Diagnosis of lung cancer using salivary miRNAs expression and clinical characteristics

Lung cancer (LC), the primary cause for cancer-related death globally is a diverse illness with various characteristics. Saliva is a readily available biofluid and a rich source of miRNA. It can be collected non-invasively as well as transported and stored easily. The process is also reproducible and cost-effective. The aim of this study was to evaluate the salivary expression of microRNAs let-7a-2, miR-221, and miR-20a in saliva and evaluate their efficacy, using multiple logistic regression (MLR) model, in diagnosis of lung cancer.

Samples of saliva were obtained from 40 lung cancer patients (20 lung adenocarcinoma and 20 lung squamous cell carcinoma) and 20 healthy controls. The levels of let-7a-2, miR-221, and miR-20a expression in saliva were assessed by RT-qPCR. Receiver operating characteristic (ROC) curve was utilized to assess the potential significance of miRNAs in saliva for lung cancer diagnosis with the use of multiple logistic regression (MLR), principal component analysis, and machine learning methods.

Diagnostic odds ratio (DOR) of miR-20a in lung adenocarcinoma diagnosis versus healthy control was higher than miR-221, and DOR of miR-221 was higher than let-7a-2. miR-20a demonstrated a higher DOR for small cell lung carcinoma versus healthy control compared to let-7a-2, which in turn exhibited a higher DOR than miR-221. MLR of miR-221, let-7a-2, miR-20a, and smoking habit using main effects led to accuracy of 0.725 (sensitivity: 0.80, specificity: 0.65) and AUC = 0.795 for differentiation of small-cell lung carcinoma from lung adenocarcinoma. Our results showed that MLR based on salivary miRNAs could diagnose LUAD and SCLC from healthy control using main effects and two-way interactions with the accuracy of 0.90 (sensitivity = 0.95 and specificity = 0.85).

A salivary miRNA-based MLR model is a promising diagnostic tool for lung cancer, offering a non-invasive screening option for high-risk asymptomatic individuals.