Presented at the 16th Annual NEALS Meeting in Clearwater, Florida, October 4, 2017, and also at the 28th International Symposium on ALS/MND in Boston, Massachusetts, on December 9, 2017.
Background: Disease heterogeneity is widely believed to be a confounding factor in the analysis of ALS clinical trials. In particular, deaths and slowly progressing patients may be the root causes of the observed heterogeneity. As a step towards solving this problem, we report on several ALS predictive models and use them in simulations where we compare randomizing on predicted outcomes to traditional methods.
Objectives: Validate the models using BENEFIT-ALS placebo arm data and apply the models to simulations.
Conclusions:
- A GBM model platform capable of predicting several key ALS disease progression metrics, including ALSFRS-R, vital capacity and survival, using only data available at baseline without a run-in period has been developed using the PRO-ACT ALS database and rigorously validated using the external, contemporary BENEFIT-ALS placebo arm data set.
- Internal validations provide evidence of reproducibility
- External validations provide evidence of generalizability
- Two useful drug development applications were developed using the ALSFRS-R, survival and vital capacity models:
- Patients were stratified into slowly, average and rapidly progressing subgroups
- Tertiles were used to create randomization strata that proved superior to traditional methods of randomization in terms of decreasing sample size
Authors: Albert A. Taylor, Danielle Beaulieu, Samad Jahandideh, Lisa Meng, Amy Bian, Jinsy Andrews, David L. Ennist