S100A2 is a predictive biomarker of adjuvant therapy benefit in pancreatic adenocarcinoma

Background Prognosis of patients with pancreatic adenocarcinoma (PAC) remains poor. S100A2 has been recently suggested as a negative prognostic biomarker in PAC. We aimed to investigate its prognostic and/or predictive value in a large independent multicentric cohort of patients with resected PAC. Methods Sequential samples of 471 patients were retrospectively collected; 142 patients did not receive adjuvant treatment (30%) and 329 (70%) received an adjuvant treatment. We measured protein levels of S100A2 by semiquantitative immunohistochemistry with tissue microarrays and correlated with patients’ overall survival (OS) and disease-free survival (DFS). Results S100A2 protein status was obtained in 462 (98%) patients. Its expression was low, moderate or high in 59%, 12% and 2% of cases, respectively. It was not correlated with DFS or OS in the whole population, neither in the subgroup of patients who did not receive adjuvant treatment. However among patients who received an adjuvant therapy, moderate/high levels of S100A2 were significantly associated with longer OS and DFS in multivariate analysis (hazard ratios of 0.63, p = 0.022 and 0.67, p = 0.017, respectively), whereas low S100A2 was not. Interaction tests for adjuvant therapy were statistically significant both for the OS and the DFS ( p = 0.001 and p = 0.023, respectively). On multivariate analysis, S100A2 retained independent predictive values (OS: p p = 0.003) with a significant benefit of adjuvant therapy for those patients with moderate/high S100A2. Conclusions S100A2 expression predicts longer DFS and OS in patients treated with adjuvant therapy and should be evaluated as a predictive biomarker.