Management of primary refractory acute myeloid leukemia in the era of targeted therapies

Abstract

Primary refractory acute myeloid leukemia (AML), or primary induction failure, represents a continued challenge in clinical management. This review presents an overview of primary refractory disease and a discussion of risk factors for induction failure, including current evidence regarding the impact of karyotype and molecular mutation status on responsiveness to chemotherapy. We review the evidence for various treatment options for refractory AML including salvage chemotherapy regimens, allogeneic hematopoietic stem cell transplantation, targeted agents, and non-intensive therapies such as hypomethylating agents. A therapeutic approach to this patient population is presented, and several new and emerging therapies are reviewed.

INTRODUCTION

Approximately 10–40% of adults with acute myeloid leukemia (AML) will have persistent leukemia following intensive induction chemotherapy[1–4]. Despite significant progress that has been made in AML therapy, the treatment of primary refractory disease (PRD) remains challenging due to relatively low response rates to salvage chemotherapy and poor overall survival (OS) rates. Allogeneic hematopoietic stem cell transplantation (HSCT) represents the best hope for long-term cure in this group of patients, although many patients with PRD are not candidates for transplantation due to factors such as poor performance status or advanced age. This review discusses risk factors for PRD and provides an update on current treatment options and promising emerging therapies.

DEFINITION

For the purposes of this review, we will define PRD--also known as primary induction failure--as the lack of a complete remission after two courses of standard-dose cytarabine-based induction chemotherapy (e.g. cytarabine and anthracycline “7+3”)[5,6] or at least one course of high-dose cytarabine-based induction[7]. For this definition, complete remission (CR) is defined as fewer than 5% blasts on morphologic examination of the bone marrow along with recovery of the absolute neutrophil count (ANC) to greater than 1.0 × 10⁹/L and recovery of the platelet count to at least 100 × 10⁹/L. Patients with all criteria for CR except either residual neutropenia or thrombocytopenia (CR with incomplete blood count recovery, CRi)[5,8] do have inferior outcomes compared to those who achieve a full CR[9], but patients with CRi are not included in the primary refractory group.

It should be noted that the definition of a complete remission is evolving as the prognostic significance of measurable residual disease (MRD) by multiparameter flow cytometry (MFC) or molecular-based assays becomes more clear[10,11]. The 2017 European LeukemiaNet (ELN) recommendations include CR without MRD (CR_MRD-) as a separate response category[5]. Patients who achieve a CR_MRD- have improved outcomes compared to those who achieve a CR with persistent MRD, including those who undergo allogeneic HSCT[12–18]. A lack of standardization among laboratories and variation in sensitivity depending on the type of assay used has slowed widespread adoption of MRD monitoring in AML [10,19]. Therefore, for the remainder of this review article we will continue to use the traditional definition of PRD as persistent leukemia by light microscopy.

RISK FACTORS

Risk factors for primary refractory AML include a complex or monosomal karyotype, advanced age, an increased time to blast clearance in the bone marrow or peripheral blood, a high white blood cell (WBC) count at diagnosis, secondary AML, and the presence of certain molecular mutations such as TP53[20–23].

Cytogenetics

Karyotype is the most important factor in estimating prognosis. The revised MRC cytogenetic classification system stratifies karyotypes into favorable, intermediate and adverse risk, which are associated with 10-year OS rates of 69%, 33%, and 12%, respectively[24,25]. Karyotype is also an important predictor of whether a patient will achieve a CR with induction chemotherapy[20,21,26]. A large prospective analysis of 1213 adults with de novo AML who were treated with “7+3” induction chemotherapy on 5 Cancer and Leukemia Group B (CALGB) studies found that only 30% of subjects with ≥ 5 unrelated cytogenetic abnormalities and 47% of subjects with 3–4 cytogenetic abnormalities achieved a CR, as compared to 68% of subjects with a normal karyotype (p<0.001 and p=0.002, respectively)[20,21].

Among the subgroup of patients with adverse cytogenetics, inv(3)(q21q26), t(3;3)(q21;q26), and other abnormalities of 3q (excluding t(3;5)(q21~25;q31~35), which are relatively rare and occur in approximately 2% of adults with AML, are well-known risk factors for PRD[20,21,24,27–30]. A monosomal karyotype (MK), defined as having at least 2 autosomal monosomies or a single autosomal monosomy in the presence of at least 1 other structural chromosomal abnormality, is also associated with a very poor prognosis and an increased rate of PRD[22,31–33]. Breems et al. found that among 1975 patients with AML treated on various Dutch-Belgian Haemato-Oncology Cooperative Group (HOVON)/Swiss Group for Clinical Cancer Research trials, 52% of subjects with a MK failed to achieve a CR compared to 18% of patients overall, and subjects with a MK had a 4-year OS of only 4%[22]. Of note, HOVON conducted a prospective randomized trial in which patients with newly diagnosed AML were randomized to either standard-dose cytarabine (200mg/m2/day by continuous infusion on days 1–7) or high-dose cytarabine (HiDAC) (1000mg/m2 every 12 hours on days 1–5), both in combination with idarubicin[34]. Although there was no difference in CR rate or OS overall at a median follow-up of 5 years, a sub-group analysis found that subjects with a monosomal karyotype had improved 5-year event-free survival (13% vs. 0%) and OS (16% vs. 0%) with HiDAC, suggesting a HiDAC-based induction regimen be considered in patients with a known monosomal karyotype[34]. There has also been some enthusiasm for considering the inclusion of cladribine in the induction regimen of patients with adverse-risk karyotypes, though whether this benefits monosomal karyotype is uncertain[35–39].

Age

An older age is associated with an increased likelihood of PRD, which is primarily related to biological differences in AML which tend to occur with increasing age, including an increased incidence of complex and monosomal karyotypes[31,40]. Patients with an older age also have an increased likelihood of having a poor performance status (PS) or multiple other medical comorbidities. In addition to increasing the risk of treatment-related mortality, an Eastern Cooperative Oncology Group (ECOG) PS ≥ 2 has been independently associated with failure to achieve CR with induction therapy even in the absence of early death[23].

Time to blast clearance

The time to blast clearance in the bone marrow and peripheral blood after starting induction chemotherapy has also been recognized as a predictor of PRD[41–45]. Several studies have found that a higher blast percentage on a nadir bone marrow biopsy performed on day 14 to 16 of the first cycle of induction therapy is associated with an increased incidence of PRD[41–43].

Molecular mutation status

Among the recurrent somatic mutations commonly found in AML[27,46], some have been associated with a more favorable prognosis and others with a less favorable outcome[47]. With respect to achieving a CR with induction chemotherapy, both gene fusions and recurrent mutations have been predictive of induction success rates. For example, patients with core binding factor fusions have extremely high rates of CR, as do those with cytogenetically normal AML who have either a mutation in NPM1 or double mutations in CEBPA[27]. In addition, patients with these genotypes also have a relatively high 4-year OS rate (around 60%) in the absence of FLT3-ITD mutation[48]. On the other hand, TP53 mutations are associated with complex karyotype and a significantly worse outcome. This reflects both a low rate of initial remission as well as a high relapse rate, regardless of post-remission therapy delivered [49,50]. An analysis of adults with AML who were treated on various studies of the German-Austrian AML study group found that among 234 patients with complex karyotypes, those with mutations in TP53 were less likely than those who were TP53-wild type to achieve a CR with induction chemotherapy (28% vs. 50%, p=0.01) and had a dramatically worse 3-year OS (3% vs. 28%, p<0.0001)[49]. Interestingly, a non-randomized study of frontline decitabine therapy in TP53-mutated AML showed more favorable response rates and short-term outcomes in this group, suggesting hypomethylating agents might represent preferred agents for these patients[51]. The prognostic significance of other somatic mutations with respect to initial induction chemotherapy response is less clear[47,52–54].

TREATMENT

The treatment of patients who are refractory to standard induction chemotherapy remains extremely challenging and outcomes are overall poor. Therapeutic options include more intensively dosed or timed salvage chemotherapy, direct allogeneic HSCT, targeted agents, a hypomethylating agent (HMA), other non-intensive therapies such as low-dose cytarabine (LDAC), and enrollment to clinical trials.

Salvage chemotherapy

Many salvage chemotherapy regimens have been studied in primary refractory AML, including HiDAC, HAM, MEC, FLAG, FLAG-Ida, and CLAG-M (Table 2), among others. It is somewhat challenging to compare the outcomes of studies of salvage regimens due to both significant heterogeneity in the patients included in the studies and in the definitions by which PRD was defined. In general, CR rates with intensive salvage chemotherapy are in the range of 20 to 35%, although this varies based on the patient population studied as older patients and patients with high-risk cytogenetics are less likely to achieve a CR[55–58]. A large study of 1025 patients with PRD who were treated on German-Austrian AML Study Group trials and underwent salvage chemotherapy with various regimens including HAM (high-dose cytarabine and mitoxantrone), A-HAE (high dose-cytarabine, etoposide, and ATRA), and GO-A-HAM (gemtuzomab ozogamicin, ATRA, and HAM), found an overall CR/CRi rate of 36%[58]. There is not a standard of care regarding which salvage chemotherapy regimen should be chosen first line for a patient with PRD, but we favor a HiDAC-based regimen if the patient has not previously received high-dose cytarabine during the initial induction attempts [5,59].

Table 2.

Selected salvage chemotherapy regimens

Abbrev.	Regimen	Total no. patients	Overall CR rate	No. pts. with PRD	CR rate for pts with PRD	OS	Early mortality rate/TRM	Ref.
HiDAC	Ara-C 3g/m2 q12h d1–6	81	26/81 (32%)	27	Not reported	Med OS 8 mo	10/81 (12%)	[55]
HAM/ S-HAM	Ara-C 3g/m2 q12h d1–3 MIT 12mg/m2 d2–3 Ara-C 3g/m2 q12h d1,2,8,9 MIT 10mg/m2 d3,4,10,11	150	42/150 (28%)	150	42/150 (28%)	Not reported	3/150 (2.0%)	[58]
GO-A-HAM	GO 2mg/m2 (5mg max) d1 ATRA 45mg/m2 po d4–6 and 15mg/m2 po d7–28 +HAM as above	140	70/140 (50%)	140	70/140 (50%)	Not reported	2 (1.4%)	[58]
MEC	MIT 8mg/m2 d1–5 Etoposide 100mg/m2 d1–5 Ara-C 1g/m2 d1–5	63	16/63 (25%)	17	3/17 (18%)	Med OS 5.4mo Med DFS 9.3mo	7/63 (11%)	[56]
GCLAC	G-CSF 5mcg/kg d0-recovery Clofarabine 25mg/m2 d1–5 Ara-C 2g/m2 d1–5	50	21/46 (46%)	18	12/18 (67%)	Med OS 8.8mo	0/50 (0%)	[57,127]
FA/FLAG	Fludarabine 30mg/m2 d1–5 Ara-C 2g/m2 d1–5 ±G-CSF 400mcg/m2 d0-recovery	FA: 81 FLAG: 20	FA: 22/81 (27%) FLAG: 4/20 (20%)	FA: 20 FLAG: 3	FA: 2/20 (10%) FLAG: 0/3 (0%)	FA: Med OS 3.4mo FLAG: Med OS 3.8mo	FA: 18/81 (22%) FLAG: 4/20 (20%)	[57]
FLAG-Ida	Fludarabine 25mg/m2 d1–5 Ara-C 2g/m2 d1–5 G-CSF d6-recovery	34	15/34 (53.6%)	11	7/11 (62.5%)	Med OS 22wks	6/34 (17.6%)	[128]
CLAG-M	Cladribine 5mg/m2 d1–5 Ara-C 2g/m2 d1–5 G-CSF 300mcg d0–5 MIT 10mg/m2 d1–3	118	66/118 (58%)	75	16/75 (21%)	4y OS 14% 4y DFS 30%	8/118 (7%)	[36]

Abbreviations: CR, complete remission; OS, overall survival; PRD, primary refractory disease; TRM, treatment-related mortality; Ara-C, cytarabine; MIT, mitoxantrone; GO, gemtuzumab ozogamicin; ATRA, all-trans retinoic acid; G-CSF, granulocyte-colony stimulating factor

Unfortunately, the results of recent prospective randomized clinical trials that attempted to improve on these outcomes have been largely negative. The VALOR study, which was a phase III double-blind study of high-dose cytarabine alone or in combination with the quinolone derivative vosaroxin in patients with relapsed and refractory AML, found no improvement in OS among patients that received vosaroxin compared to placebo[60]. Likewise, although there has been some enthusiasm for the use of clofarabine in the refractory setting, the CLASSIC I trial showed no difference in median OS for intermediate-dose cytarabine alone or in combination with a single dose of clofarabine in patients ≥ 55 years old with relapsed and refractory AML[61]. An international phase III study of 381 patients with relapsed or refractory AML who were treated with elacytarabine versus investigator choice of 7 different salvage regimens (HiDAC, MEC, FLAG/FLAG-Ida, LDAC, HMA, hydroxyurea, or best supportive care) found no difference in median OS in the elacytarabine arm compared to the control arm (3.5 months vs. 3.3 months, p=0.96); there was also no difference in the CR rate in the elacytarabine arm compared to the control arm (23% vs. 21%)[62]. Importantly, a subgroup analysis also revealed no significant differences in outcome among any of the treatment options in the investigator choice arm[62]. While no subgroup analysis for PRD was included, the overall short survivals with all salvage arms makes it unlikely that a particular approach was associated with markedly better outcomes.

Hypomethylating agents

The HMAs azacitidine and decitabine are frequently used to treat patients with PRD, especially those patients who are older or less fit or who are also refractory to intensive salvage chemotherapy. Although most prospective randomized studies of HMAs have been conducted in the frontline setting[63–65], several smaller studies have also suggested that patients with PRD benefit from treatment with HMAs. A retrospective analysis of 47 patients with relapsed or refractory AML who were treated with azacitidine found that 21% achieved a CR; median OS was 9 months[66]. Another study of 130 patients over the age of 50 years with refractory or relapsed AML who received azaciditine as part of a “compassionate use” program in France found similar outcomes with a CR/CRi rate of 17% and median OS of 8.4 months[67]. Although the reported CR rates with HMAs in the relapsed/refractory settings are widely variable (3.6%−21%), partial responses (PR) or stabilization of disease may also provide a clinical benefit or improvement in quality of life[59]. HMAs also have the benefit of significantly less toxicity compared with intensive salvage regimens and are typically given in the outpatient setting. Available data make a compelling argument for their use, especially in the context of patients not expected to bridge to transplant.

Allogeneic hematopoietic stem cell transplant

Allogeneic HSCT is critically important for fit patients with primary refractory AML as salvage chemotherapy alone is not sufficient for long-term disease control. A possible exception are the rare PRD patients with core-binding factor AML who potentially can experience long-term disease control with HiDAC-based salvage and post-remission therapy. A study of 150 patients with PRD who were treated on a recent SWOG study (S0106) found that the 4-year OS rate was only 4% for the 86 patients who did not undergo allogeneic HSCT compared to 48% for those subjects who underwent transplantation[68]. A similar study of patients with PRD who were treated at the M.D. Anderson Cancer Center found that the median OS of patients who received salvage chemotherapy alone was 2.9 months and 3-year OS was only 2%, compared to 39% for patients who underwent an up-front allogeneic HSCT (p<0.001)[69]. Thus, all patients with PRD who are eligible for allogeneic HSCT should be referred to a transplant center with expeditious completion of HLA typing and initiation of a donor search.

Consideration for immediate allogeneic HSCT is reasonable for eligible patients with PRD, especially if a well-matched donor has been identified, as reported 3-year OS rates for patients who undergo allogeneic HSCT with active AML are in the range of 14–39%[58,69–76]. A review of registry data from the Center for International Blood and Marrow Transplant Research (CIBMTR) from 1440 subjects with PRD specifically who underwent allogeneic HSCT found a 5 year OS rate of 21% (95% CI, 19–23%)[76]. The preferred preparative regimen for patients with active AML at the time of transplantation varies by center. At our institution, we are currently using a regimen containing clofarabine and busulfan based on promising results reported in several small studies. Among 71 subjects included in a prospective phase II study of patients with relapsed and/or refractory AML who underwent myeloablative allogeneic HSCT with clofarabine and busulfan conditioning, the 2-year OS rate was 26% while the non-relapse mortality at 2 years was 25%[73]. Of note, the 2-year event-free survival (EFS) was significantly better among the subgroup of patients with PRD compared to those with relapsed AML (2-year EFS 34% vs 8%, p<0.01) and there was also a trend toward improved OS in this group (2-year OS 34% vs. 24%, p=0.09)[73].

Because a number of studies have found that a lower bone marrow blast percentage prior to allogeneic HSCT is associated with significantly improved outcomes after transplantation[71,72,77–81], at our institution we typically offer salvage therapy to patients with PRD prior to proceeding with allogeneic HSCT in an attempt to maximally decrease the burden of disease prior to transplant. It should be noted, however, that because CR rates to salvage are low, salvage may largely select for a lower risk transplant population. Although a randomized trial to determine whether salvage improves overall survival as compared to direct allogeneic transplant is logistically challenging, indeed this question remains unanswered. The SIERRA trial (NCT02665065) will address this question by randomizing older patients with either PRD or relapsed/refractory AML to salvage chemotherapy followed by conventional allogeneic transplant vs. immediate reduced intensity transplant using a novel radioimmunotherapeutic preparative regimen.

Several scoring systems have been developed to predict outcomes for patients who undergo allogeneic HSCT with active AML[70,74], including the model of Todisco et al. developed specifically for patients with PRD. These authors found that age > 60, ≥ 25% bone marrow blasts at the time of transplant, > 2 prior cycles of chemotherapy, and intermediate-2 or adverse cytogenetics by ELN criteria were independently associated with an increased risk of death after transplant[74]. Subjects with 0–1 risk factors had a 3-year OS rate of 32%, while those with 2 risk factors had a 3-year OS of 10% and those with 3–4 risk factors had a 3-year OS of only 3%[74]. Overall, allogeneic HSCT with active AML is feasible and should be considered in select patients. The use of predictive scoring systems may help identify those patients who are most likely to benefit from transplantation with active disease[70,74].

NEW AND EMERGING THERAPIES

Given the poor prognosis of patients with PRD, we encourage consideration of enrollment on a clinical trial whenever feasible, especially those who have either failed or are not candidates for salvage chemotherapy and allogeneic HSCT. Molecular testing by PCR and/or next-generation sequencing (NGS) techniques should be performed for all patients with PRD in order to evaluate for the presence of specific mutations for which oral small molecule inhibitors have been developed, namely FLT3, IDH1, and IDH2 mutations. Several new AML therapies that have either recently been approved or are currently being investigated in clinical trials are discussed here.

FLT3 inhibitors

FLT3-ITD mutations occur in approximately 23% of adult AML, while mutations in the tyrosine kinase domain (TKD), most commonly FLT3-D835, are found in about 7%[82,83]. Both types of mutations cause constitutive kinase activation [84,85], but FLT3-ITD mutations in particular are associated with a significantly worse prognosis due to relatively high rates of relapse[86,87]. Additionally, the presence of a high FLT3-ITD:WT allelic ratio or certain ITD insertion regions (e.g. Beta-1 sheet of tyrosine kinase 1 domain) predict higher rates of PRD[83,88]. Because refractory and/or relapsed FLT3-ITD+ AML seldom responds durably to salvage chemotherapy[89], we recommend that that these patients be referred for evaluation for enrollment on a clinical trial of a FLT3 inhibitor when possible.

A number of FLT3 inhibitors have been investigated, including the multi-kinase inhibitors sorafenib, lestaurtinib, and midostaurin as well as the more potent and FLT3-selective inhibitors crenolanib, gilteritinib (ASP2215), and quizartinib (AC220). A phase I/II study of sorafenib in combination with idarubicin and intermediate-dose cytarabine, which primarily included patients with newly diagnosed AML, found that 18/19 subjects (95%) with FLT3-ITD mutations achieved CR[90,91]. Although this suggested that the addition of sorafenib could potentially decrease the incidence of PRD in patients with FLT3-ITD mutations, subsequent randomized trials that evaluated sorafenib versus placebo in combination with standard “7+3” found that sorafenib increased toxicity but did not improve OS[92,93]. By contrast, the addition of midostaurin to standard induction and consolidation chemotherapy for newly-diagnosed FLT3-mutated patients under age 60 did not increase toxicity and was associated with a statistically significant improvement in OS compared to placebo (51.4% vs. 44.3 at 4 years, respectively; HR 0.78, p=0.009)[94]. A randomized phase III study to compare crenolanib versus midostaurin when given in combination with standard induction chemotherapy (NCT03258931) will be underway soon.

While none of the multi-kinase inhibitors has had substantial, durable activity in relapsed/refractory patients, the clinical activity of the more potent, selective inhibitors have been promising, even as single agents. Among 169 patients with relapsed and/or refractory FLT3-mutated AML who received gilteritinib on a phase I/II dose-escalation study at a dose of at least 80 mg daily, 43% had elimination of all circulating or extramedullary blasts and reduction in marrow blasts to < 5%, with variable peripheral recovery and very modest toxicity[95]. The median duration of response to gilteritinib at these doses was 20 weeks with a median OS of 31 weeks (range 1.7–61 weeks)[95]. Pivotal phase III randomized trials comparing gilteritinib (NCT02421939) and quizartinib (QuANTUM-R; NCT02039726) to salvage chemotherapy for relapsed and refractory FLT3-mutated AML have been conducted. The QuANTUM-R study demonstrated an improvement in OS with quizartinib monotherapy compared to salvage chemotherapy in patients with relapsed/refractory FLT3-ITD-mutated AML (HR 0.76, 95% CI 0.58–0.98). U.S. Food and Drug Administration (FDA) review of gilteritinib and quizartinib is either underway or expected soon. FLT3 inhibitors therefore appear likely to soon become the standard of care for patients with PRD who have a FLT3 mutations. Studies combining FLT3 inhibitors with other active agents for relapsed/refractory AML may improve upon these responses and survival, and such studies are either underway or planned[96,97].

Currently, if a patient with refractory FLT3-ITD+ AML is unable to enroll on a clinical trial evaluating a selective FLT3 inhibitor, a reasonable salvage option is the combination of azacitidine and sorafenib[98]. A phase II study of this combination in 43 patients (93% of whom had FLT3-ITD mutations) found a CR/CRi rate of 43% (16/43 subjects) with a median duration of response of 2.3 months (range 1–14.3 months) and median OS of 6.2 months[98]. Of note, although transplant can be performed in substantial numbers of relapsed and refractory patients following response to FLT3 inhibitors, post-transplant relapse rates remain high[99]. Therefore, when feasible, we recommend restarting the FLT3 inhibitor following donor engraftment.

Ivosidenib and enasidenib

Somatic point mutations in isocitrate dehydrogenase (IDH) 1 or IDH2 occur relatively frequently in AML and result in gain-of-function enzymatic activity leading to the conversion of alpha-ketoglutarate to a new metabolite, (R)-2-hydroxyglutarate (2-HG)[100] [101,102]. 2-HG accumulates in cells, leading to a variety of metabolic and epigenetic changes that result in a block in cellular differentiation and promote leukemogenesis[103–105]. Several oral inhibitors of the mutant IDH enzymes have been developed, including ivosidenib which selectively inhibits mutant IDH1 and enasidenib which targets mutant IDH2. In patients with relapsed and/or refractory AML with IDH2 mutations, a phase I/II study of enasidenib found that 40.3% (71/176) of subjects had an objective response including 19.3% (34/176) who achieved a CR, with a median OS of 9.3 months[106]. Likewise, a phase I study of ivosidenib found a CR/CRi rate of 30.4% (38/125) with a median duration of response of 8.2 months in patients with relapsed and/or refractory IDH1-mutant AML who received a dose of 500mg daily[107]. Of note, grade 3–4 IDH-inhibitor-associated differentiation syndrome was observed in 3–5% of patients in both studies, and managed with corticosteroids and/or temporary holding of the drug in some instances[106,107]. Based on these studies, the FDA approved enasidenib in August 2017 and is currently reviewing a new drug application for ivosidenib for the treatment of relapsed and/or refractory AML with mutations in IDH2 and IDH1, respectively.

CPX-351

CPX-351 is a novel formulation of cytarabine and daunorubicin in a liposomal carrier at a fixed 5:1 molar ratio, which in vitro studies have suggested generates maximal synergy of the two drugs[108]. CPX-351 was recently approved by the FDA for the treatment of newly diagnosed AML that is either therapy-related or secondary to prior myelodysplastic syndrome (MDS). This approval was based on the results of a phase III randomized controlled trial of CPX-351 versus standard “7+3” in 309 older patients (ages 60–75) with previously untreated secondary AML. On this trial, patients treated with CPX-351 had a superior CR/CRi rate (47.7% vs 33.3%, p=0.016) and an improved median OS (9.56 vs. 5.95 months, p=0.005)[109]. Although this frontline study is promising, currently there are insufficient data to recommend the use of CPX-351 over other cytotoxic regimens in the context of PRD.

Venetoclax

Venetoclax is an oral small molecule inhibitor of the anti-apoptotic protein BCL-2 and is currently FDA-approved for the treatment of chronic lymphocytic leukemia[110]. Venetoclax has received breakthrough therapy designation from the FDA for the treatment of patients with newly diagnosed AML who are older or ineligible for intensive induction in combination with HMA or LDAC. This was based on promising preliminary results from several studies, including a phase Ib dose-escalation study of venetoclax in combination with azacitidine or decitabine in patients ≥ 65 years old with previously untreated AML (NCT02203773) which demonstrated a CR/CRi rate of 61% (35/57)[111,112]. Randomized phase III studies are ongoing (NCT02993523 and NCT03069352). If these high response rates are confirmed then it is possible that the use of venetoclax could lead to a reduction in induction failure rates in the elderly population that is at high risk for PRD. Of note, the activity of venetoclax in the relapsed and/or refractory setting appears to be much more modest, with objective response rates around 20%, although the available data is limited[113,114]. Thus, the role of venetoclax in the setting of disease that is refractory to induction chemotherapy remains to be defined.

Gemtuzumab ozogamicin

Although the interesting history of gemtuzumab ozogamicin (GO), an antibody-drug conjugate that consists of an antibody targeting CD33 linked to a calicheamicin antibiotic, has been reviewed elsewhere[115], GO is now back on the market after being granted full FDA approval in September 2017 for both newly diagnosed and relapsed and/or refractory, CD33-positive AML. The approval for relapsed and/or refractory disease is based on the MyloFrance-1 study which included patients with AML in first relapse only who were treated with GO 3mg/m2 on days 1, 4, and 7. A CR or CRi was achieved in 33% subjects and median relapse-free survival was 11 months[116]. On this study, patients were required to have a prior first remission of at least 3 months, as patients with chemorefractory AML seldom respond to GO monotherapy. Still, combination salvage regimens that contain GO might show better outcomes and should be studied in PRD[58].

Novel immunotherapeutic approaches

Immune checkpoint inhibitors targeting PD-1 and CTLA-4 have revolutionized the treatment of many solid malignancies, but have thus far appeared less promising as monotherapy for relapsed and refractory AML. However, both nivolumab and ipilimumab can occasionally induce remissions in patients with AML who relapse after allogeneic HSCT, and ongoing trials are evaluating the role of checkpoint blockade in combination with chemotherapy (NCT02768792) and HMAs (NCT02397720)[117–119]. The role of checkpoint inhibitors in combination therapy for PRD specifically remains to be evaluated.

Alternative antibody-based immunotherapy approaches may be more promising for AML, including flotetuzumab, which is a humanized dual-affinity molecule that recognizes and redirects T cells to target cells expressing CD123, the alpha chain of the interleukin-3 (IL-3) receptor[120]. CD123 is expressed by a majority of AML cells as well as normal immature hematopoietic cells [121]. A preliminary analysis of a phase I dose-escalation study in patients with relapsed and/or refractory AML and MDS demonstrated that flotetuzumab has an acceptable tolerability and safety profile[120]. The most common adverse events were infusion reactions and cytokine release syndrome (CRS). Of 14 subjects who received a dose of at least 500 ng/kg/day and had available response data, the ORR was 43% (6/14 subjects), with a CR/CRi rate of 28% (4/14 subjects)[120].

Tagraxofusp (SL-401) is another biologic agent targeting CD123 that consists of diphtheria toxin conjugated to interleukin 3 (IL-3)[122]. Preliminary data from 17 subjects with AML and blastic plasmacytoid dendritic cell neoplasm (BPDCN) enrolled in the lead-in phase of a study of tagraxofusp demonstrated feasibility and safety, although two BPDCN patients experienced capillary leak syndrome as a dose-limiting toxicity (DLT)[122]. No dose-limiting toxicity was identified for the AML patients, however, and dose escalation is still on-going (NCT02113982)[122]. Another study is evaluating tagraxofusp in combination with azacitidine for relapsed and refractory AML (NCT03113643).

With regard to cellular therapies, the past year has seen tremendous advances for the treatment of B-cell malignancies, with the FDA approval of tisagenlecleucel (Kymriah) and axicabtagene ciloleucel (Yescarta) for acute lymphoblastic leukemia and large B-cell lymphomas. Chimeric antigen receptor (CAR) T cell therapy for AML, however, has lagged behind, which is at least partially related to the lack of a truly AML-specific target, as most AML surface molecules are also expressed by normal hematopoietic stem and progenitor cells (HSPCs) and/or myeloid progenitor cells[123]. Budde et al. recently presented preliminary safety and feasibility results from a first-in-human trial of CD123-specific CAR T cells in patients with AML and BPDCN [124]. Surprisingly, myeloablation was not observed[124]. Of the 6 subjects in the AML cohort, all of whom had relapsed and refractory disease after allogeneic HSCT, 2 subjects had a CR or CRi, 1 subject achieved a morphologic leukemia-free state, and 2 subjects had stable disease[124]. Although these data are preliminary, they illustrate the transformative potential of cellular therapies for PRD.

APPROACH TO PRIMARY REFRACTORY AML

All patients with PRD who are candidates for allogeneic HSCT should be referred to a transplant center and undergo expedited HLA typing and donor identification, as HSCT represents the best chance for long-term disease-free survival[125]. A decision for direct allogeneic versus salvage chemotherapy as an interim step should be made expeditiously so as to not lose a window for immediate transplant, should this be a possibility. All patients should undergo molecular testing via PCR and/or NGS, which not only provides important prognostic information but might also reveal the presence of a targetable mutation in FLT3, IDH1, or IDH2.

For fit patients who are not candidates for a molecularly targeted agent, the decision of whether to attempt to achieve a CR with salvage chemotherapy prior to proceeding with allogeneic HSCT is not straightforward; risks and benefits of both options need to be weighed carefully. Because a lower bone marrow blast percentage has been associated with improved outcomes following allogeneic HSCT and because pre-transplant testing and donor identification takes time, we typically offer patients salvage chemotherapy. If the patient has not previously received high-dose cytarabine, then a combination regimen such as MEC is reasonable. For patients who still have persistent disease following salvage chemotherapy, the likelihood of achieving a CR with additional salvage chemotherapy is low[126]. In this case, proceeding directly to an allogeneic HSCT if a donor has been identified or enrolling the patient on a clinical trial is ideal.

For patients with PRD who are elderly or have a poor performance status following induction chemotherapy, we typically treat with azacitidine or decitabine if the patient is not enrolled on a trial or otherwise eligible for targeted agents. We are also more likely to consider a HMA in patients with mutations in TP53[51]. If a response is obtained then we continue the HMA indefinitely as long as the patient has a continued clinical benefit or, if the performance status has improved, may consider proceeding to a RIC allogeneic HSCT, particularly if the patient has entered CR or CRi. While our institution does not have a defined maximum age for allogeneic transplantation, successful transplant outcomes in patients over 70 years old are uncommon and require very careful patient selection. Frail patients with PRD and those with poor performance status may benefit more from supportive/palliative therapy and/or low-dose Ara-C or HMA than transplant.

Our current approach to the treatment of primary refractory AML is summarized in Figure 1.

Figure 1. Approach to management of primary refractory AML.

Our approach to primary refractory AML takes into account patient eligibility for allogeneic HSCT and salvage chemotherapy, clinical trial availability, and molecular mutation status.

Abbreviations: AML, acute myeloid leukemia; HLA, human leukocyte antigen; FLT3-ITD, fms-like tyrosine kinase 3-internal tandem duplication; HSCT, hematopoietic stem cell transplantation; HMA, hypomethylating agent; LDAC, low-dose cytarabine

CONCLUSION

Primary refractory AML represents a continued challenge in clinical management. Risk factors for PRD include an advanced age, elevated WBC count at diagnosis, complex or monosomal karyotype, the presence of a TP53 mutation, inv(3q), and an increased time to blast clearance after the initiation of induction chemotherapy. Treatment options for PRD include salvage chemotherapy, allogeneic HSCT, non-intensive therapies such as a HMA or LDAC, molecularly targeted agents, and enrollment on clinical trials. Allogeneic transplantation represents the best chance for long-term survival, either immediately after declaring refractoriness to induction or after salvage therapy.

Table 1.

Risk factors for primary refractory AML.

Age > 60 years

Cytogenetics

Adverse risk karyotype

inv(3)(q21q26)/t(3;3)(q21;q26)

Monosomal karyotype

TP53 mutation

High WBC count at diagnosis

Secondary AML High allelic ratio FLT3-ITD:WT ratio

Blast clearance

Nadir marrow biopsy results

Time to peripheral blast clearance

Poor performance status

Abbreviations: TP53, tumor protein p53; WBC, white blood cell; AML, acute myeloid leukemia; FLT3-ITD, fms-like tyrosine kinase 3-internal tandem duplication; WT, wild-type