Bayesian adaptive designs provide a critical approach to improve the efficiency and success of drug development that has been embraced by the US Food and Drug Administration (FDA). This is particularly important for early phase trials as they form the basis for the development and success of subsequent phase II and III trials. The objective of this book is to describe the state-of-the-art model-assisted designs to facilitate and accelerate the use of novel adaptive designs for early phase clinical trials. Model-assisted designs possess avant-garde features where superiority meets simplicity. Model-assisted designs enjoy exceptional performance comparable to more complicated model-based adaptive designs, yet their decision rules often can be pre-tabulated and included in the protocol—making implementation as simple as conventional algorithm-based designs. An example is the Bayesian optimal interval (BOIN) design, the first dose-finding design to receive the fit-for-purpose designation from the FDA. This designation underscores the regulatory agency's support of the use of the novel adaptive design to improve drug development. Features Represents the first book to provide comprehensive coverage of model-assisted designs for various types of dose-finding and optimization clinical trials Describes the up-to-date theory and practice for model-assisted designs Presents many practical challenges, issues, and solutions arising from early-phase clinical trials Illustrates with many real trial applications Offers numerous tips and guidance on designing dose finding and optimization trials Provides step-by-step illustrations of using software to design trials Develops a companion website (www.trialdesign.org) to provide freely available, easy-to-use software to assist learning and implementing model-assisted designs Written by internationally recognized research leaders who pioneered model-assisted designs from the University of Texas MD Anderson Cancer Center, this book shows how model-assisted designs can greatly improve the efficiency and simplify the design, conduct, and optimization of early-phase dose-finding trials. It should therefore be a very useful practical reference for biostatisticians, clinicians working in clinical trials, and drug regulatory professionals, as well as graduate students of biostatistics. Novel model-assisted designs showcase the new KISS principle: Keep it simple and smart!

Reliably optimizing a new treatment in humans is a critical first step in clinical evaluation since choosing a suboptimal dose or schedule may lead to failure in later trials. At the same time, if promising preclinical results do not translate into a real treatment advance, it is important to determine this quickly and terminate the clinical evaluation process to avoid wasting resources. Bayesian Designs for Phase I–II Clinical Trials describes how phase I–II designs can serve as a bridge or protective barrier between preclinical studies and large confirmatory clinical trials. It illustrates many of the severe drawbacks with conventional methods used for early-phase clinical trials and presents numerous Bayesian designs for human clinical trials of new experimental treatment regimes. Written by research leaders from the University of Texas MD Anderson Cancer Center, this book shows how Bayesian designs for early-phase clinical trials can explore, refine, and optimize new experimental treatments. It emphasizes the importance of basing decisions on both efficacy and toxicity.

This book provides a comprehensive introduction to statistical methods for designing early phase dose-finding clinical trials. It will serve as a textbook or handbook for graduate students and practitioners in biostatistics and clinical investigators who are involved in designing, conducting, monitoring, and analyzing dose-finding trials. The book will also provide an overview of advanced topics and discussions in this field for the benefit of researchers in biostatistics and statistical science. Beginning with backgrounds and fundamental notions on dose finding in early phase clinical trials, the book then provides traditional and recent dose-finding designs of phase I trials for, e.g., cytotoxic agents in oncology, to evaluate toxicity outcome. Included are rule-based and model-based designs, such as 3 + 3 designs, accelerated titration designs, toxicity probability interval designs, continual reassessment method and related designs, and escalation overdose control designs. This book also covers more complex and updated dose-finding designs of phase I-II and I/II trials for cytotoxic agents, and cytostatic agents, focusing on both toxicity and efficacy outcomes, such as designs with covariates and drug combinations, maximum tolerated dose-schedule finding designs, and so on.

The late-onset outcome issue is common in early phase dose- nding clinical trials. This problem becomes more intractable in phase I/II clinical trials because both toxicity and e cacy responses are subject to the late-onset outcome issue. The existing methods applying for the phase I trials cannot be used directly for the phase I/II trial due to a lack of capability to model the joint toxicity{e cacy distribution. We propose a conditional weighted likelihood (CWL) method to circumvent this issue. The key idea of the CWL method is to decompose the joint probability into the product of marginal and conditional probabilities and then weight each probability based on each patient's actual follow-up time. We further extend the proposed method to handle more complex situations where the late-onset outcomes are competing risks or semicompeting risks outcomes. We treat the late-onset competing risks/semi-competing risks outcomes as missing data and develop a series of Bayesian data-augmentation methods to e ciently impute the missing data and draw the posterior samples of the parameters of interest. We also propose adaptive dose- nding algorithms to allocate patients and identify the optimal biological dose during the trial. Simulation studies show that the proposed methods yield desirable operating characteristics and outperform the existing methods.

In this research, two new designs in clinical trials are proposed. The first problem is a new Bayesian adaptive dose-finding design and its application in an oncology clinical trial. This design is used for phase IB studies with the biomarker as the endpoint and with the fewer patients. The second problem is another new Bayesian adaptive dose-finding design with longitudinal analysis and its application in phase II depression clinical trial. This design is best fit for phase II dosing-finding clinical trials with clinical endpoints. MTD information has been obtained before the trials. In adaptive dose-finding clinical trials, the strategy is to reduce the allocation of patients to non-informative doses and also save the trial cost. Bayesian adaptive dose finding design has the ability to utilize accumulating data obtained in real time to alter the course of the trial, thereby enabling dynamic allocation to different dosing groups and dropping of ineffective dosing group earlier. In this research, Bayesian adaptive method is used as a new and useful approach that applies to phase IB and phase II dose-finding clinical trials to evaluate safety and efficacy of the study treatment. Response model and Normal Dynamic Linear Models (NDLMs) are applied in stages 1-4. Conditional probability for each parameter in the model is derived using appropriate prior distributions. Markov Chain Monte Carlo (MCMC) method is used to do the simulation. Model parameters with meaningful prior distributions and the posterior quantities are obtained to evaluate the trial results and they help to determine the optimal dose level which can be used in later studies. Simulations are done for different scenarios in the two designs and used to validate the model. Five-thousand simulation trials are conducted to verify the repeatability of the results. The posterior probability of success for the trial is greater than 90% based on the simulation results. The results give clearer idea if one needs to go further to test new dose levels based on the thorough evaluation of the existing partial data. Compared with the other adaptive dose finding strategy, the proposed Bayesian adaptive designs are sensitive and robust to help the investigators draw conclusion as early as possible. The designs can also reduce sample size substantially which in turn leads to savings in cost and time. Continuous-time Markov model has the advantage over the traditional survival model and can be used to describe disease as a series of probable transitions between health states. This is an attractive feature since it provides the ability to describe the course of disease over time. It can also describe and estimate expected survival in clinical cohort. In this research, continuous-time Markov model is used in the time-to-event analysis in a phase II oncology trial. Six states are defined in the Markov chain which is used in time to progression analysis for 36 patients with neuroendocrine carcinoma. The transition probability matrix P is defined and used to iterate future transition and survival probabilities. The estimate from matrix analysis shows that the results are reliable and comparable with the Kaplan-Meier results.

This dissertation, "Bayesian Adaptive Methods for Phase I Clinical Trials" by Ruitao, Lin, 林瑞涛, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: The primary objective of phase I dose-finding trials is to determine the maximum tolerated dose (MTD), which is typically defined as the dose with the dose-limiting toxicity probability closest to the target toxicity probability. The American Society of Clinical Oncology (ASCO) recently published an update of the ASCO policy statement to call for new phase I trial designs to allow for more efficient escalation to the therapeutic dose levels in order to cope with the changing landscape in cancer research. In this thesis, innovative and robust designs for single- or multiple-agent phase I dose-finding trials are studied. To enhance robustness and efficiency, two nonparametric methods are proposed to locate the MTD in single-agent phase I clinical trials without imposing any parametric assumption on the underlying distribution of the toxicity curve. First, a uniformly most powerful Bayesian interval (UMPBI) design is developed for dose finding, where the optimal interval is determined by the rejection region of the uniformly most powerful Bayesian test. UMPBI is easy to implement and can be nicely interpreted. Compared with existing interval designs, the proposed UMPBI design exhibits a unique feature of convergence to the MTD. Next, a nonparametric overdose control (NOC) method is proposed by casting dose finding in a Bayesian model selection framework. Each dose assignment under NOC is determined such that the posterior probability of overdosing is controlled. In addition, NOC is incorporated with a fractional imputation method to deal with late-onset toxicity outcomes. Both of the UMPBI and NOC designs are flexible, as well as possessing sound theoretical support and desirable numerical performance. In the era of precision medicine, combination therapy is playing an increasingly important role in drug development. However, drug combinations often lead to a high-dimensional dose searching space compared to conventional single-agent dose finding, especially when three or more drugs are combined for treatment. Most of the current dose-finding designs aim to quantify the toxicity probability space using certain prespecified yet complicated models. Not only do these models characterize each individual drug's toxicity profile, but they also need to quantify their interaction effects, which often leads to multi-parameter models. In order to stabilize the current practice of dose finding in drug-combination trials with limited sample sizes, a random walk Bayesian optimal interval (RW-BOIN) design and a Bootstrap aggregating continual reassessment method (Bagging CRM) are proposed. RW-BOIN is built on the basis of the single-agent BOIN design, and it can be utilized to tackle high-dimensional dose-finding problems. A convergence theorem is established to ensure the validity of RW-BOIN. On the other hand, Bagging CRM implements a dimension reduction technique and some ensemble methods in machine learning, so that the toxicity probability space can be stably reduced to a one-dimensional searching line. Simulation studies show that both RW-BOIN and Bagging CRM have comparative and robust operating characteristics compared with existing approaches under various dose-toxicity scenarios. All of the proposed methods are exemplified with real phase I dose-finding trials. Subjects: Bayesian statistical decision theory Clinical trials - Statistical methods

This book begins with an introduction of pragmatic cluster randomized trials (PCTs) and reviews various pragmatic issues that need to be addressed by statisticians at the design stage. It discusses the advantages and disadvantages of each type of PCT, and provides sample size formulas, sensitivity analyses, and examples for sample size calculation. The generalized estimating equation (GEE) method will be employed to derive sample size formulas for various types of outcomes from the exponential family, including continuous, binary, and count variables. Experimental designs that have been frequently employed in PCTs will be discussed, including cluster randomized designs, matched-pair cluster randomized design, stratified cluster randomized design, stepped-wedge cluster randomized design, longitudinal cluster randomized design, and crossover cluster randomized design. It demonstrates that the GEE approach is flexible to accommodate pragmatic issues such as hierarchical correlation structures, different missing data patterns, randomly varying cluster sizes, etc. It has been reported that the GEE approach leads to under-estimated variance with limited numbers of clusters. The remedy for this limitation is investigated for the design of PCTs. This book can assist practitioners in the design of PCTs by providing a description of the advantages and disadvantages of various PCTs and sample size formulas that address various pragmatic issues, facilitating the proper implementation of PCTs to improve health care. It can also serve as a textbook for biostatistics students at the graduate level to enhance their knowledge or skill in clinical trial design. Key Features: Discuss the advantages and disadvantages of each type of PCTs, and provide sample size formulas, sensitivity analyses, and examples. Address an unmet need for guidance books on sample size calculations for PCTs; A wide variety of experimental designs adopted by PCTs are covered; The sample size solutions can be readily implemented due to the accommodation of common pragmatic issues encountered in real-world practice; Useful to both academic and industrial biostatisticians involved in clinical trial design; Can be used as a textbook for graduate students majoring in statistics and biostatistics.