Existing agents, novel agents, or transplantation for high-risk MDS

Although allogeneic hematopoietic stem cell transplantation (alloHCT) is the only cure for myelodysplastic syndrome (MDS), it carries risk for treatment-related mortality (TRM) and morbidity. The major alternative option is hypomethylating agents (HMAs), which have a lower TRM and less morbidity, but they are not curative and offer limited improvement in overall survival (OS). Despite several failed initial attempts to improve outcomes with HMA combination strategies, there is much hope with newer targeted therapies. For example, oral administration of HMA allows increased exposure duration, potentially improving outcomes. Regardless of initial therapeutic choice, relapse after HMA and alloHCT is associated with short survival and a lack of available treatment options. Enrollment in clinical trials should be strongly supported by clinicians, MDS support networks, and medical societies because it remains the only way to improve historical outcomes. The treatment of MDS requires a unified approach incorporating cellular-based therapies, HMA combination approaches, and novel targeted agents.

AlloHCT for patients with advanced MDS shows a 5-year OS of 58%, 39%, and 23% for patients with intermediate-, high-, and very high-risk disease, respectively, by the revised International Prognostic Scoring System (IPSS-R).¹  Relapse is the most common cause of failure after alloHCT. Multiple retrospective studies compared outcomes with alloHCT versus non-alloHCT approaches.^2-4  These studies show increased OS for patients with advanced MDS (intermediate-, high-, and very high-risk MDS by the IPSS-R) who undergo alloHCT. They are limited by the retrospective nature, leading to selection bias and the foundation of the Markov decision models, which assume a stochastic approach.

Recently, 2 large multicenter studies have evaluated alloHCT for patients with MDS prospectively.^5,6  The French trial enrolled 162 patients aged 50 to 70 years with advanced MDS into a prospective biological assignment trial. Patients with matched related donors (MRDs) or matched unrelated donors (MUDs) were planned to undergo alloHCT; there was no plan for patients to receive cord blood or haploidentical donor alloHCT.⁵  There were 54 patients with an MRD, 58 patients with an MUD, and 50 patients with no donor. The majority of patients received HMAs in both the donor (71%) and no-donor (88%) groups. Outcomes were assessed based on intent to treat, with 72% of patients in the donor group receiving alloHCT and 22% of patients in the no donor group receiving alternative donor alloHCT. The patients in the no-donor group who ultimately underwent alloHCT were censored at time of alloHCT. With a median follow-up of 43 months, the OS was 27% in the no-donor group and 63% in the donor group. Four-year survival was 24% in the no-donor group and 37% in the donor group, which was significantly different (P = .02) (Figure 1). Of note, the survival curves did not separate until 2 years after enrollment (landmark analysis 3 months after enrollment to account for donor search). Thus, the benefit of alloHCT for patients with an expected survival of <2 years without alloHCT would be questionable. A similar study through the Bone Marrow Transplant Clinical Trials Network (BMT-CTN) 1102 has completed accrual, but final results have not been released.⁶ 

The definition of risk in the context of alloHCT for MDS is mutable. BMT-CTN 1102 defined risk based solely on IPSS with intermediate-2 (int-2) or high risk. In contrast, the French trial defined the alloHCT population of interest by IPSS int-2, high, intermediate-1 (int-1) with poor-risk cytogenetics, low-risk patients with severe thrombocytopenia, and patients with chronic myelomonocytic leukemia (CMML) and ≥2 of the following: splenomegaly, thrombocytopenia, or leukocytosis. Patients with severe thrombocytopenia or neutropenia who might otherwise be in lower-risk categories should be considered for alloHCT because of the life-threatening complications of cytopenias and the lack of ability to offer continued transfusion support. Patients whose HMA treatment fails have a median OS of 5.6 months and 17 months for high/int-2 and low/int-1 risk MDS, respectively.^7,8  Both of these retrospective studies showed greater OS for patients who received alloHCT compared with conventional care after HMA failed. This is a highly selective population of patients, and no prospective trials have specifically addressed outcomes for patients with HMA failure who subsequently undergo alloHCT. However, a retrospective analysis from the Fred Hutchinson Cancer Research Center demonstrated that the 3-year relapse-free survival after alloHCT was 23.8% for patients with MDS for whom HMA therapy failed but was 42% for patients with MDS for whom HMA therapy succeeded (hazard ratio, 1.88; 95% confidence interval, 1.19-2.95; P < .01).⁹  Although outcomes after failure of HMA might be better with alloHCT than with non-alloHCT approaches, these results suggest that it is better to proceed with alloHCT while patients are responding to HMA rather than wait for failure. Additional disease-specific risk factors would include the type and total number of underlying mutational abnormalities (Figure 2).^10,11  Younger patients with high-risk mutations (as shown in Figure 2) should be considered for alloHCT even if otherwise considered to be at low risk because of the poor outcomes with standard treatments. High-risk mutations are also associated with inferior outcomes after alloHCT, primarily because of higher rates of relapse.^12-15  Reported relapse rates as high as 80% after alloHCT in patients with TP53 may argue against using this modality. However, the published outcomes are poor regardless of chosen intervention, and alloHCT remains the single most potent antimyeloid therapy available. Furthermore, alloHCT is not a static field; there are constant attempts to improve outcomes, especially for patients who are at extremely high risk of relapse.

Most patients with advanced MDS do not undergo alloHCT. A single-center study indicated that 65% of transplant-eligible patients with MDS are referred and only 33% underwent alloHCT.¹⁶  Multiple factors underlie this difference, including advanced age, comorbidities, and lack of donor availability. Although age alone is not a contraindication for alloHCT, it is certainly a consideration. A prospective observational study compared the outcomes of patients with MDS after alloHCT based on age (55 to 64 vs ≥65 years).¹⁷  A total of 688 patients aged ≥65 years and 592 patients aged 55 to 64 years underwent alloHCT. The median age in the older age group was 68 (range, 65-79) years. With a median follow-up of 47 months, the 3-year TRM/OS was 28%/37% and 25%/42% for patients aged ≥65 and 55 to 64 years, respectively (Figure 3). There was no significant difference in OS between the 2 cohorts, as measured by multivariate analysis adjusted for excess risk for mortality. Although there is no strict age cutoff for alloHCT, I generally recommend alternative approaches in patients aged ≥75 years. Multiple studies have demonstrated that the HCT comorbidity index (HCT-CI) predicts TRM.¹⁸  Patients with HCT-CI of ≥4 are considered for alternative approaches exclusive of alloHCT.

One factor that has changed recently is donor availability. A retrospective analysis was performed of 228 patients with MDS who underwent haploidentical HCT; 102 received post-HCT cyclophosphamide for graft-versus-host disease prophylaxis.¹⁹  With a median follow-up of 18 months, the patients who received post-HCT cyclophosphamide had a 3-year OS and TRM of 38% and 41%, respectively. A similar retrospective analysis with cord blood transplantation for 176 patients with MDS showed a 3-year OS of 31% and 3-year TRM of 40%.²⁰  Based on current donor algorithms, which are inclusive of haploidentical, cord blood, and mismatched unrelated donors, most patients who are considered candidates for alloHCT have identifiable donors.

Patient 1 is a 52-year-old man who developed easy bruising with epistaxis. A complete blood count with differential shows a total white blood cell count of 8,200, hemoglobin 10 g/dL, platelet count 29,000, 8% peripheral myeloid blasts, and an absolute neutrophil count of 4,800. Bone marrow aspirate and biopsy shows 10% blasts by morphology, increased nucleated red blood cells with megaloblastoid maturation, and left-shifted myeloid maturation. The diagnosis is MDS-EB2. Standard karyotype shows multiple chromosomal abnormalities with a complex monosomal karyotype including monosomies 5, 13, 18, and 20, trisomy 8, and deletion 4q. Genotype sequencing shows mutation with TP53 (p.V143M, NM_000546.5:c.427G>A) and PDGFRA (p.P581S, NM_006206.4:c.1741C>T). IPSS-R score is 8, very high risk; HCT-CI score is 0.

Patient 2 is a 77-year-old man who presents with dizziness and shortness of breath that is worse with exertion. A complete blood count with differential shows a white blood cell count of 1,400, hemoglobin 7.9 g/dL, platelet count 41,000, absolute neutrophil count 210, and no peripheral blasts. He receives a transfusion of 2 U packed red blood cells. A bone marrow aspirate and biopsy demonstrate 6% myeloid blasts by morphology; myeloid and erythroid lineages show megaloblastoid changes. Standard karyotype shows multiple chromosomal changes with trisomy 1, 2, 6, 11, 14, 15, 22, del5q in 11 metaphases; 2 metaphases with a normal male karyotype; and a separate clone with sole del5q in 7 metaphases. Mutational analysis reveals TP53 (p.Cys176Trp, NM_000546.5:c.528C>G). IPSS-R is 9, very high risk. He has multiple comorbidities including coronary vascular disease with stent placement and current atrial fibrillation, morbid obesity, and type 2 diabetes mellitus; HCT-CI = 4.

Retrospective studies giving intensive chemotherapy (IC) to patients with advanced MDS before alloHCT showed minimal or no benefit.^21-23  All are hampered by selection bias and inclusion bias, including patients who ultimately underwent alloHCT. Randomized clinical trials (RCTs) comparing IC with no IC before alloHCT for patients with advanced MDS have failed. Many physicians were unwilling to randomly assign patients with ≥10% bone marrow myeloblasts to proceed directly to alloHCT despite the absence of evidence showing benefit due to the known higher post-alloHCT relapse rates. Retrospective studies comparing IC and HMA as a pre-alloHCT debulking strategy showed similar long-term OS after alloHCT.^24-26  A phase 2 RCT (NCT01812252) comparing IC with HMA before alloHCT is currently enrolling patients.

Patient 1 enrolls in an RCT comparing IC with HMA as a pre-alloHCT debulking strategy. He achieves complete remission after treatment with 3 cycles of azacitidine (aza) and venetoclax but with minimal identifiable disease (MID) detected by high-resolution flow cytometry, with 0.6% abnormal myeloid blasts. Standard karyotype was normal, but fluorescence in situ hybridization testing showed 11.3% of interphase cells with deletion 5q. The patient undergoes an MUD alloHCT with a 4-day busulfan conditioning regimen, with plans for post-HCT maintenance APR-246 (mutant p53 reactivating small molecule) on a clinical trial.

Unless patients are candidates for alloHCT, I generally do not recommended IC for patients with advanced MDS. Although there is an expected complete remission rate of 62%, the duration is ∼1 year without consolidative alloHCT, and IC has significant morbidity and mortality.²⁷  The role for IC in the setting of pre-alloHCT cytoreduction would be through the potential elimination of MID. An ongoing multicenter phase 2 trial is evaluating CPX-351 as a pre-alloHCT debulking strategy for patients with advanced MDS who are candidates for alloHCT (NCT03572764).

HMAs remain the only treatment approved by the US Food and Drug Administration for patients with advanced MDS; however, the benefit for most patients remains marginal. Two RCTs with decitabine (dec) showed no significant OS benefit with dec compared with best supportive care (BSC); however, significantly improved responses and decreased rates of acute myeloid leukemia (AML) transformation were reported.^28,29  Two RCTs (CALGB and AZA001) compared aza with BSC or conventional care regimens (IC, low-dose cytarabine, or BSC) and showed significantly increased leukemia-free survival and OS, respectively.^30,31  On average, the median OS was significantly higher by 9 months with aza. The median response time with aza is 3 months, with 96% of responders doing so by 6 cycles. With dec, the median response time is 1.7 months, with most patients who do respond responding by 4 months. Many patients with MDS are undertreated with HMA therapy because treatment is stopped prematurely or significant dose reductions are performed. I recommend that patients who are treated with aza receive ≥6 cycles and that patients treated with dec receive ≥4 cycles unless there is clear evidence of progression. I do not delay or dose reduce cycles for cytopenias unless there is concurrent evidence of infection.

Given the modest results seen with HMAs, there are ongoing attempts to improve outcomes. These efforts are hindered by a lack of understanding of the mechanism of action and appropriate dosing. Oral administration may lead to increased activity through prolonged exposure. CC-486, an oral formulation of aza, has adequate absorption and an acceptable side effect profile and is effective for patients with MDS.³²  Ongoing trials are evaluating CC-486 as post-alloHCT maintenance (NCT04173533) or as treatment of anemia for patients with low-risk MDS (NCT01566695). Oral HMA may be inactivated by cytidine deaminase (CDA) in the gastrointestinal tract. Cedazuridine (ced) is an oral CDA inhibitor shown to increase HMA exposure after oral administration. A phase 2 RCT with oral ced/dec and intravenous dec for patients with advanced MDS or CMML indicates similar pharmacokinetic profiles and similar OS (Figure 4).³³  Based on the results, the US Food and Drug Administration has granted priority review of ced/dec. Given the marginal results with HMAs, patients who are not candidates for alloHCT should be encouraged to enroll into clinical trials evaluating HMA combination therapy or novel agents.

Multiple attempts to improve outcomes with HMA combination therapy have largely failed. Although numerous phase 2 trials have shown potential increased response rates with novel agents, none to date have increased efficacy in an RCT.^34-37  Many agents chosen for addition to HMAs increase hematopoietic toxicity, leading to cycle delay and dose decrease, thereby limiting the efficacy of HMAs. The design of clinical trials to improve HMA outcomes is complicated by a lack of knowledge about the mechanism of action and appropriate dosing. Current RCTs evaluating combination HMA therapy are summarized in Table 1.

Patient 2 is enrolled into a phase 3 multicenter RCT comparing aza with aza + APR-246 (NCT03745716) and randomly assigned to the aza-alone arm. He has received 9 cycles of aza. Most recent bone marrow analysis shows complete remission but with MID, with 0.9% abnormal myeloblasts detected by flow. He has not needed transfusion support in 6 months.

Guadecitabine is a next-generation HMA with deoxyguanosine added to dec, which limits the activity of CDA. This prolongs the half-life of dec and decreases peak plasma exposure, leading to potentially higher response rates and lower toxicity even for patients who are refractory or resistant to HMAs. A phase 2 trial conducted in the United States with guadecitabine showed an overall response rate of 43% and a median OS of 1 year, with a 2-year survival rate of 25% (14%-38%) for patients for whom HMA failed.³⁸  A similar trial conducted in France showed a median OS of 7.1 months, with a 1-year survival of 33%.³⁹  A phase 3 RCT (NCT020907359) comparing guadecitabine with conventional care regimens has completed accrual and is in follow-up. Rigosertib, a small-molecule RAS activation inhibitor, has been compared with BSC in a phase 3 RCT for patients for whom HMA failed.⁴⁰  There was no significant difference in OS; however, a subset analysis of patients with very high-risk disease showed an OS benefit with rigosertib (Figure 5). A subsequent phase 3 trial (NCT02562443) in a restricted population of patients with MDS has completed accrual but not follow-up. An RCT using oral rigosertib with or without aza for upfront therapy for patients with advanced MDS is planned. Given the dearth of effective treatment options in patients for whom HMA has failed, enrollment into clinical trials in both the non-alloHCT and alloHCT settings should be encouraged.

There are currently 347 clinical trials for MDS listed on clinicaltrials.gov that are actively accruing patients. It is likely that most will fail to complete accrual. To date no RCT with HMA combination therapy has shown improved efficacy. A collaborative effort by investigators, patients, MDS support networks, and medical societies will be needed to move the field forward.

The author thanks Helen Crawford for help with manuscript preparation.

None disclosed.

Bart L. Scott, Clinical Research Division, Fred Hutchinson Cancer Research Center, University of Washington, 1100 Fairview Ave North, D1-100, Seattle, WA 98109; e-mail: bscott@fredhutch.org.