April 16, 2004 | Many people don't realize the agency has the world's largest repository of in vitro and animal results that are linked with actual human outcomes data. The truth is that the FDA vaults are bulging with information about the mysterious and treacherous paths to blockbuster and budgetbuster results. The problem is this vast experience mostly informs what the FDA does — but not many others.
Now, at the prodding of outgoing commissioner Mark McClellan, the FDA wants to change its miserly ways — how is not yet clear — and share its accumulated wisdom as part of a focused attack on costly, unpredictable product development. At a hastily arranged press conference on the eve of St. Patrick's Day, McClellan and other top FDA brass released a major initiative they're calling "Innovation or Stagnation: Challenge and Opportunity on the Critical Path to New Medical Products."
"Agency reviewers see the successes and associated best practices as well as the failures, slowdowns, barriers, and missed opportunities that occur during the course of product development. [D]ata on product testing, safety evaluation, and clinical trials are stored in the millions of pages of FDA files," the report states.
Imagine combing through all those data to tease out best practices or to produce improved predictive safety models. Both are tantalizing prospects put forward by the FDA.
|Imagine combing through all those data to tease out best practices or to produce improved predictive safety models. Both are tantalizing prospects put forward by the FDA.
Doing basic science isn't the problem, the agency insists. Rather, it's the general failure to turn new insights and past experiences into predictive development tools that is problematic. The report also takes a swipe at traditional animal models, which it characterizes as poorly understood and often irrelevant to compound response in humans.
So what's the solution?
"A new product development toolkit — containing powerful new scientific and technical methods such as animal or computer-based predictive models, biomarkers for safety and effectiveness, and new clinical evaluation techniques — is urgently needed to improve predictability and efficiency along the critical path from laboratory concept to commercial product," the FDA says.
That certainly would help, even if the big improvement is simply learning how to fail bad compounds sooner. The FDA estimates that a 10-percent improvement in predicting failures before clinical trials could save $100 million in development costs per drug.
The devil, of course, is in the details. Right now, the agency is a little fuzzy on how it will tackle tool development and best practices determination. A collaboration with industry that resembles the FDA rule-setting process is likely. But without a champion relentlessly pushing the initiative, it could quickly stall. It's easy to imagine the economist side of McClellan chomping at the bit to turn all those FDA data into efficiencies. Now that he's moved on, it's not clear who at the agency shares his zeal. A project like this needs a big, squeaky wheel to push it forward.
Difficulties aside, the new focus on tools is refreshing. They can help. Toxicity problems, for example, kill many drug candidates rather late in development. The report cites a claim by one pharmaceutical company that clinical failures based on liver toxicity cost it more than $2 billion in the past decade. The FDA suggests it may be possible to mine its clinical data to construct models for early toxicity screening.
Culling out best practices from the mountain of applications the FDA has processed is also an appealing, if difficult, proposition. Bio·IT World attempts to do this on a far more modest scale with our annual Best Practices Awards program. (The deadline for entries in the 2004 competition is May 7. Please contact me at firstname.lastname@example.org for additional information or guidance on filing a late entry. Winners will be announced in a June ceremony at the National Press Club in Washington, D.C.) It's amazing what can be learned from others' successes and failures, although dealing with proprietary information will be a challenge.