A Metabolic Gene Signature to Predict Overall Survival in Head and Neck Squamous Cell Carcinoma

Abstract EN

Background.

Head and neck squamous cell carcinoma (HNSCC) is a common malignancy that emanates from the lips, mouth, paranasal sinuses, oropharynx, larynx, nasopharynx, and from other pharyngeal cancers.

The availability of high-throughput expression data has made it possible to use global gene expression data to analyze the relationship between metabolic-related gene expression and clinical outcomes in HNSCC patients.

Method.

In this study, we used RNA sequencing (RNA-seq) data from the cancer genome atlas (TCGA), with validation in the GEO dataset to profile the metabolic microenvironment and define potential biomarkers for metabolic therapy.

Results.

We extracted data for 529 patients and 327 metabolic genes (198 upregulated and 129 downregulated genes) in the TCGA database.

Carbonic anhydrase 9 (CA9) and CA6 had the largest logFCs in the upregulated and downregulated genes, respectively.

Our Cox regression model data showed 51 prognostic-related genes with lysocardiolipin acyltransferase 1 (LCLAT1) and choline dehydrogenase (CHDH) being associated with the highest risk (HR=1.144, 95% CI=1.044~1.251) and the lowest risk (HR=0.580, 95% CI=0.400~0.839) in HNSCC, respectively.

We next used the ROC curve to evaluate whether the differentially expressed metabolic-related genes could serve as early predictors of HNSCC.

The findings showed an AUC of 0.745 and 0.618 in the TCGA and GEO analysis, respectively.

Besides, the ability for the genes to predict clinicopathological HNSCC status was analyzed and the data showed that the AUC for age, gender, grade, stage, T, M, and N was 0.520, 0.495, 0.568, 0.606, 0.577, 0.476, and 0.673, respectively, in the TCGA dataset.

On the other hand, the AUC for age, gender, stage, T, M, N, smoking, and HPV16-pos was 0.599, 0.531, 0.622, 0.606, 0.616, 0.550, 0.614, 0.519, and 0.397, respectively, in the GEO dataset.

Conclusion.

Taken together, our study unearths a novel metabolic gene signature for the prediction of HNSCC prognosis based on the TCGA dataset.

Our signature might point out the metabolic microenvironment disorders and provides potential treatment targets and prognostic biomarkers.