METHODS

Subjects

There were three age and sex matched groups: OCD, schizophrenia, and normal subjects. Each group consisted of 22 subjects (15 men and 7 women), who were all right handed, and had a mean age of 26.7 (SD 7.2), 26.6 (SD 6.5), and 26.2 (SD 6.1) years, respectively. Patients were recruited from Seoul National University Hospital, and none had a dual diagnosis of OCD and schizophrenia. Normal volunteers were recruited from the community through newspaper advertisements. Mean period of education and IQ were statistically different among the groups, but mean parental socioeconomic status was not significantly different. This study was carried out under guidelines for the use of human subjects established by the institutional review board. Written informed consent was obtained from all subjects.

Magnetic resonance image acquisition and processing

Three dimensional T1 weighted spoiled gradient echo MR images were acquired on a 1.5 tesla GE SIGNA scanner, using imaging parameters of 1.5 mm sagittal slices, TE = 5.5 ms, TR = 14.4 ms, NEX = 1, rotation angle = 20°, FOV = 21×21 cm, and a matrix of 256×256. MR images were processed using an image processing software package, ANALYZE (version 3.0, Mayo Foundation, USA). Images were resampled, reoriented, spatially realigned to the conventional position, and filtered using the anisotropic diffusion method with five iterations to improve the signal to noise ratio. Intracranial volume (ICV) was calculated after the brain extraction and segmentation.

ROI tracing

The neuroanatomical regions of interest (ROIs) were traced manually on all coronal slices with the references of sagittal or axial planes using the ANALYZE ROI module. Boundaries of the hippocampus were defined using a minor modification of the method previously reported by Pantel and colleagues.⁶ Tracing was performed initially on the sagittal slices for a reliable separation, and proceeded to the coronal plane applying different boundaries according to the positions of the head, body, or tail of the hippocampus. Boundaries of the amygdala were preliminarily drawn on the axial plane, using a modification of the delineating methods of Honeycutt and colleagues⁷ and Convit and colleagues⁸; the main tracing was performed on the coronal view. Boundaries of the thalamus were based on the detailed guidelines of Portas and colleagues.⁹ The medial and lateral geniculate bodies were included as a part of the thalamus.

In order to assess inter-rater reliability, a tracing was performed independently on a set of 10 MR scans by two raters (YIK and CWK) and volumes of the traced regions were measured automatically. The intra-class correlation coefficients (ICC) were calculated and found to be high: 0.89/0.90, 0.87/0.91, and 0.95/0.97 for the left/right side of the hippocampus, amygdala, and thalamus, respectively. These high reliability scores corresponded to relatively small percentage volume differences—that is, the smallest 1.8% (SD 1.1) for the right thalamus and the largest 6.7% (SD 5.3) for the left amygdala. In the case of the between group measures, one rater (YIK) traced the hippocampus and the amygdala, and the other rater (CWK) traced the thalamus.

Statistical analysis

Primary analysis was performed using a mixed model ANCOVA with one between subjects factor and one within subjects factor for each region. This analysis was followed by one way ANCOVA with Bonferroni post hoc tests for the left and right of each individual. One way ANOVA was used to test for ICV group differences. The ICV was used as covariates for all ANCOVAs. All statistical analyses were two tailed, and p < 0.05 was used as a level of significance.

RESULTS

As presented in table 1, mean ICV was not statistically different among the three groups. A significant main group effect was found for hippocampal volumes (F_{2, 63} = 7.30, p < 0.01), and significant group differences were revealed in both sides of the hippocampus (left: F_{2, 63} = 6.92, p < 0.01; right: F_{2, 63} = 6.29, p < 0.01); both OCD and schizophrenia groups had bilaterally smaller hippocampal volumes than the normal comparison subjects. No significant main effect of side, and interaction effect of side-by-group were found.

A significant main group effect was also found for amygdala volumes (F_{2, 63} = 3.50, p < 0.05). A significant group difference was revealed in the left amygdala only (F_{2, 63} = 4.17, p < 0.05); the OCD group had a greater left amygdala volume than both the schizophrenia and the normal groups. No significant main effect of side, and interaction effect of side-by-group were found. In case of thalamic volume, any significance was not found in a main group effect, group differences for each side, and a main effect of side and interaction effect of side-by-group.

DISCUSSION

Our findings of a bilaterally smaller hippocampus in both OCD and schizophrenia than in normal subjects show that hippocampal volume reduction is not specific to each illness. Although hippocampal volume reduction has been one of the most consistent abnormal morphological findings in schizophrenia, a comparison of schizophrenia with affective psychosis also fails to show its specificity.¹⁰ Therefore, these non-specific findings are likely to link to clinical overlap among the psychiatric illnesses, although its mechanism is unclear.

Contrary to the hippocampus, findings from the amygdala support our hypothesis that OCD and schizophrenia reveal different abnormal patterns. Enlargement of the left amygdala in OCD may relate to amygdala hyperfunction, and supports the amygdalocentric model regarding the pathophysiology of OCD.¹¹ The amygdala may be a key neuroanatomic substrate of the anxiety that perpetuates compulsions in OCD. Given that the anxiety mediated by amygdala is inhibited by feedback from the orbitofrontal cortex,¹² the normal cortical-to-amygdala inhibition that prevents the amygdala from being overactive is likely to malfunction in OCD.

Although the thalamic volume did not show any group differences, it may be too large as an ROI to detect subtle changes confined to a specific thalamic nucleus. Significant thalamic abnormalities in schizophrenia have only emerged when single specific thalamic nuclei have been considered.⁵ Subregional thalamic abnormality in OCD was found in our previous study⁴ using a voxel based morphometric approach, and was too small to detect using the present ROI method. Therefore, a subregional approach is needed to identify a possible thalamic abnormality in OCD, as attempted in schizophrenia.

To our knowledge, this is the first report that compares volume measures of the hippocampus-amygdala complex in OCD and schizophrenia. Further research to determine the nature of the different involvements of OCD and schizophrenia with the abnormal fronto-temporal network may shed further light on the pathophysiology of these illnesses.