“It has really been incredible what has happened in the last 20 years, but particularly in the last 10,” said Roy S. Herbst, MD, PhD, a professor of Medicine and chief of Medical Oncology, Yale Cancer Center and Smilow Cancer Hospital at Yale New Haven. “We have moved from the era of targeted therapy [in select patients] and then, of course, to immunotherapy.”
Herbst spoke on initial treatment with immunotherapy in patients with lung cancer during a presentation at the 11th New York Lung Cancer Symposium, providing a look back on the immunotherapy approvals and pivotal data for nivolumab (Opdivo), pembrolizumab (Keytruda), and atezolizumab (Tecentriq), along with a glance at what the field can expect next.
Choosing the Best Frontline Option
In October 2016, the FDA approved pembrolizumab in the frontline setting for patients with NSCLC, based on findings from the phase III randomized KEYNOTE-024 study, which showed that single-agent pembrolizumab led to a 40% reduction in the risk of death compared with doublet chemotherapy for untreated patients with advanced disease with ≥50% PD-L1 expression on cells.1 The median progression-free survival (PFS) was also improved by 4.3 months.
“It does not get much better than this,” said Herbst. “This is historic.”
Pembrolizumab is the sole checkpoint inhibitor approved for frontline NSCLC, with nivolumab and atezolizumab both available for use after patients progress on chemotherapy.
“The world changed exactly 1 month ago, and this is KEYNOTE-024,” Herbst said. “It is the most clinically developed trial at this point because it really has moved pembrolizumab to the frontline setting in untreated patients with lung cancer.”
When it comes to nivolumab in the first-line setting, however, the phase III multicenter, open-label CheckMate-026 study results had the opposite impact. In CheckMate-026, there was not a PFS benefit observed with nivolumab over chemotherapy.2 Patients who received nivolumab had a median PFS of 4.2 months compared with 5.9 months for patients treated with a chemotherapy doublet chosen by the treating physician. Additionally, there was no significant difference observed between the 2 arms for overall survival (OS) and response rate.
“Then, there’s this trial (CheckMate026), and you have got to scratch your head,” Herbst reflected. “As enthusiastic as we were, it would have been nice to see a second drug [succeed in frontline]. Most of us think that nivolumab and pembrolizumab are really not that different.”
However, CheckMate-026 was designed differently, as it included patients with ≥1% PD-L1 expression on cells. In a subgroup of patients with ≥5% PD-L1 expression, the median OS was 14.4 months with nivolumab and 13.2 months with chemotherapy. The 1-year survival was 56.3% and 53.6% with nivolumab and chemotherapy, respectively.
“Could it be that there were other factors involved?” Herbst questioned. “That is still under review.”
PD-L1 as a Biomarker: Is it Worthwhile?
Herbst spoke on biomarker results from the phase III CheckMate-057 trial, which compared nivolumab with docetaxel in patients with advanced NSCLC.3 Here, it was found that nivolumab-treated patients who harbored PD-L1 expression levels of ≥1%, ≥5%, and ≥10% had longer OS, PFS, and median duration of response, as well as higher objective response rates (ORRs).
“I have been working on this for a few years, and I have evolved in my thinking,” Herbst said. “Based on those data, the FDA and all of us said, ‘we don’t really need a biomarker to give nivolumab in second-line; we will give it to almost everyone.'”
PD-L1 expression in the KEYNOTE-001 and KEYNOTE-010 trials, which explored pembrolizumab versus docetaxel in advanced NSCLC had similar results.4,5 PFS and OS were shorter among patients with a PD-L1-expression score of 1% to 49% or a score of less than 1% than among those with a score of at least 50% expression on cells.
Expression of PD-L1 is heterogeneous and varies with the antibody used, Herbst explained, suggesting it to be an imperfect biomarker. Additionally, the role of PD-L1 as a prognostic factor is not yet well defined, he added.