Skip to main content

2014 | Book

Practical Considerations for Adaptive Trial Design and Implementation

Editors: Weili He, José Pinheiro, Olga M. Kuznetsova

Publisher: Springer New York

Book Series : Statistics for Biology and Health


About this book

This edited volume is a definitive text on adaptive clinical trial designs from creation and customization to utilization. As this book covers the full spectrum of topics involved in the adaptive designs arena, it will serve as a valuable reference for researchers working in industry, government and academia. The target audience is anyone involved in the planning and execution of clinical trials, in particular, statisticians, clinicians, pharmacometricians, clinical operation specialists, drug supply managers, and infrastructure providers. In spite of the increased efficiency of adaptive trials in saving costs and time, ultimately getting drugs to patients sooner, their adoption in clinical development is still relatively low. One of the chief reasons is the higher complexity of adaptive design trials as compared to traditional trials. Barriers to the use of clinical trials with adaptive features include the concerns about the integrity of study design and conduct, the risk of regulatory non-acceptance, the need for an advanced infrastructure for complex randomization and clinical supply scenarios, change management for process and behavior modifications, extensive resource requirements for the planning and design of adaptive trials and the potential to relegate key decision makings to outside entities. There have been limited publications that address these practical considerations and recommend best practices and solutions. This book fills this publication gap, providing guidance on practical considerations for adaptive trial design and implementation. The book comprises three parts: Part I focuses on practical considerations from a design perspective, whereas Part II delineates practical considerations related to the implementation of adaptive trials. Putting it all together, Part III presents four illustrative case studies ranging from description and discussion of specific adaptive trial design considerations to the logistic and regulatory issues faced in trial implementation.

Bringing together the expertise of leading key opinion leaders from pharmaceutical industry, academia, and regulatory agencies, this book provides a balanced and comprehensive coverage of practical considerations for adaptive trial design and implementation.

Table of Contents


Design Considerations

Chapter 1. The Need for and the Future of Adaptive Designs in Clinical Development
There has been much progress in the development and implementation of adaptive designs over the past 20 years. A major driver for this class of novel designs is to increase the information value of clinical trial data to enable better decisions, leading to more efficient drug development processes and improved late-stage success rates. In this chapter, we review common types of adaptive designs that have been developed and the frequently encountered challenges associated with their implementations. We discuss reasons why, in our opinion, the interest in adaptive designs will continue to rise. Furthermore, we describe what still needs to be done to move adaptive designs into our standard toolbox of design options. We emphasize the importance to implement adaptive designs with thorough upfront planning. The business case mandates that we treat the opportunities offered by adaptive designs carefully so that we can successfully foster a broad acceptance of properly designed and executed adaptive designs, when they represent the best design options based on their performance characteristics to address the need of a particular situation.
Christy Chuang-Stein, Frank Bretz
Chapter 2. Regulatory Guidance Documents on Adaptive Designs: An Industry Perspective
Adaptive designs have the potential to be a transformative methodology in clinical drug development, but acceptance by regulatory agencies is a prerequisite for their broader adoption and success, especially in the context of confirmatory studies. Both FDA and EMA have published guidance documents focusing on adaptive designs, which have been neither discouraging nor clearly supportive of the approach in their assessments and recommendations. As a result, the interpretation of the regulatory position on adaptive designs also has been mixed, with some citing the guidance documents as evidence that health authorities do not accept adaptive designs, while others mentioning the same documents as indication that regulators support their use in drug development, when properly planned, conducted, and analyzed. This chapter reviews and discusses the two main regulatory documents on adaptive designs issued by the time this book was published: the reflection paper by EMA (Reflection paper on methodological issues in confirmatory clinical trials with flexible design and analysis plan (draft CHMP/EWP/2459/02, 23-Mar-2006), 2007) and the draft guidance by FDA (Adaptive design clinical trials for drug and biologics draft guidance, 2010). Reactions from the biopharmaceutical industry to both documents, collated by industry trade groups, are also presented and discussed.
José Pinheiro
Chapter 3. A Commentary on the U.S. FDA Adaptive Design Draft Guidance and EMA Reflection Paper from a Regulatory Perspective and Regulatory Experiences
Fixed design confirmatory trials rely on emerging and reliable prior data and knowledge to provide necessary assumptions about the key design parameters including nuisance parameters. Traditionally, a fixed design has been the gold standard for its simplicity, validity, and ability to provide an unbiased estimate of the treatment effect. To allow for pre-specified flexibility in an ongoing trial, a simple two-arm controlled trial with a single primary efficacy endpoint, the repeated significance testing involving multiplicity adjustment becomes more complex than a fixed design approach. The repeated significance testing recognized in group sequential design and analysis was proposed as early as the randomization ratio adaptation in late 1960s (Zelen 1969), e.g., Armitage et al. (1969).
Sue-Jane Wang
Chapter 4. Considerations and Optimization of Adaptive Trial Design in Clinical Development Programs
Although the efficiency of adaptive design on the trial level is well recognized, its impact is even greater when applied at the program or portfolio level. Besides its simplest form of sample size reestimation or early stopping in a given trial, the adaptive design achieves efficiency by combining in a single trial objectives that are usually addressed in two separate conventional studies. Another feature of adaptive design is population enrichment where drug response can be optimized to specific patient subpopulations that respond better to treatment. More complex adaptive strategies integrate the development of several compounds and/or indications into one process. We provide an overview of these types of adaptive designs and illustrate their value added in a case study of an adaptive “compound” finder that investigates several compounds in Alzheimer’s disease area simultaneously approaching the proof-of-concept stage.
Michael Krams, Vladimir Dragalin
Chapter 5. Optimal Cost-Effective Go–No Go Decisions in Clinical Development
In late-stage drug development, drug developers have to make two critical Go–No Go decisions. The first one is whether to proceed to the definitive Phase III investigation after a Phase II proof-of-concept (POC) trial. The second one is whether to stop a Phase III confirmatory trial for futility after an interim analysis of the data. In practice, the two decisions are heuristically made with limited statistical input, usually amounting to statistical characterization of proposed options. We propose to find the optimal decisions by explicitly maximizing a benefit–cost ratio function, which is often the implicit objective in an otherwise qualitative decision-making process. The numerator of the function represents the benefit (proportional to the expected number of truly active drugs identified for Phase III development in the POC setting; proportional to the expected power for successful completion of Phase III in the interim analysis setting), and the denominator represents the expected total late-stage development cost. The method is easy to explain and simple to implement. The optimal design parameters provide a rational starting point for decision makers to consider. As an illustration, the method developed herein is applied to examples from the oncology therapeutic area including an adaptive seamless Phase II/III design. The same idea is applicable to any disease area where cost-effectiveness of a Go–No Go decision is a major concern.
Cong Chen, Robert A. Beckman, Linda Z. Sun
Chapter 6. Timing and Frequency of Interim Analyses in Confirmatory Trials
In many pivotal clinical trials, timing and frequency of interim analyses are important for ethical treatment of patients and for practical and regulatory purposes. It is often desirable to evaluate a large trial of a new treatment that has some safety risk in order to stop or modify the trial based on the emerging risk–benefit profile compared to control treatment. Statistical considerations would suggest not stopping too soon in order to avoid large Type I or Type II error or basing a decision on inadequate data. Regulators often prefer to minimize interim analyses of efficacy due to presumed bias created by early stopping and an inability to adequately evaluate important secondary efficacy endpoints, safety, or the general risk–benefit profile for the new treatment. For practical purposes, analyses must be done soon enough to have a meaningful impact on the trial. For the same reason, limiting enrollment rates and ensuring prompt collection and analysis of data are important. We discuss tradeoffs between these factors in deciding when to perform interim analyses. In addition to formal evaluations for early positive efficacy findings, there are different considerations for trials early in the development process, for safety monitoring during a trial, and for futility analyses. We consider logistical and regulatory issues throughout.
Keaven M. Anderson
Chapter 7. Approaches for Optimal Dose Selection for Adaptive Design Trials
Adaptive designs use accumulating data to modify in a prospectively planned manner certain design aspects of a clinical study without undermining its validity and integrity. The aim of this chapter is to review adaptive design approaches for dose finding and optimal dose selection and to demonstrate that adaptivity is a fundamentally important concept, which can be applied to dose selection in different stages of clinical development. We review the major statistical methods available for planning and analyzing adaptive designs in Phase I, II, and III. To illustrate the ideas, we refer to examples and case studies from the literature, where available.
David Lawrence, Frank Bretz
Chapter 8. A Review of Available Software and Capabilities for Adaptive Designs
This chapter provides a brief review of methodologies and software solutions for several types of adaptive designs: the traditional and adaptive group sequential designs including sample size reestimation, multistage adaptive designs with arm and subpopulation selection at interim analyses, and adaptive designs for dose-finding studies.
Yevgen Tymofyeyev
Chapter 9. Randomization Challenges in Adaptive Design Studies
Adaptive design studies often face randomization challenges. Adaptive dose-ranging studies require randomization techniques that, in a small cohort, approximate reasonably well an inconveniently skewed allocation ratio to several treatment arms. When a small interim analysis sample needs to be balanced in several important predictors, dynamic allocation might be required to achieve this goal. Accelerated drug development often necessitates a large number of centers to speed up the study enrollment. When the drug is limited or costly, as is often the case with adaptive design studies conducted early in drug development, advanced randomization techniques are needed to efficiently manage the drug supplies in multicenter trials. In open-label adaptive design trials randomization procedures less predictable than permuted block randomization help reduce potential for selection bias. Randomization techniques developed for equal allocation to several treatment arms help dealing with the randomization challenges in equal allocation adaptive design studies. When these techniques are expanded to unequal allocation common to adaptive designs, care should be taken to preserve the allocation ratio at every allocation step. In this chapter we review randomization techniques useful in adaptive design studies, including those developed in recent years to specifically address the needs above.
Olga M. Kuznetsova
Chapter 10. Response-Adaptive Randomization for Clinical Trials
Response-adaptive randomization in clinical trials uses accumulated patient response data to adjust the allocation probability for the next patient, so that a particular objective, for example, more patients assigned to the better performing treatment arm, can be achieved. This ethically appealing randomization procedure has gained significant attention in academia, regulatory agencies, and industry in light of widespread of adaptive clinical trial designs with the FDA’s Critical Path Initiative (FDA: Innovation or stagnation: challenge and opportunity on the critical path to new medical products, 2004). However, this procedure has also generated unmatched controversy since its first application in the ECMO trial (Bartlett et al., Pediatrics 76:479–487, 1985). In this chapter, we will describe response-adaptive randomization procedures from both frequentist and Bayesian perspectives and provide a comprehensive assessment on situations where such procedures should be applied.
Lanju Zhang, William F. Rosenberger

Trial Implementation Considerations

Chapter 11. Implementing Adaptive Designs: Operational Considerations, Putting It All Together
The use of adaptive clinical trial designs for a drug development program has clear advantages over traditional methods, given the ability to identify optimal clinical benefits and make informed decisions regarding safety and efficacy earlier in the clinical trial process. However, operational execution can be challenging due to the added complexities of implementing adaptive designs. These complexities deserve additional attention. Key operational challenges occur in several areas: availability of statistical simulation tools for clinical trial modeling at the planning stage; the use of trial simulation modeling approaches to ensure that the trial is meeting expected outcomes; and challenges regarding rapid data collection, clinical monitoring, resourcing, minimization of data leakage, IVRS, drug supply management, and systems integration. The purpose of this chapter is to highlight several operational challenges that must be taken into consideration in conducting an adaptive clinical trial. Adaptive design implementation strategies are also discussed in this chapter.
Olga Marchenko, Christy Nolan
Chapter 12. Implementation Issues in Adaptive Design Trials
In this chapter we discuss operational challenges that are specific to adaptive trials (as well as complex nonadaptive trials): essentially, the need to validate the design, to control the trial centrally, to collect and analyze key data rapidly, to preserve the trial blinding and integrity, and to document all important adaptive decisions taken. We illustrate these challenges using an actual phase I trial in oncology, and argue that the issues can be addressed through proper planning, choice of experienced vendors and independent groups (coordinating center and DSMB), statistical teams with adequate expertise in the design chosen (randomization and CRM), recourse to efficient computer technology (IWRS, EDC, automated e-mailing), and oversight by a team that must be as flexible as the trial design!
Linda Danielson, Jerome Carlier, Tomasz Burzykowski, Marc Buyse
Chapter 13. Implementing Adaptive Designs: Using Technology to Protect Trial Integrity, Reduce Operational Bias, and Build Regulatory Trust
Experimental design and the execution environment have to go hand in hand to enable successful implementation of adaptive designs: the more complex, dynamic, and closer to real time the adaptation(s), the greater the demand on the execution environment.
As industry has dipped its toes into the adaptive pond, it has done so cautiously, partly handicapped by a traditional execution environment not designed to entertain real time learning and decision making. Therefore, the majority of adaptive designs currently implemented are of the simpler kind: one or two interim analyses, and limited adaptations: sample size adjustments, possibly dropping a dose, in other words: whatever a traditional and antiquated execution environment can support without requiring more than minor work-around solutions.
The opportunity space for adaptive designs from an experimental design perspective in pharmaceutical drug development is of course much wider (as discussed in other chapters). Here we present a conceptual view of a scalable execution environment, designed to support the full opportunity space of adaptive designs, and highlight the role of enabling technology and integrated processes.
This chapter looks at the role of technology today, but perhaps more importantly, identifies the role and need for technology in providing scalable and more efficient solutions that not only enable larger uptake of the simpler adaptive trials of today, but also support the more operationally demanding of adaptive trials, as well as enabling custom designs to be readily implemented. Systems and processes under stress introduce an increased risk of mistakes that may threaten trial integrity and introduce operational bias. In this future landscape, technology will play a more significant role not only to increase efficiency, but also to ensure trial integrity is maintained and operational bias is minimized. However technology is only a tool that is managed by humans, and as such, it is critical to remember that technology must go hand in hand with appropriate processes to control human behavior.
Judith Quinlan, Michael Krams
Chapter 14. Considerations for Interim Analyses in Adaptive Trials, and Perspectives on the Use of DMCs
A particularly critical issue for adaptive clinical trials, with potentially great impact on how large a role these trials will come to play in confirmatory stages of clinical development, involves the processes by which accruing data are collected and analyzed, and by which adaptation decisions are made and implemented. The importance of this issue arises from the sensitivity of unblinded interim results and the potential, reflected in current conventions in nonadaptive trials, for access to interim results to introduce biases into the trial conduct and its results. This issue is intertwined with the role of independent Data Monitoring Committees, commonly the only party granted access to interim comparative results in current practice. We discuss the issues of who should be involved in data review for adaptation decisions, how the data flow and access to results is controlled, and the specific role that Data Monitoring Committees might play in this process.
Paul Gallo, David DeMets, Lisa LaVange
Chapter 15. Approaches for Clinical Supply Modelling and Simulation
Clinical supply is impacted by decisions and events at every stage of a clinical trial. Protocol design, logistics planning, and operational dynamics pose challenges to the management of clinical supply in terms of complexity and uncertainty. In this chapter, we propose a simulation modelling approach to address these issues and support decision-makers in effectively managing clinical supply. The approach is comprehensively described in terms of underlying structure and process, and is illustrated with adaptive trials involving dropping of arms and a Bayesian responsive-adaptive design for dose finding.
Nitin R. Patel, Suresh Ankolekar, Pralay Senchaudhuri
Chapter 16. Approaches for Patient Recruitment Modeling and Simulation
Accurate enrollment information is critical for timely decision-making and execution for clinical trials. Enrollment must be carefully planned and monitored in order to maximize business benefit and to achieve study objectives. This is particularly true in adaptive designs (AD) trials, where too slow or too fast patient enrollment along with inaccurate enrollment prediction will imperil the timing of and/or invalidate the planned adaptations in AD trials. This chapter will discuss the key considerations for patient enrollment management and present and discuss different patient recruitment models.
Weili He, Xiting Cao

Case Studies

Chapter 17. A Case Study for Adaptive Trial Design Consideration and Implementation
The use of adaptive designs in dose ranging studies can increase the efficiency of drug development by improving our ability to efficiently learn about the dose–response and better determine whether to take a drug forward into confirmatory phase testing and at what dose. This approach can maximize the ability to test a larger number of doses in a single trial while simultaneously increasing the efficiency of the trial in terms of making better go–no-go decisions about continuing the trial and/or the development of the drug for a specific indication.
We show in a real case study of a dose ranging trial in patients with acute exacerbations of schizophrenia how such an adaptive design explicitly addresses multiple trial goals, adaptively allocates subjects according to ongoing information needs, and allows termination for both early success and futility.
Vladimir Dragalin, Michael Krams
Chapter 18. Case Study: Design Considerations for a Phase Ib Randomized, Placebo-Controlled, 4-Period Crossover Adaptive Dose-Finding Clinical Trial
A new formulation of a test drug was to be studied in patients for the first time. Preclinical studies yielded a wide range of effective doses across species, so there was a desire to include at least 6 doses in the dose-finding trial. Adaptive design was chosen to focus the dose assignments on those that yielded at least 75 % of maximal response (ED75). A 4-period crossover in which each of 68 patients received 3 doses of test drug and placebo was chosen since it optimized performance characteristics. Frequent interim analyses were performed to optimally choose the 3 doses of test drug to maximize dose assignments to the doses estimated to be closest to ED75 based on analysis of study data accumulated up to each interim analysis time. The completed study yielded minimal assignment of doses away from ED75 and successfully identified, and focused dose assignment around, the target dose. The table below shows the numbers of assignments to each dose and the respective observed mean responses.
James A. Bolognese, Yevgen Tymofyeyev
Chapter 19. Continual Reassessment Method for a First-in-Human Trial: From Design to Trial Implementation
We present a case study of a Phase 1 oncology dose-escalation trial utilizing modified Continual Reassessment Method (CRM). Learning about the dose–toxicity relationship and choosing the correct Maximum Tolerated Dose (MTD) to take forward into Phase II is one of the most challenging research questions in Phase 1 oncology trials. CRM is a Bayesian adaptive design targeting a specific Dose Limiting Toxicity (DLT) rate, e.g., 25 %. Similar to the traditional 3 + 3 designs used in oncology Phase 1 trials, learning about drug’s toxicity profile with CRM occurs in real time. However, since CRM algorithm incorporates dose–toxicity modeling in the learning process, its ability to identify the correct Maximum Tolerated Dose is substantially improved, compared to the traditional 3 + 3 design. Such design also results in more patients being allocated to tolerable doses with therapeutic potential than would be the case in a more traditional 3 + 3 dose-escalation trial. This trial was designed and executed using a custom-developed and validated software package which helped to alleviate substantial increase in overhead cost typically associated with planning and implementation of such designs. We present the whole “story” of the trial from beginning to end, including selection of study design, assessment of its operating characteristics via simulations, execution, study results, and lessons learned.
Inna Perevozskaya, Lixin Han, Kristen Pierce
Chapter 20. Practical Considerations for a Two-Stage Confirmatory Adaptive Clinical Trial Design and Its Implementation: ADVENT Trial
In this chapter, we provide the details of an innovative two-stage, seamless adaptive clinical trial called ADVENT. This trial was conducted as a “final phase 3” clinical trial to establish the safety and efficacy of a first-in-class antidiarrheal agent, crofelemer, for the symptomatic relief of diarrhea in HIV patients receiving anti-retroviral therapy. Given that this was a trial with two-stage design that included a dose selection, it was necessary to demonstrate the strong control of Type 1 error. This was accomplished with a close testing procedure applied to combination tests that utilized the inverse normal combination function. We developed a one-sided significance testing procedure that ensures strong control of the Type 1 error at level 0.025. Using appropriate statistical methodology for combining the results from the two stages, a statistically significant outcome was obtained for the primary efficacy endpoint and crofelemer received marketing approval based on the ADVENT trial. While the authors acknowledge the importance of statistical methodology required to analyze the data from the ADVENT trial, this chapter also provides significant details on the clinical and regulatory challenges that were demanded for the conduct of this innovative, two-stage, adaptive clinical trial.
Pravin R. Chaturvedi, Zoran Antonijevic, Cyrus Mehta
Practical Considerations for Adaptive Trial Design and Implementation
Weili He
José Pinheiro
Olga M. Kuznetsova
Copyright Year
Springer New York
Electronic ISBN
Print ISBN

Premium Partner