Background: Accurate prediction of prognosis for patients with intrahepatic cholangiocarcinoma (ICC) remains a challenge. We sought to define a preoperative risk tool to predict long-term survival after resection of ICC. Study Design: Patients who underwent hepatectomy for ICC at 1 of 16 major hepatobiliary centers between 1990 and 2015 were identified. Clinicopathologic data were analyzed and a prognostic model was developed based on the regression β-coefficients on data in training set. The model was subsequently assessed using a validation set. Results: Among 538 patients, most patients had a solitary tumor (median tumor number 1; interquartile range 1 to 2) and median tumor size was 5.7 cm (interquartile range 4.0 to 8.0 cm). Median and 5-year overall survival was 39.0 months and 39.0%, respectively. On multivariable analyses, preoperative factors associated with long-term survival included tumor size (hazard ratio [HR] 1.12; 95% CI 1.06 to 1.18), natural logarithm carbohydrate antigen 19-9 level (HR 1.33; 95% CI 1.22 to 1.45), albumin level (HR 0.76; 95% CI 0.55 to 0.99), and neutrophil to lymphocyte ratio (HR 1.05; 95% CI 1.02 to 1.09). A weighted composite prognostic score was constructed based on these factors: [9 + (1.12 × tumor size) + (2.81 × natural logarithm carbohydrate antigen 19-9) + (0.50 × neutrophil to lymphocyte ratio) + (−2.79 × albumin)]. The model demonstrated good performance in the testing (area under the curve 0.696) and validation (0.691) datasets. The model performed better than both the T categories (area under the curve 0.532) and the cumulative stage classifications in the American Joint Committee on Cancer staging manual, 8th edition (area under the curve 0.559). When assessing risk of death within 1 year of operation, a risk score ≥25 had a positive predictive value of 59.8% compared with a positive predictive value of 35.3% for American Joint Committee on Cancer staging manual, 8th edition T4 disease and 31.8% for stage IIIB disease. Conclusions: Postsurgical long-term outcomes could be predicted using a composite weighted scoring system based on preoperative clinical parameters. The preoperative risk model can be used to inform patient to provider conversations and expectations before operation.
ASJC Scopus subject areas