Benefit of second-line chemotherapy for advanced biliary tract cancer.

Abstract

e15621 Background: Second-line chemotherapy (2LCTX) is increasingly applied in patients (pts) with advanced biliary tract cancer (aBTC), although no randomized trial has so far demonstrated the benefit of this intervention over best supportive care (BSC) alone. In the absence of randomized data, we thus conducted a comparative effectiveness analysis of survival outcomes in aBTC pts treated with BSC±2LCTX. Methods: In this single-center cohort study, we retrospectively included 80 pts with metastatic, recurrent, or inoperable aBTC who completed 1st-line CTX at our department between 2003 and 2016. Thirty-eight of these pts (48%) received 2LCTX+BSC (Fluoropyrimidine (FP) mono: n = 26 (68%), FP-based combination CTX: n = 9 (24%), Others: n = 3 (8%)). Primary endpoint was 18-month overall survival (OS). An inverse-probability-of-treatment-weighted analysis (IPTW) was implemented to rigorously account for imbalances in prognostic variables between the two study groups. Results: During a median follow-up of 14.8 months, we observed 49 deaths. Six-month, 12-month, and 18-month OS estimates were 77%, 53% and 23% in the BSC+2LCTX group, and 29%, 21%, and 14% in patients in the BSC group, for a univariable hazard ratio (HR) for OS of 0.36 (95%CI: 0.20-0.64, p = 0.001). However, pts receiving 2LCTX+BSC had a significantly higher prevalence of favorable prognostic variables, such as a higher Karnofsky Index (p = 0.0001), lower serum bilirubin (p = 0.03), higher hemoglobin (p = 0.002), and higher serum albumin (p = 0.0007). After careful adjustment for these imbalances using IPTW, 2LCTX+BSC was not associated with an OS benefit over BSC alone (Adjusted HR = 0.62, 95%CI: 0.30-1.29, p = 0.201). In IPTW analysis, 6-, 12-, and 18-month OS were 51%, 33% and 14% in the BSC+2LCTX group, and 35%, 29%, and 19% in patients in the BSC group. The beneficial association of 2LCTX with OS was highly time-dependent, with IPTW HRs of 0.07 (p = 0.002), 0.42 (p = 0.05), and 0.53 (p = 0.11) after 3, 6, and 12 months, respectively. Conclusions: Within the limitations of a non-randomized study, our data support the concept that 2LCTX is associated with a short-term OS benefit in pts with aBTC.