What Ever Happened to Critical Path

Call it a case of bad news, then good news, then—distressingly—no news at all. In March 2004, FDA issued a white paper entitled "Innovation or Stagnation: Challenge and Opportunity on the Critical Path to New Product Development." It painted a picture of a drug-development system that was breaking down.

Janet Woodcock, FDA's deputy commissioner for operations

Not only had the arrival of genomics not created its expected bounty of new drugs, but there were signs that things were getting worse. Spending by both the industry and NIH had more than doubled since 1993, but license applications for new drugs and biologics were down worldwide. One might have expected that the cost of discovery was driving the change—but in fact, almost all of the growth in the expense of developing a new drug came later, in the clinical phases of development that FDA labeled "the critical path."

Douglas Throckmorton, deputy director of the Center for Drug Evaluation and Research

Worst of all, there were no signs that the industry and the agency were benefiting from their experience. "A new medicinal compound entering Phase 1 testing, often representing the culmination of upwards of a decade of preclinical screening and evaluation, is estimated to have only an eight percent chance of reaching the market," wrote the paper's authors. "This reflects a worsening outlook from the historical success rate of about 14 percent. In other words, a drug entering Phase 1 trials in 2000 was not more likely to reach the market than one entering Phase 1 trials in 1985."

Lawrence Lesko, director of CDER's Office of Clinical Pharmacology and Biopharmaceutics.

That was the bad news. The good was that FDA planned an ambitious Critical Path Initiative to improve and accelerate drug development by partnering with companies and academic institutions to discover and validate biomarkers, create innovative protocols for clinical trials, and invent models and predictive tools—in short, to help create a robust science of drug development.

The white paper was written with a sense of urgency, and issued with fanfare. A list of research priorities was promised within a matter of months. And then . . .

And then, seemingly nothing. The "research opportunities" list never appeared, and Critical Path seemed to drop off the radar as public concern veered in the direction of drug safety, and politicians railed about the "too-cozy" relationship between FDA and the industry. As recently as this fall, Critical Path looked dead or close to it.

But is it? The past few months have seen a flurry of announcements about projects related to Critical Path, and at several meetings Acting Commissioner Andrew von Eschenbach called the initiative one of his highest priorities. At press time, FDA said the long-awaited research opportunity list was (again) just days away from release.

The long silence, explains Deputy Commissioner for Operations Janet Woodcock, was not a period of inactivity, but a time of serious conversations with stakeholders about sensitive issues, especially the problem of sharing data and scientific insight while preserving intellectual property. "People shouldn't underestimate the difficulties of creating new models," she says. "It involves a lot of lawyers and signing papers and agreements and working your way through questions: how would we do this, and who would benefit if we discover something, and so forth. A lot of things that are worth doing can't happen quickly."

And some can't happen at all. With the agency's reputation at a low point, this is either a very good or a very bad time to go before the public with Critical Path. In the current political climate, "cooperative research" is likely to sound like "more cozy deals," and "more efficient" R&D will be heard as "fast and sloppy." But the alternative to taking a risk is the near certainty of seeing drug development flounder, with companies unable to take advantage of new scientific developments because the agency just hasn't kept up.

"We have the tools to solve some of these problems," says Woodcock. "We need to get a move on it."

A Science of Development

To explain how drug companies came to their predicament, Woodcock focuses on a key difference between pharma and other industries: "If you look at electronics, for example, it's standard to have certain basic understandings in the platform from which they make their innovations," she says. "They don't keep everything secret. They only keep secret the part that's their innovation. That hasn't happened in biomedicine. Innovations have been kept locked up, and there's no research infrastructure."

There are several reasons for this: Until recently, the biological knowledge to truly explore disease pathways and biomarkers wasn't available. Then, as bioscience exploded in the 1980s and 1990s, companies—and even academics—tended to treat many discoveries as intellectual property to be exploited, rather than as science to be shared. The regulatory process, as FDA concedes, has sometimes discouraged companies from innovating in R&D and manufacturing.

But there are other reasons, says Woodcock. One is rooted in the culture of basic biological research. "When I talk to some basic research scientists, they say, 'Well, first we have to totally understand the body, molecular level, genetic level, cellular level, organ system level, the whole organism level, and the disease level.' I don't think they get it: You could do theoretical physics and understand quantum electrodynamics and still have no Internet or plasma television. You need a whole spectrum."

Most important, though, until the past few years, the industry was too successful to seriously contemplate change. "There would have been no need to talk about this ten years ago," says Woodcock. "It would have fallen on deaf ears if we tried." Today, with pipelines weak and major patents expiring, companies are more willing to listen.

FDA is getting visible support from its sister agencies, which also advocate a more informed and efficient drug development process. At NIH, director Elias Zerhouni also is under pressure to demonstrate that the billions the agency spends on basic and clinical research will pay off in breakthrough treatments for critical diseases. This has generated support for "translational research" under the NIH Roadmap process.

The connection to translational research—the science of moving new discoveries from the laboratory into clinical development—is especially important, since FDA's aim is to create a similar science devoted to the motion of drugs from preclinical development to the patient.

Meanwhile, the Centers for Medicare and Medicaid Services (CMS), headed by former FDA commissioner Mark McClellan, is advocating increased collaboration with FDA to ensure that new drugs are safe and effective, and that they also provide evidence of value to justify coverage by Medicare and other programs. McClellan supported the original Critical Path report when he ran FDA; he considers it vital for the agency to take the lead in spurring new approaches to drug development.

The Critical Path report and opportunities list aim to clarify FDA's involvement in issues that on the surface are problems for industry and academia to tackle. FDA officials believe that the agency has a responsibility to examine whether regulatory requirements inhibit drug development.

The main problem, explains Woodcock, is that FDA has to use the science that's available, now 30 years old. She adamantly wants a better tool kit to reduce the uncertainty of the drug approval process.

"We're beat up for not approving things that ultimately prove to be beneficial because we're uncertain about the benefits," she says. "We're beat up if we approve something and then a subgroup turns out to be harmed by it. It's a no-win situation for the regulators, because there are always going to be people who benefit and people who are harmed."

She explains that the industry and the agency need to move from a population-based model, in which drugs are tested on broad pools of patients, to a more targeted approach in which clinical development focuses on patients most likely to benefit from a drug. But to do that will require new libraries of biomarkers, new models for clinical trials, and other tools—all of which Critical Path hopes to develop. FDA does not fund research, Woodcock acknowledges, but it has a responsibility to improve its rules and evaluation processes.

The potential payoff? A solution to falling R&D productivity, new tools for improving safety, and in the long run, a chance to make drug development more predictable and manageable. "I think we bring a knowledge of where the issues are." says Douglas Throckmorton, deputy director of FDA's Center for Drug Evaluation and Research (CDER), "because we sit at the intersection between basic science and product development and marketing. "We know where the critical lesions are, if you will—the things that if made better could materially affect drug development. We also bring the regulatory piece. At the end of the day, we can say, 'Yes, this is acceptable. This biomarker is useful in this setting. We'll use it as a surrogate.' That clarity, of course, is terribly important for sponsors.

"What we can't bring are monetary resources and personnel resources. So what we can offer is to work with the people that can do those things—clearly identify the tasks, the tests, and the research, so at the end of the day, we, in fact, can make that regulatory decision."

Science Plus Process

"For me," says Lawrence Lesko, director of CDER's Office of Clinical Pharmacology and Biopharmaceutics, "Critical Path is two broad things. One is the science—the innovations that would improve productivity and success. That's what most people have focused on. But it's just as important to focus on the process by which FDA interacts with industry and vice versa, because you could have all the innovative science you want, but if you don't have venues to have conversations about that science, I believe it would be business as usual."

As an example of what he means, Lesko points to FDA's Voluntary Genomic Data Submission program (VGDS), which began in 2004. Created to provide an avenue for the use of innovative science, the program offers companies a safe harbor to discuss genomic data with the agency without fear of repercussions in the review phase.

Companies apply to participate in VGDS, and FDA selects the applications most likely to yield new insight. The company submits a packet of data—perhaps a validation package about a genomic expression, or data related to the selection of a biomarker that the company thinks might be associated with an adverse event. FDA reviews the data, and meets with the company to discuss how to move forward.

"Because it's exploratory data, we're free to explore all possibilities," says Lesko. "The strength of the meeting is the ability to dialogue in a way that everyone knows is nonbinding." FDA has conducted about 30 VGDS meetings involving a dozen companies in the past year—about as many as the agency can handle, Lesko says.

Working with individual companies, however, is much easier than collecting data from multiple companies or working with consortia. FDA has projects under way that will help it gain experience in sharing selected data while maintaining the intellectual property of the companies that produced it.

For example, the agency has established a Cooperative Research and Development Agreement (CRADA) with the IT firm Mortara Instrument to operate a databank of electrocardiogram (ECG) data from pharma companies. In the past, sponsors provided FDA summary analyses of ECG data. Now the need to evaluate specific drug-induced cardiac toxicity evidence, has prompted FDA to seek more specific ECG waveform data. Norman Stockbridge, director of FDA's division of cardiovascular and renal products, explained the technical logistics for submitting such data at a workshop held in September.

The real value of the data will come when FDA is able to share them with outside researchers. For example, the Duke Clinical Research Institute seeks to establish a collaborative effort to gain access to the ECG warehouse to study drug effect on QT prolongation and other issues. The aim is to address the growing debate over whether QT data really reflects risk for cardiac adverse effects, such as torsades de pointes. This proposal has been the subject of intense discussion in recent months, as pharma companies have expressed concerns about sharing ECG information with anyone outside FDA. FDA wants eventually to link ECG data to patient outcomes and PG/PD data from QT trials.

A meeting sponsored by Duke in October revealed the difficulties with collaborative efforts that involve pharma companies sharing data bases linked to specific drugs, particularly in the cardiovascular area. Companies appear reluctant to share even data from placebo studies. FDA would like to know why different drugs have different effects and why a small percentage of patients have problems in this area.

"Whoever owns the data has to have an opportunity to look at the project that's going to be done with the data and say, 'You're appropriately respective of proprietary and patient confidentiality issues,'" explains Throckmorton. "We've done that in the past. That's been a pretty standard thing."

In Search of Markers

One central Critical Path theme is to build a scientific framework for using more biomarkers in product development. Better biomarkers can more precisely identify animal toxicities and early screens for human toxicities, but FDA recognizes that it is necessary to develop new models for biomarker qualification.

Pharmaceutical companies regularly develop biomarkers for their own internal use in clinical evaluation, but they do not share this knowledge with competitors. FDA sees broader use of biomarkers as key to streamlining clinical trials, but recognizes that joint approaches are needed to avoid duplicative efforts and reduce the cost of validating biomarkers.

At a recent advisory committee meeting, Larry Lesko noted that a few validated biomarkers exist, namely those for anti-coagulants, viral load and blood glucose. Biomarkers are key to model-based drug development, enhancing drug safety by avoiding adverse drug events, managing drug risk, and making product labels more informative as to appropriate dose.

New biomarkers also may play a role in retrospective drug rescue scenarios. For example, by identifying the patient population likely to benefit from a drug without suffering from safety problems, a manufacturer might be able to return a troubled product, such as Vioxx (rofecoxib), to the market.

To clarify what needs to be done to validate new biomarkers, FDA seeks to identify endpoints already in use. FDA issued draft guidance for cancer trials in April 2005 on what endpoints besides survival or irreversible morbidity can be used—tumor size/shrinkage, symptom relief, quality of life—and how to measure them. The agency plans to follow this general guidance with a series of guidances for specific cancer types.

To obtain a broader picture of what biomarkers are being used by the research community, FDA is surveying its medical officers about biomarkers that have been filed in applications over the years and accepted as surrogate endpoints by reviewers. Remarkably, there was no central repository of the information; indeed, when the survey began, the agency learned that there was not even substantial agreement about the definition of "surrogate endpoint."

Another major Critical Path effort involves using models and new approaches to make the clinical research process more efficient and less costly. A major symptom of current problems is a continued 50 percent failure rate of Phase III clinical trials. A recent McKinsey & Co. analysis of almost 300 Phase III failures at major pharma companies revealed that half failed to show efficacy compared to placebo. McKinsey analyst Rodney Zemmel says that only five percent should fail in Phase III for that reason. The data suggest that something is seriously wrong with earlier stages of development.

Thus, many FDA initiatives support strategies to provide more information about a drug early in the development process so that later trials are more likely to show efficacy and safety. For example, FDA issued a draft guidance on exploratory investigational new drug (IND) studies in April 2005 encouraging "early into man" Phase 1 trials. Using this approach, sponsors give very small doses to small patient populations to determine early on if a drug shows any effect.

Model-based drug development is a continuing theme among clinical research statisticians who are developing methods to use Bayesian statistical models to better predict clinical trial outcomes. This involves developing models of disease as influenced by a drug to establish the proper dose response needed to show efficacy, explains CDER scientific advisor Donald Stanski. Researchers can simulate early trials and then obtain feedback from FDA that can help define uncertainties and safety concerns.

Getting the dose right early, Stanski explains, can support decisions to fail a drug early—but with confidence—and thus reduce Phase III failures.

Robert Temple, director of CDER's office of medical policy, has long been an advocate of clinical trial enrichment approaches to improve the research process. This involves selecting clinical trial participants who are more likely to respond to a drug, or less likely to experience safety problems. This approach may indicate safety and efficacy more quickly and efficiently, although it may result in narrower labeling.

Perhaps the most visible sign of FDA's interest in improving performance in Phase III can be seen in the agency's new ideas about how to respond to Phase II research.

Traditionally, the agency meets with companies after Phase II is complete to discuss the data and the company's strategies for moving forward. But by that time, argues Lesko, it is hard to have much impact on Phase III. "It's too late," he says. "The trial is already designed, the sites are lined up, the dosage forms are made."

To correct the situation, the agency has created a Critical Path initiative to hold informal end-of-Phase IIa meetings—at which FDA uses disease state computer models, virtual clinical trials, in-house data, and other tools to dig deep into the question of how companies should proceed. The meetings are voluntary—at least for now; a draft concept paper was issued several years ago, and a draft guidance is in the works.

Promoting Public–Private Partnerships

Because FDA lacks resources to fund research, the agency seeks to spur examination of Critical Path opportunities through joint projects with pharma companies and research organizations, either individually or in groups.

Much of the work is expected to take place under CRADAs. In a typical CRADA, a manufacturer or group of companies supplies the funding and resources, including data on drug characteristics, research results, or innovative manufacturing approaches. FDA offers advice on study designs and agrees to evaluate results and use them to support a regulatory test or standard. Academic or nonprofit organizations offer research facilities and a neutral party to manage funding and protect proprietary information.

Establishing consortia takes a long time, Woodcock says. To identify and qualify a new biomarker, for example, may involve evaluating assays from several partners, each anxious to guard its intellectual property. Members of a consortium get an early look at how it is performing. But the goal, says Woodcock, is to improve FDA standards and develop predictive evaluation tools. "That's what FDA gets out of this," she says.

FDA plans to work with the consortia being established to develop new predictive tools. If the science is done right, says Woodcock, the tools will be incorporated into FDA standards, as was done in the past for stability data and animal toxicology tests.

An early example of such efforts is a collaboration with the National Cancer Institute (NCI), which was launched in 2003 by McClellan and von Eschenbach, when he headed NCI. Now, as the head of FDA, von Eschenbach is promoting expanded FDA collaboration with NIH entities.

As a first step, both FDA and NCI have assigned scientists at both agencies to temporarily swap positions so they become better acquainted with the opportunities and challenges of their colleagues. The group also is working on clinical trial automation and development of standard clinical trial data reporting forms to permit electronic submission of NCI trial data to FDA. NCI recently said it would use the e-signature model developed by pharma for cancer trials.

Anotehr exciting NCI initiative is to examines the use of imaging technologies to monitor the impact of therapies on cancer tumors in drug development as well as treatment. An NCI group is examining how FDG-PET scans could be used in very early clinical trials to help screen molecules for action in humans. This data could then be used to identify possible surrogate endpoints.

FDA Launches Collaborations

Several joint research initiatives have been announced in recent months:

Liver toxicology biomarker FDA announced in November that its National Center for Toxicological Research (NCTR) and Massachusetts-based BG Medicine have signed a CRADA for a project that will determine whether a standard test used in early drug development can discover early signs of human liver toxicity, one of the main causes for drug development failure. The study uses data from pharma companies, which, in turn, gain access to project data and the opportunity to license discovered biomarkers.

Manufacturing survey CDER's Office of Pharmaceutical Science is establishing a CRADA with Conformia Software of California to conduct a confidential survey of industry R&D and regulatory personnel on the causes of drug development bottlenecks. Conformia will seek information from 20 manufacturers on development information problems, pilot plan information management, pre-clinical development challenges, and manufacturing science. The goal is to gain a clearer picture of current industry development practices and problems.

Simulation software The Office of Biostatistics is collaborating with PharSight Corporation to develop software that will emulate clinical trial outcomes based on pharmacodynamic data.

C-Path Institute, a non-profit organization founded by Ray Woosley with support from FDA, SRI International, and the University of Arizona, has launched a Toxicogenomic Cross Validation Consortium to bring together pharma companies to validate each other's safety test methods. Six leading firms are participating, and FDA hopes to use the results to develop guidance on applying new technology for pre-clinical safety screening. C-Path also partners with an Arizona pharmacy chain to create a public network for reporting adverse events, which aims to help FDA improve its drug safety monitoring system.

Novartis is working on three research collaborations with FDA. One examines how to apply process analytical technology (PAT) to identify new methods for assuring quality manufacturing. FDA and Novartis are negotiating a CRADA calling for joint evaluation of research findings, which will give FDA experience implementing and validating proposals under its PAT guidance.

Another project involves a CRADA to develop preclinical biomarkers that evaluate renal toxicity. And a third collaboration will address hurdles related to developing a diagnostic kit with a drug product. Novartis plans to conduct an observational study involving several compounds in its pipeline where a genomic diagnostic may be of value.

National Institute for Pharmaceutical Technology & Education is a non-profit consortium formed by 11 universities, headed by Purdue, to address manufacturing and scientific issues affecting variability in drug development. In June, NIPTE signed an agreement with FDA to work jointly with manufacturers and FDA on research and educational projects to improve pharmaceutical manufacturing science and technologies. FDA will provide scientific expertise to NIPTE and consider its recommendations in developing new policies and guidances. The group seeks $25 million a year for five years from Congress and other government agencies to establish Centers of Excellence that will launch specific projects to reduce the high cost of drug development.

Moving Forward

Critical Path faces many hurdles. It's not clear how the concept will play on Capitol Hill: Will legislators zero in on the element of industry-agency collaboration and resist it, or will they perceive it (as FDA does) as a key to improving drug safety? Many Critical Path initiatives are funded with money from industry or other outside sources, but there will be a high resource burden on the agency—especially in the form of post–Phase IIa meetings and analyses of voluntary genomic submissions. And companies need to buy into the idea that sharing science is more productive than hoarding it.

If Critical Path succeeds, pharma gains something it has never had: predictability. "You can design an airplane on a computer and the airplane flies," says Woodcock. "You can design a skyscraper, a bridge, or whatever, and there are mechanistic predictive tools to evaluate materials and design. The pharmaceutical industry does not have ways of doing this. I would say [industry has] been in an existential despair because the science wasn't there. But it's over. We need to focus on applying the science for drug development in a way that's going to work."