Although anti-PD-1/PD-L1 monotherapy has achieved clinical success in non-small cell lung cancer (NSCLC), definitive predictive biomarkers remain to be elucidated. In this study, we performed whole-transcriptome sequencing of pretreatment tumor tissue samples and pretreatment and on-treatment whole blood samples (WB) samples obtained from a clinically annotated cohort of NSCLC patients (n = 40) treated with nivolumab (anti-PD-1) monotherapy. Using a single-sample gene set enrichment scoring method, we found that the tumors of responders with lung adenocarcinoma (LUAD, n = 20) are inherently immunogenic to promote antitumor immunity, whereas those with lung squamous cell carcinoma (LUSC, n = 18) have a less immunosuppressive tumor microenvironment. These findings suggested that nivolumab may function as a molecular targeted agent in LUAD and as an immunomodulating agent in LUSC. In addition, our study explains why the reliability of PD-L1 expression on tumor cells as a predictive biomarker for the response to nivolumab monotherapy is quite different between LUAD and LUSC.

e21097 Background: Although anti-PD-1/PD-L1 monotherapy has achieved clinical success in non-small cell lung cancer (NSCLC), definitive predictive biomarkers remain to be elucidated. We assumed that by combining gene expression signatures with patient clinical data, we could identify a novel promising biomarker to predict response to anti-PD-1/PD-L1 monotherapy in NSCLC patients. Moreover, the characterization of these signatures will help us to decipher the complexity of tumor-immune interactions and better understand the tumor microenvironment (TME) that favors clinical response to nivolumab monotherapy. Methods: From clinically annotated NSCLC patients (n = 40) with nivolumab monotherapy in the second- or later-line settings, we prospectively collected tumor tissues and peripheral blood mononuclear cells (PBMCs) before first dose of nivolumab and PBMCs after first 4 or 5 doses of nivolumab. All tumor tissue and PBMC samples obtained were applied to whole-transcriptome sequencing (RNA-seq). We extracted transcriptomic datasets of lung adenocarcinoma (LUAD) (n = 20) and lung squamous cell carcinoma (LUSC) (n = 18) from the results, separately analyzed each histological subtype. To elucidate biological processes associated with clinical outcomes, we performed a supervised gene set enrichment analysis (GSEA) approach and an unsupervised single sample scoring approach. Results: In LUAD, we observed that gene sets related to interferon (type I and II) signaling (‘IFN signatures’) and antigen processing and presentation (‘APP signatures’) were significantly enriched in pre-treatment PBMCs of responders. IFN and APP signatures, which are closely related to each other, functionally cooperate to activate anti-tumor immune response. The enrichment of IFN and APP signatures provides the possibility that responders have a pre-existing anti-tumor immunity prior to nivolumab monotherapy. In LUSC, neither IFN nor APP signatures were enriched in pre-treatment tumor tissues and PBMCs of responders. Instead, gene sets related to the regulation of the TME (‘TME signatures’) are significantly enriched in pre-treatment tumor tissues of non-responders. The enrichment of TME signatures suggested that non-responders have an extremely immunosuppressive TME. These findings highlighted that responsive LUAD inherently have a high immunogenicity to elicit effective anti-tumor responses, whereas responsive LUSC have a similar level of immunogenicity as non-responsive LUSC but are free from an extremely immunosuppressive TME. Conclusions: We found that nivolumab enhanced anti-tumor immunity in patients with LUAD in a quite different way from patients with LUSC. Our study provides a blueprint for innovating combinational immunotherapy and supporting patient selection and treatment strategies on long-term clinical outcomes.

No abstract