Abstract Regorafenib is a small molecule inhibitor of multiple transmembrane and intracellular kinases involved in normal cellular functions and in pathologic processes such as oncogenesis, tumor angiogenesis, metastasis, and tumor immunity. Regorafenib is approved for the treatment of patients with metastatic colorectal cancer (CRC) who have been previously treated with fluoropyrimidine-, oxaliplatin- and irinotecan-based chemotherapy, an anti-VEGF therapy, and, if RAS wild-type, an anti-EGFR therapy or with locally advanced, unresectable or metastatic gastrointestinal stromal tumor (GIST) who have been previously treated with imatinib mesylate and sunitinib malate. Recently an overall survival benefit has been shown in patients with hepatocellular carcinoma (HCC) who had previously been treated with sorafenib (RESORCE). OncoTrack is an Innovative Medicines Initiative (IMI) sponsored project with the goal to improve the basis for identification of biomarkers based on the mechanisms of action of therapies approved for this indication. For this purpose, a panel of CRC-PDX xenografts was generated by the OncoTrack project. At the time of the analysis reported here fifty xenografts had been treated with regorafenib at a dose of 10 mg/kg/d or with vehicle for 24 days. The analysis of tumor growth rates (TGR) showed pronounced differences between different tumors and between vehicle and regorafenib treated models. The relative antitumor activity (relative TGR) of regorafenib varied between -0,13 (good response) and 0.0 (no response). Investigations of relative TGR in relationship to (non-) clinical parameters of the primary tumor such as age, gender, sidedness and tumor histology identified a marginally significant (p= 0.04) better response in tumors from younger patients. No other correlations were detected to this end, which may be due to the small sample number. To correlate antitumor activity of regorafenib with gene expression, RNA was isolated from sections of selected vehicle and regorafenib treated xenografts and hybrized on Affymetrix HuGene-2.1_st human transcriptome arrays. Expression profiles were subsequently analyzed using the Random Forests algorithm to identify gene expression signatures predicting response to regorafenib. The best signatures did not perform better than signatures derived after randomizing responses, i.e. no predictive signature could be identified. Further studies with larger samples sizes are necessary to improve the outcome of such an approach; however one should acknowledge that to date no predictive gene signatures could be identified for multikinase inhibitors, which may be intrinsic to their complex mechanism of action. The research reported here received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement 115234 (OncoTrack). Citation Format: Henrik Seidel, Jens Hoffmann, Ralf Lesche, Sylvia Grünewald, David Henderson, Dieter Zopf. Correlation of preclinical antitumor activity of regorafenib in CRC-PDX xenografts with gene expression and clinical parameters of the primary tumor [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 5036. doi:10.1158/1538-7445.AM2017-5036