miRNAs as Biomarkers for Diagnosing and Predicting Survival of Head and Neck Squamous Cell Carcinoma Patients

Abstract

Head and Neck Squamous Cell Carcinoma (HNSCC) is the sixth most common cancer worldwide. These tumors originate from epithelial cells of the upper aerodigestive tract. HNSCC tumors in different regions can have significantly different molecular characteristics. While many microRNAs (miRNAs) have been found to be involved in the regulation of the carcinogenesis and pathogenesis of HNSCC, new HNSCC related miRNAs are still being discovered. The aim of this study was to explore potential miRNA biomarkers that can be used to diagnose HNSCC and prognose survival of HNSCC patients. For this purpose, we chose a panel of 12 miRNAs: miR-146a-5p, miR-449a, miR-126-5p, miR-34a-5p, miR-34b-5p, miR-34c-5p, miR-217-5p, miR-378c, miR-6510-3p, miR-96-5p, miR-149-5p, and miR-133a-5p. Expression of these miRNAs was measured in tumor tissue and neighboring healthy tissue collected from patients diagnosed with HNSCC (n = 79) in either the oral cavity, oropharynx, or larynx. We observed a pattern of differentially expressed miRNAs at each of these cancer locations. Our study showed that some of these miRNAs, separately or in combination, could serve as biomarkers distinguishing between healthy and tumor tissue, and their expression correlated with patients' overall survival.

Keywords: head and neck squamous cell carcinoma; laryngeal cancer; miRNA; oral cancer; oropharyngeal cancer.

Figure 1

The scatter plot (A) indicating that the non-detect rate increases with average ΔCt by location, tissue type and gene. The boxplots showing the comparison of −ΔCt residues between observed expressions and non-detects, (B) if conventional imputation of setting the Ct of each non-detect to 40 was conducted, and (C) if expectation-maximization (EM) algorithm was used to estimate the Ct of each non-detect.

Figure 2

Expression of miRNAs in tumor tissue and nearby healthy tissue of HNSCC patients. (A) Boxplots showing the comparison of expression of 12 miRNAs between tumor tissue and nearby healthy tissue in the oral cavity, and larynx. Expression was calculated using −∆∆Ct method; values above 0 denote upregulation, and values below 0 denote downregulation of miRNA, compared with healthy tissue. Wilcoxon signed-rank test was performed. (B) Heatmap and dendrogram showing the overall expression profile of the 12 miRNAs in patient samples. Samples were grouped by tissue type and tumor location. The colors in the heatmap represent the value of −∆∆Ct ranging from downregulation (dark blue) to upregulation (red). (C) Summary of findings on miRNA expressions as biomarkers in diagnosis of HNSCC (“↓” denotes downregulation and “↑” denotes upregulation).

Figure 3

ROC curves showing miRNA biomarkers with potential in distinguishing tumor tissue from nearby healthy tissue in oral (A) and laryngeal (B) cancer.

Figure 4

ROC showing accuracy of combined miRNA expression as a biomarker in distinguishing oral (A) and laryngeal (B) tumor tissue compared with neighboring healthy tissue. (A) Combined expression of miR-6510-3p and miR-34c-5p as a biomarker distinguishing oral tumor tissue from nearby healthy tissue with a threshold of −0.194 determined by the best model: logit(tumor) = −2.239 − 1.127(miR-6510-3p) + 1.392(miR-34c-5p). (B) Combined expression of miR-449a-5p, miR-6510-3p, and miR-149-5p as a biomarker in distinguishing laryngeal tumor tissue from nearby healthy tissue with a threshold of −0.212 determined by the best model: logit(tumor) = −1.855 + 0.438(miR-449a-5p) − 1.246(miR-6510-3p) + 1.473(miR-149-5p). In both models, if logit(tumor) score is greater than its corresponding threshold, the diagnosis is positive, otherwise the diagnosis is negative.

Figure 5

Relation between miRNA expression and overall survival. (A,B) Kaplan Meier survival curves showing comparison of overall survival between low and high miRNA expression defined by optimal cutoff of −∆∆Ct in oral (A) and laryngeal (B) cancers. (C) Summary of multiple Cox regression analysis for oral and laryngeal cancer investigating hazard ratio (HR) of high expression vs. low expression of each miRNA in separate survival models with age at diagnosis, T staging and N staging under control. HR < 1 denotes high miRNA expression at lower risk compared to low expression. HR > 1 denotes high miRNA expression at higher risk than low expression.