Diagnosis of malignant versus tuberculous ascites using tumor markers and globulin ratios in serum and ascites: A Fisher discriminant model.

BACKGROUND AND STUDY AIMS This study was conducted to investigate the significance of tumor and biochemical markers in serum and ascitic fluid in the differential diagnosis of tuberculous and malignant ascites. PATIENTS AND METHODS Based on findings from natural orifice transluminal endoscopic surgery and postoperative pathology or cytology of 63 patients, they were divided into the malignant group (31 patients) and the tuberculous group (32 patients). Levels of tumor markers, albumin, globulin, and lactate dehydrogenase were measured simultaneously. Data were statistically analyzed, and a Fisher discriminant model was established. The receiver operating characteristic curve was constructed to confirm the discriminant value. RESULTS The levels of carcinoembryonic antigen (CEA), cancer antigen 125 (CA125), cancer antigen 19-9 (CA 19-9), and globulin in serum and ascitic fluid were different between the tuberculous and malignant ascites groups (P < .05). The ratios of ascites-to-serum levels of CEA, CA125, and CA 19-9, as well as the ratio of serum-to-ascites of globulin levels, were different between the two groups (P < .05). The Fisher discriminant model was established based on the ascites-to-serum ratios of CEA, CA125, and CA 19-9 levels and the serum-to-ascites ratio of globulin levels. The area under the curve was 0.908, the sensitivity was 0.838 (26/31), and the specificity was 0.875 (28/32). CONCLUSION A Fisher discriminant model can be established using serum and ascites tumor markers and globulin ratios, which is valuable in the differential diagnosis of tuberculous versus malignant ascites.