JOURNAL OF CHILD AND ADOLESCENT PSYCHOPHARMACOLOGY

Volume 21, Number 3, 2011

ª Mary Ann Liebert, Inc.
Pp. 229–236
DOI: 10.1089/cap.2009.0121

Aripiprazole in the Treatment of Irritability in Pediatric

Patients (Aged 6–17 Years) with Autistic Disorder:
Results from a 52-Week, Open-Label Study

Ronald N. Marcus, M.D.,1 Randall Owen, M.D.,2 George Manos, Ph.D.,1

Raymond Mankoski, M.D., Ph.D.,3 Lisa Kamen, M.H.A.,1 Robert D. McQuade, Ph.D.,4
William H. Carson, M.D.,4 Patricia K. Corey-Lisle, Ph.D.,1 and Michael G. Aman, Ph.D.5

Abstract
Aim: To report the long-term efficacy of aripiprazole in the treatment of irritability in children and adolescents (ages 6–17
years) with autistic disorder.
Methods: This was a 52-week, open-label, flexible-dose (2–15 mg/day) study of aripiprazole for the treatment of children and
adolescents with irritability associated with autistic disorder. Eligible subjects were enrolled from two 8-week randomized
trials or were enrolled as de novo subjects. ‘‘Prior aripiprazole’’ subjects had received treatment with aripiprazole for 8 weeks
before entering this study. Evaluation of efficacy, a secondary objective after evaluation of safety and tolerability in this study,
was conducted using the caregiver-rated Aberrant Behavior Checklist–Irritability subscale and the clinician-rated Clinical
Global Impression–Improvement score.
Results: Three hundred thirty subjects received treatment (de novo, n ¼ 86; prior aripiprazole, n ¼ 174; prior placebo, n ¼ 70)
and 199 subjects (60.3%) completed 52 weeks of treatment. At their last study visit, 38.2% of subjects were receiving
concomitant central nervous system medications (commonly antidepressants, 13.4%; psychostimulants, 11.5%; antiepi-
leptics, 5.9%). At week 52 (observed cases data set), the mean change from baseline in Aberrant Behavior Checklist
Irritability subscale scores was 8.0 in de novo subjects and 6.1 in prior placebo subjects; prior aripiprazole subjects
maintained symptom improvement that was achieved with treatment in the prior study. At endpoint, the majority of subjects
had a Clinical Global Impressions–Improvement score of 2 (much improved) or 1 (very much improved).
Conclusion: Aripiprazole reduced symptoms of irritability associated with autistic disorder in pediatric subjects ages 6–17
years who were studied for up to 1 year.

Introduction munication, or symbolic or imaginative play must have an onset

before 3 years of age for a diagnosis to be made (American Psy-

A utistic disorder is characterized by impairments in social

interaction and communication, as well as restricted, repeti-
tive, and stereotyped patterns of behavior, activities, or interests.
chiatric Association 2000). Understandably, these symptoms can
have a substantial impact on the individuals and their families. This
impact can be further increased by the presence of associated be-
To qualify for a Diagnostic and Statistical Manual of Mental haviors such as irritability, which may manifest as tantrums, ag-
Disorders, Fourth Edition, diagnosis of autistic disorder, individ- gressiveness, self-injurious behavior, and sudden mood changes, all
uals must have qualitative impairment in at least six items, with at of which can have a significant impact on education and social
least two social interaction items, one communication item, and development (Volkmar et al. 1999).
one restricted, repetitive, and stereotyped patterns of behavior item Although there are no approved pharmacologic treatments that
(American Psychiatric Association 2000). These core symptoms of target the core deficits of autistic disorder, associated secondary
autistic disorder often become apparent during early childhood, and symptoms such as irritability may be ameliorated by a combination
delays or abnormal functioning in social interaction, social com- of behavioral and pharmacologic approaches, including the use of

Clinical trials information: NCT00365859; registry: www.clinicaltrials.gov

1
Bristol-Myers Squibb, Wallingford, Connecticut.
2
Formerly, Bristol-Myers Squibb, Wallingford, Connecticut.
3
Bristol-Myers Squibb, Plainsboro, New Jersey.
4
Otsuka Pharmaceutical Development and Commercialization, Inc., Princeton, New Jersey.
5
The Nisonger Center UCEDD, The Ohio State University, Columbus, Ohio.

229
230 MARCUS ET AL.

atypical antipsychotics (Myers and Johnson 2007). Risperidone This study enrolled three groups of subjects: (a) subjects who
and aripiprazole are approved by the U.S. Food and Drug were previously randomized to placebo in one of two 8-week
Administration for the treatment of pediatric patients with irrita- multicenter, randomized, double-blind, placebo-controlled trials
bility associated with autistic disorder, including symptoms of (prior placebo subjects); (b) subjects who were previously ran-
aggression toward others, self-injuriousness, temper tantrums, and domized to aripiprazole in one of the 8-week trials (prior ar-
quickly changing moods (Aripiprazole 2009; Risperidone 2009). ipiprazole subjects); and (c) de novo subjects who had not been
Recently, aripiprazole was shown to be efficacious and generally enrolled in these previous studies.
safe and well tolerated for the treatment of children and adolescents
(ages 6–17 years) with irritability associated with autistic disorder in Prior aripiprazole and prior placebo subjects. Subjects
two 8-week, multicenter, randomized, double-blind, placebo- entering open-label treatment from the previously completed trials
controlled studies (Marcus et al. 2009; Owen et al. 2009). These had all met inclusion criteria specified in the original trials. All
studies were similar in design; one was a fixed-dose study in which subjects were required to have a Clinical Global Impressions–
subjects were randomized to receive aripiprazole (5, 10, or 15 mg/ Severity (CGI-S) score  4 and a caregiver-rated ABC (Aman et al.
day) or placebo (Marcus et al. 2009), and the other was a flexible- 1985a, 1985b) Irritability subscale score  18 at baseline. CGI-S
dose study in which subjects were randomized to receive ratings were focused on the severity of irritability (e.g., tantrums,
aripiprazole (2–15 mg/day, titrated to clinical effect, with dose op- aggression, self injury, and quickly changing moods) associated
tions of 5, 10, or 15 mg/day) or placebo (Owen et al. 2009). In the with autistic disorder, as this was the target symptom for treat-
fixed-dose study, all aripiprazole doses produced significantly greater ment. The ABC Irritability subscale is a 15-item subscale that in-
improvement than placebo in mean Aberrant Behavior Checklist– cludes items describing symptoms of irritability, such as temper
Irritability (ABC-I) subscale scores (5 mg/day, 12.4; 10 mg/day, tantrums, aggression, mood changes, and self-injury. Subjects who
13.2; 15 mg/day, 14.4; vs. placebo, 8.4; all p < 0.05) at week 8; completed either previous double-blind study were eligible if
in the flexible-dose study, aripiprazole also produced significantly continuation of treatment was warranted, as judged by the inves-
greater improvement than placebo (12.9 vs. 5.0; p < 0.001). tigator, and they had no significant protocol violations or clinically
Although these 8-week studies demonstrated the short-term ef- relevant adverse events that would preclude their inclusion in the
ficacy, safety, and tolerability of aripiprazole, there remains a need study.
for longer-term data to inform clinical decision making. As such, a
52-week, open-label, multicenter study was conducted to evaluate
De novo subjects. De novo subjects who had not been en-
the long-term safety and tolerability of flexibly dosed aripiprazole
rolled in one of these placebo-controlled studies were also per-
for the treatment of irritability associated with autistic disorder. The
mitted to enroll in this open-label study. Although de novo subjects
safety and tolerability findings, which were the primary objective
had to demonstrate current behavioral problems or a history of
of this study, are reported in detail elsewhere (Marcus et al. In
behavioral problems that were currently being treated with psy-
press). Here we report the efficacy findings from this study, which
chotropic medication, there was no requirement of a minimum
were included as secondary objectives.
ABC-I or CGI-S score. Study sites were not permitted to enroll
de novo subjects in this study until they had completed enrollment
Methods
for the randomized, controlled studies, unless approved by the
Study design and subjects study sponsor. De novo subjects with a history of serious behavioral
problems who were being treated with aripiprazole and were doing
This was an open-label, multicenter, 52-week study conducted at
well were eligible for enrollment if, in the investigator’s clini-
53 sites in the United States between September 2006 and June
cal judgment, they would benefit from study participation. For
2009 in accordance with the Declaration of Helsinki and in ac-
these subjects (n ¼ 6) prior aripiprazole was discontinued at least 9
cordance with Good Clinical Practice as defined by the Interna-
days (usually longer) before baseline assessment and entry into the
tional Council on Harmonization. The Institutional Review Board
current trial.
at each site approved the protocol. All parents/guardians provided
written informed consent to participate, and subjects provided
written, informed assent when possible. Study treatments
Subjects were children and adolescents, aged 6–17 years, with a Aripiprazole treatment. All subjects received open-label
diagnosis of autistic disorder and with behavioral problems such as treatment with flexibly dosed (2–15 mg/day) aripiprazole (oral
irritability, agitation, self-injurious behavior, or a combination of tablet) taken once daily at the same time each day without regard to
these symptoms. Diagnostic and Statistical Manual of Mental meals. Aripiprazole was initiated at 2 mg/day, with a target dose of
Disorders, Fourth Edition, Text Revision, criteria for autistic dis- 5, 10, or 15 mg/day (maximum dose, 15 mg/day). All subjects, both
order were confirmed using the Autism Diagnostic Interview– de novo and rollover, who participated in the study started therapy
Revised (ADI-R) (Lord et al. 1994), which was administered by an with aripiprazole at 2 mg/day. All rollover subjects had study
experienced interviewer who had been previously trained and ap- medication re-titrated starting at 2 mg/day at entry into open-label
proved as ‘‘Research Reliable’’ on the ADI-R or who had been treatment, as both subjects and clinicians were blinded to treatment
approved after successful completion of a 2-day rater training assignment in the previous double-blind study. Rollover subjects
course conducted by an ADI-R-certified trainer. took their first dose of study medication during open-label treat-
Exclusion criteria included a current diagnosis of bipolar dis- ment (day 1) on the day after their week 8 visit of the antecedent
order, psychosis, schizophrenia, or major depression, Fragile-X double-blind studies. All dose increases, if deemed appropriate by
Syndrome, or diagnosis of another disorder on the autism spec- the investigator based on the patient’s clinical response and med-
trum (pervasive developmental disorder–not otherwise specified, ication tolerability, were made no more frequently than every 4
Asperger’s disorder, Rett’s disorder, or childhood disintegrative days and were incremental from the current dose level to the next
disorder). (dose levels were 2, 5, 10, or 15 mg/day). Dose adjustments could
ARIPIPRAZOLE EFFICACY IN AUTISTIC DISORDER 231

be made at any time based on efficacy and tolerability at the current instrument assessing the impact that caring for children and ado-
dose. lescents with serious emotional, mental, and behavioral disturbances
has on the family over the preceding 3 months. The PedsQL-com-
Prohibited concomitant therapies. Clonidine, guanfacine, bined scales total score was used in this study; this was calculated as
guanabenz, carbamazapine, oxcarbazepine, and antipsychotics the sum on the PedsQL emotional (5 items), social functioning (5
other than aripiprazole, that is, medications with the greatest po- items), and cognitive functioning (6 items from the PedsQL Multi-
tential to impact the assessment of the effects of aripiprazole for dimensional Fatigue scale) subscale items divided by the number of
the treatment of irritability associated with autistic disorder, were items answered. Scores are presented on a 0–100 scale, where higher
expressly prohibited during open-label treatment. De novo subjects scores indicate better health-related quality of life. The CGSQ
receiving these medications were washed out at least 4 days before Global score is presented as the sum of the three subscales, which
the interim screening visit. Subjects receiving long-acting (depot) were calculated as the averages of the corresponding individual
neuroleptics were discontinued for at least 1 full cycle plus 1 week items. CGSQ Global scores are on a 0–15 scale, where higher scores
before interim screening visit. Prior aripiprazole and prior placebo indicate a higher degree of caregiver strain.
subjects underwent medication washout of these agents before Subjects had study visits at the end of weeks 1, 2, and 4 (2 days
participating in the previously conducted trials. from the baseline visit) and at the end of weeks 8, 14, 20, 26, 34, 42,
and 52 (7 days from the end of baseline visit). CGI-S and CGI-I
Permitted concomitant therapies. Subjects were permitted ratings were obtained at all study visits; ABC-I was evaluated at
to receive concomitant medications for the treatment of comorbid week 4 and all visits thereafter. All other efficacy evaluations were
psychiatric disorders. As such, psychostimulants, antidepressants, assessed at baseline and endpoint.
anxiolytics, and mood stabilizers were permitted during open-label All analyses were conducted using the efficacy sample, which
treatment unless contraindicated or likely to result in an adverse included all patients who received at least one dose of study
drug interaction with aripiprazole. Although these concomitant medication during open-label treatment and who had at least one
psychotropic medications were prohibited or restricted during the postbaseline efficacy evaluation and corresponding baseline value
previous placebo-controlled studies, a wider spectrum of concom- (if applicable).
itant treatments was permitted in this open-label study to improve Three subject groups based on previous treatment were defined
patient retention and more accurately reflect the actual medication- for the purpose of the analyses presented herein: (a) prior ar-
use patterns in this population. ipiprazole subjects, (b) prior placebo subjects, and (c) de novo
In this open-label study, and the previous double-blind studies, subjects. For each subgroup, descriptive statistics were calculated
subjects were permitted to receive antianxiety medication (such as for the mean change from baseline for each of the efficacy evalu-
benzodiazepines) and sleep aids (limited to diphenhydramine, non- ations, or for mean score in the case of CGI-I. For the purpose of
benzodiazepine hypnotics such as zolpidem or eszopiclone or analysis, all baseline values were derived from week 0 (open-label
melatonin) at the investigator’s discretion. Diphenhydramine could treatment baseline visit) for de novo subjects or the last evaluation
also be given for serious behavior problems (up to 50 mg/day), and during double-blind treatment for prior aripiprazole/placebo sub-
psychotropic medication was permitted for the acute treatment of jects. The exception to this was CGI-I scores for prior aripiprazole/
unforeseen medical events. Anticholinergic therapy (benztropine) placebo subjects in which improvement in symptoms was rated
or propranolol was permitted for the treatment of extrapyramidal relative to symptom baseline before receiving dosing in the double-
symptoms, although administration of movement rating scales was blind treatment phase of the antecedent study.
not to occur if these medications had been received in the last 12 As this was an open-label study, with no placebo or active
hours. control group, no formal statistical testing was undertaken; all re-
Nonpharmacologic therapy (e.g., psychotherapy and behavior sults are reported as descriptive statistics. Summary statistics are
modification) was permitted provided it was stable before screen- presented, including mean and standard deviation for continuous
ing and consistent throughout the study. variables, and frequency and percent frequency for categorical
variables. Of note, prior aripiprazole subjects’ baseline measure-
ments were taken at the last evaluation during double-blind treat-
Assessments and statistical analyses
ment after having received aripiprazole for 8 weeks. As such, mean
Evaluating the long-term efficacy of flexibly dosed aripiprazole changes in efficacy scores in this group were expected to be small
for the treatment of irritability associated with autistic disorder was over the course of long-term treatment and would merely indicate
the secondary objective of this study. Efficacy was evaluated using that previous symptom control was maintained.
the caregiver-rated ABC, which has five subscales designated as
Irritability (15 items), Lethargy/Social Withdrawal (16 items), Results
Stereotypic Behavior (7 items), Hyperactivity (16 items), and In-
Subject disposition and demographics
appropriate Speech (4 items). Each individual item on the ABC is
scored on a Likert scale from 0 (not at all a problem) to 3 (the In total, 330 subjects entered open-label treatment in this study.
problem is severe in degree). Ninety-five percent (n ¼ 70/74) of placebo subjects and 97%
Participants were also assessed on the clinician-rated CGI- (n ¼ 174/179) of aripiprazole subjects completing the short-term
Improvement (CGI-I) score and CGI-S score (Guy 1976) and the study chose to enter open-label treatment, giving a total of 244
clinician-rated Children’s Yale–Brown Obsessive Compulsive subjects who had previously participated in either of the two 8-
Scale (CY-BOCS) (Scahill et al. 1997) (compulsion scale only). week studies. Eighty-six de novo subjects entered the open-label
Quality of life assessments included the caregiver-rated Pediatric treatment phase; an additional 23 were screened but did not meet
Quality of Life Inventory (PedsQL) (Varni et al. 2001, 2002), entrance criteria. Six de novo subjects had received aripiprazole
which rates health-related quality of life, and the Caregiver Strain treatment before study enrollment. The duration of aripiprazole
Questionnaire (CGSQ) (Brannan et al. 1997), a 21-item self-report ranged from 24 to 317 days in three subjects; in the remaining
232 MARCUS ET AL.

Table 1. Subject Disposition

n (%) De novo (n ¼ 86) Prior placebo (n ¼ 70) Prior aripiprazole (n ¼ 174)

Completed 55 (64.0) 37 (52.9) 107 (61.5)

Discontinued 31 (36.0) 33 (47.1) 67 (38.5)
Lack of efficacy 8 (9.3) 5 (7.1) 7 (4.0)
Adverse event 9 (10.5) 11 (15.7) 15 (8.6)
Withdrew consent 7 (8.1) 5 (7.1) 15 (8.6)
Lost to follow-up 2 (2.3) 8 (11.4) 21 (12.1)
Poor/noncompliance 2 (2.3) 1 (1.4) 2 (1.1)
Administrative reason by sponsor 1 (1.2) 1 (1.4) 2 (1.1)
Other 2 (2.0) 2 (2.9) 5 (2.9)

three subjects the exact duration of prior aripiprazole treatment was n ¼ 64 (32.0%); 15 mg/day, n ¼ 90 (45.0%) (based on the most
unknown. Prior doses received were aripiprazole 2 mg/day (n ¼ 2), frequent dose received during week 52; in case of ties, the highest
5 mg/day (n ¼ 2), and 10 mg/day (n ¼ 1). One subject received dose was used).
aripiprazole 10 mg/day with a dose reduction to aripiprazole 5 mg/
day. All six subjects discontinued aripiprazole at least 9 days, Concomitant medications
usually longer, before baseline assessments and treatment.
At their last study visit, 38.2% of subjects were receiving con-
Subject disposition during the 52-week, open-label, multicenter
comitant CNS medications; the most commonly used medica-
study is shown in Table 1. The 52-week, open-label treatment pe-
tions included antidepressants (13.4%), psychostimulants (11.5%),
riod was completed by 60.3% of subjects. Completion rates were
and antiepileptics (5.9%). For subjects who completed the study
broadly similar among subject subgroups. The most common rea-
(n ¼ 199), the most commonly used CNS medications (excluding
son for discontinuation was adverse events (10.6% of all subjects);
analgesics and antipyretics) were anxiolytics (20.1%), antidepres-
discontinuation due to adverse events was lower in prior ar-
sants (16.6%), psychostimulants (15.1%), antiepileptics (6.5%),
ipiprazole subjects (8.6%) than in prior placebo (15.7%) or de novo
and anticholinergics (6.0%).
subjects (10.5%). In all, 6.1% of all subjects discontinued due to
Fifteen subjects (4.5%) took concomitant medication for the
lack of efficacy. Additional reasons for discontinuation, generally
treatment of EPS at some point during the study. These included
unrelated to study medication, included loss to follow-up (9.4% of
anticholinergics (3.9%), beta-blocking agents (0.9%), and dopa-
all subjects), withdrawal of consent (8.2%), inadequate compliance
minergic agents (amantadine; 0.3%). In addition, 114 subjects
(1.5%), administrative reasons (1.2%), and other (2.9%). A total of
(34.5%) received concomitant antihistamine medication (including
330 subjects received at least one dose of study medication, of
diphenhydramine) during the study; most commonly for the treat-
whom 322 (97.6%) had at least one postbaseline efficacy assess-
ment of adverse events (19.4%).
ment; these were included in the efficacy sample.
Subject baseline demographic characteristics have been pre-
sented previously (Marcus et al. In press) and were similar among Efficacy assessments
subgroups. The mean age for all subjects was 9.6 years and the
Prior placebo and de novo subjects. Baseline ABC-I and
majority of subjects in all subgroups (79.7%) were aged 6–12 years.
CGI-S scores were slightly higher in de novo subjects than prior
The majority of subjects were male (87.0%) and Caucasian
placebo subjects. The mean change in ABC Irritability subscale
(71.2%). Before study entry, 57.9% of subjects had received prior
scores by study week for subjects continuing with aripiprazole
central nervous system (CNS) medications (de novo, 77.9%; prior
treatment is shown in Figure 1. For prior placebo subjects and
placebo, 44.3%; prior aripiprazole 53.4%). The most commonly
de novo subjects, improvement in ABC-I scores occurred early in
used CNS medications (>10% of subjects) before study enrollment
the course of treatment and was maintained to study endpoint. At
were antipsychotics (24.5%), psychostimulants (17.9%), anxiolyt-
week 52 (observed case [OC] data set), the mean change from
ics (17.6%), and antidepressants (16.7%). Baseline mean (standard
baseline ABC Irritability scores was 6.1 (11.9) and 8.0 (10.1) in
deviation [SD]) ABC Irritability subscale scores at the start of the
placebo and de novo subjects, respectively. Similar improve-
open-label treatment phase were 23.2 (8.9) in de novo subjects
ments in ABC Irritability scores were seen for the last observation
(n ¼ 80), 21.5 (9.8) in prior placebo subjects (n ¼ 68) and 15.0 (9.2)
carried forward (LOCF) data set at week 52 (Fig. 1). In the ante-
in prior aripiprazole subjects (n ¼ 166).
cedent placebo-controlled trials, the mean baseline ABC-I
scores for the subjects randomized to placebo (the ‘‘prior placebo’’
Aripiprazole dosing
subjects in this trial) were 26.9 and 30.8 in the fixed-dose and
Overall, 200 subjects received at least one dose of aripiprazole flexible-dose studies, respectively, and mean improvement in
during the last week of the study; the mean daily dose of ABC-I scores was 8.4 and 5.0 points after 8 weeks of double-blind
aripiprazole during the final week of treatment (week 52) in these treatment.
subjects was 10.6 mg/day (range: 1.1–15.0 mg/day). The mean The mean changes in CGI-S scores for these subgroups are
daily dose of aripiprazole on the last day of dosing for all subjects, shown in Table 2. Both prior placebo and de novo subgroups
including those who discontinued prematurely, was 9.6 mg/day showed mean improvements from baseline in CGI-S scores at week
(range: 1.1–15 mg aripiprazole daily). At week 52, the distribution 52 (OC) and endpoint (LOCF).
of aripiprazole dosing was as follows: 0 mg/day, n ¼ 1 (0.5%); Mean (SD) CGI-I scores showed that symptom improvement
2 mg/day, n ¼ 10 (5.0%); 5 mg/day, n ¼ 35 (17.5%); 10 mg/day, was maintained over the 52-week treatment period for prior
ARIPIPRAZOLE EFFICACY IN AUTISTIC DISORDER 233

De novo (n=48 [80]) Prior placebo (n=36 [68]) Prior aripiprazole (n=94 [166])

Mean change from baseline in ABC-I score

0

-2

-4

-6

-8

-10

-12
0 4 8 12 16 20 24 28 32 36 40 44 48 52 LOCF
Week

FIG. 1. Mean change in ABC Irritability subscale scores by week (OC data set, also showing LOCF endpoint), efficacy sample. ABC
Irritability subscale scores at baseline: de novo, 23.2 (8.9); prior placebo, 21.5 (9.8); prior aripiprazole, 15.0 (9.2). Prior aripiprazole
subjects had previously received aripiprazole for 8 weeks before open-label baseline assessments; thus, lower baseline values in this
group reflect less severe impairment at baseline and mean change scores represent the change during long-term treatment, subsequent to
improvement experienced during double-blind treatment. ABC ¼ Aberrant Behavior Checklist; LOCF ¼ last observation carried for-
ward; OC ¼ observed case; n ¼ Week 52 OC patient numbers (week 52 LOCF patient numbers).

placebo (2.4 [1.2]) and de novo (2.7 [1.3]) subjects (week 52; aripiprazole group at week 52 (LOCF), indicating that symptom
LOCF). At endpoint, the majority of prior placebo and de novo improvement was maintained. Distribution of CGI-I scores (Fig. 2)
subjects had a CGI-I score of much improved or very much im- showed that the majority of prior aripiprazole subjects had a CGI-I
proved compared with their condition before treatment. score of much improved or very much improved at endpoint.
Mean (SD) change from baseline in additional efficacy out- For all ABC subscales, open-label baseline values for the
comes at week 52 (OC) and endpoint (LOCF) is also shown in prior aripiprazole group were lower than for the prior placebo and
Table 2. Both prior placebo and de novo subject groups showed a de novo subject groups, as would be expected given that prior
mean improvement on ABC Social Withdrawal, Stereotypic Be- aripiprazole subjects had previously received active treatment for 8
havior, Hyperactivity, and Inappropriate Speech subscale scores at weeks (Table 2). Prior aripiprazole subjects showed a mean im-
week 52 (OC) and endpoint (LOCF). provement in ABC Social Withdrawal, Stereotypic Behavior, Hy-
Prior placebo and de novo subjects showed a mean improvement peractivity, and Inappropriate Speech subscale scores at week 52
from baseline in CY-BOCS Compulsion Scale score at week 52 for patients who remained on treatment (OC data). Mean changes in
(OC) and endpoint (LOCF) and showed increases in PedsQL scores ABC Social Withdrawal, Stereotypic Behavior, Hyperactivity, and
from baseline to endpoint, indicating improvement in quality of Inappropriate Speech subscale scores for prior aripiprazole subjects
life. Mean change in CGSQ scores at endpoint (LOCF) showed at endpoint (LOCF) were small.
improvement in caregiver strain in the prior placebo and de novo Prior aripiprazole subjects showed a slight mean improvement
subject groups. from baseline in CY-BOCS Compulsion Scale score at week 52
(OC) and endpoint (LOCF). Prior aripiprazole subjects experienced
Prior aripiprazole subjects. For prior aripiprazole subjects, mean increases in PedsQL scores from baseline to endpoint, indi-
ABC Irritability scores remained constant over the course of long- cating improvement in quality of life, although no overall im-
term treatment, indicating that prior improvement experienced with provement in CGSQ scores at endpoint (LOCF) was seen for prior
aripiprazole treatment during the randomized treatment period was aripiprazole subjects.
maintained during the 52-week open-label treatment period; at
week 52 (OC data set), the mean change from baseline in ABC
Discussion
Irritability scores was þ 0.7 (10.2). In the antecedent placebo-
controlled trials, the mean baseline ABC-I scores for the subjects Results from this long-term open-label study suggest that
randomized to aripiprazole (the ‘‘prior aripiprazole’’ subjects in aripiprazole, for up to 1 year, can reduce symptoms of irritability in
this trial) were between 28 and 30 at the original baseline at the start children and adolescents with autistic disorder who also demon-
of double-blind treatment, and there was a mean improvement in strated behaviors such as irritability, agitation, and self-injurious
ABC-I scores of 12 to 14 points after 8 weeks of double-blind behavior or a combination of these. Symptoms of irritability, as-
treatment. sessed using the caregiver-rated ABC Irritability subscale score,
CGI-S scores at week 52 (OC) and endpoint (LOCF) were were improved early in the course of treatment and these im-
generally unchanged (Table 2), although the mean CGI-S score at provements were maintained over the 52-week study period in
open-label baseline was 3.9 points, compared with a mean baseline patients receiving aripiprazole for the first time during this study
score of around 5 at baseline in the antecedent placebo-controlled (prior placebo subjects and the majority of de novo subjects). The
trials. The mean (SD) CGI-I score was 2.5 (1.2) for the prior pattern of improvement seen in the first few weeks of aripiprazole
234 MARCUS ET AL.

Table 2. Mean (SD) Change from Baseline in Additional Efficacy Outcomes at Week 52 (Observed Case)
and Endpoint (Last Observation Carried Forward)

De novo Prior placebo Prior Aripiprazolea

Variable mean (SD) (n ¼ 84) (n ¼ 69) (n ¼ 169)

Clinical Global Impressions—Severity

Baseline 4.8 (1.0) 4.2 (1.0) 3.9 (1.1)
Week 52 (OC) 1.0 (0.8) 0.6 (1.2) 0.1 (1.0)
Endpoint (LOCF) 0.8 (0.9) 0.4 (1.1) 0.0 (1.0)
Aberrant Behavior Checklist
Irritability
Baseline 23.2 (8.9) 21.5 (9.8) 15.0 (9.2)
Week 52 (OC) 8.0 (10.1) 6.1 (11.9) 0.7 (10.2)
Endpoint (LOCF) 6.5 (11.1) 6.1 (11.3) 0.7 (9.7)
Lethargy/social withdrawal
Baseline 14.6 (8.6) 11.3 (9.2) 10.4 (8.9)
Week 52 (OC) 6.4 (7.9) 4.1 (7.2) 2.3 (6.4)
Endpoint (LOCF) 5.4 (9.1) 3.0 (7.0) 1.8 (6.1)
Stereotypic behavior
Baseline 8.1 (5.2) 8.1 (5.6) 6.4 (5.5)
Week 52 (OC) 2.7 (3.1) 1.9 (4.1) 0.5 (4.4)
Endpoint (LOCF) 2.5 (4.7) 1.9 (4.5) 0.1 (4.6)
Hyperactivity
Baseline 28.4 (10.9) 25.8 (13.2) 18.4 (12.0)
Week 52 (OC) 12.3 (8.5) 9.1 (11.5) 0.6 (10.3)
Endpoint (LOCF) 10.0 (10.6) 8.3 (10.9) 0.3 (11.2)
Inappropriate Speech
Baseline 5.8 (3.2) 5.7 (4.2) 4.2 (3.6)
Week 52 (OC) 2.0 (2.5) 1.8 (3.0) 0.3 (2.4)
Endpoint (LOCF) 1.9 (2.7) 1.8 (2.9) 0.3 (2.5)
CY-BOCS (Compulsions)
Baseline 12.6 (4.6) 12.1 (4.0) 10.4 (3.9)
Week 52 (OC) 2.8 (3.5) 2.6 (5.4) 0.2 (4.1)
Endpoint (LOCF) 2.0 (3.7) 2.4 (5.1) 0.2 (3.8)
Pediatric Quality of Life Inventory mean combined total scores
Baseline 43.1 (14.4) 50.4 (17.2) 57.4 (15.3)
Endpoint (LOCF) 12.2 (16.1) 2.9 (16.4) 1.0 (15.9)
Caregiver Strain Questionnaire
Baseline 9.0 (1.9) 8.3 (2.7) 7.0 (1.9)
Endpoint (LOCF) 1.6 (2.3) 1.2 (2.5) 0.0 (1.7)

For the Pediatric Quality of Life Inventory combined total scores, positive scores signify improvement.
a
Prior aripiprazole subjects had previously received aripiprazole for 8 weeks before open-label baseline assessments; thus, lower baseline values in this
group reflect less severe impairment at baseline and mean change scores represent the change during long-term treatment, subsequent to improvement
experienced during double-blind treatment.
CY–BOCS ¼ Children’s Yale–Brown Obsessive Compulsive Scale; LOCF ¼ last observation carried forward; OC ¼ observed case; SD ¼ standard
deviation.

treatment for placebo rollover and de novo subjects was similar to bility for up to 52 weeks in the majority of patients. Further,
that observed with aripiprazole in the double-blind studies: early completion rates in this study were high, with 60% of subjects
symptom improvement within the first few weeks followed by finishing 52 weeks of treatment and few subjects (6.1%) dis-
further improvement up to week 8 (Marcus et al. 2009; Owen et al. continuing long-term treatment due to lack of efficacy. Dis-
2009). For individuals who had previously received 8 weeks of continuation rates due to lack of efficacy were also low in the
aripiprazole treatment during the prior double-blind studies, im- double-blind studies; no aripiprazole-treated subjects in the fixed-
provements in irritability symptoms were maintained over the 52- dose study (Marcus et al. 2009) and 2.1% of aripiprazole-treated
week treatment period. However, it should be considered that this subjects in the flexible-dose study discontinued treatment due to
study was not specifically designed to assess maintenance of effect lack of efficacy (Owen et al. 2009).
or the prevention of relapse. Consistent with improvement in In addition to the improvements in irritability observed in this
symptoms assessed using the caregiver-rated ABC Irritability long-term study, prior placebo and de novo subjects also demon-
subscale scores, the distribution of clinician-rated CGI-I score at strated improvements on the other ABC subscales. It is also inter-
endpoint suggests that aripiprazole improved symptoms of irrita- esting to note that aripiprazole treatment resulted in improvements
ARIPIPRAZOLE EFFICACY IN AUTISTIC DISORDER 235

100 with consideration of several limitations such as the open-label

study design, along with the challenges of interpreting long-term
90
study data. Further, this study permitted the use of a range of con-
Percentage of patients (%)

80 comitant psychotropic and nonpsychotropic medications. At their

last visit, 38.2% of subjects were receiving a concomitant CNS
70
medication, most frequently an antidepressant (13.4%), a psy-
60 chostimulant (11.5%), or an antiepileptic (5.9%). While this con-
comitant medication use is more likely to reflect real-life clinical
50
practice, it is important to recognize the rate of concomitant med-
40 ication use when interpreting the efficacy findings reported. The
rate of concomitant medication use for the treatment of EPS re-
30
ported here (5%) was similar to that reported in subjects receiving
20 aripiprazole in the short-term fixed-dose (6%) (Marcus et al. 2009)
10
and flexible-dose studies (2%) (Owen et al. 2009).
Additional limitations should also be considered. First, all
0 patients were re-titrated to aripiprazole 2 mg/day at entry into open-
De novo (n=84) Prior placebo (n=69) Prior aripiprazo le label treatment. While this may have had an impact on prior ar-
(n=169)
ipiprazole patients who were receiving higher aripiprazole doses
Much or very much worse Minimally worse No change during the previous studies, we did not systematically evaluate the
Minimally improved Very much or much improved effects of a reduction in dose to 2 mg/day. It was necessary to dose
in this manner, as the treatments patients were receiving during
FIG. 2. Distribution of CGI-I score at endpoint (LOCF), effi- double-blind treatment were not known at entry into open-label
cacy sample. CGI Improvement was rated relative to the symp-
treatment. This necessitated re-titration to therapeutic dose.
toms at open-label baseline for de novo subjects and relative to
double-blind treatment baseline for prior aripiprazole/placebo Nevertheless, the long half-life of aripiprazole (*75 hours)
subjects. CGI ¼ Clinical Global Impressions. makes discontinuation symptoms somewhat unlikely during re-
titration. It is worth noting, however, that the majority of subjects in
this study were taking 10 or 15 mg/day at the end of the study.
in quality of life measures at endpoint in these groups, as measured Second, the majority of subjects in these studies were male, as
on the PedsQL and CGSQ scales, suggesting that improvements would be expected based on the higher prevalence of autistic dis-
in irritability associated with autistic disorder during 52 weeks of order in males than in females. As such, the generalizability of our
treatment may translate into improved day-to-day functioning for findings to a predominantly female population is unknown. Third, a
the children and their families. Improvement in these measures was washout period of 4 days was used in this study, as some patients
also seen in subjects receiving aripiprazole during the short-term could potentially decompensate substantially with longer washout
studies (Marcus et al. 2009; Owen et al. 2009), and this improvement periods. Although the washout period for prior medications used
was maintained over the 52-week treatment period in this study. may be considered too short for previous antipsychotic washout,
The safety and tolerability findings from this study are reported the first ABC-I assessment was at week 4, by which time all pro-
elsewhere (Marcus et al. In press). Briefly, aripiprazole was gen- hibited antipsychotics would have been cleared. Fourth, as this
erally well tolerated as a long-term treatment in this population; study included only subjects with autistic disorder, the generaliz-
discontinuations due to adverse events were 10.6%, and the most ability of these results to other pervasive developmental disorders is
common AEs leading to discontinuation were aggression and unknown. Finally, as this was not a comparison study, judgments
weight increase (Marcus et al. In press). The most frequently re- regarding the relative efficacy of aripiprazole compared with other
ported AEs (occurring in  10% of subjects) in the total study antipsychotics in this population cannot be made.
population during long-term treatment were weight increase (23%),
vomiting (19%), nasopharyngitis (13%), increased appetite (13%), Conclusions
pyrexia (12%), upper respiratory tract infection (12%), and in-
somnia (10%). EPS-related AEs occurred in 15% of subjects Aripiprazole reduced symptoms of irritability associated with
(Marcus et al. In press). Nine (2.7%) subjects reported serious AEs; autistic disorder in pediatric subjects aged 6–17 years who were
only aggression was reported for more than one subject (Marcus studied for up to 1 year.
et al., In press). Increased weight gain, though present, appeared to
plateau over time. As with all atypical antipsychotics, individuals Clinical Significance
receiving treatment with aripiprazole should be monitored for For chronic conditions that may require long-term treatment,
weight gain, and this should be proactively managed should it occur there is a need for data regarding treatment effects that go beyond
(Correll 2008). As with all pharmacologic treatment generally, the the acute phase of an illness to help inform clinical decision
decision to initiate medication treatment in pediatric patients with making. The results from this study confirm that there can be
irritability associated with autistic disorder should be made be- continued benefit—for up to 1 year—from the use of aripiprazole
tween healthcare providers and caregivers only after a thorough for the treatment of the symptoms of irritability associated with
diagnostic evaluation and discussion of both the benefits and the autistic disorder in pediatric subjects aged 6 to 17 years.
risks.
The findings of this study are strengthened by the large patient
Disclosures
population that was enrolled and the relatively high numbers of
subjects who continued to receive aripiprazole for the 52-week Michael Aman was a consultant to and received grant support
study duration. However, the efficacy findings should be interpreted from Bristol-Myers Squibb.
236 MARCUS ET AL.

Lisa Kamen, Raymond Mankoski, George Manos, and Ronald study of the safety and tolerability of aripiprazole flexibly dosed in
N. Marcus are employees of Bristol-Myers Squibb. the treatment of irritability in pediatric patients (6–17 years) with
Patricia Corey-Lisle and Randall Owen are former employees of autistic disorder. J Clin Psychiatry In press.
Bristol-Myers Squibb. Marcus RN, Owen R, Kamen L, Manos G, McQuade RD, Carson
Robert D. McQuade and William H. Carson are employees of WH, Aman MG: A placebo-controlled, fixed-dose study of ar-
Otsuka Pharmaceutical Development & Commercialization, Inc. ipiprazole in children and adolescents with irritability associated
with autistic disorder. J Am Acad Child Adolesc Psychiatry 48:
1110–1119, 2009.
Acknowledgments
Myers SM, Johnson CP: Management of children with autism spec-
This study was supported by Bristol-Myers Squibb (Princeton, trum disorders. Pediatrics 120:1162–1182, 2007.
NJ) and Otsuka Pharmaceutical Co., Ltd. (Tokyo, Japan). Editorial Owen R, Sikich L, Marcus RN, Corey-Lisle P, Manos G, McQuade
support for the preparation of this article was provided by Ogilvy RD, Carson WH, Findling RL: Aripiprazole in the treatment of
Healthworld Medical Education; funding was provided by Bristol- irritability in children and adolescents with autistic disorder. Pe-
Myers Squibb. diatrics 124:1533–1540, 2009.
Risperidone: (Risperdal) U.S. Full Prescribing Information. Titusville
NJ: Janssen, Division of Ortho-McNeil-Janssen Pharmaceuticals
References
Inc.; 2009.
Aman MG, Singh NN, Stewart AW, Field CJ: The Aberrant Behavior Scahill L, Riddle MA, McSwiggin-Hardin M, Ort SI, King RA,
Checklist: A behavior rating scale for the assessment of treatment Goodman WK, Cicchetti D, Leckman JF: Children’s Yale-Brown
effects. Am J Ment Defic 89:485–491, 1985a. Obsessive Compulsive Scale: Reliability and validity. J Am Acad
Aman MG, Singh NN, Stewart AW, Field CJ: Psychometric charac- Child Adolesc Psychiatry 36:844–852, 1997.
teristics of the Aberrant Behavior Checklist. Am J Ment Defic 89: Varni JW, Burwinkle TM, Katz ER, Meeske K, Dickinson P:
492–502, 1985b. The PedsQLTM in pediatric cancer: Reliability and validity of the
American Psychiatric Association: Diagnostic and Statistical Manual Pediatric Quality of Life InventoryTM Generic Core Scales, Mul-
of Mental Disorders, 4th ed., Text Revision (DSM-IV-TR). Wa- tidimensional Fatigue Scale, and Cancer Module. Cancer 94:2090–
shington, DC: American Psychiatric Association; 2000. 2106, 2002.
Aripiprazole: (Abilify) U.S. Full Prescribing Information. Princeton, Varni JW, Seid M, Kurtin PS: PedsQLTM 4.0: reliability and validity
NJ: Bristol-Myers Squibb and Rockville MD: Otsuka America of the Pediatric Quality of Life InventoryTM version 4.0 Generic
Pharmaceutical, Inc.; 2009. Core Scales in healthy and patient populations. Med Care 39:800–
Brannan MA, Heflinger CA, Bickman L: The Caregiver Strain 812, 2001.
Questionnaire. Measuring the impact on the family of living with a Volkmar F, Cook EH, Jr., Pomeroy J, Realmuto G, Tanguay P:
child with serious emotional disturbance. J Emot Behav Disord 5: Practice parameters for the assessment and treatment of children,
212–222, 1997. adolescents, and adults with autism and other pervasive develop-
Correll CU: Assessing and maximizing the safety and tolerability of mental disorders. American Academy of Child and Adolescent
antipsychotics used in the treatment of children and adolescents. J Psychiatry Working Group on Quality Issues. J Am Acad Child
Clin Psychiatry 69 Suppl 4:26–36, 2008. Adolesc Psychiatry 38:32S–54S, 1999.
Guy W: ECDEU Assessment Manual for Psychopharmacology—
Revised. Rockville, MD: US Department of Health Services; Address correspondence to:
1976. Michael Aman, Ph.D.
Lord C, Rutter M, Le Couteur A: Autism Diagnostic Interview- The Nisonger Center UCEDD
Revised: A revised version of a diagnostic interview for caregivers The Ohio State University
of individuals with possible pervasive developmental disorders. J 1581 Dodd Drive
Autism Dev Disord 24:659–685, 1994. Columbus, OH 43210
Marcus RN, Owen R, Manos G, Mankoski R, Kamen L, McQuade
RD, Carson WH, Findling RL: A 52-week, open-label, multicenter E-mail: [email protected]