The primary objective of this proposed research is to provide practical models and methods for the design, conduct, and analysis of complex clinical trials not accommodated by conventional trial designs. These are trials of experimental treatment regimens for which little is known about the effects of dose, dose-schedule combination, or dose sequence in multiple cycles, or about short-term or long-term adverse events (toxicities) or efficacy. All models and methods are Bayesian, which facilitates adaptive decision making and accounts for multiple sources of variability. A major difficulty is the inheren tension between the physician's desire to treat the patients enrolled in a trial ethically and the scientific goal to obtain information useful for treatment regime evaluation and refinement to benefit future patients. Consequently, to ensure ethical trials, the designs include sequential, outcome-adaptive learn-as-you go decision rules. In the settings considered here, this is complicated by the complexity of patient outcomes, which include both efficacy and toxicity variables, each possibly binary, ordinal, or continuous with possible interval censoring, in some cases observed over multiple treatment cycles. For most of the designs and methods, the desirability of each combination of possible outcomes is quantified by elicited numerical utilities which are used as the basis for adaptive decision-making. This formalizes the idea, inherent in virtually all medical decision-making, of trade-offs between desirable and undesirable therapeutic outcomes, and it provides a method for reducing multiple outcomes to a single decision criterion, the posterior mean utility of each treatment regime, quantifying the relative importance of the outcomes. Methods will be developed for eliciting numerical utilities and establishing utility functions, in the cases of bivariate or trivariate discrete outcomes, and for bivariate event times.
Specific Aim 3 is more general in that it accommodates any optimality criterion, including posterior mean utility, efficacy-toxicity probability trade-off, or the continal reassessment method criterion. In some trials with adaptive treatment assignment, the exploration versus exploitation problem may occur, wherein a greedy algorithm that always chooses the empirically optimal regime may get stuck at a regime that actually is inferior. To deal with this, we will explore hybrid algorithms that include adaptive randomization to reduce the risk of choosing inferior regimes. Computer simulation will be used as a design tool to calibrate design parameters and prior parameters to obtain designs with good frequentist operating characteristics.
The Specific Aims are to develop (1) a phase I-II design to optimize (schedule, dose) regimes based on the joint utility of the times to response and toxicity, (2) a family of phase I-II designs, based on bivariate binary or bivariate ordinal (toxicity, efficacy), o optimize the dose sequence of an experimental agent given sequentially in multiple cycles, (3) an algorithm to improve the logistics of outcome-adaptive clinical trials by future data weighted randomization, and (4) phase I-II design, based on three outcomes, to optimize the sedation dose for preterm infants being intubated to treat respiratory distress syndrome. Efficient, user-friendly, freely available computer programs will be developed to facilitate widespread application.

Public Health Relevance

The proposed research will provide designs for complex clinical trials that are more efficient, more scientifically valid, and more ethical than conventioal designs by (1) using more patient outcome data actually available during the trial to make real-time adaptive decisions, (2) incorporating elicited prior information, (3) basing adaptive decisions on utilities of patient clinical outcomes, (4) when appropriate, evaluating schedule-dose combinations or multi-cycle regimes rather than single treatments, and (5) combining separate phases of conventional treatment evaluation. The proposed methods will increase the chance of favorable therapeutic outcomes for patients in the trial, and benefit future patients by accelerating and improving treatment regime development, increasing the chances of identifying regimes providing substantive therapeutic improvements, and dropping unpromising or unsafe regimes.

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Research Project (R01)
Project #
4R01CA083932-15
Application #
9020208
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Witherspoon, Kim
Project Start
2000-04-03
Project End
2018-04-30
Budget Start
2016-05-01
Budget End
2017-04-30
Support Year
15
Fiscal Year
2016
Total Cost
Indirect Cost
Name
University of Texas MD Anderson Cancer Center
Department
Biostatistics & Other Math Sci
Type
Hospitals
DUNS #
800772139
City
Houston
State
TX
Country
United States
Zip Code
77030
Murray, Thomas A; Yuan, Ying; Thall, Peter F et al. (2018) A utility-based design for randomized comparative trials with ordinal outcomes and prognostic subgroups. Biometrics 74:1095-1103
Chapple, Andrew G; Thall, Peter F (2018) A Hybrid Phase I-II/III Clinical Trial Design Allowing Dose Re-Optimization in Phase III. Biometrics :
Thall, Peter F; Mueller, Peter; Xu, Yanxun et al. (2017) Bayesian nonparametric statistics: A new toolkit for discovery in cancer research. Pharm Stat 16:414-423
Morita, Satoshi; Thall, Peter F; Takeda, Kentaro (2017) A simulation study of methods for selecting subgroup-specific doses in phase 1 trials. Pharm Stat 16:143-156
Thall, Peter F; Ursino, Moreno; Baudouin, Véronique et al. (2017) Bayesian treatment comparison using parametric mixture priors computed from elicited histograms. Stat Methods Med Res :962280217726803
Chapple, Andrew G; Vannucci, Marina; Thall, Peter F et al. (2017) Bayesian variable selection for a semi-competing risks model with three hazard functions. Comput Stat Data Anal 112:170-185
Thall, Peter F; Nguyen, Hoang Q; Zinner, Ralph G (2017) Parametric Dose Standardization for Optimizing Two-Agent Combinations in a Phase I-II Trial with Ordinal Outcomes. J R Stat Soc Ser C Appl Stat 66:201-224
Xu, Yanxun; Thall, Peter F; Müller, Peter et al. (2017) A Decision-Theoretic Comparison of Treatments to Resolve Air Leaks After Lung Surgery Based on Nonparametric Modeling. Bayesian Anal 12:639-652
Wathen, J Kyle; Thall, Peter F (2017) A simulation study of outcome adaptive randomization in multi-arm clinical trials. Clin Trials 14:432-440
Murray, Thomas A; Thall, Peter F; Yuan, Ying et al. (2017) Robust treatment comparison based on utilities of semi-competing risks in non-small-cell lung cancer. J Am Stat Assoc 112:11-23

Showing the most recent 10 out of 73 publications