A model predicting postoperative survival less than the natural lifespan of hilar cholangiocarcinoma: a multicenter study evaluating surgery necessity.

Abstract

Abstract Background For patients suffering from hilar cholangiocarcinoma (HCCA), it would be no sense if radical surgery didn’t result in better survival than natural lifespan. Therefore, establishing a model predicting postoperative survival less than natural lifespan may assist in surgery decision-making. Methods HCCA patients receiving radical surgery from three centers were enrolled retrospectively. Twelve months was designated the natural lifespan of HCCA, and patients were divided into long-term survival (LTS) and short-term survival (STS) groups by this cutoff value. A nomogram was established from a randomly selected training cohort comprised of 70% of patients and verified on the validation cohort comprised of the remaining 30%. The model was internally validated using 5-fold cross-validation. Decision curve analysis (DCA) and calibration curve were also used to assess the model. In-hospital costs of both groups were compared. Results Out of 245 patients, 67 (27%) had STS, with a mean survival time of 5.67 months, while 178 (73%) had LTS, with a mean survival time of 33.03 months. Among the STS group, 17 died of complications, and the others died of recurrence. Multivariable logistic regression revealed that risk factors for STS were age (OR: 1.06, 95% CI: 1.00-1.11), malnutrition (OR: 3.01, 95% CI: 1.12–8.07), CA125 (OR: 1.01, 95% CI: 1.00-1.02), vascular invasion (OR: 3.50, 95% CI: 1.33–9.21), poor differentiation (OR: 8.51, 95% CI: 2.38–30.44) and lymph node metastasis (OR: 5.04, 95% CI: 1.73–14.66). The C-index of nomogram was 0.834. DCA convinced the power of our model. The in-hospital cost of the STS group was higher than that of the LTS group (mean, 169250.8 and 138118.0, respectively, P = 0.006). Conclusions A model predicting poor prognosis could avoid unnecessary surgery and provide postoperative treatment references.