The PancreasView gemcitabine transcriptomic signature predicts response to gemcitabine in patients with resected pancreatic ductal adenocarcinoma

Abstract

Introduction: In the current study, we aimed to assess the efficacy of a gemcitabine response predictive signature that is part of the PancreasView transcriptomic predictive tool (Gem + or Gem−). Methods: We used a cohort of pancreatic ductal adenocarcinoma patients treated from the Massachusetts General Hospital who underwent upfront resection. Results: In this cohort of 43 patients, 20 (46.5%) received adjuvant gemcitabine (GEM arm) and 23 (53.5%) did not receive any adjuvant chemotherapy. Among the 43 patients, the Gem signature defined a subgroup of 16 patients (37.2%) who were sensitive (Gem+) and 27 (62.8%), who were resistant to gemcitabine (Gem−). The Gem+ patients who received adjuvant gemcitabine had significantly better median disease-free survival (DFS) compared to the Gem− patients (NR until 72 months of follow-up vs 19.0 months; stratified hazard ratio [HR]: 0.19; 95% CI, 0.04- 0.86; P = .032) and longer median cancer-specific survival (CSS) (NR until 96 months of follow-up vs 37.0 months; stratified HR: 0.18; 95% CI, 0.04-0.85; P = .030) when treated with gemcitabine. The gemcitabine signature remained an independent predictive factor for DFS (HR: 0.41; 95% CI, 0.19-0.89; P = .024) and CSS (HR: 0.47; 95% CI, 0.22-1.23; P = .059) after adjusting for clinicopathological characteristics in an unstratified univariate Cox hazard model. Conclusions: This validation of the gemcitabine predictive transcriptomic signature in an independent cohort from Massachusetts General Hospital reinforces the robustness and reliability of this tool. This study highlights the potential of the signature to aid in the personalization of chemotherapy and enhance patient outcomes in pancreatic ductal adenocarcinoma.