Rituximab for remission maintenance in relapsing antineutrophil cytoplasmic antibody–associated vasculitis

Authors

  • Rona M. Smith,

    Corresponding author
    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    • Vasculitis and Lupus Clinic, Addenbrooke's Hospital Cambridge, Hills Road, Cambridge CB2 0QQ, UK
    Search for more papers by this author
    • Drs. R. M. Smith and R. B. Jones contributed equally to this work.

    • Dr. R. M. Smith has received speaking fees (less than $10,000) as well as research grant funding from Roche.

  • Rachel B. Jones,

    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
    • Drs. R. M. Smith and R. B. Jones contributed equally to this work.

    • Dr. Jones has received speaking fees from Roche (less than $10,000) and travel expenses from Ortho Biotech (less than $10,000).

  • Mary-Jane Guerry,

    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
  • Simona Laurino,

    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
  • Fausta Catapano,

    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
  • Afzal Chaudhry,

    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
    • Dr. Chaudhry has received consulting fees, speaking fees, and travel expenses from Roche (less than $10,000).

  • Kenneth G. C. Smith,

    1. Addenbrooke's Hospital, Cambridge Institute for Medical Research, and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
  • David R. W. Jayne

    1. Addenbrooke's Hospital and University of Cambridge School of Clinical Medicine, Cambridge, UK
    Search for more papers by this author
    • Dr. Jayne has received consulting fees and speaking fees (less than $10,000) as well as research grant funding from Roche.


Abstract

Objective

Rituximab is effective induction therapy in refractory or relapsing antineutrophil cytoplasmic antibody–associated vasculitis (AAV). However, further relapse is common, and maintenance strategies are required. The aim of this study was to reduce relapse rates using a fixed-interval rituximab re-treatment protocol.

Methods

Retrospective, standardized collection of data from sequential patients receiving rituximab for refractory or relapsing AAV at a single center was studied. Group A patients (n = 28) received rituximab induction therapy (4 infusions of 375 mg/m2 or 2 infusions 1 gm) and further rituximab at the time of subsequent relapse. Group B patients (n = 45) received routine rituximab re-treatment for 2 years: 2 doses of 1 gm each for remission induction, then 1 gm every 6 months (total of 6 gm). Group C patients (n = 19) comprised patients in group A who subsequently relapsed and began routine re-treatment for 2 years.

Results

Response (complete/partial remission) occurred in 26 of the 28 patients (93%) in group A, 43 of the 45 patients (96%) in group B, and 18 of the 19 patients (95%) in group C. At 2 years, relapses had occurred in 19 of 26 patients (73%) in group A, 5 of 43 (12%) in group B (P < 0.001), and 2 of 18 (11%) in group C (P < 0.001). At the last followup (median of 44 months), relapses had occurred in 85% of those in group A (22 of 26), 26% of those in group B (11 of 43; P < 0.001), and 56% of those in group C (10 of 18; P = 0.001). Glucocorticoid dosages were decreased and immunosuppression therapy was withdrawn in the majority of patients. Routine rituximab re-treatment was well tolerated, and no new safety issues were identified.

Conclusion

Two-year, fixed-interval rituximab re-treatment was associated with a reduction in relapse rates during the re-treatment period and a more prolonged period of remission during subsequent followup. In the absence of biomarkers that accurately predict relapse, routine rituximab re-treatment may be an effective strategy for remission maintenance in patients with refractory and relapsing AAV.

Antineutrophil cytoplasmic antibody–associated vasculitis (AAV), including granulomatosis with polyangiitis (Wegener's) (GPA) and microscopic polyangiitis (MPA), is characterized by leukocyte infiltration of blood vessel walls, fibrinoid necrosis, and vascular damage and is associated with the presence of circulating autoantibodies to neutrophil cytoplasmic antigens (ANCAs). Prior to the availability of effective treatment, AAV was associated with a mortality rate of 93% within 2 years, primarily due to renal and respiratory failure (1). The introduction of glucocorticoids and cyclophosphamide has transformed survival times, with 5-year survival rates approaching 80% (2). AAV has become a chronic relapsing disorder, with progressive organ damage and disability affecting >95% of patients. The cumulative exposure to glucocorticoids and immunosuppressive drugs contributes to toxicity and organ damage and is of particular concern to the 50% of patients who experience a relapse within 5 years of initial remission and the 10% who have a refractory disease course (3–5).

The chimeric anti-CD20 monoclonal antibody rituximab depletes pre–B cells and mature B cells but not plasma cells. Rituximab has efficacy equivalent to that of cyclophosphamide for remission induction in new AAV patients, and superior efficacy in those with relapsing disease (6, 7). However, the majority of patients with refractory/relapsing AAV experience another relapse after rituximab, regardless of whether immunosuppression is withdrawn (8). Time to relapse is variable, with the median being 12 months. Further rituximab courses are effective, but it is uncertain whether repeated treatment to avoid relapse should be guided by the peripheral blood B cell counts or the ANCA levels or whether a routine, time-based re-treatment regimen should be followed.

Routine re-treatment with rituximab (1 gm at 24 weeks) in rheumatoid arthritis (RA) patients has been shown to be effective in the prevention of relapse and is associated with overall improvement in disease control (9). In AAV, relapses may result in loss of vital organ function, thus necessitating exposure to further high-dose glucocorticoids and immunosuppressive agents. We compared routine, time-based rituximab re-treatment to administration of further rituximab at the time of relapse to ascertain whether relapse rates could be reduced and prolonged treatment-free remissions secured in patients with refractory/relapsing AAV.

PATIENTS AND METHODS

Patient population.

In 2002, rituximab administration as off-label, compassionate use therapy for refractory/relapsing AAV was begun at our institution. Initially, patients received a single course of rituximab and only received subsequent rituximab at the time of relapse. A routine, time-based re-treatment regimen was implemented in 2006. In the present study, the outcomes in patients treated from 2002 to 2006 according to a non–routine re-treatment protocol were compared to those in patients treated after 2006 according to the routine rituximab re-treatment protocol. In accordance with the UK National Health Service Research Ethics Committee guidelines, ethics approval was not required because this work comprises retrospective data and all treatment decisions were made prior to our evaluation.

Study inclusion required a diagnosis of AAV (GPA or MPA) (10). Patients with Churg-Strauss syndrome were excluded. A total of 69 of 73 patients (95%) were ANCA positive during their prior disease course. Four with ANCA-negative vasculitis had histologic confirmation of GPA. All had refractory/relapsing disease. This included patients in whom remission had never been achieved and those experiencing a relapse despite remission maintenance immunosuppression, as well as patients in whom standard therapies were contraindicated and those with partial remission in whom prednisolone >10 mg/day was required for disease control (11). Patients given rituximab as first-line therapy were excluded, as were those with <6 months of followup.

Clinical data.

Data collection included pre-rituximab demographic characteristics, disease activity assessments, medications, and laboratory results at each clinic assessment and severe adverse events since rituximab initiation. This study included 35 patients described in a previous report (8). Disease activity was graded according to the Disease Extent Index (DEI) (12) and by investigators' assessments of disease activity as either remission, partial remission, progressive disease (treatment failure), or relapse. Full remission was defined as a DEI score of ≤2 and a prednisolone dosage of ≤10 mg/day. Partial remission was defined as a >50% reduction in the DEI score. Treatment failure occurred when remission was not achieved and disease activity progressed, necessitating additional immunosuppression. Major relapse was defined as vital organ–threatening vasculitic activity, such as, disease of the eyes, lungs, kidneys, or subglottis that would historically have been managed with cyclophosphamide. Minor relapse was defined as other manifestations attributable to active vasculitis plus an increase in therapy: either unscheduled rituximab, addition of another immunosuppressive agent, intravenous (IV) methylprednisolone, or oral prednisolone at a dosage of >20 mg/day. Severe adverse events were those resulting in hospitalization, intravenous therapy, malignancies, life-threatening situations, or death. B cell depletion was defined as counts <0.01 × 109/liter and B cell return as counts ≥0.01 × 109/liter.

Treatment protocol.

Twenty-eight patients initially received rituximab, either 375 mg/m2/week for 4 weeks or 2 doses of 1 gm each (with a 2-week interval) and received further rituximab only at the time of clinical relapse (group A). Forty-five patients followed the routine repeat rituximab re-treatment regimen, which consisted of 2 rituximab doses of 1 gm each (with a 2-week interval) followed by a single 1-gm dose every 6 months for 2 years (total of 6 gm over 24 months), after which routine administration of rituximab was discontinued (group B).

Practice for the use of premedications and withdrawal of concomitant immunosuppression was similar for all groups; however, this was standardized for group B patients, in whom IV hydrocortisone 100 mg, IV chlorpheniramine 10 mg, and oral acetaminophen 1 gm were administered before the rituximab infusion. Other immunosuppressive agents were discontinued at the time of first the rituximab dose in this group. The dosage of oral glucocorticoids was reduced, aiming for withdrawal by 24 months. A total of 19 of 28 patients in group A began routine re-treatment after its introduction in 2006, with group A followup being censored in 2006, and all subsequent data reported in a third group (group C), rather than by their being included in group B.

Outcomes and compliance.

Outcomes were assessed as remission rates, DEI scores, and time to relapse. Compliance with the routine re-treatment protocol was assessed according to a reduction in the prednisolone dosage, withdrawal of concomitant immunosuppression, and timing of subsequent rituximab infusions.

Statistical analysis.

Statistical analysis was performed using SPSS version 18 software. Results are expressed as actual values and percentages for categorical variables and as medians and ranges for continuous variables. For missing data (e.g., if a patient did not attend clinic that month), the last observation carried forward method was used until the end of the individual patient's followup. The median interval between assessments was 3 months, with a median of 13 assessments per patient (range 1–27), and the total patient-months of followup was 3,782. Changes in variables, including the DEI score and the prednisolone dosage, were compared by Mann-Whitney U test. The proportions of patients, including those who were ANCA positive and those who experienced adverse events, were compared using chi-square tests. 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

Patient characteristics.

This study included 73 patients. Twenty-eight patients received an initial rituximab course, with further rituximab doses only if there was a clinical relapse (group A). Upon entry into group A, 6 of 28 patients (21%) received rituximab for consolidation of remission rather than for active disease, in light of prior repeated relapses or refractory disease. Forty-five patients received routine re-treatment for 2 years (group B). From 2006 onward, 19 of 28 patients in group A switched to routine re-treatment, either for treatment of a relapse (15 of 19 patients) or for maintenance of remission (4 of 19 patients) (group C). Those entering group C for remission maintenance were perceived to be at high risk of further relapse (n = 2) or they had ongoing disease activity that was not sufficient to meet the criteria for full relapse (n = 2).

Demographic characteristics were similar across the 3 groups (Table 1). The median age at the first rituximab dose was 52 years (range 16–77 years), 41% (30 of 73) were male, and the majority (61 of 73 [84%]) had GPA. The median disease duration prior to the first rituximab infusion was longer in group A (87 months [range 6–360 months]; P = 0.013) and group C (114 months [range 37–399 months]; P < 0.001) than in group B (43 months [range 8–220 months]), reflecting the fact that patients from groups A and C received their first rituximab course between 2002 and 2005.

Table 1. Characteristics and treatments in patients with AAV, by RTX treatment group*
 Group A (n = 28)Group B (n = 45)Group C (n = 19)Groups A and B (n = 73)
  • *

    Treatments in the antineutrophil cytoplasmic antibody–associated vasculitis (AAV) patients were as follows: group A received further rituximab (RTX) therapy only at the time of relapse; group B received routine RTX re-treatment with 2 doses of 1 gm each, followed by 1 gm at 6-month intervals for 2 years (total of 6 gm); and group C (19 patients from group A) began RTX re-treatment after a relapse of AAV. Followup in group C began at the start of re-treatment. All values are at/from the time of the first-ever RTX dose, except in group C, which are at/from the time of entry into the routine re-treatment protocol. GPA = granulomatosis with polyangiitis (Wegener's); MPA = microscopic polyangiitis; CYC = cyclophosphamide; anti-TNF = anti–tumor necrosis factor; IVIG = intravenous immunoglobulin; MMF = mycophenolate mofetil; ANCA = antineutrophil cytoplasmic antibody; PR3 = proteinase 3; MPO = myeloperoxidase; cANCA = cytoplasmic ANCA; pANCA = perinuclear ANCA; DEI = Disease Extent Index; BSA = body surface area.

  • One patient in group A and 1 patient in group C received tacrolimus. Three patients in group B received other immunosuppressive agents: abatacept in 1, antithymocyte globulin in 1, and leflunomide in 1.

  • One patient in group B died at 4 months and is included in the analysis. All others had at least 6 months of followup.

Age at first RTX dose, median (range) years53 (16–71)51 (16–77)61 (34–74)52 (16–77)
No. (%) male12 (43)18 (40)7 (37)30 (41)
Diagnosis, no. (%)    
 GPA20 (71)41 (91)15 (79)61 (84)
 MPA8 (29)4 (9)4 (21)12 (16)
Disease duration, median (range) months87 (6–360)43 (8–220)114 (37–399)57 (6–360)
Prior CYC therapy    
 No. (%) of patients27 (96)41 (91)18 (95)68 (93)
 Cumulative dose, median (range) gm29.0 (0–150)14.9 (0–130)30.0 (0–77.5)19.5 (0–150)
Prior therapies, no. (%)    
 Alemtuzumab5 (18)2 (4)5 (26)7 (10)
 Anti-TNF8 (29)5 (11)6 (31)13 (18)
 Azathioprine21 (75)33 (73)14 (74)54 (74)
 Gusperimus4 (14)9 (20)3 (16)13 (18)
 IVIG5 (18)12 (27)3 (16)17 (23)
 MMF22 (79)34 (76)14 (74)56 (77)
 Methotrexate10 (36)15 (33)7 (37)25 (34)
 Plasma exchange3 (11)7 (16)1 (5)10 (14)
 Other1 (4)3 (7)1 (6)4 (5)
No. of prior immunosuppressive agents (excluding corticosteroids), median (range)4 (2–8)3 (1–8)4 (2–7)4 (1–8)
Historical ANCA status, no. (%)    
 PR314 (50)34 (76)10 (52)48 (66)
 MPO4 (14)7 (16)2 (11)11 (15)
 cANCA6 (22)2 (4)4 (21)8 (11)
 pANCA2 (7)1 (2)2 (11)3 (4)
 Negative2 (7)1 (2)1 (5)3 (4)
Indication for RTX therapy, no. (%)    
 Active disease22 (79)42 (93)15 (79)64 (88)
 Consolidation of remission6 (21)3 (7)4 (21)9 (12)
No. of active organ systems at study entry, median (range)2 (0–4)2 (0–7)2 (0–3)2 (0–7)
DEI score at study entry, median (range)4 (0–8)3 (0–13)3 (0–6)4 (0–13)
ANCA status at study entry, no. (%)    
 PR39 (32)26 (57)7 (37)35 (47)
 MPO4 (14)3 (7)2 (11)7 (10)
 cANCA4 (14)6 (13)5 (11)10 (14)
 pANCA2 (8)3 (7)2 (4)5 (7)
 Negative9 (32)7 (16)3 (16)16 (22)
Followup, median (range) months18 (7–102)40 (4–61)55 (19–62)36 (4–102)
Response at 6 months, no. (%)    
 Complete remission23 (82)38 (85)17 (90)61 (84)
 Partial remission3 (11)5 (11)1 (5)8 (11)
 Treatment failure2 (7)2 (4)1 (5)4 (5)
RTX dose, median (range)    
 Total dose, gm2.75 (2.0–11.5)6 (2.0–11)7 (2–12)6 (2–12)
 Dose/BSA1.5 (0.96–6.0)3.35 (1.11–6.65)3.57 (1.27–6.52)2.97 (0.96–6.65)
 Dose/BSA/year1.11 (0.18–2.57)1.06 (0.49–3.36)0.86 (0.27–1.52)1.04 (0.18–3.36)

At the time of the first rituximab dose, a median of 2 organ systems (range 0–7) showed active disease in the entire cohort of patients. The distribution of organ involvement in the 73 patients was as follows: ear nose and throat (ENT) involvement in 41 patients (56%), lungs in 25 (34%), joints in 20 (27%), kidneys in 12 (16%), eyes in 9 (12%), skin in 9 (12%), peripheral nervous system (PNS) in 5 (7%), and central nervous system in 3 (4%).

The median rituximab dose/body surface area (BSA)/year was similar between groups: 1.11 gm/m2/ year (range 0.18–2.57) in group A, 1.06 gm/m2/year (range 0.49–3.36; P = 0.838 versus group A) in group B, 0.86 gm/m2/year (range 0.27–1.52; P = 0.392 versus group A) in group C, and 1.04 gm/m2/year (range 0.18–3.36) overall. Both the total rituximab dose and the total rituximab dose/BSA were greater in groups B and C (P < 0.001 for both variables for each comparison) than in group A. Given that the BSA/year values were similar between groups, these differences partly reflect the longer median followup in group B (40 months [range 4–61 months]; P = 0.038) and in group C (55 months [range 19–62 months]; P = 0.012) as compared to that in group A (18 months [range 7–102 months]) and are not solely due to participation in the routine re-treatment regimen.

Initial treatment response.

Rituximab therapy led to complete remission by 6 months in 23 of the 28 patients (82%) in group A, 38 of the 45 patients (85%) in group B, and 17 of the 19 patients (90%) in group C. Partial remission occurred in 3 patients (11%) in group A, 5 (11%) in group B, and 1 (5%) in group C. Only 2 patients (7%) in group A, 2 (4%) in group B, and 1 (5%) in group C experienced treatment failure. These patients had refractory GPA with progressive granulomatous disease manifestations: retroorbital granuloma in one and endobronchial, nodular, or cavitating lung lesions in the others. The DEI score fell from a median of 4 (range 0–13) at entry to 0 (range 0–8) at 6 months, which was sustained until 24 months in all groups (Figure 1A).

Figure 1.

Response to rituximab in patients with antineutrophil cytoplasmic antibody–associated vasculitis. A, Disease Extent Index at 0, 6, 12, 18, and 24 months after the first scheduled rituximab dose. B, Daily oral prednisolone dose at 0, 6, 12, 18, and 24 months after the first scheduled rituximab dose. Data in A and B 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 5th to the 95th percentiles. Solid symbols indicate outliers. C, Percentage of patients receiving an immunosuppressive agent in addition to rituximab (excluding corticosteroids) at 0, 6, 12, 18, and 24 months.

Concomitant therapies.

The median prednisolone dosage fell from 11 mg/day (range 0–45) at study entry to 5 mg/day (range 0–25; P = 0.012) at 24 months in group A, from 10 mg/day (range 0–40) to 2.75 mg/day (range 0–40; P < 0.001) in group B, and from 5 mg/day (range 0–20) to 1 mg/day (range 0–10; P = 0.024) in group C (Figure 1B). Glucocorticoids were withdrawn in 6 of the 28 patients (21%) in group A, 17 of the 45 patients (38%) in group B, and 9 of the 19 patients (47%) in group C by the 24-month time point. The majority of patients were receiving immunosuppressive agents at study entry: 26 (93%) in group A, 40 (89%) in group B, and 19 (100%) in group C. Immunosuppressive agents were withdrawn at the time rituximab was begun in 16 of 26 patients (62%) in group A, 38 of 40 patients (95%) in group B, and 19 of 19 patients (100%) in group C. At 24 months, 1 of 40 patients (3%) in group B was receiving intravenous immunoglobulin (IVIG), 2 of 18 patients (11%) in group C were receiving azathioprine (1 patient) and IVIG (1 patient) (Figure 1C), and 6 of 13 patients (46%) in group A were receiving immunosuppressive agents.

Adherence to rituximab regimen.

Adherence to the routine re-treatment regimen was satisfactory, with 82% of rituximab doses in group B and 85% in group C given between 5 and 7 months from the previous dose. Eight patients in groups B and C had 1 of their rituximab doses administered earlier than 6 months: to hasten the achievement of complete remission in 5 whose disease was in partial remission, to treat relapse in 1, and for an unclear indication in 2.

Two-year routine re-treatment was completed by 38 of the 45 patients (84%) in group B and 17 of the 19 patients (89%) in group C. Failure to complete was due to patient withdrawal for treatment failure (2 in group B and 1 in group C), death (3 in group B), or insufficient followup (2 in group B and 1 in group C; median of 19 months). After 24 months of routine re-treatment, 4 of 17 patients (24%) in group C continued to receive rituximab for a perceived high risk of relapse.

Relapse rates.

During the 24-month routine re-treatment period, 5 of the 43 patients (12%) in group B who had achieved remission and 2 of the 18 patients (11%) in group C experienced relapse. In comparison, 19 of 26 patients (73%) in group A had a relapse within 24 months (P < 0.001 versus group B and versus group C). At 48 months, relapses had occurred in 21 of 26 patients (81%) in group A, as compared to 11 of 43 patients (26%) in group B (P < 0.001) and 7 of 18 patients (39%) in group C (P < 0.001) (Figure 2). By the end of followup, relapses had occurred in 22 of 26 patients (85%) in group A, as compared to 11 of 43 patients (26%) in group B (P < 0.001) and 10 of 18 patients (56%) in group C (P = 0.001). The median time to first relapse was 12 months (range 5–76 months) in group A, 29 months (range 5–48 months) in group B, and 34.5 months (range 5–53 months) in group C.

Figure 2.

Relapse-free survival rates up to 48 months in patients with antineutrophil cytoplasmic antibody–associated vasculitis treated with rituximab. Red lines represent group A, green lines group B, and blue lines group C. Numbers under each graph represent the number of patients for whom followup information was available at each time point, with censoring at the end of followup or at the time of treatment failure. Vertical broken line represents the end of the 24-month fixed-interval rituximab re-treatment protocol. P < 0.001 for group A versus group B and versus group C, by log rank analysis.

After the 24-month treatment period, 38 patients in group B who had completed 24 months of routine re-treatment were followed up for a further 15.5 months (range 0–42 months) since the last rituximab dose. In group C, 17 patients completed 24 months of routine re-treatment, and followup data after the last rituximab dose were available for 13 of them for a further 31 months (range 0–44 months); the remaining 4 patients continued rituximab beyond 24 months. After the end of the 24-month re-treatment period, 8 of 38 patients (21%) in group B and 8 of 13 patients (62%) in group C had a relapse. Relapses during this period were less frequent in group B (P = 0.002) and group C (P = 0.028) patients as compared to the period following the initial rituximab course in group A patients.

Management of relapse.

Group A.

Two patients experienced treatment failure in group A. Thirty-four relapses occurred in 22 of 26 patients (85%) (Figure 3A). Of 29 relapses for which there was detailed information, 15 were major and 14 were minor. Organ involvement at the time of relapse was the ENT for 15 of the 29 relapses (52%), the joints for 10 (34%), the lungs for 8 (28%), the eyes for 6 (21%), the kidneys for 4 (14%), the skin for 3 (10%), and the PNS for 1 (3%). Of the 22 patients who experienced a relapse, 19 began the 24-month routine re-treatment protocol (forming group C).

Figure 3.

Management of relapses in patients with antineutrophil cytoplasmic antibody–associated vasculitis treated with rituximab (RTX). A, Management of the first relapse during the initial 24 months of followup in group A. Two or more relapses occurred in 6 of 24 patients. B, Management of the first relapse during the 24-month routine rituximab re-treatment regimen in group B. A total of 38 of 43 patients completed the re-treatment protocol, and 8 of 38 patients subsequently experienced a relapse. C, Management of the first relapse during the 24-month routine rituximab re-treatment regimen in group C. A total of 17 of the 19 patients completed the re-treatment protocol, and 8 of 13 patients then experienced a relapse. Four patients continued on the protocol because of a perceived high risk of relapse.

Group B.

Two patients experienced treatment failure in group B. Of the remaining 43 patients, 5 (12%) experienced a relapse within 24 months, but the disease was controlled by continuing the treatment regimen (Figure 3B). After the 24-month protocol, 8 of 38 patients (21%) subsequently experienced relapse; 6 of them restarted rituximab, 1 started methotrexate, and 1 started azathioprine. Six of the 13 relapses were deemed to be major. Organ involvement at the time of relapse was the ENT for 11 of the 13 relapses (85%), the joints for 4 (31%), the lungs for 7 (54%), and the eyes for 1 (8%). Three patients experienced >1 relapse.

Group C.

One patient experienced treatment failure in group C. Relapses occurred in 2 of 18 patients (11%) during the 24-month protocol but were controlled by continuing the treatment regimen (Figure 3C). Relapses after the 24-month protocol occurred in 8 of 13 patients (62%), of whom 7 restarted rituximab and 1 started mycophenolate mofetil. Four relapses were major. Organ involvement at the time of relapse was the ENT for 7 of 10 relapses (70%), the joints for 4 (40%), the PNS for 2 (20%), the eyes for 2 (20%), the lungs for 1 (10%), and the kidneys for 1 (10%).

Laboratory findings.

B cell depletion.

B cell depletion occurred in all patients following the first rituximab course. Transient return of B cells (B cell count ≥0.1 × 109/liter) was documented in 11 of the 45 patients (24%) in group B and 13 of the 19 patients (68%) in group C at least once during the routine re-treatment and tended to occur just prior to the next rituximab dose, after which further depletion occurred. At the end of followup, 19 of the 28 patients (68%) in group A, 18 of the 45 patients (40%) in group B, and 8 of the 19 patients (42%) in group C had a return of B cells, with a median time to return of 11 months after the last rituximab dose (range 2–30 months) in group A, 7.5 months (range 1–32 months) in group B, and 7.5 months (range 1–14 months) in group C. B cell counts had been performed at the time (or within the preceding 3 months) of relapse in 10 of 10 patients (100%) in group B; 4 of them (40%) had detectable B cells. Of the first relapses in group A, 18 of 22 patients (82%) had B cell counts performed at the time of relapse, with detectable B cells in 12 of them (67%). In group C, 9 of 10 patients (90%) had B cells measured at the first relapse, and B cells had returned in 5 of them (56%).

ANCA.

A history of ANCA positivity was noted in 69 of the 73 study patients (95%). At the first rituximab infusion, ANCA positivity by both indirect immunofluorescence (IIF) and enzyme-linked immunosorbent assay (ELISA) was noted in 13 of 28 patients (46%) in group A, 29 of 45 (64%) in group B, and 9 of 19 (47%) in group C (Table 1). ANCA positivity by IIF was identified in 19 patients (68%) in group A, 38 (84%) in group B, and 16 (84%) in group C. Remission and relapse rates were similar between the ANCA-positive group and the ANCA-negative group at the first rituximab infusion. Between 0 and 24 months after beginning rituximab, the proportion positive by IIF and ELISA fell (P = 0.894 for group A, P = 0.002 for group B, and P = 0.005 for group C). At the time of relapse, 19 of 57 patients (33%) were ANCA positive by IIF and ELISA; 12 of the 57 patients (21%) were persistently positive, and 7 (12%) experienced seroconversion from ANCA negative to ANCA positive before the relapse occurred. ANCA status at the time of relapse did not differ between groups. ANCA negativity at the time of relapse was identified by both IIF and ELISA in 67% and by IIF alone in 54% of the group B and group C patients.

Safety.

Severe adverse events.

Severe adverse events occurred in 9 of 28 patients (32%) in group A (16 events), 21 of 45 (47%) in group B (45 events), and 7 of 19 (37%) in group C (20 events) (P = 0.326 for group A versus group B and P = 0.984 for group A versus group C) (Table 2). Four patients died at 3, 7, 12, and 59 months, respectively, after the last rituximab infusion. Three were over the age of 70 years and had extensive comorbid conditions. Causes were intraabdominal sepsis in 1 patient, chest sepsis associated with chronic kidney disease and ischemic heart disease in 1 patient, and unknown cause in 2 patients (1 of which occurred 7 months after the patient was transferred to another hospital). Severe infections occurred in 6 of 28 patients (21%) in group A (10 events), 12 of 45 (27%) in group B (30 events), and 5 of 19 (26%) in group C (14 events) (P = 0.822 for group A versus group B and P = 0.970 for group A versus group C) (Table 2).

Table 2. Severe adverse events, by RTX treatment group
 Group A (n = 28)Group B (n = 45)Group C (n = 19)
  • *

    Thirteen episodes of chest infection occurred in 2 patients.

  • A patient who underwent renal transplantation 32 months following rituximab (RTX) therapy developed both symptomatic cytomegalovirus viremia (at 7 months) and Pneumocystis jiroveci pneumonia (at 11 months) posttransplantation.

  • Other hospitalizations included the following: in group B, 2 separate admissions for hyponatremia and hypoglycemia in 1 patient, 2 separate admissions for management of leg ulceration and a below the knee amputation for peripheral vascular disease in 1 patient, and 2 admissions for diverticulitis in 1 patient.

  • §

    In Group B, a 71-year-old woman was diagnosed as having peritoneal adenocarcinoma 29 months after the first RTX dose. In Group C, a 52-year-old woman was diagnosed as having breast cancer 52 months after the first RTX dose.

  • Causes of death were as follows: in group A, intraabdominal sepsis in 1 patient, and in group B, chest sepsis associated with chronic kidney disease and ischemic heart disease in 1 patient, and unknown cause in 2 patients (1 of which occurred 7 months after care was transferred to another hospital).

All severe adverse events   
 No. of adverse events164520
 No. (%) of patients9 (32)21 (47)7 (37)
All serious infections, no. of events103014
 Chest924*10
 Skin020
 Campylobacter020
 Cytomegalovirus001
 Pneumocystis jiroveci pneumonia001
 Swine influenza virus A H1N1011
 Other (not specified)111
All other hospitalizations, no. of events5115
 Fracture/joint replacement353
 Renal transplant/dialysis201
 Other061
Cancer§011
Death130

Immunoglobulin levels.

At the time of the first rituximab dose, 7 of 28 patients (25%) in group A and 8 of 45 patients (18%) in group B had IgG levels <7 gm/liter (the lower limit of normal). Upon initiation of routine re-treatment, 5 of 19 patients (26%) in group C had IgG levels <7 gm/liter. The median IgG level at study entry was 10.0 gm/liter (range 3.9–14.8) in group B and 8.2 gm/liter (range 4.2–17.9) in group C (P = 0.339). After 24 months, IgG levels <7 gm/liter occurred in 13 of 40 patients (33%) in group B and 8 of 18 patients (44%) in group C, with median levels of 8.25 gm/liter (range 3.2–18.8; P = 0.046) and 7.0 gm/liter (range 2.6–11.5; P = 0.056), respectively. By 36 months, median IgG levels in group B were 8.35 gm/liter (range 3.4–13.7; P = 0.089), with 10 of 28 patients (36%) having IgG levels <7 gm/liter, whereas in group C, there was a further trend for IgG levels to decline, with a median of 6.55 gm/liter (range 2.8–12.5; P = 0.011) and with 12 of 18 patients (67%) having IgG levels <7 gm/liter.

At entry, IgM levels <0.3 gm/liter (lower limit of detection) were seen in 2 of 45 patients (4%) in group B and 2 of 19 patients (11%) in group C. By 24 months, low IgM levels were present in 8 of 40 patients (20%) in group B and 6 of 18 patients (33%) in group C. Similar to trends in IgG levels between 24 and 36 months, IgM levels in group B trended toward an increase, whereas the levels remained low in group C, with IgM <0.3 gm/liter in 3 of 27 patients (11%) in group B and in 6 of 18 patients (33%) in group C.

DISCUSSION

Rituximab is effective for the induction of remission in patients with AAV (6, 7). Nevertheless, most patients with refractory/relapsing disease experience another relapse after rituximab therapy (8, 13, 14). Routine rituximab re-treatment was associated with a reduction in relapse rates as compared to single rituximab courses and allowed early withdrawal of immunosuppression and a reduction in the glucocorticoid dosage or withdrawal. Beyond 2 years and after discontinuation of rituximab, further relapses were seen in 25% of the patients after a longer period of remission than that seen after a single rituximab course.

There have been concerns regarding the risk of infection, loss of humoral immunity, and cost implications with rituximab re-treatment in AAV (15). Nevertheless, the use of rituximab for refractory/relapsing AAV is becoming widespread, and maintenance strategies are required. Potential approaches include the decision to give further rituximab at the time of relapse, to base re-treatment on the levels of biomarkers, ANCA levels, or peripheral B cell counts, or to follow a routine, fixed-interval re-treatment regimen. Waiting for clinical relapse exposes patients to risks of further disease activity and increased glucocorticoid exposure; we have not found ANCA or B cell counts to be sufficiently reliable to guide repeat rituximab dosing (8). Appreciating that some patients may receive more immunosuppression than is absolutely needed, as already occurs with current remission maintenance strategies using azathioprine or methotrexate, we believe that in the absence of better biomarkers, routine re-treatment is the only strategy likely to be effective. Should improved biomarkers become available, closer adjustment of drug dosing to the individual patient's need will be achievable.

The therapeutic mechanism of rituximab in AAV is mediated by depletion of autoreactive B cells that appear to support T cell autoreactivity (16). B cells at sites of active inflammation are relatively protected from depletion and can persist after rituximab therapy despite peripheral blood depletion (17, 18), thus providing the nucleus for a subsequent relapse. We suspect this is the case for many patients with GPA who have granulomatous lesions. Indeed, in RA and systemic lupus erythematosus, a memory B cell phenotype (CD27high), as compared to a naive B cell phenotype, in the re-constituting B cell blood pool is associated with an increased likelihood of relapse (19, 20). Repeating rituximab therapy when the disease is inactive is more likely to remove persisting autoreactive cells, thereby securing longer lasting remissions and increasing the chance of cure through re-establishment of physiologic immune regulation. Thus, the aim of our routine re-treatment regimen was both the reduction in the risk of relapse during the regimen and the reduction of the subsequent risk of relapse after rituximab withdrawal.

Our own data in AAV suggested that rituximab re-treatment of relapse effectively induces second and subsequent remissions, and exposure to repeated rituximab courses, albeit at varying time intervals, did not highlight new safety concerns (8). Rhee et al (13) repeated rituximab at 4-month intervals in 39 AAV patients, and short-term results demonstrated low relapse rates during treatment. In RA, 2 courses of rituximab given 6 months apart resulted in improved and sustained efficacy at 1 year as compared to a single course, with a similar safety profile (9, 21).

We adopted a 6-month time interval for rituximab re-treatment for our maintenance regimen, as the median time to relapse post-rituximab has been shown to be 12 months, with the earliest relapses occurring at 6 months (8). Given the improvements in disease control observed following second rituximab courses in RA patients (9, 21), we proposed that in AAV, a 6-monthly 1-gm rituximab dose would be sufficient for relapse prevention without increasing the annual rituximab exposure. For newly diagnosed AAV, remission induction/maintenance regimens are typically of 2 years' duration, after which cessation of therapy is considered (22).

Because of the retrospective nature of our study, our results have limitations and are open to bias. Group A comprised the earliest patients to receive rituximab, and they had a longer disease course compared to group B. The inclusion of group C, a subset of group A receiving the same regimen as group B, acted as a control for group B. There could have been positive or negative bias for relapse in group C patients due to their greater previous cumulative rituximab exposure or the self selection of a subgroup with high risk of relapse. The low risk of relapse in group C patients, which was similar to that in group B, strengthens our conclusions concerning the efficacy of routine rituximab re-treatment. Adherence to the protocol was good, but events such as the continuation of repeat rituximab dosing in some group C patients and the re-introduction of immunosuppressive agents are possible confounding factors. However, in the analysis, we took such confounders into account, calculating relapse rates following rituximab withdrawal from the time of the final dose instead of for the 24-month time point and focusing on first relapses, which were unaffected by the re-introduction of immunosuppressive agents at the time of relapse.

Beyond 24 months, relapses occurred following rituximab withdrawal, but at lower rates and with longer time to relapse than was seen following a single rituximab course in group A. Our results support the hypothesis that rituximab re-treatment during the remission phase has prolonged benefits after completion of the re-treatment period, being mediated by a more-complete depletion of persisting autoreactive B cells. This now requires confirmation in a prospective trial.

Most patients had a relapse after a single rituximab course, necessitating further multiple rituximab courses (group A), and despite the re-treatment regimen, 12% had a relapse within the 2-year treatment period (group B), mainly within the 2 months preceding a dose of rituximab, suggesting that the majority of re-dosing in our cohort was necessary. Because the majority of relapses in groups B and C were controlled by further rituximab, higher rituximab dosing during the maintenance phase is likely to further reduce relapse rates. Frequently, patients were ANCA negative and B cell depleted at the time of relapse, confirming our previous findings that ANCA levels and B cell counts lack sufficient sensitivity to predict relapse. The potential dangers of higher rituximab exposure need to be balanced against the benefits of glucocorticoid and immunosuppressive withdrawal and improved disease control.

Annual rituximab exposure was not greater in patients receiving routine re-treatment as compared to those who were re-treated based on relapse, although cumulative rituximab exposure was greater in groups B and C, reflecting longer followup. Numerically more-severe infections occurred in patients undergoing routine re-treatment; this was also attributable to longer followup. The overall proportions of patients with severe infections were similar between groups. In RA, stable rates of infection have been observed with up to 5 courses of rituximab in 2,578 trial patients. The frequency of severe infections was similar to that in other studies of refractory/relapsing AAV conducted with mycophenolate mofetil (23), alemtuzumab (24), and deoxyspergualin (25) at our center. We did not routinely prescribe antibiotic, antiviral, or fungal prophylaxis, and no unusual trends in infections were identified. However, Pneumocystis jiroveci has been reported following rituximab, and the Food and Drug Administration approval of rituximab for the treatment of GPA/MPA recommends prophylaxis. Low IgG levels were present in 21% of the patients at baseline, reflecting prior immunosuppression, and there was a trend toward lower immunoglobulin levels with the repeat rituximab dosing. In this study, low IgG levels at baseline and during followup were not clearly associated with increased infections. Infection rates are higher in AAV compared to RA, and factors such as chronic lung disease, prior immunosuppressive load, and concomitant glucocorticoids may have a greater influence on infection risk.

In conclusion, a 2-year, fixed-interval routine rituximab re-treatment regimen was associated with lower relapse rates as compared to a single rituximab course, both during the 2-year treatment regimen and in the early phase following rituximab withdrawal. Reduction of the glucocorticoid dosage and withdrawal of immunosuppressive agents was possible, and no new safety issues were identified. Thus, in the absence of biomarkers that accurately predict relapse, routine rituximab re-treatment appears to be an effective strategy for the maintenance of remission in patients with refractory/relapsing AAV.

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. R. M. Smith 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. R. M. Smith, Jones, Laurino, K. G. C. Smith, Jayne.

Acquisition of data. R. M. Smith, Jones, Guerry, Laurino, Catapano, Chaudhry, K. G. C. Smith, Jayne.

Analysis and interpretation of data. R. M. Smith, Jones, Laurino, Chaudhry, K. G. C. Smith, Jayne.

Acknowledgements

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

Ancillary