USEFULNESS AND LIMITS OF TUMOR-MARKER ASSAYS IN PLEURAL EFFUSIONS

The concentrations of CEA, SCC antigen, and NSE were determined in 190 pleural effusions of various origin. Contrary to SCC antigen and NSE, CEA distinguished significantly (1) effusions secondary to metastatic disease from benign pleural fluids, (2) cytopathologically establishe d malignant effusions from those with negative cytopathology, and (3) malignant effusions due to metastatic disease from those due to mesoth elioma. The cutoff levels for recognizing malignant effusions were obt ained by optimizing the Youden-indices of the marker assays and found to be 13.5 ng/ml for CEA, 6.9 ng/ml for SCC antigen, and 13.5 ng/ml fo r NSE. In effusions secondary to metastatic bronchial carcinomas, the positivity rates were 61.7 % for CEA, 23.4 % for SCC antigen, and 34 % for NSE; in effusions secondary to extrapulmonary tumors, they were 5 0 % for CEA, 12.5 % for SCC antigen, and 30 % for NSE. In tumor patien ts with paramalignant effusions, i.e. those without cytopathologic evi dence of metastases to the pleura, the rates of false-positive marker elevations were 18.5 % for CEA, 25.9 % for SCC antigen, and 37.0 % for NSE. The detection rate of malignancy increased to 87.5 % when cytolo gy (78.6 %) and CEA measurement were used in adjunction. It is conclud ed that SCC antigen and NSE are useless for the diagnosis of malignant effusions. Although the best performance in terms of sensitivity (56. 3 %) and specificity (97.4 % versus benign effusions) was attained wit h the CEA assay, its value must remain equivocal due to the false-posi tive elevations in paramalignant effusions. CEA, however, provided a v aluable aid for distinguishing metastatic carcinoma from malignant mes othelioma.