Cognitive Impairment in Bipolar Disorder and Schizophrenia: A Systematic Review

Aims: Previous comparisons of cognitive decline among patients with bipolar disorder (BD) and schizophrenia (SZ) have found somehow quite similar profiles of deficits, but results have varied between studies. Therefore an extensive and thoughtful systematic review of the matter is warranted. Methods: Studies were found through systematic search (PubMed) following PRISMA guidelines. To be included, studies must have assessed the following cognitive functions: executive functions, memory, IQ, attention-concentration, and perceptuomotor function. In order to make comparison between the two entities, studies should include BD patients with operationally defined euthymia, schizophrenic patients in remission, and third group of healthy control patients. Comparisons were made after controlling for years of schooling and residual affective symptoms. Results: We found that overall both SZ and BD patients present deficits on all neurocognitive measures compared to healthy controls. In particular, SZ patients show more severe and pervasive cognitive deficits while BD patients present a milder and more confined impairment. In addition, evidence from the literature suggests that SZ and BD patients share a similar cognitive impairment profile with different degrees of deficits. Therefore, the difference between the two groups seems to be more quantitative (degree of deficit) rather than qualitative (profile), supporting a dimensional approach to the two clinical entities. Limitations of the present review includes the impossibility to control for effects of medication, varying time required for assessment across studies, illness diagnosis reliability, and course severity. Conclusion: Patients with BD might exhibit a cognitive impairment that could be similar to SZ in terms of their profile, although patients with SZ may have more severe and widespread impairments.


INTRODUCTION
In his landmark paper "The Diagnosis and Prognosis of Dementia Praecox" presented in Heidelberg in 1898 (1), German psychiatrist Emil Kraepelin discussed the results of his methodical observation of patients by stating that in dementia praecox "The prognosis, however, is really by no means simple. Whether dementia praecox is susceptible of a complete and permanent recovery . . . is still very doubtful, if not impossible. But improvements are not at all unusual . . .. It is a more serious matter that in most of these cases the improvement is only temporary, and that such patients are in great danger of relapsing sooner or later, without any particular cause, and then generally suffer more serious injury from their illness." He distinguished dementia praecox from an apparently less severe course illness that was reported in 13% of his patients. The idea that an illness with a course potentially severe as dementia praecox and a "spontaneous recovery" was difficult to accept in the nineteenth century Germany psychiatry. In addition Kraepelin saw Dementia Praecox as a "disease process in the brain, involving the cortical neurones [and brought about] by an autointoxication . . . as a result of a disorder of metabolism" (2).
Although the course of illness, as understood by Kraepelin, pointed to a differentiate Manic Depressive Insanity (current bipolar disorder -BD) from Dementia Praecox (current Schizophrenia -SZ) in terms of degree of cognitive deterioration (Dementia), research in the last decade has begun to shed some light into the long term course of BD. In addition, they have also reported that in BD different cognitive domains deteriorate in remarkable similar ways as SZ. In other words, although Kraepelin's great insight has led to the proper qualitative differentiation between these remarkably similar illnesses, recent prospective observational studies (3) have shown that BD and SZ share similar dimensional characteristics in terms of neurocognitive domains. Data from recent meta-analysis conducted in BD and SZ are in line with these findings. Depp et al. and Fett et al. reported a similar mean correlation between overall neurocognitive ability and everyday functioning (4,5). Mann-Wrobel et al., similarly though less severely than in SZ (6), reported a broadly generalized cognitive impairment in euthymic BD patients with verbal ability as a possible area of preserved performance (7). Neuropsychological deficits represent a core feature of BD and SZ; however, their onset and progression differ between diagnostic groups. (8). For instance, although cognitive deterioration can appear from the outset in both disorders, in BD, as compared to SZ, it follows a progressive longitudinal worsening related to the number and the intensity of mood episodes (9). This might reflect different brain structural and functional alterations related to BD exacerbations. Imaging studies have shown alterations in regions related to emotional and cognitive functioning. A meta-analysis by Bora et al showed gray matter reductions are more circumscribed to these areas when controlling by gender. Modest differences between both disorders were found, with SZ patients showing higher dorsomedial and dorsolateral prefrontal cortex affectation compared with BD (10). In functional studies frontal-subcortical structures show abnormalities in both disorders with common hyper-activation of emotional limbic regions in relation to prefrontal-cortical regions. In addition, BD has shown more pronounced chronic hypo-activation of ventral-medial regions (11). The end result of this apparent connectivity abnormality may result in a glutamatergic transmission excess in key memory and processing regions with concomitant excitotoxicity and neurocognitive sequelae in BD (12). Moreover, modern techniques in molecular biology and neuroanatomical immunocytochemistry have established that both disorders share neuronal pathological markers in brain areas as hippocampus that correlate with neurocognitive sequelae (13,14).
The aim of this systematic review is twofold: (i) describe and compare the neuropsychological dysfunction between BD and SZ patients as addressed by included studies; (ii) investigate the course of illness to better differentiate and predict onset and progression of these two disorders.

METHODS
To identify studies reporting cognitive impairment in BD and SZ we searched PubMed database for the period between 2008/01/31 and 2013/01/31, following PRISMA guideline recommendations (15). We combined the search strategy of free text terms and exploded MESH headings as following: [((SZ AND bipolar AND control) AND (neurocogniti * OR cognitive OR neuropsycholog * )) NOT (Gen * (MeSH Major Topic) OR Neuroimag * (MeSH Major Topic))]. Additional studies were identified by reviewing the references of each article found, method deemed appropriated by recent articles seeking similar objectives as ours (4,10). The search cutoff date was the last day of January 2013. Criteria for inclusion in this study were:(a) publication Full text available in English, (b) inclusion of a BD group with operationally defined euthymia, a schizophrenic group in remission and a healthy comparison group (studies that did not directly compare the three groups were excluded), (c) recruitment of adult population (18 years or older), (d) meta-analysis, controlled clinical trial, systematic reviews, comparative study, and randomized controlled trial, (e) assessment of the following cognitive functions: executive functions, memory, IQ or educational level, attention-concentration, and perceptuomotor function. Key words and application of inclusion/exclusion criteria resulted in 18 papers. Additionally we performed a second search using the algorithm ("premorbid" OR "pre-morbid" OR "at risk" OR "prodrom") AND (schizoph * AND bipolar) AND (cognition OR neurocogniti * OR neuropsycholog * ) AND [Meta-Analysis(ptyp) OR Randomized Controlled Trial(ptyp) OR Review(ptyp) AND "last 5 years" (PDat)]. This algorithm yielded to one additional review (8) with same aims as ours. Additional 22 records were identified trough bibliographic cross-referencing. After removing 1 duplicate study, 39 records were screened. Application of inclusion/exclusion criteria resulted in exclusion of 29 studies: 16 were out the search interval dates, 2 lacked either schizophrenic or euthymic bipolar group, and 1 lacked of a control group, 3 were focused on differential neuroimaging. 2 were focused on genetics, and 4 on the effect of treatment alternatives (CBT, Befriending, ECT, and Ziprasidone), 1 studied only bipolar patients and discussed comparatively with SZ. As shown in Table 1, key words and application of inclusion/exclusion criteria finally resulted in 10 papers.
Two reviewers extracted all data independently (Sergio A. Barroilhet and Paul A. Vöhringer). Discrepancies between reviewers were resolved through discussion until consensus was reached.

RESULTS
Findings from studies that analyzed the cognitive profile of patients with SZ and BD when compared with healthy controls (HC) are summarized in Table 2.
Reviewed studies found that overall both SZ and BD patients present deficits on almost all neurocognitive measures compared to HC, with few exceptions in some domains (16,17) for some sample comparisons (18, 19). In particular, SZ patients show more severe and pervasive cognitive deficits while BD patients present a milder and more confined impairment (20). In addition, evidence from the literature suggests that SZ and BD patients share a similar cognitive impairment profile with different degrees of deficits. Therefore, the difference between the two groups seems to be more quantitative (degree of deficit) rather than qualitative (profile), supporting a dimensional approach to the two clinical entities (21).
With regard to the course of the disease, authors report that in BD patient's cognitive deficits persist during periods of euthymia, which implies that cognitive deficits are not uniquely associated Frontiers in Psychiatry | Schizophrenia      (17,19,(22)(23)(24)(25).
In particular, a study conducted in Australia on 32 SZ and 28 BD patients explored the semantic retrieval (Generate condition) and the semantic recognition (Forced choice condition): SZ patients performed significantly poorer compared to HC on semantic retrieval (MRC Oxford Psycholinguistic Database Generate condition total score: 49.7 (±20.4) and 74.1 (±17.8), respectively, p < 0.001; after adjusting for premorbid IQ) (25). Similar findings were reported in other studies (26,27). On the contrary, BD patients did not present significant impairment. With regard to semantic recognition, both clinical groups presented similar performance with difficulties in identifying the correct word meaning in the presence of alternatives (MRC Oxford Psycholinguistic Database Forced choice condition total score: 72.7 (±16.0) and 75.1 (±17.0), respectively vs. 88.0 (±5.6) in HC, p < 0.001; after adjusting for premorbid IQ) (25).
Sanchez-Morla et al. examined visual memory on 88 SZ and 73 BD patients using Rey-Osterreith Complex Figure Test (ROCFT). They found significant differences in immediate and delayed visual memory among the three groups, with BD patients performing intermediately between SZ patients and HC (23).

IQ
Several studies analyzed premorbid IQ in SZ and BD patients using tests based on measures of current ability though to be relatively insensitive to brain dysfunction (17,23).

ATTENTION
Some authors reported that attention deficits in SZ and BD patients persisted during the entire course of the disease including during periods of euthymia (16,23,24,29,30).
Researchers described similar degree of sustained attention impairment in SZ and BD patients. In particular, a study conducted in Spain in 2009 measured -using the Continuous Performance Test (CPT) -lower number of hits and a longer reaction time in both SZ and BD patients as compared to HC (CPT: hits: 26.2 (±10.4), 30.7 (±8.7), and 34.6 (±6.2), respectively, p < 0.037; Furthermore, the reaction time was longer for SZ and BD compared with HC (ms): 550 (±94), 540 (±98), and 480 (±65), respectively, p < 0.0001; after adjusting for the effect of age, years of education, gender, subclinical depressive symptomatology, and premorbid IQ) (23). Others reported similar finding using different tests that required keeping attention focused for an extended amount of time (Repeatable Battery for the Assessment of Neuropsychological Status; RBANS, Subtest Attention, Digit Spam, and Coding) (24).
Based on results of the Stroop Test, Brissos et al. described increased slowness and perseveration in incongruent tasks in SZ patients as compared to BD patients (with or without history of psychotic symptoms) and HC and interpreted it as a proxy of deficient inhibitory control, strongly related with poor selective attention (20).

PERCEPTUOMOTOR FUNCTIONS
Simonsen et al. divided the BD study population in psychotic and non-psychotic subjects. Both categories reported lower perceptuomotor functions as compared to controls. No significant differences were shown between them. Other patients with a history of psychosis (SZ, schizoaffective disorder) performed poorly than HC on measures of processing speeding.

DISCUSSION
Overall, patients with BD exhibit neurocognitive impairment qualitatively similar to but quantitatively less pronounced than patients with SZ. In a general sense, it seems that while emotional and interpersonal difficulties in childhood reflect a general predisposition to adult psychiatric illness, early developmental impairments in psychomotor, language, and cognitive function show some specificity to SZ outcomes and are not seen in those who later develop BD. (31) A previous meta-analysis showed that SZ patients demonstrate worse cognitive performance than BD patients in most domains, with substantial heterogeneity. This study focused on performance differences between patients groups, without normal controls, thus deviations from the norm were not analyzed (32). The course of cognitive impairment also differs between groups: premorbid cognitive impairment is present early in life in SZ (often childhood/adolescence), preceding onset of psychosis, but cognitive impairment typically does not occur until later in adulthood in BD, after many years of mood episodes (8).
Despite overlap between BD and SZ in genetics and phenomenology of psychosis (31), these cognitive impairment data indicate that BD and SZ are different in severity and course of cognition.
During adulthood, BD patients have shown impairment on neuropsychological measures sensitive to frontal lobe dysfunction during the adolescence, usually before the diagnosis is made. This may suggest the presence of some kind of neurodevelopmental vulnerability, expressed initially as attentional deficits or comorbidity with attention deficit-hyperactivity disorder (33). Nevertheless, there is lack of systematic studies regarding time varying patterns upon developmental deficits, starting on pre-morbid stage, and continuing during the illness phases of both conditions. Regarding specific domains of cognitive impairment, in SZ working memory was impaired, while in BD attention was impaired (34) The usual clinical symptomatology showed by these illnesses can also modify cognitive functioning. For instance, severity of negative symptoms in SZ may have some correlation with poor neurocognitive performance, while on the contrary, severity of depressive symptoms in BD showed no association with performances on any neuropsychological test (28). A high level of negative symptomatology might be seen as predictor of worse response (17,19); there are prior data suggesting that negative symptoms had strong predictive capacity of neurocognitive outcomes associated with measures of executive functioning and selective attention, but not associated with visual or verbal learning and memory, verbal fluency, or processing speed (35). Further, changes in depressive symptoms were correlated with poorer verbal fluency, while no relationship between manic symptoms and neuropsychological performance was detected (36).
Evidence suggests that patients were usually taking psychiatric medication while cognitively assessed, eventually biasing the measure of differences between groups. This lack of clean assessment, impede drawing any definitive conclusion of the effects of medication on cognitive function. One study suggested the presence of the pharmacological treatment over cognitive functioning. (37) concluded that schizophrenic patients in polypharmacy had significantly lower cognitive functions than those in the second-generation antipsychotics monotherapy (37). Additionally, moderate improvement was found in cognitive test performance of patients on their first episode of SZ treated with second-generation antipsychotics (38). Similarly, untreated bipolar patients showed better cognitive performance than patients on atypical antipsychotics (39).
Interesting evidence on the effect of psychopharmacologic agents over cognitive function has been developed with lithium. Studies suggest that cognitive deficits are not explained by this medication, but by the condition itself (15,40). With respect to other drugs, cognitive side effects may occur as a direct consequence of the mechanism of action of the psychotropic agent, or indirectly through side effects such as sedation (41). Nevertheless, post-onset neuropsychological evaluation may be confounded by medication, making it difficult to comment about the relevance of neuropsychological deficits when talking about the etiology or development of these conditions. We think that further randomized research on drug naive patients should clarify the role of medication on cognitive deficits. Moreover, typically early models for the etiology of psychotic illnesses focused on dopamine neurotransmission because of the powerful antipsychotic action of dopamine antagonists giving support to the so-called dopamine hypothesis of psychosis. Nevertheless, recent evidence increasingly supports a primarily glutamatergic dysfunction in this condition, where dopaminergic disbalance is a secondary effect. (42) In fact, a more recent model for the pathophysiology of SZ and BD involves a pathophysiologic mechanism by which the cognitive decline is representing the core of the clinical morbidity and in this context impairment of glutamate and more specifically N -methyl-d-aspartate (NMDA) function has been considered of major importance. Given that astrocytes are crucial in controlling glutamate homeostasis, seems to logical to assign a major role to glial-neuronal interactions in the pathophysiology of the cognitive decline in these conditions (43) Few studies evaluated psychosocial function of patients. When assessed, verbal memory impairment seemed to correlate with a poorer psychosocial functionality. Furthermore, BD patients scored lower than SZ patients; we thought that depressive or sub threshold depressive symptoms might play a role on these results. Executive function had a directly positive effect on psychosocial adjustment in both patients with BD and SZ. There were positive correlations between insight and executive function among subjects with SZ and BD, suggesting that the ability to recognize their own experiences of illness seems to be the first step to developing an effective strategy for supplementing their deficiencies in psychosocial adjustment (44). The possibility of a shared general cognitive impairment might yet be possible as an explanation, due to comparable qualitative scores shown by the two conditions; nevertheless these abnormalities oftentimes are larger in SZ patients. Following the same vein, a study suggests the existence of a continuum between SZ and affective disorders but with substantial difference between the courses of their cognitive deficits, being more severe in SZ and milder in BD (28). On the other hand, a study contradicts this theory suggesting that BD patients usually present impaired performance on multiple cognitive domains, whereas their first-degree relatives shown performances comparable to HC on most cognitive tasks (3) additional research is warranted in order to reach a sound conclusion on this matter.
Evidence regarding the effect of past psychotics symptoms on the neurocognitive performance remains unclear. A study found similar performance in all BD patients; no differences were found depending on the history of the disease (16). On the contrary, there is evidence concluding that patients with a psychotic bipolar condition represent a group that is neurocognitive closer to the SZ than the non-psychotic bipolar subjects. This study was carried out with symptomatic patients and therefore this clinical condition may confound these results (19).
The overall homogeneity among these findings argues favorably for the validity of a milder neuropsychological impairment in BD compared with SZ and yet other considerations like the eventual cognitive continuum between BD and SZ, the role of psychotropic medication, the pre-morbid deficit and the possible effect of psychotic symptomatology over the neuropsychological status remains unclear.
Limitations of the present systematic review includes heterogeneity of samples and assessments, the impossibility to control for acute or chronic effects of medication, varying time required for assessment across different studies, illness diagnosis stability, and course severity. Another limitation to be considered pertains with the time limit that was set up for the systematic review. In effect, to gain novelty about the selected studies, search time was set up with year 2008 as the lower limit and important papers about glial research have appeared before the 2008, thus some relevant evidence regarding this pathophysiological mechanism may be left out of the overall search. www.frontiersin.org