Aripiprazole in the real-world treatment for irritability associated with autism spectrum disorder in children and adolescents in Japan: 52-week post-marketing surveillance

Patients newly treated with aripiprazole for irritability associated with ASD in children and adolescents (≥6 years and < 18 years) were included in the surveillance. The target sample size was calculated to be 300 patients on the assumption of detecting adverse drug reactions (ADRs) occurring at a frequency of 1% with a confidence of at least 95%.

This surveillance was conducted as a multicenter, prospective, non-interventional, observational study for 52 weeks (1 year). The registry and case reports were encoded by the physicians through Electronic Data Capture system during the period from April 2017 to September 2019.

This surveillance was conducted in compliance with the Ministerial Ordinance on Standards for Conducting Post-marketing Surveys and Studies on Drugs; MHLW Ordinance No. 171 issued on December 20, 2004 (GPSP). As the surveillance is a non-interventional study in accordance with GPSP, the need for ethics approval and consent were waived. The surveillance was designed by Otsuka, reviewed by the Japanese Pharmaceutical and Medical Devices Agency (PMDA), and also registered at Clinical Trials.gov (identifier: NCT03179787).

Aripiprazole was administered orally once daily with a starting dose of 1 mg daily and a maintenance dose of 1–15 mg daily according to the package insert in Japan [17]. The dose could be adjusted according to the severity of patient’s symptoms, however the dose increase per day was to be 3 mg or less and the daily dose was not to exceed 15 mg [17].

Using caregiver-rated Aberrant Behavior Checklist-Japanese version (ABC-J), the degree of aberrant behaviors was scored on four levels ranging from 0 ‘not at all a problem’ to 3 ‘the problem is severe in degree’, totaling 58 items. These items were classified into five subscales: irritability (15 items), lethargy/social withdrawal (16 items), stereotypy (7 items), hyperactivity (16 items), and inappropriate speech (4 items), and scores were calculated [18].

Physician-rated Clinical Global Impression-Improvement (CGI-I) and Severity of illness (CGI-S) scales quantified the physician’s impression at baseline (CGI-S only), week 4, week 8, week 16, week 24, and week 52, or at the time of discontinuation of aripiprazole administration. The CGI-I scale was scored from 1 ‘very much improved’ to 7 ‘very much worse’, and CGI-S was scored from 1 ‘normal’ to 7 ‘very much severely ill’.

Caregiver-rated Strengths and Difficulties Questionnaire for Children (SDQ) was used to score the adaptation and mental health conditions on three levels: 0 ‘not true’, 1 ‘somewhat true’, and 2 ‘certainly true’, with a total of 25 items. These items consist of five subscales: conduct problems (5 items), hyperactivity (5 items), emotional problems (5 items), peer problems (5 items), and prosocial behavior (5 items), each yielding scores between 0 and 10. The Total Difficulties Score ranging from 0 to 40 was derived from the sum of the four subscale scores, excluding prosocial behavior [19]. Banding of raw scores obtained with the five SDQ subscales and the Total Difficulties Score, the scale properties were assessed as ‘Low Need’, ‘Some Need’, or ‘High Need’, applying recommended banding of raw scores by Matsuishi T et al. [20] both in ≤13 years and in > 13 years old.

In addition, the average sleep time duration in the last 4 weeks, which was reported by patients or caregivers and not actigraphy-measured data, was recorded at week 24, week 52, and/or end of study. Recommended sleep duration by age group was predefined using modified method reported by Hirshkowitz M et al. [21], and the shift in sleep duration from baseline to end-point (last observation carried forward; LOCF) was calculated.

Results were summarized using descriptive statistics. Continuous variables were described as mean ± standard deviation (SD). The LOCF method was used with imputation of the latest observed values. An applicable paired t-test or Wilcoxon signed-rank sum test was applied to compare the pre-dose and post-dose effect. The level of statistical significance was set at two-sided 5% and the confidence interval at two-sided 95%.

The treatment continuation rate was analyzed by Kaplan-Meier method, with the reason for discontinuation of aripiprazole censored as “transfer,”, “lost to follow-up,” or “improved symptoms”.

The Japanese version of the ICH Medical Dictionary for Regulatory Activities (MedDRA/J version 22.1) was used in coding and classifying adverse events. Adverse events for which a causal relationship to aripiprazole could not be ruled out were considered as ADRs. All statistical analyses were performed using SAS software (version 9.3; SAS Institute Inc.).

Five hundred and twenty-eight patients were enrolled at 100 sites across Japan, and 526 case reports were collected during the period of April 2017 to September 2019. The safety analysis population consisted of 510 patients. Among excluded patients, 15 patients were lost to follow-up, and one patient was not treated with aripiprazole. After excluding 21 patients with the protocol deviations (including all effectiveness measures were not assessed after treatment: n = 11, all effectiveness measures were assessed over 7 days after last dose: n = 7, the daily dose was > 15 mg: n = 2, all effectiveness were assessed after day 393: n = 1), the effectiveness analysis population consisted of 489 patients (Fig. 1).

Baseline patient characteristics are summarized in Table 1. Among 510 patients included in the safety analysis, the mean age of the patient population at baseline was 10.4 ± 3.1 years, and the majority of the patients were males (73.7%) and aged 6 years or older but younger than 13 years (71.8%).

Major comorbidities (≥1%) were ADHD in 53.1% (n = 271), sleep disorders in 17.6% (n = 90), learning disorders (LD) in 11.2% (n = 57), and tic disorders in 6.9% (n = 35). There were 24.1% (n = 123) of patients with intellectual disabilities.

The most common concomitant medications (drugs for ASD irritability, ≥1%) were antidepressants in 7.1% (n = 36), risperidone in 5.1% (n = 26), and ADHD medications in 1.2% (n = 6).

On the physician-rated CGI-S scale, most patients were assessed as being moderately ill (55.3%, n = 282), followed by markedly ill (29.4%, n = 150).

The mean duration of aripiprazole treatment was 290.6 ± 151.5 days and the mean daily dose was 2.20 ± 1.85 mg.

A total of 201 patients (39.4% of the safety population) discontinued aripiprazole treatment during the surveillance period. Of the 201 patients who discontinued treatment, the most common reason for discontinuation (multiple responses were allowed) was ‘request for discontinuation from the patient or family’ in 33.8% (n = 68) (Table 2) and the mean duration of treatment prior to discontinuation was 133.6 ± 113.4 days.

The continuation rate of aripiprazole treatment was 84.6% at day 168 (week 24) and 78.1% at day 364 (week 52) from the first dose (see Additional file 1).

In 510 patients included in the safety analysis, adverse events occurred in 24.5% (n = 125) and ADRs based on physician-assessed causality occurred in 22.7% (n = 116) (Table 3). The most common ADRs with an incidence of ≥1% were somnolence (9.4%, n = 48), followed by weight increased (3.3%, n = 17), nausea (1.4%, n = 7), increased appetite and headache (1.2% each, n = 6), and obesity (1.0%, n = 5). Fig. 2 shows the duration to onset date of ADRs from the first dose, and the summary statistics (median [min-max]) were headache (3.5 days [1–284 days]), nausea (8.0 days [2–82 days]), somnolence (32.5 days [1–383 days]), increased appetite (120.5 days [29–179 days]), weight increased (168.0 days [1–320 days]), and obesity (192.0 days [76–365 days]).

Serious ADRs were epilepsy, partial seizures, and renal impairment (0.2% each).

Most common ADRs leading to discontinuation with an incidence of > 0.5% were somnolence (2.0%), weight increased (0.8%), and insomnia (0.6%).

The most common ADRs (≥0.5%) associated with extrapyramidal symptoms were tic disorders and akathisia (0.8% each), both of which were non-serious.

The mean percentile of weight was 48.36% ± 31.86% at baseline, 52.35% ± 32.86% at end-point (LOCF), and the mean change from baseline to end-point was 3.99% ± 11.65%. Percentile body weight categorical shift data are presented in Table 4. All patients with a weight in the > 90th percentile at end-point (LOCF) were in the > 75th percentile at baseline, and all patients with a weight in the ≤10th percentile at end-point (LOCF) were in the ≤25th percentile at baseline. There were no patients in which the percentiles of weight differed greatly between before and after aripiprazole treatment.

The mean of ABC-J irritability subscale scores at baseline was 19.8 ± 9.5 (n = 396), decreased to 14.7 ± 8.8 (n = 288) at week 4 and to 12.9 ± 8.1 (n = 316) at week 8, remained at 12.7 to 13.1 at week 16 to 52, and was 13.0 ± 9.0 (n = 396) at end-point (LOCF). Mean changes of ABC-J irritability subscale scores from baseline were − 5.7 ± 6.8 at 4 weeks, − 7.0 to − 8.3 after week 8 and − 6.8 ± 8.3 at end-point (LOCF), with significant reductions in scores compared with baseline at each assessment point (p < 0.0001). Patients showed a mean improvement in all ABC-J subscales at all assessment points and end-point (LOCF) compared with baseline (Fig. 3). Interestingly, based on multiple regression analysis, comorbid ADHD was not selected as the variable affecting the ABC-J irritability subscale score at end-point (see Additional file 2).

From the results of the improvement assessment by CGI-I, the percentage of patients scoring 3 ‘minimally improved’ or less was 74.3% (n = 306/412) at week 4, 79.4% (n = 342/431) to 85.4% (n = 239/280) after week 8, and 77.0% (n = 376/488) at end-point (LOCF), with an increasing trend throughout the treatment period (see Additional file 3).

From the results of the severity assessment by CGI-S, the percentage of patients who were assessed to be 1 ‘normal’ to 3 ‘mildly ill’ was 8.8% (n = 43/489) at baseline, 43.9% (n = 180/410) at week 4, 51.3% (n = 220/429) at week 8, and 59.7% (n = 292/489) at end-point (LOCF) and tended to increase throughout the treatment period (see Fig. 4 and Additional file 3).

The mean baseline SDQ Total Difficulties Score was 21.1 ± 5.5 (n = 309); mean changes from baseline were − 3.3 ± 4.9 (n = 215) at week 24, − 4.3 ± 6.1 (n = 200) at week 52, and − 3.8 ± 5.6 (n = 309) at end-point (LOCF) (p < 0.0001) (Fig. 5). The mean SDQ prosocial behavior subscale score was 3.5 ± 2.5 (n = 316) at baseline, and mean changes from baseline were 0.6 ± 1.7 (n = 217) at week 24, 0.8 ± 2.1 (n = 200) at week 52, and 0.7 ± 2.0 (n = 316) at end-point (p < 0.0001) (Fig. 5). All SDQ subscale scores and Total Difficulties Score were significantly decreased at all assessment points compared with baseline. The shift in SDQ scale properties from baseline to end-point (LOCF) is shown in Table 5. For conduct problems, the percentage of patients changing from ‘High Need’ to ‘Some Need’ or ‘Low need’ was 38.8% (n = 83/214) and the percentage of patients changing from ‘Some Need’ to ‘Low need’ was 51.1% (n = 24/47). In contrast, the percentage of patients changing from ‘Low Need’ to ‘Some Need’ or ‘High need’ was 23.1% (n = 12/52) and the patients changing from ‘Some Need’ to ‘High need’ was 19.1% (n = 9/47). Also, similar trends were shown in hyperactivity/inattention, emotional symptoms, and prosocial behavior subscales.

The shift in mean sleep time duration from baseline to end-point (LOCF) is shown in Fig. 6. The percentage of patients changing from ‘Too short’ to ‘May be appropriate’ or ‘Recommended’ was 55.6% (n = 15/27). In contrast, the percentage of patients changing from ‘Recommended’ or ‘May be appropriate’ to ‘Too short’ was 1.6% (n = 4/258). Only one patient changed from ‘Too long’ to ‘Recommended’. No patients had a mean sleep time duration of ‘May be appropriate (longer)’ or ‘Too long’ at end-point (LOCF).