Durable Responses With PD-1 Inhibition in Lung and Kidney Cancer and the Ongoing Search for Predictive Biomarkers.

Abstract

In non–small-cell lung cancer (NSCLC), tremendous advances have been made in the treatment of tumors harboring mutations in genes like EGFR, ALK, and ROS1, but there has been little progress in prolonging survival for the majority of patients with lung cancer who do not have these targetable oncogenic mutations. The response rate to second-line chemotherapy is less than 10% with a 1-year survival rate of 30%, and administration of more than two lines of chemotherapy has not demonstrated a survival benefit in NSCLC. In advanced renal cell carcinoma (RCC), despite a number of approved anti–vascular endothelial growth factor and mammalian target of rapamycin inhibitors with antitumor activity, durable responses are rare and treatments can be associated with serious adverse events. In metastatic RCC, we have known for more than 20 years that treatment with high-dose interleukin-2 can lead to longterm complete responses, but this therapy is only effective in a small fraction of patients and its significant toxicity profile limits widespread use among RCC patients. Disruption of programmed cell death protein 1 (PD-1) pathway signaling with immune checkpoint inhibitors represents a major breakthrough in the treatment of an increasing number of tumor types including melanoma, lung cancer, RCC, bladder cancer, and Hodgkin lymphoma. Although a minority of patients with lung and kidney cancer responded to the fully human immunoglobulin G4 antibody nivolumab in early trials, many of these patients displayed durable responses. Now, the articles by Gettinger et al and McDermott et al that accompany this editorial present the long-term safety and efficacy data for nivolumab in NSCLC and RCC, respectively. In the NSCLC trial by Gettinger et al, one of three doses of nivolumab was administered every 2 weeks for up to 96 weeks in 129 patients, more than half of whom had received more than three lines of prior systemic therapy. The response rate was 17% and the 3-year overall survival rate was an impressive 18%, exceeding what has otherwise been observed for such a heavily pretreated group. In the McDermott et al trial of 34 pretreated RCC patients, the response rate to nivolumab was 29%, with a 3-year survival of 44%. Interestingly, nine of 18 NSCLC patients and three of five RCC patients who stopped treatment with nivolumab due to adverse events or study completion continued to experience a response for longer than 9 months after drug discontinuation. Furthermore, the tails of the Kaplan-Meier survival curves in both trials appear to plateau at approximately 3 years, raising the hopes for long-term benefit or even cure in these diseases. Grade 3 to 4 treatment-related adverse events occurred in 14% to 18% of patients, and there were three treatmentrelated deaths associated with pneumonitis in the lung cancer cohort. We eagerly await the overall survival results from ongoing randomized phase III studies in pretreated NSCLC and RCC comparing nivolumab to docetaxel and everolimus, respectively. Are there clinical features or pathologic biomarkers to help oncologists decide which of their patients with lung and kidney cancer should be treated with PD-1 inhibitors? In the RCC study of 34 patients, clinical and pathologic predictors of treatment response were not described, and tumor programmed cell death protein ligand 1 (PD-L1) expression status was not reported. In the lung cancer study, response to nivolumab did not seem to be impacted by clinical factors such as age or sex, but patients who were current or former smokers had higher response rates than light or never-smokers, supporting the hypothesis that tumors with higher mutational load are more likely to create neoantigens that can be recognized by the immune system. In terms of pathologic characteristics, nivolumab efficacy appeared to be no different among patients with squamous compared with nonsquamous tumor histology, and too few cases were genotyped to know whether mutational subsets (eg, KRAS, EGFR) responded differently to nivolumab. Somewhat dishearteningly, the current lung cancer study showed no associated between tumor cell PD-L1 immunohistochemical staining and response to nivolumab. This contrasts with recent data in NSCLC with the PD-1 inhibitor pembrolizumab where there appears to be a more direct correlation between the degree of tumor cell PD-L1 expression and response rate. These differing results reflect several of the technical and biologic issues currently plaguing the field of immuno-oncology in its feverish search to identify the predictive biomarkers for immunotherapies. There are numerous PD-L1 immunohistochemistry (IHC) antibodies but their test characteristics have not been compared with one another, nor are they available for academic study, because many are in development as proprietary companion diagnostics. In addition, there are no clearly validated diagnostic cutoffs for what constitutes clinically-meaningful PD-L1 positivity by IHC. Biologically, PD-L1 expression on infiltrating JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 33 NUMBER 18 JUNE 2