Treatment Planning for Resectable Pancreatic Cancer Based on Prediction of Postoperative Early Recurrence

Abstract

Background: Resectable pancreatic cancer (R-PC) often recurs early after radical resection, which is associated with poor prognosis. Predicting early recurrence preoperatively is useful for determining the optimal treatment. Methods: 178 patients, diagnosed with R-PC on the CT image and undergoing radical resection at our institution from 2005 to 2019 were retrospectively analyzed. Patients with recurrence within six months after resection formed the early recurrence (ER) group, while other patients constituted non-early recurrence (Non-ER) group. Early recurrence prediction score was developed using preoperative parameters. Results: ER was observed in 45 patients (25.3%). The ER group had significantly higher preoperative CA19-9 (p=0.03), serum SPan-1 (p=0.006), and CT tumor diameter (p=0.01) compared with Non-ER group. The ROC curve analysis identified cut-off values for CA19-9 (133 U/mL), SPan-1 (78.2 U/mL), and preoperative tumor diameter (23 mm). When the parameter exceeded the cut-off level, 1 point was given, and the total score of the three factors was defined as the early recurrence prediction score. Multivariate analysis for ER-related perioperative and surgical factors identified the score of 3 (odds ratio [OR], 4.63 [95% CI, 1.82–11.78], p=0.0013) and R1 resection (OR, 3.20 [1.01–10.17], p=0.049) as independent predictors of ER. Conclusions: For R-PC, ER could be predicted by the scoring system using preoperative serum CA19-9 and SPan-1 levels and CT tumor diameter, which may have great significance in identifying patients with poor prognoses and determining the optimal treatment strategy.