Abstract Pancreatic cancer is a devasting disease with an incidence of 0.1% and - in most cases - only palliative treatment options. The standard 1st line chemotherapy in inoperable, locally advanced (stage II and III) and metastatic (stage V) pancreas adenocarcinoma is Gemcitabine as a single agent with a median survival of about 6 month. In addition, Gemcitabine is indicated as adjuvant chemotherapy after surgery. Although the antimetabolite 5-Flourouracil (5-FU) and the EGF-R inhibitor Tarceva (Erlotinib) have been approved for 2nd line treatment, new and more efficacious treatments are urgently needed. In the present study, more than 60 primary pancreatic carcinoma samples from patients were transplanted subcutaneously (s.c.) into NMRI nude mice. In most cases, tumor material from chemonaive patients with defined histology and staging was used for implantation. Up to now, 14 pancreatic carcinoma models named PAXF and passaged in nude mice have been established and characterized comprehensively. Tumor bearing animals were treated in-vivo with 5-FU (100 or 75mg/kg, q7dx3, i.p.), Gemcitabine (240mg/kg, q7dx3, i.v.) and Tarceva (25 and 50mg/kg, qdx21, p.o.). In most studies, tumor growth was inhibited with best T/C values > 50%, highlighting the general chemoresistance of pancreatic cancer. Only for PAXF1872 and PAXF1998, a high sensitivity towards Gemcitabine was evident with best T/C values of 8% and 3.8%, respectively. In general, there was no correlation of response to Tarceva with EGFR expression status and best T/C values ranging from 89.6% (PAXF1876) to 38.5% (PAXF1982). A more broad chemosensitivity profile was established with Gemcitabine, 5-FU, Tarceva, Taxol, Cisplatin, Rapamycin, Sorafenib and Sunitinib in the ex-vivo clonogenic assay. Interestingly, 2 out of 12 pancreatic cancer models responded with high sensitivity towards treatment with Rapamycin (IC50 <= 10nM) with a mean IC50 = 0.18 µM over all models. Finally, the established tumor models were characterized by using Affymetrix HG U133 plus2.0 gene expression arrays, exon sequencing (Ki-ras, B-raf, p53) and the Sequenom MassARRAY® System - OncoCarta(TM) panel. In all tumors analysed so far, an activating Ki-ras (H-ras, B-raf) mutation was identified. Mutational analysis of p53 (exons 4 to 10) revealed only 4 out of 13 models as p53 wild-type. For the PAXF1998 tumor model with animals cured by Gemcitabine therapy, mutations in Ki-ras (G12V, heterozygous) and p53 (V274A, homozygous) have been identified. In summary, a unique collection of patient-derived xenograft models of high clinical relevance has been established. These pancreatic carcinoma models are available for translational research studies including in-vivo efficacy testing of new investigational drugs. Citation Information: Mol Cancer Ther 2009;8(12 Suppl):A23.