- Top of page
- Supporting Information
Primary immune thrombocytopenia (ITP) is an acquired immune-mediated disorder with absence of any underlying cause. Corticosteroids are the standard initial treatment. Splenectomy is the main second-line treatment. A trend to delay or avoid splenectomy has developed thanks to new agents like rituximab. Few studies have assessed the response rate to rituximab in chronic ITP. We performed the first meta-analysis of randomized clinical trials and observational studies on rituximab as an effective splenectomy-avoiding option in adult chronic ITP. Overall methods were adapted from published guidelines for meta-analysis (meta-analysis of observational studies in epidemiology and preferred reporting items for systematic reviews and meta-analyses). Two haematologist investigators carried out study selection and data extraction independently, recording overall response rate (ORR) and complete response (CR) as primary assessment criteria. Of 364 records were identified through electronic databases. Of 19 retrospective or prospective observational studies were retained after removing duplicate studies and full-text analyses. The ORR was 57% (95% confidence interval [CI]: 48–65), for 368 non-splenectomized patients after rituximab; CR was 41% (95% CI: 0·33–0·51) for 346 patients. Results were stable for ORR and CR in all sub-analyses. In univariate or multivariate mixed-effect meta-regression, age was the most relevant effect. According to our results, rituximab should be used in earlier in non-splenectomized patients.
Primary immune thrombocytopenia (ITP) is an acquired immune mediated disorder characterized by thrombocytopenia, defined as a peripheral blood platelet count <100 × 109/l, with the absence of any obvious initiating and/or underlying cause of this thrombocytopenia. (Rodeghiero et al, 2009) The pathogenesis of ITP involves the generation of autoreactive antibodies against platelet antigens, which may result in both accelerating platelet destruction, essentially located in the spleen, and reducing platelet production with impaired megakaryocyte maturation (Stasi et al, 2002). It is the most common autoimmune haematological disease, with an estimated incidence of 1 in 50 000 persons (Sailler, 2008). ITP is usually chronic (>12 months) in adults and after this period, the probability of spontaneous remission is low (Rodeghiero et al, 2009). Treatment is usually carried out in patients with platelet counts <50 × 109/l or with bleeding signs (Provan et al, 2010). Corticosteroids are the standard initial treatment. Intravenous IVIg or anti-D therapy may be effective in emergency settings, and is appropriate in combination with first-line therapies (Provan et al, 2010). Less than half of patients are ‘cured’ with first-line treatment (Provan et al, 2010). For the last 20–30 years splenectomy has been proposed as the main second-line treatment for relapse or first-line refractory patients. The beneficial effects of splenectomy are due to the removal of the sites of antibody-coated platelet destruction. However, controversies exist concerning the role of splenectomy, which has evolved from being the cornerstone of ITP management to a therapy best avoided. Now, a trend to exhaust all available medical treatment options before performing such an irreversible and immunocompromising procedure as splenectomy is debated (Stasi et al, 2010). In the literature, older studies reported splenectomy rates of 50–60% or higher, whereas it is around 20–25% in more recent cohorts (Sailler, 2008). As increasing numbers of patients are reluctant to undergo splenectomy and physicians are hesitant to recommend it, a trend to delay or avoid splenectomy has developed in the USA and Europe thanks to new agents, such as rituximab, a chimeric monoclonal antibody targeting CD20 B-cell surface antigen. Rituximab has shown activity in miscellaneous autoimmune disorders by reducing circulating B cell counts (Cines & Blanchette, 2002; Robak, 2004). Its effects might be strictly related to B-cell depletion and the consequent inhibition of several B-cell pathological activities, such as the production of autoantibodies specific for platelet and megakaryocyte glycoproteins (glycoproteins IIb/IIIa and Ib/IX) (Cines & Blanchette, 2002; Robak, 2004). Moreover rituximab has been shown to up-regulate regulatory T-cells (Stasi et al, 2008; Li et al, 2011).
To date, a few studies have assessed the response rate following rituximab treatment in chronic ITP (overall response rate [ORR] 605, complete response [CR] 40%) (Provan et al, 2010). Early anti-CD20 data outcome is also inferior to outcomes seen after splenectomy. However, all but two published studies (Stasi et al, 2001; Cooper et al, 2004) have reported only a small number of patients, and most cohorts have included a mix of splenectomized and/or non-splenectomized patients. Although a recent French pilot study assessed rituximab efficacy and safety in non-splenectomized adults with chronic ITP (Godeau et al, 2008), it remains important to determine whether rituximab is an effective treatment for delaying or avoiding splenectomy in chronic ITP.
To clarify the use of rituximab in chronic ITP management, we performed the first meta-analysis of randomized controlled trials (RCTs) and observational studies on the effect of rituximab in adult non-splenectomized primary ITP patients.
- Top of page
- Supporting Information
The originality of this report is the selection of only non-splenectomized primary chronic ITP patients receiving rituximab. The goal is to highlight rituximab's place in persistent or chronic ITP treatment, especially before splenectomy. We did not compare rituximab to splenectomy.
Splenectomy has the best success rate in the treatment of ITP: with immediate response rates of 80%. Around two-thirds of patients remain in response at long-term follow up and 15–40% of patients are refractory or relapse (Mazzucconi et al, 1999; Schwartz et al, 2003; Kojouri et al, 2004; Vianelli et al, 2005). Splenectomy can be associated with immediate (general risks of surgery and perioperative complications as bleeding) and long-term morbidity (long-term vascular complications and risk of fatal bacterial infection often underestimated, requiring lifelong continuing vigilance): 10% for laparoscopic surgery to 20% for laparotomy (Mazzucconi et al, 1999; Schwartz et al, 2003; Kojouri et al, 2004; Vianelli et al, 2005). Rituximab has been used as a treatment for the last 10 years in miscellaneous immunological disorders, such as ITP. The treatment seems to be safe and effective (Cooper et al, 2004; Dolan et al, 2008; Godeau et al, 2008). However, several cases of progressive multi-focal leucoencephalopathy (PML) have been reported with monoclonal antibodies including rituximab. Carson et al (2009) reported 57 cases of PML in patients treated with rituximab over an 11-year period (from 1997 to 2008), with a 90% mortality rate. All had been treated with other immunosuppressive regimens and the majority of the patients also suffered from pathologies predisposing to PML. However, it seems that ITP is not a pathology predisposing to PML and the number of cases of ITP in is extremely low (one case each) in reported series (Arnold et al, 2007; Berger et al, 2009; Carson et al, 2009). Nevertheless, given that many patients with chronic ITP are reluctant to undergo splenectomy and prefer rituximab as second-line therapy, the rate of splenectomy has decreased from 50–60% to 20–25% (Sailler, 2008). The recent International Consensus Report (Provan et al, 2010) and ASH guidelines (Neunert et al, 2011) have considered a tailored sequence of treatment. Consequently, it is important to define the place of new medicinal treatments earlier in the treatment of ITP.
The results of this meta-analysis are descriptive and not comparative. Relatively few rituximab treated non-splenectomized patients have been individually described in the literature to date. No randomized study has been designed to compare rituximab to splenectomy in chronic ITP patients. The four RCTs retained in this meta-analysis compare dexamethasone or other regimen to rituximab + dexamethasone (Hasan et al, 2009; Zaja et al, 2010) or different rituximab dosing schemes (Zwaginga et al, 2010; Li et al, 2011). Their design was different and it was difficult to compare them. In the literature, data often included splenectomized and non-splenectomized patients or primary and secondary ITP, resulting in confounding conclusions. Moreover, some reports have only been published in abstract form and did not provide sufficient details to be used in this meta-analysis or the data for non-splenectomized patients was often not extractable.
In our report, rituximab treatment in non-splenectomized chronic primary ITP patients resulted in an ORR of 57% and a CR of 41%. In all sub-analyses, the ORR was stable at around 57% and the CR at around 41%. As expected, CR was higher when defined by a platelet count >100 × 109/l. Response duration reached 49 weeks (95% CI: 17–60), time to response was 6·34 weeks (95% CI: 2·83–9·83) and the median follow-up 9 months was (95% CI: 2·3–65). Most patients received 4 weekly infusions of 375 mg/m2.
In all studies recorded for this meta-analysis, the response rate was defined according to the achievement of a platelet count >50 × 109/l for the ORR and >100 × 109/l or 150 × 109/l for the CR. The standardization of outcome criteria in adult ITP (Rodeghiero et al, 2009) was not applied in this meta-analysis because, many studies recorded in this meta-analysis were published before 2009 and the same CR criteria were not always applied. Furthermore, the lack of individual data in many studies prevented the calculation of a CR rate according to the new criteria. Moreover, certain reports used additional criteria, such as the discontinuation of steroids (Saleh et al, 2000; Zaja et al, 2003; Medeot et al, 2008), or specified the time to response measurement (Godeau et al, 2008; Audia et al, 2010).
Our results are consistent with the literature, in which only three meta-analyses evaluating the role of rituximab in ITP were found. All of these included both splenectomized and non-splenectomized patients. Zaja et al (2003) reported an ORR of 48% and a CR of 32%. Arnold et al (2007) reported an ORR of 62·5% and CR of 46·3% of CR, whereas the remaining meta-analysis showed a CR of 44% (Dolan et al, 2008).
Time to response (6·34 weeks) was only available in five studies, involving a small number of patients (n = 36). These results seemed to be longer than those previously reported in the literature (Zaja et al, 2003). One analysis reported a time to response of only 5·5 weeks among six studies (Arnold et al, 2007) and the earlier meta-analysis found that approximately 2 thirds of patients responded before 2 weeks (Zaja et al, 2003).
While our result has to be balanced (high heterogeneity and small sample size), duration response (49 weeks) is consistent with other studies including splenectomized patients (Zaja et al, 2003; Arnold et al, 2007), although the response duration interval was larger (4–117 weeks) in one study (Zaja et al, 2003). Estimation of duration of response in observational meta-analysis is not consensually defined and others authors used different methods.
A total of 13 prospective studies (including four RCTs) among 19 publications and use of individualized data for 10 studies gives strength to this meta-analysis. Formal meta-analytic methods can only use RCTs. Currently, there is a lack of RCTs regarding the use of rituximab in chronic ITP. We used rigorous methods to analyse and synthetize the evidence for rituximab efficacy in non-splenectomized patients. This meta-analysis used a large population that was not only selected by RCT. No study had a particular impact, as shown by the ‘one study removed’ Approach. Analysis of bias of publication revealed a lack of three studies for ORR and one study for ORR at 1 year. ORR was not modified and remained at around 57%. CR also remained stable at around 41% and the bias of publication was even lower (just two studies in analyses of randomized and prospective trials).
Funnel plot analysis helped to identify studies with precise effect size estimates (ORR) that were unexpectedly high or low. Two studies showed a high response rate. The first included treatment-naive patients (Li et al, 2011); the second was a repeat treatment of rituximab in a patient who previously responded (Hasan et al, 2009).
The heterogeneity in the design of the studies may be a limit to this study. The same rates of overall response and CR were found in the sub-analysis with the RCTs and prospective studies or with individual data (ORR 57%; CR 41% and 43%, respectively). The heterogeneity of the studies also concerns the CR definitions. However, the results for the CR150 sub-analysis (were the same as the global analysis. One of the strengths of this meta-analysis is the stability of the results. The heterogeneity (I 2) of the various analyses was taken into account by the use of a random model.
Patients were heterogeneous in the duration of ITP, age, sex and previous treatment. We could not explore previous treatment because data were not reported separately, but it was possible for the other parameters. Effect of age and sex was found in a univariate model of meta-regression. The effect of the patient's sex is more debatable: one study gives leverage (Alasfoor et al, 2009). However, a trend persisted when the regression was carried out without this study (P = 0·06 vs. P = 0·01) and sex was relevant for ORR in a multivariate meta-regression model. The ITP duration had no effect on ORR or CR as opposed to the literature (Cooper et al, 2004). But, this analysis only contained 36 patients and there could be a lack of substance. The age effect could be reflected by a shorter ITP duration; patients of a younger age may have a shorter ITP duration than older patients.
Recent studies should give better results than older studies and may be due to the use of rituximab in non-heavily pre-treated patients. Smaller studies tended to report greater response than larger studies probably due to the lack of representativeness of the small sample.
Our study enabled us to consider the results of rituximab treatment in primary adult non-splenectomized ITP patients. Rituximab was associated with an ORR of 57% and a CR rate of 41%. The heterogeneity of studies included could be balanced against the stability of results obtained in different sub-analysis.
Can this study change the clinical management of the ITP? We think so. We tried to highlight the interest of rituximab early during the history of ITP after steroids by emphasizing the population of non-splenectomized patients. We believe that this meta-analysis has generated clearer information regarding a large sample of about 300 ITP patients and which confirms the results of other meta-analyses, which also included splenectomized patients.
Because it is seems difficult to set up a RCT that could compare rituximab to splenectomy, we need meta-analysis to clearly identify the place of rituximab before splenectomy in management of chronic adult ITP. It is the reason why our results could modify the management of these patients and place rituximab as a second-line therapy.
Recent ASH recommendations (Neunert et al, 2011) placed rituximab therapy (grade 2C) after the splenectomy, and the International Consensus Report (Provan et al, 2010) placed rituximab second-in-line, at the same level as splenectomy. While thrombopoietin receptor agonists are recommended early in treatment, rituximab, a well-known effective drug that has been used in practice for several years, is still placed after splenectomy. This does not reflect actual ITP management, as shown by the recent decrease in the number of splenectomies performed (Sailler, 2008). It appears that the popularity of splenectomy has decreased dramatically in the past few years despite increasing reports of the high and durable rate of response with surgery (Sailler, 2008) and the few reported cases of PML with rituximab (Carson & Bennett, 2009).
The results of this analysis confirms, although lacking a license, the possibility and efficacy of the early use of rituximab in ITP. The aim of rituximab treatment is to obtain a long-term CR, which occurs in around 20% of patients (Cooper et al, 2007; Clausen et al, 2011). Even though it seems to be more profitable in young subjects, rituximab should probably be proposed in older patients for whom surgery is difficult. Durations of response are short (around 43 weeks) and some patients relapse. A second course of rituximab could be proposed, for which response rates seem to be the same (Hasan et al, 2009). But we think, at this time, splenectomy should be preferred.
The financial aspect of treatment remains the crucial point. A Spanish study reported the mean cost per treatment episode for ITP at 11 683 euros (Conde García et al, 2009). Although, we have to temperate use of rituximab according to local resources, we cannot only compare cost of splenectomy versus rituximab. Only a well-driven cost-efficiency study, including direct and indirect costs, with a temporal horizon covering all the consequences of both treatments, could identify optimization of financial resources. Furthermore, it would be necessary to include a study of quality-of-life study to consider the risks and constraints bound to splenectomy.
In the absence of RCTs, we have to base our treatment strategies on this type of analysis. Rituximab could be used early, before splenectomy, in chronic or persistent ITP, particularly in patients who are at medically high risk, or are unwilling to undergo splenectomy. Rituximab is used, at least in real life, to delay splenectomy, also knowing that splenectomy will be avoided in 20% of cases, i.e., for those patients who achieve a long-term response with rituximab.
In conclusion, rituximab has a place early in the tailored strategy of treatment of ITP, especially before splenectomy.