A multicenter survey of rituximab therapy for refractory antineutrophil cytoplasmic antibody–associated vasculitis

Authors


Abstract

Objective

B cell depletion with rituximab has allowed remissions in relapsing or refractory antineutrophil cytoplasmic antibody (ANCA)–associated vasculitis in small studies. The aim of this study was to determine the efficacy and safety of rituximab for ANCA-associated vasculitis in a larger multicenter cohort. This permitted comparison of rituximab dosing regimens, the value of continuing immunosuppression, and investigation of ANCA and B cell levels as re-treatment biomarkers.

Methods

Retrospective, standardized data collection from 65 sequential patients receiving rituximab for refractory ANCA-associated vasculitis at 4 centers in the UK was used.

Results

All patients achieved B cell depletion. Complete remission occurred in 49 of the 65 patients (75%), partial remission in 15 (23%), and no response in 1 (2%). The prednisolone dosage was reduced from 12.5 mg/day (median) to 9.0 mg/day at 6 months (P = 0.0006). Immunosuppressive therapy was withdrawn in 37 of 60 patients (62%). Twenty-eight of 49 patients who achieved full remission (57%) experienced relapse (median 11.5 months). B cell return preceded relapse in 14 of 27 patients (52%). Although ANCA levels fell after rituximab therapy, relapse was not associated with ANCA positivity or a rise in ANCA levels. Neither the initial rituximab regimen (4 infusions of 375 mg/m2 each given 1 week apart or 2 infusions of 1 gm each given 2 weeks apart) nor withdrawal of immunosuppressive therapy (37 of 60 patients [62%]) influenced the timing of relapse. Thirty-eight patients received ≥2 courses of rituximab, and complete remission was induced or maintained in 32 of them (84%). IgM levels fell, although IgG levels remained stable. Forty-six serious adverse events occurred, including 2 episodes of late-onset neutropenia, which were attributed to rituximab.

Conclusion

Rituximab was effective remission induction therapy for refractory ANCA-associated vasculitis in this study. There was no difference in efficacy between the 2 main treatment regimens. Continuing immunosuppression did not reduce relapses. Relapses occurred, but re-treatment was effective and safe. There was no clear influence of rituximab on the frequency of serious adverse events. ANCA and B cell levels lacked sufficient sensitivity to guide the timing of re-treatment.

Antineutrophil cytoplasmic antibody (ANCA)–associated vasculitis is a multisystem autoimmune disease characterized by ANCA production and small-vessel inflammation. The ANCA-associated vasculitides comprise Wegener's granulomatosis (WG), microscopic polyangiitis (MPA), and Churg-Strauss syndrome (CSS), which have similar clinical and serologic features and similar treatment responses.

Standard immunosuppressive therapies for ANCA-associated systemic vasculitis (AAV), such as cyclophosphamide (CYC) with high-dose glucocorticoids, achieve initial remission rates of 70–90%. The remaining patients run a persistent disease course, and 50% of those who achieve initial remission will relapse within 2 years of diagnosis. Therapy-related toxicities, including sepsis and myelosuppression, remain a major problem and contribute to mortality (1). Newer therapies are needed.

Rituximab is a chimeric monoclonal antibody that targets the B cell–specific CD20 calcium channel, resulting in depletion of peripheral B cells, but not plasma cells. The drug was licensed in the US and Europe in 1997 for the treatment of B cell lymphomas (2) and in 2006 for rheumatoid arthritis (RA) (3–6). Preliminary studies have reported benefit with rituximab treatment in other autoimmune diseases, including systemic lupus erythematosus (7–9) and multiple sclerosis (10, 11).

In studies of small numbers of patients, rituximab has been associated with high rates of remission, often >80%, in patients with vasculitis that had been refractory to standard therapies (9, 12–19). Relapses of vasculitis following rituximab therapy have occurred, although efficacy with second and subsequent courses of rituximab has been reported (9, 13, 14). The value of B cell and ANCA levels as re-treatment guides in this context has not been determined. Rituximab dosing regimens have varied, with the majority of investigators using either 4 infusions of 375 mg/m2 given 1 week apart (the “lymphoma regimen”) or 2 infusions of 1 gm each given 2 weeks apart (the “RA” regimen). Until now, outcomes following different dosing regimens in vasculitis patients have not been compared. In previous studies, the majority of vasculitis patients received additional immunosuppressive therapies along with the rituximab, and in some studies, these therapies were continued after rituximab to prevent relapse. The safety of withdrawal of immunosuppressive therapy at the time of rituximab withdrawal has not been evaluated. We thus performed a multicenter survey to investigate these areas of controversy as well as to provide further safety and efficacy data with rituximab from a larger cohort of vasculitis patients with refractory disease.

PATIENTS AND METHODS

Participating centers and inclusion criteria.

Four UK vasculitis centers participated in this study: Addenbrooke's Hospital in Cambridge, the University Hospital in Birmingham, Great Ormond Street Hospital for Children in London, and the Hammersmith Hospital in London. One investigator per site completed standardized data sheets for sequential patients who met the inclusion criteria. Data collection included information on demographics, disease activity, and safety results at the time of each rituximab infusion and every 6 months thereafter. For the purpose of the rituximab response analysis, patient followup was censored at the time of rituximab re-treatment in patients receiving multiple courses. Included in this study were 11 patients from an earlier study (9) and 2 patients whose cases have previously been described (19, 20).

Study inclusion required a diagnosis of AAV (WG, MPA, or CSS) (21). Fifty-five of the 65 study patients (85%) had a history of ANCA positivity at some time in their disease course. It was not a requirement that ANCA be present at initiation of rituximab. All patients received rituximab as off-label therapy for refractory disease, encompassing those whose disease had failed to remit or had relapsed despite standard immunosuppression, as well as patients in whom standard therapies were contraindicated (22). Patients treated with rituximab as first-line therapy were excluded, as were those with fewer than 6 months of followup after a first rituximab course.

Clinical and laboratory assessments.

Complete remission was defined as the absence of disease symptoms and signs with a reduction in the steroid dosage (22). Partial remission was defined as a >50% reduction in disease activity, as assessed by the Disease Extent Index (DEI) (23). The DEI comprises 10 organ systems and an additional category for constitutional symptoms. Organ systems affected by active vasculitis are scored as 2 points and constitutional symptoms as 1 point. The maximum score is 21. Relapse was defined as a recurrence of symptoms attributable to vasculitis that warranted therapy escalation beyond a temporary increase in the corticosteroid dosage. Serious adverse events were those resulting in hospitalization, intravenous (IV) therapy, life-threatening situations, or death.

B cell depletion was defined as counts that were <0.02 × 109/liter, as measured by fluorescence-activated cell sorting (FACS), and B cell repopulation was defined as counts that were ≥0.02 × 109/liter. The ANCA staining pattern (cytoplasmic [cANCA] or perinuclear [pANCA]) was assessed by indirect immunofluorescence, and ANCA specificity for myeloperoxidase (MPO) or proteinase 3 (PR3) was detected by enzyme-linked immunosorbent assay (ELISA). The upper limit of normal for the binding levels of both MPO ANCA and PR3 ANCA ranged from 6 units/ml to 25 units/ml between centers. The MPO ANCA and PR3 ANCA results were interpreted according to local reference ranges.

Disease activity was classified according to the investigators' global assessment as new disease activity/worsening of disease activity, stable persistent disease activity for at least 3 months, or remission, as well as according to the DEI (23). The DEI score was chosen in preference to the Birmingham Vasculitis Activity Score (BVAS) (24) in order to minimize inaccuracies with interpretations of retrospective data.

Patient population.

This study included 65 patients, 40 from Addenbrooke's Hospital in Cambridge, 16 from the University Hospital in Birmingham, 5 from the Hammersmith Hospital in London, and 4 from the Great Ormond Street Hospital in London (Table 1). The median age of the patients at the first course of rituximab was 47 years (age range 7–77 years), 52% of them were male, and the majority had WG (71%). The median disease duration prior to rituximab therapy was 72 months (range 1.5–360 months). The frequency of new or worse disease activity was 85%, and 15% had persistent disease activity (Table 1).

Table 1. Characteristics and treatments in the patients with ANCA-associated vasculitis, by rituximab treatment group*
 Rituximab protocolAll patients (n = 65)
Two infusions of 1 gm each given 2 weeks apart (n = 32)Four infusions of 375 mg/m2 each given 1 week apart (n = 26)Other (n = 7)
  • *

    Except where indicated otherwise, values are the median (range). ANCA = antineutrophil cytoplasmic antibody; WG = Wegener's granulomatosis; MPA = microscopic polyangiitis; CSS = Churg-Strauss syndrome; PR3 = proteinase 3; MPO = myeloperoxidase; cANCA = cytoplasmic ANCA; pANCA = perinuclear ANCA; CYC = cyclophosphamide; anti-TNFα = anti–tumor necrosis factor α; IVIG = intravenous immunoglobulin; BSA = body surface area; ECMS = enteric-coated mycophenolate sodium.

  • Other rituximab protocols were as follows: 2 infusions of 750 mg/m2 each given 2 weeks apart; 3 infusions of 1 gm each given 1 week apart; and 1 infusion of 3.2 gm.

  • Values for the entire study group represent the number of patients; median dosage (range).

  • §

    Values are the median (range).

  • Patients received either intravenous (IV) corticosteroids (1–3 gm total) or high-dose daily oral corticosteroids (1 mg/kg for 5 days).

Patient characteristics    
 Age at first rituximab course, years49 (16–77)47 (11–69)25 (7–56)47 (7–77)
 No. (%) male16 (50)13 (50)5 (71)34 (52)
 Diagnosis, no. of patientsWG in 26, MPA in 4, CSS in 2WG in 16, MPA in 6, CSS in 3, unclassified in 1WG in 4, unclassified in 3WG in 46, MPA in 10, CSS in 5, unclassified in 4
 History of ANCA by diagnosis, no. of patientsPR3 in 21, MPO in 2, cANCA in 2, pANCA in 1PR3 in 11, MPO in 5, cANCA in 5, pANCA in 1PR3 in 5, cANCA in 1, MPO in 1PR3 in 37, MPO in 8, cANCA in 8, pANCA in 2
 Disease duration before rituximab, months72 (2–288)84 (10–360)60 (7–216)72 (1.5–360)
 Cumulative CYC dose (oral plus IV), gm28 (1–150)17 (0–90)15 (0–100)24 (0–150) (2 patients took no CYC)
 Previous immunotherapy, no. (%)    
  Anti-TNFα11 (34)8 (31)5 (71)24 (38)
  IVIG6 (19)10 (38)1 (14)17 (26)
  Azathioprine26 (81)17 (65)6 (86)49 (77)
  Methotrexate10 (31)4 (15)4 (57)18 (28)
  Mycophenolate mofetil23 (72)19 (73)5 (71)47 (74)
  Other10 (31)13 (50)2 (29)25 (38)
  Total no. of previous therapies/patient4 (2–8)4 (2–8)4 (2–6)4 (2–8)
 No. of previous relapses/patient3 (1–16)3 (2–11)6 (5–7)3 (1–16)
 ANCA status at study entry by diagnosis, no. of patientsPR3 in 13, MPO in 1, cANCA in 1PR3 in 8, MPO in 5, cANCA in 1PR3 in 4PR3 in 25, MPO in 6, cANCA in 2, negative in 32
 Total followup after rituximab, months21 (3–39)17 (3–52)21 (12–40)20 (3–55)
 No. of organs with active disease at first rituximab course2 (1–4)2 (1–6)3 (1–4)2 (1–6)
 Indication for rituximab, no. (%)    
  New/worse disease23 (71)26 (100)6 (86)55 (85)
  Persistent disease9 (28)0 (0)1 (14)10 (15)
 BSA, m21.85 (1.57–2.61)1.87 (0.89–2.38)
 Total rituximab dose adjusted for BSA at the first  rituximab course, mg1,081 (766–1,273)1,500 (1,500–1,500)
Treatments    
 Immunotherapy before the first rituximab course, no. (%)    
  Mycophenolate mofetil12 (38)15 (58)1 (14)28; 2 gm/day (0.5–3 gm/day)
  Azathioprine3 (9)2 (8)2 (29)7; 75 mg/day (12.5–150 mg/day)
  IV pulse CYC6 (19)006
  Oral CYC3 (9)02 (29)5; 150 mg/day (25–200 mg/day)
  Deoxyspergualin2 (6)2 (8)04
  Methotrexate2 (6)1 (4)03; 20 mg/day (17.5–25 mg/day)
  Plasma exchange1 (3)0 (0)01
  ECMS01 (4)01; 2.16 gm/day
  Alemtuzumab01 (4)01; 11 months before rituximab
  Combination of 2 or 3 therapies1 (3)2 (8)1 (14)4
  None2 (6)2 (8)1 (14)5
  Oral corticosteroid dosage, mg/day§12.5 (0–60)12.5 (0–60)20 (0–60)12.5 (0–60)
 Rituximab protocols, no. of patients    
  2 infusions of 1 gm each given 2 weeks apart320032 (total dose 2 gm)
  4 infusions of 375 mg/m2 each given 1 week apart026026 (total dose ∼3 gm)
  2 infusions of 750 mg/m2 each given 2 weeks apart0033 (total dose ∼3 gm)
  3 infusions of 1 gm each given 1 week apart0011 (total dose 3 gm)
  1 infusion of 3.2 gm0033 (total dose 3.2 gm)
 Additional therapy with the first rituximab course, no. (%)    
  CYC12 (38)12 (46)4 (57)28; 0.9 gm (0.2–1.85 gm)
  IV corticosteroids/5 days 1 mg/kg oral10 (31)1 (4)6 (86)17
  None17 (53)13 (50)022

At the time of the first rituximab course, a median of 2 organs (range 1–6) showed active disease. The frequency of initial organ activity is illustrated in Figure 1A. The relatively high rate of endobronchial disease and retroorbital granuloma reflected the most common diagnosis (WG) and the complex treatment-resistant nature of this patient cohort. Just prior to rituximab administration, 92% were receiving continuous therapy with immunosuppressive agents (Figure 1B).

Figure 1.

Characteristics of the 65 patients with antineutrophil cytoplasmic antibody–associated vasculitis refractory to treatment and response to therapy with rituximab. A, Proportion of patients with specific organ system activity at the first rituximab course. ENT = ear, nose, throat; LN = lung nodules; LO = lung other (pulmonary fibrosis and asthma); EB = endobronchial; ROG = retroorbital granuloma; PNS = peripheral nervous system; GIT = gastrointestinal tract; LH = lung hemorrhage; other = muscle. B, Proportion of patients receiving specific immunosuppressive agents prior to the first rituximab course. MMF = mycophenolate mofetil; AZA = azathioprine; IV = intravenous; CYC = cyclophosphamide; PO = oral; DSG = deoxyspergualin; MTX = methotrexate; PEX = plasma exchange; ECMS = enteric-coated mycophenolate sodium; CAM = Campath 1H (Bayer, Bury St. Edmonds, UK; alemtuzumab); multiple = 2 or more concomitant therapies. C, Proportion of patients with specific disease activity states at 0, 6, and 12 months following the first rituximab course. D, Disease Extent Index at 0, 6, and 12 months following the first rituximab course and at the last followup visit. Data are shown as box plots. Each box represents the 25th to 75th percentiles. Lines inside the boxes represent the median. Lines outside the boxes represent the 10th and the 90th percentiles. Circles indicate outliers. E, Daily oral dose of prednisolone at 0, 6, and 12 months following the first rituximab course and currently. Values are the mean and SD. F, Proportion of patients receiving concomitant therapy with immunosuppressive agents at 0, 6, and 12 months following the first rituximab course and currently.

Five patients (8%) were not taking immunosuppressive therapy at the time of the first rituximab course, and all of them had active disease. Four of these 5 patients could not tolerate other immunosuppressive therapies: 2 of them had WG, and immunosuppressive therapy was withdrawn 1 month prior to the first rituximab course because of recurrent chest infections, and the other 2 patients had prolonged recurrent leukopenia. One patient was not receiving immunosuppressive agents during recovery from chemotherapy for non-Hodgkin's lymphoma.

Treatment protocol and followup.

Rituximab was administered to 26 patients as 4 doses of 375 mg/m2 each given 1 week apart, as used to treat lymphoma, and to 32 patients as 2 doses of 1 gm each given 2 weeks apart, as used to treat RA. In 7 patients, other dosing regimens were used for the first rituximab course.

In addition to the rituximab, 28 patients received IV CYC (median total dose 0.9 gm [range 0.2–1.85 gm]); another 8 patients had stopped taking CYC just prior to initiation of rituximab. A total of 17 patients received additional high-dose oral or IV glucocorticoids at the time of rituximab. No additional therapy was administered to 22 patients.

The median followup after rituximab was for 20 months (range 3–55 months). All patients had at least 6 months of followup, except for 1 patient who died at 3 months of followup. The median followup after the first rituximab course, with censoring at the time of the second rituximab course, was for 14 months (range 3–52 months).

After the first rituximab course, some investigators administered a second course of rituximab if relapse occurred, whereas others administered a second course on a preemptive basis, either at a set time point (e.g., every 6 months) or following B cell restoration. One investigator did not administer a second course of rituximab, but switched to other therapies if relapse occurred. Of the 38 patients (59%) who received a second rituximab course, the rituximab was given preemptively in 6, after relapse in 27, and to further improve disease control after either partial remission (4 patients) or persistently active disease (1 patient) following the first rituximab course in 5 patients.

The results after the first rituximab course, with followup censored at the time of the second rituximab course, provided the most information with regard to the timing of the return of B cells, increases in ANCA levels, and relapse of disease. Results following multiple additional courses of rituximab were informative with regard to long-term changes in immunoglobulin levels and severe adverse events.

Statistical analysis.

Statistical analysis was performed using SPSS version 11 (SPSS, Chicago, IL) and GraphPad Prism version 4 (GraphPad Software, San Diego, CA) software packages. Changes in variables, including the DEI and the prednisolone dosage, were compared by Mann-Whitney U test. Relapses were analyzed using Kaplan-Meier survival analysis, with log rank analysis for significance. P values less than 0.05 were considered significant for all statistical tests.

RESULTS

Efficacy of rituximab.

B cell depletion and remission.

Depletion of B cells (<0.02 × 109/liter) was achieved in all patients after the first rituximab course. B cell repopulation occurred in 28 of 64 patients (44%) at a median of 11 months (range 6–37 months) after the first rituximab course (B cell followup data not available in 1 patient). Of the remaining 36 patients, 19 were given a second rituximab course before the return of B cells, and 17 remained persistently B cell depleted at a median followup of 15 months (range 3–52 months). After a second rituximab course in 38 patients, B cell depletion did not occur in 1 of them.

At the time of the first course of rituximab, all patients had active disease. Fifty-five of the 65 patients (85%) had new or worse disease activity, and the remaining 10 patients (15%) had stable, persistent disease activity for at least 3 months. The median DEI score at the time of the first rituximab course was 4 (range 2–11) in patients classified as having new or worse disease and 2 (range 2–6) in patients with persistent disease activity. Figures 1C and D show the changes in disease activity over time. The DEI fell from a median of 4 (range 2–11) to a median of 0 (range 0–4) at 6 months (P < 0.0001) and remained low at 12 months (median 0 [range 0–7]; P < 0.0001).

Complete remission occurred in 49 of the 65 patients (75%), with a further 15 patients (23%) experiencing partial remission. One patient with a retroorbital granuloma failed to respond to rituximab therapy. This patient had previously failed to respond to CYC (>100 gm), azathioprine (AZA), and mycophenolate mofetil (MMF). The median time to remission was 2 months (range 1–5 months). Remission was accompanied by a decrease in the prednisolone dosage from a median of 12.5 mg/day (range 0–60 mg/day) initially to a median of 9.0 mg/day (range 0–80 mg/day) at 6 months (P = 0.006) and a median of 7.5 mg/day (range 0–20 mg/day) at 12 months (P < 0.001). Figure 1E shows the changes in the mean prednisolone dosage over time.

Of the 60 patients who were receiving immunosuppressive agents just prior to the first rituximab course, the immunosuppressive therapy was withdrawn in 37 of them (62%) (Figure 1F). Treatments that were withdrawn were MMF in 16 patients, CYC in 8, deoxyspergualin in 4, AZA in 2, methotrexate (MTX) in 2, enteric-coated mycophenolate sodium (ECMS) in 1, plasma exchange in 1, alemtuzumab in 1, infliximab plus cyclosporine in 1, and MMF plus cyclosporine in 1. In a further 4 patients, the dosage of immunosuppressive therapy was reduced (AZA in 1, MMF in 2, and MTX in 1). In 3 patients, the CYC was switched to another agent (MMF in 2 and AZA in 1). In 2 patients, part of their combination therapy was reduced (CYC withdrawn with dapsone continued in 1, and etanercept and MMF withdrawn with IV immunoglobulin [IVIG] continued in 1). Ten patients continued taking MMF and 4 continued taking AZA, with unchanged dosages.

Fifteen patients received only 1 course of rituximab and remained relapse-free thereafter. The median followup in these patients was 21 months (range 3–52 months).

Relapse and re-treatment.

Relapse occurred in 28 of the 49 patients who experienced complete remission initially (57%). The median time to relapse was 11.5 months (range 4–37 months). No differences in disease subtype, initial ANCA status (positive versus negative or MPO versus PR3 ANCA), or organ system involvement were noted between patients who experienced relapse and patients who did not.

Thirty-eight patients received a second course of rituximab. Of these 38 patients, 27 were treated for disease relapse, 6 were treated preemptively to avoid relapse, either following B cell restoration or at a set time point (e.g., every 6 months) while they were still B cell depleted, and 5 were treated to further improve disease control because of either partial remission (n = 4) or persistently active disease (n = 1). After the second course of rituximab, 84% experienced a second complete remission or maintained remission. The median time to remission after a second rituximab course was 1.5 months (range 1–4 months). Three or more rituximab courses were administered to 20 patients, with 1 patient receiving a total of 7 courses. Figure 2 shows the rituximab re-treatment, response, and relapse patterns.

Figure 2.

Rituximab courses, disease remission, and disease relapse in 65 patients with antineutrophil cytoplasmic antibody–associated vasculitis refractory to treatment. Values are the total number of patients receiving each treatment course (number treated for active disease/number treated preemptively for relapse) and the number (%) of patients. One patient received 7 courses of rituximab and was still in remission at the last followup visit.

In 2006, 1 of the study centers started routine preemptive re-treatment with 1 gm of rituximab given in 1 infusion every 6 months. Fifteen patients received at least 1 preemptive course of rituximab (maximum 3 courses). Following initiation of the preemptive re-treatment practice, no relapses occurred (median followup 11 months [range 5–23 months]).

Predictors of remission and relapse.

B cell depletion after the initial rituximab course was followed by B cell return at a median of 11 months in 28 patients, 25 of whom experienced complete remission. Of the 28 patients who had a relapse following the initial complete remission, 14 had a relapse at the time of or after B cell return, 3 had a relapse before B cell return, and 10 had a relapse but remained persistently B cell depleted for the remainder of followup (B cell data missing in 1 patient who had a relapse). Therefore, 48% of patients (13 of 27) had a relapse before B cell repopulation, and 32% of patients (8 of 25) in whom B cells returned following an initial complete remission did not have a relapse. Figures 3A and B illustrate the relationship between relapse and B cell return in individual patients achieving complete remission after the initial course of rituximab. Twenty-five patients experienced restoration of B cells (Figure 3A), and 23 experienced continued depletion of B cells (Figure 3B).

Figure 3.

Predictors of remission and relapse in patients with antineutrophil cytoplasmic antibody (ANCA)–associated vasculitis refractory to treatment. A and B, Relationship between complete remission in 49 patients (B cell data unavailable in 1), relapse, and B cell repopulation (≥0.02 × 109/liter). Shown are the duration of followup, the timing of B cell return, and the timing of relapse in 25 patients with complete remission and B cell repopulation occurring during followup (14 with relapse at the time of or after B cell return, 3 with relapse before B cell return, and 8 without relapse) (A), as well as the duration of followup and the timing of relapse in 23 patients with complete remission who remained B cell depleted throughout followup (10 with relapse, and 13 without relapse) (B). C, Relapse-free survival curves in 63 patients who were positive (n = 33) or negative (n = 30) for ANCA at the time of the first rituximab course. D, Levels of proteinase 3 (PR3) ANCA and myeloperoxidase (MPO) ANCA binding at 0, 6, and 12 months following the first rituximab course and currently in the 31 patients who were initially positive for PR3 ANCA or MPO ANCA by enzyme-linked immunosorbent assay (ELISA). Values are the median, with 25th and 75th percentiles. Horizontal line shows the lowest value for the upper limit of normal (6 units/ml).

At the time of initial treatment, 33 patients were ANCA positive (25 had PR3 ANCA, 6 had MPO ANCA, and 2 had cANCA alone), and 32 were ANCA negative. The initial median DEI score was lower in ANCA-negative patients (3.8 versus 4.5 in ANCA-positive patients; P = 0.03). The remission and relapse rates were similar in the ANCA-positive and ANCA-negative groups (Figure 3C). The mean levels of MPO ANCA and PR3 ANCA binding, as determined by ELISA, were decreased at 6 months after the first rituximab course (P = 0.001) (Figure 3D). Twenty- eight patients experienced a disease relapse after an initial complete remission; 12 of them were ANCA-positive at the time of rituximab treatment, as determined by ELISA. Of these 12 patients, 2 showed a clear increase in ANCA levels before relapse, as determined by ELISA. The majority of patients therefore relapsed without a major change in the ANCA-binding level.

Factors influencing remission and relapse.

The predominant rituximab protocols used were 4 infusions of 375 mg/m2 each given 1 week apart (the lymphoma regimen; n = 26) and 2 infusions of 1 gm each given 2 weeks apart (the RA regimen; n = 32). Both regimens induced depletion of peripheral blood B cells in all patients and resulted in similar rates of remission (81% and 75%, respectively). Neither the duration of B cell depletion nor the duration of disease remission was influenced by the different dosing regimens (Figures 4A and B). More of the patients who received the 2 1-gm infusions of rituximab were treated with a course of CYC just prior to the rituximab course. Other factors that may have influenced outcome, including additional CYC with the rituximab, the average daily corticosteroid dose, withdrawal of immunosuppressive agents, the initial DEI scores, and the disease subtype, were similar in the 2 main treatment protocol groups (Table 1).

Figure 4.

Factors influencing remission and relapse in patients with antineutrophil cytoplasmic antibody–associated vasculitis refractory to treatment. A, B cell depletion survival curves following the first rituximab course, by rituximab treatment protocol (2 main protocols used): 4 infusions of 375 mg/m2 each given 1 week apart (n = 26) and 2 infusions of 1 gm each given 2 weeks apart (n = 30; B cell data unavailable in 1 patient). B, Relapse-free survival curves following the first rituximab course, by rituximab treatment protocol (2 main protocols used): 4 infusions of 375 mg/m2 each given 1 week apart (n = 26) or 2 infusions of 1 gm each given 2 weeks apart (n = 31). C, Relapse-free survival curves following the first rituximab course in patients who did (n = 10) and those who did not (n = 33) continue immunosuppressive therapy. Patients in whom an initial remission did not occur and those who were not receiving immunosuppressive agents at the time of the first rituximab course were excluded from the analysis.

In 60 patients who were receiving immunosuppressive agents just prior to the first rituximab course, 62% (37 patients) had concomitant immunosuppressive therapy withdrawn following the initial rituximab course (MMF in 16, CYC in 8, deoxyspergualin in 4, AZA in 2, MTX in 2, ECMS in 1, plasma exchange in 1, alemtuzumab in 1, and combination therapy with infliximab plus cyclosporine in 1 and MMF plus cyclosporine in 1). The withdrawal of immunosuppressive therapy occurred in 32 of the 60 patients by 6 months and in 37 of the 60 patients by 12 months. No significant difference in time to relapse was seen in patients in whom immunosuppressive agents were withdrawn as compared with patients in whom immunosuppressive agents were continued (P = 0.79) (Figure 4C). Median steroid dosages were similar between these 2 patient groups, as were disease demographics and changes in the DEI score.

Rituximab safety.

Rituximab was well tolerated, and there were no severe infusion reactions. During an overall total of 1,548 months of followup, 45 serious adverse events occurred in 25 patients (Table 2). Sixteen serious infections occurred a median of 8 months after the first rituximab course (range 0.5–49 months). Seventeen events, including 1 death, were related to active vasculitis. In 2 patients, neutropenia (<0.5 × 109/liter) occurred at 3 months and 5 months, respectively, after the second rituximab course. Both episodes were short lived, and there was spontaneous recovery of the neutrophil count, without sepsis. Although concomitant medications might have been responsible (trimethoprim/sulfamethoxazole and AZA in one patient and cyclosporine plus amlodipine in the other), rituximab was considered the likely cause. With the exception of neutropenia, no other serious adverse event appeared to be directly attributable to rituximab (Table 2).

Table 2. Serious adverse events occurring in the 65 study patients*
Event category, serious adverse eventNo. (%) of patientsNo. of eventsNo. of months since first rituximab courseNo. of rituximab courses before the event
  • *

    Serious adverse events were those that resulted in hospitalization, intravenous therapy, a life-threatening situation, or death. A total of 46 serious adverse events occurred in 25 patients. Some patients experienced more than 1 serious adverse event. ENT = ear, nose, throat; IVIG = intravenous immunoglobulin.

Disease activity    
 Endobronchial disease5 (7.5)61, 4, 4, 9, 38, 471, 1, 1, 1, 2, 3
 Asthma2 (3.0)62, 2, 6, 7, 9, 111, 1, 1, 1, 1, 1
 Peripheral neuropathy1 (1.5)141
 ENT disease2 (3.0)24, 61, 1
 Pulmonary fibrosis, respiratory failure, and death1 (1.5)181
 Retroorbital granuloma1 (1.5)1272
Infection    
 Pneumonia (organism not specified)7 (11.0)100.5, 1, 2, 4, 5, 5, 9, 14, 24, 491, 1, 1, 1, 1, 1, 2, 2, 1, 2
 Aspergillus lung infection1 (1.5)182
 Pseudomonas lung infection1 (1.5)192
 Line sepsis1 (1.5)131
 Sepsis2 (3.0)22, 101, 2
 Cellulitis1 (1.5)1202
Other    
 Pulmonary embolism1 (1.5)1241
 Sudden unexplained death1 (1.5)131
 Addisonian crisis2 (3.0)25, 271, 1
 Hip surgery1 (1.5)151
 Aseptic meningitis post-IVIG therapy1 (1.5)161
 Aortic valve replacement1 (1.5)181
 Neutropenia2 (3.0)214, 182, 2
 Failure of B cells to deplete1 (1.5)1192
 Ankle surgery1 (1.5)1241
 Eye surgery1 (1.5)1423

Two deaths occurred. One sudden unexplained death occurred 3 months following a rituximab course in a 59-year-old patient whose disease was in remission. A postmortem examination was not performed. The second death occurred 8 months after a rituximab course in a 7-year-old patient. This boy had severe pulmonary fibrosis and MPO ANCA positivity (with unclassified AAV) prior to treatment with rituximab. The cause of death was believed to be disease-related and not the consequence of rituximab per se.

IgM levels fell from a median of 0.8 gm/liter (range 0.2–3.3 gm/liter) at 0 months to a median of 0.52 gm/liter (range 0.1–1.7 gm/liter) at 6 months (P = 0.002), with a further nonsignificant decline continuing over the remainder of followup (0.50 gm/liter [range 0.09–1.5 gm/liter] at the last followup; normal range 0.4–2.2 gm/liter). The median IgG level was 7.9 gm/liter (range 1.5–28.9 gm/liter) initially and 8.0 gm/liter (range 3.7–17.4 gm/liter) at the last followup (P = 0.65) (normal range 6–13 gm/liter).

DISCUSSION

B cell depletion with rituximab therapy was associated with complete remission in 75% of adults and children with refractory AAV. Remissions were sustained after a single rituximab course, although 58% of these patients subsequently experienced a relapse. Multiple courses of rituximab effectively induced further remissions following relapses, and re-treatment on a preemptive basis resulted in sustained remissions. These results are consistent with those of previous studies (9, 12–20). This is the largest series reported to date, and it pools the results from 4 UK centers. All patients had active disease at the time rituximab was initiated, and previous disease courses were characterized by poorly controlled disease, multiple relapses, and treatment with multiple immunosuppressive agents.

Some centers followed the practice of temporary use of IV CYC or corticosteroids at the time of rituximab therapy to aid early disease control. Given that complete remissions occur an average of 2 months after the rituximab course, this approach seems reasonable in patients with rapidly progressive disease. It is unlikely that this temporary increase in immunosuppressive agents would account for the sustained remissions in patients with active disease who were already receiving continuous therapy with immunosuppressive agents. Furthermore, a significant number of patients had sustained remissions without additional therapies.

A positive outcome bias may occur in retrospective studies, and this possibility was minimized by including in our study all eligible patients who were treated with rituximab during the study period and by including data from 4 independent centers. The overall response to rituximab in this patient series was superior to that seen with alternative therapies in similar cohorts of patients with refractory vasculitis. In a retrospective series of patients treated with MMF, a reduction in disease activity occurred in 86% (19 of 22); however, few achieved complete remission, and 47% experienced a relapse by 14 months, which required other therapies (25). In a placebo-controlled trial, 82% of patients (14 of 17) responded to IVIG therapy; however, the response was not maintained beyond 3 months (26). In a long-term followup of 71 patients treated with alemtuzumab, 85% responded; however, relapses occurred in 72% at a median of 9.2 months (27). Thus, the quality and duration of responses following rituximab treatment appear to be superior to those following alternative therapies for refractory disease.

One patient with a retroorbital granuloma failed to respond to rituximab. This patient had a long disease course and was exposed to a high cumulative dose of CYC. Aries et al (28) reported improvement in only 3 of their 8 patients who had disease manifestations and chronicity similar to those in our patient. On the other hand, our results show that rituximab is effective therapy for granulomatous manifestations, such as retroorbital granuloma and endobronchial disease, in the majority of patients. These differences may be partly due to the rituximab dosing regimen used by Aries et al (an infusion of 375 gm/m2 given every fourth week), as well as the longer followup in our series of patients, which allowed for sufficient time to observe the maximum benefit of rituximab therapy.

The earlier use of rituximab in our patient series was associated with continued concomitant immunosuppression. After observing sustained remission, some investigators reduced the dosage and then withdrew the concomitant therapies, and our results showed that the rates of relapse were not higher in patients in whom these treatments were withdrawn. A selection bias in patients from whom immunosuppressive therapy was withdrawn is likely, with investigators preferring to withdraw therapies when the perceived risk of relapse is low. However, in 1 center (Addenbrooke's Hospital; 40 patients), the decision to withdraw immunosuppressive therapy was dictated by local policy and not solely by the characteristics of the patients. In the majority of patients in this study, the steroid dosages were tapered, but the treatment was not withdrawn. The continuation of low-dose corticosteroids may have protected patients from experiencing a relapse.

No dose-ranging studies have been performed in vasculitis patients to ascertain the optimum rituximab dosage, and previous small, single-center studies have been unable to compare different dosing regimens. The pharmacokinetics and pharmacodynamics of rituximab are variable and are influenced by factors such as polymorphisms in Fcγ receptor IIIa (29), proteinuria, and human antichimeric antibodies (30). The rituximab dosing regimen used to treat lymphoma (4 infusions of 375 mg/m2 each given 1 week apart) results in an average total dose of 2.6 gm in a person weighing 70 kg. In a study of 203 lymphoma patients, the mean maximum serum concentration (Cmax) following the fourth rituximab infusion was reported to be 486 μg/ml (31). In comparison, the dosing regimen used to treat RA (2 infusions of 1 gm each given 2 weeks apart) results in a lower total dose of rituximab (2 gm) that is administered over a shorter time period, and a mean Cmax of 370 μg/ml following the second infusion has been reported (31). The latter regimen is more attractive in terms of patient convenience, cost, and ease of administration. An alternative rituximab regimen of 2 infusions of 750 mg/m2 given 2 weeks apart was used in the pediatric patients in our study; this regimen provided the same cumulative dose as used in lymphoma patients, with less inconvenience to the patient.

Pharmacokinetic modeling in RA suggested that a body surface area–calculated dose would not improve the predictability of drug exposure (32), and clinical trials in RA have found that 2 infusions of 1 gm given 2 weeks apart is both effective (5, 6) and safe with repeated use (33). The more extensive organ involvement that usually occurs in vasculitis as compared with RA raises the question of whether 2 infusions of 1 gm given 2 weeks apart would allow sufficient tissue penetration for B cell depletion from lesional tissue, perhaps allowing sustained remissions. The results from our series of patients showed no difference in the duration of B cell depletion or the therapeutic effect between the 2 main treatment regimens (2 infusions of 1 gm each given 2 weeks apart and 4 infusions of 375 mg/m2 each given 1 week apart), which suggests that the dosing regimen in which 2 1-gm infusions are given 2 weeks apart may be sufficient in AAV.

The apparent benefit of second and subsequent rituximab courses observed in our study suggests a role of rituximab re-treatment in AAV. Preemptive re-treatment with rituximab in adult patients may be preferable to re-treatment following relapse, although data relating to this issue in children are not provided by this study. Given the variable timing of relapses, it would be ideal to target the therapy to the individual patient to avoid unnecessary treatments and costs. We evaluated the role of ANCA and B cell levels as predictors of relapse and re-treatment. ANCA was present in just over one-half of the patients at the time of initial treatment, consistent with ANCA positivity in other cohorts of patients with refractory AAV (34, 35). Levels of MPO and PR3 ANCA fell following rituximab treatment; however, clinical remissions occurred to the same extent in both ANCA-positive and ANCA-negative patients, and relapses were preceded by an increase in ANCA levels in only a few patients. From these data, we can conclude that ANCA was not sufficiently sensitive to guide the timing of re-treatment.

Similar to the findings in RA patients treated with rituximab (33), the repopulation of B cells in AAV patients was not a reliable marker of relapse. Just under one-half of the relapses occurred before B cell return, and 32% of the patients whose B cells returned did not experience a relapse. Therefore, repopulation of B cells in the peripheral blood is not sufficiently reliable to guide re-treatment alone. Possible explanations for relapse in the absence of apparent peripheral blood B cells are that our FACS analysis did not detect B cell levels that were <0.02 ×109/liter. A more sensitive analytical method may have shown closer correlations with disease activity. In addition, rituximab results in the reduction of CD20-positive B cells in tissues as well as in blood (36, 37), which may not be complete in tissues (38, 39). Certainly, tissue B cell repopulation can occur prior to detectable return to the peripheral blood (20). Furthermore, variability in the phenotype of the repopulating B cells has been demonstrated following rituximab therapy in patients with systemic lupus erythematosus (40), and this may contribute to the variable duration of clinical effect.

Rituximab was well tolerated, and there were no severe infusion reactions. The frequency of serious adverse events was similar to that seen in other studies of vasculitis patients and appeared to largely reflect the influence of concurrent or previous immunosuppressive therapy, underlying disease, and corticosteroid therapy. A review of 5 trials of rituximab treatment in patients with non-Hodgkin's lymphoma found no additional risk of infection with rituximab and chemotherapy versus chemotherapy alone (41); however, it remains unclear whether rituximab increases the risk of infection in patients with vasculitis. Two patients experienced transient neutropenia. Rituximab-associated late-onset neutropenia has been described not only in association with the use of high-dose immunosuppressive therapy in lymphoma patients (42), but also in association with concomitant low-dose immunosuppressive therapy in patients with pemphigus vulgaris (as in our patients) (43). Arrest of promyelocyte maturation has been demonstrated in late-onset neutropenia following rituximab therapy (42); however, the mechanism is unclear. The neutropenia in our patients resolved spontaneously, without the use of granulocyte colony-stimulating factor and without sequelae from infection. Two deaths occurred. In neither case was a causal association between rituximab and death found.

A significant fall in IgM levels occurred by 6 months. IgG levels were maintained within the normal range. These results are consistent with short-lived plasma cells making a greater contribution to serum IgM than to serum IgG; conversely, most of the IgG that is present in serum is produced by long-lived bone marrow plasma cells (44). Rates of serious infection were not increased with prolonged followup or after successive courses of rituximab. Similar decreases in IgM levels have been reported with multidose rituximab regimens in RA (33), and this may complicate long-term therapy in some patients. Multiple courses of rituximab in this cohort appeared to be safe, although it is important to emphasize that patients who received multiple courses tended to be in remission, receiving little or no concomitant immunosuppressive therapy, and taking low doses of corticosteroids and were therefore generally at a lower risk of infection than other vasculitis cohorts. In order to minimize the risk of underreporting of infections, only data concerning serious infections were recorded. It remains to be seen whether more prolonged use of rituximab will ultimately result in significant declines in immunoglobulin levels, as may be expected if a necessary replenishment of long-lived plasma cells is prevented.

In conclusion, our findings add to previous evidence of the high level of efficacy of rituximab therapy in patients with refractory AAV. This treatment should now be considered for patients with this presentation. While an additional course(s) of rituximab is also effective, the value of a re-treatment protocol in preventing disease flares remains under investigation. Immunosuppressive therapy can be withdrawn and steroid dosages tapered in patients achieving complete clinical remission, without jeopardizing the relapse risk, as long as preemptive re-treatment is considered as a long-term strategy. The optimum timing of a subsequent course of rituximab is unclear, and better disease biomarkers are needed.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Jones had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Jones, Brogan, Savage, Jayne.

Acquisition of data. Jones, Ferraro, Chaudhry, Brogan, Salama, Smith, Jayne.

Analysis and interpretation of data. Jones, Chaudhry, Brogan, Savage, Jayne.

Acknowledgements

We thank all of the nurses and physicians involved in the care of the patients included in this study.

Ancillary