Considerations regarding maintenance therapy for acute myeloid leukemia in remission

ABSTRACT

Introduction

For most adult patients with acute myeloid leukemia, relapse is characteristic of the disease. When allotransplant in first complete remission is administered as consolidative therapy, relapse is still common, affecting 20–40% of recipients. Maintenance of remission with low-dose treatments may hold promise in preventing relapse.

Areas Covered

Improvements in the detection of clinical and biological variants of disease allow the practitioner to identify which patients, based on disease features, may benefit from therapy directed at residual clonal elements that might contribute to relapse. Along with improvements in methods of detecting residual disease, novel agents are under investigation as a platform in order to maintain remission and may contribute to prolonged survival. In this manuscript, we review literature available through PubMed regarding the use of maintenance therapy, described as post-remission or post-transplant treatment intended to delay or prevent relapse.

Expert Opinion

Although results of randomized studies are limited, a role for maintenance therapy, particularly directed at molecular targets, in distinct settings of post-remission management is recommended. We also advise that randomized studies of immune therapy along with opportunities for further evaluation of risk-agnostic interventions be a focus of cooperative groups.

KEYWORDS:

• AML
• maintenance
• remission
• targets
• therapy
• allogeneic
• transplantation

1. Introduction to the problem of relapse in AML

Despite developments in the evaluation and treatment of newly diagnosed acute myeloid leukemia, relapse remains the most common reason for the high mortality associated with the disease [1]. Allogeneic hematopoietic cell transplant has been applied in first remission as a form of immunotherapy to address the risk of relapse, but for many patients, relapse still occurs. In fact, when one considers the improvements in detection of clinical and biological variants of AML that predict for relapse, practitioners now have tools to judge which of their patients, based on disease-features, may derive benefit from the prolonged pancytopenia and hospitalization associated with cytarabine-based induction chemotherapy and post-remission allogeneic transplantation. Further assessment based on the interactions of molecular markers, the occurrence of co-mutations, and the presence of measurable residual disease following induction therapy all collectively modify risk. The identification of these mutations has been a driver for drug development targeting mutations, and may serve as a potential platform for maintenance therapy to prolong survival among patients with AML [2–5].

2. The role of post-remission maintenance with or without consolidative allogeneic transplant

Several concepts arise when considering post-remission maintenance therapy after allotransplant in AML [6]. The same concepts apply to maintenance therapy for AML without consolidative transplant. Most clinical trials have focused on the larger population of AML patients at risk for relapse, and have not specifically addressed the post-transplant setting. Nonetheless, the concepts derived from studies of maintenance therapy in AML apply to the post-transplant setting, and a practitioner is compelled to consider the efficacy of maintenance therapy based on studies that might exclude allotransplant. One concept is to consider the historical experience of maintenance based on the history of prolonged, low-dose therapy intended to reduce relapse. Until the approval of oral azacytidine as maintenance therapy, no therapy had been approved by the United States FDA for such management [7,8]. Before completion of that trial, there had been no long-term, prospective, randomized trial in AML that could demonstrate a survival advantage conferred by maintenance therapy in any post-remission setting. The reasons for this are many, and include logistical hurdles, such as the impact of consolidative allogeneic transplantation from diverse donor sources. Other reasons include the complex biological and clinical risk factors that characterize adult AML in first remission. Early studies of prolonged, low-dose chemotherapy were done at a time when biological risk was not well-studied, and stratification for clinical variables was applied inconsistently. That did not mean that maintenance chemotherapy did not work; it meant that the designs of studies were such that efficacy could not be demonstrated. Once enriched for specific subtypes of disease, like acute promyelocytic leukemia, maintenance was demonstrated to provide benefit in leukemia-free survival for some patients, and for some disease variants [9]

In the large phase 3 QUAZAR trial, patients age ≥55, with AML in first complete remission with leukemia characterized by intermediate- or poor-risk karyotype, and ineligible for allogeneic hematopoietic cell transplantation at the time of screening, were randomized to receive 300 mg daily oral azacytidine or placebo for 14 days in a 28-day treatment cycle [7,10]. In order to manage differences in leukemia biology, the randomization to treatment groups was stratified by age, cytogenetic risk group, prior history of myelodysplasia or CMML, and by history of consolidation therapy. In addition to relapse and survival endpoints, patients were also evaluated for measurable disease by flow cytometry of bone marrow at baseline and upon at least one visit while on study. As expected, patients entered on the oral-azacytidine study with measurable residual disease at baseline study were more likely to have shorter survival when compared with those whose disease was MRD-negative at a threshold of > 0.1%. Notwithstanding the clinical impact of measurable disease, oral-azacytidine maintenance improved survival regardless of the MRD status. Moreover, compared with placebo, randomization to the experimental arm was also associated with a significantly greater likelihood of conversion to MRD-negative status, and this was associated with longer survival. Predictors for an MRD response included prior consolidation therapy before entering the study and a longer interval from first remission to study entry. Patients in the placebo arm, however, received minimal post-remission therapy, in general, and the inconsistency of consolidation therapy may have led to an improvement in outcome for patients assigned to the experimental arm.

The success of this trial in reaching the study endpoints of improved median relapse-free and overall survival may have as much to do with the structure of the trial as with the efficacy of post-remission maintenance with the putative mechanism of action of oral azacytidine to induce a hypomethylated molecular phenotype. Firstly, restricting study entry to those deemed ineligible for allogeneic transplant removed the impact of allogeneic transplant. Allogeneic transplant has a significant effect on both non-relapse mortality and the incidence of relapse. Allotransplant is administered for its intended immunotherapeutic effects and is believed to induce its therapeutic effects through acquired immune response to residual leukemia. The restriction in the QUAZAR study to those not treated with transplant rendered the population under study more homogenous clinically and biologically. Patients eligible for the study were older, were more likely to have leukemia characterized by intermediate cytogenetic risk rather than poor-risk, and less likely to have secondary AML that would have militated for allotransplant. A smaller proportion of patients were deemed ineligible for transplant based on no available donor or patient decision. Since patients were screened after they had achieved remission, even poor performance status may have been restricted from study-entry since subjects had proven their ability to make it through intensive-chemotherapy induction, and for many, intensive consolidation therapy. Thus, the selection criteria selected for a ‘best-of-the-worst’ patient population, fit enough to have made it through induction therapy that pre-dated therapy with hypomethylating agent-venetoclax, but not fit enough to undergo allotransplantation. Treatment on the investigational arm led to a statistical improvement in overall survival, and an improvement in relapse-free survival, but at later time-points in the study, the probability of survival for the two groups became quite similar [10]. After nearly 5 years of median study follow-up, the probability of survival in the oral-azacytidine treatment and placebo groups was 26.5% and 20.1%, respectively [11]. The major impact of maintenance was seen in the early time-points, yielding a significant improvement in median survival in favor of the experimental arm. Despite the drop-off in effect over time, this large maintenance study achieved important results by demonstrating a conversion in MRD-positivity rate and in delaying disease progression for a significant population of AML patients that fit the study criteria. Whether these results would apply in the post-transplant setting is not clear. Further, the QUAZAR study demonstrated the path forward with other interventions.

3. Molecular targets of maintenance therapy

Another approach to maintenance directed at residual leukemia would be to direct therapy to distinctive features of the clone rather than broadly attacking epigenetic modifications. A trial of sorafenib as maintenance, in this case after allogeneic transplantation, assigned patients with flt3-ITD-mutated AML in first complete remission after allogeneic transplant to sorafenib versus placebo [12,13] Although the population of patients enrolled in the Swiss trial was small (30 patients retrospectively analyzed), the authors demonstrated a significant reduction in risk of relapse at 24 months, and suggested that maintenance may have had a significant impact since median overall survival was not reached with a median follow-up of 12 months [12] In a retrospective trial from the European Bone Marrow Transplant registry, 28 of 462 patients with flt3-mutated AML undergoing allogeneic transplantation received sorafenib initiated at a median of 55 days after transplant. Dose-modification was required in 12 patients, mostly for skin rash, skin GvHD, and hematologic toxicity, but also for other reasons. The median duration of the therapy was 446 days for those treated prophylactically (n = 18) and 385 days for those treated preemptively for positive MRD findings (n = 9). The authors demonstrated that posttransplant sorafenib maintenance reduced the risk of relapse and impacted the adverse effect of MRD-positivity in this small group of patients [13]. The study did not suggest that allogeneic transplant could be replaced by targeted therapy as maintenance. More recently, a randomized, placebo-controlled trial by the German–Austrian Study group gave 24 months of sorafenib maintenance versus placebo for patients with flt3-ITD-mutated AML treated with allogeneic transplant in remission. The trial demonstrated an improvement in leukemia-free survival in favor of the sorafenib maintenance arm. The greatest benefit accrued to those whose disease was MRD-positive after transplant [14]. The EBMT now recommends maintenance treatment with sorafenib for at least two years after allogeneic transplant for patients with AML characterized by flt3 ITD [15].

The administration of maintenance therapy directed at the flt3-ITD mutation cannot easily be assessed by excluding the impact of allogeneic transplant [16,17]. Since the flt3 mutation is associated with an intermediate- to high-risk of relapse, initial therapy for eligible patients typically includes allogeneic transplant in first remission. Differences in preparative conditioning, donor source, and immunoprophylaxis against graft-versus-host disease all impact the interpretation of results. In a meta-analysis of 12 studies, Fei and colleagues identified studies in which over 2000 patients with AML characterized by flt3 mutations were treated with allogeneic transplant; 635 were treated with maintenance therapy directed at the flt3 mutation (ITD and TKD) [14–16]. Only three trials were randomized [14,18,19]. Among the studies reviewed, only five required that patients be in complete remission at the time of transplant; the others included patients with variable disease status. Nonetheless, their meta-analysis demonstrated a significant improvement in both relapse-free and overall survival among those who received maintenance therapy, and this improvement was attributed to a decreased cumulative incidence of relapse. Among studies that evaluated non-relapse mortality, the use of maintenance after allogeneic transplant did not confer an increased risk of mortality attributed to transplant complications. Although there was an increased incidence of skin toxicity in the maintenance group, there was no evidence of an increased risk of acute or chronic GvHD associated with such therapy.

The majority of published studies of maintenance therapy for flt3-mutated AML have used sorafenib or midostaurin, and the timing when therapy was initiated varied. Midostaurin, a drug that has been approved by the European Medical Agency as maintenance, has been incorporated into treatment after allotransplant on both sides of the Atlantic. Fewer studies have reported the impact of other agents such as quizartinib or gilteritinib. In randomized trials of the latter agents against salvage chemotherapy, a significant improvement in initial response rate to re-induction therapy permitted more patients assigned to the targeted agents to undergo post-remission allogeneic transplantation in complete remission. Many patients in those trials continued on the flt3-directed maintenance drugs after transplant. The impact of continued quizartinib or gilteritinib administered after transplant cannot be determined, but the drugs seemed to be well tolerated, and their use associated with long-term survival [20,21]. Several trials are in progress to assess the efficacy of these agents in the post-transplant maintenance setting.

Following the approval of other drugs that target distinct mutations in AML, maintenance strategies are under evaluation at early time points in therapy. Both ivosidenib and enasidenib have single-agent activity in relapsed or chemotherapy-refractory AML characterized by mutated IDH 1 and 2, respectively, and have been moved to initial treatment combined with chemotherapy, hypomethylating agents, and venetoclax [22–24]. In this setting, the use of these agents is associated not only with remission induction, but with clearance of the pathogenic mutation by digital PCR, and survival probabilities of over 75% at 1 year. The challenge will be in determining the contribution of consolidation chemotherapy, inhibitor therapy, and prolonged maintenance to these survival statistics. In the absence of randomized, controlled trials, the contribution of maintenance therapy with the targeted agent to long-term leukemia-free survival will be hard to assess, since even brief exposure to the targeted agent in induction and/or consolidation cycles may be sufficient to achieve the maximum response conferred by the drug.

4. Risk-agnostic strategies are required for most patients receiving maintenance therapy

Incorporating novel agents that have been utilized in the induction of remission for patients ineligible for intensive chemotherapy has gained favor in the post-allogeneic transplant setting. Venetoclax combinations, although myelosuppressive, are often utilized, off-label, in patients at high risk of relapse after allogeneic transplant for AML. At present, few studies have been published to determine the safety of such approaches, the need for dose attenuation for myelosuppression, and the impact, if any, on relapse during maintenance with venetoclax and hypomethylating agents [24].

Another form of maintenance may come from agents that induce a sustained immunologic response to residual leukemia. Allogeneic hematopoietic cell transplant is generally not considered a form of maintenance therapy since it involves intense pre-transplant conditioning and supportive care through a period of prolonged myelosuppression. Of course, the sustained leukemia-free survival achieved by many patients after allotransplant may be a result of disease surveillance maintained by donor immunity [25]. An attempt to mobilize immunoreactive cells to address the potential for relapse associated with a reservoir of persistent leukemia could include maintenance treatment with pharmacologic agents or cell therapy based on the allotransplant model. Immune-checkpoint blockade with several agents has been shown to be tolerable, but of limited efficacy in the treatment of overt leukemia [26,27]. These agents have also been explored in patients with relapsed/refractory AML after treatment with high-dose cytarabine. Interferon has been studied alone as maintenance in two randomized trials that failed to demonstrate efficacy, although a more recent study showed efficacy in favorable-risk AML (see Table 1) [30]. Lenalidomide has also been studied after allogeneic transplant, but toxicity and leukemia relapse demonstrate that this drug does not have the potential in mobilizing an immune response to AML. Several randomized trials of IL-2 likewise have shown no benefit, however when combined with histamine, a phase 3 trial showed improvement in leukemia-free survival, without an impact on survival. In the current era, a variety of T- or NK cell-therapy products directed at myeloid antigens are under evaluation, but until a leukemia-specific target can be identified, the potential for prolonged myelosuppression, and the limited activity in the relapsed setting, has led to doubts about the potential efficacy of engineered T or NK cells in AML. Vaccine therapy after allogeneic transplant has been considered to offer a mechanism to specifically induce a donor-derived cellular immune response directed to myeloid leukemia antigens [31]. Alternatively, inactivated leukemia cells can be transduced to this antigen during development in order to support myeloid cell migration secrete cytokines in the immediate, post-transplant setting [32]. In a small trial of 30 patients, a strategy of post-transplant immune-cell induction with GM CSF-secreting leukemia cells led to a higher incidence of both acute and chronic GvHD, without enhanced antileukemia efficacy of the graft.

Table 1. Categories of interventions that may serve as maintenance therapy in the setting of Acute Myelogenous Leukemia in remission.

Pharmacologic Intervention	Maintenance Used	Outcomes	References
Hypomethylating Agents	Oral azacytidine	Oral-azacytidine maintenance improved survival regardless of the MRD status	[7,8]
FLT3 Targets	Sorafenib	A significant reduction in risk of relapse at 24 months	[12–15,18]
	Midostaurin	Increased relapse-free survival in midostaurin group compared to SOC group	[19]
	Quizartinib	Significant increased median overall survival in quizartinib group vs. placebo group.	[29]
	Gilteritinib	Significant median OS of gilteritinib group compared to salvage chemotherapy group	[21,28]
Mutated IDH1 Targets	Ivosidenib	Improved CR and reduction of cells with mIDH1 for patients treated with ivosidenib	[21,24]
Mutated IDH2 Targets	Enasidenib	Modest improvement in outcome of patients taking enasidenib and AZA with better outcomes for those adding Venetoclax	[22,23]
Immunologic Targets	Interferon	Efficacy demonstrated only for favorable-risk	[30]
	Lenalidomide	Toxicity and leukemia relapse demonstrate that this drug does not have potential in mobilizing an immune response to AML studied after allogeneic transplant	[27]
	IL-2	No benefit	[26]
	IL-2+histamine	A phase 3 trial showed improvement in leukemia-free survival, without an impact on survival	[26]
	Vaccine	No benefit and increased GvHD	[31,32]
Allogeneic Transplant		Significant impact in the post-remission setting in case-control studies	[25,28,29]

Another immunotherapeutic intervention would be antibody-directed, either to distinct clonal elements or to something universal among myeloid antigens, or directed to an immune-responsive cell population distinct from adaptive immune cells. Among antibody-directed therapies, gemtuzumab ozogamicin, a drug already approved for management of acute myeloid leukemia, could be considered in a trial of maintenance therapy. Gemtuzumab is a monoclonal antibody directed against CD33, an antigen expressed broadly in myeloid cells. Of course, such treatment would be expected to induce marrow suppression; the drug has proven difficult to dose in a way that minimizes marrow-toxic effects. Furthermore, it is not selective, and elements of the clone could escape the antibody with decreased CD33 expression. Furthermore, the drug has well-known hepatotoxicity, although this adverse effect is dose-related. Other monoclonal antibodies to cell-surface markers in myeloid cells include tagraxofusp, directed at CD123, and Actimab, directed to CD33. Both agents have a cytotoxic payload; the former has a truncated diphtheria toxin, and the latter a radionuclide.

Immune-therapy in the form of monoclonal antibodies may deliver its activity by modulating other cell populations. Magrolimab is a monoclonal antibody directed at CD47, a receptor for thrombospondin that has a role in vascular development. The target also acts to inhibit phagocytosis through its interaction with signal-regulatory protein alpha, a transmembrane receptor on myeloid cells. Myeloid cells express this antigen during development in order to support myeloid cell migration [33]. Antagonism of that receptor leads to enhanced phagocytosis, and has been exploited in the management of myelodysplasia treated with hypomethylating agents. The potential for drugs directed at inhibiting CD47 to be utilized in the management of multiple different tumor types, including relapsed/refractory AML. The drug could also be included in approaches to deliver maintenance in the setting of minimal residual disease.

5. Expert opinion

The challenges to developing, implementing, and evaluating any maintenance therapy in the setting of AML in first or greater remission are as likely to be structural as pharmacological. In this document, we have identified several structural, or logistical barriers to documenting efficacy for a maintenance regimen. No matter how plausible or credible a regimen must be, there are few methods by which one may isolate the effect of maintenance. Firstly, AML is a disease characterized by diverse biological features. Distinct cytogenetic and molecular features can be clustered, but still yield subtypes of disease that may influence response to treatment, and duration of remission. Isolating these in a clinical trial is bound to be challenging, given the limited number of patients with specific disease subtypes enrolled in a clinical trial. Further, overlapping features may introduce further heterogeneity in a trial that enrolls on the basis of a specific molecular or cytogenetic characteristic. Superimposed on defining a specific set of disease features are clinical features such as age, gender, and co-morbid medical conditions that have all been demonstrated independently to have a significant impact on disease recurrence and survival. Additionally, other interventions, distributed in a non-randomized way, may have the most important effect on clinical endpoints [22]. Specifically, the use of allogeneic transplant defines a group of patients whose disease biology may be intermediate- or high-risk, but whose clinical features, by virtue of the selection process involved in the allocation of transplant resources, may be more favorable than those not proceeding to transplant.

Clinical trials have attempted to address the heterogeneity of disease biology and clinical management. Many trials are restricted to an older population, not because the read-out may be accomplished more quickly, but because disease biology is generally higher risk, and allogeneic transplantation is less likely to intervene, thus isolating the potential effect of a maintenance strategy. The problem here is that there are far too many trials that are so constructed, and results may not be generalizable to a different population of patients. Studies of agents that target a molecular profile may produce results influenced not only by the use of transplant, but how transplant is done, what kind of donors are used and what kind of immune-suppressive prophylaxis is administered. These specifics may be responsible for outcomes that differ by geography and may vitiate a favorable response to a targeted maintenance strategy on one continent when compared to another region. Additionally, different populations may have different approaches to managing adverse effects, or may utilize agents that affect the pharmacological properties of maintenance drugs.

Recommendations for bringing a maintenance regimen to practice could include changing the marker of response, such as the elimination of measurable residual disease. The endpoint of decreasing disease measurable by either multiparameter flow cytometry or next-generation sequencing was a regulatory strategy by which Blinatumumab was approved in the United States. Such an approach, however, may not be available to newer agents since approval may depend on demonstrating improved survival, and not conversion of a laboratory feature like MRD from positive to negative. Another approach would be to isolate a group of patients for whom therapeutic options are limited, such as patients with relapsed AML after allogeneic transplant, for whom a subsequent remission is unlikely to be sustained either with or without further cellular therapy. Although these patients have disease characterized by a high risk of recurrence, inversion of a second remission, or sustained remission with a maintenance agent, could be considered a regulatory endpoint. At the other end of the spectrum, treatment in first remission for a group of patients likely to go onto allogeneic transplant could isolate an effect of maintenance if the group is identified as having disease with a high likelihood of recurrence even after allotransplant [23]. Finally, a single-agent treatment with a small molecule or with immunotherapy initiated for either early relapse, or molecular evidence of recurrence after allogeneic transplant would lend itself to long-term therapy to maintain prolonged disease control.

Although maintenance approaches that can be delivered across the spectrum of distinct biological features of AML have not produced success, the immunotherapeutic effect of allogeneic transplant can be identified statistically among allotransplant recipients who sustain chronic GvHD, and in the laboratory. The absence of a strong effect of maintenance immunotherapy may be a result of targets that are acquired late in the outgrowth of a malignant clone, a problem of drug distribution to the bone marrow niche, the resistance conferred by prolonged quiescence of elements within the leukemia clone that contribute to relapse, and elements that suppress an immune response. The same mechanisms of resistance may apply to pharmacological agents, but if these agents are well-tolerated, and if they can be administered over long periods of time, maintenance therapy has the potential to overcome the leading cause of treatment failure in leukemia, relapse, and resistance to therapy.

Article highlights

• The role of maintenance therapy for Acute Myeloid Leukemia in remission requires discrete targets.

• The distinction of maintenance therapy in the post-allogeneic transplant setting follows the model of dose-intensive chemotherapy and adoptive immunotherapy.

• What can we learn from the introduction of oral azacytidine as an approved maintenance therapy in patients with AML in first remission ineligible for dose-intensive management and allogeneic transplantation is that prolonged, low-dose medication may delay or prevent disease relapse for a subset of patients.

• Potential maintenance strategies directed against leukemia-specific mutations after allogeneic transplantation have demonstrated the feasibility, and likely the efficacy of post-transplant maintenance.

• Immunotherapy as a risk-agnostic approach to maintenance after allogeneic transplantation may be the only option available to the majority of patients whose disease is not characterized by unique disease targets.

Declaration of interest

GJ Schiller has received grants or contracts from; AbbVie, Actinium, Actuate, Agios, Arog, Astellas, Amgen Aptevo, AltruBio, AVM Bio, Bristol Myers Squibb/Celgene, Biopath, BioMea, Biosight, Cellularity, Celator, Constellation, Cogent, Cellectis, Daiichi-Sankyo, Deciphera, Delta-Fly, Fate, Forma, FujiFilm, Gamida, Genentech-Roche, Glycomimetics, Geron, Gilead, Incyte, Karyopharm, Kiadis, Kite/Gilead, Kronos Bio, Kura, Janssen, Immunogene, Loxo, Marker, Mateon, Onconova, Pfizer, PrECOG, Regimmune, Samus, Sangamo, Sellas, Stemline, Syros, Takeda, Tolero, Trovagene, Agios, Amgen, Jazz, Orca, Ono-UK, Novartis. GJ Schiller has also received consulting fees from; Bristol Myers Squibb, Curios, and Daiichi, has received honoraria and is on the speakers’ bureau for Agios, Gamida, Gilead, Incyte, Amgen, Bristol Myers Squibb, Novartis, Ono Pharma, AVM Biotech, and GSK, is on the speakers’ bureau for AbbVie, Astellas, Celgene, Karyopharm, and Stemline and has received honoraria from AZ. GJ Schiller serves on the board/advisory committee member for Agios, Gamida, Gilead, Incyte, Amgen, Bristol Myers Squibb, Novartis, Ono Pharma, AVM Biotech, GSK and AZ and holds stock in Amgen, Bristol Myers Squibb, and Janssen/J&J. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or material discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This paper was not funded.