Diagnostic Certainty of a Mesothelioma Diagnosis Based on Immunophenotype
Frank Schneider, Agnieszka Onisko, Rajmohan Murali. University of Pittsburgh Medical Center, Pittsburgh, PA; Memorial Sloan-Kettering Cancer Center, New York, NY
Background: A diagnosis of mesothelioma is usually made only after establishing the immunophenotype of the neoplastic cells. The International Mesothelioma Interest Group suggested using two antibodies each for mesothelioma and the diagnoses in the differential diagnosis. We present here a Bayesian model to generate post-test probabilities (PTP) for various stain combinations using epithelioid mesothelioma (EpM) as an example.
Design: Pre-test probabilities for 42 different neoplasms were determined by counting their relative occurrence in 300 random pleural biopsies and resections during an 8-year period (32 different tumors) and by adding to that list other tumors less often seen in the pleura. A model was created using SMILE and GeNIe, an inference engine and development environment for probabilistic models (Decision Systems Laboratory, Pittsburgh, PA), considering the immunophenotype of the various differential diagnoses. Under the assumption that a cytokeratin AE1/3 stain is positive, the PTP of EpM was calculated for combinations of commonly used stains.
Results: The four most common pleural tumors were carcinomas of breast, lung, ovary and EpM (pre-test probabilities 37%, 24%, 11%, 5%, respectively) in women, and carcinomas of lung, esophagus, EpM and biphasic mesothelioma (pre-test probabilities 26%, 6%, 19%, 6%, respectively) in men. PTPs varied slightly based on patient gender. A combination of positive mesothelioma markers Calretinin and CK5/6 with negative adenocarcinoma markers TTF-1 and either MOC-31, BerEp4 or B72.3 (4-stain panel) provided a PTP of 90%. Using CEA instead of the latter three markers reduced the PTP to 70%. When a 6-stain panel was employed by adding D2-40 or WT-1 and using two of the adenocarcinoma makers in addition to TTF-1, the PTP increased to 98% or greater. In the four stain panel, having one stain show an unexpected result decreases the PTP to below 20%, but PTPs can be restored to >80% by adding a both another mesothelioma and adenocarcinoma marker.
Conclusions: A combination of two mesothelial markers and two adenocarcinoma markers provides sufficient diagnostic certainty for EpM, especially since in that situation the PTPs for the differential diagnoses are far below 10%. A panel of at least three stains each is useful if one of the markers show unexpected staining. Calculation of the PTP can be performed for any stain combination and result, and could be used as marker of diagnostic certainty in the clinical and medico-legal setting.
Wednesday, March 21, 2012 9:30 AM
Poster Session V # 298, Wednesday Morning