Eisai’s development plan for Halaven was derailed by FDA’s approval of Ixempra in the same metastatic breast cancer indication, precluding the planned accelerated approval filing and redirecting the sponsor to a full approval track, but it also meant eribulin received the first approval in its highly refractory patient population ever to be based on overall survival data.

Halaven’s survival advantage gave the NDA an advantage throughout the regulatory review process, as FDA made a priority of getting the drug to patients without delay. Issues that can easily lead to a “complete response” letter, including a dispute over manufacturing and FDA doubts about the choice of a complex active control arm in the pivotal trial, were resolved in the interest of providing patients with a therapeutic option that could extend survival.

Eisai effectively traded the questionable value of its Phase II response rate data for the regulatory “gold standard” of improved survival.

Robust overall survival data is a rarity in trials in advanced metastatic cancer patients, where FDA accepts many applications that rely on less clinically meaningful endpoints because of the exigencies of the disease. Bristol-Myers Squibb’s Ixempra received a full, not accelerated, approval based on progression-free survival and response rate data, even though FDA reviewers doubted that the drug’s impact on surrogate endpoints would translate into a benefit in overall survival (Also see "FDA Cleared Ixempra Despite Doubts Over Chance Of Showing Survival Benefit" - Pink Sheet, 1 June, 2008.).

Eisai effectively traded the questionable value of its Phase II response rate data for the regulatory “gold standard” of improved survival.

The Impact Of Ixempra

The accelerated approval plan counted on Halaven being the first drug approved for an unmet medical need, in this case, metastatic breast cancer patients “with at least 2 but not more than 4 prior therapies,” Division of Biologic Oncology Products Director Patricia Keegan said in her Nov. 11, 2010 summary review. “If another agent were granted regular approval for this population, accelerated approval could not be granted,” she noted. And Ixempra (ixabepilone) was approved for that population Oct. 16, 2007.

Eisai was publicly conscious of the delay that shifting from an accelerated to a regular approval track entailed. At a March 4, 2011, analyst event, Eisai categorized Halaven as a project that had “lagged behind the schedule” because the company had “previously assumed filing under Subpart H.”

But the delay proved fortuitous. Eisai ended up working smoothly with FDA for an efficient single-cycle approval based on the “gold standard” of overall survival in 2010, instead of prosecuting a marginal case for approval on a response rate surrogate endpoint.

The Plan For Accelerated Approval

Eisai had planned to base an accelerated approval filing on demonstration of durable objective tumor responses, relying on the results of two Phase II single-arm trials, Studies 201 and 211, in the third-line (or greater) treatment setting; the trials required patients to have previously received an anthracycline and a taxane. Verification of clinical benefit was to come from a randomized trial, Study 301, comparing eribulin to capecitabine for second-line use. The primary endpoint for Study 301 was progression-free survival.

The 201 and 211 studies were submitted to the eventual NDA as supportive trials, and the objective response rate data they produced suggests that a rocky road was ahead for any application relying on the two trials.

“With regard to the acceptability of the NDA package to support accelerated approval, FDA noted that the ORR … appeared low and that adequacy to support approval would be a review issue,” FDA told Eisai in a pre-NDA meeting Aug. 23, 2007, Keegan recounts.

Overall response, comprised entirely of partial responses, was 11.5% in study 201 and 9.3% in study 211. At the pre-NDA meeting, “the agency stressed that an objective response rate of 15% might not be sufficient to be considered reasonably likely to predict clinical benefit,” Keegan said.

In contrast, the survival advantage in the subsequent Study 305, also known as the EMBRACE trial, turned out to be robust, Keegan said, in that the finding was consistent across relevant subgroups based on patient demographics and tumor prognostic characteristics and the overall survival effects were statistically significant at multiple timepoints.

“A clinically meaningful prolongation of overall survival is the preferred basis for drug approval in the treatment of metastatic breast cancer, because improved survival reflects both safety and efficacy (deaths are caused by toxicity or progressive disease, or both),” medical reviewer Martha Donoghue stated in her Aug. 31, 2010 review. “Overall survival is an unequivocal, direct measurement of clinical benefit.”

It may be the gold standard, but OS is not mainstream for oncology approvals. An FDA analysis of oncology approvals from 2006 to 2010 found that only 17.1% of FDA drug actions were based on overall survival, compared with overall response rate and progression-free survival, each used in 39.5% of FDA actions (Also see "FDA Official Stresses "Flexibility" On Speeding Approval Of Drugs That Truly Stand Out" - Pink Sheet, 27 October, 2010.).

In highly refractory breast cancer patients, an OS effect was unprecedented. “Unlike this application, none of the approvals in the anthracycline and taxane-refractory population were based upon a demonstrated improvement in overall survival,” clinical reviewer Donoghue said (see chart, below).

Regimen	Approval	Primary Endpoint
Capecitabine monotherapy	Accelerated approval, 4/30/1998	Response rate
Lapatinib plus capecitabine	Accelerated approval, 3/13/2007	Time to progression
Ixabepilone monotherapy & combination with capecitabine	Regular approval, 10/16/2007	Progression-free survival (combo); response rate (monotherapy)
Eribulin	Regular approval, 11/15/2010	Overall survival

A Boost From The European Regulatory Process

Halaven’s overall survival numbers would not have been so convincing if FDA had not asked the sponsor to submit the results of an updated analysis requested by the European Medicines Agency (Eisai submitted eribulin simultaneously in Europe, the U.S. and Japan). “The effect on OS was confirmed by an updated analysis showing the treatment effect to be persistent and statistically robust,” Office of Oncology Drug Products Director Richard Pazdur said in a Nov. 12, 2010 summary review.

Cross-disciplinary team leader Steven Lemery explained the importance of the updated unplanned survival analysis. “The results of the primary analysis of OS were not necessarily statistically persuasive,” he said in his Nov. 9, 2010 review. “However, the results of the updated OS analysis after approximately 75% of events appeared more favorable for OS with limited loss of follow-up.” The p-value for the primary analysis was 0.041; the p-value for the later updated data analysis was 0.014. Eribulin patients’ median survival beat the treatment of physician’s choice arm in the initial NDA by 2.5 months, edging up to a 2.7 month difference in favor of eribulin in the updated analysis. The updated results are included in labeling.

Consensus In Favor Of Approval At FDA

“All members of the review team recommended approval of this application,” Keegan noted. That level of concurrence within the agency helped Halaven avoid advisory committee review, even though most first-in-class NMEs make that trip. Indeed, during a May 5, 2010 teleconference FDA told Eisai to expect review during the Oncologic Drugs Advisory Committee’s tentatively scheduled meeting Sept. 1-2, 2010. A few weeks later, on May 25, 2010, FDA informed Eisai that it had “decided not to take” Halaven to the committee.

“There were no controversial issues identified by the review team or me that would have benefitted from an advisory committee discussion,” Pazdur explained. The decision was agreed upon across the clinical and statistical review team and division/office management. The “primary justification for this decision relates to the primary endpoint of Study 305,” Lemery explained. “A statistically significant improvement in overall survival is the ‘gold standard’ for the approval of oncology drugs.”

In lieu of a panel review, the review team brought in three consultants to give outside perspective. According to FDA memoranda of the discussions, which took place Oct. 19-20, 2010, the experts – M.D. Anderson Cancer Center breast cancer expert Aman Buzdar, ODAC Chair Wyndham Wilson, and patient representative Debra Madden – all agreed that the improvement in overall survival was a meaningful benefit and identified no major concerns regarding the NDA.

One Pivotal Trial Is Enough

“The major issue regarding the consideration of efficacy for this application was whether FDA could rely on the results of a single study to support the approval of eribulin,” Lemery stated. “An additional issue … was the choice of the Study 305 control arm (single agent therapy chosen by the investigator).”

Eribulin’s effect on overall survival helped the NDA past those concerns as well.

Medical reviewer Donoghue acknowledged “the inherent limitations of relying on the results of a single randomized, well-controlled study for approval,” but concluded that the NDA had sufficient scientific and legal basis for approval under FDA’s “Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products” guidance, based on the FDA Modernization Act of 1997.

The guidance limits approval based on a single study “to situations in which a trial has demonstrated a clinically meaningful effect on mortality, irreversible morbidity, or prevention of a disease with potentially serious outcome,” Donoghue quotes. “Confirmation of the result in a second trial would be practically or ethically impossible.”

She further noted that Study 305 had many of the characteristics of a desirable single study per the guidance. “It is a large, multicenter trial that demonstrated consistent results across most patient subsets, and achieved more than one endpoint. Furthermore, the demonstration of a clinically meaningful, statistically significant overall survival benefit in a refractory population that has very limited treatment options renders the conduct of a second confirmatory randomized controlled trial practically or ethically impossible” because patients cannot be expected to risk randomization to a treatment with inferior survival.

A Control Arm For Very Experienced Patients

A perennial challenge for the design of trials for advanced cancer patients with extensive experience of prior therapy is selection of a control arm, because the gold standard of placebo control is unethical and unfeasible in patients with progressive, life-threatening disease. Eisai decided to leave the selection of an active control to the treating physicians, who selected a single agent as their control prior to randomization.

FDA had been leery of the treatment of physicians’ choice design during the IND phase. Donoghue noted that in a Jan. 22, 2007 letter to Eisai: “FDA expressed concern that Study 305 may not support full approval because the use of different therapies in the control arm would make it difficult to determine whether the toxicities of the control arm had an adverse impact on survival.”

Lemery commented that the control arm in Study 305 complicated the efficacy and safety analysis because true treatment effects could not be determined without a placebo control. “There is the theoretical chance that therapies in the control arm could have increased toxicity with no benefit,” he noted in his review memo, making it impossible for the trial to “definitively prove that eribulin is better than placebo.”

The team leader allowed that while the physicians' choice control meant a less clear analysis, it was acceptable for the situation.

The specific setting for the clinical trial is an important consideration in choosing an active control arm, Lemery explained. “In settings where there is a well-recognized standard of care (i.e., first line therapy of Hodgkin’s lymphoma), the clear choice is to compare the experimental regimen to the standard of care regimen (often this would be an add-on design).”

But in settings like ≥3^rd line metastatic breast cancer or other settings without clear effective therapy and where patients have already been treated with multiple different drugs, "there is reasonable justification for use of a ‘physicians’ choice’ control arm,” Lemery stated.

"In summary,” he said, “despite the potential disadvantages of bias and increased complexities of data analysis posed by the ‘physicians choice’ design, the control arm chosen for Study 305 was acceptable for the following reasons: inability to enroll patients in a true placebo-controlled trial; lack of an established standard of care in the ≥3^rd line metastatic breast cancer setting; overall survival endpoint; and the potential advantage of choosing an individualized regimen with a reasonable risk-benefit profile for each patient that simulates a ‘real-world’ situation.”

Control Arm’s Complexity Limited Indication

“However, because Study 305 was a single study with a complex control arm in a highly refractory population, it cannot be determined if eribulin would be superior to currently used drugs in earlier stages of breast cancer,” Lemery stated. “Thus, the review team recommended limiting the indication to patients who have received an anthracycline, a taxane and at least two chemotherapy regimens in the metastatic setting.”

The complex control arm led FDA to restrict the indication.

The label submitted by Eisai presented an indication for the treatment of patients with locally advanced or metastatic breast cancer who have previously received at least two chemotherapeutic regimens, Lemery reported.

The approved indication limits the submitted indication slightly. The indication was “revised to reflect the population enrolled in Protocol 305,” Keegan said. “This population was limited to patients with metastatic disease (no locally advanced disease patients).” FDA also tightened the history of chemotherapy criterion to bring it in line with the enrollment criteria: patients who have previously received an anthracycline, a taxane, and at least two chemotherapeutic regimens for the treatment of metastatic disease. “Prior treatment with an anthracycline or taxane may have occurred in the adjuvant or neoadjuvant setting,” Keegan noted.

FDA's Division of Drug Marketing, Advertising and Communication recommended a further limitation of the indication to “patients who experienced disease progression within 6 months of their last chemotherapeutic regimen,” an inclusion criterion of Study 305, Lemery reported. “The clinical review team, however, did not agree with the DDMAC recommendation.”

Lemery explained his opposition to the DDMAC recommendation, observing that “differences in timing of obtaining imaging in clinical practice compared to a clinical trial makes determination of the exact timing of progression difficult.” At any rate, he noted, the typical course of metastatic breast cancer moots the importance of the six-month criterion because “most women will be expected to progress within six months of their prior regimen.”

Lemery also highlighted the lack of a survival benefit for other cytotoxic drugs administered to women with metastatic breast cancer following treatment with taxanes and anthracyclines (and two therapies in the metastatic setting).

The Halaven review offers a vote of confidence in the FDA oncology division’s ideal of a demonstrated effect on survival as the gold standard for approval, even though the agency has regularly cleared NDAs and BLAs based on surrogate endpoints in difficult advanced cancer settings in the 15 years that the accelerated approval program has applied to oncology.

Even as Subpart H submissions have become routine, FDA’s top oncology staff has questioned the value of accelerated approval. Pazdur has championed a move away from accelerated approval submissions that are based on final data from single-arm Phase II studies, which can only support claims in refractory disease, in favor of filings based on interim Phase III data.

The Office of Oncology Drug Products has focused on the problem of sponsors not completing the confirmatory trials they are supposed to complete post-marketing to confirm the clinical benefit suggested by surrogate endpoints and convert the accelerated to a full approval ([A#06140080001]).

The ongoing debate over Avastin’s accelerated approval in breast cancer (and the potential for FDA to vacate that approval) dramatizes the potential risks of approvals based on data that does not meet the gold standard of survival. CDER is pinning the integrity of the Subpart H pathway on the June hearing. In its written summary of arguments, CDER argues that maintaining the breast cancer indication while Genentech conducts an additional confirmatory trial would turn accelerated approval into a pathway that was a lower standard for approval (Also see "CDER Says Keeping Avastin Breast Cancer Claim Would Undermine Accelerated Approval Process" - Pink Sheet, 16 May, 2011.).

Halaven's approval shows the advantage of targeting overall survival in the first place. While Eisai was all set to pursue accelerated approval, gaining that approval based on somewhat marginal response rate data in two single-arm Phase II trials could have been a difficult task and would require a confirmatory survival trial. But when circumstances forced Eisai to abandon that plan and conduct a trial with a survival endpoint, the sponsor ended up with full approval, clean labeling, and co-operation from FDA in overcoming problems with the NDA.

By Bridget Silverman

Halaven Survival Data Was Gold-Standard Lining In Forced Shift From Accelerated To Full Approval

More from Archive

More from Pink Sheet