By Turna Ray
The drug industry has long maintained that its primary motivation for using pharmacogenomic strategies in drug development is to advance treatments with superior efficacy and low toxicities in subpopulations of patients. However, recent interactions between the US Food and Drug Administration and industry suggest that both parties may need to make some adjustments in order to make the best use of genomic discoveries in oncology development: Not only does pharma need to integrate biomarker analysis earlier in the drug development process, but the FDA's evidence requirements for personalized cancer drugs may also need to evolve.
"With genomics, and pharmacogenomics, and all the other aspects of the cancers, we have to think of a different way than the randomized trial," Patrick Loehrer, interim director of Indiana University's cancer center and a member of FDA's Oncology Drugs Advisory Committee, said during a meeting last week. "If I get colon cancer, and [someone else] gets colon cancer, just because we are different people, we're going to have different responses. Now the randomized trial is supposed to correct for that, but ultimately, we need to correct for the randomized trial by coming up with better genetic markers."
The ODAC meeting was held last week to discuss potential improvements to the FDA's accelerated approval process for oncologics. The committee heard from five sponsors who had outstanding post-marketing commitments related to drugs that had previously received accelerated approval from the agency. Three among these firms, Amgen, Eli Lilly, and Novartis, discussed completed and ongoing studies related to accelerated approval commitments for personalized medicine products.
Lilly presented data related to a biologics licensing application for Erbitux in combination with irinotecan for the treatment of EGFR-expressing metastatic colorectal cancer, and Amgen discussed data on its BLA for Vectibix in chemotherapy-experienced metastatic colorectal cancer patients with tumors expressing EGFR. The FDA in 2009 relabeled Vectibix and Erbitux to restrict their administration in colorectal cancer patients expressing KRAS mutations in codons 12 and 13 — a subpopulation that doesn't respond to EGFR-inhibiting monoclonal antibodies.
Novartis, meantime, presented its post-marketing commitments related to a supplemental NDA for Gleevec as an adjuvant treatment for patients who have had removed gastrointestinal stromal tumors expressing a protooncogene protein called CD117.
Following sponsors' presentations, ODAC discussed the circumstances under which the agency should accept single-arm trials for accelerated approvals; whether FDA should ask sponsors to submit two confirmatory trials, instead of one randomized study as is the standard for oncology drug approvals; whether confirmatory trials should already be underway when accelerated approval is granted for a drug; and the extent to which sponsors should work with cooperative groups to conduct confirmatory studies for accelerated approval commitments.
ODAC members did not vote on these questions. In general, committee members felt that they'd like to see fewer single-arm studies and preferred that sponsors conduct randomized-controlled trials when possible. Furthermore, panelists felt that ideally companies should conduct at least two randomized studies to confirm the safety and efficacy of drugs that have received accelerated approval. In the case of drugs for rare diseases, the ODAC panel felt the FDA might provide exceptions to this requirement.
With regard to the timing of studies, committee members felt that for accelerated approval to be granted, a sponsor must at least have a specific plan for conducing post-marketing studies, if randomized studies aren't already underway. Furthermore, while FDA and the advisory panel recognized that cooperative groups are a valuable resource for sponsors in conducting drug development trials, they stressed that the onus is on the sponsor to ensure timely completion of confirmatory studies required under accelerated approvals.
The Overall Survival Challenge
The FDA is increasingly demanding that new cancer drugs show a statistically significant improvement in overall survival compared to standard treatments. Even with a pharmacogenomic strategy, sponsors aren't having an easier time showing that patients are living longer on their drugs.
For example, AstraZeneca recently pulled its accelerated approval application for Iressa in non-small cell lung cancer, and said that it will not re-file a new application to try to launch the drug in a genomically-defined subpopulation. The company said that after conversations with the agency, it became clear that AstraZeneca wouldn't be able to meet the agency's requirements for accelerated approval (PGx Reporter 02/09/11).
Although the company wouldn't divulge the details of its discussions with FDA regarding Iressa, it is worth noting that a large randomized trial showed that Iressa did not significantly improve overall survival in NSCLC patients with EGFR mutations.
[ pagebreak ]
Ultimately, if FDA adopts ODAC's recommendations to require multiple randomized-controlled trials, personalized cancer drugs may have trouble passing muster, since it is difficult to recruit patients for trials conducted in molecularly defined subpopulations, and as a result such trials tend to be small.
The agency has made exceptions when it comes to personalized medicine products, particularly in the case of Amgen and Lilly, when new data emerged suggesting that patients with KRAS-mutated tumors don't respond to EGFR-inhibiting antibodies. In 2008, the agency held an ODAC meeting to review the available data from the sponsors of Vectibix and Erbitux, and decided on a set of criteria they would need to meet for submitting PGx analysis from samples collected in earlier studies (PGx Reporter 12/17/2008).
However, since both of these drugs were given accelerated approval status, the sponsors had to submit post-marketing studies showing that these treatments affected patient outcomes in genetically-defined patient populations. However, the data presented by Amgen at last week's meeting showed that Vectibix did not impact survival (PGx Reporter 02/09/11).
At the meeting, Amgen announced it would conduct another study that would be designed to ensure that the survival data from Vectibix is not confounded by non-responsive patients crossing over to another treatment. Still, advisory committee members were dubious that Amgen would be able to show any more improvement in survival than it already had.
"It's unlikely that these trials are going to show more than a few months of improvement in survival. And for the average patient this has very little impact," Loehrer said at the meeting.
Since the agency awards expedited review for treatments that serve an unmet medical need or treat a rare disease, Loehrer urged the FDA to better define what it means by "rare disease." Several committee members agreed, pointing out that when considering diseases characterized by molecular markers, the patient population becomes even smaller.
"For KRAS mutant patients, for example, in whom we know EGFR antibodies don't work, … if we had an EGFR antibody that suddenly had a 30 percent response rate, that would be meaningful — maybe not enough to be approved, but it would be meaningful," Loehrer said. "Certainly in rare diseases we're going to have to use these genetic markers to help us."
Trial Designs
Richard Pazdur, director of FDA's Office of Oncology Drug Products, said during the meeting that sponsors need to plan out their clinical trials earlier, and discuss the design of the trials with the FDA. Regulatory officials have encouraged sponsors developing drug/diagnostic products to meet with the FDA early and often to discuss biomarker strategies and regulatory requirements, which are still evolving at the agency for products that need to be simultaneously codeveloped by pharma and diagnostics firms.
FDA has several mechanisms to encourage industry to incorporate genomic biomarkers early in drug development. The Voluntary Exploratory Data Submissions program allows companies to discuss with FDA early biomarkers in drug development without regulatory repercussions. Several drugmakers that have filed new drug applications or are in the process of filing applications with the FDA for pharmacogenomically guided products have used this program.
Issam Zineh, associate director of genomics at FDA's Office of Clinical Pharmacology, previously told PGx Reporter that the genomics group in his office has seen a "dramatic increase" in the number of applications that have been filed containing genomic biomarkers. In 2008, the agency received around 60 drug submissions with genomic data. Last year, that number jumped to 210, and applications for oncology drugs represented half of the submissions.
Zineh said he sees sponsors struggling with sample acquisition in drug development trials, which is strongly encouraged by the FDA but is not required. Most sponsors make sample donation voluntary for patients participating in a study. However, if a gene marker is identified later in clinical trials, requiring a stratified trial, companies may find they don't have enough samples to conduct such analysis.
[ pagebreak ]
"Another challenge is simply one of certainty," Zineh previously noted. "Many times, drug developers feel they have insufficient data at the end of Phase II in drug development to confidently select out a target population to be tested in the definitive efficacy studies." He added that he'd like sponsors to experiment with "N-of-1 trials or cluster randomization designs to really get at whether pharmacogenomics can be done at a clinical practice level."
For example, Pfizer studied its ALK inhibitor crizotinib in two patients as a proof of concept showing that a patient with ALK rearrangements in lung cancer tumors had a sustained partial response to the drug, while a patient without an ALK translocation had no response to the investigational drug. Instead of taking the accelerated approval pathway, Pfizer has received Fast Track status for crizotinib, which allows the company to submit data on a rolling basis. This may be a better method for approval for this PGx-guided drug, since it allows more interaction and step-by-step planning for not only the development of the drug, but also the companion tests that will gauge ALK rearrangements in patients (PGx Reporter 01/19/11).
Lax Requirements?
The overwhelming sentiment among ODAC members at last week's meeting was that the agency has been far too lax when it comes to approving cancer drugs. Musa Mayer, the patient representative on ODAC, urged FDA not to lower its standards in the name of compassion.
"I am very disappointed that this process of approval has really become a screening process ... Everyone is interested in what is the least that we have to offer," said Silvana Martino, committee member and director of the breast cancer program at the Angeles Clinic and Research Institute. "I think this committee, the FDA, and others have allowed that to become the mood of science within the field of oncology … where we are willing to accept drugs with the most minimal evidence that they do anything at all, and moved them forward."
FDA's Pazdur acknowledged that sponsors often ask him what minimum requirements they can meet to gain approval of cancer treatments. In general, he appeared to be leaning toward urging industry to do at least one randomized trial, rather than a single-arm trial, to gain accelerated approval.
"Rather than accruing a large number of patients in a single-arm trial … you might have been better off just doing a randomized study relatively early on," Pazdur said, adding that there "is a role for single-arm studies, particularly in unique diseases, where one has high response rates."
Additionally, Pazdur noted that the Office of Oncology Drug Products would hold annual ODAC meetings to discuss sponsors' outstanding post-marketing commitments under the accelerated approval program. In this way, the agency can keep better track of sponsors' progress, and take action against those who are not completing their post-marketing commitments with due diligence.
Whatever new requirements the FDA embraces, it is clear that the agency will have to address its drug approval requirements in cancer with pharmacogenomics in mind. In this regard, the FDA at the meeting did not state any specific requirements it was considering for sponsors developing personalized cancer drugs.
Jay Tenenbaum, CEO of CollabRx and founder of a new collaborative research effort called Cancer Commons, believes that when it comes to cancer treatments, the answers to questions of whether a drug works in a patient and why won't come from the types of randomized trials that FDA is fond of.
[ pagebreak ]
"Clinical trials are an inefficient way to answer the questions that FDA is seeking. A better way to do it is to be able to study the individual impact on basically every patient being treated on this drug, as well as patients not being treated on the drug," Tenenbaum told PGx Reporter. "The top-level question for me is: Are any breast cancer patients being helped by, say, Avastin? If the answer is even one, then I'd like to know what makes that person unique."
System Changes
Although the majority of ODAC members and FDA seemed to feel that the agency has lowered its approval standards for oncology drugs to the detriment of patients, many cancer patients would disagree with this view. FDA's decision to revoke its accelerated approval for Genentech's Avastin in metastatic breast cancer due to its limited effect on survival has raised an outcry among patient groups and among some members of Congress.
"The fact that the FDA is saying that overall survival is the only thing that matters is absurd," Tenenbaum said, noting that the fact that Avastin reached late-stage studies suggests that some patients are responding to the drug. As such, Tenenbaum feels that industry and the FDA should shift their focus from trying to approve or fail drugs, to trying to figure out the disease that a patient has that the drug might work in.
"Don't kill the drug — just figure out what it's useful for. The future model isn't going to be that this drug is going to be useful for a large number of patients, but it's likely that it's going to be useful in a small cohort in combination with other drugs," Tenenbaum said.
Tenenbaum is a melanoma survivor who participated in a trial for an investigational drug that eventually failed late-stage trials. The sponsor, to minimize legal repercussions, destroyed all available supplies of the drug, even though some people in the trial were benefitting from it. "Does that make any sense?" Tenenbaum posited. "The system is wacko."
In Avastin's case, Genentech has pharmacogenomic data suggesting that certain patients may have a better response than the general breast cancer population. The company, however, only proposed to further study these markers after FDA threatened to remove the drug from the market (PGx Reporter 01/26/11).
According to Tenenbaum, pharmacogenomically guided treatments are unlikely to make it through the regulatory system using FDA's randomized trial gold standard. "Cancer is hundreds of unique diseases … There's no bloody way you're going to have enough dollars or time to populate the whole matrix of randomized trials in the way that FDA would like to see it," Tenenbaum said.
He proposes instead that sponsors conduct small pilot studies to show that a drug might work off-label or in a pharmacogenomically targeted population, and then work out a "clear expectation" of the types of outcomes FDA wants to see.
Tenenbaum's pet project, Cancer Commons, is an effort to change the current process by which cancer patients receive drugs and research is conducted. The web-based system aims to link physicians, researchers, and patients to collectively figure out which drug works for which patients based on their molecular subtypes.
Cancer Commons' editorial board develops molecular disease models that try to match patients with the most validated tests, treatments, and trials based on a particular molecular cancer type. After applying a model and putting a patient on a particular drug, doctors and researchers can then report their observations on patient outcomes back to Cancer Commons.
The program is currently focused on melanoma. Tenenbaum noted that Cancer Commons plans on speaking to drug developers with late-stage investigational drugs to include information about their drugs in the program. By the end of the year, he hopes to expand the program into breast and lung cancer.
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at tray [at] genomeweb [.] com.