Integrated multi-omics analysis identifies CD73 as a prognostic biomarker and immunotherapy response predictor in head and neck squamous cell carcinoma

Background Advances in tumor immunotherapy have been developed for patients with advanced recurrent or metastatic (R/M) HNSCC. However, the response of most HNSCC patients to immune checkpoint inhibitors (ICI) remains unsatisfactory. CD73 is a promising target for tumor immunotherapy, but its role in HNSCC remains insufficient. In this study, we aim to explore the function of CD73 in HNSCC. Methods Transcriptomic and clinical data of TCGA-HNSC were downloaded from UCSC Xena for analysis of CD73 mRNA expression and prognosis. Immunohistochemical assay were performed to validate the expression of CD73 in tumor tissues and its relationship with CD8+ T cells. GSEA analysis was performed with the “clusterProfiler” R package. Immune infiltration analysis was calculated with ESTIMATE, CIBERSORT and MCP-counter algorithms. Single-cell transcriptomic data was originated from GSE103322. Cell clustering, annotation and CD73 expression were from the TISCH database. Correlation data between CD73 and tumor signatures were obtained from the CancerSEA database. Somatic mutation data were obtained from TCGA-HNSC and analyzed by “maftools” R package. Immune efficacy prediction was performed using TIDE algorithm and validated with the IMvigor210 cohort. Results Compared with normal tissues, both mRNA and protein expressions of CD73 were elevated in tumor tissues (P = 9.7×10-10, P = 7.6×10-5, respectively). Kaplan-Meier analysis revealed that patients with high expression of CD73 had worse overall survival (log-rank P = 0.0094), and CD73 could be used as a diagnostic factor for HNSCC (AUC = 0.778). Both bulk RNA-seq and single-cell RNA-seq analysis showed that high CD73 expression can promote EMT and metastasis, samples with high CD73 expression had reduced CD8+ T cells. Furthermore, it was found that CD73-high group was more prone to have mutations in TP53, HRAS and CDKN2A, and were negatively correlated with TMB (P = 0.0055) and MSI (P = 0.00034). Mutational signature analysis found that CD73 was associated with APOBEC signature. Immunotherapy efficacy analysis showed that CD73-high group was less sensitive to immune efficacy. Conclusions Our results demonstrate that CD73 has an inhibitory effect on the tumor microenvironment, and is more likely to be unresponsive to ICI therapy. Collectively, targeting CD73 may provide new insights for tumor targeted therapy and/or immunotherapy.