A review of treatment options employed in relapsed/refractory AML

ABSTRACT Introduction: Acute myeloid leukemia [AML] is a heterogenous group of primary hematopoietic neoplasms arising from myeloid precursor cells. Up to 50% of patients failed to achieve remission with initial therapy and go on to develop refractory AML. Whenever possible, enrollment in a clinical trial in view of the paucity of evidence surrounding a clearly superior treatment modality is recommended, and the therapy which provides the best chance for cure post remission is allogeneic hematopoietic stem cell transplantation [HCT], with much of everyday clinical decision-making in relapsed/refractory (R/R) AML surrounding the choice of the least toxic regimen that could achieve remission and enable prompt HCT. Discussion: We discuss a variety of treatment modalities employed in the R/R AML setting beginning with traditional cytotoxic regimens. We then turn our attention to targeted therapies that have shown efficacy in specific patient populations such as the IDH inhibitors and FLT3 inhibitors and lastly, we turn our attention to immunotherapeutic agents employed in the R/R in the setting, such as CD33 inhibitors and bispecific antibodies. Conclusion: It appears increasingly clear that approaching AML as a homogenous disease entity is unsatisfactory in view of the variations in such disease factors as cytogenetic and molecular markers, age, and disease severity at presentation; all of which contribute significantly to heterogeneity of the disease. Moving forward, treating AML would likely require tailored therapy following advances in technology such as molecular profiling, drug sensitivity and resistance testing.


Introduction
Acute myeloid leukemia (AML) is a heterogeneous group of primary hematopoietic neoplasms arising primarily from cells committed to the myeloid line of cellular development.
AML is diagnosed with a blast threshold ≥ 20% in the bone marrow or blood or >10% in the presence of genetic abnormalities that define specific AML subtypes, (AML with BCR: ABL1 is a notable exception that still requires a ≥20% cutoff) [1].Risk stratification of patients diagnosed with AML takes into account numerous disease factors such as the presence/ absence of adverse cytogenetic features, germline predisposition, prior exposure history to cytotoxic agents or radiotherapy, and prior history of myelodysplasia or myeloproliferative neoplasm amongst others.Patient factors such as age and poor performance status must also be considered and are considered the strongest clinical predictors of early death [2,3].
The goal of therapy for AML would be eradication of the disease, if possible, and is accomplished by inducing a complete remission (CR) with initial therapy followed by consolidation and/or maintenance therapy to maximize treatment response.The response to treatment and overall prognosis is variable, dependent on several patient and tumor specific factors such as age, performance status and karyotype.10-40% of patients with newly diagnosed AML will fail to attain a complete remission (CR) with intensive chemotherapy [4].Additionally, up to 50% of patients who initially achieve a CR subsequently develop relapsed AML [5].These patients generally have a poor prognosis.Important criteria used to evaluate treatment responses along with treatment failure criteria and their definitions are summarized in Table 1.
The use of techniques such as multiparametric flow cytometry and real-time quantitative PCR have improved detection of measurable residual disease, with numerous studies showing their prognostic value for relapse and overall survival (OS) [6,7].
Although there are no validated criteria to consider a patient unfit for intensive chemotherapy, the choice of therapy is dictated by many of the same principles that govern risk stratification.The presence/absence of target mutations (e.g.FLT3, IDH1, IDH2) are important considerations in determining the proper choice of regimen.Patients considered fit for intensive therapy are managed with comparatively more aggressive induction regimens that include anthracyclines and cytarabine.Patients considered unsuitable for intensive therapy on the other hand are managed with lower intensity regimens incorporating hypomethylating agents (HMAs) with venetoclax or low dose cytarabine.It is important to note that molecular re-evaluation is necessary at relapse to identify the emergence of actionable targets owing to the clonal evolution of AML at this stage as these may be suitable for targeted salvage therapies [8,9].
Whenever possible, it is recommended to enroll patients with AML in a clinical trial.AML remains the most frequent indication for allogeneic hematopoietic stem cell transplantation (HCT), and outcomes posttransplantation in AML continue to improve with newer methods to prevent and treat complications such as graft-versus-host disease (GVHD) and infections.Most patients do not undergo the procedure owing to factors such as lack of access, patient/physician choice and medical comorbidities and consequently, the role of HCT in clinical practice is primarily for patients with poor risk AML [10,11].In general, following cytoreduction in relapsed/refractory AML (r/r AML), allogeneic HCT is recommended, although low-intensity maintenance regimens may be favored in older patient populations.
A wide variety of different treatment modalities have been studied to improve outcomes in patients with r/r AML.We will review traditional chemotherapeutic agents as well as targeted single agents and immunotherapeutic agents.

Traditional cytotoxic therapy
Many of these agents are still used in combination with targeted therapies for remission-reinduction in r/r AML.A review of the mechanism of action of these agents and notable adverse events is highlighted in Table 2.
The salvage regimens discussed below may be considered as induction therapy in r/r AML.A subset of patients may benefit from direct allogeneic HCT without salvage therapy and is an area of investigation [23].By and large anthracyclines and high dose cytarabine form the backbone of these regimens.Unfortunately, none of these regimens have clearly demonstrated superiority over the others, and if possible, enrollment in a clinical trial should always take precedence.In general, salvage regimens attempt to incorporate agents that were not used during initial induction cycles.Current directions are focused on combining traditional cytotoxic regimens with novel targeted agents.
HiDAC: In general, patients may receive the HiDAC regimen (high dose cytarabine) in the salvage setting if they have not received it previously for induction either with or without an anthracycline.An exception can be made in patients who suffer a late relapse more than 18 months from their initial CR, as they may achieve a second CR by retreatment with HiDAC even if initially treated with the same.Reported CR rates are 32-47% in this setting [24,25].Interestingly, the combination of HiDAC plus mitoxantrone may produce higher response rates with side effects appearing to be similar in both regimens, although toxicity is prohibitively high in patients over the age of 60 [24].Unlike HiDAC monotherapy, reinduction with cytarabine plus anthracycline may be employed for those who receive it initially and subsequently achieve a CR that persists for more than 1 year.CRs in this setting have been reported at approximately 50% [25].HiDAC is also commonly employed for consolidation therapy following induction, although recent evidence has challenged the need for exceptionally high dose levels owing to greater toxicity in favor of intermediate dose cytarabine [26].Regimens employing high dose cytarabine are dangerously toxic in elderly populations, owing to compromised organ function and comorbidities in these populations [27].
FLAG/FLAG-IDA: Fludarabine is most often employed as part of the FLA/FLAG regimens.Idarubicin may also be added in the widely used FLAG -IDA regimen.with reported CR rates of around 50% [28].FLAG-IDA is a frequently used control arm in clinical trials for r/r AML.In at least one study no significant difference in CR rates were observed between FLAG and FLAG-IDA [29].Fludarabine has been shown to augment the rate of synthesis of cytarabine in blast cells when administered prior to cytarabine [13].
Despite the addition of G-CSF to FLA, reported CR rates did not differ significantly, being 61% and 58% respectively between FLA and FLAG [30].The addition of idarubicin may be important however, as FLA alone was inferior to cytarabine, daunorubicin and etoposide reinduction [30].MEC: Mitoxantrone with etoposide is commonly used for salvage therapy, with CR rates of approximately 40% being reported [31].The MEC regimen (mitoxantrone, etoposide, cytarabine) has been studied with CR rates of 66% being reported [32].Mitoxantrone has also been studied in combination with high and intermediate dose cytarabine (HAM regimen), and patients over 60 years achieved a CR rate of 76% with this regimen [33].
FLAD: The FLAD regimen (Fludarabine, cytarabine and liposomal daunorubicin) achieved an overall CR rate of 53% in one study, and as 58% of these patients went on to undergo allogeneic HSCT, this regimen may have utility as a bridge to transplant [34].
FLAM: The FLAM regimen (Flavopiridol, cytarabine and mitoxantrone) has demonstrated efficacy as compared to carboplatin-topotecan and sirolimus-MEC per a phase II study, with a reported CR rate of 28% vs 14% and 15% [35].
ADE: The ADE regimen (Cytarabine, daunorubicin, etoposide) may be administered together as part of a standard regimen or sequentially.The standard administration has been shown to have improved CR rates (54% vs 34%) with improved 3-year survival (12% vs 6%) and may therefore be superior [30].
Clofarabine + IDAC: Clofarabine may be administered with intermediate dose cytarabine, with CR rates of 35% to 51% being reported, as evidenced by at least one study [15].
Etoposide + cyclophosphamide: High dose etoposide has been evaluated to have demonstrable efficacy in Table 2. Mechanisms of action and notable adverse effects of cytotoxic agents employed in the management of r/r AML.

Agent
Mechanism of action Notable adverse effects Cytarabine (Ara-C) Pyrimidine analog that is converted to its active compound, azacitidine triphosphate, following which it is incorporated into DNA.Its primary action is inhibition of DNA polymerase resulting in decreased synthesis and repair of DNA and is specific to the S phase of the cell cycle [12].

Fludarabine
Inhibits DNA synthesis through inhibition of DNA polymerase and ribonucleotide reductase, as well as inhibition of DNA primase and DNA ligase I [13] Myelosuppression, neurotoxicity, hemolytic anemia, Evan's syndrome [14].

Clofarabine
Ribonucleotide reductase inhibition, inhibits the DNA repair process and alters mitochondrial membrane permeability releasing inducing factor and cytochrome C [15].

Cladribine
Cell cycle nonspecific.Incorporates into DNA and results in strand breakage and shutdown of DNA synthesis and repair [16].
Daunorubicin, idarubicin Inhibits DNA and RNA synthesis by intercalation between DNA base pairs and causing steric hindrance of the local uncoiling of the double helix [17].

Mitoxantrone
Structurally related anthracenedione that functions similarly to anthracyclines to cause DNA strand cross linkage, strand breaks and nucleic acid synthesis inhibition through steric hindrance [17].
Is preferred over other anthracyclines due to reduced incidence of cardiotoxicity and GI toxicities [17].

Aclarubicin
Inserts its trisaccharide chain into the minor DNA groove to poison the action of topoisomerase I and II.It has also been shown to evict histones from nucleosomes and is also an inhibitor of the 20S proteasome [18].

Etoposide
Type II topoisomerase inhibitor.Appears to cause DNA strand breakage, with subsequent cell cycle arrest in the S phase [19].

Vosaroxin
First in class quinolone derivative that induces replicationdependent DNA damage by intercalating into DNA and blocking topoisomerase II-mediated re-ligation, ultimately leading to apoptosis [20] Nephrotoxicity, myelosuppression, low potential for drugdrug interactions as it is minimally metabolized [20] Omacetaxine mepesuccinate (Homoharringtonine) A plant alkaloid derived from Cephalotaxus fortune.It is a protein translation inhibitor and inhibits correct positioning of amino acids during the initial step of protein translation [21].
Flavopiridol (alvocidib) Synthetic analog of a naturally occurring flavonoid alkaloid isolated from Dysoxylum binectariferum tree bark.It has properties as a cyclin-dependent kinase inhibitor and leads to inhibition of positive transcription elongation factor b [22].
conjunction with cyclophosphamide, with a reported CR rate of up to 42% [39].Mucosal toxicity was found to be dose limiting.
Elacytarabine: A relatively newer agent, elacytarabine, which is an acid ester of cytarabine was studied in a randomized phase III study of patients with r/r AML, and no significant differences in CR or OS rates were found between treatment and control arms which included other commonly employed cytotoxic regimens [40].
Vosaroxin + cytarabine: Aside from being investigated as a therapeutic target for primary AML [20], vosaroxin has been found to improve CR rates in combination with cytarabine in r/r AML as compared to cytarabine with placebo (30% vs 16% respectively) in the VALOR trial, although treatment-related adverse events and deaths were also higher in the treatment groups [41].
CPX-351: CPX-351 is a liposomal formation of cytarabine and daunorubicin in a 5:1 molar ratio.Although no statistical difference in 1-year survival between CPX-351 and standard salvage chemotherapies was found, CPX-351 excited attention as it was found to increase CR rates in subgroups with poor risk features including the elderly and patients with adverse cytogenetics compared to standard chemotherapy (47.7% and 33.3% respectively) [42], although this study did not specifically evaluate the drug in the r/r AML setting.
5+2: The traditional 7+3 regimen of 7 days of cytarabine continuous infusion with IV daunorubicin on days 1-3 is one of the oldest standardized induction regimens for de novo AML.The 5+2 regimen utilizes the same agents with dose reduction given over a shorter time frame (5 days' continuous infusion of cytarabine with IV daunorubicin on days 1 and 2).The non-randomized portion of a phase II study evaluated the 5+2 regimen as re-induction after patients underwent initial treatment with the 7+3 regimen and were found to have residual leukemia.6 of these 13 patients (46%) studied achieved CR [43].
LDAC: LDAC (Low dose cytarabine) has been evaluated in the setting of relapsed AML with results from a retrospective analysis of older patients with first relapse AML showing a CR rate of 17% and a median post-relapse survival of 5.6 months, which was not much better than for patients with best supportive care (3.2 months) [44].LDAC may have utility as maintenance therapy after remission induction in r/r AML however, as evidenced by a study which showed a median disease-free survival of 7.4 months in patients treated with repeated courses of LDAC maintenance therapy compared to observation alone [45].
A table highlighting some of the chemotherapeutic regimens discussed has been provided for ease of reference (Table 3).

Targeted therapies
A deeper understanding of the biology of AML has led to the identification of deregulated pathways that drive the proliferation of blasts.In turn, these discoveries have led to the development of specific agents that target these molecular pathways and have revolutionized the care of patients with AML.
In patients with the requisite mutations, these agents can be considered for induction therapy in combination with intensive cytotoxic regimens, as well as for consolidation and/or maintenance therapy e.g.midostaurin in FLT3 mutated AML.A review of the mechanism of action of these agents as well notable adverse effects is provided in Table 4.

BCL2 inhibitors: venetoclax
In patients considered unfit for intensive chemotherapy, substantial progress has been made with the addition of venetoclax to the HMA azacitidine.Venetoclax with azacitidine is considered a new standard of care in the management of older patients or those considered unfit for more intensive therapies.As compared to azacitidine alone, the addition of venetoclax improved CR/CRi rates (66.4% vs 28.3%) and median OS (14.7 vs 9.6 months) [56].
The PETHEMA registry experience evaluated the effectiveness of venetoclax in the r/r AML setting, which showed higher CR rates for venetoclax in combination with azacitidine than with decitabine or low dose cytarabine (17.9% vs 6.7% and 0% respectively) [57].The potential role of mutations as predictive factors for response to venetoclax-based therapy has been assessed in a cohort of 40 R/R AML patients.The presence of NPM1, RUNX1, or SRSF2 mutations have been associated with higher CR/CRi rates, and patients harboring RUNX1 mutations had a longer OS [58].
Although venetoclax monotherapy has shown impressive results in the treatment of CLL, monotherapy for r/r AML is underwhelming, with an overall CR rate of 19% [59].

Hypomethylating agents: azacitidine, decitabine
Azacitidine has shown efficacy in combination with venetoclax both in de novo as well as r/r AML treatment of patients unfit for intensive chemotherapy.Dose reductions may be necessary if clinically significant bone marrow suppression develops, and the risk of tumor lysis syndrome with this regimen may necessitate prophylactic administration of urate lowering drugs with close electrolyte monitoring [47].
Azacitidine may also be combined with targeted therapies such as IDH inhibitors in patients with the requisite mutations.In r/r AML patients fit for intensive therapy, azacitidine monotherapy may be employed as maintenance therapy after consolidation, although data regarding the role of oral azacitidine maintenance therapy in younger patients (< 55 y) or patients with core-binding factor AML are lacking [1].In addition, data is lacking for oral azacitidine after GO-based or CPX-351 induction/consolidation therapy [1].
An outcomes review of r/r AML patients treated with azacitidine at three different French institutions showed a CR rate of 21%, with a bone marrow blast percentage less than 20% being identified as the only independent prognostic factor irrespective of age or performance status [60].A study of decitabine administration in a cohort of 102 patients with r/r AML showed a CR rate of 15.7% with a median overall survival of 177 days [61].Myelosuppression, nephrotoxicity and hepatotoxicity (azacitidine) [47] Ivosidenib Inhibitor of the mutant isocitrate dehydrogenase 1 [IDH1] enzyme which results in reduced blast counts and induces myeloid differentiation.IDH1 mutations in AML lead to increased levels of 2-hydroxyglutarate in cells, which inhibits alpha-ketoglutarate-dependent enzymes that subsequently results in impaired hematopoietic differentiation.[48].

Enasidenib
Inhibitor of IDH2 with subsequent reduction in blast count and increased maturation of myeloid cells [49].
Differentiation syndrome, bilirubin elevation, resistance may emerge through development of second site IDH 2 mutations which inhibit drug binding [9].Midostaurin, sorafenib, gilteritinib, quizartinib Inhibitors of FMS-like tyrosine kinase receptor 3 (FLT3).Interact with the ATP-binding site of the intracellular tyrosine kinase domain and competitively inhibit ATP binding, thereby preventing receptor autophosphorylation and activation of downstream signaling [50].

SNDX-5613
Inhibitor of the interaction of menin with an epigenetic entity based on the MLL/KMT2A gene.This leads to downregulation of transcriptional process essential for leukemogenesis [51].
-Vorinostat Inhibition of histone deactylase enzymes HDAC1, HDAC2, HDAC3 and HDAC6, which results in altered chromatin structure and transcription factor activation with subsequent cell growth termination and apoptosis [52].
-Tosedostat M1 aminopeptidase enzyme family inhibitor that functions in a tumor selective manner and causes depletion of cellular amino acid pools, with resultant apoptosis.[53].
Bortezomib Inhibits chymotrypsin-like activity at the 26S proteasome, with subsequent activation of signaling cascades that result in apoptosis of the cell [54].
Sirolimus, everolimus Inhibitors of mammalian target of rapamycin (mTOR), a serine/ threonine kinase involved in the regulation of cell growth and proliferation, which lead to reduced activity of downstream targets that are constitutively activated in AML [55].

GI toxicity, pulmonary toxicity [55]
HEMATOLOGY 3.3.Isocitrate dehydrogenase inhibitors (IDH1, IDH2): ivosidenib, enasidenib IDH1 mutations in AML may have a frequency of up to 33% [62] and as such present an attractive therapeutic target.In contrast to ivosidenib, enasidenib has been noted to work independent of the mutational load of IDH2.
In patients with IDH1 mutated R/R AML, monotherapy with ivosidenib yielded CR rates of 21.8% in a phase 1 dose escalation and expansion trial, with an overall response rate of 41.6% [48].Notably, transfusional independence was attained in 35% of patients who had achieved a response.Similarly, enasidenib as monotherapy for IDH2 mutant R/R AML was shown to achieve an overall response rate of 40.3% with a CR rate of 20.2% in a phase 1/2 dose escalation and expansion trial.19.3% of those who attained CR had an overall survival of 19.7 months [49].
It is noteworthy that the addition of azacitidine to ivosidenib has been shown to improve median OS in newly diagnosed AML patients with IDH1 mutation compared to monotherapy (24 vs 7.9 months) [63].

FMS-like tyrosine kinase receptor 3 (FLT3) inhibitors: midostaurin, sorafenib, gilteritinib, quizartinib
FLT3-ITD and FLT3-TKD oncogene mutations are common in AML with frequencies of up to 25% and 10% respectively [64].First-generation FLT3 inhibitors midostaurin and sorafenib function relatively nonselectively against FLT3, and although midostaurin has been approved as first-line therapy, both have little activity as monotherapy in relapsed patients.The addition of midostaurin for FLT3 mutated AML to daunorubicin-cytarabine induction and high-dose cytarabine consolidation improved 4-year OS by 7.1% [65].Although it may subsequently be continued as maintenance therapy in this situation, its value is uncertain.In contrast, next-generation TKI's which include gilteritinib and quizartinib have shown antileukemic single agent activity [66], and gilteritinib is also FDA approved for r/r FLT3 mutated AML.
The ADMIRAL trial compared outcomes in patients with r/r FLT3 mutated AML who were administered either gilteritinib or standard salvage chemotherapy, with gilteritinib administration resulting in significantly longer survival than chemotherapy [9.3 versus 5.6 months], as well as a high percentage of patients with complete remission compared to salvage chemotherapy [21.1% versus 10.5%] [67].Serious adverse effects were also reported less frequently in the gilteritinib group, the most common of which were cytopenias.Although quizartinib is not FDA approved for r/r FLT3 mutated AML, the QuANTUM-R trial comparing quizartinib and salvage chemotherapy showed a median overall survival benefit with quizartinib of 6.2 months versus 4.7 months with chemotherapy [68].A phase 1 dose escalation and expansion study of quizartinib for r/r FLT3 -ITD mutated AML elicited a complete remission rate of 37.5% in study participants [69].
'Triple therapy' with venetoclax, decitabine and specific FLT3 inhibitors (sorafenib, gilteritinib and midostaurin) was studied in a phase II trial that included a cohort of 13 r/r FLT3 mutated AML patients, with a reported composite CR rate of 62%.The subset of patients that had prior exposure to FLT3 inhibitors reported a similar CR rate of 63% [70].

Menin inhibitors: SNDX-5613
The AUGMENT-101 study was the first in-human evaluation of the menin inhibitor SNDX-5613 and was conducted in r/r AML patients with NMP1 mutations and KMT2A gene rearrangements.Of note, the patients enrolled had failed multiple lines of therapy, including allogeneic HCT [71].The overall response rate was 53%, and a CR rate of 20% was achieved in this study.The major dose-limiting toxicity was QT prolongation in about 10% of the treatment group.Another novel menin inhibitor, KO-539, is being investigated in the KOMET-001 study [72].

Histone deacetylase inhibitors: vorinostat
Vorinostat is a histone deactylase inhibitor approved for cutaneous T-cell lymphoma which, as monotherapy in RR -AML, was found to have a CR rate of only 4.5% in a group of 22 patients [73].It fared better in a trial of combination therapy with cytarabine and etoposide with a CR rate of 46% [74].

Aminopepetidase inhibitors: tosedostat
Tosedostat monotherapy in the r/r AML setting has been evaluated with CR rates of 10% in a phase II study [53].A trend for a higher response rate and survival was observed in patients with prior MDS or who had received hypomethylating agents for previous first induction, as well as patients who had multilineage AML compared to other AML types.

Proteosome inhibitors: bortezomib
An expanded phase I trial of bortezomib + MEC (mitoxantrone, etoposide, cytarabine) in r/r AML patients demonstrated CR/CRi (complete remission with incomplete count recovery) rates of 52% [54].Interestingly, of the five patients with RUNX1 mutations (a mutation associated with unfavorable outcomes and shorter survival times in MDS patients [75]), three (60%) achieved CR/ CRi, suggesting that bortezomib may have possible benefit in this difficult subset of patients.

Mammalian target of rapamycin (mTOR) inhibitors: sirolimus, everolimus
Sirolimus monotherapy in one study was shown to have only a partial response rate of 44% in patients with r/r AML and poor risk AML [76].A phase Ib/II study of everolimus in combination with azacitidine in r/r AML patients showed a CR/CRi rate of 12.5%, with dose-limiting toxicity being observed in only 2 of the 40 patients in the study population [77].In contrast, a phase Ib study of everolimus with conventional 7+3 daunorubicin and cytarabine induction chemotherapy in patients with first relapse AML demonstrated a CR rate of 68%, with <10% toxicity reported mainly involving the GI tract and lungs [51].

Immunotherapies
In contrast to chemotherapy which primarily targets dividing tumor cells or targeted therapies that target specific pathways responsible for blast proliferation, the following immunotherapeutic interventions aim to redirect the host immune response towards eradication of the tumor cells.Some of the agents that have been investigated in the r/r AML setting are outlined below:

CD33 inhibitors: gemtuzumab ozogamicin (GO), SGN-CD33A
CD33 is expressed in over 80% of patients with AML [78], and as such is a promising target for immunotherapy.Gemtuzumab ozogamicin (GO) is a humanized CD33 directed monoclonal antibody-drug conjugate, which is composed of the IgG4 kappa antibody gemtuzumab linked to a cytotoxic antibiotic calicheamicin derivative.GO binds to the CD33 antigen which results in internalization of the antibody-antigen complex and subsequent release of the calicheamicin derivative [78].This derivative binds to DNA causing double-stranded breaks, inducing cell cycle arrest and apoptosis.Potentially fatal hepatic sinusoidal obstruction syndrome has been reported with use of GO.
Although the drug was initially withdrawn over safety concerns with full dosing, a phase II study of patients with CD33 positive first relapse AML who were administered fractional doses of GO demonstrated an excellent safety profile with a CR rate of 26% [78].Of note, remission rates correlated with Pglycoprotein and MRP1 activity.Fractionated GO with standard dose cytarabine achieved even more impressive response rates in one study of late first relapse CD33 positive AML, with CR and CR with incomplete platelet recovery rates of 75% [79].
SGN-CD33A, another CD33 inhibitor, is a monoclonal antibody conjugated to a novel synthetic pyrrolobenzodiazepine dimer, which is a potent DNA cross-linking cytotoxin.A phase 1 interim analysis of a study of patient with r/r AML who received SGN-CD33A as monotherapy achieved a CR rate of 28% [80].Pulmonary embolism and myelosuppression were reported as dose-limiting toxicities.

Bispecific antibodies: flotetuzumab, AMG 330
Bispecific antibodies, a newer approach to immunotherapy, are antibodies which contain two antigen recognition sites that can be specific to two different antigens or two different epitopes of the same antigen.Bi-specific T-cell engager (BiTE) antibodies are bispecific antibodies constructed by linking the scFv fragments of two different monoclonal antibodies with a short flexible peptide linker, thereby allowing simultaneous binding of T-cell surface molecules and tumor cell antigens to promote tumor lysis [81].AMG 330 is one such BiTE directed against CD3+ T cells and CD33+ AML tumor cells with resultant T-cell activation and cytotoxicity.
A phase I dose escalation study of AMG 330 r/r AML patients proved to be well tolerated.Although cytokine release syndrome proved to be the most frequent adverse event (67%), it was reversible and occurred in a dose-dependent manner.17% of patients developed CR or CRi, with 57% of these patients having an adverse cytogenetic risk profile [82].
Dual affinity retargeting proteins (DARTs) are bispecific antibodies formed by the heterodimerization of two Fv fragments (Fv1 contains the heavy chain region of antibody 1 and light chain region of antibody 2, whilst Fv2 contains the heavy chain region of antibody 2 and the light chain region of antibody 1).Like BiTE antibodies, they are bispecific, but have been shown to be more potent and stable than BiTEs [83].
Flotetuzumab is an investigational DART that engages CD3 on T-cells and CD123, an IL3-α receptor subunit expressed in 60-80% of AML patients.A phase II study of flotetuzumab in patients with r/r AML showed a combined CR/CRh (complete remission with partial hematologic recovery) rate of 26.7%, with median overall survival of 10.2 months.The most frequently reported adverse events were infusionrelated reactions and cytokine release syndrome that was largely mitigated with dexamethasone and tocilizumab pretreatment [84].Of note, bone marrow transcriptomic analysis revealed a 10-gene signature that predicted response to flotetuzumab in this subset.

CD47 inhibitors: magrolimab
Magrolimab is a first in class monoclonal antibody against CD47 that interferes with macrophage recognition by the SIRPα receptor, thereby blocking the 'don't eat me' signal that tumor cells use to avoid immune surveillance [85].A phase 1b trial looked at 29 patients with r/r AML who received 'triple therapy' with venetoclax, azacitidine and magrolimab.Interestingly, patients without prior venetoclax exposure responded significantly as compared to the cohort of patients that had prior venetoclax exposure (75% vs 12% CR/CRi respectively) [85].

Donor lymphocyte infusions
A donor lymphocyte infusion (DLI) refers to the transfusion of unstimulated lymphocyte concentrates, collected from the original stem cell donor as buffy coat preparations.DLIs for the treatment of relapsed leukemia was introduced primarily for chronic myeloid leukemia.Although less successful in AML, remissions have been observed in select cases, and it is considered primarily for relapse after allogeneic HCT [86].
Randomized trial data regarding the use of DLIs do not yet exist.The largest study evaluating the use of DLIs in relapsed AML after allogeneic HCT is a retrospective analysis of 399 patients, with the authors recommending against DLI after first relapse posttransplantation in favor of cytoreductive approaches owing to the inferior results of DLI with active disease (i.e. a 2-year OS of only 9%) [86].Those who achieve a remission can then be considered for a DLI.Another retrospective analysis of DLIs in relapsed patients after allogeneic HCT identified that patients who achieved a CR after DLI (only 8% of patients in the analysis) almost exclusively had the best OS [87].

Chimeric antigen receptor-transduced T cell (CAR-T) therapy
CAR-T cells are engineered T cells that can redirect antigen specificity and are being actively investigated as a promising therapeutic modality in the management of r/r AML.CAR-T therapy has already shown impressive results in the management of B-cell malignancies, and several specific tumor antigens are being explored as therapeutic targets for r/r AML, such as CD33, CD123, CLL-1, NKG2D and TIM-3, amongst others [88].
CD33 directed CAR-T therapy was evaluated for r/r AML in a clinical trial involving a 41-year-old male patient with a marked decrease in blast fraction two weeks after the therapy from >50% to <6%.However, cytokine release syndrome subsequently developed, and the patient died of disease progression 12 weeks later [89].CD123 is also being explored as a highly desirable target and may potentially be superior to CD33 directed therapy owing to their lower toxicity profile [88].A number of phase I/II studies are ongoing to investigate the efficacy of CD33 and CD123 directed CAR-T therapy as a potential target (e.g.NCT01864902, NCT03795779, NCT03190278).First results from a phase I clinical study evaluating CYAD-02, an NKG2D ligand-directed CAR-T therapy in 6 r/r AML patients demonstrated safety and tolerability with no dose-limiting toxicities being observed [90].

Discussion
Treatment options for de-novo AML have evolved greatly with the introduction of targeted therapies.For example, patients who harbor a FLT3 mutation and are eligible for intensive therapy are treated firstline with midostaurin in combination with 7 + 3 [65], and gemtuzumab ozogamicin can be considered with intensive chemotherapy for patients with CD33 positive AML [79].
Despite such advancements, AML remains a disease entity with a poor clinical outcome.In an analysis of more than 3012 patients treated in frontline ECOG AML studies, 58.9% of patients experienced relapse at a median of 7.2 months after achieving a CR [91].Even those patients treated with curative intent induction chemotherapy and having achieved CR per currently defined endpoints only had a median overall survival of 20 months in one review of over 4000 patients with AML [92].The management of r/r AML remains a clinically challenging one.
Patients who present with a suspected relapse will need to be thoroughly evaluated to adequately inform treatment decisions.A bone marrow aspiration for immunophenotyping, cytochemistry and cytogenetics with a mutation analysis should be performed even if they were performed at initial diagnosis owing to the possible emergence of new clones [8,9].HLA typing should be performed in potential candidates for HCT, and a donor search initiated for HCT to aid with the decision-making process.An assessment of the performance status of the patient and medical fitness will need to be made, factoring in medical comorbidities and relevant biochemical and/or imaging investigations.
The choice of therapy depends on several factors.Notwithstanding the above, other factors that will need to be considered include access to an experienced transplant center, resources for home care, and importantly the patient's own wishes.A patient's prior experience with induction chemotherapy may also be relevant if they were unable to tolerate it without significant toxicity.The European Prognostic Index for patients with AML in first relapse identified four risk factors (relapse-free interval from first remission, cytogenetics at the time of diagnosis, age at first relapse, HCT before relapse) from which three risk groups have been identified (favorable, intermediate, and high risk) [28].A retrospective study of AML patients who underwent salvage therapy found that a first CR duration of <12 months and a second CR duration of <6 months adversely affected OS [93].

Management of patients considered fit for intensive therapies
Although current treatment paradigms for r/r AML have shifted in favor of utilizing targeted agents, intensive regimens utilizing cytotoxic chemotherapy continue to play a role, especially in the management of younger and fit patients as a bridge to HCT.As discussed above, numerous different regimens have been studied in this setting, such as FLAG-IDA and MEC.Unfortunately, no particular intensive regimen has been found to be superior in the r/r setting.
Current directions are focused on investigating the combination of traditional cytotoxic regimens with targeted therapies.For example, FLA-V-IDA (FLAG-IDA plus venetoclax given on days 1-7) has demonstrated an improved overall response rate (ORR) compared to FLAG-IDA alone (69% vs 47% respectively) in a phase Ib/II study, with no excess hematologic toxicity being reported by the addition of venetoclax [94].
Patients with r/r AML are especially encouraged to participate in clinical trials investigating targeted therapies based on the mutation profile for the patient's particular AML.Allogeneic HCT should be strongly considered for patients with r/r AML, although as previously discussed, transplant eligibility is contingent on several other factors such as patient's age and comorbidities.Aggressive therapeutic regimens that may be considered for appropriate patients with r/r AML per the National Comprehensive Cancer Network (NCCN) guidelines are listed below [95].These recommendations are category 2A unless otherwise indicated: . CLAG ± idarubicin, CLAG-M. .HiDAC (must not have previously received such treatment) ± mitoxantrone/idarubicin. .FLAG, FLAG-IDA. .MEC.
The guidelines also recommend targeted therapy options in specific r/r AML populations as follows: AML with FLT3-ITD mutation: .Gilteritinib (category 1 recommendation).

Management of patients considered unfit for intensive therapies
In recent years, significant progress has been made in the management of AML patients considered unfit for intensive chemotherapy, such as the addition of the BCL2 inhibitor venetoclax to azacitidine.CR rates compared to azacitidine alone markedly improved CR rates, (66.4% vs 28.3%) and helped establish a new standard of care for AML patients unfit for intensive therapy [56].
Targeted therapy options are often employed in this setting owing to their favorable side effect profile, such as HMAs, BCL2 inhibitors and IDH inhibitors.Several other factors, notably the cytogenetic makeup at relapse, would influence this decision making.Molecular re-evaluation at this stage may detect actionable targets such as IDH1/IDH2 or FLT3-ITD clones.
Overall, approved therapeutic options in the r/r AML setting for unfit patients are very limited.It is recommended to enter these patients into clinical trials whenever possible.Less aggressive therapeutic options recommended per the NCCN guidelines are listed below [95]: . Venetoclax + HMA (azacitidine/decitabine)/low dose cytarabine .HMAs (azacitidine/decitabine) .LDAC (category 2B recommendation) As with the intensive treatment population, targeted therapies for specific AML populations may be considered.

Other considerations and concluding remarks
Although not discussed in this review as we attempted to limit the scope to therapeutic agents, allogeneic hematopoietic stem cell transplantation must be given due consideration in r/r AML, even though 5-year survival rates in such patients are reported at only 25% [5].Allogeneic HCT following a second CR is associated with relatively lower rates of relapse and represents the only potentially curative option in r/r AML patients who manage to achieve a second CR [1].Eligibility criteria for HCT vary significantly but patients with poor overall performance status, or moderate to severe organ impairment are usually considered ineligible.Achievement of a CR with cytoreductive or myeloablative therapy prior to HSCT is generally preferred.Following salvage therapy for r/r AML, the addition of myeloablative conditioning regimens such as busulfan plus cyclophosphamide may improve OS prior to HCT [96].
It is possible that r/r AML patients are by and large manifesting a refractory clone that has persisted despite successful initial therapy, and in view this, it should not be assumed that the initial treatment was successful, but rather that it was not as effective as could be hoped.It seems increasingly evident that the concept of 'remission' in these patients is predicated on outmoded criteria, as they do not provide a sensitive enough assessment of AML disease burden [97].Measurable residual disease assays have helped to highlight this discrepancy and explain the fact that despite the apparent success of many patients achieving a complete remission after induction therapy, their median overall survival remains less than two years [5].It is worth noting that measurable residual disease assays may also provide insufficient lead time to detect relapse with certain AML subsets, such as core-binding factor AML [98].
Although AML has often been approached as a homogenous disease entity in the past, variations in such disease factors as cytogenetic and molecular markers, age at presentation and disease severity on presentation all contribute to significant heterogeneity in this disease.It is increasingly being understood that tackling AML would require individualized therapy, and advances in technologies such as molecular profiling and drug sensitivity and resistance testing would perhaps be increasingly employed to tailor such therapy.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Table 1 .
[1]ponse and treatment failure criteria with their definitions[1].CRh or CRi at the response landmark, i.e. after 2 courses of intensive induction treatment or a defined landmark, e.g. 180 d after commencing less-intensive therapy Relapsed disease (after CR, CRh or CRi) Bone marrow blasts ≥ 5%; or reappearance of blasts in the blood in at least 2 peripheral blood samples at least one week apart; or development of extramedullary disease.

Table 3 .
Different chemotherapeutic regimens discussed in the management of relapsed/refractory acute myeloid leukemia.

Table 4 .
Mechanisms of action and notable adverse effects of targeted agents employed in the management of r/r AML.