Clinical trajectories from first-line chemotherapy to best supportive care in unresectable pancreatic cancer: a 10-year real-world retrospective study.

BACKGROUND

Real-world treatment trajectories and predictors of transition to best supportive care (BSC) in unresectable pancreatic cancer (UPC) remain underexplored.

METHODS

This single-center retrospective study (2014-2025) included 274 patients. Overall survival (OS) and time-to-BSC were evaluated using Kaplan-Meier methods and a complete-case multivariable Cox model. Proportional hazards were assessed using Schoenfeld residuals, and treatment-line exposure using a 90-day landmark analysis.

RESULTS

Among 274 patients, 248 (90.5%) initiated chemotherapy; 61.3% and 21.8% received ≥2 and ≥3 lines, respectively. Median OS was 9.7 months and increased across treatment-line categories (p < 0.001), and 84.7% transitioned to BSC. Higher C-reactive protein-to-albumin ratio (CAR) was associated with shorter BSC-free survival (p < 0.001). Although baseline CAR did not differ significantly between the chemotherapy and upfront-BSC groups (p = 0.076), it was independently associated with worse OS (HR 2.75), together with ECOG performance status ≥2 (HR 2.64) and liver metastasis (HR 1.62), and its association with mortality attenuated over time. In the 90-day landmark cohort (n = 215), receipt of ≥2 lines by day 90 was associated with longer subsequent OS (13.0 vs. 8.3 months, p = 0.012).

CONCLUSIONS

Real-world treatment trajectories in UPC involved frequent transitions to BSC across successive therapy lines. Baseline CAR was independently associated with worse OS and earlier BSC transition and may provide prognostic information beyond performance status. CAR should not be used alone for treatment decisions but may identify patients at risk of early deterioration and prompt timely goals-of-care discussions. Prospective validation, including serial assessment, is warranted.