Eltrombopag versus romiplostim in treatment of children with persistent or chronic immune thrombocytopenia: a systematic review incorporating an indirect-comparison meta-analysis

In absence of direct comparison, we conducted an indirect-comparison meta-analysis to evaluate the efficacy and safety of thrombopoietin-receptor agonists(TPO-RAs) in treatment of pediatric persistent or chronic immune thrombocytopenia(ITP). PubMed, Embase, Cochrane Library, Clinical Trials.gov, China National Knowledge Infrastructure, and Chinese Biomedical Literature Database were searched from their earliest records to May 2017. Randomized controlled trials comparing the TPO-RAs with placebo in pediatric ITP were included. Outcomes included overall response rate(primary), durable response, overall or clinically significant bleeding, the proportion of patients receiving rescue medication, and safety. Five randomized placebo-controlled studies(N = 261) were analyzed. The overall response[Risk Ratio(RR) 0.57, 95% confidence interval(CI) 0.21–1.56], the incidence of adverse events (RR 0.96, 95%CI 0.66–1.39), durable response(RR 2.48, 95%CI 0.31–19.97), and the proportion of patients receiving rescue treatment(RR 0.73, 95%CI 0.20–2.73) were similar between eltrombopag and romiplostim group. Nevertheless, eltrombopag might have lower risk of overall bleeding(RR 0.43, 95%CI 0.23–0.80) and clinically significant bleeding(RR 0.33, 95%CI 0.12–0.89) than romiplostim. This meta-analysis suggests that eltrombopag might be similar to romiplostim in efficacy and safety, but seems to reduce the risk of bleeding compared to romiplostim. Furthermore, cost of the treatment, comorbidity of patients and drug compliance should also be considered in clinical decision making.

Immune thrombocytopenia (ITP) is an immune-mediated disease characterized by transient or persistent decrease in the platelet count and increased risk of bleeding 1 . The pediatric ITP incidence reported in recent population-based studies varied greatly in different countries: 1.91 per 10 5 in Japan 2 , 2.83 per 10 5 in France 3 , 4.2 per 10 5 in the United Kingdom 4 and 14.5 per 10 5 in Korea 5 , respectively. Most children with ITP achieve remission spontaneously over time, however, 23.1%~47.3% of them develop a disease course of more than 6 months 6 .
For management of pediatric ITP, the first-line treatments include observation, corticosteroids, intravenous immunoglobulin (IVIg) and anti-D immunoglobulin [7][8][9] . If previous treatments fail, subsequent treatments may include splenectomy, rituximab, thrombopoietin receptor agonists (TPO-RAs), or more potent immunosuppression 7 . Eltrombopag (ELT), an oral non-peptide TPO-RA which could facilitate the proliferation and differentiation of megakaryocytes and increase platelet production 10 , has been approved by the United States Food

Results
Search result and characteristics of included studies. A total of 3,499 citations were obtained from the literature search and Fig. 1 showed the selection process. Five randomized, placebo-controlled studies (261 participants) 18,19,[26][27][28] were included in this systematic review. Agreement between two reviewers for study selection was excellent (K = 0.89). As shown in Table 1, four multicenter studies were conducted in 16 countries or regions (USA, UK, Canada, Spain, France, the Netherlands, Argentina, Czech Republic, Germany, Hong Kong, Israel, Italy, Russia, Taiwan, Thailand and Australia) 18,19,27,28 except one single-center study conducted in Egypt 26 .
All participants were aged 1-17 years old, with disease duration more than 6 months and baseline platelet count less than 30 × 10 9 /L. Two studies (159 participants) evaluated the efficacy and safety of ELT in comparison to placebo 27,28 (Table 2). The initial dose of ELT was based on age and weight, and the following dose was adjusted according to individual platelet counts, with a target of 50-200 × 10 9 /L 27 . The initial dosage and adjustment were shown in Table 3. Three studies evaluated the efficacy and safety of ROM 18,19,26 . It was administrated at the initial dose of 1 μg/kg and was also adjusted according to platelet counts ( Table 3). The double-blind phase of included studies ranged from 7 to 24 weeks. Fig. 2, four studies 18,19,27,28 had low risk of selection bias for central randomization while the other one 26 was unclear because the method of randomization and allocation concealment were not reported. Four studies 18,19,27,28 had low risk of performance bias and detection bias, as both participants and study personnel were masked, however, this risk was not clear in one study 26 for failing to report who was blinded. All studies 18,19,[26][27][28] had low risk of attrition bias, as there was no loss to follow-up or they dealt with the missing data properly (e.g. applying intention-to-treat analysis which could underestimate the efficacy of the interventions). Four studies 18,19,27,28 had low risk of reporting bias since they were registered in ClinicalTrials.gov and had reported all predesigned outcome, however, the other study 26 did neither mention registration information nor have an available protocol, so it was unclear whether all the pre-designed outcomes in this study had been reported. Four studies 18,19,27,28 were supported by pharmaceutical industry, the bias caused by conflict of interest was unclear.

Risk of Bias assessment. As shown in
Overall Platelet Response. The overall platelet response was reported in all studies (two 18,19 for ELT and three [26][27][28] for ROM) including 261 participants. The heterogeneity was not statistically significant (I 2 = 26%, P = 0.24 and I 2 = 0%, P = 0.55, respectively). The pooled results with a fixed-effect model (Table 4) showed that proportion of participants achieving overall response was significantly higher in the TPO-RAs group than in the placebo group (RR = 2.64, 95%CI: 1.58-4.44, P < 0.05 for ELT and RR = 5.05, 95%CI: 2.21-11.53, P < 0.05 for ROM). The result of indirect comparison (Fig. 3) indicated that the overall response between ELT and ROM was not significantly different (RR = 0.57, 95%CI: 0.21-1.56, P > 0.05).

Adverse events (AEs).
Four studies (197 participants) 19,26-28 reported the overall incidence of any adverse events reported in participants receiving TPO-RAs or placebo. The pooled analysis (Table 4) showed that the incidence was not significantly different between two groups (RR = 0.97, 95%CI: 0.72-1.29, P > 0.05 for ELT and RR = 1.00, 95%CI: 0.69-1.45, P > 0.05 for ROM). And the result of indirect comparison (Fig. 4) also showed that the overall incidence of any AEs in ELT group was similar to that in ROM (RR = 0.96, 95%CI: 0.66-1.39, P > 0.05).
SAEs were reported in all studies 18,19,[26][27][28] , and the results of both direct and indirect comparison (Table 4 and The most common AEs in TPO-RAs and placebo were headache, vomiting, upper respiratory tract infection, pyrexia, cough, epistaxis, oropharyngeal pain, abdominal pain, and upper abdominal pain. However, the results of head-to-head comparison (Table 4) of the incidences demonstrated no significant difference between TPO-RAs and placebo, except that vomiting was less likely to occur in ELT than in placebo (RR = 0.31, 95%CI: 0.12-0.82, P < 0.05). According to the result of indirect comparison (Fig. 4), the incidences of these AEs in ELT were similar to that in ROM, except that cough was more frequent in ELT than in ROM (RR = 14.40, 95%CI: 1.21-171.13, P < 0.05).

Discussion
This is the first systematic review incorporating an indirect-comparison meta-analysis summarizing the evidence of efficacy and safety of TPO-RAs in children with persistent or chronic ITP. Our study suggests that the use of TPO-RAs may improve the durable and overall platelet response compared with placebo, while ELT resembles ROM in efficacy. It's shown that ELT might be superior to ROM or placebo in reducing bleeding risk (including clinically significant bleeding). The incidences of AEs (including SAEs) of ELT, ROM and placebo were similar. Neither ELT nor ROM could reduce the need for rescue treatment. A systematic review on the efficacy and safety of TPO-RAs in adults and children concluded that TPO-RAs were effective and safe second-line options for primary ITP patients 20 . According to its subgroup analysis for children, TPO-RAs could significantly improve platelet response (RR = 2.49, 95%CI: 1.46-4.23, P < 0.05) and durable response (RR = 7.64, 95%CI: 2.73-21.36, P < 0.05), without increasing the incidence of AEs (including SAEs) compared with placebo or standard care. Those findings are mostly consistent with the results of this review on children. However, our research indicated that ELT might reduce the risk of bleeding compared to ROM or placebo, and the TPO-RAs did not reduce the need for rescue treatment, which are inconsistent with previous study results. These differences might be caused by the different ways in pooling data: the previous study pooled the results from studies on ELT and ROM together, while our study conducted analysis on the two drugs separately.
An indirect comparison conducted in adult ITP patients demonstrated that ROM significantly improved overall platelet response compared to ELT, while the durable platelet response of the two TPO-RAs was similar 25,29 . Our research drew a different conclusion that ELT and ROM were similar in overall or durable platelet response for children with persistent or chronic ITP, probably due to population difference.
Nonetheless, an updated systematic review and meta-analysis of RCTs suggested that TPO-RAs were associated with higher risk of thromboembolic events compared with placebo or standard care 32 . And another meta-analysis of three large, population-based observational studies concluded that the risk of arterial and venous thromboembolism should be considered when evaluating the risk of thromboembolism attributed to ITP treatments (e.g. TPO-RAs) 33 . A disproportionality analysis in the World Health Organization global individual case safety report (ICSR) database (VigiBase) suggested the presence of a signal for an increased risk of thrombosis with ELT compared to ROM (adjusted reporting odds ratio = 1.72, 95%CI 1.47-2.02, P < 0.05) 34 . Although these AEs were not reported in the RCTs due to small sample size, physicians should be cautious when administering TPO-RAs to patients with higher risk of thromboembolism, especially for ELT.
In spite of increasing pharmacoeconomic studies revealing the advantage of ELT in cost-effectiveness for adult patients with ITP [35][36][37] , further studies are still needed to compare the benefits and cost of different TPO-RAs for pediatric persistent or chronic ITP to facilitate better clinical decision making.
There are several limitations in this study. As we only included RCTs in this review, the results may not have good generalizability for strict inclusion criteria and small sample size in those studies. These studies were also not sensitive enough to find rare AEs related to the drugs, as the sample size was relatively small. On the other hand, the results of indirect comparisons between ELT and ROM should be interpreted with caution because of low power of test and heterogeneity caused by different study designs, patient populations, and outcome definitions.
In summary, ELT and ROM might lead to similar overall and durable platelet response and incidence of adverse events in children with persistent or chronic ITP, but patients receiving ROM might have lower risk of bleeding compared to those receiving ROM. Cost of treatments, comorbidity and drug compliance should also be considered when making decisions in the management of pediatric persistent or chronic ITP with thrombopoietin-receptor agonists.

Methods
We followed the standard set by Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) in this systematic review ( Table 1). The study was registered in PROSPERO International Prospective Register of Systematic Review (PROSPERO 2017: CRD42017068657).

Searching. Pubmed, Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) published in
Cochran Library were searched using the search strategies detailed in Table S2, from their earliest records to May 2017. Clinical Trials.gov was searched using the terms "immune thrombocytopenia", "children", "eltrombopag", and "romiplostim". The China National Knowledge Infrastructure(CNKI) and Chinese Biomedical Literature Database(CBM) were also searched in Chinese.

Eligibility Criteria.
All included studies met the following criteria: (1) randomized controlled studies; (2) participants were children <18 years with ITP lasting for 6 months or longer; (3) the intervention was ELT or ROM irrespective of dosage and schedule; (4) the comparison was placebo; (5) studies included at least one of the following outcomes: overall platelet response (primary outcome), defined as achieving at least once platelet response (50 × 10 9 /L or more) during treatment; incidence of overall and serious adverse events (SAEs); durable platelet response, defined as achieving platelet counts of 50 × 10 9 /L or more for 4 or more weeks; incidence of clinically significant bleeding (WHO Grade 2~4); all bleeding events; and the proportion of patients who received rescue treatment[e.g. receiving any unscheduled or new treatment (including new drugs, increase the dose of a concomitant drug from baseline, platelet transfusion or splenectomy) for immediate risk or treatment failure]; (6) publication written in English or Chinese. We excluded the patients with Evans syndrome or secondary ITP and the studies including both children and adults when children data could not be extracted separately.

Study Selection and Data Extraction. Two authors independently screened the titles and abstracts of all
studies identified by the search strategies, and assessed the studies using predetermined inclusion criteria. The full texts of all potentially relevant articles were retrieved for detailed review. We resolved any disagreements by discussion until consensus was achieved. We used a pre-designed data collection form to extract data from each eligible study. The following data were extracted: (1) authors; (2) year of publication; (3) country or region where the study conducted; (4) study design and use of control; (5) number of participants randomized into each group; (6) gender, age, and disease duration of participants; (7) baseline platelet count, previous ITP medication, and splenectomy status; (8) dose and schedule of TPO-receptor agonists; (9) outcomes of each study and their definitions; (10) numerical data for assessment of included outcomes; (11) sources of funding.
Risk of Bias Assessment. Two authors independently assessed the risk of bias of each included study using the checklist developed by Cochrane Collaboration 38 . The items included random sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other bias. We categorized the judgments as low, high or unclear risk of bias and created plots of risk of bias assessment in Review Manager Software (RevMan 5.3). Statistical Synthesis. We calculated a kappa statistic for measuring the agreement level between two authors making decisions on study selection. The value of kappa (K) between 0.40 and 0.59 was considered as fair agreement, between 0.60 and 0.74 as good and 0.75 or more as excellent.
If more than one study reported the same outcome, the pairwise meta-analysis was conducted to calculate the pooled estimate of the risk ratio (RR) of different TPO-RAs versus placebo by RevMan 5.3. Statistical heterogeneity among studies was examined by the Chi-square test and quantified by the I 2 statistic 39 . We used a fixed-effect model to synthesize data when heterogeneity was not significant (P > 0.1 and I 2 < 50%). When heterogeneity was significant (P ≤ 0.1 and I 2 ≥ 50%) and could not be explained by subgroup analyses or in terms of clinical or methodological features of the trials, the random-effect model was used. If both the ELT-placebo and ROM-placebo trials reported the same outcome, the relative treatment effect (RR) for ELT versus ROM was estimated using indirect procedure of Stata12.0 software 40  The subgroup analysis was also conducted according to the different types of adverse events.