CheckMate-026 Underscores Predictive Value of High PD-L1 Expression

Edward Garon, MD

The value of PD-L1 expression when using checkpoint inhibitors in non—small cell lung cancer (NSCLC) is underscored by the just-announced disappointing progression-free survival (PFS) findings from the phase III CheckMate-026 study of frontline nivolumab (Opdivo) versus physician's choice of combination chemotherapy, Edward B Garon, MD, said in his keynote talk at the 2016 International Lung Cancer Congress.

Garon, of the David Geffen School of Medicine at UCLA, discussed the strong correlation between response rates and PD-L1 expression in an overview on immunotherapeutic advances in NSCLC. Patients enrolled in the CheckMate-026 study had PD-L1 expression on ≥5% of tumor cells—a much broader range than in the successful KEYNOTE-024 trial of pembrolizumab (Keytruda) as frontline monotherapy for patients with high PD-L1—expressing NSCLC. Patients with tumors showing ≥50% PD-L1 were selected for that study, and improvements in overall survival (OS) and PFS were seen.

“In light of the news, I think it only underscores the value, potentially, of that 50% cutpoint,” Garon said.

Lung cancer has not historically been considered an immunogenic disease, but excitement is growing as advances in NSCLC accumulate, Garon said. “When I was in my training, “interleukin-2 (IL-2) had some preclinical and correlative data that supported its use potentially in lung cancer, but unlike renal cell carcinoma and melanoma, there really were no positive results that we saw from this.” Trials were dogged by poor response, poor patient compliance, and high toxicity, he said.

He mentioned the MAGRIT trial, which investigated efficacy of MAGE-A3 antigen-specific cancer agents following tumor resection in patients with MAGE-A3-positive stages IB, II, and IIIA NSCLC. No significant OS was found and no predictive gene signature was identified. The START trial of MUC1 antigen-specific cancer immunotherapy tecemotide following chemotherapy also failed to improve OS.

It was just 4 years ago that nivolumab as monotherapy in NSCLC succeeded in blocking PD-1 immunosuppression, leading to durable response, Garon said. In that study,¹ the overall response rate (ORR) was 17.1%, although a higher general dosage might have shown the potential of this drug more clearly, Garon remarked. “The ORR appeared to be somewhat dragged down by the 1mg/kg cohort that did not do as well. In fact, at what is now the approved dose of nivolumab, there was a 23.4% response rate.”

Durable response is also key, he noted, and that same phase I nivolumab study demonstrated 27% OS at 3 years out among patients on a 3mg/kg dose—“which of course is very encouraging. So PD-1 inhibitors are now no longer a research topic. They are clearly a clinical topic.”

The CheckMate-017 study drove the point home in its randomization of patients with squamous cell NSCLC to either nivolumab or docetaxel. At 1 year, PFS was 21% for with nivolumab and 6.4% with docetaxel. The median PFS was 3.5 months versus 2.8 months, respectively (HR, 0.62; CI 95%, 0.47-0.81; P = .0004). The median OS was 9.2 months with nivolumab and 6 months with docetaxel (HR, 0.59; CI 95%, 0.44-0.79; P = .00025). “What you can see is a nearly unprecedented survival benefit,” Garon said.

That brought Garon to the question of the predictive value of PD-L1. CheckMate-017 results do show greater median OS based on higher PD-L1 expression. At ≥10% PD-L1, nivolumab achieved an 11-month OS versus 7.1 months for docetaxel. At <10% PD-L1, the respective numbers were 8.2 months and 6.1 months.

Garon said that in his early work with NSCLC patients on the PD-1 inhibitor pembrolizumab, those who responded had the highest level of expression for PD-L1, so researchers evaluated to see the extent to which PD-L1 would correlate with response rate.

“We looked at a whole bunch of different ways of assessing PD-L1—how much staining there was, whether it was intense, whether it was on most of the cells&mdash;and in the end, we went with staining in basically half of the cells as being what we thought was predictive of outcome.” Their hypothesis was supported by study results that showed a fourfold increase in the likelihood of response among patients who had high level staining versus patients who had none.²

Good tolerance by patients is a further recommendation for checkpoint inhibitors in NSCLC, Garon said. Atezolizumab (Tecentriq), although not approved so far in NSCLC, has similar promise, he said.

Mutational load is another marker of response that deserves consideration, Garon said. “Tumors with high mutational loads tend to be the ones associated with better outcomes with PD-L1 inhibitors. There was an early suggestion in lung cancer data—pembrolizumab, atezolizumab, and other agents&mdash;that showed smoking status was associated with improved outcome. Patients were more likely to respond if they were current or former smokers. Of course, smoking would be associated with mutational load.”

Garon also discussed what he called “combination mania.” He said the demonstrated potential of inhibitors makes it unlikely there is not a vast amount and diversity of combination therapy in progress; and yet AstraZeneca’s recent decision to halt 2 trials of the third-generation EGFR TKI osimertinib (Tagrisso) in combination with the PD-L1 inhibitor durvalumab, based on reports of lung disease, indicates that there are limitations.

Garon also cautioned that increasing pressure for rapid progress in drug development is putting goals ahead of well-paced medical inquiry. Before follow-up studies can be published, fresh trials are launched that might have benefited from a more in-depth look at foundation data, he said.

References

1. Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, and immune correlates of anti-PD-1 antibody in cancer. N Engl J Med. 2012;366(26):2443-2454.
2. Garon EB, Rizvi N, Hui R, et al. Efficacy of pembrolizumab (pembro; MK-3475) and validation of PD-L1 expression as a biomarker in patients (pts) with non-small cell lung cancer (NSCLC): findings from KEYNOTE-001. Presented at: 2015 AACR Annual Meeting; April 18-22; Philadelphia, PA. Abstract 9109.

<<<

View more from the 2016 International Lung Cancer Conference