Salvage Therapy for Relapsed or Refractory Acute Myeloid Leukemia

Abstract

There are a significant number of patients diagnosed with acute leukemia who either fail to achieve remission or who relapse thereafter. Challenges in treating this patient population include accurately assessing prognosis of disease and whether remission can be achieved; assessing the ability of patients to tolerate aggressive salvage therapies; choosing a salvage therapy that is most likely to succeed; and identifying suitable patients for hematopoietic stem cell transplantation. Despite the development of a variety of new investigational therapies, relapsed or refractory acute myeloid Leukemia remains a difficult clinical problem. Clinicians will need to consider all currently available approaches, including cytotoxic chemotherapy, targeted agents, and allogeneic stem cell transplantation, to optimize outcomes.

Introduction

According to the Surveillance, Epidemiology and End Results (SEER) database, approximately 12,300 adults in the United States will be diagnosed with acute myeloid leukemia (AML) in 2010. Of patients who are fit enough to receive standard induction therapy, accumulated data demonstrate that about 60-80% of younger adults and 40-50% of older adults achieve a complete remission, leaving a substantial population of surviving patients who are refractory to initial induction therapy. Patients whose disease does not respond to the first cycle of induction chemotherapy are sometimes placed in the refractory category. However a retrospective analysis of six Eastern Cooperative Oncology Group (ECOG) studies that included both younger and older adults demonstrated that 26% of patients achieving complete remission (CR) following anthracycline- and cytarabine-based induction therapy required a second cycle of identical induction therapy to do so [Rowe et al. 2010]. For our purposes, we will consider refractory disease to be disease that did not respond to up to two cycles of first-line induction therapy.

There is an additional large population of patients whose disease relapses after achieving first CR. In a study of 1069 patients who achieved first CR at MD Anderson Cancer Center between 1991 and 2003 and did not undergo allogeneic stem cell transplantation (SCT) at that time, the probability of relapse-free survival at 3 years was only 29% [Yanada et al. 2007]. The patients had a median age of 55 years and included 22% with favorable cytogenetics, 64% with intermediate risk cytogenetics, and 14% with adverse cytogenetics. Younger age and more favorable karyotype were associated with significantly increased rates of relapse-free survival at 1 year.

While there is the potential for long-term disease-free survival (DFS) for some patients with relapsed or refractory disease treated with chemotherapy alone, it is thought that prolonged DFS is more likely with hematopoietic SCT (HSCT). Although HSCT in early first relapse may be successful in some patients, identifying suitable patients and proceeding to transplant in a timely fashion usually makes such an approach unfeasible. Therefore when treating patients with relapsed or refractory disease some of the challenges include accurately assessing prognosis of disease and whether remission can be achieved; assessing the ability of patients to tolerate aggressive salvage therapies; choosing a salvage therapy that is most likely to succeed; and identifying suitable patients for HSCT.

Prognostic factors for remission following salvage therapy

Achieving a first CR in patients whose disease has not responded adequately to standard induction regimens or achieving a second CR (CR2) in patients whose disease has relapsed present difficult therapeutic challenges. Although there can be considerable heterogeneity in patients, factors have been identified that are of prognostic significance. Data clearly show that patients whose first CR lasted longer than 1 year were more likely to achieve CR2: CR2 rates of 60% in these patients have been reported [Keating et al. 1989]. Conversely, CR2 rates of less than 20% were typical when the duration of the first CR was less than 6 months (reviewed by Estey and Craddock and colleagues) [Craddock et al. 2005; Estey, 2000]. Several investigators have devised systems to identify prognostic factors associated with decreased survival in relapse (Table 1). The Dutch—Belgian and Swiss cooperative groups defined the European Prognostic Index (EPI) for patients with AML aged 15-60 in first relapse [Breems et al. 2005]. The EPI was based on a multivariate analysis of 667 young adult patients with AML in first relapse and identified four clinically relevant adverse parameters: older age; shorter relapse-free interval after first CR; unfavorable karyotype at the time of original diagnosis; and HSCT prior to first relapse. Three risk groups were defined: a favorable group with overall survival (OS) of 70% at 1 year and 46% at 5 years; an intermediate risk group with OS of 38% at 1 year and 12% at 5 years; and a poor risk group with OS of 17% at 1 year and 6% at 5 years. Cytogenetics and relapse-free interval were the two factors that carried the greatest weight and even normal cytogenetics carried significant adverse prognostic significance. The EPI was subsequently validated in a cohort of 599 patients aged 60 years or younger treated at the MD Anderson Cancer Center [Giles et al. 2006]. More recently, the impact of fms-like tyrosine kinase 3 (FLT3) mutations in relapsed disease was analyzed. Patients with the FLT3 internal tandem duplication (ITD) mutation were found to have a shorter OS, consistent with the known adverse impact of FLT3 ITD mutations on OS at diagnosis [Ravandi et al. 2010].

Table 1.

Prognostic factors associated with decreased survival in relapse.

Study	Prognostic factor	Number of patients	1 year OS (%)	p value
[Breems et al. 2005]	CR ≤ 6 months	299	14	<0.000001
	CR 7–18 months	270	36
	CR > 18 months	98	57
	Age ≤ 35	172	36	0.00014
	Age 36–45	151	30
	Age ≥45	359	25
	t (16;16) or inversion 16	33	72	<0.000001
	t (8;21)	29	54
	Intermediate cytogenetics	422	25
	Adverse cytogenetics	96	19
	No prior SCT	507	31	0.0032
	Previous auto SCT	102	21
	Previous allo SCT	58	22
[Ravandi et al. 2010]	FLT3 wild type	80	<40%	<0.0001
	FLT3 ITD mutation	47	<20%

allo, allogeneic; auto, autologous; CR, complete remission; FLT3, Fms-like tyrosine kinase 3; ITD, internal tandem duplication; OS, overall survival; SCT, stem cell transplant.

Which salvage therapy? A look at standard agents

Once the decision to proceed with salvage therapy has been made, the next challenge is choosing the salvage regimen. There have been few randomized trials comparing salvage regimens in AML. There is therefore no clear evidence of superiority of any regimen and choice of salvage regimens is often based on clinician preference. Cytarabine (Ara-C) has long been a mainstay of salvage therapy in AML and a review of the AraC literature gives an idea of the kind of results that are typical in trials of salvage regimens (Table 2). Published data on Ara-C as salvage therapy in AML go back to the 1980s. An early study using 3 g m² of cytarabine for 12 doses reported CR rates of about 60% in a population of younger adults (median age 37 years) with relapsed disease [Herzig et al., 1985].

Table 2.

Cytarabine-based salvage regimens in acute myeloid leukemia.

Reference and study design	Regimens	Number of patients	Median age (years)	%CR
[Herzig et al. 1985] nonrandomized	Ara-C versus	78	37	63 versus 65
	Ara-C + DXR or DNR
[Vogler et al. 1994] randomized phase III	HiDAC versus	131	—	31 versus 38
	HiDAC + Etoposide
[Kern et al. 1998] randomized phase III	HiDAC + MTZ versus	186	50	47
	IDAC + MTZ
[Ferrara et al. 1999] nonrandomized	FLAG	26	39	50
[Estey et al. 1993] phase II	Flu, Ara-C	59	52	36
[Pastore et al. 2003] nonrandomized	Flu, Ara-C, Ida and G-CSF	46	41	52
[Martin et al. 2009] nonrandomized	Flu, Ara-C, Ida ± GO with concurrent or sequentiaL G-CSF	71	48	29 (+GO) versus 39 (−GO)
[Litzow et al. 2010] randomized phase II	IDAC + GO versus	82	60	12 versus
	IDAC + liposomal DNR versus		52	7 versus
	Ara-c, CTX, Topotecan		53	4
[Spadea et al. 1993] nonrandomized	MEC*	74	36	55
[Ohno et al. 1994] randomized phase III	MTZ, Etoposide, Ara-C + G-CSF versus	50	43	54 versus 42
	MTZ, Etoposide, Ara-C*		47
[Thomas et al. 1999] randomized phase III	EMA* + GM-CSF versus	192	47	65 versus 59
	EMA		46
[Archimbaud et al. 1995] phase II	EMA	133	43	60

Ara-C, cytarabine; CTX, cyclophosphamide; DNR, daunorubicin; DXR, doxorubicin; EMA, etoposide, mitoxantrone, Ara-C; FLAG, fludarabine, Ara-C, G-CSF; Flu, fludarabine; G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte—macrophage colony stimulating factor; GO, gemtuzumab ozogamicin; HiDAC, high-dose Ara-C; Ida, idarubicin; IDAC, intermediate dose Ara-C; MEC, mitoxantrone, etoposide, Ara-C; MTZ, mitoxantrone.

^*EMA and MEC use the same three drugs but differ in dosing and administration.

Numerous studies investigating the effect of additional agents on the efficacy of a cytarabine-based salvage regimen are also shown in Table 2, including several studies looking at the combination of fludarabine, cytarabine, and granulocyte colony-stimulating factor (G-CSF) (FLAG). The addition of anthracyclines to cytarabine-based therapy has also been investigated, and the combination of mitoxantrone and etoposide, with or without cytarabine, is among the most studied salvage regimens. Notably, however, randomized controlled trials to determine if the notably high CR rates reported in the EMA86 and EMA91 studies [Thomas et al. 1999; Archimbaud et al. 1995] are reproducible have not been conducted. In fact, the two-drug regimen of mitoxantrone and etoposide was shown many years ago to be an effective salvage regimen [Ho et al. 1988] and randomized comparisons between the two- and three-drug regimens have not been performed. Emerging from the literature of the 1980s and 1990s were several effective salvage regimens for relapsed or refractory AML, with Ara-C and anthracycline-like compounds as the backbone of most regimens. Combinations that included fludarabine and/or combinations of both Ara-C and anthracycline-like compounds also demonstrated efficacy [Leopold and Willemze, 2002]. Clinicians may choose one regimen over the other based on time since last cytarabine but this approach has not been validated.

A more recent trend has been to add novel agents to the familiar anthracycline and ara-C backbone. For example, a phase I study was completed at the University of Pennsylvania in which sirolimus, a mammalian target of rapamycin (mTOR) inhibitor, was administered with mitoxantrone, etoposide and Ara-C [Perl et al. 2009]. This was based on preclinical data showing activation of the mTOR pathway in the primary cells of patients with AML and decreased survival of these cells when incubated with sirolimus, and enhanced efficacy of chemotherapy with sirolimus [Xu et al. 2005]. A 30% response rate was noted in the 10 patients treated at the highest dose level in the phase I study. Another variation on the theme is the mitoxantrone, cytarabine, and flavopiridol regimen, which is currently being tested along with mitoxantrone, etoposide, cytarabine plus sirolimus in a randomized phase II trial run by the ECOG. The third arm of the trial involves the combination regimen of carboplatin and topotecan.

A look at newer agents

While the 1980s and 1990s were focused on the development of new chemotherapeutic strategies to improve outcomes in relapsed or refractory AML, the last decade has focused more on targeted therapies, a broad term that encompasses both antibody-mediated and molecular-based therapies. Such therapies have been used alone or in combination with traditional, chemotherapy-based salvage regimens. However, there has been at least one additional chemotherapeutic agent developed in the last decade that has had significant efficacy in relapsed or refractory AML. Clofarabine is a nucleoside analogue that has a structural similarity to both fludarabine and cladribine. Like fludarabine, it inhibits DNA elongation, but in addition it also inhibits ribonucleotide reductase, like cladribine (discussed by Kantarjian and colleagues) [Kantarjian et al. 2003]. Clofarabine was developed at the MD Anderson Cancer Center and was evaluated in a phase II study of relapsed and refractory AML, high-risk myelodysplastic syndrome (MDS), acute lymphoblastic leukemia, or chronic myelogenous leukemia in blast crisis [Kantarjian et al. 2003]. There were 31 patients with AML. Eight of 19 patients in first salvage achieved CR with a dose of 40 mg m² for 5 days, and five of 12 patients undergoing second or subsequent salvage achieved a CR. The most notable toxicity was grade 3–4 transaminitis, which occurred transiently in 15% of patients. Skin rashes and hand-foot syndrome were observed in 10-15% of patients. There is currently an ongoing phase II study to compare clofarabine, idarubicin and cytarabine versus clofarabine and idarubicin in relapsed or refractory AML [Faderl et al. 2008]. In addition, the results of a phase II study of clofarabine and high-dose cytarabine with G-CSF priming have been reported [Becker et al. 2009]. In 38 patients with relapsed or refractory AML, the CR plus CR without platelet recovery rate was 64%. Also of interest is a small study from the University of Chicago that showed clofarabine could be used effectively in association with allogeneic SCT [Locke et al. 2010]. Seventeen patients with relapsed or refractory AML received clofarabine with a plan to initiate conditioning for SCT at nadir. Effective cytoreduction (<20% cellularity in bone marrow with <10% blasts by day 12) occurred in 10 of 17 patients, and 16 patients went on to allogeneic SCT. Two-year transplant-related mortality was 36% and nine of 16 patients relapsed after SCT. There are plans to study this approach in a forthcoming trial.

Other investigational agents

There are a number of other classes of agents outside the realm of traditional cytotoxic chemotherapy that are being tested alone or in combination with chemotherapy, including trials in previously untreated AML. DNA methyltransferase inhibitors in relapsed or refractory AML have for the most part been disappointing. A large French study of 184 patients with relapsed or refractory disease who were treated with azacytadine only demonstrated a CR rate of only 7% [Itzykson et al. 2009]. A retrospective study looking at the combination of decitabine with the conjugated antibody gemtuzumab ozogamicin in 79 patients with relapsed or refractory AML showed a CR rate of 16% [Ritchie et al. 2009]. Some authors have indicated that this would be expected from gemtuzumab ozogamicin alone, raising the question of whether decitabine is adding any efficacy at all [Zhu et al. 2010]. Gemtuzumab ozogamicin, which is a recombinant humanized antibody to CD33 linked to calicheamicin, showed early promise in previously untreated older patients and was subsequently studied in the refractory setting [reviewed in Pagano et al. 2007; Sievers et al. 1999]. However, this agent was withdrawn from the market in 2010 because of toxicity concerns.

Intriguingly, a recent case series from Johns Hopkins University suggested that azacytadine may have activity in relapsed disease following allogeneic SCT [Bolanos-Meade et al. 2010]. Ten patients with myeloid malignancies who relapsed after allogeneic HSCT received singleagent azacytadine. Six patients achieved CR, one had stable disease, and in three disease progressed after a median of six cycles of azacytadine. Nine of the ten patients were alive with a median OS of over 400 days at the most recent follow up.

The success of DNA methyltransferase inhibitors in high-risk MDS has generated interest in testing other epigenetic modifiers in AML. For example, the histone deacetylase inhibitors have been studied in combination with azacytadine [Silverman et al. 2008; Garcia-Manero et al. 2007].

In addition, there has been interest in expanding the role of lenalidomide to include treatment of AML. Lenalidomide has been used with great success in multiple myeloma, but its use in myeloid malignancies has been largely limited to lower-risk MDS associated with the 5q— cytogenetic abnormality. Its mechanism of action is unclear but it may have pleiotropic effects on cell growth, viability, cytokine production, and the immune system [Kotla et al. 2009]. In a newly reported phase I dose-escalation study of 31 patients with relapsed or refractory AML (median age 63), a CR rate of 16% was achieved [Blum et al. 2010]. The patients who achieved CR did not have the 5q— cytogenetic abnormality and responses lasted up to 14 months. Two of four patients who were treated with lenalidomide following relapse after allogeneic HSCT achieved CR. These patients did not receive donor leukocyte infusions but did develop cutaneous graft versus host disease. The highest dose tested was 50 mg daily, which is the same dose that was tested in a recently reported phase II study of lenalidomide as frontline therapy for older patients with AML [Fehniger et al. 2010].

Other pathways of interest include the retinoic acid X receptor signaling pathway, and at the authors’ institution, a phase I study has recently been completed in refractory or relapsed non-M3 AML using bexarotene, a retinoic acid X receptor agonist [Tsai et al. 2008]. In 27 patients with a median age of 69 years, 15% achieved bone marrow blast percentages of less than 5%. A phase II trial has been opened.

Therapies for FLT3 internal tandem duplication positive disease

The recognition that FLT3 ITD mutations confer poor prognosis in AML previously characterized as intermediate risk is an important development in the past decade. Numerous studies have confirmed the adverse prognostic impact of the FLT3 ITD mutation on cytogenetically normal AML (reviewed by Small) [Small, 2006]. There is also evidence that the poor prognosis of the FLT3 ITD mutation holds at the time of first relapse. In a study of 80 FLT3 wild type (WT) patients and 47 patients with FLT3 ITD mutations, the CR rates after first relapse were 41% and 24% respectively for patients receiving salvage chemotherapy, and almost all of these patients received an intensive salvage regimen or allogeneic transplant [Ravandi et al. 2010]. The OS from time of first relapse was 37 weeks for FLT3 WT patients and 13 weeks for patients with FLT3 ITD mutations. These observations have naturally brought about the development of agents designed to target FLT3 for therapeutic benefit. These agents have been tested both as monotherapy and in combination with conventional chemotherapy. Sorafenib and sunitinib were the first kinase inhibitors with significant activity against FLT3 to be tried in the clinic (reviewed by Wiernik) [Wiernik, 2010]. Other agents active against FLT3 include midostaurin, lestaurtinib, KW-2449, and AC220. Studies involving FLT3 inhibitors are presented in Table 3.

Table 3.

FLT3 inhibitors as salvage therapy in acute myeloid leukemia.

Reference and study design	Regimen	Number of patients	Median age (years)	Relevant results
[Crump et al. 2010] phase I	Sorafenib	42	71	300 twice daily maximally tolerated dose. One CR in a FLT3 ITD + patient. Untreated MDS or secondary AML
[Metzelder et al. 2009a] nonrandomized	Sorafenib	6	–	Two survived >200 days after relapse following allo transplant. Two more patients remained in molecular remission at time of publication
[Metzelder et al. 2009b] nonrandomized	Sorafenib	18	–	Relapsed/refractory AML. 16/18 patients cleared peripheral blasts. 39% developed resistance by 180 days.
[Schroeder et al. 2009] nonrandomized	Sorafenib	4	–	All had relapsed after allo transplant. Two patients achieved CR, including one molecular CR.
[Fischer et al. 2010] phase IIb	Midostaurin (PKC412)	95	>60	Relapsed/refractory/unfit. 35 patients with FLT3 mutations. 71% of these patients had BM blasts reduced by > 50%
[Levis et al. 2009] phase II	Lestaurtinib + MEC or HiDAC	220	55	FLT3+, first relapse. No improvement with addition of lestaurtinib to chemotherapy
[Pratz et al. 2008] phase I	KW-2449	–	–	Only transient reduction in blast counts and correlative studies show FLT3 was being inhibited <20% of time
[Cortes et al. 2009] phase I	AC220	76	60	Unselected for patients with FLT3 mutations, but 56% of patients with FLT3 mutations achieved at least a PR

allo, allogeneic; AML, acute myeloid leukemia; BM, bone marrow; CR, complete response; FLT3, Fms-like tyrosine kinase 3; ITD, internal tandem duplications; MDS, myelodysplastic syndrome; PR, partial response.

Allogeneic stem cell transplantation in relapsed disease

Salvage chemotherapy regimens are considered to be effective based on remission rates alone. Although few studies provide survival data, those that do typically report OS in the range of 5-15 months [Thomas et al. 1999; Archimbaud et al. 1995; Vogler et al. 1994]. This calls into question whether intensive salvage therapy should be administered at all if allogeneic SCT is not the goal.

Allogeneic SCT has a role in the treatment of relapsed or refractory AML either as salvage therapy or as subsequent therapy following CR2 achieved by salvage chemotherapy. It is known that outcomes for allogeneic SCT for AML are superior when the transplant is performed in CR2 rather than during first relapse. Data from the International Bone Marrow Transplant Registry (IBMTR) on over 3500 allogeneic SCTs performed on patients with AML between 1989 and 1995 showed 3-year leukemia-free survival rates of approximately 60%, 35%, and 25% for transplants performed during first CR, subsequent CR, or in relapse respectively [Horowitz and Rowlings, 1997]. Similar data have been reported by the transplantation group at Seattle for younger patients (median age 30 years) [Sierra et al. 2000]. The Seattle data and IBMTR data reinforce the idea that allogeneic transplantation in CR2 or subsequent CR can lead to long-term leukemia-free survival in a significant cohort of patients, something which has not been demonstrated using nontransplant salvage therapies. An important question that arises is whether to proceed directly to allogeneic transplant in first relapse or to administer salvage chemotherapy in an attempt to achieve CR2. A study by the IBMTR designed to identify patients who would benefit from allogeneic transplantation while in relapse [Duval et al. 2010] examined 2255 patients with acute leukemia who underwent fully myeloablative allogeneic HSCT from matched related or unrelated donors in first or later relapse or after primary induction failure. There were 1673 patients with AML included in the study, and on multivariate analysis, five variables were identified that favored superior survival. A scoring system based on those five variables created four distinct prognostic groups. The favorable variables were primary induction failure or duration of first CR greater than 6 months; good or intermediate cytogenetics prior to transplant; HLA identical sibling or well matched or partially mistmatched unrelated donor as stem cell source; absence of circulating blasts; and Karnofsky score 90-100. Patients were given one point for each unfavorable variable, and patients who underwent transplantation with a related donor who was not HLA identical were given one additional point. Patients with 0 points had an OS rate of 42% at 3 years, while patients with 1, 2, or >3 points had a 3-year OS rate of 28%, 15%, and 6% respectively. This study is important because it identifies patients in whom allogeneic SCT with active disease is a viable salvage strategy. Not surprisingly, the characteristics of patients with active leukemia who would benefit from allogeneic SCT are also those who have a good chance of achieving CR with salvage chemotherapy. Given the clear survival advantage of transplantation in CR2 over transplantation in first relapse, the authors favor administering salvage therapy to the subset of patients who have a reasonable chance of achieving CR2, which is largely influenced by the duration of remission as described earlier. Conversely, when the likelihood of success of salvage therapy is poor, proceeding directly to allogeneic transplantation if an HLA identical sibling or well matched unrelated donor is available is a reasonable strategy, at least for younger patients who are able to tolerate myeloablative conditioning. As the study by Duval and colleagues points out, the presence of an HLA identical sibling or well matched unrelated donor, absence of circulating blasts, and a good performance status provide the best chance of success for a transplant in a patient with active disease who had poor cytogenetics and a short CR1 [Duval et al. 2010]. Interestingly, for a subgroup of patients defined by the presence of relapsed leukemia at the time of transplantation, a recent study from the University of Freiburg in Germany suggests that matched unrelated donor transplantation may lead to superior outcomes compared with matched related donor transplantation [Denz et al. 2010]. This observation has yet to be validated in a large randomized study comparing unrelated and related SCT in high-risk patients.

There have been additional efforts to expand the pool of patients for whom allogeneic transplantation after first relapse is feasible, particularly the group of patients who cannot tolerate myeloablative allogeneic SCT but whose disease burden is too high to proceed to a reduced-intensity conditioning transplant. A German—Austrian protocol for primary refractory AML consisted of fludarabine, cytarabine, and amasacrine chemotherapy followed 4 days later by reduced-intensity conditioning with 4 Gy total body irradiation, cyclophosphamide, and antithymocyte globulin and allogeneic transplantation from a matched related or unrelated donor. A total of 103 patients were enrolled, and median marrow blasts were 30%. Those patients without graft versus host disease at day +120 received donor lymphocyte infusions at escalating doses. Four-year leukemia-free survival rate was 30% [Schmid et al. 2006].

A prospective-randomized trial of allogeneic SCT versus salvage chemotherapy alone with an endpoint of OS has never been conducted. However, the MD Anderson group analyzed this question in a retrospective fashion [Armistead et al. 2009]. They studied 396 patients who underwent allogeneic SCT or chemotherapy without SCT as second salvage after first salvage failed to achieve CR, or in patients in first salvage-induced CR. Median survival was 5.1 months for allogeneic SCT as second salvage (n = 84) versus 2.3 months (n = 200) for chemotherapy as second salvage. Median survival was 11.7 months for allogeneic SCT after first salvage-induced CR (n = 46) versus 5.6 months for chemotherapy alone (n = 66). The differences were statistically significant in all cases. However, the ultimate goal for allogeneic SCT remains long-term DFS. Therefore a benefit in OS of several months in this population is a much less compelling reason to offer this modality than the possibility that a minority of patients may achieve long-term DFS, an endpoint which is not achievable in relapsed AML with chemotherapy alone.

Summary

Despite the development of a variety of new investigational therapies, relapsed or refractory AML remains a difficult clinical problem. The past decade has brought many new targeted therapies and therapies outside the realm of traditional cytotoxic chemotherapy, as well as a wealth of new prognostic and genetic information that will hopefully lead to more individually tailored therapy for relapsed disease in the future. A significant unmet challenge remains improving our understanding of leukemia stem cells and the mechanisms underlying relapse rates, which remain disappointingly high. In the absence of such an improved understanding, clinicians will need to consider all currently available approaches, including cytotoxic chemotherapy, targeted agents, and allogeneic SCT, or a combination of these, to optimize outcomes.