Cytological diagnosis of pleural mesothelioma (PM) is controversial even using ancillary markers (BAP1, MTAP and CDKN2A). Here, we aimed to prospectively validate a previously developed 117-gene expression panel for the differential cytological diagnosis of epithelioid, biphasic PM and mesothelial hyperplasia. Seventy-seven pleural effusions were classified using the 117-gene expression levels (NanoString system). Sixty-eight cases were also screened for ancillary markers. The performance of both gene panel and ancillary markers was evaluated using ROC metrics. A score using the top consistently deregulated genes between epithelioid and biphasic PM was built to subtype malignant effusions. The panel alone reached a diagnostic accuracy (0.89) comparable to the best marker combination (BAP1 plus MTAP: 0.88). Ancillary tests missed 8 PMs, seven of which were correctly classified by the panel. The score built by averaging the expression levels of MSLN, CLDN15 and CFB showed an accuracy of 0.80 in subtyping epithelioid and biphasic effusions. The 117-gene panel is effective for PM cytological diagnosis of epithelioid and biphasic PM. This tool can be complementary to ancillary markers, reducing invasive procedures and allowing an earlier diagnosis. Finally, the possibility to subtype PM on effusions strengthens the panel role in PM diagnosis and management.

Pleural mesothelioma (PM) comprises three main subtypes: epithelioid, biphasic and sarcomatoid, which have a different impact on prognosis and treatment definition. However, PM subtyping can be complex given the inter- and intra-tumour morphological heterogeneity. We aim to use immunohistochemistry (IHC) to evaluate five markers (Mesothelin, Claudin-15, Complement Factor B, Plasminogen Activator Inhibitor 1, and p21-activated Kinase 4), whose encoding genes have been previously reported as deregulated among PM subtypes. Immunohistochemical expressions were determined in a case series of 73 PMs, and cut-offs for the epithelioid and non-epithelioid subtypes were selected. Further validation was performed on an independent cohort (30 PMs). For biphasic PM, the percentage of the epithelioid component was assessed, and IHC evaluation was also performed on the individual components separately. Mesothelin and Claudin-15 showed a good sensitivity (79% and 84%) and specificity (84% and 73%) for the epithelioid subtype. CFB and PAK4 had an inferior performance, with higher sensitivity (89% and 84%), but lower specificity (64% and 36%). In the biphasic group, all markers showed a different expression when comparing epithelioid with sarcomatoid areas. Mesothelin, Claudin-15 and CFB can be useful in subtype discrimination. PAI1 and PAK4 can improve component distinction in biphasic PM.