencies (in the opposite direction) were also noted, but were less significant. These relationships persisted when confounders were statistically controlled for. The results are consistent with previous findings of evoked potential correlates of clinical symptomatology, especially those finding EP latency correlates of psychosis severity and affective blunting. The findings are discussed in relationship to concepts relevant to psychosis, including arousal, sensory gating, and the dopamine hypothesis.
Introduction A number of studies have reported significant relationships between schizophrenic symptom patterns and EP parameters. Shagass (1980), in a review of findings, noted that schizophrenic patients who were chronic, "floridly psychotic," and less depressed, exhibited greater amplitudes and reduced wave shape variability in the first 100 msec after stimulus (visual and somatosensory modalities included) compared with other schizophrenic patients. This finding was recently replicated in a study dividing carefully diagnosed schizophrenic patients into psychotic versus nonpsychotic subgroups based on a self-administered personality inventory (Josiassen et al. 1986). Differences between schizophrenic subgroups have also been reported in a number of
From the Ohio State University College of Medicine, Columbus, Ohio (S.B.S., H.A.N.); and University of Rochester, Rochester, New York (J.S.L., M.J., C.F.K., M.H.) Address reprint requests to Dr. Steven B. Schwarzkopf, Ohio State University College of Medicine, 473 W. 12th AVE., Columbus, OH 43210-1228. Received August 27, 1988; revised May 7, 1989.
© 1990 Societyof BiologicalPsychiatry
0006-3223/90/$03.50
VEPs and Positive/Negative Symptoms
BIOL PSYCHIATRY 1990;27:400-410
401
studies using EP paradigms with stimuli of varying intensity. Using this type of EP paradigm, acute and chronic patients have been reported to differ on VEP measures (Landau et al. 1975). Connolly et al. (1983) also using visual stimuli of differing intensities, found topographical differences in the locatioa of abnormalities, depending on the patient's clinical symptoms. A left temporal a~normality was noted in "nuclear" schizophrenics and a left occipital deviation was noted in patients with hypomania and anxiety symptoms. Clinical correlates relevant to "positive" and "negative" psychotic symptoms have been observed in the NI00 and P200 EP latencies. Studies have found reduced latencies in schizophrenic patients with more "positive" symptoms (Saletu et al. 1973; Roth et al. 1980; Schlor et al. 1985) in college students with evidence of looseness of associations (Catts et al. 1986) and in controls with high ratings on psychoticism (Schlor et al. 1985). Saletu et al. (1973) also noted prolonged latencies in patients with "blunt affect," one of the primary negative symptoms. VEP latency differences have also been noted in schizophrenic patients with versus without a first degree family history of psychiatric illness (Romani et all. 1996): Based on these studies, we tested the hypothesis that patients with predominantly positive symptoms would have reduced VEP latencies compared with patients who have predominantly negative symptoms. Similarly, we tested for significant correlations between clinical symptom ratings and latency measures. VEP latencies (N140 and P200 wave) at fot~" intensities of light were measured as the variables of interest. VEPs and assessments for positive and negative symptoms were collected for 14 medicated male schizophrenic patients. Potential confounders, including neuroleptic dose, age, and chronicity, were examined and statistically controlled for.
Methods
Subjects Fourteen consecutively admitted, physically healthy, medicated, male schizophrenic or schizoaffective patients (by DSM III-R, structured clinical interview [SCID] [Spitzer et al. 1987]) consented to participate in the study. Patients were part of an ongoing study of neurophysiological correlates of psychotic symptoms and treatment response. The mean age for the sample was 28.1 years (SD 6.2), mean lifetime hospitalization was 22.5 months (SD 22.4), and patients were on an average neuroleptic dose of 1059.8 mg chlorpromazine equivalents (SD 690) (Baldessarini 1985). Demographic information is shown in Table 1. Means and SDS are shown for all patients and for patients divided by predominance of positive and negative symptomatology (defined using a median split on a combined positive and negative symptom score). Patients with predominantly positive symptoms did not differ significantly from patients with predominantly negative symptoms on age, chronicity, and neuroleptic dose.
Symptom Ratings Positive and negative symptoms were evaluated using the Brief Psychiatric Rating Scale (BPRS) (Overall and Gorham 1962) and the Scale for the Assessment of Negative Symptoms (SANS) (Andreasen 1982). Ratings were completed by a researcher blind to evoked potential findings. Factor scores consistent with positive symptoms were computed from
402
BIOL PSYCHIATRY 1990;27:400-410
S.B. Schwarzkopf et al.
Table 1. Descriptive Statistics For All Patients and Patients Grouped by Symptoms (High POS-NEG, Low POS-NEG) Variable Age (in years) Chronicity (mo. hosp.)b Neuroleptic dose (CPZ equiv.y Proportion schizoaffective
All pts (N = 14) 28.07 (6.17) 22.46 (22.43) 1059.79 (690.70) 3/14
High POS-NEG" (N = 7) 27.57 (6.19) 16.71 (i4.03) 1052.57 (608.37) 2/7
Low POS-NEG a (N = 7) 28.57 (6.60) 29.17 (29.51 ) 1067.00 (814.5) 117
Mean SD. °High POS-NEG patients are those with positive-negtive symptom scores above the median, and Low POS-NEG patients have scores below the median score. High POS-NEG generally with predominantly positive symptoms, Low POS-NEG generally with predominantly negative symptoms. aChronicity in total months hospitalized. CChlorpromazine equivalents of neuroleptic medication.
the BPRS (factors: hostility, thought disorder, and activation), (Overall et al. 1967). The average of the three factor scores was used as the positive symptom score (POS SX). The average of the five global subscores of the SANS was used as the negative symptom score (NEG SX). Clinical variables for analysis included POS SX score, NEG SX score, and a contrast variable, POS-NEG, calculated by subtracting the NEG SX score from the POS SX score. The POS-NEG score is larger and positive if the patient has predominantly positive symptoms, and small or negative if negative symptoms predominate.
Testing Procedure A flash evoked potential (VEP) with varying stimulus intensity was chosen because of its previous use in differentiating subtypes of schizophrenic patients, recent focus on sensory gating in schizophrenia, and the potential for a more informative and powerful measure when using a within-subjects measure (i.e., change from one condition to another within the same subject). VEPs were recorded at CZ (monopolar recording, linked ear reference) to four intensities of light as detailed by Landau et al. (1975). Gold electrodes were used with skin impedance kept below l0 Kohm. The photostimulator was custom built to specifications of Landau et al. (1975), using a fluorescept screen (40 × 20 cm), with flash intensities of 3, 30, 80, and 240 foot iamberts measured at 50 cm from subject (distance at time of testing). Hash duration was 500 msec, interstimulus interval 988 msec, with intensities presented in a "pseudorandom" order (each intensity preceded each other intensity an equal number of times). A minimum of 50 nonartifacted trials were collected (EOG recorded with lateral and superior orbital electrodes). Electrooculograph (EOG) was monitored to rule out eye blink artifact using 100 IxV as the rejection criterion. Amplitude rejection criterion for CZ was also set at i00 ~V. Visual fixation on the center of the screen was the only task during testing. VEP latei~cies were measured using the following criteria: (1) P 100 as the maximal positivity from 60 msec to 140 msec, (2) Nl¢0 as Re maximal negativity from 80 msec to 180 msec, (3) P200 as the maximal positivity between 140 msec and 280 msec. NI40 and P200 peaks were easily identified for all patients. All recording involved online
BIOL PSYCHIATRY 1990;27:400-410
VEPs and Positive/Negative Symptoms
403
P2 A . . . . . . . . . . .
High
;
~
5'0
/
1(}0
Figure 1. Average flash VEPs at lowest and highest flash intensities for a single subject. P2 Low is the P200 latency at low intensity, and P2 A is the change in P200 latency between the low and high intensity condition (Low P2-High P2 latency in msec).
~i!ii
150 200 250 After Flash
300
Milliseconds
digitizing at a rate of 250 Hz (high and low pass filters set at 0.3 and 100 Hz, respectively) using Grass 1)511 amplifiers. After off-line averaging with elimination of trials with eye-blink artifact, the averaged wave form was plotted for each of the four intensities. The NI40 and P200 latencies were recorded by a researcher blind to clinical ratings of the patient. Interrater reliability for latencies was tested and found to be very high (r > 0.95). The latency change between the lowest intensity flash VEP and the highest intensity flash VEP was calculated (Figs. 1 and 2, P2 A). This variable indicates the within-subject change in latency from the low to high intensity flash condition (intensity l-intensity 2 latency). It was hoped that this measure might indicate both the baseline status (latency at low intensity) and the individuars response to increasing sensory input (latency at highest intensity). It was also hoped that this measure would control for interindividual variability at any particular flash intensity. Figure 1 shows two representative VEP wave forms. The lower wave form is averaged for the lowest flash intensity (30 foot lamberts) and the top wave form is averaged for the highest flash intensity (240 foot lamberts). Indicated in the figure are two variables of particular ic~terest, the P200 latency at the lowest flash intensity (P2 Low), and the P200 latency change from the lowest to the highest flash intensity (P2 A, P200 latency
P2 A
P Intensity
0
50
Figure 2. Average flash VEPs at lowest and highest flash intensities for a subject with very high ratings on positive symptoms. P2 Low is the P200 latency at low intensity, and P2 A is the change in P200 latency between the low and high intensity condition (Low P2High P2 latency in msec). Note the subject's much smaller P2 Low and I)2 A for thi~ sub ject compared with those in Figure 1.
IX~ !ii__\X 100 150 200 250 Milliseconds After Flash
300
404
BIOL PSYCHIATRY 1990;27:400-410
S.B. Schwarzkopf et al.
at lowest intensity-P200 latency at highest intensity) shown as the shaded area in the figure. This subject has a relatively prolonged P2 Low and a large positive P2 A. This pattern was characteristic of patients with predominantly negative symptoms in this study. Figure 2 shows the VEP wave forms from a patient with predominantly positive symptoms. The P2 Low for this subject and the P2 A score are smaller than those of the first subject. Although the majority of patients showed a "typical" pattern of reduced latency with increase in intensity, a number of high positive symptom patients had very small P2 Low values and a seemingly paradoxical increase in latency at the highest intensity. This pattern resulted m negative P2 A scores for these patients (4 of 7 in the high POS-NEG group, none of the 7 in the low POS-NEG group).
Data Analysis Analyses included initial testing of ci~nical and VEP variables for the assumption of normality of distribution using a standard statistical package (SAS, univariate procedure 1986). Patients were classified as either high POS-NEG or low POS-NEG based on a median split (upper 7 p~ients versus lower 7 patients on POS-NEG scores). Differences in group means for the latency measures were assessed using t-tests. In addition, Pearson's correlation coefficients between VEP latency measures and clinical rating scores were used to test for significance of association. One-tail testing was used because of the directional nature of the hypotheses. Significant correlations for the average clinical scores were followed up with examination of the individual subscales to further define which symptoms contributed most to the relationships. Correlations between possible confounders, including medication dosage, chronicity, and age, were calculated and tested for significant effects. The influence of potential confounders was also assessed using multiple regression and partial correlations, statistically controlling for the influence of each of the variables.
Results When patients were divided into those with highest versus lowest POS-NEG scores, no significant N140 latency differences were noted between the groups, although there was a consistent trend for the high POS-NEG patients to ha,,e shorter latencies. No significant correlations were noted between any of the NI40 latency values and symptom rating scales, though a strong trend was found for a correlation between POS SX and N I40 latency at the lowest intensity (negative correlation, p < 0.10). Table 2 shows the means and standard deviations of the P200 latencies at each of the four intensities (3, 40, 80, 240 foot lamberts), and for the P200 A. Mean values are shown for all patients as well as values for patients divided by their POS-NEG score, as in Table 1. The mean values show the high POS-NEG patients to have consistently reduced latencies compared with the low POS-NEG group, except at the very highest intensity, where the ~ u e s are virtually identical. These differences reached statistical significance for the latencies at the two lowest flash intensities and for the P200 A score. The mean P200 ~ score for high positive symptom patients was actually negative because of 4 patients who had larger latencies at intensi~" 4 than at intensity 1. The table shows that the mean P200 latency at intensity 1 for the high POS-NEG patients is less than the mean latency value at the highest intensity in the
VEPs and Positive/Negative Symptoms
BIOL PSYCHIATRY
405
1990;27:400-410
Table 2. P200 Latency Measures For All Patients and Patients Grouped by Symptoms (High POS-NEG, Low POS-NEG) Variable (msec) P200 Latency
All pts
High PGS-NEG ~
(N = 14)
(N = 7)
196.85 (28.47) 188.57 (27.48) 185.86 (20.71) 189.43 (20.49) 7.43 (28.15)
intensity 1
P200 Latency intensity 2
P200 Latency intensity 3 P200 Latency intensity 4
P200 Ad
Low POS-NEGe (N = 7)
178.29 (20.38) 173.71 (20.51) 181.14 (22.71) 189.71 (26.52) - 11.43 (19.92)
215.43 b (23. i4) 203.43 c (26.48) 188.57 (19.52) 189.14 (14.37) 26.2~ (22.13)
Mean SD. °High POS-NEG patientsare those with positive-negativesymptom scores above the median, and Low POS-NEG patients have scores below the median score. High POS-NEG patients have predominantly positive symptoms, Low POS-NEG have predominantly negative symptoms. bHigh POS-NEG patients significantly smaller than Low POS-NEG patients, p < 0.005, one-tailed t-test. thigh POS-NEG patients significantly smaller than Low POS-NEG patients, p < 0.05, one-tailed t-test. 'tp200 change score (P200 latency intensity l--P200 latency intensity 4)
low POS-NEG patients. This finding suggests a floor effect, with high POS-NEG patients being near their minimum latency even at the lowest flash intensity. The Pearson correlation coefficients are shown in Table 3 for the P200 latency variables versus the clinical symptom scores. Correlations were calculated between clinical ratings (POS SX, NEG SX, POS-NEG) and the P200 latencies (at each of the 4 intensities, 3, 30, 80, 240 foot lamberts) and the latency change score (P200 A). A number of significant negative correlations between POS SX and the P200 latency measures are shown (Table 3). These correlations were significant at the lowest intensity conditions (3 and 30 foot lamberts). The latency change variable, P200 A, showed even Stronger correlations than the individual latencies (p < 0.01). The NEG SX score (as me:,sured by the SANS) was not significantly correlated with the P200 latencies. There was, however, a significant positive correlation between NEG SX and the latency change variable, P200 A (p < 0.05). A number of the correlations between the symptom contrast variable, POS-NEG, and P200 latency measures were significant and negative. These correlations were slightly stronger but comparable with the POS SX correlations.
Table 3. Pearson Correlations Between P200 Latency Variables and Clinical Symptom Scores P200 Latencies by Intensity~
POS SX NEG SX POS-NEG
3
30
80
240
P200Ab
- 0.522 c 0.264 - 0.549 ~
- 0.577c 0.185 - 0.54Y
- 0.0t3 - 0.126 0.062
0.200 --0.267 0.305
- 0.674 c 0.462 ~ - 0.777 ~
alntensity in foot lamberts. bp200 latency change (low int. - high ;~nt.). Cp < 0.05,
dp
BIOL PSYCHIATRY !990;27:400--410
Visual Evoked Potential Correlates of Positive/Negative Symptoms in Schizophrenia Steven B. Schwarzkopf, J. Steven Lamberti, Maureen Jiminez, Catherine F. Kane, Michael Henricks, and Henry A. Nasrallah
Previous studies of schizophrenic patients have found evoked potential (EP) correlates of clinical symptomatology, including EP differences between subtypes of schizophrenia. In the current study, 14 medicated male schizophrenics underwent flash visual evoked potentials (VEP) and were clinically rated for positive and negative symptoms. We tested the hypothesis that positive symptoms would be associated with VEP latency reduction and negative symptoms with latency prolongation. Patients were divided into predominantly positive symptom and predominantly negative symptom groups using a combination of positive and negative symptom ratings. Patients with predominantly positive symptoms exhibited reduced latencies when compared with predominantly negative ~m~tom patients. Similarly, significant negative correlations between positive symptom ratings and P200 latency variables were found. Correlations between negative symptom measures and P200 latencies (in the opposite direction) were also noted, but were less significant. These relationships persisted when confounders were statistically controlled for. The results are consistent with previous findings of evoked potential correlates of clinical symptomatology, especially those finding EP latency correlates of psychosis severity and affective blunting. The findings are discussed in relationship to concepts relevant to psychosis, including arousal, sensory gating, and the dopamine hypothesis.
Introduction A number of studies have reported significant relationships between schizophrenic symptom patterns and EP parameters. Shagass (1980), in a review of findings, noted that schizophrenic patients who were chronic, "floridly psychotic," and less depressed, exhibited greater amplitudes and reduced wave shape variability in the first 100 msec after stimulus (visual and somatosensory modalities included) compared with other schizophrenic patients. This finding was recently replicated in a study dividing carefully diagnosed schizophrenic patients into psychotic versus nonpsychotic subgroups based on a self-administered personality inventory (Josiassen et al. 1986). Differences between schizophrenic subgroups have also been reported in a number of
From the Ohio State University College of Medicine, Columbus, Ohio (S.B.S., H.A.N.); and University of Rochester, Rochester, New York (J.S.L., M.J., C.F.K., M.H.) Address reprint requests to Dr. Steven B. Schwarzkopf, Ohio State University College of Medicine, 473 W. 12th AVE., Columbus, OH 43210-1228. Received August 27, 1988; revised May 7, 1989.
© 1990 Societyof BiologicalPsychiatry
0006-3223/90/$03.50
VEPs and Positive/Negative Symptoms
BIOL PSYCHIATRY 1990;27:400-410
401
studies using EP paradigms with stimuli of varying intensity. Using this type of EP paradigm, acute and chronic patients have been reported to differ on VEP measures (Landau et al. 1975). Connolly et al. (1983) also using visual stimuli of differing intensities, found topographical differences in the locatioa of abnormalities, depending on the patient's clinical symptoms. A left temporal a~normality was noted in "nuclear" schizophrenics and a left occipital deviation was noted in patients with hypomania and anxiety symptoms. Clinical correlates relevant to "positive" and "negative" psychotic symptoms have been observed in the NI00 and P200 EP latencies. Studies have found reduced latencies in schizophrenic patients with more "positive" symptoms (Saletu et al. 1973; Roth et al. 1980; Schlor et al. 1985) in college students with evidence of looseness of associations (Catts et al. 1986) and in controls with high ratings on psychoticism (Schlor et al. 1985). Saletu et al. (1973) also noted prolonged latencies in patients with "blunt affect," one of the primary negative symptoms. VEP latency differences have also been noted in schizophrenic patients with versus without a first degree family history of psychiatric illness (Romani et all. 1996): Based on these studies, we tested the hypothesis that patients with predominantly positive symptoms would have reduced VEP latencies compared with patients who have predominantly negative symptoms. Similarly, we tested for significant correlations between clinical symptom ratings and latency measures. VEP latencies (N140 and P200 wave) at fot~" intensities of light were measured as the variables of interest. VEPs and assessments for positive and negative symptoms were collected for 14 medicated male schizophrenic patients. Potential confounders, including neuroleptic dose, age, and chronicity, were examined and statistically controlled for.
Methods
Subjects Fourteen consecutively admitted, physically healthy, medicated, male schizophrenic or schizoaffective patients (by DSM III-R, structured clinical interview [SCID] [Spitzer et al. 1987]) consented to participate in the study. Patients were part of an ongoing study of neurophysiological correlates of psychotic symptoms and treatment response. The mean age for the sample was 28.1 years (SD 6.2), mean lifetime hospitalization was 22.5 months (SD 22.4), and patients were on an average neuroleptic dose of 1059.8 mg chlorpromazine equivalents (SD 690) (Baldessarini 1985). Demographic information is shown in Table 1. Means and SDS are shown for all patients and for patients divided by predominance of positive and negative symptomatology (defined using a median split on a combined positive and negative symptom score). Patients with predominantly positive symptoms did not differ significantly from patients with predominantly negative symptoms on age, chronicity, and neuroleptic dose.
Symptom Ratings Positive and negative symptoms were evaluated using the Brief Psychiatric Rating Scale (BPRS) (Overall and Gorham 1962) and the Scale for the Assessment of Negative Symptoms (SANS) (Andreasen 1982). Ratings were completed by a researcher blind to evoked potential findings. Factor scores consistent with positive symptoms were computed from
402
BIOL PSYCHIATRY 1990;27:400-410
S.B. Schwarzkopf et al.
Table 1. Descriptive Statistics For All Patients and Patients Grouped by Symptoms (High POS-NEG, Low POS-NEG) Variable Age (in years) Chronicity (mo. hosp.)b Neuroleptic dose (CPZ equiv.y Proportion schizoaffective
All pts (N = 14) 28.07 (6.17) 22.46 (22.43) 1059.79 (690.70) 3/14
High POS-NEG" (N = 7) 27.57 (6.19) 16.71 (i4.03) 1052.57 (608.37) 2/7
Low POS-NEG a (N = 7) 28.57 (6.60) 29.17 (29.51 ) 1067.00 (814.5) 117
Mean SD. °High POS-NEG patients are those with positive-negtive symptom scores above the median, and Low POS-NEG patients have scores below the median score. High POS-NEG generally with predominantly positive symptoms, Low POS-NEG generally with predominantly negative symptoms. aChronicity in total months hospitalized. CChlorpromazine equivalents of neuroleptic medication.
the BPRS (factors: hostility, thought disorder, and activation), (Overall et al. 1967). The average of the three factor scores was used as the positive symptom score (POS SX). The average of the five global subscores of the SANS was used as the negative symptom score (NEG SX). Clinical variables for analysis included POS SX score, NEG SX score, and a contrast variable, POS-NEG, calculated by subtracting the NEG SX score from the POS SX score. The POS-NEG score is larger and positive if the patient has predominantly positive symptoms, and small or negative if negative symptoms predominate.
Testing Procedure A flash evoked potential (VEP) with varying stimulus intensity was chosen because of its previous use in differentiating subtypes of schizophrenic patients, recent focus on sensory gating in schizophrenia, and the potential for a more informative and powerful measure when using a within-subjects measure (i.e., change from one condition to another within the same subject). VEPs were recorded at CZ (monopolar recording, linked ear reference) to four intensities of light as detailed by Landau et al. (1975). Gold electrodes were used with skin impedance kept below l0 Kohm. The photostimulator was custom built to specifications of Landau et al. (1975), using a fluorescept screen (40 × 20 cm), with flash intensities of 3, 30, 80, and 240 foot iamberts measured at 50 cm from subject (distance at time of testing). Hash duration was 500 msec, interstimulus interval 988 msec, with intensities presented in a "pseudorandom" order (each intensity preceded each other intensity an equal number of times). A minimum of 50 nonartifacted trials were collected (EOG recorded with lateral and superior orbital electrodes). Electrooculograph (EOG) was monitored to rule out eye blink artifact using 100 IxV as the rejection criterion. Amplitude rejection criterion for CZ was also set at i00 ~V. Visual fixation on the center of the screen was the only task during testing. VEP latei~cies were measured using the following criteria: (1) P 100 as the maximal positivity from 60 msec to 140 msec, (2) Nl¢0 as Re maximal negativity from 80 msec to 180 msec, (3) P200 as the maximal positivity between 140 msec and 280 msec. NI40 and P200 peaks were easily identified for all patients. All recording involved online
BIOL PSYCHIATRY 1990;27:400-410
VEPs and Positive/Negative Symptoms
403
P2 A . . . . . . . . . . .
High
;
~
5'0
/
1(}0
Figure 1. Average flash VEPs at lowest and highest flash intensities for a single subject. P2 Low is the P200 latency at low intensity, and P2 A is the change in P200 latency between the low and high intensity condition (Low P2-High P2 latency in msec).
~i!ii
150 200 250 After Flash
300
Milliseconds
digitizing at a rate of 250 Hz (high and low pass filters set at 0.3 and 100 Hz, respectively) using Grass 1)511 amplifiers. After off-line averaging with elimination of trials with eye-blink artifact, the averaged wave form was plotted for each of the four intensities. The NI40 and P200 latencies were recorded by a researcher blind to clinical ratings of the patient. Interrater reliability for latencies was tested and found to be very high (r > 0.95). The latency change between the lowest intensity flash VEP and the highest intensity flash VEP was calculated (Figs. 1 and 2, P2 A). This variable indicates the within-subject change in latency from the low to high intensity flash condition (intensity l-intensity 2 latency). It was hoped that this measure might indicate both the baseline status (latency at low intensity) and the individuars response to increasing sensory input (latency at highest intensity). It was also hoped that this measure would control for interindividual variability at any particular flash intensity. Figure 1 shows two representative VEP wave forms. The lower wave form is averaged for the lowest flash intensity (30 foot lamberts) and the top wave form is averaged for the highest flash intensity (240 foot lamberts). Indicated in the figure are two variables of particular ic~terest, the P200 latency at the lowest flash intensity (P2 Low), and the P200 latency change from the lowest to the highest flash intensity (P2 A, P200 latency
P2 A
P Intensity
0
50
Figure 2. Average flash VEPs at lowest and highest flash intensities for a subject with very high ratings on positive symptoms. P2 Low is the P200 latency at low intensity, and P2 A is the change in P200 latency between the low and high intensity condition (Low P2High P2 latency in msec). Note the subject's much smaller P2 Low and I)2 A for thi~ sub ject compared with those in Figure 1.
IX~ !ii__\X 100 150 200 250 Milliseconds After Flash
300
404
BIOL PSYCHIATRY 1990;27:400-410
S.B. Schwarzkopf et al.
at lowest intensity-P200 latency at highest intensity) shown as the shaded area in the figure. This subject has a relatively prolonged P2 Low and a large positive P2 A. This pattern was characteristic of patients with predominantly negative symptoms in this study. Figure 2 shows the VEP wave forms from a patient with predominantly positive symptoms. The P2 Low for this subject and the P2 A score are smaller than those of the first subject. Although the majority of patients showed a "typical" pattern of reduced latency with increase in intensity, a number of high positive symptom patients had very small P2 Low values and a seemingly paradoxical increase in latency at the highest intensity. This pattern resulted m negative P2 A scores for these patients (4 of 7 in the high POS-NEG group, none of the 7 in the low POS-NEG group).
Data Analysis Analyses included initial testing of ci~nical and VEP variables for the assumption of normality of distribution using a standard statistical package (SAS, univariate procedure 1986). Patients were classified as either high POS-NEG or low POS-NEG based on a median split (upper 7 p~ients versus lower 7 patients on POS-NEG scores). Differences in group means for the latency measures were assessed using t-tests. In addition, Pearson's correlation coefficients between VEP latency measures and clinical rating scores were used to test for significance of association. One-tail testing was used because of the directional nature of the hypotheses. Significant correlations for the average clinical scores were followed up with examination of the individual subscales to further define which symptoms contributed most to the relationships. Correlations between possible confounders, including medication dosage, chronicity, and age, were calculated and tested for significant effects. The influence of potential confounders was also assessed using multiple regression and partial correlations, statistically controlling for the influence of each of the variables.
Results When patients were divided into those with highest versus lowest POS-NEG scores, no significant N140 latency differences were noted between the groups, although there was a consistent trend for the high POS-NEG patients to ha,,e shorter latencies. No significant correlations were noted between any of the NI40 latency values and symptom rating scales, though a strong trend was found for a correlation between POS SX and N I40 latency at the lowest intensity (negative correlation, p < 0.10). Table 2 shows the means and standard deviations of the P200 latencies at each of the four intensities (3, 40, 80, 240 foot lamberts), and for the P200 A. Mean values are shown for all patients as well as values for patients divided by their POS-NEG score, as in Table 1. The mean values show the high POS-NEG patients to have consistently reduced latencies compared with the low POS-NEG group, except at the very highest intensity, where the ~ u e s are virtually identical. These differences reached statistical significance for the latencies at the two lowest flash intensities and for the P200 A score. The mean P200 ~ score for high positive symptom patients was actually negative because of 4 patients who had larger latencies at intensi~" 4 than at intensity 1. The table shows that the mean P200 latency at intensity 1 for the high POS-NEG patients is less than the mean latency value at the highest intensity in the
VEPs and Positive/Negative Symptoms
BIOL PSYCHIATRY
405
1990;27:400-410
Table 2. P200 Latency Measures For All Patients and Patients Grouped by Symptoms (High POS-NEG, Low POS-NEG) Variable (msec) P200 Latency
All pts
High PGS-NEG ~
(N = 14)
(N = 7)
196.85 (28.47) 188.57 (27.48) 185.86 (20.71) 189.43 (20.49) 7.43 (28.15)
intensity 1
P200 Latency intensity 2
P200 Latency intensity 3 P200 Latency intensity 4
P200 Ad
Low POS-NEGe (N = 7)
178.29 (20.38) 173.71 (20.51) 181.14 (22.71) 189.71 (26.52) - 11.43 (19.92)
215.43 b (23. i4) 203.43 c (26.48) 188.57 (19.52) 189.14 (14.37) 26.2~ (22.13)
Mean SD. °High POS-NEG patientsare those with positive-negativesymptom scores above the median, and Low POS-NEG patients have scores below the median score. High POS-NEG patients have predominantly positive symptoms, Low POS-NEG have predominantly negative symptoms. bHigh POS-NEG patients significantly smaller than Low POS-NEG patients, p < 0.005, one-tailed t-test. thigh POS-NEG patients significantly smaller than Low POS-NEG patients, p < 0.05, one-tailed t-test. 'tp200 change score (P200 latency intensity l--P200 latency intensity 4)
low POS-NEG patients. This finding suggests a floor effect, with high POS-NEG patients being near their minimum latency even at the lowest flash intensity. The Pearson correlation coefficients are shown in Table 3 for the P200 latency variables versus the clinical symptom scores. Correlations were calculated between clinical ratings (POS SX, NEG SX, POS-NEG) and the P200 latencies (at each of the 4 intensities, 3, 30, 80, 240 foot lamberts) and the latency change score (P200 A). A number of significant negative correlations between POS SX and the P200 latency measures are shown (Table 3). These correlations were significant at the lowest intensity conditions (3 and 30 foot lamberts). The latency change variable, P200 A, showed even Stronger correlations than the individual latencies (p < 0.01). The NEG SX score (as me:,sured by the SANS) was not significantly correlated with the P200 latencies. There was, however, a significant positive correlation between NEG SX and the latency change variable, P200 A (p < 0.05). A number of the correlations between the symptom contrast variable, POS-NEG, and P200 latency measures were significant and negative. These correlations were slightly stronger but comparable with the POS SX correlations.
Table 3. Pearson Correlations Between P200 Latency Variables and Clinical Symptom Scores P200 Latencies by Intensity~
POS SX NEG SX POS-NEG
3
30
80
240
P200Ab
- 0.522 c 0.264 - 0.549 ~
- 0.577c 0.185 - 0.54Y
- 0.0t3 - 0.126 0.062
0.200 --0.267 0.305
- 0.674 c 0.462 ~ - 0.777 ~
alntensity in foot lamberts. bp200 latency change (low int. - high ;~nt.). Cp < 0.05,
dp
