Development of a novel head and neck squamous cell carcinoma prognostic signature by bulk/single-cell sequencing data integration.

Abstract

Identifying cell subpopulations conferring unfavorable prognosis in cancer holds clinical significance. Here, we sought to identify prognostic cell subsets and develop a novel, prognostic signature for head neck squamous cell carcinoma (HNSCC). Highly prognostic cell subpopulations in HNSCC were identified by integrating single-cell and bulk transcriptomic datasets. The prognostic signature and nomogram were developed by least absolute shrinkage and selection operator and multivariate Cox regression analyses based on significantly upregulated genes in this specific cell subpopulation, respectively. The qRT-PCR experiments were utilized for independent validation in our patient cohort. A specific cancer cell subset associated with unfavorable prognoses was identified. Functional dissections revealed that its transcriptional programs were significantly enriched in E2F, epithelial-to-mesenchymal transition, and glycolysis. A novel prognostic signature comprising six genes was developed and further validated. Risk scores based on qRT-PCR data robustly stratified patients into subgroups with distinct prognoses. A nomogram integrated from this signature and clinical stage had superior performance. Our model derived from integrative analyses of single-cell and bulk RNA-sequencing is a novel, robust prognostic biomarker for HNSCC.