Background: Cisplatin–gemcitabine–durvalumab (CGD) has established a new first-line standard in advanced biliary tract cancers (BTC). However, the extent to which trial results translate into real-world populations remains uncertain. We assessed whether the survival outcomes observed in the TOPAZ-1 CGD arm were reproduced in a cohort of patients with advanced BTC treated with CGD in routine clinical practice. Methods: We conducted a retrospective, international, multicenter, real-world study including patients treated with CGD across 55 centers in 12 countries. Reconstructed individual patient-level data from TOPAZ-1 were used as comparator. Real-world patients were classified as TOPAZ-1-in or TOPAZ-1-out according to trial eligibility criteria. Entropy balancing aligned baseline characteristics between cohorts. Overall survival (OS) was compared using weighted pooled survival models and restricted mean survival time (RMST) analyses. Results: Overall, 1382 real-world patients were included in the primary OS analysis. TOPAZ-1 eligibility status was evaluable in 1358 patients (912 TOPAZ-1-in and 446 TOPAZ-1-out); eligibility-stratified survival analyses included 1,351 patients with complete required data (910 TOPAZ-1-in and 441 TOPAZ-1-out). After calibration, baseline characteristics were well balanced. OS was longer in the real-world cohort than in TOPAZ-1 in unweighted (HR 0.83, p = 0.012) and calibrated analyses (HR 0.84, p = 0.018). TOPAZ-1-in patients had longer observed OS than the trial population (HR 0.76, p < 0.001), whereas TOPAZ-1-out patients had similar observed outcomes (HR 0.98, p = 0.85), with consistent calibrated results. Exploratory subgroup analyses showed the largest observed OS differences for ECOG PS 0 (calibrated HR 0.61, p < 0.001) and locally advanced disease (calibrated HR 0.38, p < 0.001); RMST analyses showed increasing absolute between-cohort survival differences over time. Conclusion: The survival outcomes observed in the TOPAZ-1 CGD arm were broadly reproduced in this large real-world cohort. Findings for ECOG PS 0 and locally advanced disease are exploratory and should be interpreted cautiously because of reduced effective sample size and limited empirical overlap.
From trial efficacy to real-world effectiveness: evaluating the generalizability of cisplatin–gemcitabine–durvalumab in advanced biliary tract cancers / Camera, S., Rimini, M., Ikeda, M., Abidoye, O., Lucchetti, J., Antonuzzo, L., Kim, J.W., Nichetti, F., Saborowski, A., Gerdes, C., Pressiani, T., Vivaldi, C., Rapposelli, I.G., Peeters, F., Braconi, C., Pinato, D.J., Tamburini, E., Pircher, C., Foltran, L., Tamberi, S., et al.. - In: JOURNAL OF GASTROENTEROLOGY. - ISSN 0944-1174. - (2026). [10.1007/s00535-026-02535-w]
From trial efficacy to real-world effectiveness: evaluating the generalizability of cisplatin–gemcitabine–durvalumab in advanced biliary tract cancers
Parisi, Alessandro;Lancianese, Alessia;
2026-01-01
Abstract
Background: Cisplatin–gemcitabine–durvalumab (CGD) has established a new first-line standard in advanced biliary tract cancers (BTC). However, the extent to which trial results translate into real-world populations remains uncertain. We assessed whether the survival outcomes observed in the TOPAZ-1 CGD arm were reproduced in a cohort of patients with advanced BTC treated with CGD in routine clinical practice. Methods: We conducted a retrospective, international, multicenter, real-world study including patients treated with CGD across 55 centers in 12 countries. Reconstructed individual patient-level data from TOPAZ-1 were used as comparator. Real-world patients were classified as TOPAZ-1-in or TOPAZ-1-out according to trial eligibility criteria. Entropy balancing aligned baseline characteristics between cohorts. Overall survival (OS) was compared using weighted pooled survival models and restricted mean survival time (RMST) analyses. Results: Overall, 1382 real-world patients were included in the primary OS analysis. TOPAZ-1 eligibility status was evaluable in 1358 patients (912 TOPAZ-1-in and 446 TOPAZ-1-out); eligibility-stratified survival analyses included 1,351 patients with complete required data (910 TOPAZ-1-in and 441 TOPAZ-1-out). After calibration, baseline characteristics were well balanced. OS was longer in the real-world cohort than in TOPAZ-1 in unweighted (HR 0.83, p = 0.012) and calibrated analyses (HR 0.84, p = 0.018). TOPAZ-1-in patients had longer observed OS than the trial population (HR 0.76, p < 0.001), whereas TOPAZ-1-out patients had similar observed outcomes (HR 0.98, p = 0.85), with consistent calibrated results. Exploratory subgroup analyses showed the largest observed OS differences for ECOG PS 0 (calibrated HR 0.61, p < 0.001) and locally advanced disease (calibrated HR 0.38, p < 0.001); RMST analyses showed increasing absolute between-cohort survival differences over time. Conclusion: The survival outcomes observed in the TOPAZ-1 CGD arm were broadly reproduced in this large real-world cohort. Findings for ECOG PS 0 and locally advanced disease are exploratory and should be interpreted cautiously because of reduced effective sample size and limited empirical overlap.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


