Correspondence address: YasuhiroKawasaki MD Department of Neuropsychiatry, Faculty of Medicine, Toyama Medical and Pharmaceutical University, 2630 Sugitani, Toyama 930-0194, Japan. Email: firstname.lastname@example.org
Abstract Volumes of the medial temporal lobe structures (i.e. the amygdala, hippocampus, and parahippocampal gyrus), Sylvian fissure, and inferior horn of the lateral ventricle relative to the cerebral hemisphere were measured in 24 patients with schizophrenia and 23 normal controls using magnetic resonance imaging. The patients had significantly larger Sylvian fissures and inferior horns bilaterally than the controls. In the patients the right Sylvian fissure size showed a significant positive correlation with the duration of illness. Moreover, earlier onset of illness was significantly correlated with decreased volume of the left medial temporal lobe structures. These results replicate previous finding of inferior horn enlargement and suggest the significance of the Sylvian fissure and the medial temporal lobe structures in pathophysiology of schizophrenia.
There is a growing body of evidence for morphological brain changes in patients with schizophrenia. Recent magnetic resonance imaging (MRI) studies have shown a reduction in volume of medial temporal lobe structures (i.e. amygdala, hippocampus, and parahippocampal gyrus) in patients with chronic schizophrenia, in accordance with post-mortem brain studies.1–12 Volume reduction of medial temporal lobe structures has also been found in the first-episode patients, which may be developmental in origin.13–18
With regard to the clinical symptoms, some computed tomography studies have shown dilatation of the left Sylvian fissure to be related to positive symptoms such as thought disorders, and auditory hallucinations.19,20 In contrast, dilatation of the right or bilateral Sylvian fissures was related to negative symptoms.19,21 With MRI, volume reductions in the left superior temporal gyrus are reportedly correlated with severity of auditory hallucinations and thought disorders, and volume reductions in medial temporal lobe structures with the severity of the positive symptoms.3,4,22–24
We previously reported that asymmetry of the lateral ventricle was inversely correlated with age at the onset of schizophrenia.25 In the present study, to further explore the clinical significance of the morphological abnormalities in the brain of patients, we performed volume measurements of the temporal lobe structures by MRI, and examined the relationships between the neuroanatomic measurements and clinical variables in 24 patients with schizophrenia.
Subjects and methods
Twenty-four patients meeting DSM-III-R26 criteria for schizophrenia were studied. The patients were inpatients or outpatients of Toyama Medical and Pharmaceutical University Hospital, Toyama, Japan, who gave informed consent for this study. Only patients under the age of 40 were selected to eliminate possible effects of ageing. Patients were excluded if they had a history of any organic mental disorder, alcohol abuse, neuroleptic malignant syndrome, neurologic disease, or drug treatment known to affect the brain (e.g. cortisol). Ten of the subjects were males and 14 were females. Their mean (SD) age at the time of the MRI scans was 21.8 (3.7) years (range, 15.9–29.6), and all were being treated with typical neuroleptics at the time of the MRI scans [12.1 (10.2) mg/day in haloperidol equivalents].
Age at the onset of the prodromal phase and the active phase was determined according to DSM-III-R26 criteria by one of the authors (MS) blind to the MRI results. Mean age at the onset of the prodromal phase was 17.4 (3.1) years (range, 13–26), and at the onset of the active phase, 18.3 (3.1) years (range, 14–26). Mean duration of illness, defined as the period which had elapsed since the onset of the prodromal phase, was 4.4 (2.9) years (range, 0.5–11.6). There were no significant differences between the male and female patients in age at the onset of the prodromal or the active phase. Twenty of the patients were assessed for severity of clinical symptoms with the Scale for the Assessment of Negative Symptoms (SANS)27 and the Positive and Negative Syndrome Scale (PANSS).28
The control group consisted of 23 healthy volunteers who were recruited from within the hospital community. Each subject was given precise information about the intention and the procedure of the MRI examination prior to the study. Informed consent was obtained orally from all the subjects, and was recorded in the individual clinical document of the patients or was preserved as a written consent from control subjects. The control subjects were screened for medical and psychiatric illness that would exclude them. Ten were males and 13 females. Their mean age was 21.5 (2.6) years, and did not differ from that of the patients.
Magnetic resonance imaging scans were performed using a Siemens Magnetom 1.5 tesla scanner (Erlangen, Germany). All subjects were aligned in the MRI scanner by using a laser alignment system to avoid lateral tilting of the head. T1-weighted coronal slices (repetition time, 525 ms; echo time, 15 ms) were obtained. The coronal slices were parallel to the floor of the fourth ventricle seen on the scout mid-sagittal image. Slice thickness was 5 mm, and the slices were separated by 0.5 mm. Images from MRI films were converted to a computer system (Macintosh IIci; Apple Computer Inc., Cupertino, CA, USA) using a CCD video camera, light viewbox, and graphic display board (RasterOps 24STV; RasterOps Corporation, Santa Clara, CA, USA). Quantitative measurements of regions of interest were made using the NIH Image 1.52 software program (National Institute of Health, Bethesda, MD, USA).
Lateral tilting of the head of the patients during the MRI scanning procedure was assessed according to the method of Zipursky et al.29 Details of the method are described elsewhere.25 Consequently, none of the subjects showed a significant slope with this method, and all subjects were considered eligible. Quantitative measurements were then performed on five contiguous slices starting at the level at which the amygdala was clearly visible and extending caudally. Area was measured in the cerebral hemispheres, sylvian fissures, inferior horns of the lateral ventricle and medial temporal lobe structures (MTS) including the amygdala, hippocampus, and parahippocampal gyrus. Area measurements were made on all five slices in the cerebral hemispheres and inferior horns. The area of the sylvian fissures was measured in three consecutive slices, beginning anteriorly where the temporal stem clearly appeared. The boundaries between the brain substance and Sylvian fissure or inferior horns were drawn semi-automatically using the threshold method. Area measurements of MTS were made on five slices in which the rostral two slices were mainly composed of the amygdala and parahippocampal gyrus, and caudal three slices were of the hippocampus and parahippocampal gyrus. The measured areas were then summed for each structure. Representative MRI images with delineated areas for all the measurements are shown in Fig. 1.
Interrater reliability was computed for all regions of interest by two raters (MA and HH) with five cases. The interclass correlation coefficients (ICC) for the two raters were as follows: left cerebral hemisphere, 0.99; right cerebral hemisphere, 0.96; left MTS, 0.98; right MTS, 0.97; left Sylvian fissure, 0.99; right Sylvian fissure, 0.96; left inferior horn, 0.71; right inferior horn, 0.66.
To control for variations in brain size, MTS volume relative to hemicerebral volume [i.e. the MTS–brain ratio defined as (the sum of MTS areas in the defined slices)/(the sum of hemicerebral areas in the corresponding slices) × 100] was calculated. The Sylvian fissure–brain ratio and the inferior horn–brain ratio were obtained in the same manner.
Statistical analyses were performed using the Macintosh STATISTICA 4.1 J software package (StatSoft, Tulsa, OK, USA). A three-factor repeated-measures analysis of variance (ANOVA) with diagnosis (i.e. patient and control; between-subjects factor), gender (i.e. male and female; between-subjects factor), and hemisphere (i.e. right and left; within-subjects factor) was used to examine which factor interferes specifically with hemicerebral areas. Subsequently, a series of two-factor repeated-measures ANOVA was performed to test for main effects and interactions of diagnosis (between-subjects factor) and hemisphere (within-subjects factor) in each region of interest by employing the relative volumes of temporal lobe structures to the hemicerebral volumes. Additionally, follow-up analysis included post-hoc Tukey's honestly significant difference test for individual region with right and left separately. Correlations between MRI measurements and clinical variables as well as among the MRI measurements were assessed using Pearson's product-moment correlation. This statistical approach using the parametric test had been validated by the Kolmogorov–Smirnov test to confirm normal distribution of the present MRI measurements and clinical variables. The level of statistical significance was determined as P < 0.05, in view of the exploratory nature of this study.
Temporal lobe structure–brain ratios
The sum of hemicerebral areas in five defined slices did not differ between schizophrenic and control groups (F = 1.12, d.f. = 1,43, P = 0.295) or between right and left hemispheres (F = 3.54, d.f. = 1,43, P = 0.067). However there was a significant main effect of gender (F = 41.09, d.f. = 1,43, P < 0.001). Post-hoc test revealed that male subjects have a greater hemispheric area than female subjects in both hemispheres of both diagnostic groups (P < 0.001).
Table 1 summarizes the regional brain measures and the results of ANOVA. There was a significant main effect of hemisphere in the MTS–brain ratio (F = 12.62, d.f. = 1,45, P = 0.001). Post-hoc test indicated that a laterality difference (i.e. the right being larger than the left) was observed only in the patient group (P = 0.003). The Sylvian fissure–brain ratio differed between patient and control groups (F = 39.77, d.f. = 1,45, P < 0.001) and between right and left hemispheres (F = 4.54, d.f. = 1,45, P = 0.039). Post-hoc test indicated that the schizophrenic patients had larger fissure than the controls in both hemispheres (P < 0.001) and that a laterality difference (i.e. the right being smaller than the left) was observed only in the control group (P = 0.046). In the inferior horn–brain ratio an ANOVA revealed significant group difference (F = 7.56, d.f. = 1,45, P = 0.009) and hemisphere difference (F = 8.70, d.f. = 1,45, P = 0.005). Post-hoc test revealed that the inferior horn–brain ratio was greater in the patients compared with the controls in both hemispheres (right, P = 0.013; left, P = 0.035), however, any laterality differences were not observed statistically.
Table 1. Temporal lobe structure–brain ratios in patients with schizophrenia and controls
Interaction (d.f. = 1,45)
Main effects (d.f. = 1,45)
Patients (n = 24)
Controls (n = 23)
Diagnosis × Hemisphere
ANOVA, analysis of variance; SD, standard deviation; MTS, medial temporal lobe structures.
P < 0.05,
P < 0.01 compared with controls (post-hoc Tukey's honestly significant difference test).
P < 0.05,
P < 0.01 compared with left hemisphere (post-hoc Tukey's honestly significant difference test).
In the schizophrenic patients, the left MTS–brain ratio was positively correlated with age at the onset of the prodromal phase (r = 0.479, P = 0.018) (Fig. 2a) and with age at the active phase (r = 0.454, P = 0.026). The right Sylvian fissure–brain ratio showed a significant positive correlation with the duration of illness (r = 0.424, P = 0.039) (Fig. 2b). None of the regional brain measures was correlated with either age at the time of MRI scanning or the extent of exposure to neuroleptic drugs.
The MTS–brain ratio did not correlate with the Sylvian fissure–brain ratio or the inferior horn–brain ratio, but with the contralateral MTS–brain ratio (r = 0.77, P < 0.001). There were significant relationships between the Sylvian fissure–brain ratio and the ipsilateral inferior horn–brain ratio (right, r = 0.513, P = 0.010; left, r = 0.456, P = 0.025). With regard to the relationship between both hemispheres, the right and left Sylvian fissure–brain ratios revealed a significant relationship (r = 0.597, P = 0.002), whereas the inferior horn–brain ratio did not correlate with the contralateral inferior horn (r = 0.376, P = 0.071).
Among 20 of 24 patients the mean composite SANS score (SD) for the patients was 58.7 (20.1). The mean PANSS scores were 13.7 (3.7) (positive symptoms), 13.1 (3.9) (negative symptoms), and 11.2 (1.8) (general psychopathology). There was not a significant correlation between the regional brain measures and the clinical symptoms.
It has been generally assumed that large variations in normal brain size need to be taken into account when looking at smaller areas within the brain. The present result of the sum of hemicerebral areas indicated that gender difference is not negligible. This finding is in keeping with that of previous reports concerning the gender difference in brain morphology.13,30,31 Therefore, statistical correction of brain size is essential for comparative purposes, especially using mixed subjects.
In the present study, the patients with schizophrenia had larger bilateral Sylvian fissures and inferior horns than the controls, which was consistent with findings of many previous computed tomography (CT) and MRI studies.4,19,32–35 The dilatation of the Sylvian fissure suggests a reduction in the volume of the surrounding brain tissue. Consistent with this assumption, decreased volume in the superior temporal gyrus, and the insular cortex has been reported.22,23,36–38 Thus, Sylvian fissure size may be a sensitive indicator for the subtle changes in the surrounding structures.
It may be of significance that duration of illness was correlated with the right Sylvian fissure size. Turetsky et al.39 reported a significant relationship between right frontal lobe volume reduction and duration of illness. Whether the brain morphological changes seen in schizophrenic patients are static or progressive remains to be clarified; some prospective follow-up MRI studies have found a reduction of total cerebral volume, a decrease in cortical grey matter volume during adolescence, and a progressive ventricular enlargement in at least a subgroup of schizophrenic patients.40–43 Moreover, progressive reduction of the superior temporal gyrus volume has been reported.44 One would expect that a subtle active process of altered brain cellular growth and repair may be continuing through the course of the illness in some patients.
The present results showed that age at the onset of illness correlated positively with the left MTS–brain ratio in the schizophrenic patients. There were no significant correlations between the left MTS–brain ratio and duration of illness or age at the time of MRI scanning, corroborating the results of other studies.1,4,13,45 Our previous study showed the inverse correlation of asymmetry of lateral ventricle size with age at the onset of illness.25 Jacobsen et al.46 and Fukuzako et al.47 reported that earlier onset of illness correlates with a lack of normal hippocampal asymmetry (i.e. right larger than the left). These findings support the view that the brain morphological change is a factor which constitutes vulnerability to the illness. The morphological changes in the medial temporal lobe may be developmental in origin, but the precise course of these changes in the premorbid period remains to be clarified.17,48
The present study could not replicate the finding of significant volume reduction in MTS in schizophrenic patients as had several other studies.1–4,23,46,49–51 These morphological changes have been reported to be more evident in male patients than female, and often only the male patients have been studied.3,4,13,23,52 In the present study, a greater number of female patients are included than male patients, and the hippocampus, amygdala and parahippocampal gyrus were measured as a single structure. Thus the subject selection and the method of measurement may be the reason for the failure to replicate the previous findings in this study.
It is of interest to investigate whether reduced volume in the MTS was accompanied by an increase in volume of the inferior horn of the lateral ventricle. Shenton et al.23 reported a significant relationship between these two brain structures. The present study, in accordance with previous studies by Bogerts et al.13 and Kawasaki et al.,4 failed to disclose any significant relationship between them. Contrary to earlier presumption that could be deduced from the progressive neurodegenerative process, it is more plausible that a lack of close relationship is consistent with a developmental disturbance or hypoplasia.
In conclusion, the present study suggests the significance of the Sylvian fissures in schizophrenia and that earlier onset of illness may be associated with smaller medial temporal lobe structures. It would be of crucial significance to study morphological changes in these structures during adolescence in normal controls and subjects in the prodromal or premorbid phase of this disease with longitudinal research.43,53
This study was supported in part by grants from the Japanese Ministry of Health and Welfare.