FLT3 (Fms-like tyrosine kinase 3) internal tandem duplication (ITD) is a frequently observed and prognostically significant aberration in acute myeloid leukemia (AML) with incidence of 20-25%. Recently two FLT3 inhibitors, namely gilteritinib and midostaurin, were approved for treatment of FLT3-mutated AML patients in combination with chemotherapy. Regardless of the confirmed survival benefit of FLT3 inhibitors, many FLT3-ITD+ patients are unresponsive and most patients eventually develop drug resistance. Here, our aim was to determine whether FLT3-ITD mutational load or length of ITD predicts ex vivo response to FLT3 inhibitors in AML.

Fragment analysis was performed on 119 samples from 87 AML patients using 10 ng of genomic DNA with primers 5’-6-FAM GCAATTTAGGTATGAAA GCCAGC-3’ and 5’-CTTTCAGCATTTTGACGGCAACC-3’. PCR-products were diluted with MQ-water (1:100-200) and 2 μl of each dilution was added to a 10 μl reaction mixture containing Hi-Di and a size standard. Samples were screened with ABI3730xl instrument (Illumina) and analyzed in GeneMapper. Ficoll-separated bone marrow mononuclear cells were available for drug testing from 56 patients (74 samples). Drug testing was performed by seeding primary BM MNCs (10,000/well) in mononuclear cell medium on 384-well plates containing ten FLT3 inhibitors in five concentrations (0,1 – 1,000 nM). The drug plates were kept in a humidified incubator at +37°C for 72 hours. After drug exposure, cell viabilities were measured with CellTiter-Glo® reagent (Promega) using a PHERAstar® FS microplate reader (BMG Labtech). Luminescence from each well was normalized to signal coming from positive (BzCl) and negative (DMSO) control wells. Normalization was necessary to produce dose-response curves based on normalized survival. A drug sensitivity score for each FLT3 inhibitor was calculated from the area under the curves as described (Yadav et al, 2013). Statistical analyses were performed in GraphPad Prism software.

By fragment analysis, we identified FLT3-ITD from 51 samples collected from 38 patients. The median allelic burden of FLT3-ITD in the positive samples was 0.338 (range 0.015 – 1.0), while ITD lengths varied between 17 and 213 nucleotides (median 44.5). We initially assessed FLT3 inhibitor responses between FLT3-ITD+ (n = 23) and FLT3-ITD– (n = 46) AML samples. Five FLT3-ITD– samples from the available data were excluded due to presence of FLT3-TKD mutation. We found that ex vivo response to all FLT3 inhibitors was significantly higher in the FLT3-ITD+ patients compared to FLT3-ITD– samples with P-values ranging from 0.01 (lestaurtinib) to < 0.0001 (e.g. quizartinib, sorafenib). By assessing whether FLT3-ITD length affects sensitivity to FLT3 inhibitors, we found no correlation for any of the FLT3 inhibitors. Intriguingly, by analyzing the role of FLT3-ITD mutational load on drug response, we discovered strong correlations. The correlation coefficients were the highest for second-generation FLT3 inhibitors, namely crenolanib (R = 0.7622) and quizartinib (R = 0.7109), and the lowest for the less-specific first-generation FLT3 inhibitors such as ponatinib (R = 0.5306) and lestaurtinib (R = 0.3920). In our study, samples with higher FLT3-ITD mutational load also tended to have higher blast counts (R = 0.4864), while FLT3-ITD length showed no correlation with the amount of blast cells.

Our results suggest that FLT3-ITD mutational load could be an important parameter for deciding whether to administer FLT3-ITD+ AML patients a specific or non-specific FLT3 inhibitor in combination with chemotherapy. The data implies that FLT3-ITD+ patients with high FLT3-ITD mutational burden could benefit more from specific FLT3 inhibitors, while patients with low amounts of FLT3-ITD could respond better to non-specific FLT3 inhibitors targeting several receptor tyrosine kinases.