Effects of the multi‐kinase inhibitor midostaurin in combination with chemotherapy in models of acute myeloid leukaemia

Abstract Recently, several targeted agents have been developed for specific subsets of patients with acute myeloid leukaemia (AML), including midostaurin, the first FDA‐approved FLT3 inhibitor for newly diagnosed patients with FLT3 mutations. However, in the initial Phase I/II clinical trials, some patients without FLT3 mutations had transient responses to midostaurin, suggesting that this multi‐targeted kinase inhibitor might benefit AML patients more broadly. Here, we demonstrate submicromolar efficacy of midostaurin in vitro and efficacy in vivo against wild‐type (wt) FLT3‐expressing AML cell lines and primary cells, and we compare its effectiveness with that of other FLT3 inhibitors currently in clinical trials. Midostaurin was found to synergize with standard chemotherapeutic drugs and some targeted agents against AML cells without mutations in FLT3. The mechanism may involve, in part, the unique kinase profile of midostaurin that includes proteins implicated in AML transformation, such as SYK or KIT, or inhibition of ERK pathway or proviability signalling. Our findings support further investigation of midostaurin as a chemosensitizing agent in AML patients without FLT3 mutations.

As midostaurin has not been investigated largely in wt FLT3expressing AML pre-clinically, we sought to explore its activity in this context, alone and in combination, to better understand the observed clinical efficacy and inform best combinations for future clinical investigation. It has been shown that for AML patients to achieve maximum clinical benefit, it is imperative that midostaurin be administered in combination with other anticancer agents. 6,8 Here, we investigate the effects of midostaurin alone against AML characterized as expressing either wt or mutant FLT3, as well as in combination with standard chemotherapy and targeted inhibitors.
Our results suggest that midostaurin, with its unique kinase profile and broad-spectrum activity, could potentially be used clinically as a chemosensitizing agent for AML patients expressing either mutated or non-mutated FLT3.

| Chemical compounds and biologic reagents
Details are provided in the Appendix S1.

| Cell proliferation studies and apoptosis studies
Details are provided in the Appendix S1.

| Immunoblotting
Protein lysate preparation and immunoblotting were carried out as previously described. 5

| AML patient cells
Methodology is described in detail in Appendix S1.

| Drug combination studies
Methodology for drug combination studies 24 is described in detail in Appendix S1.

| Non-invasive in vivo bioluminescence studies
All animal studies were performed according to protocols approved by the Dana-Farber Cancer Institute's Institutional Animal Care and Use Committee.
Bioluminescence imaging was carried out as previously described. 25 Briefly, for administration to female NSG mice (6-8 weeks of age; The Jackson Laboratory, Bar Harbor, ME), virus-and Mycoplasmafree SKNO-1-luc+ cells or OCI-AML3-luc+ cells were washed and resuspended in 1X PBS and administered via IV tail vein injection (2 × 10 6 cells/250 µL PBS). A sample size of at least 8 mice per treatment group was chosen to ensure statistical significance. Anaesthetized mice were imaged 3 days after IV injection of cells to generate a baseline used to establish treatment cohorts and matched leukaemia burden (mice were randomized). Drug treatments were also started 3 days after inoculation. Total body luminescence was measured as previously described. 26 Mice were treated with vehicle (n = 10) or midostaurin (100 mg/kg) (n = 10) for 28  For all in vivo studies, P < .05 was considered to be statistically significant. The data had similar variance and met the assumptions of the tests carried out. For in vivo studies investigating the single-agent effects of midostaurin, the Mann-Whitney test (two-tailed) was carried out to assess differences in leukaemia burden between vehicle and drug-treated mice and the Gehan-Breslow-Wilcoxon test was carried out for survival curve comparisons.

| Comparison of anti-proliferative effects of midostaurin and FLT3 inhibitors against wt FLT3expressing AML cell lines
Clinical trial results demonstrate efficacy of midostaurin in wt FLT3-expressing AML. 6,8 Significantly, midostaurin and its major metabolites have been reported to have a number of kinase targets implicated in transformation and chemoresistance. 12 Although considerable pre-clinical and clinical data have been generated showing efficacy of midostaurin against mutant FLT3-positive AML, comparatively little is known about the effectiveness of midostaurin against AML expressing wt FLT3 and other driver oncogenes. Midostaurin and other, more targeted FLT3 inhibitors were thus investigated to better understand important targets.
We measured the efficacy of midostaurin and several advanced stage FLT3 inhibitors as single agents against human AML cell lines expressing either oncogenic FLT3 or wt FLT3. Midostaurin and the other FLT3 inhibitors, as expected, most potently suppressed the growth of FLT3-ITD-positive MOLM14 cells; however, the inhibitors effectively killed wt FLT3-expressing AML lines at submicromolar concentrations ( Figure 1A-E). As predicted, cell lines expressing an Asn822Lys mutation of KIT (SKNO-1-luc+ and Kasumi-1-luc+) 27,28 were particularly sensitive to FLT3 inhibitors that have been reported to display strong mutant KIT inhibitory activity, including sorafenib and quizartinib 29,30 (Figure 1C-E).
The targeted SYK inhibitor, PRT062607, was included as a negative control in that it displays no FLT3 inhibitory activity. 22 As expected, mutant FLT3-expressing MOLM14 cells were significantly growth-inhibited by the panel of FLT3 inhibitors; however, PRT062607 showed much less potency against these cells ( Figure 1F).
To better understand the mechanism of midostaurin against AML cells expressing either mutated or wt FLT3, we next examined its effects on signalling molecules, including those implicated in inhibiting apoptosis as well as those contributing to cellular transformation. Consistent with the characterization of midostaurin as pro-apoptotic, 5  Midostaurin also inhibited S6 phosphorylation in OCI-AML2 cells in a concentration-dependent manner, with robust inhibition observed at 500 nmol/L, although there was no inhibitory effect observed on AKT or MAPK phosphorylation in this cell line ( Figure S1H). The utility of pS6 as a biomarker for the efficacy of midostaurin against wt FLT3-expressing AML was investigated by comparing drug effects on pS6 in AML cell lines that were around 2-fold or more sensitive to midostaurin (Kasumi-1-luc+, K052, OCI-AML2) than other AML cell lines (HEL, OCI-AML3). A decrease in pS6 was observed in midostaurin-treated Kasumi-1-luc+, K052 and OCI-AML2 at concentrations 250-500 nmol/L ( Figure 2F). In contrast, there was no decrease in pS6 in HEL or OCI-AML3 cells treated at the same concentrations ( Figure 2F). These results suggest that pS6 is a good biomarker for midostaurin activity and reflect the inhibition of MAPK signalling by midostaurin as a key component of its mechanism. Indeed, we have recently shown that pS6 is an excellent biomarker for activity of a targeted small molecule inhibitor of ERK1/2. 31 In addition to human cell lines, midostaurin and the other FLT3 inhibitors were also tested for efficacy against primary AML cells.
The inhibitors were observed to kill primary human AML cells having a high (>90) per cent of blasts and expressing either wt FLT3 or oncogenic FLT3 in a concentration-dependent fashion and to a greater extent than PBMCs from a normal donor, and with higher potency than PRT062607, used as a negative control because of its lack of FLT3 inhibitory activity ( Figure

| SYK is a target of midostaurin but not of other FLT3 inhibitors in clinical development
The multi-targeted nature and unique properties of midostaurin offer potential clinical benefits that other inhibitors may not, including suppression of kinases other than FLT3, such as mutant KIT, which play a role in aberrant signalling characteristic of the transformed phenotype or that are implicated in stromal cell-mediated chemoresistance. 12 A recent study reported that SYK, a protein implicated in AML transformation and drug resistance, was among a panel of signalling molecules, including VEGFR2, LYN, IGF1R, RET, PDPK1 and TRKA, which were potently suppressed not only by midostaurin but also by its metabolites. 12 Consistent with this, we have previously reported that SYK is a target of midostaurin. 22  Figure S4J).

F I G U R E 3
Anti-proliferative effects of FLT3 inhibition and SYK inhibition on primary AML samples. (A-C) Primary AML cells were treated for approximately three days. Targeted SYK inhibitor, PRT062607, was tested in parallel with the FLT3 inhibitors. Patient characteristics are summarized in Table S1. For AML2, no FLT3-ITD mutation was detected. For AML3, a FLT3-ITD mutation was detected (insertion of 12 nt 3′ to nt 1826, followed by a duplication of nt 1788-1826 (total = 51 bp); the ITD is 7% of the total flt3 alleles in the specimen; this normally corresponds to 14% blasts. However, the specimen had 93% blasts on the day of this analysis. There is a minor clone with a 51bp ITD. For AML5, no FLT3-ITD mutation was detected.  Figure S7). Taken together, these data support the notion of targeted SYK suppression by midostaurin but not by the other FLT3 inhibitors.

| Midostaurin potentiates effects of standard chemotherapy
Midostaurin has high synergizing potential and is able to positively combine with a number of agents, including 5-azacytidine, decitabine and daunorubicin, as well as inhibitors of proviability signalling molecules such as the Bcl-2 inhibitor ABT-199 (venetoclax), the Bcl-xL and   Figure S8).
As such, we similarly investigated the effects of combining midostaurin with the same chemotherapy agents and proviability signal-

| D ISCUSS I ON
Midostaurin has a potentially useful combination of targets that are involved in multiple key signalling pathways, and this multi-targeted behaviour may account for some of the observed efficacy of the inhibitor against numerous AML genotypes. In addition, considering that overexpression of wt FLT3 is characteristic of the majority of AML cases, the importance of inhibiting wt FLT3 may have been previously unappreciated. Early investigations revealed the potential of AML patients without mutations in FLT3 to be clinically responsive to midostaurin and are consistent with the results presented herein using our cell line-based and murine models of wt FLT3 AML. 8,9 Specifically, in a Phase IIb study, haematologic improvement was observed in 16/35 AML patients harbouring FLT3 mutations and 20/57 AML patients expressing wt FLT3, and more than half of patients exhibited peripheral blood or bone marrow blast reduction independent of FLT3 mutational status, although with a higher overall incidence observed for mutant FLT3-positive patients. 8 Similarly, in a Phase I study, 100 mg midostaurin treatment twice daily led to complete responses in 8/23 wt FLT3-expressing patients and 5/56 mutant FLT3expressing patients, with 9 wt FLT3 patients and 2 mutant FLT3 patients surviving over 4 years. 9 For patients treated with 50 mg TA B L E 1 Combination indices generated by Calcusyn software for midostaurin combined with indicated chemotherapy agents or inhibitors of proviability signalling molecules against FLT3-ITD-positive MOLM14 cells (panel 1, top); combination indices generated by Calcusyn software for midostaurin combined with indicated chemotherapy agents against wt FLT3 AML cell lines (panel 2, middle); combination indices generated by Calcusyn software for midostaurin combined with inhibitors of proviability signalling molecules against wt FLT3 AML cell lines (panel 3, bottom) Whereas overall response rates of first-generation FLT3 inhibitors such as midostaurin as single agents have ranged from 0% to 3%, 8 overall response rates have been 20%-40% when these inhibitors were combined with standard chemotherapy. 36,37 This is compared to an overall response rate of second-generation FLT3 inhibitors as single agents of 40%-50%; however, duration of response is short and haematologic recovery is incomplete for these inhibitors. 38 Other limitations of second-generation FLT3 inhibitors include a short (6-8 hours) half-life for crenolanib, requiring thrice-daily dosing, and a lack of activity of quizartinib against FLT3-TKD mutants such as those occurring at the D835 or F691

Combination Indices
residues. 17 We demonstrated the ability of midostaurin to behave as a relatively high resistance to venetoclax as a single agent as has been previously reported. 39 We generally observed some combinations between midostaurin and standard chemotherapeutic agents to show stronger synergy than others, with comparatively weaker combination effects observed for the simultaneous administration of midostaurin+ Ara-C against both mutant FLT3-and

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.