Malignancy-related ascites: a diagnostic pitfall of spontaneous bacterial peritonitis by ascitic fluid polymorphonuclear cell count.

To define patients with an ascitic fluid polymorphonuclear cell count > or = 250 cells/mm3 or > or = 500 cells/mm3 but without spontaneous bacterial peritonitis, 166 patients with sterile cirrhotic ascites, 46 patients with spontaneous bacterial peritonitis, 123 patients with hepatocellular carcinoma, 67 patients with peritoneal carcinomatosis or massive liver metastasis and 12 patients with other miscellaneous diseases were studied. The sensitivity, specificity and accuracy of the diagnosis of spontaneous bacterial peritonitis were 100, 86 and 88% with the cut-off value of an ascitic fluid polymorphonuclear cell count > or = 250 cells/mm3; and were 93, 91 and 92% with that value > or = 500 cells/mm3, respectively. With the cut-off value > or = 250 cells/mm3 or > or = 500 cells/mm3, the prevalence was 18% or 14% in hepatocellular carcinoma; and 30% or 19% in peritoneal carcinomatosis or massive liver metastasis. The ascitic fluid lactate concentration was insensitive and nonspecific. Among the patients with an ascitic fluid polymorphonuclear cell count greater than the cut-off values, an ascitic fluid erythrocyte count > or = 10,000 cells/mm3, a ratio of ascitic fluid erythrocyte to total leukocyte count > or = 100, and the ratio of ascitic fluid polymorphonuclear cell to total leukocyte count < or = 75% indicated hepatocellular carcinoma, while serum to ascites albumin gradient < or = 1.1 g/dl and a ratio of ascitic fluid polymorphonuclear cell to total leukocyte count < or = 75% indicated peritoneal carcinomatosis or massive liver metastasis.