Article Text

Download PDFPDF

Extended report
Randomised controlled trial of prolonged treatment in the remission phase of ANCA-associated vasculitis
  1. Alexandre Karras1,2,
  2. Christian Pagnoux3,
  3. Marion Haubitz4,
  4. Kirsten de Groot5,
  5. Xavier Puechal6,
  6. Jan Willem Cohen Tervaert7,
  7. Mårten Segelmark8,
  8. Loic Guillevin2,6,
  9. David Jayne9
  10. On behalf of the European Vasculitis Society
  1. 1Department of Nephrology, AP-HP, Hôpital Européen Georges Pompidou, Paris, France
  2. 2Université Paris Descartes, Paris, France
  3. 3Vasculitis Clinic, Division of Rheumatology, Mount Sinai Hospital, Toronto, Ontario, Canada
  4. 4Department of Nephrology and Hypertension, Center for Internal Medicine and Medical Clinic III, Klinikum Fulda, Fulda, Germany
  5. 5IIIrd Medical Department, Klinikum Offenbach, Offenbach, Germany
  6. 6Department of Internal Medicine, National Referral Center for Rare Autoimmune and Systemic Diseases, AP-HP, Hôpital Cochin, Paris, France
  7. 7Department of Immunology, Maastricht University, Maastricht, The Netherlands
  8. 8Department of Medical and Health Sciences and Department of Nephrology, Linköping University, Linköping, Sweden
  9. 9Department of Medicine, University of Cambridge, Cambridge, UK
  1. Correspondence to Pr Alexandre Karras, Department of Nephrology, AP-HP, Hôpital Européen Georges Pompidou, Paris 75015, France; alexandre.karras{at}


Objectives A prospective randomised trial to compare two different durations of maintenance immunosuppressive therapy for the prevention of relapse in anti-neutrophil cytoplasmic antibodies (ANCA)-associated vasculitis (AAV).

Methods Patients with AAV were recruited 18–24 months after diagnosis if they were in stable remission after cyclophosphamide/prednisolone-based induction followed by azathioprine/prednisolone maintenance therapy. They were randomised (1:1) to receive continued azathioprine/prednisolone to 48 months from diagnosis (continuation group) or to withdraw azathioprine/prednisolone by 24 months (withdrawal group). The primary endpoint was the relapse risk, from randomisation to 48 months from diagnosis.

Results One hundred and seventeen patients were randomised and 110 remained to the trial end. At entry, median serum creatinine was 116 μmol/L (range 58–372), 53% were ANCA positive. The percentage of patients presenting with relapse was higher in the withdrawal than in the continuation treatment group (63% vs 22%, p<0.0001, OR 5.96, 95% CI 2.58 to 13.77). ANCA positivity at randomisation was associated with relapse risk (51% vs 29%, p=0.017, OR 2.57, 95% CI 1.16 to 5.68). Renal function, ANCA specificity, vasculitis type and age were not predictive of relapse. Severe adverse events were more frequent in the continuation than withdrawal groups (nine vs three events), but the continuation group had better renal outcome (0 vs 4 cases of end-stage renal disease), with no difference in patient survival.

Conclusions Prolonged remission maintenance therapy with azathioprine/prednisolone, beyond 24 months after diagnosis reduces relapse risk out to 48 months and improves renal survival in AAV.

Trial registration number ISRCTN13739474

  • ANCA
  • maintenance therapy
  • relapse
  • azathioprine
  • vasculitis

Statistics from


Anti-neutrophil cytoplasmic antibodies (ANCA)-associated vasculitis (AAV) are a group of autoimmune systemic diseases that are associated with a necrotising, pauci-immune, vasculitis of small blood vessels and the presence of circulating autoantibodies to myeloperoxidase (MPO-ANCA) or proteinase 3 (PR3-ANCA). The major subgroups of AAV are microscopic polyangiitis (MPA), granulomatosis with polyangiitis (GPA, Wegener’s) and eosinophilic GPA.1 Renal involvement is manifested by a necrotising, crescentic glomerulonephritis and results in end-stage renal disease (ESRD) in up to 20% of patients.2

Therapeutic management of MPA and GPA is divided into induction and maintenance phases. Remission induction is achieved in most cases with cyclophosphamide or rituximab in combination with high-dose glucocorticoids and sometimes plasma exchanges, while remission maintenance regimens have employed an oral immunosuppressive, such as azathioprine or methotrexate, or repeat-dose rituximab, with or without low-dose glucocorticoids.3 4 The optimal duration of remission maintenance therapy remains unknown, with current consensus recommendations suggesting at least 24 months, once remission has been obtained.5

Relapse occurs in 30%–50% of patients by 5 years and has been associated with a diagnosis of GPA, PR3-ANCA specificity, the presence of ear nose and throat involvement, persisting ANCA positivity after induction therapy, a lower serum creatinine at diagnosis and withdrawal of glucocorticoids or immunosuppressives.6 The consequences of relapse are additional accrual of disease and treatment-related damage and morbidity, and renal relapse is associated with an increased risk of ESRD.7 Continuing remission therapy increases exposure to the toxicities of immunosuppressives and glucocorticoids.

This study tested whether continued azathioprine/prednisolone was more effective in preventing relapse than their withdrawal at 24 months from diagnosis in patients with AAV.


Study design and patients

We hypothesised that prolonged maintenance therapy with low-dose prednisolone and azathioprine reduces the frequency of relapse, when compared with withdrawal of immunosuppression 2 years after diagnosis. This trial (REMAIN, prolonged REmission-MAINtenance therapy in systemic vasculitis) was conducted by the European Vasculitis Society (EUVAS) and recruited patients from 33 centres in 11 European countries. The study was approved by the local ethics committee in each participating centre, and all patients provided written informed consent, according to the Declaration of Helsinki.

Criteria for inclusion were (1) a diagnosis of MPA, GPA or renal-limited vasculitis; (2) renal involvement and/or other threatened loss of function of a vital organ (lung, brain, eye, motor nerve or gut) and ANCA positivity, and ANCA-negative patients were eligible for enrolment in the study only when there was histological confirmation of pauci-immune vasculitis; (3) remission-induction therapy with cyclophosphamide and prednisolone for at least 3 months, with or without plasma exchanges; and (4) stable remission on azathioprine/prednisolone. They were recruited and randomised 18–24 months from commencement of therapy.

Exclusion criteria were age under 18 years, pregnancy, previous malignancy, known HIV infection, previous life-threatening relapse, ESRD at inclusion and allergy to study medications. Patients not in stable remission for at least 6 months at 18 months after commencement of therapy and patients who had discontinued azathioprine and/or prednisolone were excluded from the study.

Disease definitions

Diagnostic definitions were initially based on the 1994 Chapel Hill Consensus Conference on the Nomenclature of Systemic Vasculitides8 and re-evaluated after completion of the study, according to the recently revised criteria.1 Remission was defined as a 1994 Birmingham Vasculitis Activity Score (BVAS) of 0, indicating the absence of new or worse disease activity, with persistent disease activity for no more than one item.9 Major relapse was defined by the recurrence or first appearance of at least 1 of the 24 items of the BVAS, which are indicative of threatened function of a vital organ (kidney, lung, brain, eye, motor nerve or gut) attributable to active vasculitis. Minor relapse was defined by the recurrence or first appearance of at least three other BVAS items. Determinations of remission and relapse were made by the investigator and validated retrospectively by an independent observer.

Drug regimens

Patients were randomly assigned at entry, 18–24 months after initiation of immunosuppression, in a 1:1 ratio, to continuation or withdrawal treatment groups (table 1).

Table 1

Drug doses according to treatment arm


Study assessments were performed at entry, then every 3 months until 30 months from entry for non-relapsing patients and until the time of relapse for relapsing patients. The assessments included BVAS,9 serum creatinine, C-reactive protein, ANCA positivity (requiring both positivity by indirect immunofluorescence and by PR3-ANCA or MPO-ANCA assay, performed in local laboratories), drug doses and adverse events. Glomerular filtration rate (eGFR) was estimated using the modification of diet in renal disease (MDRD) formula.10 All cause damage since the vasculitis diagnosis was assessed by the Vasculitis Damage Index (VDI)11 at entry then every 6 months to the trial end.


The primary endpoint was the percentage of patients presenting a relapse of vasculitis, including major and minor relapses, during the study period. Secondary endpoints were incidence of major and minor relapses, mortality, adverse events of therapy, rise in cumulative damage score (VDI), deterioration of eGFR, incidence of ESRD and ANCA status during follow-up.

Statistical analysis

Randomisation was performed centrally. Treatment allocation was done by block randomisation (permuted blocks of four) per country. Primary data were collected locally in record books and subsequently submitted for centralised validation and analysis.

Based on a one-tailed design, with a significance level of 5% and a power of 0.8, the inclusion of 116 patients was required to demonstrate a 20% lower relapse rate in the maintenance therapy group during the study period (assuming a threefold increase of relapse risk in the discontinuation arm).

The demographic characteristics of the two groups were compared with the use of Student’s t-test or a Wilcoxon rank-sum test for continuous measures and a χ2 test or Fisher’s exact test for categorical variables. The effect of treatment on time to relapse was examined by Kaplan-Meier analysis, with the use of the log-rank test. The two groups were compared in terms of the secondary endpoints reached between the time of remission and the end of the study. The rates of adverse events were compared with the use of two-by-two tables and Fisher’s exact test. The values of eGFR and VDI were compared with the use of the Wilcoxon rank-sum test.



Between September 1998 and March 2010, 121 patients were enrolled. Four were excluded due to ineligibility, because of early relapse (n=3) or malignancy (n=1). One hundred and seventeen patients were randomised with a mean period of 18.8±1.8 months after initiation of induction therapy: 61 to the continuation and 56 to the withdrawal groups. During the study, three patients withdrew (patient’s choice (n=2) or physician’s decision (n=1)) and four patients were lost to follow-up (figure 1). Complete data for analysis were available for 110 patients.

Figure 1

Flow diagram of the REMAIN study, summarising enrolment, intervention allocation, follow-up and data analysis.

Demographic characteristics of the study population are detailed in table 2. Fifty-two patients (47%) had GPA and 58 (53%) had MPA. ANCA specificity was PR3 in 52% of cases and MPO in 44%, whereas ANCA were negative or without specificity in 4%. Almost all patients (96%) were enrolled in this study after the first remission of newly diagnosed AAV. The median follow-up was 925 days (IQR 878–970), after randomisation.

Table 2

Demographics of randomised patients according to treatment arm, 18–24 months after diagnosis

Protocol treatment

At randomisation, the mean daily azathioprine dose was 99±37 mg and the mean daily prednisolone dose was 5.9±2.2 mg. Immunosuppression was rapidly tapered according to the study protocol in the withdrawal group, whereas it was continued until the end of study in the continuation group. Median and mean daily doses of azathioprine and prednisolone are detailed in supplementary figures 1a and 1b.

Efficacy assessment

Primary endpoint

Thirty-two patients (62.7%) in the withdrawal group experienced a relapse as compared with 13 (22.0%) in the continuation group (log rank test p<0.0001) (figure 2a). The patients in the withdrawal group had a 2.84-fold higher relative risk of relapse (95% CI 1.72 to 4.9) when compared with patients continuing immunosuppression. Interestingly, 78% of relapses in the withdrawal group occurred after removal of azathioprine versus 8% in the continuation group. The median daily azathioprine dose at relapse was 75 mg (IQR 50–100) mg in the continuation group versus 0 mg (IQR 0–0) in the withdrawal group.

Figure 2

Kaplan-Meier analysis of the study population showing relapse-free survival according to treatment group (2a), major relapse-free survival according to treatment group (2b) and relapse-free survival according to positivity or negativity of ANCA at randomisation (2c). Daily dose of prednisolone and azathioprine in each group is shown above. C, continuation subgroup; W, withdrawal subgroup.

Of note, primary endpoint difference between treatment groups remained highly significant (p<0.001) if we excluded patients of the withdrawal group that continued small doses of azathioprine beyond month 6 (n=5) or if we included patients that withdrew their consent or were lost to follow-up (n=7).

Secondary endpoints

A major relapse occurred in 18 patients (35.3%) of the withdrawal and in eight patients (13.5%) of the continuation groups (p=0.007) (figure 2b). The eGFR at last follow-up was 52.5±26.7 mL/min/1.73 m2 in the withdrawal and 54.1±24.7 mL/min/1.73 m2 in the continuation groups (p=0.78). Nevertheless, the ΔeGFR between randomisation and end of study in the withdrawal group was −3.3±14.9 mL/min/1.73 m2, whereas it was +2.5±9.8 mL/min/1.73 m2 in the continuation group (p=0.01). This difference was not found when the analysis was restricted to patients that had completed follow-up without relapse (+0.8±15.4 vs +2.6±8.0 mL/min/1.73 m2, p=0.53). Four patients (7.8%) of the withdrawal group developed ESRD during follow-up as compared with none in the continuation group (p=0.012). Of note, median eGFR at randomisation, for these four patients, was 29.5 mL/min/1.73 m2 (range 15–59). There were two deaths (3.9%) in the withdrawal and five (8.5%) in the continuation groups (p=0.32). Causes of death were cancer in three cases, cardiovascular disease in 2 and undetermined for two patients.

Although there was no difference in the percentage of ANCA-positive patients at randomisation (51% in the continuation vs 56% in the withdrawal groups), patients in the withdrawal group had more frequent reappearance of ANCA after randomisation. By month 6, 72% of patients in the withdrawal group were ANCA positive compared with 52% in the continuation group (p=0.04) (see online supplementary figure 2a). There was no difference in the final VDI score between the two groups (see online supplementary figure 2b).

Supplementary Material

Supplementary figure 2

Predictors of relapse

Univariate analysis revealed that withdrawal of immunosuppression (p<0.0001) and ANCA positivity at randomisation (p=0.017) were the only predictors of relapse during follow-up (figure 2c and table 3). Multivariate analysis confirmed that these factors were independently associated with risk of relapse. ANCA specificity at diagnosis (PR3 vs MPO), disease phenotype (GPA vs MPA), age or renal function at randomisation were not predictive of relapse in this study. Similar results were obtained when studying the risk of major relapse during follow-up.

Table 3

Risk factors associated with AAV relapse

Kaplan-Meier analysis shows that the difference between survival according to treatment arm persists across different subgroups of patients, such as patients with MPO-ANCA or PR3-ANCA specificity (see online supplementary figure 3a), as well as in patients with or without ANCA positivity at randomisation (see online supplementary figure 3b).

Supplementary Material

Supplementary figure 3

Severity of relapse

Characteristics of the relapses occurring during the study period are detailed in supplementary table 1. Relapse severity was not different between treatment arms. Mean BVAS at relapse was 7.1±4.1 in the withdrawal group and 8.7±4.4 in the continuation group (p=0.29). Organ involvement at relapse was not different between the two groups, except eye, nose, throat (ENT) involvement, which was more frequent in the continuation group (69% vs 34%, p=0.048).

Supplementary Material

Supplementary Table 1

Adverse events

Seventy-one adverse events were reported in 46 patients (table 4). Severe or life-threatening adverse events occurred in 13 patients (12%). There was no statistical difference between the two groups in the prevalence or severity of adverse events. The most frequent events were infections, occurring in 13 (22%) patients of the continuation and 10 (19%) of the withdrawal group. Haematological disorders and cardiovascular events were more frequent among patients of the continuation group.

Table 4

Adverse events (AEs)


ANCA-associated vasculitis carries a substantial and often unpredictable risk of relapse, and prolonged relapse prevention therapy is recommended that may itself contribute to organ damage, morbidity and patient mortality.7 We have shown that continuation of treatment with azathioprine and prednisolone beyond 24 months from diagnosis was more effective at preventing relapse than withdrawal of these agents for AAV patients with GPA/MPA. These results confirm the ability of the azathioprine/prednisolone combination to influence relapse risk and suggest that treatment should be continued for at least 48 months from diagnosis, especially in those with persistent ANCA positivity after induction therapy.

Despite the use of azathioprine/prednisolone-based remission maintenance therapy, the risk of relapse after induction of remission with cyclophosphamide was 38% at 5 years in the recent meta-analysis of previous EUVAS trials.6 This risk remains high when rituximab is given for induction of remission instead of cyclophosphamide, with a 32% risk of relapse at month 18.12 Previous relapse prevention studies in AAV have demonstrated an equivalence of methotrexate to azathioprine, a lower efficacy with mycophenolate mofetil, and no efficacy of etanercept.13–15 In view of the high proportion of patients with renal disease in this study, azathioprine was selected over methotrexate. Repeat-dose rituximab has recently been shown to be superior at preventing relapse than azathioprine,16 but this drug was not is use in AAV at the time this trial was designed.

The optimal duration of immunosuppressive therapy after induction of remission is unknown. Only one previous study has compared different regimens, in patients with persistent anti-PR3-ANCA positivity at remission. Sanders et al17 randomised 45 patients to receive either standard (1 year after diagnosis and subsequent tapering) or extended (4 years after diagnosis and tapered thereafter) azathioprine maintenance therapy. Although 46% of patients relapsed in the standard therapy group versus 24% in the extended therapy group, the difference was not statistically different.

This study clearly demonstrates that continuation of glucocorticoids and azathioprine beyond 2 years is associated with a threefold reduction of relapse risk. Moreover, extension of immunosuppression is associated with a better renal survival, as illustrated by the fact that all patients that reached ESRD during follow-up had discontinued azathioprine a few months before. Of note, previous studies have shown that every renal relapse is associated with an eGFR decrease of 8–12 mL/min18 19 and that those patients are 4.7 times more likely to progress to ESRD.20

The overall frequency of relapse seen in this study, 41% at 48 months, was similar to that reported in other AAV studies (46% in the IMPROVE trial,14 35% in the WEGENT trial,13 45% in the azathioprine arm of the MAINRITSAN trial).16 In the present study, relapse risk (62.7%) in the control group was higher than usually reported, but in previous trials the follow-up period after cessation of immunosuppression was significantly shorter. Interestingly, there was good compliance of clinicians to the study drug regimen in the withdrawal group and even a trend to underdosing in the continuation group that may have reduced the magnitude of the treatment effect of azathioprine/prednisolone. The relative contributions of azathioprine or prednisolone to the treatment effect are not known, although an earlier systematic review has highlighted the increase in relapse risk that follows glucocorticoid withdrawal.6

In our study, continuation of immunosuppressive therapy was associated with more frequent adverse events such as malignancies or infections. The safety profile was not statistically different between the two groups, but our study was underpowered to detect significant differences in adverse events between groups. Importantly, patient survival as well as VDI score, which reflects the cumulative organ damage due to vasculitis and/or treatment, was similar. Of note, the final VDI score in the whole study population was 2.2, similar to that described (2.66) in the long-term follow-up of the EUVAS trials.7 Despite small numbers, we found that cytopaenias and cardiovascular complications were more common in the continuation group, possibly in relation with the known toxicity of azathioprine and the metabolic effects of glucocorticoids.

The main question raised by this study is whether we should recommend an extended duration of immunosuppression for all patients with AAV after achievement of sustained remission. Numerous studies have demonstrated that the risk of relapse is more important among patients with GPA and/or anti-PR3,6 21 suggesting that this subgroup of patients should receive prolonged immunosuppressive therapy. Our study was underpowered to confirm this hypothesis (see online supplementary figure 3a). Nevertheless, our data show that persistent ANCA positivity, 2 years after initiation of immunosuppression predicts a higher risk of relapse, suggest that this subgroup of patient may require a different immunosuppressive regimen. Other studies have observed a reduced risk of relapse was associated with negativity of ANCA at the time of switching to maintenance therapy.22 23 However, in our study, relapse occurred in 29% of patients with negative ANCAs at randomisation, when assigned to azathioprine discontinuation (supplementary figure 3b), revealing that even those patients have an important risk of relapse when stopping the immunosuppressive drugs too early. Of note, 83% of relapsing patients with ANCA negativity at inclusion had positive ANCA testing at relapse.

Our study has several limitations. First, this trial was open label and the absence of placebo might have led to overestimation of the relapse rate in patients having discontinued azathioprine and corticosteroids. BVAS is a semiobjective tool, although many relapses were renal assessed by objective criteria. Nevertheless, prognosis was still different between the two groups, even when robust criteria were used to define severe flare of disease or ESRD. Second, this study was designed and conducted before widespread use of rituximab, and it is difficult to extrapolate these results to patients receiving B-cell depleting agents as induction and/or maintenance therapy. Further studies are determining if, similarly to what we have shown with azathioprine, an 18-month duration is less effective than a 48-month maintenance therapy with rituximab, for prevention of AAV relapse (MAINRITSAN 3three trial, NCT02433522). Third, the study design may have induced some bias in the study population, excluding patients with intolerance to azathioprine and patients with more severe disease, such as patients with life-threatening vasculitis or patients who experienced early relapse, during the initial 18 months azathioprine therapy preceding randomisation. We did not have detailed information from diagnosis to precisely describe organ distribution such as ENT or cardiovascular involvement, baseline renal function, cyclophosphamide exposure (total dose, intravenous/oral administration), factors known to influence relapse risk. Fourth, the 12 years’ duration of study enrolment could have influenced the results. Nevertheless, the relapse rate was stable throughout the study period, and patients who entered the study before 2006 had an overall relapse rate of 40% versus 42% for those who were enrolled after this date. Finally, we were not able to calculate and compare cumulative glucocorticoid dose in each group, as treatment data were not collected for all patients, after occurrence of relapse. This point is important as it has been shown that part of the long-term metabolic and cardiovascular toxicity of immunosuppression is due to excessive cumulative doses of glucocorticoids, given either as a preventive remission maintenance therapy or as curative induction treatment during repeated flares of AAV.24

In conclusion, we suggest that at least some of the patients that have reached remission of AAV require long-term immunosuppressive therapy to prevent recurrence of the disease. The challenge of future studies will be to define the best immunosuppressive scheme, providing both efficacy and limited toxicity and to find clinical or biological markers that will identify high-risk patients who will require prolonged therapy.

Supplementary Material

Supplementary figure 1a

Supplementary Material

Supplementary figure 1b

Supplementary Material

Supplementary figure 4


We are grateful to all the physicians who participated in this trial:

A Bruchfeld, Karolinska, Sweden; F Chantrel, Colmar, France; J Dadoniene, Vilnius, Lithuania; A Ekstrand, Helsinki, Finland; P Eriksson, Linkoping, Sweden; M Essig, Limoges, France; A Fernstrom, Karolinska, Sweden; J Floege, Aachen, Germany: P Gobert, Avignon, France; P Godmer, Vannes, France; C Hanrotel-Saliou, Brest, France; L Harper, Brimingham, UK; T Hauser, Basel, Switzerland; O Lidove, Paris, France; F Maurier, Metz, France; A Mertens, Aachen, Germany; E Mirapeix, Barcelona, Spain; J Odum, Wolverhampton, UK; T Quemeneur, Valenciennes, France; L F Quintana, Barcelona, Spain; H D Rupprecht, Erlangen, Germany; I Rychlik, Prague, Czech Republic; D Selga, Lund, Sweden; G Sterner, Malmö, Sweden; N Tieule, Nice, France; P Vanhille, Valenciennes, France; P van Paassen, Maastricht, The Netherlands; K Verburg, Leiden, The Netherlands; S Weidner, Munich, Germany; K Westman, Lund, Sweden.



  • Contributors AK contributed to data collection, data analysis and interpretation,manuscript preparation and review. CP contributed to data generation, collection,analysis and interpretation, manuscript preparation and review. MH contributed to study design and set-up, data generation, analysis and interpretation,manuscript preparation and review. KdG contributed to data generation, analysis and interpretation, manuscript preparation and review. XP contributed to data generation, analysis and interpretation, manuscript preparation and review. JWCT contributed to data generation, analysis and interpretation, manuscript preparation and review. MS contributed to study design and set-up, data generation, analysis and interpretation, manuscript preparation and review. LG contributed to data generation, analysis and interpretation, manuscript preparation and review. DJ contributed to study design and set-up, data generation and collection, analysis and interpretation, manuscript preparation and review.

  • Competing interests AK has received lecture fees from Roche/Genentech. DJ has received research grants and lecture fees from Roche/Genentech. CP has received research grants and lecture fees from Roche/Genentech and advisory board fees from ChemoCentryx and Sanofi.

  • Patient consent Patient.

  • Ethics approval NHS Executive North West MREC, Gateway House, Piccadilly South, Manchester M60 7LP, Reference Number: MREC/00/8/74.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All study data are included in this manuscript.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.