Evaluation of 12 antibodies for distinguishing epithelioid mesothelioma from adenocarcinoma: identification of a three-antibody immunohistochemical panel with maximal sensitivity and specificity.

We evaluated the sensitivity and specificity of 10 monoclonal and two polyclonal antibodies for distinguishing epithelioid mesothelioma from adenocarcinoma (AdCA) using immunohistochemistry (IHC). The antibodies were directed against the mesothelial-associated antigens mesothelin, calretinin, cytokeratin 5, thrombomodulin, Wilms' tumor-1 (WT-1) gene product and HBME-1, and the nonmesothelial antigens Lewis-Y blood group (antibody BG8), MOC-31, BerEp4, CD15, and carcinoembryonic antigen (CEA) family. The 133 tumors evaluated included 65 malignant epithelioid mesotheliomas, 22 lung AdCAs, 27 ovarian serous carcinomas, 24 breast carcinomas, and five gastric carcinomas. Diagnoses were based on clinical, histologic, ultrastructural, and/or IHC findings. Calretinin had the best sensitivity for mesothelioma (95%), followed by HBME-1 (84%), WT-1 (78%), cytokeratin 5 (76%), mesothelin (75%), and vimentin and thrombomodulin (68%). Thrombomodulin had the best specificity for mesothelioma (92%), followed by cytokeratin 5 (89%), calretinin (87%) vimentin (84%), and HBME-1 (45%). When ovarian carcinomas were excluded from the analysis, the specificity of mesothelin and WT-1 for the diagnosis of mesothelioma increased to 90 and 81%, respectively. The sensitivity of the nonmesothelial antigens for AdCA was organ dependent, with BG8 performing best in the breast cancer group (96%), and BerEp4, BG8, MOC-31 performing best in the lung cancer group (100%). The specificity of the nonmesothelial antigens for AdCA was 98% for BG8 and CEA, 97% for CD15, 95% for BerEp4, and 87% for MOC-31. A novel statistical analysis technique employing logic regression analysis identified a three-antibody immunohistochemical panel including calretinin, BG8, and MOC-31, which provided over 96% sensitivity and specificity for distinguishing epithelioid mesothelioma from AdCA.