BackgroundPatients with acute myeloid leukemia and internal tandem duplication of fms-like tyrosine kinase receptor-3 (FLT3+ AML) make up about 25% of newly diagnosed AML and have a poor prognosis. As a tyrosine kinase, FLT3 is an attractive therapeutic target, however the majority of patients develop resistance. Although FLT3 inhibitors initially induce a rapid decline of circulating leukemia cells, the decline of blasts in the bone marrow occurs more slowly suggesting that leukemia cells within the bone marrow microenvironment are less sensitive to the effects of FLT3 kinase inhibition. We tested multiple cytokines, growth factors and extracellular matrix proteins from the microenvironment to determine which proteins were most protective of FLT3+ AML cell lines in vitro. ResultsWe exposed the FLT3+ AML cell line, MOLM-14, to the FLT3 inhibitor quizartinib (AC220) and found that two of the most protective proteins in our screen were fibroblast growth factor 2 (FGF2) and, consistent with previous reports, FLT3 ligand (FL). FGF2 is expressed in bone marrow stromal cells and promotes resistance to kinase-inhibitors in other malignancies, including chronic myeloid leukemia (Traer et al, submitted). FGF2-mediated resistance to quizartinib could be overcome by addition of the specific FGF receptor (FGFR) inhibitor PD173074. To identify the FGFR responsible for resistance, we targeted FGFR1-4 with siRNAs and evaluated the protective effects of FGF2 in the presence of quizartinib. Only siRNA targeting FGFR1 was able to attenuate the protective effects of FGF2. FGF2 activation of FGFR1 led to downstream activation of the MAPK pathway, circumventing the effects of quizartinib on FLT3.Since patients develop resistance over several months, we tested the protective effects of FGF2 on MOLM-14 cells in extended cultures. After 8 weeks, all cultures supplemented with FGF2 eventually resumed growth (n=4). In contrast, only 2 of 4 long-term MOLM-14 cells cultured in quizartinib alone resumed growth after 12 weeks. We sequenced the FLT3 kinase domain and found no mutations in 3 of 4 FGF2-supplemented cultures. However, one FGF2-supplemented culture developed a kinase domain mutation after >3 months suggesting continued clonal evolution. In contrast, the resistant cultures grown in quizartinib alone both developed resistance through mutation of the FLT3 kinase domain, providing evidence that the microenvironment affects the kinetics and mechanism of resistance.We then examined FGF2 in the bone marrow in response to quizartinib, using biopsies from FLT3+ AML patients on the phase II clinical trial. By immunohistochemistry, FGF2 expression was only modestly increased in patients prior to initiation of quizartinib (31.4% compared to 23.2% of normal marrow samples, p=0.19). However, FGF2 increased significantly during treatment with quizartinib (51%, p=0.0013) and remained elevated until disease relapse. Thus, residual leukemia cells exist in an environment with relatively increased FGF2 and this population eventually develops overt kinase resistance, similar to our in vitro model. As previous reports have shown that FL also increases during therapy with FLT3 inhibitors, we suspect that FL and FGF2 may work cooperatively to promote resistance in the microenvironment. Since FLT3 inhibitors that also target FGFR, such as ponatinib, are currently in clinical trials it will be interesting to compare patterns of resistance in these patients. ConclusionsOur data supports a model in which paracrine FGF2 from bone marrow stroma promotes survival of residual leukemia cells, eventually leading to quizartinib resistance and disease relapse. In vitro resistance occurs via FGF2 ligand-induced activation of the FGFR1/MAPK pathway, and can be overcome by concomitant inhibition of FLT3 and FGFR, at least initially, however continued clonal evolution can lead to development of kinase domain mutations. Overall, our results illustrate the importance of ligand-induced resistance and the need to develop rational inhibitor combinations, particularly in diseases like FLT3+ AML that routinely develop resistance to kinase inhibitors. Disclosures:Kovacsovics:Ambit: Investigator on clinical trial for Ambit. Dr. Kovacsovics received no research support from Ambit. Other. Tyner:InCyte Corporation: Research Funding. Druker:Ambit Biosciences: Consultancy.