Multi-omics comprehensive analyses of programmed cell death patterns to regulate the immune characteristics of head and neck squamous cell carcinoma

Transl Oncol. 2024 Mar:41:101862. doi: 10.1016/j.tranon.2023.101862. Epub 2024 Jan 18.

Abstract

Head and neck squamous cell carcinoma (HNSCC) is a heterogeneous cancer with high morbidity and mortality. Triggering the programmed cell death (PCD) to enhance the anti-tumor therapies is being applied in multiple cancers. However, the limited understanding of genetic heterogeneity in HNSCC severely hampers the clinical efficacy. We systematically analyzed 14 types of PCD in HNSCC from The Cancer Genome Atlas (TCGA). We utilized ssGSEA to calculate the PCD scores and classify patients into two clusters. Subsequently, we displayed the genomic alteration landscape to unravel the significant differences in copy number alterations and gene mutations. Furthermore, we calculated the IC50 values of targeted drugs to predict the differences in sensitivity. To identify the immune-related prognostic types, we comprehensively estimated the relationship between immune indicators and all prognostic PCD in three datasets (TCGA, GSE65858, GSE41613). Finally, 7 regulators were filtered. Subsequently, we integrated 10 machine learning algorithms and 101 algorithm combinations to test the clinical predictive efficacy. Using WGCNA as a basis, we built a weighted co-expression network to identify modules involved in the immune landscape with different colors. Meanwhile, our results indicated that blue and red modules containing crucial regulators closely related to the CD4+, CD8+ T cells, TMB or PD-L1. FCGR2A from blue module, CSF2, INHBA, and THBS1 from the red module were determined. After verifying in vivo experiments, FCGR2A was identified as hub gene. In conclusion, our findings suggest a potential role of PCD in HNSCC, offering new insights into effective immunotherapy and anti-tumor therapies in HNSCC.

Keywords: Genetic heterogeneity; Head and neck squamous cell carcinoma; Immune-infiltrating characteristics; Machine learning; Programmed cell death.