Abstract:
|
Pre-clinical experimentation in mice is a hallmark of pharmaceutical drug development, contributing to in vivo proof of efficacy, first in human (FIH) dose projection, safety evaluation, as well as clues about drug mechanism of action (MOA) and biomarkers. Pre-clinical investigations of anti-cancer immuno-therapeutics utilize syngeneic mouse tumor models hosting mouse immune systems to achieve many early development objectives. Syngeneic mouse tumors grown in in-bred mouse strains have been observed to display heterogeneous responses, whereby a fraction of tumors show resistance, partial, or complete response, to the same immuno-intervention. We propose a Bayesian hierarchical mixture response model, for the design and analysis of syngeneic mouse anti-cancer immuno-therapeutic studies. At the core of our methodology is flexible and powerful Bayesian learning to measure posterior beliefs about scientific hypotheses with, possibly multiple, endpoint decision criteria. The relative qualitative advantages, power, and limitations for identifying superior treatments and combinations are demonstrated with examples from syngeneic mouse immuno-intervention studies and by simulation.
|