Designs for Single‐ or Multiple‐Agent Phase I Trials

Abstract

Phase I trials of cytotoxic agents in oncology are usually dose-finding studies that involve a single cytotoxic agent. Many statistical methods have been proposed for these trials, all of which are based on the assumption of a monotonic dose-toxicity curve. For single-agent trials, this is a valid assumption. In many trials, however, investigators are interested in finding the maximally tolerated dose based on escalating multiple cytotoxic agents. When there are multiple agents, monotonicity of the dose-toxicity curve is not clearly defined. In this article we present a design for phase I trials in which the toxicity probabilities follow a partial order, meaning that there are pairs of treatments for which the ordering of the toxicity probabilities is not known at the start of the trial. We compare the new design to existing methods for simple orders and investigate the properties of the design for two partial orders.

An adaptive phase I design for identifying a biologically optimal dose for dual agent drug combinations

“…Dose level (3,5) has the highest success probability Table III. Simulation results based on 500 simulated trials for scenario 3.…”

“…This dose level was recommended in 48 per cent of the simulated trials. Dose levels (2, 5) and (3, 4) satisfy the success region criteria, and hence the overall recommendation rate is 70 per cent including dose levels (2, 5), (3,5) and (3,4). The no recommendation rate is 0.2 per cent and the average sample size is 33.…”

“…Despite the increased testing of multiple agents in phase I oncology trials, few designs for dose escalation of two or more agents have been proposed [2][3][4]. A two-stage Bayesian design for identifying one or more acceptable dose pairs with informative priors characterizing the single agent toxicity curves and non-informative priors for characterizing agent-agent interactions was proposed by Thall et al [2]; a quasi-Bayesian design based on partial ordering of the toxicity probabilities (i.e.…”

“…A two-stage Bayesian design for identifying one or more acceptable dose pairs with informative priors characterizing the single agent toxicity curves and non-informative priors for characterizing agent-agent interactions was proposed by Thall et al [2]; a quasi-Bayesian design based on partial ordering of the toxicity probabilities (i.e. when the monotonicity of the dose-toxicity curve is not completely known at the start of the trial) was recommended by Conaway et al [3]; and a Bayesian design to identify the MTD of one agent in combination with each of the possible doses of the second agent was proposed by Wang and Ivanova [4]. These designs continue to base their dose finding algorithm on the dose-toxicity relationship on the assumption that higher dose levels will provide the maximal therapeutic efficacy.…”

Operating Characteristics of Restrictions on Skipping Dose Level for Adaptive Dose-Finding Method in Two-Agent Phase I Trials

“…Thall et al (2003) proposed a six-parameter model for the toxicity probabilities of the dose combinations and a toxicity equivalence contour for two-agent combinations. Conaway et al (2004) determined the simple and partial orders of the toxicity probabilities by defining the nodal and non-nodal parameters. Wang and Ivanova (2005) proposed a logistic-type regression for dose combinations that used the doses of the two agents as the covariates.…”