FormalPara Key Points

Our meta-analysis indicates that patients treated with an additional eltrombopag to IST may have a higher rate of haematologic response.

Our study suggests that the application of eltrombopag in treatment can be beneficial to patients with SAA.

1 Introduction

Severe aplastic anaemia (SAA), a syndrome of bone marrow failure that is a significant threat to human health and can be fatal if left untreated, manifests as marrow hypoplasia and peripheral blood cytopenia [1,2,3]. Historically, treatment approaches for SAA involved immunosuppressive therapy (IST) and haematopoietic stem cell transplantation (HSCT) [4]. Currently, in patients with SAA aged ≤40 years and with a human leukocyte antigen (HLA) identical sibling donor, HSCT is still the preferred treatment for patients with SAA, with a high overall survival rate of more than 80% [5, 6]. However, patients are often unfit for HSCT because of their poor physical condition or donor unavailability. Furthermore, HSCT may be accompanied by complications, such as graft-versus-host disease (GVHD) and infection, which can compromise the quality of life of a patient [7]. Standard IST, including anti-thymocyte globulin (ATG) and cyclosporin A (CSA), may be an alternative treatment option for patients with no appropriate donor or for those who are ineligible for HSCT [8, 9]. In 2020, a study reported that the haematological response rate of patients treated with IST ranged from 60 to 70% [10]. However, more than 25% of patients fail to respond to IST, and approximately 75% have unsatisfactory long-term outcomes [11].

Eltrombopag (EPAG), an oral synthetic thrombopoietin receptor agonist, has been reported to be effective for both refractory and newly diagnosed SAA in adults [12, 13]. One possible mechanism is that thrombopoietin (TPO) enhances the haematopoietic stem cell function. Another possible mechanism is that TPO induces platelet maturation and release by binding to the c-MPL receptor on megakaryocytes [14]. However, adverse events such as liver toxicity and clone [15, 16] evolution of EPAG limited its use in patients with SAA [17]. Some studies have reported that patients with SAA treated with EPAG + IST have a higher haematological response rate than those not treated with EPAG [18,19,20]. However, a study conducted by Groarke et al reported that the addition of EPAG to IST did not improve the overall response rate at 6 months in children [21]. Moreover, two studies reported the overall response rates at 3 months in children and adults treated with EPAG + IST are 35.7% and 59.0%, respectively [22, 23]. The efficacy of EPAG is still disputable, and there have been some differences in the haematologic response rate between adults and children who were treated with EPAG. Therefore, we conducted this systematic review and meta-analysis to compare the efficacy of EPAG + IST with that of IST alone in patients with SAA and assessed the differences in the efficacy of EPAG in adults and children.

2 Methods

This study was performed, and the results are reported in accordance with the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [24].

2.1 Literature Search Strategy

Relevant articles up to 1 January 2023 were independently retrieved by two authors (Zhang and Wang) from PubMed, EMBASE, and the Cochrane Library. A combination of ‘Aplastic Anaemia’ and ‘Eltrombopag’ was used as the search term to retrieve relevant studies. The search strategy for PubMed is described in the Supplementary file. Moreover, we manually searched the references of all included studies to identify any other relevant studies. A third author (Junjie Fan) was consulted to address any disagreements.

2.2 Outcomes and Study Selection

The primary outcomes were as follows: overall response at 6 months (defined as complete response or partial response); complete response at 6 months (defined as an absolute neutrophil count [ANC] ≥ 1.0 × 109/L, haemoglobin [Hb] levels ≥ 100 g/L and a platelet count ≥ 100 × 109/L); partial response at 6 months (defined in each included study); and nonresponse at 6 months (defined as failing to achieve a response by 6 months). The secondary outcomes were overall response at 3 months, complete response at 3 months, partial response at 3 months, and nonresponse at 3 months. Two authors (Wang and Cui) independently screened the titles and abstracts of all studies and subsequently examined the full texts of eligible studies. If there were any disagreements in the study selection process, a consensus was reached through group discussion with a third author (Shaoyan Hu). The studies included in this meta-analysis had to meet the following criteria: (1) patients were divided into two groups: the EPAG + IST group and the IST alone group; (2) each clinical trial included more than five patients; (3) all patients with SAA were diagnosed according to the WHO diagnostic criteria; (4) retrospective or prospective original studies. Studies were excluded if they met any of the following criteria: (a) reviews, letter, meeting abstract, supplement, comment, or case report; (b) single-arm study; (c) different studies used the same datasets; (d) studies that did not report clinical outcomes of patients with SAA.

2.3 Data Extraction and Quality Assessment

Two authors (Zhang and Cheng) independently collected the baseline characteristics of each included study using a standardised Excel table. The baseline characteristics were as follows: first author, publication year, study design, number of patients, median age, number of females, disease severity, and paroxysmal nocturnal haemoglobinuria (PNH) clone. Two authors (Fan and Wang) independently used the Newcastle-Ottawa Scale (NOS) to evaluate the quality of nonrandomised controlled trials (RCTs). In contrast, the Cochrane Collaboration tool was used for RCT. Any disagreements in this process were addressed through discussion or adjudication by a third author (Shaoyan Hu).

2.4 Statistical Analysis

We used RevMan version 5.4 to perform the statistical calculations involved in this meta-analysis. A random-effects model was used to calculate odds ratios (ORs) with 95% confidence intervals (CIs) for primary and secondary outcomes. Moreover, we used I2 statistics to evaluate the heterogeneity of all the included studies. When I2 was greater than 50%, significant heterogeneity was observed among the included studies. Publication bias in primary outcomes was evaluated using funnel plots. Additionally, we performed a subgroup analysis based on age (children vs children and adults) (patients aged 0–18 years were defined as “children”).

3 Results

3.1 Study Selection and Study Characteristics

Initially, 847 potential clinical articles were retrieved and extracted through literature search. Of these, 197 were excluded after finding duplicates. A further 608 studies were excluded after reading the titles and abstracts of the remaining 650 studies according to the exclusion criteria. We then read the full texts to screen the remaining 42 articles, and 8 studies met the inclusion criteria. However, only 6 studies were ultimately included in our meta-analysis [22, 23, 25,26,27,28] because 2 studies [29, 30] included the same patients who were treated with EPAG + IST as one of our included studies [25]. A selection flowchart of the included studies is shown in Fig. 1. Of the included studies, three were prospective studies (two were RCTs) [23, 25, 28] and three were retrospective studies [22, 26, 27]. In total, 699 patients with SAA were included in our meta-analysis, including 364 treated with EPAG + IST and 335 treated with IST alone. The baseline characteristics of each study are summarised in Table 1. The results of the primary outcomes and their subgroup analyses are presented in Table 2, and those of the secondary outcomes and their subgroup analyses are shown in Table 3. There was no significant bias in the primary outcomes across the included studies because the funnel plots were roughly symmetrical (Supplementary Figs 1–4). The quality assessments of non-RCTs and RCTs are shown in Supplementary Tables 1 and 2, indicating medium and high quality, respectively.

Fig. 1
figure 1

The selection flow chart of included studies

Full size image
Table 1 Baseline characteristics of each study included in this meta-analysis
Full size table
Table 2 Evaluation of AA at 6 months for studies included in this meta-analysis
Full size table
Table 3 Evaluation of AA at 3 months for studies included in this meta-analysis
Full size table

3.2 Overall Response at 6 Months

The overall response at 6 months was evaluated in six studies, including 699 patients with SAA. Our analysis showed that the overall response rate of patients in the EPAG + IST group was higher than that of patients in the IST-alone group. (OR = 2.25; 95% CI, 1.60–3.16; p < 0.00001, I2 = 0%) (Fig. 2). In the subsequent subgroup analysis, we found that the results in the “children and adults” subgroup were consistent with the initial pooled results (OR = 2.58; 95% CI, 1.72–3.86; p < 0.00001, I2 = 0%), whereas no significant difference was identified in the children subgroup (OR = 1.63; 95% CI, 0.87–3.05; p = 0.13, I2 = 0%) (Table 2).

Fig. 2
figure 2

Overall response at 6 months. CI confidence interval, EPAG eltrombopag, IST immunosuppressive therapy

Full size image

3.3 Complete Response at 6 Months

The complete response at 6 months was evaluated in 6 studies that included 699 patients with SAA. Our analysis results of 6 studies suggested that the complete response rate was higher in the EPAG + IST group than in the IST alone group (OR = 2.61; 95% CI, 1.82–3.74; p < 0.00001, I2 = 0%) (Fig. 3). Then, the results of both the children and “children and adults” subgroups were similar to the previous analysis results (Table 2).

Fig. 3
figure 3

Complete response at 6 months. CI confidence interval, EPAG eltrombopag, IST immunosuppressive therapy

Full size image

3.4 Partial Response at 6 Months

Partial response at 6 months was evaluated in 4 studies, including 632 patients with SAA. There was no significant difference between the EPAG + IST group and the IST alone group according to the pooled results (OR = 0.94; 95% CI, 0.49–1.81; p = 0.85, I2 = 68%) (Fig. 4). The results of subgroup analysis were similar to those of the previous analysis (Table 2).

Fig. 4
figure 4

Partial response at 6 months. CI confidence interval, EPAG eltrombopag, IST immunosuppressive therapy

Full size image

3.5 No Response at 6 Months

No response at 6 months was reported in 4 studies that included 632 patients with SAA. We found that the nonresponse rate of the EPAG + IST group was lower than that of the IST alone group according to the pooled results (OR = 0.32; 95% CI, 0.19–0.52; p < 0.00001, I2 = 30%) (Fig. 5). The results of the subgroup analysis in the “children and adults” subgroup were similar to our previous analysis results. By contrast, no significant differences were identified in the children subgroup (Table 2).

Fig. 5
figure 5

No response at 6 months. CI confidence interval, EPAG eltrombopag, IST immunosuppressive therapy

Full size image

3.6 Secondary Outcomes of this Meta-Analysis

Our analysis results suggested that there was no significant difference between the EPAG + IST group and the IST alone group in the 3-month overall response (OR = 1.52; 95% CI, 0.39–6.00; p = 0.55, I2 = 80%). In the subsequent subgroup analysis, the results for the children subgroup were consistent with the initial pooled results. In contrast, the occurrence rate of the 3-month overall response in the “children and adults” subgroup was higher in the EPAG + IST group.

Moreover, our pooled analysis indicated that the 3-month complete response rate (OR = 2.14; 95% CI, 1.09–4.23; p = 0.03, I2 = 0%) and 3-month partial response rate (OR = 2.10; 95% CI, 1.21–3.65; p = 0.008, I2 = 0%) were higher in the EPAG + IST group between the two groups. The subgroup analysis indicated that the 3-month complete response rate and 3-month partial response rate in the “children and adults” subgroup were consistent with the initial pooled results, whereas significant difference was noted in the results of the children subgroup between the two treatment groups.

In addition, the 3-month nonresponse rate was lower in the EPAG + IST group according to the pooled results (OR = 0.31; 95% CI, 0.18–0.53; p < 0.0001, I2 = 0%). The results in the “children and adults” subgroup were consistent with the initial pooled evidence, but no significant difference was noted in the results of the children subgroup (Table 3).

3.7 Adverse Events

Adverse events reported in patients treated with EPAG include elevated liver enzymes, infection, bilirubin increase, jaundice, hypertension, seizure, febrile neutropenia, rash, diarrhoea, gastrointestinal disorders, myalgia and joint pain, renal failure, cardiac disorders, skin hyperpigmentation, blurred vision, and ileus. The most common adverse events are gastrointestinal disorders, increased bilirubin, and elevated liver enzyme levels [22, 23, 26,27,28]. In a study conducted by Goronkova et al. [28], the incidence of liver abnormalities was higher in the EPAG group. A study conducted by Lesmana et al. [27] indicated that the incidence of renal insufficiency was higher in patients treated with EPAG.

Most studies did not report the discontinuation of EPAG due to severe adverse events. A study conducted by Peffault et al. [23] indicated that EPAG was discontinued in ten patients because of elevated liver enzymes (four patients), bone marrow reticulin deposition, cytomegalovirus infection, rhabdomyolysis, pulmonary embolism, and skin toxidermia. The study conducted by Goronkova et al. [28] reported that the only adverse events that led to the discontinuation of EPAG were liver test abnormalities. Liver toxicity could be a crucial adverse event of EPAG and the primary reason for its discontinuation. Therefore, monitoring liver function during EPAG administration is important.

4 Discussion

In 2019, Hong et al. conducted a meta-analysis to explore the efficacy and safety of EPAG for aplastic anaemia (AA) and found that EPAG was effective for both refractory AA and ATG-naïve AA. In addition, their study indicated that EPAG had different effects in patients with AA treated with different regimens, accompanied by clone evaluation and adverse effects (such as haemorrhage and infection) during the treatment period [31]. Several studies have reported the safety and efficacy of EPAG + IST versus IST alone for the treatment of SAA [22, 23, 25,26,27,28]. Their findings are disparate, and the application of EPAG in children and adults remains controversial. Therefore, we performed this meta-analysis to compare the efficacy of EPAG + IST with IST alone in patients with SAA. Our meta-analysis included six studies involving 364 patients with SAA in the EPAG + IST group and 335 patients with SAA in the IST alone group. Most comparison outcomes were significantly different. Based on the primary and secondary outcomes, we concluded that the efficacy of IST+ EPAG is superior to that of IST alone. We noted that the overall response rate, an important efficacy evaluation indicator, differed when analysed at 6 and 3 months. The 6-month overall response rate was better than that at 3 months. This difference can be explained by the fact that the onset time of EPAG on SAA may be relatively long, and it can be used for longer than 3 months, even if no improvement is achieved [32].

Subgroup analysis was performed to investigate whether there was a difference in efficacy between children and adults (children vs children and adults). The reason we divided the patients into the children and “children and adults” subgroups was that four of our included articles studied children, and two of them studied the combination of children and adults. Nevertheless, we can still estimate how age affects the efficacy of EPAG using a subgroup analysis. The results of the subgroup analysis indicated that EPAG + IST could be better than IST alone in adult patients with SAA; however, for paediatric patients, the addition of EPAG may not significantly improve the efficacy. We speculate that this discrepancy may be due to the initial ANC. As reported, patients with SAA and a higher initial ANC responded better than those with a lower ANC, whereas more adult patients with SAA had a higher ANC [30]. Another possible reason is that paediatric patients respond better to IST than adult patients. Subsequently, based on the IST, an additional EPAG can further improve the efficacy in adult patients [33,34,35,36]. This difference between different age groups may be helpful for future protocols and decision making regarding SAA treatment.

The addition of EPAG resulted in a better response, especially in adult patients with SAA. The adverse effects of EPAG, such as liver toxicity and clone evolution, should not be ignored. Because EPAG plays a role in haematopoietic stem cell proliferation, patients with SAA treated with EPAG + IST may be more vulnerable to other haematological diseases, such as PNH, acute myeloid leukaemia, and myelodysplastic syndrome [20]. A prospective study conducted by Townsley et al. [12] reported that 18% of patients who underwent EPAG + IST had liver test abnormalities, indicating that medical staff should regularly monitor the liver function of patients. Relapse is another complication of EPAG + IST. Another prospective study published in BLOOD reported that patients with SAA treated with EPAG + IST relapsed with a cumulative rate of 39%, mostly occurring within the first 2 years after the use of anti-thymocyte globulin [25]. Therefore, preventive strategies should be further investigated to reduce the rate of relapse.

Recently, several studies have explored the efficacy of other thrombopoietin receptor agonists such as romiplostim and avatrombopag in patients with AA. One multicentre study indicated that the haematological response rates of 27 weeks and 53 weeks in refractory patients with SAA treated with romiplostim were 84% and 81%, respectively [37]. Another retrospective study conducted in France reported only one response among 14 patients with refractory SAA, but no severe adverse events arising from the application of romiplostim [38]. A retrospective single-centre study conducted by Chi et al. [39] explored the efficacy of avatrombopag in chemoradiotherapy-induced AA and found that the overall response rates to avatrombopag at 3 and 6 months were 55.9% and 58.8%, respectively. However, studies on the efficacy of other thrombopoietin receptor agonists in patients with AA are still lacking and should be further investigated.

Moderate heterogeneity existed in several outcomes, such as the 6-month partial response and 3-month overall response, which could limit the reliability of our results. However, we could not further analyse the source of heterogeneity because of the limited number of studies included. We assumed that the baseline data differences among the included studies, such as the PNH clone and initial absolute reticulocyte count (ARC), may be the source of heterogeneity. In 2021, Tu et al. conducted a meta-analysis to evaluate the role of PNH clones in patients with AA. The pooled results showed that the 6-month overall response rate in the PNH clone-positive group was higher than that in the PNH clone-negative group, which may indicate that the PNH clone is predictive of the haematologic response in patients with AA [40]. Moreover, another recent retrospective study reported that patients with SAA who responded to EPAG + IST had a higher initial ARC than those who did not, suggesting that the initial ARC of patients treated with EPAG + IST may be related to these haematological responses [30]. Unfortunately, not all data were available in the included studies; therefore, we could not avoid heterogeneity caused by these factors.

This meta-analysis had some limitations that should be emphasised. First, due to the limited number of included studies and sample sizes, our results may have limited reliability. Second, the definition of a partial response differed among the included studies, which may have caused potential bias. For example, the definition of partial response in a randomised prospective multicentre trial by Goronkova et al., was an ANC ≥ 0.5 × 109/L, Hb levels ≥ 85 g/dL, and platelet counts ≥ 30 × 109/L, whereas the definition in a retrospective study by Jie et al., was an ANC ≥ 0.5 × 109/L, Hb levels ≥ 80 g/dL, and platelet counts ≥ 20 × 109/L. Third, some baseline characteristics of the included studies could not be determined, which may have led to bias. Fourth, there were three retrospective studies and three prospective studies among the six studies included; therefore, the mixed results of the different types of studies may have affected the conclusions. Fifth, the conclusion may be inaccurate due to the subgroup analysis of “children” and “children and adults” groups. Further studies should be performed to explore the reliability of the conclusions of our study and to identify whether there is a difference in the efficacy of EPAG between children and adults.

5 Conclusions

Our results showed that the addition of EPAG to IST resulted in a better haematologic response rate in patients with SAA than in those treated with IST alone. However, this conclusion is not reliable in children. Large-scale, high-quality RCTs are required to further investigate the efficacy of EPAG because of the limitations of our study.