Design and analysis of early phase adaptive platform trials in the era of precision oncology

We need efficient, innovative and robust early phase I/II trial designs to swiftly test new cancer therapies and ensure safe and effective ones are being progressed to later stage trials. Recent advances in precision oncology (which aims to match cancer treatments to the patient's unique form of cancer) have motivated innovative trial designs, particularly the idea of master protocol (e.g., basket, umbrella and platform trial), for the evaluation of molecularly targeted cancer therapies. Early phase adaptive platform trials (EP-APTs) are such innovative approaches and indisputably can provide efficiency improvements, statistically or operationally or both: (a) statistically - allowing within-trial information sharing of safety and preliminary efficacy data (and hence increase in precision or power); (b) operationally - evaluating several targeted therapies for one (or more) diseases concurrently, and accepting additions of new treatment arms or patient population during the trial[1]. EP-APTs often also incorporate pre-specified changes (adaptations) to trial aspects to be made by analysing interim data. Such adaptations include early stopping of treatment arm(s) as soon as enough evidence is gathered or allocating more patients to treatments showing greater benefits.