Search

Background Previous studies on borderline resectable (BR) pancreatic cancer (PC) included patients with heterogenous preoperative states; however, the definition of resectability for PC has evolved. We aimed to investigate the prognostic factors for PC other than anatomical resectability in those who underwent upfront resection and discuss the optimal treatment strategy for PC. Methods We retrospectively examined 431 patients who underwent upfront surgery with curative intent between 2007 and 2014. The association between clinical characteristics and survival outcomes was assessed by stratifying patients according to risk factors. The patients were categorized into four groups based on anatomical (resectable [R]/BR) and biological features (CA19-9 ≤500/>500 U/mL): anatomical R with CA19-9 ≤500 U/mL (favorable-R); anatomical BR with CA19-9 ≤500 U/mL (favorable-BR); anatomical R with CA19-9 >500 U/mL (risky-R); and anatomical BR with CA19-9 >500 U/mL (risky-BR). Results Overall, 320 and 111 patients had anatomical R- and BR-PC, respectively. A modified Glasgow prognostic score = 2 (hazard ratio [HR]: 1.73), NLR>5 (hazard ratio [HR]: 1.54), CA19-9 >500 U/mL (HR: 1.86), and anatomical BR (HR: 1.38) were independent prognostic factors for overall survival. The risky-R group had likely worse prognosis (16 months vs. 19 months, P = 0.0605) and a significantly higher early recurrence rate (36% vs 18%, P = 0.0231) than the favorable-BR group. Conclusions It is essential to stratify and distinguish PC patients at a high risk of worse prognosis. Risky-R was an unfavorable prognostic factor and should thus be considered in the decision-making for treatment with neoadjuvant chemotherapy, in addition to anatomical BR-PC.