Acute myeloid leukemia: 2013 update on risk-stratification and management


  • Elihu H. Estey

    Corresponding author
    1. Division of Hematology, University of Washington
    2. Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA
    Search for more papers by this author

  • Conflict of interest: Nothing to report


Disease overview

Acute myeloid leukemia (AML) results from accumulation of abnormal blasts in the marrow. These cells interfere with normal hematopoiesis, can escape into the peripheral blood, and infiltrate CSF and lung. It is likely that many different mutations, epigenetic aberrations, or abnormalities in micro RNA expression can produce the same morphologic disease with these differences responsible for the very variable response to therapy, which is AMLs principal feature.


This rests on demonstration that the marrow or blood has > 20% blasts of myeloid lineage. Blast lineage is assessed by multiparameter flow cytometry, with CD33 and CD13 being surface markers typically expressed by myeloid blasts. It should be realized that clinical/prognostic considerations, not the blast % per se, should be the main factor determining how a patient is treated.

Risk Stratification

Two features determine risk: the probability of treatment-related mortality (TRM) and, more important, even in patients aged >75 with Zubrod performance status 1, the probability of resistance to standard therapy despite not incurring TRM. The chief predictor of resistance is cytogenetics, with a monosomal karyotype (MK) denoting the disease is essentially incurable with standard therapy even if followed by a standard allogeneic transplant (HCT). The most common cytogentic finding is a normal karyotype(NK) and those of such patients with an NPM1 mutation but no FLT3 internal tandem duplication (ITD), or with a CEBPA mutation, have a prognosis similar to that of patients with the most favorable cytogenetics (inv 16 or t[8;21]) (60–70% cure rate). In contrast, NK patients with a FLT3 ITD have only a 30–40% chance of cure even after HCT. Accordingly analyses of NPM1, FLT3, and CEBPA should be part of routine evaluation, much as is cytogenetics. Risk is best assessed considering several variables simultaneously rather than, for example, only age. Increasing evidence indicates that other mutations and abnormalities in microRNA (miRNA) expression also affect resistance as do post treatment factors, in particular the presence of minimal residual disease. These newer mutations and MRD are discussed in this update.

Risk-adapted therapy

Patients with inv (16) or t(8;21) or who are NPM1+/FLT3ITD—can receive standard therapy (daunorubicin + cytarabine) and should not receive HCT in first CR. It seems likely that use of a daily daunorubicin dose of 90 mg/m2 will further improve outcome in these patients. There appears no reason to use doses of cytarabine > 1 g/m2 (for example bid X 6 days), as opposed to the more commonly used 3 g/m2. Patients with an unfavorable karyotype (particularly MK) are unlikely to benefit from standard therapy (even with dose escalation) and are thus prime candidates for clinical trials of new drugs or new approaches to HCT; the latter should be done in first CR. Patients with intermediate prognoses (for example NK and NPM and FLT3ITD negative) should also receive HCT in first CR and can plausibly receive either investigational or standard induction therapy, with the same prognostic information about standard therapy leading one patient to choose the standard and another an investigational option. This update discusses results with newer agents: quizartinib and crenolanib, gemtuzumab ozogamicin, clofarabine and cladribine, azacitidine and decitabine, volasertib, and means to prevent relapse after allogeneic transplant.

The diagnosis of AML essentially is made as it was in 2012. Thus this review will emphasize new developments in risk stratification and treatment using as references many papers published in 2012. Am. J. Hematol. 88:318–327, 2013. © 2013 Wiley Periodicals, Inc.

Risk Stratification

TRM vs. resistance

It is important to re-emphasize that the major cause of therapeutic failure in AML is resistance to treatment rather than treatment-related mortality (TRM) [1, 2]. Although various criteria for TRM exist, Walter et al.[3] using data from 2,238 MD Anderson Cancer Center (MDA) and 1,127 Southwest Oncology Group (SWOG) patients given 3 days of an anthracycline + either standard doses of ara-C (100 mg/m2 daily X 7, i.e. “3+7”) or higher doses noted a sharp decline in weekly mortality rate once 4 weeks had elapsed from start of induction therapy suggesting that patients who died in these 4 weeks were a distinct group. Using 4-weeks as a criterion they developed (by bootstrapping) a model that predicts TRM. Of note covariates such as performance status, creatinine, albumin, platelet count, WBC and blast count and de novo vs. secondary AML can replace age in the model by with little loss in accuracy. Subsequently using this same dataset Othus et al. [4] found that TRM rates in SWOG and at MDA had declined considerably over time (Table 1): The decline was independent of the younger age and better performance status of patients treated in later years and was observed regardless of age, performance status, or WBC. Improvements in antifungal therapy are the most likely explanation for this general decline. These data suggest that, even in older patients, AML induction treatment should be selected with the primary goal of decreasing resistance rather than decreasing TRM. Resistance manifests either as failure to enter CR despite not incurring TRM or, more commonly, as relapse from CR. Induction therapy can not only affect CR rate, but also relapse rate even after allogeneic hematopoietic cell transplant (HCT) [5] Selection of AML therapy fundamentally entails a choice between standard therapy, such as “3+7”, decitabine, or azacitidine, and a clinical trial. Because the results of a trial are unknown when a patient presents, the choice between standard and investigational largely depends on the likelihood of resistance with standard therapy. While cytogenetics remains the single best predictor of such resistance [6], molecular information has become increasingly important.

Table 1. Decrease in TRM with Time
Pts. TRMPts. TRMPts. TRMPts. TRM
SWOG392 18%406 13%113 12%498 3%
MDA102 16%638 14%732 9%470 4%

European LeukemiaNet (ELN) system verified in independent data set

The 2012 update [7] emphasized the role of mutations (hereafter denoted by “+”) in NPM1, double mutations in CEBPA, and internal tandem duplications (ITD) in FLT3 in forecasting resistance in patients, particularly those with a normal karyotype (NK). In such patients cytogenetic information is much less informative than in patients with inv 16, t(8;21)[collectively CBF AML], a monosomal or a complex karyotype. Thus for example as implied by the ELN classification [8] (Table 1 in 2012 update) the risk of relapse in NPM+/FLT3 ITD wildtype (hereafter denoted by “-“) (or CEBPA++) NK AML is sufficiently low (and similar to that in CBF AML) as to not warrant the potential complications of HCT. The opposite is true in patients with other combinations of NPM1 and FLT3 or in patients who are CEBPA-. Examining 1550 CALGB patients with de novo AML Mrozek et al. [9] have confirmed the independent prognostic significance of the ELN classification (“favorable” vs. intermediate 1 vs. intermediate 2 vs. adverse) in patients aged < 60, although the two intermediate groups had similar outcomes in older patients.

Inclusion of other mutations

The ELN remains the most practical means to assess the probability of resistance and thus NPM, FLT3, and CEBPA status should routinely be assessed much as is cytogenetics. However, it is becoming increasingly clear that other mutations affect the probability of resistance and that the effect of a given mutation depends on the presence/absence of others. Analyzing 657 adults age < 60 with de novo AML given 3+7 (with a daunorubicin dose of 45 or 90mg/m2 as determined by randomization) on ECOG trial E1900, Patel et al. [6] classified patients with NK (or other intermediate prognosis karyotypes) into favorable, unfavorable, or intermediate groups (Fig. 1). The former comprised: NPM+/FLT3 ITD – as previously but only if an IDH1 or IDH2 mutation was present. Unfavorable NK patients were (a) FLT3 ITD− but either TET2, ASXL1 or PHF6+ or with an MLL partial tandem duplication (PTD) or (b) FLT3 ITD+ but only if TET2 or DNMT 3a + without mutant CEBPA, or with MLL-PTD or +8. Other patients were intermediate Contrasted with cytogenetics alone, this system reduces the proportion of patients with an intermediate prognosis to 35% from 63% using cytogenetics alone, thus facilitating the choice between standard (for “favorable” risk) vs. investigational (for unfavorable risk) therapy.

Figure 1.

Integrated genetic analysis to improve risk stratification in de novo AML. [Color figure can be viewed in the online issue, which is available at]

When defining “unfavorable” risk groups, most prognostic systems typically imply what not to do (give standard therapy) rather than what to do. While the prognostic import of the Patel et al. system [6] awaits verification, the authors found that use of the 90 mg/m2 daily daunorubicin dose improved outcome in patients with DNMT3a mutations, MLL-PTD, or NPM mutations. While this finding did not reflect a multivariate analysis it is of conceptual interest because dose escalation in AML has generally been found to only benefit already favorable patients (unlike those with DNMT3a mutations) [10]. Furthermore, DNMT3a, MLL-PTDs, and NPM mutations were mutually exclusive, suggesting a biologic basis for sensitivity to higher daunorubicin doses [6].

Beyond mutations

The effect of a given mutation (IDH is an example) is inconstant across different reports. One possible explanation is that the effect of a given mutation may depend on the presence/absence of other mutations, as noted above. A second explanation invokes differences in the “downstream” effects of a genetic abnormality, and in particular whether the abnormality is “expressed” and eventually translated into protein. Expression is regulated by DNA methylation patterns in AML blasts, as well as by expression of various micro RNAs (miRNAs). miRNAs hybridize to target messenger RNAs (mRNAs) thereby preventing their translation into protein. Studying miR-155 expression in 363 patients with newly-diagnosed NK AML treated on various Cancer and Acute Leukemia Group B (CALGB) protocols generally employing standard therapy but without HCT in first CR, Marcucci et al. [11] found that, with a median follow-up of 7 years, higher expression was associated with an approximately 50% reduction in odds of CR, a 60% increase in probability of death, and in patients under age 60, a 2-fold reduction in disease-free survival, a reliable measure of therapeutic resistance. The patients studied for miR-155 expression had similar outcomes as patients treated on the same CALGB protocols but not studied, suggesting the former were a representative population. The effect of miR-155 expression was independent of many of the genetic abnormalities noted previously. Finally, the deleterious effect of higher expression was limited to the ELN's favorable group in younger patients (age < 60) and to its favorable and intermediate-1 group in older patients. It is not difficult to foresee a time when AML patients will be tested both for a range of genetic abnormalities, such as mutations, and relevant to the abnormalities' downstream effects, profiled for DNA methylation, mRNA, miRNA and protein expression (“proteomics”). Not only will the ability to predict resistance after standard therapy likely be improved but, the discovery of new molecular markers may suggest new therapeutic targets. For example, as noted by Marcucci et al. [11], inhibitors of miRNAs will soon be available to overcome the effect of miR-155; these authors have also suggested that higher levels of miR-29b predict response to decitabine [12].

How practical is this?

Results of some molecular tests do not become available for days-weeks contrasting with a common belief that treatment of AML must begin sooner, even in patients who present with stable WBC < 50,000 However, based on data from 599 patients receiving "intensive" chemotherapy for a new diagnosis of AML, Bertoli at al. reported that, after accounting for standard prognostic covariates, time from diagnosis to treatment (TDT, median 8 days, patients with > 90 days excluded) had no effect on survival, CR, or early death rates [13]. The same was true as true both in patients age < and > 60 and in patients with WBC < vs. > 50,000. It is of course plausible that patients with longer TDTs were inherently more favorable than patients with shorter TDTs, leading to the decision that treatment could be delayed and leading to the similar outcomes in patients with longer and shorter TDTs. However since a randomized trial to address this possibility is unlikely to be conducted, Bertoli et al.'s data suggest that delay in initiation of therapy to await results of various molecular tests is feasible. Although initiation of “investigational therapy” in patients whose molecular results augur a poor outcome with standard therapy might await achievement of remission with the latter, in some patients the chance of CR with standard therapy is <50% and induction therapy can influence length of CR as well as its achievement [5].

Minimal residual disease (MRD)

While incorporation of the pre-treatment molecular information described above will likely improve prognostic accuracy information obtained post treatment will almost certainly also be relevant. Of particular note are assessments of MRD in patients in CR by standard criteria (given in reference 8). MRD can be determined by PCR to detect (a) leukemia fusion genes (such as those characteristic of inv 16, t(8;21), or t(15;17)), (b) mutations, (for example in NPM1), or (c) over expression of genes such as WT1. Substantial data suggest that these techniques can distinguish patients at different risks of relapse although nominally still in CR [14, 15].

MRD can also be examined by multiparameter flow cytometry (MPFC), which relies on identification of patterns of cell surface antigens characteristic of the patient's AML. Buccisano et al. have shown that patients with “good” (inv 16, t8;21) or intermediate karyotypes who have MRD detected by MPFC (MRD+) at the end of post remission therapy have outcomes resembling those in patients with FLT3 ITD+ or unfavorable karyotype AML [16] (Fig. 2). In turn patients who are FLT3ITD + but without MRD fare better than those who are FLT3 ITD + with MRD (Fig. 2). Walter et al. have reported that MPFC detectable MRD prior to HCT is independently associated with relapse after HCT [17]. Since MRD prior to HCT reflects ineffectiveness of chemotherapy prior to HCT these results suggest that ineffective chemotherapy begets ineffective HCT, although the effect pre HCT of a given increment in MPFC detectable MRD seems greater, at least in children, in ALL than in AML [18].

Figure 2.

MRD at completion of consolidation therapy to improve risk stratification. [Color figure can be viewed in the online issue, which is available at]

The sensitivity of MRD monitoring motivates several questions. In patients where both MPFC and PCR are applicable which should be used [19]? Can blood be used rather than bone marrow? Paietta for example has suggested that blood would be appropriate in cases where relapse is likely to be delayed, for example in CBF AML [20] Should MRD be assessed at multiple time points, for example at CR, immediately after completion of post remission therapy and beyond? What levels of MRD should motivate a change in therapy? Are levels reproducible at different centers? And most importantly will reduction in MRD lead to better relapse free or overall survival or is MRD merely a surrogate of refractory AML? Rubnitz et al. administered a 2nd course of induction therapy early to children with MRD as assessed by MPFC and included gemtuzumab ozogamicin (GO) on this course in such children. Children with MRD after three courses were assigned to HCT regardless of cytogenetic risk and the availability of a matched sibling donor [21]. Plausibly as a result of these interventions results were superior to those observed in historical controls. Analogous results have been reported in adults [16]. Particularly valuable information is likely to come from the MRC/NCRI AML17 study in the UK in which patients are being randomized to MRD monitoring or not.


Better prognosis patients- those without “unfavorable” cytogenetic abnormalities or FLT3 ITD [8]


GO combines an antibody directed against the cell surface antigen CD33 with the toxin calechiamicin. The US Food and Drug Administration (FDA) approved GO for treatment of relapsed/refractory AML in adults aged > 60 but mandated randomized trials. A US + Canada Intergroup trial (SO106) randomized 506 newly diagnosed patients under age 60 to receive or not receive GO (6 mg/m2 day 4) during “3+7” induction, although with only 45mg/m2 daunorubicin in the no GO arm vs. 60mg/m2 in the GO arm, and re-randomized patients in remission to GO during post-remission chemotherapy. Addition of GO in either induction or post-remission setting had no overall effect on CR rate, response duration, or survival (OS) but was associated with an excess of deaths within 30 days of starting chemotherapy [22]. These results prompted withdrawal of GO. However this decision appears debatable in light of three published randomized trials. The MRC/NCRI AML15 trial randomized 1113 adults, 86% under age 60, with newly diagnosed AML to receive one of 3 induction regimens of varying intensity +/− GO (3mg/m2 day 1) [23]. Patients were re-randomized to receive GO on their first course of post-remission therapy. While, there were no differences in CR rate, relapse-free survival (RFS), or OS, much as in SO106, an internally validated prognostic index indicated that approximately 75% of patients would live longer if given GO. Included were all patients with CBF AML (inv 16 or t8; 21), many with “intermediate prognosis” cytogenetics, but none with adverse cytogenetics. The effect of induction GO was similar regardless of chemotherapy regimen, suggesting that the drug was not a non-specific means of dose intensification and regardless of randomization to GO post-remission. Randomizing 1115 newly-diagnosed older patients (median age 67, 98% age > 60) considered “fit for intensive chemotherapy” to one of two induction chemotherapy regimens each+/− GO the MRC/NCRI AML16 trial found similar CR rates but lower cumulative incidence of relapse (CIR) and superior RFS and OS with GO again regardless of induction chemotherapy received and again with intermediate cytogenetics deriving more OS benefit than adverse cytogenetics (17% and 5% reductions in risk of death respectively) [24]. Figure 3 depicts a meta-analysis of the MRC/NCRI AML 15 (younger patients) and 16 (older patients) trials The third randomized trial assigned 280 patients aged 50–70 (median 62) to receive three courses (the 1st to induce remission, the 2nd and 3rd to “consolidate” it) of daunorubicin + cytarabine +/− GO (3 mg/m2 day 1,4,7 induction, d1 consolidation). Although remission rates were similar in the 2 arms, the GO group had better RFS and OS, and as in AML 15, benefit was observed in patients with “favorable” or intermediate, but not adverse cytogenetics [25]. In none of these trials was there an increase in 30-day death rate. Furthermore because AML is, as indicated in the “Risk Stratification” section, in reality several diseases defined by cytogenetic and, molecular characteristics, it appears disingenuous to assume that a beneficial therapy will be equally effective in all types of AML, or, as a corollary, will necessarily show an overall survival benefit. Yet such an assumption appears implicit in the decision to withdraw GO. Accordingly several papers have been written to encourage the drug's re-introduction [26, 27].

Figure 3.

Meta-analysis of NCRI/MRC AML 15 (younger patients) and AML 16 (older patients) Trials. [Color figure can be viewed in the online issue, which is available at]

Arsenic trioxide (ATO) + all-transretinoic acid (ATRA) without chemotherapy for newly diagnosed APL

Conventional treatment (anthracycline + ATRA) for APL is usually reported to cure >80% of patients, although, because of early death, this may occur in only 60–70% of patients in community settings [28]. To dispense with chemotherapy MD Anderson investigators administered, in a single-arm trial, only ATO+ATRA to patients presenting with WBC < 10,000 [29]. The seeming ability to obtain results similar to those with chemotherapy led to a 159 patient Italian-German randomized trial comparing the MD Anderson regimen to idarubicin + ATRA in patients aged < 70 with WBC < 10,000.With a median follow-up of 34 months, and as presented at the plenary session of the 2012 American Society of Hematology (ASH) meeting event-free survival (EFS), the study's primary endpoint, was superior with ATO+ATRA (Fig. 4) as was OS [30]. Given that it has less hematologic toxicity it is plausible that ATO+ATRA without chemotherapy will replace anthracyclines + ATRA in APL patients with WBC < 10,000, particularly if, as seems likely, oral ATO becomes available.

Figure 4.

Event-free survival in APL; Idarubicin + ATRA vs. ATO + ATRA. [Color figure can be viewed in the online issue, which is available at]

Worse prognosis patients

FLT3 aberrations

Quizartinib (formerly AC220) is the most potent available FLT3 ITD inhibitor and, although inhibiting CKIT, is more selective than other inhibitors such as midostaurin, lestuartinib, or sorafenib. Results of two single arm quizatinib trials were presented at the 2012 ASH meeting: one enrolled 92 FLT3ITD + patients age ≥ 60 receiving the drug as 1st salvage therapy for refractory or relapsed AML [31], the second enrolled 99 FLT3ITD+ patients age ≥ 18 receiving the drug as 2nd salvage [32]. The “composite” CR rate (including CR, CRp, and CRi) was 54% in the first study and 44% in the second, with CRi denoting <5% marrow blasts but an absolute neutrophil count < 1000 and, typically, platelet count < 100,000. The great majority of responses were such CRi (0 CR, 3% CRp in the first, 4% CR, 0CRp in the second study). Median survival was about 6 months in both studies. A principal objective was to see whether some responding patients might subsequently receive HCT resulting in prolonged OS and indeed 23% of patients (34% in the second study) received HCT. Figure 5 shows OS in the latter; it is not clear whether account was taken of the time needed to be alive before HCT could be done (“guarantee time”) or how these results compare to those in FLT3TD+ patients receiving HCT at such a late stage of their disease. The failure of count recovery in quizartinib- induced CRi may reflect persistent AML or a CKIT-mediated effect on normal hematopoiesis. An ongoing study using lower doses of quizartinib in an effort to decrease the incidence of asymptomatic QTc interval prolongation may help answer the question. Trials combining quizartinib with standard chemotherapy and to prevent relapse post HCT have also been initiated. Crenolanib is a FLT3 ITD inhibitor that while less active vs. FLT3 ITDs than quizartinib is active vs. the D835 mutation commonly seen at treatment failure after quizartinib [33]. Hence a combination of quiazartinib + crenolanib is of interest. Sorafenib is the only FLT3 ITD inhibitor that is currently available. A randomized trial (276 patients) presented at ASH 2012 noted that addition of sorafenib to standard chemotherapy in patients aged <60 did not affect CR rate, improved 1-year EFS (64% vs. 50% median follow-up 18 months) to greater extent than OS (72% vs. 66%) [34]; a breakdown by FLT3 status was not provided.

Figure 5.

OS in patients receiving or not receiving HCT after quizartinib. [Color figure can be viewed in the online issue, which is available at]


Several randomized trials have not found that this nucleoside analog more effective than standard therapy. The CLASSIC1 trial [35] involved 326 patients aged > 50 and approximately equally divided between relapsed and primary refractory assigned to “high-dose” ara-C (HiDAC: 1g/m2 days 1–5) +/− clofarabine 40mg/m2 days 1–5. While CR rate was superior with clofarabine (35% vs. 23%) RFS was not and together with a higher TRM rate (16% vs. 5%) resulted in similar OS (medians 6.6 and 6.3 months). Application of the TRM model [3] might identify patients less likely to incur TRM and hence more likely to have improved OS with clofarabine + HiDAC. The intensive arm of the MRC/NCRI AML16 trial randomized 806 older patients considered “fit for intensive therapy” between daunoubicin (50mg/m2 an days 1–3) together with either clofarabine (20 mg/m2 on days 1–3) or ara-C (100 mg/m2 on days 1–10) [36]. CR rates (66–71%), 3- year CIR (68–74%) and OS (22–23%) did not differ between the arms nor was there a suggestion that certain groups, for example as defined by cyogenetics or concurrent receipt of GO, benefited more from one arm than the other. AML16 randomized 406 patients (median age 74) considered unfit for intensive therapy between low dose ara-C (LDAC) and clofarabine (20mg/m2 days 1–5) [37]. While clofarabine produced a higher CR rate (22% vs. 12%), OS in patients not achieving CR and OS from relapse were both better with LDAC leading to similar OS in the two arms, again with no interactions between treatment and covariates such as cytogenetics or administration of GO.


A Polish study randomized 652 patients aged < 60 among DA)(ara-C 200 mg/m2 Days 1–7 + daunorubicin 60 mg/m2 Days 1–3) DAF (DA + fludarabine) or DAC (DA + cladribine (5 mg/m2 Days 1–5). CR was more common with DAC than DA (68% vs. 56%) as was CR after the first course (62% vs. 51%), reflecting less resistance and translating into improved survival despite similar CR duration (Fig. 6) [38]. The effect of cladribine was most obvious in patients age 50–60 with “unfavorable” cytogenetics and seemed independent of HCT. The applicability of the study to general practice can be criticized because of the low CR rate in the DA arm perhaps reflecting a relatively long interval (5–7 weeks) between the first and a 2nd course of chemotherapy with the latter making a relatively small contribution to overall CR rate [39]. Nonetheless the results mark cladribine, which is considerably less costly than clofarabine as an interesting drug in AML.

Figure 6.

OS and CR duration in patients given daunorubicin + Ara-C (DA), DA+ fludarabine, or DA + cladribine. [Color figure can be viewed in the online issue, which is available at]

Azacitidine and Decitabine

It is becoming apparent that results with these drugs used alone rival those with “more intense” therapies, at least in older patients where azacitidine and decitabine are used most commonly. For example Quintas-Cardama et al. found similar OS among 557 patients aged ≥ 65 given regimens generally containing ara-C at 1–2 g/m2 daily + idarubicin or fludarabine ± other agents and 114 patients given decitabine (n = 67) or azacitidine (n = 47) despite higher CR rates with the more intense therapies [40] (Fig. 7), thus again, as in the AML16 clofarabine trial [37] questioning the role of CR. Results remained the same after accounting for known prognostic covariates, although it is not clear whether these included year of treatment [4], More formally an international study randomizing 485 patients age ≥ 65 found median survival of 7.7 months for decitabine 20 mg/m2 daily X 5 vs. 5.0 months for “treatment choice” (88% low dose ara-C, 12% supportive care only) [41]. A focus on whether the difference is “statistically significant” (P = 0.037 on an unplanned analysis after the planned number of events had occurred) loses sight of the unsatisfactory results in both arms, with patients losing >90% of their normally remaining life expectancy. An analogous trial involving azacitidine leads to the same conclusion [42]. Thus a fundamental question is whether results with either drug can be improved. One possibility is identification of patients more likely to respond either based on molecular characteristics [12] or less active disease. Examples of the latter are patients in CR, although use of decitabine to delay relapse appears to have been unsuccessful [43], or patients with MRD after HCT (see below). A second possibility is combination with other drugs, although attempts to replace ara-C with azacitidine in the ICE regimen for AML induction were unsuccessful [44] A combination of decitabine with deoxyguanosine (SGI-110) increases exposure to decitabine and has been noted to produce responses in AML or MDS unresponsive to decitabine [45].

Figure 7.

OS and RFS according to type of therapy received: azacitidine or decitabine (epigenetic) or “more intensive” (chemo). [Color figure can be viewed in the online issue, which is available at]


This is a polo-like kinase inhibitor that was combined with LDAC and, in a randomized trial, compared with LDAC alone in 87 patients (median age 75, 1/3 with unfavorable cytogenetics) considered unfit for intensive therapy. CR +CRi rate were higher with the combination (31% vs. 11%) as was EFS (P = 0.08, medians of 69 vs. 169 days) [46].


Peripheral blood vs. marrow as source of cells from unrelated donors

The majority of HCT currently use cells from matched unrelated donors. The ease with which cells can be collected from peripheral blood (PB) has led to this being the most common source of donor cells. A multicenter study randomized 551 patients (approximately half with AML of whom half were in 1st CR) to receive cells from unrelated donors' PB or bone marrow [47]. There were no differences in OS with graft failure less likely in patients receiving PB who were however more likely to have chronic graft vs. host disease (GVHD). As discussed in the editorial accompanying the paper [48], these data might motivate a change in practice such that PB would be used only in those patients with a need for rapid engraftment, for example those with severe infections.

Prevention of relapse

Relapse remains the most common cause of failure after HCT. In addition to use of different preparative regimens attempts to prevent relapse have focused on the post-HCT setting. Examples include (a) various cellular therapies to augment the graft-vs.- leukemia (GVL) effect without simultaneously worsening acute GVHD [49], (b) quizartinib in FLT3ITD+ AML pre HCT and (c) azacitidine either as prophylaxis in high risk patients [50], in patients with MRD, or in patients in whom decreasing proportions of CD34 + donor cells signal decreasing GVL and portend relapse [51].

Summary of Treatment Recommendations

These have not changed materially since the 2012 update [7]. The prognostic system noted in table 2 continues based on cytogenetics, NPM, FLT3, and CEBPA. Patients with the best prognoses can be treated with standard therapy emphasizing higher doses of ara-C post CR (HiDAC) [10], although (a) it remains unknown if the same is true in patients considered best risk based on NPM+/FLT3− rather than cytogenetics and (b) the exact dose of HiDAC is not clear; however, doses of 1.0−1.5 g/m2 seems as effective as 3 g/m2 [52]. Patients who have inv [16] or t(8;21) together with a CKIT mutation have worse prognoses than patients without a CKIT mutation [7] and thus might be candidates for HCT from a matched sibling donor or for clinical trials involving the CKIT inhibitor dasatinib. The choice of 3+7 rather than a clinical trial for induction in the intermediate-1 and intermediate-2 groups depends on whether patients see their prognosis with the former as good enough to be reluctant to undertake a clinical trial, which might produce a worse outcome. I would however strongly recommend a trial combining a FLT3 inhibitor such as quizartinib with standard therapy in the intermediate-2 (FLT3 ITD+) group. In contrast CR rates are <50% in the intermediate and worst groups (even in younger patients) with standard therapy making a clinical trial particularly important and recalling that a monosomal karyotype is the single most important negative prognostic covariate following standard therapy even following HCT [53, 54]. The general consensus is that patients in the intermediate 1, 2, 3 and worst groups should receive HCT in first CR. However in view of the greater likelihood of chronic GVHD with an unrelated donor many would reserve unrelated donor HCT only for patients in the int-2, int-3 and worst groups. Results with HCT are still sufficiently poor that clinical trials involving new preparative regimens or means to prevent post HCT relapse should be considered in these 3 groups as should a clinical trial if in patients in whom HCT is not feasible [55].

Table 2. Summary of Treatment Recommendations
Prognostic groupSubsetsInductionPost-Remission
Bestinv (16 or t(16;16); t(8;21) NK with NPM1 mutation and no FLT3 ITD; NK with double mutated CEBPA3+7Ara-C at 1 g/m2 BID daily X 6 [52];
 Dasatinib in clinical trial if inv 16, or t(8;21) with CKIT mutated
Inter-mediate 1NK w/o NPM mutation or FLT3 ITD;3+7;HCT from matched sibling donor (MSD);
Cytogenetic abnormalities other than best or unfavorableClinical trialAra-C as above or clinical trial (e.g. CSL362) [53] if not HCT candidate
Inter-mediate 2FLT3-ITD+3+7;HCT from MSD or matched unrelated donor; consider trial with quizartinib post HCT; Ara-C as above or clinical trial (e.g. CSL362) [55] if not HCT candidate
 Clinical trial involving FLT3 inhibitor, e.g. quizartinib (see text)
Inter-mediate 3Unfavorable cytogenetics without monosomal karyotypeClinical trialHCT from MSD or matched unrelated donor; consider trial involving new preparative regimen or means to prevent relapse after HCT (see text); clinical trial if not HCT candidate
WorstMonosomal karyotype [54, 55]Clinical trialAs in intermediate −3

Dose reductions or use of decitabine or azacitidine rather than 3+7 should be based not only on age but on other covariates such performance status and others described in Ref. [3].