How I treat AML relapse after allogeneic HSCT Free

Allogeneic hematopoietic stem cell transplantation (HSCT) can be a curative therapy for acute myeloid leukemia (AML).¹ The therapeutic efficacy of HSCT is based on the eradication of leukemic stem cells with conditioning chemotherapy given before the infusion of donor cells, as well as a powerful graft-versus-leukemia (GVL) effect mediated by recognition and killing of cancerous cells by immune effector cells in the donor graft.² The evidence for this GVL effect has been inferred clinically based on the efficacy of reduced-intensity conditioning (RIC) transplants,³ an inverse correlation between graft-versus-host disease (GVHD) and relapse,^4,5 increased relapse rates in some T-cell–depleted transplants,⁶ and the success of donor lymphocyte infusions (DLIs) in a subset of patients attaining sustained remissions after post-HSCT relapse.⁷ 

Relapse after HSCT is a catastrophic event with poor prognosis, and relapse rates have remained unchanged over the last 3 decades despite an evolving understanding of the immunobiology of GVL and immune escape mechanisms that lead to post-HSCT relapse.⁸ In a large Center for International Blood and Marrow Transplant Research analysis, 1-year overall survival (OS) from relapse was only 23%, with the time from HSCT to relapse being the strongest predictor of outcome.⁹ Higher disease burden at relapse, adverse cytogenetics, and older age were also poor prognostic factors. Ley et al described the genomic landscape of AML, identifying driver mutations and showing that the mutation rate in AML is relatively low and a complex interplay of genetic and epigenetic events lead to the emergence of disease.¹⁰ It is now standard of care to obtain mutational analysis using next-generation sequencing (NGS) on every patient with AML, with direct implications for treatment selection (targeting pathogenic mutations such as FLT3, IDH1, and IDH2) and prognostication.¹¹ In myelodysplastic syndrome (MDS) and AML, patients with TP53 alterations are at the highest risk of relapse (∼80%).^12,13 Additionally, the recognition of measurable residual disease (MRD) before HSCT as an important risk factor for relapse has been a key advancement, acknowledging that the methods of MRD measurement and their standardization are in evolution.¹⁴ Among transplant-related factors, the role of conditioning intensity has been critical. A randomized study from the Blood and Marrow Transplant Clinical Trials Network comparing transplant outcomes between myeloablative conditioning (MAC; more intensive) and RIC (less intensive) demonstrated that relapse rates were lower and OS was higher in patients with AML receiving MAC HSCT,¹⁵ further confirmed with long-term follow-up¹⁶ although a smaller European trial using MAC and RIC regimens different from those in the CTN trial found no difference in long-term outcomes.¹⁷ In the CTN trial, improved outcomes with conditioning intensification were found to be predominantly in those with pretransplant MRD, pointing to a critical effect of conditioning intensity in MRD⁺ disease control.¹⁸ 

GVL is primarily mediated by cytotoxic T-lymphocytes (CTLs) in the donor graft and those generated from engrafted donor stem cells.¹⁹ The T-cell receptor (TCR) on donor T-lymphocytes recognizes a complex formed by recipient HLA allele and an antigen presented on leukemic cells, with subsequent immunologic killing.²⁰ The identity of the antigenic targets for GVL has been elusive, although data suggest these could be mismatched coding germ line variants between the donor and recipient, which are expressed on leukemic cells (GVL) or normal recipient tissue (GVHD) and are called minor histocompatibility antigens.^21-23 There is less evidence to support the role of neoantigens^24,25 as well as leukemia-associated antigens.^26,27 In recent years, there has also been interest in the role of natural killer (NK) cells,²⁸ B-lymphocytes,²⁹ and the innate immune system in the GVL effect. In seminal studies, AML relapse after HSCT was found to be associated with dysregulation of immune function pathways, specifically the downregulation of major histocompatibility complex class II genes involved in antigen presentation, rather than acquisition of new mutations in leukemic blasts.³⁰ Vago et al found that in HLA-mismatched haploidentical transplants, mutated leukemic cells in post-HSCT relapse samples lose the HLA haplotype that differs from the donor’s haplotype by acquired uniparental disomy of chromosome 6p, leading to the nonrecognition of mutant leukemic cells by the donor’s T-lymphocytes and eventual relapse.³¹ This appears to be a mode of immune escape chiefly in mismatched transplants and is seen at much lower frequencies in HLA-matched HSCTs.³² Finally, upregulation of coinhibitory ligands (PD-L1 and B7-H3) has been shown on relapsed leukemic blasts, with corresponding upregulation of TCRs on donor T cells.³³ Understanding the biologic principles associated with relapse could, and should, guide therapeutic choices for treatment.

We now discuss 3 illustrative cases, with a review of therapy options for AML relapse after HSCT, and construct a suggested treatment algorithm for posttransplant relapse.

A 70-year-old fit man was diagnosed with AML, with initial workup demonstrating a white blood cell (WBC) count of 67 × 10³/μL, with 85% myeloblasts. Bone marrow biopsy (BMBx) revealed a normal 46,XY karyotype. NGS panel revealed a FLT3-ITD with a variant allele frequency of 47% (allelic ratio, 0.89), as well as mutations in DNMT3A and WT1. He was categorized as intermediate risk by European LeukemiaNet risk classification by genetics. He received intensive induction chemotherapy with cytarabine and daunorubicin (7 + 3) with midostaurin per the RATIFY trial³⁴ and achieved a complete remission (CR). He was treated with 2 cycles of intermediate-dose cytarabine and proceeded to a haploidentical RIC HSCT with fludarabine/cyclophosphamide/total body irradiation conditioning and posttransplant cyclophosphamide, tacrolimus, mycophenolate mofetil for GVHD prophylaxis. He was initially treated with maintenance sorafenib starting on day +60, which was discontinued after a month of therapy due to poor tolerance. Four months after HSCT, he relapsed with 23% blasts in the aspirate smear. An NGS panel redemonstrated the initial FLT3-ITD clone at 11% VAF.

Considerations for early relapse after HSCT include the following: (1) tapering of immunosuppression (IS) if still on immunosuppressive agents; (2) addressing the leukemia itself with chemotherapy or targeted agents based on fitness and complete genomic reassessment³⁵; and (3) consolidation with cellular therapy in the form of DLIs or second HSCT.

1. For patients with relapsed disease who are still on IS without active GVHD, the initial intervention is typically a brisk taper of immunosuppressive agents to generate a GVL effect by reactivation of donor T-lymphocytes. A retrospective analysis showed that IS tapering alone without additional therapeutic interventions (chemotherapy, DLI, or second HSCT) led to clinical responses in a subset of patients with far more frequent responses in RICs than in MACs (32.7% vs 4.5%; P = .0007).³⁶ Most responses (97%) were accompanied by a GVHD flare. IS taper alone is unsuccessful in ∼80% of patients and is not sufficient in the face of proliferative disease; hence, targeted or nontargeted chemotherapeutic agents may be needed either concomitantly or shortly after IS taper, followed by consolidative cellular therapy. IS taper is also used with clinical signs of imminent relapse, such as dropping myeloid or T-cell chimerism or reappearance of molecular mutations, without overt morphologic relapse.

The patient’s tacrolimus was tapered and discontinued. Three weeks after IS discontinuation, the patient reported a rash, consistent with GVHD, which was well controlled with topical corticosteroids. He started gilteritinib 120 mg daily. He achieved CR and was consolidated with 3 DLI doses. Further DLI was withheld for transaminase elevations concerning for liver GVHD.

1. The choice of chemotherapeutic agent in this setting depends on a number of factors, including the presence of a targetable mutation, disease burden and tempo, and patient fitness. A number of targeted agents are now available, which in select cases have improved outcomes compared with chemotherapy in relapsed AML. The ADMIRAL trial compared single-agent gilteritinib, a FLT3-inhibitor active against internal tandem duplications (ITD) and tyrosine kinase domain (TKD) mutations, with investigator-choice salvage chemotherapy in FLT3⁺ (ITD allelic ratio ≥ 0.05) relapsed/refractory (R/R) AML, including in 20% of patients who relapsed after HSCT.³⁷ Both remission rates (34% vs 15.3%) and median OS (9.3 vs 5.6 months) favored gilteritinib. Real-world evidence suggests that there is clinical activity in patients treated with prior FLT3 inhibitors, as was the case in our patient.³⁸ Quizartinib has also shown survival benefit compared with investigator-choice chemotherapy in FLT3-ITD (VAF ≥ 3%) mutated R/R AML, although it is not approved by the US Food and Drug Administration (FDA) for this indication.³⁹ Hence, if a FLT3 mutation is present, a FLT3 inhibitor (typically gilteritinib) is part of the initial chemotherapeutic strategy. Our patient was also started on maintenance therapy before relapse with the multikinase inhibitor sorafenib, based on 2 randomized trials demonstrating relapse-free survival and OS benefit.^40,41 Additionally, Mathew et al have shown in murine models that sorafenib can potentiate GVL via off-target interleukin-15 production by FLT3-ITD–mutated leukemic blasts synergizing with the allogeneic CD8⁺ T-cell response of GVL.⁴² Interestingly, the recently reported MORPHO trial using gilteritinib maintenance suggests that maintenance may only be necessary in patients who are MRD⁺ before or after HSCT, although these results may be specific to this drug.⁴³ Similar to our patient, the most common reason for treatment discontinuation in the SORMAIN trial was adverse events, a recurrent real-world problem.

2. In patients who are not candidates for induction chemotherapy or as an adjunctive therapy to cytotoxic chemotherapy, there are an increasing number of targeted agents, including mutant isocitrate dehydrogenase (IDH)1 and IDH2 inhibitors. There are no prospective studies focused exclusively on posttransplant relapse, with this population often representing a minority of cases in R/R AML clinical trials. Hence, the decision of agent and expected efficacy are often extrapolated from results seen in R/R trials. See Table 1 for a summary of key clinical trials in R/R AML.

3. We pursue cytotoxic chemotherapy in selected younger and fit individuals with chemotherapy-sensitive disease, recognizing that response rates are low, and toxicities may be considerable. In prospective randomized trials, cytotoxic chemotherapy alone leads to remissions in 10% to 40% of cases, with a median OS in the range of 3 to 6 months, with similarly disappointing results with lower-intensity therapy with hypomethylating agents alone.^45,51,57-59 Combination with venetoclax (Ven) has revolutionized the upfront treatment of older patients with AML, particularly for non–TP53-mutated high-risk disease, and is increasingly used in the R/R setting with similar remission rates as cytotoxic chemotherapy, albeit without randomized evidence.^60,61 The addition of Ven to high-dose chemotherapy may be a useful strategy in select cases.^62,63 Thus, we select high- vs lower-intensity therapy based predominantly on patient factors, the nature of prior therapy, as well as our growing understanding of which disease subtypes might respond most favorably to Ven-based vs cytotoxic chemotherapy.⁶⁴ 

4. Consolidative cellular therapy in the form of either DLI or a second HSCT can achieve sustained remission in a subset of post-HSCT relapses. DLI consists of infusion of unmanipulated donor peripheral blood T-lymphocytes and is typically administered in several aliquots in serially escalating doses, with close monitoring for GVHD, the most frequent toxicity of this intervention.⁶⁵ DLI was originally demonstrated to be effective in patients with chronic myelogenous leukemia (CML) who had undergone T-cell–depleted transplants^66,67 but subsequently has been used with varying success in other hematologic malignancies. In AML, an important retrospective analysis in 399 patients in first hematologic relapse after HSCT showed a 2-year OS of 21% with DLI vs 9% without; multivariate analysis of the DLI group revealed a 2-year survival of 56% vs 15% for DLI done in remission/favorable karyotype vs done in relapse or with an aplastic marrow, respectively.⁶⁸ Low disease burden at DLI was also independently prognostic for survival in DLI recipients. Hence, although DLI is beneficial, patient selection is crucial for its success. A total nucleated cell dose of 1 × 10⁷ cells per kg was found to be effective with acceptable GVHD in CML, and a later study in AML demonstrated that doses >10 × 10⁷ were associated with a greater risk of GVHD but similar responses.⁶⁹ Hence, for HLA-matched donors, we start at 1 × 10⁷ and escalate to a maximum of 3 to 4 doses as tolerated and recommend slightly lower doses (1 × 10⁶) in haplotransplants or if DLI is being used in a preemptive or even prophylactic fashion.⁷⁰ Although the precise mechanism by which DLI is effective remains unclear, Claret et al showed a significant improvement in TCR diversity months after successful DLI,⁷¹ whereas Bachireddy et al have shown in relapsed CML that response to DLI is associated with preexisting marrow-infiltrating CD8⁺ T cells and local reversal of T-cell exhaustion.⁷² The mechanistic underpinnings of DLI in AML remain incompletely understood, with recent work from our group implicating ZNF683⁺ GZMB⁺ CD8⁺ CTLs in DLI responders and TIGIT expression CTLs in DLI nonresponders.⁷³ A better understanding of the interplay between adoptive cellular therapy products and the bone marrow microenvironment may explain the clinical differences in response between DLI in CML and AML.

An alternative to DLI is a second HSCT, discussed in more detail with our second patient. There are no prospective studies to help us decide between a second HSCT and DLI as our choice of consolidative cellular therapy. However, several large retrospective analyses, including one by Kharfan-Dabaja et al, have shown a 2-year OS in the 25% range with either intervention as well as better outcomes in patients who relapse >6 months after HSCT and those who achieve low disease burden before cellular therapy with either intervention.⁷⁴ Retrospective data suggest that similar outcomes are obtained with different donor types with slightly increased nonrelapse mortality (NRM) with haploidentical donors.⁷⁵ With the growing understanding that loss of the unshared haplotype is an important mechanism of immune escape in haploidentical transplants,³¹ assays to assess HLA loss are being developed. If uniparental disomy is indeed the mechanism of relapse, DLI from the original donor would not be expected to work, and a second transplant would be a better option. These assays are largely investigational at this time. In summary, we prefer DLI if relapse has occurred within 6 months, if a modest degree of donor T-cell chimerism persists, and if the disease burden is not overwhelming. Conversely, we consider second HSCT for fit patients, preferably in MRD⁻ remissions, who have relapsed >6 months from transplantation.

A 34-year-old woman presented with a WBC count of 170 × 10³/μL and skin lesions. BMBx revealed de novo AML (normal karyotype with mutations in NPM1, DNMT3A, and NRAS). A biopsy of a skin lesion showed leukemia cutis. She underwent induction chemotherapy with 7 + 3 and achieved a morphologic remission with a negative NGS panel but with evidence of MRD by flow cytometry. She then underwent a cycle of consolidation with high-dose cytarabine, with conversion to flow-MRD negativity, followed by MAC HSCT from an HLA-matched unrelated donor with tacrolimus/methotrexate for GVHD prophylaxis. Sixteen months after HSCT, she presented with pleuritic chest pain. A computed tomography chest revealed multiple new soft tissue masses involving the left anterior inferior chest wall, left pleura, left ventricular apex, and atrial septum. A positron emission tomography/computed tomography scan revealed widespread FDG-avid extramedullary (EM) disease. A biopsy from a right axillary mass confirmed myeloid sarcoma with an NPM1 mutation. Concomitantly performed BMBx showed no evidence of morphologic relapse, but an NGS panel redemonstrated the diagnostic NPM1 mutation. She was immediately reinduced with mitoxantrone/etoposide/cytarabine, complicated by a significant fungal pneumonia and deconditioning. DLI was not an option due to donor unavailability, and she was not a candidate for a second HSCT given an Eastern Cooperative Oncology Group performance status of 3. She remained in remission for 3 months and then relapsed once more with leukemia cutis. She was then treated with ipilimumab (Ipi) on trial, with resolution of the lesions for 6 months, after which she progressed in the marrow and passed away from relapsed disease.

Isolated EM relapses are recognized as a distinct mode of relapse after HSCT, with prevalence ranging from 4% to 10% in various studies and with relapse typically occurring at a significantly longer interval from HSCT than marrow relapse.^76,77 Predictors of outcome include adverse-risk cytogenetics, previous EM disease before HSCT, as well as AML French-American-British M4 and M5 disease.⁷⁶ Reported OS ranges from 20% to 30% within 3 years and is generally better than with marrow relapse. These differences with marrow relapse suggest disparate mechanisms of immune escape, leading to variations in outcomes.

Patients are typically off IS when late relapses occur, and hence, taper of IS is usually not an option. Key considerations in the event of a late EM relapse include the following: (1) presence of targetable mutations; (2) estimating the burden of disease, which would determine the need for immediate chemotherapy vs consideration of immunotherapies, such as checkpoint inhibitors; (3) need for localized radiation or central nervous system–directed therapy; and (4) consideration of consolidative DLI or a second HSCT. Because there are no targetable mutations with FDA-approved agents in this case, and the patient relapsed with aggressive EM disease, compromising end organ function, debulking with cytotoxic chemotherapy was appropriate, with adjunctive local radiation to preserve end organ function or provide symptom relief as indicated.

Consolidative cellular therapy, particularly a second HSCT, would have been reasonable after chemotherapy, given the long interval between HSCT and relapse. However, the patient’s performance status was not adequate, a not uncommon scenario in real-world relapses. Second HSCTs are typically more toxic than first transplants. A recent retrospective analysis by Penak et al of 3356 patients with diverse hematologic malignancies suggested an OS of 38%, progression-free survival of 28%, NRM of 22%, and relapse incidence of 50%⁷⁸ which were more encouraging than prior reports.^79-81 Older age, low performance score, high disease-risk index, early relapse after the first HSCT (<6 months), unrelated/haploidentical donor, GVHD before second alloHSCT, and disease burden at transplant were associated with outcomes. In an AML-focused study, Schmalter et al identified donor type, age, and disease status at HSCT2, in vivo TCD, and Karnofsky index as associated with NRM.⁸² DLI has been shown to be less effective with EM relapse in some studies.⁸³ 

An interesting feature in this case is the recurrence of leukemia cutis after remission of EM disease and treatment with Ipi. CTL-associated protein 4 (CTLA-4) and programmed death 1 (PD-1) receptors on T cells are engaged by their respective ligands (B7-1/B7-2 and PD-L1/PD-L2), expressed on tumor cells, inhibiting T-cell effector function, which can theoretically lead to immune escape and relapse.⁸⁴ Experiments in murine models have shown that CTL-associated protein 4 blockade with Ipi (human immunoglobulin G1 kappa monoclonal antibody) in late relapses after HSCT can lead to augmentation of GVL without causing GVHD.⁸⁵ This was translated in a phase 1/1b dose escalation study (n = 28) using Ipi for the treatment of relapse after HSCT.⁸⁶ CRs were documented in 4 patients with EM AML, of whom 3 had leukemia cutis. Immune adverse effects were seen in 21% of patients and GVHD in 14%, typically responsive to corticosteroids. Although we do not routinely use immunotherapy in EM relapse outside of clinical trials, the enhanced efficacy of checkpoint blockade in leukemia cutis is intriguing and suggests tissue-specific effects of the tumor microenvironment and should be a focus for future work.

A 68-year-old man with a history of prostate adenocarcinoma treated with radiation and androgen-deprivation therapy was found to be pancytopenic on a routine preoperative checkup. A BMBx revealed AML with 21% blasts as well as significant dysplasia. Karyotype analysis revealed a complex karyotype, including deletion 17p. An NGS panel revealed a TP53 mutation and NRAS mutations. He achieved CR after treatment with 2 cycles of decitabine (DEC)/Ven and proceeded to RIC HSCT from an HLA-matched unrelated donor with fludarabine/busulfan conditioning and posttransplant cyclophosphamide, tacrolimus, mycophenolate mofetil for GVHD prophylaxis. Three months after HSCT, he was found to have dropping donor chimerism, with recurrence of the original TP53 mutation by clinical NGS as well as complex cytogenetics, prompting IS taper. Shortly thereafter, dropping chimerism prompted preemptive treatment with a hypomethylating agent (HMA) and DLI. Unfortunately, disease progressed, and he passed away after 2 cycles of therapy.

There is no standard approach to the treatment of molecular relapse after HSCT; however, there is now widespread recognition that prognosis is poor once recurrent mutations or dropping T-cell chimerism are seen,⁸⁷ emphasizing the need to adopt novel approaches for these scenarios. Preemptively, we prioritize patients who are at high risk for relapse (MRD⁺ peri-HSCT and high-risk genomics) for clinical trials of conditioning intensification, posttransplant maintenance, use of peri-transplant adoptive cellular therapies, or novel graft engineering to maximize GVL, although, with the exception of FLT3 inhibitor maintenance, we await data from randomized trials to conclusively demonstrate survival benefit with any of these approaches. We have reviewed maintenance strategies elsewhere⁸⁸ and note that these efforts are largely either prophylactic in high-risk patients or preemptive when falling chimerism, MRD detection, or molecular relapse is found using either targeted agents or mutation-agnostic strategies.^40,43,89,90 The frequency of MRD assessments after HSCT has not been well defined; we would suggest monitoring chimerism monthly and MRD every 3 months, guided by European LeukemiaNet MRD Working Group, expert consensus and institutional guidance.⁹¹ Outside of a clinical trial, for patients who have dropping chimerism, detectable MRD, or molecular relapse, our approach is a combination of IS taper when feasible, hypomethylating agents,⁹² and DLI, which are used with varying degrees of success. Of note, EBMT best practice recommendations include lower DLI doses in the prophylactic/preemptive setting.⁹³ 

In TP53-mutated AML, once morphologic relapse is detected, HMAs (DEC or azacitidine) intercalated with DLI is an option, although the efficacy in those previously exposed to these agents is unclear. In the pre-HSCT setting, therapy with HMAs in TP53-mutated oligoblastic AML and MDS resulted in CR rates of 15% to 20%, similar to those with WT TP53 disease; however, remissions are typically short-lived, with a median OS in the 5- to 12-month range.⁹⁴ A 10-day vs 5-day schedule of DEC was not found to be superior in a phase 2 randomized study.⁹⁵ Extrapolating from these data, a 5- to 7-day course of an HMA would be our preferred initial therapy, with the addition of DLI after each cycle, to complete 4 to 5 cycles of DLI. The addition of DLI is most likely to be successful if administered with low disease burden; hence, patients will often receive a few cycles of HMA alone before DLI is added. In the event that CR is achieved with HMA plus DLI, HMA maintenance would be continued per physician discretion; we would recommend at least 2 years of therapy.

The addition of a second agent to HMAs has been attempted with variable efficacy. Ven is a B-cell lymphoma-2 inhibitor, which acts on dysregulated apoptotic pathways in malignant cells, thus restoring activation of these pathways.⁹⁶ The combination of Ven with lower-intensity chemotherapy regimens has become the standard frontline therapy for older unfit adults with AML⁹⁴; in addition, the DEC/Ven combination is effective in high-risk AML.⁹⁷ However, evolving data suggest that Ven may not be as effective in TP53-mutated myeloid disease.⁹⁸ 

Many open questions remain in the pathobiology of posttransplant AML relapse, including translating our growing understanding of relapse biology and leukemic immune escape into clinical assays and therapeutic interventions. Important early steps after post-HSCT relapse identification include fitness determination and complete disease reassessment with NGS, along with thoughtful discussions regarding patient preferences and goals of care. We would frame future directions in the following categories.

1. Prognostication and relapse prediction. In the near future, advances in standardization and interpretation of NGS-MRD will allow for the detection of and potentially intervention for incipient relapse in nearly all AML subtypes.^99,100 In older patients, pretransplant NGS-MRD often reflects chemotherapy-resistant biology encoded at the time of diagnosis (ie, MDS associated and TP53 mutations)¹⁰¹; ongoing work seeks to clarify how to respond to pretransplant MRD and when to monitor and intervene in post-HSCT molecular relapses. Additionally, MRD assays are imperfectly sensitive for relapse risk, and advances in functional assays may offer additive prognostic and therapeutic information.¹⁰² There is considerable interest in using targeted or lower-intensity therapy as post-HSCT maintenance to reduce relapse risk, in the hopes of identifying patient subgroups who might preferentially benefit from posttransplant maintenance based on MRD or other factors.

2. Novel agents. We anticipate that the number of approved targeted agents for relapsed AML will continue to expand. These include small molecules and antibody-based strategies. The menin inhibitor revumenib is now approved for R/R KMT2A-rearranged acute leukemia and has been studied in NPM1-mutated leukemias, with some experience in the posttransplant setting.^56,103,TP53-mutated AML remains a key challenge. APR-246 (eprenetapopt), a putative TP53 stabilizer, showed promising activity in the post-HSCT setting with azacitidine¹⁰⁴; however, this agent is not available due to an indefinite FDA hold. Identifying antigen targets preferentially present on blasts compared with healthy hematopoietic progenitors has been a challenge in developing antibody-based and chimeric antigen receptor therapies in myeloid malignancies. Bispecific antibodies to CD3 and CD123 are under development.¹⁰⁵ 

3. Innovations in adoptive cellular therapy. There are no clear data informing whether DLI, second HSCT, or both should be pursued in posttransplantation relapse. The limited survival with any of these approaches has prompted efforts to optimize adoptive cellular therapy platforms. Cytokine-induced memory–like NK cells are allogeneic donor-derived NK cells treated with a cytokine cocktail to allow for expansion and persistence in the recipient during infusion and have been found to have antileukemic activity before transplant with emerging data after transplant.¹⁰⁶ Other modulations include adding checkpoint inhibitors to regulatory T cells (Treg)-depleted DLI (ClinicalTrials.gov identifier: NCT03912064), combination with HMA and lenalidomide,¹⁰⁷ better elucidation of minor histocompatibility antigens as GVL antigenic targets,¹⁰⁸ autologous chimeric antigen receptors,¹⁰⁹ or the use of allogeneic donor-derived TCR-engineered T cells that target HA-1 and HA-2 hematopoietic cell antigens, respectively, both presented on HLA-A∗02:01, a promising new frontier.¹¹⁰ 

In conclusion, relapse of AML after HSCT is a significant challenge with poor prognosis. In a relatively small subset of cases, the combination of IS taper, chemotherapy/targeted therapy, and consolidative adoptive cellular therapy (DLI and/or second HSCT) can lead to prolonged remissions with good quality of life. The treatment algorithms presented here (Figure 1) are fundamentally limited by the lack of prospective data in the posttransplant setting. For this reason, prioritizing enrollment on clinical trials and thoughtful discussions regarding goals of care, particularly in frail patients or those with highly treatment-resistant disease who might be best served with comfort-focused care, are universal principles.

This work is supported by National Institutes of Health, National Cancer Institute grant PO1CA229092 and Leukemia and Lymphoma Society SCOR grant 7030-33.

Contribution: M.G., H.M.M., and R.J.S. drafted and critically reviewed the manuscript and approved the final version of the manuscript.

Conflict-of-interest disclosure: R.J.S. serves on the board of directors for Kiadis and Be The Match/National Marrow Donor Program; has provided consulting for Gilead, Rheos Therapeutics, Vor Biopharma, and Novartis; and serves on the data safety monitoring board for Juno/Celgene/Bristol Myers Squibb. The remaining authors declare no competing financial interests.

Correspondence: Robert J. Soiffer, Department of Hematology/Oncology, Dana-Farber Cancer Institute, 450 Brookline Ave, Boston, MA 02115; email: robert_soiffer@dfci.harvard.edu.