Detecting cholangiocarcinoma in patients with primary sclerosing cholangitis

Background: Primary sclerosing cholangitis is a progressive cholestatic liver disease associated with cholangiocarcinoma. Brush cytology and serum tumor markers (carcinoembryonic antigen, carbohydrate antigen 19-9 [CA19-9]) have been used to diagnose cholangiocarcinoma, but there are few data comparing their effectiveness. Methods: The effectiveness of brush cytology, carcinoembryonic antigen, and CA19-9 for the diagnosis of cholangiocarcinoma was retrospectively studied by review of patients with primary sclerosing cholangitis. Receiver operator curves were used to identify cutoff points for carcinoembryonic antigen and CA19-9. Results: Of 692 patients with primary sclerosing cholangitis screened, adequate follow-up was obtained in 333, 44 (13%) of whom had a diagnosis of cholangiocarcinoma. Three hundred eighteen brush cytology specimens were obtained in 151 patients; serum carcinoembryonic antigen and CA19-9 levels were obtained in 144 and 55 patients, respectively. The overall sensitivity and specificity of brush cytology were, respectively, 46.4% (95% CI [27.5, 64.5]) and 100% (95% CI [97.2, 100]). A carcinoembryonic antigen >5.2 ng/mL had a sensitivity of 68.0% (95% CI [47.5, 83.9]) and specificity of 81.5% (95% CI [73.9, 87.7]). A CA19-9 >180 U/mL had a sensitivity of 66.7% (95% CI [34.9, 87.7]) and specificity of 97.7% (95% CI [88.2, 99.9]). In the subset of patients in which all 3 tests were obtained, (n = 45, cholangiocarcinoma = 8) the combination of an abnormal carcinoembryonic antigen or CA19-9 had the highest sensitivity: 100% (95% CI [65.1, 100.0]) with a specificity of 78.4% (95% CI [63.1, 89.7]). The combination of a positive brush cytology or an abnormal CA19-9 had a sensitivity and specificity of, respectively, 87.5% (95% CI [50.0, 99.4]) and 97.3% (95% CI [86.2, 99.9]). Conclusions: Screening patients with primary sclerosing cholangitis for cholangiocarcinoma with CA19-9 and carcinoembryonic antigen is reasonable, but the ideal intervals at which to obtain these tests and the cost-effectiveness require further study. (Gastrointest Endosc 2002;56:40-7.)