Covariate-adjusted response-adaptive randomization for multi-arm clinical trials using a modified forward looking Gittins index rule

Leveraging machine learning: Covariate-adjusted Bayesian adaptive randomization and subgroup discovery in multi-arm survival trials

Clinical trials evaluate the safety and efficacy of treatments for specific diseases. Ensuring these studies are well-powered is crucial for identifying superior treatments. With the rise of personalized medicine, treatment efficacy may vary based on biomarker profiles. However, researchers often lack prior knowledge about which biomarkers are linked to varied treatment effects. Fixed or response-adaptive designs may not sufficiently account for heterogeneous patient characteristics, such as genetic diversity, potentially reducing the chance of selecting the optimal treatment for individuals. Recent advances in Bayesian nonparametric modeling pave the way for innovative trial designs that not only maintain robust power but also offer the flexibility to identify subgroups deriving greater benefits from specific treatments. Building on this inspiration, we introduce a Bayesian adaptive design for multi-arm trials focusing on time-to-event endpoints. We introduce a covariate-adjusted response adaptive randomization, updating treatment allocation probabilities grounded on causal effect estimates using a random intercept accelerated failure time BART model. After the trial concludes, we suggest employing a multi-response decision tree to pinpoint subgroups with varying treatment impacts. The performance of our design is then assessed via comprehensive simulations.

Covariate adjustment in Bayesian adaptive randomized controlled trials

In conventional randomized controlled trials, adjustment for baseline values of covariates known to be at least moderately associated with the outcome increases the power of the trial. Recent work has shown a particular benefit for more flexible frequentist designs, such as information adaptive and adaptive multi-arm designs. However, covariate adjustment has not been characterized within the more flexible Bayesian adaptive designs, despite their growing popularity. We focus on a subclass of these which allow for early stopping at an interim analysis given evidence of treatment superiority. We consider both collapsible and non-collapsible estimands and show how to obtain posterior samples of marginal estimands from adjusted analyses. We describe several estimands for three common outcome types. We perform a simulation study to assess the impact of covariate adjustment using a variety of adjustment models in several different scenarios. This is followed by a real-world application of the compared approaches to a COVID-19 trial with a binary endpoint. For all scenarios, it is shown that covariate adjustment increases power and the probability of stopping the trials early, and decreases the expected sample sizes as compared to unadjusted analyses.

Using biomarkers to allocate patients in a response-adaptive clinical trial

In this paper, we discuss a response adaptive randomization method, and why it should be used in clinical trials for rare diseases compared to a randomized controlled trial with equal fixed randomization. The developed method uses a patient's biomarkers to alter the allocation probability to each treatment, in order to emphasize the benefit to the trial population. The method starts with an initial burn-in period of a small number of patients, who with equal probability, are allocated to each treatment. We then use a regression method to predict the best outcome of the next patient, using their biomarkers and the information from the previous patients. This estimated best treatment is assigned to the next patient with high probability. A completed clinical trial for the effect of catumaxomab on the survival of cancer patients is used as an example to demonstrate the use of the method and the differences to a controlled trial with equal allocation. Different regression procedures are investigated and compared to a randomized controlled trial, using efficacy and ethical measures.

Covariate-adjusted response-adaptive designs based on semiparametric approaches

We consider theoretical and practical issues for innovatively using a large number of covariates in clinical trials to achieve various design objectives without model misspecification. Specifically, we propose a new family of semiparametric covariate-adjusted response-adaptive randomization (CARA) designs and we use the target maximum likelihood estimation (TMLE) to analyze the correlated data from CARA designs. Our approach can flexibly achieve multiple objectives and correctly incorporate the effect of a large number of covariates on the responses without model misspecification. We also obtain the consistency and asymptotic normality of the target parameters, allocation probabilities, and allocation proportions. Numerical studies demonstrate that our approach has advantages over existing approaches, even when the data-generating distribution is complicated.

A novel statistical test for treatment differences in clinical trials using a response-adaptive forward-looking Gittins Index Rule

The most common objective for response-adaptive clinical trials is to seek to ensure that patients within a trial have a high chance of receiving the best treatment available by altering the chance of allocation on the basis of accumulating data. Approaches that yield good patient benefit properties suffer from low power from a frequentist perspective when testing for a treatment difference at the end of the study due to the high imbalance in treatment allocations. In this work we develop an alternative pairwise test for treatment difference on the basis of allocation probabilities of the covariate-adjusted response-adaptive randomization with forward-looking Gittins Index (CARA-FLGI) Rule for binary responses. The performance of the novel test is evaluated in simulations for two-armed studies and then its applications to multiarmed studies are illustrated. The proposed test has markedly improved power over the traditional Fisher exact test when this class of nonmyopic response adaptation is used. We also find that the test's power is close to the power of a Fisher exact test under equal randomization.

Nonmyopic and pseudo-nonmyopic approaches to optimal sequential design in the presence of covariates

In sequential experiments, subjects become available for the study over a period of time, and covariates are often measured at the time of arrival. We consider the setting where the sample size is fixed but covariate values are unknown until subjects enrol. Given a model for the outcome, a sequential optimal design approach can be used to allocate treatments to minimize the variance of the estimator of the treatment effect. We extend existing optimal design methodology so it can be used within a nonmyopic framework, where treatment allocation for the current subject depends not only on the treatments and covariates of the subjects already enrolled in the study, but also the impact of possible future treatment assignments within a specified horizon. The nonmyopic approach requires recursive formulae and suffers from the curse of dimensionality. We propose a pseudo-nonmyopic approach which has a similar aim to the nonmyopic approach, but does not involve recursion and instead relies on simulating trajectories of future possible decisions. Our simulation studies show that, for the simple case of a logistic regression with a single binary covariate and a binary treatment, and a more realistic case with four binary covariates, binary treatment and treatment-covariate interactions, the nonmyopic and pseudo-nonmyopic approaches provide no competitive advantage over the myopic approach, both in terms of the size of the estimated treatment effect and also the efficiency of the designs. Results are robust to the size of the horizon used in the nonmyopic approach, and the number of simulated trajectories used in the pseudo-nonmyopic approach.

Challenges and opportunities in biomarker-driven trials: adaptive randomization

In an era of precision medicine, as advanced technology such as molecular profiling at individual patient level has been developed and become increasingly accessible and affordable, biomarker-driven trials have been received a lot of attention and are expected to receive more attention in order to integrate clinical practice with clinical research. Biomarkers play a critical role to identify patients who are expected to get benefit from a treatment, and it is important to effectively incorporate the biomarkers into clinical trials to understand the biomarker-treatment relationship and increase the efficiency. We investigate incorporating biomarkers in adaptive randomization to identify patients who would respond better to the treatment and optimize the treatment allocation. The covariate-adjusted variants of the existing response-adaptive randomization are used to implement biomarker-driven randomization, and the performance of the biomarker-driven randomization is compared with the existing randomization methods, such as traditional fixed randomization with equal probability and response-adaptive randomization without incorporating biomarkers, under the group sequential design allowing early stopping due to superiority and futility. Various scenarios are taken into account to see the impact of the biomarker-driven randomization in the simulation study. It shows that the overall type I error rate is likely to be inflated by the effect of prognostic biomarkers. Several suggestions and considerations for the challenges are discussed to maintain the type I error rate at the nominal level.

A Contextual-Bandit-Based Approach for Informed Decision-Making in Clinical Trials

Clinical trials are conducted to evaluate the efficacy of new treatments. Clinical trials involving multiple treatments utilize the randomization of treatment assignments to enable the evaluation of treatment efficacies in an unbiased manner. Such evaluation is performed in post hoc studies that usually use supervised-learning methods that rely on large amounts of data collected in a randomized fashion. That approach often proves to be suboptimal in that some participants may suffer and even die as a result of having not received the most appropriate treatments during the trial. Reinforcement-learning methods improve the situation by making it possible to learn the treatment efficacies dynamically during the course of the trial, and to adapt treatment assignments accordingly. Recent efforts using multi-arm bandits, a type of reinforcement-learning method, have focused on maximizing clinical outcomes for a population that was assumed to be homogeneous. However, those approaches have failed to account for the variability among participants that is becoming increasingly evident as a result of recent clinical-trial-based studies. We present a contextual-bandit-based online treatment optimization algorithm that, in choosing treatments for new participants in the study, takes into account not only the maximization of the clinical outcomes as well as the patient characteristics. We evaluated our algorithm using a real clinical trial dataset from the International Stroke Trial. We simulated the online setting by sequentially going through the data of each participant admitted to the trial. Two bandits (one for each context) were created, with four choices of treatments. For a new participant in the trial, depending on the context, one of the bandits was selected. Then, we took three different approaches to choose a treatment: (a) a random choice (i.e., the strategy currently used in clinical trial settings), (b) a Thompson sampling-based approach, and (c) a UCB-based approach. Success probabilities of each context were calculated separately by considering the participants with the same context. Those estimated outcomes were used to update the prior distributions within the bandit corresponding to the context of each participant. We repeated that process through the end of the trial and recorded the outcomes and the chosen treatments for each approach. We also evaluated a context-free multi-arm-bandit-based approach, using the same dataset, to showcase the benefits of our approach. In the context-free case, we calculated the success probabilities for the Bernoulli sampler using the whole clinical trial dataset in a context-independent manner. The results of our retrospective analysis indicate that the proposed approach performs significantly better than either a random assignment of treatments (the current gold standard) or a multi-arm-bandit-based approach, providing substantial gains in the percentage of participants who are assigned the most suitable treatments. The contextual-bandit and multi-arm bandit approaches provide 72.63% and 64.34% gains, respectively, compared to a random assignment.

Multi-arm covariate-adaptive randomization

Simultaneously investigating multiple treatments in a single study achieves considerable efficiency in contrast to the traditional two-arm trials. Balancing treatment allocation for influential covariates has become increasingly important in today's clinical trials. The multi-arm covariate-adaptive randomized clinical trial is one of the most powerful tools to incorporate covariate information and multiple treatments in a single study. Pocock and Simon's procedure has been extended to the multi-arm case. However, the theoretical properties of multi-arm covariate-adaptive randomization have remained largely elusive for decades. In this paper, we propose a general framework for multi-arm covariate-adaptive designs which also includes the two-arm case, and establish the corresponding theory under widely satisfied conditions. The theoretical results provide new insights into the balance properties of covariate-adaptive randomization procedures and make foundations for most existing statistical inferences under two-arm covariate-adaptive randomization. Furthermore, these open a door to study the theoretical properties of statistical inferences for clinical trials based on multi-arm covariate-adaptive randomization procedures.

Personalized Risk-Based Screening Design for Comparative Two-Arm Group Sequential Clinical Trials

Personalized medicine has been emerging to take into account individual variability in genes and environment. In the era of personalized medicine, it is critical to incorporate the patients’ characteristics and improve the clinical benefit for patients. The patients’ characteristics are incorporated in adaptive randomization to identify patients who are expected to get more benefit from the treatment and optimize the treatment allocation. However, it is challenging to control potential selection bias from using observed efficacy data and the effect of prognostic covariates in adaptive randomization. This paper proposes a personalized risk-based screening design using Bayesian covariate-adjusted response-adaptive randomization that compares the experimental screening method to a standard screening method based on indicators of having a disease. Personalized risk-based allocation probability is built for adaptive randomization, and Bayesian adaptive decision rules are calibrated to preserve error rates. A simulation study shows that the proposed design controls error rates and yields a much smaller number of failures and a larger number of patients allocated to a better intervention compared to existing randomized controlled trial designs. Therefore, the proposed design performs well for randomized controlled clinical trials under personalized medicine.

Designing and conducting adaptive trials to evaluate interventions in health services and implementation research: practical considerations

Randomized controlled clinical trials are widely considered the gold standard for evaluating the efficacy or effectiveness of interventions in health care. Adaptive trials incorporate changes as the study proceeds, such as modifying allocation probabilities or eliminating treatment arms that are likely to be ineffective. These designs have been widely used in drug discovery studies but can also be useful in health services and implementation research and have been minimally used. As motivating examples, we use an ongoing adaptive trial and two completed parallel group studies and highlight potential advantages, disadvantages, and important considerations when using adaptive trial designs in health services and implementation research. In addition, we investigate the impact on power and the study duration if the two completed parallel-group trials had instead been conducted using adaptive principles. Compared with traditional trial designs, adaptive designs can often allow one to evaluate more interventions, adjust participant allocation probabilities (e.g., to achieve covariate balance), and identify participants who are likely to agree to enroll. These features could reduce resources needed to conduct a trial. However, adaptive trials have potential disadvantages and practical aspects that need to be considered, most notably outcomes that can be rapidly measured and extracted (e.g., long-term outcomes that take significant time to measure from data sources can be challenging), minimal missing data, and time trends. In conclusion, adaptive designs are a promising approach to help identify how best to implement evidence-based interventions into real-world practice in health services and implementation research.

Response adaptive designs for Phase II trials with binary endpoint based on context-dependent information measures

In many rare disease Phase II clinical trials, two objectives are of interest to an investigator: maximising the statistical power and maximising the number of patients responding to the treatment. These two objectives are competing, therefore, clinical trial designs offering a balance between them are needed. Recently, it was argued that response-adaptive designs such as families of multi-arm bandit (MAB) methods could provide the means for achieving this balance. Furthermore, response-adaptive designs based on a concept of context-dependent (weighted) information criteria were recently proposed with a focus on Shannon's differential entropy. The information-theoretic designs based on the weighted Renyi, Tsallis and Fisher informations are also proposed. Due to built-in parameters of these novel designs, the balance between the statistical power and the number of patients that respond to the treatment can be tuned explicitly. The asymptotic properties of these measures are studied in order to construct intuitive criteria for arm selection. A comprehensive simulation study shows that using the exact criteria over asymptotic ones or using information measures with more parameters, namely Renyi and Tsallis entropies, brings no sufficient gain in terms of the power or proportion of patients allocated to superior treatments. The proposed designs based on information-theoretical criteria are compared to several alternative approaches. For example, via tuning of the built-in parameter, one can find designs with power comparable to the fixed equal randomisation's but a greater number of patients responded in the trials.

A response-adaptive randomization procedure for multi-armed clinical trials with normally distributed outcomes

We propose a novel response‐adaptive randomization procedure for multi‐armed trials with continuous outcomes that are assumed to be normally distributed. Our proposed rule is non‐myopic, and oriented toward a patient benefit objective, yet maintains computational feasibility. We derive our response‐adaptive algorithm based on the Gittins index for the multi‐armed bandit problem, as a modification of the method first introduced in Villar et al. (Biometrics, 71, pp. 969‐978). The resulting procedure can be implemented under the assumption of both known or unknown variance. We illustrate the proposed procedure by simulations in the context of phase II cancer trials. Our results show that, in a multi‐armed setting, there are efficiency and patient benefit gains of using a response‐adaptive allocation procedure with a continuous endpoint instead of a binary one. These gains persist even if an anticipated low rate of missing data due to deaths, dropouts, or complete responses is imputed online through a procedure first introduced in this paper. Additionally, we discuss how there are response‐adaptive designs that outperform the traditional equal randomized design both in terms of efficiency and patient benefit measures in the multi‐armed trial context.

Rethinking the Gold Standard With Multi-armed Bandits: Machine Learning Allocation Algorithms for Experiments

In experiments, researchers commonly allocate subjects randomly and equally to the different treatment conditions before the experiment starts. While this approach is intuitive, it means that new information gathered during the experiment is not utilized until after the experiment has ended. Based on methodological approaches from other scientific disciplines such as computer science and medicine, we suggest machine learning algorithms for subject allocation in experiments. Specifically, we discuss a Bayesian multi-armed bandit algorithm for randomized controlled trials and use Monte Carlo simulations to compare its efficiency with randomized controlled trials that have a fixed and balanced subject allocation. Our findings indicate that a randomized allocation based on Bayesian multi-armed bandits is more efficient and ethical in most settings. We develop recommendations for researchers and discuss the limitations of our approach.

The Performance of Fixed-Horizon, Look-Ahead Procedures Compared to Backward Induction in Bayesian Adaptive-Randomization Decision-Theoretic Clinical Trial Design

Designing optimal, Bayesian decision-theoretic trials has traditionally required the use of computationally-intensive backward induction. While methods for addressing this barrier have been put forward, few are both computationally tractable and non-myopic, with applications of the Gittins index being one notable example. Here we explore the look-ahead approach with adaptive-randomization, with designs ranging from the fully myopic to the fully informed. We compare the operating characteristics of the look-ahead designed trials, in which decision rules are based on a fixed number of future blocks, with those of trials designed using traditional backward induction. The less-myopic designs performed well. As the designs become more myopic or the trials longer, there were disparities in regions of the decision space that are transition zones between continuation and stopping decisions. The more myopic trials generally suffered from early stopping as compared to the less myopic and backward induction trials. Myopic trials with adaptive randomization also saw as many as 28 % of their continuation decisions change to a different randomization ratio as compared to the backward induction designs. Finally, early stages of myopic-designed trials may have disproportionate effect on trial characteristics.