Rituximab for eosinophilic granulomatosis with polyangiitis: a systematic review of observational studies

Abstract

Objective

To analyse the available evidence about the use of rituximab (RTX) and other biologic agents in eosinophilic granulomatosis with polyangiitis (EGPA) patients and to provide useful findings to inform the design of future, reliable clinical trials.

Methods

A systematic review was performed. A systematic search was conducted in PubMed/MEDLINE, Scopus, Web of Science and the Cochrane library databases on RTX, and an extensive literature search was conducted on other biologic agents.

Results

Forty-five papers pertinent to our questions were found: 16 retrospective cohort studies, 8 case series, 3 prospective cohort studies and 18 single case reports, for a total of 368 EGPA patients. More than 80% of evaluable patients achieved complete or partial remission with a tendency towards a higher rate of complete response in the pANCA-positive subgroup.

Conclusion

Although the majority of the evaluable EGPA patients treated with RTX appears to achieve complete remission, we strongly believe that a number of sources of heterogeneity impair a clear interpretation of results and limit their transferability in clinical practice. Differences in design, enrolment criteria, outcome definition and measurement make a comparison among data obtained from studies on RTX and other biologic agents unreliable.

Introduction

Eosinophilic granulomatosis with polyangiitis (EGPA, formerly known as the Churg–Strauss syndrome) is a rare systemic necrotizing and eosinophil-rich vasculitis affecting small- to medium-sized vessels, characterized by asthma, sinusitis, pulmonary infiltrates and neuropathy [1–3].

The ACR established a six-item classification criteria [4] and additional criteria were added by the Chapel Hill consensus conference [5]. EGPA is currently classified among ANCA-associated vasculitis (AAV), sharing features with granulomatosis with polyangiitis (GPA, formerly Wegener’s granulomatosis) and microscopic polyangiitis (MPA) [6].

Due to the rarity of AAV and the inherent diagnostic difficulties in these complex diseases, clinical research is scarce, particularly for EGPA, whose prevalence (10–24/million) is 4 times lower than that of GPA [7] and incidence is 10 times lower than MPA [8]. Studies on the efficacy of treatments considering EGPA an individual entity are few, small-sized and underpowered [9]. Previous systematic reviews (SRs) [10–12] analysed evidence on the therapeutic use of biologic agents for all AAV and not specifically for EGPA. Rituximab (RTX) is approved for GPA and MPA, and its use has been proposed for EGPA [13]. Other biologics, such as mepolizumab (MPZ, monoclonal anti-IL-5 antibody) and omalizumab (OMZ, monoclonal anti-IgE antibody), the first biologic drug approved to treat EGPA, have been used in patients with EGPA, particularly those with uncontrolled or glucocorticoid (GC)-dependent asthma [14]. However, data on efficacy and safety of RTX use for EGPA come from uncontrolled studies and anecdotal reports, information on OMZ treatment remain limited and the effective risk–benefit balance on MPZ treatment remains unknown [14, 15]. Moreover, heterogeneity among studies, such as the disease’s definition and staging, activity and outcome assessment, treatment schedule, adverse events’ definitions and reporting, impairs extrapolation of results to practice [16].

Based on this background, we performed a SR about the use of RTX and an extensive literature search on other biologic agents in EGPA patients in order to: (i) critically summarize actual evidence about RTX efficacy and safety in EGPA; (ii) compare evidence about RTX with the data available on other biologics efficacy and safety; (iii) evaluate the weakness of available research, providing useful findings to make uniform the design of future clinical trials.

Methods

Protocol and registration

This study has been registered on PROSPERO (registration number 137629). Regarding to the SR, search strategy, clinical study selection as well as data extraction and analysis were performed and reported according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (supplementary Table S1, available at Rheumatology online). Institutional review board approval was not required.

Eligibility criteria

We included clinical studies (clinical trials, cohort observational studies, case series and case reports) reporting: (i) adult patients affected by EGPA; (ii) treatment with RTX, MPZ or OMZ, or other biologics; (iii) any clinical outcomes defined as clinical remission, relapse or mortality. Research questions were formulated according to the PICO format: Population, Intervention, Comparator, Outcomes. Observational studies reported in subsequent publications through years, but describing the same cohort of patients, have been detected and the most recent article has been selected and included in the systematic review, in order to avoid duplicates.

Information sources

A systematic search was conducted on RTX in PubMed/MEDLINE, Scopus, Web of Science and the Cochrane library databases up to the end of January 2019, through a comprehensive search strategy without language restriction, combining MeSH terms and free terms (supplementary Table S2, available at Rheumatology online). Reference lists of all pertinent retrieved clinical studies were also analysed through a manual search, in order to identify additional relevant papers. Moreover, conference abstracts were searched in the Scopus database and screened for pertinence.

Regarding other biologics, we conducted an extensive literature search on other biologic agents (last update August 2020), combining MeSH and free terms, and a manual search was carried out as well (supplementary Table S2, available at Rheumatology online).

Study selection and data extraction

Four blinded investigators (V.G.M., D.O., G.R. and A.A.) independently screened titles and abstracts to identify potentially relevant articles. Duplicate publications were actively searched and excluded. Full-texts of potentially pertinent articles were obtained and analysed by four independent investigators and data were extracted in a pre-designed structured form including patients’ characteristics, study design, outcomes and results. Any disagreement about a paper’s inclusion or data extraction was resolved by discussion with a fifth independent reviewer (G.P.).

In detail, data from each study were extracted as follows: first author name and year of publication, study design, patients’ characteristics (age, sex), sample size, diagnostic criteria used, baseline vasculitis activity score, ANCA status, disease extension (organ involvement) and refractoriness, RTX schedule of administration, concomitant treatments, median follow-up (months), complete and partial remission definitions and other evaluated outcomes, number of patients that achieved remission and steroid reduction (prednisone ≤7.5 mg/day), number of relapses, adverse events (AEs) and serious AEs (SAEs), infections, neoplasms and deaths reported.

Missing information or data were requested from corresponding authors, contacting them by e-mail. A total of 113 e-mails were sent, aiming to obtain more details about patients’ characteristics, diagnostic criteria used, outcome definitions and results, but only 13 corresponding authors answered.

Statistical analysis

In order to identify factors influencing the response to rituximab therapy, five different logistic regression models have been performed, using complete remission, partial remission, steroid reduction, death and relapse as binary dependent variables, respectively. The following variables have been evaluated: patients’ age, sex, concomitant treatments, RTX schedule of administration, ANCA status and peripheral nervous system involvement. Missing data have been replaced using multiple imputation procedure. Odds ratios (ORs) (together with their 95% CIs) were directly estimated from regression coefficients. A significance level of 0.05 was used for all the statistical tests. Multiple imputation has been executed by a dedicated spreadsheet and all statistical analyses have been performed through SPSS Statistics, version 23.0 (New York, USA).

Results

RTX

The extensive search of the literature followed by a careful manual screening of retrieved articles led to finding 45 papers pertinent to our questions (supplementary Tables S1 and S3, available at Rheumatology online). The search algorithm is detailed in Fig. 1. According to our aims, all retrieved studies have been included for data extraction regardless of their methodological quality.

In detail, we found 16 retrospective cohort studies [17–32] describing 296 EGPA patients, 8 case series [33–40] (48 patients), 3 prospective cohort studies [41–43] that enrolled 6 cases, and 18 single case reports [44–61], for a total of 368 EGPA patients. Supplementary Table S1, available at Rheumatology online shows characteristics and results of all studies. Of note, more than one-third of the studies (including about 80% of patients) do not provide specific critical information about subjects affected by EGPA (Table 1).

More than 80% of evaluable patients achieved complete or partial remission (Table 2). However, a significant percentage of studies reported results obtained from EGPA patients together with other AAV, preventing a separate analysis. Multivariate analysis, although strongly limited by the high rate of missing data, shows a trend to complete remission in pANCA-positive patients [OR 3.97 (95% CI 0.98, 16.01); P= 0.053]. However, combining partial and complete remission, no correlation was found [OR 0.667 (95% CI 0.037, 12.16); P= 0.78]. (Table 3). Moreover, studies enrolling mixed AAV populations show higher remission rates. The response rate does not seem to be influenced by type of diagnostic criteria, concomitant therapies, RTX schedules or extent of organ involvement.

About one-third of treated patients are evaluable for AEs. AEs occurred in 13 of 115 evaluable cases (11%). In particular, five pneumological complications and eight infusion reactions happened. Among mild AEs, we notice hypogammaglobulinemia (1%), transient visual disturbance (1%) and nausea (1%). In addition, we recorded infections separately from the AEs; we found reports of respiratory infections in 45 cases (17%), one herpes zoster infection (0.4%), one septic shock (0.4%), one invasive fungus infection (0.4%), two pyelonephritis (0.8%) and one cellulitis (0.4%). Finally, neoplasms were observed in 12 of 117 evaluable patients (10%): an astrocytoma and 11 urological tumors.

Other biologics used in EGPA patients

The search exclusively led to the identification of papers on MPZ and OMZ, whereas no results for anti-TNFα drugs were found. The articles were analysed by four blinded and independent investigators, leading to the extraction of the same set of data analysed for RTX.

Seven articles—one randomized controlled trial (RCT) [62], one post hoc analysis [63] of the same RCT, three uncontrolled prospective cohort studies [64–66] and two case reports [67, 68] describing results on 105 EGPA patients treated with MPZ—have been identified. The overall remission rate was 64% (63 out of 98 evaluable patients), whereas the reduction of steroid under the threshold of 7.5 mg/day prednisone was achieved in 60% of patients (62/103). The relapse rate was 40% (39/98). In particular, the MIRRA trial [62] led to MPZ approval for EGPA therapy (Food and Drug Administration and European Medicines Agency). In this trial 53% (36/68) of refractory or relapsed EGPA patients treated with MPZ vs 19% (13/68) in the placebo group achieved remission [OR 5.91 (95% CI 2.68, 13.03)]. Moreover, 44% (30/68) of the patients in the experimental group met the criteria for steroid reduction [OR 0.20 (95% CI 0.09, 0.41)]. The relapse rate was lower in the MPZ group [56 vs 80%; OR 0.32 (95% CI 0.21, 0.50)]. According to EULAR recommendations [69], remission defined as BVAS = 0 and prednisone ≤7.5 mg/day was adopted in most of the analysed studies [62, 63, 65, 66].

Regarding OMZ, 23 articles were reviewed. Duplicates (one article) were removed and 12 articles were excluded: one because EGPA diagnostic criteria were not met, one due to the absence of data regarding the response and nine articles because of EGPA onset under OMZ treatment. Ten articles (one uncontrolled prospective cohort study [70], two retrospective cohort studies [71, 72] and seven case reports [45, 73–78]) describing 48 EGPA subjects treated with OMZ were analysed. The overall remission rate was 56% (27 out of 48 valuable patients), whereas steroid reduction (≤7.5 mg/day prednisone) was achieved in 57% (27/47) of patients. The relapse rate was 38% (18/47). In the two larger retrospective cohorts [45, 72] clinical remission was reached in 56% and 35% of the patients, respectively. Both authors applied standardized remission criteria according to EULAR recommendations. The rate of adverse effects found in the literature was 13% (6/46 evaluable patients), with no SAEs. We did not find any reports of death. Data regarding the incidence of infections and neoplasia are missing.

Discussion

According to the EGPA Consensus Task Force [79], patients should initially be treated with GCs alone in limited disease [80] and with a combination of GCs and an immunosuppressant, mainly CYC, in severe forms, defined as the presence of a Five Factor Score ≥1 [80–82]. Remission can be achieved in >85% of patients after these first-line treatments, but it is noteworthy that 85% of them cannot stop GC treatment because of asthma and/or ENT manifestations [83, 84], with a high rate of side effects [85, 86]. Moreover, relapses occur in more than one-third of cases during GC tapering. Given these important limits, there is a need for additional, more effective and safer therapies. From the first description in 2001 [87] of the use of biologic agent in a patient with AAV, substantial progress has been made. For the last 10 years, RTX has been used in the treatment of AAV and many RCTs have been successfully conducted, mainly enrolling patients affected by GPA and MPA [88–91].

While hundreds of EGPA patients have also been treated, data on RTX efficacy in this population are scarce and of poor quality, and come only from case reports, small case series and retrospective cohort studies. Indeed, RTX use for EGPA patients was recently recommended by an expert consensus to treat patients with EGPA with renal involvement or refractory disease [17], but with a low grade of recommendation according to the European Vasculitis Society (EUVAS) survey participants [80, 92].

To our knowledge, this is the first SR of observational studies including only adults with EGPA. Ramos-Casals et al. [93] in 2008 published a SR about the use of biologic agents in adult patients with systemic autoimmune diseases and concluded that experience with RTX in EGPA was anecdotal (three cases reported), and no recommendations could be made. Muñoz et al. [94] performed a search about RTX in the treatment of EGPA in MEDLINE and LILACS until 2014 and included 27 patients, reporting clinical remission in 16 and clinical response in 8. In another SR about the role of RTX in treatment of some vasculitis [12], the authors suggested that, considering the reported general efficacy in AAV, RTX was likely effective in EGPA, but data were limited to three articles only. The recent SR by Ayan et al. [95] underlined many uncertainties on optimal use of RTX in AAV, but there was no specific comment about its use in EGPA patients.

Our SR pointed out several methodological issues. First of all, more than one-third of the studies (about 80% of patients), did not provide specific information about subjects affected by EGPA. Second, we noticed a wide heterogeneity in disease definition and stages, activity status, outcomes definitions and measures, schedules of administration, follow-up duration, adverse events definition and reporting, and use of concomitant drugs.

Regarding disease definition, we found that only in half of the retrieved studies (and in one-quarter of the total patients) were the criteria used for EGPA diagnosis reported. The 1990 ACR classification criteria [4] are the most popular for this disease, even if a formal validation has never been performed [96]. Other classifications, such Lanham criteria [97], the Chapel Hill consensus conference system [5] or the European Medicines Agency (EMA) algorithm [98], presented even more limits, and new classification criteria for EGPA using the Classification of Vasculitis Study (DCVAS) dataset [99] is waiting for final endorsement by EULAR and the ACR. Nevertheless, the ACR criteria for EGPA were used in only one-third of the evaluable patients identified by our SR.

In our analysis >80% of evaluable patients achieved complete or partial remission, but we strongly believe that the above-mentioned limits hamper the reliability of these data. In fact, a significant percentage of studies reported results obtained from EGPA patients together with other AAV, preventing a separate analysis. Moreover, studies enrolling mixed AAV populations reported higher remission rates, as shown by the multivariate analysis in which there is a trend of association between studies that enrolled patients from AAV cohort as covariate and complete remission, with an OR of 3.55 (95% CI 1.14, 11.09). Differences among criteria used to define outcomes further impede the interpretation and comparison of results. In fact, similar to the diagnostic criteria, only one-fifth of the studies reported data about complete or partial remission that met the ACR criteria. After the conclusion of this revision, data for about 147 EGPA patients coming from a retrospective European collaborative study were published [100]. Remission and partial response were reported in 49 and 24%, respectively, of the 63 EGPA patients treated with RTX; 17 patients (27%) experienced AEs. These data support the conclusions of our SR [100].

We found that >60% of patients were ANCA positive, a percentage clearly higher than observed in clinical practice, limiting transferability of the results. ANCA positivity was indeed used as inclusion criteria to enroll patients in most of the studies on the use of RTX in EGPA patients.

As in previous studies [100–102], our SR confirms, but without reaching the statistical significance threshold, a higher rate of complete remission in the pANCA-positive subgroup [OR 3.97 (95% CI 0.98, 16.01); P= 0.053]. It is noteworthy that Lyons et al. [103] performed a first genome-wide association study with 684 EGPA patients which suggesting that treatment strategies might be different between ANCA-positive and -negative EGPA patients.

A minority of studies in our SR reported AEs: only one-third of treated patients are evaluable for this important outcome, with an overall 11% of AEs, but more studies are needed to come to better conclusions about this topic.

Other biologics used in EGPA patients

MPZ and OMZ are the two other biologics tested in this condition. Available data show that MPZ is characterized by a better overall response and a greater GC-sparing effect than OMZ [15]. In particular, a recent report [100] confirmed a remission rate of 78 vs 15% in EGPA patients treated with MPZ and OMZ, respectively. Moreover, up to 22% of OMZ-treated patients experienced mild to moderate AEs, mainly asthenia. The role of OMZ as a steroid-sparing agent in EGPA patients with severe asthmatic manifestations is therefore questionable. Of interest, concerns about the risk of OMZ-treated asthmatic patients of developing EGPA have been raised [15].

Overall, the retrieved reports suffer from incomplete description of characteristics of the enrolled population (e.g. ANCA status), and heterogeneity in the experimental drugs schedule and in the definition and reporting of adverse effects. Finally, differences in design, enrolment criteria, outcome definition and measurement make a comparison among data obtained from studies on MPZ, OMZ and RTX unreliable. However, it is worth highlighting that the ongoing extension of the study by Wechsler and colleagues on MPZ [62] could help to shed more light on potential benefits in EGPA patients.

Limits

The main limitations of our SR are the heterogeneity of the included studies in terms of the characteristics of populations, concomitant immunosuppressive therapies, RTX protocols, outcome measures and remission definitions, and the high prevalence of missing data, which was only marginally corrected by contacting corresponding authors. In particular, multivariate analysis was strongly limited. Of the covariates stated, there is a very high percentage of missing data, such as 87% for age and ANCA status or 83% for RTX regimen. We performed a multiple imputation according to Rubin [104], considering that our case is reasonably an MAR (missing at random). Unfortunately, it was impossible to perform the same analysis on the database including only data from patients, because of the crossed missing data (i.e. there were no cases with complete data for all the considered variables). Moreover, the available studies are very small-sized. The low methodological quality prevents any attempt to verify the consistency of the results and to generate reliable summary measures of efficacy. Thus, the pooled response rate shown should be interpreted as the theoretical average response rate perceived by readers of literature about use of RTX in EGPA, rather than the true effect. Perhaps the two ongoing RCTs evaluating the efficacy of RTX for EGPA as remission induction therapy vs cyclophosphamide (CTX) [the REOVAS (Rituximab in Eosinophilic Granulomatosis With Polyangiitis) trial; NCT02807103] and as remission maintenance therapy vs AZA [the MAINRITSEG (Maintenance of Remission With Rituximab Versus Azathioprine for Newly-diagnosed or Relapsing Eosinophilic Granulomatosis With Polyangiitis) trial; NCT03164473], respectively, will shed more light on this important topic.

Conclusion

In conclusion, our SR identified major flaws in the available literature. Accordingly, our key recommendation for further research is to conduct prospective cohort studies, using validated criteria for disease and outcomes definition. Embedded trials, possibly RCTs, conducted with homogeneous therapeutic schedules, stratification for ANCA status, appropriate follow-up and a careful data reporting, could then definitely clarify the effectiveness of RTX, as well as of other biologics, in EGPA.

Acknowledgements

V.G.M., G.P. and A.G. conceived this study. V.G.M., G.R., D.O. and A.A. conducted data collection. G.R. was responsible for data management. V.G.M., M.R. and G.P. analysed all data. V.G.M. drafted the article, and all authors contributed substantially to its revision. V.G.M. takes responsibility for the paper as a whole. Our research is a systematic review. We declare no prior publications of this article. We have no financial support or funding sources for the work to disclose. We do not have any benefits from commercial sources to declare. We do not identify any situation that might be perceived as a potential conflict of interest or the appearance of a conflict of interest with regard to the work. We declare no copyright constraints.

Funding: No funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out this.

Disclosure statement: The authors declare no conflicts of interest.

Data availability statement

The data underlying this article will be shared on reasonable request to the corresponding author.

Supplementary data

Supplementary data are available at Rheumatology online.

References