4528 Background: Novel drugs are urgently needed for pancreatic ductal adenocarcinoma (PDA). Current preclinical models may not recapitulate the biology of human cancers. Direct xenografts have the potential to better represent the biology of human cancer. We have completed a clinical trial in which surgically resected PDAs were xenografted into nude mice thus generating an in vivo platform for drug development and biomarker identification. Methods: PDAs resected at Johns Hopkins Hospital are implanted into nude mice. From November 2005 to December 2008, 240 patients were consented, 236 operated, 102 were resectable PDAs thus fulfilling the inclusion criteria. 104 patients had other histologies and 34 were unresectable PDAs. Xenografted tumors were treated with gemcitabine and the activity in the xenograft correlated with patient clinical outcome. Results: Out of these 102 cases, 77 were xenografted, 53 were engrafted and reached the treatment phase. 35 out of the 53 originator patients were clinically treated with gemcitabine. The median overall survival of resected patients was 646 days (similar to 660 days from CONKO-001 trial). Patients whose carcinomas engrafted had a shorter median overall survival (319 vs. 728 days, p=0.002), probably reflecting a more aggressive tumor biology in the engrafted group. The response rate of gemcitabine in xenografts based on RECIST criteria was 8% (similar to response rate to gemcitabine in phase III trial by Moore et al.). In the adjuvant setting, the median disease free survival (DFS) was significantly longer in those patients predicted as sensitive by the xenograft model (568 vs. 286 days, p=0.037). In the metastatic setting, the median time to progression (TTP) was longer in those patients predicted as sensitive by the model (126 vs. 73 days) though this difference was not statistically significant. Overall the correlation between gemcitabine activity in this model and clinical outcome as measured by DFS was 0.71 in the subgroup of patients whose xenografts had a TGI lower than -10%. Conclusions: This preclinical model predicts gemcitabine outcome in the clinic in the adjuvant setting. Further studies to analyze the causes of shorter survival in patients whose PDA engrafted are warranted. No significant financial relationships to disclose.