Current Hematologic Malignancy Reports

, Volume 9, Issue 2, pp 186–192

Treatment of Relapse of Acute Myeloid Leukemia After Allogeneic Hematopoietic Stem Cell Transplantation


  • Amir T. Fathi
    • Center for LeukemiaMassachusetts General Hospital
    • Bone Marrow Transplant Unit, Cox 108Massachusetts General Hospital
Acute Leukemias (R Stone, Section Editor)

DOI: 10.1007/s11899-014-0209-2

Cite this article as:
Fathi, A.T. & Chen, Y. Curr Hematol Malig Rep (2014) 9: 186. doi:10.1007/s11899-014-0209-2


Disease relapse remains a major cause of mortality for patients with acute myeloid leukemia (AML) undergoing allogeneic hematopoietic stem cell transplantation (HSCT). Historically, patients who experience disease relapse after HSCT have a dismal prognosis with very few long-term survivors. There is no standard treatment for patients in this situation given the variability in patient characteristics, disease biology, complications such as graft-vs.-host disease (GVHD) and infections, donor availability, and patient choice. Here, we discuss the current options for treatment of relapsed AML after HSCT including conventional chemotherapy, novel agents, donor leukocyte infusion, second allogeneic HSCT, and emerging therapies.


Acute myeloid leukemiaAllogeneic hematopoietic stem cell transplantationRelapseGraft-vs.-host disease (GVHD)Emerging therapies


Allogeneic hematopoietic stem cell transplantation (HSCT) is incorporated into treatment for many patients with acute myeloid leukemia (AML). In the last two decades, an increasing number of patients with AML are undergoing HSCT due to several factors including better disease prognostication, expansion of the unrelated donor pool, improvements in HLA-typing, decreases in early transplant-related mortality [1], the advent of reduced intensity conditioning [2], safer protocols for alternative donors [3, 4], and a growing experience with elderly patients [5, 6]. Nevertheless, despite an increasing number of patients with AML undergoing HSCT, disease relapse remains the major cause of mortality for the majority of patients [7].

The prognosis for patients with AML whose disease relapses after HSCT remains poor, and there is no standard therapeutic strategy. An individualized approach is essential as many factors need to be taken into account including: 1) patient performance status and comorbidities, 2) complications experienced such as graft-vs.-host disease (GVHD), infections, and organ injury, 3) disease biology including targetable mutations, 4) donor chimerism, 5) donor availability, and 6) patient choice.

Here, we review the current options for treatment of AML which has recurred after an allogeneic HSCT. We discuss several chemotherapy-based approaches including conventional chemotherapy, hypomethylating agents, and emerging novel therapies, and then discuss cellular therapy including donor leukocyte infusion and second allogeneic stem cell transplantation.

Chemotherapy-Based Approaches

Re-induction Chemotherapy

The traditional therapeutic approach to relapsed AML following HSCT has been multi-agent re-induction chemotherapy in order to achieve a second complete remission (CR) as transition to subsequent efforts at salvage and cure. Unfortunately, such re-induction approaches are unsuccessful in the vast majority of patients, either due to disease refractoriness or early relapse. Various chemotherapy regimens have been studied in relapsed AML, but long-term survival has been demonstrated to be as low as 5 % [8]. The exception to this rule may apply to patients with core-binding alterations or those with isolated NPM1 mutations [9, 10], in whom a second remission is more likely to occur and more likely to be sustained.

High-dose cytarabine (HiDAC) and HiDAC-containing regimens have been studied and employed in patients with relapsed AML for almost three decades [11••, 12, 13]. The maximally-tolerated dose of bolus cytarabine in these studies was determined to be 3 g/m2 twice daily for six days, and this regimen has been used effectively in patients with resistant disease. Nevertheless, irreversible cerebellar toxicity is a significant concern at this dose, especially among the elderly [14], and the majority of regimens which contain HiDAC employ lower doses, which are better tolerated.

Pre-clinical studies suggested that the addition of other chemotherapeutic agents to cytarabine may enhance the cytotoxic effects on myeloblasts [15, 16]. These led to subsequent clinical study and incorporation of agents such as mitoxantrone, cladribine, fludarabine, and idarubicin, in addition to high dose cytarabine, into re-induction chemotherapeutic regimens [1721]. These combinations are fairly well-tolerated in most patients, and are associated with initial efficacy, but with poor long-term outcomes. More recently, clofarabine has been found to be efficacious in relapsed AML [22, 23]. Like other purine analogues, clofarabine exerts its cytotoxicity by increasing ara-cytosine triphosphate (ara-CTP) in myeloblasts, but also inhibits ribonucleotide reductase, thereby decreasing the pool of nucleotides which are incorporated into DNA during cell cycling [24]. Clofarabine combined with cytarabine has been shown to be an effective re-induction approach across various risk-stratified AML groups, as well as among the elderly [25, 26]. A recent study found an improved remission rate and event-free survival in those treated with the combination of cytarabine and clofarabine, in comparison to those receiving cytarabine alone [27].

Despite the presence of multiple chemotherapy regimens for relapsed disease, to date, advanced-phase prospective trials comparing the efficacy among them are lacking. As a result, there is minimal guidance as to which of many studied regimens is optimal in relapsed patients, an issue further complicated by the substantial heterogeneity of relapsed disease and its preceding treatment [28]. Indeed, comparing various studies is quite challenging, as rates of CR have been reported as low as 20 % and as high as 80 % among re-induction regimens. Nevertheless, almost all trials have reported poor long-term survival with traditional chemotherapy [29]. In addition, conventional re-induction chemotherapy may not be a viable option for certain patients who have experienced significant complications of HSCT, are elderly, or have relapsed soon after HSCT.

Hypomethylating Agents

In recent years, hypomethylating agents such as azacitidine and decitabine have been studied in patients who relapse following SCT. In addition to their suspected promotion of differentiation in myeloblasts, such agents are thought to have an immunomodulatory effect through enhancing immunogenicity and the “graft versus leukemia” (GVL) effect [3032]. Multiple small studies have assessed the role and efficacy of these agents in the relapsed setting following HSCT.

Goodyear and colleagues first evaluated the role of azacitidine in modulating the expression of MAGE antigens and enhancing MAGE-specific cytotoxic T-cell responses against leukemia cells. Eight of 11 patients with AML or high-risk myelodysplastic syndromes with circulating MAGE-reactive T cells responded to a combination of azacitidine and valproic acid, a known inhibitor of histone deacetylase [33]. Others have also suspected an immunomodulatory role for hypomethylating agents, given the modulation of the HLA-DR antigen by azacitidine in leukemic cell lines [31]. In one study, responses were observed in five of nine patients who receive low doses of azacitidine for recurrent disease following HSCT, with three of the responses being categorized as CR [34]. In another small study of ten relapsed patients, six were found to achieve a remission, with prolonged median survival [35].

Hypomethylating agents have also been combined with donor lymphocyte infusion (DLI) to enhance the GVL effect and clinical responses following relapse. In a study of 26 patients with relapsed disease, low-dose azacitidine administration was followed by DLI, and 66 % of the patients clinically benefited, including four who achieved a CR [36]. Others have also reported the efficacy of azacitidine, alone or in combination with DLI, as therapy for patients who relapse following stem cell transplant [3739]. A recent study demonstrated a significant expansion of CD4-expressing regulatory T cells in those treated with azacitidine for relapsed disease following allogeneic SCT [40], a potentially important mechanism mediating the efficacy of the drug in this setting. Therefore, these less intensive approaches at managing relapse with hypomethylating agents may hold promise for the treatment of relapse following HSCT.

Emerging Therapies

The immunomodulator lenalidomide has been recently studied as therapy for relapsed and refractory AML, with a minority of patients experiencing durable remissions [41]. However, the experience of lenalidomide in those patients who relapse after HSCT has been limited. In an initial case report, a patient with AML exhibiting a 5q chromosomal deletion, whose disease had relapsed a year after SCT, was treated with lenalidomide. After four weeks of treatment, she was found to have evidence of skin GVHD, after ten weeks, normalization of peripheral blood cell counts were noted, and after 12 weeks, a bone marrow biopsy revealed CR [42]. Larger studies of lenalidomide in AML have suggested an immunomodulatory and efficacious role for lenalidomide in the post-HSCT setting. In a study of 31 patients with relapsed and refractory leukemias treated with lenalidomide, four had experienced relapse following transplantation. Of these, two achieved a durable remission shortly following the emergence of cutaneous GVHD [41]. Additional studies of lenalidomide, alone or in combination with other immunomodulating agents, have been reported for relapsed/refractory AML, with mostly modest results [4345]. Lenalidomide and other immunomodulatory agents are attractive in the post-HSCT setting because it appears that they can stimulate an effective graft-vs.-host reaction in a certain subset of patients and can potentially augment any GVL effect. The mechanism behind this is entirely unclear, but certainly merits further investigation.

Approximately a third of patients with AML have internal tandem duplication (ITD) mutations involving the FLT3 gene [46], an abnormality associated with higher propensity for relapse and poor outcomes [47, 48]. As a result, most patients with FLT3/ITD AML, who achieve remission with conventional upfront treatment and are otherwise eligible, proceed to SCT as the preferred approach at consolidation. However, subsequent relapse remains a common problem, and as a result, novel and effective therapeutic approaches are in need. In recent years, FLT3 tyrosine kinase inhibitors have been under development and study.

Sorafenib, a commercially available multi-kinase inhibitor employed in the treatment of renal cell and hepatocellular carcinomas [49, 50], is also a potent inhibitor of FLT3 kinase. In initial case reports, sorafenib demonstrated activity in a small number of FLT3/ITD patients who had relapsed following HSCT [51, 52], producing durable remissions. In subsequent larger studies of relapsed disease, the impact of sorafenib in the post-transplant setting was shown to be quite impressive in a subset of treated patients. Metzelder and colleagues studied 65 patients with relapsed/refractory FLT3/ITD AML, of which 29 had been previously transplanted. Of this latter group, 28 patients achieved a response, and nine accomplished a complete molecular remission, with many experiencing sustained remissions [53•]. Others have noted a significantly lower level of activity for sorafenib [54], although different patient populations may account for these discrepancies. Durable responses have also been recently reported with another potent FLT3 inhibitor, quizartinib, in the relapsed and refractory setting [55, 56]. A significant number of patients on these studies had prior stem cell transplantation. Therefore, targeted FLT3 tyrosine kinase inhibition may play a future role in the treatment of relapsed patients following stem cell transplantation.

Cellular-Therapy Based Approaches

Regardless of the response achieved with chemotherapy-based approaches, it is increasingly accepted that an augmentation of the graft-vs.-leukemia effect is necessary for achievement of durable remission. This is usually attempted through the delivery of cellular therapy, which can be either in the form of donor leukocyte infusion or a second allogeneic HSCT. Unlike with the experience with relapsed chronic myeloid leukemia (CML), results with DLI or second HSCT in the setting of overt relapsed AML are uniformly disappointing both as a result of rapid growth kinetics and decreased susceptibility to alloimmune-mediated effects. Thus, cellular therapy is usually recommended after some disease response is achieved through chemotherapy [57].

Donor Leukocyte Infusion

Relapse after a previous allogeneic HSCT avails the possibility of further donor cellular therapy in the form of donor leukocyte infusion (DLI), provided the donor remains available. Hence, DLI is not an option for recipients after umbilical cord blood transplantation or in those where the initial donor becomes ineligible or refuses. To be considered a candidate for DLI, patients should be free of clinically significant acute or chronic GVHD, and immunosuppression should be withdrawn at least several weeks prior. The theoretical mechanism of DLI is enhancement of the graft-vs.-leukemia effect through the delivery of mature cytotoxic donor cells. The risks of DLI include inciting acute and chronic GVHD, which usually correlates with disease response, or inducing a state of hematopoietic aplasia, which usually resolves with time [58]. DLI is collected through a standard leukapheresis procedure. Some centers recommend stimulating the donor again with GCSF, and this may help if the recipient is in a state of aplasia, but there is no evidence to suggest that routine use of GCSF increases the success of DLI.

There has been one large retrospective analysis from the European Blood and Marrow Transplant Group (EBMT), which has focused on DLI for relapsed AML. It compared 399 patients who did (n = 171) or did not (n = 228) receive DLI in this setting. There were 124 patients who had received chemotherapy prior to DLI, but 84 still had active AML at the time of DLI. Patients who received DLI had an improved 2-year overall survival compared to those who did not receive DLI (21 % vs. 9 %, p < 0.001). For those who did receive DLI, multivariate analysis showed that those with a lower tumor burden at relapse, favorable cytogenetics, and remission at time of DLI were associated with improved survival [59•].

The dose of DLI is usually quantified by the number of CD3+ cells / kg of recipient weight; however, there is no universally accepted standard dose. A recent study analyzed 225 patients receiving DLI from related or unrelated donors for relapsed hematological malignancies after HSCT, with multivariate analysis suggesting that an initial DLI CD3+ dose of >10 × 107 cells / kg was associated with an increased risk of subsequent GVHD, and moreover, did not decrease the risk of relapse nor improve overall survival [60]. Another study examined a series of 40 patients receiving haploidentical DLI, suggesting a reasonable starting dose was 1 × 106 CD3+ cells / kg. Currently, our standard is 5 × 106 to 1 × 107 CD3+ cells / kg for HLA-identical DLI and 1 × 105 to 1 × 106 CD3+ cells / kg for mismatched DLI, with dose escalations separated by at least four weeks if no GVHD is observed.

Clearly, there does appear to be a role of DLI in inducing effective GVL in a certain subset of patients with relapsed AML after HSCT. Those who have long initial remissions and can achieve another remission prior to DLI have the best chance of response and long-term survival; however, this is a distinct minority of patients. For patients who relapse quickly or possess refractory disease, novel approaches are needed. These include combining DLI with newer immunomodulatory agents, ex vivo activation of DLI, manipulating DLI to be able to exploit the GVL action of specific cellular subsets, and generating tumor-specific DLI to maximize GVL without causing significant GVHD [58, 61, 62]. One active area of investigation which merits specific mention is engineered chimeric antigen receptor T-cell (CART) technology, which has produced remarkable responses in certain patients with relapsed lymphoid malignancies [63], and may emerge as an effective method of manipulation of allogeneic T-cells. The challenge here, which is not new in AML, is identifying a feasible targetable antigen.

Second Allogeneic Hematopoietic Stem Cell Transplantation

Historically, a second HSCT was performed with myeloablative conditioning and the same matched related donor. However, small series illustrated very high rates of transplant-related mortality ranging between 25 % and 50 %, even in this highly selected population of patients [6468]. The CIBMTR (Center for International Blood and Marrow Transplant Research) analyzed data from 279 patients with acute and chronic leukemia (AML, n = 125) who underwent second allogeneic HSCT after disease relapse following a matched sibling HSCT. Of these, 84 % of patients were treated with a myeloablative conditioning regimen and 85 % of patients received a graft from the same donor. TRM was 30 % at 5 years with PFS and OS both being 28 % at five years. Patients who had better prognoses were younger, had longer initial durations of remission, and those who were in remission at time of second HSCT.

The use of reduced intensity conditioning (RIC) regimens has decreased TRM and has allowed older patients, or those with a greater burden of comorbidity, to undergo allogeneic HSCT. Given the high rates of TRM observed with second myeloablative HSCT, many centers now routinely use RIC when performing second HSCT, although there is little published data supporting this practice. An English national registry study analyzed outcomes in 71 patients with a variety of hematological malignancies undergoing second HSCT for relapse using RIC. The 2-year TRM was 27 % with lower rates in patients who had longer initial remissions. The 2-year overall survival was 28 % with better outcomes in patients who had longer initial remissions and in those who developed chronic GVHD. Interestingly, a couple of centers have illustrated that a subset of patients even with active disease can achieve long-term remissions after RIC HSCT, suggesting that the conditioning chemotherapy does play a role [69, 70]. Overall, it is still unclear if employing RIC leads to different outcomes for second HSCT, and longer follow-up will hopefully clarify this question.

In addition to the choice and intensity of the conditioning regimen, it is also unclear which donor should be used for the second HSCT. Recently, a German registry study analyzed 179 consecutive cases of second HSCT for patients with relapsed acute leukemia treated with a variety of conditioning and GVHD prophylaxis regimens, comparing results from using the same donor (n = 82) versus a different donor (n = 97) for the second HSCT. In those where an alternative donor was chosen, options included a different matched related donor (n = 8), an unrelated donor when a related donor was used initially (n = 29), and a different unrelated donor (n = 44). When surveyed, 60 % of centers used a donor change whenever possible, 6 % used the same donor as standard, and 34 % had no standard strategy, basing it on the individual patient situation. In multivariate analysis, donor change had no significant effect on outcomes including overall survival or non-relapse mortality, but initial remission duration and disease status at second HSCT clearly did. Overall survival at 2-years was 25 %, and superior for those who underwent initial HSCT from a matched related donor. The authors concluded that second HSCT is a reasonable strategy for the appropriate patient and both initial and different donors could be considered in this setting. Importantly, this was the largest experience showing that unrelated donors could be feasibly used as a donor source for second transplants [71•].

In general, much like the experience with DLI, patients who seem to benefit the most from second HSCT are: 1) those with chemo-sensitive disease who are able to achieve remission prior to second HSCT, and 2) those who had a long remission after initial HSCT (>6 or >12 months). To try and decrease historically high rates of non-relapse mortality, reduced intensity conditioning is becoming increasingly popular in this setting. If using the same donor for the second HSCT, a common strategy is to accelerate the pace of immunosuppression withdrawal to try and cultivate more GVL, although this is unproven. While it may be intuitive to use a different donor in attempts to harness a more potent GVL effect, the limited published data suggests that donor change does not influence overall outcomes. The ideal donor for second HSCT will be an important issue to resolve, as an increasing number of unrelated donor options become available and safer protocols are developed for use with alternative donors such as mismatched unrelated, haploidentical, and umbilical cord blood.

Lastly, it is completely unclear how to choose between DLI and second HSCT for an individual patient. If the patient’s initial donor is not available or if the patient has recovered from significant GVHD and still experienced disease relapse, it makes intuitive sense to change donors and proceed to a second HSCT. Similarly, if only high-risk donor options are available for a second HSCT, then it seems prudent to employ a DLI-based strategy. If both options are available, there is no data to suggest one strategy over another, and the characteristics associated with prognosis appear to be identical between the two treatments. Certainly, cost, resource allocation, quality of life and risks to donors are all important factors, which should be considered in future studies.


Despite the many improvements in various aspects of HSCT for patients with AML, disease relapse remains the primary cause of overall mortality. Many research efforts are focused on the prevention of relapse and these include graft engineering, pre-emptive therapy for detection of minimal residual disease, prophylactic DLI and the use of maintenance therapies after HSCT. When disease relapse does occur after HSCT, there is no standard treatment, and many factors need to be considered when choosing the appropriate therapy.

The overall goal of therapy is to stimulate more effective GVL in attempts to achieve a durable remission. If patients are still on systemic immunosuppression, these agents are usually withdrawn, and rarely, that itself will bring about a major disease response. More commonly, some systemic therapy is given followed by allogeneic cellular therapy. Systemic therapy ranges from conventional combination induction chemotherapy to newer targeted or immunomodulatory agents. Cellular therapy is usually in the form of DLI or a second allogeneic HSCT. Patients who have the best prognosis are not surprisingly, those who have had a long remission before relapse and those who can achieve remission prior to the delivery of cellular therapy. For this selected subset of patients, it appears that durable remission rates can approach 20–25 %.

Newer approaches include using agents such as lenalidomide, which appear to be able to cultivate GVL without cellular therapy, and CART technology, which can potentially direct GVL to avoid GVHD. Results of these approaches are eagerly awaited. Importantly, given that the general prognosis for patients with relapsed AML following HSCT remains quite poor and that the likelihood of durable remission are quite low, palliative measures are appropriate and should be offered to patients whose disease relapses early and aggressively after HSCT, or to those who have suffered significant morbidity from complications of HSCT. Ongoing and future research will hopefully result in better prevention of disease relapse after HSCT, identify novel ways to better exploit effective GVL, and in time, lead to higher rates of durable remission and cure.

Compliance with Ethics Guidelines

Conflict of Interest

Dr. Amir T. Fathi served on the advisory board for Agios Pharmaceutical and Seattle Genetics.

Dr. Yi-Bin Chen served as a consultant for Otsuka Pharmaceuticals and Seattle Genetics. Dr. Chen received grants from Otsuka, Seattle Genetics, Bayer/Onyx, and Novartis, Inc.

Human and Animal Rights and Informed Consent

This article does not contain any studies with human or animal subjects performed by any of the authors.

Copyright information

© Springer Science+Business Media New York 2014