Abstract:

Objective

To develop a prognostic model for head and neck squamous cell carcinoma（HNSCC） utilizing immune and metabolic-related genes， and to assess its efficacy in predicting the prognosis of HNSCC patients.

Methods

The HNSCC patient dataset was obtained from the TCGA database， while the GSE65858 dataset was sourced from the GEO database. Additionally， immune and metabolic-related gene sets were acquired from the IMMPORT and MsigDB databases， respectively. Subsequently， the TCGA dataset was partitioned into a training cohort and a validation cohort. Univariate and multifactor Cox regression，along with LASSO regression， were employed to develop prognostic models for immune-related gene signatures （IMRGs） in the training cohort. Tissue expression of prognostic genes was detected by qPCR.Internal validation was carried out in the TCGA validation cohort， followed by external validation in the GSE65858 dataset. Immune infiltration analysis， Gene Set Enrichment Analysis （GSEA） pathway enrichment analysis， and tumor mutation load analysis were conducted to delineate pathway mechanism disparities between high and low-risk subgroups.

Results

Utilizing TCGA training set data， a prognostic model for head and neck squamous cell carcinoma was developed， comprising 10 genes （HLA-F， SLC11A1， GNRH1，LGR5， ICOS， and HPRT1 were highly expressed， while DES， BTC， DDO， and CDO1 were lowly expressed in tumor tissues）. The survival duration of the low-risk subgroup significantly exceeded that of the high-risk group. Both univariate and multivariate COX analyses confirmed the risk score as an independent prognostic determinant. The combination of a risk score and various clinical features proves effective in predicting patient prognosis， as evidenced by an area under the receiver operating curve of 0.825 and a C-index of 0.751.GSEA enrichment and immune cell infiltration analyses revealed elevated B cell expression in the low-risk group. Moreover， gene correlation and tumor mutation load analyses indicated heightened levels of DNA replicators （MCM6 and POLD3）， mismatch repair gene （MSH6）， epithelial mesenchymal transformation gene （LOXL2）， and tumor mutation rate in the high-risk group.

Conclusion

The IMRGs model for HNSCC patients constructed based on 10 genes can effectively predict the prognosis of patients， and provide a new target for clinical prognosis identification and treatment.

Key words: Squamous cell carcinoma of the head and neck, Immune, Metabolism, Risk model