Skip to main content
Premium Trial:

Request an Annual Quote

AACR: Study Shows PD-L1 Expression Can ID Best Responders to Merck's anti-PD-1 Immunotherapy in NSCLC


PHILADELPHIA – Non-small cell lung cancer patients who had PD-L1 expression in at least half of their tumor cells saw their tumors shrink in greater numbers and lived longer after treatment with Merck's Keytruda (pembrolizumab) than patients with lower levels of expression, researchers reported yesterday at the American Association for Cancer Research annual meeting.

The study, led by Edward Garon of the University of California, Los Angeles, and published in the New England Journal of Medicine, represents the largest trial of a PD-1 blocking agent in lung cancer to date. The results, according to the study authors, demonstrated that PD-L1 expression may be a predictive biomarker used to identify NSCLC patients most likely to respond robustly to anti-PD-1 immunotherapy.

However, complicating matters is the fact that PD-L1 expression is not a straightforward marker for deciding whether or not to give patients Keytruda, like the presence or absence of BRAF V600E mutations are for discerning who should receive or forego a BRAF inhibitor. While PD-L1 expression may be particularly helpful for deciding which lung cancer patients should receive Keytruda in earlier treatment settings, experts at the meeting said additional trials are needed to figure out how to boost the immune systems of patients with lower PD-L1 expression so that they may also receive Keytruda in combination with other agents.

The US Food and Drug Administration granted accelerated approval last year to Keytruda as a treatment for advanced melanoma patients who have stopped responding to other therapies. The drug blocks the interaction between the PD-1 receptor and its ligands PD-L1 and PD-L2.

When PD-1 proteins interact with their ligands, it hinders the body's immune system from attacking cancer cells. The idea behind anti-PD-1 drugs is to inhibit this interaction, allowing T cells to attack cancer cells. Moreover, studies have shown that when PD-L1 is upregulated, certain tumor types, such as NSCLC, are more responsive to anti-PD-1 agents.

In the latest study investigating Keytruda in NSCLC patients, researchers gauged PD-L1 status using a prototype immunohistochemistry assay. Merck inked a companion diagnostic deal last year with Agilent Technologies' subsidiary Dako to develop a test that gauges PD-L1 expression for Merck's anti-PD-1 drug development program.

On the day of the AACR presentation, Merck announced that it had filed a supplemental biologics application for Keytruda in NSCLC, and Dako had filed for premarket approval for the IHC companion test that gauges PD-L1 expression.

In the study presented at AACR, researchers led by Garon enrolled around 500 patients into two groups — a training set involving approximately 182 patients where the PD-L1 expression cut point was selected and a validation set of 313 patients — to test out how well the patients responded to Keytruda based on the cut points. Patient responses were further evaluated according to whether they were previously treated with systemic therapy, which most were, or treatment naive. 

Using the IHC assay, Garon's group established three PD-L1 expression groups: a low-expression group that had less than 1 percent of PD-L1-positive tumor cells; an intermediate-expression group that had between 1 percent and 49 percent of PD-L1-positive cells; and a high-expression group that had 50 percent or more PD-L1-positive tumor cells. Approximately 25 percent of patients in the trial had PD-L1 expression in at least 50 percent of their tumor cells.

In the study, this latter group with the strongest PD-L1 upregulation had significantly higher overall response rate, progression-free survival, and overall survival compared to the other two lower PD-L1 expressing groups. In the validation set, 45 percent of subjects in the PD-L1 high group had tumor shrinkage compared to 17 percent in the intermediate- and 11 percent in the low-expressing groups. In the overall study population, the overall response rate was 19 percent.

Regardless of prior treatment status, median progression-free survival was 6.3 months for those with high PD-L1-positive status, and 3.3 months and 2.3 months in the intermediate and low PD-L1-positive groups, respectively. However, in treatment-naive high PD-L1 positive patients, median PFS was 12.5 months compared to 6.1 months in previously treated patients.

Median overall survival, as of the analysis cut off date of Aug. 29, 2014, had not been reached for high PD-L1-positive patients and was 8.8 months for both intermediate and low-expression groups. When considered in terms of prior treatment status, median overall survival for treatment-naive patients who had high PD-L1 positive status had not been reached, but was 16.2 months for the intermediate and 10.4 months for the low PD-L1 group. In previously treated patients, similarly, median survival hadn't been reached in the high PD-L1-positive group, but was 7.3 months and 8.6 months in intermediate and low groups.

Based on the results, Garon said his team could confidently conclude that for previously treated, advanced NSCLC patients who had PD-L1 expression in at least half of their tumor cells, Keytruda was likely to be "associated with superior clinical outcomes" compared to cytotoxic chemotherapy. "Researchers also noted a manageable toxicity profile for the drug and reported low rates of immune-mediated adverse reactions in Keytruda treated patients.

Garon added that for those with lesser degrees of PD-L1 staining, Keytrudamay be a better option for them in certain clinical scenarios compared to other agents, such as standard chemo. But more definitive data is needed in this regard from randomized trials comparing Keytruda against these other agents, he said.

'Blurry' biomarker

The results of this trial were particularly "exciting" for Suzanne Topalian of Johns Hopkins Medicine, who after Garon's presentation highlighted that more than 75 percent of the study population had progressed on prior systemic therapy. In advanced NSCLC, this is a very difficult-to-treat patient population, Topalian said. "The impact of second- and third-line agents is generally not expected to prolong survival."

Although in cancer there is a "big push" to use predictive biomarkers to distinguish responders from non-responders to treatment, she said, not all biomarkers can be implemented in such a straightforward fashion. "Certainly, that situation is more clear cut with the kinase inhibitor class of drugs where the biology tells us if the drug is really not likely to work if the patient doesn't have a particular mutation," Topalian said. "Here, today, we heard about a potential biomarker to select patients with lung cancer for anti-PD-1 therapy, but this kind of biomarker is a lot more blurry. And you saw that even patients with lower levels of expression of this marker still had notable response rates."

Merck is exploring how PD-L1 expression impacts Keytruda response in patients with different types of cancers. However, the rates of patients with different levels of PD-L1 expression may vary across tumor types and some of the test characteristics could differ by histology, Garon pointed out.

For example, Antoni Ribas from UCLA presented data at AACR from the KEYNOTE-006 trial comparing two dosing schedules of Keytruda against Bristol-Myers Squibb's Yervoy (ipilimumab) in advanced melanoma patients. In that study, Keytruda beat out Yervoy in terms of overall survival, progression-free survival, and response rates, leading Ribas to conclude that the data was sufficient to shift the standard of care in first-line advanced melanoma from Yervoy to Keytruda.

However, the impact of PD-L1 expression wasn't clear. While the hazard ratios for progression-free survival in both PD-L1 positive and negative patients seemed to favor Keytruda over Yervoy, the hazard ratios for overall survival of PD-L1 negative patients didn't appear to tilt toward one drug or the other.

"There has been a significant amount of study done on the importance of tumor expression of PD-L1 as a predictive biomarker, and while tumor expression of PD-L1 can effectively enrich cohorts of patients, it is not a binary or static predictive marker. Patients with PD-L1 negative tumors can still respond, and that is demonstrated here," Jedd Wolchok of Memorial Sloan Kettering said at the meeting, reviewing the PFS data from KEYNOTE-006.

But then, considering overall survival, in the PD-L1 negative group, "we no longer have favoring for pembrolizumab," Wolchok said, explaining that this could mean that PD-L1-negative patients have a shorter response to Keytruda or that post-progression therapy could be influencing the results.

Although PD-L1 will be challenging to implement as a predictive marker, there are a number of companies advancing PD-1 and PD-L1 inhibiting drugs that are poised to enter the market, each with their own companion tests. This may be confusing for the medical community. For example, different tests might have different cutoffs for defining PD-L1-positive and –negative patients, and this would impact which patients are identified as potential best responders to the drug and which aren't.

The US Food and Drug Administration held a meeting recently where four drugmakers — AstraZeneca, Roche/Genentech, Bristol-Myers Squibb, and Merck — announced they would compare the IHC tests they were using to stratify patients in trials for anti-PDL1 immunotherapy treatments and characterize the analytical variability between the platforms.

While presenting data on Keytruda from the NSCLC study, Garon said he couldn't speak to how insurers might employ the PD-L1 marker in coverage schemes for Keytruda, but he still believed the marker was important to have in the clinicians' armamentarium. "I'm pleased to have this data available for practitioners and for patients to be able to make appropriate risk stratifications," he said. "I think there are certainly clinical scenarios in which knowing the differences between these groups in terms of likelihood of response will be very significant."

He was hopeful, however, that in the future there would be interventions available for low-PD-L1 expressing patients that could make their tumor microenviroment more favorable for responding to a PD-1 inhibitor.