Disease‐modifying treatments for primary autoimmune haemolytic anaemia

Abstract

Background

Primary autoimmune haemolytic anaemia (AIHA) is an autoantibody mediated condition characterised by a variable disease course. A myriad of immunomodulatory agents have been employed but there is a paucity of evidence to support their use or compare their effectiveness.

Objectives

To determine the effects of various disease‐modifying treatment modalities in people with AHIHA.

Search methods

We searched MEDLINE (Ovid) (1946 to 2021), Embase (Ovid) (1974 to 2021), Latin American and Caribbean Health Sciences Literature (LILACS) (1982 to 2021), and the Cochrane Library (CENTRAL). Clinical trial registries and relevant conference proceedings were also reviewed. Records were included as of 7 March 2021. We did not impose any language restrictions.

Selection criteria

Randomised controlled trials (RCTs) comparing immunosuppressive or immunomodulatory treatments against no treatment, placebo, or another immunosuppressive or immunomodulatory treatment, for people of all age with idiopathic AIHA.

Data collection and analysis

We used standard methodological procedures expected by Cochrane. The prioritised pre‐defined outcomes included complete haematological response at 12 months, frequency of adverse events at two, six and 12 months, partial haematological response at 12 months, overall survival at six and 12 months, relapse‐free survival (RFS) at six and 12 months, red blood cel (RBC) transfusion requirement after treatment at 12 months, and quality of life (QOL) as measured by validated instruments at 12 months. Based on data availability, we were only able to perform meta‐analysis on frequency of complete haematological response.

Main results

Two trials were included, enrolling a total of 104 adult participants (96 randomised) with warm AIHA in the setting of tertiary referral centres, both comparing the effectiveness between rituximab (375 mg/m² weekly for four weeks, or 1000 mg for two doses two weeks apart) plus glucocorticoid (prednisolone 1.5 or 1mg/kg/day with taper) and glucocorticoid monotherapy. The average age of participants in the two trials were 67 and 71, respectively. One of the included studies had good methodological quality with low risk of bias, whereas the other study had high risk of performance and detection bias due to lack of blinding. Compared with glucocorticoid alone, adding rituximab may result in a large increase of complete response at 12 months (n = 96, risk ratio (RR) 2.13, 95% confidence interval (CI) 1.34 to 3.40, GRADE: low‐certainty evidence).

Rates of adverse effects at prespecified time‐points were not reported.

Limited data on partial haematological response were reported. The evidence is very uncertain about the effect of adding rituximab to glucocorticoids on partial haematological response at 12 months (n = 32; study = 1; RR 3.00, 95% CI 0.13 to 68.57; GRADE very low‐certainty evidence).

RBC transfusion need at 12 months was reported in one study, with four participants (mean number of packed red cell units 4.0 ± 2.82) from the rituximab group and five participants from the placebo (corticosteroid only) (mean number of packed red cell units 5.6 ± 4.15) group requiring transfusion, indicating very uncertain evidence about the effect of adding rituximab to glucocorticoids (n = 32, RR 0.80, 95% CI 0.26 to 2.45, GRADE very low‐certainty evidence). The other study did not report transfusion requirement at prespecified time points but reported no difference in transfusion requirement between the two groups when comparing responders from enrolment to end of response or to the end of study follow‐up (34 units versus 30 units, median [range]: 0 [1 to 6] versus 0 [1 to 5], P = 0·81).

Overall survival and RFS rates at prespecified time‐points were not explicitly reported in either study. Data on QOL were not available.

Authors' conclusions

Available literature on the effectiveness of immunomodulatory therapy for primary AIHA is restricted to comparison between rituximab plus glucocorticoid and glucocorticoid alone, in patients with newly diagnosed warm AIHA, calling for need for additional studies. The current result suggests that combinatory therapy with rituximab and glucocorticoid may increase the rate of complete haematological response over glucocorticoid monotherapy.

Plain language summary

Treatment for primary autoimmune haemolytic anaemia

What is primary autoimmune haemolytic anaemia (AIHA)?

Red blood cells carry oxygen to all parts of the body. In healthy people, red blood cells live about 120 days. In AIHA the immune system does not work properly and destroys red blood cells faster than rhe body can make them. Primary AIHA is when a person does not have another condition that caused their AIHA.

Primary AIHA develops differently in different people. People often need to have repeated blood transfusions to keep their red blood cell levels normal. The disease may be life‐threatening for some people.

What is the aim of this review and why is it important?

The aim of this review is to find out what research studies tell us about the safety and effectiveness of treating primary AIHA with immune modulating agents. Immune modulating agents change how your immune system works. They are not a cure, but they can slow AIHA.

But how well do these agents work? Are some safer than others? The answers are important to help doctors and patients make treatment decisions based on evidence.

What studies did the review find?

We looked at studies that were finished by 7 March 2021. We found two studies that looked at an immune modulating agent called rituximab, which targets immune cells, called B‐cells.

The studies compared two groups of adults with primary AIHA who were treated:

· with rituximab and steroid

· with steroid alone

It’s important to know that these studies were very small, including a total of 96 people. And the design of one study had a problem that may have affected the results.

Recovery after treatment

· At 12 months after beginning treatment, rituximab and steroid appear to help more people recover from primary AIHA than steroid alone. Researchers think the quality of this research finding is low. They think future research is likely to change these results.

· At six months after beginning treatment, there was no difference in recovery between the adults treated with rituximab and a steroid and those treated with a steroid alone.

Side effects from treatment

· Side effects between the two groups were the same.

What’s the bottom line of this review?

Rituximab and steroid may help more people have a complete recovery from newly diagnosed, primary warm AIHA than steroid alone. Side effects appear to be about the same for both treatments. But more research is needed to know for sure.

Doctors offer other treatments for primary AIHA that may be helpful, but there is a lack of research about how well they work. For now, based on the little information we have, the patient and doctor will have to carefully make decisions together about treating primary AIHA.

Summary of findings

Summary of findings 1. Summary of findings.

Rituximab plus corticosteroid compared with corticosteroid (+/‐ placebo) for warm AIHA
Patient or population: adults with newly diagnosed warm AIHA Settings: referral centres Intervention: rituximab plus corticosteroid Comparison: corticosteroid only
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Corticosteroid only	Rituximab and corticosteroid only
Complete haematological response at 12 months	313 per 1000	667 per 1000	RR 2.13 (1.34 to 3.40)	96 (2)	⊕⊕⊝⊝ low#
Frequency of adverse events at two, six and 12 months	See comment	See comment	Not estimable	See comment	See comment	The included studies did not report frequency of adverse events at two, six an 12 months
Partial haematological response at 12 months	0 per 1000	63 per 1000	RR 3.00 (0.13 to 68.57)	32 (1)	⊕⊝⊝⊝ very low$	Only reported by Michel 2017
Overall survival at six and 12 months	See comment	See comment	Not estimable	See comment	See comment	The included studies did not report OS
Relapse‐free survival at six and 12 months	See comment	See comment	Not estimable	See comment	See comment	The included studies did not report RFS at specific time points
RBC transfusion requirement after treatment at 12 months	313 per 1000 (mean number of packed red cell units 5.6 ± 4.15)	250 per 1000 (mean number of packed red cell units 4.0 ± 2.82)	RR 0.80 (0.26 to 2.45)	32 (1)	⊕⊝⊝⊝ very low	Only reported by Michel 2017. The included studies did not report transfusion requirement at two and six months.
QOL as measured by validated instruments at 12 months	See comment	See comment	Not estimable	See comment	See comment	The included studies did not report QOL measures
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; OS: overall survival; QoL: quality of life; RBC: red blood cell; RCT: randomised controlled trial; ; RFS: relapse‐free survivalRR: Risk Ratio
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

#Evidence from RCT downgraded because of high risk of bias in study design and small sample size.

$Evidence from RCT downgraded because of imprecision in results and small sample size.

Background

Description of the condition

Autoimmune haemolytic anaemia (AIHA) is a condition characterised by the presence of autoantibodies which bind to surface of the patient's own red blood cells (RBCs) leading to premature destruction (Gehrs 2002). Such destruction of RBCs (haemolysis) can occur within the circulation (intravascular haemolysis) or within the spleen (extravascular haemolysis). Incidence of AIHA ranges from 0.8 to 3 new cases per 100,000 individuals per year, and the estimated prevalence is 17 per 100,000 individuals, more commonly occurring in individuals after 50 years of age (Aladjidi 2011; Eaton 2007; Gehrs 2002; Klein 2010; Sokol 1992). AIHA may run an acute or chronic course; severity of the condition can range from fully compensated disease to life‐threatening anaemia, and can be classified by the temperature reactivity of RBC autoantibodies, as well as by the underlying aetiology. Warm AIHA is the most common type; it is associated with autoantibodies, often immunoglobulin G (IgG), being most reactive at 37^oC and can be secondary to rheumatological disorders and lymphoproliferative conditions. Haemolysis is usually extravascular and takes place in the spleen, with a disease course that is typically chronic and relapsing. Cold AIHA, on the other hand, is a result of autoantibodies, usually immunoglobulin M (IgM), with the highest affinity at 0^oC to 4^oC and can be associated with underlying lymphoproliferative conditions and infections (e.g. mycoplasma, Ebstein‐Barr virus). IgM causes complement fixation and results in intravascular haemolysis, which tends to be abrupt but self‐limiting. Paroxysmal cold haemoglobinuria is a rare subtype of cold AIHA caused by IgG that preferentially binds at a lower temperature, mostly in children following infections. Infrequently, mixed‐type AIHA may occur in an individual where a combination of warm and cold autoantibodies exist. Drug‐induced immune haemolytic anaemia is a distinct entity that may be associated with both warm and cold AIHA (Johnson 2007). This review focuses on the treatment for primary, or idiopathic AIHA, where no underlying systemic disease can be identified.

Diagnosis of primary AIHA depends on the demonstration of haemolysis, serologic evidence of autoantibody against the individual's own RBCs and exclusion of secondary causes (identifiable in 20% to 80% of cases) (Gehrs 2002). The direct antiglobulin test (DAT) is commonly used to demonstrate the presence of autoantibody‐coated patient RBCs, although it is important to bear in mind alternative causes for a positive DAT including the use of intravenous immunoglobulin, drug‐induced autoantibodies, haemolytic transfusion reaction, thalassaemia, sickle cell disease, and multiple myeloma (Clark 1992). Haemolysis on the other hand, is suggested clinically when yellowish discolouration of the skin (jaundice), together with pallor is detected, with or without the presence of enlargement of the spleen (splenomegaly). In terms of laboratory evaluation, complete blood count with peripheral smear, serum bilirubin, lactate dehydrogenase (LDH), haptoglobin, methaemalbumin and urine haemoglobin are useful in determining the presence and type of haemolysis (presence of schistocytes, low haptoglobin, raised methaemalbumin and urine haemoglobin in intravascular haemolysis; presence of spherocytes, raised unconjugated bilirubin in extravascular haemolysis). In warm AIHA, anti‐IgG anti‐sera is typically detected in DAT while in cold AIHA, anti‐C3d anti‐sera is usually present due to IgM‐mediated haemolysis.

Individuals with AIHA may required immunomodulatory therapies and more than half of the people, especially those with severe anaemia and mixed serological type, may require multiple agents to achieve treatment response (Barcellini 2014). They also experience morbidities including thrombosis, acute renal failure, infection and acute renal failure due to the disease itself and complications from immunomodulatory therapies including splenectomy. Between a quarter and half of all individuals who achieve remission might develop disease relapse (Barcellini 2014). Non‐responders or those who relapse may be dependent on regular transfusions to alleviate the symptoms of anaemia. Estimated mortality due to AIHA is 3% to 4% (Barcellini 2018).

Description of the intervention

Treatment depends on the severity of haemolysis. Supportive treatment including folic acid supplementation, red cell transfusion and avoidance of cold exposure in individuals with cold AIHA are not within the scope of the current review (Gehrs 2002). In AIHA with identifiable secondary causes, treatment should be targeted towards the underlying condition.

Specific treatment for idiopathic AIHA includes immunosuppressive or immunomodulatory therapy. Traditionally, glucocorticoids (prednisolone 1 mg to 1.5 mg/kg/day for one to three weeks then tapered) are the first‐line therapy for people with AIHA with up to three quarters of people demonstrating improvement within three weeks. Relapse is nevertheless common (15% to 40%) after tapering of glucocorticoids in the first six months to one year, and thus the majority of people responding to glucocorticoids would have to be continued with a lower maintenance dose (Aladjidi 2011; Sankaran 2016; Zanella 2014). Adverse effects from prolonged glucocorticoid use include Cushingoid changes, hypertension, hyperglycaemia, peptic ulcers, and reduced bone mineral density. Splenectomy has been considered the second‐line therapy for people failing glucocorticoid therapy. Response rates after splenectomy ranged from 60% to 75%, but carry the risk of thrombosis as well as infection due to encapsulated bacteria and parasites including malaria (Davidson 2001; Katkhouda 1998).

In people with refractory disease, a number of interventions have been tested with various degree of success. These include immunosuppressive agents including cyclophosphamide, danazol, cyclosporine, and mycophenolate mofetil (Hershko 1990; Howard 2002; Moyo 2002; Pignon 1993), as well as monoclonal antibodies including anti‐CD20 (rituximab) and anti‐CD52 (alemtuzumab) (Zecca 2003; Cheung 2006). With a response rate of 80% to 90%, rituximab is increasingly being used in people with AIHA failing glucocorticoids. Each medication possesses its own side effects which are detailed in the next section (Cheung 2006; Hershko 1990; Howard 2002; Moyo 2002; Pignon 1993, Zecca 2003). Interventions targeted at removing circulating autoantibodies include intravenous immunoglobulin and plasma exchange (Flores 1993; Smith 2003). Accessibility to the above measures may be limited in resource‐poor countries.

How the intervention might work

Immunosuppressive or immunomodulatory therapy interfere with the immune destruction of RBCs.

1. Glucocorticoids inhibit Fc receptor‐mediated removal of sensitised RBCs, and in the longer term reduce production of autoantibodies (Zanella 2014).

2. Splenectomy removes a major site of RBC destruction and autoantibody production (Zanella 2014).

3. Rituximab is an anti‐CD20 monoclonal antibody given as four‐weekly intravenous infusions. The antibody targets and depletes host B cells, which are responsible for the generation of autoantibodies (Zecca 2003). Side effects include infusion reactions, immunosuppression and hepatitis B reactivation.

4. Alemtuzumab is an anti‐CD52 monoclonal antibody targeting CD52, which is an antigen expressed by mature lymphocytes. Lymphodepletion with alemtuzumab would effectively devoid the host's capability of autoantibody generation (Cheung 2006). Side effects include infusion reactions and immunosuppression.

5. Cyclophosphamide is an alkylating agent with potent myelosuppressive effect. Lymphocytes are highly sensitive to the drug and depletion would disrupt the autoimmunity involved (Moyo 2002). Nausea, vomiting, alopecia, myelosuppression, haemorrhagic cystitis, and gonadal toxicity are potential adverse effects with the drug.

6. Danazol is a semi‐synthetic androgen with immunomodulatory effect via decreasing IgG production and cell‐bound IgG and complement (Pignon 1993). Side effects include its androgenic effects and derangement in liver function.

7. Cyclosporine is a calcineurin inhibitor that suppresses T helper cells activities, decreasing autoantibodies synthesis (Hershko 1990). Individuals should be monitored for hair growth, gum hypertrophy, renal impairment, hypertension and rarely, neurological complications.

8. Mycophenolate mofetil (MMF) reversibly inhibits inosine monophosphate dehydrogenase and in turn inhibits purine synthesis required in the proliferation of B and T cells (Howard 2002). Adverse effects are gastrointestinal disturbances and myelosuppression.

9. Use of intravenous immunoglobulin (IVIG) in AIHA is based on its effectiveness in immune thrombocytopenia, but its efficacy with AIHA is more controversial. The mechanism of action is hypothesised to be through competitive inhibition of autoantibody adsorption to patient blood cells, as well as by decreasing uptake of autoantibody‐coated blood cells by the reticuloendothelial system through blockage of macrophage Fc receptors (Flores 1993). Infusion reactions, fever, arthralgia, haemolysis, and aseptic meningitis are possible complications.

10. Plasma exchange is usually reserved for people with fulminant AIHA. The process effectively removes host circulating immunoglobulins and complements that are responsible for the immune destruction of RBCs (Smith 2003). Side effects are attributed to the insertion of the apheresis catheter, and the process of apheresis which may cause haemodynamic disturbances, hypocalcaemia and coagulopathy.

Why it is important to do this review

Various immunosuppressive medications and other specific treatment modalities are available for the management of idiopathic AIHA. As an uncommon condition, it remains uncertain whether high‐quality evidence exists to support any treatment regimen and what the more effective regimens are. This systematic review will provide an evidence base for the selection of immunomodulatory treatment that most likely benefits people with idiopathic AIHA.

Objectives

To determine the effectiveness and safety of various disease‐modifying treatment modalities in people with autoimmune haemolytic anaemia (AIHA).

Methods

Criteria for considering studies for this review

Types of studies

We included only randomised controlled trials (RCTs).

Types of participants

We included male and female participants of all ages suffering from idiopathic AIHA. Individuals with newly diagnosed, relapsing or refractory disease; warm or cold AIHA were included. Studies on people with secondary AIHA were excluded. There was no limitation on study settings.

Types of interventions

We included trials evaluating specific, immunosuppressive or immunomodulatory treatments for AIHA, encompassing but not limited to: corticosteroids, intravenous immunoglobulin (IVIG), rituximab, alemtuzumab, azathioprine, cyclosporine, mycophenolate mofetil (MMF), plasma exchange and splenectomy. We excluded trials of supportive treatment such as folic acid supplement and transfusion alone. The control interventions could be another specific treatment,supportive treatment alone, placebo treatment, or no treatment. We also included trials comparing different dosing regimens of the same treatment.

Types of outcome measures

We did not use the outcomes listed below as criteria for study inclusion.

Primary outcomes

1. Frequency of complete haematological response (complete response (CR), defined as normalisation of haemoglobin concentration, reticulocyte count, and indirect (or unconjugated] bilirubin level) at months two, six and 12.

2. Frequency of adverse events at two, six and 12 months.

Secondary outcomes

1. Frequency of partial haematological response (partial response (PR), defined as improvement in haemoglobin concentration) at two, six and 12 months

2. RBC transfusion requirement after treatment (measured as units of RBCs transfused per month or millilitres (mL) per kg body weight) at two, six and 12 months

3. Frequency of direct antiglobulin test (DAT) positivity after treatment at two, six and 12 months

4. Overall survival at six and 12 months

5. Relapse‐free survival (RFS) at six and 12 months

6. Frequency of relapse at six and 12 months

7. Quality of life (QOL) as measured by validated instruments at 12 months

Search methods for identification of studies

Electronic searches

We searched MEDLINE (Ovid) (1946 to 2021) (Appendix 1), EMBASE (Ovid) (1974 to 2021) (Appendix 2), LILACS (Latin American and Caribbean Health Sciences Literature) (1982 to 2021) (Appendix 3), and Cochrane Library (CENTRAL) (latest issue) (Appendix 4). The search strategies for the different electronic databases using a combination of controlled vocabulary and text word terms are shown in the appendices (Appendix 1; Appendix 2; Appendix 3; Appendix 4).

We did not incorporate search in the China Journal Net database (search date 7 March 2021) after submission of the protocol (Liu 2017) due to the extreme brevity of full‐text articles published which made interpretation of both quality and results impractical.

Searching other resources

We searched the following clinical trial registries on the Internet using the search terms " haemolytic anaemia" or "hemolytic anemia":

1. WHO International Clinical Trials Registry Platform (ICTRP) (http://apps.who.int/trialsearch/)

2. United States Clinical Trials Registry (https://clinicaltrials.gov/)

3. European Union Clinical Trial Register (http://www.clinicaltrialsregister.eu/)

We also searched conference proceedings of the following scientific meetings using the search terms " haemolytic anaemia" or "hemolytic anemia" and "trials":

1. Annual Meeting of American Society of Haematology (ASH) (2004 to 2021)

2. Annual Congress of the European Haematology Association (EHA) (2006 to 2021).

We searched reference lists of relevant articles identified in all the searches. We contacted authors of included studies to enquire about additional studies. We did not apply any language restriction in our searches.

Data collection and analysis

Selection of studies

Both review authors independently examined identified studies to select studies meeting the inclusion criteria. Discrepancies were resolved by discussion. We reported the flow of studies as per the PRISMA statement in a flow chart (Moher 2009), which included data on the number of records identified through database searching, number of additional records identified through other sources, number of records after duplicates removed, number of records screened, number of records excluded, number of full‐text articles examined for eligibility, number of full‐text articles excluded with reasons, and numbers of studies included in qualitative and quantitative syntheses (Higgins 2011a); see Figure 1.

1.

Study flow diagram.

Data extraction and management

Both review authors independently extracted data using a standardised data collection form (Higgins 2011a). We extracted data on study design and methods including study dates, location, setting, type of study (parallel group or cross‐over or other designs), stratified randomisation variables, random sequence generation, allocation concealment, and blinding methods. We extracted data on participant inclusion and exclusion criteria, number of participants in each intervention and control group, and for each group, their demographics (age, sex, ethnicity), baseline laboratory parameters (haemoglobin level, reticulocyte count, bilirubin level, direct anti‐globulin test (DAT) positivity), baseline transfusion requirements, associated conditions and comorbidities, and previous intervention received (drug and regimen). We extracted data on details of intervention for each group, and dropouts of each group and follow‐up duration. We extracted outcomes measured in each study that matched with our pre‐specified outcomes. No transformation of reported data was required. We also extracted data on funding source and declaration of interests. Data were entered into Review Manager software (RevMan 5.3) (RevMan 2014) by one review author (APYL) and checked by the other review author (DKLC).

Assessment of risk of bias in included studies

Both review authors independently assessed the risk of bias of included studies with the Cochrane 'Risk of bias' tool (Higgins 2011b). The following domains were assessed: random sequence generation, allocation concealment, blinding of participants, study personnel and outcome assessors, incomplete outcome data, selective reporting, and other sources of bias. For each domain for each study, the risk of bias was graded as "high", "unclear" or "low" according to the criteria stated in the CochraneHandbook for Systematic Reviews of Interventions. Discrepancies between the review authors were resolved by discussion.

Measures of treatment effect

We estimated risk ratio (RR) with 95% confidence intervals (CI) for dichotomous outcomes. We used hazard ratio (HR) with 95% CI for time‐to‐event outcomes and planned to use mean difference (MD) with 95% CIs for continuous outcomes. For continuous outcomes using different scales (e.g. quality of life (QoL) data), we planned to use standardised mean difference (SMD) with 95% CI (Deeks 2011).

Unit of analysis issues

We planned to use appropriate unit of analysis for cluster‐randomised trials and cross‐over trials according to the CochraneHandbook for Systematic Reviews of Interventions (Deeks 2011), if these types of trials had been found and included.

Dealing with missing data

We requested missing data from the corresponding authors of included studies (Birgens 2013; Michel 2017) and nonetheless failed to acquire additional information with no response received. We included all participants with intention‐to‐treat (ITT) analyses.

Assessment of heterogeneity

We used the I² statistic to evaluate the degree of heterogeneity of treatment effects. We followed the guide on interpretation of the I² statistic suggested by the CochraneHandbook for Systematic Reviews of Interventions (Deeks 2011) as follows:

1. 0% to 40%: may not be important;

2. 30% to 60%: may represent moderate heterogeneity;

3. 50% to 90%: may represent substantial heterogeneity;

4. 75% to 100%: considerable heterogeneity.

We also used the Chi² test of homogeneity to assess the strength of evidence regarding heterogeneity.

Assessment of reporting biases

We planned to assess publication bias using a funnel plot (estimated differences in treatment effects against their standard errors) in case 10 or more studies are identified for a given outcome, although this was not probable as only two studies were included (Sterne 2011).

Data synthesis

With the assumption that the different studies would likely be estimating different, yet related, intervention effects, we used the random‐effects model for meta‐analysis to obtain the mean effect across studies(Deeks 2011). In case different interventions are identified in future updates, they will be summarised individually. Analysis was performed using Cochrane's statistical package Review Manager (RevMan 5.3) (RevMan 2014).

Subgroup analysis and investigation of heterogeneity

We planned to perform subgroup analyses (Deeks 2011) for the following subgroups as they may have different prognosis and response to treatment (different size of treatment effect).

1. Different age groups (children < 18 years or adults >= 18 years).

2. With or without associated conditions (such as thrombocytopenia or neutropenia, or other autoimmune phenomena).

3. Different degrees of anaemia (transfusion dependent or transfusion independent).

4. Newly diagnosed versus refractory disease.

We also planned to assess subgroup differences by examining the I² statistic and performing a Chi² test for homogeneity across subgroup results. In case heterogeneity was present, we planned to investigate the probable reasons for heterogeneity by examining the distribution of important participant factors between trials (age, associated conditions, degrees of anaemia, and previous interventions) and trial factors (randomisation concealment, blinding, dropouts, intervention regimens).

As the above are currently not applicable based on the studies included, we will perform the above analyses in future updates as necessary.

Sensitivity analysis

We planned to perform sensitivity analysis to assess the impact of excluding studies with high risk of bias (Deeks 2011). We also planned to perform sensitivity analysis to assess the impact of excluding studies with significant amount (> 30%) of missing outcome data.

Summary of findings and assessment of the certainty of the evidence

We produced a 'Summary of findings' table according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). The pre‐defined outcomes included were:

• frequency of complete haematological response (defined as normalisation of haemoglobin concentration, reticulocyte count, and indirect bilirubin level) at months two, six and 12 months;

• frequency of adverse events at two, six and 12 months;

• overall survival at six and 12 months;

• relapse‐free survival at six and 12 months;

• RBC transfusion requirement after treatment (measured as units of red cells transfused per month or mL per kg body weight) at two, six and 12 months; and

• QOL as measured by validated instruments at 12 months.

These tables summarised the results for the prioritised pre‐defined outcomes, namely:

• complete haematological response at 12 months;

• frequency of adverse events at two, six and 12 months;

• partial haematological response at 12 months;

• overall survival at six and 12 months;

• relapse‐free survival at six and 12 months,

• RBC transfusion requirement after treatment at 12 months; and

• QOL as measured by validated instruments at 12 months;

and provide grading of the certainty of evidence for each outcome according to the GRADE system as recommended by the CochraneHandbook for Systematic Reviews of Interventions (GRADEpro 2008; Lefebvre 2011; Schünemann 2011).

Results

Description of studies

Results of the search

We obtained a total of 11980 results from the electronic search of the databases, after de‐duplication, 10610 articles remained. We screened the titles and abstracts of these against the inclusion and exclusion criteria for study selection and excluded 10564 references based on titles or abstracts alone. We obtained the full texts of the remaining 46 articles and assessed for eligibility. Two studies fulfilled the inclusion criteria and were included in the review (Birgens 2013, Michel 2017), with one secondary citation to an included study also identified (Michel 2015). The flow of records is summarised in Figure 1.

Included studies

Two studies were included (Birgens 2013, Michel 2017) with details summarised in the ‘Characteristics of included studies’ table. Both trials were multi‐centre, Phase III randomised controlled trials (RCTs) comparing the efficacy of rituximab plus glucocorticoid versus glucocorticoid monotherapy in adults with warm, primary, autoimmune haemolytic anaemia (AIHA).

In the study by Birgens and colleagues (Birgens 2013), 65 participants with newly diagnosed warm AIHA (steroid use for <1 week was allowed) were recruited in eight haematological centres in Denmark between March 2005 and June 2012. With one patient withdrawing before randomisation (1:1 using pre‐coded envelopes), 32 were randomised to receive rituximab plus glucocorticoid (12 females [37.5%], median age [range] = 65 [41‐89]) and 32 to receive glucocorticoid monotherapy (15 females [46.9%], median age [range] = 67 [35‐90]. All participants received oral prednisolone 1·5 mg/kg/day for two weeks followed by tapering according to this schedule: 0·75 mg/kg/day for one week (week three), thereafter 0·5 mg/kg/day for one week (week four), followed by a gradual reduction over the next four to eight weeks to the lowest dose that was effective in maintaining a normal haemoglobin level. In the group allocated to a combination of prednisolone and rituximab, rituximab was given at a dosage of 375 mg/m² as an intravenous infusion once a week for four weeks. The primary objective of the study was to analyse differences in treatment responses between the two groups. Responses were evaluated at three, six and 12 months after treatment was initiated. Complete response (CR) was defined as normalisation in haemoglobin concentration without any ongoing immunosuppressive treatment and without any biochemical signs of haemolytic activity. Partial response (PR) was defined as being similar to CR but requiring continued low‐dose prednisolone (< 10 mg/day), or appearing as compensated haemolytic anaemia entailing a stable, acceptable haemoglobin level without any need of treatment except < 10 mg/day prednisone. Secondary objectives were to evaluate differences in relapse‐free survival (RFS), red blood cell transfusion requirement after treatment, the need for splenectomy, and safety profiles associated with the treatments up to 12 months after enrolment. This trial was investigator‐initiated but was supported in part by research funding from Roche A/S, Denmark. The design of the study, protocol writing, interpretation of the data, and preparation of the manuscript were conducted without any interference from Roche.

In the study by Michel and colleagues (Michel 2017), 39 participants with newly diagnosed warm AIHA (steroid use for < 6 weeks was allowed) were screened from 13 centres in France from 2011 to 2015. Thirty‐two participants were randomised to receive rituximab (n = 16) or placebo (n = 16) (17 females [53%], mean age at inclusion 71  ±  16 years) on top of glucocorticoid therapy in a double‐blind manner. At inclusion, all participants received oral prednisone at a daily dose of 1 mg/kg for two weeks, and then tapered according to a pre‐defined recommended reduction scheme. Besides prednisone, randomised participants received two infusions of rituximab or placebo at a fixed dose of 1000 mg two weeks apart. The primary endpoint was overall response rate (CR + PR) in anITT analysis at one year. CR in this study was defined as a haemoglobin (Hb) level ≥ 11 g/dL (women) or ≥ 12 g/dL (men) without features of ongoing haemolysis (including normal haptoglobin level), without any ongoing treatment for warm AIHA on two different occasions four weeks apart in the absence of any recent transfusion. PR was defined as Hb level ≥ 10g/dL with at least a 2 g increase from baseline (i.e. at warm AIHA diagnosis) without any other treatment than prednisone given at a daily dose 10 mg or recent transfusion. Secondary objectives were PR, cumulative dose of steroids, and number of transfusions and hospitalisations at one year and incidence and severity of adverse events in both groups and CR/PR at two years. This study was promoted by the Direction de la Recherche Clinique, Assistance Publique‐Hôpitaux de Paris and supported by the Etablissement Français du Sang and by Roche.

The median observation time was not reported in either study.

Excluded studies

The current review excluded 41 studies as they were not RCTs in design (Abdwani 2009; Almomen 2013; Aqrabawi 2012; Aqrabawi 2013; Barcellini 2012; Barcellini 2013; Bartko 2017; Bartko 2018; Berentsen 2017; Berentsen 2017a; Choudhry 1996; Crickx 2019; Dierickx 2015; Heidel 2007; Hill 2018; Jager 2017; Jilma 2016; Lehrer‐Graiwer 2016; Li 2020; Liu 2001; Mehmood 2016; Miano 2016; Motto 2002; Moyo 2002; Osterborg 2009; Panicker 2016; Panicker 2016a; Rai 2018; Raschetti 2012; Reynaud 2015; Rice 2012; Rodrigo 2015; Rossi 2018; Roth 2016; Shah 2013; Wang 2005; Williams 2020; Yang 2020; Zaja 2002; Zanella 2013; Zecca 2003). Two ongoing RCTs on the use of fostamatinib (oral spleen tyrosine kinase inhibitor) and M281 (human, anti‐neonatal Fc receptor antibody) respectively for warm AIHA are identified but excluded from the current analysis (Cooper 2020; NCT04119050).

Risk of bias in included studies

One of the two included studies (Michel 2017) had good methodological quality with low risk of bias in all areas; the other study (Birgens 2013) had high risk of performance and detection bias due to lack of blinding, but otherwise low risk in other areas. These findings are summarised in the table Characteristics of included studies and Figure 2.

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Both included trials involved random allocation of participants, using pre‐coded envelops (Birgens 2013) and central computer randomisation (Michel 2017), respectively.

Blinding

Blinding was not performed in the study by Birgens and colleagues (Birgens 2013).

Incomplete outcome data

There were no dropouts in both included studies (Birgens 2013, Michel 2017).

Selective reporting

All predefined outcomes were accounted for in both included studies (Birgens 2013, Michel 2017); there was no evidence of selective reporting.

Other potential sources of bias

No other potential sources of bias were identified.

Effects of interventions

See: Table 1

Both included studied compared rituximab plus corticosteroid compared with corticosteroid alone (+/‐ placebo) and were combined in a meta‐analysis. All effect sizes were calculated with the fixed‐effect model unless otherwise specified. See Table 1 for further details.

Primary outcome

Frequency of complete haematological response (defined as normalisation of haemoglobin (Hb) concentration, reticulocyte (immature red blood cells (RBCs)) count, and indirect (or unconjugated] bilirubin level) at months two, six and 12.

Compared with glucocorticoid alone, the evidence is very uncertain about the effect of adding rituximab to glucocorticoids on complete haematological response at six months (n = 64, study = 1; RR 1.73, 95% CI 0.99 to 3.02; participants = 64; RR and CI calculated by review authors; GRADE very low certainty), but it may result in a large increase of complete response at 12 months (n = 96; studies = 2; pooled RR 2.13, 95% CI 1.34 to 3.40; I² = 0%; GRADE low certainty) (Figure 3, Table 2). No heterogeneity was observed. Both trials (Birgens 2013, Michel 2017) did not report complete response rate at two months and one study (Michel 2017) did not report response at six months.

3.

Forest plot of comparison: 1 Rituximab with steroid versus steroid, outcome: 1.1 Complete haematological response at 12 months.

1. Primary outcome ‐ complete haematological response.

	Number of participants with complete response
	Rituximab+corticosteroid			Corticosteroid only
	2 months	6 months	12 months	2 months	6 months	12 months
Birgens 2013 (n = 64 randomised)	n.r.	19	21	n.r.	11	10
Michel 2017 (n = 32 randomised)	n.r.	n.r.	11	n.r.	n.r.	5

Frequency of adverse events at two, six and 12 months.

Both studies (Birgens 2013, Michel 2017) did not report adverse effects at the specified time points.

Cumulative adverse effects and serious adverse events with a non‐fatal or fatal outcome did not differ between the two groups in the study by Birgens and colleagues (Birgens 2013, Table 3). The study by Michel and colleagues (Michel 2017) described no post‐infusion reactions in both groups, but four and 10 episodes of severe adverse events in participants who received rituximab with corticosteroid and corticosteroid alone, respectively (Table 4). Reduced gammaglobulin level was also reported to be associated with rituximab treatment (comparing mean level at inclusion 10.8 ± 4.0 g/L versus that at 12 months 8.1 ± 2.2 g/L for rituximab group), although no difference was observed in evaluable participants from either group at 12 months (P = 0.499).

2. Primary outcome ‐ frequency of adverse events (Birgens 2013).

	Frequency of adverse events (CTCAE grade 1 or above) in study by Birgens 2013
	Rituximab+corticosteroid (%)	Corticosteroid only (%)	P value
Fever	3.3	6.7	0.6
Nausea	0	10	0.1
Dyspnea	13.3	16.7	0.7
Dyspepsia	3.3	13.3	0.2
Restless legs	6.7	0	0.5
Vertigo	0	10	0.1
Increased appetite	0	10	0.1
Insomnia	10	10	1.0
Arthralgia	6.7	3.3	1.0
Headache	6.7	13.3	0.4
Fatigue	13.3	13.3	1.0
Hypotension	6.7	3.3	1.0
Palpitation	0	6.7	0.7
Hyperglycaemia	0	6.7	0.7

3. Primary outcome ‐ frequency of adverse events (Michel 2017).

	Frequency of severe adverse events in study by Michel 2017
	Rituximab+corticosteroid	Corticosteroid only
Severe adverse events	4 (3 patients/participants)	10 (7 patients/participants)
Severe infections (non‐fatal)	2 (pneumonia, including one patient with probable pneumocystis infection)	4 (pneumonia = 1; pulmonary abscess =1; diverticulitis = 1; prostatitis = 1)
Non‐infectious (non‐fatal)	2 (neutropenia)	2 (severe pulmonary embolism = 1; spontaneous vertebral fractures = 1)
Fatal events	0	4 (Intracranial tumour = 1; septic shock post‐colonic perforation after colonoscopy = 1; septic and cardiogenic shock post‐amputation for critical limb ischaemia and osteitis; pneumonia = 1)
Gammaglobulin level at 12 months	8.1  ±  2.2 g	7.7  ±  1.5 g

Secondary outcome

Frequency of partial haematological response (defined as improvement in haemoglobin concentration) at two, six and 12 months

Both trials (Birgens 2013, Michel 2017) did not report complete response rate at two months (Table 5). One study (Birgens 2013) reported six‐month partial response in figure format, but the corresponding data at 12 months were not reported. The other study (Michel 2017) did not report response at six months. The evidence is very uncertain about the effect of adding rituximab to glucocorticoids on partial haematological response reported at six months (n = 64; study = 1; RR 0.80, 95% CI 0.24 to 2.71; RR and CI calculated by review authors; GRADE very low certainty) or 12 months (n = 32; study = 1; RR 3.00, 95% CI 0.13 to 68.57; GRADE very low certainty).

4. Secondary outcome ‐ partial haematological response.

	Number of participants with partial response
	Rituximab+corticosteroid			Corticosteroid only
	2 months	6 months	12 months	2 months	6 months	12 months
Birgens 2013 (n = 64 randomised)	n.r.	4 (based on figure from study)	n.r.	n.r.	5 (based on figure from study)	N/A
Michel 2017 (n = 32 randomised)	n.r.	n.r.	1	n.r.	n.r.	0

RBC transfusion requirement after treatment (measured as units of RBCs transfused per month or millilitres (mL) per kg body weight) at two, six and 12 months

Study by Michel and colleagues (Michel 2017) reported RBC transfusion received by participants up to 12 months, with four participants (mean number of packed red cell units 4.0 ± 2.82) from the rituximab group and five participants from the placebo (corticosteroid only) (mean number of packed red cell units 5.6 ± 4.15) group requiring transfusion, the evidence is very uncertain about the effect of adding rituximab to glucocorticoids (n = 32, RR 0.80, 95% CI 0.26 to 2.45, GRADE very low certainty). Birgens and colleagues (Birgens 2013) did not report transfusion requirement at the predefined time points, but described no difference in transfusion requirement between the rituximab/corticosteroid and the corticosteroid monotherapy group when comparing responders (CR and PR) from enrolment to end of response or to the end of study follow‐up (34 units versus 30 units, median [range]: 0 [1 to 6] versus 0 [1 to 5], P = 0·81).

Frequency of DAT positivity after treatment at two, six and 12 months

Study by Michel and colleagues (Michel 2017) reported positive DAT at 12 months in 5/13 (38%) participants who received rituximab and corticosteroid and compared to 8/10 (80%) participants who received placebo (corticosteroid only). DAT positivity was not reported by study by Birgensand colleagues (Birgens 2013).

Overall survival at six and 12 months

Overall survival at these specific time points was not reported by either study. Michel and colleagues (Michel 2017) reported the death of six participants treated with corticosteroid only versus none with combination of rituximab and corticosteroid at two years (P = 0.017).

Relapse‐free survival (RFS) at six and 12 months

RFS rates at six and 12 months was not explicitly reported by either study. In the study byBirgens and colleagues Birgens 2013), RFS in all responders was higher in the participants receiving rituximab and prednisolone combined (P = 0·02) corresponding to a hazard ratio (HR) of 0·33 (95% Cl 0·12 to 0·88; RR and CI calculated by review authors). In the study by Michel and colleagues (Michel 2017), RFS was better for the rituximab group (P = 0.023).

Frequency of relapse at six and 12 months

Frequency of relapse at six an 12 months was not reported in either study. Birgens 2013 reported that at 36 months, about 70% of the participants who showed either CR or PR were still relapse‐free in the combination treatment group, versus about 45% in the prednisolone monotherapy group (P = 0.02).

Quality of life (QOL) as measured by validated instruments at 12 months

QOL measures were not reported by either included study.

Discussion

Summary of main results

In this review, two randomised controlled trials (RCTs) with 96 participants suffering from primary warm‐antibody autoimmune haemolytic anaemia (AIHA) were included (Birgens 2013, Michel 2017). These studies compared the efficacy of treatment with rituximab combined with corticosteroid versus with corticosteroid alone in adults. No RCTs studying other interventions for AIHA were found.

The addition of rituximab to corticosteroid resulted in higher complete haematological response rates at 12 months (two trials). Relapse‐free survival (RFS) was better in the rituximab group with follow‐up up to four and eight years, respectively in two trials. The proportion of participants with partial haematological response reported at six months (one trial) and 12 months (one trial) did not differ with the addition of rituximab to corticosteroid. REd blood cell (RBC) transfusion requirement was also similar between the two study arms (two trials). Lower rate of direct antiglobulin test (DAT) positivity at 12 months was reported in the rituximab arm (one trial). Similar frequencies of adverse effects and serious adverse events were observed between the two groups (one trial).

Overall completeness and applicability of evidence

While the included studies offered evidence for the efficacy of adding rituximab to corticosteroid in adults with warm‐type AIHA for induction of complete or partial haematological response, the reported data did not allow some other predefined primary and secondary objectives to be addressed. In particular, there is a lack of data on patient‐relevant outcome including adverse effects, overall survival (OS,) RFS and quality of life (QOL) of participants involved. This patient‐centered information would be crucial in the decision‐making process for people with AIHA considering the use of rituximab, beyond rate of recovery in haematological parameters. On the other hand, RCTs are only available on comparing the use of rituximab in combination with corticosteroid against corticosteroid alone, this did not provide evidence for the use of other immunomodulatory therapies such as corticosteroid alone, splenectomy, mycophenolate mofetil (MMF), or intravenous immunoglobulin (IVIg). This represents a major limitation of our review and offers no support for physicians to consider alternative agents or in people who fail with rituximab and steroid therapy. Such observation also implies that most immunomodulatory agents have been used based non‐randomised trials and as off‐label indications. AIHA also refers to a broad spectrum of entities. The included RCTs only recruited adults and those with newly diagnosed, warm‐type AIHA, limiting generalisability especially for people with relapsed disease, and those with cold‐type AIHA where the pathogenic mechanism differs The disease in the paediatric age group and as well as effective strategies for cold‐type or relapsing AIHA remain to be evaluated in high‐quality clinical trials. The lack of data on QOL measures should be noted as AIHA is a chronic condition. Overall, high‐quality evidence on the management of AIHA is still lacking.

Quality of the evidence

One of the two included studies had low risk of bias, the other study had high risk of performance and detection bias due to lack of blinding. The overall certainty quality of evidence for complete haematological response was low, as one of the included trials had high risk of bias and the overall sample size was small. The overall certainty of evidence for partial haematological response and RBC transfusion requirement were very low, as both outcomes were based on only one trial and sample size was small. This calls for the need for further blinded RCTs to consolidate the available data on the benefit of combining rituximab with corticosteroid over corticosteroid monotherapy.

Potential biases in the review process

We performed a independent,extensive literature search by two review authors for published RCTs. It is possible that studies may be missed by our current strategy. The limited number of studies included did not allow us to analyse for publication bias using a funnel plot.

Agreements and disagreements with other studies or reviews

Our findings are in line with a meta‐analysis of 21 studies (Reynaud 2015), of which 20 were non‐randomised observational studies and one being an RCT included in our review (Birgens 2013), which suggested efficacy and safety in the use of rituximab for AIHA.

Authors' conclusions

Implications for practice.

The addition of rituximab to corticosteroid may improve complete response rate in adults with newly diagnosed warm, primary autoimmune haemolytic anaemia (AIHA), in comparison with the use of corticosteroid alone. There is a lack of high‐certainty evidence to support the use of other immunosuppressive therapy.

Implications for research.

Further studies are required to consolidate the evidence for combining rituximab with corticosteroid in warm AIHA. In addition, high‐quality RCTs are needed to address the efficacy of the numerous immunosuppressive/immunomodulatory approaches in managing AIHA beyond rituximab. Studies should also target various populations based on patient age and differences in disease pathophysiology. Studies with adequate follow‐up, report on survival and quality of life (QOL) measures should be prioritised in such chronic conditions.

History

Protocol first published: Issue 1, 2017
Review first published: Issue 3, 2021

Date	Event	Description
26 March 2020	Amended	Incorporated feedback from editor
14 October 2019	Amended	Review draft completed

Appendices

Appendix 1. MEDLINE search strategy

1. exp ANEMIA, HEMOLYTIC, AUTOIMMUNE/

2. autoimmune hemolytic anemia.mp.

3. autoimmune haemolytic anaemia.mp.

4. (haemolyt$ or hemolyt$).tw.

5. Evans.tw.

6. AIHA.tw.

7. ((agglutinin* or antibod*) adj2 cold* adj2 diseas*).tw,kf,ot.

8. or/1‐7

9. randomized controlled trial.pt.

10. controlled clinical trial.pt.

11. randomi?ed.ab.

12. placebo.ab.

13. drug therapy.fs.

14. randomly.ab.

15. trial.tw.

16. groups.ab.

17. or/9‐16

18. human.sh.

19. 17 and 18

20. 19 and 8

Appendix 2. Embase search strategy

1. exp ANEMIA, HEMOLYTIC, AUTOIMMUNE/

2. autoimmune hemolytic anemia.mp.

3. autoimmune haemolytic anaemia.mp.

4. (haemolyt$ or hemolyt$).tw.

5. Evans.tw.

6. AIHA.tw.

7. ((agglutinin* or antibod*) adj2 cold* adj2 diseas*).tw.

8. or/1‐7

9. (randomi$ or placebo$ or single blind$ or double blind$ or triple blind$).ti,ab.

10. RETRACTED ARTICLE/

11. or/9‐10

12. (animal$ not human$).sh,hw.

13. (book or conference paper or editorial or letter or review).pt. not exp randomized controlled trial/

14. (random sampl$ or random digit$ or random effect$ or random survey or random regression).ti,ab. not exp randomized controlled trial/

15. 11 not (12 or 13 or 14)

16. 8 and 15

Appendix 3. LILACS search strategy

db:("LILACS") AND type_of_study:("clinical_trials") AND ((autoimmune hemolytic anemia) OR (autoimmune haemolytic anaemia) OR hemoly* OR haemoly* OR Evans OR AIHA)

Appendix 4. CENTRAL search strategy

#1 MeSH descriptor: [Anemia, Hemolytic, Autoimmune] explode all trees
#2 autoimmune hemolytic anemia 
#3 autoimmune haemolytic anaemia 
#4 (haemolytic* or hemolytic*) 
#5 ((agglutinin* or antibod*) near/2 cold near/2 diseas*) 
#6 AIHA 
#7 evans:ti,ab,kw
#8 #1 or #2 or #3 or #4 or #5 or #6 or #7 in Trials

Data and analyses

Comparison 1. Rituximab with steroid versus steroid.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Complete response at 12 months	2	96	Risk Ratio (M‐H, Fixed, 95% CI)	2.13 [1.34, 3.40]

1.1. Analysis.

Comparison 1: Rituximab with steroid versus steroid, Outcome 1: Complete response at 12 months

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Birgens 2013.

Study characteristics
Methods	Parallel group randomised trial
Participants	People treated at eight haematological centres in Denmark participated in the study. Eligible participants had newly diagnosed and previously untreated warm AIHA, participants with primary AIHA, as well as those with a concomitant autoimmune disease or a low‐grade B‐cell lymphoproliferative neoplasia, were also included. Participants with overt drug‐induced immune haemolytic anaemia were not included. Participants were allowed to have received oral prednisolone for up to 1 week before enrolment. The exclusion criteria were as follows: age under 18 years; Eastern Cooperative Oncology Group performance status above 2; previous rituximab treatment; immunosuppressive or anti‐neoplastic drugs within the last 3 months; haemolytic anaemia secondary to autoimmune disease within the last 6 months; other serious diseases, including malignancy; pregnancy or breast‐feeding; hypersensitivity to rituximab, and active infection requiring systemic treatment. In addition to careful clinical assessment, a bone marrow aspirate and biopsy and an abdominal ultrasound scan or a computerised tomography scan were performed on all to determine whether the AIHA was primary or secondary to a B‐cell lymphoproliferative disease.
Interventions	Eligible participants were randomised 1:1 by use of pre‐coded envelopes. All participants received prednisolone 1·5 mg/kg/day for 2 weeks followed by tapering according to this schedule: 0·75 mg/kg/day for 1 week (week 3), thereafter 0·5 mg/kg/day for 1 week (week 4), followed by a gradual reduction over the next 4–8 weeks to the lowest dose that was effective in maintaining a normal Hb level. In the group allocated to a combination of prednisolone and rituximab, the participants received prednisolone at the same dose and schedule as in the monotherapy group and were given rituximab at a dosage of 375 mg/m2 as an intravenous infusion once a week for 4 weeks. All participants received oral folic acid 5 mg/day, and those given a rituximab infusion also received premedication with acetaminophen 1 g and clemastine 2 mg intravenously 30 to 60 minutes before the infusion.
Outcomes	The participants underwent a full clinical examination and complete blood counts including haemolytic parameters (Hb, reticulocyte count, total bilirubin, lactate dehydrogenase, and haptoglobin) at enrolment, on days +7, +14, +21, +28, +42, +56, +70 and +84, then monthly until month 6, and finally, every third month until the end of follow‐up. This applied to all participants until they showed a lack of response necessitating either a switch to some other immunosuppressive treatment or splenectomy, or they relapsed after an initial positive response. The minimum and maximum follow‐up times after initiating treatment were 12 and 48 months, respectively. The primary objective of the study was to analyse differences in treatment responses between the two groups. Responses were evaluated at 3, 6 and 12 months after treatment was initiated. Complete response (CR) was defined as normalisation in Hb concentration without any ongoing immunosuppressive treatment and without any biochemical signs of haemolytic activity. Partial response (PR) was defined as being similar to CR but requiring continued low‐dose prednisolone (<10 mg/day), or appearing as compensated haemolytic anaemia entailing a stable, acceptable Hb level without any need of treatment except < 10 mg/day prednisone. Secondary objectives of our investigation were to evaluate differences in relapse‐free survival (RFS), red blood cell transfusion requirement after treatment, the need for splenectomy, and safety profiles associated with the treatments up to 12 months after enrolment.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised by pre‐coded envelopes
Allocation concealment (selection bias)	Low risk	Randomised by pre‐coded envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unblinded
Blinding of outcome assessment (detection bias) All outcomes	High risk	Unblinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcome data accounted for
Selective reporting (reporting bias)	Low risk	Results reported matching methodology listed on protocol
Other bias	Low risk	No other potential sources of bias were identified

Michel 2017.

Study characteristics
Methods	Parallel group, double‐blind, placebo‐controlled, randomised trial
Participants	Inclusion criteria were (1) age ≥18 years at inclusion; (2) warm AIHA defined at time of diagnosis by Hb level ≤10 g/dL with signs of haemolysis (at least haptoglobin level<4 mg/L), and a positive direct antiglobulin test (DAT) result (IgG or IgG1 complement C3d pattern); (3) warm AIHA diagnosed and treated<6 weeks before inclusion; (4) serum gammaglobulin level >5 g/L at inclusion; (5) absence of detectable lymph nodes on total body CT‐scan (performed before inclusion if not performed at warm AIHA diagnosis); (6) Evans’ syndrome accepted with presence of the other inclusion criteria and platelet count>30x109/L at inclusion; (7) females of childbearing age with negative pregnancy test results and effective contraceptive method within at least 6 months after inclusion; and (8) informed and written consent.
Interventions	Participants were randomised to receive rituximab or placebo on top of corticosteroid. Premedication with 100 mg intravenous methylprednisolone was systematically administered before rituximab or placebo as recommended. At inclusion, whatever the dose and type of corticosteroids treatment received before entering the study, all participants were given prednisone at a daily dose of 1.0 mg/kg for at least 2 weeks. After Day 15, with at least partial response (PR) achieved (defined by Hb level>10 g/dL with at least a 2 g increase from baseline), the dose of prednisone had to be tapered by 10 mg every 10 days up to 20 mg, then by 5 mg every 10 days from 20 to 10 mg, then by 2.5 mg from 10 to 5 mg and then stopped after 10 days with lasting response. With lack of response within 2 weeks after inclusion, the daily dose of prednisone could be increased up to 1.5 or 2 mg/kg at the investigator’s discretion. In the absence of at least PR within 6 weeks after inclusion despite an increased dose of prednisone, the use of other drugs such as danazol or immunosuppressors and/or an indication for splenectomy were left to the investigator’s discretion. Each treatment and dosage was specifically reported in the case report form, and the patient was considered a non‐responder in the ITT analysis. The need for transfusion was left to the investigator’s discretion.
Outcomes	The primary outcome was the efficacy of rituximab by comparing the overall response rate (PR+complete response [CR]) at 1 year in both groups. Secondary objectives were PR, cumulative dose of steroids, and number of transfusions and hospitalisations at 1 year and incidence and severity of adverse events in both groups and CR/PR at 2 years. CR was defined as Hb level ≥11 g/dL (women) or ≥12 g/dL (men) without features of ongoing haemolysis (including normal haptoglobin level), without any ongoing treatment for wAIHA on two different occasions 4‐weeks apart in the absence of any recent transfusion. CR could be considered even with a positive DAT result. PR was defined as Hb level ≥10g/dL with at least a 2 g increase from baseline (i.e. at warm AIHA diagnosis) without any other treatment than prednisone given at a daily dose 10 mg or recent transfusion. Failure was considered CR or at least PR not achieved at 1 year or if the patient had received any therapy (other than prednisone and transfusions) known to be active in warm AIHA (danazol, immunosuppressor) or had undergone splenectomy within the year after inclusion. CD19+ B lymphocytes in peripheral blood were assessed at baseline and then repeatedly at Weeks 12, 28, and 52. Investigators were blinded to results until the end of the study.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Central computerised randomisation procedure used
Allocation concealment (selection bias)	Low risk	Central computerised randomisation procedure used
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel blinded
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Statistician blinded to study conduct and monitoring
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcome data accounted for
Selective reporting (reporting bias)	Low risk	Results reported matching methodology listed on protocol
Other bias	Low risk	No other potential sources of bias were identified

AIHA: autoimmune haemolytic anaemia; CT: computerised tomography; DAT: direct antiglobulin test ; Hb: haemoglobin;ITT: intention‐to‐treat; RFS: relapse‐free survival

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Abdwani 2009	Not RCT
Almomen 2013	Not RCT
Aqrabawi 2012	Not RCT
Aqrabawi 2013	Not RCT
Barcellini 2012	Not RCT
Barcellini 2013	Not RCT
Bartko 2017	Not RCT
Bartko 2018	Not RCT
Berentsen 2017	Not RCT
Berentsen 2017a	Not RCT
Choudhry 1996	Not RCT
Crickx 2019	Not RCT
Dierickx 2015	Not RCT
Heidel 2007	Not RCT
Hill 2018	Not RCT
Jager 2017	Not RCT
Jilma 2016	Not RCT
Lehrer‐Graiwer 2016	Not RCT
Li 2020	Not RCT
Liu 2001	Not RCT
Mehmood 2016	Not RCT
Miano 2016	Not RCT
Motto 2002	Not RCT
Moyo 2002	Not RCT
Osterborg 2009	Not RCT
Panicker 2016	Not RCT
Panicker 2016a	Not RCT
Rai 2018	Not RCT
Raschetti 2012	Not RCT
Reynaud 2015	Meta‐analysis, not an interventional trial
Rice 2012	Review of literature, not an interventional trial; compares supportive care (splenectomy only)
Rodrigo 2015	Not RCT
Rossi 2018	Phase II trial, not RCT
Roth 2016	Not RCT
Shah 2013	Not RCT
Wang 2005	Not RCT
Williams 2020	Not RCT
Yang 2020	Not RCT
Zaja 2002	Not RCT
Zanella 2013	Not RCT
Zecca 2003	Not RCT

RCT: randomised controlled trial

Characteristics of ongoing studies [ordered by study ID]

Cooper 2020.

Study name	Fostamatinib for the treatment of warm AIHA: A phase 3, randomised, double‐blind, placebo‐controlled, global study
Methods	Phase 3, randomised, double‐blind, placebo‐controlled trial
Participants	Adults with primary or secondary warm AIHA who failed one or more prior therapy
Interventions	Fostamatinib or placebo for 24 weeks
Outcomes	Haemoglobin response (haemoglobin level ≥10 g/dL with a ≥2 g/dL increase from baseline in the absence of rescue therapy); duration of haemoglobin response; need for warm AIHA rescue treatment; and incidence of adverse events.
Starting date	1 April 2019 (first patient recruited)
Contact information	Rigel Pharmaceuticals
Notes	NCT03764618

NCT04119050.

Study name	Efficacy and safety of M281 in adults with warm autoimmune haemolytic anaemia
Methods	Phase 3, randomised, double‐blind, placebo‐controlled study
Participants	Adults with primary or secondary warm AIHA
Interventions	M281 (human anti‐neonatal Fc receptor antibody) or placebo
Outcomes	Primary outcome: Number of Participants That Attain Hemoglobin (Hgb) Response [ Time Frame: Up to Week 20 ] Secondary outcomes: Percentage of Participants That Achieve any Reduction in the Daily Dose of Corticosteroids While Maintaining Hgb Response [ Time Frame: Baseline (Day 1, Week 0) through 24 weeks ] Percent Reduction in Daily Corticosteroid Dose [ Time Frame: Baseline (Day 1, Week 0) through 24 weeks ] Hgb Range at Steady State [ Time Frame: Baseline (Day 1, Week 0) through 24 weeks ]It will be estimated using a model‐based longitudinal analysis of Hgb/hemolysis parameters in relationship to IgG level and dose regimen. Percentage of Participants With Hgb Within the Normal Range for Their Gender [ Time Frame: Baseline (Day 1, Week 0) through Week 16 and Week 24 ] Change From Baseline in Hgb [ Time Frame: Baseline (Day 1, Week 0) through Week 16 and Week 24 ] Percentage of Participants who Experience at Least a 2 g/dL Increase in Hgb From Baseline and Normalization of Lactate Dehydrogenase and Haptoglobin [ Time Frame: Baseline (Day 1, Week 0) through Week 24 ] Mean Time During Which the Primary Endpoint is Maintained [ Time Frame: Baseline (Day 1, Week 0) through Week 24 ] Change From Baseline in the Total Score From the Functional Assessment of Chronic Illness Therapy (FACIT) Fatigue Scale [ Time Frame: Baseline (Day 1, Week 0) through Week 24 ]
Starting date	8 October 2019 (first posted)
Contact information	Momenta Pharmaceuticals, Inc.
Notes	NCT04119050

Differences between protocol and review

We did not incorporate search in the China Journal Net database (search date 7 March 2021) after submission of the protocol (Liu 2017) due to the extreme brevity of full‐text articles published which made interpretation of both quality and results impractical.

Contributions of authors

APY Liu: conceiving of the review, protocol development, searching for trials, selection of studies, quality assessment of trials, data extraction, data entry, data analyses, data interpretation, development of final review, disagreement resolution, review updates, corresponding author.

DKL Cheuk: conceiving of the review, protocol development, searching for trials, selection of studies, quality assessment of trials, data extraction, data entry, data analyses, data interpretation, development of final review, disagreement resolution, review updates.

Sources of support

Internal sources

• The University of Hong Kong, Hong Kong

External sources

• No sources of support supplied

Declarations of interest

APY Liu: none known.

DKL Cheuk: none known.

New