Generalized Statistical Approaches for the Design for Phase I Trials.

Phase I Trials;Dose-finding Algorithms;Time-to-Event Trials;Bayesian Statistics;Two-Agent Trials;Public Health;Health Sciences;Biostatistics

Jia, NanWang, Lu ;

University of Michigan

关键词: Phase I Trials;    Dose-finding Algorithms;    Time-to-Event Trials;    Bayesian Statistics;    Two-Agent Trials;    Public Health;    Health Sciences;    Biostatistics;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/93913/jnan_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

This research examines new design approaches for Phase I clinical trials, which are designed to study doses of the same agent or dose combinations of two agents in a small group of patients to determine the maximum tolerated dose or dose combination.Our first focus is to propose an adaptive accelerated Biased Coin Design (aaBCD) that generalizes the traditional BCD design algorithm by incorporating an adaptive weight function based upon the amount of follow-up of each enrolled patient, so that the dose assignment for each eligible patient can be determined immediately with no delay, leading to a shortertrial overall.We later focus on a generalized version of the Continual Reassessment Method (CRM), denoted gCRM, for identifying the maximum tolerated dose combination of two agents. For each dose of one agent, we apply the traditional CRM to study doses of the other agent; each of these CRM designs assumes the same dose-toxicity model, as well as the value of the parameter used in the model. However, each model includes a second parameter that varies among the models in an effort to allow flexibility when modeling the probability of dose-limiting toxicity (DLT) of all combinations, yet borrow strength among neighboring combinations as well.We lastly extend the gCRM by incorporating results for patients with incomplete follow-up into the decision rule for the assignment of a dose combination to the next available patient. We derive methods that account for the differing amounts of follow-up that could occur for the two agents and propose the use of a copula function to adjust for early- or late-onset DLTs. We show an optimal weight via theory and simulations that when the DLT times are distributed the same as assumed model, the optimal weight performs best among all the weight functions under consideration.