Peripheral immune signature resembles tumor microenvironment and predicts clinical outcomes in head and neck squamous cell carcinoma

Abstract

The contour of the tumor immune microenvironment (TIME) is very important for tumor prognostic prediction but hard to be characterized in clinical practice. It is unclear practice whether the peripheral immune signature (pIS) reflects the TIME as a feasible prognostic indicator for head and neck squamous cell carcinoma (HNSCC) patients. Here, we enrolled 599 HNSCC patients from three domestic institutes to explore the relationship between the pIS and survival. The peripheral neutrophil-to-lymphocyte ratio (pNLR) was screened out as a significant prognostic variable through multivariable COX regression analyses. An inverse correlation between pNLR and survival was found in the data of these 599 patients. Meanwhile, the bulk tumor RNA-seq data of 913 cases were downloaded from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases to identify the prognosis-associated TIME features. The TIME feature was consistent to the finding of clinical data, in which high tissue NLR predicted a poor prognosis. Differentially expressed immune-related gene (DEIRG) enrichment analysis also showed a trend that the gene sets in patients with a good prognosis were enriched in lymphocyte-related functions, while those with a poor prognosis were enriched in neutrophil-related functions. At the same time, the well prediction performance of our model based on DEIRGs was verified in both TCGA and GEO cohorts. Finally, the correlation between pIS and the TIME was confirmed in a small independent cohort of 30 HNSCC patients. A positive correlation was confirmed prospectively between the pNLR and the TIME pattern in our independent cohort. Our findings provide evidence that the pNLR is a feasible prognostic signature that reflects the TIME patterns to some extent in HNSCC.