Clinical outcomes and risk stratification in unresectable biliary tract cancers undergoing radiation therapy.

ABSTRACT

BACKGROUND: Biliary tract cancers (BTC) are rare and aggressive malignancies originating from intrahepatic and extrahepatic bile ducts and the gallbladder. Surgery is the only curative option, but due to late-stage diagnosis, is frequently not feasible, leaving chemotherapy as the primary treatment. Radiotherapy (RT) can be an effective alternative for patients with unresectable, non-metastatic BTC despite the generally poor prognosis and significant variability. To help manage patients with unresectable BTC who receive RT, we aimed to identify prognostic markers that could aid in predicting overall survival (OS). METHODS: A retrospective cohort study was conducted at the University of Pennsylvania, involving seventy-eight patients with unresectable BTC treated with definitive intent RT. Comprehensive demographic, clinical, and treatment-related data were extracted from the electronic medical records. Univariate and multivariate Cox regressions were employed to identify predictors of OS after RT. A biomarker model was developed for refined survival prediction. RESULTS: The cohort primarily comprised patients with good performance status without significant hepatic dysfunction at presentation. The predominant treatment approach involved hypofractionated RT or concurrent 5FU-based chemoRT. Median OS after RT was 12.3 months, and 20 patients (15.6%) experienced local progression with a median time of 30.1 months. Univariate and multivariate analyses identified CA19-9 (above median) and higher albumin-bilirubin (ALBI) grades at presentation as significant predictors of poor OS. Median OS after RT was 24 months for patients with no risk factors and 6.3 months for those with both. CONCLUSIONS: Our study demonstrates generally poor but significantly heterogeneous OS in patients with unresectable BTC treated with RT. We have developed a biomarker model based on CA19-9 and ALBI grade at presentation that can distinguish sub-populations with markedly diverse prognoses. This model can aid the clinical management of this challenging disease.