OBJECTIVE: Clinical trial simulation (CTS) was used to select a robust design to test the hypothesis that a new treatment was effective for Alzheimer's disease (AD). Typically, a parallel group, placebo controlled, 12-week trial in 200-400 AD patients would be used to establish drug effect relative to placebo (i.e., Ho: Drug Effect = 0). We evaluated if a crossover design would allow smaller and shorter duration trials. MATERIALS AND METHODS: A family of plausible drug and disease models describing the time course of the AD assessment scale (ADAS-Cog) was developed based on Phase I data and literature reports of other treatments for AD. The models included pharmacokinetic, pharmacodynamic, disease progression, and placebo components. Eight alternative trial designs were explored via simulation. One hundred replicates of each combination of drug and disease model and trial design were simulated. A 'positive trial' reflecting drug activity was declared considering both a dose trend test (p < 0.05) and pair-wise comparisons to placebo (p < 0.025). RESULTS: A 4 x 4 Latin Square design was predicted to have at least 80% power to detect activity across a range of drug and disease models. The trial design was subsequently implemented and the trial was completed. Based on the results of the actual trial, a conclusive decision about further development was taken. The crossover design provided enhanced power over a parallel group design due to the lower residual variability. CONCLUSION: CTS aided the decision to use a more efficient proof of concept trial design, leading to savings of up to US 4 M dollars in direct costs and a firm decision 8-12 months earlier than a 12-week parallel group trial.
We have not found any resources mentioned in this publication.
SciCrunch is a data sharing and display platform. Anyone can create a custom portal where they can select searchable subsets of hundreds of data sources, brand their web pages and create their community. SciCrunch will push data updates automatically to all portals on a weekly basis. User communities can also add their own data to SciCrunch, however this is not currently a free service.