Distinguishing benign from malignant mesothelial cells in effusions by Glut-1, EMA, and Desmin expression: an evidence-based approach

Abstract

Distinguishing malignant mesothelioma (MM) from reactive mesothelial hyperplasia (RM) may be difficult in effusions. This study tested the hypothesis that immunocytochemistry (IC) in effusion cell blocks (CB) can distinguish MM from RM and that the results may be applied to individual specimens. External validation of a risk score (RS) model associating sensitivity and specificity was applied to an external set of MM and RM specimens from a separate institution. Forty three effusion cytology CBs of 25 confirmed malignant mesotheliomas were compared to CBs of 23 benign mesothelial effusions without inflammation and 13 reactive mesothelial proliferations associated with inflammation. Glut-1, EMA, and Desmin expression were evaluated by immunocytochemistry on CBs. Each antibody was compared using ROC values, where the area under the curve (AUC) was 0.90, 0.82, and 0.84 for Glut-1, EMA, and Desmin, respectively. Logistic regression (LR) analysis was applied to a combination of Glut-1 and EMA. A combined ROC curve was modeled for Glut-1 and EMA (AUC = 0.93). A RS = 2 × (Glut-1%) + 1 × (EMA%) was created from this ROC curve. When applied to an external set of MM and RM, the RS resulted in an ROC with AUC = 0.91. In conclusion, a RS derived from a LR of Glut-1 and EMA IC greatly improves the distinction between MM from RM cells in individual effusions. The study illustrates principles of evidence-based pathology concerning internal and external test performance in the differential diagnosis of MM versus RM.