Objectives: Current methods for diagnosing lung cancer (LC) have varying degrees of risks and complications. MicroRNA (miRNA) is a small molecule noncoding RNA with gene regulation functions. Many studies have shown that miRNA can be used for the diagnosis of LC, but there are differences in diagnostic accuracy. Therefore, we aim to systematically review and meta-analyze published articles to comprehensively evaluate the diagnostic value of miRNA for LC.
Materials and methods: We searched the PubMed, Embase, and Cochrane databases, and calculated the area under the curve (AUC) by plotting the summary receiver operator characteristic curve using the sensitivity and specificity of each included study. The AUC was calculated and the likelihood ratio was plotted to assess the diagnostic accuracy of miRNA. We used QUADAS-2 in Review Manager 5.3 to evaluate the quality of all the articles. The other analyses were performed using the STATA 12.0 software.
Results: We included a total of 29 articles, 98 studies, and the qualities of all the articles were satisfactory. The overall pooled parameters calculated from all studies were as follows: sensitivity = 0.77, specificity = 0.83, positive likelihood ratio (PLR) = 4.6, negative likelihood ratio (NLR) = 0.28, and AUC = 0.87 for miRNA diagnosis. It had significant advantages over other biomarkers. Subgroup analysis showed that when combined four or more miRNA for the diagnosis of LC, the parameters were as follows: sensitivity = 0.90, specificity = 0.93, PLR = 13.2, NLR = 0.11, and AUC = 0.97.
Conclusion: Four or more miRNA combination could be used for the diagnosis of LC. Besides this, we also found that miRNA showed a greater advantage in distinguishing LC from benign lung diseases than distinguishing between LC and normal people. Our findings provided a new way of thinking about the clinical diagnosis of LC from a nonmorphological aspect.
Keywords: AUC; diagnosis; lung neoplasms; microRNAs; sensitivity and specificity.
© 2019 Wiley Periodicals, Inc.