Midostaurin, enasidenib, CPX-351, gemtuzumab ozogamicin, and venetoclax bring new hope to AML

Once touted as the “boulevard of broken dreams,” acute myeloid leukemia (AML) has been a therapeutic graveyard for a litany of failed drug development programs attempting to reform the AML treatment landscape.¹  In this article, we highlight a selection of emerging drugs starting to make an impact in the care of patients with AML.

Midostaurin, an N-benzoyl staurosporine analog derived from Streptomyces staurosporeus, was initially characterized as an inhibitor of protein kinase C. Found to inhibit C-FOS and MAPK, it was also shown to have multikinase inhibitory activity against platelet-derived growth factor receptor, CDK1, KIT, and vascular endothelial growth factor.²  A drug screen identified midostaurin to have FLT3 inhibitory activity, which led to its repurposing as a drug for FLT3 mutant AML.³  A decade-long clinical development journey culminated in a pivotal trial (Randomized AML trial in FLT3 in patients less than 60 years [RATIFY]), which combined midostaurin or placebo with standard induction and consolidation chemotherapy, followed by 12 months of midostaurin or placebo maintenance for adults with FLT3-ITD and FLT3-TKD mutant AML.⁴  This global effort, which involved 13 AML cooperative groups, demonstrated that midostaurin significantly improved 4-year overall survival (OS) from 44.3% to 51.4% (hazard ratio, 0.78; P = .009), compared with placebo.⁴  This led to US Food and Drug Administration (FDA) approval of midostaurin, with the benefit of midostaurin observed in patients with FLT3-ITD low (0.05-0.7) and high allelic ratio (>0.7), as well as in patients with FLT3-TKD.⁴  These mutations are collectively found in up to 40% of the AML population.^5-7  Although FLT3 testing in the RATIFY trial was performed by academic laboratories, LeukoStrat CDx FLT3 Mutation Assay (Invivoscribe Technologies, Inc.) was approved in parallel as a companion assay for the detection of FLT3 mutant AML in the United States by the FDA. Rapid screening for both FLT3-ITD and FLT3-TKD at diagnosis will now be routinely required to effectively incorporate midostaurin into the standard of care for AML. Furthermore, the FDA label does not restrict use to patients 18 to 59 years of age, as occurred in RATIFY. Therefore, routine screening and treatment of mutant FLT3 is likely to extend to older populations fit for intensive chemotherapy.

Several questions remain, however, regarding the optimal use of midostaurin in AML. The median exposure of patients to midostaurin was only 42 days, suggesting that the main benefit was derived early on in treatment.⁴  The magnitude of benefit and optimal duration of midostaurin as maintenance therapy after completion of chemotherapy is contentious. Further study to demonstrate a significant survival outcome is likely to require a formidable number of patients and time. A more feasible objective may be demonstration that maintenance treatment can effectively eliminate minimal residual disease and prolong relapse-free survival.

Concurrent NPM1 mutation partially mitigates the adverse prognostic impact of FLT3-ITD,⁸  and future post hoc analyses to examine the magnitude of midostaurin benefit in various FLT3-ITD/NPM1 subgroups is warranted. The RATIFY study also suggested greater benefit for midostaurin in males, but not females, with FLT3-ITD and conversely, for females, but not males, with FLT3-TKD.⁴  The data also indicated that males had a worse baseline outcome than females with FLT3-ITD. Further work is needed to confirm and unravel these gender conundrums.

In the RATIFY study, posttransplant survival was marginally better in those receiving prior midostaurin (P = .07), with benefit limited to patients transplanted in first remission.⁴  Future work should verify whether midostaurin delivers more patients to transplant in first remission without minimal residual disease and determine whether the addition of midostaurin or other FLT3 inhibitors to the postallogeneic stem cell transplant (SCT) setting will lead to further improvements in survival.

The recent FDA approval of midostaurin as frontline therapy for FLT3-mutant AML may create hurdles for new FLT3 inhibitors seeking first-line drug registration because the control arm will need to include midostaurin, thereby setting a higher clinical bar for new investigational agents to surpass. With midostaurin as the comparator, it remains to be seen whether the greater FLT3 potency associated with newer generation inhibitors, such as quizartinib, crenolanib, and gilteritinib, will have more clinical relevance than the multikinase effects of midostaurin in randomized head-to-head studies. To date, it is not clear why midostaurin succeeded, whereas other multikinase FLT3 inhibitors, such as lestaurtinib did not.⁹  Midostaurin’s multikinase mechanism of action is likely to see it explored in combination with standard chemotherapy in FLT3 wild-type AML. This will follow the lead of sorafenib, another multikinase inhibitor, which improved event-free survival in unselected adults with AML when combined with intensive chemotherapy.¹⁰ 

First identified from the whole genome sequence of an index patient with AML in 2009,¹¹  recurrent hotspot mutations affecting the catalytic domains of IDH1 (Arg132) and isocitrate dehydrogenase 2 (IDH2 [Arg140 and Arg172]) occur in ∼8% and ∼12% of cases, respectively.^7,12,13  Prognostic impact of mutant IDH2 is contentious (reviewed by Medeiros et al),¹⁴  and isolated IDH2-R172 has been linked to a favorable outcome in 1 study.⁶  Studies linking mutant IDH1/2 to the neometabolite 2-hydroxyglutarate and arrested myeloid differentiation,^15-17  propagated the development of targeted inhibitors to mutant IDH2 (AG-221; enasidenib)^18,19  and IDH1 (AG-120; ivosidenib and BAY1436032).^20,21  The allosteric IDH2 inhibitor enasidenib effectively suppressed production of 2-hydroxyglutarate, releasing myeloid blasts from differentiation block.¹⁹  As an orally administered drug in relapsed/refractory IDH2 mutant AML, enasidenib (100 mg daily) produced complete remission (CR) and CR with incomplete hematologic recovery (CRi) in 26.6% of patients.¹⁸  An additional 12% achieved either partial remission or morphologic leukemia-free state, giving an overall response rate of 38.5%.¹⁸  Median duration of response was 5.6 months (8.8 months if CR achieved) and OS 9.3 months (19.7 months if CR achieved), in contrast to ∼3 months with standard therapies.²²  Interestingly, despite reduced affinity for IDH2-R172, the overall response rate to enasidenib was 53.3%, compared with 35.4% for IDH2-R140.²³  Clinical responses to enasidenib were observed without reduction in the IDH mutant allele burden, reflecting conversion from predominantly undifferentiated, to differentiated clonal hematopoiesis.¹⁹  The median time to best response was 3.7 months, with 82% of responses observed by cycle 7.¹⁸  Responses have also been observed in patients with very small IDH clone sizes, raising the possibility that the drug may have additional paracrine effects on non-IDH mutant blasts.^19,24  Enasidenib has a distinct toxicity profile, the most important being IDH inhibitor–associated differentiation syndrome (IDH-DS), which occurs in 14% (grade 3+ severity in 7%).¹⁸  IDH-DS may develop with or without concurrent hyperleukocytosis, and as late as 5 months after therapy initiation. Rapid myeloid proliferation manifesting as noninfectious leukocytosis has also been observed, requiring hydroxyurea administration and occasionally measures to mitigate tumor lysis syndrome.^18,25  The terminal half-life of enasidenib is ∼5.7 days¹⁸ ; thus, drug cessation alone may not ameliorate IDH-DS. A high degree of clinical vigilance is needed and empirical use of steroids frequently warranted in cases of suspected IDH-DS, emphasized by a black box warning in the product information.

A particular challenge for patients and doctors will be attempts to maintain patients with persistent AML on IDH inhibitors for prolonged periods while waiting for a clinical response. Identifying predictive biomarkers of IDH inhibitor response will be highly valuable in justifying the decision to maintain patients on therapy. Preliminary results suggest a low likelihood of response when mutant NRAS is present.¹⁹  Further work is needed to confirm the predictive value of this and other molecular markers. Enhancing clinical outcomes by combining IDH inhibitors with other drugs, such as hypomethylating agents (HMAs) or standard chemotherapy is already being evaluated. Registration studies in the first-line setting, however, will be challenged by the need for real-time mutation screening, the relatively low frequency of mutant IDH1 and IDH2 in the AML population and increasing commercial competition from non-IDH targeted drug options currently in development for AML that are also active in this patient subgroup.

Prior preclinical studies demonstrating that cytarabine and daunorubicin delivered at molar ratios between 1:1 and 10:1 were synergistic, whereas lower ratios (1:5-1:10) were antagonistic, led to the development of CPX-351, which encapsulates cytarabine and daunorubicin at a fixed 5:1 molar ratio.²⁶  This ratiometric liposomal delivery system enhanced drug concentration in bone marrow and drug uptake into AML blasts, promoting superior antileukemic efficacy in vivo.^27,28  In human trials, the mean elimination half-life for CPX-351 was 25 hours for daunorubicin and 37 hours for cytarabine, substantially longer than pharmacokinetic exposures to free drugs. This could explain the longer time to neutrophil (36 vs 32 days) and platelet (37 vs 28 days) recovery with CPX-351, compared with conventional 7+3 chemotherapy.²⁹  Despite greater marrow suppression, a randomized phase 2 trial in patients 60 to 75 years did not show a significant increase in 30-day treatment-related mortality with CPX-351 (3.5%), compared with 7+3 (7.3%) as first-line therapy.³⁰  Although CPX-351 failed to increase OS in the overall study population, a preplanned analysis identified a superior CR rate and OS for CPX-351 in patients with secondary AML.³⁰  A pivotal phase 3 study therefore recruited 309 patients aged 60 to 75 years with a history of prior cytotoxic treatment, antecedent myelodysplastic syndrome (MDS) or chronic myelomonocytic leukemia, or AML with World Health Organization–defined MDS-related cytogenetic abnormalities. This study confirmed the higher CR rate (47.7% vs 33.3%, P = .016) and OS (median, 9.56 vs 5.95 months; hazard ratio, 0.69; P = .005) for CPX-351 over 7+3 (60 mg/m² daunorubicin),³¹  prompting the FDA to approve CPX-351 for therapy-relatedAML (t-AML) and AML with myelodysplasia-related changes (AML-MRC).

The road to CPX-351 approval highlights the importance of studying subgroup responses to investigational therapies within clinical trials. The virtue of identifying a responder population within the context of a randomized phase 2 study followed by validation with a targeted phase 3 study was successfully demonstrated. Perplexingly, the rationale for benefit in t-AML and AML-MRC remains an open question, with one hypothesis that liposomal drug delivery may overcome Pgp-mediated drug resistance.

The FDA label includes patients with t-AML and AML-MRC, which therefore extends the eligible population to include (1) younger patients and (2) AML with multilineage dysplasia (MLD), both of which were not specifically examined in the pivotal study. The morphologic definition of MLD is subject to interobserver bias,³²  and the revised World Health Organization 2016 definition of MLD now excludes patients with NPM1^MUT and biallelic CEBPA^MUT.³³  Del(9q) has also been removed from the list of MDS-related cytogenetic abnormalities.³³  Cytogenetic information may not be readily available when treatment needs to commence; therefore, it remains to be seen how these nuances will affect prescribing and drug provision practices. AML with preceding myeloproliferative neoplasm was not included in the definition of AML-MRC.³⁴  In first relapse, a randomized phase 2 study failed to show benefit for CPX-351 in AML compared with other salvage regimens, except in patients with a high European Prognostic Index risk score.³⁵  Therefore, future research will be important to validate the role of CPX-351 in the salvage setting, as well as in defined cytogenetic and molecular AML subgroups not adequately powered in the primary study to define the magnitude and consistency of benefit. Safety in combination with other novel therapies will also be an area of increasing future interest.

Gemtuzumab ozogamicin (GO) is a humanized immunoglobulin G₄ antibody (hP67.6) directed against CD33 and conjugated via a hydrolysable linker to the DNA toxin calicheamicin. GO/CD33 complexes are internalized into lysosomes, releasing calicheamicin and promoting single and double-strand breaks and cellular death. GO initially received accelerated FDA approval in 2000 for the treatment of CD33⁺ AML aged ≥60 years in first relapse, with the requirement that the company undertake a confirmatory postmarketing study.^36,37  A phase 3 study (S0106) was conducted by SWOG in untreated de novo AML, comparing daunorubicin/cytarabine (DA, 45 mg/m² daunorubicin) plus GO 6 mg/m² on day 4 with DA alone (60 mg/m² daunorubicin). The GO arm had higher induction mortality (5.5% vs 1.4%), without improving CR or relapse-free survival.³⁸  Based on these negative results, Pfizer was forced to withdraw GO from the market on 21 June 2010. Over the next decade, 4 additional investigator-led randomized studies in Europe (GOELAMS AML2006IR,³⁹  MRC AML15,⁴⁰  ALFA-0701,⁴¹  and NCRI AML16⁴² ) were completed. ALFA-0701 randomized 278 patients with untreated de novo AML aged 50 to 70 years to DA (60 mg/m² daunorubicin) alone or in combination with a fractionated GO induction schedule (3 mg/m² on days 1, 4, and 7).⁴¹  A single dose of GO (3 mg/m²) was also given on day 1 of each of 2 consolidation cycles. Although CR with or without platelet recovery and early deaths were similar, patients in the GO arm had significantly improved median event-free survival (19.6 vs 11.9 months; P = .00018) and OS (34 vs 19.2 months; P = .046), with a subanalysis revealing benefit limited to patients with favorable and intermediate-risk karyotype.⁴¹  A meta-analysis of 3325 patients from 5 randomized studies in untreated AML (aged 18-84) concluded that GO improved OS in patients with favorable and intermediate-risk karyotype when combined with standard induction chemotherapy.⁴³  Rates of veno-occlusive disease (VOD) and 30- and 60-day mortality were lower with 3 mg/m² vs 6 mg/m² GO.⁴⁴  The MyloFrance-1 study also gave 3 mg/m² on days 1, 4, and 7 to patients with AML at first relapse.⁴⁵ 

The first approved dose of GO (9 mg/m² repeated after 2 weeks) was associated with grade 3-4 hyperbilirubinemia (23%) and elevated transaminases (17%), as well as prolonged severe myelosuppression.⁴⁶  At 9 mg/m², GO supersaturated CD33 binding sites, even after 2 weeks, resulting in internalization of GO/CD33 complexes and reduced availability of unbound target antigen.⁴⁷  Because only one-half of the antibody pool is actually conjugated to calicheamicin, increased binding site competition from unconjugated and inactive drug was another concern.³⁶  Delivering GO using a fractionated dosing schedule substantially improved the safety profile without compromising clinical outcomes.^41,45  A major concern for patients receiving GO is the risk of VOD, especially among patients who received SCT within 3 months. Revised dosing schedules appear to have lowered rates of VOD (<5% in ALFA-0701⁴⁸  and none were reported in the MyloFrance-1 study⁴⁵ ). Experience from these studies, however, remains limited, and clinicians should remain alert to the risk of VOD by avoiding concurrent hepatotoxic medications, minimizing SCT within 3 months of GO and continuing to monitor rates of VOD in the postmarketing period.

Although the US label for GO is broad, many unanswered questions remain. The limited GO activity in adverse karyotype AML requires further investigation, but may relate to reduced CD33 expression in adverse karyotype and lower rates of GO response in patients with increased expression of Pgp or MDR1.⁴⁵  In contrast, GO has pronounced efficacy in acute promyelocytic leukemia, which strongly expresses CD33.^49,50  For consolidation therapy in the ALFA-0701 study, GO was combined with DA,⁴¹  which differs from higher dose cytarabine-based consolidation regimens used extensively in other parts of the world. Further research to explore the safety and additional efficacy of fractionated-dose GO in combination with higher dose cytarabine in consolidation are therefore warranted. Although GO has been included in the FDA label for patients with relapsed and refractory disease, the MyloFrance-1 study was a phase 2 uncontrolled trial,⁴⁵  making it difficult to determine the superiority of GO over other conventional salvage options.

Much remains to be learnt regarding the potential for GO to be combined with other novel drugs currently in development for AML. For example, NPM1 mutant AML exhibits high levels of CD33 and BCL-2 expression, which may make it attractive to combine GO with venetoclax if this combination is found to be tolerable.^51,52  The frequent association between NPM1 and FLT3 mutations may also stimulate studies combining GO with FLT3 inhibitors. Therefore, despite almost 2 decades since its initial approval, research into the full use and potential of GO is just beginning.

Approximately one-third of elderly patients (>75 years) with AML are palliated without active therapy.⁵³  Increasing medical comorbidities, a higher frequency of poor risk gene mutations, and prior HMA failure are some of the diverse challenges limiting progress in elderly patients with AML.^54,55  Clinical responses (CR/CRi) to standard AML therapies used in the elderly, such as azacitidine (28%), decitabine (26%), or low-dose cytarabine (LDAC, 11% to 18%) are modest.^23,56,57  Increased expression of the pro-survival protein BCL-2 relative to the pro-apoptotic protein BAX is associated with reduced CR rates, earlier relapse, and inferior OS in patients receiving intensive chemotherapy for AML.⁵⁸  The BCL-2 inhibitor venetoclax was only modestly effective as monotherapy in relapsed/refractory AML (19% CR/CRi).⁵⁹  Recent phase 2 studies in elderly patients unfit for intensive chemotherapy have combined venetoclax with either HMAs or LDAC, producing CR/CRi rates of 62% to 68% and 12-month survival outcomes of 50% to 70%.^60,61  Responses were achieved rapidly (median, 1 month) and early mortality was low (2%). Registration studies are currently under way to validate the benefit of venetoclax in combination with standard therapies in elderly patients with AML (NCT02993523 and NCT03069352).

The robust activity of venetoclax in combination with low-intensity therapies in elderly patients with AML provides a competitive alternative to other inhibitors, such as FLT3 and IDH inhibitors, without the need for preemptive mutation screening. This may increase the difficulty of patient recruitment into registration studies of elderly patients with AML targeting a specific subgroup in the frontline setting. The response rates and 12-month OS for venetoclax/HMA or LDAC also compare favorably with results achieved using intensive chemotherapy in the elderly AML population.⁶²  Therefore, the distinction between “fit” and “unfit” older patients when selecting therapy may lose relevance if a highly active treatment with relatively low toxicity becomes available. In younger adults, and in relapsed/refractory AML, future studies will likely determine if venetoclax can be safely combined with more intensive chemotherapy approaches. The best outcomes for venetoclax/HMA or LDAC appear to be in patients with NPM1 mutant AML, which notably express high levels of BCL-2.⁵²  Future research should also seek to understand mechanisms of clonal resistance, and the potential for BH3-mimetics targeting other pro-survival proteins, such as MCL1 to be combined with BCL-2 targeting in AML.^63,64 

Despite a relatively unchanging therapeutic landscape for several decades in AML, the stage has now been revitalized by the debut of 4 new FDA approvals within the space of just 6 months in 2017 for patients with FLT3 mutant AML, IDH2 mutant AML, CD33 positive AML, t-AML, and AML-MRC (Table 1). The majority of the AML population may now have treatment outcomes augmented by the addition of a novel drug in the clinic. Additionally, venetoclax is also making solid strides toward a possible drug registration in elderly patients with AML. Although cytarabine and daunorubicin will continue to play an important role in AML, patients and physicians will now have the help of several new recruits in their fight against this lethal blood cancer.

Contribution: A.H.W. designed, wrote, and edited the paper. I.S.T. wrote and edited the paper.

Conflict-of-interest disclosure: A.H.W. is on the advisory board and has received honoraria and research funding from Novartis, Celgene, and AbbVie. I.S.T. declares no competing financial interests.

Correspondence: Andrew Wei, Department of Clinical Haematology, Alfred Hospital and Australian Centre for Blood Diseases, Monash University, Commercial Road, VIC 3004, Australia; e-mail: andrew.wei@monash.edu.