Bayesian dose‐finding phase I trial design incorporating pharmacokinetic assessment in the field of oncology

Abstract

The main purpose of a phase I dose-finding study in the field of oncology is to evaluate toxicity and pharmacokinetic (PK) data and to estimate the optimal dose (OD) for subsequent clinical trials. From a pharmacological perspective, PK information is considered as an appropriate indicator for evaluating the degree of drug intervention in humans. Dose proportionality is typically assessed to investigate the PK properties of a drug. If we rely solely on the dose-exposure relationship, then when this relationship is not proportional, eg, if there is saturation of drug elimination or absorption, the performance of OD selection may be affected. This may be because any exposure ratio between two dose levels differs from the dose ratio between the dose levels. In addition, large inter-individual variability in exposure affects the occurrence of toxicity. Therefore, incorporating PK assessment in a phase I dose-finding trial may enable us to more precisely estimate OD. In most oncology dose-finding clinical trials, however, the result of PK analysis is not explicitly incorporated in the dose-finding determination analysis. In this study, we propose a Bayesian approach to incorporating PK assessment into OD estimation in a dose-finding trial. Our proposed approach incorporates into the statistical model an adjustment based on the latest area under the time-concentration curve assessment. The simulation study shows that compared with standard methods, the proposed method may be better able to select the correct ODs across a variety of realistic settings.

Comparison of design methods for a safety run-in phase of a phase II clinical trial

“…Although Rolling 6 is our recommendation for a Phase II safety run-in that requires pharmacokinetic assessment, for standard Phase I trials methods have been proposed to incorporate pharmacokinetic assessment. 28…”

Pharmacometrics meets statistics—A synergy for modern drug development

“…It was found that, for the studied scenarios, trial efficiency (as measured by the number of observed dose‐limiting toxicities or the probability of correct MTD selection) is not improved, but the estimation of the dose–toxicity curve may be better compared to the designs that do not utilize PK data. Another recent paper 50 proposed a Bayesian dose finding design using dose‐exposure data in the escalation rule and found that it may improve the probability of correct dose selection. An important and potentially useful consideration is the assessment of distributed designs (evaluating a range of doses) versus concentrated design (testing only placebo and the maximum dose).…”

Exposure driven dose escalation design with overdose control: Concept and first real life experience in an oncology phase I trial

“…However, it has been shown by Ursino et al [ 23 ] in a simulation study that none of these approaches has a significant edge over previous methods. Ursino et al [ 23 ] and more recently Takeda et al [ 24 ] have proposed more elaborate methods with attempts of a better integration of the PK. However in both manuscripts simple estimation of the drug exposure were used while the between patients variability was not appropriately characterized.…”

“…The potential variability in bioavailability or of absorption in case of flip-flop kinetics [ 25 , 26 ] are ignored. Takeda et al [ 24 ] considered simplified non-linear kinetics not reflecting what is observed in case of saturated absorption for instance or for target mediated drug disposition antibody. Furthermore, the authors used inflated measurement errors in lieu of between patients variability.…”