Genome-wide microRNA expression profiling in malignant pleural effusion to identify a ten-microRNA signature.

Abstract

e23123 Background: Pleural carcinosis caused by tumors of the chest (e.g., lung and breast cancer) or elsewhere in the body (e.g., ovarian carcinoma) that metastasize to the visceral and/or parietal pleura. Recurrent malignant pleural effusion due to pleural carcinosis is one of the most common findings in oncology. The aim of the study was to identify a miRNA signature associated with clinicians’ prescription of patients with MPE. Methods: We used qRT-PCR assay to evaluate a wide range miRNAs (1920 miRNAs) signature measured from pleural effusion between patients with cancer and healthy controls. From the data of our previous studies we selected a panel of 96 miRNAs related to malignant pleural effusion. Data were compared by three classification methods software for statistics and modeling. We used a first set of 77 pleural effusion samples as training group, including 27 patients affected by Lung Adenocarcinomas (LAC), 9 with other lung cancer, 17 with other tumors and 24 inflammatory patients as negative controls. Moreover, we used more 64 pleural effusion samples for double-blind predictive study. Data analysis was performed using a machine learning approach of a Support Vector Machine classifier with a Student's t-test feature selection procedure. Results: We identified a panel of ten miRNAs with optimum classification performance. The combination of these 10-miRNAs alone could discriminate MPE cases from negative controls with an AUC of 0.969 (accuracy = 93.5%; specificity = 91.7%). The selected panel of another 10-miRNAs could separate lung cancer cases from negative controls with an AUC of 0.973 (accuracy = 94.8%; specificity = 95.8%), and a small panel of 4-miRNAs could good discriminate LAC cases from negative controls with an AUC of 0.946 (accuracy 88.9%; specificity = 100%). The accuracy rate of the double-blind predictive sensitivity value was 90.9% and specificity value was 95.8% for MPE patients with lung cancer of miRNA signature. Conclusions: Our miRNAs profile may serve as a new biomarker for MPE diagnosis. Otherwise, we identified a 10-miRNAs signature for MPE patients with lung cancer and a 4-miRNAs for MPE patients with lung adenocarcinoma.