Implementation of Evidence-Based Immunostain Panels for Diagnosis of Pleural Effusions with Atypical Epithelioid Cells: Theory vs. Practice
AM Marchevsky, AE Walts. Cedars-Sinai Medical Center, Los Angeles, CA
Background: Based on the incidence of carcinomas, cost-effectiveness of various immunostains and application of Bayesian statistics, we previously developed 3 evidence-based immunohistochemical (EB-IHC) panels for the diagnosis of pleural effusions (PE) with atypical epithelioid cells (AEC).
Design: The EB-IHC panel to distinguish benign from malignant PE included MOC31, BER-EP4, calretinin, and CK5/6. The panel for origin of tumor in the malignant PE included TTF-1, PSA, CDX2, calretinin for male and TTF-1, ER, CA125, and calretinin for female patients.The rationale for utilization of these panels was presented to our 5 cytopathologists and ordering the panels was facilitated by computer codes. Use of the panels was recommended but not mandated. One year later we assessed utilization and diagnostic applicability of the panels and reasons for non-compliance with their usage.
Results: 955 PE were diagnosed during the year. Immunostains were used in 122 PE with AEC (48 benign and 74 malignant). Malignant PE in females (N=46) included 16 breast, 10 lung, 9 mullerian, 3 neuroendocrine/SCC lung, 2 cervix, 1 case each of colon, stomach, duodenum/pancreas, bile duct, and 2 unknown primaries. Malignant PE in males (N=28) included 8 lung, 5 colon, 3 prostate, 3 neuroendocrine/SCC lung, 2 pancreas, 1 each of stomach, urothelium, renal cell, larynx, breast, mesothelioma, and unknown primary. The distribution of primaries was similar to that in the previous study. Complete EB-IHC panels were used in 16% (20) cases (15 benign, 5 malignant). Stains selected from the panels were used in different combinations in 21% (26) cases (13 benign, 13 malignant). In the remainder 62% (76) cases (20 benign, 56 malignant), 1 to 11 additional IHC (Add-IHC) were used per case. Use of Add-IHC identified the tumor origin in only 7 (9%) of these 76 cases (6 neuroendocrine/SCC and 1 renal cell carcinoma). The most frequently used Add-IHC that did not yield additional diagnostic information were CK7, CK20, WT-1, GCDFP-15, mammaglobin, and B72.3.
Conclusions: Reasons for poor compliance included limitations in EB-IHC panel design and reluctance of pathologists to modify their practice. A more pragmatic approach that combines clinical parameters, immunostains selected from the panels rather than sequential use of complete panels, targeted use of neuroendocrine markers, and an algorithm for diagnosis of PE with AEC with periodic reassessment of its clinical utility has been adopted.
Wednesday, March 24, 2010 1:00 PM
Poster Session VI # 59, Wednesday Afternoon