NEW YORK (GenomeWeb) – The US Food and Drug Administration has been discussing with industry a new category of tests, dubbed complementary diagnostics, for guiding treatment strategies and identifying best responders to drugs.
On one hand, the emergence of this category suggests a growing sophistication within industry when it comes to personalized medicine in an increasingly crowded marketplace of molecularly targeted drugs and tests. On the other hand, the regulatory nuance may not help doctors who have to decide whether or not to test a patient before prescribing a drug, and may be lost on payors who decide test coverage based on medical necessity.
In October 2015, the FDA for the first time approved a complementary diagnostic, Agilent subsidiary Dako's PD-L1 IHC 28-8 pharmDx, as a tool to identify non-squamous, non-small cell lung cancer patients who might benefit most from Bristol-Myers Squibb's Opdivo (nivolumab). Last month, the agency expanded the intended use for the complementary test to advanced melanoma patients considering treatment with Opdivo in combination with Yervoy (ipilimumab) or just Opdivo.
The FDA doesn't consider complementary testing necessary for the safe and effective use of a drug like it does companion diagnostics. In the case of Opdivo, the drug's indication isn't restricted to a biomarker-defined patient population. So, one could argue that regardless of a patient's PD-L1 expression status, "Wouldn't you give the drug anyway?" said Elizabeth Mansfield, deputy office director for personalized medicine at FDA's device division.
"But now that there are a growing number of choices in the way that you might want to treat a patient, also taking into account a patient's preference, for example, for the cost of the drug, if you have the test you might say, 'Well, you're not the group of people who is predicted to do better, then we might look at some other options for you.'" Mansfield told GenomeWeb. "It just gives additional information for physicians and patients to make decisions about therapy."
The FDA doesn't consider pricing in its regulatory decisions, but the pharmaceutical industry has come under fire from patient groups for the high price of targeted drugs. According to published reports, the price of a year's worth of the Opdivo/Yervoy combination for melanoma is more than $250,000, while Opdivo for NSCLC is priced at around $150,000. The other immunotherapy option on the market is Merck's Keytruda (pembrolizumab) also priced at around $150,000 a year, and approved with a companion diagnostic in the NSCLC setting. Although both drugs are projected to achieve multi-billion, blockbuster status by 2020, market analysts expect that in the short term Opdivo may have the market advantage since there is no required testing.
Anticipating more complementary tests to enter the market, FDA is crafting guidelines on the topic. Dako's PD-L1 complementary test required pre-market approval, but this may not be the regulatory bar in every situation. Drug and diagnostic companies expect the complementary category will comprise a wide range of tests for risk assessment, diagnosis, prognosis, monitoring, and guiding therapeutic decisions.
Based on analysis by consulting firm Diaceutics, nearly 75 percent of molecules in 14 top pharma pipelines last year were potentially associated with a biomarker, were being developed with a biomarker strategy, or could benefit from such a strategy. Pharma's growing investment in personalized medicine has resulted in the availability of multiple treatments for the same molecularly defined subset, with different FDA-approved companion tests and unapproved LDTs claiming to gauge the same analyte.
The growing complexity in the personalized medicine space is certainly not lost on the FDA. The agency held a meeting last year to discuss the challenge of having multiple tests for the class of anti-PD-1 and anti-PD-L1 drugs. Diaceutics CEO Peter Keeling anticipates that standardized testing in the immunotherapy setting is several years away. In the near term, regardless of whether or not testing is required from a regulatory standpoint, Keeling said drugmakers need to educate physicians and alleviate the confusion around biomarkers associated with drug response.
In the messy world of patient care, healthcare providers may not understand the scientific rationale for complementary testing as described in labeling, and instead view it as a "nice-to-have" option and not order it when it could benefit patients. Similarly, payors may not cover a test if they determine it isn't medically necessary.
You can create as much gray as you want, but there's a black-and-white decision to be made.
"I know what FDA is trying to do is show some flexibility here," said Girish Putcha, who is director of laboratory science at Medicare contractor Palmetto's MolDx program, but who spoke to GenomeWeb as an industry observer. "But at the end of the day there's a patient and a physician who's asking should the patient get the test and the drug," noted Putcha. "You can create as much gray as you want, but there's a black-and-white decision to be made."
A new category
Among pharmaceutical companies, particularly those with internal diagnostics groups, the term "complementary diagnostic" has been in common parlance for at least five years. Novartis, for example, was publicly discussing this category of tests in 2011. That same year, J&J created Janssen Diagnostics with an eye toward advancing diagnostic products that could "enable" its drugs.
"From the very beginning we were discussing this distinction between companion and complementary diagnostics," said Jami Taylor, senior director of global access policy at Janssen Diagnostics. "It encompassed a whole category of important products that simply wasn't receiving due attention in the larger personalized medicine dialogue and certainly not at the policy and regulatory level."
At the time, companion diagnostics were getting all the shine. The FDA released its draft guidance on the topic, and was pushing the idea of drug and test codevelopment and parallel approval. That summer the agency also approved the first personalized melanoma and lung cancer drugs with companion diagnostics.
Within Janssen Diagnostics there was early recognition of the advantages of complementary diagnostics, noted Peter Hoehn, who leads the group. Since companion diagnostics are strictly tied to a drug, in some ways it is easier to show their value to pharma, and relatively straightforward to manage adoption and reimbursement. "But we also realized that complementary diagnostics … could provide benefit to a drug or to a disease class, and that's really important, and could add incremental value to patients and for us," said Hoehn. Janssen Diagnostics is working on complementary diagnostics in a number of areas, including tests for earlier diagnosis of Alzheimer's patients so they can be treated before symptoms progress and for identifying patients more likely to respond immunology drugs.
Diaceutics' pharma clients are inking Rx/Dx collaborations earlier and thinking more about how biomarkers can influence who gets their drugs. "Ultimately, our clients are led by what the data say" about whether a test is necessary for their drug, Keeling said, but he added there are mixed opinions within pharma about complementary diagnostics. Some of Diaceutics' clients recognize diagnostics as increasingly important and want to actively embed them in drug programs. "Then there are clients that feel testing is still an impediment to access to their drugs, and they are seeking to find ways in which the drug can be prescribed without dependence on the test," he said.
"If we have information we want it to be available, but we're not sure how you should use it and we certainly don't have a lot of control over whether a doctor will use it."
When FDA approves a drug with a companion diagnostic, the drug is indicated in labeling for a biomarker-defined population, and the test's labeling notes its use to identify patients who should or shouldn't receive the drug. With Opdivo's approval in NSCLC and melanoma, however, the agency did not restrict it in any way for a biomarker-defined patient subset.
Although the indication for Opdivo doesn't mention the PD-L1 marker, the "clinical studies" section of the label features graphs displaying how PD-L1 expression is associated with outcomes in Opdivo-treated patients. Meanwhile, the role of PD-L1 is explained up front in the "intended use" portion of the complementary diagnostic's label. In the non-squamous NSCLC setting, the label for Dako's PD-L1 IHC 28-8 pharmDx describes it as a test for gauging PD-L1 expression, which "may be associated with enhanced survival from Opdivo." (The FDA hasn't yet posted the updated label with the test's expanded intended use in melanoma.)
Labeling that accurately describes the relationship between a drug and a complementary diagnostic is a challenging area to work through, according to Hoehn. "In some of the programs we've done we've had conversation with the FDA to really try to understand what would the diagnostic label be, and what, if anything, would go into the drug label," Hoehn said. "Would it be in the clinical trials section, would it change the indication, and what does that mean for our sales representatives' ability to promote it?"
Ultimately, labeling dictates how medical products can be marketed, and if the labels of the treatment and the complementary test place different emphasis on the role of the predictive biomarker this may hinder drug and diagnostic firm sales reps from communicating a unified message to doctors. In October, when Opdivo was first approved in the NSCLC setting, BMS noted the simultaneous approval of Dako's complementary test and explained that testing wasn't required. BMS' latest press release announcing the melanoma indication for Opdivo doesn't mention a complementary test or PD-L1 expression.
Given the complexity of the information with regard to PD-L1, "I would imagine it would be a challenge for the sales organization to really have a clear message around when you should test and why you need to test and the benefits of it," said Hoehn, adding that his group has had discussions with the FDA about what drug sales reps are allowed to say to doctors about complementary testing since it's not in the indication of the drug.
Sales reps have their work cut out for them in the immunotherapy space, where the FDA has also approved Keytruda in NSCLC, but with a companion diagnostic called PD-L1 IHC 22C3 pharmDx, also developed by Dako. In a recent survey, Diaceutics asked 30 US lung cancer specialists if they understood the differences in the companion and complementary PD-L1 test labels; 51 percent said they did and 34 percent said they didn't. However, according to Diaceutics' analysis, 15 percent of those who said they understood, incorrectly elaborated on their answer. Keeling suspects that a more detailed survey would likely reveal that a greater portion of doctors didn't catch the nuanced differences between the companion and complementary test labels.
Physicians aren't required to follow product labeling, since FDA doesn't regulate the practice of medicine, but labels do contain information about the safety and efficacy of a product. Historically, it hasn't been easy for doctors to figure out what, if any, action they should take based on biomarker information in FDA-approved drug labeling. In a 2014 study published in JAMA Internal Medicine, researchers from Harvard Medical School evaluated more than 100 drug labels with biomarker information, and found a minority of them had information on the clinical utility of testing for those markers. This, the authors concluded, may make it confusing for doctors to interpret labeling.
The way the FDA is thinking about complementary diagnostics, Mansfield expects "there will be a more utilitarian tie" between the test and the drug since they were developed together. "I know people have complained that in the drug labels, the FDA puts the information [about biomarkers] out there but we don't say how to use it," she said. "FDA's position on this is that … if we have information we want it to be available, but we're not sure how you should use it and we certainly don't have a lot of control over whether a doctor will use it."
We would like to avoid having four separate drugs for four different diagnostics that could be interpreted differently.
Even when a companion diagnostic is unequivocally required in drug labeling, not all doctors order testing for a variety of reasons, including lengthy turnaround times for results and difficulty getting a tumor sample. A global survey last year by drugmaker Boehringer Ingelheim found that 81 percent of newly diagnosed advanced NSCLC patients received testing for EGFR mutations, an improvement from an earlier survey that showed around half of patients were getting tested. However, 51 percent of surveyed oncologists indicated that patients' EGFR mutation status did not impact their therapeutic strategy.
Boehringer anticipated that the adoption of a companion diagnostic would take time, and launched the "Lets Test" education campaign to raise awareness of EGFR and ALK markers in NSCLC, several months before its NSCLC treatment Gilotrif (afatinib) was approved for patients with EGFR mutations. "If something is of strategic importance, we'd have to partner with companies to find ways to help with that adoption," Hoehn said.
A spokesperson for BMS didn't highlight any specific physician or patient education efforts around PD-L1, but told GenomeWeb that "while testing is not required" for the Opdivo/Yervoy combination in melanoma, "we understand healthcare professionals may find value in knowing a patient's PD-L1 expression status," and "we are committed to providing diagnostic tools which may help provide additional information."
Diagnostic firms are also taking action to educate patients and providers about testing in personalized medicine. Foundation Medicine, for example, has partnered with cancer advocacy groups to educate patients about comprehensive molecular screening, which gauges hundreds of markers at once, instead of one or a handful.
Industry observers expect the present confusion around companion and complementary testing won't be an issue when the one-drug-one-test model gives way to molecular panels that can gauge multiple markers at once and guide treatment strategies for entire drug classes. "We would like to avoid having four separate drugs for four different diagnostics that could be interpreted differently," Christopher Fikry, general manager Quest Diagnostics' oncology business, told GenomeWeb. "Ideally, we want the manufacturers and regulatory authorities to come together and collaborate on what the optimal diagnostic looks like that would service all the needs of the patient."
For now, he said most of responsibility falls on the diagnostic partner to make sure doctors are ordering the right test. "We're trying to build up as much infrastructure as we can, not just to support the tests, but to support the physicians in interpreting the results afterwards," he said.
Quest, which provides clinical lab testing services to half the physicians and hospitals in the US, was one of the first labs in the country to offer Dako's PD-L1 complementary and companion tests. The national reference lab has a vast test menu, but Quest also employs 700 pathologists and physicians, as well as a team of genetic counselors, to assist doctors.
"What we'll share with the client is what's in the FDA-approved label, help them understand if it's an absolute requirement or not, and then we talk through whether it would be appropriate for an individual patient," Fikry said. "It's as much art as it is science."
It's difficult to demonstrate utility of a test if there isn't one clear option to use once the test is run.
Value to payors
Putcha anticipates that the labeling language around complementary diagnostics may also be challenging for payors to interpret. "You've given me a [drug] label that says the diagnostic isn't necessary," he said in reviewing the Opdivo label. "It doesn't determine either selection or dosing of the drug. So how does a diagnostic manufacturer come back and make an argument for why this diagnostic should be paid for?"
When the FDA approves a drug with a companion test, payors generally cover the test. But in the case of complementary diagnostics, test developers may have to provide additional evidence around utility. Putcha explained that labs and manufacturers would have to address standard questions about who should be tested; how test results are supposed to change and whether they actually change patient management; and whether these changes improve health outcomes.
However, the incongruent way drugs and tests are reimbursed in this country can be a barrier for personalized medicine. When the FDA approves a drug for a patient population, payors pay for it in that context. In the case of Opdivo, because the drug is indicated for advanced melanoma and NSCLC patients, "I suspect payors would find it difficult to not pay for the drug, especially in the labeled population, with an understanding that some off-label use will probably still happen for a variety of reasons," Putcha said. "But the irony is that the same restrictions about on-label diagnostic use and reimbursement don't seem to apply to drugs, which is bizarre, since drugs are actually interventional and diagnostics merely guide the intervention."
Given these realities, the Opdivo label hamstrings payors and to some extent the diagnostic codeveloper, Putcha observed. "I think you have to look at that label [in NSCLC] and say [that] a payor would have to at least cover that drug in some patients who based on the sponsor’s own studies can be identified and very likely will have no survival benefit versus docetaxel," he said. "On top of that I shouldn't be paying for the diagnostic because you just told me you don't need it, either for drug selection or dosing."
According to Fikry, Quest to date hasn't had huge pushback from payors in terms of reimbursing complementary testing. "The examples we have today, these tend to be sicker patients on second or third line therapy, so there is a lower bar to get some of these things reimbursed," he said.
The looser regulatory ties between a complementary diagnostic and a specific drug may not be that much of a roadblock to reimbursement if the test has clinical utility. Medicare expert Bruce Quinn expects there will be some complementary diagnostics important for patient management that aren't tied to one drug, while other tests will garner FDA approval but won't be much better than cheaper alternatives on the market.
Quinn, senior director at the international law firm FaegreBD, pointed out that there are many tests, such as those for transplant organ and blood donor matching, which are necessary for making medical decisions but aren't companion diagnostics. "You prove the necessity to the payor by having a test that has no cheaper alternative and that is really needed for important management choices," said Quinn.
Still, in the personalized medicine setting, payors are used to thinking about test utility in the context of what drug is being prescribed. "It's difficult to demonstrate utility of a test if there isn't one clear option to use once the test is run," Ralph Riley, health economics and pricing leader at Janssen Diagnostics, told GenomeWeb.
Janssen Diagnostics, as part of a coalition within the European Personalized Medicine Association, is drafting a white paper to make the case that payors should consider broader patient and cost metrics when deciding coverage of complementary diagnostics. "There are benefits in this from a payor perspective as well," Riley said, but he noted that a lot more work needs to be done to help them recognize the value of complementary testing.