Midostaurin In Acute Myeloid Leukemia: An Evidence-Based Review And Patient Selection

Abstract: Fms-related-tyrosine kinase 3 (FLT3) mutations occur in approximately a third of acute myeloid leukemia (AML) patients and confer an adverse prognosis. Numerous studies have evaluated FLT3 targeting as single agent and in combination approaches in frontline and relapsed AML. At this time, midostaurin, a multikinase inhibitor, is the only FLT3-inhibitor that is US FDA approved to be used in combination with induction therapy in the frontline FLT3-mutated AML setting based on improved overall survival noted in the RATIFY Phase III trial. The utility of midostaurin in maintenance post stem cell transplanta- tion has shown promising results and further studies are still ongoing. In this review, we discuss the studies that led to the inception of midostaurin as a targeted kinase inhibitor, its evaluation in AML, the early clinical trials and the large Phase III clinical trial that led to its eventual US FDA-approval in FLT3-mutated AML. Our review also discusses data on midostaurin adverse effects, mechanisms of resistance and limitations of its utility. We further discuss emerging second-generation FLT3 inhibitors, with a focus on quizartinib and gilteritinib and future directions to enhance FLT3-inhibitor efficacy and overcome mechanisms of resistance.

Introduction
As mentioned earlier, HP-β-CD and SBE-β-CD are safe derivatives of β-CD and are approved for clinical applications and are the most commonly studied CDs in recent years (Brewster & Loftsson, 2007), we tested nintedanib complexation using these two CDs. Phase-solubility diagrams of nintedanib with HP-β-CD and SBE-β-CD are presented in Fig. 1A. Solubilization capability of the β-CDs can be quantitated using the data obtained from these phase solubility studies. From the diagrams it is depicted that SBE-β-CD showed AL-type phase solubility, i.e., solubility of nintedanib increased linearly with increasing concentration of CD (Higuchi & Connors, 1965). The apparent solubility of nintedanib was found to be increased to around 500 µM. The AL-type phase diagram revealed 1:1 stoichiometry between nintedanib and SBE-β-CD during formation of inclusion complexes. This is also supported by the slope value obtained from the linear phase solubility diagram, which was found to be lower than one (0.9499) indicating the formation of 1:1 complexation between the drug molecules and SBE-β-CD. The stability constant (Ks) for SBE-β- CD based inclusion complex was found to be 689 M-1, which is well in the range (100-1000 M-1) required for appropriate stability of inclusion complexes and also required to improve oral bioavailability of therapeutics (Devasari et al., 2015; Loftsson et al., 2005; Yang et al., 2009). It can also be observed that at lower Nintedanib: SBE-β-CD molar ratio high amount of nintedanib solubilizes whereas this effect is more prominent for HP-β-CD at higher molar ratio.

The inclusion complex with HP-β-CD showed linearity at lower concentration of CD, however phase solubility diagram with higher concentration showed AP-type phase solubility which may be ascribed to thehigher (1:2 or 1:4) stoichiometry between nintedanib and HP-β-CD at higher concentration (Higuchi & Connors, 1965). Although the solubility of nintedanib increased up to 700 µM, this AP-type phase solubility resulted in the requirement of higher amount of HP-β-CD to form stable inclusion complex ultimately leading to the increased bulk of the complex. However, for the formulation of dosage form for oral drug administration this might result in increased size of dosage form leading to various formulation issues associated with designing, packaging and storage along with poor patient compliance. As discussed earlier, our aim to prepare inclusion complex was to form a stable complex utilizing minimum quantity of CDs. Hence, based on the results of phase solubility studies, we chose SBE-β-CD nintedanib inclusion complex for further characterization and evaluation of therapeutic activity and intestinal permeability.It is known that an alteration in the spectra (measured as shift in λmax) of drug molecule is observed after inclusion owing to the modified microenvironment of solvent due to complexation. The molar ratio at which maximum shift in the λmax of drug occurs is considered the stoichiometric ratio. The Job’s plot is represented in Fig. 1B. Results of the study showed maximum peak at 0.5 mole fraction value which indicated 1:1 stoichiometry during the formation of inclusion complex. Similar type of stoichiometry has also been observed from phase solubility diagram.

Hence, nintedanib and SBE-β-CD were used in 1:1 ratio for the formation of inclusion complex for further characterization and therapeutic evaluation.The FT-IR spectra of nintedanib, SBE-β-CD, physical mixture of nintedanib and SBE-β-CD, and nintedanib-CD complex are presented in Fig. 2-(A). The IR spectrum of nintedanib showed characteristic peaks at about 2945, 2933 (C-H stretching, CH3), 2358, 2341, 1705 (C=O stretch,ester), 1653 (C=O stretch, Amide), 1506, 1442, 1292 (C-N stretch), 1222, 1147 and 1089 cm-1 (Fig. 2-(A)-i). The FT-IR spectrum of SBE-β-CD exhibited characteristic peaks at about 3431, 2934 and 1022 cm-1 which correspond for O-H stretching, C-H stretching and C-O stretching vibration, respectively (Fig. 2-(A)-ii). The spectrum of physical mixture of nintedanib and SBE- β-CD demonstrated a superposition spectrum of both compounds, however less intense absorption peaks of nintedanib at around 1292, 1222, 1147 and 1089 cm-1 demonstrated some interaction between nintedanib and SBE-β-CD during formation of physical mixture (Fig. 2-(A)-iii). The FT- IR spectrum of nintedanib-CD complex clearly showed that characteristic peaks of nintedanib are disappeared or some less intense peaks are observed at around 1653, 1506 and 1442 cm-1 (Fig. 2- (A)-iv). These results clearly depicted that some functional groups of nintedanib are included in the cavity of SBE-β-CD to form molecular complex.Fig. 2-(B) shows the 1H NMR chemical shifts of nintedanib, SBE-β-CD, the physical mixture of nintedanib and SBE-β-CD, and the inclusion complex of nintedanib-CD. Most of the proton chemical shifts observed for the drug alone were very similar to those observed for the physical mixture and inclusion complex (<0.01 ppm change).

However, larger chemical shifts were observed for H1, H2 and H3 as shown in Fig. 2-(B). Specifically, H1 and H2 exhibited a modest change of 0.21 ppm, while H3 showed a substantial change in chemical shift of 0.38 ppm. These chemical shift changes suggest that this aspect of nintedanib scaffold finds itself within the cavityof SBE-β-CD while the rest of nintedanib exists outside of the SBE-β-CD cavity for the inclusion complex. The presence of nintedanib within SBE-β-CD is further supported by large chemical shifts observed for C1 and C2 (0.46 and 0.05 ppm, respectively) when comparing SBE-β-CD alone to the inclusion complex. Molecular modeling studies, discussed later in the manuscript, have also suggested that piperazine ring of nintedanib trapped in cyclodextrin cavity could form stable inclusion complex.Thermal behavior of pure drug and inclusion complex was investigated by thermogravimetric method (Fig. 3-(A)). The thermogram of nintedanib showed a sharp and intense endotherm peak at 255°C ((A)-i), which corresponds to melting point of the drug. Sharp and intense peaks depicted crystalline characteristic of pure drug. The DSC thermogram of SBE-β-CD showed two peaks, one broad peak at 120°C and other peak at 270°C ((A)-ii). First peaks represents to the removal of water molecules from the cavity whereas later presents to the decomposition of CD. Our results are in accordance with previous report (Devasari et al., 2015).

The DSC thermogram of physical mixture of nintedanib and SBE-β-CD showed peaks of both molecules, however less intense peaks of both molecules were observed at 255°C and 270°C, and this might be because of possible interaction during formation of physical mixture ((A)-iii). However, sharp peak of nintedanib is disappeared from the DSC thermogram of nintedanib-CD complex, which clearly depicted the formation of inclusion complex and conversion of crystalline nintedanib to amorphous state after complex formation ((A)-iv). Further, in the DSC thermogram of nintedanib-CD complex an endotherm peak at around 355°C was observed, which might be ascribed to the decomposition of complex. Hence, DCS thermogram not only demonstrated the formation of inclusion complex but also showed that after formation of inclusion complex thermo-stability of nintedanib is improved.Powder XRD (P-XRD) studies were performed to detect the crystallinity of the pure drug and complexed drug. As shown in Fig. 3-(B), nintedanib exhibited several intense and sharp peaks, which confirm the crystalline nature of nintedanib. However, an XRD spectrum of SBE-β-CD has not shown any sharp peak which showed amorphous nature of SBE-β-CD. Further, XRD spectra of nintedanib-CD inclusion complex showed that there are no sharp peaks corresponding to the nintedanib, which depicted that nintedanib might be incorporated in the cavity of CD during complexation and changed to the amorphous state during freezing process (Vangara et al., 2014).For the stability studies, we have also used water soluble salt of nintedanib, nintedanib EHS, which is available commercially by prescription. Results of the stability studies in different biological fluids are shown in Fig. 4.

Results of the studies showed that although very soluble, nintedanib EHS is highly unstable in PBS (pH 7.4) and is also significantly unstable in simulated intestinal fluid (SIF). Most of the drug (more than 60%) was degraded in PBS (pH 7.4) (Fig. 4A) whereas more than 10% drug was degraded in SIF after 4 hours of incubation at 37°C (Fig. 4B). Nintedanib free base also degraded in PBS however relatively less as compared to EHS salt. In SIF, we did not find stability issues with nintedanib base whereas in simulated gastric fluid (SGF), both base and EHS salt were found stable (Fig. 4C). When nintedanib-CD complex was incubated with SIF and PBS, it was observed that nintedanib was stable for longer time in both fluids as compared to plain drug (Figs. 4A-C). Hence, it could be depicted that stability of nintedanib improved because of its inclusion in the cavity of CD and CD complex would be a better alternative to water soluble EHS salt for dosage form development. By improving stability of nintedanib in SIF and PBS, bioavailability of drug could also be improved.GOLD scores for nintedanib with various substituted CDs are given in Fig. 5B. Docked images of nintedanib over HP5βCD and SBE7βCD are shown in Fig. 5C and 5D.CDs form a tube like structure where drug molecule is trapped or loaded. As substituents on CD are added, based on the number and size, the length of the tunnel is also increased accordingly.

Nintedanib is bound to all CDs through a number of hydrophobic interactions. Having multiple aromatic rings, nintedanib was able to show good binding affinity represented by GOLD scores. There are no H-binding interactions were observed with any of CDs. In comparison, HP5CD showed several unfavorable steric interactions over SBE7CD. In all docking poses with CDs, piperazine ring of the nintedanib was trapped in the CD tunnel which may provide additional metabolic stability to this drug when compared to administering free drug along with its increased solubility. In case of SBE7βCD isomer1, where all 7 sulfobutyl ether groups were on the C-6 primary hydroxyl groups reduced docking scores which may be due to steric hindrance by these groups. When some of these sulfobutyl ether groups were moved to other side of CD as seen in isomer 2, 3 and 4 docking scores were significantly increased. Overall SBE7βCD may have either better drug loading and release characteristics compared to HP5CD drug complexes. Findings from molecular modeling studies are consistent with phase solubility and other experimental studies.The major focus of present study was to assess the feasibility of enhancing nintedanib transport across intestinal epithelium by reducing the efflux. Nintedanib is a known substrate for p- glycoprotein (p-gp) mediated efflux which may significantly contribute to its low oral bioavailability of 4-5% (Dallinger et al., 2016).

Interestingly, in addition to providing solubilityenhancement for pharmaceuticals, -cyclodextrin is also known for modulating p-gp mediated efflux by inhibiting p-gp ATPase activity (Zhang et al., 2013). To assess the capability of CD complexation on the above mentioned characteristics, permeability studies were conducted on the epiIntestinal tissue model which consists of monolayer of human primary small intestine epithelial cells on the cell culture inserts. Till date, most of the in-vitro transport/permeability studies to predict in-vivo intestinal permeability have been performed using Caco-2 cells or Caco-2/HT-29 co-cultures (Artursson, Palm, & Luthman, 2001; Hilgendorf et al., 2000; Nollevaux et al., 2006). However, several reports have suggested limited in-vitro-in-vivo correlation (Gupta, Doshi, & Mitragotri, 2013). Hence, in our study we measured efficacy of CD complexation in modulating permeability of prepared nintedanib using epiIntestinal tissue model. This model recapitulates the physiology, 3D tissue architecture, and function of small intestine. For this study, we determined cumulative drug transport (both A to B; and B to A), % of loaded transported. Using the permeability data, apparent permeability constants (Papp) for both apical to basal and basal to apical transport, and efflux ratio were calculated using the equations described in the Methods section. To confirm the role of CD complex on the permeability of nintedanib, we have also included water soluble nintedanib EHS in this study. Results of the study demonstrated cumulative amount of drug transported was significantly higher in basolateral to apical direction for all the groups (nin, nin-EHS, and nin-CD complex) (Fig. 6A). However, nin-EHS demonstrated the most prominent increase in B to A transport.

At the same time, CD complexation improved the A to B transport 3- folds, and reducing the B to A transport by 2.5 folds as compared to nin-EHS (Fig. 6A). Total % dose transported over 2 hours also showed similar trends with nin-CD demonstrating approximately 4-fold enhancement of A to B transport and 2-fold reduction of B to A transport (Fig. 6B). Nintedanib EHS showed higher Papp for basal to apical transport (6.3x10-7 cm2/sec) ascompared to 3.1x10-7 cm2/sec for nin-CD complex (Fig. 6C), This difference in Papp clearly underlined the efficacy of CD-complex in averting B to A efflux for nintedanib, which is depicted in the efflux ratio calculations which showed an efflux ratio of more than 2 for both free base and EHS salt, demonstrating active efflux. CD complexation however resulted in reduction of efflux ratio to 0.4, which is 6-8 fold lesser than nin and nin-EHS, thus demonstrating the p-gp modulating capabilities of cyclodextrin (Fig. 6D). Earlier reports have suggested that efflux ratio more than 2 represents p-gp efflux (Zhang et al., 2018; Zheng, Chen, & Benet, 2016). Our results are in accordance with previous report showing nintedanib as p-gp substrate (Luedtke et al., 2018). The formation of CD complex resulted in the increase of Papp for apical to basal transport. Moreover, Papp for basal to apical transport decreased when CD complex was used. Earlier, it has been reported that CD complexation could help in reducing p-gp efflux of drugs. As discussed above, CD being hydrophilic in nature, is not substrate for p-gp. Hence, drugs complexed in the interior of CD could also avoid p-gp efflux.

Results of our study clearly showed that CD complexation increased transport of nintedanib across intestinal membrane simultaneously reducing p-gp efflux. Increased permeability of nintedanib may be ascribed to both enhanced aqueous solubility as well as inhibition of p-gp activity by SBE-β-CD (Zhang et al., 2011). However, further mechanistic studies are warranted to assess the effect of CD complexation on the transport mechanism and metabolic profile of nintedanib.After preparing drug-CD inclusion complex, we tested the bio-activity of inclusion complex in terms of anti-proliferative activity, effect on cells migration, and effect on the synthesis of collagen in growth factors stimulated lung fibroblast cells. As nintedanib is FDA-approved for idiopathic pulmonary fibrosis (IPF), we chose different growth factors which are reported to be involved inthe pathogenesis and progression of IPF, including PDGF-BB, bFGF, and TGF-β (Chaudhary et al., 2007; Coward et al., 2010; Richeldi et al., 2014). The activity of CD complex was compared with plain nintedanib.As reported earlier that growth factors stimulated fibroblast proliferation by activating growth factor receptors. Nintedanib has been reported to reduce the growth of activated fibroblasts by blocking growth factor induced proliferation (Wollin, Maillet, Quesniaux, Holweg, & Ryffel, 2014). Results showed that PDGF-BB (Fig. 7A) and bFGF (Fig. 7B) stimulated the proliferation of fibroblast cells, which was reduced by the pre-incubation (30 min) with nintedanib and nintedanib-CD in dose-dependent manner.

Interestingly, nintedanib-CD showed significantly higher anti-proliferative activity when compared to plain nintedanib at each tested concentration.The effect of nintedanib on the migration of fibroblast cells was evaluated by scratch assay (wound healing). Fig. 8 shows representative images of wound created by scratching serum starved cells with a sterile p200 microtip, following treatment with different compounds for 48 hours. Results of the study depicted that incubation of cells with PDGF-BB stimulated the migration of cells (complete closure of the wound after 48 hours) whereas incubation of cells with PDGF-BB in the presence of nintedanib and nintedanib-CD reduced/blocked the migration of cells. Nintedanib-CD was found to be more potent than nintedanib that may be correlated to better stability of the drug in CD inclusion complex.To further confirm the effect on fibrogenesis, we investigated the effect of nintedanib and nintedanib-CD inclusion complex on the growth factor stimulated collagen deposition in thefibroblast cells. It has already been reported that PDGF-BB, bFGF and VEGF have not shown any effect on stimulation of collagen deposition in fibroblast cells, whereas TGF-β stimulated cells demonstrate collagen deposition significantly higher than normal cells (Hostettler et al., 2014). Hence, in our studies we used TGF-β to stimulate collagen deposition. Results of the study demonstrated that TGF-β stimulated collagen deposition by 127% as compared to control cells (cells without any treatment) (Fig. 9A).

Further, it has also been found that pre-treatment (30 min) of cells with nintedanib or nintedanib-CD complex significantly suppressed the stimulated deposition of collagen in the fibroblast cells (Figs. 9A&B). Microscopic observation further confirmed the effect of nintedanib and nintedanib-CD on the collagen deposition. As shown in the Fig. 9B, fibers stained with red color (collagen deposition) are increased in TGF-β stimulated cells whereas cells treated with nintedanib and nintedanib-CD showed significantly less red color fibers thus confirming the reduction of collagen deposition by the treatment.Acute myeloid leukemia (AML) represents a malignant clonal disorder of myeloid cells that impairs normal hematopoiesis. Age, history of pre-leukemic hematologic disorders, karyotype and mutation profile provide significant prognostic information that dictate therapeutic decisions in AML.1,2 Treatment of AML traditionally encom- passes induction with high-dose anthracyclines, such as idarubicin (or daunorubicin), and cytarabine arabinoside-based regimens. Consolidation therapy traditionally includes high-dose chemotherapy (cytarabine) alone (HiDAC) or in combination with anthracycline (also referred to as “2+5”) depending on regional and institutional preferences in low-risk patients, or allogeneic transplantation in the majority of intermediate and in high-risk patients, respectively.3Traditionally, normal karyotype in AML conferred intermediate-risk disease.

However, with the advent of mutational analysis, it became evident that there were certain clonal mutations that could significantly alter the AML pathogenesis and prognosis. Some mutations such as nucleophosmin 1 (NPM1) and CCAAT/enhancer- binding protein-α (CEBPA) confer favorable risk. On the other hand, mutations in genes such as Fms-related-tyrosine kinase 3 (FLT3), runt-related transcription factor 1 (RUNX1), DNA methyl transferase 3 alpha (DNMT3A), tumor protein p53 (TP53)php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php).and others confer unfavorable risk.2 An internal tandem duplication (ITD) mutation in FLT3 gene on chromosome 13q12 is one of the most common mutations noted in AML, occurring in approximately a third of newly diagnosed adults with AML.4–6 FLT3 encodes a class III receptor tyrosine kinase (RTK), which is normally expressed in CD34+ hematopoietic stem cells and promotes diverse downstream pathways depending on the impact of co-occurring signals.7 For instance, binding of the FLT3-ligand (FL) to FLT3 receptor in the absence of other growth factors induces monocytic differentiation of hema- topoietic progenitors.8 However, FLT3 activation induces proliferation and maintenance of progenitors when inter- leukin-3, stem cell factor and FL are engaged.

8–Mutations in FLT3 commonly occur in patients with AML who have diploid cytogenetics indicating that the mutated FLT3 is the driver for leukemogenesis.11 There are two major classes of activating FLT3 mutations reported in AML patients. The first class of mutations is 3–400 base pair in-frame duplications detected in 20% to 25% of patients with AML and referred to as internal-tandem duplications (FLT3-ITD).12 FLT3-ITD mutations lead to constitutive acti- vation of the FLT3-signaling cascade with subsequent stimu- lation of downstream signaling pathways including signal transducer and activator of transcription 5 (STAT5), phos- phatidyl-inositol 3-kinase (PI3K) and protein kinase B (AKT) pathways.13,14 The second class of FLT3 mutations occurs as point mutations, most commonly a substitution of tyrosine for aspartic acid at codon 835 (D835Y), in the tyrosine kinase domain (FLT3-TKD).15 These occur in 5% to 10% of patients with AML. Similar to FLT3-ITD, these point mutations lead to downstream activation of prolifera- tive pathways.11 Patients with FLT3-ITD mutations have similar complete remission (CR) rates compared with non- FLT3 mutated patients, but inferior outcomes due to shorter CR duration, high relapse rates, and inferior overall survival when treated with induction therapy alone, without the addi- tion of a TKI.4,11 FLT3-TKD mutations on the other hand have been generally noted to have a neutral impact on overall survival (OS).16 The significance of these mutations in clin- ical practice has been further fortified by the emergence of effective targeted therapies to these mutations and their implications on survival.AML patients with FLT3-ITD mutations have shorter CR durations, higher rates of recurrence, and inferior OS compared to patients without FLT3 mutations. Ofsignificance, the FLT3 allelic burden is also important and has a prognostic impact.

Polymerase chain reaction (PCR) technique is the most commonly used tool to assay for FLT3 allelic burden. However, because of the competition from wildtype allele, the sensitivity of the PCR assay is low which may be overcome by using patient-specific primers.20 A more recent assay detects FLT3-positive minimal residual disease AML with higher sensitivity and specificity. The proposed approach starts with a PCR amplification step, followed by next-genera- tion sequencing, and uses a unique software program to quantify the findings.21 Further, developing assays to mea- sure the inhibitory effects of an oral drug is also important. Plasma inhibitory activity is the first assay to measure the efficacy of target inhibition of FLT3.22Given the significance of FLT3 mutations in AML, there has been significant interest in developing and apply- ing targeted therapies for FLT3-mutated AML patients in the induction, consolidation, and/or maintenance phases, to decrease the risk of relapse and improve OS, as well as in relapsed FLT3-mutated AML.23 Thomas and Campbell (2019) recently reviewed four agents that showed promis- ing results in AML FLT3 inhibition.24 Briefly, ponatinib, sunitinib, and sorafenib are non-specific tyrosine kinase inhibitors approved in multiple solid tumor malignancies with known FLT3-inhibitory activity.25 Ponatinib showed modest activity in a small cohort of AML patients with overall response rate of 25% but had significant adverse events.26 Sunitinib showed activity in inhibiting FLT3 in AML patients but the duration of response was short lived.25,27 Both ponatinib and sunitinib have not been widely incorporated in AML therapy. Sorafenib, while not approved in AML, has been used effectively for many years as a maintenance post stem cell transplantation in FLT3-mutated AML patients,28 as well as in combination with induction chemotherapy in newly diagnosed FLT3-ITD mutated AML and in combination with azaci- tidine in frontline and relapsed older FLT3-mutated AML.

Interestingly, sorafenib improved event-free survival (EFS), but not OS when added to 3+7 induction regimen even among FLT3-wildtype patients, likely more through its multi-kinase inhibitory activity rather than its direct FLT3-inhibitory activity,31 and this study is ongoing with follow-up data eagerly awaited. Of note, gilteritinib a potent second-generation FLT3-inhibitor was also recently approved as single-agent therapy for patients with relapsed/refractory FLT3 mutated (both ITD and TKD) AML based on improved OS and response rates comparedto conventional cytotoxic or low-intensity therapies in a randomized Phase III setting.32 The only agent that is currently FDA approved in combination with induction and consolidation therapy for newly diagnosed FLT3- mutated AML patients is midostaurin, and will be the focus of this review article.Midostaurin, formerly known as PKC412, is an orally admi- nistered multi-targeted tyrosine kinase inhibitor that inhibits multiple kinases including proto-oncogene c-Kit (KIT), pla- telet-derived growth factor (PDGFR)-α/-β, protein kinase C (PKC), spleen associated tyrosine kinase (SYK), cellular Src kinase (SRC), and vascular endothelial growth factor recep- tor (VEGFR)-1/-2.23 Midostaurin was developed as a ther- apeutic alternative to the naturally available staurosporine in order to inhibit PKC activity.33,34 Several in vitro and animal studies demonstrated the efficacy of midostaurin in halting cellular proliferation.35–38 These findings led to a Phase I study in 32 patients with advanced solid tumors whereby midostaurin demonstrated a relatively safe profile with low- grade gastrointestinal and hematological toxicities.39 A treat- ment dose of 150 mg/day was considered adequate for further investigation although subsequent studies used dif- ferent regimens. Subsequent clinical trials investigated the utility of midostaurin in solid cancers and lymphomas but failed to replicate the preclinical findings.

Specifically, mid- ostaurin was tested in lymphoproliferative disorders as a single agent,40 different solid tumors in combination with 5-fluorouracil,41 non-small cell lung cancer in combination with gemcitabine and cisplatin,42 and metastatic melanoma as a single agent,43 without demonstrating significant activity but notably with a maintained low toxicity profile. Further, the anti-angiogenic activity of midostaurin via inhibition of vascular endothelial growth factor (VEGF) in animal models led to clinical studies that examined its role in diabetic retinopathy and macular edema.44,45 Midostaurin at a lower dose of 100 mg/day demonstrated activity in reducing dia- betic retinopathy and macular edema but the gastrointestinal toxicity profile noted in these patients with chronic use of midostaurin limited its further clinical development in dia- betic patients.46The first evidence of the activity of midostaurin in AML was in 2002. In a drug screening assay, Weisberg et aldemonstrated the efficacy of midostaurin in inducing G1 arrest and apoptosis in FLT3-mutated Ba/F3 leukemia cell lines and mouse models.47 In a subsequent study, Grundler et al demonstrated that midostaurin can inhibit FLT3-ITD encoded protein, which by then was known to be one of the most common and aggressive mutations in AML.48 Combining midostaurin with the histone deacetylase inhi- bitor LAQ824, now known as dacinostat, demonstrated synergistic activity in inhibiting FLT3-mutated AML cell lines, which was one of the first preclinical experiments suggesting a role for combining midostaurin with other established leukemia drugs.49 Also, midostaurin demon- strated more potent synergism when combined with con- ventional anti-leukemic agents such as cytarabine, doxorubicin and idarubicin in inhibiting FLT3 mutated, compared with FLT3 wildtype leukemia cell lines in vitro.50,51 Interestingly, midostaurin elicited apoptotic cell death in FLT3-mutated AML cell lines, while it induced cell cycle arrest in FLT3-wildtype cell lines.

Hence, the context in which midostaurin was used likely mattered in dictating its response. These preclinical studies in AML and the relatively safe profile of midostaurin in solid tumor clinical trials supported the development of clinical trials to evaluate midostaurin in the treatment of patients with AML.The relatively safe profile of midostaurin in prior clinical trials in non-leukemia patients paved the way for the initia- tion of a Phase II clinical trial in AML. To that end, a total of 20 patients with relapsed/refractory AML with an FLT3-ITD or FLT3-D835Y mutation received midostaurin at a dose of 75 mg orally for 3 times daily (TID).53 The drug was rela- tively well tolerated. The most common toxicities were gas- trointestinal side effects including nausea, vomiting and diarrhea. Grade 1/2 gastrointestinal toxicities occurred in 65% of the patients on trial. Of note, 3 patients developed fatal pulmonary events in the context of progressive leuko- cytosis, pulmonary infiltrates of unclear etiology and pneu- monia, respectively. Midostaurin demonstrated measurable responses. Specifically, 14/20 (70%) and 6/20 (30%) of trea- ted patients had 50% reduction in peripheral blood and bone marrow blasts, respectively. However, none of the patient attained CR, despite the significant blast reductions, although 1 patient had <5% blasts in a hypocellular bone marrow and was documented as a partial remission. A subsequent PhaseIIB trial treated 95 patients with relapsed/refractory AML and myelodysplastic syndrome (MDS) with mutant or wild- type FLT3 with midostaurin as a single agent.54 Similar to the prior study, there was a reduction in blasts on treatment in 71% of FLT3-mutant and 42% of FLT3 wildtype-treated patients.

Blast reduction of 50% of more was observed in 42% of treated patients. However, only one patient experi- enced a partial remission and no patients experienced a CR or CR with incomplete hematologic recovery (CRi) suggesting that midostaurin would likely not be sufficient as a mono- therapy agent. Subsequent studies explored combining mid- ostaurin with other agents. In a Phase I/II trial, the combination of midostaurin with the hypomethylating agent 5-azacitidine in 54 untreated and relapsed/refractory AML and high-risk MDS patients showed a modest overall response rate of 26% (1/54 CR, 6/54 CRi), 6/54 morphologic leukemia-free state (MLFS) and 1/54 partial remission).55 The median response duration was 20 weeks and median overall survival was 22 weeks at a median follow-up of 15 weeks (range, 1–85 weeks). The longest response duration was noted in patients without prior exposure to FLT3 inhibi- tors and patients who did not have a previous bone marrow transplantation.55With the emergence of greater understanding of potential efficacy requisites with FLT3 inhibition in AML, Levis et al demonstrated that the lack of response to midostaurin may be due to the lack of a sufficient inhibitory effect on FLT3 as demonstrated by plasma inhibitory levels.22 This led to con- siderations that utilizing midostaurin in an FLT3-inhibitor naïve population, ensuring adequate inhibition of FLT3 by monitoring plasma inhibitory levels, and combining midos- taurin with other agents in frontline setting may be a better approach to use this agent in patients with AML. In a Phase Ib study, midostaurin at different dosing schedules in combi- nation with chemotherapy in 79 younger adults (<60 years of age) newly diagnosed AML patients with mutant or wild- type FLT3 demonstrated CR rates of 80% in the 50 mg twice a day dosing schedule cohort (40 patients).

The OS prob- abilities at 1 and 2 years were 85% and 62% in patients with FLT3-mutated AML, and 78% and 52% in patients with FLT3-wildtype AML, respectively. Interestingly, the median OS of FLT3-mutated patients was similar to that of the FLT3 wildtype patients, leading to the hypothesis that the addition of TKI midostaurin could potentially neutralize the adverse impact of the FLT3 mutation and improve the outcomes of these patients. Midostaurin was not well tolerated when administered at a dose of 50 mg twice a day or 100mg twice a day starting from Day 1 of induction due tosignificant gastrointestinal side effects during days 1–7 when patients received midostaurin concomitantly with the cytotoxic therapy (3+7). The tolerance improved signifi- cantly when patient received 3+7 alone on Days 1 to 7 and the midostaurin was introduced from Day 8 onwards, espe- cially with the 50 mg twice a day dose. Collectively, these data supported the potential benefit of midostaurin in combi- nation with induction therapy in younger patients with FLT3-mutated AML, and was the basis for the Phase III RATIFY trial.The RATIFY trial enrolled FLT3-ITD or FLT3-TKD patients with newly diagnosed AML 18–60 years of age from 13 AML cooperative groups (225 sites). A total of 3277 patients were screened and 717 were randomized patients to receive either midostaurin or placebo with 3+7 (daunorubicin with cytarabine) induction and high-dose cytarabine consolidation therapy (up to 4 consolidations), followed by 12 months of maintenance with either midos- taurin or placebo.

Midostaurin was administered at a dose of 50 mg twice a day on Days 8–21 in induction and each consolidation cycle. During maintenance midostaurin 50 mg twice a day was administered continuously from Cycle 1 Day 1 onwards, without interruption. In order to ensure rapid FLT3 mutational testing and support enroll- ment to the trial across multiple sites in different countries, a large-scale cooperative effort established an efficient poly- merase chain reaction-based FLT3 mutation assay with turnaround time of less than 48 hrs.58 This milestone allowed rapid accrual of patients into the trial, frequently within 4 to 5 days of AML presentation.Patients were stratified based on the type and frequencyof FLT3 mutation into 3 groups: FLT3-TKD, high allelic ratio FLT3-ITD (>0.70) and low allelic ratio FLT3-ITD (

Post- ASCT maintenance with midostaurin was not a part of the RATIFY trial. The cut-off for entry into the RATIFY study was an FLT3 allelic ratio of >/=0.05. On a post-hoc analysis using an arbitrarily selected FLT3 allelic ratio cut-off of 0.7 (to separate what the RATIFY authors called high >/=0.7 versus low allelic ratio <0.7) it was noted that statistically, a similar degree of benefit appeared to be noted in both “high” and “low” allelic ratio. At this time what we can conclude from the RATIFY data is that there is no data regarding the use of midostaurin in patients with an FLT3 allelic ratio<0.05 at diagnosis, and that addition of midostaurin to induc- tion therapy is recommended in all patients with an FLT3 allelic ration >0.05, irrespective of the allelic ratio.RATIFY represented a landmark study as this was the first-ever study to demonstrate the effectiveness of targeted biomarker-driven therapy in patients with AML, a critical proof of concept success that has galvanized the further development and approval of numerous additional-targeted therapies in AML. The RATIFY findings led to US Food and Drug Administration (FDA) approval of midostaurin in com- bination with cytarabine and daunorubicin induction and cytarabine consolidation in newly diagnosed adult AML patients of all ages with FLT3-ITD or FLT3-TKD mutations in April 2017. Of significance, midostaurin was the first drug to be approved for AML in 15 years. A companion FLT3 diagnostic testing was developed and also US FDA approved through a partnership between Invivoscribe and Novartis.58 Accordingly, FLT3 mutational analysis and incorporation of midostaurin into frontline therapy have been integrated into the standard of care work up and treatment for patients with AML in US and Europe. The recommended dosage of mid- ostaurin is 50 mg twice a day with food on Days 8 to 21 of each induction therapy cycle with cytarabine and daunorubi- cin, and on Days 8 to 21 of each consolidation cycle with high-dose cytarabine.

We discuss maintenance in this section using currently available (and unfortunately insufficient) data. We agree that the efficacy of midostaurin in maintenance therapybased on currently evaluable data is unclear and unfortu- nately no matter how we dissect and analyze the RATIFY data this is a topic on which we (or others) will not be able to make any definitive conclusions or recommendations,59 until post-consolidation randomized trials of maintenance that include pre-maintenance and sequential MRD assess- ment (ideally both flow and NGS based) both prior to and post stem cell transplant are conducted and analyzed. Such trials should ideally help establish not only whether main- tenance is beneficial or not in a general fashion, but more importantly help identify patients who benefits most from maintenance versus those who do not benefit or benefit marginally: is it MRD-negative patients who benefit from maintenance by further suppressing emergence of a resis- tance clone or is it in fact MRD+ patients who can be converted to MRD-negative status by maintenance improving their survival with or without a subsequent SCT. Similarly, could maintenance obviate the need for SCT in a subset of patients versus conversely would main- tenance not obviate SCT but actually serve as a bridge to make SCT better by reducing preSCT disease burden? These are questions being studied in Ph+ ALL, a disease with similar development history to FLT3, but with a longer history and more advanced longitudinal datasets. An example of such an ongoing well-designed trial with rich correlative analysis including sequential MRD assess- ment is the MORPHO-trial of gilteritinib versus observa- tion in the post-SCT setting in FLT3 (ITD or D835) mutated patients who under allogeneic SCT.

At this time with the currently available-limited randomized data in FLT3 AML maintenance and based on our experience with TKIs in other disease such as CML, Ph+ ALL, sorafenib in AML our groups recommendation has been to continue maintenance and furthermore to consider pro- longed maintenance rather than 1 year maintenance given that late relapses have been seen at 2 and 3 years post- induction and consolidation and these may be prevented by prolonged maintenance. At this time this is a recommendation without available randomized clinical trial data. Recently, the German-Austrian AML Study Group investigated the efficacy of midostaurin plus intensive chemotherapy, followed by allogeneic hematopoietic stem cell transplantation and single-agent midostaurin maintenance therapy in FLT3-ITD-mutated AML patients.60 In this Phase II study, 284 AML patients received induction therapy and 76.4% (217/284) attained CR/CRi at first or second induction, of which 134/217 (61.7%) underwent allogeneic hematopoietic stem celltransplantation as consolidation. Seventy-five of 134 patients (56%) eventually received maintenance midos- taurin. The median overall survival was 26 months (95% CI, 18.8–36 months). In univariable analysis, patients who started maintenance therapy with midostaur in within 100 days of consolidation had better OS compared to those who did not. While older patients (61–70 years) also benefited from this regimen, there were more frequent cardiac toxicities (22%) and induction death rate (10.5%) in this population.

This Phase II study demonstrated an efficacious use of midostaurin in maintenance therapy which also extended to older patients, albeit in a non- randomized fashion. However, a randomized controlled study is needed to further validate these findings.The most common adverse effects of midostaurin were nausea, vomiting, diarrhea, fatigue and headaches. Given the gastrointestinal side effect profile of midostaurin, pro- phylactic anti-emetics, such as ondansetron, olanzapine or lorazepam, are recommended prior to its administration. While midostaurin was not associated with QTc prolonga- tion in healthy individuals, 10.1% of AML patients on midostaurin had QTc prolongation, compared to 5.7% on placebo.61 No clinical events related to QTc prolongation were noted. It is recommended to assess QTc intervals in AML patients receiving midostaurin, especially concomi- tantly with other QTc-prolonging medications such as some of the anti-emetics, quinolones, and azoles and if possible to avoid concomitant QTc prolonging medications or replace them with suitable alternatives, when feasible.Midostaurin is metabolized to its active metabolites GGP6221 and CGP52421 via CYP3A4 in the liver, and then excreted through feces.24 Midostaurin levels signifi- cantly increased when concomitantly administered with ketoconazole, posaconazole or voriconazole, and levels decreased when rifampicin was co-administered.24,62 Since most AML patients are usually on second-genera- tion anti-fungals as a prophylaxis or as treatment, dose adjustments and monitoring for toxicities are warranted.

In patients taking a concomitant strong CYP3A4 inhibitor (such as posaconazole or voriconazole), it is recommended that the midostaurin dose be reduced to 25 mg twice a day. Some experts may recommend using isavuconazole as the preferred azole in this scenario as it is considered a mod- erate inhibitor of CYP3A4 and is still quite an effectiveanti-fungal.63 While this may be a reasonable option there are no clear consensus guidelines on this topic.Further correlative analysis on midostaurin-treated patients and subset analysis of the RATIFY study set have attempted to elucidate the mechanisms of primary and secondary resistance to midostaurin in patients with AML. Upregulation of myeloid cell leukemia 1 protein (MCL-1) was shown to be an important mechanism of primary resistance to midostaurin in AML.64 Similarly, upregulation of anti-apoptotic genes and down-regulation of proapoptotic genes were demonstrated to be associated with acquired resistance to midostaurin in AML.65 A sin- gle amino acid substitution at position 676 (N676K) within the FLT3 kinase domain was identified to be the cause of resistance in 1 of 6 evaluable relapsed/refractory AML patients who relapsed while on midostaurin treatment in the Phase II trial of single-agent midostaurin. Subsequent work demonstrated that the allelic burden of mutant FLT3 dictated responses and established the basis for assessing the allelic frequency in AML.67

These find- ings were the impetus for attempting to combine midos- taurin with other drugs and to generate other, potentially more potent, specific and better tolerated FLT3 inhibitors.Better understanding of the underlying AML biology, clonal evolution and selection, emergence of a number of effective- targeted therapies and combinations with FLT3 and isocitrate dehydrogenase (IDH) inhibitors have transformed the way we treat AML. Recently the US FDA approved second-genera- tion FLT3 inhibitor gilteritinib as a single agent in patients with FLT3-mutated relapsed/refractory AML (ITD or D835) based on impressive CR/CRi and OS compared with conven- tional chemotherapy (cytotoxic combination chemotherapy or hypomethylating agent therapy) in a randomized Phase III (ADMIRAL trial).68 Another highly potent and very selective FLT3 inhibitor quizartinib improved CR/CRi and OS in a randomized Phase III trial in relapsed/refractory FLT3-ITD mutated AML and is anticipated to have US FDA approval in the near future.69,70 A third second-generation FLT3-inhi- bitor is crenolanib and this agent targets both FLT3-ITD and-D835 potently (Figure 1).71,72 These FLT3 inhibitors have a high FLT3-specificity and high response rates as single agents compared with first-generation FLT3-inhibitors such as mid- ostaurin and sorafenib with a relatively good tolerability pro- file, and as such, are being evaluated in frontline combinations with induction therapy and hypomethylating agent-based ther- apy in newly diagnosed FLT3-mutated AML.

Clinical trials in newly diagnosed AML with quizartinib (NCT01390337, NCT03723681, NCT02668653, NCT02834390), gilteritinib (NCT03836209, NCT02752035, NCT02236013, NCT02310321) and crenolanib (NCT03258931, NCT02 283177) in combination with induction therapy (3+7 or HMA) are ongoing, and preliminary-reported data appear encouraging.73–75 An emerging question is regarding how these more selective, potent TKIs such as quizartinib, gilter- itinib and crenolanib will compare to broad, non-selective multikinase inhibitors such as midostaurin and sorafenib when combined with frontline induction therapy in FLT3- mutated AML. It has been postulated that newly diagnosed FLT3-mutated AML is less addicted to FLT3 signaling but more dependent on multiple kinase pathways for growth and proliferation suggesting that broad kinase inhibitors may be of benefit and potentially preferable over selective TKIs. This is in contrast to the time of relapse wherein FLT3 is often selected out or emerges as the resistance driver clone and more specific, potent FLT3 inhibitors may be more effective as single agents or even more so in combinations. This assumption is being explored in ongoing-randomized studies of frontline induction with second-generation FLT3 TKIs (including crenolanib and gilteritinib) versus frontline induc- tion with midostaurin (NCT03258931, NCT03836209).

Given the nature of midostaurin as a broad, multi- kinase inhibitor, and the recognition that it inhibits num- ber of kinases and pro-survival pathways that are essential to leukemic cell growth, survival, and prolifera- tion, and taking into account the clinical activity seen with midostaurin in patients with wildtype FLT3,54 it is possible that the on-leukemia, off-FLT3 effects of mid- ostaurin afford benefit regardless of the FLT3 status. The impact of midostaurin in FLT3 wildtype AML is cur- rently being explored in an ongoing Phase III clinical trial (NCT03512197).
Midostaurin was approved in young patients fit for high- intensity chemotherapy who received 3+7 regimen based on the RATIFY trial results. The impact of adding midostaurin to low-intensity therapies such as hypomethylating agents (azacitidine or decitabine) or to low dose cytarabine in older patients with AML who are not fit for high-intensity che- motherapy is yet to be defined. In vitro the combination of decitabine and midostaurin was synergistically active against FLT3-ITD mutation expressing AML cells, advo- cating for such combinations.76 In a Phase I/II trial of midostaurin combined with azacitidine the combination appeared to be effective and safe in patients with AML and high-risk MDS.55 A number of midostaurin containing low-intensity combinations in elderly unfit population are actively been investigated (NCT01130662, NCT01093573, NCT01093573, NCT02634827, NCT01846624).

Another potential use of midostaurin in AML is in core binding leukemia (CBF), with or without FLT3 mutations, as KIT mutations are commonly seen and thought to impact prognosis in CBFs.77,78 This is currently being explored in two clinical trials (NCT03686345, NCT01830361). Post- transplant maintenance with midostaurin is also of interest in FLT3-mutated AML and is being evaluated in an ongoing- randomized Phase II clinical trial (NCT01883362; RADIUS trial) with preliminary data suggesting a trend for improved EFS and OS with post-allogeneic SCT midostaurin maintenance,79 similar to what was noted with sorafenib in the SORMAIN trial, but not yet achieving statistical signifi- cance in the RADIUS trial. It must be noted though that midostaurin and sorafenib maintenance, particularly post- allogeneic SCT, are associated with more toxicities and require frequent and liberal dose adjustments to allow patients to stay on therapy. As FLT3 inhibitors are more frequently being used in FLT3-mutated AML, the role of stem cell transplant needs to be better defined. In the RATIFY trial, patients who received midostaurin with induction/consolidation and then underwent subsequent SCT in first remission had the best outcomes, supporting the choice of transplant in the frontline setting once remission was achieved. Of note, midostaurin was recommended to be used as a maintenance for 1 year in this clinical trial preASCT or in patients who did not got to ASCT, but post-ASCT midostaurin maintenance was not part of the RATIFY trial. The duration of midostaurin and other TKI usage (such as BCR-ABL inhibitors in Ph+ ALL) is not yet definitely defined, both post-transplant and in patients who do not go for transplant. It is possible that longer FLT3 inhibitor maintenance beyond 1 year may change the prognostic sig- nificance of FLT3-ITD mutation and potentially even the need for SCT (especially is specific molecularly select, favorable groups) and this is being evaluated in ongoing Phase II studies.

Additional expected developments in the usage of midos- taurin, particularly as we better understand the underlying mechanisms of resistance to TKIs in general and FLT3 inhi- bitors specifically, are rationally designed treatment combina- tions with agents targeting known resistance pathways. One expected combination is with Bcl-2 inhibitor venetoclax and HMA or LDAC (triple combination). As was discussed ear- lier, up-regulation of anti-apoptotic and down-regulation of pro-apoptotic proteins is a known major mechanism of resis- tance to FLT3 inhibitor therapy that may be circumvented with the addition of venetoclax. One report showed that FLT3-ITD cells have higher level of Bcl-2 compared to FLT3 wildtype cells.80 In addition, venetoclax primary and secondary resis- tance appears to be driven by FLT3-ITD mutation.81 MCL-1 has been reported to be an essential effector of FLT3-ITD- mediated drug resistance.82 A number of FLT3 inhibitors down-regulate MCL1, and may thus reduce resistance to BCL2 inhibitors.83 Hence, in addition to synergism with HMAs, the triple combination of midostaurin plus venetoclax with HMA (or LDAC) in FLT3-mutated AML is rational and its evaluation in clinical trial is anticipated in the near future. In summary, a number of active trials combining midos- taurin with other agents are ongoing (Table 1). We hope to identify and further refine in the coming years the patient subgroups with the ASP2215 greatest potential benefit from midostaurin, disease biomarkers of response and resistance, the optimal time of usage, and rationally designed treatment combinations that will result in improved efficacy and survival.