For decades, platinum-based chemotherapy has been the first-line standard-of-care for non-small cell lung cancer (NSCLC). However, its high rate of adverse events and limited success against advanced cancers have left those of us who treat NSCLC searching for alternatives.
One such alternative is nivolumab, a PD-1 immune checkpoint inhibitor antibody approved by the FDA in 2014. Recent phase III studies, such as CheckMate 017, evaluated NSCLC patients who had completed or were receiving platinum-based chemotherapy and then received follow-on treatment with either nivolumab or docetaxel. For the patients on nivolumab, overall survival significantly improved.
The global trial we led at The Ohio State University Comprehensive Cancer Center – Arthur G. James Cancer Hospital and Richard J. Solove Research Institute, CheckMate 026, evaluated progression-free and overall survival in a broad population of NSCLC patients who had not undergone chemotherapy, comparing nivolumab monotherapy to platinum doublet chemotherapy. We were particularly interested in understanding how patient PD-L1 expression impacted outcomes, and conducted an exploratory analysis to determine if outcomes differed for patients with high numbers of tumor mutations that could result in new immune targets, called “mutation burden.”
We found that as a first-line treatment for stage IV/recurrent NSCLC, nivolumab monotherapy did not improve progression-free survival versus platinum-based chemotherapy, even for patients with at least 5 percent PD-L1 expression. However, outcomes were significantly improved for patients with higher tumor mutation burden, particularly for those with both high PD-L1 and high mutation burden. Patients on nivolumab experienced fewer adverse events—nearly two-thirds fewer for grade 3 or 4 events.
An open-label, randomized, phase III study for patients with at least 1 percent PD-L1 tumor expression was conducted. All patients were adults with histologically confirmed squamous or nonsquamous stage IV/recurrent NSCLC, ECOG PS 0-1, with tumors assessed following RECIST 1.1.
Patients provided tumor and blood samples sufficient for whole exome sequencing, which we used to determine tumor mutation burden, the total number of somatic missense mutations. Following procedures outlined in CheckMate 017, we collected fresh or archival tumor-biopsy specimens within 6 months of each patient's enrollment and had the specimens evaluated by a centralized laboratory using the 28-8 antibody. Patients were grouped according to total mutation burden tertile distribution at boundaries of 0 to <100 mutations (low), 100 to 242 mutations (medium), and ≥ 243 mutations (high).
None of the patients had received prior systemic anticancer therapy for their disease. If a patient's cancer had spread to their central nervous system, they remained eligible to participate if they were undergoing adequate treatment and had returned to neurologic baseline at least 2 weeks before randomization.
Patients who had received prior palliative radiotherapy were permitted to enroll if that therapy had been completed at least 2 weeks prior to randomization, and recurrent patients had to have completed adjuvant or neoadjuvant chemotherapy at least 6 months prior to enrollment. Overall, 38.6 percent of patients had prior radiotherapy. Patients also had to be off of corticosteroids or on a stable or decreasing dose of 10 mg, or less, daily prednisone (or equivalent). The study excluded patients with autoimmune disease, known EGFR mutations or ALK translocations sensitive to available targeted therapy.
We enrolled 1,325 patients and, of those, 784 (59%) were not randomized due to non-evaluable PD-L1 samples (6%), PD-L1 expression less than 1 percent (23%), or failure to meet other study criteria (30%). Of the 541 (40.8%) who were enrolled, 271 received 3 mg/kg nivolumab once every 2 weeks, while 270 received platinum doublet chemotherapy every 3 weeks for 4-6 cycles.
We allowed maintenance pemetrexed for patients with nonsquamous NSCLC who had stabilized after 4 cycles of doublet chemotherapy, which ultimately included 38.0 percent of treated patients. We also allowed for corticosteroid treatment in courses of less than 3 weeks for non-autoimmune conditions, including potentially immunologic treatment-related adverse events. Of the patients receiving chemotherapy, 128 (60%) crossed over to receive nivolumab after progression.
We allowed dose delays, along with two or fewer dose reductions, for chemotherapy to address toxicity. For nivolumab, we allowed dose delays for toxicity, but not dose reductions.
From diagnosis to randomization, median time was 1.9 months for patients who would take nivolumab and 2.0 months for patients who would receive chemotherapy. Overall, 530 patients (98.0% of all randomized patients) received treatment. We stratified the patients by PD-L1 expression (less than 5% vs. at least 5%) and tumor histology (squamous vs. nonsquamous). More than 78 percent of the randomized patients had 5 percent PD-L1 expression or greater.
The two treatment arms shared similar baseline characteristics, except the chemotherapy arm had higher proportions of female patients (45.2% vs. 32.1% in the nivolumab arm) and patients with at least 50 percent PD-L1 expression (46.7% vs. 32.5% in the nivolumab arm). Additionally, the nivolumab arm had more patients with liver metastases (19.9% vs. 13.3% in the chemotherapy arm) and greater tumor burden.
Among patients with at least 5 percent PD-L1 expression receiving nivolumab, 43.6 percent later received systemic cancer therapy, and 18.7 percent of treated patients remained on nivolumab at the time of database lock. For the patients in the chemotherapy arm, 64.2 percent received subsequent systemic therapy, including 60.4 percent who received nivolumab as crossover treatment within the study (57.5%) and/or later in clinical practice (3.3%).
For our trial, the primary endpoint was progression-free survival based on the assessment of a blinded independent central review (BICR) of patients with at least 5 percent PD-L1 expression. Secondary endpoints included progression-free survival per BICR for all randomized patients, overall survival among patients with at least 5 percent PD-L1 expression and among all randomized patients, and objective response rate per BICR among patients with at least 5 percent PD-L1 expression.
For the 423 patients with at least 5 percent PD-L1 expression—more than 78 percent of all randomized patients—progression-free survival with nivolumab was not significantly improved versus chemotherapy. Median progression-free survival was 4.2 months with nivolumab and 5.9 months with chemotherapy, with median overall survival following a similar pattern of 14.4 months with nivolumab and 13.2 months with chemotherapy.
Those results were similar for all randomized patients. Median time to response was also similar for both treatment arms, though duration was more than twice as long with nivolumab (12.1 months) than with chemotherapy (5.7 months).
In fact, across most of the predefined study subgroups, progression-free and overall survival were comparable with two exceptions: Patients with squamous histology had improved outcomes, as did patients with high tumor mutation burden.
While we did not observe a significant association between tumor mutation burden and PD-L1 expression, objective response rate in patients with high tumor mutation burden was significantly higher in the nivolumab arm (46.8%) than in the chemotherapy arm (28.3%) and progression-free survival was improved with nivolumab (9.7 months) versus chemotherapy (5.8 months). Tumor mutation burden was independent of PD-L1 and those with both high TMB and PD-L1 did not reach the median PFS, with 70 percent progression-free at 18 months. It is important to note that most of the patients in the chemotherapy arm who had high tumor mutation burden (65%) received nivolumab after the study crossover and would be expected to have done well.
Our trial also found that rates of adverse events for nivolumab and the chemotherapy regimen were similar, with events affecting more than 71 percent of patients treated with nivolumab and more than 92 percent of those treated with chemotherapy.
However, for serious events—grade III or IV—occurrence was significantly lower with nivolumab (17.6%) than chemotherapy (50.6%). As important, those events were less likely to cause a patient to leave the study in the nivolumab cohort (9.7%) versus the chemotherapy cohort (13.3%).
As a first-line treatment, the nivolumab regimen was comparable to the chemotherapy standard of care in the pre-specified study population of PD-L1 greater than 5 percent. Results from CheckMate 017 have suggested that nivolumab can prolong survival for previously treated patients with advanced NSCLC, and in this study the lack of a survival improvement in this population could be due to the high frequency of follow-on nivolumab treatment may have enhanced the rates of overall survival in the chemotherapy cohort. The chemotherapy arm also may have been imbalanced due to overall better prognostic disease characteristics (the aforementioned lower tumor burden, fewer liver metastases, and higher proportion of females).
In our effort to better understand biomarkers beyond PD-L1 expression as guides for treatment, we found evidence that high tumor mutation burden may, in fact, increase a patient's likelihood of benefitting from immunotherapy. Naiyer Rizvi and colleagues looked at this in a cohort of patients treated with immunotherapy and suggested this may be because high tumor mutation burden appears to enhance tumor immunogenicity by increasing the number of neoantigens that T cells can identify as threats, improving immune response. Our study provides strong support for this hypothesis since we studied samples from a much larger, chemotherapy-controlled, randomized study. Thus, we think this hypothesis warrants prospective exploration.
Taken broadly, our findings suggest nivolumab monotherapy is a promising option for first-line cancer therapy in selected patients. We are still learning the best biomarkers or combination of biomarkers to select patients most likely to benefit. Combination studies like CheckMate 012, which combines nivolumab with ipilimumab, another immune checkpoint, suggests that such an approach may be particularly effective as a first-line treatment. A phase III study now underway, CheckMate 227, is following such a strategy, evaluating efficacy and safety of nivolumab plus ipilimumab versus chemotherapy. The role of tumor mutation burden in this context also deserves exploration.
DAVID P. CARBONE, MD, PHD, is the Barbara J. Bonner Chair in Lung Cancer Research and Director of the James Thoracic Center at The Ohio State University Comprehensive Cancer - Arthur G. James Cancer Hospital and Richard J. Solove Research Institute (OSUCCC - James).