High-Dose Propranolol for Severe and Chronic Aggression in Autism Spectrum Disorder

Abstract

Background

Despite the use of behavioral interventions and psychotropic medications, many individuals with autism spectrum disorder (ASD) who engage in severe aggression remain refractory to conventional treatment. Propranolol, a beta-blocker, has accumulated much anecdotal evidence as a promising option. However, well-designed studies are rare, and the apprehension about cardiovascular side effects from large doses continues to exist.

Purpose

The aims of this study were (1) to demonstrate the feasibility of treating aggression with high-dose propranolol using telehealth study visits and (2) to document cardiac safety.

Methods

This study utilized a randomized, double-blind, placebo-controlled, crossover design. Dosing was titrated up in a flexible but stepwise fashion until therapeutic response was obtained or up to 200 mg tid. Following washout, those who were assigned propranolol were crossed over to placebo and vice versa. Six participants between the ages 12–19 participated. The primary outcome measures were the final Clinical Global Impression Improvement Scale (CGI-I) and the Aberrant Behavior Checklist—Community Irritability (ABC-C/I) scores at 200 mg tid.

Results

The CGI-I indicated a 50% reduction in symptoms in the propranolol phase, while the ABC-I indicated a 37% reduction in comparison to placebo. The effect sizes (r) for the CGI-I and the ABC-C/I were large, −0.74 and −0.64, respectively. The average blood pressure was 122/68 during the placebo phase and 109/72 during the propranolol phase. All Holter monitor exams were unremarkable.

Conclusion

These results suggest that propranolol is an effective option in decreasing aggression in individuals with ASD. As this was a small study, a larger clinical trial is needed.

The symptoms of irritability associated with autism spectrum disorder (ASD) include aggression, tantrums, self-injury, and disruption. These challenging behaviors occur in as high as 68% of the population^1,2 and have been found to persist for over a decade,^3,4 indicating that these symptoms could even be lifelong. These behaviors pose a significant barrier to pursuing quality of life⁵ and are a common chief concern for caregivers, above the core symptoms of ASD.^1,6–8 Psychiatric hospitalization is much more common in this population, with 10.8% of children with ASD having had a hospitalization, most commonly for self-injury or aggression.⁹ Medicaid expenditures are 10 times higher for those diagnosed with ASD compared to other children, with much of the difference attributable to inpatient psychiatric care.¹⁰ As children age, the expenditure for behavioral symptoms increases, and there is a concomitant decrease in the available expenditure for other services.^10,11 At a personal level, this translates to adolescents and adults being restricted to school or community placements that provide less assistance or accommodation for their core difficulties as well as fewer opportunities that align with their individual strengths. Inadequate treatments can also result in trauma from the use of acute interventions such as manual restraint, takedowns, and 4-point restraints.^12–14

Two antipsychotic medications, risperidone and aripiprazole, are the only 2 medications that are FDA-approved for the treatment of irritability associated with ASD. While very successful for some, their efficacy is far from optimal with 1 study showing 40% of the sample being resistant to any pharmacotherapy.¹⁵ These medications are also associated with many adverse effects making their long-term use problematic,¹⁶ including 3 times the risk of diabetes¹⁷ and weight gain.¹⁸ Presently, the treatments for these symptoms continue to yield only partial benefit for many. Better treatment for severe problem behaviors is therefore urgently needed.

Other than the 2 FDA-approved medications, well-designed treatment studies for the severe challenging behaviors associated with ASD continue to be relatively scarce. There have been many anecdotal or small case series reports that showed promise. One of these strategies involves the beta-adrenergic blocking agent propranolol. The literature on the use of propranolol for these symptoms in a variety of psychiatric disorders is limited by lack of large rigorous studies. Most of the literature is made up of case studies, although there have been 6 small double-blind and/or placebo-controlled studies.¹⁹ While primarily targeting aggression, self-injury, and destructive behaviors, these studies administered propranolol to individuals with various psychiatric diagnoses, including intermittent explosive disorder (with and without known brain damage), conduct disorders, attention deficit disorder, schizophrenia, atypical psychosis, schizoaffective disorder, drug abuse, seizure disorder, borderline personality disorder, various types of intellectual disabilities with various etiologic origins, and dementia.^20–23 Some of these studies used mixed diagnostic populations with a wide range of participant ages, ranging from children to geriatric cases and over a wide range of cognitive levels.

In individuals with ASD, a case series of 8 adults with violent or self-injurious behaviors was reported.²⁴ The authors described a remarkable effect on previously intractable aggression, which was consistent with similar findings for those with schizophrenia, brain damage, and severe intellectual disability. In a recent review of the effect of beta-adrenergic blockers on challenging behaviors of people with developmental disability, it was reported that many had positive outcomes, which supported the efficacy of propranolol for this indication, although the authors point out that the overall quality of the research was poor with the absence of randomized controlled trials.²² In a review of studies on brain-injured patients with aggression, it was concluded that beta blockers had the best evidence for efficacy of all medications, despite the lack of quality studies.²⁵

The largest case series to date²⁶ reported on a retrospective analysis of 46 individuals (38 males, 8 females) with a mean age of 16 years (range = 8 to 32 years). The mean propranolol dose was 462 mg/d (range = 120–960 mg). The mean dosage for males was 481 mg (range = 120–960 mg) and for females was 372 mg (range = 120–640 mg). Of the 46 patients, 39 (85%) were found to be much improved or very much improved (CGI-I 1 or 2) based on their aggressive symptoms, self-injurious behaviors, or severely disruptive behaviors. Two out of 46 patients were slightly improved (CGI-I = 3). Five (11%) of 46 cases were not improved or worsened (CGI-I ≥ 4). Although not quantified, there was little to no benefit to other symptoms present, such as hyperactivity, repetitive behaviors, inattention, and mood dysregulation. Only 2 patients (4.2%) needed to be discontinued due to side effects (both bronchial constriction). Thirty patients on high-dose propranolol had cardiology workup, and other than some bradycardia and reduction in pulse (not clinically significant), there were no adverse effects on the Holter monitor or echocardiogram.

Despite the literature on the use of propranolol for aggression across a range of diagnoses, it is not commonly prescribed in individuals with ASD. In the UK, adrenergic blocking agents are rarely prescribed, with only 2% of individuals with ASD receiving any type of beta-blocker.²⁷ In a study of pharmacological treatments in Germany, no beta blockers were listed in the most common 25 medications prescribed for ASD.²⁸

In the case series of 46 patients discussed above,²⁶ the average number of drugs previously tried was 6.2, indicating that several other medications without adequate evidence are typically prescribed before propranolol is considered. It is also likely that concerns regarding the safety of high-dose propranolol have limited its clinical application. Well-designed clinical trials demonstrating the efficacy and safety of high-dose propranolol are needed to inform clinical practice and to improve the quality of life for an underserved subset of individuals with ASD.

Therefore, the goals of this study were as follows: (1) The first goal is to demonstrate the feasibility of treating and evaluating severely behaviorally impaired individuals using a double-blind, placebo-controlled, crossover study of high-dose propranolol. Along with the usual rigor needed for a gold standard clinical trial, this indication for individuals with ASD who engage in severe challenging behaviors presents other challenges. Attending many clinic visits involves an undue burden for the families, including scheduling changes and safety issues related to transportation. Therefore, we piloted methods that would minimize this burden on the families through the use of telehealth for behavioral and cardiologic follow-up.

METHODS

All procedures were approved by the IRB of the host institution, and written informed consent was obtained from the parents of all subjects. In this study, none of the participants were deemed competent to sign informed consent.

Participants

Children and young adults with a diagnosis of ASD who engaged in severe and chronic aggression, self-injury, and disruptive behaviors that interfered with day-to-day activities between the ages of 12 and 30 were recruited. All participants were required to have tried at least 2 psychotropics, including 1 antipsychotic medication that yielded inadequate outcome. In addition, a Clinical Global Impression Severity scale (CGI-S) score of 6 or 7 (severely impaired or most severely impaired) and an Aberrant Behavior Checklist—Community Irritability (ABC-C/I) scale score at or above 18 were required. Psychotropic medications and behavior interventions were kept stable throughout the study.

Design and Dosing

A double-blind, placebo-controlled, crossover design was used. That is, each participant served as their own control. A university research pharmacy randomized the assignment and compounded the drug. Capsules were prescribed for those who could swallow them, but compounded liquid was also offered, and 1 participant used the liquid preparation. Following a 2-week baseline, all participants began at dose 1 (10 mg tid) and titrated up in a flexible but stepwise fashion. The standard dosing was 10 mg tid, 40 mg tid, 80 mg tid, 120 mg tid, 160 mg tid, and 200 mg tid or matching placebo. The dose was titrated up until adequate therapeutic response was obtained, as agreed by the parents and psychiatrist, or up to a dose of 600 mg/d. When the dose was held steady for 6 weeks, it was considered to be the maintenance dose. Following the first phase, titration off of the placebo or propranolol occurred every 3 days until the dose reached 0 mg/d. Then, the participant entered the second phase. Those who were assigned propranolol were assigned a placebo and vice versa. Following the double-blind phase, the participant had the opportunity to continue in the optional 4-week open-label extension phase.

Study Visits and Drug Delivery

The study was designed as a weekly telehealth study prior to COVID-19 in consideration of difficulties with schedule disruption and traveling in this study population. Therefore, all but the initial 1–2 visits were conducted via a HIPAA-compliant telehealth platform. Study drugs were formulated by a research pharmacy and delivered via courier.

Safety Monitoring

A full physical exam, ECG, Holter monitoring, and echocardiogram were conducted at baseline and after propranolol titration. Pulse and blood pressure were taken at screening and baseline and then 1 hour after the administration of the medication for the first 3 days of each new dose. Due to the known cardiologic effects of propranolol, each participant was fitted with a portable 24-hour Holter monitor. The Holter monitor exam was completed at dose 3 (80 mg tid) and again at maintenance dose to ensure no significant bradycardia or other cardiac issues ensue.

Demographic Assessments

All participants submitted a documented diagnosis of ASD prior to study enrollment, and this was confirmed by a licensed psychologist using the Social Communication Questionnaire (SCQ). All but 1 participant was administered a nonverbal cognitive assessment (Leiter-3) by the same psychologist; the remaining participant had completed a recent comprehensive cognitive assessment (WISC-V) that was accepted instead of the Leiter-3. Parents of all participants completed the Vineland Adaptive Behavior Scale-III as a measure of self-sufficiency in Communication, Daily Living Skills, and Socialization.

Primary Dependent Measures

Clinical Global Impression—Improvement (CGI-I)²⁹ was used to judge the overall clinical condition relative to baseline using the same scale as the CGI-S. The physician rated the improvement weekly from baseline on a scale of 1 = very much improved since the initiation of treatment; 2 = much improved; 3 = minimally improved; 4 = no change from baseline (the initiation of treatment); 5 = minimally worse; 6 = much worse; 7 = very much worse since the initiation of treatment.

Aberrant Behavior Checklist—Community (ABC-C)³⁰ is a global behavior checklist that measures drug and other treatment effects in people with developmental disabilities. It is made up of 5 subscales, including Irritability, Lethargy, Inappropriate Speech, Hyperactivity, and Stereotypy, based on 58 items that describe various behavioral problems. The score for each item ranges from 0 indicating “no problem” to 3 indicating “severe problem.” The participant's parent completed the ABC-C weekly. The rate of a positive response is commonly defined as at least a 25% decrease in the Irritability subscale score.³¹

Data Analysis

The primary efficacy parameter was the change in CGI-I and ABC-C/I from baseline to study endpoint (ie, the last data point in each phase). A Wilcoxon’s signed rank test was conducted to evaluate if there was a statistically significant difference in CGI-I and ABC-C/I between placebo and drug phase. The equivalent non-parametric test of paired samples t-test was selected, due to the small sample size (N = 6). Descriptive statistics and effect sizes were also consulted.

RESULTS

Participants

Six children and young adults with ASD (2 females, 4 males) presenting with severe and chronic aggression between the ages 12–19 participated (mean = 16 years). Three of 6 (50%) were people of color (2 Asian Americans; 1 African American) whose mean nonverbal IQ was 63 (range = 32–110). The mean Vineland Adaptive Behavior Scales Composite score was 50 (range = 27–74) with the Communication Domain mean of 53 (range = 20–88), Daily Living Skills Domain mean of 54 (range = 35–79), and Socialization Domain mean of 36 (range = 20–66), reflecting individuals who are severely affected by ASD. At enrollment, the mean CGI-S score was 6.17 (range = 6–7), and the mean ABC-C/I score was 28 (range = 19–33), well above the score of 18 required for inclusion.

Titration and Tolerability

Propranolol or placebo was titrated flexibly to identify the maximum tolerated dose and then continued treatment at that dose for 4 weeks. In the propranolol phase, the longest total duration of treatment was 6 weeks, and all participants completed 4 weeks on the maximum dose. In contrast, in the placebo phase, 5 of 6 participants dropped out of treatment before completing a full 4 weeks at the maximum dose. This was true when a placebo was in either arm of the study.

Safety Monitoring

The average blood pressure on admission was 119/82, with an average blood pressure of 122/68 during the placebo phase and 109/72 during the propranolol phase. All Holter monitor exams were unremarkable, and there was no alteration in the study due to cardiac effects. Two adverse events were reported in the propranolol group: lethargy that was self-limited and loose stool without diarrhea. Neither adverse event was rated as related to the study drug. No other adverse events were reported.

Primary Outcome

The primary outcome measures were the final CGI-I and the ABC-I scores at 200 mg tid for both propranolol and placebo. For both measures, a lower score indicates fewer symptoms. The Wilcoxon signed rank test indicated a non-statistically significant difference between placebo and propranolol phase, for both CGI-I (T = 2.00, p = .071) and ABC-C/I (T = 3.00, p = .12). Nevertheless, the descriptive analysis showed substantial differences between placebo and propranolol phase for both measures. Across participants, the mean CGI-I score for placebo was 5.0 (SD = 1.3) and for propranolol was 2.5 (SD = 1.4), an average of 2.5-point (50%) reduction (see Figure 1). The mean ABC-C/I score for placebo was 19.2 (SD = 7.6) and for propranolol was 12.2 (SD = 5.5), a 7-point (37%) reduction (see Figure 2). Furthermore, the median CGI-I score for placebo was 5.5 and for propranolol was 2.00, a median of 3.5-point (64%) reduction. The median ABC-C/I for placebo was 18.5 and for propranolol was 14.0, a 4.5-point (24%) reduction (see Table 1 for descriptive analysis). Finally, effect size estimates (r) for the CGI-I and the ABC-C/I were large, -0.74 and -0.64, respectively.

FIGURE 1.

Endpoint CGI-I per study phase.

FIGURE 2.

Endpoint ABC-C/I per study phase.

TABLE 1.

Descriptive Analysis on CGI-I and ABC-C/I Between Placebo and Propranolol (Drug) Phase

	M	SD	Median	25th Quartile	50th Quartile	75th Quartile
CGI-I
Placebo	5.0	1.3	5.5	3.8	5.5	6.0
Propranolol	2.5	1.4	2.0	1.8	2.0	3.5
ABC-C/I
Placebo	19.2	7.6	18.5	12.3	18.5	27.3
Propranolol	12.2	5.5	14.0	6.3	14.0	16.5

CGI-I indicates Clinical Global Impression Improvement Scale; ABC-C/I, Aberrant Behavior Checklist—Community Irritability Subscale.

DISCUSSION

This randomized, double-blind, placebo-controlled, crossover study further suggests that propranolol is effective in decreasing challenging behaviors in children and adolescents with ASD. The CGI-I showed a 50% reduction and the ABC-C/I showed a 37% reduction in challenging behaviors when compared with placebo, with good reliability observed across both study physician (CGI-I) and parents (ABC-C/I). In addition, the effect sizes (r) were large, −0.74 for the CGI-I and −0.64 for the ABC-C/I. It should be noted that the 37% reduction in challenging behaviors was somewhat skewed as 1 case in which these behaviors had been present for years showed a dramatic decrease during phase A (on drug). The lack of symptoms continued through the washout into the placebo phase lasting for about 8 weeks. At that point, the challenging behaviors returned to full severity, and in 2 more weeks, the parents decided to withdraw from the study.

In comparison, previous studies on irritability in ASD with risperidone in 101 children resulted in a 56.9% reduction in the ABC-C/I score, as compared with a 14.1% reduction in the placebo group. On the CGI-I, 75.5% of those on risperidone showed a score of 1 (very much improved) or 2 (much improved), while 11.5% showed similar improvements on placebo.³¹ The effect size (Cohen’s d) for the CGI-I was 1.4 and for the ABC-C/I was 1.2.³²

In our clinical, unpublished retrospective data, we have a total of 26 echocardiograms and over 140 Holter monitor exams without any clinically significant findings; our data suggest that echocardiogram and Holter monitoring are not necessary in the absence of cardiovascular symptoms. We found a consistent change in blood pressure and pulse in the propranolol group although with rare exception, not clinically significant.

Telehealth study visits allowed participants who would otherwise not be able to participate in a research study to attend weekly study visits. This study demonstrated the feasibility of such clinical research studies in ASD, requiring low resource while allowing for safety and convenience suitable for this population. The blood pressure and pulse monitoring likewise were successful as this is now often a routine procedure in clinical medicine.

Although beyond the scope of the data obtained in this report, the mechanism by which propranolol might be working in order to reduce the symptoms which we are targeting is of interest. The simplest explanation would be a peripheral mechanism utilizing sympathetic inhibition. This would most likely occur at low doses and so it does not explain the apparent need for higher doses. A central beta-adrenergic blockade could be another straightforward hypothesis. Direct inhibition of areas such as the amygdala³³ would block the fear or rage, which is likely to accompany the targeted behaviors. The amygdala, hippocampus, and cortical circuits are believed to trigger fear memories, and this can be blocked with propranolol.³⁴ Direct inhibition of the locus coeruleus would block hyperarousal, and if the locus coeruleus is firing in a hypertonic frequency, inhibition could allow for improved phasic functioning with the result that cortically based strategies could allow for solutions to problems rather than frustration leading to rage.³⁵ Consolidation and reconsolidation of memories are dependent on the functioning of the noradrenergic system. This mechanism has been studied in post-traumatic stress disorder (PTSD). In our largely nonverbal subjects, it is difficult to ascertain if there are memories that are triggering rage, however, in more verbal subjects, there has been shown to be a high correlation of PTSD and ASD.^36,37 Again, it would be likely that this effect would be seen at lower doses due to the lipophilic properties of propranolol, and so it is possible that other mechanisms may explain our findings. Although there has not been much research reported on high dose of propranolol, there have been 2 animal studies that reported its blocking voltage-gated sodium channels (which was not the case for metoprolol or nadolol), thus resembling local anesthetics by hyperpolarizing and slowing the recovery of the axons. The blockade of these Na channels is also present in antiepileptic medications such as valproic acid and topiramate, which may explain the behavioral effects.^38,39 If propranolol is shown in more definitive studies to have the effects reported on here, then the understanding of the mechanism of action would be of great benefit in the use and design of medications for this indication.

Limitations

As this was a feasibility pilot study, we had a small sample that limits the generalization of our results to a wider population, and therefore, until a placebo-controlled, double-blind study with an adequate number of subjects is conducted, the clinical value of this report is limited. However, the order of the treatment assignment was randomized, and the study itself utilized a double-blind, placebo-controlled design. Moreover, our findings are consistent with our retrospective study as well as other case reports in the literature. Therefore, this study provides further evidence justifying a large, gold standard clinical trial. In addition, because this trial aimed to reduce chronic and severe challenging behaviors, the crossover study design was ethically suitable for this population, allowing all participants whose severe symptoms were debilitating (ie, an average ABC-C/I score of 28 at enrollment) to have access to the study drug.

A carryover drug effect was observed between the washout from drug to placebo in at least 1 participant, which confounded the initial estimate of the treatment effects. Future studies can avoid such carryover effects by sufficiently extending the washout period between treatment phases or by employing parallel groups.

Future Direction

Considerable progress has been made in evidence-based interventions for challenging behaviors in ASD with the use of antipsychotic medications and function-based interventions. Despite the widespread use of these and other interventions in clinical settings, the literature on these methods continues to exist as silos. As a result, concerted interdisciplinary research efforts in alleviating challenging behaviors are lacking. A large multisite randomized clinical trial of risperidone in combination with parent training found that such training reduced the dose of risperidone needed.⁴⁰ It remains to be seen if similar findings can be found with propranolol. As ASD is a neurobiological condition and challenging behaviors are thought to be a manifestation of both environmental (learned) and physiological variables, an examination of combined behavioral and pharmacological intervention could have a targeted impact on these behaviors that interfere with learning, independence, and quality of life.