- Email: [email protected]
Ventricular Enlargement in Poor-Outcome Schizophrenia
ORIGINAL ARTICLES Ventricular Enlargement in Poor-Outcome Schizophrenia Kenneth L. Davis, Monte S. Buchsbaum, Lina Shihabuddin, Jacqueline Spiegel-Cohen, Michael Metzger, Ede Frecska, Richard S. Keefe, and Peter Powchik Background: A subset of patients with schizophrenia, defined on the basis of longitudinal deficits in self-care, may show a classic (“Kraepelinian”) degenerative course. An independent validator of the phenomenologically defined Kraepelinian subtype might be provided by a structural indicator of possible brain degeneration: ventricular size as measured by computed tomography (CT). Methods: To examine whether Kraepelinian patients would show a differential increase in ventricular size over time, two CT scans were conducted at intervals separated by .4 years, an average of 5 years. Fifty-three male patients with DSM-III-R diagnoses of chronic schizophrenia were subdivided into Kraepelinian (n 5 22; mean age 5 42 6 8.6 years) and non-Kraepelinian (n 5 31; mean age 5 38 6 12.2 years) subgroups. Kraepelinian patients were defined on the basis of longitudinal criteria: .5 years of complete dependence on others for life necessities and care, lack of employment, and sustained symptomatology. Thirteen normal elderly volunteers (mean age 5 60 6 17.8) were also scanned at 4-year intervals. CT measurements were made by raters without knowledge of subgroup membership. A semiautomated computer program was used to trace the anterior horn, lateral ventricles, and temporal horns for each slice level on which they were clearly seen. Results: The ventricles showed a bilateral increase in size over the 4-year interval in the Kraepelinian subgroup, more marked in the left hemisphere than the right. By contrast, neither the non-Kraepelinian subgroup nor the normal volunteers showed significant CT changes from scan 1 to scan 2. Conclusions: Thus, the longitudinal dysfunctions in selfcare that characterize the Kraepelinian patients were associated with an independent indicator of brain abnormality. Biol Psychiatry 1998;43:783–793 © 1998 Society of Biological Psychiatry
From the Department of Psychiatry, Mount Sinai School of Medicine, New York, New York. Address reprint requests to Dr. Kenneth L. Davis, Department of Psychiatry, Box 1230, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029-6574. Received August 14, 1997; revised December 10, 1997; accepted December 17, 1997.
© 1998 Society of Biological Psychiatry
Key Words: Kraepelinian subtype, brain atrophy, skull size, temporal lobe, brain asymmetry
Introduction
A
lthough there is a general consensus that ventricular enlargement is present in patients with schizophrenia (for review, see Shelton and Weinberger 1986), there is less agreement in more recent reviews about whether this enlargement is a static change early in the course of illness (Vita et al 1997) or a progressive lesion (DeLisi et al 1997). Longitudinal studies are critical to answer this question, since in cross-sectional studies the patient’s age, age of illness onset, duration of illness, and duration of treatment are confounded. Some longitudinal studies have found no change in ventricular size over time (Nasrallah et al 1986; Illowsky et al 1988; Vita et al 1988, 1994; Degreef et al 1991; Sponheim et al 1991; Vita 1991; Jaskiw et al 1994), whereas others have found increased size with repeat scans (Kemali et al 1989; Woods et al 1990; DeLisi et al 1995, 1997; Nair et al 1997). The possibility that only a subgroup of patients with more severe or unremitting illness might show progressive ventricular enlargement is suggested by the finding of ventricular enlargement on repeat scans in patients who had undergone repeated hospitalizations (Woods et al 1990), those who failed to remit (Frecska et al 1994; Lieberman et al 1996), and those who were less medication compliant (DeLisi et al 1997). In contrast, patients studied in their first episodes of schizophrenia, for whom the chronicity of course is as yet unknown, were more likely to show no progressive change in the ventricular system (Degreef et al 1991; Sponheim et al 1991; Jaskiw et al 1994; Vita et al 1994; DeLisi et al 1995). Recent cluster-analytic strategies to assess the rate of ventricular volume change also support the existence of a subgroup with enlarging ventricles (Nair et al 1997). Computed tomographic (CT) studies, including the current report, typically produce somewhat less reliable measurement of ventricular size than high-resolution magnetic resonance imaging (MRI) due to relatively thick (8 –10 mm) slices; however, 1-mm-thick MRI planes were 0006-3223/98/$19.00 PII S0006-3223(97)00553-2
784
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
not available 10 years ago when this study began, and at present only CT is available for long-term follow-up. Relatively short follow-up intervals may also have diminished the statistical power of some studies to detect progressive change and favored the null hypothesis. Higher resolution volumetric studies with larger sample sizes and longer follow-up intervals have tended to be more likely to reveal progressive enlargement. For example, a 2-year follow-up by DeLisi et al (1992) did not reveal the progressive enlargement that became apparent at 4 (DeLisi et al 1995) and 5 years (DeLisi et al 1997). Thus, the issue of progressivity versus stability of ventricular size remains not entirely resolved. Based on these data, a reformulated “progressivity of ventricular enlargement hypothesis” would seek to contrast poor and better prognosis schizophrenic patients. Such a design would have the advantage of controlling for nuances involved in repeated CT scans, by contrasting two sets of schizophrenic patients who have undergone two scans on the same scanner. This hypothesis would be best tested if the scanning interval were relatively long in duration and the sample comprised patients with ages toward the fifth decade of life, the time at which agerelated changes in ventricular size begin to accelerate (Waddington et al 1991). Over the last decade, a group of schizophrenic patients with very poor outcome, termed “Kraepelinian,” have been characterized (Keefe et al 1987). These are patients who have been shown to have at least 5 years of continuous inability to provide themselves with food, clothing, or shelter without the help of a caretaker. Evidence indicates that Kraepelinian patients differ from less chronic (non-Kraepelinian) patients in being less responsive to neuroleptics, having more severe positive and negative symptoms, and having a poorer level of premorbid social and sexual adjustment (Keefe et al 1988, 1989, 1990, 1991, 1993, 1996; Losonczy et al 1986a, 1986b). Hence, this population is a reasonable one in which to test the hypothesis that poor-outcome patients constitute the schizophrenic subgroup that shows progressive ventricular enlargement. Thus, the present study compared a relatively large group of Kraepelinian and non-Kraepelinian schizophrenic patients who underwent two CT scans on the same scanner, with at least 4 years between scans, and who were entering the fifth decade of life.
Methods and Materials Subjects CT scans were acquired from 53 male patients with chronic schizophrenia and 13 normal volunteers on two occasions at least 4 years apart. Patients were recruited from the inpatient and outpatient programs of the Schizophrenia Biological Research
Center of the Bronx Department of Veterans Affairs Medical Center, Pilgrim Psychiatric Center, and Mount Sinai Medical Center. All patients met DSM-III-R criteria for chronic schizophrenia (American Psychiatric Association 1987) as determined by a two-member diagnostic team that used the Schedule for Affective Disorders and Schizophrenia (Endicott and Spitzer 1978). The diagnostic team had excellent interrater reliability [intraclass correlation coefficient (ICC) 5 .90]. Patients with significant medical illnesses, substance abuse, or other neurological illness were excluded. Data from a subset of these patients not followed longitudinally constituted a separate report (Frecska et al in preparation). The schizophrenic patients were divided into Kraepelinian and non-Kraepelinian clinical subgroups on the basis of the following criteria: 1) continuous hospitalization or complete dependence on others for food, clothing, and shelter; 2) no useful employment or work; and 3) no evidence of remission of symptoms. Twenty-two patients (mean 6 SD age 5 42 6 8.6 years) met the criteria for the Kraepelinian subtype, and 31 (mean age 5 38 6 12.2 years) did not (not significantly different by t test). Two independent diagnosticians achieved a high degree of reliability in the determination of Kraepelinian/non-Kraepelinian status (ICC 5 .95). Demographic data included age of illness onset, number of months of hospitalization, marital status, employment, and years of education. Clinical evaluations with the Scale for the Assessment of Negative Symptoms (SANS; Andreasen 1983) were repeated every 2 years during the follow-up period. The interval between CT scans was 62.4 months (SD 5 12.7) in the non-Kraepelinian and 58.5 months (SD 5 13.1) in the Kraepelinian patients. Patients provided informed consent for each CT scan and each clinical evaluation. Normal older men (n 5 13, mean age 5 60 years, SD 5 17.8) were also scanned at 4-year intervals (mean 5 63 months, SD 5 10.2) using the same exclusion criteria and medical screening in addition to the requirement of no psychiatric illness in self or first-degree relatives.
CT Measures All CT scans were collected on a Picker 2000 SX scanner with a high-resolution matrix and 512 projections per slice using a constant technique for the entire period to make longitudinal comparison possible. Scan slices were taken with the head parallel to the orbitomeatal line at a thickness of 8 mm per slice, which was standard technique at study inception in 1987. The CT transaxial images were compared to plates in the atlas of Matsui and Hirano (1978) illustrating CT and postmortem brain slices obtained at 20 degrees from the canthomeatal line; each CT slice intercepting the ventricle was labeled with the corresponding level in the Matsui–Hirano atlas identified by percentage of head height (Figure 1). Subjects with any missing or defective slices were excluded. All scans were analyzed by tracers without knowledge of subgroup diagnosis. First, a computer-automated edging program was used to obtain an outlined edge and whole brain slice volume for each slice. The anterior horn, lateral ventricles, and temporal horns were traced for each of the two slice levels on which they were clearly seen (Figure 1). Each region of interest was traced using a Sobel filter that enhances the
Ventricular Enlargement and Schizophrenia
regional borders and increases interrater reliability. The lateral ventricle was outlined in eight subjects by two tracers, each without knowledge of the results obtained by the other (ICC for size 5 .979). The temporal horn was outlined on two levels in 10 subjects by two tracers for the left and right hemisphere (ICC for size .87–.94). After a region had been outlined, the coordinates, and normalized area (area/whole slice area or ventricle– brain ratio, VBR), and volumetric (mm3) data were stored for later analysis. By analysis of relative data (region/whole slice area), “global scaling factors” or the constant individual differences in whole brain size were removed. Alternative statistical adjustment methods for cerebral volume are discussed in Lange et al (1997), but may require larger samples.
Statistical Methods Data on area of the ventricles at every slice level were available to provide volumetric analysis. Since different ventricular regions might reveal changes in brain size in different brain areas, both regional and total ventricular changes over time were considered. Separate statistical evaluation of every brain region has the hazard of Type I statistical error. For this reason, the data from group comparisons were first examined by an omnibus repeated-measures multivariate analysis of variance (ANOVA) (BMDP 4V; Dixon et al 1985). Groups were independent dimensions (Kraepelinian, non-Kraepelinian, normal volunteer). There were four repeated-measures dimensions: date (baseline, second scan), regions (anterior, lateral, temporal), hemispheres (right, left), and slice levels (ventral or 28% and dorsal or 34% of head height for the anterior and temporal horns, and ventral or 41% and dorsal or 48% of head height for the lateral ventricles). Our main hypothesis was tested by the group by region interaction (one single F and p value for the entire study), and higher order interactions were examined to establish regional differences. All significant effects are presented. Since the simplest null hypothesis is that neither schizophrenia subgroup nor time interval was a significant source of variation, we first examined the main effects of group and time; when this analysis proved statistically significant, we present the multiway interactions and simple contrasts (follow-up ANOVA) for group, time, and associated interactions. Correlations between ventricular mea-
785
BIOL PSYCHIATRY 1998;43:783–793
Table 1. Ventricle–Brain Ratios at the Dorsal and Ventral Brain Levels Subtype Kraepelinian Ventral Dorsal Non-Kraepelinian Ventral Dorsal Normal older volunteers Ventral Dorsal
Anterior
Temporal
Lateral
0.76 1.19
(0.35) (0.41)
0.99 1.26
(0.61) (0.63)
3.40 1.08
(1.21) (1.10)
0.66 0.96
(0.35) (0.47)
0.55 0.91
(0.35) (0.47)
3.30 1.28
(1.16) (1.14)
0.57 0.94
(0.30) (0.35)
0.52 0.85
(0.39) (0.42)
3.25 1.76
(1.04) (1.46)
Measurements are averages across both scan assessments. Two-group ANOVA (Kraepelinian vs. non-Kraepelinian), group 3 ventricular region 3 slice level interaction: F 5 4.62, df 5 2,50, p 5 .02. Three-group ANOVA, group 3 ventricular region 3 slice level interaction: F 5 4.44, df 5 2.78,87.4, p 5 .007. Simple interactions for each structure, temporal, group effect: F 5 9.22, df 5 2,63, p 5 .0003; lateral, slice 3 group: F 5 6.06, df 5 2,63, p 5 .0039. Simple interactions for normal subjects vs. Kraepelinian group, group 3 region 3 slice: F 5 7.30, df 5 1.39,45.86, p 5 .0048. Simple interactions at baseline only for normal vs. patient groups, group 3 region: F 5 5.08, df 5 2.60,81.7, p 5 .0043; group 3 slice 3 region: F 5 3.23, df 5 2.76,86.8, p 5 .029. Standard deviations are shown in parentheses.
sures and clinical variables should be considered exploratory. We present the correlations with the ventricular region that showed the greatest statistically confirmed group difference in enlargement (on follow-up ANOVA), the left lateral ventricle, and the total ventricular measure to minimize Type I statistical error associated with testing all ventricular regions.
Results Ventricular Size in Schizophrenia Subtypes The VBRs were entered in a five-way analysis of variance (ventricular size 3 hemisphere 3 slice level 3 date 3 patient group). At the time of the first scan on entrance into the study, the Kraepelinian and non-Kraepelinian subgroups did not differ in total relative ventricular volume when averaged across ventricular region and hemisphere (1.32, SD 5 1.06 vs. 1.27, SD 5 1.16, respectively). When regional differences were examined, however, Kraepelinian patients had larger values in the anterior and temporal ventricles, whereas the lateral ventricles showed no differences (Table 1; simple interaction for year 1, region 3 group interaction: F 5 5.08, df 5 2.60,81.7, p 5 .0043).
Ventricular Size Increase over Follow-up Interval Figure 1. CT slices intersecting lateral, anterior, and temporal divisions of the ventricular system. CT scans with areas of ventricle filled (green) on four contiguous slices demonstrate division of ventricles into the three areas entered in ANOVA (see Tables 1 and 2).
The Kraepelinian subgroup showed an increase in total size of the VBR over the 5-year period (from 1.32 to 1.58), unlike the non-Kraepelinian group (from 1.27 to 1.28; group 3 time interaction: F 5 7.62, df 5 1,51, p 5 .008). This change was confirmed in the lateral ventricle alone (year 3 group simple interaction; Table 2), whereas
786
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
Table 2. Ventricle–Brain Ratios of the Anterior, Temporal, and Lateral Horns by Hemisphere Anterior Subtype Kraepelinian Baseline Left Right L1R Second scan at mean interval of 5 years Left Right L1R Non-Kraepelinian Baseline Left Right L1R Second scan at mean interval of 5 years Left Right L1R
Temporal
Lateral
Total
Mean
(SD)
Mean
(SD)
Mean
(SD)
Mean
(SD)
0.95 0.89 0.92
(0.38) (0.40) (0.41)
1.01 1.02 1.01
(0.53) (0.63) (0.53)
2.02 2.03 2.02
(1.48) (1.50) (1.48)
1.33 1.31 1.32
(1.05) (1.07) (1.06)
1.07 0.98 1.03
(0.46) (0.46) (0.46)
1.27 1.20 1.24
(0.69) (0.74) (0.71)
2.60 2.32 2.46
(1.85) (1.69) (1.77)
1.64 1.50 1.58
(1.34) (1.23) (1.29)
0.85 0.78 0.82
(0.46) (0.41) (0.43)
0.68 0.70 0.69
(0.44) (0.40) (0.42)
2.41 2.22 2.31
(1.48) (1.41) (1.44)
1.31 1.23 1.27
(1.21) (1.12) (1.16)
0.84 0.78 0.81
(0.45) (0.43) (0.44)
0.79 0.75 0.77
(0.53) (0.43) (0.48)
2.27 2.23 2.25
(1.61) (1.62) (1.61)
1.30 1.25 1.28
(1.22) (1.21) (1.21)
Group 3 date interaction: F 5 7.62, df 5 1,51, p 5 .008; group 3 date 3 hemisphere interaction: F 5 11.41, df 5 1,51, p 5 .00014; group 3 date 3 ventricular region 3 hemisphere: F 5 6.20, df 5 1.58,80.6, p 5 .0059. Simple interactions with group and date for each structure tested separately, lateral horn for group 3 date 3 hemisphere: F 5 12.44, df 5 1,51, p 5 .0009. Simple interactions with group and date for each hemisphere tested separately, left hemisphere for group and date: F 5 12.43, df 5 1,51, p 5 .0009. Simple interactions with date for each group tested separately, Kraepelinian group, main effect of date: F 5 13.58, df 5 1,51, p 5 .0006; date 3 hemisphere: F 5 13.22, df 5 1,51, p 5 .0006; date 3 region 3 hemisphere: F 5 3.46, df 5 1.58,80.6, p 5 .0468. Simple interactions with group for each date tested separately, baseline, region 3 group: F 5 4,71, df 5 1.27,64.9, p 5 .02; second scan, group 3 region by hemisphere: F 5 3.79, df 5 1.81,92.4, p 5 .02.
similar follow-up analyses for anterior and temporal ventricular regions were not statistically significant. Since the effect was confirmed for the left lateral ventricle in follow-up ANOVA, we examined the frequency distributions for the baseline, follow-up, and change scores. As shown in Figure 2, the relative size of the lateral ventricle in the entire patient group showed a distribution that did not differ significantly from normal at baseline but was bimodal at the follow-up scan (Kolmogorov–Smirnov test, x2 5 25.7, df 5 4, p 5 .000037). The difference scores showed proportionately greater enlargement in the Kraepelinian group (Figure 3), but the frequency distribution of the difference scores for the entire group did not show a significant deviation from normality on the Kolmogorov– Smirnov test. Because we anticipated that the Kraepelinian but not the non-Kraepelinian subgroup would increase VBR over time, we examined the correlation between duration of illness and total VBR and observed a significant association with duration of illness in the Kraepelinian group (r 5 .35, p , .05, one-tailed) but not the non-Kraepelinian group (r 5 .06, p 5 ns). Hemispheric differences in the degree of enlargement of the total VBR were also found, with the Kraepelinian subgroup showing a greater ventricular size increase over
the 5-year interval on the left than on the right side (Figures 4 –7); this asymmetrical enlargement was significantly more prominent in the lateral horns (Table 2). We also expressed ventricular enlargement over the 5-year period as a percentage of initial size and observed similar findings. Examining the left dorsal lateral ventricle (selected for confirmation with the follow-up ANOVA in Table 2), we obtained the expected correlation with Kraepelinian status (r 5 .35 for the dorsal slice and r 5 .38 for the ventral slice). Percent ventricular enlargement was also correlated significantly with months of hospitalization in all patients (r 5 .34, p , .05), but not with age, duration of illness, or the other clinical variables in Table 3. Because we expected duration of illness possibly to be correlated with ventricular enlargement in Kraepelinian patients only, we examined this correlation in each group separately (r 5 .22 in Kraepelinian patients and r 5 .22 in non-Kraepelinian patients for left dorsal lateral ventricle percentage increase).
Ventricular Size in Normal Volunteers Normal older individuals had no significant change in ventricular size over the 5-year interval between scans (four-way ANOVA on normal subjects; no interaction that
Ventricular Enlargement and Schizophrenia
BIOL PSYCHIATRY 1998;43:783–793
787
Figure 2. Frequency histograms of left lateral ventricle in all patients at baseline, year 5, and difference. obs, observations. Non-Kraepelinian, n 5 31; Kraepelinian, n 5 22.
contained date reached statistical significance). When compared with the relatively younger schizophrenic patients, the older normal subjects had generally smaller ventricles, especially for the anterior and temporal horns; only the dorsal lateral ventricle was larger (Table 1).
Skull Shape The possibility that the Kraepelinian and non-Kraepelinian subgroups might be characterized by a different skull contour was examined in a five-way analysis of variance
Figure 3. Individual ventricular size at baseline and year 5 in Kraepelinian and non-Kraepelinian patients.
788
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
(group 3 slice level 3 hemisphere 3 anteroposterior segment 3 date). Assessment of the relative skull width measurements at the seven anteroposterior positions at the five slice levels used in the VBR analysis failed to reveal any significant group effect or interaction.
Clinical Correlations with Ventricular Size in Kraepelinian Patients Patients with larger ventricles had spent more months in the hospital; this correlation was significant (p , .05) for both Kraepelinian (scan 1: r 5 .43; scan 2: r 5 .54) and non-Kraepelinian patients (scan 1: r 5 .39; scan 2: r 5 .45). The correlation for scan 2 remained significant (p , .05) for Kraepelinian patients after the effects of age were partialled out.
Clinical Correlations in Non-Kraepelinian Patients
Figure 4. Tally plots of the Kraepelinian patients for the sizes and differences in ventricular shape at year 1 and year 5 at the 41% and 47% of head height levels (Matsui and Hirano 1978; slices 8 and 7, respectively) or approximately 24- and 20-mm levels (Talairach and Tournoux 1988). The left side of the picture is the patient’s left. Top color bar shows number of subjects for each pixel position whose ventricular outline is outside this x–y location. Thus the center of the structure is white (100%) or red (93– 80%), indicating that at this position, nearly all subjects’ outlines were outside this area (all subjects have pixels in this area). The bottom row shows the difference, pixel by pixel, between year 5 and year 1. Red and yellow show areas that were inside each subject’s outline for year 5 but outside for year 1, indicating enlargement of the ventricle at this point. Red areas appear laterally in the most dorsal location, appear medially and posteriorly in the more ventral slice, and are more prominent in the left hemisphere.
As was true for the Kraepelinian patients, larger ventricular size in non-Kraepelinian patients was correlated with age at both scans 1 (r 5 .34, p , .05) and 2 (r 5 .38, p , .05). There were no significant correlations between total SANS change scores and ventricular change.
Discussion Poor-outcome (i.e., Kraepelinian) schizophrenic patients showed significantly greater increases in total ventricular volume over an approximate 5-year interval between baseline and follow-up CT scans than did better-prognosis (i.e., non-Kraepelinian) patients. Although all areas of the ventricular system showed increases in size in the Kraepelinian subgroup, only the lateral ventricles were statistically significantly enlarged compared with values in the non-Kraepelinian subgroup over the 5-year interval. Fur-
Figure 5. Ventricular size in Kraepelinian and non-Kraepelinian patients. Increase in total ventricle/brain ratio is greater in Kraepelinian than non-Kraepelinian patients over 5-year interval (see marginal means and ANOVA in Table 2).
Ventricular Enlargement and Schizophrenia
Figure 6. Lateral ventricle contours. Green line shows contour that includes 80% of subjects at baseline, and red line shows 80% contour at year 5. Note enlargement on lateral and posterior margins of ventricle especially on the left. This simplified view of Figure 2 is complementary in demonstrating that the enlargement is not due only to a few outliers, since the extreme fifth of the population is not shown. The contour line is found by testing outward from the center of each ventricle at 720 angular positions and counting the number of ventricular outlines that are intersected. Axial level given above AC (anterior commissure)PC (posterior commissure) line.
thermore, within the Kraepelinian subgroup, enlargement was significantly larger in the left lateral ventricle than in the right. A group of older normal volunteers, whose age coincided with the period of life at which age-related ventricular enlargement peaks (Waddington et al 1991), did not show a statistically significant ventricular size increase over the same time interval and with the same scanner that detected ventricular changes in the patients with the Kraepelinian subtype of schizophrenia. Although some key features of this longitudinal study of ventricular enlargement are unique, it is by no means the first longitudinal study of ventricular enlargement in schizophrenia; however, it is the first study to have compared subgroups that were characterized, a priori, as comprising better- vs. poor-outcome patients. The sample is also the largest one reported to date in which patients were followed for a mean duration of 5 years. The characteristics of our study design (large sample size, long follow-up interval, older patients, better- vs. poor-outcome subgroups) may help shed light on some of the divergent findings reported in the literature. Heterogeneous sampling of patients could have accounted for only subgroups of schizophrenic patients having shown progressive ventricular enlargement on repeat scanning in some studies (Vita et al 1988; Nasrallah et al 1986; DeLisi et al 1992). Indeed, samples in those studies that have previously found progressive ventricular enlargement were composed of very chronic patients (Kemali et al 1989; Woods et al
BIOL PSYCHIATRY 1998;43:783–793
789
1990); however, the stability of ventricular size found in one study of 15 patients with chronic schizophrenia, rescanned after 7–9 years of continuous illness, including 7 continuously hospitalized patients, remains a result that cannot be reconciled with the current or other data (Illowsky et al 1988). An intriguing hypothesis could be suggested that more severe illness with more inpatient care and the associated diminution in sensory stimulation could result in brain tissue loss. This question is partly addressed by the CT scans in the Genain quadruplets, identical quadruplets with schizophrenia (Buchsbaum et al 1984). Their life histories in the two decades before scanning varied widely, with one quad hospitalized for 15 years and one hospitalized little, but living at home with spouse and children; however, their CT scans were remarkably similar, suggesting greater genetic than sensory environment effects on ventricular size. Before the advent of CT studies, pneumoencephalography (PEG) was used to study longitudinal changes in the ventricular system. The first report of progressive ventricular enlargement derived from an investigation by Moore et al (1935). Heterogeneity of ventricular enlargement with PEG was apparent, with the phenomenon being linked to progressive cognitive and clinical deterioration (Haug 1962; Huber 1957). Taken together, therefore, the combined findings of PEG and CT studies lead to the credible assertion that there exists a subgroup of pooroutcome schizophrenia characterized by progressive ventricular enlargement. One cannot help but wonder if such patients followed into older age would not be those who were viewed as having a degenerating course (Kraepelin 1913, 1919) and have been more recently described in their seventh, eighth, and ninth decades of life as being frankly demented (Davidson et al 1995). A host of potential artifacts can influence studies using CT methodology in schizophrenia. When subjects are scanned twice, the angle of the head and the axial head position are only approximately equal. The resultant variability could account for random differences in ventricular volume between the first and second scans that would increase Type II statistical error. There is no reason to suspect, however, that there should be a systematically different bias between Kraepelinian and non-Kraepelinian patients on head position and slice level over time. The use of two contemporaneously scanned comparison groups for the Kraepelinian patients—namely, age-matched nonKraepelinian patients and aged normal subjects—provides additional information on random changes in position and slice level variability between scans. Ventricular enlargement has been noted in anorexia (e.g., Swayze et al 1996) and in alcoholism (Pfefferbaum et al 1993). Therefore, if Kraepelinian patients had sub-
790
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
Figure 7. Individual ventricles at baseline and year 5 of study. Top row shows 4 Kraepelinian patients at baseline and year 5 with marked ventricular enlargement prominent in temporal horn. Bottom row shows stability in ventricular size in non-Kraepelinian patients.
stantially poorer nutritional status or alcoholism at the time of their second scan, this could have accounted for the findings of larger ventricles. That explanation seems unlikely for the following reasons. All first scans were conducted in patients when they were inpatients and generally experiencing an exacerbation of illness. In contrast, second scans were performed at times when patients were not experiencing an acute exacerbation and when their symptoms were generally stable. Hence, anorexia, poor nutrition, and the stress of psychotic exacerbation would be more likely to have been factors during the first scan than the second scan, biasing against the results seen
here. Patients with alcoholism were excluded both for the first and second scans, so this source of potential ventricular enlargement is also unlikely to be a major contributor to the effect. Age affects the total volume of the ventricular system. The relationship between age and ventricular volume is quadratic, concave up (Waddington et al 1991; Sullivan et al 1993), with the greatest rate of increase in size per year occurring after the fifth decade of life. The mean age of the Kraepelinian schizophrenic patients in this study, at the time of their second scan, enters the inflection point for change of ventricular size. Kraepelinian and non-Kraepe-
Table 3. Patient Description Kraepelinian
Age Mother’s age at patient birth Years of education Number of hospitalizations Months of hospitalization Age at illness onset Currently employed
Non-Kraepelinian
All patients
Mean
SD
Mean
SD
Mean
SD
42.0 29.2 12.1 8.6 132.2 20.9
8.6 8.1 2.0 9.8 99.7 4.9
37.5 24.5 11.6 6.6 29.6 23.6
12.1 4.6 2.5 4.6 49.0 6.0
39.5 25.6 11.8 8.0 66.4 23.1
11.6 5.8 2.4 7.4 85.7 5.69
0.0%
20.0%
13%
Ventricular Enlargement and Schizophrenia
linian patients did not significantly differ in age, and both had larger ventricles than the older controls; however, some earlier longitudinal CT studies rescanned patients at an age when relatively small volumetric changes were occurring, and as a consequence any progressive enlargement of ventricular volume would not easily be detected. While it would have been useful to have a completely age-matched normal sample, obtaining CT scans in normal subjects is not without difficulty. Continuation studies with high-resolution MRI and without radiation exposure will yield more definitive age-rate comparisons. An obvious criticism of the current study is that it employed CT scanning rather than higher resolution MRI. A false-positive result, however, would not derive from a lower resolution imaging technique, which would be more likely to be affected by random sampling error and fail to find changes over time. Higher resolution MRI used with the current cohort would probably have revealed the same effects, although better definition of the temporal horns might have not only revealed the baseline differences we did observe but made a more definitive regional longitudinal finding possible. Schizophreniform patients, scanned at the time of their first episode, tended to show larger left than right ventricles, unlike control subjects (DeLisi et al 1992). Patients with schizotypal personality disorder assessed with highresolution MRI also showed left rather than right temporal horn enlargement (Buchsbaum et al 1997). Enlargement also appeared larger in the left posterior lateral ventricle in individuals carrying a chromosome-5 marker allele who were members of a pedigree in which the linkage marker was associated with schizophrenia (Shihabuddin et al 1996). This asymmetry in ventricular size suggests that ventricular enlargement might, in part, be a lateralized process. Consistent with this possibility is that left cerebral hypodensity has also been seen on CT scans (Reveley et al 1987), and neuropathological changes in postmortem tissue are often more pronounced on the left (Brown et al 1986; Jakob and Beckmann 1986). Such asymmetry has even extended to dopamine in the left amygdala (Reynolds 1983). Neurochemical abnormalities may also extend to left temporal glutamatergic dysfunction in schizophrenia (Deakin et al 1989). The possibility that schizophrenia is a disease of asymmetry has been hypothesized (Crow et al 1989). Taken together, these data would support the notion that a more active pathological process occurs on the left side of the brain in these very chronic Kraepelinian schizophrenic patients, and could account for the greater enlargement seen on the left than on the right in the current study. Progressive enlargement in ventricular volume can be seen as evidence of a dynamic, rather than a static lesion; however, this interpretation is in no way incompatible with
BIOL PSYCHIATRY 1998;43:783–793
791
the possibility that schizophrenia is associated with a neurodevelopmental abnormality (Murray and Lewis, 1987; Weinberger 1987). Conceivably, neurodevelopmental abnormalities could predispose to increased sensitivity to age-related effects. To view neurodevelopmental etiologies and progressivity as mutually exclusive is to lose sight of how genetically determined diseases such as trisomy 21 and Huntington’s disease display neurodevelopmental consequences at a relatively early age, with progressivity and even neurodegeneration much later in life. The cellular basis for the progressivity in ventricular size detected over approximately 5 years in very poorprognosis schizophrenic patients is clearly a matter of speculation. The absence of pronounced gliosis in postmortem studies of schizophrenia certainly suggests apoptotic rather than atrophic processes. The possibility that such effects may in part be mediated by excitotoxicity secondary to hyperglutamatergia has been suggested (Coyle 1976; Olney and Farber 1995). A challenge to these views is the ability to integrate hypoglutamatergia, to which is attributed some symptoms of schizophrenia, with hyperglutamateriga, to which is attributed cell death. One possibility is to implicate the loss of gamma-aminobutyric acidergic inhibitory interneurons (Olney and Farber 1995; Benes et al 1991). Many questions remain to be answered as the phenomenon of progressive ventricular enlargement in pooroutcome schizophrenia is further elucidated. For example, is the process linear or curvilinear? Are there associated and identifiable changes in tissue volume? What are the effects of neuroleptics upon the progression of ventricular changes? If the discourse on this topic can move away from the simple dichotomy of whether schizophrenia is a static or progressive lesion and move toward the elucidation of the phenomonology, it may be possible to begin to answer some of these questions.
This research was supported in part by the Schizophrenia Biological Research Center and the Program in Biological Psychiatry (grants from the Veterans Administration to K.L. Davis), and MH40071 to M.S. Buchsbaum from the National Institute of Mental Health. Cheuk Tang, Vladimir Tkach, and Dr. Tse-Chung Wei provided programming support, and Rita Amato and Granada Stephens assisted in the graphical analysis of the images. Bradley R. Buchsbaum assisted in additional graphics and statistical analysis. The consent was approved by the Mount Sinai Institutional Review Board.
References American Psychiatric Association (1987): Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R), 3rd ed rev. Washington, DC: American Psychiatric Press.
792
BIOL PSYCHIATRY 1998;43:783–793
Andreasen NC (1983): The Scale for the Assessment of Negative Symptoms. Iowa City, IA: University of Iowa. Benes F, McSparren J, Bird E, San Giovanni J, Vincent S (1991): Deficits in small interneurons in prefrontal and cingulate cortices of schizophrenic and schizoaffective patients. Arch Gen Psychiatry 48:996 –1001. Brown R, Colter N, Corsellis JAN, et al (1986): Post-mortem evidence of structural brain changes in schizophrenia. Differences in brain weight, temporal horn area, and parahippocampal gyrus compared with affective disorder. Arch Gen Psychiatry 43:36 – 42. Buchsbaum MS, Mirsky AF, DeLisi LE, et al (1984): The Genain quadruplets: Electrophysiological, positron emission, and X-ray tomographic studies. Psychiatry Res 13:95–108. Buchsbaum MS, Yang S, Hazlett EA, et al (1997): Ventricular volume and asymmetry in schizotypal personality disorder and schizophrenia assessed with magnetic resonance imaging. Schizophr Res 27:45–54. Coyle JT (1976): Neurotoxin action of kainic acid. J Neurochem 41:1–11. Crow TJ, Ball J, Bloom SR, et al (1989): Schizophrenia as an anomaly of development of cerebral asymmetry: A postmortem study and a proposal concerning the genetic basis of the disease. Arch Gen Psychiatry 46:1145–1150. Davidson M, Harvey PD, Powchik P, et al (1995): Severity of symptoms in chronically institutionalized geriatric schizophrenic patients. Am J Psychiatry 152:197–207. Deakin JFW, Slater P, Simpson MDC, et al (1989): Frontal cortical and left temporal glutamatergic dysfunction in schizophrenia. J Neurochem 52:1781–1786. Degreef G, Ashtari M, Wu H, Borenstein M, Geisler S, Lieberman J (1991): Follow-up MRI study in first episode schizophrenia. Schizophr Res 5:1204 –1206. DeLisi LE, Stritzke P, Riordan H, et al (1992): The timing of brain morphological changes in schizophrenia and their relationship to clinical outcome. Biol Psychiatry 31:241–254. DeLisi LE, Tew W, Xie S, et al (1995): A prospective follow-up study of brain morphology and cognition in first-episode schizophrenic patients: Preliminary findings. Biol Psychiatry 38:349 –360. DeLisi LE, Sakuma M, Tew W, Kushner M, Hoff AL, Grimson R (1997): Schizophrenia as a chronic active brain process: A study of progressive brain structural change subsequent to the onset of schizophrenia. Psychiatry Res Neuroimaging 74: 129 –140. Dixon WJ, Brown MB, Engelman L (1985): BMDP Statistical Software. Berkeley: University of California Press. Endicott J, Spitzer RL (1978): A diagnostic interview: The Schedule for Affective Disorders and Schizophrenia. Arch Gen Psychiatry 35:837– 844. Frecska E, Losonczy MF, Keefe RSE, et al (1994): Agedependent change of ventricular measures in schizophrenia: Cross-sectional analysis [abstract of the Society of Biological Psychiatry, Philadelphia, Pennsylvania]. Biol Psychiatry 35: 704. Frecska E, Losonczy MF, Keefe RSE, et al (in preparation): Ventricular enlargement in schizophrenia: Support for asymmetric age-dependent change.
K.L. Davis et al
Haug OJ (1962): Pneumoencephalographic studies in mental disease. Acta Psychiatr Scand 165:1–114. Huber G (1957): Pneumoencephalographische und Pychopathologische bilder bei endogen Psychosen. Berlin: Springer International. Illowsky BP, Juliano DM, Bigelow LB, Weinberger DR (1988): Stability of CT scan findings in schizophrenia: Results of an 8 year follow-up study. J Neurol Neurosurg Psychiatry 51:209 –213. Jakob H, Beckmann H (1986): Prenatal developmental disturbances in the limbic allocortex in schizophrenics. J Neural Transm 65:303–326. Jaskiw GE, Juliano DM, Goldberg TE, Hertzman M, UrowHamell E, Weinberger DR (1994): Cerebral ventricular enlargement in schizophreniform disorder does not progress: A seven year follow-up study. Schizophr Res 14:23–28. Keefe RSE, Mohs RC, Losonczy MF, et al (1987): Characteristics of very poor outcome schizophrenia. Am J Psychiatry 144:889 – 894. Keefe RSE, Mohs RC, Davidson M, et al (1988): Kraepelinian schizophrenia: A subgroup of schizophrenia? Psychopharmacol Bull 24:458 – 462. Keefe RSE, Mohs RC, Losonczy MF, et al (1989): Premorbid sociosexual functioning and long-term outcome in schizophrenia. Am J Psychiatry 146:206 –211. Keefe RSE, Mohs RC, Silverman JM, et al (1990): Characteristics of Kraepelinian schizophrenia and their relation to premorbid sociosexual functioning. In: Angrist B, Schulz SC, editors. The Neuroleptic-Nonresponsive Patient: Characterization and Treatment. Washington, DC: American Psychiatric Press, pp 3–21. Keefe RSE, Lobel DS, Mohs RC, et al (1991): Diagnostic issues in chronic schizophrenia: Kraepelinian schizophrenia, undifferentiated schizophrenia, and state-independent negative symptoms. Schizophr Res 4:71–79. Keefe RSE, Frecska E, Apter SH, et al (1996): Clinical characteristics of Kraepelinian schizophrenia: Replication and extension of previous findings. Am J Psychiatry 153:806 – 811. Kemali D, Maj M, Galderisi S, Milici N, Salvati A (1989): Ventricle-to-brain ratio in schizophrenia: A controlled follow-up study. Biol Psychiatry 26:753–756. Kraepelin E (1913): Lectures on Clinical Psychiatry New York: William Wood & Company. Kraepelin E (1919): Dementia Praecox and Paraphrenia Together with Manic-Depressive Insanity and Paranoia. Edinburgh: E&S Livingstone. Lange N, Giedd JN, Castellanos FX, Vaituzis AC, Rapoport JL (1997): Variability of human brain structure size. Psychiatry Res Neuroimaging 4:1–12. Lieberman JA, Alvir JA, Koreen A, et al (1996): Psychobiologic correlates of treatment response in schizophrenia. Neuropsychopharmacology 14:13S–21S. Losonczy MF, Song IS, Mohs RC, et al (1986a): Correlates of lateral ventricular size in chronic schizophrenia: I. Behavioral and treatment response measures. Am J Psychiatry 143:976 – 981. Losonczy MF, Song IS, Mohs RC, et al (1986b): Correlates of lateral ventricular size in chronic schizophrenia: II. Biological measures. Am J Psychiatry 143:1113–1118.
Ventricular Enlargement and Schizophrenia
Matsui T, Hirano A (1978): An Atlas of the Human Brain for Computerized Tomography. Tokyo: Igaku-Shoin. Moore MT, Nathan D, Elliot AR, et al (1935): Encephalographic studies in mental disease. Am J Psychiatry 92:43– 67. Murray RM, Lewis SW (1987): Is schizophrenia a neurodevelopmental disorder? Br Med J 295:681– 682. Nair TR, Christensen JD, Kingsbury SJ, Kumar NG, Terry WM, Garver DL (1997): Progression of cerebroventricular enlargement and the subtyping of schizophrenia. Psychiatry Res Neuroimaging 74:141–150. Nasrallah HA, Olson SC, McCalley-Whitters M, Chapman S, Jacoby CG (1986): Cerebral ventricular enlargement in schizophrenia. Arch Gen Psychiatry 43:157–159. Olney JW, Farber NB (1995): Glutamate receptor dysfunction and schizophrenia. Arch Gen Psychiatry 52:998 –1007. Pfefferbaum A, Sullivan EV, Rosenbloom MJ, Shear PK, Mathalon DH, Lim KO (1993): Increase in brain cerebrospinal fluid volume is greater in older than in younger alcoholic patients. Psychiatry Res 50:257–274. Reveley M, Reveley AM, Baldy R (1987): Left cerebral hemisphere hypodensity in discordant schizophrenic twins: A controlled study. Arch Gen Psychiatry 44:625– 632. Reynolds GP (1983): Increased concentrations and lateral asymmetry of amygdala dopamine in schizophrenia. Nature 305: 527–529. Shelton RC, Weinberger DR (1986): X-ray computed tomographic studies in schizophrenia: A review and synthesis. In: Nasarallah HA, Weinberger DR, editors. Handbook of Schizophrenia, vol 1. Amsterdam: Elsevier, pp 207–250. Shihabuddin L, Silverman JM, Buchsbaum MS, et al (1996): Ventricular enlargement associated with linkage marker for schizophrenia: Related disorders in one predigree. Mol Psychiatry 1:215–222. Sponheim RS, Iacono WG, Beiser M (1991): Stability of ventricular size after the onset of psychosis in schizophrenia. Psychiatry Res Neuroimaging 40:21–29. Sullivan EV, Shear PK, Mathalon DH, et al (1993): Greater
BIOL PSYCHIATRY 1998;43:783–793
793
abnormalities of brain cerebrospinal fluid volumes in younger than in older patients with Alzheimer’s disease. Arch Neurol 50:359 –373. Swayze VW, Andersen A, Arndt S, et al (1996): Reversibility of brain tissue loss in anorexia nervosa assessed with a computerized Talairach 3-D proportional grid. Psychol Med 26:381– 390. Talairach J, Tournoux P (1988): Co-planar Sterotaxic Atlas of the Human Brain. Three-Dimensional Proportional System: An Approach to Cerebral Imaging [translated by M. Rayport]. New York: Thieme Medical Publishers. Vita A (1991): Cerebral ventricular abnormalities in schizophrenia: Evidence of their early origin and stability over time. Schizophr Res 5:189 –190. Vita A, Sacchetti E, Valvassori G, Cazzullo CL (1988): Brain morphology in schizophrenia: A 2-to-5-year CT scan follow-up study. Acta Psychiatr Scand 78:618 – 621. Vita A, Giobbio GM, Dieci M, et al (1994): Stability of cerebral ventricular size from the appearance of the first psychotic symptoms to the later diagnosis of schizophrenia. Biol Psychiatry 35:960 –962. Vita A, Dieci M, Giobbio GM, Tenconi F, Invernizzi G (1997): Time course of cerebral ventricular enlargement in schizophrenia supports the hypothesis of its neurodevelopmental nature. Schizophr Res 23:25–30. Waddington JL, O’Callaghan E, Buckley P, et al (1991): The age dependencies of MRI abnormalities in schizophrenia suggest early ventricular enlargement but later prominence of cortical atrophy. Schizophr Res 5:188 –189. Weinberger DR (1987): Implications of normal brain development for the pathogenesis of schizophrenia. Arch Gen Psychiatry 44:660 – 669. Woods BT, Yurgelun-Todd D, Benes FM, Frankenburg FR, Pope HG Jr, McSparren J (1990): Progressive ventricular enlargement in schizophrenia: Comparison to bipolar affective disorder and correlation with clinical course. Biol Psychiatry 27:341–352.
From the Department of Psychiatry, Mount Sinai School of Medicine, New York, New York. Address reprint requests to Dr. Kenneth L. Davis, Department of Psychiatry, Box 1230, Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029-6574. Received August 14, 1997; revised December 10, 1997; accepted December 17, 1997.
© 1998 Society of Biological Psychiatry
Key Words: Kraepelinian subtype, brain atrophy, skull size, temporal lobe, brain asymmetry
Introduction
A
lthough there is a general consensus that ventricular enlargement is present in patients with schizophrenia (for review, see Shelton and Weinberger 1986), there is less agreement in more recent reviews about whether this enlargement is a static change early in the course of illness (Vita et al 1997) or a progressive lesion (DeLisi et al 1997). Longitudinal studies are critical to answer this question, since in cross-sectional studies the patient’s age, age of illness onset, duration of illness, and duration of treatment are confounded. Some longitudinal studies have found no change in ventricular size over time (Nasrallah et al 1986; Illowsky et al 1988; Vita et al 1988, 1994; Degreef et al 1991; Sponheim et al 1991; Vita 1991; Jaskiw et al 1994), whereas others have found increased size with repeat scans (Kemali et al 1989; Woods et al 1990; DeLisi et al 1995, 1997; Nair et al 1997). The possibility that only a subgroup of patients with more severe or unremitting illness might show progressive ventricular enlargement is suggested by the finding of ventricular enlargement on repeat scans in patients who had undergone repeated hospitalizations (Woods et al 1990), those who failed to remit (Frecska et al 1994; Lieberman et al 1996), and those who were less medication compliant (DeLisi et al 1997). In contrast, patients studied in their first episodes of schizophrenia, for whom the chronicity of course is as yet unknown, were more likely to show no progressive change in the ventricular system (Degreef et al 1991; Sponheim et al 1991; Jaskiw et al 1994; Vita et al 1994; DeLisi et al 1995). Recent cluster-analytic strategies to assess the rate of ventricular volume change also support the existence of a subgroup with enlarging ventricles (Nair et al 1997). Computed tomographic (CT) studies, including the current report, typically produce somewhat less reliable measurement of ventricular size than high-resolution magnetic resonance imaging (MRI) due to relatively thick (8 –10 mm) slices; however, 1-mm-thick MRI planes were 0006-3223/98/$19.00 PII S0006-3223(97)00553-2
784
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
not available 10 years ago when this study began, and at present only CT is available for long-term follow-up. Relatively short follow-up intervals may also have diminished the statistical power of some studies to detect progressive change and favored the null hypothesis. Higher resolution volumetric studies with larger sample sizes and longer follow-up intervals have tended to be more likely to reveal progressive enlargement. For example, a 2-year follow-up by DeLisi et al (1992) did not reveal the progressive enlargement that became apparent at 4 (DeLisi et al 1995) and 5 years (DeLisi et al 1997). Thus, the issue of progressivity versus stability of ventricular size remains not entirely resolved. Based on these data, a reformulated “progressivity of ventricular enlargement hypothesis” would seek to contrast poor and better prognosis schizophrenic patients. Such a design would have the advantage of controlling for nuances involved in repeated CT scans, by contrasting two sets of schizophrenic patients who have undergone two scans on the same scanner. This hypothesis would be best tested if the scanning interval were relatively long in duration and the sample comprised patients with ages toward the fifth decade of life, the time at which agerelated changes in ventricular size begin to accelerate (Waddington et al 1991). Over the last decade, a group of schizophrenic patients with very poor outcome, termed “Kraepelinian,” have been characterized (Keefe et al 1987). These are patients who have been shown to have at least 5 years of continuous inability to provide themselves with food, clothing, or shelter without the help of a caretaker. Evidence indicates that Kraepelinian patients differ from less chronic (non-Kraepelinian) patients in being less responsive to neuroleptics, having more severe positive and negative symptoms, and having a poorer level of premorbid social and sexual adjustment (Keefe et al 1988, 1989, 1990, 1991, 1993, 1996; Losonczy et al 1986a, 1986b). Hence, this population is a reasonable one in which to test the hypothesis that poor-outcome patients constitute the schizophrenic subgroup that shows progressive ventricular enlargement. Thus, the present study compared a relatively large group of Kraepelinian and non-Kraepelinian schizophrenic patients who underwent two CT scans on the same scanner, with at least 4 years between scans, and who were entering the fifth decade of life.
Methods and Materials Subjects CT scans were acquired from 53 male patients with chronic schizophrenia and 13 normal volunteers on two occasions at least 4 years apart. Patients were recruited from the inpatient and outpatient programs of the Schizophrenia Biological Research
Center of the Bronx Department of Veterans Affairs Medical Center, Pilgrim Psychiatric Center, and Mount Sinai Medical Center. All patients met DSM-III-R criteria for chronic schizophrenia (American Psychiatric Association 1987) as determined by a two-member diagnostic team that used the Schedule for Affective Disorders and Schizophrenia (Endicott and Spitzer 1978). The diagnostic team had excellent interrater reliability [intraclass correlation coefficient (ICC) 5 .90]. Patients with significant medical illnesses, substance abuse, or other neurological illness were excluded. Data from a subset of these patients not followed longitudinally constituted a separate report (Frecska et al in preparation). The schizophrenic patients were divided into Kraepelinian and non-Kraepelinian clinical subgroups on the basis of the following criteria: 1) continuous hospitalization or complete dependence on others for food, clothing, and shelter; 2) no useful employment or work; and 3) no evidence of remission of symptoms. Twenty-two patients (mean 6 SD age 5 42 6 8.6 years) met the criteria for the Kraepelinian subtype, and 31 (mean age 5 38 6 12.2 years) did not (not significantly different by t test). Two independent diagnosticians achieved a high degree of reliability in the determination of Kraepelinian/non-Kraepelinian status (ICC 5 .95). Demographic data included age of illness onset, number of months of hospitalization, marital status, employment, and years of education. Clinical evaluations with the Scale for the Assessment of Negative Symptoms (SANS; Andreasen 1983) were repeated every 2 years during the follow-up period. The interval between CT scans was 62.4 months (SD 5 12.7) in the non-Kraepelinian and 58.5 months (SD 5 13.1) in the Kraepelinian patients. Patients provided informed consent for each CT scan and each clinical evaluation. Normal older men (n 5 13, mean age 5 60 years, SD 5 17.8) were also scanned at 4-year intervals (mean 5 63 months, SD 5 10.2) using the same exclusion criteria and medical screening in addition to the requirement of no psychiatric illness in self or first-degree relatives.
CT Measures All CT scans were collected on a Picker 2000 SX scanner with a high-resolution matrix and 512 projections per slice using a constant technique for the entire period to make longitudinal comparison possible. Scan slices were taken with the head parallel to the orbitomeatal line at a thickness of 8 mm per slice, which was standard technique at study inception in 1987. The CT transaxial images were compared to plates in the atlas of Matsui and Hirano (1978) illustrating CT and postmortem brain slices obtained at 20 degrees from the canthomeatal line; each CT slice intercepting the ventricle was labeled with the corresponding level in the Matsui–Hirano atlas identified by percentage of head height (Figure 1). Subjects with any missing or defective slices were excluded. All scans were analyzed by tracers without knowledge of subgroup diagnosis. First, a computer-automated edging program was used to obtain an outlined edge and whole brain slice volume for each slice. The anterior horn, lateral ventricles, and temporal horns were traced for each of the two slice levels on which they were clearly seen (Figure 1). Each region of interest was traced using a Sobel filter that enhances the
Ventricular Enlargement and Schizophrenia
regional borders and increases interrater reliability. The lateral ventricle was outlined in eight subjects by two tracers, each without knowledge of the results obtained by the other (ICC for size 5 .979). The temporal horn was outlined on two levels in 10 subjects by two tracers for the left and right hemisphere (ICC for size .87–.94). After a region had been outlined, the coordinates, and normalized area (area/whole slice area or ventricle– brain ratio, VBR), and volumetric (mm3) data were stored for later analysis. By analysis of relative data (region/whole slice area), “global scaling factors” or the constant individual differences in whole brain size were removed. Alternative statistical adjustment methods for cerebral volume are discussed in Lange et al (1997), but may require larger samples.
Statistical Methods Data on area of the ventricles at every slice level were available to provide volumetric analysis. Since different ventricular regions might reveal changes in brain size in different brain areas, both regional and total ventricular changes over time were considered. Separate statistical evaluation of every brain region has the hazard of Type I statistical error. For this reason, the data from group comparisons were first examined by an omnibus repeated-measures multivariate analysis of variance (ANOVA) (BMDP 4V; Dixon et al 1985). Groups were independent dimensions (Kraepelinian, non-Kraepelinian, normal volunteer). There were four repeated-measures dimensions: date (baseline, second scan), regions (anterior, lateral, temporal), hemispheres (right, left), and slice levels (ventral or 28% and dorsal or 34% of head height for the anterior and temporal horns, and ventral or 41% and dorsal or 48% of head height for the lateral ventricles). Our main hypothesis was tested by the group by region interaction (one single F and p value for the entire study), and higher order interactions were examined to establish regional differences. All significant effects are presented. Since the simplest null hypothesis is that neither schizophrenia subgroup nor time interval was a significant source of variation, we first examined the main effects of group and time; when this analysis proved statistically significant, we present the multiway interactions and simple contrasts (follow-up ANOVA) for group, time, and associated interactions. Correlations between ventricular mea-
785
BIOL PSYCHIATRY 1998;43:783–793
Table 1. Ventricle–Brain Ratios at the Dorsal and Ventral Brain Levels Subtype Kraepelinian Ventral Dorsal Non-Kraepelinian Ventral Dorsal Normal older volunteers Ventral Dorsal
Anterior
Temporal
Lateral
0.76 1.19
(0.35) (0.41)
0.99 1.26
(0.61) (0.63)
3.40 1.08
(1.21) (1.10)
0.66 0.96
(0.35) (0.47)
0.55 0.91
(0.35) (0.47)
3.30 1.28
(1.16) (1.14)
0.57 0.94
(0.30) (0.35)
0.52 0.85
(0.39) (0.42)
3.25 1.76
(1.04) (1.46)
Measurements are averages across both scan assessments. Two-group ANOVA (Kraepelinian vs. non-Kraepelinian), group 3 ventricular region 3 slice level interaction: F 5 4.62, df 5 2,50, p 5 .02. Three-group ANOVA, group 3 ventricular region 3 slice level interaction: F 5 4.44, df 5 2.78,87.4, p 5 .007. Simple interactions for each structure, temporal, group effect: F 5 9.22, df 5 2,63, p 5 .0003; lateral, slice 3 group: F 5 6.06, df 5 2,63, p 5 .0039. Simple interactions for normal subjects vs. Kraepelinian group, group 3 region 3 slice: F 5 7.30, df 5 1.39,45.86, p 5 .0048. Simple interactions at baseline only for normal vs. patient groups, group 3 region: F 5 5.08, df 5 2.60,81.7, p 5 .0043; group 3 slice 3 region: F 5 3.23, df 5 2.76,86.8, p 5 .029. Standard deviations are shown in parentheses.
sures and clinical variables should be considered exploratory. We present the correlations with the ventricular region that showed the greatest statistically confirmed group difference in enlargement (on follow-up ANOVA), the left lateral ventricle, and the total ventricular measure to minimize Type I statistical error associated with testing all ventricular regions.
Results Ventricular Size in Schizophrenia Subtypes The VBRs were entered in a five-way analysis of variance (ventricular size 3 hemisphere 3 slice level 3 date 3 patient group). At the time of the first scan on entrance into the study, the Kraepelinian and non-Kraepelinian subgroups did not differ in total relative ventricular volume when averaged across ventricular region and hemisphere (1.32, SD 5 1.06 vs. 1.27, SD 5 1.16, respectively). When regional differences were examined, however, Kraepelinian patients had larger values in the anterior and temporal ventricles, whereas the lateral ventricles showed no differences (Table 1; simple interaction for year 1, region 3 group interaction: F 5 5.08, df 5 2.60,81.7, p 5 .0043).
Ventricular Size Increase over Follow-up Interval Figure 1. CT slices intersecting lateral, anterior, and temporal divisions of the ventricular system. CT scans with areas of ventricle filled (green) on four contiguous slices demonstrate division of ventricles into the three areas entered in ANOVA (see Tables 1 and 2).
The Kraepelinian subgroup showed an increase in total size of the VBR over the 5-year period (from 1.32 to 1.58), unlike the non-Kraepelinian group (from 1.27 to 1.28; group 3 time interaction: F 5 7.62, df 5 1,51, p 5 .008). This change was confirmed in the lateral ventricle alone (year 3 group simple interaction; Table 2), whereas
786
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
Table 2. Ventricle–Brain Ratios of the Anterior, Temporal, and Lateral Horns by Hemisphere Anterior Subtype Kraepelinian Baseline Left Right L1R Second scan at mean interval of 5 years Left Right L1R Non-Kraepelinian Baseline Left Right L1R Second scan at mean interval of 5 years Left Right L1R
Temporal
Lateral
Total
Mean
(SD)
Mean
(SD)
Mean
(SD)
Mean
(SD)
0.95 0.89 0.92
(0.38) (0.40) (0.41)
1.01 1.02 1.01
(0.53) (0.63) (0.53)
2.02 2.03 2.02
(1.48) (1.50) (1.48)
1.33 1.31 1.32
(1.05) (1.07) (1.06)
1.07 0.98 1.03
(0.46) (0.46) (0.46)
1.27 1.20 1.24
(0.69) (0.74) (0.71)
2.60 2.32 2.46
(1.85) (1.69) (1.77)
1.64 1.50 1.58
(1.34) (1.23) (1.29)
0.85 0.78 0.82
(0.46) (0.41) (0.43)
0.68 0.70 0.69
(0.44) (0.40) (0.42)
2.41 2.22 2.31
(1.48) (1.41) (1.44)
1.31 1.23 1.27
(1.21) (1.12) (1.16)
0.84 0.78 0.81
(0.45) (0.43) (0.44)
0.79 0.75 0.77
(0.53) (0.43) (0.48)
2.27 2.23 2.25
(1.61) (1.62) (1.61)
1.30 1.25 1.28
(1.22) (1.21) (1.21)
Group 3 date interaction: F 5 7.62, df 5 1,51, p 5 .008; group 3 date 3 hemisphere interaction: F 5 11.41, df 5 1,51, p 5 .00014; group 3 date 3 ventricular region 3 hemisphere: F 5 6.20, df 5 1.58,80.6, p 5 .0059. Simple interactions with group and date for each structure tested separately, lateral horn for group 3 date 3 hemisphere: F 5 12.44, df 5 1,51, p 5 .0009. Simple interactions with group and date for each hemisphere tested separately, left hemisphere for group and date: F 5 12.43, df 5 1,51, p 5 .0009. Simple interactions with date for each group tested separately, Kraepelinian group, main effect of date: F 5 13.58, df 5 1,51, p 5 .0006; date 3 hemisphere: F 5 13.22, df 5 1,51, p 5 .0006; date 3 region 3 hemisphere: F 5 3.46, df 5 1.58,80.6, p 5 .0468. Simple interactions with group for each date tested separately, baseline, region 3 group: F 5 4,71, df 5 1.27,64.9, p 5 .02; second scan, group 3 region by hemisphere: F 5 3.79, df 5 1.81,92.4, p 5 .02.
similar follow-up analyses for anterior and temporal ventricular regions were not statistically significant. Since the effect was confirmed for the left lateral ventricle in follow-up ANOVA, we examined the frequency distributions for the baseline, follow-up, and change scores. As shown in Figure 2, the relative size of the lateral ventricle in the entire patient group showed a distribution that did not differ significantly from normal at baseline but was bimodal at the follow-up scan (Kolmogorov–Smirnov test, x2 5 25.7, df 5 4, p 5 .000037). The difference scores showed proportionately greater enlargement in the Kraepelinian group (Figure 3), but the frequency distribution of the difference scores for the entire group did not show a significant deviation from normality on the Kolmogorov– Smirnov test. Because we anticipated that the Kraepelinian but not the non-Kraepelinian subgroup would increase VBR over time, we examined the correlation between duration of illness and total VBR and observed a significant association with duration of illness in the Kraepelinian group (r 5 .35, p , .05, one-tailed) but not the non-Kraepelinian group (r 5 .06, p 5 ns). Hemispheric differences in the degree of enlargement of the total VBR were also found, with the Kraepelinian subgroup showing a greater ventricular size increase over
the 5-year interval on the left than on the right side (Figures 4 –7); this asymmetrical enlargement was significantly more prominent in the lateral horns (Table 2). We also expressed ventricular enlargement over the 5-year period as a percentage of initial size and observed similar findings. Examining the left dorsal lateral ventricle (selected for confirmation with the follow-up ANOVA in Table 2), we obtained the expected correlation with Kraepelinian status (r 5 .35 for the dorsal slice and r 5 .38 for the ventral slice). Percent ventricular enlargement was also correlated significantly with months of hospitalization in all patients (r 5 .34, p , .05), but not with age, duration of illness, or the other clinical variables in Table 3. Because we expected duration of illness possibly to be correlated with ventricular enlargement in Kraepelinian patients only, we examined this correlation in each group separately (r 5 .22 in Kraepelinian patients and r 5 .22 in non-Kraepelinian patients for left dorsal lateral ventricle percentage increase).
Ventricular Size in Normal Volunteers Normal older individuals had no significant change in ventricular size over the 5-year interval between scans (four-way ANOVA on normal subjects; no interaction that
Ventricular Enlargement and Schizophrenia
BIOL PSYCHIATRY 1998;43:783–793
787
Figure 2. Frequency histograms of left lateral ventricle in all patients at baseline, year 5, and difference. obs, observations. Non-Kraepelinian, n 5 31; Kraepelinian, n 5 22.
contained date reached statistical significance). When compared with the relatively younger schizophrenic patients, the older normal subjects had generally smaller ventricles, especially for the anterior and temporal horns; only the dorsal lateral ventricle was larger (Table 1).
Skull Shape The possibility that the Kraepelinian and non-Kraepelinian subgroups might be characterized by a different skull contour was examined in a five-way analysis of variance
Figure 3. Individual ventricular size at baseline and year 5 in Kraepelinian and non-Kraepelinian patients.
788
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
(group 3 slice level 3 hemisphere 3 anteroposterior segment 3 date). Assessment of the relative skull width measurements at the seven anteroposterior positions at the five slice levels used in the VBR analysis failed to reveal any significant group effect or interaction.
Clinical Correlations with Ventricular Size in Kraepelinian Patients Patients with larger ventricles had spent more months in the hospital; this correlation was significant (p , .05) for both Kraepelinian (scan 1: r 5 .43; scan 2: r 5 .54) and non-Kraepelinian patients (scan 1: r 5 .39; scan 2: r 5 .45). The correlation for scan 2 remained significant (p , .05) for Kraepelinian patients after the effects of age were partialled out.
Clinical Correlations in Non-Kraepelinian Patients
Figure 4. Tally plots of the Kraepelinian patients for the sizes and differences in ventricular shape at year 1 and year 5 at the 41% and 47% of head height levels (Matsui and Hirano 1978; slices 8 and 7, respectively) or approximately 24- and 20-mm levels (Talairach and Tournoux 1988). The left side of the picture is the patient’s left. Top color bar shows number of subjects for each pixel position whose ventricular outline is outside this x–y location. Thus the center of the structure is white (100%) or red (93– 80%), indicating that at this position, nearly all subjects’ outlines were outside this area (all subjects have pixels in this area). The bottom row shows the difference, pixel by pixel, between year 5 and year 1. Red and yellow show areas that were inside each subject’s outline for year 5 but outside for year 1, indicating enlargement of the ventricle at this point. Red areas appear laterally in the most dorsal location, appear medially and posteriorly in the more ventral slice, and are more prominent in the left hemisphere.
As was true for the Kraepelinian patients, larger ventricular size in non-Kraepelinian patients was correlated with age at both scans 1 (r 5 .34, p , .05) and 2 (r 5 .38, p , .05). There were no significant correlations between total SANS change scores and ventricular change.
Discussion Poor-outcome (i.e., Kraepelinian) schizophrenic patients showed significantly greater increases in total ventricular volume over an approximate 5-year interval between baseline and follow-up CT scans than did better-prognosis (i.e., non-Kraepelinian) patients. Although all areas of the ventricular system showed increases in size in the Kraepelinian subgroup, only the lateral ventricles were statistically significantly enlarged compared with values in the non-Kraepelinian subgroup over the 5-year interval. Fur-
Figure 5. Ventricular size in Kraepelinian and non-Kraepelinian patients. Increase in total ventricle/brain ratio is greater in Kraepelinian than non-Kraepelinian patients over 5-year interval (see marginal means and ANOVA in Table 2).
Ventricular Enlargement and Schizophrenia
Figure 6. Lateral ventricle contours. Green line shows contour that includes 80% of subjects at baseline, and red line shows 80% contour at year 5. Note enlargement on lateral and posterior margins of ventricle especially on the left. This simplified view of Figure 2 is complementary in demonstrating that the enlargement is not due only to a few outliers, since the extreme fifth of the population is not shown. The contour line is found by testing outward from the center of each ventricle at 720 angular positions and counting the number of ventricular outlines that are intersected. Axial level given above AC (anterior commissure)PC (posterior commissure) line.
thermore, within the Kraepelinian subgroup, enlargement was significantly larger in the left lateral ventricle than in the right. A group of older normal volunteers, whose age coincided with the period of life at which age-related ventricular enlargement peaks (Waddington et al 1991), did not show a statistically significant ventricular size increase over the same time interval and with the same scanner that detected ventricular changes in the patients with the Kraepelinian subtype of schizophrenia. Although some key features of this longitudinal study of ventricular enlargement are unique, it is by no means the first longitudinal study of ventricular enlargement in schizophrenia; however, it is the first study to have compared subgroups that were characterized, a priori, as comprising better- vs. poor-outcome patients. The sample is also the largest one reported to date in which patients were followed for a mean duration of 5 years. The characteristics of our study design (large sample size, long follow-up interval, older patients, better- vs. poor-outcome subgroups) may help shed light on some of the divergent findings reported in the literature. Heterogeneous sampling of patients could have accounted for only subgroups of schizophrenic patients having shown progressive ventricular enlargement on repeat scanning in some studies (Vita et al 1988; Nasrallah et al 1986; DeLisi et al 1992). Indeed, samples in those studies that have previously found progressive ventricular enlargement were composed of very chronic patients (Kemali et al 1989; Woods et al
BIOL PSYCHIATRY 1998;43:783–793
789
1990); however, the stability of ventricular size found in one study of 15 patients with chronic schizophrenia, rescanned after 7–9 years of continuous illness, including 7 continuously hospitalized patients, remains a result that cannot be reconciled with the current or other data (Illowsky et al 1988). An intriguing hypothesis could be suggested that more severe illness with more inpatient care and the associated diminution in sensory stimulation could result in brain tissue loss. This question is partly addressed by the CT scans in the Genain quadruplets, identical quadruplets with schizophrenia (Buchsbaum et al 1984). Their life histories in the two decades before scanning varied widely, with one quad hospitalized for 15 years and one hospitalized little, but living at home with spouse and children; however, their CT scans were remarkably similar, suggesting greater genetic than sensory environment effects on ventricular size. Before the advent of CT studies, pneumoencephalography (PEG) was used to study longitudinal changes in the ventricular system. The first report of progressive ventricular enlargement derived from an investigation by Moore et al (1935). Heterogeneity of ventricular enlargement with PEG was apparent, with the phenomenon being linked to progressive cognitive and clinical deterioration (Haug 1962; Huber 1957). Taken together, therefore, the combined findings of PEG and CT studies lead to the credible assertion that there exists a subgroup of pooroutcome schizophrenia characterized by progressive ventricular enlargement. One cannot help but wonder if such patients followed into older age would not be those who were viewed as having a degenerating course (Kraepelin 1913, 1919) and have been more recently described in their seventh, eighth, and ninth decades of life as being frankly demented (Davidson et al 1995). A host of potential artifacts can influence studies using CT methodology in schizophrenia. When subjects are scanned twice, the angle of the head and the axial head position are only approximately equal. The resultant variability could account for random differences in ventricular volume between the first and second scans that would increase Type II statistical error. There is no reason to suspect, however, that there should be a systematically different bias between Kraepelinian and non-Kraepelinian patients on head position and slice level over time. The use of two contemporaneously scanned comparison groups for the Kraepelinian patients—namely, age-matched nonKraepelinian patients and aged normal subjects—provides additional information on random changes in position and slice level variability between scans. Ventricular enlargement has been noted in anorexia (e.g., Swayze et al 1996) and in alcoholism (Pfefferbaum et al 1993). Therefore, if Kraepelinian patients had sub-
790
K.L. Davis et al
BIOL PSYCHIATRY 1998;43:783–793
Figure 7. Individual ventricles at baseline and year 5 of study. Top row shows 4 Kraepelinian patients at baseline and year 5 with marked ventricular enlargement prominent in temporal horn. Bottom row shows stability in ventricular size in non-Kraepelinian patients.
stantially poorer nutritional status or alcoholism at the time of their second scan, this could have accounted for the findings of larger ventricles. That explanation seems unlikely for the following reasons. All first scans were conducted in patients when they were inpatients and generally experiencing an exacerbation of illness. In contrast, second scans were performed at times when patients were not experiencing an acute exacerbation and when their symptoms were generally stable. Hence, anorexia, poor nutrition, and the stress of psychotic exacerbation would be more likely to have been factors during the first scan than the second scan, biasing against the results seen
here. Patients with alcoholism were excluded both for the first and second scans, so this source of potential ventricular enlargement is also unlikely to be a major contributor to the effect. Age affects the total volume of the ventricular system. The relationship between age and ventricular volume is quadratic, concave up (Waddington et al 1991; Sullivan et al 1993), with the greatest rate of increase in size per year occurring after the fifth decade of life. The mean age of the Kraepelinian schizophrenic patients in this study, at the time of their second scan, enters the inflection point for change of ventricular size. Kraepelinian and non-Kraepe-
Table 3. Patient Description Kraepelinian
Age Mother’s age at patient birth Years of education Number of hospitalizations Months of hospitalization Age at illness onset Currently employed
Non-Kraepelinian
All patients
Mean
SD
Mean
SD
Mean
SD
42.0 29.2 12.1 8.6 132.2 20.9
8.6 8.1 2.0 9.8 99.7 4.9
37.5 24.5 11.6 6.6 29.6 23.6
12.1 4.6 2.5 4.6 49.0 6.0
39.5 25.6 11.8 8.0 66.4 23.1
11.6 5.8 2.4 7.4 85.7 5.69
0.0%
20.0%
13%
Ventricular Enlargement and Schizophrenia
linian patients did not significantly differ in age, and both had larger ventricles than the older controls; however, some earlier longitudinal CT studies rescanned patients at an age when relatively small volumetric changes were occurring, and as a consequence any progressive enlargement of ventricular volume would not easily be detected. While it would have been useful to have a completely age-matched normal sample, obtaining CT scans in normal subjects is not without difficulty. Continuation studies with high-resolution MRI and without radiation exposure will yield more definitive age-rate comparisons. An obvious criticism of the current study is that it employed CT scanning rather than higher resolution MRI. A false-positive result, however, would not derive from a lower resolution imaging technique, which would be more likely to be affected by random sampling error and fail to find changes over time. Higher resolution MRI used with the current cohort would probably have revealed the same effects, although better definition of the temporal horns might have not only revealed the baseline differences we did observe but made a more definitive regional longitudinal finding possible. Schizophreniform patients, scanned at the time of their first episode, tended to show larger left than right ventricles, unlike control subjects (DeLisi et al 1992). Patients with schizotypal personality disorder assessed with highresolution MRI also showed left rather than right temporal horn enlargement (Buchsbaum et al 1997). Enlargement also appeared larger in the left posterior lateral ventricle in individuals carrying a chromosome-5 marker allele who were members of a pedigree in which the linkage marker was associated with schizophrenia (Shihabuddin et al 1996). This asymmetry in ventricular size suggests that ventricular enlargement might, in part, be a lateralized process. Consistent with this possibility is that left cerebral hypodensity has also been seen on CT scans (Reveley et al 1987), and neuropathological changes in postmortem tissue are often more pronounced on the left (Brown et al 1986; Jakob and Beckmann 1986). Such asymmetry has even extended to dopamine in the left amygdala (Reynolds 1983). Neurochemical abnormalities may also extend to left temporal glutamatergic dysfunction in schizophrenia (Deakin et al 1989). The possibility that schizophrenia is a disease of asymmetry has been hypothesized (Crow et al 1989). Taken together, these data would support the notion that a more active pathological process occurs on the left side of the brain in these very chronic Kraepelinian schizophrenic patients, and could account for the greater enlargement seen on the left than on the right in the current study. Progressive enlargement in ventricular volume can be seen as evidence of a dynamic, rather than a static lesion; however, this interpretation is in no way incompatible with
BIOL PSYCHIATRY 1998;43:783–793
791
the possibility that schizophrenia is associated with a neurodevelopmental abnormality (Murray and Lewis, 1987; Weinberger 1987). Conceivably, neurodevelopmental abnormalities could predispose to increased sensitivity to age-related effects. To view neurodevelopmental etiologies and progressivity as mutually exclusive is to lose sight of how genetically determined diseases such as trisomy 21 and Huntington’s disease display neurodevelopmental consequences at a relatively early age, with progressivity and even neurodegeneration much later in life. The cellular basis for the progressivity in ventricular size detected over approximately 5 years in very poorprognosis schizophrenic patients is clearly a matter of speculation. The absence of pronounced gliosis in postmortem studies of schizophrenia certainly suggests apoptotic rather than atrophic processes. The possibility that such effects may in part be mediated by excitotoxicity secondary to hyperglutamatergia has been suggested (Coyle 1976; Olney and Farber 1995). A challenge to these views is the ability to integrate hypoglutamatergia, to which is attributed some symptoms of schizophrenia, with hyperglutamateriga, to which is attributed cell death. One possibility is to implicate the loss of gamma-aminobutyric acidergic inhibitory interneurons (Olney and Farber 1995; Benes et al 1991). Many questions remain to be answered as the phenomenon of progressive ventricular enlargement in pooroutcome schizophrenia is further elucidated. For example, is the process linear or curvilinear? Are there associated and identifiable changes in tissue volume? What are the effects of neuroleptics upon the progression of ventricular changes? If the discourse on this topic can move away from the simple dichotomy of whether schizophrenia is a static or progressive lesion and move toward the elucidation of the phenomonology, it may be possible to begin to answer some of these questions.
This research was supported in part by the Schizophrenia Biological Research Center and the Program in Biological Psychiatry (grants from the Veterans Administration to K.L. Davis), and MH40071 to M.S. Buchsbaum from the National Institute of Mental Health. Cheuk Tang, Vladimir Tkach, and Dr. Tse-Chung Wei provided programming support, and Rita Amato and Granada Stephens assisted in the graphical analysis of the images. Bradley R. Buchsbaum assisted in additional graphics and statistical analysis. The consent was approved by the Mount Sinai Institutional Review Board.
References American Psychiatric Association (1987): Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R), 3rd ed rev. Washington, DC: American Psychiatric Press.
792
BIOL PSYCHIATRY 1998;43:783–793
Andreasen NC (1983): The Scale for the Assessment of Negative Symptoms. Iowa City, IA: University of Iowa. Benes F, McSparren J, Bird E, San Giovanni J, Vincent S (1991): Deficits in small interneurons in prefrontal and cingulate cortices of schizophrenic and schizoaffective patients. Arch Gen Psychiatry 48:996 –1001. Brown R, Colter N, Corsellis JAN, et al (1986): Post-mortem evidence of structural brain changes in schizophrenia. Differences in brain weight, temporal horn area, and parahippocampal gyrus compared with affective disorder. Arch Gen Psychiatry 43:36 – 42. Buchsbaum MS, Mirsky AF, DeLisi LE, et al (1984): The Genain quadruplets: Electrophysiological, positron emission, and X-ray tomographic studies. Psychiatry Res 13:95–108. Buchsbaum MS, Yang S, Hazlett EA, et al (1997): Ventricular volume and asymmetry in schizotypal personality disorder and schizophrenia assessed with magnetic resonance imaging. Schizophr Res 27:45–54. Coyle JT (1976): Neurotoxin action of kainic acid. J Neurochem 41:1–11. Crow TJ, Ball J, Bloom SR, et al (1989): Schizophrenia as an anomaly of development of cerebral asymmetry: A postmortem study and a proposal concerning the genetic basis of the disease. Arch Gen Psychiatry 46:1145–1150. Davidson M, Harvey PD, Powchik P, et al (1995): Severity of symptoms in chronically institutionalized geriatric schizophrenic patients. Am J Psychiatry 152:197–207. Deakin JFW, Slater P, Simpson MDC, et al (1989): Frontal cortical and left temporal glutamatergic dysfunction in schizophrenia. J Neurochem 52:1781–1786. Degreef G, Ashtari M, Wu H, Borenstein M, Geisler S, Lieberman J (1991): Follow-up MRI study in first episode schizophrenia. Schizophr Res 5:1204 –1206. DeLisi LE, Stritzke P, Riordan H, et al (1992): The timing of brain morphological changes in schizophrenia and their relationship to clinical outcome. Biol Psychiatry 31:241–254. DeLisi LE, Tew W, Xie S, et al (1995): A prospective follow-up study of brain morphology and cognition in first-episode schizophrenic patients: Preliminary findings. Biol Psychiatry 38:349 –360. DeLisi LE, Sakuma M, Tew W, Kushner M, Hoff AL, Grimson R (1997): Schizophrenia as a chronic active brain process: A study of progressive brain structural change subsequent to the onset of schizophrenia. Psychiatry Res Neuroimaging 74: 129 –140. Dixon WJ, Brown MB, Engelman L (1985): BMDP Statistical Software. Berkeley: University of California Press. Endicott J, Spitzer RL (1978): A diagnostic interview: The Schedule for Affective Disorders and Schizophrenia. Arch Gen Psychiatry 35:837– 844. Frecska E, Losonczy MF, Keefe RSE, et al (1994): Agedependent change of ventricular measures in schizophrenia: Cross-sectional analysis [abstract of the Society of Biological Psychiatry, Philadelphia, Pennsylvania]. Biol Psychiatry 35: 704. Frecska E, Losonczy MF, Keefe RSE, et al (in preparation): Ventricular enlargement in schizophrenia: Support for asymmetric age-dependent change.
K.L. Davis et al
Haug OJ (1962): Pneumoencephalographic studies in mental disease. Acta Psychiatr Scand 165:1–114. Huber G (1957): Pneumoencephalographische und Pychopathologische bilder bei endogen Psychosen. Berlin: Springer International. Illowsky BP, Juliano DM, Bigelow LB, Weinberger DR (1988): Stability of CT scan findings in schizophrenia: Results of an 8 year follow-up study. J Neurol Neurosurg Psychiatry 51:209 –213. Jakob H, Beckmann H (1986): Prenatal developmental disturbances in the limbic allocortex in schizophrenics. J Neural Transm 65:303–326. Jaskiw GE, Juliano DM, Goldberg TE, Hertzman M, UrowHamell E, Weinberger DR (1994): Cerebral ventricular enlargement in schizophreniform disorder does not progress: A seven year follow-up study. Schizophr Res 14:23–28. Keefe RSE, Mohs RC, Losonczy MF, et al (1987): Characteristics of very poor outcome schizophrenia. Am J Psychiatry 144:889 – 894. Keefe RSE, Mohs RC, Davidson M, et al (1988): Kraepelinian schizophrenia: A subgroup of schizophrenia? Psychopharmacol Bull 24:458 – 462. Keefe RSE, Mohs RC, Losonczy MF, et al (1989): Premorbid sociosexual functioning and long-term outcome in schizophrenia. Am J Psychiatry 146:206 –211. Keefe RSE, Mohs RC, Silverman JM, et al (1990): Characteristics of Kraepelinian schizophrenia and their relation to premorbid sociosexual functioning. In: Angrist B, Schulz SC, editors. The Neuroleptic-Nonresponsive Patient: Characterization and Treatment. Washington, DC: American Psychiatric Press, pp 3–21. Keefe RSE, Lobel DS, Mohs RC, et al (1991): Diagnostic issues in chronic schizophrenia: Kraepelinian schizophrenia, undifferentiated schizophrenia, and state-independent negative symptoms. Schizophr Res 4:71–79. Keefe RSE, Frecska E, Apter SH, et al (1996): Clinical characteristics of Kraepelinian schizophrenia: Replication and extension of previous findings. Am J Psychiatry 153:806 – 811. Kemali D, Maj M, Galderisi S, Milici N, Salvati A (1989): Ventricle-to-brain ratio in schizophrenia: A controlled follow-up study. Biol Psychiatry 26:753–756. Kraepelin E (1913): Lectures on Clinical Psychiatry New York: William Wood & Company. Kraepelin E (1919): Dementia Praecox and Paraphrenia Together with Manic-Depressive Insanity and Paranoia. Edinburgh: E&S Livingstone. Lange N, Giedd JN, Castellanos FX, Vaituzis AC, Rapoport JL (1997): Variability of human brain structure size. Psychiatry Res Neuroimaging 4:1–12. Lieberman JA, Alvir JA, Koreen A, et al (1996): Psychobiologic correlates of treatment response in schizophrenia. Neuropsychopharmacology 14:13S–21S. Losonczy MF, Song IS, Mohs RC, et al (1986a): Correlates of lateral ventricular size in chronic schizophrenia: I. Behavioral and treatment response measures. Am J Psychiatry 143:976 – 981. Losonczy MF, Song IS, Mohs RC, et al (1986b): Correlates of lateral ventricular size in chronic schizophrenia: II. Biological measures. Am J Psychiatry 143:1113–1118.
Ventricular Enlargement and Schizophrenia
Matsui T, Hirano A (1978): An Atlas of the Human Brain for Computerized Tomography. Tokyo: Igaku-Shoin. Moore MT, Nathan D, Elliot AR, et al (1935): Encephalographic studies in mental disease. Am J Psychiatry 92:43– 67. Murray RM, Lewis SW (1987): Is schizophrenia a neurodevelopmental disorder? Br Med J 295:681– 682. Nair TR, Christensen JD, Kingsbury SJ, Kumar NG, Terry WM, Garver DL (1997): Progression of cerebroventricular enlargement and the subtyping of schizophrenia. Psychiatry Res Neuroimaging 74:141–150. Nasrallah HA, Olson SC, McCalley-Whitters M, Chapman S, Jacoby CG (1986): Cerebral ventricular enlargement in schizophrenia. Arch Gen Psychiatry 43:157–159. Olney JW, Farber NB (1995): Glutamate receptor dysfunction and schizophrenia. Arch Gen Psychiatry 52:998 –1007. Pfefferbaum A, Sullivan EV, Rosenbloom MJ, Shear PK, Mathalon DH, Lim KO (1993): Increase in brain cerebrospinal fluid volume is greater in older than in younger alcoholic patients. Psychiatry Res 50:257–274. Reveley M, Reveley AM, Baldy R (1987): Left cerebral hemisphere hypodensity in discordant schizophrenic twins: A controlled study. Arch Gen Psychiatry 44:625– 632. Reynolds GP (1983): Increased concentrations and lateral asymmetry of amygdala dopamine in schizophrenia. Nature 305: 527–529. Shelton RC, Weinberger DR (1986): X-ray computed tomographic studies in schizophrenia: A review and synthesis. In: Nasarallah HA, Weinberger DR, editors. Handbook of Schizophrenia, vol 1. Amsterdam: Elsevier, pp 207–250. Shihabuddin L, Silverman JM, Buchsbaum MS, et al (1996): Ventricular enlargement associated with linkage marker for schizophrenia: Related disorders in one predigree. Mol Psychiatry 1:215–222. Sponheim RS, Iacono WG, Beiser M (1991): Stability of ventricular size after the onset of psychosis in schizophrenia. Psychiatry Res Neuroimaging 40:21–29. Sullivan EV, Shear PK, Mathalon DH, et al (1993): Greater
BIOL PSYCHIATRY 1998;43:783–793
793
abnormalities of brain cerebrospinal fluid volumes in younger than in older patients with Alzheimer’s disease. Arch Neurol 50:359 –373. Swayze VW, Andersen A, Arndt S, et al (1996): Reversibility of brain tissue loss in anorexia nervosa assessed with a computerized Talairach 3-D proportional grid. Psychol Med 26:381– 390. Talairach J, Tournoux P (1988): Co-planar Sterotaxic Atlas of the Human Brain. Three-Dimensional Proportional System: An Approach to Cerebral Imaging [translated by M. Rayport]. New York: Thieme Medical Publishers. Vita A (1991): Cerebral ventricular abnormalities in schizophrenia: Evidence of their early origin and stability over time. Schizophr Res 5:189 –190. Vita A, Sacchetti E, Valvassori G, Cazzullo CL (1988): Brain morphology in schizophrenia: A 2-to-5-year CT scan follow-up study. Acta Psychiatr Scand 78:618 – 621. Vita A, Giobbio GM, Dieci M, et al (1994): Stability of cerebral ventricular size from the appearance of the first psychotic symptoms to the later diagnosis of schizophrenia. Biol Psychiatry 35:960 –962. Vita A, Dieci M, Giobbio GM, Tenconi F, Invernizzi G (1997): Time course of cerebral ventricular enlargement in schizophrenia supports the hypothesis of its neurodevelopmental nature. Schizophr Res 23:25–30. Waddington JL, O’Callaghan E, Buckley P, et al (1991): The age dependencies of MRI abnormalities in schizophrenia suggest early ventricular enlargement but later prominence of cortical atrophy. Schizophr Res 5:188 –189. Weinberger DR (1987): Implications of normal brain development for the pathogenesis of schizophrenia. Arch Gen Psychiatry 44:660 – 669. Woods BT, Yurgelun-Todd D, Benes FM, Frankenburg FR, Pope HG Jr, McSparren J (1990): Progressive ventricular enlargement in schizophrenia: Comparison to bipolar affective disorder and correlation with clinical course. Biol Psychiatry 27:341–352.