A Randomized, Double-Blind, Placebo-Controlled Study of Risperidone Maintenance Treatment in Children and Adolescents With Disruptive Behavior Disorders

Children and adolescents (ages 5–17 years) without moderate or severe intellectual impairment (IQ ≥55, as determined with assessments [11–13] given at screening or within the preceding 3 years) were eligible for enrollment if they met DSM-IV criteria for conduct disorder, oppositional defiant disorder, or disruptive behavior disorder not otherwise specified, with the diagnosis confirmed by the Schedule for Affective Disorders and Schizophrenia for School-Age Children–Present and Lifetime Version (K-SADS-PL). The K-SADS-PL has been validated in children ages 7–17 years, with excellent reliability for diagnosing disruptive behavior disorders (14). Inclusion required that the conduct problem be serious enough to warrant clinical treatment with risperidone and be associated with a score ≥24 on the conduct problem subscale of the Nisonger Child Behavior Rating Form—parent version (15, 16) at both screening and treatment initiation. Children and adolescents with comorbid ADHD were not excluded, but those with other serious medical or psychiatric conditions such as schizophrenia or bipolar disorder were excluded. Concomitant therapy with stable psychostimulant dosing was permitted (i.e., patients must have been receiving a stable dose of psychostimulants for at least 30 days before study entry and that dose must have been maintained by the clinician). Treatment with additional antipsychotics, lithium, anticonvulsants, or antidepressants was not permitted. Medications to treat extrapyramidal symptoms were allowed only after dose reduction was attempted but were generally discouraged. In addition, each patient was required to have a responsible caregiver to provide assessments and ensure medication compliance. After provision of a complete description of the study, written informed consent was obtained from the parents or guardians of each patient before study enrollment and assent was gained from every patient.

Before treatment, all patients were evaluated with standardized and validated measures of disruptive behavior disorder symptoms and general function, which were translated and back-translated into the languages of the various trial centers. These were the Nisonger Child Behavior Rating Form (in particular, the conduct problem subscale, which includes items such as “explosive, easily angered,” “knowingly destroys property,” and “threatens people”), the Clinical Global Impression (CGI) severity scale (17), the Children’s Global Assessment Scale (18), and a visual analogue scale for rating the most troublesome symptom. In addition, at day 7 from the start of treatment and at subsequent assessments, CGI change ratings were obtained. Cognitive function was assessed with a continuous performance test (10) and a modified version (19) of the California Verbal Learning Test–Children’s Version (MVLT-10 for patients with chronological or mental age <6, MVLT-15 for those with chronological or mental age ≥6). All patients also underwent a physical examination (including assignment of Tanner stage), laboratory testing (hematology, chemistry, and urinalysis), and ECG.

Patients were treated with risperidone, 1 mg/ml oral solution, initially as a once-daily dose in the evening, but this could be divided into twice-daily dosing if indicated (e.g., if sedation or breakthrough symptoms occurred). Treatment was divided into three sequential phases: acute treatment (6 weeks of open-label risperidone), continuation treatment (6 weeks of single-blind risperidone), and maintenance (6 months of double-blind risperidone or placebo). The single-blind phase was employed to ensure that the patient and caregiver were unaware of the exact timing of the randomization to risperidone or placebo. Patients and caregivers were informed, before the onset of the trial, that the randomization would take place at some point during the second 6-week treatment period (continuation phase). During the acute treatment phase, patients were treated with flexible risperidone doses depending on their body weight. Patients with a body weight up to 50 kg started at 0.25 mg/day, and those with a body weight of 50 kg or more were treated initially with 0.50 mg/day. If deemed necessary by the treating physician, this dose was increased by 0.25 mg (for patients <50 kg) or 0.50 mg (those ≥50 kg) on the third and fifth day of treatment to a maximum daily dose of 0.75 mg (patients <50 kg) or 1.5 mg (those ≥50 kg). Response during acute treatment was defined as achieving either ≥50% reduction in score on the Nisonger Child Behavior Rating Form conduct problem subscale or a score ≤17 on the conduct problem subscale with a score of 1 or 2 (“much” or “very much” improved) on the CGI change scale. Responders from the acute treatment phase entered into the continuation treatment phase and remained at the dose level used in acute treatment. Dose changes to manage adverse events were allowed during this phase, but every clinically acceptable attempt had to be made to keep the dose constant. Sustained response was defined as having at week 12 no deterioration beyond 3 points on the Nisonger Child Behavior Rating Form conduct problem subscale compared with the end of the acute phase (week 6). Also required was a score <24 on the conduct problem subscale at weeks 8 and 10 and exhibited improvement compared with screening. Patients exhibiting sustained response during continuation treatment were randomly assigned in a 1:1 ratio to one of two maintenance phase conditions: risperidone at the same dose used in the continuation phase or placebo. Maintenance therapy continued for up to 6 months. There was no down-titration of medication for those switched to placebo. The randomization code was generated by the study sponsor, with treatment numbers allocated at each investigative center in chronological order. Placebo and risperidone oral solutions were identical in appearance and flavor.

The primary efficacy measure was time to symptom recurrence, defined during the double-blind maintenance phase as a deterioration of  ≥2 points on the CGI severity scale or ≥7 points on the conduct problem subscale at two consecutive visits 6–8 days apart. The CGI severity scale was chosen as a measure for symptom recurrence rather than the CGI change scale because time has less of an impact on the CGI severity scale, given that the rater is not required to recall the condition at baseline. In addition, the CGI severity scale was considered to be less subject to variability introduced by potential changes in raters. As soon as patients met the criteria for symptom recurrence (the second confirmatory visit), they were discontinued from the trial. Secondary efficacy measures included rates of discontinuation due to symptom recurrence, and change from screening or baseline on the Nisonger Child Behavior Rating Form subscales, CGI severity and change scales, and visual analogue scale rating of the most troublesome symptom, all assessed biweekly during acute and continuation treatment and monthly during maintenance treatment. The Children’s Global Assessment Scale score was recorded at completion of the continuation and maintenance treatment phases.

Safety and tolerability were assessed by recording the occurrence of spontaneously reported adverse events. Treatment-emergent adverse events (i.e., new in onset or changed in severity) were tabulated separately for each phase of the study. Cognitive function, laboratory values, ECG, and vital signs were measured at screening and the completion of the continuation and maintenance treatment phases. A physical examination, including assignment of Tanner stage, was performed at screening and completion of maintenance treatment. Height and weight were measured at screening, on completion of acute and continuation treatment phases, and after 3 and 6 months of maintenance treatment. To place weight changes into the context of normal growth, results at each visit for height, weight, and body mass index were transformed to z scores using the SAS code available at http://www.cdc.gov/growthcharts. These z scores indicate the extent that a child’s weight, height, or body mass index differs from the average for children of the same age and sex, using a U.S. reference population.

A power analysis identified that 208 patients needed to participate in maintenance therapy to achieve a 90% power for detecting a difference in time to symptom recurrence between treatment groups. Accounting for anticipated dropouts with no postbaseline data, 225 patients were required to begin maintenance therapy, necessitating at least 364 patients for acute treatment, since 65% of patients were expected to respond during the acute phase, and 95% of these were expected to fulfill the requirements for randomization.

Descriptive statistics were used to evaluate demographic and baseline data as well as change from screening and baseline for the total and subscale scores of the Nisonger Child Behavior Rating Form, visual analogue scale ratings of the most troublesome symptom, CGI change and severity ratings, and Children’s Global Assessment Scale score. Screening occurred in the 7 days before enrollment in acute therapy. Baseline was considered to be initiation of double-blind maintenance treatment. Efficacy analyses were evaluated for the intent-to-treat dataset, defined as all patients who received at least one dose of placebo or risperidone during the double-blind maintenance phase. A Cox proportional hazard model was used to compare time to symptom recurrence between treatment groups. The Cochran-Mantel-Haenszel test for general association, with country controlled, was used for assessing symptom recurrence rate. All statistical tests were interpreted at the 5% significance level (two-tailed). Safety analyses were performed for all patients who received at least one dose of study drug. Prevalence of adverse events and change from screening and baseline in vital signs, laboratory tests, ECG, and cognitive tests were summarized with descriptive statistics. Changes from screening and baseline in height and weight (transformed into z scores) were also summarized by descriptive statistics.

A total of 575 patients were screened and of these, 527 patients were enrolled in the 6-week, open-label, acute treatment phase. Overall, 436 patients continued single-blind treatment for an additional 6 weeks, and 335 were randomly assigned to 6 months of double-blind maintenance treatment (Figure 1). Patient demographics are shown in Table 1. The majority of patients were Caucasian (87%, N=459) and of normal (IQ>84) intellect (63%, N=333); a significant proportion (44%, N=231) were adolescents. Most (66%, N=349) had comorbid ADHD.

The average Nisonger Child Behavior Rating Form conduct problem subscale score of 35.2 (SD=6.29) at screening indicates that patients had severe symptoms. The symptom rated most often with the visual analogue scale as the most troublesome symptom was categorized as either aggression (46%) or oppositional defiant behavior (43%). Mean Children’s Global Assessment Scale score at screening was 47.0 (SD=9.9), corresponding to a moderate interference in most social areas or severe impairment in one area.

Treatment was discontinued before entering the double-blind maintenance phase by 192 (36%) patients, with most discontinuations due to patients not meeting the criteria for either initial response (10%, N=52) or sustained response (12%, N=61). Other reasons for discontinuation during the acute and continuation phases were adverse events (N=12), patient lost to follow-up (N=20), patient withdrawal of consent (N=15), patient noncompliance (N=14), patient ineligibility for study continuation (N=4), and other reasons (N=14). Double-blind maintenance treatment was completed by 162 of the patients who started this phase (48%), with most discontinuations due to meeting symptom recurrence criteria. Mean risperidone dosage was approximately 0.02 mg/kg per day (Table 2). Prior therapy was used by 351 (67%) of patients enrolled into the acute treatment phase, most commonly psychostimulants (39%), antipsychotics (26%), and anticonvulsants (7%). During acute and continuation treatment phases, 59% of all patients received concomitant medications (24% psychostimulants, 16% analgesics). During the maintenance phase, concomitant medications were taken by 55% of risperidone-treated patients (21% psychostimulants, analgesics 15%) and 50% of placebo-treated patients (22% psychostimulants, 12% analgesics).

At the end of the maintenance phase, there was a significant difference in rate of symptom recurrence, based on efficacy scales and investigator impression, between those assigned to risperidone (27.3%, N=47) and those assigned to placebo (42.3%, N=69) (c²=10.04, df=1, p=0.002). Time to symptom recurrence was significantly shorter with placebo compared with maintenance risperidone (c²=18.45, df=1, p<0.001) (Figure 2). Symptom recurrence occurred in 25% of patients after 119 days with risperidone and 37 days with placebo. Six-month Kaplan-Meier symptom recurrence estimates were 29.7% for risperidone and 47.1% for placebo. The instantaneous risk-to-symptom recurrence (hazard ratio) was 2.24 (95% CI=1.54–3.28) times higher after switching to placebo compared with continuing risperidone. Since less than 50% of patients met the criteria for symptom recurrence, the median time to symptom recurrence was not estimable. For patients who discontinued due to symptom recurrence, the range of time to symptom recurrence was 12–159 days among those receiving placebo and 14–181 days for those receiving risperidone. Risk for symptom recurrence was not affected by age, gender, diagnosis, or baseline disruptive behavior disorder severity.

Risperidone was associated with a significantly lower rate of symptom recurrence than placebo (27.3% versus 42.3%) (c²=10.04, df=1, p=0.002). In the intent-to-treat population, conduct problem scores decreased (i.e., condition improved) significantly within 2 weeks of treatment initiation and continued to improve further throughout acute and continuation treatment (Figure 3). During the maintenance phase, mean scores increased compared with the start of the maintenance phase in both groups, but deterioration at endpoint of the maintenance phase was significantly higher in the placebo group (F=13.17, df=1,321, p<0.001). Change from the beginning of the maintenance phase to endpoint for all Nisonger Child Behavior Rating Form subscales, the most troublesome symptom visual analogue scale rating, and the global measurements (CGI severity and Children’s Global Assessment Scale) favored risperidone, with significant differences (p≤0.01) between risperidone and placebo on all measures except the insecure/anxious, self-injury/stereotypic behavior, self-isolated/ritualistic, and overly sensitive subscales of the Nisonger Child Behavior Rating Form. The changes in secondary efficacy measures in the intent-to-treat population from screening to the end of the continuation phase and from the beginning of the maintenance phase until endpoint are presented in Table 3.

Patients maintained on risperidone for up to 9 months had significant improvements (by two decile categories) in mean Children’s Global Assessment Scale scores, corresponding to a change from moderate or severe impairment to generally functioning well.

Safety data were available for all 527 patients enrolled in the acute treatment phase. Treatment-emergent adverse events were reported more frequently during acute treatment (54.8%) compared with the continuation phase (34.9%) and maintenance phase (47.7% with risperidone versus 36.2% with placebo) (Table 4). The most frequently reported treatment-emergent adverse events were headache, somnolence, fatigue, and increased appetite. Most treatment-emergent adverse events were mild or moderate in severity. Treatment-emergent adverse events resulted in treatment discontinuation in 1.7% of patients during acute treatment, 2.1% during the continuation phase, and 1.7% with risperidone and 0.6% with placebo during the maintenance phase. Treatment-emergent adverse events that resulted in treatment discontinuation during the maintenance phase with risperidone were involuntary muscle contractions, abnormal ECG, and paranoid reaction, each of which occurred in one patient. One patient receiving placebo discontinued because of an implantation complication related to treatment for a different condition. The incidence of extrapyramidal symptoms was low (Table 5), and antiextrapyramidal-symptom medications were used to treat extrapyramidal symptoms in only two patients. There were no reports of tardive dyskinesia. Potentially prolactin-related treatment-emergent adverse events were infrequent and occurred during acute and continuation therapy in 10 patients (six male [gynecomastia] and four female [amenorrhea in one, breast pain and lactation in one, and lactation only in two]) and during risperidone maintenance in three male subjects (gynecomastia) and two female subjects (amenorrhea and lactation in one, and breast discharge in one). There were no potentially prolactin-related treatment-emergent adverse events among those receiving placebo during the maintenance phase. Somnolence was most commonly reported during the acute phase of treatment (11.6%) and dramatically decreased thereafter. In general, somnolence was of mild-to-moderate severity and time-limited to treatment initiation or dose increase.

There was no clinically significant change in mean fasting glucose levels during treatment. Neither were there clinically meaningful changes in other laboratory values with the exception of prolactin plasma levels. Mean prolactin plasma levels increased during acute and continuation phases from 8.3 ng/ml (SD=6.6) at screening to 27.2 ng/ml (SD=20.1) at the end of the continuation phase. In the intent-to-treat population, mean levels decreased during the maintenance phase both for patients treated with risperidone (baseline: 29.4 ng/ml [SD=21.9]; endpoint: 20.3 ng/ml [SD=21.3]) and those receiving placebo (baseline: 29.7 ng/ml [SD=18.0]; endpoint: 9.6 ng/ml [SD=9.5]).

Mean weight increased from screening to the end of the continuation phase (week 12) by 3.2 kg (SD=2.49). Subsequently, from the beginning to the end of the maintenance phase, mean weight increased by 2.1 kg (SD=2.7) for risperidone-treated patients, whereas patients receiving placebo experienced a decrease in mean weight of –0.2 kg (SD=2.2). Mean height and body mass index increased over the first 12 weeks of treatment by 1.7 cm (SD=2.1) and 1.0 kg/m² (SD=1.1), respectively. From the beginning to the end of the maintenance phase, height increased in both groups (risperidone: mean=2.5 cm, SD=1.9; placebo: mean=1.5 cm, SD=1.7), and body mass index increased in risperidone-treated patients (0.3 kg/m² [SD=1.1]) but decreased for patients switched to placebo (–0.4 kg/m² [SD=0.9]). Among patients receiving risperidone in all study phases, mean weight and body mass index z scores increased during the first 12 weeks of treatment (weight: mean change=0.3 [SD=0.3]; body mass index: mean change=0.3 [SD=0.4]). During maintenance treatment no further weight gain beyond natural growth was observed (change in weight z scores from maintenance baseline: mean=0.0, SD=0.3, at endpoint and mean=0.1, SD=0.3, at other time points). In placebo patients, mean weight z scores decreased compared with maintenance baseline by 0.1 (SD=0.2). Mean body mass index z scores showed a similar pattern, with no increase in risperidone patients (mean change from baseline to endpoint=0.0, SD=0.4) and a decrease in placebo patients (mean change=–0.2, SD=0.3). Figure 4 indicates that weight increase was an initial effect, was not continued with prolonged treatment, and was reversed with discontinuation of risperidone treatment.

There were no clinically significant changes in vital signs and no significant changes in QT intervals during any treatment phases. Most mean changes in ECG parameters were small, and the incidences of high heart rate (>100 bpm) and low PR interval (≤120 msec) during the double-blind maintenance phase were similar between the risperidone and placebo groups.

Changes in cognitive testing scores from beginning to end of the double-blind maintenance phase were small and similar for the risperidone and placebo groups. Mean changes on the MVLT-15 (total correct scores) were 0.1 (SD=1.7) for placebo (N=139) and –0.3 (SD=3.8) for risperidone (N=138); on the MVLT-10 these were 4.3 (SD=4.9) for placebo (N=5) and 2.0 (SD=2.1) for risperidone (N=9). The scores on the continuous performance test were also similar between placebo- and risperidone-treated patients (mean change in number of correct hits: placebo, mean=0.1 [SD=0.4], N=136; risperidone, mean=0.1 [SD=0.4], N=139). These differences were not statistically significant.

As expected, Tanner stages increased during the study, with similar changes in patients receiving risperidone and placebo during the maintenance phase (Figure 5).