Structural Brain Abnormalities in Patients With Schizophrenia and Their Healthy Siblings

Cumulative evidence from family, twin, and adoption studies suggests that genetic factors play an important role in the risk of developing schizophrenia (1). Although the mode of inheritance is still unknown, 85% of the susceptibility to schizophrenia may be genetic in origin (2). Apart from an underlying genetic basis, additional environmental effects may contribute to the etiology of schizophrenia (3).

Results from structural imaging studies indicate that brain abnormalities play an important role in the pathology of schizophrenia. The most consistent morphological findings are lateral ventricle enlargement, which is indicative of reduced brain volume, and third ventricular enlargement (for review see reference 4). Other results suggest volume reductions in cortical gray and white matter (5–12) and in specific brain regions such as the frontal lobes (6, 8, 12–14), the thalamus (15–20), and limbic structures, including the amygdala and hippocampus (for review see reference 21). Morphological studies of the basal ganglia have been conflicting, with some describing volume increases (14, 22, 23) and others describing volume decreases (24, 25) of these structures, which may be related to differences in type of antipsychotic medication used (24). Studies of the cerebellum, and more specifically the cerebellar vermis, have also produced inconsistent results (for review see reference 26).

Whether structural brain abnormalities are also found in relatives of schizophrenic patients, who are thought to have a greater genetic liability for schizophrenia than nonrelated comparison subjects, is less clear. Family studies so far mainly reported lateral ventricular enlargement (27–30) and higher CSF volumes (5, 27) in first-degree relatives of patients. Lateral ventricular enlargement, however, is not found consistently (5, 19) or only in subgroups of relatives, such as those with multiple affected relatives (31) or relatives with a schizotypal personality disorder (32).

Few studies have measured brain structures other than the lateral ventricles or CSF in relatives of schizophrenic patients. Those that did reported enlargement of the third ventricle (19), volume decrease of the thalamus (17, 19), reductions of cortical gray matter (5), and volume loss of the amygdala-hippocampal complex (19).

Previously we reported that healthy siblings of patients with schizophrenia partially share the thalamic abnormalities of their affected relatives (17). The goal of the current study was to investigate in this same group the contribution of genotype to the morphology of specific brain structures other than the thalamus. Volumes of the cerebrum, gray and white matter, cortical CSF, lateral and third ventricles, frontal lobes, caudate nucleus, amygdala, hippocampus, parahippocampal gyrus, and the cerebellum were measured in patients with schizophrenia, their healthy siblings, and comparison subjects.

METHOD

Subjects

Thirty-two same-sex siblings (24 men and eight women) discordant for schizophrenia and 32 normal comparison subjects, matched for age, gender, and handedness, participated in this study after written informed consent was obtained.

Subjects were similar in mean age (schizophrenic patients: 40.6 years [SD=8.2], healthy siblings: 40.9 years [SD=8.6], and comparison subjects: 40.3 years [SD=9.3]). Four schizophrenic patients were left-handed, while all healthy siblings were right-handed.

The diagnosis of DSM-IV schizophrenia was determined by using the Comprehensive Assessment of Symptoms and History (33) and the Schedule for Affective Disorders and Schizophrenia—Lifetime Version (34). Siblings with personality disorders according to the Structured Clinical Interview for DSM-IV Personality Disorders (35) were excluded. All of the healthy siblings needed to be at least 8 years older than the age the affected sibling developed the first symptoms of schizophrenia, a criterion that suggests that they would be very unlikely to develop schizophrenia in the future (36).

Subjects with a major medical or neurological illness, IQ below 80, history of having received ECT, or history of substance dependence (substance abuse section of the Comprehensive Assessment of Symptoms and History) were excluded. No significant differences in educational level (defined as the total number of years of education) or parental educational level were found among the three groups.

Patients’ mean age at onset of illness was 21.9 years (SD=5.0), and the mean duration of illness was 20.1 years (SD=8.0). Fourteen patients were taking clozapine (mean=228 mg/day, SD=89), one was taking risperidone (6 mg/day), and one was taking bromperidol (4 mg/day). Of the 15 patients who were taking atypical antipsychotic medication, 10 had used it for at least 1 year.

Magnetic Resonance Imaging (MRI) Acquisition

MRIs were obtained on a Philips 1.5-T scanner. Volumes of interest were acquired by using T₁-weighted scans with 1.2-mm contiguous coronal slices (TE=4.6 msec, TR=30 msec, flip angle=30º, field of view=256256 mm) and dual contrast turbo spin-echo scans with 1.6-mm thick contiguous coronal slices (TE1=14 msec, TE2=80 msec, TR=6350 msec, field of view=256256 mm) and processed on a Unix-9000 workstation. All scans were blinded for diagnosis, left or right side of the brain, and put into Talairach frame. Intracranial, cerebral gray and white matter, lateral ventricles and third ventricle, and cerebellum volumes were automatically measured after correction for scanner nonuniformity by using histogram analysis algorithms and a series of mathematical morphological operators to connect all voxels of interest (37–39).

Intracranial volume was segmented on the dual contrast turbo spin-echo scans, with the foramen magnum being used as inferior boundary. Total brain volumes were segmented on the T₁-weighted scans and contained gray and white matter tissue only. In lateral ventricle segmentation, automatic decision rules bridged connections not detectable and prevented “leaking” into cisterns. The third ventricle was limited by coronal slices that clearly showed the anterior and posterior commissures; the upper boundary was a plane through the plexus choroideus ventriculi tertii in the midsagittal slice perpendicular to this slice. The cerebellum was limited by the tentorium cerebelli and the brain stem.

The following structures were measured manually in an anterior to posterior direction by using ANALYZE (Mayo Clinic, Rochester, Minn.). The frontal lobes were limited by the frontal pole, the lateral fissure, and the interhemispheric, circular insular, precentral, and cingulate sulci (40). The caudate nucleus segmentation started in the first slice, where it was clearly visible and was limited by the lateral ventricle, the internal capsule, the nucleus accumbens, and the slice that contained the posterior commissure (adapted from Chakos et al. [23]). Amygdala segmentation started in the coronal slice in which the optic tract is situated above the amygdala. Segmentation of the hippocampus started in the coronal slice in which the mammillary bodies were visible and stopped when the fornix was visible as continuous tract (41). Parahippocampal gyrus segmentation began simultaneously with segmentation of the amygdala. The posterior commissure was its posterior border.

The reliability of the measurements was determined by the intraclass correlation coefficient (ICC). The scans of 10 subjects were independently measured by two trained raters to calculate the ICCs of the volumes of interest: intracranial volume (0.99), total brain (0.99), cerebellum (0.95), lateral ventricles (0.99), and the third ventricle (0.95). Left and right intrarater ICCs were also calculated for the frontal lobes (0.98 and 0.99, respectively), caudate (0.88 and 0.95), amygdala (0.80 and 0.87), hippocampus (0.92 and 0.75), and the parahippocampal gyrus (0.97 and 0.93).

Statistical Analyses

Volumetric differences between the schizophrenic patients, their healthy siblings, and the comparison subjects (group) were analyzed for each structure by using repeated measures analysis of covariance (ANCOVA), with a group by side (left versus right) and, if applicable, by matter (gray and white) design.

Intracranial volume was used as a covariant for the cerebrum, gray and white matter, the ventricles, and CSF volumes, whereas whole brain volume was used as covariant for the frontal lobes, caudate nuclei, cerebellum, amygdala, hippocampus, and parahippocampal gyrus.

Two-tailed t tests were used to evaluate which group contributed most to significant ANCOVA effects. Differences between schizophrenic patients and their siblings were analyzed by means of paired t tests, whereas differences between patients and comparison subjects and differences between healthy siblings and comparison subjects were analyzed by means of independent t tests.

A second analysis was performed to control for possible confounding effects from the dependency within the siblings. ANCOVAs were used to compare patients with healthy comparison subjects, while excluding the healthy siblings from this comparison, in order to determine which abnormalities may be related to schizophrenia. Similar to the previous analysis, two-tailed t tests among patients, their healthy siblings, and comparison subjects were used to determine which abnormalities were possibly related to genotype.

When applicable, individual differences of brain structures between patients and their healthy siblings, or between healthy siblings and their matched comparison subjects, were analyzed by means of chi-square tests.

RESULTS

Raw volumes are listed in table 1.

No significant main effects or interactions were found for intracranial volume, cortical CSF, cerebellum, amygdala, hippocampus, or the parahippocampal gyrus.

A significant main effect of group was found for volume of the cerebrum (F=3.2, df=2, 60, p<0.05), which could mainly be attributed to a significantly lower volume in the schizophrenic patients than in the comparison subjects (t=2.5, df=46, p<0.01), whereas the volumes of the healthy siblings did not significantly different from either the patients (t=0.5, df=15, p<0.60) or the comparison subjects (t=1.0, df=46, p<0.30). No significant interactions were found.

A significant main effect of group was found for lateral ventricle volume (F=7.2, df=2, 60, p<0.01), which could mainly be attributed to volumes in schizophrenic patients that were significantly higher than those of their healthy siblings (t=3.2, df=15, p<0.01) and the comparison subjects (t=3.6, df=46, p<0.01); no difference was found between healthy siblings and comparison subjects (t=0.2, df=46, p<0.80), and no significant interaction was found.

A significant main effect of group was found for third ventricle volume (F=8.9, df=2, 60, p<0.01) (figure 1), which could mainly be attributed to significantly lower volumes in the comparison subjects than in both the schizophrenic patients (t=3.7, df=46, p<0.01) and the healthy siblings (t=2.2, df=46, p<0.04); the patients and their healthy siblings did not differ (t=1.8, df=15, p<0.10). In 12 of 16 patients, third ventricle volumes were larger than those of their healthy siblings (χ²=4.0, df=1, p<0.05), and in 13 of 16 healthy siblings third ventricle volumes were larger than those of their individually matched comparison subjects (χ²=6.3, df=1, p<0.05).

No significant main effect of group for frontal lobe volume was found (figure 2). However, a significant interaction of group by matter was found (F=4.6, df=2, 60, p<0.01), which could mainly be attributed to gray matter volumes that were significantly lower in the schizophrenic patients than in both the healthy siblings (t=2.1, df=15, p<0.05) and the comparison subjects (t=2.8, df=46, p<0.01). No significant differences in frontal white matter volumes were found between patients and comparison subjects (t=1.0, df=46, p<0.30), healthy siblings and comparison subjects (t=0.7, df=46, p<0.50), or between patients and their healthy siblings (t=1.5, df=15, p<0.20). No significant group-by-side or group-by-side-by-matter interactions were found.

A significant main effect of group was found for caudate nucleus volume (F=3.5, df=2, 60, p<0.04), which could mainly be attributed to significantly higher volumes in the schizophrenic patients than in both the healthy siblings (t=2.4, df=15, p<0.03) and the comparison subjects (t=2.6, df=46, p<0.01). No significant differences were found between healthy siblings and comparison subjects (t=0.2, df=46, p<0.80). No significant interaction of group by side was found.

The second analysis, used to control for dependency between the patients and their healthy siblings, did not change the findings.

DISCUSSION

The goal of the current study was to investigate the contribution of genotype on structural brain abnormalities in schizophrenia. To address this issue, volumes of the cerebrum, gray and white matter of the cerebrum, cortical CSF, lateral and third ventricles, frontal lobes, caudate nucleus, amygdala, hippocampus, parahippocampal gyrus, and cerebellum were measured in patients with schizophrenia, their healthy siblings, and matched healthy comparison subjects.

This study found that healthy siblings share third ventricle enlargement with their affected relatives and partially display a reduction in cerebral volume but show no other structural brain abnormalities. In contrast, their schizophrenic relatives display, in addition to enlargement of the third ventricle and a lower cerebral volume, higher lateral ventricle and caudate nucleus volumes and a volume reduction of frontal lobe gray matter in excess of their lower cerebrum volume. No abnormalities were found in intracranial volume, CSF volume, or volumes of the cerebellum, amygdala, hippocampus, or the parahippocampal gyrus.

Our finding of third ventricle enlargement is consistent with results reported by Lawrie et al. (19), although some methodological differences between the two studies exist. First, our group of first-degree relatives consisted of subjects without any psychiatric disorder, whereas the relatives in the study by Lawrie et al. (19) were not necessarily healthy relatives and were selected for having at least two family members with schizophrenia. Second, the relatives in our study were the patients’ own relatives, which implies that shared brain abnormalities between patients and relatives are possibly related to their partially shared genotype. In contrast, the study by Lawrie et al. compared patients with high-risk subjects who were not part of their family. Despite these methodological differences, results from both studies suggest that third ventricular enlargement may be related to the genetic vulnerability to develop schizophrenia.

Third ventricle enlargement suggests a reduction in brain tissue volume that reflects volume loss of diencephalic structures, such as the thalamus and the hypothalamus (4, 42). Indeed, the most prominent integrating center within the diencephalon, the thalamus, was reported to be smaller in patients with schizophrenia (15–20) and in subjects at risk for schizophrenia (17, 19).

In addition to third ventricle enlargement, the healthy siblings’ cerebral volumes were of intermediate size and did not significantly differ from those of the schizophrenic patients, who showed the smallest cerebral volumes, or the comparison subjects, who displayed the largest cerebral volumes. This finding is in agreement with results from a previous study that found reduced cortical gray matter volumes in siblings of patients with schizophrenia (5).

In the study by Lawrie et al. (19), however, whole brain volumes were not found to differ among patients, high-risk subjects, and comparison subjects, possibly because these measures were not corrected for intracranial volume as they were in our study. Indeed, without correction for intracranial volume, no significant group differences in cerebral volume would have been found in our study.

Apart from third ventricle enlargement and cerebral volume loss, the present study showed no familial effect for other brain structures. Since volume decrease of frontal gray matter, lateral ventricular enlargement, and volume increase of the caudate were typically found in the patients, these structural abnormalities may not directly be related to a genetic risk, but rather may be the consequence of the illness process itself or related to environmental effects. Results from a recent study by Sharma et al. (31) suggest that “obligate carriers” (relatives thought to have a higher genetic load for schizophrenia) but not other relatives (nonobligate carriers) have enlarged lateral ventricles. That study included subjects from families multiply affected by schizophrenia, whereas ours did not. Therefore, the volume measures of our healthy siblings can best be compared with those of the nonobligate carriers from the study by Sharma et al. Indeed, both the healthy siblings from our study and the nonobligate carriers from the study by Sharma et al. showed normal lateral ventricular volumes.

The cerebral volume loss in our schizophrenic patient group is consistent with results from a substantial number of imaging studies that have indicated that patients with schizophrenia have brain volume reductions of about 3% (for review see reference 43). In our study, this volume loss appears to be the result of both gray and white matter loss, which confirms findings from a recent study (5).

We found a reduction of the frontal lobe gray matter in the patients with schizophrenia that was in excess of whole brain volume reduction, which is in agreement with findings from previous imaging studies (6, 8, 12–14). Converging lines of research suggest that prefrontal gray matter volume reductions in schizophrenia may be related to neuronal abnormalities. Results from postmortem studies suggest a neuronal shrinkage in finding that the prefrontal neuronal density is greater in schizophrenia, whereas the number of prefrontal neurons is not (44, 45). Our finding of prefrontal gray matter volume reduction and previously reported thalamic volume loss in patients with schizophrenia appears to confirm the hypothesis that specific brain circuits that connect the frontal lobes, the basal ganglia, and the thalamus may be abnormal in schizophrenia (46).

In addition to the reductions in prefrontal gray matter, our schizophrenic patient group showed an increase in caudate volume. Several MRI studies of the basal ganglia have been published with largely equivocal results, which may reflect differences in patients’ medication status (14, 22–25). It is of interest that most of the patients in our study were being treated with atypical antipsychotic drugs and still showed caudate volume enlargement, which suggests that atypical antipsychotic medication may at least to some extent also induce a caudate enlargement.

In contrast to the reported abnormalities, the cerebellum, amygdala, hippocampus, and parahippocampal gyrus volumes were not different in schizophrenic patients. Previous reports that measured cerebellar volume are inconsistent, finding both enlargements and reductions in volume (for review see reference 26), whereas we found no significant differences among the groups. Similarly, our analyses of the amygdala, hippocampus, and parahippocampal gyrus volumes revealed no significant differences between schizophrenic patients and comparison subjects. In a recent meta-analytic study it was concluded that the hippocampus is bilaterally reduced in schizophrenia (21). Differences in methods such as variations in magnetic field strength, slice thickness, and segmentation criteria, may account for the inconsistency between our findings and those of that study. Furthermore, part of the reported findings of reduced hippocampal volumes may be related to complications during pregnancy or at time of birth since volume reductions of these structures may reflect hypoxia late in gestation or at time of birth (47).

Our results have to be interpreted carefully. The number of subjects is relatively small, and multiple statistical tests were required to interpret our data. Although the statistical power of our significant findings was adequate, we cannot exclude the possibility that the power of some of our nonsignificant findings may have been relatively small. Furthermore, the healthy siblings only partially share their genotype with their affected relative. In a polygenic disorder such as schizophrenia, the probability of a sibling inheriting the full combination of risk genes depends on the number of genes involved in the disorder (48). Since in schizophrenia a substantial number of genes may be involved, it is likely that some of the genes that may be related to schizophrenia are not present in the healthy siblings. Consequently, this may result in less pronounced brain abnormalities.

Nevertheless, our findings allow us to conclude that third ventricular enlargement, and to some extent cerebral volume reductions, may be related to genetic defects that produce a susceptibility to schizophrenia, whereas volume decrease of the prefrontal gray matter and volume increase of the lateral ventricles and the caudate nucleus may be disease-related or related to additional environmental effects that occur exclusively in patients with schizophrenia.

Received May 17, 1999; revisions received Aug. 30 and Oct. 12, 1999; accepted Oct. 22, 1999From the Department of Psychiatry, University Hospital Utrecht, the Netherlands. Address reprint requests to Dr. Staal, Department of Psychiatry MH-CRC, University of Iowa, 200 Hawkins Dr., Iowa City, IA 52242.

TABLE 1

Enlarge table