NEW YORK (GenomeWeb) – The twists and turns leading to the simultaneous review and approval of AstraZeneca's PARP inhibitor Lynparza and Myriad Genetics' BRACAnalysis CDx tell the story of a pharmaceutical company trying to develop a drug in the era of cancer genomics and a diagnostic firm advancing a companion test in an uncharted regulatory landscape.
The FDA ended 2014 with the approval of Lynparza (olaparib), a treatment for advanced ovarian cancer patients who have germline BRCA mutations and have received three or more prior lines of chemotherapy. At the same time, the agency approved Myriad's BRACAnalysis CDx for gauging which patients have germline BRCA mutations and therefore would likely benefit from Lynparza. Representatives from AstraZeneca, Myriad, and the FDA discussed the development and review process for these products at a meeting hosted by the Drug Information Association last week.
The approval of BRACAnalysis CDx marked the first time that a laboratory-developed test (LDT) went through the premarket approval (PMA) process and the first time an LDT was approved as a companion test. Since 1996, Myriad had been performing BRACAnalysis as an LDT that gauged the risk of hereditary breast and ovarian cancer.
The approval of the BRACAnalysis CDx comes as the FDA is seeking to regulate all LDTs. The message in FDA's LDT draft guidances and its final guidance on companion diagnostics is that a diagnostic that is essential for the safe and effective use of a drug would be a high risk test that requires premarket review. Although certain aspects of the FDA's LDT guidance is highly controversial among lab industry players, the agency has been talking about regulating such tests for years and Myriad has been paying attention.
Within Myriad, company leaders began thinking about how prepared the company was to meet FDA oversight when the time came, Russel Henderson, a project manager at Myriad, said at the meeting. The Lynparza development program offered Myriad a chance to demonstrate how a lab might take an LDT through FDA approval.
In doing so, Myriad encountered head on some of the questions that labs have with FDA regulation. Within the lab testing process, what is the device that would be regulated by FDA? How burdensome will FDA's quality system regulations be? After getting approval or clearance for an LDT, what types of tweaks or changes will trigger additional FDA submissions?
Meanwhile, AstraZeneca's experience developing Lynparza points to the difficulties of drug development in the genomics era. As new markers and pathways come to light, fine tuning, and sometimes upending, researchers' understanding of disease biology, drugmakers must figure out quickly what that means for ongoing drug programs. The patient population that the firm originally thought would receive the drug may not remain the same or as big. And increasingly pharmas need tools to home in on the population that will receive their drugs.
This demands a certain level of flexibility from the FDA, too. Not only does the agency have to ensure the safety and efficacy of drugs and tests, it has to do it in a way that keeps up with advancing science and the latest technological innovations.
Identifying responders
Lynparza came to AstraZeneca through its acquisition of UK spinout KuDOS in 2006, while the drug was in early studies. As it progressed through clinical studies and AstraZeneca came to better understand the drug's mechanism, it became clear that some patients responded particularly well. The challenge was how to identify these patients.
Olaparib works by binding to PARP enzymes and preventing them from repairing DNA breaks. This type of DNA damage can lead to tumor cell death, except cells can continue to proliferate by repairing DNA breaks through the process of homologous recombination.
BRCA genes are crucial to homologous recombination, but when patients have mutations in these genes the repair mechanism is broken. And so, ovarian cancer patients with BRCA mutations respond well to Lynparza because their DNA repair process is hobbled to start. When they receive the PARP inhibitor, their tumor cells die because they are overwhelmed by DNA damage.
We anticipated, though, that many people may have taken our approval … as going against the advice of the ODAC.
Early in Lynparza's development, the importance of defective homologous recombination on treatment response was evident, and as early as 2004 there was data suggesting BRCA mutations could be predictive for response to PARP inhibitors. But the best way to identify responders wasn't clear. In the Phase II Study 19 comparing Lynparza against placebo, AstraZeneca tried to enrich for patients who were sensitive to platinum-based treatments because they appeared to be deficient in homologous recombination. (Myriad is currently fine tuning such a test, called myChoice HRD.)
The sponsor was even working on an assay to measure homologous recombination repair deficiency as a tool for identifying best responders to Lynparza. By the time researchers were ready to unblind Study 19 and analyze the primary endpoint, however, the assay hadn't advanced enough to be taken through regulatory review. At the DIA meeting, Maria Orr, a diagnostic team leader at AstraZeneca, noted that the firm was simultaneously looking at gene arrays and immunohistochemistry testing, but those also didn't pan out.
"So, it was decided that the safest population to define was the BRCA-mutated population," Orr said. AstraZeneca had initially determined BRCA mutation status in just under 100 of 265 patients in Study 19. Then, when the progression-free survival benefit appeared particular robust in this subset of patients, researchers went back to test additional blood and tumor samples in 96 percent of participants.
The agency granted AstraZeneca priority review based on Study 19, but requested experts from its Oncologic Drugs Advisory Committee to weigh in on the data. Although Study 19 showed Lynparza improved progression-free survival over placebo, ODAC was concerned about the lack of survival benefit seen in the study and the reliability of the treatment effect seen in the BRCA mutation-positive subset.
Studies had also shown an increased risk for myelodysplastic syndromes and acute myeloid leukemia following treatment with a PARP inhibitor. Committee members had reservations about putting patients in the maintenance setting — where patients would receive Lynparza with the aim of keeping their cancer from coming back — at risk for incurable leukemias.
Furthermore, Gwynn Ison, an FDA reviewer involved with the Lynparza application, noted at the meeting that ODAC was worried that granting accelerated approval based on Study 19 would hinder accrual for the confirmatory SOLO2 trial, which would provide more definitive data on the drug's safety and efficacy in the BRCA mutated population. So, the committee voted 11 to 2 against the accelerated approval for Lynparza.
Subsequently, AstraZeneca and the FDA had extensive discussions and identified an alternative accelerated approval pathway, focusing primarily on data from another trial, Study 42. This single-arm study was investigating Lynparza in around 300 patients with multiple tumor types, who had received three or more lines of prior chemotherapy, and who had germline BRCA mutations. Based on this, the NDA was amended and the PDUFA data was extended to Jan. 3, 2015.
Among the evaluable ovarian cancer patients in this study, there was a 34 percent overall response rate and the duration of response was 7.9 months. "What we ended up doing was approving the drug in this more refractory population," Ison said. "We anticipated, though, that many people may have taken our approval … as going against the advice of the ODAC."
But this wasn't the case, in Ison's view, because after factoring in ODAC's concerns, the FDA didn't approve Lynparza in the maintenance setting. Instead, the agency decided to approve it in a refractory population based on Study 42, which showed the drug to have better efficacy in the fourth-line setting but a similar toxicity profile compared to other available therapies.
Another important point, Ison noted, was that by the time FDA approved Lynparza based on Study 42, the SOLO2 trial was finished enrolling patients. "So, there was no longer a concern about hindering accrual of that confirmatory study," she said.
Defining the device
The Rx/Dx collaboration between AstraZeneca and Myriad came together in 2009, after the FDA's device division informed the drug firm during a Phase II meeting that Lynparza would require a companion test. According to the FDA, an LDT wouldn't suffice, and Myriad's BRCA test would be a Class III, high-risk device requiring a PMA.
In August 2013, Myriad received investigational device exemption for BRACAnalysis to be used in two Phase III studies, which began in September. At that time, the PMA was planned to track with the Phase III studies slated for completion over the next two to three years. But later that year, AstraZeneca and FDA agreed that the drug would be reviewed at an accelerated clip based on data from the Phase II Study 19, which set the PDUFA date for Oct. 3, 2014.
"That threw us in a spin at Myriad," Jolette Franco, regulatory affairs manager at Myriad Genetic Laboratories, said at the meeting. "Now we had gone from being able to take two to three years, to having to submit a PMA on a laboratory-developed test in about a year." Pursuing the modular PMA track, Myriad submitted four modules — the first in December 2013 and the last in September 2014. But because the end goal is to get the drug and test on the market together, Myriad had to start new PMA modules as it was fixing deficiencies in prior ones to stay on track.
After the firm filed its first module, the FDA told Myriad to go back and reevaluate what it considers a device. "We thought it was the whole thing, from beginning to end, our entire laboratory process," Franco said.
Previously, labs have taken kits through FDA approval, even though they may have been originally performing these tests as LDTs. Kits can be performed at multiple labs. But Myriad chose to take its LDT through a PMA, which meant that after FDA approval, it would be performed only at Myriad's lab in Utah.
"All of our tests are operated as a laboratory-developed test. It's not within Myriad's business model to launch simplistic tests," Henderson said. "We focus on the more complicated, technically challenging assays. For that reason we maintain those in house as lab-developed tests."
An FDA official involved with the PMA said at the meeting that given all the pieces of instrumentation, internal validation, and variant classification expertise involved in performing BRACAnalysis CDx in the lab, "it would not be practical" to offer it as a kit.
In fact, Myriad has been performing BRACAnalysis in its lab, certified under the Clinical Laboratory Improvement Amendments and accredited by the College of American Pathologists, for around two decades. In that time, the lab has performed more than 1.5 million BRACAnalysis tests. Its lab and quality systems are inspected every other year and each of its LDTs are proficiency tested every six months.
When you get approval for a PMA, your life is not done.
Myriad had no shortage of data on its LDTs, but FDA regulates devices. Taking an LDT through the PMA process was uncharted territory, and for Myriad it would be even more complicated because BRACAnalysis would accompany a drug.
Myriad wasn't used to thinking about its lab testing procedures as devices. "We had to figure out what our device was," Franco said. "When a clinical laboratory develops its own test and uses it, in no way do they view what they're developing as a device or as a kit. It's a lab test, right?"
To get over this hurdle, Franco engaged her team in a thought experiment. They imagined a box and all the things they would put inside it in order to develop a kit. They decided to put in the box formulated reagents, Myriad-developed customized software, critical equipment, the variant classification process, and instructions for use. These are the things, Franco said, a clinical lab "manufactures" in order to be able to run its own test, and this distinction between manufacturing and the lab process was critical to defining the device.
"The light kind of came on," Franco recalled. "Once we figured out our device, life was good."
Myriad also has 15,000 variants in its database, of which 4,000 are deleterious or suspected deleterious. Since it would be impossible to test all of them in a regulatory context, Myriad had to come up with a representative set of variants and samples that had those markers to demonstrate the analytical capabilities of the CDx.
The firm also separated its CDx efforts from its commercial lab where it tests around 1,500 samples a day, and launched a dedicated lab for developing and running the companion test according to FDA standards. "This was a good strategy because it would isolate our CLIA, high-throughput test that's broad in its application … and compliance [for the CDx] would be a lot easier obtained if we had an independent laboratory," Henderson said.
Meanwhile, on the drug development side, ODAC had recommended against accelerated approval for Lynparza as a maintenance therapy. After the ODAC meeting, FDA and AstraZeneca decided to amend the NDA and this resulted in a PDUFA extension to Jan. 3, 2015. Ultimately, Lynparza and BRACAnalysis CDx were approved on Dec. 19, 2014.
Was it worth it?
The PDUFA extension was what Myriad was most happy about, Franco said. Reena Phillips, director of the Division of Molecular Genetics and Pathology within FDA's device division, admitted that the agency's reviewers also struggled with the pace of the applications and welcomed the additional time. The total review time for the drug was 319 days, but only 86 days for the CDx.
"There were some uncertainties in the beginning because it was an LDT, but Myriad was very cooperative," Phillips said. "They had the background of offering [BRACAnalysis] for 20 years. So, we took that also into consideration when we were making some decisions."
In the end, Franco said it took hundreds of thousands of man hours to submit the PMA. "It was a lot of work. [It involved] many hands, many resources, a lot of money, and lots of testing," she reflected.
Even after getting FDA approval, there is work to be done. AstraZeneca has to submit data from confirmatory studies, and this is the first time the FDA approved a CDx with post-approval conditions. The agency had previously approved companion tests with specific variants, but it approved BRACAnalysis CDx based on a representative set.
As such, in the post-market setting, Myriad has to provide updates so the agency can gauge how often classifications change and how often de novo variants are detected. Myriad also has to conduct additional validation testing to cover a larger spectrum of variant types and submit data from confirmatory trials.
"When you get approval for a PMA, your life is not done," Franco said. "You have an approved device and now you have to look at every single thing that happens to that device." Clinical labs make changes every day to their testing processes, she noted, but all changes now have to be evaluated to determine if it impacts the safety and effectiveness of the device. If it does, then the lab has to file PMA supplements to the agency.
Franco was recently asked whether all this effort was worth it. She admitted that "financially, it will be a few years before it will be worth it." But by engaging in the drug/diagnostic collaboration with AstraZeneca, Myriad played a part in bringing a drug to market for a population of women who don't have many choices. "It's a phenomenal thing to be a part of," she said.