• immune thrombocytopenia;
  • non-splenectomized;
  • rituximab


  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. 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.


  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Data sources and searches

Overall methods were adapted from the MOOSE (Meta-analysis Of Observational Studies in Epidemiology) and PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines for meta-analyses (Stroup et al, 2000; Moher et al, 2009). A literature search was conducted to identify RCTs and observational studies of rituximab treatment efficacy in adult non-splenectomized primary ITP. On May 31, 2011, similar electronic search strategies were performed using the National Library of Medicine PubMed database (from 1960), the Cochrane Central Register of Controlled Trials bases, the American Society of Hematology (ASH) Library and the International Society on Thrombosis and Haemostasis (ISTH) annual meetings from 2000 to 2011 to retrieve articles. The search criteria were Primary Autoimmune Thrombocytopenia [Medical Subject Heading, MeSH] or Primary Autoimmune Thrombocytopenic Purpura [MeSH] or Primary Idiopathic Thrombocytopenic Purpura [MeSH] or Primary Purpura, Thrombocytopenic or ITP and rituximab/rituxan with adult and human as limits (Table SI).

Study selection

Searches, study selection and data extraction were carried out independently by two haematologists; discrepancies were solved by discussion between authors until consensus.

Studies were included in the meta-analysis if they matched all pre-specified eligibility criteria.

Studies were excluded if they recruited only patients who (i) had secondary causes of thrombocytopenia, such as hepatitis B or C virus infection, bone marrow failure syndrome, drug induced thrombocytopenia, malignancies such as chronic lymphocytic leukaemia and non-Hodgkin lymphoma, Evan syndrome, (ii) were children <18 years old and (iii) had undergone splenectomy.

Studies that included a mix of splenectomized and/or non-splenectomized patients were included if the data of non-splenectomized adult ITP patients were extractable from articles. Finally, studies enrolling <5 patients were not included (risk of extreme bias). Searches were restricted to articles published in English or French. There was no restriction in study design and reports in abstract form were included. Where duplicate publication was found, the latest version was retained. For the first selection, the investigator read titles and abstracts of all articles. Then, the full-text of each retained article was read to determine the inclusion and exclusion criteria; the final selection included articles that contained relevant data.

Data extraction and quality assessment

From eligible studies, two haematologists reviewed and independently extracted data using piloted forms. Inter-study variables included information on study characteristics (author(s), publication year and journal, study design, follow-up duration), sample characteristics (country, type of ITP, duration of ITP from diagnosis to rituximab treatment, number of prior treatments, sample size and number of non-splenectomized patients), type of treatment (drug and administration route) and outcome definition. For each study, baseline demographics (mean age, gender), biological parameters (mean platelet count before and after Rituximab) and response rate (ORR, CR), time to response and duration of response variables were extracted if available. Individual data were used where possible. No classical tool (Newcastle-Ottawa Scale (NOS) or Jadad scale) was formally adapted for quality assessment of all of the included studies. Therefore, both investigators decided on a quality assessment score by combining items extracted from classical quality assessment scales (study design, randomization, and individual data). This quality assessment is detailed in Table SII. Study quality was reported, independently in a blinded fashion by, author, journal, year of publication, and results by both reviewers.

Data synthesis and analysis

Here we present the mean ORR and CR rate, together with their (95% confidence intervals, 95%CI) obtained for the 368 non-splenectomized chronic ITP patients treated with rituximab.

The majority of studies (80%) included were not randomized studies. Accordingly, we have no comparator arms.

We chose to record ORR and CR in primary assessment criteria. Definition of ORR was similar between the studies (platelet count >50 × 109/l) while the definition of CR criteria differed. Consequently, in seven studies, the CR was defined by a platelet count >150 × 109/l (CR150), while for the remaining studies the CR was defined by a platelet count >100 × 109/l (CR100). We chose to consider the CR rate reported in each study, according to author criteria, because the international criteria of CR (platelet count ≥100 × 109/l) were only published in 2009 (Rodeghiero et al, 2009) and were not applicable to all studies due to lack of individual data. We then reported the CR rate separately, according to definition (CR150 or CR100) in an analysis.

The criteria of secondary judgment were ORR and CR at 1 year, response time, mean platelet count at response and duration of response. For response time, mean platelet count at response and duration of response, we extracted individualized data from each study where available according to following criteria: adult non-splenectomized ITP patient. For platelet count or time to response, we noted or calculated the mean and standard deviation in each study by pooling the results of patients who corresponded to the criteria. The results of each study were then pooled in a random effects model meta-analysis. For duration of response, we used individualized patient data that corresponding to inclusion criteria, which were extracted from studies where available. Median of duration of response was estimated using the Kaplan Meier method, pooling extracted data of responding patients.

We determined estimates of rituximab effect by calculating the weighted median proportion. To account for expected heterogeneity, the random-effects model was used. Summary estimates of event rates were obtained using the fixed effects model method when heterogeneity was null. The percentage of variability beyond chance was estimated using the I 2 statistic (Higgins & Thompson, 2002; Higgins et al, 2003). An I 2 statistic with values over 50% (P value <0·1) may indicate substantial heterogeneity. To explain heterogeneity, the effects of inter-and intra-study covariates on study risk ratios were investigated using univariate and multivariate logarithmic mixed-effects meta-regressions. Qualitative inter-study covariates were coded into binary variables.

The risk of publication bias was assessed using the one-tailed Egger's test (Egger et al, 1997) and determined graphically by the funnel plot, which plots the natural log of study risk ratios versus their standard error (Egger et al, 1997). Additionally, Duval and Tweedie's trim-and-fill method was used (Duval & Tweedie, 2000). A P-value below 0·05 was considered statistically significant in all analysis. In sensitivity analyses, the impact of study selection was addressed by a ‘one-study removed’ meta-analysis approach.

Secondary analysis included estimation of ORR and CR with randomized (arm with rituximab) or prospective studies and with studies providing individualized data (Table SII). We also studied the CR in subgroup studies, considering CR if the platelet count was >150 × 109/l (according to the authors definition) (CR150).

All analysis were performed using comprehensive meta-analysis Version 2.2.048 (Biostat, Englewood, NJ, USA) and sas, version 9.2 (SAS Institute Inc., Cary, NC, USA).


  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Results of the search

Study selection

The flow diagram of the search results is presented in Fig 1. A total of 364 records were identified through electronic databases. Removing search overlap and irrelevant studies based on title and abstract review reduced the results to 33 records. After full-text analysis, 14 records were excluded because they were irrelevant studies (one article), duplicate publications (two articles), did not provide sufficient sizes of sample (<5 cases) (one article) or because data of non-splenectomized patients were not extractible or concerned non primary ITP (10 records). Published abstracts were included in the meta-analysis. Consequently, 19 studies were included in the meta-analysis; details on study design, populations, interventions and outcome are given in Table 1. One study (Hasan et al, 2009) comprises two parts: results of follow up on prior treatment of patients with rituximab and the retreatment of these patients.


Figure 1. Flow diagram of the trial selection process.

Download figure to PowerPoint

Table 1. Study characteristics
StudyArticle typeProspective designRandomizationNTime from diagnosis to treatment (months)Median age (years)Proportion of males (%)ORR (%)ORR at 1 year (%)CR (%)CR150 (%)CR at 1 year (%)Time to response (weeks)Platelet count at response (×109/l)Duration of response (weeks)
  1. (a) Hasan et al (2009), results for patients with second treatment of rituximab; (b) Hasan et al (2009), results for patients after the first treatment of rituximab; ORR, overall response rate; CR, complete response; CR150, complete response defined as platelet count >150 × 109/l.

Saleh et al (2000)Article1 6   16 00  148 
Stasi et al (2001)Article1 1719·2944·174141 29   197·1264·4
Stasi et al (2001)Article1 418392575 50  10·650237
Zaja et al (2003)Article1 1371·4657·071546 31  4·29 59·74
Cooper et al (2004)Article1 26   61·5 34·634·6    
Narat et al (2005)Letter  3 6133·3366·6 66·6  6·5283 
Peñalver et al (2006)Article  42   67 50     
Provan et al (2007)Article  722·635·2828·557·128·657·157·128·68289·75 
Schweizer et al (2007)Article  9 62·442256 33·33  2·4121·216
Garcia-Chavez et al (2007)Article1 362·6663·3333·333330 0864 
Godeau et al (2008)Article1 601·94823404035 354  
Medeot et al (2008)Article1 22   68·2       
Alasfoor et al (2009)Article1 78·2844·28861005786 57  52·61
Audia et al (2010)Article  15   33·3 13·3     
Hasan et al (2009) (a)Article11548·4482060 4040  143·6650·55
Hasan et al (2009) (b)Article1196552·1133·3377·7 44·444·4  125·339
Zaja et al (2010)Article114904945638553     
Badar et al (2010)Abstract  51·3414·872080 8080    
Zwaginga et al (2010)Abstract1135  3740 2020    
Li et al (2011)Article11310·252641·980·667·767·7 484   

Included studies were retrospective or prospective observational studies conducted in adults with chronic primary ITP receiving rituximab before splenectomy.

Study design and description

Across the 19 included studies, 368 non-splenectomized patients were recorded. Rituximab treatments were generally administered by intravenous injection at 375 mg/m2 weekly for 4 weeks. The dose was different (100 mg/m2) in two studies (Provan et al, 2007; Li et al, 2011), consisted of dose escalation (from 35 to 375 mg/m2) in one study (Saleh et al, 2000) and the schedule was different in three studies (1–4 cycles) (Peñalver et al, 2006; Audia et al, 2010; Zwaginga et al, 2010).

Overall study quality was largely reduced to only four randomized trials (Hasan et al, 2009; Zaja et al, 2010; Zwaginga et al, 2010; Li et al, 2011). Among the non-randomized studies, 13 were prospective with four Phase I/II trials (Saleh et al, 2000; Stasi et al, 2001; Cines & Blanchette, 2002; Zaja et al, 2003, 2010; Cooper et al, 2004; Garcia-Chavez et al, 2007; Godeau et al, 2008; Medeot et al, 2008; Alasfoor et al, 2009; Hasan et al, 2009; Zwaginga et al, 2010; Li et al, 2011). Other studies were retrospective (six articles and one abstract presented at anASH meeting) (Narat et al, 2005; Peñalver et al, 2006; Provan et al, 2007; Schweizer et al, 2007; Audia et al, 2010; Badar et al, 2010). Individual data were available in 10 studies (Stasi et al, 2001; Cines & Blanchette, 2002; Zaja et al, 2003; Narat et al, 2005; Garcia-Chavez et al, 2007; Provan et al, 2007; Schweizer et al, 2007; Alasfoor et al, 2009; Hasan et al, 2009; Badar et al, 2010). Five studies enrolled only non-splenectomized patients (Provan et al, 2007; Godeau et al, 2008; Badar et al, 2010; Zaja et al, 2010; Zwaginga et al, 2010).

Different strategies were used to check for possible confounding bias (Table SII). The studies were classified according to their design and the presence of individual data. Two sub-analysis were carried out: one of the RCTs and prospective studies and one with studies that provided individual data. The median follow-up was 9 months (2·3–65).

Effect size estimates

Overall response rate

The ORR was 57% (95% CI: 48–65) for 368 non-splenectomized patients after rituximab (Fig 2) and 57% (95% CI: 35–76) at 1 year for 157 patients (Fig 3). Heterogeneity was moderate in the first analysis (I 2 < 50%, P = 0·008) and high in the second (I 2 > 50%, P < 0·0001). In the analysis of RCTs and prospective studies (287 patients), ORR was stable at 56·6% (95% CI: 46–66) (Figure S1). Heterogeneity was high (I 2 > 50%, P < 0·01). In the same way, analysis of studies providing individual data (82 patients) showed an ORR of 57% (95% CI: 45–67). There was no heterogeneity (I 2 = 0%, P = 0·55).


Figure 2. Forrest plot of ORR after Rituximab treatment in non-splenectomized patients. Heterogeneity is estimated with I2 statistic according to Higgins et al (2003) and Higgins and Thompson (2002). Black diamond, global event rate; horizontal line, 95% confidence interval of event rate in study, vertical line, mean of event rate in study. (A) Hasan et al (2009), results for patients with second treatment of rituximab; (B) Hasan et al (2009), results for patients after the first treatment of rituximab.

Download figure to PowerPoint


Figure 3. Forrest plot of ORR at 1 year after Rituximab treatment in non-splenectomized patients. Heterogeneity is estimated with I2 statistic according to Higgins et al (2003) and Higgins and Thompson (2002). Black diamond, global event rate; horizontal line, 95% confidence interval of event rate in study, vertical line, mean of event rate in study.

Download figure to PowerPoint

Complete response rate

CR was 41·5% (95% CI: 33–50) for 346 non-splenectomized patients (Fig 4) and 40% (95% CI: 31–49) at 1 year for 108 patients (Fig 5). CR150 was 36% (95% CI: 24–59) (153 patients), and CR100 was 45·6% (95% CI: 34·5–57) (193 patients) (Fig 4). Heterogeneity was moderate (I 2 around 50%, P ≈ 0·01) for overall CR and CR100 but lower for CR150. There was little inter- and intra-group heterogeneity. Analysis of RCTs and prospective studies (265 patients) showed a 40% CR (95% CI: 30–51) (Figure S2) and heterogeneity was high (I 2 > 50%, P = 0·007). In the analysis of studies providing individual data (82 patients), the CR was 43% (95% CI: 31–54) with no heterogeneity (I 2 = 7·7%, P = 0·37).


Figure 4. Forrest plot of CR after Rituximab treatment in non-splenectomized patients according to definition of CR. Heterogeneity is estimated with I2 statistic according to Higgins et al (2003) and Higgins and Thompson (2002). Figure shows CR defined by platelet counts >100 × 109/l or >150 × 109/l, and overall CR. Black diamond, global event rate (area is proportional to study weight in the meta-analysis); horizontal line, 95% confidence interval of event rate in study, vertical line, mean of event rate in study.

Download figure to PowerPoint


Figure 5. Forrest plot of CR at 1 year after Rituximab treatment in non-splenectomized patients. Heterogeneity is estimated with I2 statistic according to Higgins et al (2003) and Higgins and Thompson (2002). Black diamond, global event rate; horizontal line, 95% confidence interval of event rate in study, vertical line, mean of event rate in study.

Download figure to PowerPoint

Duration, time to response and platelet count after treatment

Mean of time to response was 6·34 (2·83–9·85) weeks for 36 patients (Figure S3). Mean of platelet count increased up to 200 (129–271) ×109/l at response times for 54 patients (Figure S4). In all analysis, heterogeneity was high (I 2 > 50%, P < 0·01). Median of duration of response was 49 (17–60) weeks for 36 responding patients (Figure S5).

Investigating heterogeneity

Differences in patient characteristics

In univariate mixed-effect meta-regressions, ORR and CR were significantly associated with differences in average age (both P < 0·01) and sex (both P < 0·01) and for CR with difference in year of publication (P = 0·05). The association between age and ORR or CR is depicted in Fig 6. Best response rates were obtained in young people, males and in recent publications. ORR was marginally associated with a difference in the year of publication (P = 0·08) and with the number of patients included (P = 0·09). Other covariates – mean duration of ITP – were not significantly different (P = 0·11 with CR and P = 0·7 with ORR) (Table 2). For CR, the multivariate model included age, sex and year of publication. Only age was relevant (P = 0·03): young people had a better response after rituximab treatment. For ORR, age, sex, year of publication, number of patients and duration of ITP were included in the multivariate model. Only the first four were relevant (respectively P = 0·02, 0·005, 0·06, 0·02). Therefore, young, male patient, recent publication and small size study had better results (Table 3).


Figure 6. Measured (circle) and predicted (line) ORR and CR depending on the difference in baseline age by univariate mixed-effect meta-regression. (A) Effect of Age on ORR; (B) Effect of Age on CR.

Download figure to PowerPoint

Table 2. Univariate associations of inter-study or intra-study covariates and ORR and CR estimated from mixed-effects meta-regressions
CovariateORR P-valueCR P-value
  1. ITP, immune thrombocytopenia; ORR, overall response rate; CR, complete response.

Inter-study variables
Sample size0·090·88
Follow-up duration0·250·74
Intra-study variables
Duration of ITP0·700·11
Table 3. Multivariate associations of inter-study or intra-study covariates and ORR and CR estimated from mixed-effects meta-regressions
CovariateORR P-valueCR P-value
  1. ORR, overall response rate; CR, complete response.

Inter-study variables
Year of publication0·06 
Sample size0·02 
Intra-study variables
Trial and intervention quality

We could not identify any associations between trial quality and ORR or CR using univariate mixed-effect meta-regressions. In particular, using a prospective or retrospective design and abstract or article did not impact on the response rate. The impact of each study was equivalent on ORR and CR; a ‘one study removed’ approach did not modify the results for these parameters (Figure S6).

Publication bias

Egger's test showed no evidence of publication bias of studies on ORR (P = 0·21), ORR at 1 year (P = 0·5), CR (P = 0·37), CR at 1 year (P = 0·39), CR150 (P = 0·32) and CR in randomized or prospective trials (P = 0·31). This was consistent with the shape of funnel plots showing good symmetry. There were three missing studies for ORR showing a lower event rate (Fig 7), one for ORR at 1 year. Adding the virtual missing study from the trim and fill method only tended to find a lower ORR to rituximab treatment (53%; 95% CI: 44–62), greater ORR at 1 year (61%; 95% CI: 40–79) and remained unchanged for CR. Concerning CR, the funnel plot shape showed good symmetry. There was one missing study for CR at 1 year inclining CR at 1 year to 40% (95% CI: 31–50) by the trim and fill method. In the analysis Of RCTs and prospective studies for CR, with two virtual missing studies (with greater event rate), CR was 42% (95% CI: 34–53).


Figure 7. Funnel plot of natural logarithms of ORR and CR (dots) and resulting random-effects meta-analysis estimate (diamond). Results are based on the 20 included studies for ORR (one study had two results) and 19 for CR (one study had two results) (open circles) and with the addition of one virtual study identified by ‘trim and fill’ analysis (black circles) (Egger et al, 1997; Duval & Tweedie, 2000). (A) Funnel plot of ORR; (B) Funnel plot of CR.

Download figure to PowerPoint


  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. 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.


  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
  • Alasfoor, K. , Alrasheed, M. , Alsayegh, F. & Mousa, S.A. (2009) Rituximab in the treatment of idiopathic thrombocytopenic purpura (ITP). Annals of Hematology, 88, 239243.
  • Arnold, D.M. , Dentali, F. , Crowther, M.A. , Meyer, R.M. , Cook, R.J. , Sigouin, C. , Fraser, G.A. , Lim, W. & Kelton, J.G. (2007) Systematic review: efficacy and safety of rituximab for adults with idiopathic thrombocytopenic purpura. Annals of Internal Medicine, 146, 2533.
  • Audia, S. , Lakomy, D. , Guy, J. , Leguy-Seguin, V. , Berthier, S. , Aho, S. , Lorcerie, B. & Bonnotte, B. (2010) Treatment of immune thrombocytopenia: a retrospective study of 40 patients. La Revue De Médecine Interne, 31, 337344.
  • Badar, T. , Ziauddin, S. , Albeirouti, B. , Abdelhameed, A. & Al-Humaidi, A. (2010) Efficacy of early use of rituximab in steroid refractory severe immune thrombocytopenic purpura in adult patients: single institution experience from Saudi Arabia. Blood (ASH Annual Meeting Abstracts), 116, 4681.
  • Berger, J.R. , Houff, S.A. & Major, E.O. (2009) Monoclonal antibodies and progressive multifocal leukoencephalopathy. mAbs, 1, 583589.
  • Carson, K.R. & Bennett, C.L. (2009) Rituximab and progressive multi-focal leukoencephalopathy: the jury is deliberating. Leukemia & Lymphoma, 50, 323324.
  • Carson, K.R. , Evens, A.M. , Richey, E.A. , Habermann, T.M. , Focosi, D. , Seymour, J.F. , Laubach, J. , Bawn, S.D. , Gordon, L.I. , Winter, J.N. , Furman, R.R. , Vose, J.M. , Zelenetz, A.D. , Mamtani, R. , Raisch, D.W. , Dorshimer, G.W. , Rosen, S.T. , Muro, K. , Gottardi-Littell, N.R. , Talley, R.L. , Sartor, O. , Green, D. , Major, E.O. & Bennett, C.L. (2009) Progressive multifocal leukoencephalopathy after rituximab therapy in HIV-negative patients: a report of 57 cases from the Research on Adverse Drug Events and Reports project. Blood, 113, 48344840.
  • Cines, D.B. & Blanchette, V.S. (2002) Immune thrombocytopenic purpura. The New England Journal of Medicine, 346, 9951008.
  • Clausen, M.R. , Segel, E. , Brandsborg, M. & d'Amore, F. (2011) Very long-term remission induced by short-term rituximab monotherapy in a patient with heavily pretreated, chronic immune thrombocytopenic purpura. European Journal of Haematology, 86, 256259.
  • Conde García, M.C. , Fernández Feijoo, M.A. & Calleja Hernández, M.A. (2009) Study of rituximab efficacy, cost, safety, and compliance of its package leaflet in a tertiary hospital]. Farmacia Hospitalaria: Órgano Oficial De Expresión Científica De La Sociedad Española De Farmacia Hospitalaria, 33, 305311.
  • Cooper, N. , Stasi, R. , Cunningham-Rundles, S. , Feuerstein, M.A. , Leonard, J.P. , Amadori, S. & Bussel, J.B. (2004) The efficacy and safety of B-cell depletion with anti-CD20 monoclonal antibody in adults with chronic immune thrombocytopenic purpura. British Journal of Haematology, 125, 232239.
  • Cooper, N. , Evangelista, M.L. , Amadori, S. & Stasi, R. (2007) Should rituximab be used before or after splenectomy in patients with immune thrombocytopenic purpura? Current Opinion in Hematology, 14, 642646.
  • Dolan, J.P. , Sheppard, B.C. & DeLoughery, T.G. (2008) Splenectomy for immune thrombocytopenic purpura: surgery for the 21st century. American Journal of Hematology, 83, 9396.
  • Duval, S. & Tweedie, R. (2000) Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56, 455463.
  • Egger, M. , Davey Smith, G. , Schneider, M. & Minder, C. (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ (Clinical Research Ed.), 315, 629634.
  • Garcia-Chavez, J. , Majluf-Cruz, A. , Montiel-Cervantes, L. , Esparza, M.G. , Vela-Ojeda, J. & Mexican Hematology Study Group (2007) Rituximab therapy for chronic and refractory immune thrombocytopenic purpura: a long-term follow-up analysis. Annals of Hematology, 86, 871877.
  • Godeau, B. , Porcher, R. , Fain, O. , Lefrère, F. , Fenaux, P. , Cheze, S. , Vekhoff, A. , Chauveheid, M.P. , Stirnemann, J. , Galicier, L. , Bourgeois, E. , Haiat, S. , Varet, B. , Leporrier, M. , Papo, T. , Khellaf, M. , Michel, M. & Bierling, P. (2008) Rituximab efficacy and safety in adult splenectomy candidates with chronic immune thrombocytopenic purpura: results of a prospective multicenter phase 2 study. Blood, 112, 9991004.
  • Hasan, A. , Michel, M. , Patel, V. , Stasi, R. , Cunningham-Rundles, S. , Leonard, J.P. & Bussel, J. (2009) Repeated courses of rituximab in chronic ITP: three different regimens. American Journal of Hematology, 84, 661665.
  • Higgins, J.P.T. & Thompson, S.G. (2002) Quantifying heterogeneity in a meta-analysis. Statistics in Medicine, 21, 15391558.
  • Higgins, J.P. , Thompson, S.G. , Deeks, J.J. & Altman, D.G. (2003) Measuring inconsistency in meta-analyses. BMJ (Clinical Research Ed.), 327, 557560.
  • Kojouri, K. , Vesely, S.K. , Terrell, D.R. & George, J.N. (2004) Splenectomy for adult patients with idiopathic thrombocytopenic purpura: a systematic review to assess long-term platelet count responses, prediction of response, and surgical complications. Blood, 104, 26232634.
  • Li, Z. , Mou, W. , Lu, G. , Cao, J. , He, X. , Pan, X. & Xu, K. (2011) Low-dose rituximab combined with short-term glucocorticoids up-regulates Treg cell levels in patients with immune thrombocytopenia. International Journal of Hematology, 93, 9198.
  • Mazzucconi, M.G. , Arista, M.C. , Peraino, M. , Chistolini, A. , Felici, C. , Francavilla, V. , Macale, E. , Conti, L. & Gandolfo, G.M. (1999) Long-term follow-up of autoimmune thrombocytopenic purpura (ATP) patients submitted to splenectomy. European Journal of Haematology, 62, 219222.
  • Medeot, M. , Zaja, F. , Vianelli, N. , Battista, M. , Baccarini, M. , Patriarca, F. , Soldano, F. , Isola, M. , De Luca, S. & Fanin, R. (2008) Rituximab therapy in adult patients with relapsed or refractory immune thrombocytopenic purpura: long-term follow-up results. European Journal of Haematology, 81, 165169.
  • Moher, D. , Liberati, A. , Tetzlaff, J. , Altman, D.G. & PRISMA Group (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Journal of Clinical Epidemiology, 62, 10061012.
  • Narat, S. , Gandla, J. , Hoffbrand, A.V. , Hughes, R.G. & Mehta, A.B. (2005) Rituximab in the treatment of refractory autoimmune cytopenias in adults. Haematologica, 90, 12731274.
  • Neunert, C. , Lim, W. , Crowther, M. , Cohen, A. , Solberg, Jr, L. , Crowther, M.A. & American Society of Hematology (2011) The American Society of Hematology 2011 evidence-based practice guideline for immune thrombocytopenia. Blood, 117, 41904207.
  • Peñalver, F.J. , Jiménez-Yuste, V. , Almagro, M. , Alvarez-Larrán, A. , Rodríguez, L. , Casado, M. , Gallur, L. , Giraldo, P. , Hernández, R. , Menor, D. , Rodríguez, M.J. , Caballero, D. , González, R. , Mayans, J. , Millán, I. , Cabrera, J.R. & Multi-institutional Retrospective Spanish Study Group on the Use of Rituximab in Refractory ITP (2006) Rituximab in the management of chronic immune thrombocytopenic purpura: an effective and safe therapeutic alternative in refractory patients. Annals of Hematology, 85, 400406.
  • Provan, D. , Butler, T. , Evangelista, M.L. , Amadori, S. , Newland, A.C. & Stasi, R. (2007) Activity and safety profile of low-dose rituximab for the treatment of autoimmune cytopenias in adults. Haematologica, 92, 16951698.
  • Provan, D. , Stasi, R. , Newland, A.C. , Blanchette, V.S. , Bolton-Maggs, P. , Bussel, J.B. , Chong, B.H. , Cines, D.B. , Gernsheimer, T.B. , Godeau, B. , Grainger, J. , Greer, I. , Hunt, B.J. , Imbach, P.A. , Lyons, G. , McMillan, R. , Rodeghiero, F. , Sanz, M.A. , Tarantino, M. , Watson, S. , Young, J. & Kuter, D.J. (2010) International consensus report on the investigation and management of primary immune thrombocytopenia. Blood, 115, 168186.
  • Robak, T. (2004) Monoclonal antibodies in the treatment of autoimmune cytopenias. European Journal of Haematology, 72, 7988.
  • Rodeghiero, F. , Stasi, R. , Gernsheimer, T. , Michel, M. , Provan, D. , Arnold, D.M. , Bussel, J.B. , Cines, D.B. , Chong, B.H. , Cooper, N. , Godeau, B. , Lechner, K. , Mazzucconi, M.G. , McMillan, R. , Sanz, M.A. , Imbach, P. , Blanchette, V. , Kühne, T. , Ruggeri, M. & George, J.N. (2009) Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura of adults and children: report from an international working group. Blood, 113, 23862393.
  • Sailler, L. (2008) Rituximab off label use for difficult-to-treat auto-immune diseases: reappraisal of benefits and risks. Clinical Reviews in Allergy & Immunology, 34, 103110.
  • Saleh, M.N. , Gutheil, J. , Moore, M. , Bunch, P.W. , Butler, J. , Kunkel, L. , Grillo-López, A.J. & LoBuglio, A.F. (2000) A pilot study of the anti-CD20 monoclonal antibody rituximab in patients with refractory immune thrombocytopenia. Seminars in Oncology, 27 (Suppl 12), 99103.
  • Schwartz, J. , Leber, M.D. , Gillis, S. , Giunta, A. , Eldor, A. & Bussel, J.B. (2003) Long term follow-up after splenectomy performed for immune thrombocytopenic purpura (ITP). American Journal of Hematology, 72, 9498.
  • Schweizer, C. , Reu, F.J. , Ho, A.D. & Hensel, M. (2007) Low rate of long-lasting remissions after successful treatment of immune thrombocytopenic purpura with rituximab. Annals of Hematology, 86, 711717.
  • Stasi, R. , Pagano, A. , Stipa, E. & Amadori, S. (2001) Rituximab chimeric anti-CD20 monoclonal antibody treatment for adults with chronic idiopathic thrombocytopenic purpura. Blood, 98, 952957.
  • Stasi, R. , Stipa, E. , Forte, V. , Meo, P. & Amadori, S. (2002) Variable patterns of response to rituximab treatment in adults with chronic idiopathic thrombocytopenic purpura. Blood, 99, 38723873.
  • Stasi, R. , Cooper, N. , Del Poeta, G. , Stipa, E. , Evangelista, M.L. , Abruzzese, E. & Amadori, S. (2008) Analysis of regulatory T-cell changes in patients with idiopathic thrombocytopenic purpura receiving B cell–depleting therapy with rituximab. Blood, 112, 11471150.
  • Stasi, R. , Newland, A. , Thornton, P. & Pabinger, I. (2010) Should medical treatment options be exhausted before splenectomy is performed in adult ITP patients? A debate Annals of Hematology, 89, 11851195.
  • Stroup, D.F. , Berlin, J.A. , Morton, S.C. , Olkin, I. , Williamson, G.D. , Rennie, D. , Moher, D. , Becker, B.J. , Sipe, T.A. & Thacker, S.B. (2000) Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA: The Journal of the American Medical Association, 283, 20082012.
  • Vianelli, N. , Galli, M. , de Vivo, A. , Intermesoli, T. , Giannini, B. , Mazzuconi, M. , Barbui, T. , Tura, S. & Baccaranion, M. (2005) Efficacy and safety of splenectomy in immune thrombocytopenic purpura: long-term results of 402 cases. Haematologica, 90, 7277.
  • Zaja, F. , Vianelli, N. , Sperotto, A. , De Vita, S. , Iacona, I. , Zaccaria, A. , Masolini, P. , Tomadini, V. , Tani, M. , Molinari, A.L. , Baccarani, M. & Fanin, R. (2003) B-cell compartment as the selective target for the treatment of immune thrombocytopenias. Haematologica, 88, 538546.
  • Zaja, F. , Baccarani, M. , Mazza, P. , Bocchia, M. , Gugliotta, L. , Zaccaria, A. , Vianelli, N. , Defina, M. , Tieghi, A. , Amadori, S. , Campagna, S. , Ferrara, F. , Angelucci, E. , Usala, E. , Cantoni, S. , Visani, G. , Fornaro, A. , Rizzi, R. , De Stefano, V. , Casulli, F. , Battista, M.L. , Isola, M. , Soldano, F. , Gamba, E. & Fanin, R. (2010) Dexamethasone plus rituximab yields higher sustained response rates than dexamethasone monotherapy in adults with primary immune thrombocytopenia. Blood, 115, 27552762.
  • Zwaginga, J. , Van der Holt, R. , Biemond, B. , te Boekhorst, P. , Levin, M. , Vreughdenhil, A. , Huijgens, P.C. , Brand, A. , van der Griend, R. , Luten, M. , Pruijt, H. , de Weerdt, O. , van Pampus, E. , Zweegman, S. , Hollestein, R. & Koene, H. (2010) Interim analysis on a Dutch HOVON multicenter randomised open label phase II trial on 3 rituximab dosing schemes in chronic ITP patients. Blood (ASH) Annual Meeting Abstracts, 116, 2514.

Supporting Information

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information
bjh9169-sup-0001-FigS1.pdfapplication/PDF164KFig S1. Forrest plot of ORR after Rituximab treatment in non-splenectomized patients according randomized trials and prospective studies.
bjh9169-sup-0002-FigS2.pdfapplication/PDF156KFig S2. Forrest plot of CR after Rituximab treatment in non-splenectomized patients according to randomized trials and prospective studies.
bjh9169-sup-0003-FigS3.pdfapplication/PDF80KFig S3. Forrest plot of time to response (in weeks) after Rituximab treatment in non-splenectomized patients.
bjh9169-sup-0004-FigS4.pdfapplication/PDF97KFig S4. Forrest plot of platelet count (G/L) after Rituximab treatment in non-splenectomized patients.
bjh9169-sup-0005-FigS5.pdfapplication/PDF173KFig S5. Kaplan Meier of duration of response (weeks) after Rituximab treatment in non-splenectomized patients.
bjh9169-sup-0006-FigS6a.pdfapplication/PDF5KFig S6. One study removed for ORR (a) and CR (b).
bjh9169-sup-0008-TableS1-S2.docxWord document24KTable SI. Article request strategies used. Table SII. Strategies used to control confounding in studies.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.