Differential Ex Vivo Sensitivity Patterns to Small-Molecule Inhibitors and Combination Therapies Among Acute Myeloid Leukemia Patients Harboring Both FLT3 and NPM1 Mutations

Background: One of the daunting challenges for the use of targeted therapies in the treatment of acute myeloid leukemia (AML) has been the marked heterogeneity of the disease, wherein rapidly evolving subclonal populations harboring different molecular variants and microenvironmental factors contribute to transient response and subsequent relapse on treatment. Genomics-based efforts over the past decade have defined the most frequently mutated genes in AML, among which FLT3 and NPM1 are each detected in approximately 25-30% of patients and co-occur in many individuals (TCGA, NEJM 2013). Among FLT3 mutations, the most common is an internal tandem duplication (ITD) which results in constitutive activation of the receptor. Given recent evidence suggesting potential impact of co-occurring mutations on clinical prognosis in AML (Papaemmanuil et al., NEJM 2016), we used functional screening assays to identify small-molecule targeted inhibitors and inhibitor combinations demonstrating selective efficacy in genetic subgroups of AML based on FLT3 and NPM1 mutational status. Methods: Primary mononuclear cells isolated from AML patients were plated in the presence of graded concentrations of a >120 member panel of multiple classes of single-agent inhibitors (kinase, bromodomain, BCL2; n=210 patients) or 48 pairwise combinations among agents of different classes/pathways in fixed molar dose series (n=133 patients). Area under the curve (AUC) values were derived from probit-based regression for each response curve. FLT3 -ITD and NPM1 mutational status were compiled from clinical labs or by capillary electrophoresis using a QIAxcel instrument, and patients were segregated into four subgroups based on the presence or absence of variants in these genes. Enhanced efficacy of combinations was measured by combination ratio (CR), defined as the AUC of the combination divided by the lowest AUC of the corresponding single agents. CR values were compared within each subgroup by Wilcoxon Sign Rank test; single and combination drug treatment AUC values were compared across groups by one-way ANOVA, with p-values adjusted for false discovery rate. Results: Among FLT3 inhibitors on the single-agent panel, nearly all demonstrated significantly greater sensitivity in patients harboring FLT3-ITD mutations, independent of NPM1 status. However, several of these agents including sorafenib showed trends toward a further efficacy in patients carrying both mutations. Overall, combinations involving a FLT3 inhibitor had efficacy in both FLT3-ITD-mutant and FLT3-ITD-negative patients. Intriguingly, however, among the tested combinations, three non-FLT3 inhibitor-inclusive combinations demonstrated significantly enhanced efficacy exclusively in the FLT3-ITD/NPM1 double mutant patients: ARRY-382 - panobinostat (median AUC CR: 0.601; p=0.046), idelalisib - panobinostat (median AUC CR: 0.443; p=0.003), and trametinib - venetoclax (median AUC CR: 0.393; p=0.001). By contrast, only venetoclax among the inhibitors represented in these combinations demonstrated significant differences in AUC as a single agent across the four subgroups, exhibiting lowest AUC in the FLT3-ITD mutant/NPM1 wild-type patients. Lastly, among these three combinations that were selectively effective in the FLT3-ITD/NPM1 double mutant subgroup, comparison of AUC values for single-agent and combination treatments demonstrated significantly lower AUC for each combination relative to each drug alone. Additional studies incorporating the association of RNA-seq-based gene expression in the context of these mutations are ongoing.Conclusions: Our findings suggest that, although the efficacy of FLT3 inhibitors is modestly enhanced in patients with co-occurring FLT3-ITD and NPM1 mutations, drug combination treatments that include a FLT3 inhibitor may offer broader efficacy among patients and afford strategies for staving off certain resistance mechanisms. Furthermore, select combinations that do not target FLT3 directly (such as the MEK inhibitor trametinib and the BCL2 inhibitor venetoclax) may offer opportunity for biomarker-driven enhanced efficacy and improved outcomes in patients harboring both mutations.