BACKGROUND: Carbohydrate associated antigen (CA19-9) has been approved by the FDA as a biomarker for monitoring treatment effect in pancreatic cancer. However, the value of serum CA19-9 as a biomarker of response to chemotherapy in bile duct cancer is unclear. The aim of this study was to determine if a decline in CA19-9 (CA19-9 response) during chemotherapy is predictive of survival in patients with inoperable bile duct cancer.

METHODS: Consecutive patients with inoperable bile duct cancer treated at a University Hospital were retrospectively included in an investigational cohort (n = 212). Three validation cohorts were established including patients 1) participating in phase I/II trials at a Danish Hospital (n = 71), 2) identified retrospectively in a Canadian cohort (n = 196) and 3) randomized in the ABC-02 trial (n = 410). Patients with a baseline CA19-9 and at least one CA19-9 value measured 10-12 weeks after the start of chemotherapy were included. Multivariate Cox regression analyses were performed.

RESULTS: Patients meeting the criteria to be included were 54 in the investigational cohort and 34, 68 and 148 in the three validation sets, respectively. Multivariate analysis included radiological response, performance status, bilirubin, gender, site of cancer, extend of disease, CA19-9 at baseline and age. A hazard ratio (HR) of 0.60 (95%CI: 0.44-0.80, p = 0.0005) for death in CA19-9 responders was reached in the investigational cohort. The predictive value of CA 19-9 response was confirmed in all three validation cohorts.

CONCLUSIONS: CA19-9 response is a robust predictor of survival in patients with inoperable bile duct cancer in four independent data sets.