Boosting Adaptive Clinical Trials

Best Practices Winner: Wyeth
Project: Adaptive Design Explorer
Category: Clinical Trial Design
Nominated by: Tessella

By Deb Borfitz

July 23, 2009 | Wyeth Research and partners are developing trial simulation software to streamline the process of designing and running adaptive clinical trials (ACTs).

The Adaptive Design Explorer (ADE) is essentially an “educational platform” allowing biostatisticians to quickly and easily identify potential design options and compare their performance, says Michael Krams, VP of adaptive clinical trials and applied program strategies. “The beauty of [ADE] is that it enables simulation-guided clinical trial design, assessing our ability to accurately estimate the dose response and determine the target dose in an efficient manner,” says Krams.

ACTs are inherently complex in that they use accumulating data to decide in a predefined manner how to modify key aspects of a study, including sample size and allocation of patients to different doses, says Vladimir Dragalin, Wyeth’s assistant VP of statistical research and applications in the division of global biostatistics and programming. “Simulating clinical trials allows us to formally compare and contrast the operating characteristics of the different design options.”

One of a Kind
The ADE, now in its third major release, serves as a catalogue of adaptive and non-adaptive designs against which simulations can be run and candidate designs can be assessed on a common platform with a friendly graphical user interface, explains Dragalin. “There’s nothing on the market like it.”

Making comparisons across a “level playing field” helps identify when an adaptive design approach is justifiable and which design is the most appropriate for a particular trial, says Dragalin. Adaptive design programs were heretofore created one at a time to answer a particular research question. The ADE “enables planning and implementing adaptive designs in a much more resource-efficient manner, allowing scalability.”

According to Krams, “response adaptive dose-finding designs” are now routinely considered across all therapeutic areas at Wyeth. The aim is to strike a balance between needed dose response information and increased costs.

Design options within the ADE are based on different what-if scenarios and can be “fine tuned” to reflect user-specified trial characteristics, says Dragalin. Users also have the option of developing an entirely new adaptive design by choosing from a la carte menu items. These include trial stopping rules (for efficacy, harm, or futility), allocation rules (assignment of subjects to available treatment arms), and sampling rules (number of subjects involved in the next stage). “Different designs have one such rule or several.”

Importantly, ADE’s graphical user interface is largely standardized across designs with a common workflow: specifying the nature and target(s) of the trial, the trial execution model and scenarios for simulations; selecting designs and specifying their parameters; running simulations; and evaluating and comparing the designs’ performance over the scenarios.

Development of the “design engine” began in the fall of 2006 as a team effort involving Wyeth’s in-house researchers and ACT specialists at Berry Consultants. The British consultancy Tessella was tasked with the development of the software-user interface. ADE underwent pilot testing about a year later. The ADE includes a database that holds all executed simulations, created designs, and the corresponding results. The database is connected to a grid of computers that perform the simulations.

Ultimately, through an “adaptive execution environment,” the ADE will be communicating with every other information system at Wyeth, says Dan Burns, program manager of applied program strategies at Wyeth. “We are [currently] integrating the IT systems relevant in the conduct of ACTs.” Up to now, ACTs have been run without the benefit of this integrated system, “but we want to enable scalable implementation of adaptive dose-finding studies, running up to 30 ACTs simultaneously.”

Linking drug supply software with the ADE will bring real-time supply chain management capabilities to the implementation of ACTs, “ensuring the right drug and doses arrive at clinical trial sites at the right time,” says Krams. ACTs routinely require many more treatment arms than traditional trials and can thus become the more costly alternative unless the drug supply problem is addressed.

Adaptive designs address the pressing need, referred to in the FDA’s 2006 Critical Path Opportunities Report, to more efficiently develop drugs and learn about dose response, says Dragalin. For Wyeth, ADE is a Critical Path Opportunities enabler. ADE has proven essential to elevating adaptive designs from a being a “small specialist niche” to a technology that can be deployed across a significant proportion of the company’s trials.

Krams says the ROI is potentially huge. “Each design proposal we bring forward is accompanied by a formal business case analysis. Millions of dollars can be saved by stopping trials early and tens of millions of dollars can be banked by avoiding having to do rework. Our ability to accurately determine the correct dose to take forward into confirmatory phase III trials has improved considerably.”

“The results Wyeth achieved in transforming clinical processes is a great case study of what can be accomplished by a focused, dedicated and determined team; and ought to be a wake-up call to the rest of the industry,” said Grant Stephen, CEO Tessella. “It is satisfying to see that their success has been recognized by the BIO-IT judging panel and Tessella is proud to have been part of the program team. If you are serious about improving how clinical trials are done, Wyeth’s comprehensive approach to adaptive trials demands careful study.”