Innovative Statistical Approaches for Clinical Trials

A fixed-sample-size clinical trial defines, prior to study initiation, a number of patients who will be enrolled in the trial and the doses that they will receive. The trial begins and enrollment continues until all the patients have been enrolled and assigned to treatment, at which the trial is stopped, that data are collected, cleaned, unblinded, and analyzed, and the results are made known. There are designs, however, that do not fix beforehand the number of subjects and/or the doses or treatments that will be used. Such designs are called adaptive designs, since the design of the trial will be adapted during the trial to information that is collected during the trial. In this module, we will explore adaptive trial designs, from simple variance-estimation strategies to play-the-winner methods, to deterministic and Bayesian methods in oncology. Advantages and pitfalls of each of the designs will be discussed.

• What are adaptive designs? Why use adaptive designs?
• N-adjustable designs -- sample size re-estimation
• Drop-the-loser
• Adaptive dose finding
• Biomarker-adaptive/Enrichment designs
• Seamless phase 1/2 or phase 2/3 designs
• Phase I designs (3+3, CRM, mTPI) to find maximum tolerated dose
• Examples