Practical Considerations For Adaptive Designs In Clinical Trials
By Julia Ogier, consultant, and Matthew Cardinal, associate principal consultant, Halloran Consulting
The life sciences industry and academic world seem to produce incredible scientific breakthroughs on a daily, if not hourly, basis these days. The pace of scientific breakthrough is mesmerizing, as a dazzling variety of technologies and studies have helped humans understand the underlying causes of disease. Whether those causes are genetic, environmental, or behavioral, it seems that we have an arsenal of tools to understand much more than ever how we can meet unmet human health needs. From the ambitious (at the time) Human Genome Project to the current application of machine learning and artificial intelligence to vast integrated data sets, it would seem that humans could be on the cusp of fundamentally altering the quality and longevity of human life.
However, a curious thing has happened. Despite all the data at our disposal, there has been very little change in the pace of new therapies approved by the FDA. Multiple credible sources have noted little change in the average annual number of new molecular entities approved by the FDA from 2006 to2014.1 Many of us who actively develop drugs are surprised by this data, as we have seen an incredible uptick in patient engagement over the past 10 years. Patients are educated about their options and often bring publications (legitimate and not so legitimate) to their doctors. There also has been a remarkable uptick in patient advocacy groups getting involved, directly investing in the development of drugs for specific groups of patients. Industry has seen a committed and willing partner in large patient foundations like the Michael J Fox Foundation and the Leukemia and Lymphoma Society to much smaller, tenacious organizations like the National Hemophilia Foundation and Parent Project Muscular Dystrophy (PPMD).
As sponsor companies seek to evaluate compounds that might meet unmet medical needs, they have found limited success in reducing clinical development expenditures and increasing the likelihood of success. The traditional model of clinical development is very prescriptive, pun intended, with each trial designed to answer a very specific question. The word prospective is synonymous with “better.” However, the most comprehensive and most relevant clinical data is often the data coming in real time out of ongoing clinical evaluations. An adaptive clinical trial design allows a company to examine a clinical trial while it is ongoing and adjust accordingly instead of following the traditional stepwise process of: 1) design the trial, 2) conduct the trial, and 3) analyze the data. Companies use learned information about a drug to inform the next iteration of testing in real time – in early phases, this aligns with the goal of quicker approximation of the maximum tolerated dose (MTD), avoiding exposing volunteers or end-stage patients to potentially unsafe or ineffective doses of drugs; in later phases, there are opportunities to reduce sample sizes, adjust dosing schedules or regimens, manage safety risks, or even adjust the timing of efficacy evaluations all to allow a better, more comprehensive understanding of patient safety and potential clinical benefit. There has been buzz regarding adaptive design for some years now and statisticians can explain the nuances with remarkable clarity, but deploying an adaptive trial carried substantial risks due to the lack of guidelines from regulators. However, a new guidance document issued in November 20192 now allows companies to use adaptive design with some direction from the FDA.
There is a lot more up-front effort when planning for a trial that includes an adaptive element (e.g., changes in treatment allocation, increases or decreases in sample size, immediate continuation from one phase to the subsequent phase, and modification of eligibility criteria3) as compared to a traditional clinical trial. However, this up-front work can directly lead to benefits both for drug developers and, most importantly, the patients participating in clinical trials. For example, the continual reassessment method (CRM) is a Bayesian adaptive methodology for assessing dose-toxicity relationships and estimating the maximum tolerated dose in Phase 1 safety and pharmacology studies, predominantly in oncology trials. This method may allow for faster escalation to higher doses, which can shorten the duration of the trial and reduce the sample size required to determine the MTD. In addition, CRM designs may have improved accuracy for estimating the MTD, which can reduce the likelihood of exposures to excessively toxic doses in Phase 1 studies and more effectively establish the limits of tolerability for later stage development. 4,5
Although adaptive designs can ultimately save companies time, money, and resources by shortening the clinical program, it is important to thoroughly vet this decision. An experienced biostatistician should test multiple types of trial methodologies to ensure an adaptive design approach makes more sense than a standard 3+3 design or a single-ascending dose design. One of the best scenarios for adaptive design is implementing the use of data from an earlier trial that can inform the probability of Phase 3 trial outcomes. By using this data in a probabilistic method to inform the study design, sponsors can reduce trial sample size, which may shorten the duration of the study. However, if sponsors attempt to do this without proper preparation and information, they can make inaccurate assumptions and increase the likelihood of trial failure. Understanding these complexities and acknowledging the potential bias involved in these trials are crucial.
Engaging expert statisticians, communicating with stakeholders, and planning for a longer lead time are all important considerations, but companies tend to be aware and prepared for these. The two things that are often overlooked are: 1) investigators’ involvement and understanding of the additional challenges and complexities they will face during clinical trial conduct and 2) communication of the potential design adaptations to the trial participants.6
Making sure investigators understand and are on board with the adaptive design is critical as it requires an extra level of involvement from them when interim analyses are taking place throughout a trial. Also, they need to understand the limitations that go along with the data, what can and cannot be inferred from the results, and how to communicate the results.7 This is often made clear to the stakeholders of the sponsor company, but not to the investigators, and it can lead to problems with trial execution, which can have disastrous consequences down the development and regulatory roads.
With potential trial participants, it is just as important to communicate the potential adaptations of a trial: what might change, what might not, and how those changes could affect them individually. The best way to do this is by creating additional materials for sites to provide during the screening process8 and by making sure that the clinical trial staff fully understands the design. While this is extremely important from an ethical standpoint, it can also benefit the sponsor company by encouraging enrollment. Patients are more likely to sign up for the trial if they understand that potential adjustments to the treatment regimen may give them a higher chance of receiving effective treatment rather than placebo. As stated previously, the goal is to expose more clinical trial participants to pharmacodynamically active dose levels in a shorter amount of time.
As adaptive design becomes more common, examples where those designs have delivered clinical or financial success are more readily available online. The best way to avoid issues in outlining your adaptive design is to find practical, successful examples and to engage people who have experience in this area. The more innovative we become as an industry, the better the outcomes for everyone involved. Adaptive design should be used more often by more sponsor companies, and the FDA’s new guidance document and its focus on this topic in recent years is proof that it agrees.
About The Authors:
Julia Ogier has experience in strategic clinical development, cross-functional project management and clinical operations, particularly focusing on study start-up in early-stage trials. At Halloran, she provides clinical operations support and clinical development expertise to pharma and biotech clients including project management, CRO/vendor management, clinical site feasibility and start-up, operational oversight, and ensuring adherence to GCP, ICH guidelines, and applicable regulations. Prior to joining Halloran, Ogier was an account executive at MacDougall Biomedical Communications, where she managed client relationships and developed and executed corporate communications strategies relating to the design, status, safety and efficacy of Phase 1 through 3 clinical trials.
Matthew Cardinal has more than 15 years of experience in biologics, drug, and medical device development. At Halloran, he functions as a team leader in the design, planning, and management of development strategies for hematology, immunology, gene therapy, regenerative medicine, and infectious disease/vaccine programs. Cardinal supports all development lifecycles, from pre-IND through registration, with a focus on translational clinical research and quantitative drug development. He has a specific background and interest in indications for rare diseases and AAV gene therapy, including hematology and rare immunodeficiencies. Prior to joining Halloran, Cardinal was a clinical development lead for early phase rare disease programs at Pfizer.