Prediction of survival with second-line therapy in biliary tract cancer: Actualisation of the AGEO CT2BIL cohort and European multicentre validations

Background: The benefit of second-line chemotherapy (L2) over standard first-line (L1) gemcitabine plus cisplatin (GEMCIS) or oxaliplatin (GEMOX) chemotherapy in advanced biliary tract cancer (aBTC) is unclear. Our aim was to identify and validate prognostic factors for overall survival (OS) with L2 in aBTC to guide clinical decisions in this setting. Methods: We performed a retrospective analysis of four prospective patient cohorts: a development cohort (28 French centres) and three validation cohorts from Italy, UK and France. All consecutive patients with aBTC receiving L2 after GEMCIS/GEMOX L1 between 2003 and 2016 were included. The association of clinicobiological data with OS was investigated in univariate and multivariate Cox analyses. A simple score was derived from the multivariate model. Results: The development cohort included 405 patients treated with L1 GEMOX (91%) or GEMCIS. Of them, 55.3% were men, and median age was 64.8 years. Prior surgical resection was observed in 26.7%, and 94.8% had metastatic disease. Performance status (PS) was 0, 1 and 2 in 17.8%, 52.4% and 29.7%, respectively. Among 22 clinical parameters, eight were associated with OS in univariate analysis. In multivariate analysis, four were independent prognostic factors (p < 0.05): PS, reason for L1 discontinuation, prior resection of primary tumour and peritoneal carcinomatosis. The model had the Harrell's concordance index of 0.655, a good calibration and was validated in the three external cohorts (N = 392). Conclusion: We validated previously reported predictive factors of OS with L2 and identified peritoneal carcinomatosis as a new pejorative factor in nearly 800 patients. Our model and score may be useful in daily practice and for future clinical trial design.