Serum miRNAs as biomarkers for early diagnosis of non-small cell lung cancer (NSCLC).

Lung cancer is the leading cause of cancer-related deaths worldwide. Late diagnosis, attributable at least partly to the lack of non-invasive screening methods, contributes to its high mortality. Recently, microRNAs (miRNAs) attracted the attention of scientists as possible lung cancer biomarkers since they are stable, retrievable and dysregulated in many types of cancers. We performed a critical review of the literature and selected 8 miRNAs, 4 of which with high sensitivity/AUC and 4 with high specificity. These miRNAs compose a two-step screening: in the first step the high sensitivity miRNAs are measured (miR-223, miR-20a, miR-145, miR-448), whereas in the second step the high specificity miRNAs are measured (miR-210, miR-628-3p, miR-29c, miR-1244). We aimed to identify the two best performing miRNAs. All analyzed miRNAs were expressed at significantly higher levels in NSCLC patients compared to controls. Among the high-sensitivity miRNAs, miR-223 performed best; among high-specificity miRNAs, miR-29c performed best. We then analysed miR-223 and miR-29c to determine the best combination and cut-offs for early lung cancer diagnosis. We measured the levels of miR-223 and miR-29c in a Training set and tested combinations and cut-off values that allowed to obtain the best separation between patients and controls. We found that the formula “miR-223>500 copies/μl OR miR-29c>50 copies/μl” had the highest sensitivity (75%) and acceptable specificity (50%). The formula was then applied to a blind Validation set and we found that it had a very high sensitivity of 92.5%, despite a poor specificity of 37.5%. Our results suggest that our test holds great potential for screening of patients at risk for stage I-II NSCLC.