Enrollment, allocation, follow-up, and analysis. Quetiapine was given as quetiapine fumarate; ziprasidone, as ziprasidone hydrochloride.
Least-squares mean improvement in neurocognitive composite score after 2 months of antipsychotic treatment, adjusted for baseline score and whether the patient required crisis stabilization in the 3 months prior to study entry. Patients with tardive dyskinesia were not included in the data presented in this figure (data set 1). Only the ziprasidone hydrochloride data were from data set 3, collected when ziprasidone became available, after 40% of the patients had already been entered into the study. Quetiapine was given as quetiapine fumarate.
Least-squares mean improvement in neurocognitive domain scores after 2 months of antipsychotic treatment, adjusted for baseline score and whether the patient required crisis stabilization in the 3 months prior to study entry. Patients with tardive dyskinesia were not included in the data presented in this figure (data set 1). Only the ziprasidone hydrochloride data were from data set 3, collected when ziprasidone became available, after 40% of the patients had already been entered into the study. Quetiapine was given as quetiapine fumarate.
. Least-squares mean improvement in neurocognitive composite score after 18 months of antipsychotic treatment, adjusted for baseline score and whether the patient required crisis stabilization in the 3 months prior to study entry. Patients with tardive dyskinesia were not included in the data presented in this figure (data set 1). Only the ziprasidone hydrochloride data were from data set 3, collected when ziprasidone became available, after 40% of the patients had already been entered into the study. Quetiapine was given as quetiapine fumarate.
Keefe RSE, Bilder RM, Davis SM, Harvey PD, Palmer BW, Gold JM, Meltzer HY, Green MF, Capuano G, Stroup TS, McEvoy JP, Swartz MS, Rosenheck RA, Perkins DO, Davis CE, Hsiao JK, Lieberman JA, CATIE Investigators and the Neurocognitive Working Group. Neurocognitive Effects of Antipsychotic Medications in Patients With Chronic Schizophrenia in the CATIE Trial. Arch Gen Psychiatry. 2007;64(6):633–647. doi:10.1001/archpsyc.64.6.633
Neurocognitive impairment in schizophrenia is severe and is an important predictor of functional outcome. The relative effect of the second-generation (atypical) antipsychotic drugs and older agents on neurocognition has not been comprehensively determined.
To compare the neurocognitive effects of several second-generation antipsychotics and a first-generation antipsychotic, perphenazine.
Randomized, double-blind study of patients with schizophrenia assigned to receive treatment with olanzapine, perphenazine, quetiapine fumarate, or risperidone for up to 18 months as reported previously by Lieberman et al. Ziprasidone hydrochloride was included after its approval by the Food and Drug Administration.
Fifty-seven sites participated, including academic sites and treatment mental health facilities representative of the community.
From a cohort of 1460 patients in the treatment study, 817 completed neurocognitive testing immediately prior to randomization and then after 2 months of treatment.
Main Outcome Measures
The primary outcome was change in a neurocognitive composite score after 2 months of treatment. Secondary outcomes included neurocognitive composite score change at 6 months and 18 months after continued treatment and changes in neurocognitive domains.
At 2 months, treatment resulted in small neurocognitive improvements of z = 0.13 for olanzapine (P<.002), 0.25 for perphenazine (P<.001), 0.18 for quetiapine (P<.001), 0.26 for risperidone (P<.001), and 0.12 for ziprasidone (P<.06), with no significant differences between groups. Results at 6 months were similar. After 18 months of treatment, neurocognitive improvement was greater in the perphenazine group than in the olanzapine and risperidone groups. Neurocognitive improvement predicted longer time to treatment discontinuation, independently from symptom improvement, in patients treated with quetiapine or ziprasidone.
After 2 months of antipsychotic treatment, all groups had a small but significant improvement in neurocognition. There were no differences between any pair of agents, including the typical drug perphenazine. These results differ from the majority of previous studies, and the possible reasons are discussed.
Neurocognition is severely impaired in patients with schizophrenia1,2 independent of phase of illness.3,4 While these impairments may appear prior to the onset of psychosis,5- 7 their severity in patients with chronic schizophrenia is about 1.58- 10 to 2.011 SDs lower than the healthy population. Neurocognitive impairment is a central clinical feature of schizophrenia, because it is strongly associated with functional outcomes.12,13
The importance of neurocognition in schizophrenia is emphasized by large industry and government initiatives to develop new compounds to target neurocognitive impairment,14 yet there are currently no approved treatments for neurocognitive impairment in schizophrenia and related disorders. The effect of newer-generation, “atypical” antipsychotic medications on neurocognition in patients with schizophrenia is controversial. While many studies10,15- 50 and meta-analyses51- 53 have suggested that second-generation antipsychotic treatment provides greater neurocognitive benefit to patients with schizophrenia than first-generation, “typical” antipsychotics, many of these studies have had substantial weaknesses, such as small sample sizes, short duration of treatment, no comparator or a comparator of relatively high doses of first-generation antipsychotic treatment, and inattention to important clinical factors such as the relationship of cognitive improvement with symptom change, anticholinergic treatment, and change in extrapyramidal symptoms.51- 54 Further, many of these trials were industry sponsored, which has been argued recently to exert a biasing effect.55 The impact of these various design weaknesses on study results is controversial.52- 54
The observed associations between cognitive impairment and functional outcomes in patients with schizophrenia have offered promise that neurocognitive improvement will lead to functional benefit in these patients. Little is known about this relationship, and it is not clear if the neurocognitive benefit that may be provided by antipsychotic treatment will lead to improved treatment continuation.56
This National Institute of Mental Health–sponsored study Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) compared the neurocognitive effects of olanzapine, perphenazine, quetiapine fumarate, risperidone, and ziprasidone hydrochloride in patients with chronic schizophrenia. Our primary hypothesis was that neurocognitive response would be significantly different between these treatments. To maximize the statistical power for testing this hypothesis, our primary statistical tests focused on change in a single composite score after 2 months of treatment, during which improvement in neurocognition was expected based on previous studies. An additional hypothesis was that neurocognitive improvement in the early stages of treatment would predict treatment effectiveness as measured by time to all-cause discontinuation.
The CATIE study was initiated by the National Institute of Mental Health to determine the comparative effectiveness of antipsychotic drugs. Its rationale, design, and methods were previously described57- 62 and the treatment effects on discontinuation rates and symptoms have been reported.63 The study was conducted between January 2001 and December 2004 at 57 US clinical sites (16 university clinics, 10 state mental health agencies, 7 Veterans Affairs Medical Centers, 6 private nonprofit agencies, 4 private practice sites, and 14 mixed-system sites). Patients were randomized to receive olanzapine, perphenazine, quetiapine, risperidone, or ziprasidone under double-blind conditions and studied for up to 18 months or until treatment was discontinued for any reason (phase 1). Patients whose assigned treatment was discontinued could receive other treatments in phases 2 and 3.64,65 The present report is limited to phase 1 results.
Eligible patients were 18 to 65 years of age; had received a diagnosis of schizophrenia, as determined on the basis of the Structured Clinical Interview of the DSM-IV; and were able to take oral antipsychotic medication. Patients were excluded if they had a diagnosis of schizoaffective disorder, mental retardation, or other cognitive disorders; had a history of serious adverse reactions to the proposed treatments; had had only 1 schizophrenic episode; had a history of treatment resistance, defined by persistence of severe symptoms despite adequate trials of 1 of the proposed treatments or prior treatment with clozapine; were pregnant or were breastfeeding; or had a serious and unstable medical condition.
The study was approved by the institutional review board at each site, and written informed consent was obtained from the patients or their legally authorized representatives.
Identical-appearing capsules contained olanzapine (Zyprexa; Eli Lilly and Co, Indianapolis, Ind) (7.5 mg), quetiapine fumarate (Seroquel; AstraZeneca Pharmaceuticals LP, Wilmington, Del) (200 mg), risperidone (Risperdal; Janssen Pharmaceutica Products, Titusville, NJ) (1.5 mg), perphenazine (Trilafon; Schering-Plough, Kenilworth, NJ) (8 mg), or (after January 2002) ziprasidone hydrochloride (Geodon; Pfizer Inc, New York, NY) (40 mg). The packaging was done by Quintiles Inc, Research Triangle Park, NC. The dose of the medications was flexible, ranging from 1 to 4 capsules daily, based on the study physician's judgment. Relative tablet strength was reviewed by senior representatives from each drug manufacturer. Overlap in the administration of the antipsychotic agents that patients received before study entry was permitted for the first 4 weeks after randomization to allow a gradual transition to study medication. Concomitant medications were permitted throughout the trial, except for additional antipsychotic agents. Patients had monthly visits with study doctors.
Because of product labeling, quetiapine and ziprasidone were given twice daily and olanzapine, perphenazine, and risperidone, once daily. To protect blinding, half the patients randomly assigned to perphenazine, olanzapine, and risperidone treatment were assigned to twice-daily dosing and half to once-daily dosing. To minimize initial adverse effects, patients assigned to quetiapine fumarate began treatment by receiving one 100-mg capsule on days 1 and 2, 1 twice daily on day 3, and 1 for the first dose of day 4. All patients assigned to twice-daily dosing received 5 identical-appearing capsules to begin treatment. Patients with current tardive dyskinesia (TD) as determined by a physician using Schooler-Kane criteria63 were randomized to treatments other than perphenazine.
We hypothesized that there would be significant differences between olanzapine, perphenazine, quetiapine, risperidone, and ziprasidone in improvement in cognition as measured by a neurocognitive composite score. As described in detail previously,58,59 neurocognitive assessment with 11 tests was completed to produce a valid neurocognitive composite score in 1331 of 1460 patients at baseline. The 129 patients who did not complete baseline testing were very similar to those who completed testing on all symptom and demographic variables.59 After 2 months in the study, 906 patients were still taking the medication to which they had been randomized, and 817 of them completed neurocognitive testing again (Figure 1). The neurocognitive tests were chosen by a group of advisors based on published standards, including sensitivity to impairment in schizophrenia, relation to functional outcome, potential sensitivity to treatment, and practicality for a large multisite schizophrenia antipsychotic clinical trial.52,58,59 Testers were certified at an investigators' meeting, with follow-up certification sessions throughout the study. Data quality was reviewed at a coordinating center.58,59 Five neurocognitive domain scores were calculated from 9 neurocognitive test summary scores and standardized to create z scores (mean [SD], 0 ) for each domain. A neurocognitive composite score was calculated by creating a z score of the average of the 5 standardized domain scores.59 Neurocognitive testing was also completed after 6 months and 18 months of continuing treatment with the same antipsychotic. The Pearson correlation between baseline and month 2 composite score was 0.85, indicating high within-subject consistency of the measure. Additional outcomes, collected at months 1, 3, 6, 9, 12, 15, and 18, included the incidence of neurologic adverse effects and scores on the Positive and Negative Syndrome Scale66 (PANSS). Positive and Negative Syndrome Scale scores can range from 30 to 210, with higher scores indicating more severe psychopathology. Scores on the Wide Range Achievement Test, third edition,67 reading subtest were used to estimate premorbid intellectual abilities.
The statistical plan was finalized prior to the initiation of analyses and posted on the CATIE Web site (http://www.catie.unc.edu/schizophrenia/documents/NeurocoganalysisplanFINALNov42005.pdf). Baseline characteristics for the 817 patients with a month 2 assessment were compared with the 514 patients who had baseline neurocognitive testing and did not complete month 2 testing in phase 1 using χ2 tests or t tests. Treatment groups were also compared on these same parameters for the patients with month 2 data using χ2 and analysis of variance tests. Within-group improvement in cognitive performance over time was evaluated with 1-sample t tests.
The primary objective was the comparison of treatment groups for the change in neurocognitive composite score from baseline to month 2. Secondary objectives included treatment comparisons for change in the 5 standardized domain scores at month 2. Statistical testing for changes at months 6 and 18 and for individual items is descriptive in nature. Treatment groups were compared using analysis of covariance (ANCOVA), adjusting for baseline score, whether the patient had required crisis stabilization in the 3 months prior to study entry, and TD status where applicable. All comparisons involving perphenazine excluded patients with TD. Because ziprasidone was added after approximately 40% of the patients had been enrolled, ziprasidone comparisons are secondary and were limited to the cohort of patients who underwent randomization after ziprasidone was added.
The composite score was evaluated for overall statistical significance between the 4 treatment groups at month 2 relative to P = .05 with the use of a test with 3 df, excluding patients with TD (data set 1). If the overall test result was significant, perphenazine was then compared with each of the atypical antipsychotics by a Hochberg modification68 of the Bonferroni correction for multiple treatment comparisons, in which the largest P value was compared with .05 and the smallest P value was compared with .05/3 = .017. In addition, the 3 atypical drugs were compared with each other relative to P≤.05 via step-down testing; pairwise comparisons were evaluated only if the P value from the 2 df test was ≤.05 (data set 2, patients with TD included). The ziprasidone group was compared with perphenazine and the other 3 atypical drugs within the ziprasidone cohort using a Hochberg adjustment for 4 treatment comparisons, in which the smallest P value was compared relative to .05/4 = .01 (data set 3, patients with TD excluded for comparison with perphenazine, and data set 4, patients with TD included).
Statistical significance of the domain scores was applied in the same manner as the composite, with 1 extra step. Overall significance for each domain was determined by using a Hochberg adjustment for the number of domains, in which the largest P value was compared with .05 and the smallest P value was compared with .05/5 = .01. For each domain score, further adjustment for multiple treatment comparisons was then applied to the significance level assigned in the overall stage.
A sensitivity analysis of the ANCOVA model for the primary outcome evaluated the effects of potentially important baseline covariates such as site, demographic variables, clinical characteristics, alcohol and substance abuse, baseline medications, and their interactions with treatment group. The effect of postbaseline changes in PANSS, Abnormal Involuntary Movement Scale,69 and Barnes and Simpson-Angus scale70 scores; compliance; and the addition of anticholinergic treatment on the primary outcome at month 2 were examined using ANCOVA and Pearson correlation. Lastly, Cox proportional hazards regression was used to evaluate whether change in the composite score from baseline to month 2 was a predictor of subsequent time to discontinuation for any reason and for lack of efficacy. The analysis followed the same strategy described earlier in the primary ANCOVA model and also adjusted for baseline composite score, as well as PANSS baseline score and change from baseline to month 1.
The power for finding a meaningful difference in neurocognitive change between any pair of treatment groups, identified as one quarter of the baseline standard deviation, was approximately 97% for the primary composite score, based on a sample size of 150 per treatment group and a standard deviation for change from baseline of 0.55 (obtained from month 2 results). For the domain scores, power estimates for finding any pairwise treatment difference ranged from 66% (SD 0.90%) to 82% (SD 0.75%).
The enrollment, allocation, and follow-up of study patients were described previously.63 One thousand four hundred ninety-three patients were enrolled in the study and randomized to treatment. Data from 33 subjects at 1 site were excluded prior to analysis for poor integrity. The 1331 patients who were tested at baseline were found in previously published analyses to be very similar on demographic measures to the total number of 1460 patients who were entered into the study,59 and this conclusion applies to the current cohort of 817 patients who completed neurocognitive testing at baseline and at the primary end point, 2 months postbaseline. This group of patients is the primary cohort for this report. At baseline, 25% of patients reported being antipsychotic-free, 60% of patients were taking second-generation antipsychotics, 10% were taking first-generation antipsychotics, and 5% were taking a combination of first- and second-generation antipsychotics. The percentages of patients who were randomized to the same drug they reported taking at baseline are as follows: olanzapine, 30%; perphenazine, 0%; quetiapine, 10%; risperidone, 24%; and ziprasidone, 4%. The baseline demographic and clinical characteristics of the primary cohort and those who were tested at baseline but not at 2 months (n = 514) are described in Table 1. Patients who were tested at baseline and month 2 had lower PANSS scores and higher doses during the course of phase 1. The other measures, including neurocognitive composite and domain scores, were not different between groups. The primary cohort was representative of the population of patients with chronic schizophrenia except there were fewer women (25%). Patients who completed testing at baseline and 18 months were very similar to the primary cohort.
Mean modal doses during the entire course of phase 1 for the patients who had neurocognitive test data at baseline and 2 months were olanzapine, 21.0 mg/d; perphenazine, 21.5 mg/d; quetiapine fumarate, 566.3 mg/d; risperidone, 4.1 mg/d; and ziprasidone hydrochloride, 121.9 mg/d. Mean modal doses at month 2 were very similar but slightly lower: olanzapine, 19.6 mg/d; perphenazine, 20.3 mg/d; quetiapine fumarate, 528.3 mg/d; risperidone, 3.9 mg/d; and ziprasidone hydrochloride, 117.4 mg/d. The doses of the second-generation drugs are similar to what is found in clinical practice,71,72 although the doses of olanzapine and quetiapine are about 30% higher. Seventy-four percent of patients discontinued their assigned treatment in phase 1 prior to 18 months, with a median time of 6 months. Because treatment discontinuation differed between treatments,63 the percentage of patients who provided neurocognitive data at the 2-month assessment also differed between treatments as follows: olanzapine, 68%; perphenazine, 65%; quetiapine, 59%; risperidone, 58%; and ziprasidone, 55%. Mean modal doses for the patients who had neurocognitive test data at baseline and 18 months were olanzapine, 20.5 mg/d; perphenazine, 22.0 mg/d; quetiapine fumarate, 596.3 mg/d; risperidone, 4.1 mg/d; and ziprasidone hydrochloride, 118.7 mg/d.
Change in the neurocognitive composite score from baseline to 2 months, the primary outcome measure in this study, showed improvement in each of the treatment groups. In the primary set of patients without TD (data set 1), treatment resulted in the following composite score improvements: olanzapine, 0.13 (P<.002); perphenazine, 0.25 (P<.001); quetiapine, 0.18 (P<.001); and risperidone, 0.26 (P<.001). (Least-squares means are presented in Figure 2; means are presented in Table 2). There was no overall difference between the treatment groups (P = .20). When patients with TD were included in the analysis (data set 2) and TD was used as an additional covariate in the analysis, the results were similar, with composite score improvements of 0.13 for olanzapine, 0.14 for quetiapine, and 0.22 for risperidone. Patients with TD had less overall improvement than those without TD (t573 = 2.33; P = .02). However, the TD × treatment group interaction for neurocognitive composite score change was not statistically significant (P = .27). Within the cohort of 463 patients who underwent randomization after ziprasidone was added to the trial, the results were similar and not statistically significant between groups. The ziprasidone group had improvements of 0.12 (P<.06) when patients with TD were excluded (data set 3) and 0.18 (P<.001) when they were included (data set 4).
Treatment analyses were also completed with site, Wide Range Achievement Test reading score, years of education, and baseline alcohol and substance use included in the model. These covariates did not produce statistics that differed from the unadjusted analyses.
A model of prediction of improvement in neurocognitive composite score from baseline to 2 months (R2 = 0.19) suggested that lower baseline composite score (P<.001), higher Wide Range Achievement Test reading score (P<.001), presence of substance abuse at baseline (P = .002), better compliance (P = .003), greater improvement in PANSS Negative Symptom Scale scores at 1 month (P = .007), absence of TD (P = .01), and site (P = .01) were all significant predictors of greater cognitive improvement.
Figure 3 and Table 2 present the changes in the 5 neurocognitive domains from baseline to 2 months. There was no significant disparity between the groups in improvement across the neurocognitive domains (all P values >.08).
Since there were no differences between treatment groups on the neurocognitive composite and domain scores, presentation of the individual measures at baseline and month 2 includes the entire cohort collapsed across treatment groups (Table 3). Change in the individual measures was small but consistently positive.
There were 523 patients who completed neurocognitive testing at baseline and then 6 months later while still receiving the medication to which they had been randomized. In the original analytic plan, the comparisons at 6 months were considered exploratory and thus not eligible for statistical thresholds for significance. The neurocognitive composite score improved (P<.001) for each of the treatment groups from baseline to 6 months of treatment. There were no differences between the groups on the change in the neurocognitive composite score (P = .35) or any of the neurocognitive domains (all P values >.01).
A total of 303 patients were tested at 18 months while still receiving the medication to which they had originally been randomized. Thus, 37% of the patients from the 2-month analyses were included. The 303 patients represent only 21% of the 1460 patients entered into the trial. Comparisons were made to determine if the 303 patients who continued for 18 months taking the medication to which they had been randomized were different from the other patients (n = 1028) on the neurocognitive scores at baseline. There were no significant differences on the composite score or any of the individual domains (all P values >.05), suggesting that the patients in these analyses are representative of the entire cohort in terms of baseline neurocognition. The percentage of patients who provided neurocognitive data at the 18-month assessment differed between treatments as follows: olanzapine, 29%; perphenazine, 23%; quetiapine, 18%; risperidone, 21%; and ziprasidone, 18%.
In the original analytic plan, the comparisons at 18 months were considered exploratory and thus not eligible for statistical thresholds for significance. In patients without TD, there were improvements in the neurocognitive composite score from baseline in all of the treatment groups (Table 4 and Figure 4). The improvement in the composite score from month 2 to month 18 was 0.11 (t273 = 3.20; P ≤ 0.01), suggesting that most of the cognitive improvement occurred in the first 2 months of treatment.
There were overall differences between treatments in composite score change (P<.05). Pairwise comparisons suggested that improvement in the neurocognitive composite score was greater in the perphenazine group (0.49) than in the olanzapine group (0.15; P = .002) or the risperidone group (0.28; P = .04). The ziprasidone and quetiapine groups did not differ from any of the other treatments.
Some of the overall differences between groups were explained by the differences in the reasoning domain, although this difference would not have met the criteria for formal statistical significance after controlling for multiple comparisons. The perphenazine group tended to improve more than the other treatment groups (Table 4). In comparisons between the second-generation treatments that included patients with TD, there were no overall differences between the groups.
The Pearson correlations between change in neurocognitive composite score from baseline and change in PANSS symptom factors from baseline are listed in Table 5. Negative correlations indicate that cognitive improvement was associated with symptom reduction, but the magnitude of these correlations was quite small. The correlations between change in neurocognitive composite score from baseline to 2 months and the 1-month and 3-month changes from baseline in extrapyramidal symptoms (Simpson-Angus scale mean score), TD (Abnormal Involuntary Movement Scale global severity score), and akathisia (Barnes scale global score) were r<0.05 with all P values >.05. The correlation between neurocognitive change and compliance was 0.07 (P = .04). The 151 patients who were receiving anticholinergic medications at baseline had a mean neurocognitive score change of 0.169, which did not differ from the 666 patients who were not receiving anticholinergic medications at baseline, who had a mean score change of 0.224. However, for the 38 patients who had anticholinergic medications added during the first 2 months of treatment, composite scores worsened (mean [SD] change, −0.049 [0.46]) at the 2-month assessment, compared with other patients whose composite scores improved (mean [SD] change, 0.190 [0.56]) (P≤0.01). No interaction between neurologic or anticholinergic measures and treatment group was found for the composite score.
Cox proportional hazards regression analyses suggested that change in the neurocognitive composite score from baseline to 2 months was not a significant predictor of time until all-cause discontinuation. However, there was a suggestion of a treatment interaction for this analysis (P<.10). Neurocognitive improvement predicted time to all-cause discontinuation in patients treated with quetiapine (P = .02; hazard ratio for 0.25-SD improvement in composite score, 0.98 [95% confidence interval, 0.954-0.996]) and ziprasidone (P = .009; hazard ratio, 0.957 [95% confidence interval, 0.925-0.989]) but not the other treatments. Inclusion of change in PANSS score, compliance, and other baseline covariates found previously to predict time to discontinuation did not reduce the statistical significance for the quetiapine group. The prediction of time to discontinuation due to inefficacy produced similar results.
All of the antipsychotic treatment groups had a small improvement in neurocognition as measured by change in a composite score derived from 11 neurocognitive tests assessed at baseline and after 2 months of treatment. However, there was no significant difference between the groups. Exploratory analyses suggested that after 18 months of treatment, there might be differences between the treatments, with the older antipsychotic perphenazine demonstrating the most neurocognitive improvement. Neurocognitive improvement was found to contribute to the overall effectiveness of some of the antipsychotic treatments as measured by time to discontinuation. A variety of demographic and clinical factors contributed to neurocognitive improvement.
The current results are in contrast to numerous published studies and 3 meta-analyses suggesting neurocognitive advantages of the second-generation antipsychotic medications compared with first-generation treatments.10,15- 53 This failure to document a neurocognitive advantage of second-generation antipsychotics suggests that the positive findings from prior reports may not generalize well to the type of everyday clinical practice examined in the CATIE trial. The contrast between the current results and those reported previously may be explained in part by methodological differences between the studies; thus, careful scrutiny is warranted.51- 54
First, the current study included neurocognitive treatment data on 817 patients, more than twice as many as the largest previous trial. It is possible that smaller studies are susceptible to results that are less stable and generalizable.
Second, in contrast to many previous studies that used high dosages of first-generation antipsychotics, usually haloperidol, creating an unfair comparison because of the increased risk of extrapyramidal symptoms and anticholinergic treatment, which may impair cognition,51- 54 this study used relative tablet strengths that were reviewed and approved by scientific experts and leaders from each of the pharmaceutical companies that manufacture the compounds. Further, the choice of perphenazine as the representative of the older antipsychotic medications may have reduced the extrapyramidal symptoms and need for anticholinergic treatment associated with haloperidol and other high-potency first-generation antipsychotics,63,82 especially since the dose range of 8 to 32 mg, equivalent to 2.5 to 10 mg of haloperidol, was lower than in most early studies.83- 85 Perphenazine, a phenothiazine, has been available since the late 1950s. It was selected for CATIE because it is a moderate-potency first-generation antipsychotic and causes only mild sedative effects and extrapyramidal symptoms in the dose range selected. Although it has never been considered to be an atypical antipsychotic, one of its metabolites, N-dealkylperphenazine, has relatively high affinity for serotonin 2A receptors, which by some definitions may confer atypical properties.86 Therefore, prior studies may have documented, at least in part, negative effects associated with high-dose haloperidol treatment rather than an intrinsic, specific cognitive enhancement effect associated with some of the second-generation drugs vs the older medications. It will be important for health care professionals using perphenazine or other first-generation antipsychotics to carefully consider these dosage implications.
Third, compared with most clinical trials,87 the present study had broad inclusion and minimal exclusion criteria and allowed comorbid conditions, concomitant medications, and current substance abuse. It was conducted at a variety of clinical settings where people with schizophrenia are treated. These “real-world” features of the study, which were intended to make the results more widely applicable,87 may also account for the differences in results between this and previous studies. While single-site trials that are limited to research centers and highly screened patients may be more sensitive to the cognitive changes associated with treatment and may offer hope for the potential neurocognitive benefit of these medications, the current results may more accurately reflect clinical reality.
Fourth, at least 60% of patients in this study reported receiving atypical antipsychotic treatment prior to randomization, which is substantially higher than in many of the earlier studies, which were completed when treatment with second-generation antipsychotics was less common. However, only 16% of patients were randomized to the same drug they reported taking at baseline. Although the inclusion in our statistical analyses of patients' reports of their antipsychotic medication prior to randomization did not change the study results, it is possible that some of the potential neurocognitive benefit of the second-generation treatments examined in this study had already been realized prior to the initiation of the randomized treatment. If so, treatment effects may have been attenuated, but this would not explain the surprisingly beneficial effect of perphenazine.
Since the neurocognitive benefit of treatment in this study was small, the possibility that neurocognitive improvement was due solely to practice effects or expectation biases cannot be ruled out. In studies of antipsychotic medications, practice and placebo effects in schizophrenia are impossible to disentangle from treatment effects. However, 1 recent double-blind study comparing the effects of donepezil hydrochloride and placebo in a highly refined sample of 226 patients with schizophrenia stabilized while taking second-generation antipsychotics suggested that patients taking placebo who had negative symptom improvement also had neurocognitive effect size improvements of 0.22 SD after being tested twice over 6 weeks on the same test battery used in this study, suggesting a practice or placebo effect (R.S.E.K, Anil Malhotra, MD, H.Y.M., John Kane, MD, Robert Buchanan, MD, Anita Murthy, PharmD, Mindy Sovel, MA, Chunming Li, PhD, Robert Goldman, PhD, unpublished data, July 2004) consistent with the improvements reported herein. However, higher doses of haloperidol may blunt working memory performance88 and procedural learning, likely mediators of practice-related performance improvements,89,90 which may have conferred an advantage for the newer drugs in some previous studies, while patients treated with lower doses of haloperidol have demonstrated learning improvements, especially in first-episode samples.10,91 Thus, while practice effects or expectation biases may be present in patients with schizophrenia on some tests, they may be adversely affected by certain treatments.
The elements of treatment compliance are complex,92 and the current data support a weak relationship between neurocognitive improvement and treatment compliance because greater neurocognitive benefit was slightly associated with greater compliance. Although tenuous, these data may suggest either that increased compliance to treatment enhances cognitive benefit or that cognitive improvement encourages treatment compliance.56
Patients who had anticholinergic medication added during the first 2 months of treatment demonstrated no cognitive benefit, which was significantly less than those patients who did not have anticholinergics added. This result does not suggest that anticholinergic treatments should be withheld to aid cognition, however, since the extrapyramidal symptoms that are the targets of anticholinergic treatment are also associated with worse cognitive performance.10,22
In patients treated with quetiapine or ziprasidone, cognitive improvement predicted overall treatment effectiveness as measured by time to treatment discontinuation. These results suggest that treatment effectiveness is multiply determined and that cognitive improvement is one of the significant determinants. Cognitive change was a predictor of treatment discontinuation in patients receiving quetiapine and ziprasidone, which produced relatively weak symptom response in this study.63 The differences between treatment groups in the contribution of cognitive improvement to longer periods to discontinuation suggest that cognitive improvement may be a contributor to treatment continuation in patients who are not demonstrating other symptom improvement or that the absence of cognitive improvement (or worsening) may be particularly important in the decision to discontinue treatment. Conversely, neurocognitive change may be a less important contributor to treatment continuation if symptom reduction is substantial. Finally, the amount of cognitive improvement that confers to improved functional outcomes such as quality of life is unknown. The various contributors to treatment discontinuation and quality of life in the CATIE trial will be evaluated comprehensively in a future report.
Several design features in this study limit the interpretation of the findings. First, the patient population in this study did not include patients with schizoaffective disorder, those with only 1 schizophrenic episode, or those who had adverse reactions to any of the treatments or were treatment resistant and was unlikely to include patients who had responded well to their current treatment. Thus, although this trial likely included a population that was more generalizable than most clinical trials,87 this group of patients was still selective compared with the entire population of patients treated with antipsychotic medications, and the results must be viewed in this context. Second, a disadvantage to the random assignment of medications in this study is that if a medication is more effective for certain subgroups of patients who are recognizable by health care professionals, the effectiveness of this treatment will be reduced. Third, the randomization scheme prevented patients with TD from receiving treatment with perphenazine. Although all comparisons between perphenazine and the other treatments eliminated patients with TD, this differential screening process may still have given an advantage to perphenazine with regard to neurocognitive efficacy. The randomization of patients with TD, who may be less responsive to first-generation antipsychotic treatment,93,94 may have inadvertently assigned to the perphenazine group patients who were more likely to respond to the only first-generation antipsychotic in the study. Finally, while the design of the CATIE trial allowed inclusion of patients not usually studied in clinical trials, only 817 (56%) of the 1460 patients in the trial generated neurocognitive data at baseline and after having received treatment with the same antipsychotic for 2 months, and only 303 patients (21%) generated neurocognitive data after having been treated with the same antipsychotic for 18 months. Furthermore, because of the differential discontinuation rate between the treatments,63 a smaller percentage of patients were tested at end point with some of the treatments. Thus, the generalizability of these findings to the large community of patients receiving antipsychotics must be considered.
In sum, neurocognitive improvement associated with antipsychotic treatment in patients with schizophrenia was small and not different between treatments, including the first-generation antipsychotic perphenazine. A variety of issues may explain the contrast between the current findings and previous studies that had suggested an advantage of second-generation treatments, such as the relatively low dose of perphenazine, the overall high discontinuation rate in the study, and the lower percentage of patients entering the study while taking first-generation antipsychotics compared with earlier studies. Neurocognitive changes may be important predictors of treatment discontinuation with certain antipsychotics.
Correspondence: Richard S. E. Keefe, PhD, Department of Psychiatry, Duke University Medical Center, Box 3270, Durham, NC 27710 (email@example.com).
Submitted for Publication: May 18, 2006; final revision received August 9, 2006; accepted October 7, 2006.
Financial Disclosure: Dr Keefe has received grant/research support from AstraZeneca Pharmaceuticals LP, Eli Lilly and Co, Janssen Pharmaceutica, and Pfizer Inc, as well as provided educational services to AstraZeneca Pharmaceuticals LP, Eli Lilly and Co, Forest Labs, GlaxoSmithKline, Janssen Pharmaceutica, Johnson & Johnson, Pfizer Inc, and Repligen. He has also served as a consultant and on advisory boards for various pharmaceutical companies as follows: Abbott Pharmaceuticals (advisory board), AstraZeneca Pharmaceuticals LP (advisory board, consultant), Bristol-Myers Squibb (advisory board), Dainippon Sumitomo Pharma (consultant), Eli Lilly and Co (advisory board, consultant), Forest Labs (consultant), GlaxoSmithKline (consultant), Johnson & Johnson (advisory board, consultant), Lundbeck/Solvay/Wyeth (advisory board), Memory Pharmaceuticals (advisory board), Merck (advisory board, consultant), Orexigen (advisory board, consultant), Otsuka Pharmaceutical Co Ltd (consultant), Pfizer Inc (advisory board, consultant), Repligen (consultant), Saegis (advisory board, consultant), Sanofi/Aventis (advisory board, consultant), and Xenoport (consultant). Dr Keefe also receives royalties for the Brief Assessment of Cognition, the Brief Assessment of Cognition in Affective Disorders, and the Brief Assessment of Cognition in Schizophrenia, as well as one of the subtests from these batteries that is a part of the National Institute of Mental Health–MATRICS Consensus Battery. Dr Bilder has received grant/research support from Pfizer Inc and Pharmacia and served as a consultant and/or on advisory boards for Abbott (advisory board), ACADIA (consultant), AstraZeneca Pharmaceuticals LP (advisory board), Cypress Bioscience (consultant), Dainippon Sumitomo Pharma (consultant), Eli Lilly and Co (consultant), Janssen Pharmaceutica (advisory board, consultant), Johnson & Johnson (consultant), Lundbeck/Solvay/Wyeth (advisory board), Memory Pharmaceuticals (advisory board), Pfizer Inc (advisory board), and Pharmacia (consultant). Dr Bilder is a shareholder and consultant to Cogtest Inc and received royalties for the Clinical Global Impression of Cognition in Schizophrenia. Dr Davis is an employee of Quintiles Inc and has received consulting fees from Eli Lilly and Co and Pfizer Inc. Dr Harvey is a consultant for Janssen Pharmaceutica Products, Eli Lilly and Co, Pfizer Inc, AstraZeneca Pharmaceuticals LP, Novartis, the Sanofi-Aventis group, Solvay Pharmaceuticals, Memory Pharmaceuticals, Actelion Pharma, and Cypress Biosciences; was on the speakers' bureau for Pfizer Inc and Eli Lilly and Co; and has received grants from Bristol-Myers Squibb and Pfizer Inc. Dr Gold has served as a consultant for Eli Lilly and Co and Pfizer Inc. Dr Gold also receives royalties for the Brief Assessment of Cognition, the Brief Assessment of Cognition in Affective Disorders, and the Brief Assessment of Cognition in Schizophrenia, as well as one of the subtests from these batteries that is a part of the National Institute of Mental Health–MATRICS Consensus Battery. Dr Meltzer has received grant/research support from Abbott, ACADIA, ARYx Therapeutics, AstraZeneca Pharmaceuticals LP, Bristol-Meyers Squibb, Cephalon, Eli Lilly and Co, Janssen Pharmaceutica, Memory Pharmaceuticals, the National Institute of Mental Health, Organon, Pfizer Inc, Sepracor Inc, Solvay, and Wyeth as well as provided educational services to Janssen Pharmaceutica and Pfizer Inc. He has also served as a consultant and on advisory boards for various pharmaceutical companies as follows: Abbott Pharmaceuticals (consultant), ACADIA (advisory board), AstraZeneca Pharmaceuticals LP (consultant), Brain Cells Inc (advisory board), Bristol-Myers Squibb (advisory board), Dainippon Sumitomo Pharma (consultant), Eli Lilly and Co (advisory board, consultant), GlaxoSmithKline (consultant), Johnson & Johnson (advisory board, consultant), Lundbeck/Solvay/Wyeth (advisory board, consultant), Memory Pharmaceuticals (advisory board), Merck (advisory board, consultant), and Minster Pharmaceuticals (advisory board, consultant). Dr Green has served as a consultant to and received honoraria from Janssen FP, Pfizer Inc, Eli Lilly and Co, Otsuka Pharmaceutical Co Ltd, Lundbeck, AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Abbott Laboratories, Sanofi-Aventis Pharmaceuticals, and Memory Pharmaceuticals. He has received research support from Janssen FP. Dr Stroup has received consulting and educational fees from AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Eli Lilly and Co, Pfizer Inc, Janssen Pharmaceutica Products, and GlaxoSmithKline. Dr McEvoy has received research support from Eli Lilly, Janssen Pharmaceutica, AstraZeneca Pharmaceuticals LP, and Pfizer. He has been a consultant to GlaxoSmithKline and Janssen Pharmaceutica. He has received honoraria from speaking for Eli Lilly and Co, Janssen Pharmaceutica, Bristol-Meyers Squibb, and Pfizer. Dr Swartz has received research funding from Eli Lilly and Co and consulting and educational fees from AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Eli Lilly and Co, and Pfizer Inc. Dr Rosenheck had received research support from Eli Lilly and Co, Janssen Pharmaceutica, AstraZeneca Pharmaceuticals LP, and Wyeth. He has been a consultant to GlaxoSmithKline, Bristol-Myers Squibb, and Janssen Pharmaceutica. Dr Perkins has received research funding from AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Otsuka Pharmaceutical Co Ltd, Eli Lilly and Co, Janssen Pharmaceutica Products, and Pfizer Inc and consulting and education fees from AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Eli Lilly and Co, Janssen Pharmaceutica Products, GlaxoSmithKline, Forest Labs, Pfizer Inc, and Shire. Dr Lieberman has received research grant and contract support from AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Janssen, Eli Lilly and Co, Novartis, and Pfizer Inc and research grant support from Abbott, ACADIA, GlaxoSmithKline, Merck, Organon, and Sanofi-Synthelabo. He has acted as a consultant and an advisory board member for Abbott, AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, GlaxoSmithKline, Eli Lilly and Co, Pfizer Inc, and Solvay; has been a consultant for Novartis, Janssen, and Johnson & Johnson; and has been an advisory board member for Aventis, Lundbeck, and Organon. Dr Lieberman holds a patent for Repligen and has received honoraria from AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb, Janssen, Eli Lilly and Co, Novartis, and Pfizer Inc.
Funding/Support: This article was based on results from the CATIE project, supported by grant N01 MH90001 from the National Institute of Mental Health. AstraZeneca Pharmaceuticals LP, Bristol-Myers Squibb Co, Forest Pharmaceuticals, Inc, Janssen Pharmaceutica Products, Eli Lilly and Co, Otsuka Pharmaceutical Co Ltd, Pfizer Inc, Schering-Plough, and Zenith Goldline Pharmaceuticals, Inc, provided medications for the studies.
Previous Presentations: This work was presented in part at the annual meeting of the Society for Biological Psychiatry; May 19, 2006; Toronto, Ontario.
Acknowledgment: We are indebted to the 1493 participants in the CATIE Schizophrenia Trial for their collaboration. We gratefully acknowledge the contributions of the late Mahmoud A. Parsa, MD, of the Department of Psychiatry, Case Western Reserve University, Cleveland, Ohio; Grayson S. Norquist, MD, MSPH, previously the director, Division of Services and Intervention Research, National Institute of Mental Health, and currently the chairman, Department of Psychiatry and Human Behavior, University of Mississippi Medical Center, Jackson; Ingrid Rojas-Eloi, BS, project manager of CATIE; Tiffany Harris, staff assistant, Department of Psychiatry, School of Medicine, University of North Carolina at Chapel Hill; Trina Walker, research assistant, Duke University Medical Center; and the Quintiles CATIE project team.
The CATIE Study Investigators Group includes Lawrence Adler, MD, Clinical Insights; Glen Burnie, MD, and Mohammed Bari, MD, Synergy Clinical Research, Chula Vista, Calif; Irving Belz, MD, Tri-County Mental Health Mental Retardation Services, Conroe, Tex; Raymond Bland, MD, Southern Illinois University School of Medicine, Springfield; Thomas Blocher, MD, Mental Health and Mental Retardation Authority of Harris County, Houston, Tex; Brent Bolyard, MD, Cox North Hospital, Springfield, Mo; Alan Buffenstein, MD, The Queen's Medical Center, Honolulu, Hawaii; John Burruss, MD, Baylor College of Medicine, Houston; Matthew Byerly, MD, University of Texas Southwestern Medical Center at Dallas; Jose Canive, MD, Albuquerque VA Medical Center, Albuquerque, NM; Stanley Caroff, MD, Behavioral Health Service, Philadelphia, Pa; Charles Casat, MD, Behavioral Health Center, Charlotte, NC; Eugenio Chavez-Rice, MD, El Paso Community Mental Health Mental Retardation Center, El Paso, Tex; John Csernansky, MD, Washington University School of Medicine, St Louis, Mo; Pedro Delgado, MD, University Hospitals of Cleveland, Cleveland, Ohio; Richard Douyon, MD, VA Medical Center, Miami, Fla; Cyril D'Souza, MD, Connecticut Mental Health Center, New Haven; Ira Glick, MD, Stanford University School of Medicine, Stanford, Calif; Donald Goff, MD, Massachusetts General Hospital, Boston; Silvia Gratz, MD, Eastern Pennsylvania Psychiatric Institute, Philadelphia; George T. Grossberg, MD, St Louis University School of Medicine–Wohl Institute, St Louis; Mahlon Hale, MD, New Britain General Hospital, New Britain, Conn; Mark Hamner, MD, Medical University of South Carolina and Veterans Affairs Medical Center, Charleston; Richard Jaffe, MD, Belmont Center for Comprehensive Treatment, Philadelphia; Dilip Jeste, MD, University of California, San Diego, VA Medical Center; Anita Kablinger, MD, Louisiana State University Health Sciences Center, Shreveport; Ahsan Khan, MD, Psychiatric Research Institute, Wichita, Kan; Steven Lamberti, MD, University of Rochester Medical Center, Rochester, NY; Michael T. Levy, MD, PC, Staten Island University Hospital, Staten Island, NY; Jeffrey Lieberman, MD, University of North Carolina School of Medicine, Chapel Hill; Gerald Maguire, MD, University of California Irvine, Orange; Theo Manschreck, MD, Corrigan Mental Health Center, Fall River, Mass; Joseph McEvoy, MD, Duke University Medical Center, Durham, NC; Mark McGee, MD, Appalachian Psychiatric Healthcare System, Athens, Ohio; Herbert Meltzer, MD, Vanderbilt University Medical Center, Nashville, Tenn; Alexander Miller, MD, University of Texas Health Science Center at San Antonio; Del D. Miller, MD, University of Iowa, Iowa City; Henry Nasrallah, MD, University of Cincinnati Medical Center, Cincinnati, Ohio; Charles Nemeroff, MD, PhD, Emory University School of Medicine, Atlanta, Ga; Stephen Olson, MD, University of Minnesota Medical School, Minneapolis; Gregory F. Oxenkrug, MD, St. Elizabeth's Medical Center, Boston; Jayendra Patel, MD, University of Massachusetts Health Care, Worcester; Frederick Reimher, MD, University of Utah Medical Center, Salt Lake City; Silvana Riggio, MD, Mount Sinai Medical Center–Bronx VA Medical Center, Bronx, NY; Samuel Risch, MD, University of California, San Francisco; Bruce Saltz, MD, Henderson Mental Health Center, Boca Raton, Fla; Thomas Simpatico, MD, Northwestern University, Chicago, Ill; George Simpson, MD, University of Southern California Medical Center, Los Angeles; Michael Smith, MD, Harbor–UCLA Medical Center, Torrance, Calif; Roger Sommi, PharmD, University of Missouri, Kansas City; Richard M. Steinbook, MD, University of Miami School of Medicine, Miami; Michael Stevens, MD, Valley Mental Health, Salt Lake City; Andre Tapp, MD, VA Puget Sound Health Care System, Tacoma, Wash; Rafael Torres, MD, University of Mississippi, Jackson; Peter Weiden, MD, SUNY Downstate Medical Center, Brooklyn, NY; James Wolberg, MD, Mount Sinai Medical Center, New York, NY.