By Turna Ray
Even though a US Food and Drug Administration advisory panel this week voted unanimously to remove Avastin's indication as a breast cancer treatment, Genentech is still planning to move forward with a study that will investigate whether breast cancer patients with high levels of VEGF-A are so-called "super-responders" to the drug.
"We are willing to work with the FDA … to find a solution such as a modified or restricted label," Hal Barron, Genentech's chief medical officer, said during the second day of a two-day hearing held to discuss the agency's proposal to withdraw accelerated approval for Avastin as a treatment for metastatic breast cancer.
Barron's comments — and the Oncologic Drugs Advisory Committee unanimous vote against keeping Avastin's breast cancer indication — followed a full day of arguments from patients and the Roche subsidiary that the "super-responder" patient subgroup should not be denied access (see related article, this issue).
Genentech received accelerated approval in 2008 for Avastin in combination with paclitaxel for patients with HER2-negative metastatic breast cancer who have not yet received chemotherapy for metastatic disease. However, last December, the FDA notified the company of its intent to revoke Avastin's metastatic breast cancer indication based on confirmatory studies that showed the treatment was not efficacious or safe for this population of patients.
Genentech appealed FDA's decision and requested the public hearing, which was intended to allow Genentech, FDA reviewers, ODAC, and cancer patients to publicly discuss the scientific and legal issues at play in FDA's decision.
At the end of the hearing, the advisory committee was asked to vote on a number of questions, one of which was whether the FDA commissioner should keep Avastin on the market for breast cancer in some capacity while Genentech conducts another trial to try to prove that it is a viable treatment for that indication. All members of the advisory committee voted "no" on this question.
The additional confirmatory trial that Genentech had proposed involved a "biomarker component" in that it would use a VEGF assay to pick out whether patients with high levels of VEGF-A had a more favorable response to Avastin than did metastatic breast cancer patients without this biomarker. But since there was no mature data that such a biomarker strategy would likely prove Avastin's efficacy in a subpopulation, committee members did not want to expose patients unnecessarily to a toxic agent for many years to come.
Patients and several oncologists spoke during the hearing in favor of keeping Avastin available for women with triple-negative breast cancer. Although there is anecdotal evidence of triple-negative breast cancer patients being super-responders to Avastin, Genentech could not provide clinical evidence supporting these accounts.
"There was some hope that perhaps there was some subset of patients that responded favorably, and it seems that that's not the case," said Natalie Compagni-Portis, the patient representative on the advisory committee.
"I think when you're looking at things in a broad sense, and the agency has to look at protecting a large number of patients, sometimes they have to make a decision that doesn't favor individual patients," said committee member Ralph Freedman, a professor at MD Anderson Cancer Center.
FDA initially greenlighted the drug based on data from a clinical trial called E2100 that showed that patients on Avastin/paclitaxel combination therapy experienced a statistically significant median improvement in progression-free survival of 5.5 months over patients on the paclitaxel arm. However, in this trial, there was no improvement in overall survival with the addition of Avastin.
As a condition of accelerated approval, the agency stipulated that the company submit confirmatory studies that confirmed that Avastin improved patient outcomes over standard chemotherapy as seen in E2100. In two studies comparing Avastin plus chemotherapy to chemotherapy alone, patients in the combination arm in one trial, called RIBBON1, experienced an approximate one-month median improvement in PFS; while patients in the second trial, called AVADO, had median PFS of less than three months with the combination regimen. Neither study showed an overall survival advantage or a quality of life benefit associated with Avastin plus chemotherapy.
The advisory committee members all felt that the totality of evidence on Avastin did not prove that it is an effective metastatic breast cancer drug. Additionally, the "modest magnitude of benefit" seen in the clinical trials doesn't justify the toxicity risks and the risk of death seen in studies, the committee members concluded. "We all wanted Avastin to succeed … but these studies don't bear out that hope," Compagni-Portis said.
While FDA usually follows the advice of its advisory committees, the guidance provided by the independent body is not binding on the agency. Notably, FDA went against ODAC's recommendations when it first chose to grant Avastin accelerated approval in metastatic breast cancer. Ultimately, FDA Commissioner Margaret Hamburg will decide whether to keep Avastin as an option for breast cancer patients, to remove it in that market, or whether to amend the indication to subset of patients.
On the first day of the hearing, ODAC supported Genentech's idea for exploring who responds best to Avastin, but expressed concern with the three-and-a-half-year projected timeline for the completion of the proposed confirmatory trial with biomarker subset analysis. Instead, the committee indicated that it would be more appropriate for the company to pull the drug's metastatic breast cancer indication for the time being and refile a new application with an amended indication once the trial is completed.
[ pagebreak ]
On the second day of the hearing, Genentech continued to push the agency to maintain the drug on the market as an option for breast cancer patients. However, the company seemed open to negotiating a label change with FDA's Center for Drug Evaluation and Research, narrowing the indication of the drug to those most likely to respond.
Genentech's Barron said that the company would be willing to work with the agency on a restricted label, and noted that the firm's "primary objective is to preserve in an appropriate manner options for women with metastatic breast cancer."
Joyce O'Shaughnessy, co-chair of breast cancer research at US Oncology and a paid consultant for Genentech in breast cancer trials for Avastin, suggested that perhaps FDA could keep the drug available for advanced triple-negative breast cancer patients or women with aggressive ER-positive breast cancer, under a restricted labeling scheme.
There is a hypothesis among breast cancer researchers that because triple-negative breast cancer patients have higher VEGF levels, they may respond to Avastin, a drug which prevents tumor vessel growth by inhibiting the VEGF ligand and regulating angiogenesis. Although the clinical evidence submitted by Genentech doesn't suggest that triple-negative breast cancer patients see a pronounced response with Avastin, on the first day of the hearing several women with this specific type of cancer said they had gone many months without seeing their tumors progress.
O'Shaughenessy also recounted during the hearing that she had treated triple-negative breast cancer patients who were responding well to Avastin with manageable side effects. However, this anecdotal evidence was not corroborated by clinical evidence provided by Genentech.
The kind of labeling change proposed by O'Shaughnessy would have to be based "on data from clinical trials suggesting that a subgroup has a favorable benefit/risk profile," noted John Jenkins, director of CDER's Office of New Drugs.
The FDA conducted exploratory analysis segregating triple-negative breast cancer patients from patients who were HER2-negative but ER- or PR-positive in all the studies submitted by Genentech. "Based on survival and progression-free survival data, subgroup analysis looks similar. There are not, for the most part, differences between patients who are HER2-negative, but ER/PR-positive, and those who are triple-negative," said Patricia Keegan, director of CDER's Division of Biologic Oncology Products. "So we don't have any sense that they respond any differently than other patients."
Sandra Horning, global head of clinical development in hematology/oncology at Genentech, couldn't provide specific response data on the triple-negative breast cancer population when questioned by the committee. She noted that the company isn't necessarily saying that these patients have a more pronounced response to Avastin than do other breast cancer patients, but because this subset has a more aggressive form of the disease and limited treatment options, they have more at stake.
According to Genentech officials, the lead biomarker the firm will use to identify best responders in the proposed study is vascular endothelial growth factor A, a protein encoded by the VEGFA gene.
James Reimann, global head of oncology biostatistics at Genentech, noted that CDER has asked the company to meet with the Center for Device and Radiological Health to discuss the VEGF-A companion test prior to starting a special protocol assessment with CDER. Recruitment for the confirmatory trial will begin in the first quarter of next year, and take 3.5 years to complete.
However, during FDA's questioning of Genentech officials, FDA's Jenkins projected that it could be as long as five years before the company files the new data and the agency makes its decision. If there is another negative decision, "we could be right back here at a public hearing," one FDA official noted.
[ pagebreak ]
Along the same lines, advisory committee members expressed concern that if FDA were to keep the drug on the market for the broader population of breast cancer patients while Genentech conducted a confirmatory trial, and then the trial proved negative, then many patients will have been exposed to the toxicities associated with a drug that essentially has no efficacy.
Genentech intends to conduct interim analysis to assess whether the confirmatory trial has a good shot at coming out positive at the end of three years. If, at interim analysis, the company sees it going in an unfavorable direction, then "it could trigger a voluntary withdrawal" of Avastin, Reimann said.
However, given the preliminary nature of Genentech's biomarker data, the advisory committee wasn't convinced that Avastin's benefit/risk calculation would be markedly different in patients with high VEGF-A levels.
Genentech's strategy is based on a hypothesis identified in retrospective biomarker analysis conducted in the AVADO trial. In that trial, the sponsor collected samples from 54 percent of patients, and looked at markers "based on their role in metastasis and angiogenensis," according to Reimann. "What we found in AVADO were that high levels of VEGF-A were predictive of a larger benefit of Avastin, with a hazard ratio of 0.87 in the VEGF-A low group and 0.49 in the VEGF-A high group."
Although high VEGF-A was not found to be predictive for overall survival in AVADO, he noted that this could be due to the small number of total patients who experienced a survival advantage in the study. Researchers also found that high VEGF-A levels were "strongly prognostic" of poor outcome in both PFS and OS, and that this association was not due to baseline characteristics of patients.
"Next steps include development of a commercially available VEGF-A assay, and prospective validation within the [proposed] confirmatory study," said Reimann, adding that the company is also studying the VEGF-A marker in other tumor types.
Tough Road Ahead
ODAC member Wyndham Wilson, who is also chief of the lymphoma therapeutics section at the National Cancer Institute, cited PGx analysis conducted as part of the E2100 study to note that researchers found that 7.6 percent of patients with VEGF-1154 AA and -2578 AA genotypes had a higher median overall survival after treatment with Avastin/paclitaxel, while 21.5 percent of patients with -2578 CA and -1154 GG genotypes had worse median overall survival.
"Obviously this raises concern that any hypothesis about VEGF needs to incorporate polymorphisms, but it also raises the question that if the E2100 trial is not correct, that by allowing this trial [and] allowing this drug to stay on accelerated approval, we may be harming more people than we're helping," Wilson said.
To this, Genentech's Horning noted that the firm has not been able to confirm the response SNPs identified in the E2100 substudy in subsequent trials. According to retrospective SNP analysis conducted in 154 pancreatic cancer patients in a Phase III trial called AVITA, researchers did not report a survival association with Avastin in patients harboring the VEGF -2578 AA genotype. However, in this study, researchers did find correlation with a VEGF receptor-1 polymorphism and increase in overall survival.
"Next we went to the AVADO study where we do have plasma and we not only looked at these polymorphisms but we looked at a large number of polymorphisms. We were unable to confirm the [SNP] findings in E2100," she said, adding that this data will be presented at a conference hosted by the European Crohn's & Colitis Organization and the European Society of Medical Oncology in the fall.
Horning noted that in the E2100 genotype analysis researchers didn't have plasma samples, but analyzed tumor tissue. Additionally, the researchers only genotyped patients receiving Avastin plus paclitaxel, she said.
Given the molecular heterogeneity in breast cancer tumors, it's unlikely that there will be a straightforward biomarker associated with Avastin response. "Angiogenesis in tumors involves a complex interaction between tumor, stroma, many pro- and anti-angiogenic factors. This complexity creates considerable challenges for biomarker discovery," Horning said, adding that since 2002, Genentech and Roche have evaluated more than 100 potential markers in clinical specimens of plasma, tumor, and host DNA across seven tumor types.
"Data from more than 10 Phase III trials with adequate sample collection began to read out in 2010 and will continue through 2012," she noted. "Naturally, VEGF pathway markers have been a major focus in this work, based on the mechanism of action of Avastin."
FDA officials and advisory committee members believe that unless the agency revokes Avastin's approval in breast cancer, Genentech will not be able to recruit patients for its proposed confirmatory trial, which is very similar in design to the E2100 trial in that it would compare a combination Avastin/paclitaxel regimen to paclitaxel alone, because patients would not want to be assigned to the paclitaxel arm.
[ pagebreak ]
According the agency, there is evidence suggesting that patients are reluctant to enroll in randomized trials in which an FDA-approved drug is being investigated and the study design is similar to previous pivotal trials. "[P]atients [will not] enroll in a study where they're asked to forgo an FDA-approved treatment option," Jenkins said.
For example, AstraZeneca voluntarily withdrew its accelerated approval application for Iressa in February, probably because it was having trouble recruiting patients for studies in the US. The drug's efficacy shortcomings in the general population were highly publicized and since 2005, the treatment was available through a limited access program for those already benefitting (PGx Reporter 02/09/2011).
"I really haven't heard a really good assessment how [recruitment] is going to be impacted by the indication still being approved. It seems to me that it will be extremely difficult to accrue to such a trial in the US and in Europe," Wilson said. "Hence, will this be required to be done outside of the West, and will it be significantly longer before we have any answers and therefore expose patients to even longer periods of risk if the confirmatory trial turns out to be negative?"
Genentech said it intends to recruit the majority of patients for the confirmatory trial outside of the US. However, company officials said they conducted a preliminary feasibility assessment and found there was interest among oncologists and patients to participate in the study. Genentech plans to complete a final feasibility analysis of the confirmatory trial in July.
"We do feel there are individuals in the US — physicians and patients — who are at relative equipoise on this question, and would be willing to participate in this trial, and would be willing in their participation because a biomarker hypothesis is included," Genentech's Horning said.
If FDA decides to revoke approval for Avastin in the metastatic breast cancer setting, Genentech stands to lose around $1 billion in revenue per year. Doctors would still be able to prescribe Avastin off-label for breast cancer patients, but insurers may not cover the drug, which can cost upwards of $80,000 per year.
Avastin, which brought in revenues of around $7 billion last year, is also approved as a treatment for metastatic colorectal cancer, non-small cell lung cancer, and glioblastoma.
The FDA and Genentech plan to provide additional written submissions by July 28, and the docket will remain open for public comment until then.
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at tray [at] genomeweb [.] com.