Development of an immune escape-related gene prognostic signature using integrated single-cell and bulk RNA sequencing for immune microenvironment profiling in thyroid cancer.

Abstract

Immune escape drives cancer progression and therapy resistance, yet its prognostic role and impact on the tumor immune microenvironment in thyroid cancer remain unclear. We integrated single-cell and bulk RNA sequencing data to systematically characterize immune escape and its clinical significance. scRNA-seq analysis characterized cellular heterogeneity and quantified immune escape activity via AUCell. A prognostic gene signature was constructed from differential expression analysis combined with univariate Cox and LASSO regression, and validated using Kaplan-Meier and time-dependent ROC analyses. The immune landscape was profiled using ssGSEA, CIBERSORT, and ESTIMATE, while immunophenoscore (IPS) was used to predict immunotherapy responsiveness. Functional enrichment, CellChat, SCISSOR, tumor mutation burden (TMB), and CellMiner analyses were further performed to explore underlying mechanisms and therapeutic implications. A three-gene signature (CD9, NPC2, PSMB9) effectively stratified patients into highand low-risk groups with distinct survival outcomes. Low-risk tumors exhibited an "immune-hot" phenotype with increased CD8+ T cells and activated NK cells, higher checkpoint expression, and elevated IPS, suggesting greater immunotherapy sensitivity. In contrast, high-risk tumors showed an immune-cold microenvironment with M2 macrophage enrichment. Despite higher TMB, high-risk tumors displayed reduced immune activity, indicating impaired immune recognition. Single-cell analysis further identified MIF and CCL signaling as key mediators of multicellular immune evasion. Overall, our single-cell-informed immune escape signature provides a promising framework for thyroid carcinoma risk stratification and offers insights into personalized immunotherapy.