Negative Symptoms and Hypofrontality in Chronic Schizophrenia Adam Wolkin, MD; Michael Sanfilipo, MS; Alfred P. Wolf, PhD; Burton Jonathan D. Brodie, PhD, MD; John Rotrosen, MD \s=b\ Frontal lobe dysfunction is widely suspected to underlie negative symptoms of schizophrenia. This hypothesis is based largely on long-standing observations of the similarities between the effects of frontal lobe lesions and negative symptoms. However, there is little direct evidence specifically for such an association in schizophrenic patients. We measured the relationship between decreased relative prefrontal cortex glucose metabolism (hypofrontality) using positron emission tomography and evaluated the severity of negative symptoms in 20 chronic schizophrenics who underwent scanning while not receiving neuroleptic drugs. We found a close relationship between negative symptoms and prefrontal hypometabolism, particularly in the right dorsolateral convexity. This association was regionally spe-
cific. Furthermore, there
was no
evidence that this rela-
tionship was an artifact of age, cerebral atrophy, or severity of positive symptoms. (Arch Gen Psychiatry. 1992;49:959-965) lobe
utilization has been
frontal glucose Decreased reported many,17 though studies in
not
all,8"11 positron
of glucose metabo¬ emission tomographic (PET) lism in schizophrenia. While the pathophysiologic impor¬ tance of hypofrontality is unknown, it is suspected to re¬ late causally to negative symptoms.1-512"17 This hypothesis is largely based on the similarities between negative symptoms of schizophrenia and the sequelae of frontal lobe lesions described in numerous clinical and preclinical studies. For example, experimental lesions of the prefron¬ tal cortex in animals suggest that this area is involved in social affiliation and emotional expression; animals with lesions display marked social withdrawal and defects of emotional behavior.1718 Traumatic and surgically induced frontal lesions in humans may result in avolition, apathy, anhedonia, affective flattening, and impoverished speech.1819 Impairment in cognitive function, associated
Accepted for publication January 30, From the Department of Psychiatry, University Medical Center, New York (Drs Wolkin, Angrist, Brodie, and Rotrosen); the Psychiatry Service, New York Veterans Administration Medical Center, New York (Drs Wolkin, Angrist, and Rotrosen and Mr Sanfilipo); and the Chemistry Department, Brookhaven National Laboratory, Upton, NY 1992. New York
(Dr Wolf).
Reprint requests to the New York VA Medical York, NY 10010 (Dr Wolkin).
23rd St, New
Center (116A), 423 E
Angrist, MD;
with the negative symptom complex of schizophrenia,20"23 is also believed to reflect frontal lobe dysfunction.17,24 Herein, we report the relationship between negative symptoms and frontal lobe hypometabolism in a sample of 20 chronic schizophrenics who underwent scanning while not
receiving neuroleptic drugs.
SUBJECTS AND METHODS Twenty male subjects meeting DSM-III-R criteria for chronic schizophrenia as the only active Axis I disorder (11 paranoid, one disorganized, and eight undifferentiated) gave informed written consent to participate in our study. The subjects were inpatients at a veterans' hospital, where they were hospitalized for either acute or chronic decompensation of schizophrenia. Of the 20 subjects, 13 were black, six were white, and one was Hispanic. Based on a handedness inventory, 16 were right-handed and four
had mixed dominance or were left-handed. Overall, the mean (±SD) age was 35±5 years (range, 26 to 44 years). The mean du¬ ration of illness was 13 years (range, 7 to 24 years). Based on the findings of a history, a physical examination, electrocardiography, and routine blood chemistry analyses, all subjects were found to be in good physical health. Patients with a history of drug abuse within the preceding month were excluded from the
study. Subjects had not received oral neuroleptic drugs for a minimum of 2 weeks before PET scanning and had not received depot neu¬ roleptic drugs for a minimum of 3 months. The median washout period was 23 days (range, 14 to a coding cutoff of 365 days for three patients who had never received neuroleptic drugs). Six patients required treatment for agitation before PET scanning, in which case they received only choral hydrate (0.5 g) or lorazepam (0.5 mg) every 6 hours as needed. These medications were selected based on their relatively brief half-lives to minimize re¬ sidual drug effects at the time of PET scanning. No medication was administered in the 12 hours preceding scanning. A brief cognitive test battery was administered within 24 hours of PET scanning, including the Modified Wisconsin Card Sort Task,25 Category Retrieval Test,26 and the Digit-Span, DigitSymbol, and Vocabulary Subtests of the Wechsler Adult Intelli¬
gence Scale-Revised.27 Global psychopathologic findings were rated with the Brief Psychiatric Rating Scale (BPRS); negative symptoms were specif¬ ically assessed by the Abrams and Taylor28 Scale for Emotional Blunting. Positive symptoms ratings were based on the BPRS "schizophrenia" factor (hallucinations, unusual thought content, and conceptual disorganization). All ratings were made immedi¬ ately before PET scanning. Positron emission tomographic scans were obtained at Brookhaven National Laboratory, Long Island, NY, with a PET VI
Downloaded From: http://archpsyc.jamanetwork.com/ by a West Virginia University Library User on 06/22/2015
scans were evaluated for cortical atrophy and ventricular en¬ largement. Only those sections superior to ventricles were in¬ cluded in the analyses of cortical atrophy to avoid bias from ob¬ servation of enlarged ventricles. Cortical atrophy was assessed by two independent observers using a rating scale of 0 (none) to 4 (severe). All CT scans were rated by investigators who were blind to subject status. Ventricular enlargement was assessed based on
of frontal horn and third ventricle size. These measures included linear ratios (bifrontal and bicaudate ratios; see Wolkin et al32 for descriptions) and ventricular-to-brain ratios. The latter measures were determined via computer software that permits windowing of ventricular cerebrospinal fluid. Data were analyzed for departure from the normal distribution both visually and by Shapiro-Wilks and Lilliefors tests (comput¬ ed using SPSS/PC, version 3.0, "Examine" procedure; SPSS, Chi¬ cago, 111). Variables with nonnormal distribution were trans¬ formed either by square root or log transformations, and no departure from normality was detected by the above-mentioned tests. Because of the discontinuous rating scales, both parametric and nonparametric correlations were tested. The unique contri¬ bution of negative symptoms and hypofrontality on cognitive test performance was examined with multiple regression, whereby the independent variables (relative frontal metabolism and neg¬ ative symptoms) were entered simultaneously into the regression equation for each cognitive test. measures
Fig 1.—Schematic hemidiagrams of regions of interest used in positron emission tomographic image analysis. See text for a description of methods to identify corresponding computed tomographic sections. Cer indicates cerebellum; DLPF, dorsolateral prefrontal cortex; Mes, mesial temporal lobe; Orb, orbital-medial prefrontal cortex; Str, striatum; Tern, temporal cortex; and Tha, thalamus. in low-resolution mode (11.8 mm full width at half max¬ imum in plane of section and 14.4 mm in the axis) with 18 F-fluorodeoxyglose2'* as the metabolic tracer.3" Patients were po¬ sitioned with horizontal and vertical localizing lasers such that the head was parallel to the canthomeatal plane. Positioning was maintained by individually fitted foam-filled head holders. Scan¬ ning was performed with the subjects' eyes and ears open in a quiet, dimly lit room that did not preclude observation by the subject. Plasma sampling was accomplished through venous catheterization with use of a hand warmer to arterialize venous camera
blood.31
Computed tomographic (CT) scans were obtained with a Pickar scanner. Patients were positioned following the same procedure described for PET scanning, using their own individ¬ ually molded head holder to ensure consistency in positioning between PET and CT scanning and to maintain stability. The PET scans were analyzed based on individual anatomy as ascertained from each patient's matched CT scan. Given the sam¬ ple size and camera resolution, a conservative number of major regions of interest (ROIs) were assessed for both left and right hemispheres: dorsolateral prefrontal cortex, orbitomedial pre¬ frontal cortex, a large strip of temporal cortex, striatum, thalamus, mesial temporal cortex (amygdala/hippocampus), and cerebel¬ lum. These ROIs were selected from five CT sections (Fig 1) based on the following criteria: section I, largest area of cerebellum; sec¬ tion II, basal orbital and temporal lobes; section III, largest area of basal ganglia; section IV, thalamus most prominent; and sec¬ tion V, highest section still including lateral ventricle (transecting body of ventricle). The seven hemispheric ROIs were then 1200 CT
outlined on the monitor with the use of a mouse cursor. Outlines were constructed either geometrically so as to sample activity within anatomic areas (orbital frontal, cerebellum, mesial tempo¬ ral area, and dorsolateral-prefrontal cortex) or demarcated around the entire ROI based on visible anatomy (thalamus and
striatum).
Alignment of the axis between CT and PET scans was deter¬ mined by correspondence of basal skull bony landmark patterns. Rotation and translation variables for transverse alignment were determined by computer software based on perimeter outlines (derived from edge-finding techniques) on CT and matching PET transmission scans. Individual CT ROIs were then mapped to corresponding PET sections with computer software using these alignment variables. Absolute metabolism was determined as the average metabolic rate (in micromoles per 100 g per minute) for all pixels in the ROI. Relative metabolic rate was determined as the ratio of the ROI absolute rate to the whole-brain absolute rate. The whole-brain rate was determined as the weighted average of metabolic rates for the middle 10 PET sections (of a total of 14), excluding regions within sections representing ventricular cerebrospinal fluid (pix¬ el metabolic rates less than 18 µ /100 g per minute). In addition to their use for anatomic identification of ROIs, CT
RESULTS The results presented herein focus primarily on the relationship
between
negative symptoms and frontal metabolism and subse¬ quently on the relationship of these two variables with other clin¬ ical, structural, and neuropsychological measures. Overall, Abrams and Taylor28 negative symptoms scores ranged from 0 to 20 with a mean (±SD) of 7.2±6.4 and a median of 5.5, reflecting a positive skew. In all analyses, negative symptoms scores were transformed by square root to correct for this nonnormal distribution.
Cerebral Metabolism Correlations between Abrams and Taylor28 negative symptoms scores and relative regional glucose metabolism are presented in Table 1. Relative metabolic rates were used in these correlations because (1) variance in absolute metabolism predominantly reflects interindividual and intertest differences in overall meta¬ bolic rate rather than intracerebral distributional differences3335 and (2) hypofrontality has most often been manifest as a relative decrease in frontal lobe metabolism compared with metabolic ac¬ tivity elsewhere in the brain (see Buchsbaum36 for review). Negative symptoms were significantly and negatively corre¬ lated with frontal lobe metabolism in right dorsolateral prefron¬ tal cortex (Fig 2). In other words, the more hypofrontal the met¬ abolic distribution, the more severe the negative symptoms. Relative metabolism in this ROI accounted for nearly half of the variance in negative symptoms scores. The correlation between relative right dorsolateral prefrontal metabolism and negative symptoms was significantly larger than for the corresponding correlation for the left dorsolateral ROI (differences between de¬ pendent Pearson Correlation r's t= 2.54, P
cific. Furthermore, there
was no
evidence that this rela-
tionship was an artifact of age, cerebral atrophy, or severity of positive symptoms. (Arch Gen Psychiatry. 1992;49:959-965) lobe
utilization has been
frontal glucose Decreased reported many,17 though studies in
not
all,8"11 positron
of glucose metabo¬ emission tomographic (PET) lism in schizophrenia. While the pathophysiologic impor¬ tance of hypofrontality is unknown, it is suspected to re¬ late causally to negative symptoms.1-512"17 This hypothesis is largely based on the similarities between negative symptoms of schizophrenia and the sequelae of frontal lobe lesions described in numerous clinical and preclinical studies. For example, experimental lesions of the prefron¬ tal cortex in animals suggest that this area is involved in social affiliation and emotional expression; animals with lesions display marked social withdrawal and defects of emotional behavior.1718 Traumatic and surgically induced frontal lesions in humans may result in avolition, apathy, anhedonia, affective flattening, and impoverished speech.1819 Impairment in cognitive function, associated
Accepted for publication January 30, From the Department of Psychiatry, University Medical Center, New York (Drs Wolkin, Angrist, Brodie, and Rotrosen); the Psychiatry Service, New York Veterans Administration Medical Center, New York (Drs Wolkin, Angrist, and Rotrosen and Mr Sanfilipo); and the Chemistry Department, Brookhaven National Laboratory, Upton, NY 1992. New York
(Dr Wolf).
Reprint requests to the New York VA Medical York, NY 10010 (Dr Wolkin).
23rd St, New
Center (116A), 423 E
Angrist, MD;
with the negative symptom complex of schizophrenia,20"23 is also believed to reflect frontal lobe dysfunction.17,24 Herein, we report the relationship between negative symptoms and frontal lobe hypometabolism in a sample of 20 chronic schizophrenics who underwent scanning while not
receiving neuroleptic drugs.
SUBJECTS AND METHODS Twenty male subjects meeting DSM-III-R criteria for chronic schizophrenia as the only active Axis I disorder (11 paranoid, one disorganized, and eight undifferentiated) gave informed written consent to participate in our study. The subjects were inpatients at a veterans' hospital, where they were hospitalized for either acute or chronic decompensation of schizophrenia. Of the 20 subjects, 13 were black, six were white, and one was Hispanic. Based on a handedness inventory, 16 were right-handed and four
had mixed dominance or were left-handed. Overall, the mean (±SD) age was 35±5 years (range, 26 to 44 years). The mean du¬ ration of illness was 13 years (range, 7 to 24 years). Based on the findings of a history, a physical examination, electrocardiography, and routine blood chemistry analyses, all subjects were found to be in good physical health. Patients with a history of drug abuse within the preceding month were excluded from the
study. Subjects had not received oral neuroleptic drugs for a minimum of 2 weeks before PET scanning and had not received depot neu¬ roleptic drugs for a minimum of 3 months. The median washout period was 23 days (range, 14 to a coding cutoff of 365 days for three patients who had never received neuroleptic drugs). Six patients required treatment for agitation before PET scanning, in which case they received only choral hydrate (0.5 g) or lorazepam (0.5 mg) every 6 hours as needed. These medications were selected based on their relatively brief half-lives to minimize re¬ sidual drug effects at the time of PET scanning. No medication was administered in the 12 hours preceding scanning. A brief cognitive test battery was administered within 24 hours of PET scanning, including the Modified Wisconsin Card Sort Task,25 Category Retrieval Test,26 and the Digit-Span, DigitSymbol, and Vocabulary Subtests of the Wechsler Adult Intelli¬
gence Scale-Revised.27 Global psychopathologic findings were rated with the Brief Psychiatric Rating Scale (BPRS); negative symptoms were specif¬ ically assessed by the Abrams and Taylor28 Scale for Emotional Blunting. Positive symptoms ratings were based on the BPRS "schizophrenia" factor (hallucinations, unusual thought content, and conceptual disorganization). All ratings were made immedi¬ ately before PET scanning. Positron emission tomographic scans were obtained at Brookhaven National Laboratory, Long Island, NY, with a PET VI
Downloaded From: http://archpsyc.jamanetwork.com/ by a West Virginia University Library User on 06/22/2015
scans were evaluated for cortical atrophy and ventricular en¬ largement. Only those sections superior to ventricles were in¬ cluded in the analyses of cortical atrophy to avoid bias from ob¬ servation of enlarged ventricles. Cortical atrophy was assessed by two independent observers using a rating scale of 0 (none) to 4 (severe). All CT scans were rated by investigators who were blind to subject status. Ventricular enlargement was assessed based on
of frontal horn and third ventricle size. These measures included linear ratios (bifrontal and bicaudate ratios; see Wolkin et al32 for descriptions) and ventricular-to-brain ratios. The latter measures were determined via computer software that permits windowing of ventricular cerebrospinal fluid. Data were analyzed for departure from the normal distribution both visually and by Shapiro-Wilks and Lilliefors tests (comput¬ ed using SPSS/PC, version 3.0, "Examine" procedure; SPSS, Chi¬ cago, 111). Variables with nonnormal distribution were trans¬ formed either by square root or log transformations, and no departure from normality was detected by the above-mentioned tests. Because of the discontinuous rating scales, both parametric and nonparametric correlations were tested. The unique contri¬ bution of negative symptoms and hypofrontality on cognitive test performance was examined with multiple regression, whereby the independent variables (relative frontal metabolism and neg¬ ative symptoms) were entered simultaneously into the regression equation for each cognitive test. measures
Fig 1.—Schematic hemidiagrams of regions of interest used in positron emission tomographic image analysis. See text for a description of methods to identify corresponding computed tomographic sections. Cer indicates cerebellum; DLPF, dorsolateral prefrontal cortex; Mes, mesial temporal lobe; Orb, orbital-medial prefrontal cortex; Str, striatum; Tern, temporal cortex; and Tha, thalamus. in low-resolution mode (11.8 mm full width at half max¬ imum in plane of section and 14.4 mm in the axis) with 18 F-fluorodeoxyglose2'* as the metabolic tracer.3" Patients were po¬ sitioned with horizontal and vertical localizing lasers such that the head was parallel to the canthomeatal plane. Positioning was maintained by individually fitted foam-filled head holders. Scan¬ ning was performed with the subjects' eyes and ears open in a quiet, dimly lit room that did not preclude observation by the subject. Plasma sampling was accomplished through venous catheterization with use of a hand warmer to arterialize venous camera
blood.31
Computed tomographic (CT) scans were obtained with a Pickar scanner. Patients were positioned following the same procedure described for PET scanning, using their own individ¬ ually molded head holder to ensure consistency in positioning between PET and CT scanning and to maintain stability. The PET scans were analyzed based on individual anatomy as ascertained from each patient's matched CT scan. Given the sam¬ ple size and camera resolution, a conservative number of major regions of interest (ROIs) were assessed for both left and right hemispheres: dorsolateral prefrontal cortex, orbitomedial pre¬ frontal cortex, a large strip of temporal cortex, striatum, thalamus, mesial temporal cortex (amygdala/hippocampus), and cerebel¬ lum. These ROIs were selected from five CT sections (Fig 1) based on the following criteria: section I, largest area of cerebellum; sec¬ tion II, basal orbital and temporal lobes; section III, largest area of basal ganglia; section IV, thalamus most prominent; and sec¬ tion V, highest section still including lateral ventricle (transecting body of ventricle). The seven hemispheric ROIs were then 1200 CT
outlined on the monitor with the use of a mouse cursor. Outlines were constructed either geometrically so as to sample activity within anatomic areas (orbital frontal, cerebellum, mesial tempo¬ ral area, and dorsolateral-prefrontal cortex) or demarcated around the entire ROI based on visible anatomy (thalamus and
striatum).
Alignment of the axis between CT and PET scans was deter¬ mined by correspondence of basal skull bony landmark patterns. Rotation and translation variables for transverse alignment were determined by computer software based on perimeter outlines (derived from edge-finding techniques) on CT and matching PET transmission scans. Individual CT ROIs were then mapped to corresponding PET sections with computer software using these alignment variables. Absolute metabolism was determined as the average metabolic rate (in micromoles per 100 g per minute) for all pixels in the ROI. Relative metabolic rate was determined as the ratio of the ROI absolute rate to the whole-brain absolute rate. The whole-brain rate was determined as the weighted average of metabolic rates for the middle 10 PET sections (of a total of 14), excluding regions within sections representing ventricular cerebrospinal fluid (pix¬ el metabolic rates less than 18 µ /100 g per minute). In addition to their use for anatomic identification of ROIs, CT
RESULTS The results presented herein focus primarily on the relationship
between
negative symptoms and frontal metabolism and subse¬ quently on the relationship of these two variables with other clin¬ ical, structural, and neuropsychological measures. Overall, Abrams and Taylor28 negative symptoms scores ranged from 0 to 20 with a mean (±SD) of 7.2±6.4 and a median of 5.5, reflecting a positive skew. In all analyses, negative symptoms scores were transformed by square root to correct for this nonnormal distribution.
Cerebral Metabolism Correlations between Abrams and Taylor28 negative symptoms scores and relative regional glucose metabolism are presented in Table 1. Relative metabolic rates were used in these correlations because (1) variance in absolute metabolism predominantly reflects interindividual and intertest differences in overall meta¬ bolic rate rather than intracerebral distributional differences3335 and (2) hypofrontality has most often been manifest as a relative decrease in frontal lobe metabolism compared with metabolic ac¬ tivity elsewhere in the brain (see Buchsbaum36 for review). Negative symptoms were significantly and negatively corre¬ lated with frontal lobe metabolism in right dorsolateral prefron¬ tal cortex (Fig 2). In other words, the more hypofrontal the met¬ abolic distribution, the more severe the negative symptoms. Relative metabolism in this ROI accounted for nearly half of the variance in negative symptoms scores. The correlation between relative right dorsolateral prefrontal metabolism and negative symptoms was significantly larger than for the corresponding correlation for the left dorsolateral ROI (differences between de¬ pendent Pearson Correlation r's t= 2.54, P
