Objective: JCOG1106, a randomized phase II trial conducted to compare chemoradiotherapy (S-1 concurrent radiotherapy) with (Arm B) or without (Arm A) induction chemotherapy using gemcitabine in patients with locally advanced pancreatic cancer, showed a more favorable long-term survival in Arm A. This study was aimed at exploring whether some subgroups classified by the systemic inflammatory response might derive greater benefit from either treatment.
Methods: All subjects eligible for JCOG1106 were included in this analysis (n = 51/49 in Arm A/B). This exploratory subgroup analysis was performed by Cox regression analysis to investigate the impact of the systemic inflammatory response, as assessed based on the serum C-reactive protein, serum albumin (albumin), Glasgow Prognostic Score and derived neutrophil-lymphocyte ratio, at the baseline on overall survival. P values <0.1 for the interaction were regarded as denoting significant association.
Results: Glasgow prognostic score showed significant treatment interactions for overall survival. Hazard ratios of Arm B to Arm A were 1.35 (95% confidence interval, 0.82-2.23) in the Glasgow Prognostic Score 0 (C-reactive protein ≤10 mg/L and albumin ≥35 g/L) (n = 44/34 in Arm A/B) and 0.59 (95% confidence interval, 0.24-1.50) in the Glasgow Prognostic Score 1/2 (C-reactive protein >10 mg/L and/or albumin <35 g/L) (n = 7/15) (P-interaction = 0.06). C-reactive protein alone and albumin alone also showed significant treatment interactions for overall survival.
Conclusions: Survival benefits of induction chemotherapy in chemoradiotherapy for locally advanced pancreatic cancer were observed in patients with elevated Glasgow Prognostic Score, high C-reactive protein and low albumin. These results suggest that systemic inflammatory response might be considered to apply induction chemotherapy preceding chemoradiotherapy.
Keywords: Glasgow prognostic score; S-1 concurrent radiotherapy; gemcitabine; treatment interaction.
© The Author(s) 2023. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected].