Despite the use of radiologic, endoscopic, and laparoscopic staging techniques, the rate of nontherapeutic laparotomies in patients with hilar cholangiocarcinoma remains high. This study evaluated the accuracy of preoperative high-resolution computed tomograpy (HRCT) to determine resectability in this setting.

Preoperative helical HRCT (2 contrast phases, rapid intravenous contrast bolus, 2.5-mm section thickness) for 32 consecutive patients who underwent laparotomy for the diagnosis of hilar cholangiocarcinoma from 2000 to 2005 were reviewed by a hepatobiliary radiologist. The accuracy of HRCT was determined by comparison of the imaging interpretation to intraoperative and pathologic findings. The chi-square test was used to identify imaging findings that best predicted unresectability.

Fourteen of the 32 (44%) study patients were unresectable (extension along bile duct, 4; peritoneal metastases, 4; vascular encasement, 3; noncontiguous liver metastases, 2; N2 lymphadenopathy, 1). HRCT correctly predicted resectability in 17 of 18 patients who underwent therapeutic laparotomy (sensitivity = 94%). HRCT correctly predicted the inability to resect in 11 of the remaining 14 cases (specificity = 79%). In the 3 cases in which HRCT predicted resectability and the patient was unresectable, subcentimeter peritoneal disease, a subcentimeter liver metastasis, and distal bile duct involvement were responsible factors. The negative and positive predictive values of HRCT were 92% and 85%, respectively. Individual radiographic findings that best predicted unresectability were peritoneal spread (P = .015) and hepatic artery (P = .006) or portal vein (P = .002) involvement.

Preoperative HRCT accurately predicts resectability in patients with hilar cholangiocarcinoma. Identification of specific radiographic features, in particular major vascular involvement and peritoneal abnormalities, is now used by our group to avoid unnecessary laparotomy.