Identification of hub lncRNAs in head and neck cancer based on weighted gene co-expression network analysis and experiments

Abstract

Head and neck squamous cell carcinoma (HNSCC) ranks as the sixth most common cancer among systemic malignant tumors, with 600 000 new cases occurring every year worldwide. Since HNSCC has high heterogeneity and complex pathogenesis, no effective prognostic indicator has yet been identified. Here, we aimed to identify a lncRNA signature associated with the prognosis of HNSCC as a potential new biomarker. LncRNA expression data were downloaded from The Cancer Genome Atlas database. A polygenic risk score model was constructed by using Lasso-Cox regression analysis. Weighted gene co-expression network analysis (WGCNA) was applied to analyze the co-expression modules of lncRNAs associated with the prognosis of HNSCC. The robustness of the signature was validated in testing and external cohorts. Polymerase chain reaction was performed to detect the expression levels of identified lncRNAs in cancer and adjacent tissues. We constructed an 8-lncRNA signature (LINC00567, LINC00996, MTOR-AS1, PRKG1-AS1, RAB11B-AS1, RPS6KA2-AS1, SH3BP5-AS1, ZNF451-AS1) that could be used as an independent prognostic factor of HNSCC. The signature showed strong robustness and had stable prediction performance in different cohorts. WGCNA results showed that modules related to risk score mainly participated in biological processes such as blood vessel development, positive regulation of catabolic processes, and regulation of growth. The prognostic risk score model based on lncRNA for HNSCC may help clinicians conduct individualized treatment plans.

Keywords: The Cancer Genome Atlas; WGCNA; head and neck cancer; risk score prognostic model.

Fig. 1

Flowchart of the method of this study.

Fig. 2

(A) The Lasso regression model and cross‐validation method were used to screen lncRNAs. When the number of variables was 59, we obtained the minimum partial likelihood deviance. (B) Regression coefficient graph of lncRNAs in the Lasso regression model.

Fig. 3

(A) Distribution of risk scores of patients with HNSCC in the training cohort. (B) Risk scores and survival states of patients with HNSCC in the training cohort. (C) Heat map of risk scores based on lncRNA expression in patients with HNSCC in the training cohort. (D) ROC curve of the prognostic model constructed in the training cohort. (E) Kaplan–Meier survival curve of high‐ and low‐risk patients’ OS rates in the training cohort.

Fig. 4

(A) Distribution of risk scores of patients with HNSCC in the testing cohort. (B) Risk scores and survival states of patients with HNSCC in the testing cohort. (C) Heat map of risk scores based on lncRNA expression in patients with HNSCC in the testing cohort. (D) ROC curve of the prognostic model constructed in the testing cohort. (E) Kaplan–Meier survival curve of high‐ and low‐risk patients’ OS rates in the testing cohort.

Fig. 5

(A) Distribution of risk scores of patients with HNSCC in the external validation cohort. (B) Risk scores and survival states of patients with HNSCC in the external validation cohort. (C) Heat map of risk scores based on lncRNA expression in patients with HNSCC in the external validation cohort. (D) ROC curve of the prognostic model constructed in the external validation cohort. (E) Kaplan–Meier survival curve of high‐ and low‐risk patients’ OS rates in the external validation cohort.

Fig. 6

(A) Hierarchical cluster analysis to remove outliers. (B) Gene clustering dendrogram according to the adjacency‐based dissimilarity of hierarchical clustering. The color piece below represents the module identified by the Dynamic Cut Tree method. (C) Heat map of correlations between the module and clinical characteristics. The number represents the correlation in the color piece, and the P‐value is below. Red is positively correlated, and green is negatively correlated. (D) Chart of the results of the GO and KEGG enrichment analyses of the turquoise module. The length of the bar represents the number of genes enriched, and the names on the right are the pathway names. (E) Scatter diagram of the correlations between the turquoise module genes and fustat.

Fig. 7

GSEA results based on the training cohort samples.

Fig. 8

PCR of the eight lncRNAs in the HNSCC and normal samples. (A) Expression of LINC00567. (B) Expression of LINC00996. (C) Expression of MTOR‐AS1. (D) Expression of PRKG1‐AS1. (E) Expression of RAB11B‐AS1. (F) Expression of ZNF451‐AS1. (G) Expression of RPS6KA2‐AS1. (H) Expression of SH3BP5‐AS1. Error bars represent means ± SD.