Background: The published circulating miRNA signatures proposed for early-stage non-small cell lung cancer (NSCLC) detection are inconsistent and difficult to replicate. Reproducibility and validation of an miRNA simple signature of NSCLC are prerequisites for translation to clinical application.
Methods: The serum level of miR-223 and miR-29c, emerging from published studies, respectively, as a highly sensitive and a highly specific biomarker of early-stage NSCLC, was measured with droplet digital PCR (ddPCR) technique in an Italian cohort of 75 patients with stage I-II NSCLC and 111 tumor-free controls. By ROC curve analysis we evaluated the miR-223 and miR-29c performance in discerning NSCLC cases from healthy controls.
Results: Reproducibility and robust measurability of the two miRNAs using ddPCR were documented. In a training set (40 stage I-II NSCLCs and 56 controls), miR-223 and miR-29c, respectively, showed an AUC of 0.753 [95% confidence interval (CI), 0.655-0.836] and 0.632 (95% CI, 0.527-0.729) in identifying NSCLC. Combination of miR-223 with miR-29c yielded an AUC of 0.750, not improved over that of miR-223 alone. Furthermore, in an independent blind set (35 stage I-II NSCLCs and 55 controls), we validated serum miR-223 as an effective biomarker of stage I-II NSCLC (AUC = 0.808; 95% CI, 0.712-0.884), confirming the miR-223 diagnostic performance reported by others in Chinese cohorts.
Conclusions: Using ddPCR technology, miR-223 was externally validated as a reproducible, effective serum biomarker of early-stage NSCLC in ethnically different subjects. Combination with miR-29c did not improve the miR-223 diagnostic performance.
Impact: Serum miR-223 determination may be proposed as a tool for refining NSCLC risk stratification, independent of smoking habit and age.
©2019 American Association for Cancer Research.