© 2004 by Oxford University Press
© 2004 Oxford University Press
NEWS |
Can Disease-Free Survival Be a Surrogate for Overall Survival for Drug Approvals?
The gold standard test for clinical benefit of oncology drugs is improvement in overall survival. The U.S. Food and Drug Administration allows drugs to be approved if they improve patient survival or if there is evidence that reliably predicts clinical benefit, such as an established surrogate.
In a recent analysis of the 57 oncology drug applications approved in the last 13 years through the regular approval process (in contrast to the accelerated approval process), two-thirds of them were approved based on endpoints other than survival, underscoring the importance of establishing reliable surrogate endpoints. This is particularly true in colorectal cancer, because there are an increasing number of agents being developed for the disease, noted Richard Pazdur, M.D., director of the FDA’s Division of Oncology.
A recent review of data from more than 10,000 patients suggested that 3-year disease-free survival can be used as an appropriate endpoint to replace 5-year overall survival in adjuvant colorectal cancer trials. This would have substantial advantages. "It would allow more rapid completion of trials, more rapid reporting of trials, and more rapid introduction of new agents into community practice," said Daniel J Sargent, Ph.D., assistant professor of biostatistics, Mayo College of Medicine, Rochester, Minn., and principal author of the study that reviewed the results from 12 large randomized clinical trials on patients with Duke’s D-2 and C colon cancer conducted from 1978 to 1993 that involved oncology groups in the United States, Canada, France, and Italy.
Sargent presented the review at a fall workshop held by the FDA and the American Society of Clinical Oncology in Washington, D.C. The FDA is sponsoring a series of meetings on surrogate endpoints for cancer drugs, with a focus on lung cancer and colorectal cancer. The FDA has taken the position that disease-free survival can provide sufficient evidence for approval where the majority of recurrences are symptomatic.
In Sargent’s review, there were a total of 38 treatment arms. All of the trials from 1978 to 1990 had a control arm that received no treatment. Three studies done in the 1990s had an active-treatment control arm.
"We did a very simple thing," Sargent said. "We looked at the 3-year disease-free survival and the 5-year overall survival for each arm of each trial to see how they rated." The investigators plotted the events graphically, and the event rates between the two groups were virtually identical. "Graphically this looks pretty good," Sargent said.
Subjecting the data to statistical analysis, Sargent noted that the difference between 5-year overall survival and 3-year disease-free survival was less than three absolute percentage points. "In other words, 3-year disease-free survival equals 5-year overall survival," he said.
Sargent stressed that this review is preliminary and they intend to do more analyses. "Once we have individual patient data, we will be able to do more formal analyses," he said. "The patients who were admitted on day one may have 5 years’ worth of data by the time the last admission had 3 years’ worth of data, so we may have more than 3-year data on some patients. Then there may be other endpoints, maybe 3 years isn’t the right time, maybe 2 years is early enough, maybe you need to wait for 4 years, maybe overall survival at 3 years is a surrogate and you don’t need to worry about disease-free survival. We want to look at other endpoints as well."
Commenting on Sargent’s report, Thomas R. Fleming, Ph.D., professor and chair of the Department of Biostatistics at the University of Washington in Seattle, and a member of the panel said: "This is getting exactly at the issue of the difference between a correlate and a valid surrogate. It may well be that there is a direct relationship within treatment arms, in this case time to recurrence and the overall level of survival. So you, in essence, have validated that this is a correlate. But it doesn’t mean that it’s a valid surrogate."
Fleming pointed out that there may be other mechanisms of action that could affect the treatment group. The intervention may be able to delay recurrence of disease and so achieve a survival benefit. However, that does not address the fact that the intervention may, through an entirely different mechanism, have a negative effect on survival. "This is just a first step. It’s a beautiful correlation but that may not affect the clinical endpoint. You have ultimately got to look at whether or not you see a perfect relationship between the effect of the marker and the effect on the clinical endpoint."
More information on the FDA’s project on cancer drug approval endpoints is available at http://www.fda.gov/cder/drug/cancer_endpoints/default.htm.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||