Olanzapine and clozapine

Overall design and outcome measures.

This study used a randomized, double-blind, parallel-groups design. Patients who fit inclusion and exclusion criteria (see below) received up to 2 months of either clozapine (6.25 to 50.0 mg/d) or olanzapine (2.5 to 15 mg/d). Behavioral and motor assessments were conducted at baseline prior to randomization, and at days 7, 14, 21, 28, 35, and 63 of treatment, always by the same blinded evaluator. Study medications could be adjusted based on clinical judgment with the primary aim of diminished hallucinations. Other medications remained unchanged during the study period. Weekly complete blood counts were obtained on all subjects and were reviewed by an unblinded staff member otherwise unrelated to the study.

The primary outcome variable in this intention-to-treat trial was the score on the Scale for the Assessment of Positive Symptoms (SAPS)13 for psychotic symptoms assessed on the last study day. Secondary behavioral outcome measures included the Visual Hallucinations item (number 9) from the SAPS, and the Brief Psychiatric Rating Scale (BPRS). For assessment of parkinsonism, we used the motor subscale score of the Unified Parkinson’s Disease Rating Scale during “on” and the UPDRS activities of daily living (ADL) subscale score during “on” and “off” periods. We assessed the maximal dose of each drug, the number of patients assigned to each drug completing the full 63 days of neuroleptic treatment, and the number who completed earlier because parkinsonism was substantially worse. To analyze whether specific components of the UPDRS motor score were altered, we divided the UPDRS motor section into four subscales: bradykinesia (nine items: numbers 28 to 35 and 40); tremor (seven items: numbers 20a to e, and 22a and b); rigidity (five items: numbers 23 to 27); and gait, balance, and midline function (six items: numbers 18, 19, and 36 to 39).

Patient selection.

Inclusion criteria were PD defined by core assessment program for intracerebral transplantation (CAPIT) criteria,14 age greater than 21, hallucinations or psychosis defined by the Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (DSM-IV)15 that occurred on a weekly basis for at least 30 days, and stable dosages of antiparkinsonian medications (l-dopa, dopamine agonists, selegeline, amantadine, anticholinergic agents) for 30 days. Exclusionary criteria included exposure to any neuroleptic drug over the past 6 weeks prior to enrollment; dementia, defined by a score of less than 24 on the Mini-Mental State Examination; untreated medical illnesses; leukopenia; immunodeficiency; current exposure to any bone marrow suppressant medication, including carbamazepine and anticancer agents; and any prior history of a primary psychiatric illness including schizophrenia, psychotic depression, or bipolar disorder. No patient had hallucinations that preceded the diagnosis of PD or treatment with dopaminergic drugs.

Consecutive patients in a tertiary care practice were screened for inclusion and exclusion criteria. Patients who fulfilled entry criteria read a patient information packet, and those who were interested in study participation signed informed consent. The protocol, patient information, and consent were approved by the Investigation Review Board of Rush University.

Randomization and study medications.

Each qualified consenting patient was independently randomized to one of two treatments by the toss of a fair coin, and treatment assignments were generated in advance by a staff member who was not involved in the conduct of the study. The assignment codes were kept in a locked file and were only available to the study pharmacist who had no contact with patients or the clinical evaluators. At study initiation, patients received either clozapine 6.25 mg/d or olanzapine 2.5 mg/d as a single dose given as identically appearing capsules prepared by the study pharmacist. Patients were seen 1 week later, and, based on their responses to the SAPS, description of hallucinations over the past 5 days, an open-ended interview on motoric function, and the UPDRS motor subscale score, a blinded treating physician adjusted study drugs by increasing clozapine to 12.5 mg/d or olanzapine to 5 mg/d or by leaving the dose unchanged. The treatment adjustment was based on simultaneous control of hallucinations and avoidance of deterioration in parkinsonian motor function. On day 14, after the same data collection, if hallucinations persisted and parkinsonism was not aggravated, increases were permitted to clozapine 18.75 mg/d or olanzapine 7.5 mg/d. On days 21, 28, and 35, similar dosage adjustments were permitted, to reach a maximal dose of clozapine 50 mg/d or olanzapine 15 mg/d. From day 35 to 63, study medication did not change. Final assessment occurred on day 63 or the visit on which parkinsonism was substantially worse, as judged by the blinded treating physician.

Blinding and blood count assessments.

All treatment information was maintained in a blinded format until after the database was locked and all queries were resolved Because clozapine is associated with the risk of blood dyscrasia, complete blood counts (CBC) were obtained during screening prior to randomization, weekly throughout the study, and 1 week after study medication ended. An unblinded investigator, otherwise unrelated to the study, reviewed the weekly blood counts within 24 hours. In patients demonstrating a leukocyte count below 4,000 mm³, a repeat CBC was obtained within 72 hours. Drug therapy was suspended if the leukocyte count dropped below 3,000 mm³ or the absolute neutrophil count dropped below 1,500 mm³. In such patients, repeat CBC was obtained within 7 days. If the leukocyte level rose above 3,000 mm³, drug therapy was reinstituted at the most recent dose. Drug discontinuation and study completion occurred if the leukocyte count dropped below 2,000 mm³ or the absolute neutrophil count fell below 1,000 mm³. The blinded evaluating physician had no knowledge of the blood count data. There was no formal blindness assessment for patients or evaluators.

Sample size determination.

Because data on the use of the SAPS in patients with PD were limited,11,16 ⇓ sample size calculations were based on a larger study of schizophrenic patients examined with this scale.17 The study was designed as an equivalence trial with significance level 0.05 and with an 80% power for rejecting nonequivalence if the group mean differed by 10 points or more, assuming the SD of change scores within groups was 10 points. We selected the 10 point “tolerance” based on clinical judgment and prior publications on antipsychotic efficacy of neuroleptics.16,17 ⇓ This power requirement led to a sample size of 14 patients per treatment or 28 patients total.

Stopping rules.

Because of concerns that neuroleptics could aggravate parkinsonian motor signs, after 15 patients completed their participation, the study statistician (S.L.), otherwise uninvolved in clinical care of patients, performed an interim safety analysis to examine whether the safety measure (motor UPDRS) was behaving similarly in the two groups. To protect blinding, assignment codes were delivered by the study pharmacist to the statistician as Drug 1 and Drug 2 without reference to drug names. The change in motor UPDRS scores was compared using a Wilcoxon’s rank sum test. A significance level of 0.025 was used as partial protection for the examination of the data. A two-sample t-test was used as a secondary analysis.

Statistical analysis.

Basic summary statistics were expressed as means and SD. Within-patient changes were assessed via Wilcoxon signed rank tests for paired samples. Point estimates and CI for within-patient changes were Hodges-Lehmann estimates and CI corresponding to the nonparametric test procedures. Change scores were computed as end score minus baseline score, so that positive change scores denote deterioration and negative scores denote improvement. Groups were compared using Wilcoxon rank sum tests (change scores, and baseline UPDRS and behavioral scores); 95% Hodges-Lehmann CI for differences in change scores were presented for the primary and secondary outcome measures. Positive numbers corresponded to bigger differences for patients randomized to olanzapine; negative differences corresponded to bigger differences for patients randomized to clozapine. This nonparametric method of confidence intervals corresponded to the nonparametric tests. The associated estimator of the population difference was the median of all pairwise differences. Numbers of subjects completing drug treatment for the full 63 days were compared using Fisher’s exact test.

Subjects.

The study was completed without protocol violation, but included fewer patients than anticipated because stopping criteria were met before full enrollment was achieved. Of 34 chronically hallucinating patients screened, 15 met entry criteria, and all consented to participate in the research program. After the baseline assessment, seven men and eight women received neuroleptic treatment, olanzapine in seven and clozapine in eight. Mean age was 72.1 years (SD, 8.4), and mean disease duration was 13.5 years (SD, 8.9). All patients received l-dopa, mean dose 556 mg/d (SD, 252); 73% were on agonists, 20% on anticholinergic agents, 20% on selegeline, and 20% on amantadine. There were no significant differences between the two groups in any of these measures.

Baseline characteristics.

The mean Mini-Mental State Examination score was 26.5 (SD, 2.1). All patients experienced chronic hallucinations, their mean SAPS score was 12.7 (SD, 7.0), and their mean BPRS score was 29.3 (SD, 6.3). There were no significant differences in these behavioral measures between the subjects assigned to olanzapine or clozapine treatment. At baseline, the total group had a mean “on” period UPDRS motor score of 30.7 (SD, 14.8), a mean “on” UPDRS ADL score of 12.9 (SD, 9.5), and a mean “off” UPDRS ADL score of 24.8 (SD, 9.5). At baseline, olanzapine-assigned subjects were motorically less disabled than those assigned to clozapine (mean UPDRS motor score 21.4 [SD, 12.2] versus 38.9 [SD, 12.2]; p = 0.024).

Dosage of medications and study completions.

The mean peak doses were 25.8 mg/d (SD, 13.5) for clozapine and 11.4 mg/d (SD, 3.5) for olanzapine. Of the 15 patients enrolled, three patients, all on clozapine, terminated because of reasons unrelated to the medication trial (distance, intercurrent illness, and death from unrelated causes in one subject each). Of the 12 remaining subjects, all completed the study by reaching day 63 or by reaching the endpoint of exacerbated parkinsonism (figure). Six of seven patients assigned to olanzapine completed before the full 63 days of drug treatment because of deteriorating parkinsonism; none on clozapine completed because of this effect (Fisher’s p value = 0.015). Three patients terminated on day 35, and one each on day 21, 28, and between day 35 and 63.

• Download figure
• Open in new tab
• Download powerpoint

Figure. Flow diagram for randomized comparison of clozapine (left) versus olanzapine (right) in PD patients with hallucinations.

Motor changes.

UPDRS motor scores declined in olanzapine-treated subjects (mean change, 12.3; SD, 11.5; p = 0.016; point estimate 11.25; 95% CI, 2.5 to 23.5), but actually modestly improved in clozapine-treated subjects (mean change, −6.0; SD, 8.2; p = 0.125) (table 1). The primary domains that accounted for the olanzapine-associated decline were gait (mean change, 3.3; SD, 2.0) and bradykinesia (mean change, 6.9; SD, 6.5). UPDRS change scores (baseline to study end) between the two groups differed (p = 0.004; table 2). The change scores for the UPDRS assessments of ADL showed the same deterioration on olanzapine and improvement on clozapine (see table 2), with differences for both the “on” (p = 0.017) and “off” (p = 0.005) assessments.

Despite randomization, the olanzapine group was less motorically impaired at baseline than the clozapine group. To assess if baseline impairment strongly influenced outcome, we examined whether the patients on olanzapine whose baseline scores were close to the mean baseline score of the clozapine group behaved like the other olanzapine-treated patients (deteriorated) or the clozapine-treated patients (no decline). At baseline, three of seven olanzapine-assigned subjects’ scores were close to the mean group score (38.9) for clozapine-assigned subjects (31, 34, and 37). These subjects all followed the pattern of the other olanzapine-treated patients and declined motorically (final UPDRS scores: 34, 47, and 39). We likewise examined the one clozapine-assigned patient whose baseline score (17) was close to the mean baseline score of the olanzapine-assigned group (21.4), and this patient did not decline (final UPDRS score: 17).

Changes in hallucinations and behavior.

The study was completed before an adequate number of subjects was enrolled to test fully for significant differences between olanzapine and clozapine efficacy on hallucinations. Nonetheless, even with the small numbers, patients on clozapine showed improvement from baseline function in total SAPS (p = 0.016; point estimate = −7.0; 95% CI, −11.0 to −2.0), the specific visual hallucination items on the SAPS (p = 0.013), and BPRS (p = 0.031). Although the total SAPS and the SPS visual hallucination item improved on olanzapine, the changes were not statistically significant (see table 1). The change score comparison between olanzapine and clozapine was not statistically significant.

Blood monitoring safety.

The mean baseline leukocyte count for all patients was 6,138 (SD, 1,392; range, 3,630 to 9,160). At the nadir, the mean level was 6,127 (SD, 1,543; range, 3,400 to 7,870) on olanzapine and 5,775 (SD, 1914; range, 3,138 to 8,980) on clozapine. No patient on the program needed to have blood redrawn for changes in leukocyte counts at any time in the program.