Pharmacological methods to reduce disease recurrence

The advent of reduced-intensity conditioning (RIC) regimens coupled with the increased availability of alternative donors has transformed the management of acute myeloid leukemia (AML) and myelodysplasia (MDS) such that allogeneic stem cell transplantation (SCT) is now a central component of the treatment algorithm for the majority of adults with high-risk disease.¹  Although the toxicity of allografting has reduced in the past 3 decades, the risk of disease relapse has not decreased.²  At the same time, the outcome for patients with recurrent disease remains extremely poor. Fewer than 10% of patients who relapse after a RIC allograft will survive long term, although it is possible to identify a small population who may benefit from salvage chemotherapy and a second transplant or donor lymphocyte infusion (DLI)³  (Figure 1). Therefore, disease relapse now represents the major cause of treatment failure for patients allografted for high-risk AML/MDS. To date, disappointingly slow progress has been made in defining the optimal transplantation strategy for patients with high-risk AML/MDS with regard to either conditioning regimen or posttransplantation immunosuppression. However, the last decade has seen the emergence of technologies allowing more accurate prediction of disease relapse and, at the same time, the advent of novel conditioning regimens and adjunctive posttransplantation strategies with the potential to reduce disease recurrence and improve outcome in this challenging but sizeable patient population.

Disease recurrence occurs in 30% to 80% of patients allografted for AML or MDS. Disease-specific factors associated with an increased risk of relapse include both prior chemoresistance and underlying disease biology. The use of in vivo T-cell depletion, although associated with a substantial reduction in the risk of both acute and chronic GVHD, may be associated with an increased risk of disease relapse and requires further investigation in this setting. It has long been recognized that patients with active disease at the time of transplantation have a higher risk of disease relapse, but there is emerging evidence that outcomes in patients with primary induction failure (PIF) may differ markedly from those observed in patients with refractory relapse. Several recent studies have demonstrated long-term survival rates ranging from 20% to 40% in selected patients with PIF, in contrast to the more gloomy outlook in patients with chemorefractory relapse.^4,5  There is therefore a compelling case for early donor identification in newly diagnosed adults with AML/MDS so that patients with PIF can be taken rapidly to transplantation. There is no evidence that myeloablative conditioning (MAC) regimens are required to deliver long-term survival in patients with PIF and, indeed, the most impressive results to date gave been delivered using the FLAMSA (fludarabine [Flu], amsacrine, cytosine arabinoside) RIC regimen, which uses early posttransplantation DLI with the aim of maximizing GVL.⁶  In patients transplanted in morphological first complete remission (CR1), there is now evidence that immunophenotypic and molecular measures of minimal residual disease (MRD) are important predictors of relapse risk.⁷  If this observation is confirmed in prospective studies, MRD assessment is likely to play a central role in the choice of conditioning regimen in patients with high-risk AML/MDS in the future.

Relapse risk in patients with AML is determined by presentation karyotype and is increased in patients with an unfavorable—specifically a monosomal—karyotype^8,9  (Figure 2). More recently, it has been suggested that the increase in relapse risk posttransplant is most marked in patients with the karyotypic abnormalities abnl(17p) or −5/5q−, who appear to have a particularly poor prognosis.¹⁰  Indeed, it has been suggested that the adverse impact of a monosomal karyotype is abrogated if the particularly poor outcome of these two cytogenetic subgroups is accounted for.¹¹  It is also now clear that molecular abnormalities such as acquired mutations of the FLT3 gene, which confer an increased relapse risk in patients treated with chemotherapy, have a similar adverse impact on outcome after transplantation.¹²  The mechanism by which specific cytogenetic or molecular abnormalities increase the risk of disease relapse after transplantation remains unknown. Prosaically, the higher rates of recurrence observed in patients transplanted in CR1 may simply reflect increased MRD levels caused by genetically determined chemoresistance. Alternatively, the very high relapse rate reported in specific cytogenetic subgroups raises the possibility that the GVL effect may be blunted, or even abrogated, by specific acquired genomic abnormalities in target leukemic cells. This hypothesis is supported by reports of loss of the mismatched HLA haplotype in patients who relapse after a haploidentical transplantation.¹³  More recently acquired genomic abnormalities, which may contribute to escape from immunological surveillance of the relapsed clone, have also been documented in patients with AML who relapse after a sibling or unrelated donor transplantation.^14,15 

In patients allografted using an alemtuzumab-based RIC regimen, 85% of those destined to relapse will do so in the first year after transplantation.¹⁶  Little is known of specific cytogenetic or molecular factors determining the kinetics of disease relapse in patients receiving transplantations in CR1 using either a MAC or RIC regimen. A more detailed understanding of the determinants of relapse kinetics will be of potential importance in designing studies aimed at assessing the impact of novel posttransplantation interventions.

AML is a hierarchical disease originating from a clonally transformed leukemic stem/progenitor (LSC) cell and it has been hypothesized that chemoresistant LSCs serve as a reservoir of disease in patients destined to relapse after myelosuppressive chemotherapy.^17,18  However, very few clinical studies have been performed to confirm the proposal that AML is a paradigm of the cancer stem cell hypothesis.^19,20  There is an even greater paucity of studies addressing the impact of SCT on the LSC compartment, specifically whether quantitation of this compartment can help guide the choice of conditioning regimen or the timing and nature of posttransplantation intervention.²¹ 

Several immunodominant antigens, including minor histocompatibility antigens such as HA1, tumor antigens such as WT1, and members of the cancer testis antigen family, have been identified as putative cellular targets of the GVL response. To date, most of the underpinning immunological studies in this area have been performed on bulk leukemic blasts and it will be important, although technically challenging, to replicate these observations in the human LSC population. Pharmacological strategies aimed at augmenting a GVL effect after transplantation have focused either on up-regulating the expression of putative target antigens on target cells or manipulation of cellular effector function. DNA methyl transferase inhibitors such as azacitidine and decitabine up-regulate the expression of both minor histocompatibility antigens and putative tumor antigens, and prior exposure to these drugs augments killing of tumor targets by CD8⁺ T cells.^22-24  At the same time, azacitidine, possibly through demethylation of the foxp3 promoter, has the capacity to increase the number of circulating regulatory T cells (Tregs) in murine transplant models.²⁵  This raises the theoretical possibility that DNA methyl transferase inhibitors may be able to epigenetically manipulate the alloreactive response after transplantation by simultaneously augmenting a GVL response and reducing the risk of GVHD.²⁶  Several other biological agents, including histone deacetylase inhibitors and bortezomib, modulate alloreactivity in animal models, thereby presenting an opportunity to improve clinical outcome after transplantation.²⁷ 

In patients with myeloid malignancies transplanted using a RIC regimen, the observed inverse correlation between relapse risk and treatment toxicity first reported in patients transplanted using a MAC regimen still applies. However, a range of new drugs developed in the last decade now present the possibility of redesigning both MAC and RIC regimens with the aim of optimizing antitumor activity without increasing toxicity.

The improved pharmacokinetics of an IV formulation of busulfan (ivBu) results in significantly less toxicity than is observed with oral preparations.²⁸  This has allowed the development of a Bu/cyclophosphamide (Cy) myeloablative regimen incorporating ivBu. At the same time, its combination with the highly immunosuppressive purine analog Flu has resulted in the development of a highly promising novel myeloablative regimen known as ivBu/Flu. The results of several ongoing studies evaluating the toxicity and activity of both of these regimens are currently being reported and it is too early at present to make a definitive assessment of their role.²⁹  Recently, the alkylating agent treosulfan has been demonstrated to possess a favorable toxicity profile and to deliver encouraging preliminary results in terms of both toxicity and antitumor activity when incorporated into the preparative regimen for patients with high-risk AML/MDS.^30,31 

The antitumor activity of the differing RIC regimens currently used in patients with high-risk AML/MDS is highly variable.³²  A recent randomized trial demonstrated a lower relapse rate in patients transplanted using a Flu/Bu/antithymocyte globulin (ATG) regimen compared with patients transplanted using a low-dose (200 Gy) total body irradiation (TBI)-based regimen, and there is increasing recognition that the choice of conditioning regimen must be adjusted according to the predicted risk of disease relapse.³³  Accordingly, several novel regimens with increased antitumor activity have been developed for patients with high-risk AML/MDS (Table 1). One of the most promising of these intensified RIC regimens, FLAMSA, was developed in Germany and incorporates both intensive pretransplant cytoreduction and early DLI.³⁴  Results from phase 2 studies in patients with poor-risk disease, specifically patients with adverse risk cytogenetics or PIF have been encouraging.^6,35  Recent studies suggest that the substitution of Bu for TBI within the FLAMSA regimen may further improve outcome.^36,37  In separate studies, clofarabine, which possesses potent antileukemic activity of its own, has been combined with Bu or melphalan (Mel), with promising initial results,^38,39  and encouraging preliminary results have also been reported using a combination of Flu, Mel, and 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU).⁴⁰  The potential of adjunctive radioimmunotherapy to increase the antitumor effect without increasing toxicity is supported by the observation that relapse is reduced in patients with very high-risk AML using ¹³¹I-labeled anti-CD45 antibodies combined with a low-dose TBI RIC regimen.⁴¹  Regrettably, despite this proliferation of novel conditioning regimens, there are, to date, no randomized trials in high-risk AML/MDS, and these must now be prioritized by the transplantation community.

The presence of a potent GVL effect in AML is well established in patients transplanted using a MAC regimen, but its existence after a RIC allograft was initially questioned. However, there are now compelling data supporting the existence of a comparable GVL effect in this setting, as evidenced by the impact of both the intensity of posttransplantation immunosuppression and the occurrence of GVHD on relapse risk^16,42,43  (Figure 3). Posttransplantation immunosuppression is a particularly important predictor of relapse in patients allografted using a T-cell–depleted regimen and, therefore, its duration and intensity is an important—and manipulable—regimen-dependent determinant of relapse. At present, however, the optimal intensity and duration of posttransplantation immunosuppression remain unknown and there is a compelling case for prospective randomized trials to study this simple question. More recently, the mTOR inhibitor sirolimus, which possesses both immunosuppressive activity and antitumor activity, has been studied as a GVHD prophylaxis agent in patients allografted for lymphoma, and this conceptual advance is worthy of further exploration in the setting of high-risk AML/MDS.⁴⁴ 

The emergence of biologically targeted drugs such as tyrosine kinase inhibitors and epigenetic therapies, which exert a potent antitumor effect with limited hematopoietic toxicity, has allowed their use as adjunctive posttransplantation strategies with the aim of targeting residual disease. This concept was first developed in the context of chronic myeloid leukemia, in which the administration of imatinib after transplantation was shown to reduce the risk of disease relapse in allografted patients.^45,46  Imatinib's ability to manipulate the kinetics of disease relapse allowed the postponement of DLI administration so that it was not required in the first 12 months after transplantation. Given the significant risk of GVHD when DLI is administered in the first year after a RIC allograft, this provides a strategy by which GVHD and GVL might be dissociated. The similar use of FLT3 inhibitors such as AC220 and other targeted therapies is currently being explored in patients transplanted for high-risk AML/MDS.

The possibility that epigenetic therapies such as azacitidine might improve outcome by epigenetically manipulating the alloreactive response has been explored recently in patients undergoing a RIC allograft for high-risk AML/MDS. Azacitidine is well tolerated after transplantation⁴⁷  and appears to accelerate Treg reconstitution and induce a CD8⁺ T-cell response to candidate tumor antigens,⁴⁸  presenting a novel strategy by which both the risk of GVHD and disease relapse can be reduced. Lenalidomide, because of its capacity to activate CD8⁺ T cells, represents an alternative method of pharmacologically augmenting a GVL response after transplantation. Early-phase studies incorporating posttransplantation lenalidomide however have been complicated by a high risk for GVHD, confirming its ability to augment an alloreactive response after transplantation. However, given the excessive GVHD-related toxicity associated with early lenalidomide administration, revised treatment regimens, perhaps using T-cell depletion or concurrent azacitidine administration, are required.⁴⁹ 

Preemptive administration of DLI, guided potentially by conventional MRD assessments or LSC quantitation, represents an attractive strategy by which a GVL effect can be augmented in patients deemed to be at a high risk of disease relapse. Currently, several challenges are associated with administration of DLI in high-risk patients transplanted for AML/MDS. Because of the rapid kinetics of disease relapse in AML/MDS, DLI must be delivered early, at a time when it is associated with significant GVHD-related complications. This significantly limits the dose of donor lymphocytes that can be safely administered. Because of its ability to expand Tregs, the concurrent administration of azacitidine with DLI in patients at a high risk for disease relapse is therefore of interest.^50,51 

Reducing the risk of disease relapse is central to improving the outcome of patients allografted for high-risk AML/MDS. Characterization of the cellular basis of disease relapse after transplantation will be important in terms both of monitoring response to novel therapeutic interventions and identifying new target antigens in the LSC population. Much progress has been made in developing agents that have the potential to increase the antitumor activity of the preparative regimen without increasing its toxicity and randomized comparisons with standard MAC and RIC regimens are now a priority. The potency of the GVL effect in AML and MDS is sometimes underplayed and there is significant potential to decrease relapse risk through more precise delivery of posttransplant immunosuppression or the development of pharmacological strategies that facilitate the delivery of MRD-guided DLI. Finally, the development of posttransplant pharmacological interventions with the potential either to reduce the risk of GVHD or enhance a GVL effect represent an important new strategy through which it may be possible to dissociate GVHD and GVL in patients at high risk for disease relapse.

Because remission status at the time of transplantation is an important predictor of relapse risk, the greatest importance should be attached to ensuring that patients with high-risk AML/MDS should proceed to transplant as soon as possible once they have achieved CR. In practice, this means ensuring that all patients are tissue typed at the time of diagnosis with the simultaneous initiation of a donor search. Given the equivalence in outcome between recipients of sibling and unrelated donor transplantations and the time delays associated with identification of siblings who are fit to donate (especially in older patients), we initiate typing of available siblings and a search of adult unrelated donor panels simultaneously unless there are indications to the contrary. In patients in whom an initial search of the unrelated donor registries suggests a low likelihood of identification of a donor, an immediate search of high-quality cord blood banks should be undertaken. The optimal myeloablative preparative regimen in patients under 45 years of age with high-risk AML/MDS who are in morphological CR remains a matter of conjecture, particularly with the advent of ivBU. In our institution, Cy/TBI and ivBu/Cy are viewed as equally effective in recipients of matched sibling allografts. Patients transplanted using an unrelated donor are conditioned using a Cy/TBI regimen incorporating pretransplantation alemtuzumab,⁵²  but good results have also been reported using an ivBu/Cy regimen incorporating ATG. In older patients with fewer than 3 comorbidities, our center, like many others in the United Kingdom, use the FMA (Flu, Mel, alemtuzumab) RIC regimen in patients in CR1 with intermediate-risk cytogenetics, but good results are also reported with a Flu/Bu/ATG regimen.^16,53  Patients who are in CR with adverse risk cytogenetics or PIF are at a particularly high risk of disease relapse and are eligible for an ongoing randomized phase 2 study comparing FMA with an intensified FLAMSA regimen. In the absence of a trial, an intensified RIC regimen such as FLAMSA-Bu would be used in such patients. Particular attention is paid to the degree of posttransplantation immunosuppression in all patients allografted for high-risk AML/MDS. This entails thrice weekly cyclosporine (CsA) monitoring with the aim of achieving a trough level between 100 and 200 μ/L. Once discharged, CsA levels are measured weekly and careful attention is paid to the impact of concurrent medications such as azole antifungals. A CsA taper at a rate of 10% per week is commenced in recipients of sibling and matched unrelated donor transplantation at day 60 who have no evidence of active GVHD or history of > Grade 2 acute GVHD. A BM aspirate to confirm remission status is performed at day +100, with the aim of all patients having discontinued immunosuppression by day +120. Patients who have undergone an RIC allograft incorporating T-cell depletion who have mixed T-cell chimerism at day +180 and who have discontinued immunosuppression are candidates to receive escalating DLI at 2 monthly intervals until acquisition of full donor T-cell chimerism. Patients with high-risk AML are eligible for a multicenter randomized study of posttransplantation azacitidine with the aim of reducing the risk of disease relapse.

Conflict-of-interest disclosure: The author declares no competing financial interests. Off-label drug use: Azacitidine and lenalidomide administered after transplantation to improve outcome.

Charles Craddock, Centre for Clinical Haematology, Queen Elizabeth Hospital, Birmingham B15 2TH, United Kingdom; Phone: +44-12-1472-1311; Fax: +44-12-1697-8401; e-mail: charles.craddock@uhb.nhs.uk.