When Prolonged PFS Is Not Enough
From Cancer Discovery, May 23, 2013
Progression-free survival (PFS) is well-established as a potential endpoint in clinical trials, but judging its predictive ability for cancer drug approvals can raise some tricky questions.
Such questions, along with concerns about the design of a pivotal phase III trial, were center stage in April when the U.S. Food and Drug Administration's (FDA) Oncologic Drug Advisory Committee (ODAC) recommended against approving the VEGF inhibitor tivozanib (Aveo Oncology) for the treatment of renal cell carcinoma (RCC).
The trial of tivozanib, performed primarily with patients in Eastern Europe, revealed that the drug provided a statistically significant improvement in PFS. Patients who received tivozanib had an average PFS of 11.9 months, compared with 9.1 months among control patients treated with sorafenib (Nexavar; Onyx Pharmaceuticals), which, along with sunitinib (Sutent; Pfizer), is 1 of 2 standard therapies for RCC.
However, overall survival (OS) showed the opposite trend: patients in the control arm lived an average of 29.3 months, compared with 28.8 months among patients in the tivozanib arm.
Based on those results, and on questions about how well treatment protocols for the patients mirrored standard practice in the United States, ODAC recommended rejecting the drug by a vote of 13-1. The FDA itself is generally expected to follow suit, especially since agency officials provided a harsh analysis of the trial data at the hearing.
The worrisome OS data simplified ODAC's decision, but the application raised familiar questions about when PFS findings should justify drug approvals, as they did for sunitinib and a number of other drugs approved in recent years.
Regulators “have good reason to be skeptical about PFS,” says Mario Eisenberger, MD, a professor of oncology and urology at the Johns Hopkins Sidney Kimmel Comprehensive Cancer Center in Baltimore, MD. “There's precedence now showing that drug approvals based on PFS have had to be recalled later,” he says, such as the withdrawal of bevacizumab (Avastin; Genentech) in metastatic breast cancer.
Among the potential limitations of PFS data, its measurements not only are of shorter duration than OS, but also may vary depending on the specific schedule of tumor assessments, notes a May 2013 review in Clinical Cancer Research. Issues also can crop up in how the tumor images are acquired and how tumor progression is assessed, particularly since assessment may be done locally or by a central review group. Additionally, data censoring (for example, removing data for patients who drop out of a study) can be more prevalent on one arm of the trial than the other, potentially distorting PFS results.
Such factors can obscure the connection between treatment-related PFS improvements and extended lives, says Susan Bates, MD, a senior investigator at the National Cancer Institute's Center for Cancer Research. For that reason, regulators are looking for a substantial difference in PFS—for instance, a doubling of the measure compared to controls—that might better predict a clear OS advantage, she says.
Regulators also look for trial protocols that match those in the United States, emphasizes Mikkael Sekeres, MD, ODAC committee chair and a staff physician at Ohio's Cleveland Clinic Taussig Cancer Institute.
During the tivozanib trial, patients on the control arm could cross over to the experimental arm as second-line therapy, but not vice-versa. “Crossover should have been allowed on both arms,” as is standard practice in the United States, says Sekeres. “In our view, the treatment rationale was disingenuous. There is nothing wrong with conducting clinical trials in Eastern Europe, or elsewhere, as long as the sponsor ensures the conduct of the study is the same as it would be here.”
©2013 American Association for Cancer Research.
"By all means let's be open-minded, but not so open-minded that our brains drop out." - Richard Dawkins
