Background: DZB is an oral small-molecule Fibroblast Growth Factor Receptor 1/2/3 inhibitor (FGFRi) with clinically relevant activity in FGFR2-fusion cholangiocarcinoma. Extensive kinase profiling identified Colony-Stimulating Factor 1 receptor (CSF1R) as an additional anti-cancer target for DZB. CSF1R plays a role in the maintenance of tumor-promoting M2-macrophages; inhibition facilitates repolarization to M1-type thus restoring tumor T cell activity. Screening of urothelial cancer (UC) models both in vitro and in vivo has provided information on potential response biomarkers additional to FGFR genetic aberrations. Methods: Kinase assays used a radiometric assay and anti-proliferative activity was assessed using crystal-violet (72h incubation). Bone-marrow derived mouse macrophages were CSF1 starved (12h), pre-incubated with DZB/BLZ945 (30/10m) and stimulated with 0.3 μM CSF1 (3m). CSF1R phosphorylation (pCSF1R) was analyzed by immunoblotting. Compound docking experiments used MOE software and public X-ray structures. DZB was tested at MTD (75 mg/kg, po, qd) in UC-CDX (8 mice/group) and -PDX (3 mice/group) models with FGFR-mutations and/or differing FGFR copy-number (CN)/RNA-seq expression levels. Efficacy and tolerability were quantified at the 3-week endpoint as a dT/C (treated/control). Results: Comparative kinase IC50s showed that DZB had 1:1 nM activity against FGFR1/2/3 and CSF1R, a potency not observed for other clinically relevant FGFRi’s. Structural analyses suggested a different size of the inhibitor binding-site of FGFR- and CSF1R-structures, with DZB efficiently occupying the smaller CSF1R kinase sub-pocket. Indeed, DZB reduced ligand-stimulated pCSF1R in mouse macrophages in a concentration-dependent manner, with a maximal effect similar to the selective CSF1R inhibitor BLZ945. Based on an in vitro anti-proliferative screen across 14 UC-lines, DZB had a mean GI50 of 1.7±0.2 μM (range 0.4-3.4 μM). The most sensitive lines were RT4 and RT112/84, both of which had FGFR3-TACC3 fusions, a known oncogenic-driver. In mice bearing s.c. RT4 tumors, DZB induced tumor-stasis (dT/C<0.1) and was well tolerated (dT/C>1.0) but no response was observed in the RT112/84 model suggesting that not only FGFR mutations contribute to DZB response. An unbiased UC-PDX screen indicated efficacy in 4/17 models (dT/C≤0.4; median=0.81) with DZB-response significantly positively-associated with high FGFR expression. The most sensitive tumor had high FGFR2 RNA-expression yet low CN. Data will be presented from confirmatory efficacy experiments and bioinformatic analyses. Conclusion: DZB is a potent FGFRi and CSF1R inhibitor. Screens in UC models indicate that DZB efficacy is driven by FGFR mutation and expression and, potentially, CSF1R modulation. A clinical trial is ongoing in UC patients (NCT04045613) to assess DZB monotherapy, and combination with the PD-L1 antibody atezolizumab. Citation Format: Paul McSheehy, Felix Bachmann, Nicole Forster-Gross, Marc Lecoultre, Mahmoud El Shemerly, Mila Roceri, Stefan Reinelt, Laurenz Kellenberger, Paul R Walker, Heidi Lane. Derazantinib (DZB): A dual FGFR/CSF1R-inhibitor active in PDX-models of urothelial cancer [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr LB-C12. doi:10.1158/1535-7163.TARG-19-LB-C12

No abstract

Derazantinib (DZB) is an inhibitor of fibroblast growth factor receptors 1–3 (FGFR1–3), with additional activity against colony-stimulating-factor-1 receptor (CSF1R). We have profiled the activity of DZB in gastric cancer (GC) as monotherapy and combined with paclitaxel, and explored means of stratifying patients for treatment. The antiproliferative potency of DZB in vitro was quantified in 90 tumor cell lines and shown to correlate significantly with FGFR expression (<0.01) but not with FGFR DNA copy-number (CN) or FGFR mutations. In four GC cell lines in vitro, little or no synergy was observed with paclitaxel. In athymic nude mice, bearing cell-line derived xenografts (CDX) or patient-derived xenograft (PDX) GC models, DZB efficacy correlated highly significantly with FGFR gene expression (r2 = 0.58; P = 0.0003; n = 18), but not FGFR mutations or DNA-CN. In FGFR-driven GC models, DZB had comparable efficacy to three other FGFR inhibitors and was more efficacious than paclitaxel. DZB had dose-dependent plasma pharmacokinetics but showed low brain penetration at all doses. GC models (one CDX and six PDX) were tested for sensitivity to the combination of DZB and paclitaxel and characterized by immunohistochemistry. The combination showed synergy (5) or additivity (2), and no antagonism, with synergy significantly associated (P < 0.05) with higher levels of M2-type macrophages. The association of strong efficacy of the combination in vivo with M2 macrophages, which are known to express CSF1R, and the absence of synergy in vitro is consistent with the tumor microenvironment also being a factor in DZB efficacy and suggests additional means by which DZB could be stratified for cancer treatment in the clinic.

421 Background: DZB is an oral small-molecule Fibroblast Growth Factor Receptor 1/2/3 inhibitor (FGFRi) with clinical activity in FGFR2-fusion-positive cholangiocarcinoma. DZB was screened for activity in gastrointestinal cancer (GIC), by using a panel of GIC cell-lines, human tumor xenografts and 30 GIC patient-derived xenograft (PDX) models. Methods: DZB anti-proliferative potency was determined in 26 GIC cell lines to determine the GI50. The GIC cell-line, SNU-16 was grown s.c. in nude mice and treated daily for 3-weeks with DZB at the MTD of 75 mg/kg, p.o. Plasma and tumor were removed and analyzed for drug-levels and PD biomarkers to assess pathway inhibition. DZB (@MTD) was tested in the PDX-screen (15 biliary, 13 gastric and 2 colorectal cancer; n≥3/group) using models with FGFR-fusions, FGFR-mutations and/or differing FGFR copy-number (CN)/RNA-seq expression levels. Efficacy and tolerability were quantified as a dT/C (treated/control). Results: Cellular GI50s ranged from 0.02-20 μM; the most sensitive (GI50≤0.5 μM) had FGFR-fusions or high-expression. In mice, DZB induced stasis of SNU-16 tumors (dT/C∼0.0) and was well tolerated (dT/C > 1.0); the plasma PK was dose-dependent with a Cmax of 2 μM (4 hr), a Cmin of 0.5 μM. DZB induced dose- and time-dependent changes in the MAPK-pathway and expression of downstream genes, consistent with its mode of action. In PDX-models, efficacy varied from no-response to 100% regression. Known driver-mutations were associated with partial-responses (best dT/C = 0.42), but models with FGFR-fusions, especially FGFR2-fusions, were very sensitive leading to stasis or strong-regression, particularly in gastric cancer. High-expression of FGFR2 was also associated with strong responses. There was no direct correlation between CN and high RNA-seq values suggesting amplification was not always a predictor of high expression. Endpoint PD-analyses of the PDX-models is ongoing to identify other potential stratifiers and PD-markers of response. Conclusions: DZB showed convincing activity in GIC-models with FGFR-fusions and/or high expression. A clinical trial is planned in patients with gastric cancer to investigate DZB as mono- and combination-therapy.

Microtubules are major components of the cellular cytoskeleton, ubiquitously founded in all eukaryotic cells. They are involved in mitosis, cell motility, intracellular protein and organelle transport, and maintenance of cytoskeletal shape. Avanbulin (BAL27862) is a microtubule-targeted agent (MTA) that promotes tumor cell death by destabilization of microtubules. Due to its unique binding to the colchicine site of tubulin, differently from other MTAs, avanbulin has previously shown activity in solid tumor cell lines. Its prodrug, lisavanbulin (BAL101553), has shown early signs of clinical activity, especially in tumors with high EB1 expression. Here, we assessed the preclinical anti-tumor activity of avanbulin in diffuse large B cell lymphoma (DLBCL) and the pattern of expression of EB1 in DLBCL cell lines and clinical specimens. Avanbulin showed a potent in vitro anti-lymphoma activity, which was mainly cytotoxic with potent and rapid apoptosis induction. Median IC50 was around 10nM in both activated B cell like (ABC) and germinal center B cell like (GCB) DLBCL. Half of the cell lines tested showed an induction of apoptosis already in the first 24h of treatment, the other half in the first 48h. EB1 showed expression in DLBCL clinical specimens, opening the possibility for a cohort of patients that could potentially benefit from treatment with lisavanbulin. These data show the basis for further preclinical and clinical evaluation of lisavanbulin in the lymphoma field.