R. Marayati1, C. J. Tignanelli2, J. Yeh1,2,3 1Lineberger Comprehensive Cancer Center,Chapel Hill, NC, USA 2University Of North Carolina At Chapel Hill,Department Of Surgery,Chapel Hill, NC, USA 3University Of North Carolina At Chapel Hill,Department Of Pharmacology,Chapel Hill, NC, USA
Introduction: Pancreatic cancer is a lethal malignancy with an extremely poor prognosis and lack of effective therapy. Patient-derived xenograft (PDX) models have been used as a preclinical platform to investigate the underlying biology of pancreatic cancer and to evaluate new biomarkers and anti-cancer therapies. We sought to assess whether PDX models reliably reflect the tumor biology and course of disease in patients with pancreatic cancer.
Methods: Surgically resected pancreatic ductal adenocarcinomas were obtained from 56 de-identified patients after IRB approval. Tumors were engrafted either orthotopically or subcutaneously into immunocompromised mice and passaged over time. PDX tumors were considered successful if they reached a volume of at least 200 mm3 and were passaged at least twice. The time to 200 mm3 was defined by the number of weeks it took for a tumor to reach a volume of 200 mm3 in the initial passage. At each passage, pancreatic ductal adenocarcinoma histology was confirmed by hematoxylin and eosin staining and KRAS mutation status was determined by pyrosequencing. Overall survival (OS) and recurrence-free survival (RFS) were calculated using the Kaplan-Meier method and compared using the log-rank test. Pearson’s chi-square and Fisher's exact tests were used to compare categorical variables (IBM SPSS Statistics v20).
Results: Out of 56 engrafted patient tumors, 37 (66%) met our criteria for success. The median follow-up of our patient cohort was 34 months. Patients with successfully engrafted tumors had a significantly shorter median OS (12 months vs. 21 months, p=0.039). Successful engraftment did not correlate with known pathological variables such as differentiation, lymph node involvement, stage of disease, or tumor margin status. Furthermore, successful engraftment was independently predictive of OS in a multivariate Cox regression model that included lymph node involvement, stage, and tumor margin status, with a hazard ratio of 0.497 (95% CI [0.248, 0.997], p=0.049). Of the 37 successfully engrafted tumors, 79% had KRAS mutations. Patients whose tumors took longer than 20 weeks to reach a volume of 200 mm3 had a significantly longer median RFS (18 months) than those patients whose tumors took less than 20 weeks (9 months, p=0.013). In a multivariate Cox regression model that included KRAS mutation status, lymph node involvement, stage, and tumor margin status, time to 200 mm3 was the single independent predictor of RFS with a hazard ratio of 3.429 (95% CI [1.075, 10.936], p=0.037).
Conclusions: Our results show that PDX models accurately recapitulate the growth pattern and inherent tumor biology of patients with pancreatic cancer, in that tumors that grow faster are associated with earlier recurrences. In addition, patients with successfully engrafted tumors have a significantly poorer overall survival, suggesting that these tumors represent a more aggressive subset for which novel therapies are needed.