Predictive power to assist phase 3 go/no go decision based on phase 2 data on a different endpoint

Area of Science:

• Clinical pharmacology
• Oncology drug development
• Biostatistics

Background:

• Phase 3 go/no go decisions are critical in drug development.
• Oncology trials often use overall survival (OS) as a primary endpoint, but phase 2 trials may only be powered for progression-free survival (PFS).
• The relationship between PFS and OS can be unclear, complicating phase 2 to phase 3 transition decisions.

Purpose of the Study:

• To propose a method using predictive power to aid phase 3 go/no go decisions in oncology.
• To evaluate the strength of phase 2 efficacy to predict phase 3 OS success.
• To provide a formula for calculating predictive power using PFS, OS, or both.

Main Methods:

• Development of a predictive power formula based on phase 2 efficacy data.
• Calculation of predictive power using observed PFS effect, OS effect, or both.
• Comparative analysis of predictive power from different data combinations.

Main Results:

• Predictive power based solely on phase 2 PFS effect can be low and insensitive to trial parameters when prior OS information is limited.
• Phase 2 PFS data has limited added value to predictability when phase 2 OS data is available.
• The predictability of phase 3 OS success is enhanced when phase 2 OS data is incorporated.

Conclusions:

• Caution is advised when making phase 3 go/no go decisions based solely on phase 2 PFS outcomes.
• Phase 2 OS data should be considered whenever possible to improve decision-making accuracy.
• Integrating both PFS and OS data from phase 2 trials offers a more robust prediction of phase 3 success.