Early Interest for Early Stages
The appeal of adaptive trial design is starting to register at attentive drug companies. Eli Lilly has launched three early-stage
trials for oncology and diabetes drugs. BMS is designing an adaptive approach to dosing studies for a migraine compound. And
last year FDA began beating the drum in a big way—particularly as the adaptive approach aligns with its Critical Path initiative
aimed at speeding the development of new treatments. The agency established advisory teams, held workshops, and began drafting
a series of adaptive-trials guidance documents—due out this year—on issues such as evaluating interim data and maintaining
statistical integrity. Still, when it comes to its pivotal Phase III trials, FDA remains cool to the new approach.
The high cost of doing nothing
But what about Phase IV, when comparative real-world effectiveness data can be developed? At that point, FDA has already deemed
the drug safe and effective. A novel, scientifically sound, and cost-effective research design would seem to be just the carrot
the agency needs to get drug companies to make good on their commitment to post-marketing studies. Of course, FDA's stamp
of approval is essential to green-light drug manufacturers' use of adaptive data in their marketing materials.
Missing Evidence and Opportunity
The Centers for Medicare and Medicaid Services (CMS) has been raising the bar on proving real-world comparative value since
2004, when then-CMS head Mark McClellan announced a new Coverage with Evidence Development (CED) policy. In cases where there
is insufficient outcomes evidence to make a national determination for coverage, CED restricts it to patients in clinical
studies—pending better data. The message to pharma is clear: To get on the formulary, get on the evidence-based bus.
For example, in January 2005, CMS invoked its CED policy by requiring additional evidence before deciding about coverage for
four off-label colon cancer drugs, Avastin, Erbitux, Etoxatin, and Camptosar. Sponsored by the National Cancer Institute (NCI),
this series of classically designed trials will meet CMS' demand for real-world data distinguishing the best from the rest.
Although it remains to be seen exactly how much efficiency might have been gained, time, money, and risk of adverse findings
would all have been reduced with a Bayesian approach. NCI may be missing an opportunity to pave the way to late-stage adaptive
trials for a quicker coverage policy.
The price drug manufacturers pay by not producing their own definitive comparative-effectiveness evidence is living with evaluations
made by other sources (see "The High Cost of Doing Nothing"). Recognizing that in these crunch times, companies are not searching
for additional ways to increase R&D costs, one wonders whether, in the face of growing market demand for better real-world
drug-versus-drug information, a 30-to-50 percent increase in efficiency might "tip" manufacturers to begin investing in late-stage
Bayesian trials. It may be.
Bryan R. Luce is the senior vice president for science policy at United BioSource.
He can be reached at firstname.lastname@example.org