Matrisomal abnormality: a predictive biomarker for cancer immunotherapy

Abstract

Abstract Background Immune checkpoint inhibitors (ICI) appear particularly promising for head and neck cancers, as evidenced by recent clinical trials. Despite durable antitumor responses in a subset of advanced cancer patients, most head and neck squamous cell carcinoma (HNSCC) and nasopharyngeal cancer (NPC) are intrinsically resistant to immunotherapy. Here, we aim to reveal immune infiltrative landscape of HNSCC and NPC and further to identify predictive biomarkers for selection of patients to receive ICI-based treatments. Methods A measure of mRNA deregulation in tumor microenvironment, termed as tumor matrisome index (TMI), was assessed using RNA-seq and microarray-acquired expression datasets. Receiver operating characteristic (ROC) curves and survival data were analyzed to evaluate the diagnostic and prognostic value of TMI, respectively. Parallel analyses of signatures predictive of immunotherapy response, including innate PD-1 resistance signature (IPRES), were performed using Gene set variation analysis (GSVA). Machine learning-based deconvolution algorithm (i.e., CIBERSORT) was applied to all datasets to reveal specific immune cell types associated with TMIhigh tumors. Results TMI achieved a mean AUC value of 0.903 and 0.991 in differentiating normal tissues from NPC and HNSC, respectively. Integrative genomic analyses reveal that TMI is closely associated with IPRES signatures and CIBERSORT-estimated relative abundance of specific immune cell populations. Of clinically targetable immune checkpoints, OX-40L emerged as a promising target for anti-tumor immunity in TMIhigh NPC tumors. Conclusion TMI signatures are integral components of protumoral extracellular matrix. The developed metrics to measure the extent of such matrisomal abnormalities may represent a novel predictive biomarker for cancer immunotherapy.