Abstract

1. Introduction

With increased evidence that schizophrenia is a brain disease, an issue of considerable current debate is whether it is predominantly a neurodevelopmental or neurodegenerative disorder. Neuropsychological data continues to help elucidate this issue by providing behavioral measures that can be linked to regional brain function. Evidence for impaired learning and memory (e.g., Calev et al., 1983; Saykin et al., 1991; Goldberg et al., 1993), attention (e.g., Nuechterlein and Dawson, 1984; Harvey et al., 1990; Nestor et al., 1992), and executive functions (e.g., Goldberg et al., 1987; Morrison-Stewart et al., 1992) has supported a model of fronto-temporal dysfunction in schizophrenia. The presence of these deficits at first presentation (e.g., Bilder et al., 1992; Saykin et al., 1994), combined with the lack of correlation with measures of chronicity (e.g., Goldberg et al., 1993; Nopoulos et al., 1994) are consistent with the neurodevelopmental hypothesis that stresses the relative stability of cognitive functions after the onset of schizophrenia (e.g., Weinberger, 1987; Wyatt, 1996). The cross-sectional rather than longitudinal design used in the majority of studies, however, precludes a rigorous test of the degenerative hypothesis since cross-sectional paradigms do not compare the same individuals at different times within the course of their illness.

Few longitudinal studies have been reported, and these were limited by small samples and/or lack of a longitudinal control group. In a sample of 28 first-episode patients, studied at 6 months after entry into the study and again at a 12-month follow-up, Bilder et al. (1991) found evidence of general neuropsychological stability with specific improvements in attention, motor, and generalized memory functions. This study however, did not test controls at both time periods, raising the possibility that patient improvements resulted from practice effects or other factors unrelated to course of illness. Sweeny et al. (1991) reported in a sample of 39 patients, measured between an acute episode of schizophrenia and at 12-month follow-up, improvements in attentional and executive domains as well as in motor functions. They also reported improvement on Judgment of Line Orientation and the Rey Auditory Verbal Learning Test. However, this study utilized test norms rather than a matched longitudinal control group, raising the same problems in interpreting ‘test improvement’ as the previous study.

Hoff et al. (1991) and Saykin et al. (submitted for publication) did utilize a follow-up control reference group. Hoff and colleagues compared a sample of 15 patients and 25 controls over a 24-month period and noted improved executive and concentration speed functions for 15 of the patients and global cognitive improvement for 12 of the patients. Saykin et al. (submitted for publication) examined the differences in neuropsychological performance of 21 patients and 28 controls followed over an average of 28 months. They reported equivalent improvement in both patients and control groups.

The relationship between neuropsychological change and clinical improvement has received less attention, and current findings are mixed. Cross-sectional studies suggest that negative symptoms correlate more with neuropsychological change than do positive symptoms (Bilder et al., 1985; Andreasen et al., 1990; Shtasel et al., 1992a). Longitudinally, Saykin et al. (submitted for publication) found correlations between improvement in negative symptoms and neuropsychological change. The longitudinal study by Hoff et al. (1991), however, found no correlations between either positive or negative symptoms and neuropsychological change. In a study of shorter duration (average 4 months) Silverstein et al. (1994) found no relationship between changes in six neuropsychological summary scores and changes in Brief Psychiatric Rating Scale symptoms, although they did report that neuropsychological impairment predicted blunted affect/emotional withdrawal at discharge.

The purpose of the present study was to establish in a large sample of patients with schizophrenia whether neuropsychological performance changes or remains stable relative to a group of healthy controls over a 1.5-year period. Both first-episode (FE) and previously treated (PT) individuals with schizophrenia were examined and any changes in neuropsychological performance were correlated with clinical change. Since symptoms and course of schizophrenia have been reported to be more severe in men (Goldstein, 1988; Angermeyer et al., 1990; Cowell et al., 1996), we also examined sex differences.

2. Methods

3. Results

3.1. Neuropsychological change

As can be seen in Fig. 1 there was a main effect of diagnostic grouping (patient vs. control; F(1,65)=51.71, p=0.0001), with patients performing worse at intake and follow-up on all neuropsychological functions relative to controls at intake and follow-up. A main effect of neuropsychological function (F(7,59)=4.38, p=0.0006) was also found, indicating that the performance profile for both groups was non-flat. There was no main effect of time, or any time × group interaction, indicating that performance was stable for patients and controls. An interaction was found between neuropsychological function and diagnostic group (F(7,59)=5.79, p=0.0001), indicating that the degree of patient impairment was more pronounced for some neuropsychological functions. The only other interaction was between neuropsychological function and ethnicity (F(7,59) = 3.96, p = 0.0013), indicating that the performance on some neuropsychological functions differed between racial groups across diagnosis. As in previous studies (Saykin et al., 1994; Cannon et al., 1994), patients showed generalized impairment (i.e., > 1SD below normative values) on all functions except motor, and the greatest level of impairment (i.e., >2SD) on verbal memory function. The profile shape was similar for both intake and follow-up evaluations.

Open in a separate window

Fig. 1

Neuropsychological profiles (mean(±SEM)): z-scores of patients and controls tested at intake and at follow-up. Variables are abbreviated as follows: abstraction (ABF), attention (ATT), verbal memory (VMEM), spatial memory (SMEM), language ability (LAN), spatial ability (SPA) and motor function (MOT).

Because of the interactions between diagnostic and racial groups with neuropsychological function, post-hoc ANCOVAs were performed separately for each of the eight functions. These analyses again showed that there were no interactions between time and diagnostic or racial group for any of the neuropsychological functions. Main effects of ethnicity were observed across diagnostic groups on the neuropsychological functions of abstraction (F(1,72) = 6.07, p = 0.02), attention (F(1,91) = 6.66, p = 0.01), verbal memory (F(1,93) = 4.17, p = 0.04), spatial memory (F(1,91) = 4.95, p = 0.03), language (F(1,90) = 9.99, p = 0.0021), and spatial factors (F(1,88) = 20.79, p = 0.0001). Performance was consistent for all six functions within each ethnic group, with the Caucasian group’s scores (n = 50) ranking above the African-American group’s scores (n = 46). Group ‘Other’ was ignored in this ranking since it contained only controls and only two subjects.

No differences between men or women were found when sex was entered as an additional classification variable in the overall MANCOVA. There were also no differences between first-episode and previously treated patients on any of the neuropsychological functions at either time. The lack of differences between the two patient groups in neuropsychological performance argues that medication did not have a deleterious effect. However, to confirm this we also examined the correlations between neuroleptic dose and neuropsychological performance on the eight functions. This was done both for intake and follow-up medication and neuropsychological values. None of these correlations were significant.

3.3. Relationship between clinical and neuropsychological change

The change scores for the eight neuropsychological and eight clinical factors are presented in Table 3.

As can be seen in Fig. 2, all the significant correlations were positive, indicating that clinical improvement was generally associated with cognitive improvement. Ten of the 15 significant correlations were with negative symptom factors. Of the negative symptoms, Anhedonia showed the most consistent correlations, being associated with abstraction, attention, spatial memory, and language and spatial ability neuropsychological functions. Although the most consistent correlations were with negative symptoms, the largest correlation coefficient was obtained between bizarre ideation on the SAPS and language functioning.

Open in a separate window

Fig. 2

Correlations between neuropsychological and clinical corrected change scores. Clinical variables abbreviated as in Table 3; functions abbreviated as in Fig. 1.

4. Discussion

The results do not support a neurodegenerative model of schizophrenia. Patients’ neuropsychological performance was equally impaired for first-episode and previously treated patients, and both groups showed the same test profile. The lack of difference in patient groups is consistent with cross-sectional findings of equal impairment in first-episode and previously treated patients (Saykin et al., 1994; Sweeny et al., 1991). In addition, these data show that longitudinally, no deterioration occurs over a 19-month period: patients did not show deterioration over time but remained stable in all domains.

The clinical results also do not support a neurodegenerative model: patients improved clinically on measures of both positive and negative symptomatology. Although improvement was seen in both types of symptoms, the most consistent correlations with cognitive change were seen with negative symptom improvement. This finding is consistent with previous reports in the literature supporting the generalization that negative symptoms account for more of the neuropsychological variance than positive symptoms (Bilder et al., 1985; Andreasen et al., 1990), and that they represent core-features of schizophrenia (Carpenter et al., 1988).

No sex differences were found for either neuropsychological or clinical change in this sample, although sex differences in the severity and course of schizophrenia have been documented (Shtasel et al., 1992a). This negative finding may reflect sample size. A larger sample size may also permit a more detailed examination of individual items on the clinical scales in relation to specific neuropsychological domains. Such an analysis could help identify possible mechanisms through which clinical and neuropsychological changes covary. Clinical phenomena and neuropsychological performance ultimately do both reflect behavioral manifestations of the same disease process.

Acknowledgments

References