Now that the initial surprise has worn off regarding the FDA's decision to grant accelerated approval for Genentech Inc.'s Avastin (bevacizumab) in breast cancer, the industry can begin to digest the implications of that decision.

Top of mind, of course, is whether or not the approval indicates a shift in FDA policy on measuring the efficacy of cancer drugs.

In a draft guidance document issued last year, the agency said that overall survival (OS) is the most reliable endpoint for cancer drug approval, but tumor assessment endpoints such as progression-free survival (PFS) may be evaluated on a case-by-case basis.

The PFS case for Avastin was based on a Phase III trial known as E2100, in which treated patients achieved a median PFS of 11.3 months, compared to 5.8 months for the control arm. However, there was no statistically significant difference in median overall survival (OS), with the Avastin group surviving 26.5 months and the control group surviving 24.8 months. That discrepancy led the Oncologic Drugs Advisory Committee to vote 5 to 4 against recommending approval.

The FDA, however, granted accelerated approval - marking the first time PFS has been used alone as the basis for approval of a first-line breast cancer drug. Other breast cancer drugs approved based on PFS - such as Tykerb (lapatinib, GlaxoSmithKline plc) and Ixempra (ixabepilone, Bristol-Myers Squibb Co.) - have been for refractory rather than first-line use.

Does this signify a change at FDA?

"There is an interest in looking at PFS as an endpoint," Richard Pazdur, director of the FDA's Office of Oncology Drug Products, told BioWorld. He added that the European Medicines Agency (EMEA) has long used PFS as an acceptable endpoint for the approval of first-line cancer drugs.

One benefit of PFS often cited by industry is that the data come years ahead of OS data, providing the opportunity to more quickly assess and, if applicable, allow patient access to drugs. Additionally, PFS may better illustrate the actual effects of a drug, since it is less influenced by subsequent treatments and other long-term factors that can skew OS.

"The ultimate goal is to improve overall survival, but we have to have some flexibility in approving drugs," Pazdur said.

PFS not just a surrogate for OS

Additional evidence that PFS may be coming into its own is the fact that the Avastin accelerated approval was not based on the assumption that PFS is a surrogate endpoint likely to predict the clinical benefit of OS.

Confusion on this point arose from the FDA's use of the accelerated rather than traditional approval pathway for Avastin. Accelerated approval usually means that the drug "has an effect on a surrogate endpoint that is reasonably likely . . . to predict clinical benefit." Many people wondered if the agency expected PFS to serve as a surrogate that would predict OS in future trials, even though that had not panned out in E2100. And if those future trials failed to show an OS benefit, could approval be reversed as it was for lung cancer drug Iressa (gefitinib, AstraZeneca plc), which gained accelerated approval based on response rate but then failed to show OS in a confirmatory trial?

Pazdur clarified that the accelerated approval pathway used for Avastin was not based on the traditional surrogate endpoint analysis but on language stating that accelerated approval can be granted "on the basis of an effect on a clinical endpoint other than survival or irreversible morbidity."

"We're not saying PFS is a surrogate, we're saying it is a clinical benefit in certain circumstances," Pazdur said.

Converting the accelerated approval into a full approval will be contingent on a full FDA review of data from the recently completed AVADO Phase III trial, the ongoing RIBBON I Phase III trial, and three other ongoing or planned trials. That data will "better define" both the PFS benefit and any possible OS benefit of Avastin, Pazdur said.

Genentech spokesman Ed Lang said "the details of what will make those trials acceptable" to the FDA and support full approval are "not clear."

Pazdur declined to hypothesize about specific data scenarios but assured BioWorld that the agency is "not going to be dogmatic here" and say the trials must show OS or the drug will be yanked off the market.

Lehman Brothers Inc. analyst Jim Birchenough also predicted that "it's highly unlikely that what happened to Iressa would happen to Avastin." He reasoned that when Iressa gained accelerated approval, no one knew its impact on survival. Then Iressa failed its confirmatory survival trial while the similar drug Tarceva (erlotinib, Genentech) succeeded. Avastin is different, he explained, because everyone already knows it doesn't improve survival but there aren't any other options for HER2-negative breast cancer patients.

Looking at the totality of the data

Pazdur acknowledged that PFS can be controversial in that "it is a matter of judgment whether PFS is a clinical benefit or not." That's why the FDA's decision on Avastin was based "not just on the endpoint" but on the "totality of the data," he said.

A major factor at play was the fact that Avastin is already approved for non-small-cell lung cancer (NSCLC) and metastatic colorectal cancer (CRC).

"It's easier to get two or three scoops of ice cream than one scoop, once the ice cream is out on the counter," said analyst Mark Monane of Needham & Co. He noted that since Avastin is already on the market, the FDA has access to a wealth of safety information about the drug. Monane contrasted this with a truly new drug like Dendreon Corp.'s prostate cancer vaccine Provenge (sipuleucel T), which ODAC green-lighted but the FDA rejected after two trials had failed to improve time to progression but a post hoc analysis showed survival benefit.

Pazdur also noted that Avastin was already being used off-label in breast cancer, and the FDA prefers that the public have access to data that has been "vetted and reviewed" by the agency. For example, the label for Avastin now clearly states that it is not recommended for second- and third-line treatment of breast cancer patients who previously received anthracycline and taxane therapy, based on a failed trial in that indication.

Another "piece of the puzzle" was data from the AVADO Phase III trial, Pazdur said, which the agency knew was coming and specifically requested prior to issuing its decision. While Genentech publicly announced only that the trial demonstrated a statistically significant improvement in PFS, the FDA was privy to initial secondary endpoint data on OS, safety and other measures.

AVADO also offered the benefit of confirmatory PFS data from a randomized, double-blind, placebo-controlled, company-sponsored study. E2100, on the other hand, had come under fire from ODAC because it was an open-label study conducted by the NIH-affiliated Eastern Cooperative Oncology Group and was not originally intended to support approval. The AVADO data conveniently allowed FDA to acknowledge ODAC's concerns about the E2100 trial design without completely condemning the NIH-sponsored clinical trial system.

And there may have been other political factors at play. Stanford Group Co. analyst Gregory Frykman speculated in a report that the agency could know that the upcoming March 13 ODAC panel might end up restricting the use of erythropoiesis stimulating agents in chemotherapy-induced anemia in breast cancer - in which case, perhaps the agency "was willing to give something to the breast cancer community, in the form of an Avastin accelerated approval, to head off a potential backlash that it was being too hard on that disease."

Regardless of the reasons for the decision, "the FDA is always right," Monane said. "They're like your parents: it doesn't matter if you get the car every Friday but now suddenly you can't have the car - that's ok."

The take home message for the industry, Birchenough said, is that companies "can't take away that PFS will be an approvable endpoint every time." It certainly hasn't been in the past, and the Avastin decision doesn't signal that it will be in the future.

What the Avastin decision does signal, Birchenough said, is that the FDA "will look at PFS on a case-by-case basis." But he cautioned that companies choosing PFS as an endpoint will need to show a decent magnitude of benefit, will probably end up in front of an advisory panel, and may be viewed as "riskier" than those with OS data.