Article Text
Abstract
Objective To evaluate whether the classification of patients with anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) according to ANCA type (anti-proteinase 3 (PR3) or anti-myeloperoxidase (MPO) antibodies) predicts treatment response.
Methods Treatment responses were assessed among patients enrolled in the Rituximab in ANCA-associated Vasculitis trial according to both AAV diagnosis (granulomatosis with polyangiitis (GPA)/microscopic polyangiitis (MPA)) and ANCA type (PR3-AAV/MPO-AAV). Complete remission (CR) was defined as disease activity score of 0 and successful completion of the prednisone taper.
Results PR3-AAV patients treated with rituximab (RTX) achieved CR at 6 months more frequently than did those randomised to cyclophosphamide (CYC)/azathioprine (AZA) (65% vs 48%; p=0.04). The OR for CR at 6 months among PR3-AAV patients treated with RTX as opposed to CYC/AZA was 2.11 (95% CI 1.04 to 4.30) in analyses adjusted for age, sex and new-onset versus relapsing disease at baseline. PR3-AAV patients with relapsing disease achieved CR more often following RTX treatment at 6 months (OR 3.57; 95% CI 1.43 to 8.93), 12 months (OR 4.32; 95% CI 1.53 to 12.15) and 18 months (OR 3.06; 95% CI 1.05 to 8.97). No association between treatment and CR was observed in the MPO-AAV patient subset or in groups divided according to AAV diagnosis.
Conclusions Patients with PR3-AAV respond better to RTX than to CYC/AZA. An ANCA type-based classification may guide immunosuppression in AAV.
Trial registration number NCT00104299; post-results.
- Systemic vasculitis
- Treatment
- Cyclophosphamide
- Granulomatosis with polyangiitis
Statistics from Altmetric.com
Introduction
Rituximab (RTX) and cyclophosphamide (CYC) are considered standard of care alternatives for induction of remission in patients with granulomatosis with polyangiitis (GPA) and microscopic polyangiitis (MPA).1–4 A subgroup analysis of the Rituximab in anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) trial (RAVE) demonstrated that RTX was superior to CYC in subjects with relapsing disease. However, no other subset of patients has been recognised to derive clear benefit from one specific immunosuppressive strategy as opposed to the other.1–3 ,5–7 The identification of biomarkers that improve the individualisation of treatment decisions would represent an important contribution to the care of patients with AAV.
We evaluated two different classification systems for AAV in a well-characterised group of patients: one based on the patient's specific AAV diagnosis (GPA vs MPA), and the other on the patient's specific ANCA type (proteinase 3 (PR3)-AAV vs myeloperoxidase (MPO)-AAV). The objective of this study was to determine whether the classification of patients with AAV solely based on their ANCA type predicts treatment response and other clinical outcomes.
Methods
Patient groups and treatment regimens
The RAVE study was a double-blind, placebo-controlled trial that randomised 197 patients to receive either RTX (375 mg/m2 intravenously weekly for 4 weeks; RTX group, n=99) or CYC (2 mg/kg for 3–6 months) followed by maintenance azathioprine (AZA; 2 mg/kg, up to 150 mg/day (CYC/AZA group, n=98)). Both groups received a 5.5-month prednisone taper and were followed for 18 months. Details of the trial design and the main study results have been published.2 ,5 ,8
Definition of predictor and outcome variables
We classified patients by their specific AAV diagnoses (GPA or MPA), rendered during the conduct of the trial. These diagnoses were based upon the 1994 Chapel Hill Consensus on vasculitis nomenclature.9 We also classified patients according to whether they were PR3- or MPO-ANCA positive. We refer to the subsets generated by the classification as PR3-AAV and MPO-AAV, respectively. Within each classification system, we compared patients’ demographics, baseline clinical manifestations and response to treatment stratified by treatment assignment (RTX vs CYC/AZA). In addition, we analysed the number and severity of disease flares, time to flare, cumulative glucocorticoid use, disease damage and the percentage of patients who became ANCA-negative at 6 months.
Vasculitis activity was measured using the Birmingham Vasculitis Activity Score for Wegener's Granulomatosis (BVAS/WG).10 Patients with BVAS/WG ≥1 were considered to have active disease. Complete remission (CR) was defined as a BVAS/WG of 0 and successful prednisone taper completion. Severe flares were defined as a BVAS/WG ≥3. Flares not meeting criteria for severe were classified as non-severe. Damage was graded according to the Vasculitis Damage Index (VDI).11
Statistical analysis
Continuous and categorical variables were compared between pairs of groups (GPA vs MPA and PR3-AAV vs MPO-AAV) using the Wilcoxon rank-sum test and the χ2 and Fisher's exact tests, respectively. Multivariable logistic regression was used to account for potential confounders in the analysis of selected dichotomous outcomes. Results were expressed as OR and 95% CI.
To determine the effect of RTX versus CYC/AZA on the outcome of CR at 6, 12 and 18 months, we created logistic regression models within each classification category (GPA, MPA, PR3-AAV and MPO-AAV). The models included age, sex and the status of newly diagnosed versus relapsing disease at baseline as covariates. For each classification category, we completed subgroup analyses limited to patients with newly diagnosed and relapsed disease at baseline. VDI scores were compared over time between pairs of groups (GPA vs MPA, PR3-AAV vs MPO-AAV) using repeated measures mixed-effect models. The cut-off for statistical significance was 0.05; all p Values were two-sided. SAS V.9.2 and R V.3.0.0 were used for the statistical analyses.
Results
The RAVE trial enrolled 197 patients. When divided by AAV diagnosis, 148 patients were classified as GPA (86% anti-PR3 positive) and 48 as MPA (92% anti-MPO positive) (table 1). One subject could not be categorised clearly as either GPA or MPA. When divided by ANCA type, 131 patients were classified as PR3-AAV (97% GPA) and 66 as MPO-AAV (70% MPA) (table 1). Twenty patients with GPA (14%) had anti-MPO antibodies and four with MPA (8%) had anti-PR3 antibodies.
Demographics and baseline disease characteristics
Considerable overlap existed between both classification systems in terms of demographic data and baseline clinical manifestations (see table 1 and online supplementary text). Patients classified as GPA and PR3-AAV were more likely to have relapsing disease at baseline compared with those classified as MPA and MPO-AAV, respectively. The mean BVAS/WG at study onset was equivalent across all subsets.
Response to treatment
CR at 6 months was achieved in 115 subjects (RTX n=63; CYC/AZA n=52). Patients with PR3-AAV met this outcome significantly more often when treated with RTX as opposed to CYC/AZA (65% vs 48%; p=0.04) (table 2). Multivariate analysis adjusting for differences in age, sex and the status of new-onset versus relapsing disease at baseline yielded an OR for CR at 6 months of 2.11 (95% CI 1.04 to 4.30; p=0.04), favouring RTX over CYC/AZA (table 3). In this model, the OR for CR at 6 months associated with the status of newly diagnosed versus relapsing disease was 1.74 (95% CI 0.83 to 3.66; p=0.14). In contrast, the proportion of patients with MPO-AAV in CR at 6 months did not differ according to induction regimen (RTX 61% and CYC/AZA 64%; p=0.80) (table 2). The non-differential response to treatment among patients with MPO-AAV persisted in multivariate-adjusted analyses (data not shown).
The percentage of patients achieving CR at 6 months was equivalent between treatment arms for both patients with GPA (RTX 63% and CYC/AZA 50%; p=0.11) and those with MPA (67% RTX and 63% CYC/AZA; p=0.76) (table 2). Neither treatment was favoured among patients with GPA or MPA in multivariate-adjusted analyses (data not shown).
Eighty-one patients (62%) with PR3-AAV had relapsing disease at baseline. Of those, 42 received RTX and 39 received CYC/AZA. Univariate analyses showed that CR at 6, 12 and 18 months was achieved significantly more often in patients assigned to RTX: 6 months, RTX 67% versus CYC/AZA 36% (p<0.01); 12 months, RTX 48% versus CYC/AZA 18% (p<0.01); and 18 months, RTX 36% versus CYC/AZA 15% (p=0.04). Multivariate analyses demonstrated that the better response to RTX in this subgroup persisted following adjustments for differences in age and sex (table 3). The ORs (95% CI) for attaining CR at 6, 12 and 18 months were 3.57 (1.43 to 8.93; p<0.01), 4.32 (1.53 to 12.15; p<0.01) and 3.06 (1.05 to 8.97; p=0.04), respectively.
When divided by AAV diagnosis, 91 patients (61%) classified as GPA had relapsing disease at baseline. Of those, 47 received RTX and 44 received CYC/AZA. Univariate analyses showed that CR was achieved significantly more often among patients assigned to RTX at 6 months (RTX 66% vs CYC/AZA 39% (p<0.01)) and 12 months (RTX 47% vs CYC/AZA 23% (p=0.01)), but not at 18 months (RTX 38% vs CYC/AZA 21% (p=0.06)). In multivariate analyses, the ORs (95% CI) comparing RTX to CYC/AZA for CR at 6, 12 and 18 months among GPA patients with relapsing disease at baseline were 3.11 (1.32 to 7.35; p=0.01), 3.11 (1.23 to 7.86; p=0.02) and 2.48 (0.96 to 6.39; p=0.06), respectively.
Univariate and multivariate analysis of CR at 6, 12 and 18 months demonstrated equivalent response to RTX and CYC/AZA treatment in all other subgroups studied (ie, newly diagnosed PR3-AAV, newly diagnosed MPO-AAV, relapsed MPO-AAV, newly diagnosed GPA, newly diagnosed MPA and relapsed MPA) (data not shown).
Other outcomes
In general, there were no significant differences between both classification systems in regards to disease relapse, damage and cumulative glucocorticoid use (see online supplementary text, supplementary tables 1 and 2). Unlike patients with MPO-AAV, patients with PR3-AAV became ANCA-negative at 6 months significantly more often when treated with RTX as opposed to CYC/AZA (see online supplementary text).
Discussion
Analysis of these prospective clinical trial data demonstrates that patients with PR3-AAV were more than twice as likely to achieve CR at 6 months if treated with RTX rather than with CYC/AZA. In addition, among patients with PR3-AAV who had relapsing disease at baseline, the risk of disease flare in RTX-treated patients was lower not only at 6 months, but also at 12 and 18 months, despite the fact that patients randomised to RTX were not re-treated pre-emptively with a maintenance regimen.
Prior analyses of the RAVE cohort identified the history of disease relapse as a predictor of better response to RTX over CYC. However, AAV relapses tend to occur months after diagnosis and therapy initiation. In addition, both disease relapse and the burden of prior immunosuppression are factors associated with adverse events including infection, disease damage and mortality.12–14 Therefore, previous disease relapse is suboptimal as a biomarker to guide treatment decisions. A readily available objective measure such as ANCA testing overcomes the limitation of having to wait until patients declare themselves as relapsers, offering clinicians the opportunity to make rational treatment decisions early during the course of the disease. Additional studies are required to determine whether the institution of this approach leads to better long-term outcomes in AAV.
Our results not only have implications for remission-induction, but also for remission-maintenance. Following remission-induction with CYC, RTX dosing at fixed intervals has been shown in one study to be superior to AZA for prevention of disease relapses.3 However, the risk:benefit ratio of pre-emptive B-cell depletion for different patient subpopulations is currently uncertain. Therefore, knowing that patients with PR3-AAV constitute a subgroup with a high risk of relapse but a higher likelihood of response to RTX should inform both current clinical practice and the design of future clinical trials.15–17
Our study has potential limitations. First, although still generalisable to the majority of patients with AAV,18–20 extrapolation of our findings to ANCA-negative patients is not possible. Second, our data derive from a post hoc clinical trial analysis, and therefore, should be confirmed in additional prospective studies. Finally, the relatively small sample size might have led to low statistical power to detect treatment differences in other subgroups. Larger studies, especially of patients with MPO-AAV, are needed.
In conclusion, our analysis reinforces and expands the notion that an ANCA-based classification provides valuable predictive information in AAV. Patients with PR3-AAV are more likely to achieve and maintain CR if treated with RTX as opposed to CYC/AZA. This observation may directly influence the choice of remission induction therapy for this AAV subset.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
- Data supplement 2 - Online table 1
- Data supplement 3 - Online table 2
Footnotes
Handling editor Tore K Kvien
Contributors All authors meet the criteria for authorship and more specifically for contributors statement. JHS and SU take responsibility for this work. All other authors critically reviewed the article and agreed to publish this work. Conception and design of this work were discussed with all authors.
Funding This research was performed as a project of the Immune Tolerance Network (NIH Contract N01-AI-15416; Protocol number ITN021AI), an international clinical research consortium headquartered at the University of California San Francisco and supported by the National Institute of Allergy and Infectious Diseases and the Juvenile Diabetes Research Foundation; Genentech and Biogen-Idec. At the Mayo Clinic, the trial was supported by Clinical and Translational Science Award (CTSA) Grant Number 1 UL1 RR024150-01 (National Center for Research Resources; NCRR). At Johns Hopkins, the trial was supported by UL1 RR 025005 (NCRR) and by grants K24 AR049185 (JHS) and K23 AR052820 (PS). At Boston University, the trial was supported by CTSA grant number UL1RR 025771, NIH M01 RR00533, K24 AR02224 (PAM), and an Arthritis Foundation Investigator Award (PAM). At Cleveland Clinic, the trial was supported by UL1TR000439, National Center for Advancing Translational Sciences (NCATS), NIH. ANCA ELISA kits were provided by EUROIMMUN AG (Lübeck, Germany).
Competing interests None declared
Provenance and peer review Not commissioned; externally peer reviewed.