A Bayesian Industry Approach to Phase I Combination Trials in Oncology

Operating Characteristics 100 6.3 Methodology 1026.3.1 Base Model 102 6.3.2 Metrics for Dosing Recommendations 1046.3.2.1 Posterior Distributions 104 6.3.2.2 Predictive Distributions 106 6.3.2.3 Formal Decision Analysis 1066.3.3 Prior Distributions 106 6.3.3.1 Weakly Informative Priors 107 6.3.3.2 Prior Effective Sample Size 107 6.3.3.3 Meta-Analytic-Predictive (MAP) Priors 108 6.3.3.4 Mixture Priors 1096.3.4 Combination Models 111 6.3.4.1 Combination of Two Agents 112 6.3.4.2 Combination of Three Agents 1146.4 Applications 115 6.4.1 Dual-Combination Trial 1156.4.1.1 Study Design 115 6.4.1.2 Model and Prior Distributions 115 6.4.1.3 Trial Conduct: Dose Escalations andMTD Declaration 117 6.4.2 Triple-Combination Trial 1206.4.2.1 Study Design 120Phase I trials are still perceived by many as simple. Although this is controversial in the single-agent setting, it is certainly mistaken for combinations of two or more compounds. In oncology, the challenges are many: while keeping patient safety within acceptable limits, the trials should be small, adaptive, and enable a quick declaration of the maximum tolerable dose (MTD) and/or recommended phase II dose (RP2D).