Design Innovations in Precision Promise: Adaptive Platform Clinical Trial in Pancreatic Cancer

Precision Promise is an adaptive platform trial in pancreatic cancer. Objectives include evaluating many therapies efficiently determining which patients benefit from which therapies. The primary endpoint is overall survival. Its statistical approach is Bayesian. The “trial” is actually a therapy-screening process with built-in seamless transition into confirmation mode for those therapies that show sufficient activity in at least one patient subtype. The design is revolutionary with many innovations never before used in a phase 3 registration trial, including these:

1. Many experimental arms; treatment arms enter and leave the trial over time
2. Adaptive randomization in arm-specific stage 1
3. Seamless shift, learn in stage 1 and confirm in stage 2
4. All patients count in an arm’s final analysis versus control regardless of stage
5. Common control arm (depends on patient subtype)
6. Compare experimental arm vs all controls in the trial via “time machine”
7. Continuous learning and updating information
8. Identify arms’ indications, if any, including as defined by biomarker targets
9. Interpretation of Type I error
10. Trial driven by predictive probability
11. Patients can be re-randomized (keeping overall survival as primary endpoint)
12. Longitudinal model of disease burden over time to inform overall survival