Progression-free survival is an acceptable basis for accelerated approval of ovarian cancer agents, so long as trials are powered for survival, panelists at FDA's April 26 Ovarian Cancer Endpoint Workshop in Bethesda, Md. said.

"Increase in PFS is a valid endpoint" for first-line ovarian cancer candidates, panel member Stacy Nerenstone (Gray Cancer Center) said. "However, even if you are using PFS, you should size it for overall survival."

However, Nerenstone, a former chair of FDA's Oncologic Drugs Advisory Committee, emphasized that overall survival remains the preferred endpoint for full approval in the first-line setting, especially for maintenance therapies.

"Overall survival is felt to be the significant endpoint, especially because it entails more toxicity," she said. "Improvement in progression-free survival might be acceptable, especially for accelerated approval, especially since you get the data a little bit more quickly and if it is a non-toxic treatment."

Panelist Elizabeth Eisenhauer (National Cancer Institute-Canada Clinical Trials Group) agreed: "Overall survival is the gold standard. Trials should be powered to assess it regardless of what the primary endpoint is."

The workshop was part of a series of meetings sponsored by FDA and the American Society of Clinical Oncology concerning oncology endpoints; PFS has previously been identified as a valid primary endpoint for colorectal, lung and skin cancer trials (¹ Also see "Colon Cancer Adjuvant Trials Need 3-Year Disease-Free Survival – FDA Cmte" - Pink Sheet, 17 May, 2004., p. 41).

FDA Office of Oncology Drug Products Director Richard Pazdur said the agency has previously advised sponsors conducting trials in other settings "to consider progression-free survival or time-to-progression as an endpoint for approval, but to size it for survival."

"If we never size a trial for survival, we are shortcutting oncology because we'll never see a trial that actually demonstrates survival," he explained.

Failing to power for survival will lead to progressively smaller trials and "lo and behold, in the future, even those trials will be underpowered for time-to-progression or progression-free survival," he warned.

One challenge with assessing PFS and overall survival in a single trial in ovarian cancer is the potential for confounding when patients cross over to active therapy after the PFS analysis.

"What I struggle with is if you have an effect on PFS, and that information is known and there is a large amount of crossover then to the treatment arm where you see the effect on PFS, how then can you protect the validity of the" overall survival analysis, OODP Deputy Director Karen Weiss said.

"If you were looking at PFS, you should have measures to protect the validity of overall survival," Weiss said.

Nerenstone agreed that there is a potential for confounding, but she emphasized that "it's very important to look at [survival] subsequently, even if cross-over may interfere with that evaluation."

[Editor's note: For a ² webcast or video/DVD of this meeting, visit FDAAdvisoryCommittee.com.]

The role of progression-free survival in assessment of ovarian cancer agents has come under scrutiny in recent months.

During a March meeting, ODAC voted against recommending approval of Lilly's Gemzar for ovarian cancer despite a statistically significant benefit on PFS; panel members said that PFS had not been shown to be a surrogate for overall survival in ovarian cancer (³ Also see "Progression-Free Survival Must Be Tied To Survival For Oncologic Approval" - Pink Sheet, 20 March, 2006., p. 25).

During the endpoints meeting, Workshop Co-Chair Tate Thigpen (University of Mississippi) cited data from 15 large randomized clinical trials; improvement in PFS predicted improvement in overall survival in 13 of the trials, he said.

Panel member Marc Buyse (French National Institute of Cancer) presented a meta-analysis of four randomized ovarian cancer trials with a total of 1194 patients. "The association between PFS and overall survival was very high...showing that the two endpoints were highly correlated in this population of patients."

Pazdur said another drawback with powering trials for overall survival when PFS is the primary endpoint is that statistical significance in PFS may not reflect clinical benefit.

"As we power trials for survival...we'll get an overpowering for PFS. Just merely a statistically significant difference does not indicate a clinically significant difference," he said.

Nerenstone agreed that agents need to show more than a statistically significant benefit on PFS. "What increase is meaningful? Is a one-month, two-month, three-month increase enough to get something approved? Personally I don't think it is."

Buyse also presented an analysis of the "surrogate threshold effect" for PFS in ovarian cancer, the magnitude of effect on PFS necessary to ensure a positive effect on overall survival.

According to Buyse, "a new treatment will have to cut the risk of progression of death by at least 45% for us to be sure statistically that there will be a survival benefit to be seen. That is not so nice."

Pazdur asked the committee whether an improvement in PFS is clinically significant in the absence of a PFS effect substantial enough to indicate a survival benefit.

"To paraphrase Tom Fleming, a correlate doesn't make a surrogate," Pazdur said. "The major issue in up-front treatment is would a meaningful impact on PFS be of clinical benefit in and of itself?" Fleming, a biostatician from the University of Washington, has served on a number of FDA advisory committees including ODAC.

"Could there be a delay in development of symptoms?" Pazdur asked. "Could it prevent for example other therapies being introduced earlier?"

Thigpen cited the consensus statements from a recent Gynecologic Cancer Intergroup conference that states that PFS "in and of itself provides a measure of clinical benefit to the patient" by giving "increased time free of the adverse effects of therapy before recurrence or progression."

Eisenhauer agreed that delaying development of symptoms is clinically significant. "At least in the first-line, it is reasonable to infer that when most patients end treatment without disease and therefore without symptoms, that the longer you can put [recurrence] off...that it does make sense that there is a window of time which is prolonged in and of itself."

"It is meaningful to the patients. It's meaningful to not die of your disease. It's meaningful to have periods of time when you are feeling well," she said. "The problem is I would really be surprised if we could show that...because we haven't been able to measure it."

Eisenhauer's comments point to one of the key obstacles to using PFS in ovarian cancer: the difficulty in establishing measurable criteria for progression.

Panel members favored a "composite" measure of PFS that includes clinical and radiographic evidence as well as measurement of CA-125 (cancer antigen-125) levels. CA-125 is a protein found at elevated levels in most ovarian cancer cells.

However, panelists generally felt that the CA-125 and radiographic components needed further validation. "Progression, whether it is objective or [assessed with] CA-125, there are issues with sensitivity [as] to when you choose to do the measurement," Eisenhauer said.

Nerenstone agreed that "further prospective validation of CA-125 as a correlate for progression probably needs to be collected."

While the workshop participants generally felt PFS was an appropriate surrogate for overall survival in front-line therapy, they disagreed on its appropriateness in second- and third-line settings.

"In first-line you don't know who might be alive and well five to ten years from now," Eisenhauer explained. "Your goals from treatment are in theory to increase the proportion in that group."

"In second-line, when you have people with a disease with an end that is inevitable...it seems to be more important to show that the treatments that we are giving in fact do more than change the time [to] progression at that point than at the beginning."

Nerenstone said available data in refractory setting is insufficient to support PFS as a surrogate for survival. "In those data, they don't have as much and in refractory disease it's not as compelling," she said.

Panel member David Spriggs (Memorial Sloan-Kettering Cancer Center) disagreed: "If PFS is intrinsically beneficial as an endpoint, then it should be the same whether it is up front or recurrent disease."

Ovarian Progression-Free Survival Acceptable If Trial Is Powered For Survival

More from Archive

More from Pink Sheet