BARCELONA, SPAIN - For companies looking to adopt adaptive trial designs, Merck VP of Biostatistics and Research Decision Sciences Jerald Schindler urged sponsors to frontload most of the adaptation in the non-pivotal portion of the clinical trial program and limit their own involvement in the adaptation that does take place during the pivotal portion.

Speaking at the Drug Information Association EuroMeeting March 4, the Merck exec explained his view of an adaptive clinical development program.

He suggested viewing the process as a non-pivotal section - which would identify maximum tolerated dose, assess pharmacokinetics, explore dose response and explore potential endpoints - and a pivotal section that would explore patient populations and demonstrate efficacy with a subset of doses. He envisioned beginning the process with a multitude of options, and refining and limiting those options as development progressed.

Escaping the traditional three phases of clinical trials and merging into a malleable continuum of development is a common conceit of adaptive trial design (¹ Also see "The Next Phase: FDA Looks Beyond Three-Stage Drug Development Model" - Pink Sheet, 13 February, 2006., p. 10).

The concept of adaptive clinical trial designs that would create more efficient, productive clinical trials through flexibility and modern techniques is the cutting edge of clinical trial design, but so far it has yet to really move from idea to practice.

Although adaptive trial design covers a gamut of possibilities, a central premise is that an adaptive trial could be modified along the way to address emerging issues and yield better results. Thus, one of the primary problems posed by adaptive designs is that to adapt the design mid-trial, someone must be monitoring the results to allow for course correction.

This data integrity problem is probably the chief barrier that is keeping companies from jumping onboard the adaptive trials train. After all, trial integrity "is probably one of the greatest concerns of the regulator," Peter Thomas, manager of methodology and innovation at Novartis, UK, noted at the DIA meeting.

The European Medicines Agency addressed this concern in a reflection paper issued last year, which noted that interim analyses "always introduce the possibility of damaging the integrity of a trial" (² Also see "EMEA Looking At Limited Use Of Adaptive Design In Late-Stage Trials" - Pink Sheet, 10 December, 2007., p. 16).

To minimize this problem, EMEA says sponsor involvement in interim analyses "is discouraged."

In a white paper on the subject, a Pharmaceutical Research & Manufacturers of America working group said interim results should be reviewed by a data monitoring committee that is adequately "firewalled" from personnel involved in the trial (³ Also see "Adaptive Trials Going Mainstream With Industry White Paper, FDA Guidance" - Pink Sheet, 24 July, 2006., p. 17).

Try To Adapt With Non-Pivotal Stage

Instead of trying to conceive of ways to construct adaptive trials that would satisfy regulatory standards, at least initially, the best application of adaptive trial design might be for non-pivotal studies.

"In the non-pivotal stage ... you have a lot of flexibility to experiment, to be creative," Schindler noted. "The pivotal stage is a completely different story, and if you separate it into these two phases ... it makes it much more palatable for the companies to deal with."

During the non-pivotal phase, sponsors can "have a flexible decision process that's part of the adaptive process, so you can use almost any adaptive technique that might be comfortable ... you can add doses, subtract doses, increase the sample size, decrease the sample size," he said.

There are few, if any, limitations on company data review during the non-pivotal portion, Schindler said. Company statisticians can have "almost unlimited access to the data" and the only risk is that bad decisions are made because bias is introduced into interpreting the data. The information may support the pivotal phase, "but it won't be used for approval of your product."

Sponsor input for decisions that affect study modifications should occur only when the resultant data "are never going to go to the regulatory agencies for approval," Schindler stressed.

For a pivotal trial, "make sure that the data are as pristine as possible," he said. "Make sure there is ... no sponsor involvement for an adaptive pivotal trial."

Set Up Rules To Keep Sponsor Seperate

Schindler and other panelists suggested ways to set up an adaptive design for a pivotal trial and avoid the data integrity problem.

The key to doing this, they explained, is to use a third party (a truly independent data monitoring committee) and give them explicit directions from the outset.

"The company should have no involvement in the data monitoring committee," Schindler asserted.

"When you're in the pivotal stage, I would set up rules driven by algorithms. State them in advance, make it clear to the regulatory agencies you're working with, and make it clear to the data monitoring committee these are the only circumstances that you want to make changes to the trial."

To protect integrity, he urged that interim analysis be performed by an independent group based on up-front decisions on what changes will be made for various scenarios.

Some Decisions Need Company Participation

Novartis' Thomas also discussed the use of an independent group for interim analyses.

But, he added, what "if the data does not perform in quite the way it was anticipated? It might not fit directly into an algorithm." That is a big decision to place on the DMC and "many companies would be cautious about doing that, and that presents a barrier to using adaptive designs."

In such situations, "there is a route to consult with the company before making a decision," Thomas contended.

This includes limiting data access to the fewest number of company personnel, "one, maybe two people," and to those who are "properly distanced from trial operations." Only information relevant to the decision would be revealed and rules and procedures for company involvement would have to be set up beforehand, he said.

- Cathy Dombrowski (c.dombrowski@elsevier.com)

Time To Be Creative With Adaptive Trials May Be Non-Pivotal Portion

More from Archive

More from Pink Sheet