Gender Differences in Schizophrenia on MRI Brain Scans

Transcription

1 VOL. 16, NO. 2, 1990 Gender Differences in Schizophrenia on MRI Brain Scans 205 by Henry A. Nasrallah, Steven B. Schwarzkopf, Stephen C. Olson, and Jeffrey A. Coffman Abstract There are many reports of clinical and biological gender differences in schizophrenia. Gender differences in structural brain abnormalities in schizophrenia have been reported on both computed tomographic (CT) and magnetic resonance imaging (MRI) scans. We present here a new MRI study of cerebral structures in schizophrenia. On the basis of previous findings, we hypothesized that schizophrenic males are more likely than females to show smaller brains and larger ventricles compared to their control counterparts. Our results indicated that the opposite was true: schizophrenic females, but not schizophrenic males, had smaller craniums and brains and larger lateral and third ventricles on MRI scans. The possible significance and implications of these data are discussed. Many gender differences have been reported in schizophrenia, such as premorbid history, age of onset, symptomatology, response to treatment, and course of illness (Lewine 1981, 1988; Seeman 1982; Goldstein 1988). Nasrallah and Wilcox (1989) reported that genetic factors were more prominent etiological features in schizophrenic females, while brain injury factors early in life were more prominent in schizophrenic males. A recent review of gender differences in brain structure and function in schizophrenia describes several neurobiological variables that may be relevant to the pathogenesis of schizophrenia (Delisi et al. 1989). One of the most frequently replicated findings in schizophrenia over the past decade is ventricular dilatation, as measured on computed tomographic (CT) and magnetic resonance imaging (MRI) brain scans (Shelton and Weinberger 1986; Coffman and Nasrallah, in press). However, very few studies have addressed the issue of gender differences in lateral or third cerebral ventricular enlargement in schizophrenia (Bridge et al. 1985). In a previous MRI study (Andreasen et al. 1986), we reported that schizophrenic males, but not females, showed a significant reduction in cranial and cerebral size, suggesting neurodevelopmental cerebral hypoplasia in schizophrenic males. We also reported gender differences in corpus callosum dimensions in schizophrenia on MRI scans (Nasrallah et al. 1986). In this article we report the results of a new MRI study which we conducted with the hypothesis that schizophrenic males, but not females, would show smaller craniums and cerebrums and larger lateral and third cerebral ventricular size compared to gender-matched control subjects. Methods Fifty-six consecutively referred schizophrenic subjects (41 males, 15 females) and 35 healthy volunteers (15 males, 20 females) consented to participate in the study. All schizophrenic subjects met DSM-IH-R criteria for chronic schizophrenia (American Psychiatric Association 1987), using a standard diagnostic interview, the Structured Clinical Interview for DSM-IH-R (SCID-P; Spitzer et al. 1987). The Reprint requests should be sent to Dr. H.A. Nasrallah, Dept. of Psychiatry, The Ohio State University College of Medicine, 473 West 12th Ave., Columbus, OH

2 206 SCHIZOPHRENIA BULLETIN volunteers (from the community) responded to advertisements and were matched as closely as possible for age and education. They exhibited no psychopathology on the SCID-P for controls. The inclusion criteria for this study consisted of the following: (1) males and females, aged years, with onset of illness before age 45 years; (2) ability to give informed consent; and (3) fulfillment of DSM-I11-R criteria for schizophrenia. The exclusion criteria consisted of the following: (1) serious or debilitating medical illness including seizure disorder; (2) history of penetrating head injury, prolonged loss of consciousness, or neurosurgery; (3) metallic implants or pacemakers; and (4) severe past or present substance abuse or dependence. The protocol was identical for male and female patients, eliminating bias due to protocol differences. The excess of male subjects was a result of more male volunteers and referrals and not due to a higher dropout rate in the female group. All subjects underwent an MRI brain scan protocol with a General Electric 1.5 Tesla scanner, using a Tl-weighted pulse sequence (Tl = 800 ms, TR = 1500 ms). Eight sagittal scans per subject were obtained about the midline, with a slice thickness of 3 mm and an interslice distance of 1 mm. The slice passing through, or closest to, the septum pellucidum, was identified as midsagittal. For each subject, coronal slices were also obtained using the same pulse sequence, with a slice thickness of 5 mm and an interslice distance of 5 mm. Computerized planimetry area measurements of the cranium, cerebrum, and frontal area in the midsagittal plane were made using an averaging methodology to minimize artifacts and distortion (Coffman et al. 1989). The frontal area in this study was defined as the portion of the cerebrum anterior to a line bisecting the length of the corpus callosum. Midsagittal lateral cerebral ventricle-to-brain ratio (VBR) was obtained by dividing the area of the midsagittal lateral ventricle by the midsagittal cerebral area. Coronal measurements were made on the slice 1 cm posterior to the slice that passes through the optic chiasm, and included the measurements of the right and left cerebrum and the right and left lateral ventricles, to calculate the right, left, and total VBR. The third ventricular width was measured on the same coronal slice. Statistical Analyses. Group differences on demographic variables were assessed using t tests (for those normally distributed) and Wilcoxon two-sample tests (for variables not normally distributed). All cerebral and cranial measurements met criteria for normality (Wilks-Shapiro test, males and females examined separately) and were examined using parametric methods. VBR measures failed to meet criteria for normality, making nonparametric methods necessary. Initially, a two-way multivariate analysis of covariance (MANCOVA) was run for midsagittal and coronal MRI area measures, with independent measures being sex and diagnosis, and height used as a covariate. However, on the basis of differences between the female groups on height (with potential confounding), and heterogeneity of regression (MRI variables vs. height) between males and females, separate analyses were run for males and females in followup. For males, one-way multivariate analysis of variance (MANOVA) (equivalent to Hotellings t test) was performed on the midsagittal brain and cranial measures (dependent variables), and a second MANOVA was performed on the coronal measures. The independent variable was diagnosis (control subjects vs. schizophrenic patients). This approach protects against inflated significance values in comparisons of multiple correlated variables. The two-sample Wilcoxon test was used in assessing differences between controls and patients on the VBR measures. For females, one-way MANCOVA was performed on the midsagittal brain and cranial measures, and another MANCOVA was performed on the coronal measures. Height was used as the covariate in these analyses. The independent variable was diagnosis (control subjects vs. schizophrenic patients). The twosample Wilcoxon test was used in assessing differences between controls and patients on VBR measures. Significant multivariate results were followed up with univariate analyses to examine which dependent variables differed most between groups. Results Demographics. Table 1 shows the demographic and anthropometric values for the groups. controls were well matched with male patients on demographic factors, including age, height, and weight. s had a significantly higher educational level, though the patient group averaged over 1 year of postsecondary education.

3 VOL. 16, NO. 2, Table 1. Descriptive statistics for male and female schizophrenic patients and control subjects: Mean (SD) Variable Age Education Height Weight (n = 41) (6.40) (2.27) (6.91) (15.27) (n = 15) (7.48) (1.36) (4.83) (17.62) 1 schizophrenic patients vs. male controls, r test, 2-tailed. 2 schizophrenic patients vs. male controls, Wilcoxon 2-sample test. 3 schizophrenic patients vs. female controls, Wilcoxon 2-sample test 4 schizophrenic patients vs. female controls, f test, 2-tailed. controls were nonsignificantly older and taller than female patients. controls also had more years of education than the female patients. There were no significant differences between schizophrenic males and schizophrenic females on education (13.44 years for males vs years for females), chronicity (9.68 years for males vs years for females), or age at onset (22.54 years for males vs years for females). The female patients were significantly older than the male patients (31.63 years for males vs years for females). males and control females were well matched on age (26.6 years for males vs years for females) and education (16.1 years for males vs years for females). MRI Measures. Tables 2 and 3 show group means of MRI measures and their standard deviations. The values are shown for univariate analysis of covariance (ANCOVA) p-value < ' results following the significant MANCOVA in females. The overall two-way MANCOVA showed a significant effect for sex (p < 0.01) and height (p < 0.05), and a strong trend for a main effect of diagnosis (p = 0.096, two-tailed testing). s. There were no significant differences between the male control subjects and the male schizophrenic patients on cranial, cerebral, or VBR measures in the midsagittal or the coronal planes. patients did have nonsignificantly smaller mean cranial and cerebral areas than the control subjects. s. For females, the MAN- COVA on the midsagittal measures showed a significant effect of diagnosis after controlling for height. Followup analyses showed this to be due to smaller cranial, cerebral, and frontal areas in the schizophrenic females compared to control females. This effect was greatest for the frontal area (p = 0.009), with (n = 15) (10.03) (1.83) (6.73) (11.11) (n = 20) (8.11) (1.27) (6.47) (8.81) p-value " 0.958" strong trends for the cranial (p = 0.099) and cerebral areas (p = 0.073). There was no diagnosis effect on the coronal measures after controlling for height. patients had significantly larger measures than female controls on all VBR measures (see table 4) (p < 0.05, two-tailed Wilcoxon test, midsagittal and coronal VBR measures). Discussion Contrary to our hypothesis, the data of this study suggest that cerebral hypoplasia and ventricular enlargement are more apparent in female than male schizophrenic patients. Given the fact that other studies suggest the opposite, possible explanations of the findings should be offered. One possible explanation is the sample composition. It is not likely that gender differences in the control group accounted for the

4 208 SCHIZOPHRENIA BULLETIN Table 2. Sagittal magnetic resonance imaging brain area measures (by sex and diagnostic category): Mean (SD) Areas Cranial Cerebral average Frontal patients (10.90) (11.56) (6.49) (8.01) (3.17) (3.99) subjects (13.46) (11.70) (8.34) (6.26) (3.72) (2.86) p value Analysis of covanance, height as covariate, schizophrenic patients less than control subjects, 2-tailed test. Table 3. Coronal magnetic resonance imaging brain area measures (by sex and diagnostic category): Mean (SD) Areas Cranial Cerebral right Cerebral left patients (7.73) (8.78) (2.35) (2.44) (2.26) (2.76) subjects (11.02) (7.79) (3.39) (1.92) (2.58) (1.61) findings, because both male and female control subjects were similar in their demographic differences from their schizophrenic counterparts. It is possible that the schizophrenic females in this study were relatively more severely ill than the males. Although both male and female schizophrenic groups were chronically ill and dysfunctional, they were all living in the community and not chronically institutionalized. The mean ages of onset for the males (22.54 years) and females (23.87 years) in this study were both early. However, the literature of age of onset in schizophrenia (Angermeyer and Kiihn 1988) suggests that the mean age of onset for females in this sample was relatively much earlier than for the males. It is possible that the earlier age of onset in our sample of schizophrenic females is associated with a greater degree of structural brain abnormalities, which is supported by a recent study by Johnstone et al. (1989). It is also possible that the schizophrenic males in this study were less severely ill than is usual for VBR studies. An argument can be made that ambulatory young schizophrenic males living in the community (such as in this study) rather than in institutions may be less severely ill than typical schizophrenic males. In fact, the first wellpublicized negative study of VBR in schizophrenia 0ernigan et al. 1982) had such a sample, suggesting that ventriculomegaly in schizophrenia relates to severity and chronicity. Another consideration is that the requirements of a complex battery of tests in our research project (over 20 hours of interviews, neuropsychological testing, and brain imaging procedures) created a truncated sample by excluding the most

5 VOL. 16, NO. 2, Table 4. Nonnormally distributed brain areas on magnetic resonance imaging: Mean (SD) Areas VBR midsagittal VBR coronal (left + right) VBR left coronal VBR right coronal Third ventricle 1 Wilcoxon 2-sample test, 2-tailed. patients (0.020) (0.022) (0.016) (0.019) (0.019) (0.018) (0.015) (0.021) 0.35 (0.18) 0.39 (0.32) severely ill patients of either gender. Because females have a generally less severe illness, fewer of the "treatment-refractory" female cases may have been excluded. Alternatively, since the age of onset in schizophrenic females here was earlier than in most studies, we may need to examine why less severely ill females with later age of onset were not adequately represented in this study. This may reflect a referral bias, since some of subjects (0.024) (0.013) (0.017) (0.009) (0.015) (0.009) (0.021) (0.010) 0.32 (0.15) 0.22 (0.10) p value our subjects were brought in by families who tend to be more involved when the illness occurs earlier. Also, some female patients who are doing quite well may not have been interested in participating because of their perception that their illness was well controlled and that participation in a research study would not help them. The possibility of a relatively lower VBR in female control subjects than in male control subjects must also be ruled out as a factor. Although our control subjects clearly did not have schizophrenia, some gender-related differences in the degree of psychopathology can be postulated. Little is known about the factors that motivate individuals to respond to advertisements to participate as "normal" controls and to undergo a lengthy series of tests and examinations with a minimal financial compensation. One issue might be the availability of free time in male versus female volunteers, or whether females were more motivated by altruistic reasons while males were drawn because of some subtle psychopathology. We have no way of determining whether controls with extensive substance abuse history might have concealed that from us when they volunteered. Finally, if the findings here are true, and not a Type I statistical error, then the gender-related significance of smaller brains and larger ventricles should be investigated. Genetic, developmental, and hormonal factors should all be considered. We hope that attempts to replicate our findings take into account the various issues raised here, to render the results more comparable and easier to interpret. References American Psychiatric Association. DSM-lll-R: Diagnostic and Statistical Manual of Mental Disorders. 3rd ed, revised. Washington, DC: The Association, Andreasen, N.C.; Nasrallah, H.A.; Dunn, V.; Ehrhardt, J.C.; Grove, W.M.; Olson, S.C.; Coffman, J.A.; and Crossett, J.H. Structural abnormalities in the frontal system in schizophrenia: A magnetic resonance imaging study. Archives of General Psychiatry, 43: , 1986.

6 210 SCHIZOPHRENIA BULLETIN Angermeyer, M.C., and Kiihn, L. Gender differences in age of onset of schizophrenia. European Archives of Psychiatry and Neurological Sciences, 237: , Bridge, T.P.; Parker, E.S.; Ingraham, L.; and Bickham, C.E. Gender effects seen in the cerebral ventricular/brain ratio (VBR). Biological Psychiatry, 20: , Coffman, J.A., and Nasrallah, H.A. The application of brain imaging techniques to the study of schizophrenia. Psychiatric Medicine, in press. Coffman, J.A.; Schwarzkopf, S.B.; Olson, S.C.; and Nasrallah, H.A. Midsagittal cerebral anatomy by magnetic resonance imaging: The importance of slice position and thickness. Schizophrenia Research, 2: , Delisi, L.E.; Dauphinais, D.; and Hauser, P. Gender differences in the brain: Are they relevant to the pathogenesis of schizophrenia? Comprehensive Psychiatry, 30: , Goldstein, J.M. Gender differences in the course of schizophrenia. American journal of Psychiatry, 145: , Jernigan, T.L.; Zatz, L.M.; Moses, J.A.; and Berger, P.A. Computed tomography in schizophrenics and normal volunteers: I. Fluid volume. Archives of General Psychiatry, 39: , Johnstone, E.C.; Owens, D.G.C.; Bydder, G.M.; Colter, N.; Crow, T.J.; and Frith, CD. The spectrum of structural brain changes in schizophrenia: Age of onset as a predictor of cognitive and clinical impairments and their cerebral correlates. Psychological Medicine, 19:91-103, Lewine, R.R.J. Sex differences in schizophrenia: Timing or subtype? Psychological Bulletin, 90: , Lewine, R.R.J. Gender and schizophrenia. In: Tsuang, M.T., and Simpson, J.C., eds. The Handbook of Schizophrenia, Vol. 3. Amsterdam: Elsevier Science Publishers, pp Nasrallah, H.A.; Andreasen, N.C.; Coffman, J.A.; Olson, S.C.; Dunn, V.D.; Ehrhardt, J.C.; and Chapman, S.M. A controlled magnetic resonance study of corpus callosum thickness in schizophrenia. Biological Psychiatry, 1: , Nasrallah, H.A., and Wilcox, J.A. Gender differences in the etiology and symptoms of schizophrenia: Genetic versus brain injury factors. Annals of Clinical Psychiatry, 1:51-53, Shelton, R.C., and Weinberger, D.R. X-ray computerized tomography studies in schizophrenia: A review and synthesis. In: Nasrallah, H.A., and Weinberger, D.R., eds. The Neurology of Schizophrenia. Vol. I. Amsterdam: Elsevier Science Publishers, pp Seeman, M.V. Gender differences in schizophrenia. Canadian Journal of Psychiatry, 27: , Spitzer, R.L.; Williams, J.B.W.; and Gibbon, M. Structured Clinical Interview for DSM-III-R Patient Version (SCID-P). New York: N.Y. State Psychiatric Institute, Acknowledgments Judy McLaughlin (data base management), Judy Brandt (MRI measurements), and Mary Lynn (structural interviews) assisted in the project. The Franklin County Chapter of the Alliance for The Mentally 111 gave valuable support in recruitment. This project was supported in part by the Regents of Ohio Research Challenge Grant, and by USPHS grants MH and MH from the National Institute of Mental Health. The Authors Henry A. Nasrallah, M.D., is Professor and Chairman, and Steven B. Schwarzkopf, M.D., Stephen C. Olson, M.D., and Jeffrey A. Coffman, M.D., are Assistant Professors, Department of Psychiatry, The Ohio State University, Columbus, OH.