OA22.06 Refinement of the Prognostic miR-Score for Use in Diagnostic Specimens from Chemo-Naïve Malignant Pleural Mesothelioma Patients

Abstract

A 6-microRNA signature (miR-Score, Kirschner et al 2015) was previously demonstrated to show high prognostic accuracy in a series of surgical specimens (with and without induction chemotherapy). In the present study we investigated these microRNAs in an independent cohort of MPM patients all treated with induction chemotherapy followed by extrapleural pneumonectomy (EPP). The main focus of the study was to evaluate the possible effects of induction chemotherapy on microRNA expression and to refine and validate the miR-Score for use in chemo-naïve diagnostic specimens. We identified a cohort of 120 MPM patients who received chemotherapy followed by EPP between 1999 and 2014 at University Hospital Zurich. At present microRNA analysis (RT-qPCR) has been carried out in 34 pairs of chemo-naïve (diagnostic biopsy) and chemo-treated (EPP) specimens. Paired-samples t-test was employed to determine differences in microRNA expression pre- and post-chemotherapy. Accuracy of the miR-Score in predicting a good prognosis (>20 months survival post-surgery) was evaluated by ROC curve analysis. In addition, binary logistic regression modelling was used to build a refined miR-Score. Applying the miR-Score to chemo-naïve diagnostic specimens revealed an area under the ROC curve (AUC) of 0.65 (95% CI: 0.46-0.84), and the same analysis on the EPP specimens gave an AUC of 0.57 (95% CI: 0.37-0.77). Therefore, the accuracy of the miR-Score was lower than observed in the previous study. However, pairwise comparison of microRNA expression before and after chemotherapy showed that although not reaching statistical significance, the levels of several microRNAs were lower following induction chemotherapy. We next employed binary logistic regression modelling on microRNA levels in chemo-naïve tissue to determine whether a refined microRNA signature less susceptible to chemotherapy-induced changes could be created. A refined miR-Score consisting of miR-221 and miR-30e, the two microRNAs least affected by chemotherapy, achieved AUCs of 0.77 (95% CI: 0.61-0.94) and 0.80 (95% CI: 0.64-0.96) in diagnostic and EPP specimens, respectively. When applied to samples from the previous study, the refined score resulted in an AUC of 0.72 (95% CI: 0.54-0.90). This validation and refinement study has shown that the expression of several miR-Score microRNAs appears to be affected by standard chemotherapy. A refined miR-Score was generated which is less susceptible to the effect of chemotherapy and may have prognostic value when applied to diagnostic specimens. Further validation in additional paired samples and investigation of the effect of cisplatin, pemetrexed and gemcitabine on microRNA expression are ongoing.