Prediction of early recurrence of pancreatic ductal adenocarcinoma after resection

• DOI: 10.1371/journal.pone.0249885
• Publication Date: 2021-04-12T17:32:30Z
• Authors: Toshitaka Sugawara; Daisuke Ban; Jo Nishino; Shuichi Watanabe; Aya Maekawa; Yoshiya Ishikawa; Keiichi Akahoshi; Kosuke Ogawa; Hiroaki Ono; Atsushi Kudo; Shinji Tanaka; Minoru Tanabe
• Keywords: Male; CA-19-9 Antigen; Neutrophils; Science; Kaplan-Meier Estimate; 03 medical and health sciences; Pancreatectomy; 0302 clinical medicine; Risk Factors; Humans; Lymphocytes; Aged; Neoplasm Staging; Proportional Hazards Models; Q; R; Middle Aged; 3. Good health; Pancreatic Neoplasms; ROC Curve; Area Under Curve; Lymphatic Metastasis; Preoperative Period; Medicine; Female; Neoplasm Recurrence, Local; Research Article; Carcinoma, Pancreatic Ductal

Background Even after curative resection, pancreatic ductal adenocarcinoma (PDAC) patients suffer a high rate of recurrence. There is an unmet need to predict which will experience early recurrence resection in order adjust treatment strategies. Methods Data with resectable PDAC undergoing surgical between January 2005 and September 2018 were reviewed stratify for defined as occurring within 6 months resection. Preoperative data including demographics, tumor markers, blood immune-inflammatory factors clinicopathological examined. We employed Elastic Net, sparse modeling method, construct models predicting using these multiple preoperative factors. As result, seven selected: age, duke monoclonal antigen type 2 value, neutrophil:lymphocyte ratio, systemic immune-inflammation index, size, lymph node metastasis peripancreatic invasion. Repeated 10-fold cross-validations performed, area under the receiver operating characteristic curve (AUC) decision analysis used evaluate usefulness models. Results A total 136 was included final analysis, 35 (34%) experienced Using we found that 7 14 useful predictive model. The mean AUC all constructed repeated validation superior standard marker CA 19–9 (0.718 vs 0.657), whereas model from entire patient cohort 0.767. Decision showed had higher net benefit across majority range reasonable threshold probabilities. Conclusion can improve prediction