NOVEL BAYESIAN ADAPTIVE CLINICAL TRIAL DESIGNS IN EARLY PHASES

Early phase, or phase I and phase II, trials are the first step in testing new medicines that have been developed in the lab. The main goal of phase I clinical trials is to establish the recommended dose of new drugs for phase II trials. For the cytotoxic drugs, the goal is to find maximum tolerated dose (MTD). The guiding principle for dose escalation in phase I trials is to avoid exposing too many patients to subtherapeutic doses while preserving safety and maintaining rapid accrual. Therefore, dose escalation methods, especially Bayesian designs, are recommended to be used in phase I trials. There are many proposed Bayesian phase I adaptive designs, among them, continual reassessment method (CRM) is the firstly proposed pioneered Bayesian design. The CRM needs pre-specification of a series of prior guesses of toxicity probabilities of each investigated doses, known as the skeleton, using a parametric model, and then continuously updates the estimate of the dose-toxicity curve based on accumulating data. By using a dose-toxicity model, the CRM efficiently pools data across doses and adaptively makes the decision of dose assignment and selection. Two chapters of the thesis devote to development of the CRM design (chapter 2) and to extend the CRM design (chapter 3). Specifically, chapter 2 deals with the issue of skeleton pre-specification in the CRM design. We propose an automatic way to generate multiple skeletons for Bayesian model averaging CRM (BMA-CRM), an extension of robust version of the CRM, to avoid arbitrary specification of skeletons with improving performances compared to the original CRM and BMA-CRM designs. Chapter 3 deals with bridging studies, or follow-up trials. The emergence of bridging studies is due to different ethnic populations with different responses to a same drug and consequentially attaining different MTDs. Therefore, conventional one-size-fit-all paradigm cannot work. But, despite variations among different ethnic populations, their drug responses still show somewhat similarities. Commonly, a landmark trial has been conducted and a MTD dose has also been established for a certain population. Thus, independent conducting a trial for a new population of ignoring information of the landmark trial is wasteful. Therefore, challenges of the bridging studies include: how to effectively use/borrow information of the historical landmark trial, and how to design trials to accommodate heterogeneities of different populations. In this chapter, we develop a novel design, Bridging-CRM, B-CRM, to borrow the landmark trial information based on a proposed mixture estimator and the CRM framework, and to acknowledge different populations\u27 heterogeneities of using the idea of multiple skeletons. Chapter 4 focuses on phase II design for biosimilar drug development. Biosimilar is a term that describes the equivalence of a generic version to an innovator\u27s biologic drug product; biosimilars are close, but not exact copies of biologic drugs already on the market. Guidelines for statistical methods to establish biosimilarity remain nonspecific because of the newness of biosimilars. It is therefore of high urgency to develop appropriate and reliable statistical methodologies for developing biosimilars. Some statistical methods have been proposed to assess biosimilarity, but none of them proposed designs in this field. However, biosimilar trials come with several challenges that are beyond the scope of the conventional randomized comparative trial design. First, when a biosimilar is ready to be tested in a randomized trial, the innovative reference drug has been in the market for many years and a huge amount of data on that drug has accumulated. It is critical to incorporate these rich historical data into the biosimilar trial design to improve trial efficiency. Another challenge when designing biosimilar trials is determining how to quantify and monitor the biosimilar during the trial. To address these issues, in chapter 4, we develop a new approach, the \textit{calibrated power prior} (CPP), to allow the design to adaptively borrow information from the historical data according to the congruence between the historical data and the data collected in the current trial. We also propose the \textit{Bayesian biosimilarity index} (BBI) to assess the similarity between the biosimilar and the innovative reference drug. In our design, we evaluate the BBI in a group sequential fashion based on the accumulating interim data, and stop the trial early once there is enough information to conclude or reject the similarity