Skip to main content
Premium Trial:

Request an Annual Quote

NCCN Launches Study to Harmonize PD-L1 Assays; Lack of Concordance Highlighted in Recent Study


NEW YORK (GenomeWeb) – The National Comprehensive Cancer Network announced this week that its oncology research program will collaborate with pharmaceutical company Bristol-Myers Squibb in a new study to better understand how different assays measuring PD-L1 expression correspond to one another, and how the expression of PD-L1 can be heterogeneous depending on how samples are collected and analyzed.

The effort will be co-chaired by Ignacio Wistuba of the University of Texas MD Anderson Cancer Center and David Rimm from the Yale Cancer Center.

Rimm, along with other colleagues from Yale and from the Cleveland Clinic, recently published a study illustrating how different assays for assessing PD-L1 protein expression can yield discordant or heterogeneous results, a significant roadblock for future clinical use of these tests to guide treatment with relevant immunotherapeutic drugs.

In the report, which appeared last month in JAMA Oncology, Rimm and his colleagues measured PD-L1 protein distribution in non-small cell lung cancer samples using both conventional immunohistochemistry and a quantitative immunofluorescence method they had previously developed. The group also compared the results obtained using two different PD-L1 antibodies, one of which has been used in companion diagnostic efforts by Ventana Medical Systems.

Overall, the results showed significant heterogeneity of PD-L1 expression within NSCLC tumors as well as clear discordance between the two antibodies used, the authors reported.

According to Rimm, this discordance illustrates a pressing clinical issue wrought by the cloistered development of these assays — that the assay for one drug cannot be expected to effectively predict response to others.

Rimm and his team's work in PD-L1 assay heterogeneity began with previous research, in which the group found that a variety of available PD-L1 antibodies were not actually valid.

Then, as pharmaceutical companies began development of their own separate tests for each of their own individual drugs in the space, the group turned to the question of trying to do some quantitative comparisons of the antibodies at the heart of these companion diagnostic tests.

"We don’t think it's practical for a pathology lab to do four different assays to ask a single question," he told GenomeWeb.

Aaron Hansen and Lillian Siu of the Princess Margaret Cancer Center echoed this sentiment in a viewpoint piece running in the same issue of JAMA Oncology as the Yale group's study, arguing that cloistered PD-L1 assay development has created problems down the line for researchers interested in directly comparing the efficacy of these different drugs, and has created a duplicative testing environment that could be onerous and confusing to clinicians.

"Siloed development of a companion diagnostic for a common biomarker without harmonization will lead to treatment and testing rigidity when an assay for one [agent] cannot be applied to another [agent] of the same drug class and for the same indication. This rigidity could lead to a more complicated biomarker testing process, which would ultimately affect treatment decisions and patient access to anti–PD-1/PD-L1 therapies," Hansen and Siu wrote.

"In the phase I department at Princess Margaret Hospital in head and neck cancer we have been involved in a bunch of different immune-oncology trials with PD-1 inhibitors, and as clinicians we saw this issue sort of emerging in front of us, that positive for one might not be positive for another," Hansen told GenomeWeb.

While initial drug labels for anti–PD-1 therapies did not require or recommend measurement of PD-L1 to guide therapy, this is changing now as these therapeutics have made their way into the lung cancer space where clinical trial data has demonstrated significant differences in response based on PD-L1 expression.

This October the US Food and Drug Administration approved two anti-PD-1 drugs in non-small cell lung cancer: Merck's Keytruda (pembrolizumab) and Bristol-Myers Squibb's Opdivo (nivolumab).

Alongside Keytruda the agency also approved an IHC test called PD-L1 IHC 22C3 pharmDx, developed by Agilent Technologies subsidiary Dako to identify NSCLC patients whose tumors express high levels of the PD-L1 protein.

The indication section of Keytruda's label specifies that the drug is for PD-L1-positive patients, but doesn't provide a specific cutoff for determining when a patient's tumor is PD-L1 positive. However, the label for the Dako PD-L1 IHC 22C3 pharmDx test states that if 50 percent or more of tumor cells are stained for PD-L1 expression, then the patient's sample should be considered PD-L1 positive.

However, when the FDA approved BMS's Opdivo and simultaneously approved a different Dako test, PD-L1 IHC 28-8 pharmDx, to identify patients who might benefit most from treatment, the agency didn't say the test was essential to giving the drug safely and effectively.

According to Rimm, this presents a complex prescribing scenario for clinicians, even as the field still doesn't know what AstraZeneca or Roche plan to do with their PD-1 drugs — whether they will come to market with a PD-L1 assay in either a companion diagnostic or a complementary diagnostic function.

Without a system for transposing the results of one assay to the context of a drug developed with a completely different assay, it's not hard to imagine a scenario where clinicians might have to perform several different PD-L1 tests to determine which drug may be best for their patient, Rimm said.

According to Hansen, Princess Margaret Hospital has done work to develop its own PD-L1 testing in house, but was told by pharmaceutical companies that the results wouldn’t necessarily be applicable to guiding therapy because of differences between one assay and another.

In their study last week, Rimm and his team set out in the first step of a systematic effort to better understand the relationships of PD-L1 expression to drug response in NSCLC, but comparing two PD-L1 rabbit monoclonal antibodies, E1L3N (Cell Signaling Technology) and SP142 (Spring Bioscience), which was part of Ventana's CDx development efforts and recently was released by the company for research use.

The investigators evaluated these different antibodies using 49 NSCLC samples collected retrospectively from 2011 to 2012 in the Yale Thoracic Oncology Program Tissue Bank.

Overall, the authors wrote, the results revealed that over 25 percent of patients who were above the positive threshold by one antibody were below it by the other.

Rimm's group previously developed a quantitative immunofluorescence method that objectively measures targets in both tumor and stromal compartments, and they also compared this method in the study to standard IHC, finding that despite increased sensitivity, there was still considerable intratumoral heterogeneity of PD-L1 expression and poor concordance between the two antibodies using the quantitative method.

Rimm and his colleagues granted that their conclusions should be taken in light of the study's limitations — a small number of retrospective samples and a lack of data available on patients' responses to anti–PD-1 and anti–PD-L1 drugs. In the future, the team hopes to be able to do similar studies on material from patients actually treated with anti–PD-1 therapy, the authors wrote.

More recently, two newer antibodies PD-L1 28-8 (Abcam) and SP263 (Ventana), have been released to the public, and the group is now conducting comparisons using this more comprehensive set. PD-L1 28-8 is the antibody used in Dako's Opdivo complementary diagnostic, and SP263 is an antibody being used in a diagnostic developed by Ventana to enroll patients in clinical trials for MedImmune's MEDI4736 anti-PD-L1 therapy.

"We are working very hard to access 22C3 [the fourth antibody being used in CDx development in this space] so far without success," Rimm said, "but we are trying to incorporate that, too."

With the announcement this week by the NCCN, Rimm and colleagues also have new support as they continue in this effort.

At the same time, the cause of PD-L1 assay harmonization is also being supported by several other groups. At a workshop co-organized by scientific oncology organizations and the FDA earlier this year, a group of two diagnostic companies and four biopharma companies involved in anti-PD-1 therapeutics and companion diagnostics created a blueprint for their plans to collaboratively compare the analytical performance of PD-L1 IVDs.

In an email to GenomeWeb, spokespersons from BMS and from Ventana said that this collaboration is moving forward and data are currently being generated as part of the project's initial Phase I study.

The International Association for the Study of Lung Cancer has also launched a project to develop a PD-L1 atlas to describe various available assays, their feasibility, and the comparability of these tests in patients with NSCLC.

According to Fred Hirsch, CEO of the IASLC, the atlas portion of this effort is still in the planning stages, and studies of the comparability of PDL-1 assays to support it are ongoing.

Finally, Hansen said that regulators also appear to be keyed into this issue, and that the European Medicines Agency may be working on developing a paper that would provide some regulatory guidance on interpretation of PD-L1 expression in the context of immune-oncology.