Drugs to improve anaemia, quality of life, and physical function in people with myelodysplastic syndromes (MDS)

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To evaluate the benefits and harms of both supportive and disease‐modifying drug treatments for anaemia in people with MDS, comparing them to each other, or the standard of care with or without placebo.

Background

Myelodysplastic syndromes (MDS) are a group of acquired bone marrow failure malignancies, in which people with the condition develop progressive cytopenias and abnormal haematopoiesis. This results in progressive anaemia, thrombocytopenia and/or neutropenia, and there is an increased risk of progression to acute myeloid leukaemia (AML).

Anaemia is very common in people with MDS (1), but there is uncertainty about how to best treat it, and the full impact of different treatments on health‐related quality of life (HRQoL) and physical function is unknown. Red cell transfusion is often given to treat anaemia, and many people with anaemia develop transfusion dependence (requiring regular transfusion).(2)

Symptoms from anaemia are an almost universal feature of MDS (3), including fatigue and reduced physical capacity (4). With no definitive curative option available for many, learning to live with MDS, manage symptoms effectively, and improve HRQoL are priorities in their care. Individuals’ experience of these symptoms is variable (5), and there is limited appreciation of the heterogeneity of the disease or the trajectory of people’s journeys with MDS in current written patient information. Symptoms and experience may often be poorly documented in clinical notes.

There are multiple treatment options for MDS, ranging from supportive care interventions to ameliorate symptoms and reduce anaemia, to disease‐modifying agents, such as hypomethylating or immunomodulatory agents, which have more direct treatment effects on the disease. It can be difficult to select the best treatment options for people with MDS, therefore a better understanding of the effect of these treatments on anaemia and hence HRQoL and physical function is needed to inform treatment decisions. This review would consolidate the knowledge base in this area.

Description of the condition

MDS is a rare disease of the blood, but has an increasing incidence in the elderly population, with an average age of 71 years (6). MDS is a clonal bone marrow failure disorder characterised by cytopenias and ineffective haematopoiesis. MDS is normally suspected due to unexpectedly low blood counts (cytopenias) seen on a full blood count test, and cytopenia in at least one cell line is a prerequisite to diagnosis (7).

Formal diagnosis is complex, requiring a bone marrow biopsy showing morphological evidence of dysplastic features, along with molecular and cytogenetic testing. There are multiple subtypes of MDS predominantly depending on the genetic mutations as defined by the World Health Organization (WHO) (7), although treatments are similar across subtypes. There is one variation of MDS which is diagnosed with a specific genetic abnormality (del(5q)) and has an effective, specific treatment (lenalidomide) (8). Additionally, MDS is subclassified into low‐, intermediate‐, and high‐risk subtypes based on prognostic factors such as full blood count (FBC) variables, bone marrow blast % and cytogenetics (9). Treatment decisions tend to be based on risk category rather than disease subtype, with the exception of del(5q) MDS. (9)

MDS is more common in older individuals, with a median age of diagnosis of 68 to 75 years. This is partly due to an increasing incidence of clonal haematopoiesis (CHIP) mutations with advancing age, which increases risk of haematologic malignancies and other biologic features of ageing (10). Thus, many people with MDS cannot receive the only curative therapy (allogeneic stem cell transplantation) due to age and comorbidities. Less than 10% of people with MDS are younger than 50 years. (11) Therefore, supportive care is paramount to improve HRQoL where curative treatment is not possible. Anaemia is common and associated with a poorer HRQoL (1). 

It is unclear whether treatment for anaemia alone is effective in improving HRQoL and physical function in people with MDS. Previous research has shown that the negative impact of MDS, particularly on usual activities, self‐care, anxiety and depression, is often not mediated via anaemia (12). Possible explanations for this include pathogenic factors causing reduced HRQoL and physical function beyond anaemia, or difficulties in measuring HRQoL and physical function accurately.

Description of the intervention and how it might work

Description of the intervention

In MDS, anaemia occurs because the bone marrow is ineffective at producing mature, functional red blood cells.

The interventions to treat anaemia in MDS fall into two categories: disease‐modifying agents, and supportive care (primarily targeting symptoms such as anaemia) (13).

Disease‐modifying agents typically involve medications (Table 1) which aim to alter the natural history of the disease, such as by suppressing/destroying the abnormal clone, improve marrow function long term via epigenetic regulation or immune dysregulation or delay progression to acute leukaemia. Examples include hypomethylating agents, lenalidomide, allogeneic stem cell transplant, and targeted therapies.

1. Interventions that aim to treat anaemia in myelodysplastic syndromes (MDS).

Drugclass	Drugs	Example dose	Mechanism of action*
Hypomethylating agents	Azacitidine, decitabine	Azacitidine: 75 mg/m2 (7 days) Decitabine: 15 to 20 mg/m2 (5 days)	Disease modifying: induces hypomethylation causing derepression of tumour suppressor genes
Immunomodulatory drugs (IMiDs)	Lenalidomide	10 mg once daily	Disease modifying: immunomodulation and direct inhibition of malignant cells
Chemotherapy (including cytarabine)	Cytarabine, fludarabine/FLAG‐IDA, idarubicin, daunorubicin	Variable	Disease modifying: antimitotic and cytotoxic activity
Anti‐apoptotic agents (including venetoclax)	Venetoclax	100 mg once daily	Disease modifying: BCL‐2 inhibition
Signal transduction inhibitors (including rigosertib)	Rigosertib	Variable (650 to 1375 mg/m2)	Disease modifying: microtubule destabilising agent
Immune checkpoint inhibitors	Pembrolizumab, nivolumab, magrolimab	Pembrolizumab 10 mg/kg Nivolumab 3 mg/kg fortnightly Magrolimab 30 mg/kg	Disease modifying: block checkpoint proteins from binding with their partner proteins causing T‐cell mediated destruction
Erythropoiesis stimulating agents (ESAs)	Erythropoietin (EPO) Darbepoetin (DPO)	EPO 450 units/kg weekly DPO 500 μg 3 weekly	Anaemia targeting: stimulates red cell production
Erythroid maturation agents (EMAs) such as luspatercept	Luspatercept	1 mg/kg every 3 weeks	Anaemia targeting: ameliorates ineffective erythropoiesis
Telomerase inhibitors (e.g. imetelstat)	Imetelstat	7.1 mg/kg every 4 weeks	Disease modifying: acts as a competitive inhibitor of human telomerase enzymatic activity

*Disease‐modifying agents work by targeting the abnormal clone of cells to either destroy them or prevent replication. Supportive care interventions do not affect the abnormal clone of cells causing MDS, but instead ameliorate the symptoms by optimising haematopoiesis and targeting improvement of anaemia.

Supportive care treatments aim to improve symptoms and quality of life (QoL) without changing the underlying disease biology. Examples include red cell transfusion and growth factors. These do not suppress or destroy the abnormal clone.

A non‐exhaustive list of the different interventions is summarised in Table 1.

How the intervention might work

Disease‐modifying agents work by targeting the abnormal clone of cells to either destroy them or prevent replication. Their mechanisms of action are summarised in Table 1. Supportive care interventions do not affect the abnormal clone of cells causing MDS but instead ameliorate the symptoms by optimising haematopoiesis and targeting an improvement of anaemia.

Alongside supportive care treatments, disease‐modifying treatments are normally used in people with higher‐risk MDS who are more at risk of developing AML. By targeting the abnormal MDS clone of cells, disease‐modifying treatments can help to delay the progression to AML. Side effects of these treatments vary but often include reducing healthy blood cells, hence increasing the incidence of anaemia, bleeding, and infections and therefore reducing QoL independently of the MDS itself; it is for this reason that disease‐modifying agents are reserved until needed. This makes determining the overall benefit of these treatments on QoL difficult. Those with lower‐risk MDS usually only receive supportive care until the MDS is more advanced.

Why it is important to do this review

Patient‐reported outcome measures, particularly HRQoL, remain underreported in studies involving people with MDS (14). Additionally, physical functional outcomes are rarely reported. These represent a critical gap in the evidence base. Many people with MDS describe their care as being focused primarily on survival rather than quality of living. As curative options are unavailable for a substantial proportion of people, optimising symptom control, supporting adaptation to chronic disease, and improving HRQoL are central goals of care (15). A systematic evaluation of how current treatment strategies affect HRQoL is therefore essential to ensure that management approaches align with outcomes that matter most to people with MDS.

Anaemia – one of the most common and burdensome manifestations of MDS – is consistently associated with reduced HRQoL and increased fatigue (1). However, the relationship is multifactorial and incompletely understood. Previous systematic reviews have highlighted the inconsistent evidence behind transfusion approaches in MDS and how these can best be optimised to improve HRQoL (16, 17, 18). However, these reviews have largely focused on transfusion and/or other supportive care (such as growth factors), rather than all MDS‐related treatments. In addition, several disease‐modifying therapies may initially worsen HRQoL before potential clinical benefits are realised. Without systematically synthesised evidence, clinicians and people with MDS cannot adequately weigh these short‐ and long‐term effects when making treatment decisions. This review is important because it will clarify the balance between clinical efficacy (in treating anaemia) and patient‐experienced outcomes (QoL and physical function).

There is currently insufficient high‐quality, consolidated evidence to guide optimal treatment strategies in MDS, particularly with respect to patient‐centred outcomes. By systematically evaluating the available data, this review will help define the impact of different interventions on HRQoL and physical function. This is particularly important in a population with high transfusion requirements, where transfusion dependence can substantially impair QoL. Better defining clinical outcomes beyond laboratory parameters will support shared decision‐making, inform guideline development, and assist policymakers in prioritising interventions that improve both survival and QoL.

The importance of this topic is further underscored by its identification as a “top five” priority in the Cochrane Haematology Priority Setting Process (19). This prioritisation reflects recognition from clinicians, researchers, and people with MDS that robust synthesis of HRQoL evidence in MDS is urgently needed to improve care.

Objectives

To evaluate the benefits and harms of both supportive and disease‐modifying drug treatments for anaemia in people with MDS, comparing them to each other, or the standard of care with or without placebo.

Methods

We will conduct this systematic review according to the Methodological Expectations for Cochrane Intervention Reviews (MECIR) checklist for the conduct of the review (20) and the PRISMA 2020 checklist for the reporting (21). We will follow the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (22).

Criteria for considering studies for this review

Types of studies

We will include randomised controlled trials (RCTs) and cluster randomised trials (cRCTs). We will also include cross‐over trials, but will only use data from the first period of randomisation, as the effects of the first intervention may persist into the period of the second intervention. We will exclude non‐randomised studies of interventions (NRSI) including quasi‐randomised studies, as we expect to find sufficient RCT data for inclusion in the review.

We will include published abstracts of RCTs if they report data. We will include studies published at any date and in any language, although our search strategy will be focused on searching for English language terms. We will exclude results that have not undergone peer review (i.e. preprints or results that are only available in clinical trials registries).

Types of participants

We will include studies in participants meeting the following criteria.

• Aged ≥ 18 years  

• WHO‐defined MDS or mixed myeloproliferative/myelodysplastic neoplasm overlap syndromes (MPN/MDS), including chronic myelomonocytic leukaemia (CMML) and AML with < 30% blasts 

• Life expectancy ≥ 6 months 

 We will include studies of populations with all types of MDS, including the del(5q) mutation. In studies with mixed populations, we will include studies with > 80% of included participants meeting the above criteria. If the study includes a subpopulation of participants with any type of MDS aged over 18 years for whom the outcome data are presented separately, we will include this information. 

We will include trials carried out in both inpatient and outpatient settings, and from any country.

Types of interventions

We will investigate MDS drug treatments that aim to treat anaemia, including the following.

• Hypomethylating agents (including azacitidine, decitabine)

• Lenalidomide

• Chemotherapy (including cytarabine)

• Anti‐apoptotic agents (including venetoclax)

• Signal transduction inhibitors (including rigosertib)

• Immune checkpoint inhibitors

• Targeted antibody therapies such as magrolimab

• Other targeted therapies

• Erythropoiesis‐stimulating agents (ESAs)

• Erythroid maturation agents (EMAs)

• Imetelstat (telomerase inhibitor)

Further details of interventions can be found in Table 1.

Comparators

We will compare the drugs used to placebo or best supportive care (as defined by trialists), or to one or more drugs from a different class, as described in Table 2. Where best supportive care includes an intervention of interest in the review, we will analyse the data as the appropriate drug‐versus‐drug comparison. If we are unable to establish what constitutes best supportive care, we will analyse the data in combination with studies that compare the intervention to placebo. We will consider clinician's choice of intervention as equivalent to best supportive care. Where two or more different classes of drugs are given concurrently, we will analyse these as a separate comparison. The comparisons we plan to present in the summary of findings tables are listed in Table 2.

2. Predetermined drug comparisons.

Intervention	Comparison
Comparisons included in summary of findings table
Any drug treatment	Placebo
Any drug treatment	Control/standard of care
Hypomethylating agents, subcutaneous	Hypomethylating agents, oral
Venetoclax and azacitidine	Hypomethylating agents
Additional comparisons not presented in summary of findings table
Hypomethylating agents	Chemotherapy
Hypomethylating agents	Targeted therapies individually*
Erythropoiesis‐stimulating agent	Luspatercept
Erythropoiesis‐stimulating agent	Imetelstat

*Magrolimab and venetoclax are normally given in combination with hypomethylating agents.

If there is evidence of multiple studies of other drug combinations in the identified studies, we will add to these groups.

When assessing standard of care, it is unlikely there will be multiple drug combinations, except possibly erythropoiesis‐stimulating agent plus another. This is due to the nature of myelodysplastic syndromes (MDS) treatment, whereby one treatment is used until failure and then switched. The exception to this is venetoclax‐azacitidine and magrolimab‐azacitidine, which are listed above.

These comparisons will be done for both the specified early and late time points where data are available.

We will not consider transfusion alone to be an eligible intervention of interest in this review, but only in combination with another drug treatment. This is because transfusions are a common and expected aspect of supportive care, and there is much variability in their use and the haemoglobin transfusion threshold used. Many people with MDS are given transfusions as part of supportive care, and the number and frequency of transfusions is often not recorded in studies as it may not be considered a 'drug intervention'. It would therefore not be possible to make comparisons when transfusion data is not accurately recorded in control groups.

We will exclude studies that compare drugs from the same class of drugs (intraclass comparisons), as defined in Table 1. The exception to this is the comparison of subcutaneous versus oral hypomethylating agents, as this was highlighted by patient representatives as being of particular importance for QoL. To aid future study in this area, we will summarise all studies excluded as within‐class drug comparisons in our 'Characteristics of excluded studies' table, and we will specifically comment on whether QoL outcomes are reported in these studies.

Iron as a treatment will not be included in this review, for the following reasons.

• Iron is not a specific treatment for MDS‐related anaemia, but for iron deficiency anaemia. Although it is acknowledged that people with MDS may also develop iron deficiency anaemia, this is a separate problem. There have been multiple Cochrane reviews already on the use of iron to treat anaemia (23, 24).

• Similarly, other treatments for anaemia that are not specific to MDS‐related anaemia (e.g. vitamin B₁₂ or folate supplementation) or treatments for anaemia due to non‐MDS causes (e.g. haemolysis) are not included. This is because they are not treatments for MDS‐related anaemia.

• As iron is an over‐the‐counter medication in many countries and may have widespread use in the community, its use may not be accurately captured in studies (where it is not the primary treatment being studied). Thus, it would be difficult to draw conclusions on the use of iron in this context.

Outcome measures

We have chosen these outcomes in order to be relevant to both our review questions and to consumers, clinicians, and policymakers as proposed by COMET (Core Outcome Measures in Effectiveness Trials) in partnership with Cochrane (25). COMET has worked to develop core outcomes in MDS, of which haematological improvement and HRQoL rank highly (26). For each outcome in our review, we will extract data at the following time points:

• early term: up to six months from starting the study treatment;

• late: 12 months from starting the study treatment, and at the latest time point.

 Where a study reports multiple time points, we will extract data for the closest time points that match our prespecified definitions of early and late term. We will add footnotes to the analysis to denote any time points that do not exactly match our prespecified time points. We will report both early‐term and late time points as part of our primary analyses.

Critical outcomes

We will present all of our critical outcomes in summary of findings tables. We will create these tables using the GRADE approach (27) and GRADEpro GDT software (28). Two of three review authors (AM, MP, CK) will conduct GRADE assessments of the certainty of evidence. If the authors cannot reach a consensus on the certainty of the evidence, they will consult another review author (SS). We will present data as separate outcomes for our defined early (six months) and late (12 months) time points, with the late time point being presented in the summary of findings tables for each critical outcome. In addition, we will present the early time point as a separate outcome for the erythroid response and HRQoL outcomes, as detailed below.

1. Erythroid response at six and 12 months

We will report erythroid response as defined and reported by trialists.  We will present data as separate outcomes for the early (six months) and late (12 months) time points.

We acknowledge that different studies may use different erythroid response criteria; however, we will aim to report erythroid response as a change in haemoglobin and number of transfusions. This method has been determined by International Working Group ​​(IWG) 2006 criteria (29): a haemoglobin increase by ≥ 15 g/L lasting for at least eight weeks in transfusion‐independent patients, or a reduction in red blood cell (RBC) transfusions by at least four transfusions over eight weeks in transfusion‐dependent patients is defined as an erythroid response. Where trialists report multiple definitions of erythroid response, we will prefer the IWG 2018 erythroid response criteria (30), which are based upon the patient's baseline transfusion dependence category (non‐transfused versus low versus high transfusion burden).

HI‐E criteria depending on transfusion dependence category (2018 IWG criteria)

Transfusion dependence category (2018 IWG)	HI‐E criteria (2018 IWG)
Non‐transfused (defined as 0 RBCs in 16 weeks)	At least 2 consecutive haemoglobin measurements of 15 g/L or greater for a minimum 8‐week period, in an observation period of 16 to 24 weeks, compared to the lowest mean of 2 haemoglobin measurements (apart from any transfusion) within 16 weeks prior treatment onset. A response duration of at least 16 weeks is considered clinically meaningful.
Low transfusion burden (defined as 3 to 7 RBCs in 16 weeks in at least 2 transfusion episodes, maximum 3 in 8 weeks)	Transfusion independence, defined as absence of any transfusions for at least 8 weeks in any observation period of 16 to 24 weeks (using the same transfusion policy as the 16 weeks prior to treatment). A response duration of at least 16 weeks is considered clinically meaningful.
High transfusion burden (defined as at least 8 RBCs in 16 weeks, at least 4 RBCs in 8 weeks)	Major response: transfusion independence, defined as absence of any transfusions for at least 8 weeks in any observation period of 16 to 24 weeks (using the same transfusion policy as the 16 weeks prior to treatment). A response duration of at least 16 weeks is considered clinically meaningful. Minor response: reduction by at least 50% of RBCs over a minimum of 16 weeks (using the same transfusion policy as the 16 weeks prior to treatment).

2. Health‐related quality of life at six and 12 months

We will extract patient‐reported HRQoL using a QoL scale validated for use in people with anaemia. We will present data as separate outcomes for both early and late time points.

We will extract data for the following scales, which have been validated for use in anaemia.

•  Anaemia‐dependent QoL instruments: Functional Assessment of Cancer Therapy‐Anemia (FACT‐An) (31)

• Overall QoL – global Patient‐Reported Outcome Measure (PROM) (total score): EORTC QLQ‐C30 (32), Functional Assessment of Cancer Therapy‐General (FACT‐G) (33), 36‐item Short Form Health Survey (SF‐36) (34), EQ‐5D (35)

• Disease‐specific PROM: QOL‐E (36), Quality of Life in Myelodysplasia Scale (QUALMS) (37)

To account for the multiplicity of data when outcomes are reported using different measurement instruments, we will prioritise instruments for extraction and analysis based on their usage frequency across trials and applicability in routine clinical practice, and preferentially choose instruments that provide information on the construct of anaemia‐dependent QoL, its severity, or giving a total score for anaemia‐dependent QoL. The Core Outcome Set does not specify any preferred tools for measuring HRQoL. We will therefore extract any reporting of HRQoL, but where a study reports using more than one instrument, we will prefer tools that are validated specifically for MDS, and the most widely used tools applicable in this area. Specifically, we will prioritise the following.

• QUALMS

• QOL‐E

• FACT‐An

• QLQ‐C30

Details of the quality of life scales we expect to encounter in our included studies are available in Table 3. If studies report quality of life using more than one measurement instrument, we will report and list all instruments used. Should, within a trial, different instruments measuring the same construct report discrepant findings, we will assess the reasons for this and consider treating this study as at high risk of bias if it is clear how the discrepancies have arisen. Where we are unable to resolve serious discrepancies, we may exclude the result from the analysis.

3. Quality of life scales.

Name	Number of items	Purpose	Score interpretation – higher or lower score indicates improvement
FACT‐An	47	13 of the 20 items deal with fatigue, while the remaining 7 cover other concerns related to anaemia. The FACT‐An is a useful measure of quality of life in cancer treatment, adding more focus to the problems of fatigue and anaemia.	Higher
EORTC QLQ‐C30	30	Designed to measure physical, psychological, and social functions in people with cancer	Higher
FACT‐G	27	Designed to measure 4 domains of HRQoL in people with cancer: physical, social, emotional, and functional well‐being	Higher
SF‐36	36	Measures patient health status outcomes such as emotional and social function, mental and general health, physical role – performance, vitality (energy and fatigue), and pain	Higher
EQ‐5D	5	Standardised, generic questionnaire used globally to measure and value patient‐reported HRQoL	Higher
QOL‐E	29	HRQoL instrument designed specifically for people with myelodysplastic syndromes	Higher
QUALMS	33	Disease‐specific tool designed to measure the quality of life of people with myelodysplastic syndromes	Higher

Abbreviations: HRQoL: health‐related quality of life

We will describe the specific instrument or tool used and the timing of the assessment. We will evaluate changes in HRQoL and physical function against erythroid response, and comment on whether there is a correlation between the two (18).

3. Mean physical function score at 12 months

We will extract any objective tests of physical function at 12 months, including, but not limited to, tests of balance, daily activity levels (e.g. accelerometer devices, participant diary of physical activity), handgrip strength tests, walking tests (e.g. Timed Up and Go test, 6‐minute walk test).  We will describe the specific measurement taken and the timing of the assessment. Changes in physical function will be evaluated against erythroid response (18).

4. Risk of (or time to) transformation to AML at up to 12 months

Many people with MDS will suffer various adverse events including infections and bleeding. However, transformation to AML is a key endpoint, as approximately 30% of people with MDS will develop overt AML during the course of their disease. Many studies of disease‐modifying drugs affecting MDS use transformation to AML as a primary endpoint. We have prioritised transformation to AML at up to 12 months as a key adverse event, as this is of high clinical relevance and is frequently reported in studies of drugs used in MDS treatment. We anticipate that this outcome will be reported in a variety of ways, for example time‐to‐event, or the proportion of participants developing AML at specified time points. For time‐to‐event data, we will present hazard ratios following the methodology outlined by Tierney (38). We will combine results where it is practical to do so, and all results will be reported narratively in tables.

5. Serious adverse events (SAEs) at up to 12 months

We will extract the number of participants with one or more serious adverse events (SAEs) at up to 12 months, as defined and reported by the trialists. Where trialists report only treatment‐related SAEs (T‐SAEs), we will extract data reported as T‐SAEs. However, we will only extract T‐SAEs if full SAE reporting is not available. We will document the study's definition of SAEs or T‐SAEs in footnotes in the analyses.

Important outcomes

We will report the following outcomes in our review at both early (six months) and late (12 months) time points. Where data are not available for our chosen time point, we will extract data for the closest time point and document the precise time point using footnotes in the analyses.

1. Number of transfusions at up to six months and up to 12 months

We will report any data on the number of transfusions. We will present data as separate outcomes for the early (six months) and late (12 months) time points. Where possible, we will extract the mean and standard deviation (mean SD) for the number of transfusions per participant.

2. Haemoglobin concentration at six and 12 months

We will report any data on haemoglobin concentration at both early and late time points. Where possible, we will extract the mean SD. We will present data as separate outcomes for the early (six months) and late (12 months) time points.

3. Treatment‐related morbidities: hospital readmission at up to six months and up to 12 months

We will report any data on participants being readmitted to hospital. We will present data as separate outcomes for the early (six months) and late (12 months) time points. We anticipate that this outcome will be reported in a variety of ways, for example time to hospital readmission, number of readmissions, or proportion of participants readmitted at specified time points. For time‐to‐event data, we will present hazard ratios following the methodology outlined by Tierney (38). We will combine results where it is practical to do so, and all results will be reported narratively in tables.

4. Treatment‐related morbidities: bleeding at up to six months and up to 12 months

We will report the number of participants experiencing bleeding. We will present data as separate outcomes for the early (six months) and late (12 months) time points. Where possible, we will document bleeding according to the WHO grading system (39). We will report the number of participants at both early and late time points experiencing:

• any bleeding (defined as WHO bleeding grade 1 or above);

• clinically significant bleeding (defined as WHO bleeding grade 2 or above);

• severe bleeding (defined as WHO bleeding grade 3 or above).

5. Treatment‐related morbidities: infection at up to six months and up to 12 months

We will report the number of participants experiencing infection at both early and late time points. We will present data as separate outcomes for each time point.

6. All‐cause mortality at up to six months and up to 12 months

We will report the number of participants who died at both early and late time points. We will present data as separate outcomes for each time point.

7. Adverse events at grade 3 to 4 severity at up to six months and up to 12 months

We will report the number of participants experiencing one or more adverse events at grade 3 to 4 severity, at both early and late time points. We will present data as separate outcomes for each time point.

8. Any adverse events at up to six months and up to 12 months

We will report the number of participants experiencing one or more adverse events of any severity, at both early and late time points. We will present data as separate outcomes for each time point.

In addition, we will present the early time point data for the following critical outcomes, where the later time point is presented in the summary of findings tables.

• Mean physical function score at six months

• Risk of (or time to) transformation to AML at up to six months

• SAEs at up to six months

Search methods for identification of studies

Electronic searches

We will search for all relevant published and unpublished trials without restrictions on language, year, or publication status.  We will search for relevant systematic reviews to check their reference lists for trials that may have been missed by the search.

We will search the following electronic databases from their inception: 

• Cochrane Central Register of Controlled Trials (CENTRAL, the Cochrane Library, Wiley, current issue); 

• MEDLINE (Ovid, ALL, 1946 onwards); 

• Embase (Ovid, 1974 onwards); 

• CINAHL (Cumulative Index to Nursing and Allied Health Literature) (EBSCOhost, 1981 onwards); 

• APA PsycINFO (Ovid, 1967 onwards); 

• Transfusion Evidence Library (Evidentia Publishing, 1950 onwards). 

We will search for ongoing or unpublished RCTs in the following clinical trial registers:

• World Health Organization International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch/); 

• ClinicalTrials.gov (www.clinicaltrials.gov/). 

 The MEDLINE search strategy will be developed by an Information Specialist, reviewed by the project team, and then adapted by the Information Specialist for the other databases. There will be no restrictions on publication year. Searches will use a combination of subject headings and free‐text terms and will not be limited by language, year of publication, or publication status. We will combine our searches in MEDLINE, Embase, and CINAHL with adaptations of the SIGN systematic review filters and of the Cochrane sensitivity‐ and precision‐maximising RCT filter (40). The MEDLINE search strategy is presented in Supplementary material 1.

Searching other resources

We will check the reference lists of all identified trials and relevant review articles for further eligible studies not already found by the search. These searches will be limited to the 'first‐generation' reference lists.

Once studies have been screened for inclusion, we will also search MEDLINE, Embase, and the Retraction Watch Database (via EndNote 2025: supportcenter.clarivate.com/s/article/EndNote‐20‐Retraction‐Alerts?language=en_US) for post‐publication amendments (including errata, retractions, corrigenda, and expressions of concern) to all included studies and their related secondary published papers.

Data collection and analysis

We will employ the standard methods of the Cochrane Handbook for Systematic Reviews of Interventions (41). If an author is involved in any of the identified trials, they will not participate in decision‐making regarding inclusion or exclusion of the study in the review, and will not conduct risk of bias assessment or data extraction for that trial. 

Selection of studies

Two of four review authors (MP, CK, CP, AM) will use Covidence software (42) to independently screen the titles and abstracts identified by the search strategy for potential relevance, excluding any studies that are clearly irrelevant at this stage. We will retrieve the full‐text reports of all references deemed potentially relevant, and two review authors (MP, AM) will screen the full texts for inclusion in the review.  We will attempt to obtain a translation of the methods and results sections for studies reported in non‐English language journals before assessment. In the first instance, we will ask for translation assistance from a native speaker from Cochrane Crowd. If we are unable to source a native‐speaker translation, we will create a machine translation using online tools.

Disagreements during screening will be resolved by consensus or in consultation with another review author (SS) if necessary. We will record the reasons for exclusion of studies excluded at the full‐text stage. 

Data extraction and management

Two of three review authors (CK, MP, AM) will independently extract data from included studies following the recommendations in Chapters 5 and 9 of the Cochrane Handbook for Systematic Reviews of Interventions (43, 44). We will extract data using a form that will be piloted on 20% of the included studies.  We will resolve disagreements by consensus or through arbitration with another review author (SS). 

We will extract the following data.

• General information: name of review author carrying out the data extraction, date of data extraction, study identifier, surname and contact address of contact author, language of trial, language of publication, secondary citations, study location, country of study, single or multicentre, funding source, publication type (abstract/full text/protocol), trial registration, year of publication, study dates. 

• Trial information: RCT trial design: setting, sample size, duration of trial, power calculation, treatment arms, randomisation, inclusion and exclusion criteria, comparability of groups, length of study.

• Characteristics of participants: age, sex, weight, type of MDS, breakdown of total numbers for those randomised and analysed, dropouts (percentage in each arm) with reasons and protocol violations.

• Characteristics of intervention: number of treatment arms, description of experimental arm(s), timing, dose, and route of administration of intervention, and other differences between intervention arms, transfusion thresholds/policies (if available).

• Characteristics of control arm: number of control arms, description of control arm(s), timing, dose, and route of administration of control intervention, and other differences between control arms.

• Outcomes, primary and secondary, including timing of assessments and information about how each study defined outcomes (e.g. erythroid response and SAEs).

Risk of bias assessment in included studies

Two of three review authors (MP, AM, CK) will independently assess the risk of bias in each trial at the outcome level using the Cochrane RoB 2 tool (45, 46, 47) and assign it a classification of low risk, some concerns, or high risk (47). We will assess all critical outcomes in the summary of findings tables. We will resolve disputes initially in consultation with the third member of the team conducting risk of bias assessments, and if this is unsuccessful, by consensus with all review authors. We will record all decisions using the RoB 2 Excel tool (45).

We will assess studies by effect of assignment to the intervention, and will assess risk of bias based on the following domains:

• bias arising from the randomisation process;

• bias due to deviations from the intended interventions;

• bias due to missing outcome data;

• bias in measurement of the outcome;

• bias in the selection of the reported result.

For cRCTS, we will use the RoB 2 variant. For cross‐over trials, we will use the standard RoB 2 tool, as we will only consider the first period of such trials (45, 48).

Measures of treatment effect

For continuous outcomes, we will calculate the mean difference (MD) and the corresponding 95% confidence interval (CI), where possible. We will carry out meta‐analysis of the results if there are sufficient data. We will use the standardised mean difference (SMD) if pooled studies measure the same outcome in different ways (e.g. different HRQoL scales). We will pool analysis of the results using Hedge’s g, which provides an estimate of the SMD weighted according to the relative size of each sample and allows for different HRQoL and physical function instruments to be included in a pooled analysis.  

For dichotomous outcomes, we will use the risk ratio (RR) with 95% CI to compare outcomes for each comparison, where possible. If individual studies report odds ratios (ORs) with corresponding CIs that were adjusted for confounding variables and do not provide raw contingency tables, we will pool ORs instead of RRs using the generic inverse variance method.

Unit of analysis issues

The unit of analysis will be the individual participant. We will use the methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions to adjust sample size and measures of variation (e.g. standard errors) for level of randomisation reported in trials (48). If reported, we will use the intracluster correlation coefficient (ICC) to estimate the design effect, based on the guidance provided in the Cochrane Handbook for Systematic Reviews of Interventions (48). Where ICCs are not available, we will apply ICCs from studies of similar interventions. We will undertake sensitivity analysis based on ICCs to examine the effect of combining results from cRCTs. 

For cross‐over trials, we will only use data from the first randomisation period, as it is highly likely that the first period of treatment could affect the outcome of any subsequent randomisation. We will therefore treat the first randomisation in exactly the same way as a standard parallel‐group RCT.

Studies with multiple treatment groups 

Where a single study reports multiple study arms, we will include only the arms that meet our eligibility criteria. If two comparisons are combined in the same meta‐analysis, we will halve the comparison group to avoid double‐counting (49). We will describe studies with multiple treatment groups in the 'Characteristics of included studies' table. 

Multiple time points 

We will extract data at baseline and the closest time points that match our predefined early and late time points.

Dealing with missing data

We will examine a number of potential sources for missing data, which we will take into account at the study level, outcome level, and summary data level, as outlined in Chapter 6 of the Cochrane Handbook for Systematic Reviews of Interventions (49). In the first instance, it is important to differentiate between data 'missing at random' and 'not missing at random'. 

If data are missing for which the study protocol states would be measured, or the data exist in figures but are not documented, we will request these data from the principal investigators. If, after this, data are still missing, we will have to make explicit assumptions of any methods the included studies used. For example, we will assume that the data were missing at random, or that missing values had a particular value, such as a poor outcome. 

If any outcomes (primary and secondary) are not reported in the identified studies, we will contact the study authors to obtain these data. If the requested data are not available, we will acknowledge the missing data as a limitation of the study and identify this information as an area for future research. 

Reporting bias assessment

Two of three review authors (AM, MP, CK) will independently assess risk of bias. In order to identify selective outcome reporting, we will compare outcomes described in the study protocol or trial registration (if available) with those reported in the publication. We will also compare outcomes listed in the methods section of the final paper with those presented in the results section.

We will investigate the presence of small‐study effects in the pair‐wise meta‐analyses. If there are at least 10 studies, we will make a formal assessment of publication bias using funnel plots (50) and linear regression. We will assess for funnel plot asymmetry caused by publication bias using contour‐enhanced funnel plots (51). The contour lines in the plot determine levels of statistical significance. We will assume that a lack of studies in areas of non‐significance will show signs of publication bias.

Synthesis methods

If the individual studies are sufficiently homogeneous (52), we will pool the data in meta‐analysis. We will perform analyses according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (52). We will conduct separate meta‐analyses for each comparison. Where data are insufficient to conduct a meta‐analysis, we will report the results narratively (53).

We will analyse the data using RevMan software (54). One review author will enter the data, and a second review author will check the data for accuracy.

When meta‐analysis is feasible, we will pool data using the random‐effects model, as we anticipate that true effects may be affected by variability between the included studies. For binary outcomes, we will calculate RR and estimate the between‐study variance using the restricted maximum likelihood (REML) method. We will use the inverse variance method for continuous outcomes, outcomes that include data from cRCTs (55), or outcomes for which HRs are available. We will use subgroup analyses to explore heterogeneity above 80%. If we are unable to explain the heterogeneity, or results are not reported homogeneously, then we will not perform a meta‐analysis and instead report data narratively (53).

We will include data from all eligible studies in the meta‐analysis, but will explore the effect of risk of bias through sensitivity analysis (52).

Results will be presented on forest plots where possible to allow meta‐analysis. For HRQoL outcomes where data cannot be extracted or combined quantitively, we will synthesise the data without meta‐analysis by summarising the relationship between erythroid response after treatment and HRQoL/physical functioning using narrative synthesis techniques. We will present this information in tables. The type of disease, risk score, age range of the population, and intervention received will be tabulated alongside HRQoL findings, as these factors will influence the reported HRQoL. 

Investigation of heterogeneity and subgroup analysis

Measures and tests for heterogeneity

We will assess heterogeneity following the recommendations in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (52). We will consider clinical and methodological sources of heterogeneity, using subgroup and sensitivity analyses, and whether this may be a source of statistical heterogeneity. We will quantify heterogeneity using Tau², Cochran’s Q, and the I² statistic. We will consider an I² above 50% as indicative of a moderate level of heterogeneity, and an I² above 80% as a high level of heterogeneity.

We will pool results by treatment comparison, and will subcategorise the data by type of MDS using International Prognostic Scoring System (IPSS) risk scores (9). Many treatments are used only in certain risk categories of MDS, but where trials include participants with different levels of risk receiving the same intervention, we will subgroup the data by risk score where possible.

We will conduct a subgroup analysis if most of the data cluster around time points that differ from the prespecified time points of our review.

One additional subcategorisation will be MDS with del(5q). The reason for having this separate subcategory is that this type of MDS has a specific treatment (lenolidomide) which treats anaemia effectively, with other treatments being used less frequently in this specific subtype of MDS, and hence would need to be in its own subgroup.

We will use the formal test for subgroup differences in ReMan (54). If we identify substantial statistical heterogeneity, we will investigate the possible causes of this heterogeneity, considering in particular whether transfusion policies and thresholds may have contributed to the heterogeneity, as there is no current fixed recommended haemoglobin transfusion threshold for people with MDS.

Equity‐related assessment

We will not undertake a formal investigation into health inequity in this review, as it is outside the scope of the project. However, we plan to extract data on age, sex, and study location, and will use these data to describe and comment on representation across study populations.

Sensitivity analysis

We will conduct sensitivity analyses for our critical outcomes, to explore the influence of the following characteristics.

• Studies assessed as being at overall high risk of bias using the RoB 2 tool.

• Trials that have stopped early. We will explore the reasons for early stoppage, and consider if this may have affected the outcome of the trial.

• If we identify cRCTs, we will undertake sensitivity analysis based on ICCs to examine the effect of combining results from cRCTs. 

Certainty of the evidence assessment

We will use the GRADE approach to assess the certainty of the evidence for the critical outcomes (27). The GRADE approach uses five considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of a body of evidence for each critical outcome, and to draw conclusions about the certainty of evidence within the text of the review. Two of three review authors (MP, AM, CK) will independently carry out GRADE assessments, following the current GRADE guidance for these assessments in its entirety as recommended in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (56). If we are unable to reach a consensus on GRADE assessments, we will consult an additional review author (SS).

We will use the overall risk of bias judgement, derived from the RoB 2 Excel tool, to inform our decision on downgrading for risk of bias. For time‐to‐event outcomes, we will calculate absolute effects at specific time points as recommended in the GRADE guidance (57). We will phrase the findings and certainty of the evidence as suggested in the informative statement guidance (58). 

Consumer involvement

We have consulted a small panel of consumers for their views on this protocol. The panel commented on the objectives, prioritisation of comparisons, and selection of outcomes. We will be using core outcome sets and standardised assessment scales that have been developed with consumer involvement, and this review has been identified by participants as one of the top five priority research questions, as part of the Cochrane Haematology priority‐setting exercise. Consumers have commented on this protocol as part of the peer‐review process.

We will continue to involve a consumer panel during the conduct of the review, updating them throughout the review process, seeking their advice when necessary, presenting our findings to them, and incorporating their comments into the write‐up.

Supporting Information

Supplementary materials are available with the online version of this article: 10.1002/14651858.CD016362.

Supplementary materials are published alongside the article and contain additional data and information that support or enhance the article. Supplementary materials may not be subject to the same editorial scrutiny as the content of the article and Cochrane has not copyedited, typeset or proofread these materials. The material in these sections has been supplied by the author(s) for publication under a Licence for Publication and the author(s) are solely responsible for the material. Cochrane accordingly gives no representations or warranties of any kind in relation to, and accepts no liability for any reliance on or use of, such material.

Supplementary material 1 Search strategies

a.

These authors should be considered joint first author

b.

These authors should be considered joint last author

New

Additional information

Contributions of authors

AM: was involved in the conception of the review, co‐developed the outcomes, assisted with methods development, and wrote sections of the manuscript.

MP: was involved in the conception of the review, co‐developed the outcomes, assisted with methods development, and wrote sections of the manuscript.

CK: chaired review meetings, assisted with methods development, and wrote sections of the manuscript.

MC: provided clinical and methods support and assisted with the development of the outcomes, read and commented on the manuscript.

CD: developed and wrote the search strategy, read and commented on the manuscript.

CP: provided clinical and methods support, read and commented on the manuscript.

SS: was involved in the conception of the review, co‐developed the outcomes and shaped the objectives, read and commented on the manuscript.

ZM: was involved in the conception of the review, co‐developed the outcomes and shaped the objectives, read and commented on the manuscript.

Declarations of interest

AM: works as a haematologist in the public hospital system in Victoria, Australia. Was a sub‐investigator on a study that may potentially be included in this review, but is not listed as an author on the paper.

MP: is a clinical research fellow for NHS Blood and Transplant and holds an honorary contract as a haematology registrar in Oxford University Hospitals NHS Trust.

CK: no commercial or non‐commercial conflicts of interest relevant to this review.

MC: no commercial or non‐commercial conflicts of interest relevant to this review.

CD: no commercial or non‐commercial conflicts of interest relevant to this review.

CP: is a haematology registrar within the NHS, UK.

SS: is a Consultant Haematologist for NHS Blood and Transplant and was the lead investigator on a study that may potentially be included in this review.

ZM: works as a haematologist in the public hospital system in Victoria, Australia.

Sources of support

Internal sources

• UK Blood Services, UK

  The four UK blood services fund the work of the Systematic Review Initiative (SRI). These bodies had no involvement in the development of the protocol or in conducting the review. The views and opinions expressed therein are those of the review authors and do not necessarily reflect those of the four UK blood services.

• NHS Blood & Transplant, UK

  MP and CP are supported by NHS Blood and Transplant (NHSBT), UK, who fund their Clinical Research Fellowships. NHSBT had no involvement in the development of the protocol or in conducting the review. The views and opinions expressed therein are those of the review authors and do not necessarily reflect those of the NHSBT.

External sources

• No sources of support provided

Registration and protocol

Cochrane approved the proposal for this review in June 2025.

Data, code and other materials

We do not have any data or code to share in this article, as it is a protocol and as such no data have been generated or analysed.

Notes

Published notes in RevMan are for editor use only. Authors should leave this section blank.