In recent years, there has been a greater appreciation of the elevated prevalence of cardiovascular risk factors in the schizophrenia population and the liability some treatments have for their development. These cardiovascular risk factors, including diabetes mellitus, hypertension, dyslipidemia, and obesity, are also important risk factors in the development of dementia (1—3) as well as more subtle cognitive decrements (4). However, the impact they have on the cognitive functions of patients with schizophrenia remains underexplored. This area of investigation demands attention because identification of additional causes of cognitive impairment may lead to new developments in treatments, which might be aimed at vascular factors. Indeed, the results of studies of vascular risk factors in Alzheimer's disease have spawned a number of dementia treatment trials of medications targeting vascular risk factors (5, 6). Therefore, similar treatment targets may prove worthy of investigation in schizophrenia if there is sufficient evidence linking these factors to the cognitive impairment of schizophrenia.
The baseline data from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study (7) failed to show any relationship between the metabolic syndrome and cognitive impairment in schizophrenia. However, neither CATIE nor other investigations have focused on the association of individual vascular risk factors and cognitive impairment in schizophrenia. Indeed, the CATIE requirement that participants meet at least three of the criteria to be categorized as having metabolic syndrome (7) may have resulted in the inclusion of a substantial number of individuals with only one or two vascular risk factors in the group designated as not having metabolic syndrome. This may have diluted the measured effects of these individual risk factors on cognitive performance.
We compared the association between individual vascular risk factors and cognitive performance in a large sample of patients with schizophrenia and nonpsychiatric comparison subjects. We hypothesized that patients with schizophrenia and comparison subjects with each of these vascular risk factors would demonstrate poorer cognitive performance than their counterparts who did not have these risk factors. We also made an exploratory assessment of potential interaction effects between a diagnosis of schizophrenia and the presence of these risk factors.
Participants in this study were recruited for projects conducted by the Conte Center for the Neuroscience of Mental Disorders at the Mount Sinai School of Medicine (8). Schizophrenia patients were recruited from inpatient, outpatient, day treatment, and vocational rehabilitation services at Mount Sinai Hospital (New York), Pilgrim Psychiatric Center (West Brentwood, N.Y.), Bronx VA Medical Center (New York), Hudson Valley Veterans Affairs Medical Center (Montrose, N.Y.), and Queens Hospital Center (New York). The institutional review board of each institution approved the study. Comparison subjects were recruited from the Mount Sinai Hospital Internal Medicine Associates clinic, the surrounding Manhattan area, and Long Island communities surrounding Pilgrim Psychiatric Center. Comparison subjects underwent the same diagnostic, neuropsychological, and laboratory testing procedures as the schizophrenia patients. Signed informed consent was obtained from each participant in accordance with each institution's institutional review board policies. Schizophrenia patients were diagnosed on the basis of the Comprehensive Assessment of Symptoms and History (9). Comparison subjects had no DSM-IV axis I disorders, as determined by expert consensus of the data obtained from the Comprehensive Assessment of Symptoms and History. Schizophrenia patients and comparison subjects were excluded from the analyses if they had a positive urine test for drugs of abuse, a medical diagnosis that might include significant brain involvement (e.g., HIV infection, an episode of anoxia), a history of a neurologic disorder that might produce cognitive impairment (e.g., head injury, cerebrovascular disease, Parkinson's disease, Alzheimer's disease, a seizure disorder), an unstable medical condition (e.g., poorly controlled diabetes or hypertension, symptomatic coronary artery disease), or a reading grade equivalent of grade 8 or less based on the Wide-Range Achievement Test, 3rd ed. (10, 11).
All participants completed the tests in a fixed order. All tests were administered by raters who were trained and certified as reliable and were blind to participants' medical status at the time of assessment.
The severity of psychotic and negative symptoms was assessed with the Positive and Negative Syndrome Scale (PANSS) (12). The dependent variables in this study were the total scores on the positive, negative, and general psychopathology subscales of the PANSS.
Neuropsychological test battery.
We used the standard neuropsychological test battery compiled for all clinical studies in schizophrenia conducted by the Conte Center at the Mount Sinai School of Medicine (8). These tests, described below, were chosen to assess the wide range of cognitive domains known to be impaired in schizophrenia.
The Rey Auditory-Verbal Learning Test (13) was used to assess verbal learning and memory. This test utilizes a word list paradigm in which five separate verbal presentations of a 15-word list are given, each trial followed by a free recall test. Then an interference list of 15 words is presented followed by a free recall of the first list. After a delay of 20 minutes, participants are asked to provide free recall of the first list. They are then asked to identify the 15 words presented in the first list from a list of 50 words. The dependent variables included the total number of words recalled over trials 1—5 (immediate memory), the number of words recalled after the 20-minute delay (delayed recall), and a recognition memory score ([correct recognitions + correct rejections]/50).
The Trail Making Test, Parts A and B (14), were employed as tests of visuomotor speed and the ability to alternate between sets. To perform Part A, subjects draw lines to connect circles numbered 1—25 in ascending order. In Part B, the circles include both numbers (ranging from 1 to 13) and letters (from A to L); as in Part A, the participant draws lines to connect the circles in an ascending pattern, but with the added task of alternating between the numbers and letters (i.e., 1-A-2-B-3-C, etc.). The dependent measure for each task is the time (in seconds) taken to complete each of the trails, including the time used correcting the participants.
The category verbal fluency test (animals) (15) was administered to measure verbal productivity and the intactness of the lexical system. Participants were asked to produce the names of as many different animals as they could in 1 minute. The number of original words that are considered animals produced in that interval is the dependent measure.
The letter-number sequencing test, a subtest of the WAIS, 3rd ed. (WAIS-III) (16), was used as a measure of working memory performance. In each trial, a combination of numbers and letters is read to the participant (ranging from two to eight items), after which the participant is asked to recall the numbers first, in order, starting with the lowest number, and then the letters in alphabetical order. There are three trials at each level of difficulty, with two items, three items, four items, and so on.
The digit span distraction test (17) is a measure of attentional capacity and distractibility. The task has seven trials in both distraction and non-distraction conditions. The non-distraction trials include six target digits per trial, presented in a female voice at a rate of one item every 2 seconds. The distraction condition has seven five-digit trials with the target digits presented at the same rate in a female voice. In the distraction subtask, the 2-second interdigit interval is filled with a male voice saying four irrelevant digits. Each trial is scored for the number of items recalled correctly in order. Dependent measures were scores for the non-distraction and distraction condition.
Medical assessments were performed at the General Clinical Research Center at Mount Sinai Hospital. A research nurse blind to the cognitive performance data compiled an inventory of each participant's medical history and current medications and performed a review of systems. A physical examination was performed, along with an ECG and laboratory studies (comprehensive metabolic panel and lipid profile; CBC; thyroid-stimulating hormone, B12, and folate levels; urinalysis; and urine toxicology screen). Medical diagnoses were grouped and coded according to ICD-9-CM.
From the list of medical diagnoses, two cardiovascular risk factors were the subject of this investigation: hypertension and elevated body mass index (BMI). Because there were insufficient numbers of participants with diabetes and dyslipidemia in both the schizophrenia and comparison groups (Table 1), these risk factors were excluded from the analyses.
Comparison of Clinical and Demographic Characteristics of Patients With Schizophrenia and Comparison Subjects Age 29 Years and Older
| Add to My POL
|Characteristic||Schizophrenia Patients (N=100)||Comparison Subjects (N=53)||Analysis|
|Body mass index||29.94||7.61||29.26||5.31||−0.53||118||0.5960|
| African American||44||44||22||42|
| Overweight (body mass index ≥25)||54||54||35||66||0.0037||1||0.9517|
| Diabetes mellitus||21||21||2||4||7.9243||1||0.0049|
The designation of hypertension was based on a review of the medical history and, when the history was unclear, discussion with the participant's primary physician. All participants with hypertension included in the study were receiving antihypertensive treatment at the time of assessment, and their blood pressure was under control. Categorization of participants was not based on blood pressure reading at the time of assessment; participants with newly diagnosed and untreated hypertension or poorly controlled established hypertension were excluded from the study and referred for treatment. For the purposes of the analyses, participants were dichotomized into those with and those without hypertension.
We calculated BMI as weight in kilograms divided by the square of height in meters. A BMI ≥25 was categorized as elevated, consistent with the International Classification of Adult Underweight, Overweight, and Obesity. For the purposes of the analyses, participants were dichotomized into those with a normal BMI (<25) and those with an elevated BMI (≥25).
Scores from the Rey Auditory-Verbal Learning Test, the verbal fluency for animals test, and the letter-number sequencing test appeared normally distributed and were analyzed without transformation. The time to complete the Trail Making Tests were skewed and were analyzed after taking square-root transformations. The digit span distraction test results were also skewed, and the transformation most nearly normal appeared to be the square root of the number of incorrect answers.
For each item in the neuropsychological test battery, the individual vascular risk factors were cross-tabulated by group (schizophrenia patients and comparison subjects). We analyzed the resulting data using general linear models. We entered age, gender, education, and ethnicity in all models. In the main effects model, we also included the vascular risk factor(s) of interest (the presence of hypertension or elevated BMI) and the presence or absence of a schizophrenia diagnosis. The ratio of the estimated effect of each variable in Tables 2 and 3 divided by the square root of the mean squared error and a function of the sample sizes had a t distribution, and from this value, a p value was computed. Holm's method was used to adjust the p values for multiple comparisons. For the smallest p value, the adjustment is the same as the Bonferroni correction for the nine outcome measures being analyzed, resulting in a corrected significance level set to p=0.0056. If the smallest p value exceeds 0.0056, the process stops, but if it is smaller, the next smallest p value is multiplied by 8. The process continues in a similar manner if that p value is significant, with the next smallest multiplied by 7. Within-group comparisons were performed using a contrast statement in the linear model, which basically forms a t statistic, which is the ratio of the observed differences within the group, divided by the standard error of that difference. In the interaction model, we also included the interaction between the vascular risk factor and the schizophrenia status. Adjusted means for the neuropsychological variables were calculated from the main effects model for each of the four cells formed by schizophrenia status and vascular factor, where adjustment is for age, gender, education, and ethnicity.
Analyzable cognitive and medical data were available for 247 participants. However, given the complete absence of hypertension in participants younger than age 29, these participants were excluded from the analyses, leaving 153 participants age 29 or older (100 schizophrenia patients and 53 comparison subjects) for the analyses. Demographic and clinical details are summarized in Table 1. Schizophrenia patients demonstrated, on average, moderate symptom severity, as indicated by a mean PANSS positive score of 14.50 (SD=4.63) and a mean PANSS negative score of 16.54 (SD=6.04). Ten percent of the schizophrenia patients were receiving treatment with first-generation antipsychotics and 87% with second-generation antipsychotics; 3% were medication free at the time of assessment.
Neuropsychological Test Scores of Patients With Schizophrenia and Comparison Subjects by Group and Hypertension Diagnosisaa
| Add to My POL
|Measure||Schizophrenia Patients||Comparison Subjects||Uncorrected p||Hypertension Effect Size|
|With Hypertension (N=26)||Without Hypertension (N=74)||With Hypertension (N=21)||Without Hypertension (N=32)|
|Mean||SEM||Mean||SEM||Mean||SEM||Mean||SEM||Schizophrenia Effect||Hypertension Effect||Schizophrenia Patients||Comparison Subjects|
|Rey Auditory-Verbal Learning Test|
| Total learning subscore||33.35||2.17||37.62||1.49||43.30||2.39||50.72||1.94||<0.0001bb||0.0068bb||−0.35||−0.81|
| Delayed recall subscore||5.21||0.68||6.83||0.46||6.56||0.74||10.48||0.60||<0.0001bb||<0.0001bb||−0.44||−1.38|
| Recognition discrimination subscore||0.67||0.04||0.82||0.02||0.89||0.04||0.92||0.03||<0.0001bb||0.0041bb||−0.72||−0.37|
|Verbal fluency for animals, total score||14.57||1.13||16.35||0.78||19.36||1.24||20.00||1.01||0.0002bb||0.2163||−0.35||−0.11|
|Square root of the digit span distraction test, non-distraction subscore||3.36||0.29||3.31||0.20||2.34||0.31||2.05||0.25||<0.0001bb||0.5734||+0.04||+0.20|
|Square root of the digit span distraction test, distraction subscore||3.27||0.29||2.99||0.20||2.26||0.32||1.78||0.26||<0.0001bb||0.1775||+0.20||+0.36|
|Letter-number sequencing test, raw score||6.16||0.58||7.36||0.40||8.97||0.64||8.84||0.52||0.0004bb||0.2371||−0.39||+0.04|
|Square root of the Trail Making Test, Part A, time||8.37||0.46||7.63||0.31||6.41||0.50||6.12||0.41||0.0009bb||0.1959||+0.28||+0.21|
|Square root of the Trail Making Test, Part B, time||12.36||0.60||11.79||0.38||9.96||0.61||10.02||0.50||0.0003bb||0.5841||+0.18||−0.03|
Neuropsychological Test Scores for Patients With Schizophrenia and Comparison Subjects by Group and Body Mass Index (BMI) Cutoff of 25aa
| Add to My POL
|Measure||Schizophrenia Patients||Comparison Subjects||Uncorrected p||BMI Effect Size|
|BMI≥25 (N=54)||BMI<25 (N=19)||BMI≥25 (N=35)||BMI<25 (N=12)|
|Mean||SEM||Mean||SEM||Mean||SEM||Mean||SEM||Schizophrenia Effect||BMI Effect||Schizophrenia Patients||Comparison Subjects|
|Rey Auditory-Verbal Learning Test||35.07||1.44||36.86||2.35||45.78||1.74||52.47||2.99||<0.0001bb||0.0814||−0.15||−0.63|
| Total learning subscore||5.65||0.46||7.14||0.76||8.24||0.56||10.63||0.96||<0.0001bb||0.0071||−0.42||−0.67|
| Delayed recall subscore||0.77||0.03||0.76||0.04||0.91||0.03||0.95||0.05||0.0001bb||0.775||+0.06||−0.57|
| Recognition discrimination subscore||15.76||0.81||15.33||1.32||19.31||0.98||21.41||1.68||0.0004bb||0.6468||+0.08||−0.33|
|Verbal fluency for animals, total score||3.45||0.20||2.86||0.33||2.09||0.24||1.95||0.41||<0.0001bb||0.1533||+0.45||0.10|
|Square root of the digit span distraction test, non-distraction subscore||3.33||0.20||2.60||0.33||2.01||0.24||1.83||0.41||<0.0001bb||0.0804||+0.54||0.14|
|Square root of the digit span distraction test, distraction subscore||7.24||0.41||7.06||0.66||9.20||0.49||8.98||0.84||0.0010bb||0.7383||+0.06||0.07|
|Letter-number sequencing test, raw score||7.85||0.32||7.94||0.53||6.34||0.39||5.90||0.67||0.0005bb||0.8043||−0.03||0.30|
|Square root of the Trail Making Test, Part A, time||12.09||0.39||11.59||0.66||9.90||0.48||10.32||0.81||0.0007bb||0.8043||+0.16||−0.16|
After adjustment for the effects of age, gender, education, and ethnicity, a diagnosis of schizophrenia exerted significant negative effects on all cognitive measures (all p values significant after Holm correction) (Tables 2 and 3), with schizophrenia patients performing worse than comparison subjects.
After adjustment for the effects of age, gender, education, and ethnicity, and utilizing the Holm correction, hypertension exerted a significant negative effect on immediate, delayed, and recognition memory (Table 2). Both the hypertensive schizophrenia patients and the hypertensive comparison subjects performed more poorly than their nonhypertensive counterparts (Figure 1, Table 2). Within-group comparisons showed significant negative hyper-tension effects on delayed memory in both the schizophrenia (p=0.038) and comparison groups (p<0.0001) and on recognition memory in the schizophrenia group (p=0.008) but not in the comparison group (Figure 1). For immediate memory, within-group differences were significant for the comparison group (p=0.01) but not for the schizophrenia group. Hypertension exerted no significant effects on any of the other cognitive measures. There were no significant interaction effects between a schizophrenia diagnosis and hypertension on any of the cognitive measures, even without the Holm correction.
Immediate Memory, Delayed Recall, and Recognition Memory Performance for Patients With Schizophrenia and Comparison Subjects With and Without Hypertensiona
aMemory performance measures are from the Rey Auditory-Verbal Learning Test. Scores are adjusted for age, gender, education, and ethnicity. All participants analyzed were age 29 or older. Error bars indicate 95% confidence intervals.
After adjustment for the effects of age, gender, education, and ethnicity, and utilizing the Holm correction, a BMI ≥25 had a negative effect on delayed memory, although it did not reach statistical significance (p=0.007 before adjustment, p=0.063 with adjustment) (Table 3, Figure 2). A BMI ≥25 had no significant effect on any other cognitive measures (Table 3). No interaction effects were observed between a schizophrenia diagnosis and BMI on any cognitive measures (Table 3). When the effects of hypertension were entered into the model, the effects of elevated BMI on delayed memory were weakened.
Immediate Memory, Delayed Recall, and Recognition Memory Performance for Patients With Schizophrenia and Comparison Subjects With and Without a BMI ≥25a
aMemory performance measures are from the Rey Auditory-Verbal Learning Test. Scores are adjusted for age, gender, education, and ethnicity. All participants analyzed were age 29 or older. Error bars indicate 95% confidence intervals.
Hypertension was significantly associated with poorer verbal memory performance in patients with schizophrenia and in nonpsychiatric comparison subjects age 29 and older. Moreover, an elevated BMI was more modestly associated with poorer verbal memory performance in schizophrenia patients and in comparison subjects. Although an association between hypertension and elevated BMI and poorer cognitive test performance in persons without schizophrenia has been reported (18—30), our findings provide the first evidence for an association between the presence of individual vascular risk factors and more severe cognitive impairment in schizophrenia.
There were no significant interaction effects between a diagnosis of schizophrenia and hypertension or an elevated BMI, although the results do indicate larger negative effect sizes for these vascular risk factors in the comparison group (Tables 2 and 3). However, a diagnosis of schizophrenia alone significantly impaired cognitive performance, and hypertension and an elevated BMI exerted additional negative influences on cognitive performance in schizophrenia patients.
These findings are in agreement with other studies of nonpsychiatric populations that have shown associations between hypertension and poorer cognitive performance (references 18—22, for example). Interestingly, the effects of hypertension in our study subjects, all of whom were being treated with antihypertensive medications, were restricted to verbal memory. Other studies have similarly demonstrated a restricted pattern of verbal memory impairments in hypertensive patients receiving treatment (23). In contrast, untreated hypertensive patients show a more widespread pattern of neuropsychological impairments, including in executive functioning and motor speed, in addition to verbal memory impairments (23, 24). These findings suggest that there may be transient abnormalities related to untreated hypertension resulting in cognitive impairments that are not present once treatment is started. Should this be the case in schizophrenia patients, it would have major implications for treatment focus, especially given the high rate of untreated hypertension in this population (23). Alternatively, the deleterious effects of hypertension may be restricted to the cognitive domain of verbal memory in schizophrenia patients, regardless of antihypertensive treatment. However, even such a restricted pattern of negative effects on cognitive performance in schizophrenia patients would have significant implications for many aspects of functioning in everyday life. Indeed, evidence suggests that verbal memory is a key predictor of everyday life functioning in schizophrenia (31—34).
Although a BMI >30 has been traditionally accepted as a risk for cognitive impairment, a threshold of only 25 was sufficient to negatively influence cognitive function in the participants in our study, albeit more modestly than in investigations of nonpsychiatric subjects (26—30). Although not often reported, others have noted a BMI threshold of 25 as a risk for cognitive impairment in nonpsychiatric subjects (35).
Hypertension and obesity are well-established risk factors for atherosclerosis (see reference 36 for a review), which raises the possibility that atherosclerosis is the final common pathway through which these risk factors impair cognition. Supporting this possibility is the observation that dementia is correlated with atherosclerosis severity (37). However, nonvascular mechanisms for the association of hypertension and obesity with cognitive impairment have been suggested. For obesity, some have postulated direct actions of adiposity on neuronal tissue through neurochemical mediators produced by the adipocyte (38). Leptin, a protein secreted predominantly by adipocytes, regulates appetite and may play a role in learning and memory (39). In animal models, leptin facilitates learning, spatial memory, and long-term potentiation (40) and has been shown to enhance N-methyl-D-aspartic acid receptor function and modulate synaptic plasticity in the hippocampus (41). Interestingly, higher leptin levels have been associated with greater BMI (42), suggesting leptin resistance as a causal pathway from obesity to cognitive impairment. In support of this notion is the observation that obesity and aging are also associated with hyperleptinemia and leptin resistance (43).
For hypertension, alternative mechanisms to atherosclerosis underlying its association with cognitive impairment include oxidative stress (44) and increased activation of the renin-angiotensin system. The continuous exaggeration of the human renin-angiotensin system in transgenic mice impairs cognitive function (45). The administration of an angiotensin II type 2 receptor antagonist to these transgenic mice ameliorates this cognitive impairment and reduces blood pressure (45). In contrast, treatment with hydralazine in these mice does not reverse the cognitive impairment, although it does lower blood pressure (45). Clinical studies in humans have suggested that blockade of the renin-angiotensin system can prevent cognitive impairment associated with hypertension (46).
Our findings have important clinical implications. Patients with schizophrenia experience a much higher prevalence of vascular risk factors than does the general population (47). Although lifestyle factors contribute to this elevated prevalence (48), treatment with certain second-generation antipsychotics also increases risk, including for increased weight and hypertension (49). Compounding this problem is that patients with schizophrenia are also undertreated for these vascular risk factors relative to the general population. For example, baseline data from the CATIE study showed that rates of nontreatment for schizophrenia patients ranged from 30.2% for diabetes to 62.4% for hypertension to 88.0% for dyslipidemia (25).
Because vascular risk factors represent common and modifiable factors, these findings raise the possibility that their treatment may significantly improve cognitive outcome in schizophrenia. Indeed, antihypertensive treatment in other populations has been associated with cognitive improvement in several studies (20, 22, 46). Given the high rate of undertreatment for hypertension in schizophrenia patients and the significantly greater levels of cognitive impairment in untreated compared with treated hypertensive patients (23, 24), adequate treatment of hypertension alone could have a significant impact on cognitive outcome in the general population of patients with schizophrenia. Furthermore, caloric restriction in normal to overweight elderly individuals has been associated with improvement in memory over a 3-month interval (50), raising the possibility of another point of intervention for the cognitive impairments of schizophrenia.
While intriguing, the results presented here have limitations. First, we were unable to assess glucose intolerance because there were insufficient numbers of affected persons in the comparison group. Similarly for hyperlipidemia, there were insufficient numbers for analysis. Moreover, this study examined cross-sectional relationships between vascular risk factors and cognitive functions. These relationships in demented and nonpsychiatric groups are more complex and may be influenced by the timing and duration of exposure to these factors (21). Finally, the hypertensive participants in our study were all being treated for hypertension, which does not refiect the undertreatment observed in the general population of patients with schizophrenia. However, this bias would seem to have diluted the deleterious effects of hypertension on cognition as indicated by data from the comparison group. These issues indicate a need for further studies not only to replicate these results but to extend them to effects of other vascular risk factors. This may eventually lead to the study of drug targets not yet considered for schizophrenia.