Tyrosine Kinase Inhibitor (TKI) Combination Scheduling Impacts Secondary FLT3 Tyrosine Kinase Domain (TKD) Mutation Profiles in a Xenograft Model of FLT3-ITD+ Acute Myeloid Leukemia (AML)

A common mechanism associated with clinical resistance to FLT3 inhibitors (e.g., sorafenib and quizartinib) in FLT3-ITD+ AML is the acquisition of secondary FLT3 TKD mutations. Recently, we demonstrated that crenolanib, a potent and selective FLT3 inhibitor, has preclinical activity against FLT3 inhibitor-resistant FLT3-ITD+/TKD mutants, and the combination of sorafenib and crenolanib was more efficacious than single-agent TKIs in a FLT3-ITD+ xenograft model (Zimmerman et al. Blood 2013). Here, we evaluated continuous and intermittent schedules of sorafenib (SOR) and crenolanib (CRE) and the impact on efficacy and drug resistant profiles in vivo. We hypothesized that varying selective pressure through modulation of FLT3 inhibitor exposure (e.g., single-agent vs combination, high- vslow-dose intensity) would alter drug resistance patterns.

For in vivo evaluations, one million MOLM13-luciferase+ cells were administered by tail vein injection to NSG mice and treatment was started 10 days later. Mice (7-10 per treatment group) were treated with vehicle (median survival, 16 days), SOR 60 mg/kg once daily (qd) on Monday, Wednesday, and Friday (3/7); SOR 60mg/kg qd for 5 days per week (5/7); CRE 15mg/kg twice daily (bid) 5/7; CRE 15mg/kg qd 5/7 and SOR 60mg/kg qd 5/7 (combination A); or CRE 15mg/kg bid 5/7 and SOR 60mg/kg qd 3/7 (combination B). Mice were monitored daily for assessment of toxicity and serial pharmacokinetic sampling was performed on days 15 or 16 of therapy. Mice were treated continuously until leukemic progression, sacrificed and bone marrow was harvested for assessment of leukemic infiltration and FLT3 TKD mutations in exons 17 and 20 by deep amplicon sequencing.

Combinations A & B were tolerated and more efficacious than single-agent TKIs: median survival combination A vs sorafenib qd 5/7 (59 days vs 45 days, p<0.001) or sorafenib qd 3/7 (59 days vs. 31 days, p<0.001); median survival combination B vs crenolanib bid 5/7 (52 days vs 31 days, p<0.001) or sorafenib qd 3/7 (52 days vs 31 days, p<0.001) or sorafenib qd 5/7 (52 days vs 45 days, p<0.001). Both combinations induced regression of leukemic burden by day 20 and delayed the outgrowth of leukemic cells compared to single-agent TKIs. SOR and CRE plasma exposure was not altered in combination therapy. Greater than 80% leukemic cell infiltration was observed in bone marrow at progression. TKD mutations were not observed in MOLM-13 cells isolated from mice treated with vehicle, crenolanib bid 5/7, or sorafenib qd 3/7, presumably due to the short duration of treatment in vivo until leukemic progression. However, TKD mutations were observed in bone marrow samples from mice treated with sorafenib qd 5/7, combination A, and combination B. In the sorafenib qd 5/7 treatment group, 4/8 (50%) mice developed clinically relevant D835Y (n=3) or F691L (n=1) mutations. In contrast, a differential pattern of TKD mutations was observed in combination A: 4/9 (44%) mice developed D835Y (n=2), D835A (n=1), D839Y (n=1), or N841K (n=1) mutations. In combination B, only 1/7 (14%) of mice developed a TKD mutation.

Emergence of clinically relevant FLT3 TKD mutations during TKI treatment was modeled in a MOLM-13 bone marrow xenograft model. Although both drug combination schedules were more efficacious than single-agent TKIs, the combination incorporating more dose intensive crenolanib therapy suppressed the emergence of FLT3 TKD mutations. These results may allow for optimal clinical trial design of TKI combinations and have provided the foundation for a protocol in development for FLT3-ITD+ pediatric AML.