Tumor-infiltrating immune cells (TIICs) have become an important source of markers for predicting the clinical outcomes of cancer patients. However, measurements of cellular heterogeneity vary due to the frequently updated reference genomes and gene annotations. In this study, we systematically collected and evaluated the infiltration pattern of 65 immune cells. We constructed the Immune Cell Pair (ICP) score based on the cell pair algorithm in 3,715 samples and across 12 independent cancer types, among which, the ICP score from six cancer types was further validated in 2,228 GEO samples. An extensive tumorigenic and immunogenomic analysis was subsequently conducted. As a result, the ICP score showed a robust reliability and efficacy in predicting the survival of patients with gliomas, in pan-cancer samples, and six independent cancer types. Notably, the ICP score was correlated with the genomic alteration features in gliomas. Moreover, the ICP score exhibited a remarkable association with multiple immunomodulators that could potentially mediate immune escape. Finally, the ICP score predicted immunotherapeutic responses with a high sensitivity, allowing a useful tool for predicting the overall survival and guiding immunotherapy for cancer patients.
Keywords: cell pair algorithm; glioma microenvironment; immune cell; immunotherapy; prognostic model.
Copyright © 2021 Zhang, Wang, Dai, Wu, Cao, Li, Zhang and Cheng.