Outpatient transfusions for myelodysplastic syndromes

Anemia is essentially universal in patients with myelodysplastic syndromes (MDS), and symptoms and signs of anemia, such as fatigue and dyspnea, are common presenting features.^1-4  Palpitations, headache, anxiety, insomnia, pain, and weakness are also described by patients, and these contribute to the well-documented findings that anemia is also associated with poorer health-related quality of life (QoL) in MDS.^1,4 

Both the presence and degree of anemia are important in MDS, and these have been incorporated into prognostic scores.^5,6  Malcovati et al⁵  demonstrated that hemoglobin <90 g/L in men and <80 g/L in women was independently associated with worse overall survival and with both nonleukemic and cardiac causes of death. Oliva et al^7,8  reported that degree of anemia correlated with cardiac hypertrophy and remodeling, which may help explain the worse cardiac outcomes for anemic patients, although undoubtedly other factors contribute, including cardiac dysfunction from disease-related iron dysregulation and transfusion-related iron overload.^7-9  These data do not tell us whether we can modify these outcomes by applying a different transfusion policy.

Thrombocytopenia is also very common in MDS, with up to two-thirds of patients having platelet counts <100 × 10⁹/L at diagnosis; 5% to 20% of patients have severe thrombocytopenia depending on the definition used (<20 or <10 × 10⁹/L).^1,10,11  Thrombocytopenia in MDS is commonly multifactorial; platelet dysfunction is also common but often not recognized.¹⁰  Presence and degree of thrombocytopenia are both associated with worse prognosis in terms of survival and progression to acute leukemia. In day-to-day management, the primary concern related to thrombocytopenia is risk of bleeding. Minor bleeding is common, and bleeding is variably reported as the cause of death in 5% to 24% of patients with MDS in different studies.¹¹ 

In the context of these cytopenias, 50% to 90% of patients with MDS will need red blood cell (RBC) transfusions, and many become RBC transfusion dependent; 30% to 50% of patients will need ≥1 platelet transfusion.^1-3,12  With data indicating a high but variable burden of transfusion dependency in this patient population, this review addresses recent research findings to help clinicians answer the questions: How do I decide whether a transfusion is needed for my patient, and if it is, how is it best delivered, in accordance with the principles of patient blood management? Key considerations include the following:

• The role of transfusion among the currently available therapeutic options for MDS

• The aims of, and optimal processes for delivering and monitoring, transfusion supportive care with a focus on patient QoL

Therapeutic approaches for MDS include those directed at ameliorating the underlying bone marrow disease or managing the resulting cytopenias. These options include growth factors such as erythropoiesis-stimulating agents (ESAs) or granulocyte colony-stimulating factor; hypomethylating agents such as azacitidine; immunosuppression or immunomodulation (eg, lenalidomide); chemotherapy; and allogeneic hematopoietic stem cell transplantation, the only current curative option.^1,9  Many novel agents, including molecularly targeted therapies, are in trials or coming into clinical practice. Recently, luspatercept has been shown to minimize anemia and RBC transfusion requirements, with 38% of luspatercept-treated lower-risk patients with MDS and ring sideroblasts becoming transfusion independent for ≥8 of the first 24 weeks on study, compared with 13% in the placebo arm, and with up to one-third of responders remaining transfusion independent for ≤48 weeks.¹³  Starting disease-modifying therapy (DMT) earlier can be associated with better clinical outcomes, but these therapies are not yet widely used in routine practice, mostly because of problems with cost or tolerability.¹⁴ 

As a result, many patients with MDS, especially those with lower-risk disease, are managed with supportive care alone, including transfusion, often for months to years.^1-3,15-18  Therefore, even as new therapies emerge into practice, transfusion is likely to continue to be a central part of MDS management for the foreseeable future.

The main goals of transfusion supportive care are outlined in Table 1. RBC transfusions are given primarily to prevent serious complications of anemia, both acute and chronic, including heart failure and myocardial infarction.^1,15-17,19  Other reasons for transfusion are to manage broader consequences of bone marrow failure, including fatigue and other symptoms related to anemia, and to prevent and manage of bleeding related to thrombocytopenia, aiming to improve patient QoL. However, data on the optimal timing to start transfusions in MDS are lacking, and transfusion is usually introduced on the basis of symptoms and falling blood counts.

Various definitions of RBC transfusion dependency in MDS have been proposed, recently reviewed by Germing et al.¹  These include requirements for 2 RBCs per month, RBC transfusion three or more times in a year, and other variations which include minimum numbers of RBCs or admissions per defined time period – with these definitions being variably applicable either in the routine clinical management setting, in analysis of administrative datasets, or for determining entry or response criteria for clinical trials.^2,18,20  Transfusion intensity can be further described as low (3 to 7 RBCs per 16 weeks) or high (≥8 RBC per 16 weeks), and patients can be described either as not being transfusion dependent or as having low, medium, or high transfusion burden or dependency. These definitions can be helpful for research purposes but are probably less useful in day-to-day practice, except for the purpose of prognostication.

Management of transfusion dependency in MDS is a real conundrum: Transfusions are given with the aim of relieving symptoms, and indeed they can do so, but they provide only transient benefits related to the circulating lifespan of the transfused cells. They also carry the risks of both transfusion-related adverse effects and MDS-associated consequences: If you have MDS, being anemic or thrombocytopenic makes things worse for you, but being RBC or platelet transfusion dependent has itself been shown to be associated with worse prognosis, regardless of when it develops and even at low dose density.^1,20  This association may relate in part to the underlying disease (the worse the ineffective hemopoiesis, the worse the consequences and the worse the prognosis) but also to the transfusion and its sequelae.

The principles of patient blood management (PBM), such as those described by the International Society of Blood Transfusion²¹  as “an evidence-based, multidisciplinary approach aimed at optimising the care of patients . . . put(ting) the patient at the heart of decisions made around blood transfusion, promoting appropriate use of blood and blood components and the timely use of alternatives where available,” remind us to be patient-focused. Transfusion decisions and processes should be undertaken with the patient’s active participation. We cannot simply apply a one-size-fits-all approach, but how do we determine the right approach and know how to advise an individual patient with MDS? Furthermore, hematologists are trying to balance care of each patient with efforts to minimize risks and costs and make best use of precious community blood supplies. Unfortunately, as outlined in Table 2, we still have many more questions than answers on how best to do this.

The need for high-quality data applicable to hematology/oncology, including MDS, was noted by the International Consensus Conference on PBM²²  and is reflected in current national and international clinical guidelines, where most recommend individualizing therapy but can provide only limited specific guidance on management, including hemoglobin and platelet transfusion thresholds.^15,16,23  The impact of this current uncertainty on MDS transfusion management was documented in a recent practice survey of Australian and New Zealand hematologists.²⁴ 

Consider the following cases. How would you approach the transfusion decision for each patient? What factors would influence your decision?

• Case 1: A 80-year-old woman with low- to intermediate-risk MDS who lives at home with her husband presents to clinic for review. She has no past history of cardiovascular disease, and her renal function is normal. She denies any symptoms of fatigue or dyspnea. She has previously had a trial of erythropoietin for management of anemia, with no response. Her most recent blood tests included a hemoglobin concentration of 85 g/L. You consider whether to administer an RBC transfusion.

• Case 2: A 63-year-old man with low- to intermediate-risk MDS who lives at home with his wife presents to the clinic for review. His past history is significant for a myocardial infarction 1 year earlier. He has no chest pain, dyspnea, or fatigue. His most recent blood tests showed a hemoglobin concentration of 78 g/L and normal renal function. You consider whether to administer an RBC transfusion.

• Case 3: A 68-year-old woman with intermediate-risk MDS who lives alone presents to the clinic for review. She reports exertional dyspnea and fatigue but says these features have been long-standing. She has no history of cardiovascular disease, and physical examination reveals no signs of cardiac failure. Her renal function is normal. Her last RBC transfusion was 4 weeks ago. At what hemoglobin threshold would you initiate an RBC transfusion?

Although nearly 20,000 patients have been enrolled in randomized trials comparing different hemoglobin thresholds, these have been almost exclusively in the acute anemia setting, most often in critical care or cardiac surgery, and often aimed at addressing primary outcomes of short-term (eg, in hospital) mortality. It is inappropriate to extrapolate the results from these trials to recommend a “restrictive” policy in transfusion-dependent MDS.

Unfortunately, few transfusion trials have been conducted for patients with hematological malignancies, and no studies have looked at hemoglobin threshold or outcomes studies for chronically transfused patients with other blood diseases with cytopenias, such as aplastic anemia or myelofibrosis.^22,25  A recent study of adults undergoing hematopoietic stem cell transplantation showed that a restrictive (hemoglobin [Hb] <70 g/L) threshold for RBC transfusion was safe and delivered the same QoL outcomes as a liberal (Hb <90 g/L) strategy; however, patients were young (median age 57 years), the setting was primarily hospital-based (and median hospital length of stay was 23 days) with follow-up to 100 days, and patients had short-term transfusion needs.²⁶  There are challenges in extrapolating these trial data, or data from other chronically anemic patients (including those with renal impairment, now generally managed with ESAs) to patients with MDS in the community, because a significant number of patients with MDS have comorbidities such as cardiovascular disease. We do not know whether a more liberal transfusion policy is indicated for patients with cardiac disease, including in the setting of MDS specifically.²⁷ 

With this uncertainty in mind, an international group recently conducted the Red Blood Cell Transfusion Schedule in Myelodysplastic Syndromes (REDDS) pilot trial in transfusion-dependent MDS. Patients were assigned to a restrictive (Hb 80 g/L, to maintain hemoglobin 85 to 100 g/L) or liberal (Hb 105 g/L, maintaining 110 to 125 g/L) threshold for RBC transfusion, with health-related QoL measured via the EORTC QLQ-C30 and EQ-5D-5L.²⁸  The primary outcome (adherence to assigned study arm) was shown to be feasible, and the protocol for outpatient transfusion was successfully implemented across multiple sites internationally. The investigators noted several additional points: Patients assigned to the liberal transfusion arm received about twice as many RBCs as the restrictive arm, and the time interval between transfusions was shorter for patients in the liberal arm. This has implications for both patients and transfusion services, given the need for more visits and more blood bank activities. A post hoc exploratory analysis suggested that the 5 main QoL domains were improved for participants in the liberal arm, supporting the need for additional research to elucidate the impact of different RBC transfusion policies. Finally, the REDDS analysis highlighted the need to consider not just hemoglobin concentration but also swings of amplitude, as demonstrated in an earlier modeling analysis.²⁹ 

Additional research is needed, and a number of other studies are registered (Canada [NCT 02099669], France [NCT03643042], and the international REDDS2 trial [ACTRN12619001053112p]). However, larger studies of transfusion thresholds in MDS will be challenging, with limited numbers of potentially eligible patients at any given site.

Most trials of RBC transfusion thresholds have used short-term mortality as the primary outcome. Trials of interventions to address anemia, such as ESAs, have mostly measured increase in hemoglobin as a trial outcome. Although this measurement is convenient and inexpensive, it does not tell the whole story, and we need to look at the clinical impact on patients, particularly functional outcomes and QoL, through relief of symptoms for which the intervention is being given. This also applies to RBC transfusion. Documentation of patient-reported outcomes (PROs) via patient-reported outcome measures (PROMs) is being included more routinely in clinical trials and practice. Functional outcomes are particularly important for outpatients who are managing their physical ADLs as well as the social and community aspects of their lives.

Table 3 summarizes clinical studies that have assessed QoL and functional outcomes related to RBC transfusion for patients with MDS.^28-34  In addition to the REDDS pilot trial,²⁸  a few observational studies have measured the impact of RBC transfusion on QoL or functional outcomes, and 1 small randomized trial compared RBCs of different storage age. In the RETRO study (which included patients with a range of hematology/oncology diagnoses including MDS), RBC transfusion was associated with improvement in some (fatigue, walk distance) but not all (dyspnea) outcomes, particularly if hemoglobin was maintained at ≥80 g/L 1 week after transfusion.³⁰  In a subsequent analysis from the RETRO study, Bruhn et al³¹  studied the impact of RBC transfusions over 4 weeks on patient-reported fatigue via serial FACIT-Fatigue scores and noted improvement in early post-transfusion scores, without real change after that. Notably, responses varied widely between patients. Chan et al.³²  used a variety of tools to study QoL after RBC transfusion for medical patients, of whom some had MDS. Worse pretransfusion QoL scores predicted post-transfusion improvement. These observations are also supported by trials that have shown improved QoL for patients with MDS with higher hemoglobin, regardless of how it is achieved (eg, in the Nordic study where patients received darbepoetin with or without granulocyte colony-stimulating factor or transfusion, or both, to achieve a target Hb of ≥120 g/L).³⁵ 

In addition to inclusion of PROs in clinical trials, implementation of PROMs in routine clinical practice is increasingly gaining interest. Use of PROMs to inform cancer care has been shown to improve outcomes including QoL, survival, and emergency department visits in other cancers. However, few studies have been done in hematological cancer, and to our knowledge none are specific to transfusion management. Extrapolating from other cancers, potential benefits from routine use of PROMs include improved accuracy of symptom assessment, improved patient–clinician communication, shared medical decision making, and improved QoL.³⁶  The choice of PROM is important for both clinical studies and practice, because there is different information to be gained from generic QoL tools such as the EQ5D (which are generally simple and fast to complete and permit comparisons across different conditions) with data obtained from condition-specific (eg, MDS) or symptom-specific (eg, anemia) tools, which tend to be more targeted and comprehensive but where data may be more difficult to obtain. We should also distinguish between information gained from completing a standard questionnaire compared with providing a description of experience in the patient’s own words (eg, data that may be generated from individual interviews or focus groups).⁴  Moreover, it should be recognized that patient and physician perceptions may differ substantially, either in emphasis (what is relevant to one may not be important to the other) or the importance attributed to it. For example, in one study physicians tended to give more positive scores about QoL than did their patients with MDS, and patients placed greater importance on the disruption to daily life by frequent hospital visits for transfusion than did physicians.³⁷ 

The question of hemoglobin thresholds for RBC transfusion has been further highlighted by feedback from patients with MDS. A large, multicountry survey presented at ASH 2018 reported that 40% of patients with MDS wished they had received RBC transfusions at higher hemoglobin thresholds than they currently were, suggesting that patients are aware of and concerned about the impact of anemia and how it is managed.³⁸ 

More research is needed to elucidate the relationship between anemia and thrombocytopenia and the range of symptoms being reported (including where there may be interactions) before and after transfusion. As this field develops, we can expect to see objectively recorded functional outcomes and PROs being considered absolutely central to the evaluations of both new agents and supportive care interventions, and we can also anticipate that new technologies such as wearable devices (eg, smart watches) will be more widely used for continuous monitoring and capture of these data.

The recent REDDS trial illustrated the high frequency with which transfusion-dependent MDS outpatients need hospitalization.²⁸  Not surprisingly, the need for multiple visits is a major burden and disruption for transfusion-dependent patients with MDS. Many aspects of our current outpatient transfusion policies have been designed to fit the needs of busy clinics and transfusion departments rather than the needs of patients. However, the optimal way to deliver outpatient transfusion services is not known. The REDDS findings indicate the need for additional studies to evaluate the impacts and costs of changing transfusion policies, because changes would have important implications for patients (frequent travel and clinic visits) and hospitals alike.

For some patients, transfusions on weekends, or even at home, might be an option. It is unclear how widespread is the use of home transfusion services for patients with MDS; however, these have been described, and where adequately resourced they may help reduce the travel and attendance burden for selected patients.³⁹ 

Transfusion complications in transfused patients with MDS range from common febrile reactions to uncommon but more serious adverse effects. Few reliable data are available on rates of these complications in patients with MDS, but many cases have been reported to hemovigilance programs, perhaps reflecting the high transfusion exposures in these patients. Transfusion-associated circulatory overload (TACO) is a particular concern for older adults, many of whom have cardiorespiratory comorbidities. Whether transfusing fewer units at a time, or more slowly, or accompanied by prophylactic diuretics reduces TACO risk for patients with MDS is not known.

Identifying and managing transfusion reactions in the outpatient setting can be difficult; these events are easily missed between visits or attributed to other causes, and patients need specific instructions on whom to contact and what to report. Many transfusion adverse events are process-related, and the complexity of the multiple interconnecting clinical and laboratory processes contributes to errors and delays. Efforts to simplify the outpatient transfusion process may reduce these hazards.

Alloimmunization to RBC antigens is a major concern, with most studies describing rates of 10% to 23% in transfusion-dependent patients with MDS, but much higher rates have been reported.^40-42  Alloimmunization appears to correlate with total RBC exposure and exposure to platelet units unmatched for RBC antigens, rather than inherent immunogenicity of the RBC units or the known immune dysregulation present in the underlying MDS, but it is probably multifactorial. A significant proportion of alloimmunized patients also develop autoantibodies, which may compound hemolytic complications.⁴⁰  A wide variety of RBC antibody specificities have been reported, but antibodies to K and the Rh system antigens (especially anti-E) are the most common; therefore, providing RBCs negative for these antigens may be sufficient to minimize alloimmunization for most patients.⁴²  However, more extensive matching, including genotyping, may confer additional benefit and is being increasingly used in routine practice before chronic transfusion programs.⁴³  In retrospective studies, azacitidine use has been reported to reduce rates of RBC alloimmunization.⁴⁴ 

Iron overload is commonly identified in transfusion-dependent MDS, caused by the dual problems of ineffective hemopoiesis or deranged iron metabolism from MDS and iron loading from transfused RBCs.^1,9  The toxic effects of labile plasma iron seem to play an important role in end-organ damage. Ferritin levels have traditionally been monitored to document iron overload, and an elevated level has been shown to be an independent risk factor for mortality. However, hyperferritinemia is not specific for iron overload and is starting to be supplemented by monitoring of other parameters in trials and clinical practice.

Iron chelation has been reported to reduce transfusion requirements and should be considered early for patients who become transfusion dependent.¹  However, determining who is a candidate for chelation can be difficult; for instance, higher-risk patients needing greater transfusion intensity will become iron loaded more quickly but may not benefit from chelation because of their shorter survival. Furthermore, chelation success still depends on compliance with therapy, but delivering tolerable, clinically effective and cost-effective iron chelation has been challenging, particularly until the advent of oral agents, and all chelation agents carry costs and the risk of adverse effects.

Guidelines variably recommend commencing chelation at ferritin >1,000 μg/L or after receipt of 20 to 30 U of RBCs. The recent TELESTO trial randomly assigned 225 patients with low- to intermediate-risk MDS who had received 15 to 75 U of RBCs before study entry to deferasirox or placebo.⁴⁵  Because of slow recruitment, the trial was changed from a phase 3 to phase 2 trial. Event-free survival (worsening cardiac and liver function, need for hospitalization for congestive heart failure, acute myeloid leukemia transformation) was improved in the treatment group; adverse events were common. Longer follow-up in this study will be interesting, because few patients were followed at later time points, and it remains to be seen whether these results will be taken up in routine practice, because historically chelation rates have been low in MDS, especially for older and frail patients, even though chelation improves clinical outcomes.⁴⁶ 

Up to 50% of patients need some platelet transfusion support. Routine prophylaxis (to prevent bleeding) with platelet transfusion for patients with MDS and thrombocytopenia who are not undergoing intensive therapy is not recommended, but few reliable data exist on how commonly this is performed in practice or how effective platelet transfusions are.^10-12  There is no international definition of platelet transfusion dependency in the MDS setting.

Dosing of platelet transfusion has largely been extrapolated from studies of other hematological malignancies. In a retrospective review of outpatient platelet transfusions, where 57% of the patients had leukemia or MDS, there was no clear advantage in transfusing 2 U of platelets compared with 1 U.⁴⁷  In a recent analysis from the South Australian MDS Registry, 9% of all patients (and 30% of women receiving DMT, compared with only 5% of men receiving DMT) who received a platelet transfusion developed immune-mediated refractoriness to platelet transfusions and needed HLA-matched platelets for future transfusion support, increasing the risks of bleeding and the costs and complexity of care.¹² 

One approach to severe thrombocytopenia in MDS is to offer routine tranexamic acid. A number of randomized trials are testing the safety of tranexamic acid in acute settings of hematological malignancies, although none are specifically recruiting outpatients with MDS.

New understanding of the MDS disease process and the availability of novel therapies offer the prospect of major advances in management and outcomes for patients over the coming years. However, transfusion is still an important part of MDS supportive care. Although blood components are safe in most countries, they carry substantial risks and costs, and patients who are chronically transfused are recurrently exposed to these hazards. We need to make every management decision, including each transfusion decision, carefully, guided by the available evidence and the principles of PBM to deliver personalized MDS and transfusion therapy based on an individual patient’s needs and with their participation.⁴⁸  We also need to monitor the impact of our transfusion therapy, including the effects on QoL. Many questions remain, some of which are summarized in Table 2. By designing and contributing to high-quality, patient-centered research using a range of approaches, including clinical registries to provide real-world practice and outcome data, and performing well-conducted clinical trials incorporating PROs and functional assessments, we can strengthen the evidence base to guide our practice and improve outcomes for transfused patients with MDS.

The authors thank Simon Stanworth, Allison Mo, Robert Weinkove, and Devendra Hiwase for helpful discussions.

Erica Wood, Monash University, 553 St Kilda Road, Melbourne, 3004, Australia; email: erica.wood@monash.edu.