Optimizing Schedule of Administration in Phase I Clinical Trials
In conventional phase I studies, each patient is assigned to a dose of an experimental agent under study, receiving either a single administration or a course consisting of several administrations of the agent. Patients are followed for a relatively short period of time during which a dose-limiting toxicity (DLT) may occur. At the end of the follow-up period, a binary outcome indicating the presence or absence of DLT is recorded for each patient. Generally, the maximum tolerable dose (MTD) is considered the largest dose that is “safe,” that is, that does not present a practical limitation to therapy. Many designs using this approach exist [1–5]. These designs have seen widespread use largely because they facilitate adaptive dose-finding methods wherein doses and outcomes of previous patients are used to select doses for new patients. However, these designs have some limitations due to their simplified representation of actual clinical practice.
First, in many clinical settings, physicians administer an agent more than once to a patient and monitor the long-term toxicity related to the cumulative effects of the agent. In such settings, the nominal “dose” is actually the dose given per administration, or the total dose given over a fixed number of administrations. For example, multiple cycles of chemotherapy may be administered with the aim to effectively ...