Pressure Mounts to Streamline Trials

The escalating cost of drug development is prompting sponsors and researchers to take a serious look at long-discussed but sparsely implemented research methods that promise to reduce the research failure rate and produce more meaningful study results.

Jill Wechsler

A December (2006) report from the Government Accountability Office (GAO) describes the declining productivity of the pharmaceutical research enterprise: Despite big increases in R&D investment, few new therapies are coming to market.

Industry increased R&D spending from $16 billion in 1993 to $40 billion in 2004, and estimates put the number at $60 billion last year. But FDA approved only 22 new molecular entities in 2006. The total has been flat in the last few years, but is "way off the historic highs of the mid-1990s," reports Steven Galson, director of the Center for Drug Evaluation and Research (CDER).

Safety first

This drug development downturn is taking a serious toll on major pharmaceutical companies. Firms are closing plants, cutting work forces, and looking for new ways to conduct R&D. At the same time, FDA and industry are under intense pressure to demonstrate that those drugs on the market are safe and effective.

To this end, and in response to a scathing assessment from the Institute of Medicine (IOM) last year, FDA officials unveiled a plan in January to improve how it detects and assesses drug safety problems.

The FDA report on "The Future of Drug Safety" outlines proposals for developing safety "report cards" on important new drugs and for posting an online drug safety newsletter. FDA is clarifying conflict-of-interest policies for advisory committees and establishing a new advisory committee on communication to gain more outside expertise on conveying risk information.

The agency will formally assess Risk Management Plans (RiskMAPS) to see if they really address serious safety problems. Another strategy is to seek broader access to health system databanks able to detect safety signals more quickly. Galson also promised to address the tension between pre- and postapproval staffs by engineering a "true culture change" at CDER that involves giving more authority to safety analysts.

Support for science

An underlying theme of FDA's drug safety program is that new developments in biomedical science and in information technology offer more efficient ways to address drug safety early in development and to quickly detect safety problems after a drug goes to market.

Janet Woodcock, FDA deputy commissioner and chief medical officer, heads up a range of efforts to improve risk assessment by sponsors and FDA, many launched under the Critical Path Initiative and described more fully in a January report on progress made in implementing Critical Path opportunities during 2006.

Collaborations involving FDA, industry, and academia, for example, are developing new tests for drug organ toxicity and for identifying cardiovascular risks of drugs. The Predictive Safety Consortium is sharing data to cross-validate preclinical safety biomarkers. Another consortium based at the National Institutes of Health aims to assess biomarkers for critical diseases, beginning with lung cancer and lymphoma. A new Serious Adverse Event Consortia plans to identify and validate genetic variants that may help predict the risk of drug-induced reactions, starting with liver disease and serious skin rashes.

Projects to streamline clinical trial design also are moving forward. FDA is collaborating with industry to develop standard formats for case report forms to establish a fully electronic system for submitting all clinical study data to FDA. Another project seeks to reduce the high failure rate of late-phase trials by encouraging sponsors to use clinical trial simulations to help improve dose selection.

FDA expects many of these collaborations will yield new guidances on topics such as identifying liver toxicity during drug development and utilizing enriched clinical trial designs to gain more safety information. The goal is to "build safety into products" by better understanding the genetic basis of adverse events, explained Woodcock. These approaches can screen out toxic compounds early on and identify patients most likely to have adverse events—or to respond well—to a test medicine. The aim, she said, is to "make products safer from the get-go."

Sharing data

A researcher's ability to design better clinical trials and make go/no-go development decisions could benefit from a better understanding of prior knowledge in drug and disease areas. This can be enhanced by sharing clinical trial data held by pharma companies, according to many industry R&D leaders.

The importance of access to information on disease states and biological responses beyond what any one sponsor may have, as well as obstacles to such collaboration, was debated at a January meeting, "Sharing Knowledge to Improve Clinical Drug Development and Regulatory Decisions," co-sponsored by FDA and Drug Information Association (DIA).

John Orloff, vice president of development at Novartis, voiced support for sharing information at a pre-competitive level to establish models of disease progression and qualification. He urged regulatory authorities to further such efforts by supporting alternative statistical approaches; establishing common principles for reliance on modeling assumptions; accepting industry-generated drug and disease models; and facilitating a greater sharing of disease models, epidemiology, and safety data.

The poster child for such cooperation is the 1990s effort to validate HIV viral load as a surrogate for testing new AIDS treatments.

Prior to 1997, FDA required studies with clinical endpoints to approve HIV drugs, pointed out Jeff Murray, deputy director of CDER's Division of Antiviral Products. Prodded by very vocal patient advocates, FDA joined with industry and academia in 1996 to form a Surrogate Marker Working Group to assess correlations between viral load and clinical outcomes through meta-analysis of data from multiple trials. Collaboration was particularly important because AIDS treatment involves multiple drug combination therapies.

Agreement on the results led to FDA guidance and greatly expedited HIV drug development through smaller and shorter trials.

Necessary backing

The challenge is to transfer the HIV model to treatments for less lethal and more established diseases. A key issue is how strongly FDA officials support model-based clinical trial design and the collaborative efforts needed to establish drug and disease "libraries" of research data—the key to further expanding trial models and simulation.

For these methods to gain broader acceptance, Donald Stanski, global head of modeling and simulation at Novartis, said it was important for FDA to champion Model Based Drug Development (MBDD) and to persuade agency reviewers to have confidence in this approach.

FDA is using modeling to assist in NDA and labeling decisions, according to Bob Powell of CDER's Office of Clinical Pharmacology. The combination of modeling prior knowledge (on disease, placebo, dropouts, and drug effect) and clinical trial simulation, he noted, is beginning to be used to justify Phase IIb and Phase III study designs. However, Murray of CDER acknowledged that there is limited acceptance of trial modeling and Bayesian analysis among reviewers making final regulatory decisions. And CDER Deputy Director Douglas Throckmorton noted that FDA sees modeling approaches most often playing a "salvage role" for development programs in trouble.

Industry's attitude

And just how eager industry is to share proprietary data remains to be seen. An informal survey of pharma companies indicates general support for sharing summary-level data on placebo responses and active controls, which FDA already makes public when a new drug is approved. But many companies remain reluctant to release patient-level data from clinical trials.

Most research executives want to see the value of modeling efforts—that such approaches can lead to faster, better, more efficient programs—along with evidence that no one company can build enough models on its own to answer important questions related to drug development. And that kind of evidence is beginning to emerge.

Richard LaLonde, global head of clinical pharmacology at Pfizer, reported that its Enhanced Clinical Trial Design Initiative saved the company $75 million last year and predicted even more benefits in 2007. Although Pfizer is closing facilities and making big cuts in its workforce, LaLonde said it's looking to recruit more statisticians and pharmaco-epidemiologists for this program.

Stanski and others want to spur pre-competitive data sharing through a non-profit consortium that will address data needs and modeling approaches. The idea is to start with a specific disease state; perhaps one that raises an important safety issue involving a pediatric treatment or a serious infectious disease would spur interest and support.

Jill Wechsler is the Washington editor of Applied Clinical Trials, (301) 656-4634 jwechsler@advanstar.com