Duchenne Muscular Dystrophy (DMD) is a life-threatening, sex-linked, pediatric rare disease, primarily affecting boys. It is characterized by progressive muscle degeneration, weakness, and eventually functional loss. DMD is caused by a mutation in the dystrophin gene, a protein needed to maintain muscle integrity, and for improving signaling and growth in differentiation of the muscle tissue. DMD is diagnosed at four to five years of age. By age twelve to fourteen, patients lose the ability to walk. Ultimately, patients succumb to the disease in their mid-twenties due to complications including cardiomyopathy and impaired respiratory function.
Corticosteroids are the standard for treating symptoms and can delay loss of ambulation. More recently dystrophin-restoring treatments such as exon skipping therapies, stop codon-read-through, and gene therapies hold promise to increase DMD patients’ life expectancies. However, more research is still needed to better understand the mechanisms of muscle atrophy and defects to develop more efficacious therapies.
Because DMD is a pediatric disease, there is a developmental component to how children perform in motor function tests including the 6-minute walk test (6MWT) that are used to assess drug effects. Younger subjects show some improvement in motor function over time. However, when the disease takes over, motor function declines. Hence, the 6MWT trajectory shows an “up and down” effect. The dependence on age is important to assess in drug trials because, otherwise, a drug effect may be missed or a false conclusion of efficacy may be derived.
In a recent webinar with Dr. Joga Gobburo (University of Maryland), I reviewed how a disease progression model, developed using Phoenix NLME, was used to better understand variability in the 6MWT. The project sought to determine if a quantitative approach can discern potential drug effects on the 6MWT trajectory given the age-dependent variability. Can a disease progression model be leveraged to inform downstream trial design, examine drug effects, and then simulate potential drug effects in future trials?
Disease progression modeling for rare diseases is limited by data availability and often data from the literature is initially used. Although aggregate-level information on age, race and weight is typically summarized in the literature, the lack of detailed patient demographic information that drives variability, such as type of steroid treatment, and patient factors including dystrophin mutation, baseline cardiac and respiratory function, may hinder building a disease progression model.
Six structural models with increasing complexity were evaluated for their ability to accurately predict the training dataset and a novel dataset obtained from the literature of previously published DMD clinical trials. A Linear Model with Simultaneous Estimation using Phoenix NLME was developed to bucket various forms of 6MWT variability. The model was a series of linear models that did not require age to be fixed a priori but rather could let the data speak for itself and estimate the age dependence of the variability. The disease progression linear model did a reasonable job of accounting for variability seen in the observed training datasets and for predicting 6MWT performance in the training datasets as well as a novel dataset.
To evaluate a drug for a clinical meaningful effect, a standard sample size calculation was performed to determine how many patients would be needed to detect a drug effect when using the 6MWT as a clinical endpoint. If studying an aggregate age group, 160 subjects per trial arm would need to be recruited. This is not feasible for a rare disease such as DMD. But by using the disease progression model, a small, non-age stratified trial could detect a hypothetical drug effect using only 6 subjects per treatment arm. The model was shown to parse out the age dependence on variability and predict the dose response for the hypothetical drug.
This approach highlighted the power of modeling and simulation (M&S) for rare disease drug development. It demonstrated how a quantitative platform can be used to simulate different drug trial scenarios to assess sources of patient variability with the potential to allow for much smaller drug trials to be conducted.
To learn more about this project, please click the link below to watch the recorded webinar. A special thanks to Drs. Joga Gobburo and Suzanne Minton for helping to prepare and execute the webinar.