Patient-derived xenograft model engraftment predicts poor prognosis after surgery in patients with pancreatic cancer

Objectives: To establish and evaluate a first generation patient-derived xenograft (PDX) model in nude mice using tumors resected from pancreatic cancer (PC) patients for the identification of key factors that influence xenograft success and prediction of patient prognosis. Methods: Primary tumor samples harvested from PC patients who underwent curative resection between May 2016 and April 2018 at our hospital were xenografted into nude mice. Tumor size was evaluated for 2 months. Patients' baseline characteristics and follow-up data were analyzed. Results: Tumor xenograft models were generated from 67 patients; 30 (44.8%) were successful and 37 (55.2%) failed. Xenograft models could recapitulate the pathology and genetic information of the primary tumors. Univariate analysis identified tumor engraftment, post-operation CA19-9, tumor size, lymph node status, and lymphovascular invasion as significant predictors (P=0.000, 0.023, 0.004, 0.035 and 0.005, respectively) of disease-free survival (DFS). Multivariate Cox regression analysis confirmed tumor engraftment, tumor size and lymphovascular invasion function as independent risk factors for DFS (P=0.000, 0.039 and 0.025, respectively). The hazard ratio of tumor engraftment for DFS was 0.239 (95% confidence interval, 0.109 to 0.524). Kaplan-Meier analysis of DFS indicated an unfavorable outcome in the engraftment group compared to that in the failed engraftment group (6.2 vs. 12.2 months, log rank P=0.000). Conclusion: The pathology and genetic information of primary PC tumors are recapitulated in the PDX tumor model in nude mice. Furthermore, engraftment success is an effective predictor of disease recurrence in patients after surgery. (C) 2020 IAP and EPC. Published by Elsevier B.V.