Cognitive Impairment in Geriatric Schizophrenic Patients :

Michael Davidson and Vahram Haroutunian

The outcome of schizophrenia in old age remains among the most debated topics in schizophrenia research. The debate between the Kraepelinian pronouncement that the outcome is invariably bleak and the view that the outcome of schizophrenia in old age is variable (20, 33) focuses on the schizophrenic cognitive capacities in old age and not on the psychosis, which for many (but not all) patients ameliorates. There is consensus among investigators that nearly all young and middle-aged schizophrenic patients suffer from moderate cognitive impairment (see The Effects of Neuroleptics on Plasma Homovanillic Acid), and that a certain proportion of geriatric patients suffer from a very severe form of cognitive impairment (20, 36). There is, however, no consensus on the proportion of geriatric schizophrenic patients who suffer from the severe form of cognitive impairment, on the specific manifestations of the cognitive impairment in old age, or on how moderate impairment of specific cognitive aspects progresses, if at all, into severe and possibly global cognitive impairment.

The debate on the outcome of schizophrenia in general, and on the cognitive impairment in particular, is complicated by the fact that the criteria used to define good or poor outcome are influenced by (a) the level of care needed by the patients and (b) the type of health facility where this care is delivered. For example, good outcome is associated with patients discharged from psychiatric hospitals into the community, and with care delivered by outpatient clinics, whereas poor outcome is associated with continued inpatient status. Early in the schizophrenic illness the need for inpatient psychiatric care is determined by the severity of psychotic symptoms and cognitive impairment (13). However, as psychotic symptoms ameliorate with age, cognitive impairment may become the major determinant of the type of care necessary and of the facility where this care will be provided. For example, of the 300,000 geriatric schizophrenic patients in the United States, approximately 200,000 live in nursing homes, while only 15,000 live in psychiatric hospitals (32, 61, 63, 71). It is plausible to assume that, among many other factors, amelioration of psychotic symptoms has enabled the discharge of geriatric schizophrenics from psychiatric hospitals. It is also plausible to assume that cognitive impairment has required that these patients continue to live in protected environments such as nursing homes. Therefore, if living in a protected environment is used as a criteria for outcome, cognitive impairment may be a major determinant of outcome.

The paucity of investigations on the cognitive impairment in geriatric schizophrenics can be understood in light of the methodological difficulties inherent in studying the long-term outcome of schizophrenia in general (33) and those inherent in studying cognition in the context of a chronic illness and aging in particular. Cohort attrition, diagnostic uncertainty and heterogeneity, and changes in the social and treatment environment not accounted for in the original study design are a few of the difficulties common to schizophrenia outcome studies. Interactions of cognitive functioning with positive and negative symptoms, other schizophrenic symptoms, poor cooperation, depression, poor education, somatic treatment, length of hospitalization, and concomitant neurological illnesses are some of the specific methodological difficulties which can confound or affect cognitive performance.

Furthermore, in young and middle-aged schizophrenic patients, findings that executive and memory functions, believed to be subserved by the frontal and temporal lobes respectively, are disproportionately impaired (29, 65) have been invoked in directing neuroradiological and neuropathological research to these areas (7, 76). However, as chronic schizophrenia and aging interact, separating the domains of cognition which are disproportionately affected from those which are not may not be feasible, and the question of generalized versus specific cognitive impairment, which is still debated in young patients, becomes even more difficult to resolve in geriatric schizophrenics. Thus, the already tenuous inferences made about brain dysfunction and structural lesions based on neuropsychological test results become even more speculative in geriatric patients.

Despite methodological difficulties, investigating the cognitive impairment in geriatric schizophrenic patients is essential to elucidate the pathophysiology of this illness and improve the care given to this population. For example, investigating the lifelong course of cognitive impairment may add support to either the developmental or the degenerative hypothesis of schizophrenia. An abrupt decline in cognitive performance in late adolescence or early adulthood without additional progression, or with very slow progression over the rest of the lifespan, would lend support to the evolution of an early static lesion in the context of a chronic disease and aging (see The Neurobiology of Treatment-Resistant Mood Disorders). Conversely, periods of rapid and progressive cognitive decline, or the appearance in geriatric patients of novel aspects of cognitive impairment not found in young patients, in conjunction with a characteristic neurohistological signature, would lend support to an active, possibly degenerative type of disorder.

Equally relevant to schizophrenia research is the fact that most schizophrenic patients who come to autopsy die at a very advanced age and are cognitively impaired, making it plausible that some of the plethora of neuroanatomical and neurochemical findings which are attributed to schizophrenia are in fact causally related to the cognitive impairment of schizophrenia. It would not be implausible to suspect that not accounting for the presence and severity of the cognitive impairment and for the age-related changes in schizophrenia symptomatology in general has contributed to the inconsistencies reported in neuropathological and neurochemical studies of this illness. Comparing cognitively impaired and cognitively nonimpaired schizophrenic patients in terms of demographic correlates, associated symptoms, and course in conjunction with postmortem studies might resolve these inconsistencies and help delineate a subtype of schizophrenia with a biologically distinct substrate.

Finally, in terms of the contribution to the care of geriatric schizophrenics, characterizing the cognitive impairment and knowing its course might predict the degree and type of support needed (13, 80) and thus assist in a rational planning of care delivery. Furthermore, in order to evaluate the efficacy of the treatments currently available for geriatric schizophrenic patients and to conceptualize treatments specifically targeted at ameliorating the cognitive impairment or delaying its onset, it is essential to know its course and manifestations.

This chapter will review the course and severity of cognitive impairment, the factors associated with it (i.e., other schizophrenic symptoms, demographic variables, and somatic treatment), and the potential biological substrates which may be responsible for this condition in geriatric schizophrenic patients whose illness started before age 45 (for a review of late onset schizophrenia, see The Neurobiology of Treatment-Resistant Mood Disorders and Anatomic and Functional Brain Imaging in Alzheimer's Disease). For the purposes of this review, cognitive impairment will refer to impairment in more than one area of cognition which is sufficiently severe to affect daily social functioning.

Despite the shortcomings of long-term follow-up studies, they remain the preferred method to investigate the course of cognitive impairment and to place it in the overall context of the outcome of schizophrenia. The ideal study would administer psychometric tests to a cohort of first-episode, very young schizophrenic patients drawn from a catchment area and would then repeat the testing periodically until very old age or death. Unfortunately, published long-term studies either are (a) limited to narrative descriptions of cognitive decay without formal cognitive assessments of the entire cohort (20) or (b) cover a restricted range of patients' lives (e.g., early life or middle age; see refs. 27 and 36 for reviews). A number of "first-episode" psychosis catamnestic studies which periodically administer psychometric tests were initiated in the late 1980s (14), but their long-term results will not be available in the near future.

A more feasible approach to estimate the progression of the cognitive impairment in geriatric schizophrenics is to compare cognitive performance between geriatric (>65 years of age) and nongeriatric schizophrenic patients in cross-sectional studies. However, results of cross-sectional studies are vulnerable to "sample biases," because geriatric schizophrenic patients who participate in cross-sectional studies are recruited mostly, but not exclusively, from psychiatric inpatient facilities (27, 36). In contrast, patients who require no psychiatric care, and for whom old age is associated with independent living in the community, are difficult to locate and are rarely included in cross-sectional studies. Because the ability to live independently in the community is associated with better cognitive functioning (13), results of cross sectional studies may overestimate the prevalence of cognitive impairment in geriatric schizophrenia.

Cross-sectional studies are also subject to "birth cohort effects." If younger patients have a less severe form of illness than older patients [due, for example, to treatment early in their illness with neuroleptics (79)] or if the younger patients were more educated (a factor associated with better cognitive performance), this "birth cohort effect" could be misinterpreted as age-related deterioration in individual patients (66). Similarly, comorbid neurological conditions which affect cognitive performance and whose prevalences increase with age (such as Alzheimer's disease, multi-infarct dementia, and Parkinson's dementia) could likewise be misinterpreted as age-related deterioration in individual patients (see below for a discussion of this topic). In contrast, cross-sectional studies have the advantage of covering large age ranges and including large numbers of patients so that potential differences in cognitive functioning between age groups are not underestimated.

With the aforementioned limitations in mind, reviews of cross-sectional comparison studies have found that young schizophrenic patients performed better than geriatric patients on some (but not the majority of) psychometric tests, and they concluded that the differences between the younger and older patients were not sufficient to support a progressively deteriorating course of the already existing cognitive impairment (27, 36). However, a most comprehensive review of over 100 studies (36) and the only very long-term catamnestic study (20) agree that a progressively deteriorating course of cognitive function may occur in a subgroup of schizophrenic patients whose illness is characterized by (a) very severe and persistent symptoms throughout life and (b) lengthy or uninterrupted hospitalization.

In order to investigate if such a subgroup exists and to overcome the limitations related to too narrow an age-range comparison, we divided the entire inpatient population of chronic schizophrenics residing at a long-term psychiatric hospital into seven age groups ranging from 25+ to 85+ years of age (25-34, 35-44, 45-54, etc.) and compared their cognitive performance (22). After excluding the patients who on medical screening were found to suffer from neurological or medical conditions which could affect cognitive performance, 393 subjects remained, 308 of whom were geriatric (>65 years of age or older). All 393 patients had been admitted to this chronic care psychiatric hospital after they had failed several treatment trials, and the majority of them are very chronic and severely ill patients. A battery of tests sensitive to progressive, cognitive deterioration used by the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) (51) was administered. Analyses of variance examining the effect of age group (i.e., decade) on the performance of each of the five tests found statistically significant effects of age group for each measure (see Table 1).

The results indicated that all chronic schizophrenic patients have poor learning, memory, recognition, and praxis abilities; however, young schizophrenic patients have better cognitive functioning than do geriatric schizophrenic patients. Despite the fact that every age group obtained a better mean score than the following (older) age group on virtually every measure, in order for the age-associated differences to reach statistical significance it was necessary to compare age groups separated by at least 20 years and to include in the comparison patients older than 65 years of age. Thus, the differences between age groups would have been overlooked if the age range was not broad enough or if the older groups were left out of the comparison, suggesting that if the differences in performance between age groups reflect cognitive decline in individual patients the decline is very slow.

Despite the relatively small differences in scores between each age group of schizophrenics, these differences are much larger than the differences between the same age groups of equivalently educated normal control individuals. Furthermore, in a multiple regression analysis, the contribution of age to the variance of psychometric scores in the population of geriatric schizophrenic patients was approximately 5%, which is much lower than the contribution of age to the variance in cognitive performance in a group of geriatric normal controls (57). Together, this and other data (67) suggest that for severely ill schizophrenic patients, the age-related differences in cognitive performance are larger than differences associated with normal aging. On the other hand, the differences are significantly smaller than the decline observed in individuals suffering from a typical degenerative dementia such as Alzheimer's disease (51).

In summary, these data are consistent with a very slow decline in cognitive performance over the lifespan of severely ill, chronically hospitalized schizophrenic patients, and they are inconsistent with either normal aging or a typical degenerative disease. These results are not in conflict with three reviews of this topic (27, 36), which concluded that because the differences between younger and older schizophrenic patients are not very large, the data are not consistent with a progressive deterioration of a degenerative type. The study presented here found larger differences between age groups perhaps because it compared cognitive function over a much broader age range in larger age groups and in a more severely ill patient population. In fact, the population presented here consists of (a) young and middle-aged patients who have repeatedly failed several consecutive treatment trials and (b) geriatric patients, the overwhelming majority of whom have been in the same institution for over 20 years. The question of whether the young and middle-aged patients included in this cross-sectional study will eventually develop the same kind of severe cognitive impairment as the geriatric patients, and if they all belong to a phenomenologically and biologically distinct subtype of schizophrenia, can only be conclusively answered with a long-term follow-up study. However, the chronicity and the already impaired cognitive functioning of the young and middle-aged patients evidenced by the scores presented in Table 1 suggest an affirmative answer to this question.

Of the geriatric schizophrenic patients who reside in chronic psychiatric hospitals, over 50% suffered from severe impairment in more than one area of cognition which affected social functioning (46, 53), thus meeting DSM-III-R criteria A to D for dementia. Despite persistent suggestions that the unstimulating institutional social environment contributes to the cognitive impairment (78), and despite this suggestion's face validity, a cause-effect relationship has not been established (1, 31, 40).

To investigate the role of institutionalization in the cognitive impairment of geriatric schizophrenic patients, several strategies can be employed, all of which have serious methodological flaws. For example, a case-control study could compare cognitive performance in pairs of schizophrenic patients admitted at the same time, preferably in the same institution, and match them on most demographic and illness related variables, with the exception that one patient has been continuously institutionalized for most of his/her life and the other immediately discharged and never or only briefly readmitted. While no such study has been reported, an observation (40), though not confirmed (47), suggested that patients discharged from an institution had slightly better cognitive functioning than did patients who remained institutionalized, and it could be interpreted as evidence that institutionalization is causally related to cognitive impairment. However, this conclusion is flawed because the assignment of patients to remain in an institution or to be discharged cannot be randomized. Patients who have better cognitive performance are more likely to be discharged and to stay in the community (13); thus, it may very well be that cognitive impairment serves as a selection factor for long-term institutionalization, rather than long-term institutionalization being the cause of cognitive impairment (1).

An alternative approach to separate the contributions of institutionalization and the schizophrenic illness to cognitive impairment is to compare the cognitive performance of the geriatric schizophrenics to the performance of another diagnostic group of chronically institutionalized patients such as geriatric patients suffering from mood disorder. Of the 50 mood disorder inpatients identified by our study, most of whom have been institutionalized for more than 20 years, only 30% met DSM-III-R criteria A to D for dementia, which is significantly less than the proportion (50%) of geriatric schizophrenic patients who met the criteria and reside in the same institution. Furthermore, the average Mini Mental State Examination (MMSE) scores of the mood disorder patients were significantly higher (better) than the scores of the schizophrenic patients, and there was no correlation between years of institutionalization and MMSE scores in mood disorder patients, while there was a correlation, albeit weak, between these variables in the geriatric schizophrenic patients. Taken together, the data indicate that institutionalized geriatric patients suffering from mood disorder have better cognitive functioning than do institutionalized geriatric schizophrenic patients; however, the differences between the two diagnostic groups were not clinically substantial. Thus, while the comparison between the two diagnostic groups implicates the specific schizophrenic illness rather than a generalized institutionalization effect, it does not rule out that institutionalization contributes to the cognitive impairment of both geriatric schizophrenic and mood disorder patients. On the other hand, it could be hypothesized that cognitive impairment is a point of convergence for very severe forms of schizophrenia and mood disorder, possibly mediated by a biological abnormality shared by the two disorders (77).

In summary, evidence can be accumulated to suggest that cognitive impairment is a selection factor for long-term institutionalization and is responsible for patients' continued hospitalization, rather than supporting the notion that cognitive impairment is the inevitable result of long-term institutionalization (1, 31, 40). In fact, at least one study investigating the long-term effects of deinstitutionalization has suggested an association between deinstitutionalization and worsening in cognitive functions and positive symptoms (80). Symptom aggravation even after a long period of deinstitutionalization was observed in patients who suffered from mild-to-moderate cognitive impairment before deinstitutionalization, suggesting that these patients did not benefit from community care and supporting the notion that cognitive function is predictive of independent living (13) and level of care required.

Correlational analyses in samples of young and middle-aged schizophrenic patients have consistently revealed associations between severity of cognitive impairment and negative symptoms (5, 48). Similarly, our study of 393 schizophrenic patients between the ages 25 and 85 revealed strong correlations between cognitive impairment and negative symptoms. The correlations remained significant, and their magnitude did not change throughout the entire lifespan. These correlations between cognitive impairment and negative symptoms led to the hypothesis that the two symptoms share a common biological substrate. Components of cognitive dysfunction and of negative symptoms have both been attributed to structural and functional abnormalities of the dorsolateral prefrontal cortex and its connectivities to the medial temporal lobe (65; also see The Neurobiology of Treatment-Resistant Mood Disorders).

Common schizophrenic symptoms such as preoccupation with delusions and hallucinations, as well as poor cooperation with the testing process, have been invoked to explain the poor performance of the geriatric schizophrenic patients on cognitive testing. However, in our study of geriatric schizophrenic patients and in most other studies of schizophrenics of all ages, severity of delusions and hallucinations did not correlate with severity of cognitive impairment. Furthermore, investigations have demonstrated that the cognitive impairment in schizophrenia is not a result of poor cooperation but is, instead, a result of the inability to perform the task (29).

The severity of cognitive impairment was also found to correlate with the number of years of formal education in our sample of geriatric schizophrenics. Similar findings have emerged for younger schizophrenics (31) and for normal geriatric individuals (74). In the schizophrenic patients, these associations may reflect a shared biological substrate for (a) poor educational achievement at younger age and (b) an enhanced risk for cognitive impairment in old age. However, confounding factors such as cohort effects or anxiety with test taking in the less educated individuals cannot be ruled out.

In summary, of all schizophrenic symptoms and factors associated with them, negative symptoms seem to show the strongest association with cognitive impairment, an association which remains constant throughout all age ranges. The immutable association throughout life between cognitive impairment and negative symptoms could be the phenomenological signature of a biological lesion which mediates both cognitive impairment and negative symptoms.

The generation of schizophrenic patients who are currently in their seventh or eighth decade of life have been exposed to leukotomy, insulin coma, electroconvulsive therapy (ECT), and at least 30 years of treatment with neuroleptic and anticholinergic drugs. Furthermore, many of them are still being treated with neuroleptics and few with anticholinergic drugs. Depending on the specific aspect of cognitive functioning tested, somatic treatments have been found to have therapeutic, deleterious, or negligible effects on the cognitive performance of geriatric schizophrenic patients (36, 69; see also Anatomic and Functional Brain Imaging in Alzheimer's Disease).

Our own review of the treatments given to geriatric schizophrenic patients showed no significant effects on current MMSE scores of past treatment with leukotomy, insulin coma, or ECT or of cumulative lifetime exposure to neuroleptics. At the time of the assessment, the majority of the geriatric schizophrenic patients were still treated with neuroleptics, and their mean MMSE scores were higher (better) than the scores of the patients not currently treated with neuroleptics. However, there were no differences in the MMSE scores associated with a particular class of neuroleptics despite the fact that they differed in their anticholinergic potencies or other pharmacological properties (atypical neuroleptics were not included in the comparison because none of the geriatric patients were receiving this treatment).

These data, consistent with previous reports, suggested that past or current somatic treatment (including treatment with anticholinergic drugs) has no obvious deleterious effects on cognitive functioning (42, 68). Despite the fact that patients currently being treated with neuroleptics had better MMSE scores than the nontreated patients, and that neuroleptic medication has been demonstrated to improve attention (68), better cognitive functioning cannot easily be attributed to current neuroleptic treatment. Until a prospective study of administration or discontinuation of typical neuroleptics to geriatric schizophrenics is conducted, typical neuroleptics cannot be unequivocally "absolved" of deleterious effects nor can cognitive enhancing effects be attributed to this class of drugs.

In summary, studies suggest that chronic treatment with neuroleptics or other somatic treatments are not responsible for the cognitive impairment in geriatric schizophrenics. However, while most of the studies compared different degrees of exposure to somatic treatments, they did not compare patients exposed and not exposed to the treatments.

Presence of concomitant neurological disorders (such as seizures) or a history of severe head trauma has been suggested to contribute to the schizophrenic cognitive impairment in young and geriatric patients (30). These disorders, along with medical disorders associated with increased risk for vascular dementia such as hypertension or diabetes, have constituted exclusion criteria from samples of schizophrenic patients in whom cognitive performance was evaluated in some (27), but not all, studies (see ref. 36 for a review). Furthermore, although each of these neurological and medical conditions have been invoked as a risk factor for dementia, their actual contribution to cognitive impairment is far from established. On the other hand, if presence of a risk factor was used to account for poor cognitive performance (i.e., patients who had poor cognitive performance and diabetes or hypertension were excluded from studies assessing cognitive performance), then the true prevalence of cognitive impairment associated with schizophrenia could have been underestimated because these variables' contribution in an individual patient's cognitive impairment cannot be ascertained.

Other neurological disorders such as Alzheimer's disease, multi-infarct dementia, and Parkinson's dementia could account for the cognitive impairment in schizophrenics. Because the disorders are more prevalent in geriatric individuals in general, they could enhance the differences in cognitive performance between young and geriatric schizophrenic patients. However, our own clinicopathological studies (see following sections) and those of others indicate that degenerative or vascular disorders known to cause dementia in nonschizophrenic individuals are not responsible for the cognitive impairment in geriatric schizophrenics.

Tardive dyskinesia (TD), and particularly orofacial TD, was found to be associated with cognitive impairment in schizophrenic patients in 24 of 37 studies reviewed (56). However, because establishing the presence, severity, and characteristics of schizophrenia, cognitive impairment, and TD is not an unambiguous task, the controversy surrounding the nature of the association among the three is not surprising (see Psychotropic Drug Metabolism in Old Age: Principals and Problems of Assessment). Elucidating this association is further complicated by the fact that TD and cognitive impairment share common risk factors such as advanced age, and maybe exposure to neuroleptic treatment (45, 56). Although some animal studies and observations in humans have led to the speculation that a striatal abnormality may mediate both the cognitive impairment and the limbotruncal TD (45), there is no support for a causal relationship between cognitive impairment and TD (42).

Finally, abuse of illicit drugs could theoretically contribute to the schizophrenic cognitive impairment; however, no direct evidence for this assertion exists. Furthermore, even if illicit drug abuse contributed to cognitive impairment, it could not be responsible for cognitive impairment in geriatric schizophrenics because illicit drug abuse is not common in this generation of patients.

In summary, similar to nonschizophrenic patients, cognitive impairment in schizophrenics could be associated with medical and neurological disorders; however, there is no evidence that the prevalence of medical or neurological disorders which induce cognitive impairment or dementias is increased in schizophrenics, nor is there evidence that these disorders can account for the high prevalence of cognitive impairment among geriatric institutionalized schizophrenic patients (also see following sections).

Available anti-schizophrenia drugs target positive and some negative symptoms, and occasionally agitation. There is, however, no specific pharmacological strategy for the treatment of the cognitive impairment in schizophrenia. The realization that cognitive impairment is a major contributor to the social disability in schizophrenia (8) has prompted investigators to evaluate what effects, if any, neuroleptics have on cognitive performance, and, more importantly, to include cognitive testing in efficacy trials of new anti-schizophrenia drugs. For example, despite its marked anticholinergic activity, clozapine was shown to benefit cognitive performance (49). Clozapine has been shown to release dopamine (DA) by in vivo microdialysis studies (18), and DA deficiency in the frontal cortex has produced cognitive impairment in primates (64). Thus, clozapine might benefit cognitive performance by enhancing DA neurotransmission. However, a second study which examined the effects of clozapine treatment on cognition reported no improvement in performance of chronic schizophrenic patients (28). Furthermore, because of clozapine's propensity to produce blood dyscrasia and occasionally hypotension and tachycardia, its large-scale use in geriatric schizophrenics is unlikely; therefore, other options to improve cognitive performance should be considered.

Because preliminary data regarding the pathophysiology of the cognitive impairment in schizophrenia are barely available, it is difficult to design rational therapies for this condition. In Alzheimer's disease, for example, finding a cholinergic deficiency led to the idea that cholinomimetic drugs might ameliorate Alzheimer's symptoms (see Neurppsychological Assessment of Paitents with Alzheimer's Disease). It would not be implausible, however, to hypothesize that despite the fact that cognitively impaired schizophrenic patients do not show cortical cholinergic deficiencies (34) (also see following sections), cholinomimetic drugs might, nevertheless, benefit these patients. Cholinomimetics have been shown to improve learning behavior and attention in cholinergically intact rodents, and they have been shown to improve attention in normal volunteers (52). Furthermore, schizophrenic patients have an auditory gating deficit which in experimental animals is reversed by a nicotinic agonist (62). Moreover, in a brain region responsible for vigilance and attention, the levels of choline acetyltransferase were found to be decreased in schizophrenic patients (41). Together these data lead to the hypothesis that increasing cholinergic neurotransmission and maybe dopaminergic and noradrenergic neurotransmission (26) may improve attention, and hence cognitive functions, in schizophrenic patients.

Alzheimer's disease (AD) is among the best characterized types of dementia in terms of specific areas of neuropsychological impairments and neurohistological abnormalities. By contrasting the cognitive profile of geriatric schizophrenics to that of AD patients (22), it is hoped that insight into the neural basis of dementing illnesses will be achieved. An overlap of specific cognitive deficits would suggest that clinically similar types of dementias could be associated with different neurohistological and neurochemical abnormalities. Partial dissimilarities in neuropsychological performance, together with the neurohistological investigation (see below), would support the view that these conditions involve different neural substrates.

In a comparison between the neuropsychological performance of geriatric schizophrenics and AD patients matched on level of global cognitive impairment, age, sex, and Z score education, we found that the schizophrenic patients performed worse than the AD patients on the Boston Naming test and the Praxis test. In contrast, the schizophrenic patients performed better than the AD patients on Delayed Recall, retaining 38% of the information presented relative to 23% retained by the AD patients. On the learning task, both groups performed poorly but not differently. This pattern of performance was present at all levels of global dementia severity. In spite of very close matching on measures known to affect psychometric performance (i.e., global dementia, severity, MMSE scores, age, and education), the schizophrenics and the AD patients could still be discriminated from one another on the basis of their neuropsychological test profiles. This pattern of double dissociation of neuropsychological performance indicated that despite equivalent levels of global dementia, schizophrenic and AD patients presented divergent profiles of cognitive dysfunction consistent with the notion that each dementia is associated with different neural abnormalities which in turn affect specific cognitive functions. It could be hypothesized that the cognitive impairments of geriatric schizophrenic patients are related to abnormalities of a mesial frontal-temporal neural circuit, whereas the impairments of AD are related to a disturbed temporal-parietal circuit, linking structures in the mesial temporal lobe (hippocampus, entorhinal cortex) to associational cortices in temporal and parietal lobes (see also Pharmacotherapy of Panic Disorders and Amyloidogenisis in Alzeihmer's Disease in Animal Models).

The most widely accepted neurochemical and neuropathological lesions believed to be associated with dementia in the elderly are those attributable to AD. The best correlates of dementia in AD have been deficits in cortical cholinergic markers, lesions of the basal forebrain cholinergic system, cortical neuritic plaques and neurofibrillary tangles, and deficits in cortical synaptic vesicular marker proteins, such as synaptophysin. In general (but note exceptions below), when these variables have been studied in schizophrenia they have not been found to be severely abnormal or to account for the cognitive impairment in those patients. For example, Burton et al. (16) examined the brains of 48 schizophrenic cases ranging in age between 22 and 99 years of age and found no significant differences from age-matched controls in neuritic plaques and neurofibrillary tangles. In a similar study, the brains of 12 schizophrenic cases (mean age 77, range 54-100) with evidence of cognitive impairment were compared to the brains of age- and sex-matched AD cases (60). No evidence of AD-like neuropathology in the schizophrenic group was observed in this study, nor in another study which extended these findings and demonstrated a lack of significant AD-like neuropathology even when severely cognitively impaired schizophrenics were contrasted to cases of non-cognitively impaired schizophrenics (6).

Sixty-two schizophrenic cases were compared to 116 AD cases using the CERAD neuropathology diagnostic battery. Brain specimens from each case were assessed for over 73 separate neuropathological categories, including neuritic plaques, neuritic plaques with amyloid cores, neurofibrillary tangles, neuronal cell loss, and gliosis in the cortex (midfrontal gyrus, middle and superior temporal gyrus, inferior parietal gyrus, entorhinal cortex) and in the hippocampus, amygdala, and nucleus basalis. Both cognitively impaired and cognitively nonimpaired schizophrenic groups differed significantly from the AD group on every neuropathological category assessed. Importantly, the significant differences between schizophrenics and AD cases were maintained even when 34 schizophrenic and AD pairs were matched on the basis of the severity of their cognitive impairment. Cognitively impaired and nonimpaired schizophrenic groups did not differ in any of the neuropathological categories assessed--with the exception of neurofibrillary tangles in the entorhinal cortex, which showed higher densities in the cognitively impaired patients. These results indicate that cognitive impairment in geriatric schizophrenic patients cannot be explained by an association with AD-like neuropathological variables such as the density of neuritic plaques or the density of neurofibrillary tangles in cortical and subcortical structures.

In sharp contrast to these conclusions are the results of a large study reported by Prohovnik et al. (59). The prevalence of AD-like neuropathology was found to be significantly increased in patients dying at state institutions with a chart diagnosis of schizophrenia relative to expected rates of AD in the general population. As pointed out by the authors, the diagnoses of schizophrenia obtained from clinical reports after death may not have been accurate in all 1046 cases reviewed. Furthermore, a selection bias for the demented cases remaining institutionalized may have been operating in this population. Moreover, the neuropathology criteria applied to diagnose AD differed from the CERAD diagnostic criteria. Nevertheless, the marked increase in the prevalence of AD-like neuropathology in schizophrenia reported in this study cannot be easily dismissed. The use of prospectively diagnosed cases of schizophrenia--and, more importantly, the use of modern diagnostic criteria for AD neuropathology (50) and of antibodies which recognize AD brain pathology, such as ALZ-50--may help clarify this apparent discrepancy.

In fact, immunoreactivity with the AD-related protein antibody, ALZ-50, has been studied in a small sample of schizophrenics. Cortical specimens from geriatric cognitively impaired schizophrenic cases were compared to specimens derived from geriatric controls and AD cases (58). Despite significant cognitive impairment, the ALZ-50 immunoreactivity of the cortical specimens derived from the schizophrenic cases was significantly different from the AD cases and indistinguishable from that of normal controls. There was no overlap of values between the AD group and the schizophrenic group. Similar results have been reported when ALZ-50 immunoreactivity has been examined in the nucleus accumbens, caudate nucleus, amygdala, temporal cortex, and cingulate cortex (39).

Cortical cholinergic marker deficits and the loss of cholinergic neurons in the basal forebrain are also characteristic of AD and have been shown to be causally related to compromised cognitive functioning in animals (35). Recently, cholinergic marker activity in six different cortical regions derived from geriatric controls, chronically institutionalized geriatric schizophrenic patients, and AD patients were compared (34). Cholinergic marker activity (choline acetyltransferase and acetylcholinesterase) was significantly diminished relative to controls in the AD cohort but not in the schizophrenic cohort. Additionally, cortical choline acetyltransferase activity was significantly and negatively correlated with the severity of cognitive impairment in the AD cohort, whereas no such correlations were evident in the schizophrenic cohort, suggesting that cognitive impairment in geriatric schizophrenics is not due to diminished cortical cholinergic activity.

Additionally, cognitively impaired geriatric schizophrenics did not differ from geriatric controls with respect to the number of large neurons (>30 mm) in the nucleus basalis of Meynert (23), suggesting that the basal forebrain neurons providing cholinergic afferents to the cortex are not affected in schizophrenia. Support for these findings was recently provided by another neuroanatomical study (38) which quantitatively assessed the density of acetylcholinesterase-positive fibers in five different cortical areas in schizophrenic and schizoaffective patients. A normal, age-appropriate pattern of cortical acetylcholinesterase staining was found in the study cohort.

On the other hand, some evidence in favor of brain cholinergic system involvement in schizophrenia does exist. The numbers of cholinergic interneurons in the striatum have been reported to be diminished in schizophrenic cases (38). In some studies, choline acetyltransferase activity is reported to be diminished in the nucleus accumbens and hippocampus by approximately 20% relative to controls, still a substantially smaller deficit than that which is normally noted in AD cases. The levels of choline acetyltransferase are also reported to be significantly reduced in the pedunculopontine nucleus of schizophrenic patients (41), but the relevance of this finding to dementia is not entirely clear.

In AD patients, decreased cortical immunoreactivity with antibodies against synaptic vesicular marker proteins such as synaptophysin has been found and interpreted as evidence consistent with synaptic loss (72). Synaptophysin-like immunoreactivity was found to be selectively decreased in the prefrontal cortex of schizophrenics in one study (25), whereas in another study (39) using an enzyme-linked immunosorbent assay (ELISA) technique and a different antibody for synaptophysin (EP10), the levels of synaptophysin-like immunoreactivity were found to be increased in the nucleus accumbens and temporal cortex of schizophrenics. Differences in technique (ELISA using tissue homogenates versus photodensitometry and immunocytochemistry), brain regions, age of subjects, and small sample sizes may well account for these disparate results. These observations, especially the correlations between synaptophysin-like immunoreactivity and severity of dementia in AD (72), argue in favor of a more intensive study of synaptic marker proteins in multiple brain regions of well-characterized schizophrenic populations with and without evidence of cognitive impairment.

In summary, the majority of the evidence suggests that the dementia-like cognitive impairment of geriatric schizophrenics is not related to (a) the neuropathological features associated with dementia in AD (e.g., neuritic plaques, neurofibrillary tangles, neuronal loss in the basal forebrain), (b) diminished cortical cholinergic marker activity, or (c) increased immunoreactivity with antibodies to the AD-related protein ALZ-50. The most parsimonious conclusion which can be derived from the studies reviewed above is that the biological substrates of dementia-like cognitive impairment in schizophrenia are not similar to those of AD. This conclusion is consistent with clinical studies demonstrating (a) differences in specific areas of cognitive impairment between the two diseases and (b) the marked differences in rates of cognitive deterioration (see previous sections). The failure of the aforementioned class of central nervous system (CNS) lesions to account for the severe cognitive impairment of schizophrenia serves to direct research toward other variables which may more adequately explain the cognitive impairment of geriatric schizophrenic patients such as catecholamines, synaptophysin, and other markers of neuroanatomical lesions.

There is considerable cumulative evidence from the past century of postmortem research to suggest that there is significant, albeit inconsistent, neuropathology associated with the antemortem diagnosis of schizophrenia. This evidence could be interpreted to suggest that in schizophrenia there is an increased susceptibility to a wide range of neuropathological lesions, and that many of these lesions could lead to a general syndrome of dementia-like cognitive impairment in old age. A related "threshold" hypothesis would suggest that the presence of several concomitant lesions is necessary to produce the dementia-like clinical picture. These hypotheses, for which there is some support (see below), while partially accounting for the inconsistencies reported in neuropathological studies of schizophrenia, are unsatisfying because they fail to attribute the schizophrenic trait and the cognitive impairment associated with schizophrenia to any specific cause.

A number of findings of structural abnormalities appear to be replicable, but the specific nature of the abnormalities and of the precise anatomical locations differ from study to study (70). Other less vague abnormalities, including changes in cortical (cingulate and prefrontal) and hippocampal cytoarchitecture (9, 10, 21), await replication by independent laboratories using independent autopsy samples. However, these cytoarchitectural abnormalities are relevant because they elegantly demonstrate alterations in the fine structure of the frontal and cingulate cortices. Although the centrality of these cytoarchitectural abnormalities to the cognitive impairment of schizophrenia has not been determined, they do occur in (a) cortical regions (prefrontal cortex and cingulate cortex) known to be involved in the processes which subserve learning and memory in nonhuman primates (64) and (b) regions implicated by functional imaging studies of schizophrenia (76). As such, these cytoarchitectural abnormalities must be considered to be potential contributors to the cognitive deficits of geriatric schizophrenics.

The evidence suggests that reductions in the size or volume of the brain or certain cortical or subcortical regions and increases in ventricular volume are among the most readily replicable findings in the neuropathology of schizophrenia (37), but their relevance to the specific cognitive impairment is unclear. Typical of this class of findings are data reported from a cohort of 41 schizophrenic cases (15) showing significant ventricular enlargements, diminished overall brain weight, and thinner parahippocampal cortices relative to specimens derived from patients bearing diagnoses of affective disorder. Similar results, especially with respect to ventricular volume, have been reported when schizophrenic cases have been compared to age- and sex-matched controls without histopathological evidence of CNS disease (16, 55). It is also interesting to note that when the study samples have included a relatively large age range, significant correlations between age of death and the severity of the neuropathologic lesions have not been reported (15). These findings are clearly supported by (a) the wealth of imaging studies of schizophrenic patients who often show clear evidence of enlarged ventricles and (b) some studies that report evidence of diminished cortical mantle width and length. Whether these nonspecific structural abnormalities contribute to the development of cognitive impairment in geriatric schizophrenics must still be determined.

There is little question that innumerable neurochemical abnormalities have been identified in the schizophrenic brain, and that at least some of these abnormalities, such as those associated with the dopaminergic system and some of the peptidergic systems, have been replicated in a number of different studies (11, 54). Like the neuropathological and cytoarchitectural studies mentioned above, however, neurochemical investigations have not focused on age-related changes in schizophrenia and, with a few exceptions, have not related their findings to the cognitive impairment of geriatric schizophrenics. Once again, most specimens studied neurochemically have been derived from geriatric cases, a significant proportion of whom would be expected to suffer from severe cognitive impairment. Thus, it is possible that some of the neurochemical deficits noted in these patients and attributed to the schizophrenic illness provide, instead, a biological basis for cognitive impairment.

Different from the cholinergic system of the forebrain, which is not severely affected in geriatric schizophrenics, there is reason to believe that the catecholaminergic system of the forebrain, especially of the prefrontal and cingulate cortices, is affected and may contribute to the cognitive impairment of geriatric schizophrenics. The normal functioning of the mesocortical dopaminergic system has been found to be essential for normal cognitive functioning in nonhuman primates, and to be severely affected by the aging process (64). Given the hypothesized involvement of the mesocortical dopaminergic system in schizophrenia (11; see also Physiological Indicators of the Schizophrenia Phenotype and Schizophrenia and Glutamate), it is possible that poor cognitive functioning is among the consequences of the dopaminergic or other catecholaminergic deficits in schizophrenia. At least one study (12) reports deficits of norepinephrine in the hypothalamus and nucleus accumbens in cognitively impaired schizophrenics. Furthermore, nucleus accumbens DA and 3-methoxy-4-hydroxyphenylglyde (MHPG) were found to relate inversely to the degree of cognitive impairments, supporting a role for catecholaminergic modulation of cognitive functioning in geriatric schizophrenics. Unfortunately, this intriguing finding has not been confirmed or disconfirmed by follow-up independent studies.

Abnormalities in peptidergic systems have also been observed in schizophrenic patients (43), but their association with the cognitive impairment has not been addressed. In contrast, the involvement of peptidergic systems in dementing illnesses such as AD is not questioned (24). In a recent study, we investigated the concentrations of five neuropeptides [somatostatin (SLI), cholecystokinin (CCK), vasoactive intestinal polypeptide (VIP), corticotropin-releasing hormone (CRH), and neuropeptide Y (NPY)] in six different cortical regions representing the frontal, temporal, parietal, and occipital lobes was investigated. The specimens were derived from a cohort of cognitively impaired schizophrenics. The concentrations of these peptides in the schizophrenic group were compared to the concentration of the same neuropeptides in the same brain regions of normal geriatric controls and AD cases. Relative to geriatric controls, the concentrations of SLI, NPY, VIP, and CCK were significantly reduced in the schizophrenic cohort. Furthermore, a discriminant function analysis using NPY and SLI concentrations in the frontal and temporal lobes correctly identified 100% of the schizophrenic patients. Of particular relevance to the current discussion were the similarities and differences in the pattern of neuropeptide deficits in the cortices of the cognitively impaired schizophrenic group relative to patients suffering from AD dementia. First, although significant somatostatinergic deficits were present in the schizophrenic and AD cohorts, the SLI deficits were significantly more pronounced in the temporal cortex of the schizophrenic group than in the same cortical regions of the AD cohort, underscoring the role of the temporal lobe abnormalities in schizophrenia (7, 69). In addition, significant deficits were observed in the concentrations of VIP and CCK in the cortices of the schizophrenic cases, whereas deficits in these peptides were absent in the AD cohort. On the other hand, the concentrations of CRH were reduced in the cortices of the AD cases, but not in the cortices of the schizophrenics, indicating that the reported findings in schizophrenia were not reflective of generalized peptidergic deficits, but of deficits specific to schizophrenia. The profound reductions in the concentrations of SLI in the cortices of the schizophrenic group may be of relevance to the cognitive impairment observed in these cases because in AD the cortical concentrations of SLI have been found to correlate with the degree of cognitive impairment (4).

These results demonstrate that the pattern of neuropeptide deficits observed in cognitively impaired schizophrenics is markedly different from that in normal controls and from the pattern of neuropeptide deficits observed in AD, but that reduced SLI concentrations could be a marker for cognitive impairment shared by the two diseases. The findings further demonstrate that it is possible to isolate deficits in neurochemical variables which are relatively specific to schizophrenia and to begin to distinguish between those neurochemical deficits which can be directly attributed to known neurodegenerative processes and those which may be nondegenerative and static in nature (see The Neurobiology of Neurotensin and The role of Acetylcholine Mechanisms in Mood Disorders for related issues).

A speculative but intriguing aspect of the neuropeptide deficits observed in the schizophrenic cohort is that the decrements in the cortical concentrations of NPY and SLI in the schizophrenic cohort may be manifestations of the potential neurodevelopmental lesions hypothesized in schizophrenia (see The Neurobiology of Treatment-Resistant Mood Disorders). NPY and SLI have been shown to be colocalized in human cortical neurons exhibiting nicotinamide-adenine dinucleotide phosphate (NADPH)-diaphorase activity (44, 75). Recent evidence suggests that neurodevelopmental abnormalities may lead to the aberrant dislocation of NADPH-positive neurons to the cortical white matter in schizophrenics (2, 3!popup(ch136). It is possible that the NPY and SLI deficits noted above are manifestations of the same neurodevelopmental "lesions" which cause the dislocation of NADPH-positive neurons in schizophrenia.

In summary, because very few studies have directly addressed the question of the neurochemical correlates of the cognitive impairment of schizophrenia, it is difficult to associate the cognitive impairments of geriatric schizophrenics with any specific neurochemical abnormalities. The available results do suggest, however, that abnormalities in neuropeptide-containing neurons (such as those associated with SLI and NPY) and possible alterations in cortical catecholaminergic functioning may contribute to the cognitive impairment.

An emerging awareness of factors which contribute to variability in postmortem studies, combined with increasingly sophisticated neurochemical and neuropathoanatomical techniques, promises a bright future in the next generation of studies. Critically important to the potential for future research is (a) the accumulation of normative neurochemical and neuropathoanatomical data from prospectively studied cognitively intact populations and (b) prospective atheoretical characterization of the symptoms and histories of the cases used in the neurochemical and neuropathoanatomical studies of schizophrenia. The historical characterization of the cases studied is important because the principal goal of research in the years to come must be the elucidation and discrimination of the neurochemical and neuropathoanatomical substrates of the schizophrenic trait, independent of the prevailing state of the patients shortly prior to, or at the time of, death. There is every reason to believe that studies currently underway using brain specimens derived from schizophrenic patients with varying degrees of cognitive impairment will help us to address the question of the neurochemical and neuroanatomical substrates of cognitive impairment in geriatric schizophrenics.

Investigations reviewed in this chapter indicate that a subtype of the schizophrenic illness exists in which severe cognitive impairment in old age is a predominant and incapacitating symptom. Indirect evidence suggests that this subtype of illness is characterized by (a) moderate cognitive impairment and persistent negative symptoms early in the course of the illness, (b) age-associated worsening of cognitive impairment with persistent negative symptoms for most of the lifespan, and (c) dementia-like cognitive impairment in old age. Illness severity and chronicity in general, and cognitive impairment and negative symptoms in particular, seem to be the cause, not the result, of lengthy or uninterrupted hospitalization and of intensive somatic treatment.

Without a 60-year catamnestic follow-up study it is not possible to estimate what proportion of all individuals who in late adolescence or early adulthood receive a diagnosis of schizophrenia suffer in senescence from cognitive impairment which is severe enough to distinguish it from normal aging and to significantly affect social functioning. Because this is impractical, the only other solution to address this question is to conduct an epidemiological study of the geriatric schizophrenic patients who reside in nursing homes. The study, which would verify if indeed the estimate that the majority (200,000) of all (300,000) geriatric schizophrenic patients in the United States reside in nursing homes is accurate, would determine what proportion of these patients suffer from cognitive impairment and whether the outcome of the geriatric schizophrenic patients who have remained in long-term psychiatric hospitals is not representative of the outcome of schizophrenia in general. Investigating the proportion of cognitively impaired geriatric schizophrenic patients would help elucidate the role of cognitive impairment in determining the long-term outcome of schizophrenia in general and the need to live in protected environments such as nursing homes in particular.

When differences between age groups are not very large, this observation in conjunction with the postmortem data are not supportive of a rapid degenerative process occurring across the lifespan or during the latter decades of life of the schizophrenic patients. A number of studies have reported a slow decrease in some CNS structural and functional parameters with advancing age in "cognitively intact," normal populations (17, 73). Of particular interest to schizophrenia are (a) age-related changes reported in normals in neuronal numbers of the dopaminergic and noradrenergic systems and (b) reports that aging primates are hyporeactive to pharmacological dopaminergic challenges (17, 19; Arnsten, 1994, personal communication). Age-associated reduction in DA activity superimposed on static dopaminergic abnormalities in areas relevant to cognitive functioning could produce the appearance of slow progression of cognitive impairment and alteration of symptom clusters in schizophrenic patients. This superimposition is probably most apparent and damaging in the most severely ill schizophrenic patients with the greatest preexisting cognitive impairment and possibly dopaminergic hypoactivity. Thus, the extent to which current neuropathological studies contribute to the debate concerning the static versus degenerative nature of the disease process is to suggest that clinical evidence of age-related differences in cognitive functioning in a subgroup of patients with early, preexisting cognitive impairment is not sufficient to warrant a conclusion of progressive neuropathology. The data are rather consistent with an interaction between preexisting lesion(s) and aging in the context of chronic illness. The available data suggest that it is unlikely that the cognitive impairment in schizophrenia can be attributed to neuropathological conditions with known degenerative etiology. On the other hand, it must be kept in mind that the numbers of studies directly addressing this question are so few that this conclusion is tenuous, and far from one which comes after having taken into account all possible alternatives and problems associated with testing the null hypothesis.

This work was supported by the following: National Institute of Mental Health (NIMH) grant 46436-05, awarded to Michael Davidson, M.D.; NIMH grant 45212-02, awarded to Kenneth L. Davis, M.D.; and a Veterans Affairs Merit Review grant, 5118-011, awarded to Vahram Haroutunian, Ph.D., and Michael Davidson, M.D. The authors wish to acknowledge Paul Hartel for his assistance in the preparation of this manuscript.

published 2000