Amid continuing excitement over the potential for immunotherapies targeting the PD-1/PD-L1 pathway, two major questions have loomed over PD-L1 expression levels as a biomarker for this emerging new class of drugs.
Suzanne L. Topalian, MD
Amid continuing excitement over the potential for immunotherapies targeting the PD-1/PD-L1 pathway, two major questions have loomed over PD-L1 expression levels as a biomarker for this emerging new class of drugs: What is the clinical utility of PD-L1 expression levels and what are the practical aspects of developing assays to measure those levels?
From the clinical standpoint, leading researchers say the significance of PD-L1 expression as a marker for recommending patients for therapy has not been established. Although patients with higher PD-L1 levels demonstrate better response rates to therapy, particularly in melanoma and non—small cell lung cancer (NSCLC), clinical trial participants with low levels also have responded to the drugs, including some patients with durable responses.
While those questions likely will be studied for years, pharmaceutical developers have a more urgent mission when it comes to developing assays that will be useful to practicing oncologists in determining which of the new immunotherapies is appropriate for a particular patient.
As it stands now, the four major companies developing anti-PD-1/PD-L1 therapies are using different assays to measure PD-L1 expression levels in the drugs they are testing.
This does not pose a problem now for oncologists who are considering the two currently approved anti-PD-L1 therapies—nivolumab (Opdivo) in melanoma and squamous NSCLC and pembrolizumab (Keytruda) in melanoma—because the indications for those agents are not linked to a biomarker or companion diagnostic.
However, oncology and industry leaders are concerned that the future will bring a confusing selection of assays that would make it difficult for oncologists to match the right patient with the right drug.
As a result, pharmaceutical companies locked in hot competition to bring the new immunotherapies to market have come together in an unusual collaboration to analyze the differing assays they have been using in NSCLC clinical trials (Table).
“Different pharmaceutical companies and diagnostic companies don’t often get together premarket and discuss their issues,” Elizabeth Mansfield, PhD, an FDA official who co-chaired a March 24 workshop1 that brought together the companies, researchers, and regulators, said in an interview with OncLive.
However, she noted that “there was going to be the potential for a number of different drugs that target the same molecule to be in development at the same time and that each one of them could potentially need a companion diagnostic.”
“It could potentially get very confusing with different drugs, different tissues, different ways of measuring the biomarker,” added Mansfield, who is deputy director of personalized medicine in the Office of In Vitro Diagnostics and Radiological Health, which is part of the FDA’s Center for Devices and Radiological Health (CDRH). “We’re trying to figure out how we can out this out in a rational form so that oncologists can look at it and figure out what they need to do.”
PD-L1—positive lung tissue, shown in slide at left, was analyzed with an SP263 immunohistochemistry assay by Ventana Medical Systems. The tissue shown at right tested negative using the same assay.
Photos courtesy Ventana Medical Systems
Six Companies Cooperate on Study
The drugmakers working on the project are Bristol-Myers Squibb, Merck, Genentech, and AstraZeneca, which are developing nivolumab (Opdivo), pembrolizumab (Keytruda), atezolizumab (MPDL3280A), and durvalumab (MEDI4736), respectively.
Two diagnostic companies that have helped develop the assays used in clinical trials evaluating these drugs in NSCLC are conducting the analyses: Dako, which is owned by Agilent Technologies, and Ventana Medical Systems, which is a member of the Roche Group along with Genentech. Ventana already was partnering with MedImmune, the biologics research and development arm of AstraZeneca, on a PD-L1 assay for MEDI4736 clinical trials in NSCLC.
The project has been moving forward after the companies jointly announced a blueprint for the assay analysis during the workshop that the FDA, along with leaders from the American Society of Clinical Oncology (ASCO) and the American Association for Cancer Research (AACR), hosted in March.1
“The proposal is intended to establish how the tests compare to each other given the same sample,” said Mansfield. “What will come out of this will be a set of data that shows you that, if you use the same specimen from one patient and test it with all the different tests, the way [results] would read out with all the different tests.”
Mansfield anticipates that the industry group will report on its findings next year. However, she said it is not yet clear whether the results of this assay comparison would apply to cancer types besides NSCLC. “We hope that it will at least partially inform and will at least set up a framework if we need to do this again in other tissue types,” Mansfield said.
PD-L1 Assay Differences
Unlike other biomarker assays used in cancer care today, the development of PD-L1 expression testing is complicated by the race to exploit this exciting new area of anticancer therapies.
mAB indicates monoclonal antibody; NSCLC, non—small cell lung cancer; ORR, objective response rate; PD-L1, programmed death ligand-1.
Complexities in personalized medicine: harmonizing companion diagnostics across a class of targeted therapies. Presented at: FDA-AACR-ASCO Public Workshop; Washington, DC; March 24, 2015.
Companies engaged in parallel development programs in multiple tumor types are using their own immunohistochemistry (IHC) assays to measure PD-L1, according to Reena Philip, PhD, director of the Division of Molecular Genetics and Pathology in the CDRH’s Office of In Vitro Diagnostic Device Evaluation and Safety.
“There’s a unique scenario in this case, where each PD-1 or PD-L1 therapy has its own proprietary PD-L1 IHC assay,” Philip said during the FDA workshop.1
She said the assay used with each therapeutic product is different in terms of IHC antibody clones; staining protocols, platforms and scoring methods; clinical decision points; tumor indications; and the use of tumor cells or tumor-infiltrating lymphocytes (TILs) or both.
Philip noted that PD-L1 is expressed by both tumor cells and TILs, setting the stage for another complexity. “Some companies are measuring the staining of PD-L1 just in tumor epithelial cells while others are including tumor infiltrating lymphocytes,” she said.
The reason for the differences in PD-L1 assays stems from the differences in the agents that are being developed, stressed Steven D. Averbuch, MD, a leader in oncology drug development and head of pharmacodiagnostics at Bristol-Myers Squibb, who presented an industry point of view at the workshop. “Each biopharmaceutical company’s development of a PD-1 or PD-L1 agent is specifically tailored and informed by the clinical experience of that pharmaceutical concern,” said Averbuch. “Each molecule, while addressing the same pathway, has unique pharmacologic properties.”
Moreover, Averbuch said, PD-L1 itself is a biologically complex biomarker. “It’s dynamic. There’s a lot of heterogeneity, both spatially and temporally, and many other factors go into the complexity of the immune-related biology.” In addition, Averbuch noted that there are other technical issues such as the collection of tissue specimens from biopsies including the type, timing, and processing of samples, and the inherent limitations of IHC itself.
Given all these variables, the pharmaceutical rivals agreed to collaborate on assessing the comparative performance of each PD-L1 assay used in the NSCLC clinical trials, Averbuch said.
The plan is to deliver “a package of information and data upon which analytical comparisons of the various diagnostic assays may be conducted, potentially paving the way for postmarket standardization and/or practice guideline development,” he said.
Clinical Utility Debated
In addition to the technical aspects of developing PD-L1 assays, clinical questions of how best to use such information remain unsettled.
In NSCLC, the utility of PD-L1 expression levels as a biomarker for selecting patients for therapy has been mixed. In the CheckMate-017 clinical trial that paved the way for the approval of nivolumab in advanced squamous NSCLC, patients had similar outcomes regardless of PD-L1 expression status.2
However, in the CheckMate-057 trial, patients with nonsquamous NSCLC whose PD-L1 expression levels were higher experienced greater benefit.
The median overall survival benefit improved by 41% for patients whose PD-L1 expression level was ≥1% of tumor cells. The reduction in the risk of death was 57% and 60% for patients expressing PD-L1 on ≥5% and ≥10% of their tumor cells, respectively.3
Similarly, higher PD-L1 expression correlated significantly with response rates in the KEYNOTE- 001 trial, which evaluated pembrolizumab in patients with previously treated and treatment- naïve advanced or metastatic NSCLC. Overall response rates were 45.2%, 16.5%, and 10.7% among patients with PD-L1 expression ≥50%, 1%-49%, and <1%, respectively.4
Broader Immune Questions
From her vantage point, Suzanne L. Topalian, MD, sees the dynamic and complex nature of the immune system as a significant challenge in identifying biomarkers for immunotherapy and correlating markers with outcomes. Topalian is director of the melanoma program at Johns Hopkins Kimmel Cancer Center in Baltimore, which pioneered development of the PD-1/PD-L1 immunotherapies including exploring the potential use of PD-L1 expression as a biomarker.
“Within individual patients, there can be heterogeneous PD-L1 expression at different anatomic sites and at different times during the evolution of the cancer,” Topalian noted during her remarks during the workshop. Overall, patients whose tumors express PD-L1 are more likely to respond to therapies attacking the PD-1/PD-L1 pathway “but there is still a reasonable percentage of patients who are PD-L1 negative, and yet these patients are responding,” she said.
As a result, Topalian is among other leading immunotherapy experts who say PD-L1 expression should not be used to recommend patients for therapy. “That is what oncologists worry about with the PD-L1 tests, as they are today. We don’t want to leave any patient behind who might benefit from these drugs,” she said.
Nevertheless, there likely will be good reasons why clinicians will want to measure PD-L1 expression in patients who are candidates for these therapies. “Potential utilities for this test may be selecting individual patients, but it might more likely be identifying cancer types that we should test in the clinic for susceptibility to PD-1 and PD-L1 blocking drugs, and to enrich for responsive patients in particular cancer types that have very low overall response rates,” Topalian said.