Comprehensive characterization of the molecular feature of T cells in laryngeal cancer: evidence from integrated single-cell and bulk RNA sequencing data using multiple machine learning approaches.

Background: The clinical relevance of T cell-related molecules at single-cell resolution in laryngeal cancer (LC) has not been clarified.

Materials and methods: Three LC tissues and matching adjoining normal tissues from the hospital were used to perform 10X single-cell RNA sequencing. Hub T cell-related genes (TCRGs) were detected by applying ten machine learning (ML) techniques based TCGA and GEO databases, which were also utilized to create a prediction model (TCRG classifier) and a multicenter validation model. Lastly, we conducted a comprehensive analysis of the TCRG's correlation with immunological properties.

Results: The analysis of single-cell RNA-seq data revealed that T cells are the primary components of the tumor microenvironment (TME), are significantly involved in cell differentiation pathways, and play a considerable role in intercellular communication. Based on 10 ML approaches, TCRG classifier were identified to develop and validate. The TCRG classifier exhibited excellent prognostic values with a mean C-index of 0.66 in six cohorts, serving as an independent risk factor (p < 0.01). Additionally, the TCRG exhibited a significant relationship with immune score, immune cell infiltration, immune-associated pathways, immune checkpoint inhibitors, human leukocyte antigen, and immunogenicity. Lastly, IPS, TCIC, TIDE, and IMvigor210 cohort analysis illustrated that the immunotherapy response may be accurately predicted using TCRG.

Conclusion: A TCRG classifier is an excellent resource for predicting a patient's prognosis, potentially guiding the preservation of laryngeal function, and identifying patients who may have a positive response to immunotherapy, which might have profound effects on therapeutic practice.