home > > spring 2011 > adaptive trials designs in clinical development

Adaptive Trials Designs in Clinical Development

Vital Statistics

Are adaptive design studies a new challenge to rise up to? The author seems to think so, and highlights the benefits of these studies in clinical research – one of which is increased efficiency, especially in the learning phases – but also maintains that you must have your wits about you when embarking down this road.

In March 2006, the Food and Drug Administration (FDA) issued a Critical Path Opportunities Report calling for a judicious rationalisation of clinical development.This report insisted on the necessity of bringing in new adaptive design methods in Phases 1 and 2 – to increase their efficacy – as well as securing the subsequent development of Phase III, for which the failure rate had reached an incongruous high of 50 per cent. The FDA’s interest in these new studies materialised in February 2010 with the release of ‘Guidance for Industry:Adaptive Design Clinical Trial for Drugs and Biologics’. Its target was to guide the investigators in charge of research and development in their choice of adaptive studies and to provide them with recommendations regarding their communication policy with the FDA.

To Be Adaptive or Not To Be Adaptive

Over the last few decades, clinical research mainly based its development on a conventional probabilistic approach, known as the ‘frequentist’ approach, laying a priori hypotheses H0 and H1 that are to be rejected after data collection. This orthodox approach is based on the exploitation of the Law of Large Numbers and the central limit theorem.

Now, the limits of clinical trials performed using the classic frequentist method during the learning phases (Phases 1 and 2) are well known.Their rigidity calls for interim analyses that increase the rate of type I errors, eventually leading to multiple statistical comparisons. Therefore conventional studies are not suited to the requirements of these studies.

In an effort to improve the accrued data process, other theoretical frameworks challenging this standard approach were proposed. One of which was the Bayesian method, which advocates summing up all the knowledge available in the form of an a priori distribution of the parameters, which is assessed and can be revisited later on during the process.

When applied to clinical situations that require fast and robust responses, the Bayesian approach, along with the new frequentist methods – adaptive designs – make it possible to avoid the tentative and time-consuming process of conventional frequentist methods which are common in the learning phase.The synergy offered by these two approaches, as well as their mastery in the learning phase, should become the rule and is already proving to be very effective in terms of securing development.

Frequentist Adaptive Design Studies

While frequentist adaptive design studies proceed from the same hypothetico-deductive logic as conventional studies, they differ in the level of their flexibility; the former allows changes to be made to the original experimental plan during ongoing trials, using accrued data without undermining its validity or integrity.The frequentist adaptive trial is therefore more ethically sound compared to the conventional frequentist trial as it enables investigators to reject inefficient molecules or doses much earlier in the process, therefore limiting the number of patients who are given non optimal treatment doses.Three common types of adaptive designs are:

Play-the-Winner Designs
This type of study (adaptive randomisation design) is one of the most commonly known. In this approach, the first treatment assignment is determined at random and their response to the treatment dictates the next patient’s assignment. What is very interesting about this approach is that it allows one to quickly increase the group size of the patients who benefit from the best treatment regimen, while reducing the size of the group receiving inefficient or badly tolerated treatment.

Drop-the-Loser Designs
Drop-the-loser designs are structured in order to potentially eliminate the parallel arm(s) of the study population for whom the study drug shows insufficient efficacy or an unacceptable toxicity within the scope of the pathology studied at each interim analysis.

Seamless Studies
These usually combine two or more clinical studies corresponding to two or more successive development phases into one multi-stage trial, therefore avoiding the gap created by separate trials – for instance, Phase 2 study result analysis then Phase 3 study design and implementation.

The Bayesian Approach

Recently introduced in clinical R&D, the Bayesian approach is ideal when initiating the learning phase.This approach bases itself on the continuous update of probability distributions as data are accrued,making it possible to predict the therapeutic outcome of the study drug with more accuracy.

The frequentists’ objections to the reliability that can be granted to a priori distributions and to the opportunity of turning them into mathematical models are not well-founded. Indeed, the possible discrepancies experienced during the elaboration process of these distributions become blurred, or even fade away, as the flow of data stemming from experimentation is acquired.The Bayesian methodology can prove particularly useful in Phase 1 as it makes it possible to consider the preclinical development data during the initiation of first-in-man studies. One of the most commonly used clinical applications is the ‘continual reassessment method’ (CRM), a dose escalating method which has proven its worth especially in oncology.

Excellence in Adaptive Design is Essential

Adaptive studies are generally used in major indications such as oncology, cardiology and neurology. In most cases, the implementation of an adaptive study does not necessarily result in a reduction in costs or timelines, but primarily secured development. Indeed, these studies entail the employment of a team of state-of the- art biostatisticians who are highly skilled in these methods, and will involve a lengthy and more complex planning phase compared to conventional trials, while requiring stricter constraints from a logistical point of view.

At this stage, interacting with the regulatory agencies is unavoidable. It is crucial to submit a relevant and convincing rationale of the adaptive study to the authorities while dealing with all ethical considerations in a concise way.The report should also include the study operational feasibility criteria so as to avoid unexpected problems.

Excellence in Adaptive Management: a Priority

Whether it is related to new frequentist adaptive studies or to Bayesian studies, it is absolutely necessary for the project manager to monitor activities and ensure the required level of quality.This calls for the implementation of an optimised organisation scheme that is both adaptive and reactive; therefore requiring a certain amount of resources.Actually adaptive studies call for a manager that must combine expertise, organisational intelligence and allocation of means, monitoring, data collection, immediate analyses, decision-making and local adaptation of resources.

These studies therefore necessitate a pool of well-trained human resources who must respond conscientiously ‘on the spot’ and in a reactive way to vital issues at any time during the trial, whatever the geographic location of the participating site(s) involved. In any case, the management’s objective is to keep as close as possible to the constraints of a demanding protocol.

A High Scientific Management Level

Most adaptive protocols, especially the Bayesian methods, are highly complex, and it is therefore of paramount importance that the whole study team, including investigators, independent data monitoring committees (DMC), project teams and dedicated CRAs speak the same language.The field teams’ medical culture relating to the pathology concerned and to its therapeutic environment must be optimised.They must also have a deep knowledge of the adaptive methodology involved as well as of the study protocol in its entirety. Therefore the following are required:

  • A dedicated project team that is able to explain clearly to all investigators the relevance of using new design methods, particularly Bayesian methods, within the scope of the study concerned
  • The need to convince the investigators’ pool to what extent a precise follow-up, accurate estimates and complete data are absolutely necessary of the task; motivate the investigator and involve her/him continuously in the trial
  • The ability to sustain a medical exchange with the independent DMC members who are not necessarily versed in new probabilistic methodologies; to be able to debate on methodological issues and opportunities of use with regards to product characteristics and the clinical indication retained for the development
  • Project management and monitoring activities of adaptive clinical trials must maintain a strong link between the investigators’ pool and the project managers, making sure at all times that the study is conducted in compliance with the protocol and with Good Clinical Practices. The CRAs will have to make sure that the patients’ private rights are respected (informed consent) and be responsible for the reliability and traceability of available data
In summary, a perfect knowledge of the probabilistic and pathological environment well-assimilated by all the actors of an adaptive study enables them to gain the respect of the investigators, namely opinion leaders, which is a major element of adherence to the whole study.

Flawless Logistics

From a logistical point of view, adaptive design studies are fundamentally different from conventional studies; indeed, all the resources required to complete a conventional study are determined before initiation.The fact that an interim analysis brings about the early termination of the study does not influence the initial mobilisation of resources.

With regards to adaptive design studies, namely the ‘play-the-winner’ and the ‘dose-finding’ types, it is of crucial importance to elaborate a reactive operational intervention scheme in time and space. Hence the following requirements:

  • Dedicated CRAs and project managers that are fully informed of the action protocol inherent to logistical procedures and are able to respond in real-time to the investigator’s needs. Indeed, a single logistical problem occurring during an adaptive study can invalidate the study, or cause its termination
  • Simplifying communication via collaborative platforms is obviously an important aspect, but is not sufficient to maintain real-time study operation. Indeed, this requires the intervention of personnel who are dedicated to the study and able to respond promptly and in practical terms to changes
  • The availability of a user-friendly eCRF developed in-company by a team well aware of the requirements of each specific study to optimise data collection
  • The CRAs, who must be well experimented and previously trained to complex studies, must be able to adapt continuously; this entails a working capacity well above mere vocational conscientiousness. Indeed, every CRA must feel as though they have full responsibility, and are able to increase the monitoring power depending on the recruiting rate of the investigational sites.This means that junior CRAs should not be involved in these types of studies
  • To integrate the sub-contractors very closely into the adaptive design study in a global development scheme. Negotiations must be tight with centralised laboratories so as to obtain immediate results to define the next doses required
  • A high flying data management service versed in new methodologies which continuously analyses data as they are collected
  • High quality back-up staff that have the same knowledge of the project as the participating teams and is ready to be deployed if needed


Both frequentist and Bayesian adaptive design studies are of increasing interest in pharmaceutical development, more particularly in the learning clinical phases.However, while the emergence of these methods prove to be a real whiff of fresh air for a suffering clinical R&D, they must be used with full knowledge of the facts. Contrary to what may have been widely advocated, their implementation does not necessarily mean shorter and cheaper clinical trials. The complexity of the protocol design takes much longer – and is therefore more time-consuming than the standard conventional protocols – and the monitoring activity required calls for the intervention of skilled, dedicated project teams, with a wealth of experience in problem solving, meaning they are hardly cheaper either.

Scheduled interim analyses in the protocol, performed by an independent data monitoring committee – whose members are often key opinion leaders who are very critical in their evaluation – will encourage management excellence and limit any deviations regarding the protocol or ethics.

Implementing these adaptive studies must; therefore, take into account the return on investment provided, in terms of the information being gathered to guide the design of later studies with a view to consolidating and securing the drug development plan in which they will integrate as a whole.The reduction in the number of patients needed, their reduced exposure to unsuitable therapeutic doses as well as longer exposure to beneficial doses will obviously be a factor that counts in their favour.

At the end of the day, one must be aware of the fact that as adaptive protocols involve the scheduling of possibly numerous interim analyses whose occurrence is linked to the dynamics of accrued data, they will demand specific technical and human resources to design and implement each particular study. Bearing that in mind, it is important to choose wisely when considering a CRO as they play a pivotal role in the success of the overall study.


  1. Clinical Path Opportunities Report, Food and Drug Administration, March 2006
  2. Guidance for industry, Adaptive design clinical trials for drugs and biologics, February 2010
  3. Kramar A and Paoletti X, Analyses intermédiaires, Bull Cancer 94(11): pp965-974, 2007
  4. Chow S-C and Chang M, Adaptive design methods in clinical trials – a review, Orphanet Journal of Rare Diseases 3: p11, May 2008

Read full article from PDF >>

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

There are no comments in regards to this article.

Michel Hantzperg is Chairman of International Clinical Trials Association (ICTA). With 10 years of experience as a medical research director in the European pharmaceutical industry, Michel started ICTA, a full-service CRO based in Dijon, France, in 1983. In 1998, he turned to the European R&D sphere and strengthened ICTA’s growth with active strategic alliances. In 2004, he created Themis, a consulting company specialised in study designs, advanced biostatistics, database exploitation and publications, and developed a specific interest in the new adaptive methodologies. Michel is an MD from the University of Paris.
Michel Hantzperg
Print this page
Send to a friend
Privacy statement
News and Press Releases

Huge boom in Korean Pharma reported at CPhI Korea

Amsterdam, 19th September 2019: The recently-closed CPhI Korea – co-organised by Informa Markets and the Korea Pharmaceutical Traders Association (KPTA) – has seen a surge in growth across domestic and international pharma companies, as well as a sizable rise in overall attendees. As reported earlier this year in the provisional findings of the CPhI Pharma Index[1], Korea has seen a rapid growth in its international reputation (seeing its ‘overall competitiveness’ rise 14% in the last two years), which is now translating into a sizable growth in the market and at the event.
More info >>

White Papers
Industry Events

World Vaccine Congress Europe

18-21 October 2020, Barcelona, Spain

The World Vaccine Congress is an award-winning series of conferences and exhibitions that have grown to become the largest and most established vaccine meeting of its kind across the globe. Our credibility is show through the prestigious scientific advisory board that spend months of hard work creating a new and topical agenda, year on year.
More info >>



©2000-2011 Samedan Ltd.
Add to favourites

Print this page

Send to a friend
Privacy statement