Adaptive designs for dual-agent phase I dose-escalation studies

Jennifer A. Harrington*, Graham M. Wheeler, Michael J. Sweeting, Adrian P. Mander, Duncan I. Jodrell

*Corresponding author for this work

Research output: Contribution to journalReview articlepeer-review

43 Citations (Scopus)


Anticancer agents used in combination are fundamental to successful cancer treatment, particularly in a curative setting. For dual-agent phase I trials, the goal is to identify drug doses and schedules for further clinical testing. However, current methods for establishing the recommended phase II dose for agents in combination can fail to fully explore drug interactions. With increasing numbers of anticancer drugs requiring testing, new adaptive model-based trial designs that improve on current practice have been proposed, although uptake has been minimal. We describe the methods available and discuss some of the opportunities and challenges faced in dual-agent phase I trials, as well as giving examples of trials in which adaptive designs have been implemented successfully. Improving the design and execution of phase I trials of drug combinations critically relies on collaboration between the statistical and clinical communities to facilitate the implementation of adaptive, model-based designs.

Original languageEnglish
Pages (from-to)277-288
Number of pages12
JournalNature Reviews Clinical Oncology
Issue number5
Publication statusPublished - May 2013
Externally publishedYes

Bibliographical note

Funding Information:
The authors would like to thank the referees for their helpful comments and contribution to the manuscript. J. A. Harrington and G. M. Wheeler are both PhD students whose research is funded by Cancer Research UK and the Medical Research Council UK, respectively. G. M. Wheeler, A. P. Mander and M. J. Sweeting are supported by the Medical Research Council (Grant number G0800860).


Dive into the research topics of 'Adaptive designs for dual-agent phase I dose-escalation studies'. Together they form a unique fingerprint.

Cite this