YouTube Facebook LinkedIn Google+ Twitter Xinginstagram rss  

Ten Years After: Learn and Confirm

By Kevin Davies

Feb. 12, 2007 | "We’re in good shape, so what better time to change?”  Wyeth VP of project management, Charles Gombar, is a refreshingly engaging and frank big pharma executive, and excited about Wyeth’s new approach to drug development. “You don’t want to take swimming lessons when the ship’s going down. It gave us the opportunity to carefully think through the process; we weren’t under the gun,” he says.

Wyeth faces the same problems bedeviling the industry: the soaring cost of development is matched only by the rising attrition rate as once promising drug candidates fall by the wayside. So Wyeth undertook a major transformation of its infrastructure and corporate culture called Springboard, aided by the consulting firm McKinsey, spanning commercial, manufacturing, corporate infrastructure, human resources, and R&D. Part of that initiative was the identification of the “Learn and Confirm” strategy to drug development.

“We’re not looking for incremental gains but much more fundamental change,” says Gombar, who together with Evan Loh, VP clinical research and development, spearheaded the effort within R&D. The challenge was: “How can we do clinical development much more efficiently than now? We weren’t given cost targets, or asked to shave X millions of dollars, or X months or years. We were given carte blanche.”

Gombar and colleagues brainstormed about “why we do things and how we do it.” Gombar believes drug development has evolved into a Rude Golderg apparatus. “We like to point fingers at the regulators and say it’s all their fault, and certainly regulations have changed over the years, but when you start peeling the onion, a lot [of the problems] we do to ourselves.”

Wyeth borrowed the “Learn & Confirm” terminology from a paper published by Lewis Sheiner (see Box). The FDA started using the terminology and it struck a chord at Wyeth. “Sheiner’s paper basically says that we have to get back to science-driven development. That’s what Learn & Confirm is all about,” says Gombar. Other big pharma organizations,  notably Pfizer, have referred to Sheiner’s concept in specific projects, but none has embraced the concept as wholeheartedly as Wyeth.

“For example, we treat every clinical story as if it’s a pivotal trial to support registration — but really the only pivotal trials are confirmatory trials in traditional phase III. If that’s the case, then why don’t we look at [early-stage] data more than we do? Why be blinded to this extent?”

Gombar says he saw lots of raised eyebrows early on, particularly skepticism over the reaction of the regulators. “So we asked the FDA, and they said, ‘Is it a pivotal trial?’ If the answer is no, they said, ‘We encourage you to learn from the data and make changes to your development program.’”

Wyeth is embracing adaptive trials. “Its absurd [in phase I that] we keep ourselves blinded to this stuff. You don’t need to be triple blinded.”

Gombar candidly shares one painful example of a drug in a “dreaded potentially pivotal” phase II trial: “The trial ended, and the drug didn’t work. In doing the post mortem analysis, we could have known based on biomarkers that this drug didn’t have a chance. We could have saved a helluva lot of money and resources and that information could have been fed back 1-2 years earlier than it was.”

Liberal Thinking
Wyeth’s entire development effort has now been reorganized around Sheiner’s decade-old concept. “We’re going to start liberating the thinking, especially during early clinical development. We must get away from phase I-II-III terminology; it’s had its day. People know what it means, but it’s led to a much more cookie cutter approach. But this is science — there shouldn’t be a cookie-cutter approach.”

While companies have been organized around therapeutic areas, “the difference here is that a therapeutic area might be too broad... We want people to focus down more, so we have an Alzheimer’s disease ‘Learn’ team — they have a portfolio for compounds they’re responsible for.”

Gombar believes it will pay dividends having true disease experts get a better understanding of specific diseases not just for discovery but also for development efforts. Previously, he notes, “We’d have a new molecule, so we’d form a new team and ask what is this disease? That’s not the right way to do this.” Now, Wyeth hopes to benefit from cross learning from one molecule to another. “We [can] put more than one investigational molecule into a clinical trial [and ask] are they synergistic?”

Sheiner advocated a change in thinking that requires radical changes in process. The industry didn’t embrace it whole hog the way we’re trying to embrace it now,” Gombar says. A key component is organizational discipline, which Sheiner noted would take courage because it takes time. “For an individual molecule,” Gombar notes, “it may take a bit longer, but you’ll get a richer data set.”

Gombar admits this was a concern, as “teams were getting hung up in ‘Learn’ forever, without spitting anything out into the ‘Confirm’ phase.” Many initiatives are aimed at increasing the ability for scientists to Learn per unit time. “A great example was early clinical development centers. We took the concept of a phase I unit, and said, ‘Why don’t we have phase II units?’”

Wyeth is establishing phase II centers in India, China, Brazil, Eastern and Western Europe, and eventually the United States to rapidly perform phase II-like studies, and “get this ‘Learn’ information before going into ‘Confirm’ studies. This isn’t rocket science. It’s discipline — understanding a dose-response relationship before going into Phase III.”

Gombar says the industry has not been diligent in applying technology in the development arena. Take remote data capture — “How many companies are still using paper report forms?” EDC is where Wyeth wants to go, he says. “That’s where our ultimate goal is — the paperless clinical trail. But the world wasn’t quite ready, he says, because of insufficient standardization. Two other areas Gombar highlights are translational medicine and adaptive clinical trials. “It’s time to start doing these things. We’re making a huge effort into adaptive trial designs, and really embracing translational medicine.”

To roll out Learn and Confirm, Wyeth formed a full-time implementation team, allied to a huge communication effort. “We keep taking the pulse, doing surveys, how is it being received? To my surprise, it’s actually being embraced very well. A typical reaction was, ‘Jeez, about time...’”

Sidebar: Shine On
The late UCSF pharmacology professor Lewis Sheiner was a world-renowned expert on simulating models to measure the optimal drug dose and individual patient response. He is widely recognized as the pioneer of pharmacometrics, applying his expertise in pharmacokinetic-pharmacodynamic modeling and simulation to improve the efficiency of drug testing.

Sheiner co-developed the NONMEM (nonlinear mixed effects modeling) software package with Stuart Beal. It’s a Bayesian regression program that can predict the pharmacokinetics of a drug in a patient. NONMEM became the standard for the industry and the FDA. Sheiner also developed widely used computational tools for the analysis of pharmacodynamic data.

Later in his career, Sheiner sought to improve drug development by optimizing clinical trial designs. Sheiner believed in making clinical trials more efficient and informative, optimizing both dosage recommendations and individual therapy. In 1997 he published an important paper* proposing the “learn and confirm” paradigm of drug development, in which he wrote:

“The understandable focus of commercial drug development on confirmation, as this immediately precedes and justifies regulatory approval, has led, in my view, to a parallel intellectual focus that slights learning. The predictable result... is that clinical drug development is often inefficient and inadequate.”

Confirmation, Sheiner argued, must answer a single yes/no question: does the drug offer net benefit. But learning must address “an essentially infinite set of quantitative questions concerning the functional relationship between prognostic variables, dosage, and outcome.” 

“The learn-confirm view,” Sheiner wrote, “suggests that the intellectual focus for clinical drug development should be on understanding ... It will require not only new tools (e.g. computer software for the design and analysis of scientific studies), but a radical change in the structure of pharmaceutical preclinical and clinical research and development units: A reorientation of thinking cannot be accomplished without a reorientation of process.”

Further Reading:
*Sheiner, L.B. “Learning versus confirming in clinical drug development.” Clin Pharmacol Ther. 275-291 (1997)

Subscribe to Bio-IT World  magazine.

Click here to login and leave a comment.  


Add Comment

Text Only 2000 character limit

Page 1 of 1

For reprints and/or copyright permission, please contact Angela Parsons, 781.972.5467.