Nationwide multi-centric prospective study for the identification of biomarkers to predict the treatment responses of nivolumab through comprehensive analyses of pretreatment plasma exosome mRNAs from head and neck cancer patients (BIONEXT study)

Abstract

Background: Nivolumab paved a new way in the treatment of patients with recurrent or metastatic (RM) head and neck squamous cell carcinoma (RM-HNSCC). However, the limited rates of long-term survivors (< 20%) demand a robust prognostic biomarker. This nationwide multi-centric prospective study aimed to identify a plasma exosome (PEX) mRNA signature, which serves as a companion diagnostic of nivolumab and provides a biological clue to develop effective therapies for a majority of non-survivors.

Methods: Pre-treatment plasmas (N = 104) of RM-HNSCC patients were subjected to comprehensive PEX mRNA analyses for prognostic marker discovery and validation. In parallel, paired treatment-naïve tumor and plasma samples (N = 20) were assayed to elucidate biological implications of the PEX mRNA signature.

Results: Assays for pre-treatment blood samples (N = 104) demonstrated that a combination of 6 candidate PEX mRNAs plus neutrophil-to-lymphocyte ratio precisely distinguished non-survivors from >2-year survivors (2-year OS; 0% vs 57.7%; P = 0.000124) with a high hazard ratio of 2.878 (95% CI 1.639-5.055; P = 0.0002348). Parallel biological assays demonstrated that in the paired treatment-naïve HNSCC tumor and plasma samples (N = 20), PEX HLA-E mRNA (a non-survivor-predicting marker) was positively corelated with overexpression of HLA-E protein (P = 0.0191) and the dense population of tumor-infiltrating NK cells (P = 0.024) in the corresponding tumor, suggesting that the HLA-E-NKG2A immune checkpoint may inhibit the antitumor effect of PD-1blockade.

Conclusion: The PEX mRNA signature could be useful as a companion diagnostic of nivolumab. The combination of an anti-NKG2A antibody (i.e., monalizumab) and nivolumab may serve as a treatment option for non-survivors predicted by a RT-qPCR-based pre-treatment measurement of PEX mRNAs.

Keywords: HLA-E; biomarker; exosome; head and neck cancer; nivolumab.

Figure 1

Study design.

Figure 2

(A, B) Kaplan-Meier curves representing the overall survival of patients classified according to the best overall response. (C) Box plots representing the expression levels of 20 candidate biomarker genes in patients stratified according to the best overall response. CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; NE, not evaluated. (*) P <0.05; (**) P <0.005.

Figure 3

(A) Box plots comparing the expression levels of response-predicting biomarker genes and the neutrophil-to-lymphocyte ratios (NLR) between responders (CR/PR) and non-responders (SD/PD/NE). (B) Box plots comparing the scores of combinations calculated by biomarker genes and the NLR between responders (CR/PR) and non-responders (SD/PD/NE). (*) P <0.05; (**) P <0.005; (***) P <0.0005.

Figure 4

Survival prediction based on the identified biomarker combinations. (A) Forest plots representing the hazard ratios of the biomarker combinations. HR, hazard ratios; CI, confidence intervals. (B) Kaplan-Meier curves representing the overall survival of patients classified according to the score of biomarker combination (left panel, combination 9; right panel, combination 10). (C) Kaplan-Meier curves (left panel) representing the overall survival of patients classified according to the 2 x 2 contingency table (right panel). (*) P <0.05.

Figure 5

Correlation of PEX mRNA and the tumor immune microenvironment. (A) Correlation between the HLA-E expression levels detected by RNA-seq (vertical axis) and qPCR (horizontal axis) (N =17). R represents the Pearson correlation coefficient. (B) Representative images of immunohistochemistry staining for HLA-E in tumor tissues; HLA-E low (left) and HLA-E high (right). High-magnification images of the regions indicated by black boxes are shown. The table represents the numbers of cases and the correlation between HLA-E protein and HLA-E mRNA expression levels in PEXs (N =20). The P-value was calculated by Fisher’s exact test. (*) P <0.05; (**) P <0.005; (***) P <0.0005. (C) Box plots representing the expression levels of biomarker genes detected by RT-qPCR of exosomes extracted from peripheral blood and the NLR. BNB represents the cohort of part 2 study cohort (N =20). CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; NE, not evaluated. (D) Box plots representing the proportion of immune cells estimated by CIBERSORTx in the primary tumor tissues (N =17). Patients were classified according to the expression levels of PEX HLA-E mRNA (HLA-E high, N =9; HLA-E low, N =8). P-value was calculated by Mann-Whitney U-tests. (*) P <0.05; (**) P <0.005; (***) P <0.0005. (E) Schematic summarizing of our proposed mechanism by which the effect of nivolumab is canceled in the tumor of patient with high PEX HLA-E mRNA (left panel) and a decision-making algorithm for patients (right panel). The high PEX mRNA level reflects the vigorous HLA-E protein production in cancer cells, forming HLA-E/NKG2A checkpoint with NK and CD8+CTL cells. In this setting, administration of nivolumab alone is not effective. Addition of an anti-NKG2A antibody, monalizumab, is expected to restore the cytotoxic effects of NK and CTL cells circumvented by the dual immune checkpoints.