• NanoString system
• Malignant pleural mesothelioma
• molecular markers
• benign pleural lesions
• differential diagnosis

Background and aim: malignant pleural mesothelioma (MPM) is a rare, asbestos-related tumor, extremely aggressive and unresponsive to therapies. There are three main MPM histological subtypes: epithelioid, biphasic and sarcomatoid, with the latter one having the worst prognosis. The diagnosis of MPM is mainly based on histopathological evaluation of pleural biopsies and, for patients who cannot undergo biopsy, on cytopathological examination of pleural effusions. However, the morphological features of a lesion are not always clear and it is frequently hard to differentiate malignant from benign mesothelial proliferations.
In this study we aimed to identify a new molecular tool for the differential diagnosis of MPM and benign pleural lesions. In detail, we designed a panel of 117 genes, which had been previously reported as deregulated in MPM, and performed a gene expression profiling of malignant and benign histological and cytological pleural specimens. Furthermore, we evaluated whether the identified gene panel could be useful in the molecular characterization of MPM histotypes and in the identification of new prognostic biomarkers.
Methods: this study was retrospectively conducted on RNA from patients with MPM and benign pleural lesions. The gene expression analysis was performed by the NanoString system. In the first part of this study we determined the gene expression profile of 25 epithelioid MPM, 15 benign pleural mesothelial hyperplasia (MH) and 14 mesothelial pleural samples analysed in a blind way. Then, we analysed cytological specimens from 23 epithelioid MPM and 11 MH patients. Data analysis mainly included unsupervised cluster analysis and molecular classification by the Uncorrelated Shrunken Centroid (USC) algorithm. In the second part of this work, we extended the gene expression analysis to 8 sarcomatoid and 5 biphasic MPM tissues and expression data of MPM subtypes were compared using a non-parametric Kruskal-Wallis test and the Dunn's test with Bonferroni correction. Finally, we compared gene expression levels and follow-up data of 15 epithelioid MPM patients using the Log-rank test.
Results: the diagnostic model based on the USC algorithm correctly classified all the analysed histological samples as malignant or benign on the basis of gene expression profiles. On the other hand, the identified diagnostic model appropriately classified all the cytological specimens as malignant or benign, except for two misdiagnosed MPM cases and one MH case. The comparison between the three histotypes revealed 39 genes differentially expressed among them. Finally, a poorer overall survival was associated with a higher expression of MCM2, PLK1 and CCNB2 genes. Whereas, a poorer progression free survival was associated with a lower expression of EGR3, LGALS3, TGFBR2 and TACC1 genes and a higher expression of MKI67 and PLK1 genes.
Conclusions: we defined a new molecular tool which combines molecular data (gene expression) and computational analysis (USC algorithm) to classify a pleural mesothelial proliferation as benign or malignant, not only on histological but also on cytological specimens. Our gene panel showed an accuracy in differentiating epithelioid MPM from MH of 100% on histological samples and of 91.2% on cytology. Moreover, this panel includes genes expressed in a histotype specific manner and genes whose expression seemed to affect prognosis in epithelioid MPM.
Although absolutely promising, these results need to be validated on a larger and prospective cohort to better define the application and reliability of this new molecular instrument and to favour its introduction in the clinical practice of MPM.