Did arbaclofen really fail in autism and fragile X?
Walter Kaufmann, MD, is co-director of the Fragile X Syndrome Program and a member of the department of Neurology at Boston Children’s Hospital. He was site principal investigator for three arbaclofen trials sponsored by Seaside Therapeutics and currently advises the company on data analyses. This post is second in a two-part series on clinical trials in autism spectrum disorders. (Read part 1)
The outcomes of drug trials in autism spectrum disorder (ASD) have, to date, been mixed. While atypical neuroleptic drugs have been effective for treating disruptive behavior in people with autism and are FDA-approved for that purpose, no available psychotropic drug has improved the core symptoms of ASD, such as social interaction deficits or stereotypic behaviors.
The heterogeneity—diversity—of ASD in both causes and symptoms may explain treatment failures to some extent. However, we have also lacked drugs targeting the brain mechanisms that underlie ASD. For this reason, targeted trials in fragile X syndrome, informed by neurobiology, have raised hopes of finally addressing core autistic symptoms.
Fragile X syndrome is a genetic disorder in which ASD occurs in 15 to 40 percent of cases. Initial results from a Phase 2 trial using the GABA-B agonist arbaclofen demonstrated relatively selective improvements in social avoidance in a wide age-range sample of subjects. They spurred great optimism about the genetic disorder approach to ASD—tackling autism that occurs as part of specific genetic syndromes as a path into more common, unexplained forms—and led to Phase 2 and 3 trials of children and adolescents with “idiopathic” ASD (of unknown cause).
The idiopathic ASD trials (see here, here and here) included individuals with a social impairment profile similar to that of subjects who responded to arbaclofen in the fragile X trial. They replicated some of the improvements in social function, albeit only in patients who had the same examiner throughout the trial or who had an IQ higher than 70. These exciting results, recently presented at the 2013 meeting of the Child Neurology Society, were disappointingly not reproduced by open-label follow-up trials in fragile X, which together were intended to include nearly 400 patients. This led Seaside Therapeutics to terminate the trials, to the disappointment of many patients.
Rethinking clinical study designs in autism
The inconsistent outcomes of these trials raise several critical issues for designing appropriate treatment studies for ASD and related disorders.
First, the cohort analysis approach to drug trials may not be the most appropriate for conditions that are highly heterogeneous in their causes and symptoms. One can argue that, from the outset, only a subgroup of individuals would be expected to respond to the drug. However, that raises the question of how to distinguish the group of potential responders. Behavioral rating scales, such as those used in the arbaclofen trials, have important limitations in defining autism. For instance, social avoidance seems to better describe social anxiety than the aloofness that characterizes ASD. Moreover, there is an important element of subjectivity in behavioral evaluations, as illustrated by the need to maintain the same observer throughout the trial.
Observational research commonly employs two analytical approaches: the use of multiple complementary measures (“converging evidence”) and biomarkers (objective, instrument-based measures). These approaches are not applied to drug trials since they are not part of the FDA regulatory process. Thus, the positive Phase 2 trials of arbaclofen—while unique in the magnitude and specificity of social improvement—were considered failures, because only secondary measures or post-hoc analyses demonstrated efficacy.
Finally, we need to redefine endpoints. If the novel treatments for ASD target mechanisms of disease and not symptoms, why should we continue focusing on improving a specific symptom? For some patients with fragile X, a drug may be efficacious for ASD symptoms; for others, it may reduce attention deficits or other behavioral problems. A more personalized approach to measuring outcomes, using biomarkers, has been proposed for other disorders, such as traumatic brain injury.
Until all these issues are addressed, it will be difficult to conclude whether or not a drug trial was successful. Under the current approach, there is a potential for missing valuable new treatments—a huge loss for individuals and families with ASD and related disorders.
