Complexities of Precision Medicine: Managing Multiple Rx/Dx Combos in the Same Class

What happens when multiple targeted therapies are approved by FDA, each with a unique companion diagnostic for the same biomarker? The health care system is ill-equipped to deal with that scenario, putting patients at risk if the wrong test is used to guide a treatment decision. Industry has one solution: an analytical comparison of various diagnostic assays, starting with the anti-PD-L1 immunotherapy class.

It may be one of the few downsides of personalized medicine.

The Food & Drug Administration and other stakeholders are concerned that the pharmaceutical industry’s interest in developing targeted therapies – like those that target the PD-1/PD-L1 pathway to treat a variety of cancers – could lead to a complicated post-marketing environment with multiple drugs approved in the same class, each paired with a different diagnostic test for the same or a closely related biomarker.

With at least a half-dozen anti-PD-1/PD-L1 targeted therapies either or approved or in late stage clinical trials, each with its own companion diagnostic assay, FDA says that scenario is a distinct, near-term possibility. And while that has few implications from a regulatory perspective, it has the potential to create confusion between physicians and testing laboratories – not to mention pushback from payors – in the post-market setting.

The first two programmed death receptor-1 immunotherapy products were approved without companion diagnostics: Merck & Co. Inc.’s Keytruda (pembrolizumab) for advanced melanoma and Bristol-Myers Squibb Co.’s Opdivo (nivolumab) for melanoma and an initial indication in squamous non-small cell lung cancer. (Bristol also has a deal with Dako AS to develop a PD-L1 companion diagnostic for Opdivo in non-small cell lung cancer).

But many drugs in the next wave of programmed-death checkpoint inhibitors, which target either the PD-1 protein receptor or the PD-L1 ligand, are being developed with a distinct companion assay. And that presents the potential for a complicated post-market situation where there are multiple diagnostic tests for multiple members of the same class of drugs.

“It presents a fairly significant problem. We haven’t reached it yet...but we think it could still be a problem in this particular area,” Elizabeth Mansfield, PhD, FDA’s deputy director for personalized medicine in the Center for Devices and Radiological Health’s Office of Radiological Health, said during a March 24 public workshop convened to discuss the problem and identify potential solutions.

It is a situation FDA acknowledges it didn’t foresee when it issued its final guidance document on companion diagnostics in August 2014. But the issue itself is not regulatory in nature. FDA already has an established process for parallel reviews of a drug and companion diagnostic. Rather, stakeholders predict serious problems in the post-market setting in ensuring that the right test is conducted to guide use with the right therapy.

The health care system cannot easily accommodate a slew of different companion diagnostics for the same biomarker to direct treatment with multiple therapies in the same drug class. And the solution isn’t clear: using multiple assays to determine the best course of therapy for each patient is inefficient, costly and impractical. But finding a single, universal test is also unrealistic, given that the therapies have different modes of action, intended use populations and safety and efficacy profiles.

It is worth noting that the worst-case scenario may not be realized—or, at least, not as quickly as some fear. Of the PD-1/PD-L1 therapies in late-stage development for indications like non-small cell lung cancer and melanoma, it is possible that some (or all) may be approved by FDA without a requirement for companion diagnostic. Indeed, some stakeholders at the workshop questioned whether PD-L1 is a necessary biomarker and if other, better biomarkers exist.

But for now, FDA and industry stakeholders are trying to get ahead of what may be a tricky post-approval situation. The family of programmed-death immune checkpoint inhibitors is by far the most active area of immunotherapy development. But the problem is not isolated to the anti-PD-1/PD-L1 class. Similar cases in other classes are coming, FDA says, so how industry reacts will be a bellwether for what is to come.

Increased Interest in Targeted Therapies

FDA convened the workshop, “Complexities in Personalized Medicine: Harmonizing Companion Diagnostics Across a Class of Targeted Therapies,” to find solutions before the next wave of PD-1/PD-L1 therapies are approved for marketing. The workshop was co-sponsored by the American Association for Cancer Research and the American Society of Clinical Oncology and brought together the drug and device industries, payors, patient groups, oncologists and pathologists.

Attendance at the workshop illustrated the enormous interest in the issue. Much to FDA’s surprise, space in the meeting filled up quickly, and the agency was forced to turn away many participants. The workshop was attended by 200 individuals, with another 250 listening online.

Given the dramatic increase in biomarker-targeted drugs, the post-market scenario outlined during the workshop may have been inevitable. In the early 1990s, just 5% of applications approved by the agency were targeted therapies, CDRH Division of Molecular Genetics and Pathology Director Reena Philip, PhD, noted during the workshop. By 2013, nearly half, or 45%, of new drug approvals were targeted therapies.

And thanks to the “Breakthrough” therapies pathway, targeted drugs will get to market faster than ever before. Eighty percent of Breakthrough-designated drugs and biologics are targeted therapies, and they spend two years less in development than applications that do not carry the designation, Philip said. They are approved faster too: Opdivo, for example, was reviewed in just three months for its first lung cancer claim. (Also see "Opdivo Lung Cancer Approval Exceeds Rosiest Expectations" - Pink Sheet, 4 Mar, 2015.)

It’s important to note that pairing a targeted therapy with a companion diagnostic is nothing new. There are multiple examples of successful, straightforward uses of companion diagnostics that are relatively easy to manage in the post-market setting.

For example, Pfizer Inc.’s Xalkori (crizotinib) is approved for metastatic ALK-positive non-small cell lung cancer as detected by Abbott Laboratories Inc.’s Vysis ALK Break Apart FISH Probe Kit. In colorectal cancer, a single test from Qiagen NV (therascreen KRAS RGQ PCR Kit) is approved for use with both Bristol-Myers Squibb Co./Eli Lilly & Co.’s Erbitux (cetuximab) and Amgen Inc.’s Vectibix (panitumumab).

But things get a bit trickier when multiple tests have been approved for more than one drug, as in the case of BRAF V600E mutation-position melanoma. Genentech Inc.’s Zelboraf is paired with Roche’s cobas 4800 BRAF V600 Mutation Test, while BioMerieux Inc.’s THxID BRAF kit is used with GlaxoSmithKline PLC’s Tafinlar (dabrafenib) and Mekinist (trametinib).

An Unprecedented Post-Market Scenario

The anti-PD-1/PD-L1 market has the potential to be significantly more complicated, given the attention on the class and the number of drugs in development. “For the first time, we are facing multiple drugs, multiple tests and multiple indications,” FDA’s Philip said. Eight products are in development from six different pharmaceutical companies, according to the agency. (See Exhibit 1.)

“This is a unique scenario where each anti-PD-1 or PD-L1 therapy has its own personalized assay,” FDA’s Philip said. While similar in some respects, none of the assays are exactly the same – they have different clinical cutoffs, different assessment and scoring methods, different biopsy requirements, and different staining protocols and platforms.

With multiple drugs in late-stage trials, non-small cell lung cancer is expected to be the first major indication that will face this scenario. Gideon Blumenthal, MD, lung cancer team leader in FDA’s Office of Hematology & Oncology Products, said he has been in the “eye of the storm,” with “countless” meetings with drug sponsors. (Blumenthal noted that the “storm” analogy was not meant to be a pessimistic description.)

PD-1/PDL1 Targeted Therapies in Clinical Development

Exhibit 1

Drug	Target	Status
Bristol-Myers Squibb’s Opdivo (nivolumab)	PD-1	Approved for second-line squamous NCLC March 4. Additional indications are in development.
Merck’s Keytruda (pembrolizumab)	PD-1	Approved for advanced melanoma in September 2014 and has “Breakthrough” status in NSCLC; filing planned for mid-2015. Bladder and head and neck cancers are in Phase III. Gastric cancer is in Phase II.
Medivation/CureTech’s pidilizumab	PD-1	Development ongoing in a variety of oncology indications, including Phase II studies in colorectal cancer and melanoma.
MedImmune/AstraZeneca’s MEDI0680/AMP-514	PD-1	Phase I in advanced malignancies.
Genentech/Roche’s MPDL3280A	PD-L1	“Breakthrough” status granted for NSCLC (Phase II) and bladder cancer (Phase III). Filings expected in 2015-2016.
MedImmune/AstraZeneca’s MEDI4736	PD-L1	Filings in non-small cell lung cancer expected in 2016-2017. Phase I/II studies in solid tumors are ongoing.
Bristol-Myers Squibb’s BMS-936559/MDX-1105	PD-L1	Completed Phase I studies but focus is now on virology indications like HIV and hepatitis.
Pfizer/Merck KGAA’s avelumab (MSB0010718C)	PD-L1	Two Phase I/II clinical trials: 1) multiple tumor types and 2) non-small cell lung and ovarian cancer. A Phase II study is running in m-Merckle cell carcinoma patients.

Food & Drug Administration, Prevision Policy LLC, company reports

The problem is multi-faceted. First there is the potential for miscommunication between the physician and the testing laboratory in ensuring that a patient’s biopsied material is tested using the correct assay for the intended course of treatment. That alone has the potential for significant confusion once a plethora of drug-device combinations hit the market.

Furthermore, given that each companion diagnostic is specific to its partner therapy, a series of tests would need to be conducted to find the best course of treatment for each patient – a proposition that stakeholders say would be impractical, time-consuming and expensive.

Nancy Roach, chair of the patient advocacy group Fight Colorectal Cancer, pointed out that the testing facility may eventually run out of cancer cells, requiring additional biopsies to get more, and delaying treatment. There is also the risk of patients comparing hurdles to different tests and trying the easier tests first. And, of course, there is the question of how many diagnostic tests payors will cover. For now, all too often the answer is, “none.” (Also see "Personalized Medicine Meets Payment Policy: MEDCAC Has “Low Confidence” In Most Cancer Tests" - Pink Sheet, 2 Apr, 2015.)

But finding a single, universal diagnostic is likely impossible, at least for now. Each agent “is specifically tailored and informed by the clinical experience” of the drug company, and “each molecule, while addressing the same pathway, has unique pharmacological properties,” Bristol-Myers Squibb VP-Development, Oncology & Pharmacodiagnostics Steve Averbuch, MD, said. “Each development program is different, and each patient population – even within the same tumor type – as defined by the inclusion criteria is different.”

Doug Ward, VP-companion diagnostics at Roche's Ventana Medical Systems Inc., agreed, stressing that the tests are not designed to be interchangeable between therapies. “It’s a whole system approved, not just a test,” he said. “And we have instruments and reagents that are all a part of that system.”

Genentech’s Head of Oncology Biomarker Development and Companion Diagnostics Ian McCaffery, PhD, was the most optimistic that industry will ultimately find a “best test” for a specific biomarker like PD-1/PD-L1. “We will look back on this workshop and see this as a seminal event,” he said. The question is “what is that timeframe, and how do we move there?”

Is PD-L1 Expression a Good Biomarker?

Further complicating matters is that some oncologists are questioning whether PD-L1 expression should be used to guide oncology treatment, given the percentage of patients with tumors that test negative for the gene but still respond to treatment.

Suzanne Topalian, MD, director of the melanoma program at Johns Hopkins Kimmel Cancer Center, noted that depending on the tumor type, as many as 20% of patients that test negative for PD-L1 expression still respond to treatment. In Topalian’s specialty, for example, multiple cell types in the melanoma tumor microenvironment can express PD-L1, which can lead to a high percentage of false-negatives.

Those findings have prompted oncologists to argue that it could be unethical to limit treatment with new immune checkpoint inhibitors only to those who test positive for the PD-L1 biomarker. That argument was a common refrain from clinicians at the American Society of Clinical Oncology’s annual meeting in 2014. (Also see "Should PD-L1 Expression Be Used As A Biomarker For Immunotherapy?" - Pink Sheet, 7 Jul, 2014.)

At the March workshop, FDA reviewer Blumenthal acknowledged that the utility of PD-L1 as a biomarker is an “open question.” Some “key questions” include whether it is a necessary biomarker, and how (if at all) companion assays will be used in the clinic. Blumenthal also asked whether there are other, better predictive biomarkers, joking that the question could be the subject of next year’s workshop.

Sponsor response on the issue has been mixed. Merck has acknowledged that “tremendous pushback” from oncologists convinced the company that it could not use a companion diagnostic to limit therapy with Keytruda. “I can't personally support the idea of a companion diagnostic that would limit therapy with pembrolizumab,” Merck’s head of R&D Roger Perlmutter said at ASCO last year.

But Genentech has dismissed those concerns, arguing that drug development in the absence of predictive tests is a “fool’s errand.” Patients that test negative for PD-L1 expression will still be treated, the company has said, but they may require a more aggressive course of treatment. A PD-L1 diagnostic test is still very much a core part of development of Genentech’s MPDL3280A for bladder cancer. (Also see "Genentech’s Mellman: Moral Conflict On PD-L1 Testing Is “Red Herring”" - Pink Sheet, 7 Jul, 2014.)

There is the potential that some or all of anti-PD-1/PD-L1 therapies will be approved without the requirement for companion diagnostic. In talking about the use of companion diagnostics in oncology, FDA Office of Hematology & Oncology Drug Products Richard Pazdur, MD, has stressed that the assay must be truly essential to guide treatment. (Also see "Targeted Cancer Therapies in Context: FDA’s Pazdur on Companion Diagnostics" - Pink Sheet, 26 Sep, 2012.)

If the test doesn’t accurately predict the best population for a targeted treatment, sponsors run the risk of having their drug-device combination rejected by FDA (see: ChemGenex Pharmaceuticals Ltd.’s omacetaxine/Synribo). On the other hand, if an agent is effective in both the biomarker-positive and -negative populations, FDA might deem the diagnostic unnecessary (see: Ariad Pharmaceuticals Inc.’s ponatanib/Iclusig). (Also see "A Tale Of Two Targeted Therapies" - Pink Sheet, 26 Sep, 2012.)

Industry Blueprint Offered as One Solution

In an effort to frame the issue, an industry coalition of representatives from AstraZeneca, Merck, Bristol-Myers Squibb and Genentech-Roche developed a blueprint proposal for companion diagnostic comparability and presented the document at the workshop. While just three pages of text hammered out by the coalition over five weeks of conference calls, the blueprint was nonetheless well received by stakeholders.

The blueprint first acknowledges the problem: “The matrix of therapeutics and companion diagnostics, if each therapeutic were approved in conjunction with a companion diagnostic, may present a complex challenge for testing and decision making in the clinic, potentially putting patients at risk if inappropriate diagnostic tests were used to make treatment decisions.”

Conducting a separate diagnostic test for each individual drug is not practical, the blueprint says, nor is developing one test for all therapies possible, given that the individual therapies have different modes of action, intended use populations and safety and efficacy profiles. Furthermore, trying to align different development programs to adopt a single test could slow down development.

The industry blueprint recommends engaging an independent third party to conduct an analytical comparison of the various diagnostic assays in the anti-PD-1/PD-L1 class, “potentially paving the way for post-market standardization and/or practice guideline development as appropriate.” The analysis would start with PD-L1 assays from Dako and Ventana Medical Systems.

The blueprint can be seen as a first step toward resolving the issue, but it was clear from the workshop discussion that a solution may be a long way off. The industry coalition has not yet considered next steps, although presumably one would be determining the appropriate third party to conduct the assay analysis.

The complexity of harmonizing companion diagnostics may be a commercial challenge for sponsors in the class, but it has an offsetting commercial advantage: It will complicate any effort by payors to treat anti-PD-1 therapies as interchangeable in efforts to extract price concessions.

Those dynamics may further complicate efforts to reach a solution in the near term. Even without those commercial realities, the main stakeholders on the issue – drug developers, clinicians and laboratories – are somewhat disparate and don’t always agree.

One main message from industry: given the near-term approval of the next wave of anti-PD-1/PD-L1 targeted oncology treatments, “there isn’t time to fix everything and make it perfect.” Company executives also stressed the importance of open collaboration among all stakeholders, and the need to work together to find solutions, and not “throw away” the work that has already been done.