OBJECTIVES: Tumor marker analysis in ascites has been proposed as a measure to aid in the diagnosis of malignancy. The objectives of this study were to establish tumor marker cut-offs and determine the diagnostic performance of measuring CEA, CA 19-9 and AFP in ascites for differentiating between non-malignant and malignant etiologies. DESIGN AND METHODS: Ascites from 137 patients (83 non-malignant, 54 malignant) was assayed for CEA, CA 19-9 and AFP concentrations by immunoassay. Diagnostic cut-offs were established via ROC curve analysis. Performance was compared to cytology findings and patient history following medical chart review. Analysis based on cytological findings in combination with tumor marker testing, as well as subset analysis by tumor marker secretion was also performed. RESULTS: Concentrations of CEA, CA 19-9 and AFP were significantly higher in patients with malignant ascites versus non-malignant etiologies. The diagnostic cut-off, sensitivity and specificity for CEA were 3.5 ng/mL, 31% and 95%, respectively; for CA 19-9 were 72 U/mL, 30% and 95%; and for AFP were 5 ng/mL, 17% and 95%. Using cytological findings in conjunction with tumor marker results improved the sensitivity of CEA, CA 19-9 and AFP to 57.4%, 64.8%, and 59.3%, respectively. Improvement in sensitivity was seen when subset analysis by causative malignancy was performed. CONCLUSIONS: Tumor marker analysis in ascites, especially in subset analysis by type of malignancy, demonstrated utility for differentiating non-malignant from malignant etiologies. This analysis should not replace cytology, but offers potential for differentiation in situations where cytology is inconclusive, or in patients with suspected malignancies known to secrete these markers.
OBJECTIVES:Tumor marker analysis in ascites has been proposed as a measure to aid in the diagnosis of malignancy. The objectives of this study were to establish tumor marker cut-offs and determine the diagnostic performance of measuring CEA, CA 19-9 and AFP in ascites for differentiating between non-malignant and malignant etiologies. DESIGN AND METHODS: Ascites from 137 patients (83 non-malignant, 54 malignant) was assayed for CEA, CA 19-9 and AFP concentrations by immunoassay. Diagnostic cut-offs were established via ROC curve analysis. Performance was compared to cytology findings and patient history following medical chart review. Analysis based on cytological findings in combination with tumor marker testing, as well as subset analysis by tumor marker secretion was also performed. RESULTS: Concentrations of CEA, CA 19-9 and AFP were significantly higher in patients with malignant ascites versus non-malignant etiologies. The diagnostic cut-off, sensitivity and specificity for CEA were 3.5 ng/mL, 31% and 95%, respectively; for CA 19-9 were 72 U/mL, 30% and 95%; and for AFP were 5 ng/mL, 17% and 95%. Using cytological findings in conjunction with tumor marker results improved the sensitivity of CEA, CA 19-9 and AFP to 57.4%, 64.8%, and 59.3%, respectively. Improvement in sensitivity was seen when subset analysis by causative malignancy was performed. CONCLUSIONS:Tumor marker analysis in ascites, especially in subset analysis by type of malignancy, demonstrated utility for differentiating non-malignant from malignant etiologies. This analysis should not replace cytology, but offers potential for differentiation in situations where cytology is inconclusive, or in patients with suspected malignancies known to secrete these markers.