Combinatorial Approaches to Overcome Plasma Protein Inhibition of FLT3 Tyrosine Kinase Inhibitors

Background: The FMS-like tyrosine kinase 3 (FLT3) is the most frequently mutated gene in acute myeloid leukemia (AML) and also results in poor prognosis for adult and pediatric patients, and thus represents an attractive target for tyrosine kinase inhibitors (TKI). The activity of FLT3-targeted TKI is inhibited to varying extents by plasma protein binding. Staurosporine (STS)-derived TKI are inhibited almost exclusively by the plasma protein alpha-1 acid glycoprotein (AGP), an acute-phase reactant. We studied the impact of AGP binding on the other STS-derivatives and report the development of a novel method to overcome this binding.

Methods: We assayed the impact of human AGP upon the activity of the STS-derived TKI (midostaurin, lestaurtinib, TTT-3002) against proliferation of the FLT3-ITD dependent cell line MOLM-14 and upon the parent compound (staurosporine) against the non-FLT3-dependent cell line HL-60. These experiments were repeated, co-incubating with drugs that competitively bind AGP to identify those that may restore TKI activity.

Results: The TKI are inhibited in a linear AGP-dependent manner (fold change increase of IC₅₀ per mg/dL AGP: midostaurin 3.00-fold, lestaurtinib 11.73-fold, TTT-3002 0.33-fold) across the range of AGP concentrations observed in human plasma. These results correspond to the drug-protein binding constants for the TKI: midostaurin 12.6 µM^-1, lestaurtinib 49.2 µM^-1, TTT-3002 1.41 µM^-1, all validated by competitive fluorescence displacement of the AGP-binding dye, 1-anilino-8-naphthalenesulfonate. These results predict that in vivo IC₅₀ values for these FLT3 TKI will be significantly higher than those measured under typical (10% FCS) in vitro culture conditions: midostaurin 4.7 µM, lestaurtinib 4.8 µM, TTT-3002 34 nM. By comparison, activity of the parent compound, staurosporine, against HL60 is completely inhibited by AGP. Assays using bovine plasma, serum or purified AGP do not demonstrate similar inhibition of FLT3 TKI. We are developing a murine model to overcome this experimental limitation.

We have developed a mathematical model describing the interactions of AGP with FLT3 TKI using classical mass action relationships that match experimental results and furthermore describe the effects of competitive plasma protein binding by unrelated agents. These models predict that disinhibition of TKI may be achievable in vivo, and define the properties of such "rescue" agents.

Mifepristone binds AGP (2-10 fold greater than STS-derived TKI) and has no independent effect upon FLT3-dependent cell growth. Co-treatment with mifepristone restores the IC₅₀ of TTT-3002 from 12 nM with AGP to < 0.1 nM. Disinhibition is seen for lestaurtinib (IC₅₀ shift reduced from >1000-fold to 50-fold) and midostaurin (300-fold reduced to 80-fold). This results in predicted in vivo IC₅₀ that are clinically relevant, and serve as a proof-of-principle for this method. Using this principle we have screened a library of FDA-approved compounds for the ability to rescue TKI activity despite the presence of potentially inhibitory plasma proteins. This screen has identified 40 potential agents that may displace STS-derived TKI from AGP, and an additional 90 agents that may restore TKI activity through off-target effects. Several agents have already been validated in vitro, and found to decrease the IC₅₀ of midostaurin and other TKI to clinically achievable ranges despite the presence of inhibiting proteins.

Conclusions: The failure of FLT3 TKI in previous clinical trials has been linked to a lack of plasma drug activity. This work provides biochemical confirmation of this effect. Furthermore, these results indicate that this is a property of the class as a whole, including midostaurin. Indeed, for midostaurin, the predicted in vivo IC₅₀ is higher than steady state levels suggesting that in clinical trials it likely acts through non-FLT3 mechanisms. Disinhibition of TKIs by mifepristone suggests a novel combinatorial approach restore TKI activity using unrelated compounds. We are currently examining other agents for similar synergy. By improving TKI potency in the face of inhibitory plasma protein binding, such combinations would be expected to improve their clinical efficacy by reducing the dosages necessary to thoroughly inhibit FLT3. Finally, this report provides a method for predicting at least one factor that affects the success or failure of FLT3 TKI in clinical trials.