Trials and Tribulations: Co-Clinical Trials of Autism in Mice and Humans
Despite promise in animal models, novel therapeutics consistently fail in clinical trials of autism. Due to the heterogeneous nature of autism, treatment will likely only ameliorate symptoms in a subset of patients.
There is a need to simultaneously test promising new compounds while understanding and predicting which patients will respond to each therapy. To determine what might contribute to response susceptibility, we propose the implementation of co-clinical trials in autism. In co-clinical trials, novel compounds, with established safety profiles, are simultaneously tested in mouse models of autism as well as in Phase 2 trials in humans with autism.
A high-throughput protocol for mouse models of autism with genetic modifications in known autism genes is discussed, which encompassess chronic treatment over development, multiple magnetic resonance imaging time-points, behavioural tests before, during, and after treatment, and microbiome and epigenetic analyses. Human subjects also follow an extensive phenotyping protocol, including behavioural testing, as well as genetics, epigenetics, and imaging assessments. To ensure scientific rigor, both the preclinical and clinical studies are blinded, with placebo-treated controls, with sample size deteremined by power analyses. To establish the optimal regime, three co-clinical trials were created assessing promising therapeutics in autism: oxytocin, Tideglusib (a glycogen synthase kinase 3b inhibitor), and Arbaclofen (a GABA agonist). The protocol was modified with each iteration, optimizing the protocol based on the findings from each trial.
The first trial found no differential response to oxytocin treatment in mouse models in both neuroanatomical and behavioural phenotypes, and only some trending effects in humans on higher cognition measures. The protocol was modified for the Tideglusib trial, by narrowing the focus of the behavioural tests, increasing power by backcrossing onto the same background strain, changing the administration method, as well as by altering some other procedures. Analysis to date shows some differential treatment response was seen in the mouse models, with the Shank3 knock-out mouse model showing exacerbated hyperactivity (q=0.00) and non-anxiety (q=0.043) , and the Arid1b heterozygous deletion showing significant (q=0.01) changes in the volume of the insular cortex with treatment. The human trials also demonstrated some treatment response, with significant (q=0.05) differences on measures of social withdrawal, repetitive behaviours, as well as daily living skills, memory, and sleep quality.
A comparison across species in both drugs remains to be performed. The next co-clinical trial will be on the drug Arbaclofen, and the protocol will again be modified to address issues of the former trials. Ultimately, we hope to establish a rigorous multi-modal, multi-species approach of assessing treatment, with the intention of stratification by response susceptibility.