Efficacy and safety of thrombopoietin receptor agonists in patients with primary immune thrombocytopenia: A systematic review and meta-analysis

Immune thrombocytopenia (ITP) is an autoimmune disease characterized by increased platelet destruction and impaired platelet production. In this study, we conducted a systematic review and meta-analysis to determine the efficacy and safety of thrombopoietin receptor agonists (TPO-RAs) in primary ITP patients. Thirteen randomized controlled trials were included in this study, the pooled results of which demonstrated that TPO-RAs significantly increased platelet response (R) and durable response (DR) rates [risk ratio (RR): 2.77, 95% confidence interval (CI): 2.01–3.82, P = 5.9 × 10−10; RR: 7.52, 95% CI: 3.94–14.35, P = 9.2 × 10−10; respectively] and that TPO-RAs significantly reduced the incidences of any or severe bleeding events (RR: 0.80, 95% CI: 0.67–0.95, P = 0.013; RR: 0.52, 95% CI: 0.27–0.99, P = 0.048; respectively). Moreover, our results indicated that there was a significant reduction in the proportion of patients needing rescue medications in the TPO-RA groups compared with the control groups (RR: 0.50, 95% CI: 0.42–0.59, P = 2.0 × 10−15) and that the rates of any or severe adverse events were similar between the TPO-RA and control regimens (RR: 1.01, 95% CI: 0.92–1.10; RR: 0.74, 95% CI: 0.54–1.01; respectively). These findings demonstrate that TPO-RAs are an effective and safe second-line treatment option for primary ITP patients.

patients, namely, children, may remit spontaneously several months after diagnosis. Avoidance of splenectomy may benefit these patients 2 .
However, several issues should be noted. First, the therapeutic effects of TPO-RAs vary greatly among relevant studies, and inconsistency exists with respect to the results of these studies. Second, the safety profiles of TPO-RAs are not completely reassuring, partly due to the relatively small sample sizes of the relevant studies. Third, the efficacy and safety of TPO-RAs in pediatric ITP patients have not been comprehensively reviewed. Thus, we conducted this systematic review and meta-analysis to comprehensively evaluate the efficacy and safety of TPO-RAs in adult and pediatric primary ITP patients.

Results
Study selection and characteristics. A total of 777 articles were identified, and 653 articles remained after duplicates were removed. A total of 619 of these articles were removed after their titles and abstracts were screened. The full texts of 34 potentially relevant articles were subsequently screened, and 21 of these articles were excluded (the excluded articles and the reasons for their exclusion are listed in Supplementary Table S1). Finally, 13 studies involving 1,126 ITP patients were included in the systematic review and meta-analysis (Fig. 1).
The characteristics of the included studies are listed in Tables 1 and 2. All RCTs enrolled ITP patients with platelet counts of 30 × 10 9 /L or less, with the exception of one study that enrolled ITP patients with platelet counts of less than 50 × 10 9 /L 21 . All the included patients were classified as having persistent or chronic ITP, according to the new ITP classification system 1 . Patients were also classified as failed or relapsed following more than one previous ITP treatment. Patients were allowed to receive concomitant ITP drugs, primarily corticosteroids, as long as the doses of these drugs had been stable for one month or longer before the start of the study. All studies excluded patients who had histories of bone marrow disorders or who had recently suffered arterial or venous thrombotic events.
The sample sizes of the included RCTs ranged from 18 to 234. Four studies evaluated the efficacy and safety of romiplostim in adults 9,15,20,21 , and three studies evaluated the efficacy and safety of the same treatment in children 16,22,25 . Six studies compared romiplostim with placebo 9,15,16,20,22,25 , and one study compared romiplostim with the standard of care for ITP, based on institutional practices or therapeutic guidelines 21 . Four studies regarding eltrombopag were conducted in adults [12][13][14]24 , and two studies regarding eltrombopag were conducted in Scientific RepoRts | 6:39003 | DOI: 10.1038/srep39003 children 19,23 . All six of these studies compared eltrombopag with placebo [12][13][14]19,23,24 . The durations of romiplostim and eltrombopag treatment ranged from 6 to 52 weeks and 6 to 26 weeks, respectively.
Any or severe bleeding events. Ten of the included studies reported the incidence of any or severe bleeding events. The pooled results demonstrated that TPO-RAs significantly reduced the incidence of any or severe bleeding events (RR: 0.80, 95% CI: 0.67-0.95, P = 0.013; RR: 0.52, 95% CI: 0.27-0.99, P = 0.048; respectively Fig. 3a,b). Subgroup analysis based on TPO-RA regimens indicated that romiplostim did not substantially decrease the incidence of any or severe bleeding events (RR: 0.93, 95% CI: 0.72-1.21; RR: 0.64, 95% CI: 0.31-1.35; respectively Supplementary Fig. S4, Table 3). However, regarding eltrombopag, there were substantial reductions in the incidence of any or severe bleeding events in treated patients compared with control patients (RR: 0.68, 95% CI: 0.51-0.90; RR: 0.27, 95% CI: 0.07-1.00; respectively Supplementary Fig. S4, Table 3). Subgroup analysis based on different patient populations indicated that the incidence of any or severe bleeding events did not significantly decrease with TPO-RAs vs. placebo in pediatric ITP patients (RR: 0.78, 95% CI: 0.43-1.42; RR: 0.58, 95% CI: 0.03-12.09; respectively Supplementary Fig. S5, Table 3). However, the results of these studies indicated that there were considerable reductions in the incidence of any or severe bleeding events in adult ITP patients receiving   Table 3).
Secondary outcomes. The pooled results based on eight studies indicated that there was a significant reduction in the proportion of patients needing rescue medications in the TPO-RA groups compared with the control groups (RR: 0.50, 95% CI: 0.42-0.59, P = 2.0 × 10 −15 Fig. 4a). Moreover, there was no substantial difference in the proportions of patients needing rescue medications between the different TPO-RA regimen or patient population subgroups ( Supplementary Fig. S6, Table 3). In addition, the pooled results based on three studies indicated that there was a significant increase in the proportion of patients who were able to reduce or discontinue their concurrent ITP therapies in the TPO-RA group compared with the control group (RR: 1.85, 95% CI: 1.13-3.01, P = 0.014 Fig. 4b).
Safety profiles. Thirteen studies reported the incidence of any or severe AEs. Based on data from 10 studies, the rates of any AEs were similar between the TPO-RA and control regimens (RR: 1.01, 95% CI: 0.92-1.10, P = 0.913 Fig. 5a). Additionally, based on data from 11 studies, the rates of severe AEs tended to be lower in the TPO-RA groups than in the control groups (RR: 0.74, 95% CI: 0.54-1.01, P = 0.054 Fig. 5b). No significant differences in the rates of AEs were found between the pre-specified subgroups (Supplementary Figs S7-8, Table 3).
Elalfy et al. 22 Single-center, single-blind, RCT Starting dose of 1 μ g/kg SC weekly for 12 weeks; doses were escalated to 5 μ g/kg and then tapered.
Tarantino et al. 16 Multicenter, double-blind, RCT  The AEs of interest for TPO-RAs included thrombosis, bone marrow reticulin increases, and the generation of neutralizing antibodies to TPO-RAs or endogenous TPO 17,26 . There was no substantial difference in thrombotic events between the TPO-RA and control regimens (RR: 1.08; 95% CI: 0.40-2.93 Supplementary Fig. S9, Table 3). Neutralizing antibodies to either romiplostim or TPO were not detected in six studies 9,15,16,20,21,25 , and no bone marrow reticulin/fibrosis was detected in two studies 20,22 . One patient with an increased baseline bone marrow reticulin level exhibited an increased reticulin level during romiplostim treatment, but this level subsequently returned to baseline 15 . One patient receiving romiplostim exhibited an increased bone marrow reticulin level during a 6-month post-treatment follow-up period 21 , and two patients exhibited reversible increases in their bone marrow reticulin levels during the subsequent extension study 9 .
The AEs of interest for eltrombopag included hepatobiliary function abnormalities and increases in the incidence of cataracts 27 . There was a trend toward an increased incidence of liver function abnormalities with eltrombopag vs. placebo (RR: 2.13, 95% CI: 0.74-6.17 Supplementary Fig. S10, Table 3). There was no considerable difference between the eltrombopag and control regimens with respect to the incidence of cataracts (RR: 0.89, 95% CI: 0.42-1.91 Supplementary Fig. S10, Table 3).

Discussion
Our systematic review and meta-analysis has several advantages over the previous meta-analysis conducted in 2011 28 . First, only six RCTs were included in the meta-analysis by Zeng et al., whereas 13 RCTs were included in our study. Therefore, the results of our meta-analysis may be more reliable and comprehensive than the results of the meta-analysis by Zeng et al. For example, the results of the previous meta-analysis, based on data from three RCTs, indicated that there was no significant difference in the incidence of severe bleeding events between the TPO-RA and control groups 28 , whereas the pooled results of our meta-analysis, based on the results of six RCTs, indicated that TPO-RAs significantly decreased the incidence of severe bleeding events in treated patients compared with control subjects. Second, compared with the previous meta-analysis, our analysis included more endpoints, such as the need for rescue medications and the numbers of patients who were able to reduce or discontinue concurrent ITP therapies. Third, we conducted subgroup meta-analysis based on different TPO-RA regimens (romiplostim vs. eltrombopag) or patient populations (adults vs. children) because our study included an adequate number of studies for performing such analyses.
Our meta-analysis results indicated that TPO-RAs significantly increased the rates of R or DR and reduced the incidences of any or severe bleeding events in persistent and chronic primary ITP patients compared with control subjects. Moreover, our results indicated that TPO-RAs significantly decreased the need for rescue medications and increased the numbers of patients who were able to reduce or discontinue concurrent ITP therapies. The incidence of AEs in the TPO-RA-treated groups was similar to that in the placebo groups, and there was a decreasing trend in the incidence of severe AEs in ITP patients receiving TPO-RAs compared with control subjects.
For persistent and chronic ITP patients, reducing the risk of severe bleeding may be more important than achieving specific platelet counts 12 . Our subgroup meta-analysis revealed that romiplostim did not significantly improve the incidence of any or severe bleeding events, which may be due to the fact that the effects of TPO-RA on bleeding events are affected by rescue medication use 16,29 . Patients in the TPO-RA groups needed less rescue medications than patients in the control group, which may have confounded our results regarding the incidence of bleeding events. Thus, using a composite bleeding episode endpoint including both bleeding events and/or rescue medication use or using duration-adjusted bleeding event rates may yield more detailed and accurate results than using separate endpoints 16,29 . For example, there was no significant difference in the incidence of any bleeding events between the romiplostim and placebo groups (83% vs. 74%), whereas the duration-adjusted rate of any bleeding events or composite bleeding episodes was significantly lower with romiplostim than with placebo in the study by Tarantino et al. 16 . Moreover, the incidence of bleeding events in the romiplostim group was higher than that in the placebo group (71% vs. 40%), whereas the duration-adjusted bleeding event rates were 7.3 and 11.9 in the romiplostim and placebo groups, respectively, in the study by Bussel et al. 25 . Thus, the insignificance of the effects of TPO-RAs in pediatric ITP patients was subsequently attributed to these two studies 16,25 .
It has been noted that platelet responses following TPO-RA treatment commonly gradually decrease once medications are stopped, which has raised questions regarding whether the combination of TPO-RAs with other drugs can exert additive effects and provide better clinical benefits than TPO-RAs alone. Two RCTs conducted in China compared the concomitant use of recombinant human thrombopoietin (rhTPO) with rituximab or danazol   30,31 , the results of which indicated that concomitant rhTPO use significantly increased the rate of R and shortened the time needed to achieve this response compared with rituximab or danazol monotherapy 30,31 . Moreover, the combination of rhTPO with rituximab was associated with an increased sustained response compared with rituximab monotherapy 30 . These two studies were not included in this study due to the presence of significant heterogeneity with respect to patient characteristics and therapy schedules. Moreover, the efficacy of rhTPO monotherapy, the efficacy and safety of rhTPO in Western ITP patients, and the efficacy of the combination of romiplostim or eltrombopag with other drugs for ITP require further clarification in future studies. Several issues should be noted. First, the pooled results of this analysis must be interpreted with caution due to the heterogeneity of the included studies. All persistent, chronic and refractory adult and pediatric ITP patients were included in the analysis. Patient characteristics (age, gender, ITP duration, splenectomy status), TPO-RA regimens (TPO-RA types, starting doses, dosing algorithms, and treatment durations), and outcome of interest definitions varied among the included studies. Second, close monitoring of hepatobiliary laboratory test results is recommended during eltrombopag treatment due to an increasing trend in the incidence of liver function abnormalities among ITP patients receiving this treatment. Third, worsening thrombocytopenia may occur after TPO-RA discontinuation 9 . Therefore, close monitoring of platelet counts and bleeding risks is recommended after TPO-RA discontinuation 3,32 . Finally, whether the youngest patients have lower response rates to TPO-RAs than older patients needs to be clarified further in future studies 16,23 . In conclusion, this systematic review and meta-analysis demonstrated that TPO-RAs are effective and safe agents for the treatment of primary ITP, and the results of this analysis support the use of TPO-RAs as second-line treatments for persistent and chronic ITP patients who are unresponsive to or relapse after receiving previous first-line ITP therapies.

Methods
Search strategy. We conducted a systematic literature search of PubMed, the Excerpta Medica Database (Embase), and the Cochrane Central Register of Controlled Trials (CENTRAL) to identify RCTs evaluating the efficacy of TPO-RAs in patients with primary ITP. We searched for trials published until May 2016 and restricted our search to English language studies. We used the following combination of search terms: "thrombocytopeni*", "ITP", "romiplostim", "eltrombopag", and "thrombopoietin". Our detailed search criteria are listed in Supplementary Tables S2-4. We searched the references of the identified studies manually to find additional studies.
Selection criteria. All RCTs involving patients with a confirmed diagnosis of primary ITP that compared TPO-RAs to placebo were considered, as were RCTs that included TPO-RAs as an additive to other drugs/therapies, as long as the additional drugs/therapies were identical between the two groups. RCTs reporting at least one outcome of interest were included in the analysis. The primary outcomes of interest included the rates of R or DR and the incidences of any or severe bleeding events. R was defined as achieving platelet counts of more than 50 × 10 9 /L at any time during TPO-RA treatment, and DR was defined as maintaining R for ≥ 60% of the duration of TPO-RA treatment or maintaining R for six or more weeks during the final eight weeks of TPO-RA treatment. Bleeding severity was assessed according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), the World Health Organization (WHO) bleeding scale, or a grading scale specified for pediatric ITP patients 33 . Severe bleeding events were defined as bleeding events ≥ grade 3, according to the National Cancer Institute CTCAE and the WHO bleeding scale, or bleeding events ≥ grade 4, according to the criteria developed by Buchanan and Adix 33 .
The secondary endpoints of interest included the need for rescue medications and the numbers of patients who were able to reduce or discontinue concomitant ITP therapies. Rescue medications were defined as any new drugs or as increased doses or frequencies of concomitant drugs that were administered for the treatment of ITP, as well as platelet transfusions or splenectomies that were performed during TPO-RA treatment.
Safety profiles were assessed according to the incidence of AEs, which were graded according to the National Cancer Institute CTCAE. Data extraction and methodological quality evaluation. Data extraction was performed by two independent researchers. All related data, including reference details (the first author, publication year, and study design), participant characteristics (age, gender, ITP duration, splenectomy status, and sample size), TPO-RA regimens, the abovementioned primary and secondary endpoints, and safety profiles, were extracted. All discrepancies were resolved by discussion or consultation with a specialist. The corresponding authors of the included studies were contacted if necessary.
The methodological quality of the included studies was assessed according to the following seven items: adequacy of randomization, allocation concealment, blinding of participants, outcome assessors, incomplete data, selective reporting, and other biases, according to the Cochrane Collaboration Reviewers' Handbook (Version 5.1.0) 34 .
Statistical analysis. All statistical analyses were performed using Stata software (version 12.0, Stata Corporation, College Station, TX, USA). Risk ratios (RRs) and 95% confidence intervals (CIs) were used to pool binary outcomes. Heterogeneity between studies was quantified using the I 2 statistic, and I 2 > 50% and P < 0.10 indicated the presence of significant heterogeneity between studies 35 . Meta-analyses were conducted using random effects models, irrespective of whether heterogeneity existed or not. Pre-specified subgroup meta-analysis based on the different TPO-RA regimens (romiplostim vs. eltrombopag) and patient populations (adults vs. children) of the included studies were also conducted. P < 0.05 was considered statistically significant.