spacer
home > ict > spring 2012 > adaptive designs: early phase trials
PUBLICATIONS
International Clinical Trials

Adaptive Designs: Early Phase Trials

 

Early phase clinical trials can benefit from the flexibility offered by adaptive designs, but only if adequate planning and effective communication strategies are in place. The rising cost of R&D and earlier access to new treatments represents a public health issue. Frequently, strategic decisions, such as initiating large and extensive confirmatory Phase 3 programmes, are made while questions about dose and dose regimen, patient population, predicted efficacy and safety remain imprecisely answered from exploratory Phase 1/2 programmes. The direct consequence is a high Phase 3 failure rate. The FDA, under the Critical Path Initiative, has called for innovative advances in trial design, especially for the use of prior experience and accumulated information.

Innovative adaptive design methods and approaches in clinical trials, along with the potential use of Bayesian techniques, have received encouragement in clinical R&D, which has led to the establishment of new guidelines in this field, as well as several workshops (1-4). Introducing clinical studies with adaptive designs during the exploratory phases of clinical development allows for better characterisation of the new therapy; indication and patient population selection may stand to benefit the most. Adapting or adjusting the design of a clinical trial based on planned interim data analysis could result in changes to a number of properties: sample size, number and type of treatment categories, number of patients allocated per category, changes in dosing regimen, duration of treatment, modification of inclusion/ exclusion criteria, acceptance of co-medications and even statistical procedures (5). This is in contrast to fixed-design studies, where no changes are made from the first patient dosed to the last – other than terminating the study for lack of efficacy or safety concerns.

Overall, performing Phase 1/2 clinical trials with adaptive approaches generates a better data package for predicting the benefit and risk in selected patient populations, and therefore allows a robust decision justifying the initiation of Phase 3 activities to be made. Thus, investment in a welldesigned Phase 1/2 programme using continuous learning paradigms and adaptive approaches is a reasonable strategy for designing and conducting a successful confirmatory package, thereby lowering the risk of inappropriate patient exposure (wrong population, wrong dose, and so on). Consequently, failure at later stage – when investments are even more substantial – is reduced.

Conducting informative Phase 1/2 programmes using adaptive approaches is generally viewed positively by regulators, especially considering the totality of the data available when decisions need to be made. A typical request from regulators is to be shown evidence that the lowest effective dose (and thereby likely the safest dose) has been identified. In this respect, Bayesian statistics and PK/PD modelling allow the identifi cation of promising dose ranges to be tested, including minimum and maximum therapeutic doses. When a new treatment is first introduced to subjects (healthy or patients), approaches such as the constant reassessment method is ideal (6). This method continuously evaluates response to dose/exposure (safety, biomarkers, clinical endpoints and so on) and subject numbers, adjusted based on observed data, in order to define the maximum tolerated dose (or equivalent, such as maximum needed exposure) – and more importantly, the pharmacologically active dose range. This has been regularly performed in oncology trials and Phase 1 studies for many years.

The EMA reflection paper on adaptive design focused mainly on methodological issues in confirmatory trials, while the FDA draft guidance looked at clinical trial effectiveness to support drug marketing. However, it should be noted that the ICH-E4 guidance stated that a well-controlled doseresponse study (such as an adaptive dose ranging study) could also serve as primary evidence for effectiveness, and so potentially only one standard pivotal study might be needed in addition to the adaptive design study. This is likely to be the case for small patient populations.

Clearly, adaptive approaches are implemented to rapidly and effectively characterise and predict, with high confidence, the full potential of the new treatment during the exploratory and learning phase without compromising the validity and integrity of the intended studies. This article will review some principles that should take priority when performing clinical trials with adaptive designs and especially during the explorative phase of clinical development.

Adaptive Design in Exploratory Clinical Phase: Two Case Studies

The following two examples represent different approaches to illustrate that there are a number of ways to benefit from the planned adjustment of key parameters during the study based on previous knowledge (preclinical and clinical data). These two examples also illustrate that there is no single unique recipe for introducing adaptive design in a clinical study, especially during the exploratory phases of drug development. In these two examples, the ‘learning’ and ‘confirming’ stages were clearly delineated from each other.

The first example shows an adaptive dose-finding Phase 2 trial for a compound developed for an acute treatment (see Figure 1). The study design was randomised, double-blind, placeboand active-controlled, with parallel groups to evaluate the efficacy and tolerability of the compound. The compound was known to have an excellent safety profile from previous clinical studies mostly performed in healthy subjects, but no indication of pharmacologically active dose range could be suggested by these studies or from biomarkers or PK/PD modelling. The primary objective of the trial was to identify the least effective dose, relative to the placebo. The clinical endpoints were obtained rapidly in respect to treatment administration. Implicit objectives were to select one or two doses for Phase 3, maximise proof of concept at the end of Phase 2, minimise time before Phase 3, compare the efficacy results to an active comparator, and to adequately characterise the dose-response curve. In the first stage, a very broad range of doses were tested with seven different dose levels (with a 25-fold difference between the lowest and highest dose) in about 16 patients in each dose group, versus placebo and a positive control. The study was set up to show a difference versus placebo, but not to show difference between dose levels. After interim analysis, doses showing no effi cacy were dropped and newly enrolled patients were randomised to take the remaining pre-administered doses (three upper dose levels), placebo or positive control. The strategy was to maximise the ability to detect treatment effect and increase randomisation of high dosage while reducing randomisation of lower doses not sufficiently better than placebo. This led to a better characterisation of the efficacy of the compound at varying dose levels on the dose response curve. With an adequate control of the Type 1 error (chance of a false conclusion that the treatment is effective), trial size and time was decreased in comparison with a similarly powered non-adaptive design. Fixed total sample sizes kept the budget and timing in control.

The second example, for a chronic disease, presents a different approach in which the first stage focused on establishing proof-of-concept at one selected dose (see Figure 2). This first dose level was selected, based on preclinical and clinical data as well as biomarker and PK/PD modelling, to select a likely efficacy dose with acceptable tolerability. Then, if efficacy was observed versus placebo, the second stage would open up the trial to a larger number of doses in order to define a robust dose range as well as include a positive control. In this approach, adaptive design could rapidly conclude futility (such as no difference from placebo) at an early stage and therefore limit further investment, or in case of success validate important milestones for future development. There are multiple variants of this design; for example, using a positive control during stage 1, or if no difference from placebo is observed at interim analysis, continuing dosage and including a positive control to confirm the response potential of the selected patient population.

This interesting approach differs from the first example because there is only one dose level under investigation at the first stage, reducing upfront trial workload and investment, and continuing the trial only if efficacy is observed. This approach is appealing due to its ability to limit initial investment and, if successful interim data are generated, activate the second stage of the study to document dose ranging efficacy data.

 

Good Adaptive Principles

Early Planning

Adaptive designs do not offer a solution for poor clinical development planning or for salvaging an ongoing trial, but should be considered as an additional tool to answer working hypotheses and meet well-defined study objectives. Planning of such trials is a key factor that usually requires more workload upfront than is needed for conventional studies with fixed design. Early identification of an opportunity for adaptive design is a key to success. Consultation with key regulatory agencies can help; some regulators are willing to provide advice at the protocol design stage. These consultations may avoid a costly refusal of a trial at regulatory submission milestones.

As with any trial, the planning phase will consider recruitment rate, endpoints, route of administration, position in development plan, treatment duration and so on. For example, if recruitment is too fast, the ability to learn is hindered and slower recruitment might be better. It is necessary to base changes on available parameters that can be rapidly measured and compared to the recruitment rate. Endpoints that read out late will not permit adaptive design to be effective. Optimal recruitment strategy can be established with clinical trial simulations.

Simulations are of great help during the planning of adaptive designs and require in-depth expertise. They enable understanding of the properties of these complex pathophysiological systems, to improve design characteristics and to highlight potential operational issues and logistics challenges. Issues and related consequences can be evaluated using extensive simulations covering a broad range of scenarios, and can be included in the statistical section of the protocol or in the statistical analysis plan. As no change is permitted ad hoc during the trial, each option to implement a change must be thoroughly specified in the protocol. Strong coherence must exist between the different stages of the trial.

Taking into account the complexity of such trials, and considering all the key factors and parameters that could be adapted during the course of such study, budget estimates are more complex than with a fixed design. Budget estimations based on an adaptive design should follow the trial assumptions in considering not only the whole range of expected cost but also the worst case cost estimations; they should be evaluated properly with regard to the overall clinical development plan.

Communication Plan

As shown in Figures 3 and 4 (page 22), early planning needs to bring in a cross-functional team at a much earlier point when compared to a conventional study design. Strong interactions between functions are critical; the roles and responsibilities of each partner involved in the trial have to be clearly defined.

A non-blind independent statistical centre (ISC or equivalent) is responsible for preparing data based on interim analyses and should prepare a briefing document intended for review and discussion within the data monitoring committee (DMC) – a key partner in an adaptive trial. DMC members include therapeutic research experts, statisticians and clinical pharmacologists who are independent of both the study and the sponsor. Their main responsibility is to continuously assess scientific, clinical and ethical aspects of the trial without compromising the on-going blinding of the trial. The DMC brings recommendations to the sponsor on the basis of unblinded interim analyses produced by the ISC, but also of any data source external to the trial that could be considered necessary or useful to be taken into account. The final responsibility to respect the recommendation of the DMC or not is at the level of the sponsor.

When full double-blind is mandatory in later phases of clinical development, maintaining the blinding in exploratory trials is case dependent and is less strict (for example, it could be limited to the subjects/patients and clinical sites personnel).

Procedures and guidelines must be clearly defined to ensure the integrity and validity of the trial. It is strongly recommended that the sponsor write a DMC charter, including a decision tree, in which the whole range of scenarios with matching changes and decisions are explicitly mentioned. There should be clear stopping rules for the DMC so they can make the right decision at any interim analysis. This decision charter will also be a precious tool if the DMC has to deal with an unexpected situation, in which case some deciding authority can be delegated. This charter should be part of the submission to the regulatory authorities, when applicable.

Secured channels and firewalls must be established to protect the necessary blinding throughout the trial. Meetings are frequent during an adaptive design trial and it might sometimes be useful or necessary to refer to the competent authorities or members of various committees (scientist, DMC and so on). Adequate training of investigators and site personnel who are not familiar with the methodology of the adaptive designs must be planned.

Trial Management and Operational Challenges

Clinical trials with adaptive designs are complex, and strong project management is necessary to ensure successful implementation and realisation. Therefore, close attention must be paid to the recruitment rates, site and trial supplies, as well as global coordination between data capture tools, data analysis and decisions. Minimising the time required for interim analysis and moving forward from the decision point to the implementation of adjustments are the biggest challenges (7).

The complexity is increased by the fact that these studies are generally multi-centre. Consequently they require the use of centralised management tools and systems, as well as flexible systems and processes for an optimal implementation. The logistical and operational burden can be evaluated based on trial simulations (of data outcomes and trial planning) carried out at the time of planning. Trial simulations facilitate comprehension and communication between the various partners and vendors of the study, while showing in concrete terms how each scenario and recruitment rates at different sites will impact the trial.

In such flexible designs, drug supply-related issues are common. A reliable estimation of the amount of IMPs that have to be sent to each site is essential. Use of real-time supply chain management as well as simulation removes the need for stocking all dose strengths at all sites (7).

Not delaying interim analyses and decision making should be the golden rule in adaptive trials. To reach these objectives, it is important to assess all possible issues that could impact these timelines: early planning of meetings, anticipating changes that might be needed in some documents (informed consent, eCRFs and so on). Clinical monitoring is also critical and needs to be continuous to make sure clean data are available on time for the independent DMC. Appropriate tools for centralised monitoring (such as those linked to electronic data capture), as well as for detecting covariates, outliers and biomarker responses, and analysing subpopulation online during the study, are initiatives of great value.

The Regulatory Perspective

Currently, regulatory authorities are relatively comfortable with adaptive designs in a Phase 1/2 setting. Commonly accepted designs include response-adaptive randomisation in Phase 2 trials, blinded sample size re-estimation, and early termination of a trial for reasons of efficacy, futility or safety. Currently, adaptive designs at the junction of Phase 2/3, or so-called seamless Phase 2/3 designs, are still in an experimental and learning phase and are accepted by the authorities on a caseby- case basis. Less time spent on carefully reviewing the data between the different stages in seamless Phase 2/3 designs can sometimes be perceived as a serious disadvantage and may lead to poor efficiency eventually. Some designs remain mostly unacceptable to regulatory agencies due to the potential for introducing bias. These include unplanned enrichment of the patient population and change in primary endpoint or choice of statistical testing approaches (superiority/non-inferiority) (8).

Sponsors may seek advice or ask the authorities for a scientific opinion before submitting the trial. Sponsors must fully justify the reason for running the trial using an adaptive design, especially if stage two of the study will be part of the confi rmatory trials.

Submission requires an exhaustive explanation of the adaptive characteristics and potential changes, along with a rigorous methodology and a rationale for the adaptive design. Regulatory bodies pay careful attention to the recommendations provided by the DMC to the sponsor. The decisional charter established by the sponsor and endorsed by the DMC should be in the annex of the submission as well as simulations established during planning, which provide clarification of risks and benefi ts for the complete range of scenarios. The sponsor must also guarantee the support and logistics of the trial, whatever adaptations may be decided after the interim analysis.

 

Conclusion

The primary purpose of exploratory clinical studies in drug development is to learn about various important characteristics of the new therapy such as dosing, exposure, differential patient response, biomarker responses and so on, to increase confidence in the mechanisms of action and safety. Performing clinical trials with adaptive designs, when justified and feasible, permits research of the new therapy in a much more efficient way and offers additional opportunities that can improve later studies at the confirmatory stages. Overall, this should increase R&D efficiency and allow patients to benefit sooner from new treatments.

Acknowledgment

The authors would like to thank Quentin Watthez and Dr Harry Bleiberg for their input and review of this article.

References

  1. Draft Guidance for Industry, Adaptive Design Clinical Trials for Drugs and Biologics, United Stated Department of Health and Human Services, Food and Drug Administration – CDER and CBER, Washington, DC, 2010, available at www.fda.gov/downloads/Drugs/ GuidanceComplianceRegulatoryInformation/Guidances/ UCM201790.pdf
  2. European Medicines Agency, Refl ection Paper on Methodological Issues in Confi rmatory Clinical Trials Planned With an Adaptive Design, Doc Ref: CHMP/EWP/2459/02
  3. Guidance for the Use of Bayesian Statistics in Medical Device Clinical Trials, United Stated Department of Health and Human Services, Food and Drug Administration – CDHR and CBER, Washington, DC, 2010
  4. Report on EMA/EFPIA 2nd Workshop (2009): Adaptive Design in Confi rmatory Trials, Doc Ref: EMEA/779520/2009, available at www.efpia.eu/Content/Default.asp?PageID=606
  5. Chow SC and Chang M, Adaptive design methods in clinical trials – a review, Orphanet J Rare Diseases 3(11), 2008
  6. Faries D, Practical modifi cations of the continual reassessment method for Phase I cancer clinical trial, Journal of Biopharmaceutical Statistics 4: pp147-164, 1994
  7. Quinlan J and Krams M, Implementing adaptive designs: Logistical and operational considerations, Drug Information Journal 40: pp437-444, 2006
  8. Sietserna W, Aydernir U and Sennewald E, An introduction to adaptive clinical trial designs, Regulatory Rapporteur 7(10): pp4-6, 2010

Read full article from PDF >>

Rate this article You must be a member of the site to make a vote.  
Average rating:
0
     

There are no comments in regards to this article.

 You must be a member of the site to make a comment.
spacer

Denis Gossen is a Clinical Pharmacologist with a primary interest in the identification, development and implementation of biomarkers from early preclinical and clinical development, to proof-ofconcept studies in various therapeutic areas. He has worked for various global pharmaceutical companies for more than 15 years. In 2007, he co-founded Aepodia with Dominique Demolle.

Dominique Demolle has both strategic clinical operational and clinical pharmacologist expertise and experience. She has worked in the pharmaceutical industry for more than 20 years, assuming various leadership roles including the management of international operational groups in early clinical development (Phase 1-2) and of clinical research units.

spacer
Denis Gossen
spacer
spacer
spacer
Dominique Demolle
spacer
spacer
Print this page
Send to a friend
Privacy statement
News and Press Releases

ADM Supports Clinical Trial into Effects of Probiotic Supplements for COVID-19 Patients in Hospital Care

ADM and Hospital de Sagunto, one of the largest medical centers in the Valencia region, today announce their partnership for a clinical trial into the effects of using a food supplement containing live microbial strains to promote gut microbiome balance and help improve health functions related to outcomes for COVID-19 patients in high-risk groups. On 18th March Dr Xavier Cortés, gastroenterologist at Hospital de Sagunto, communicated with medical staff in China who had observed that many COVID-19 patients had exhibited abnormal changes in their gut microbiome.
More info >>

White Papers

Quality Systems Guide

Research Quality Association

This guide concentrates on the core, underlying principles of quality that are pivotal – explicitly or implicitly – to all quality management standards that impact on research and development.
More info >>

 
Industry Events

Clinical Operations in Oncology Trials West Coast

17-18 November 2020, Hilton San Francisco Airport Bayfront

Clinical Operations in Oncology Trials West Coast will be returning this April for another 2 day event full of thought-provoking presentations, discussions and roundtables. This years' conference highlights include the high-level, interactive immuno-oncology discussion panel where specialists from Shasta Bio Ventures, Abbvie and BeiGene shared their top tips on how to run successful and impactful immuno-oncology studies.
More info >>

 

 

©2000-2011 Samedan Ltd.
Add to favourites

Print this page

Send to a friend
Privacy statement