Reflected in the regulatory struggle over Avastin’s (bevacizumab) metastatic breast cancer claim is the uncertainty underlying drug approval decisions for FDA and sponsors, laying bare the sometimes tenuous evidence with which the agency has to work.

FDA’s June 28-29 hearing on the Center for Drug Evaluation and Research’s proposal to revoke Avastin’s accelerated approval in MBC served as a figurative trial on the VEGF-inhibitor’s efficacy and safety data.

Yet it also could be viewed as a trial of CDER’s original approval decision. The regulator was forced to defend both its initial approval (granted despite differing opinions within the oncology review division) and its later determination that the indication should be revoked.

Throughout the hearing, at which the panel voted unanimously against keeping the indication, Genentech cited statements from Office of Oncology Drug Products Director Richard Pazdur’s February 2008 approval memo in an effort to contradict CDER’s current view that Avastin’s first-line MBC claim is no longer supported and should be withdrawn.

Pazdur stood his ground in explaining the reasoning behind the original approval and why CDER has since had a change of heart in light of subsequent studies. Yet the proceedings showed how an FDA approval decision can come back to haunt the agency, particularly when there are early questions surrounding the confidence in the underlying data.

Accelerated Approval = “Appropriate Option”

CDER showed a great deal of regulatory flexibility in approving Avastin for first-line MBC in 2008.

Approval was based upon a statistically significant median progression-free survival benefit of 5.5 months in the E2100 trial, which studied Avastin in combination with paclitaxel. Nevertheless, the pivotal study was plagued by questions involving missing data and design issues. Furthermore, another study (AVF2119g) in the second- and third-line settings failed to show a benefit with Avastin.

“Given the promising but yet unconfirmed benefit seen in E2100, CDER believed that accelerated approval was an appropriate regulatory option with a requirement for a post-approval study to confirm the magnitude of PFS seen in E2100 or to demonstrate some other direct clinical benefit to patients,” Pazdur said at the hearing.

However, the decision on Avastin's use for MBC can no longer be based solely on the E2100 results, he said, but rather “must be based on the totality of the evidence from all of the controlled clinical trials.” None of these studies (E2100 and AVF2119g as well as AVADO and RIBBON1) has shown a benefit in overall survival or health-related quality of life, and none has confirmed the magnitude of PFS benefit seen in E2100.

Putting Pazdur In The Hot Seat

The MBC indication was approved despite a split among oncology review staff. Pazdur served as the signatory for the approval and authored the Feb. 21, 2008 approval memo because Division of Biologic Oncology Products Director Patricia Keegan favored a “complete response, ” believing that Genentech had failed to adequately characterize the magnitude of treatment effect (Also see "Avastin’s Breast Cancer Approval Followed CDER Rift On Efficacy" - Pink Sheet, 30 May, 2011.).

Pazdur’s memo proved to be one of Genentech’s primary exhibits during the two-day hearing. As expected, the company sought to use the reasoning in that memo to counter CDER’s current views that the E2100 results could no longer be relied upon and the claim should be withdrawn. The tactic had the effect of putting Pazdur in the hot seat.

In the memo, the oncology office head explained how the Avastin benefit in E2100 was maintained in the face of various sensitivity analyses conducted to address issues related to missing data and censoring. Ultimately, E2100 showed a “robust effect” on PFS and objective response rate, the memo said.

Citing Pazdur’s reference to sensitivity analyses and the memo’s conclusions, Genentech counsel Paul Schmidt of Covington & Burling challenged CDER’s current views that the E2100 results are an outlier and not representative of Avastin’s true treatment effect.

CDER agreed there was a treatment effect and “the robustness issue is not in debate, in our mind,” Pazdur said. However, uncertainties surrounding the magnitude of PFS benefit cannot simply be addressed through sensitivity analyses, he said.

The cumulative nature of the problems with the E2100 study demanded that Avastin’s efficacy be confirmed, CDER’s Pazdur said.

The original application for the MBC indication had a number of problems, Pazdur said. These included positive data from only a single study, a negative second-line trial, questions about reliability of radiological reviews, missing scans and the loss of about one-third of patients who were not followed until a PFS event or end of study.

Each issue, in and of itself, was not unprecedented, Pazdur said. However, “once you start building a record here of multiple issues in an application, there is a problem which requires the study then to be replicated.”

An Early Look At AVADO

The memo’s reference to the top-line AVADO results also illustrate the potential pitfalls of relying on preliminary data to support regulatory decision-making.

The agency requested preliminary results from AVADO prior to approving Avastin for MBC. It did not review the primary data from the study but relied upon approximately two dozen slides provided by Genentech.

AVADO also was conducted in a first-line setting, but used docetaxel rather than paclitaxel as the chemotherapy partner. According to the data Genentech provided to CDER, the study met its primary endpoint, demonstrating a significant 0.8-month benefit in median PFS, with a hazard ratio of 0.64.

Overall survival data, which were immature at the time, showed a hazard ratio of 0.65 and a time-to-event benefit that approached statistical significance. The top-line OS data ultimately were not reflective of the mature data, however, and Avastin was not associated with any survival benefit in the final analysis.

Genentech asserted that the Pazdur memo’s reference to the AVADO top-line data led the company to believe the agency would accept the study results as confirmation of the benefit seen in E2100 based upon the PFS hazard ratio (Also see "Avastin Debate Could Provide Clarity On Use Of Progression-Free Survival Endpoints" - Pink Sheet, 11 July, 2011.).

CDER granted accelerated approval knowing the AVADO data would not confirm the median PFS benefit seen in E2100.

The median PFS difference of 0.8 months “clearly did not replicate the 5.5 months in E2100,” Sandra Horning, Genentech’s global head of clinical development for hematology/oncology, said. “Nonetheless, CDER granted accelerated approval for Avastin plus paclitaxel and accepted AVADO as a confirmatory trial knowing that AVADO did not and would not confirm the magnitude of benefits seen in E2100 if the measure of magnitude was to be median PFS.”

“Going Out On A Limb” With Approval

Pazdur said the agency requested the top-line AVADO data not for purposes of an extensive review but, rather, to determine before announcing the MBC approval whether the study had succeeded or failed on the primary endpoint.

“Our purpose in requesting this data was not to review it. We did not know the status of the review as far as would this information hold up under FDA review,” Pazdur said. “It simply was to ask if the trial was going to be reported out as a negative or positive study. Since the controversy of the E2100 decision was going to be announced, we did not want to go out on a limb when, shortly thereafter, potentially a negative trial was going to be announced,” he said.

Pazdur repeatedly stressed that CDER sought validation of Avastin’s magnitude of benefit in any study.

“The bottom line is we are looking – and Genentech knew this – for a clinical benefit to be demonstrated in one of these trials or in any trial that they chose to give us for an accelerated approval confirmation. All we wanted was one single trial to show a clinical benefit here, and they could select anything they want or come up with new studies.”

Schmidt, however, pressed CDER on what it told Genentech in terms of the sufficiency of the company’s evidence.

“The company thought that AVADO was clinical benefit,” Schmidt said. “My question is, at any point in time, having this data in hand, having accepted AVADO as a confirmatory trial, did CDER say to Genentech that a 0.8-month median PFS difference will not be enough to support approval?”

“I do not believe we did that,” Pazdur replied.

By Sue Sutter

Avastin Breast Cancer Approval Memo Comes Back To Haunt FDA’s Pazdur

More from Archive

More from Pink Sheet