Abstract
Poor survival of patients with locally advanced head and neck squamous cell carcinoma (LA-HNSCC) is partly due to early diagnosis difficulties and the lack of reliable biomarkers for predicting treatment outcomes. In the discovery cohort, plasma-derived extracellular vesicles (EVs) from LA-HNSCC patients (n = 48) and healthy volunteers (n = 12) were used for profiling for microRNA (miRNA) expression by NanoString analysis. Ten EV-associated miRNAs were differentially expressed between LA-HNSCC patients and healthy volunteers. Subsequently, the results were validated in the individual discovery and additional cases (HNSCC, n = 73; control, n = 20) by quantitative RT-PCR. Among 10 EV-miRNAs, four (miR-27b-3p, miR-491-5p, miR-1910-5p, and miR-630) were significantly dysregulated in LA-HNSCC patients (n = 73) compared with healthy volunteers (n = 20). The miRNA prediction models were developed to discriminate HNSCC patients from healthy volunteers. The model using miR-491-5p was selected as a diagnostic biomarker for LA-HNSCC with a sensitivity and specificity of 46.6% and 100%, respectively (P < .001). The dynamic changes of miRNA model score (ΔmiRNAs) were determined using scores pre- and postdefinitive treatment to further investigate the prognostic value of miRNA prediction models. The univariate and multivariate analyses indicated that ΔmiR-491-5p was the most powerful and independent prognostic indicator for overall survival (hazard ratio [HR] 5.66, 95% confidence interval, 1.77-18.01; P = .003) and disease-free survival (HR 2.82, 95% CI, 1.13-7.05; P = .027) of HNSCC patients. In summary, the miR-491-5p prediction model could serve as a blood-based diagnostic marker for LA-HNSCC. Moreover, ΔmiR-491-5p could be a potential monitoring prognostic marker to reflect the survival of HNSCC patients.