How I treat secondary acute myeloid leukemia

Visual Abstract

Abstract

Secondary acute myeloid leukemia (sAML) has traditionally been used to designate any AML disease arising from an antecedent hematologic disorder or after prior cytotoxic or radiation therapy. We now know sAML comprises multiple disease entities with distinct clinical and biological features: AML, myelodysplastic related; myeloproliferative neoplasm-blast phase; and AML post–cytotoxic therapy. These entities largely represent adverse-risk phenotypes with the majority of patients experiencing suboptimal outcomes with standard therapeutic options. Given the aging general population and the increased life span of individuals receiving DNA-damaging agents for other medical conditions, the incidence of these diseases is steadily rising and now comprise ∼25% to 30% of all new AML diagnoses. Despite the plethora of novel agents approved for AML since 2017, many either are not applicable to sAML (ie, lacking a targetable mutation), have limited efficacy, or have not been studied in these specific entities. Furthermore, these patients are underrepresented in clinical trials, and novel therapeutic options are critically needed. Here, we present multiple patient cases exemplifying the new nomenclature and classification of the diseases comprising sAML and highlighting their diverse presentations. We provide our therapeutic approach for each clinical scenario and discuss the challenges of treatment with the currently available armamentarium.

Treatment for patients with acute myeloid leukemia (AML) is being transformed for some patients, as targeted agents increase complete remission rates and survival when appropriately applied. Although the efficacy of newer regimens has been established in major clinical trials, use in routine practice is often challenging, and careful choices need to be made. Associate Editor Selina M. Luger introduces this case-based How I Treat series designed to provide guidance in those circumstances. Wei et al discuss how to maximize the benefits when the venetoclax-azacitidine regimen was indicated, while Issa and colleagues discuss differentiation therapy, highlighting how the clinical course for patients varies from what is seen with intensive chemotherapy-based induction regimens. Green and Wang tackle the challenge of managing patients with secondary AML. Finally, Roboz and Canaani outline how maintenance can benefit selected patients not undergoing allogeneic transplant, while exploring the many unanswered questions that arise in clinical practice.

Introduction

Secondary acute myeloid leukemia (sAML) has traditionally been considered an umbrella term encompassing any AML (1) arising from an antecedent hematologic disorder including myelodysplastic syndrome (MDS), myeloproliferative neoplasm (MPN), and aplastic anemia or (2) evolving from prior exposure to cytotoxic or radiation therapy. In most cases, the diagnosis of sAML confers a worse prognosis than de novo AML, and optimal outcomes are achieved with upfront chemotherapy followed by hematopoietic stem cell transplantation (HSCT).

Making the diagnosis of sAML in the current era can be challenging for the practicing clinician because of the shifting nomenclature and criteria for this heterogeneous disease entity. All patients with suspected AML should ideally undergo detailed diagnostic workup including history, physical, and analysis of marrow samples for histopathology, flow cytometry, and cytogenetic and mutational profiling. We have summarized the current diagnostic criteria for sAML in Figure 1. These include prior history of myelotoxic therapy, prior documented myeloid disease (MDS and MPN), disease burden (≥20% blasts), and specific karyotypes and somatic mutations.

Figure 1.

Diagnostic criteria for sAML (umbrella term).

In the 2016 World Health Organization (WHO) classification, sAML comprised 2 entities: AML with myelodysplasia-related changes (AML-MRC) and therapy-related myeloid neoplasms. AML-MRC was defined as the presence of multilineage dysplasia (≥50% dysplastic cells in at least 2 cell lines in the absence of NPM1 or biallelic CEBPA mutations, a history of MDS, or the presence of an MDS-related cytogenetic abnormality.¹ Therapy-related myeloid neoplasms consisted of therapy-related AML (tAML) and therapy-related MDS.¹

Over the last several years, our understanding of the biological and molecular underpinnings of sAML has greatly expanded as reflected in the current 2022 WHO classification and the International Consensus Classification (ICC). Several distinguishing mutations (ASXL, BCOR, EZH2, SF3B1, STAG2, U2AF, and ZRSR2) have been identified almost exclusively in sAML as opposed to de novo AML (FLT3, NPM1, and DMT2A). Survival of patients with “mutation-defined” sAML after intensive chemotherapy are poorer than de novo AML and parallel those of patients diagnosed with sAML based on clinical history.^{2, 3, 4} This is reflected in the 2022 European Leukemia Net (ELN) prognostic categories in which AML characterized by secondary mutational events are considered to be “adverse risk.”³ In the 2022 WHO classification, the “AML-MRC” category has been replaced with a new entity termed “AML, myelodysplasia related” (AML-MR) based on cytogenetics and/or the presence of secondary mutations with qualifiers for prior documented MDS or MDS/MPN (Tables 1 and 2).⁵ The term “secondary” has now been reassigned to designate therapy-related myeloid neoplasms including myeloid neoplasms after cytotoxic chemotherapy or evolving from an identified germ line predisposition (not discussed here; Tables 1 and 2). Patients with pending or overt leukemic transformation of preexisting MPNs are now categorized based on blast percentages into MPN in accelerated phase (MPN-AP) with 5% to 19% blasts or MPN in blast phase (MPN-BP) with ≥20% blasts.

Table 1.

2022 WHO classification: AML-MR

Categories designated AML (if ≥20% blasts in BM/PB) or MDS/AML (if 10%-19% blasts)  AML with mutant TP53 AML with myelodysplasia-related gene mutations∗ AML with myelodysplasia-related cytogenetic abnormalities# AML not otherwise specified

Diagnostic qualifiers Therapy-related (prior therapy for nonmyeloid neoplasms) Prior chemotherapy, radiotherapy, immune interventions Progressed from MDS  MDS should be confirmed by standard diagnostics and >3 mo before AML diagnosis Progressed from MDS/MPN (specify type)  MDS/MPN should be confirmed by standard diagnostics and >3 mo before AML diagnosis Germ line predisposition (specify type)

Reference: Khoury et al.⁵

BM, bone marrow; PB, peripheral blood.

^∗Defining somatic mutations: ASXL1, BCOR, EZH2, SF3B1, SRSFS2, STAG2, U2AF1, ZRSR2.

^#Defining cytogenetic abnormalities: Complex karyotype (≥3 abnormalities); 5q deletion or loss of 5q due to unbalanced translocations; Monosomy 7, 7q deletion, or loss of 7q due to unbalanced translocation; 11q deletion; 12p deletion or loss of 12p due to unbalanced translocation; Monosomy 13 or 13q deletion; 17p deletion or loss of 17- due to unbalanced translocation; Isochromosome 17q; idic(X)(q13).

Table 2.

2022 WHO classification: secondary myeloid neoplasms and MPN-BP

Other secondary myeloid neoplasms	MPN-BP
Myeloid neoplasms after cytotoxic therapy AML, MDS, MDS/MPN diagnoses Documented history of chemotherapy (except methotrexate) or large-field radiation therapy for an unrelated neoplasms or PARP inhibitor treatment Excludes clonal cytopenia of unknown significance	Prior history of MPN (CML, PV, ET, MF, chronic neutrophilic leukemia, chronic eosinophilic leukemia, MPN-NOS) Presence of ≥20% marrow blasts
Myeloid neoplasms associated with germ line predisposition

Reference: Khoury et al.⁵

CML, chronic myeloid leukemia; ET, essential thrombocythemia; MF, myelofibrosis; MPN-NOS, myeloproliferative neoplasm, not otherwise specified; PV, polycythemia vera.

Here, we present patient cases exemplifying the new classifications and highlighting the diverse clinical presentations of sAML arising from prior MDS, MPN, and cytotoxic therapy. We provide our approach for each scenario and discuss the challenges of treatment with the currently available armamentarium.

Case 1

A 57-year-old man with no significant past medical history had been feeling poorly and was found to have significant anemia, thrombocytopenia, and 36% peripheral blasts. Bone marrow biopsy (BMBx) demonstrated AML with monocytic differentiation. Dysplasia was detected in >50% of cells in 2 lineages. Cytogenetics revealed normal karyotype. Next-generation sequencing revealed mutations in isocitrate dehydrogenase enzyme 2 (IDH2), SRSF2, and STAG2 mutations. He received CLAG-M (cladribine, cytarabine, granulocyte colony-stimulating factor, and mitoxantrone) and achieved complete remission (CR) with no evidence of measurable residual disease (MRD) by flow cytometric analysis. He proceeded onto matched unrelated myeloablative allogeneic HSCT and is now disease-free.

Management of fit patients with AML-MR

This young patient has a diagnosis of AML-MRC (2016 WHO) and AML-MR with defining somatic mutations (SRSF2 and STAG2; 2022 WHO).⁵ Because of the normal karyotype and lack of antecedent hematologic disorder, his prognosis is intermediate (ELN 2017)⁶ or adverse risk (ELN 2022).³ This highlights the critical need to expediently obtain complete cytogenetic and molecular information before treatment decisions in all AML cases. Otherwise, patients may not benefit from sAML-specific therapies including clinical trials.

Although our approach to AML-MR (Figure 2A) in fit patients favors intensive chemotherapy followed by HSCT, the optimal upfront treatment remains controversial. Multiple studies have demonstrated the suboptimal outcomes of cytarabine and anthracycline-based (7+3) chemotherapy in sAML with CR and CR with incomplete hematologic recovery (CRi) rates of 30% to 50% and median overall survival (OS) of 6 to 7 months.^{7, 8, 9, 10} HSCT improves survival after 7+3 with a median OS of 1.5 years.^11,12

Figure 2.

Suggested treatment algorithms. (A) AML-MR; (B) ts-AML; (C) MPN-BP; (D) AML-pCT.

Currently the only agent approved for treatment of AML-MR in fit individuals is CPX-351, a liposomal preparation of cytarabine and daunorubicin in a fixed 5:1 molar ratio. In a phase 3 trial of CPX-351 vs 7+3 induction in patients aged 60 to 75 years with newly diagnosed tAML, chronic myelomonocytic leukemia, or AML-MRC, CPX-351 significantly improved median OS (9.56 vs 5.95 months; P = .003) and CR/CRi rates (47.7% v 33.3%; P = .016.).¹³ Side effects include decreased alopecia and mucositis, prolonged myelosuppression and rash, and higher cost than 7+3.¹⁴ More patients receiving CPX-351 bridged to HSCT, and post-hoc analysis demonstrated improved OS after HSCT with upfront CPX-351, primarily because of lower nonrelapse mortality and relapse incidence.^15,16 Although the mutational definition of AML-MR was not used, we extrapolate the use of CPX-351 to these patients as well. Although MRD was not assessed, these outcomes are consistent with greater disease control.^17,18 For patients receiving intensive chemotherapy who are transplant ineligible, we favor oral azacitidine (AZA) maintenance.¹⁹ Although CPX-351 was approved independent of age, this trial only enrolled patients aged ≥60 years, and therefore the National Comprehensive Cancer Network lists 7+3 as “preferred” for those aged <60 years.²⁰ However, retrospective studies have indicated comparable outcomes of CPX-351 in younger individuals.^21,22 We are currently conducting a prospective phase 2 trial to address this question (ClinicalTrials.gov identifier: NCT04269213).

Upfront regimens using a purine analog combined with high-dose cytarabine have also yielded higher responses and prolonged OS over 7+3 in patients with AML-MR and adverse-risk non-TP53 mutant AML. In the MRC AML15 trial, younger patients with sAML (n = 115) were randomized to FLAG-IDA (fludarabine, cytarabine, idarubicin, and granulocyte colony-stimulating factor) vs 7+3–based therapy (cytarabine, daunorubicin with/without etoposide). Although response rates were the same (81% vs 79%), 5-year OS (37% vs 27%, P = .02) and relapse free survival (41% vs 22%; P = .04) were superior for FLAG-Ida vs 7+3.²³ Analysis of the MRC AML19 trial, which randomized patients with adverse-risk AML and high-risk MDS to intensive chemotherapy suggested no difference in OS or event-free survival between CPX-351 and FLAG-IDA in patients with clinical sAML or MDS cytogenetics. However, subgroup exploratory analysis detected an OS benefit favoring CPX-351 over FLAG-IDA in patients with AML with MDS-related mutations (hazard ratio [HR], 0.42; 95% confidence interval [CI], 0.21-0.84; P = .05).²⁴ Venetoclax (VEN) added to FLAG-IDA for patients with sAML (n = 7) and tAML (n = 5) resulted in composite complete remission (CRc) rates of 83% with 90% MRD-negative responses and a median OS that has not been reached. These outcomes were not statistically different than those of de novo AML disease; however these results obtained in a single academic center await further confirmation in other centers and in larger numbers of patients.^25,26 Retrospective data from other institutes have also demonstrated markedly improved responses in younger patients with sAML treated with CLAG-M (cladribine, cytarabine, granulocyte colony-stimulating factor, and mitoxantrone) vs 7+3.^{27, 28, 29} Which patients benefit from purine and intensified cytarabine regimens as opposed to CPX-351 remains uncertain. For patients not achieving CR after 1 intensive induction cycle, a second cycle can be given; however, we often consider switching to VEN and hypomethylating agent (VEN/HMA) to retain performance status and to challenge patients with different mechanistic drugs.

Mutational profiling significantly influences therapy. TP53 mutations are frequently identified in AML-MR and confer resistance to intensive and VEN/HMA chemotherapy. Outcomes of TP53-mutant AML, specifically those with biallelic TP53 mutations, very high variant allelic frequencies, or complex karyotype, are particularly dismal and represent a high unmet need. For fit patients with TP53-mutant disease, we favor administration of intensive chemotherapy (FLAG-IDA, CLAG-M) or VEN/HMA over AZA monotherapy to more rapidly achieve disease control and permit HSCT as soon as possible in first CR. CPX-351 does not appear to offer any advantages.²⁴

In FLT3-mutant AML-MR, response rates remained higher after CPX-351 than with 7+3 (CR/CRi, 68.2% vs 23.8%), and post-hoc analysis demonstrated a trend toward improved survival with CPX-351.¹³ The phase 1b V-FAST study evaluating CPX-351 combined with midostaurin in FLT3-mutant AML (largely de novo AML) showed tolerability and an 80% CR rate.³⁰ If a FLT3 mutation is present, we consider adding a FLT3 inhibitor (quizartinib for FLT3-ITD, midostaurin for FLT3-TKD) on cycle 1, day 8 of induction.

Case 2

A 69-year-old woman with past medical history of chronic obstructive pulmonary disease (COPD) and a cardiac pacemaker presented with anemia. BMBx demonstrated MDS-excess blasts with 6% blasts. Cytogenetics demonstrated a normal karyotype. Next-generation sequencing revealed BCOR, RUNX1, and SF3B1 mutations. She was treated with AZA. Restaging BMBx after 4 cycles demonstrated 3% blasts. After 12 cycles she developed progressive anemia and thrombocytopenia. Repeat BMBx demonstrated hypercellular marrow with trilineage dysplasia and 21% blasts suggestive of AML transformation. Karyotype remained normal. Next-generation sequencing revealed the same mutations with new CEBPA and KMT2A mutations. She was treated with VEN and decitabine (DEC). A day-21 BMBx was 30% cellular with 2% blasts. She experienced prolonged count recovery. Repeat BMBx at day 42 showed a 10% cellular marrow and 1% blasts. Flow MRD was positive at 0.2%. The plan was to continue VEN/DEC with reduction of VEN duration to 10 to 14 days and repeat MRD testing after additional cycles. She continues to be evaluated for HSCT.

Management of patients who are unfit with AML arising from prior MDS

This older patient has a diagnosis of AML-MRC (2016 WHO) and AML-MR with defining somatic mutations as well as documented prior MDS (2022 WHO).⁵ Her disease is adverse risk (ELN 2022). For unfit patients with AML-MR, we favor VEN-based lower-intensive therapy. We also used these regimens for patients with borderline fitness or medical comorbidities to minimize treatment-related toxicities and preserve functional status for potentially curative HSCT. HMA monotherapy has yielded overall poor outcomes in prospective trials.^{31, 32, 33} In the VIALE-A trial, the median OS for patients with sAML treated with VEN plus AZA (VEN/AZA) was 16.4 months compared with 10.6 months with placebo/AZA; the CR/CRi rate was 66.7% vs 22.9%.³³ Retrospective studies comparing VEN/HMA with intensive chemotherapy suggest similar outcomes pending validation in prospective trials.³⁴ For younger patients with MDS/AML (10%-19% blasts), we prefer VEN/HMA therapy over intensive induction therapy to preserve cardiac and other organ function and minimize treatment-related mortality and morbidity before HSCT. VEN plus low-dose cytarabine (VEN/LDAC) remains another option in frailer patients with sAML, based on its slightly better tolerability and home self-administration. In the VIALE-C trial, the median OS for patients with AML-MR receiving VEN/LDAC vs placebo/LDAC was prolonged (5.5 vs 3.2 months); the CR/CRi rate was 36% vs 4%.³⁵

Patients with sAML emerging from MDS with prior extensive HMA exposure, also known as treated secondary AML (ts-AML), remain a high unmet need with significantly worse survival (4.2 months) and CR (32%) than those without prior HMA, independent of age, cytogenetic risk, or treatment intensity.³⁶ For individuals who are older/less fit, constituting the majority of ts-AML, we still favor VEN-based regimens. This is based largely on retrospective studies because unfortunately patients with ts-AML have been excluded from many trials including VIALE-A.^33,37 Among 526 patients with ts-AML, VEN/HMA resulted in higher CR/CRi rates (39% and 25%; P = .02) and 1-year OS (34% and 17%; P = .05) than non-VEN–based intensive/low-intensity therapy. Encouraging results of VEN/HMA for patients with high-risk MDS (with disease that today meet ICC criteria for MDS/AML) who have failed prior HMAs further support this regimen’s efficacy in ts-AML.^38,39 Given the concern for proven HMA resistance, which may blunt response and, despite limited data, we often empirically switch the HMA, that is use DEC and VEN for patients with extensive prior AZA exposure^{40, 41, 42, 43} or reach for VEN/LDAC. In fact, a primary reason why outcomes after VEN/LDAC in VIALE-C were worse than VEN/AZA in VIALE-A was because the former allowed ts-AML whereas the latter did not. In VIALE-C, patients with ts-AML constituted 20% of all patients and achieved a median OS of 5.6 vs 4.1 months and a CR/CRi of 28.6% vs 7.1%, after VEN/LDAC vs LDAC, respectively.³⁵

In younger fit patients with ts-AML, our approach (Figure 2B) involves intensive chemotherapy with purine analogues (ie, CLAG-M and FLAG-IDA). The latter is based on the lack of superiority of CPX-351 vs 7+3 and high response rates to CLAG-M.^13,28,29 Retrospective analyses have revealed poor outcomes with 7+3 in this population. Notably, in the phase 3 study of CPX-351 vs 7+3, there was no difference in outcomes for patients with ts-AML receiving either therapy (median OS, 5.65 vs 7.43 months).¹³ CLAG-M appears more promising.²⁷ Among 242 patients with ts-AML treated with intensive regimens, the CR/CRi rates for CLAG-M, 7+3, and CPX-351 were 53%, 32%, and 41.2%, respectively; however, there was no significant difference in median OS (7 months).²⁹ At our institute, we evaluated CLAG-M in 41 patients with ts-AML and adverse karyotype; the CR/CRi rate was 64%, and median OS was 8.5 months (Table 3).²⁸

Table 3.

Clinical outcomes of patients with sAML after HMA therapy

Study	Study design	Treatment	Patient population	Number of patients after HMA	Response rates (%)	Median OS (mo)
Lancet et al45	Randomized phase 2	CPX-351 vs 7+3	60-75-y-old AML with antecedent MDS/MPN or t-AML treated with HMA	CPX: n = 13, 7+3: n = 7	CPX-351: 23%; 7+3: 29%
Lancet et al13	Randomized phase 3	CPX-351 vs 7+3	60-75-y-old AML with antecedent MDS or CMML, t-AML, or de novo AML with MDS-related cytogenetic abnormalities (per 2008 WHO criteria) treated with HMA	CPX: n = 50, 7+3: n = 55	CPX-351: 26.0%; 7+3: 18.2%	CPX-351: 5.7 mo; 7+3: 5.9 mo
Boddu et al36	Retrospective	Any	AML with antecedent MDS, MPN, or AA treated with at least 1 chemotherapy or immunomodulator therapy for their AHD	N = 254 (185 received HMA)	CR = 32% (AHD with prior therapy) vs 67.6% (t-AML) vs 79.5% (de novo AML)	Intensive chemo = 4.2 mo (AHD with prior therapy)
Bello et al46	Retrospective	Induction chemotherapy	AML with antecedent MDS or MDS/MPN treated with HMA or lenalidomide	N = 25 (24 received HMA)	CR/CRi = 32%	3.7 mo (AHD with prior therapy)
Talati et al29	Retrospective	CLAG-M vs 7+3 vs CPX-351	AML with antecedent MDS or CMML treated with HMA	N = 241	CLAG-M: CR/CRi = 53%; 7+3 = CR/CRi 32%; CPX-351 = CR/CRi 41%	CLAG-M: 7.27 mo; 7+3: 7.63 mo; CPX-351: 7.07 mo
Jaglal et al27	Case-control	CLAG-M vs 7+3	AML with antecedent MDS or MDS/MPN treated with HMA	CLAG-M: n = 28, 7+3: n = 24	CLAG-M: CR = 50%; 7+3: CR = 21%	CLAG-M: 6.73 mo; 7+3: 2.87 mo
Przespolew-ski et al28	Retrospective single institute	CLAG-M vs 7+3	AML treated with HMA with poor-risk karyotype; ND and R/R AML	CLAG-M (n = 16); CPX-351 (n = 22)	CLAG-M: ORR = 81.3%. CR/CRi = 58%; CPX-351: ORR = 40.9%; prior HMA (n = 14): ORR = 82%; CR/CRI = 64.1%; CLAG no M: ORR 64.3%	CLAG-M in patients with prior HMA: 254 d (8.47 mo)
Short et al44	Retrospective single institute	Intensive vs low-intensity chemotherapy	Treated secondary AML	Intensive chemotherapy (n = 271) vs low-intensity chemotherapy without VEN (n = 237) vs HMA/VEN (n = 54)	VEN/HMA vs other IC and low-intensity therapy: CR/CRi 39% vs 10%	VEN/HMA vs other: 1-y OS, 34% vs 17%; VEN/HMA in nonadverse-risk karyotype = mOS, 13.7 mo, 1-y OS, 54%; Any therapy in adverse-risk karyotype mOS, 3-5 mo
Wei et al35	Randomized phase 3	VEN/LDAC vs LDAC	Treatment-naïve AML not eligible for intensive chemotherapy, excluding prior MPD	LDAC: n = 14; VEN/LDAC: n = 28	LDAC: 0% vs VEN/LDAC: 7%	4.1 mo (LDAC) vs 5.6 mo (VEN/LDAC)
Dinardo et al114	Phase 2	VEN/DEC (20 mg/m2 × 10 d)	Treatment-naïve AML and relapsed/refractory patients not eligible for intensive chemotherapy, prior MDS, CMML, tAML	N = 50 with previous HMA’s	16%	6.0 mo
Roboz et al47	Phase 1	IVO	Treatment-naïve AML not eligible for intensive chemotherapy, prior MDS, MPD, or tAML	N = 15	20%	NR
Heuser et al48	Randomized phase 2	Glasdegib/LDAC vs LDAC	Treatment-naïve AML not eligible for intensive chemotherapy with AHD or tAML	Glasdegib/LDAC: n = 11 vs LDAC: n = 6	Glasdegib/LDAC: 9.1% vs LDAC: 0%	7.1 mo (Glasdegib/LDAC) vs 5.1 mo (LDAC)

AA, aplastic anemia; AHD, antecedent hematological disorder; CMML, chronic myelomonocytic leukemia; IC, intensive chemotherapy; mOS, median overall survival; MPD, myeloproliferative disorder; ND, not done; NR, not reached; ORR, overall response rate; R/R, relapsed/refractory.

Importantly, patients with ts-AML are often specifically excluded from, and therefore vastly underrepresented in, clinical trials. This was despite the fact that neither intensive chemotherapy nor VEN/HMA was effective in adverse-risk karyotype and TP53-mutant ts-AML (median OS, 2-3 months).⁴⁴ We feel that patients with ts-AML remain excellent candidates and should be prioritized for early phase therapies, even upfront (Table 4).

Table 4.

Ongoing clinical trials for newly diagnosed sAML and MDS/AML

Clinicaltrials.gov identifier	Trial phase	Trial population	Treatment	Status
NCT04269213	2	t-AML, AML-MRC, and ts-AML in patients aged <60 y	CPX-351	Recruiting
NCT03150004	2	De novo AML (relapsed/refractory), t-AML, AML secondary to MDS, ts-AML, MPN-BP, high-risk MDS after failure of HMA	CLAG-M, cladribine and cytarabine (CLLDAC)	Recruiting
NCT05780879	2	t-AML, AML-MRC, ts-AML	VEN + FLAG or CLAG	Not yet recruiting
NCT04231851	2	t-AML, AML-MRC, ts-AML	CPX-351 + glasdegib	Recruiting
NCT04802161	2	t-AML, AML-MR, ts-AML	CPX-351 + pomalidomide	Recruiting
NCT04128748	1/2	Newly diagnosed AML and MDS, R/R AML and MDS including ts-AML (do not have to be FLT3+)	CPX-351 + quizartinib	Recruiting
NCT04982354	1/2	FLT3+ AML	CPX-351 + midostaurin followed by busulfan, melphalan, fludarabine conditioning therapy and CD34+-selected allograft	Recruiting
NCT03862157	1/2	t-AML, AML-MRC, ts-AML, MDS/CMML, MDS/CMML post-HMA failure	VEN + AZA + pevonedistat	Active, not recruiting
NCT05513131	2	t-AML, AML-MRC	VEN + AZA + harringtonine	Recruiting
NCT04905810	2	ts-AML	Azacitidine or DEC + VEN	Recruiting
NCT05442216	2	ts-AML or MDS/AML after prior HMA (CD123+)	Tagraxofusp + AZA with/without VEN	Recruiting
NCT04848974	1/2	ts-AML	Uproleselan + cladribine + LDAC	Active, not recruiting
NCT05379166	2	Therapy-related MDS	VEN + Azacitidine	Recruiting
NCT03878199	1/2	MPN-AP/BP	CPX-351 + ruxolitinib	Recruiting
NCT04282187	2	MPN-AP/BP	DEC + ruxolitinib, fedratinib, or pacritinib	Recruiting
NCT04763928	2	MPN-BP	VEN + DEC	Unknown status
NCT05074355	2	MPN-AP/BP	VEN + AZA	Recruiting
NCT05735184	1/2	KMT2A-r AML	Ziftomenib + VEN + AZA vs ziftomenib + 7+3	Recruiting
NCT05886049	1	KMT2A-r AML	Revumenib + 7+3	Recruiting
NCT05360160	1	KMT2A-r AML	Revumenib + oral DEC/cedazuridine + VEN	Recruiting

Source: www.clinicaltrials.gov (accessed 1 May 2024).

CMML, chronic myelomonocytic leukemia; R/R, relapsed/refractory.

Mutation status, specifically the presence of “targetable” mutations, figure prominently in treatment selection, particularly for individuals who are frail. Patients with IDH1/2-mutant AML-MR who are unfit/borderline may benefit from IDH inhibitor–based regimens, adjusting intensity depending on frailty and transplant candidacy. For instance, we prefer ivosidenib (IVO) as monotherapy for individuals with IDH1-mutant sAML unable to receive daily HMA therapy or frequent transfusions. We similarly prefer the combination of IVO and AZA over VEN/HMA. Although not compared in a randomized controlled trial, the outcomes of IVO/AZA in the AGILE study demonstrated similar, if not greater, event-free survival (HR, 0.33) and median OS (24 vs 7.9 months; P = .0005) with less myelosuppression than VEN/AZA in the VIALE-A trial.^33,48,49 For patients who are unfit with TP53-mutant sAML who are ineligible for allogeneic SCT, we prioritize clinical trials, HMA monotherapy, or supportive care. We consider VEN/HMA in select patients who would clinically benefit from absolute neutrophil count (ANC) recovery, transfusion independence, and/or consideration of subsequent HSCT.

Case 3

A 60-year-old man with a prior history of polycythemia vera was referred for declining blood counts. Peripheral blood showed 11% circulating blasts, and BMBx revealed markedly hypercellular marrow (>95%), multilineage dysplasia, moderate reticulin fibrosis, and ∼24% blasts, consistent with AML transformation. Cytogenetics revealed 46, XY, i(17), (q10)[20] with trisomy 17q and monosomy 17p with loss of 1 TP53 locus. Next-generation sequencing reported ASXL1, JAK2 V617F, RUNX1, SRSF2, and TET2 mutations. He was enrolled in a clinical trial randomizing younger adverse-risk patients with AML to 7+3 or VEN/HMA induction with intention to proceed to HSCT in first CR. He received VEN/AZA therapy and a day-21 BMBx revealed 2% blasts with persistent cytogenetic abnormalities. After cycle 3, his counts were still not recovered enough at week 6 to start the next cycle by protocol criteria, and 11% blasts were noted in the periphery. Unfortunately, his transplant was delayed because of the development of a pericardial effusion and elevated bilirubin. He completed an additional 7 cycles of VEN/AZA with prolonged delays between treatment cycles related to cytopenias. He then developed overt relapse with 40% to 50% marrow blasts. Given his performance status and goals of care, he elected to go home on hospice.

Management of sAML arising from prior MPN

This older patient has a diagnosis of AML-MRC (2016 WHO) and AML-MR based on cytogenetic and mutational profile (2022 WHO) with ELN 2022 adverse risk based on complex cytogenetics and somatic mutations (ASXL1 and SRSF2).⁵ Given the clinical history, the most appropriate diagnosis would be blast phase MPN (MPN-BP; ≥20% blasts; WHO 2022).

Outcomes of MPN-BP are dismal, with a median OS of only ∼3 to 5 months.^{50, 51, 52, 53} HSCT yields the best outcomes with 3 to 5 year survival of ∼30%.^53,54 Our approach (Figure 2C) resembles our strategy for AML-MR, with some important caveats. For younger fit patients eligible for HSCT, we favor intensive induction to achieve higher rates of CR. However, because the landmark phase 3 trial of CPX-351 in sAML did not include patients with MPN-BP, the benefit of this agent is unclear.¹³ Of note, intensive chemotherapy without subsequent HSCT is associated with higher treatment-related mortality (15%-33%) and does not appear to improve outcomes over nonintensive approaches.^{51,53,55, 56, 57, 58, 59, 60, 61, 62} This patient was enrolled on a randomized trial comparing these 2 approaches. Given the overall poor prognosis and the lack of clarity on the depth of response needed before HSCT,^63,64 we favor directly proceeding with transplantation after upfront therapy without optimal remission achievement, an approach supported by a randomized phase 3 trial in patients with nonfavorable-risk AML.⁶⁵

Given the limitations and toxicities of intensive chemotherapy, we favor low-intensity therapy for most patients with MPN-BP including older borderline fit patients eligible for transplant as well as patients who are unfit. In retrospective analyses, HMA monotherapy yielded a median OS of 5 to 11 months and CR/CRi rate of 4% to 30%.^{53,55, 56, 57,59,60,66, 67, 68, 69, 70} However, because therapy discontinuation in many patients on long-term JAK inhibitor (JAKi) leads to profound recurrence of systemic symptoms and/or splenic enlargement, continuation of prior JAKi with addition of HMA may be preferred over VEN/HMA in some patients.

The decision to add ruxolitinib to HMA therapy either upfront or sequentially depends on the patient’s symptom burden, splenomegaly, transplant candidacy, and prior JAKi therapy. For example, if a patient has massive splenomegaly, was not on a JAKi previously, and is a transplant candidate, we would consider adding ruxolitinib to HMA upfront to facilitate engraftment and improve performance status.^71,72 DEC and ruxolitinib was studied in phase 1 and 2 trials in MPN-AP/BP.^{73, 74, 75} In the phase 2 portion, 2 of 25 (8.1%) patients achieved a CR/CRi, with a median OS of 9.5 months and spleen reduction of 70.5%. Two patients (8%) proceeded to transplant. There was no clear correlation between response and survival. Two other early phase clinical trials used ruxolitinib in combination with DEC or AZA and achieved very similar results in terms of survival, although the response rates varied considerably.^76,77 A meta-analysis found that HMA plus ruxolitinib in patients with MPN-AP/BP yielded a higher overall response rate than HMA monotherapy (45% vs 30%, P = .0395), with a trend toward improved CR/CRi rate and prolonged survival compared with HMAs alone. Because a major issue with ruxolitinib is significant myelosuppression, tapering off drug for profound cytopenias is reasonable given the lack of durable responses with this approach. Clinical trials evaluating other JAKis with HMAs are ongoing (Table 4).

Although the landmark clinical trials evaluating VEN/HMA and VEN/LDAC excluded patients with MPN-BP, VEN/HMA remains an option.^33,35,37,78 Overall, the CR/CRi rate of VEN/HMA for patients with MPN-BP appears to be higher than HMA alone (∼40%), with comparable median OS (6-9 months).^{79, 80, 81, 82, 83, 84, 85, 86} Meta-analysis of patients with MPN-AP/BP found that VEN/HMA yielded a higher CR/CRi rate than HMA alone (36% vs 19%, P = .0204) and a higher CR rate than HMA plus ruxolitinib (22% vs 8%, P = .0313), potentially expediting HSCT.⁸⁷ However, unlike other AML subtypes, MPN-BP do not demonstrate a clear B-cell lymphoma 2 dependence, possibly explaining results of clinical studies suggesting no clear survival benefit with VEN-based approaches over HMA alone. Although outcomes of patients undergoing HSCT are not adversely affected by prior lower-intensity therapy,⁸⁸ prolonged cytopenias leading to infectious and hemorrhagic complications are prevalent. Optimizing patient selection and fine-tuning dose and duration of VEN/HMA are critical.

Despite emerging results with IDH1/2 inhibitors, there is insufficient data to make firm recommendations regarding targeted therapies in MPN-AP/BP.^{89, 90, 91} Clinical trials in these patients are accruing and may benefit from focusing on survival end points rather than weakly reliable surrogates such as response depth (Table 4).

Case 4

A 63-year-old woman with a history of serous ovarian cancer recently returned from a trip to Florida with worsening shortness of breath, tinnitus, lightheadedness, and intermittent palpitations. Her ovarian cancer is in remission after 5 cycles of carboplatin and Taxol followed by 2 years of olaparib maintenance. Laboratory data show severe pancytopenia with 3% peripheral blasts. BMBx revealed AML with 71% myeloblasts. Cytogenetics were monosomy 7. Myeloid gene panel demonstrated an ASXL1 mutation. She received induction therapy with CPX-351 complicated by febrile neutropenia and prolonged cytopenia. Repeat BMBx showed morphologic remission with MRD-negative disease by flow cytometry. She received consolidation chemotherapy and proceeded onto matched unrelated HSCT.

Management of tAML after prior cytotoxic agents

According to the 2022 WHO criteria, this patient’s diagnosis is AML post–cytotoxic therapy (AML-pCT), defined as AML arising after prior cytotoxic or radiation therapy. In the 2016 WHO, this diagnosis would be tAML. AML-pCT constitutes a diverse disease entity arising from the interaction between preexisting clonal hematopoiesis and damage inflicted by myelotoxic agents on hematopoietic stem cells and the marrow microenvironment. The traditional exposures leading to AML-pCT constitute alkylating agents/ionizing radiation and topoisomerase II inhibitors. The former is associated with AML with chromosome 5 and 7 aberrations, complex karyotype, and p53 mutation; the latter results in AML with KMT2A rearrangement, t(15;17), and RUNX1 mutations.^{92, 93, 94} However, a significant minority of AML-pCT lack these abnormalities and in fact can have favorable (ie, core binding factor [CBF]) or intermediate (ie, normal karyotype) cytogenetic and molecular risk disease (ELN 2016/2022). In addition, the 2022 WHO criteria also now includes nonclassical DNA-damaging therapies such as poly(ADP-ribose) polymerase (PARP) inhibitors as qualifying criteria for AML-pCT.^5,95

Given the disease heterogeneity and reliance on data extrapolated from larger sAML studies, our treatment approach (Figure 2D) to patients with AML-pCT focuses on underlying disease biology. Despite significant overlap in cytogenetic and molecular abnormalities between AML-pCT and AML-MR, especially after prior MDS, this is not true for all patients. Because the designation of AML-pCT is based on clinical history rather than intrinsic biology, AML-pCT in some individuals may actually constitute de novo AML arising by chance and/or common carcinogenic factors independent of prior therapy.⁹⁶ This can occasionally lead to prognostic and therapeutic tension. For example, NPM1-mutated AML-pCT clinically and biologically more closely resembles de novo disease with coincidental cytotoxic therapy than “true” AML-pCT and is not considered adverse prognosis.⁹⁷ Of note, there are diverging results from retrospective analyses on whether the presence of sAML mutations negatively affects the prognosis of patients with NPM1-mutant AML receiving intensive chemotherapy.^{98, 99, 100, 101} Although CBF AML-pCT have better prognoses than AML-pCT with intermediate- or adverse-risk classification, matched analyses have suggested that the outcomes of “therapy-related” CBF AML are worse than de novo CBF AML, potentially because of older patient age and additional cryptic mutations.¹⁰² Despite this, we administer 7+3 and gemtuzumab ozogamicin and defer transplant in MRD-negative remission in patients with favorable-risk AML-pCT (ie, characterized by RUNX1::RUNX1T1–, CBFB::MYH11–, or NPM1-mutant AML without adverse-risk mutations).^{103, 104, 105}

For fit patients with nonfavorable AML-pCT, we administer CPX-351 ideally followed by HSCT. Like AML-MR, these patients have an overall poor prognosis even after accounting for other factors by multivariate analyses.^7,104 The median OS of patients with AML-pCT after 7+3 approximates 6 months.¹⁰⁶ In comparison, patients with t-AML treated on the phase 3 CPX-351 vs 7+3 trial achieved a median OS of 12.2 vs 6.0 months (HR, 0.48; 95% CI, 0.26-0.86) and CR/CRi rate of 47 vs 36%. Patients with AML-pCT (37%) who underwent subsequent HSCT after CPX-351 derived the greatest overall benefit with a median OS that was not reached.¹³

Importantly, patients with AML-pCT can present unique therapeutic challenges including a history of extensive prior anthracycline exposure, poor marrow reserve from prolonged prior myelotoxic therapy, and presence of another concurrent cancer. For these patients, an honest discussion of the risks and benefits of treatment is key. In fit patients with significant anthracycline history and favorable-risk AML-pCT, we administer 7+3 concomitantly with dexrazoxane, an US Food and Drug Administration–approved drug for preventing anthracycline-induced cardiotoxicity in children and adolescents.¹⁰⁷ For non-favorable-risk disease, we prefer CLAG without mitoxantrone, which appears to result in similar outcomes as CLAG-M.²⁸ Although no landmark trial evaluating HMA, VEN/LDAC, or VEN/HMA have reported outcomes in AML-pCT, all enrolled these patients. Therefore, we administer VEN/HMA for borderline fit patients before HSCT and in older patients who are unfit. On occasion, we have given low-intensity therapy (ie, HMA) concomitantly with other cancer-directed therapies (ie, immune checkpoint inhibitors) to patients with simultaneous diagnoses of sAML and other malignancies.^108,109 To date, the benefit of targeted therapy (FLT3 and IDH1/2 inhibitors) added to chemotherapy backbones for AML-pCT is not known and requires further study.

Rearrangements in the KMT2A gene (also known as the mixed-lineage leukemia gene) are identified in 15% of tAML, with 70% arising 1 to 2 years after topoisomerase inhibitor therapy. Menin inhibitors targeting the menin–KMT2A protein interaction have demonstrated promising monotherapy efficacy (CR/CRi, 20%-35%; median OS, 7-8 months) in heavily pretreated relapsed/refractory KMT2Ar AML.^110,111 We are prioritizing clinical trials of menin inhibitors combined with intensive and nonintensive chemotherapy for patients with KMT2Ar AML-pCT.¹¹² Other ongoing research in AML-pCT is focused on developing risk prediction models and reduction strategies based on identification of preexisting clonal hematopoiesis of indeterminate potential and molecular aberrations (Table 4).¹¹³

Conclusions

The recurrent theme across the treatment spectrum for all subtypes of sAML is the universally poor outcomes to available chemotherapy and the pressing need to pursue HSCT for prolonged OS. Given these challenges, better identification, and monitoring of patients with MPN and MDS at higher risk of leukemic transformation will be critical to achieve better outcomes. These include adoption of measures such as early referral of fit patients with higher-risk disease to HSCT. Newer classification systems such as the Molecular International Prognostic scoring system, ICC, and WHO 2022 are furthering this aim by incorporating disease biology, specifically molecular information, into routine diagnostic and prognostic workups. In addition, the development of single-cell sequencing technology has elucidated the mechanisms of leukemogenesis and clonal architecture leading to sAML. This will hopefully translate into better disease monitoring (ie, MRD), earlier identification of leukemia-promoting clones, and integration of mutation-targeted agents into future treatment approaches.

At present, there is no standard of care for sAML. Our review highlights the fact that most of the available evidence for therapy consists of subset analyses, extrapolations of larger studies, and retrospective single-institute experiences. Enrolling patients with defined biological and clinical subsets (ie, AML-MR, AML-pCT, MPN-BP, and ts-AML) in well-designed multicenter clinical trials should be prioritized and are essential to improving treatment outcomes.

Conflict-of-interest disclosure: S.D.G. served on an advisory board of, and provided consulting for, Sobi and GlaxoSmithKline. E.S.W served on an advisory board for and/or received consulting fees from, AbbVie, Blueprint, Daiichi Sankyo, Immunogen, Kite, Kura, Qiagen, Rigel, Schrodinger, Stemline, and Syndax; provided talks for and received honoraria from, Pfizer, Astellas, and Dava Oncology; is a member of data safety monitoring committees for trials funded by Gilead and AbbVie; and served as section editor for UptoDate.