Decreased Caudate Volume in Neuroleptic-Naive Psychotic Patients

Sixteen first-episode neuroleptic-naive schizophrenic patients (schizophrenia, N=13; and schizoaffective disorder, N=3), nine nonschizophrenic psychotic patients (bipolar disorder with psychotic features, N=3; major depression with psychotic features, N=3; and psychotic disorders not otherwise specified, N=3), and 17 normal comparison subjects were studied. All subjects provided fully informed written consent. The patients were diagnosed by DSM-III-R criteria at consensus meetings incorporating all clinical, follow-up, and Structured Clinical Interview for DSM-III-R (SCID) (25) data. Normal comparison subjects were recruited from neighborhoods in which the patients resided, were assessed with the SCID (nonpatient version), and were matched, as a group, with the schizophrenic patients for age, sex, race, and parental socioeconomic status (26). Exclusion criteria included significant medical illness, substance abuse or dependence, prior neuroleptic treatment, and IQ lower than 70. Illness durations were computed from the onset of prodromal symptoms to entry into the study. In order to obtain the onset dating as accurately as possible, we timed the dates of onset of the first identifiable psychotic and prodromal symptoms on the basis of a best-estimate approach using data gathered from multiple sources, including the medical record, direct patient interview, and, when possible, a family interview. Dating of the onset of the first prodromal or psychotic symptom was estimated to the closest month (estimated range of error=±1 month), when possible. The duration for the schizophrenic group (N=15) was 301.35 weeks (SD=269.14, range=8–823). For one patient, it was difficult to determine the age at onset. The illness duration in the nonschizophrenic psychotic group was 155.06 weeks (SD=159.04, range=6–463).

MRI scans were conducted with a 1.5-Tesla General Electric Signa system (General Electric Medical Systems, Milwaukee) at the University of Pittsburgh Medical Center. A spin echo sagittal scout scan and axial scans (slice thickness=2.8 mm; no interslice gap; three-dimensional spoiled gradient recalled acquisition pulse sequence; TR=40 msec; TE=20 msec) were acquired. They were identified by scan number alone to retain blindness and were analyzed on a MacIntosh workstation with IMAGE software (version 1.55) (27) by trained raters, through use of a semiautomated segmentation algorithm (28). The right and left caudate nuclei were measured by using a manual tracing technique (22, 29). Intracranial volume was calculated by summing up areas of successive axial slices including gray matter, white matter, and CSF and multiplying by slice thickness. Mean number of slices was as follows: right caudate, 11.5; left caudate, 11.5; right putamen, 7.1; and left putamen, 7.2. Intrarater reliabilities (N=10) ranged from 0.90 to 0.97 for caudate and from 0.92 to 0.94 for putamen. Interrater reliabilities in our laboratory (W. Bagwell, E. Dick, M. Zeigler, and T. Kisler) (N=10; intraclass r) were high for caudate (r=0.94), putamen (r=0.97), and intracranial volume (r=0.99) but were poor for globus pallidus (r=0.26).

Separate measurements were obtained for the left and right caudate nuclei through use of a manual tracing technique. The head and body of the caudate were measured, whereas the tail was excluded. The caudate nucleus is delineated medially and superiorly by the lateral ventricle and elsewhere by frontal white matter; it is separated from the putamen by the white matter tracts of the anterior limb of the internal capsule. We included the nucleus accumbens, which merges inferiorly with the caudate; it is hard to reliably demarcate this structure from the caudate nucleus. All regions were measured in the axial plane.

Separate measurements were obtained for the left and right putamen by using manual tracing. The putamen is delineated medially by the globus pallidus and laterally and inferiorly by white matter tracts of the external capsule. The caudate and the putamen are connected inferiorly by gray matter bridges; because of the partial volume effects, these are excluded from the measurements. Measurements of the putamen were performed in the axial plane (figures 1 and 2).

We conducted analysis of covariance (ANCOVA), with intracranial volume as a covariate, and Newman-Keuls post hoc tests to examine significant differences between diagnostic groups; we used Pearson's correlations and partial correlations to examine the relation among age, illness duration, and caudate and putamen volume. For dichotomous variables such as sex and handedness, we used the Kruskal-Wallis analysis of variance (ANOVA). All significance values reported are two-tailed.

There were no significant differences among the three groups in age, sex, education, handedness, or parental socioeconomic status (table 1). Age did not correlate with caudate and putamen volumes in any of the diagnostic groups. Intracranial volume was not significantly different among the groups (schizophrenic patients: mean=1476.99 cc, SD=185.33; nonschizophrenic psychotic patients: mean=1528.93 cc, SD=17.68; comparison subjects: mean=1576.57 cc, SD=172.97) (F=1.41, df=2,39, p=0.26). ANCOVA, with intracranial volume as a covariate, revealed significant intergroup differences in both left and right and in total caudate volume (table 2). Post hoc tests revealed that both patient groups had smaller right and left caudate volumes than the comparison subjects. Illness duration (after intracranial volume was partialled out) was unrelated to caudate or putamen in both the schizophrenic and the nonschizophrenic psychotic groups (all partial r values <0.5). No significant differences were seen between the subject groups for right or left putamen.

Significant reductions (about 14%) were seen in caudate volumes in neuroleptic-naive schizophrenic patients. These findings may be diagnostically nonspecific, since similar volume reductions were also seen in nonschizophrenic psychotic patients; reduced caudate volume has also been described in depression (30). In a prospective study of a subset of these first-episode psychotic patients, we (22) and others (10) have found that caudate volumes increase following treatment with neuroleptics. Previous observations of increased caudate volume in schizophrenia may therefore reflect neuroleptic effects. Thus, caudate volume reduction in the neuroleptic-naive patients may reflect primary pathophysiology of schizophrenia and may account for some of the cognitive and psychomotor abnormalities in this illness.

Our observations of caudate, but not putamen, volume reductions in neuroleptic-naive psychotic patients deserve comment. The putamen mainly receives projections from cortical regions involved in motor control (31, 32). On the other hand, recent studies in primates have shown that the caudate nuclei are clearly activated during working memory-related tasks (2). Thus, this structure may be part of a distributed neuronal network subserving functions associated with the dorsolateral prefrontal cortex, implicated in the pathogenesis of schizophrenia. Our observations are also consistent with functional neuroimaging research. Significantly reduced basal ganglia metabolism has been observed in unmedicated schizophrenic patients through use of positron emission tomography (33–35) and single photon emission tomography (36). On the other hand, treated schizophrenic patients had higher blood flow in basal ganglia (37, 38). Indeed, one study showed decreased caudate metabolism at baseline that reversed following institution of neuroleptic treatment (34).

The lack of association between caudate volume and illness duration should be viewed with caution, since studies of patients with longer duration of illness may reveal the effects of chronicity. Nevertheless, this finding and the observation of caudate volume reduction early in the illness provide some indirect support for neurodevelopmentally mediated pathology in schizophrenia (39) and possibly in affective disorder (40). An exaggeration of periadolescent synaptic pruning, perhaps in glutamatergic corticosubcortical neurons, may be involved (41). Reduced activity in these cortico~striatal neurons, by diminishing trophic effects on the striatum, could conceivably lead to reduced synaptic neuropil, and thereby reduced size; this view is consistent with a recent observation of reduced striatal dendritic spine size in postmortem brains of schizophrenic patients (42).

Many brain regions have been implicated in the neuroanatomy of schizophrenia. Our findings suggest that the relevant brain circuits underlying the pathophysiology of schizophrenia are likely to include the caudate nuclei; however, several other regions and abnormal patterns of interregional cross-talk may be involved in this disorder. Our findings must be considered preliminary in view of the relatively small study group size. Further studies of basal ganglia structure and metabolism, through use of state-of-the-art neuroimaging techniques, are clearly warranted.

Received May 20, 1997; revision received Nov. 10, 1997; accepted Nov. 25, 1997. From Western Psychiatric Institute and Clinic. Address reprint requests to Dr. Keshavan, Western Psychiatric Institute and Clinic, 3811 O'Hara St., Pittsburgh, PA 15213-2593; [email protected] (e-mail). Supported in part by NIMH grants MH-01372, MH-42969, and MH-45156 and by a Scottish Rite Foundation grant. Wright Bagwell, B.A., Tanya Kisler, B.S., Melissa Zeigler, B.S., and Elizabeth Dick, B.S., conducted morphometric measurements; Drs. Gretchen Haas and Nina Schooler helped in the study's clinical aspects.