Landmark survival analysis and impact of anatomic site of origin in prospective clinical trials of biliary tract cancer

•Patients with gallbladder cancer have worse overall survival than those with other primary anatomic origins of biliary tract cancer. •Reduced risk of death vs. gallbladder cancer was maintained in those receiving combination chemotherapy. •Landmark survival rates provide relevant prognostic information for patients who survive for some time. •Patients receiving combination therapy have better landmark survival than those receiving monotherapy. •Patients with intrahepatic cholangiocarcinoma or cholangiocarcinoma-not specified also have better landmark survival. Background & Aims Whether all patients with advanced biliary tract cancer (aBTC) should be included in prospective trials, irrespective of the anatomic site of origin, is debated. Herein, we aimed to assess the survival impact of anatomic site of origin in prospective clinical trials of aBTC using landmark survival analysis. Methods Patients enrolled into prospective first-line aBTC clinical trials (Jan 97–Dec 15) were included. Overall survival (OS) was analysed using Cox proportional hazard regression; landmark survival (LS) and 95% CIs were calculated. Results Overall, 1,333 patients were included: median age 63 years (range 23–85); 46% male; 84% ECOG-PS0/1; 25% with locally advanced disease, 72% with metastatic, 3% not reported (NR). Patients were treated with mono-chemotherapy (23%), cisplatin/gemcitabine (36%), other combinations (39%), or NR (2%). Median OS was 10.2 months (95% CI 9.6–10.9). All sites (treatment-adjusted) had decreased risk of death vs. gallbladder cancer (GBC) (p <0.001). This reduced risk vs. GBC was maintained in those receiving cisplatin/gemcitabine for extrahepatic cholangiocarcinoma (p <0.001) and intrahepatic cholangiocarcinoma (IHC, p <0.001), but not in cholangiocarcinoma-not specified (CCA-NS, p = 0.82) or ampullary carcinoma (p = 0.96). One-year OS rates amongst patients who survived beyond 1, 2, 3 and 4 years post-trial registration were 37%, 45%, 61%, and 63%, respectively. For patients who survived 1 year, those receiving combination therapy vs. mono (p = 0.008) (acknowledging potential selection bias) and those with IHC and CCA-NS vs. GBC had better LS (both p <0.05). Metastatic disease was associated with shorter LS than locally advanced disease (p = 0.002). ECOG-PS and gender were not associated with LS (p >0.05, p = 0.08 respectively). Conclusions GBC is associated with worse OS than other BTC sites and should be considered as a stratification factor in clinical trials. LS rates enable adjusted prognostication for aBTC survivors. Lay summary Patients with gallbladder cancer have worse overall survival compared to those with biliary tract cancers of different primary origin. Thus, gallbladder cancer should be considered as a stratification factor in future clinical trials. Landmark survival rates enable adjusted prognosis prediction for patients with advanced biliary tract cancer who survive for some time. Whether all patients with advanced biliary tract cancer (aBTC) should be included in prospective trials, irrespective of the anatomic site of origin, is debated. Herein, we aimed to assess the survival impact of anatomic site of origin in prospective clinical trials of aBTC using landmark survival analysis. Patients enrolled into prospective first-line aBTC clinical trials (Jan 97–Dec 15) were included. Overall survival (OS) was analysed using Cox proportional hazard regression; landmark survival (LS) and 95% CIs were calculated. Overall, 1,333 patients were included: median age 63 years (range 23–85); 46% male; 84% ECOG-PS0/1; 25% with locally advanced disease, 72% with metastatic, 3% not reported (NR). Patients were treated with mono-chemotherapy (23%), cisplatin/gemcitabine (36%), other combinations (39%), or NR (2%). Median OS was 10.2 months (95% CI 9.6–10.9). All sites (treatment-adjusted) had decreased risk of death vs. gallbladder cancer (GBC) (p <0.001). This reduced risk vs. GBC was maintained in those receiving cisplatin/gemcitabine for extrahepatic cholangiocarcinoma (p <0.001) and intrahepatic cholangiocarcinoma (IHC, p <0.001), but not in cholangiocarcinoma-not specified (CCA-NS, p = 0.82) or ampullary carcinoma (p = 0.96). One-year OS rates amongst patients who survived beyond 1, 2, 3 and 4 years post-trial registration were 37%, 45%, 61%, and 63%, respectively. For patients who survived 1 year, those receiving combination therapy vs. mono (p = 0.008) (acknowledging potential selection bias) and those with IHC and CCA-NS vs. GBC had better LS (both p <0.05). Metastatic disease was associated with shorter LS than locally advanced disease (p = 0.002). ECOG-PS and gender were not associated with LS (p >0.05, p = 0.08 respectively). GBC is associated with worse OS than other BTC sites and should be considered as a stratification factor in clinical trials. LS rates enable adjusted prognostication for aBTC survivors.