Case studies: Adaptive designs

Early-stage adaptive trials

Dose-escalation trials

A Phase I dose-escalation trial based on the Bayesian logistic regression model (BLRM) with overdose control was designed to provide a thorough characterization of the dose-toxicity relationship and identify the maximum tolerated dose (MTD) for a new oncology therapy. The design was initially developed for a single-agent trial and later extended to a combination therapy trial (new agent plus pembrolizumab). Mediana Inc has built a custom software tool for this trial (R package with a simulation engine written in C++) to perform an efficient evaluation of the trial designs under dozens of different dose-toxicity scenarios and provided support to Safety Review Committee meetings to analyze the toxicity data at the end of each cohort.

For information on the software tool developed for this Phase I trial, see the custom software page.

Late-stage adaptive trials

Sample size re-estimation

Multiple adaptive designs for Phase III neuroscience trials (e.g., dementia, bipolar depression) have been developed. Each adaptive design included one or two interim looks and supported an option to increase the total sample size in the trial to improve the probability of success as well as standard early stopping options (early stopping due to superior efficacy or futility). Mediana Inc has built a series of custom software tools (Windows-based applications or web applications with a graphical-user interface) that enabled the sponsor to efficiently evaluate the operating characteristics of each adaptive design under multiple treatment effect scenarios. The proposed adaptive designs were accepted by the FDA without any changes or comments.

For information on the software tools developed for these Phase III trials, see the custom software page.

Trial arm selection

Two adaptive designs for Phase III trials in rare diseases with several dose-control comparisons have been created. These designs included decision rules for selecting the best dose of an investigative treatment at the interim analysis based on the predicted probability of success. These decision rules relied on the primary efficacy endpoint or a surrogate endpoint. In addition, other flexible rules were enabled, e.g., each trial could be stopped early due to superior efficacy or lack of efficacy (futility), and a rule for potentially increasing the total number of patients was also added. Mediana Inc has developed custom software tools (Windows-based applications with a graphical-user interface) that facilitated an efficient simulation-based evaluation of numerous parameters of the decision rules at the interim analysis in each trial. The proposed adaptive designs were accepted by the FDA without any changes or comments.

For information on the custom software tools developed for these Phase III trials, see the custom software page.

Population selection

Several adaptive designs have been developed for Phase III oncology trials with biomarker-driven designs. A biomarker that was believed to be predictive of treatment response was pre-defined in each trial and the efficacy of the experimental treatment was examined in the general population and the biomarker-positive subpopulation at an interim analysis. Patients with a biomarker-positive status were expected to experience a greater treatment benefit compared to biomarker-negative patients and a decision rule was set up to select the best course of action at the interim look, e.g., continue to enroll all patients or discontinue the enrollment of biomarker-negative patients. In addition to this population selection rule, a hypothesis selection rule was proposed to improve the probability of success at the final analysis as well as a decision rule for adjusting the target number of events in the trial. Mediana Inc has built custom software tools (Windows-based applications with a graphical-user interface) to run multiple sets of simulations to better understand the performance of candidate population selection and hypothesis selection rules in each trial. The proposed adaptive designs were accepted by the FDA without any changes or comments.

For information on the custom software tools developed for these Phase III oncology trials, see the custom software page.