Computed tomography (CT) plays a central role in lung cancer diagnosis. However, CT has relatively low specificity, presenting a challenge in clinical settings. We previously identified 12 microRNAs (miRNAs) whose expressions in tumor tissues were associated with lung cancer.
Using quantitative reverse transcriptase polymerase chain reaction, we aimed to identify miRNA biomarkers in sputum that could complement CT for diagnosis of lung cancer.
In a training set consisting of 66 lung cancer patients and 68 cancer-free smokers, 10 of the 12 miRNAs were differentially expressed between the cases and controls (p ≤ 0.01). From the miRNAs, a logistic regression model was built on the basis of miR-31 and miR-210, both of which had the best prediction for lung cancer, producing an area under receiver operating characteristic curve of 0.83. Combined use of the two miRNAs yielded 65.2% sensitivity and 89.7% specificity, CT had 93.9% sensitivity and 83.8% specificity for lung cancer diagnosis. Notably, combined analysis of the miRNA biomarkers and CT produced a higher specificity than does CT used alone (91.2% versus 83.8%; p < 0.05). The diagnostic performance of the biomarkers was confirmed in a testing set comprising 64 lung cancer patients and 73 cancer-free smokers.
The sputum miRNA biomarkers might be useful in improving CT for diagnosis of lung cancer, but further independent validation on an external and prospective cohort of patients is required.