A Bayesian Approach for Dose-Escalation in a Phase I Clinical Trial Incorporating Pharmacodynamic Endpoints

Abstract

Bayesian decision procedures have already been proposed for and implemented in phase I dose-escalation studies in healthy volunteers. The procedures have been based on pharmacokinetic responses reflecting the concentration of the drug in blood plasma and are conducted to learn about the dose-response relationship while avoiding excessive concentrations.However, in many dose-escalation studies, pharmacodynamic endpoints such as heart rate or blood pressure are observed, and it is these that should be used to control dose-escalation. These endpoints introduce additional complexity into the modelling of the problem relative to pharmacokinetic responses. Firstly, there are responses available following placebo administrations.Secondly, the pharmacodynamic responses are related directly to measurable plasma concentrations, which in turn are related to dose. Motivated by experience of data from a real study conducted in a conventional manner, this paper presents and evaluates a Bayesian procedure devised for the simultaneous monitoring of pharmacodynamic and pharmacokinetic responses. Account is also taken of the incidence of adverse events. Following logarithmic transformations, a linear model is used to relate dose to the pharmacokinetic endpoint and a quadratic model to relate the latter to the pharmacodynamic endpoint. A logistic model is used to relate the pharmacokinetic endpoint to the risk of an adverse event.

Dose‐finding methods for Phase I clinical trials using pharmacokinetics in small populations

“…Three Bayesian dose‐finding methods identified from the literature and modified to be applied in the same context are described in detail. In these methods PK measurements could enter in the model as covariates (Piantadosi and Liu, ; Whitehead et al., ) or as the dependent variable in regression (Patterson et al., ; Whitehead et al., ). Two variations of such methods are proposed and two other methods have been added to test the behaviors of the two philosophies “dose ‐ tox” and “dose ‐ auc ‐ tox.” The first approach, “dose ‐ tox,” is the most common and entails estimating the relationship between the probability of toxicity, , and dose directly.…”

“…In the third method, called PKLOGIT, inspired by Whitehead et al. () we use instead of dose as a covariate in a logistic regression model for .…”

“…These priors, although not as used by Whitehead et al. (), have been chosen in order to make this method more comparable with other methods presented in this paper. Lastly, we borrowed the hierarchical model of PKLIM for that is necessary for computing the posterior probability of toxicity associated with each dose.…”

“…() and Whitehead et al. (), who proposed a Bayesian procedure with a nested hierarchical structure, in which PK/PD data are used to fit an overall dose–response relationship, so that PK/PD data are considered as dependent variables rather than as covariates. In both cases, decision theory is applied to maximize a chosen gain function in order to select the next dose.…”

Pharmacological modeling and biostatistical analysis of a new drug

“…However, presently some trials use Bayesian or other dose escalation methods to determine the MTD. [31][32][33][34] A phase I trial also evaluates the PK/PD properties of multiple doses of the drug, such as its absorption and elimination characteristics, its dosing schedule and its concentration versus effect properties. Thus, b In rare cases, the first human trial is a phase 0 or a micro-dosing study to evaluate the drug's PK/PD properties, performed in special cases where an experimental drug shows a significant potential in the preclinical stage to warrant an acceleration of its development and a reduction in its costs.…”