660 Background: Colorectal cancer liver metastasis (CRCLM) continues to be a major health problem and despite extensive efforts to develop new drugs, median survival remains at a mere 30 months. The purpose of our study is 1. Develop a precision medicine strategy for patients with CRCLM and 2. Discover novel pathways and treatments to improve outcomes. Methods: In order to develop a precision medicine strategy, 6 matched patient derived xenografts (PDX) and cell lines were established from patients undergoing resection of their CRCLM. A high-throughput drug screen containing over 100 FDA approved drugs was first used in vitro on 3 cell lines to identify therapeutic targets. The top therapeutic targets were validated in vivo. RNA seq was then performed to identify potential predictive markers of response. Results: Our high-throughput drug screen performed on 3 early passage CRC cell lines (CRC057, CRC119 and CRC240) identified ponatinib, a multi TKI, as the only agent to inhibit all 3 cell lines. Subsequent growth inhibition analysis identified the fibroblast growth factor receptor (FGFR) as the main target of ponatinib. In vitro findings were confirmed in vivo in matched PDXs. Western blot analysis post treatment showed evidence of decreased phospho-FGFRs. Further analysis of the main downstream signaling pathways of FGFR (RAS/PI3K/AKT, RAS/MEK/ERK and STAT) demonstrated that the STAT pathway was effectively targeted in all 3 cell lines. In contrast, the ERK pathway was targeted in CRC240 and CRC057 while the p-AKT was targeted in CRC119. Finally, RNAseq revealed different isoforms and mutations in these samples. Conclusions: We have developed a precision medicine strategy for patients with CRCLM using matched cell lines and PDXs coupled with high throughput drug screens and genomic analysis to identify novel targets and specifically identified the FGFR/STAT axis as a therapeutic target. Furthermore, co-targeting FGFR and its downstream pathways may provide synergy and lead to combinatorial therapies that can improve patient outcomes. Finally, RNAseq data can be used to develop predictive markers of therapy.