Our Approach
There are many new oncologic drug candidates in development with novel mechanisms of action. The selection of dose and dosing schedules during early oncology drug development remains very empirical. Phase I dose selection still relies heavily on the maximum tolerated dose paradigm, and Phase II studies are not generally well designed to assess dose-response. Moreover, the accompanying analyses of clinical trial data are often poorly informative. Longitudinal data (e.g., tumor size measurements, neutrophil count) are often summarized in a single number (e.g., objective response rate, grade 4 neutropenia). This limits the ability to learn from early trials and contributes to high failure rates in Phase III.
Drug development in oncology can benefit from quantitative application of drug-disease causal models that model longitudinal data and account for subject covariates, prognostic factors, and gene expression and protein profiles. This approach offers a powerful means to enhance learning from early clinical data by using all available information -- and therefore reducing uncertainty in understanding of drug response that contributes to better decision-making and design of Phase III trials (FDA Experience with End of Phase IIa Meetings: An Attempt to Improve Drug Development Decisions).
Project deliverables might include:
- Population PK/PD analysis (e.g., to support bridging strategies between Japan and US/EU)
- Exposure-response modeling to support regulatory decisions in later development
- Simulation of expected survival in Phase III studies based on proprietary Phase II NCE data