NEW YORK (GenomeWeb) – Four drugmakers have decided to collaborate on a study comparing two immunohistochemistry tests they are using to stratify patients in trials for anti-PDL1 immunotherapy treatments. The effort is aimed at informing healthcare providers about the analytical variability that exists between available companion tests that gauge the same analyte.
The study was discussed this week at a workshop on the challenges of standardizing companion diagnostics development when there are multiple drugs in the same class. The meeting was sponsored by the US Food and Drug Administration, the American Association for Cancer Research, and the American Society for Clinical Oncology.
In an effort to explore the challenges oncologists face in choosing which diagnostic to use in determining whether a patient will respond to a drug, meeting participants focused on real-life uncertainties the healthcare community will face when a number of new PD-L1 and PD-1 inhibitors come to market, each with their own companion tests. "For the first time we're facing the convergence of multiple drugs, multiple tests, multiple indications for the same analyte that are going on in parallel," said Reena Philip, director of the division of molecular genetics and pathology in FDA's diagnostics arm, at the meeting.
Although there are a number of possible CDx options on the market, "they are not all the same and we don't know how they compare with each other," Elizabeth Mansfield, deputy office director for personalized medicine in FDA's Office of In Vitro Diagnostics and Radiological Health, added at the meeting.
According to the FDA, a drug sponsor must develop a companion diagnostic when the test is essential for the safe and effective use of the drug, or in other words a CDx enables the drugmaker to meet FDA's safe and effective requirement by identifying a best responder population for a drug. The FDA last year finalized its companion diagnostics guidance, in which the agency reaffirmed its preference for simultaneous development of a drug alongside its companion diagnostic, but laid out a few exceptions when other development paradigms might be acceptable.
However, as more companies are advancing drugs with companion tests utilizing a precision medicine strategy, several problems have emerged. When there are multiple FDA-approved companion diagnostics for the same class of drugs, it becomes costly for hospitals to adopt all of them and confusing for doctors to figure out which test to order. Then there is the tissue problem. For certain malignancies, such as lung cancer, it is challenging enough to get sufficient tissue to run standard analyses for diagnoses, but usually not enough tissue material to run multiple, proprietary companion tests in order to figure out which drug in a class to prescribe.
Mansfield acknowledged that when drawing up the companion diagnostics guidance, the FDA didn't expect that there would be multiple drugs in the same class, each with its own companion diagnostic. She and others at the meeting characterized the problem as not so much a regulatory problem, but a downstream challenge for community oncologists who have to figure out the best test to use – amid time constraints, with limited understanding of genomics, and often with insufficient data comparing platforms – in order to determine if their patients will respond to a specific treatments.
To address some of the issues that might arise with the existence of multiple CDx for the same drug class, Philip said at the meeting that the FDA is working on a guidance for the development of me-too or follow-on assays. She highlighted a host of differences between companion diagnostics even though they aim to identify best responders to drugs in the same class.
However, the meeting was largely focused on differences in analytical validation – whether a patient has the marker of interest or not. When the specific cut off for determining marker positivity or negativity varies between companion tests, so does the intent-to-treat population for the drug. So, by using a test that hasn't been studied and optimized for use alongside a specific drug, oncologists might fail to identify those who might benefit from a treatment.
Toward this end, AstraZeneca, Roche/Genentech, Bristol-Myers Squibb, and Merck released a blueprint laying out aspects of a study comparing the analytical performance of investigational-use-only assays – developed either by Dako or Ventana Medical Systems – that they're using to test non-small cell lung cancer patients for PD-L1 expression in clinical trials involving their immunotherapies.
The study outlined in the blueprint by the four drugmakers will compare the analytical performance of Ventana's and Dako's IHC tests to gauge PD-L1 expression using an independent sample set that represents the target populations being studied by drug sponsors but that doesn't come from ongoing drug trials. The stained slides will be evaluated by the diagnostics firms and an independent third party. Finally, the findings of the study will be published and made public.
AstraZeneca is developing the PD-L1 inhibitor MEDI4736; Roche/Genentech is developing MPDL3280A, a PD-L1 inhibitor; BMS is advancing an anti-PD-L1 drug BMS-936559 and the PD-1 inhibitor Opdivo (nivolumab); Merck is developing a PD-1 inhibitor Keytruda (pembrolizumab). The FDA recently approved Opdivo in advanced melanoma and advanced squamous NSCLC; and Keytruda in advanced melanoma. These four firms are studying these immunotherapies in NSCLC with the help of an IHC test from either Ventana or Dako.
By FDA's estimation, overall between four and eight drugs targeting PD-1 and PD-L1 are currently being developed at different pharma companies for a range of cancer indications as single agents or in combination with other agents. None of the FDA approved anti-PD-1 drugs, designed to block PD-1 proteins on tumor cells, were approved alongside companion tests.
In situations where a CDx is being studied alongside the drug, however, FDA's Philip described a range of differences that might be possible among the tests. "Each PD-1 or PD-L1 therapy has its own proprietary PD-L1 IHC assay," she said. "Each therapeutic product is using a different assay in terms of different IHC antibody clones, different staining protocols, platforms, different clinical decision points, different tumor indications, different assessment methods [measuring PD-L1 expression on] tumor cells, tumor infiltrating lymphocytes, or both, [and] different scoring methods."
In discussing the blueprint, the drugmakers at the meeting also highlighted unique properties of their PD-1 and PD-L1 molecules: the different biological hypothesis, divergent development approaches, and distinct patient populations. Moreover, speakers noted that drugmakers that are developing companion tests for their immunotherapy programs are doing so with intellectual property and under contracts that hinder them from collaborating as openly with each other as would be needed for true CDx standardization.
The blueprint put forth by the drug companies is a long way from any sort of harmonization among industry players around the development of companion diagnostics under common standards or specifications and even further from developing a universal platform for drugs in the same class for the same indication.
Representatives from drug companies noted at the meeting that their blueprint was only a first step toward possible, future CDx standardization. At the same time, none of the companies wanted to engage in a way that would delay their drugs' commercialization. The analytical performance comparison, the companies wrote in the consensus document, was an effort to increase understanding of the analytical performance of the companion tests "without delaying pivotal studies and patient access to critical new therapies."
Dave Stanforth, head of R&D, companion diagnostics at Agilent, acknowledged that the drugmakers in developing the blueprint knew that they wouldn't be able to fully address the problem of multiple companion tests for the same drug class, but wanted to start from a basic level. He added that the companies chose to focus on comparing the analytical validity of IUO tests being used in clinical trials since those tests have the best shot of reaching the market. "The idea at the end was to deliver a data and information package so we can inform … the medical community about the current state of PD-L1 IHC testing and give you some idea of what's headed your way," Stanforth said.
The mechanism of PD-L1 and PD-1 inhibitors is complex, which adds to already challenging CDx development programs. PD-1 proteins bind to PD-L1 and PD-L2 ligands hindering the body's immune system from attacking cancer cells. The idea behind anti-PD-1 drugs is to block PD-1 and PD-L1 from interacting and boost the ability of T cells to attack cancer cells. Meanwhile, studies have shown that when PD-L1 is upregulated, certain tumor types are more aggressive and patients have worse prognosis. This has inspired a number of PD-L1 inhibitor drug development programs.
Companies studying their immunotherapies with companion tests are hoping to identify a best responder population. Genentech, which has breakthrough therapy status from the FDA for MPDL3280A in advanced NSCLC, conducted an early study that suggested that the level of PD-L1 expression was tied to drug response. A Phase I study in a heavily pretreated NSCLC population showed that 23 percent of patients responded to MPDL3280A regardless of their PD-L1 status determined by IHC. Among the patients whose tumors the IHC test found to be PD-L1 positive (IHC 2+ or 3+), 46 percent responded, while among those with the highest level of PD-L1 expression (IHC 3+), 83 percent responded.
However, as this study suggests, a subset of patients appear to respond even when they don't express PD-L1. This further complicates the decision for drug companies as to whether by developing a companion diagnostic for their drug they would limit patients who might benefit from getting the drug.
Some meeting participants pointed out that risk/benefit considerations might necessitate a companion diagnostic for a drug when it is used earlier in the disease paradigm, but perhaps not in later stages of advanced disease, when patients have fewer options. This is the strategy that Genentech has taken for its metastatic NSCLC drug Tarceva. The label for the drug requires testing for EGFR mutations in the first-line metastatic NSCLC setting, but not in later disease stages.
Highlighting the uncertainty that still exists in determining HER2 status for Herceptin, many meeting participants said this is the situation to avoid in the future. Herceptin was initially approved in 1998 alongside Dako's HercepTest, a semiquantitative IHC assay. Since then, Genentech has invested in launching a number of companion tests that employ IHC, FISH, and other technologies for Herceptin, but there is still discordance between labs in determining if a patient is HER2-positive or negative.
I would ask everybody who has LDTs on the top of their heads to just table that.
In the past studies have shown discordance rates as high as around 25 percent when samples deemed to be HER-positive by local labs using one testing method are retested by central reference labs by both IHC and FISH. A study published last year in Cancer and funded by Genentech found a much lower rate of discordance – 4 percent of 552 patient samples deemed to be HER2-negative when tested by one methodology in a local lab were found to be HER2-positive when retested at a central reference lab by FDA-approved IHC and FISH tests.
Current guidelines recommend that healthcare providers determine the HER2 status for all invasive breast cancer patients either by IHC or by FISH testing. The guidelines further state that labs performing HER2 testing show at least 95 percent concordance with another validated test for positive and negative assay values. Still, many healthcare providers feel there is no gold standard diagnostic method for determining best responders to the drug.
Mansfield and others at the meeting noted that the idea behind having the present discussion in the context of PD-1 and PD-L1 inhibitors is to spur the development of standards and guidelines for performing companion diagnostics early on and avoid the uncertainties experienced with HER2 testing. One idea put forth by multiple participants was the use of registries to track the drug/test combinations that work best in different clinical contexts. However, Mansfield noted that it has been challenging to keep such registries up to date with information on diagnostics.
In this regard, several efforts are underway, such as ASCO's Targeted Agent and Profiling Utilization Registry (TAPUR) and Medicare contractor Palmetto GBA's Molecular Evidence Development Consortium (MED-C). Both programs are exploring the use of registries to track patients receiving treatments, particularly off-label therapies, based on results from molecular testing.
LDTs: A discussion for another day
One topic that was off the table for discussion at the meeting this week was the regulation of lab-developed tests (LDTs). The agency has controversially released a draft plan to bring LDTs under its regulatory oversight, which would mark a shift from oversight by the Centers for Medicare and Medicaid Services' Clinical Laboratory Improvement Amendments.
"What this meeting is not is to discuss LDTs," Mansfield said at the start of the meeting. "I know that that's on a lot of peoples' minds. We've talked about that in other meetings and so whether this should be an LDT or not an LDT or whether FDA should regulate or not regulate is not the point of this meeting. The point of this meeting is what is the solution when we have multiple tests for multiple members of a class of drugs and how do we make that work in the community. I would ask everybody who has LDTs on the top of their heads to just table that and if it becomes a topic, then I will ask people to leave that topic and move on to the issues at hand today."
But the availability of LDTs, alongside FDA-approved companion diagnostics, is at the heart of the problem for oncologists. Hospitals, academic care providers, and large cancer centers usually develop their own LDTs for informing treatment decisions, and they maintain that their tests are well validated and more in line with the latest medical knowhow than FDA-approved tests that take much longer to come to market, and take even longer and additional investment to update. One presenter pointed out that while the FDA has approved a companion test that gauges KRAS mutations for identifying best responders to two colorectal cancer drugs on the market, there are currently 56 tests that can also gauge KRAS mutations listed in the McKesson Diagnostics Exchange, a catalog of molecular tests.
After the meeting, Maria Fe Paz, president of Labceutics, discussed this issue with GenomeWeb. "We believe that LDTs were excluded from the discussion mostly because of the FDA's discomfort in discussing this currently rather touchy subject in this particular context, and to avoid distracting the conversation from the PD-L1 [CDx] harmonization," she said. "But the fact is that the PD-L1 confusion among pathologists, oncologists, and laboratorians is actually often reverting back to the LDT versus IVD kit dilemma."
Labceutics – a subsidiary of personalized medicine-focused consultancy Diaceutics – aims to help pharma and diagnostics companies gain access to Europe's vast lab market and figure out how to use this untapped network to deliver tests to physicians and patients seeking to personalize treatments. "I guess everyone who knows a bit about the lab testing environment in the US … will agree on the likelihood of LDTs playing a very important role in the initial roll-out of PD-L1 testing," Fe Paz said.