To determine the safety and efficacy of additional courses of rituximab in patients with rheumatoid arthritis (RA).
To determine the safety and efficacy of additional courses of rituximab in patients with rheumatoid arthritis (RA).
An open-label extension analysis of RA patients previously treated with rituximab was conducted. Patients who had participated in any of 3 double-blind trials were eligible for additional courses (2 infusions of 1,000 mg given 2 weeks apart) if they exhibited a swollen joint count and tender joint count of ≥8 with ≥16 weeks elapsing after the previous course. Safety was assessed in patients receiving all or a portion of a rituximab course. Efficacy was assessed 24 weeks after each course, using the American College of Rheumatology 20% criteria for improvement (ACR20), ACR50, ACR70, European League Against Rheumatism (EULAR) response criteria, Disease Activity Score in 28 joints, the disability index of the Health Assessment Questionnaire, and Short Form 36 scores, stratified according to prior tumor necrosis factor (TNF) inhibitor exposure.
A total of 1,039 patients received ≥1 course of rituximab. Of these, 570 received 2 courses, 191 received 3 courses, and 40 received 4 courses, for a total of 1,669 patient-years. Irrespective of prior TNF inhibitor exposure, ACR20 responses were comparable at week 24 after course 1 and at week 24 after course 2 (65% versus 72%), as were ACR50 and ACR70 responses. EULAR moderate/good responses were also comparable in course 2 relative to course 1 (88% versus 79%), with EULAR remission occurring in a 2-fold higher proportion of patients after course 2 than after course 1 (13% versus 6%). The most common adverse events, which were mild-to-moderate acute infusion-related events, decreased with each course. The serious infection rate after course 1 (5.1 per 100 patient-years) remained stable through additional courses. The proportion of patients with circulating IgM and IgG levels below the lower limit of normal (LLN) increased with subsequent courses; however, serious infection rates in these patients (5.6 per 100 patient-years in patients with low IgM levels and 4.8 per 100 patient-years in patients with low IgG levels were comparable with those in patients with immunoglobulin levels above the LLN (4.7 per 100 patient-years). Patients with human antichimeric antibody (9.2%) did not exhibit decreasing efficacy or present additional safety concerns.
These findings indicate that patients treated with repeated courses of rituximab have sustained clinical responses with no new adverse events.
Although the precise cause of rheumatoid arthritis (RA) remains unknown, our improved understanding of its pathogenesis, combined with advances in biotechnology, has led to innovative therapies that target the pathogenetic elements of the disease. Specifically, significant progress in characterizing the cytokine networks responsible for inflammation in RA (1) has generated therapies that target tumor necrosis factor (TNF) and interleukin-1 (2). However, the need for additional effective therapies is highlighted by the failure of existing therapies to achieve clinically relevant improvement in up to 40% of patients (3–6).
Previous studies have focused on the role of B cells in the immunopathogenesis of RA, including autoantibody secretion, autoantigen presentation, proinflammatory cytokine production, and regulation of dendritic cell function (7–12). Accordingly, B lymphocytes have become an important therapeutic target in RA (8). One approach to targeting B cells in RA is the use of rituximab, a genetically engineered chimeric monoclonal antibody that selectively depletes peripheral B lymphocytes by binding CD20 on the cell surface and initiating cell-mediated and complement-dependent cytotoxicity, as well as promotion of apoptosis (13–15). Since CD20 is highly expressed on B cells but not on stem cells or plasma cells (16), B cell regeneration is not directly compromised (17). In 1997, rituximab was approved by the Food and Drug Administration for the treatment of patients with non-Hodgkin's lymphoma. Recently, rituximab in combination with methotrexate (MTX) has been approved for patients with RA who have had an inadequate response to one or more TNF inhibitors.
The efficacy and safety of a single treatment course of rituximab in RA have been demonstrated in 3 double-blind, placebo-controlled clinical trials. An initial phase IIa trial demonstrated that a single course of two 1,000-mg infusions of rituximab, alone or in combination with either cyclophosphamide or continued MTX, had significantly improved disease symptoms by weeks 24 and 48 (17, 18). A larger 24-week phase IIb trial in 465 patients with an inadequate response to disease-modifying antirheumatic drugs (DMARDs) further showed that rituximab (either 2 doses of 1,000 mg or 2 doses of 500 mg) in combination with MTX was effective and well tolerated (19). More recently, a pivotal phase III trial (Randomized Evaluation of Long-Term Efficacy of Rituximab in RA [REFLEX] Trial) in 517 patients with active RA who had an inadequate response to prior TNF inhibitor therapy demonstrated that a single course of rituximab (2 infusions of 1,000 mg given 2 weeks apart) in combination with MTX led to statistically significant and clinically meaningful improvement in disease activity at 24 weeks (20).
Since RA is a chronic disease, long-term use of treatments that target TNF or B cells will be required for continued disease control, and assessment of the safety and efficacy of repeated courses of therapy is critical. To assess the continued safety and efficacy of repeated rituximab treatment, patients from the initial randomized controlled trials were evaluated in a long-term, open-label analysis. Here, we present the aggregate safety and efficacy findings in patients who received ≥2 courses of rituximab.
The patient population was selected from those who received ≥1 course of rituximab in the phase IIa, phase IIb, or phase III trials (17, 18, 20), including patients who were initially randomized to receive rituximab in a double-blind setting and those who received open-label rituximab (either as rescue therapy, or within an open-label extension protocol). A total of 1,039 patients (the all-exposure population) received rituximab, with ∼10–15% of patients receiving an initial rituximab treatment during the open-label phase. To be eligible for a second course of rituximab, patients had to have demonstrated a ≥20% reduction in both the swollen joint count and the tender joint count at any visit 16 weeks after initial treatment or later and have active disease, defined as a swollen joint count of ≥8 (in 66 joints) or a tender joint count of ≥8 (in 68 joints). Repeat courses of treatment were administered at the investigator's discretion. Patients were eligible for subsequent courses, after the second course, if they had active disease as described, with a minimum interval between treatment courses of 16 weeks. Patients who were originally treated with rituximab in the REFLEX trial were not re-treated until week 24.
Subsequent rituximab courses were administered as two 1,000-mg intravenous (IV) infusions on days 1 and 15 following premedication with 100 mg IV methylprednisolone, together with a short course of oral glucocorticoids on the intervening days (60 mg/day on days 2–7 and 30 mg/day on days 8–13 of each treatment course). Patients received stable background MTX (10–25 mg/week). No other DMARDs were allowed before starting open-label rituximab treatment. Patients were permitted to take glucocorticoids (≤10 mg/day of prednisone or equivalent) and oral nonsteroidal antiinflammatory drugs, both of which remained at a stable dose throughout the study. In addition, all patients continued to receive a stable dose of ≥5 mg/week folic acid or equivalent, given in either a single or divided daily dose.
Patients were excluded from receiving a second course of rituximab if they received concurrent treatment with any DMARD other than stable MTX. Patients did not receive a subsequent course if, at the time of re-treatment, they had developed known contraindications to high-dose glucocorticoids (820 mg of prednisone or equivalent), an active infection, or significant abnormal laboratory values for serum creatinine level, aspartate aminotransferase level, alanine aminotransferase level, platelet count, hemoglobin value, neutrophil count, IgG level, or IgM level.
The trials that form the basis of the analyses presented here were carried out in accordance with the Declaration of Helsinki. All participating sites received approval from their governing institutional review board or equivalent, and all patients provided voluntary written informed consent.
American College of Rheumatology 20% responder status (ACR20) was determined 24 weeks after each treatment course (21). Additional end points included the proportion of patients with ACR50 and ACR70 responses, the proportion of patients with a good or moderate response according to European League Against Rheumatism (EULAR) response criteria (22), the change in Disease Activity Score in 28 joints (DAS28) (23) from original baseline (prior to initial infusion), the change in the disability index (DI) of the Health Assessment Questionnaire (HAQ) (24), and the Medical Outcomes Study Short Form 36 (SF-36) (25) physical and mental component summary scores from original baseline to week 24. The minimum clinically important difference in the DI of the HAQ is defined as an improvement of ≥0.22 from baseline (26). All end points were assessed 24 weeks after each treatment course and were calculated relative to patients' initial baseline values.
Efficacy analyses were based on observed data, and compared responses to course 1 and course 2 in patients who had received all or part of ≥2 rituximab courses and who had been followed up for ≥24 weeks after each rituximab course. In this way, responses to course 1 and course 2 were compared within the same patients. Efficacy responses were calculated from the patients' original study baselines (prior to the first infusion) to the indicated time point. For the categorical end points of ACR and EULAR response criteria, the number of patients receiving course 3 was too low for meaningful analysis; however, mean DAS28 scores for course 3 were calculated. The safety data were summarized by treatment course in the all-exposure population, which included any patient who received ≥1 rituximab infusion. Both the efficacy and the safety study populations were stratified according to prior exposure to TNF inhibitors.
The Common Terminology Criteria for Adverse Events, version 3 (27), was used to grade and record adverse events (AEs), which were monitored throughout the study and followed up until resolution. AEs were coded by system organ class and preferred terms, using the Medical Dictionary for Regulatory Activities, and by severity and relationship to the study medication. Serious or potentially serious AEs were summarized separately. A serious AE was defined as an event that was fatal, was immediately life-threatening, required inpatient hospitalization or prolongation of an existing hospitalization, resulted in persistent or significant disability or incapacity, was a congenital anomaly or birth defect, was medically significant, or required intervention to prevent 1 or more of these outcomes. A Common Terminology Criteria for Adverse Events grade 3 was defined as a severe AE and grade 4 as a life-threatening or disabling AE (28). Serious infections were defined as infections that were considered serious according to the above definition or required treatment with IV antibiotics.
Plasma immunoglobulin concentrations and human antichimeric antibody (HACA) titers were measured at each visit, approximately every 8 weeks. The normal values for IgG, IgM, and IgA levels ranged from 5.2 to 14.4, 0.5 to 3.0, and 0.5 to 4.0 gm/liter, respectively; thus, the lower limit of normal (LLN) was defined as 0.5 gm/liter for IgM and IgA and 5.2 gm/liter for IgG. HACA positivity was defined as a level of >5 reference units/ml, immunodepletable following addition of rituximab. CD19 was used as a surrogate marker for the evaluation of B cell kinetics following treatment with rituximab, since rituximab interferes with CD20 analysis by flow cytometry.
Statistical analyses of the trials have been published previously (17, 19, 20). All efficacy and safety data were reported by patient. Briefly, results were expressed as descriptive statistics, proportions, and corresponding 95% confidence intervals (95% CIs) of the difference between patient groups and treatment courses. The SEM for DAS28 was calculated at the indicated time points.
This study included a total of 1,039 patients (all-exposure population) who received ≥1 course of rituximab within the clinical program. Of these, 570 patients received 2 courses, 191 patients received 3 courses, 40 patients received 4 courses, and 3 patients received 5 courses (Figure 1). Because of the limited number of evaluable responses at the time of data cutoff, clinical efficacy results are presented through course 2 (through course 3 for DAS28), and safety results are presented through course 4. For the all-exposure safety assessment, 839 patients were followed up for >1 year, 139 for >2 years, and 89 for >3 years after initial rituximab treatment, corresponding to 1,669 patient-years.
Of the 238 patients (22.9% of the study population) who withdrew between course 1 and course 2, 112 withdrew due to an insufficient therapeutic response. The remaining withdrawals were due to safety/tolerability problems, deaths, and other reasons, including administrative reasons and loss to followup (Figure 1). In subsequent courses, the proportion of patients who withdrew decreased to 6.0% between course 2 and course 3 and to 1.6% between course 3 and course 4. Of the patients continuing in the study after the first course of rituximab, 231 of 801 (28.8%), had not yet received a second course. Similarly, following a second course, 345 of 536 patients (64.4%) had not yet received a third course, and following a third course, 148 of 188 (78.7%) had not yet received a fourth course.
The original baseline demographic and disease characteristics of the 570 patients who received at least 2 courses of rituximab were similar to those of the all-exposure population (1,039 patients) (Table 1). The 2 groups were well balanced; the majority of patients who received a second course of rituximab were women (80%), and the mean age at screening was 51.7 ± 11.4 years. Patients had active and prolonged DMARD-refractory disease (mean disease duration 10.8 ± 7.9 years) at enrollment. Of the 1,039 patients in the all-exposure population, 427 patients (41%) had never taken TNF inhibitors, while the remaining 612 patients (59%) had prior TNF inhibitor exposure. In comparison, of the patients who received a second course of rituximab, 255 (45%) had never taken TNF inhibitors and 315 (55%) had previous exposure. Overall, patients who received a second course of rituximab had been treated with a mean of 2.6 ± 1.6 DMARDs (excluding MTX) and a mean of 0.8 ± 0.9 anti-TNF inhibitors. In the 570 patients who received at least 2 courses of rituximab, the mean disease activity was lower prior to the second course of treatment than at original baseline.
|Characteristic||All-exposure population (n = 1,039)||Patients receiving ≥2 courses (n = 570)†|
|Age, years||51.9 ± 11.9||51.7 ± 11.4|
|Sex, no. (%) female/male||830 (80)/209 (20)||457 (80)/113 (20)|
|Disease duration, years||11.2 ± 8.2||10.8 ± 7.9|
|No. of prior DMARDs taken‡||2.5 ± 1.6||2.6 ± 1.6|
|No. of prior anti-TNF medications taken||0.9 ± 0.9||0.8 ± 0.9|
|Ever taken TNF inhibitors, no. (%)||612 (59)||315 (55)|
|Swollen joint count (66 joints)||22.0 ± 11.6||18.13 ± 10.4§|
|Tender joint count (68 joints)||33.5 ± 15.4||28.76 ± 14.4§|
|DAS28||6.8 ± 1.0||6.25 ± 1.12§|
The number of patients who could be evaluated for efficacy outcomes was dependent on the duration of followup after course 2, since ACR and EULAR responses were determined 24 weeks after course 1 and course 2. Of the 570 patients receiving course 2, both ACR and EULAR response data after 24 weeks were available in a total of 254 patients (155 who had been exposed to TNF inhibitors and 99 who had never taken TNF inhibitors). EULAR response data were available in 3 additional patients, all of whom had been exposed to TNF inhibitors.
In patients who had received ≥2 courses of treatment and had been followed up for 24 weeks after each course (n = 254), the proportion who met the ACR20 criteria at week 24 after course 2 was comparable with response rates after course 1, irrespective of prior TNF inhibitor exposure (65% of patients with prior TNF inhibitor exposure who completed course 1 and 72% of patients with prior TNF inhibitor exposure who completed course 2 met the criteria; 59% of patients without prior TNF inhibitor exposure who completed course 1 and 73% of patients without prior TNF inhibitor exposure who completed course 2 met the criteria) (Figures 2A and B). ACR50 and ACR70 response rates were also comparable in course 1 and course 2, irrespective of prior TNF inhibitor exposure (Figures 2A and B). An ACR70 response was achieved in 12% of the patients with prior TNF inhibitor exposure who completed course 1 versus 21% of the patients with prior TNF inhibitor exposure who completed course 2 (95% CI difference 2.5, 15.6) and 9% of the patients without prior TNF inhibitor exposure who completed course 1 versus 19% of the patients without prior TNF inhibitor exposure who completed course 2 (95% CI difference 1.9, 18.3).
Mean DAS28 scores decreased from the original baseline after the first treatment course and were generally still lower before course 2 was initiated (Figure 3). In 81 patients who had previously been exposed to TNF inhibitors and received 3 courses of rituximab, the mean DAS28 scores immediately prior to the initiation of course 1, course 2, and course 3 were 7.07, 6.17, and 6.01, respectively. Irrespective of prior TNF inhibitor exposure, changes from the original DAS28 baseline score to week 24 were maintained with each rituximab course (data not shown).
The proportion of patients who met the EULAR response criteria was comparable between course 1 and course 2, irrespective of prior TNF inhibitor exposure (Figures 2C and D). The proportions of patients with low disease activity (DAS28 ≤3.2) and whose disease was in remission (DAS28 <2.6) increased following 2 treatment courses. In patients who had prior TNF inhibitor exposure, between course 1 and course 2, 95% CI differences were (1.6, 16.4), (−4.2, 18.6), and (1.1, 12.8) for moderate/good responses, low disease activity, and remission, respectively. The proportion of patients with prior TNF inhibitor exposure whose disease was in remission, according to DAS28 score, 24 weeks following course 2 (13%) was 2-fold higher than at the same time point following course 1 (6%) (95% CI difference 1.1, 12.8). A similar increase in the proportion of patients whose disease was in remission was observed at 24 weeks following course 2 compared with course 1 in patients who had never taken TNF inhibitors (remission was achieved in 14% after course 1 and 8% after course 2) (95% CI difference −2.3, 14.4).
At 24 weeks, within-patient comparison assessments of patients' physical function using the DI questionnaire of the HAQ (24) showed that, of the 156 evaluable patients with prior TNF inhibitor exposure, 112 patients who completed course 1 (72%) and 108 patients who completed course 2 (69%) experienced a clinically meaningful improvement in function, defined as a decrease in the DI of the HAQ of ≥0.22 from baseline (26). After 24 weeks, the SF-36 (25) mental component scores in 117 evaluable patients with prior TNF inhibitor exposure increased by a mean of 7.4 following course 1 and 8.6 following course 2, relative to the original baseline (data not shown). Similarly, mean increases in the physical component scores of the SF-36 at 24 weeks were 8.6 following course 1 and 7.8 following course 2 (95% CI difference −2.3, 0.6); these differences exceeded the minimum clinically important difference for SF-36 mental and physical component scores (28).
In patients with prior TNF inhibitor exposure who received up to 3 courses of rituximab, the mean ± SD treatment intervals (weeks) between courses were similar (33.2 ± 9.5 weeks between course 1 and course 2 and 32.2 ± 10.4 weeks between course 2 and course 3). Treatment intervals between courses were also consistent in patients who had never taken TNF inhibitors (mean ± SD 45.5 ± 33.0 weeks between course 1 and course 2 and 48.3 ± 23.0 weeks between course 2 and course 3), although this population had longer mean treatment intervals compared with patients with prior TNF inhibitor exposure.
Safety data were available on the first 4 courses of rituximab (Table 2). The majority of AEs were mild or moderate (grades 1 and 2). The rate of all AEs reported during course 1 and course 2 was highest in the first 3 months and decreased thereafter. Patient-years of exposure decreased from 1,152.8 to only 12.5 across the 4 courses of treatment. The rates (and 95% CIs) of AEs per 100 patient-years were slightly decreased after course 1 but remained stable thereafter. The rates of AEs per 100 patient-years according to Common Toxicity Criteria severity (29) were highest during course 1 and generally decreased during subsequent courses (Table 2). The rates of all serious AEs per 100 patient-years were stable throughout all courses. The rate of grade 3 serious AEs per 100 patient-years during course 1 was higher than in course 2 and course 3, while the rates of grade 4 serious AEs per 100 patient-years were relatively stable. The 95% CIs calculated during course 4 were wide as a result of the limited observation time.
|Course 1 (n = 1,039)||Course 2 (n = 570)||Course 3 (n = 191)||Course 4 (n = 40)|
|Total no. of AEs†||4,570||1,358||339||48|
|AEs per 100 patient-years (95% CI)‡|
|All||396.4 (385.1–408.1)||326.0 (309.1–343.8)||383.0 (344.3–426.0)||384.2 (289.6–509.8)|
|Grade 1||184.0 (176.3–192.0)||137.3 (126.5–149.0)||154.8 (130.9–183.0)||152.1 (97.0–238.4)|
|Grade 2||155.0 (148.0–162.4)||108.0 (98.5–118.5)||87.0 (69.6–108.8)||80.0 (43.1–148.8)|
|Grade 3||33.0 (29.9–36.5)||14.9 (11.6–19.1)||15.8 (9.4–26.7)||8.0 (1.1–56.8)|
|Grade 4||1.5 (0.9–2.4)||1.2 (0.5–2.9)||1.1 (0.2–8.0)||8.0 (1.1–56.8)|
|AEs leading to withdrawal, no. (%)||29 (3)||9 (2)||0 (0)||0 (0)|
|No. of serious AEs||222||73||18||3|
|Serious AEs per 100 patient-years (95% CI)‡|
|All||19.3 (16.9–22.0)||17.5 (13.9–22.0)||20.3 (12.8–32.3)||24.0 (7.7–74.4)|
|Grade 1||0.9 (0.5–1.6)||0.7 (0.2–2.2)||1.1 (0.2–8.0)||ND|
|Grade 2||5.2 (4.0–6.7)||7.2 (5.0–10.3)||2.3 (0.6–9.0)||8.0 (1.1–56.8)|
|Grade 3||10.6 (8.9–12.6)||6.7 (4.6–9.7)||7.9 (3.8–16.6)||8.0 (1.1–56.8)|
|Grade 4||1.5 (0.9–2.4)||1.2 (0.5–2.9)||1.1 (0.2–8.0)||8.0 (1.1–56.8)|
|No. of infections||954||347||71||11|
|Infections per 100 patient-years (95% CI)||82.8 (77.7–88.2)||83.3 (75.0–92.5)||80.2 (63.6–101.2)||88.0 (48.8–159.0)|
|No. of serious infections§||59||19||5||1|
|Serious infections per 100 patient- years (95% CI)||5.1 (4.0–6.6)||4.6 (2.9–7.2)||5.6 (2.4–13.6)||8.0 (1.1–56.8)|
Overall, 38 of 1,039 patients (4%) withdrew due to AEs. Twenty-nine of 1,039 patients (3%) withdrew due to AEs during course 1, and 9 of 570 patients (2%) withdrew during course 2 (Table 2). Most AEs leading to withdrawal were infusion-associated events (33%) or RA exacerbation (20%). No patients withdrew because of AEs during course 3 or course 4. Eleven deaths, including 3 due to fatal serious infections, were reported, yielding an overall event rate of 0.66 deaths per 100 patient-years. Nine deaths occurred between course 1 and course 2, and 2 between course 2 and course 3. Deaths due to infections and malignancies are described below. Other causes of death included myelodysplastic syndrome, cerebrovascular accident, hemorrhagic stroke, and coronary artery disease, all of which occurred following course 1. There were 2 deaths from unknown causes.
Acute infusion-related reactions were the most common AE associated with rituximab treatment. Events included pruritus, fever, urticaria/rash, chills, pyrexia, rigors, sneezing, angioneurotic edema, throat irritation, cough, bronchospasm, hypotension, and hypertension. The proportion of patients experiencing an acute infusion-related event within 24 hours of the first infusion decreased from 26% for course 1 to 10–15% in subsequent courses. Serious acute infusion-related reactions were reported by <1% of patients, irrespective of treatment course (Figure 4A). Fewer acute infusion-related events, including serious AEs, were reported following the second infusion of each course, with the majority of AEs being mild or moderate (grades 1 and 2).
Of the 1,039 patients receiving course 1, 12 (1%) withdrew as a result of acute infusion-related events, and in an additional 100 patients (10%), dose modification due to AEs was required. No patients withdrew as a result of acute infusion-related events during the first infusion of course 2, course 3, or course 4, and the proportion of patients in whom dose modification was required decreased with each additional course and was less common following the second infusion of each individual course. There were no fatal infusion-related reactions, and all infusion-related reactions resolved with treatment and/or interruption or discontinuation of the infusion.
The rates of infection in the all-exposure population who received ≥1 course of rituximab were 82.8 events per 100 patient-years (all infections) and 5.1 events per 100 patient-years (serious infections) during course 1. Infection rates remained relatively stable from course 1 through course 3. However, the extent of observation decreased over time from 1,154.1 patient-years to 87.8 patient-years, with increasing 95% CIs reported (Table 2). Similar to the rates of all AEs, infection rates for course 4 were based on a limited number of patients (12.5 patient-years), and the amount of followup was insufficient to provide meaningful data and reliably estimate an infection rate.
Overall, 68 of 1,039 patients (7%) in the all-exposure population experienced a total of 78 serious infections following treatment with rituximab. Irrespective of treatment course, the infection rate was higher during the first 3 months after the infusion and declined thereafter. The most common infections reported were upper respiratory tract infection, nasopharyngitis, urinary tract infection, bronchitis, and sinusitis. Three serious infections were fatal. Two occurred between course 1 and course 2; 1 patient had bronchopneumonia and a second patient had neutropenic sepsis following concomitant treatment with trimethoprim. The third infection-related fatality occurred after course 2, from septic shock in a 54-year-old female diabetic patient with a history of sepsis and recurrent urinary tract infections. No incidences of opportunistic infections or tuberculosis were observed.
A total of 26 malignancies were reported in 22 of 1,039 patients (2%) in the all-exposure population. The most common cancers reported were skin cancer (basal cell [n = 6], squamous [n = 2], and malignant melanoma [n = 2]) and breast cancer (n = 3). No lymphomas were reported. Two malignancies (pancreatic neoplasm and intestinal adenocarcinoma) were fatal.
Prior to course 1, 203 of 986 patients (21%) had B cell counts below the LLN (<80 cells/μl). Treatment with rituximab resulted in a near-to-complete depletion of peripheral CD19+ B lymphocyte counts. In most patients, B cell counts did not return to within the normal range prior to subsequent treatment courses. Prior to course 2, 381 of 531 patients (72%) had peripheral blood B cell counts below the LLN, and prior to course 3, 123 of 171 patients (72%) had peripheral blood B cell counts below the LLN. Generally, across all treatment courses, the onset of B cell repletion was observed at least 24 weeks after administration of rituximab. In patients in whom B cells were repleted between courses of rituximab, similar patterns of B cell depletion and repletion were observed following each course (Figure 4B). No consistent correlation was seen between B cell depletion and clinical response over time following additional courses.
The proportion of patients with IgM concentrations below the LLN increased with subsequent courses. Of the patients in the all-exposure group on whom IgM data were available at week 24 (n = 1,035), the proportion of patients with IgM concentrations below the LLN at week 24 was 10.3% after course 1, 18.5% after course 2, and 23.5% after course 3. In patients with IgM levels below the LLN at 24 weeks following each treatment course, the mean IgM values were 0.41 gm/liter following course 1, 0.39 gm/liter following course 2, and 0.38 gm/liter following course 3, with an LLN of 0.5 gm/liter. A similar pattern was observed for IgG concentrations: 1.5% of the patients had levels below the LLN at week 24 after course 1, 4.3% after course 2, and 5.9% after course 3. The proportion of patients with IgA concentrations below the LLN increased slightly from 0.5% at baseline to 0.8% at weeks 16, 24, and 40. Thereafter, few patients had IgA concentrations below the LLN, and the proportion of patients with IgA concentrations below the LLN did not increase overtime.
Immunoglobulin levels were measured approximately every 8 weeks. The overall rate of serious infections in patients with at least 1 instance of a low IgM concentration (5.6 per 100 patient-years) and/or low IgG concentration (4.8 per 100 patient-years) was comparable with the rates seen in the 804 patients who had never exhibited low IgM or IgG concentrations (4.7 per 100 patient-years), as well as the all-exposure population (5.0 per 100 patient-years) (Table 3). A subgroup analysis of 207 patients with low IgM levels compared the rate of infection before a low IgM concentration was observed with that following detection of IgM levels below the LLN. The rates of all infections were 156 per 100 patient-years before a low IgM concentration was observed compared with 78 per 100 patient-years after a low IgM level was observed. Rates of serious infections were 5.1 per 100 patient-years and 5.9 per 100 patient-years, respectively.
|All-exposure population (n = 1,039)||Patients with IgM and IgG above LLN (n = 804)||Patients with IgM below LLN (n = 207)||Patients with IgG below LLN (n = 50)|
|No. (%) of patients with serious infections||68 (6.5)||45 (5.6)||18 (8.7)||6 (12.0)|
|Number of AEs/patient-years of exposure||84/1,670.8||58/1,225.1||21/376.2||6/125.4|
|Serious infection rates/100 patient-years||5.0||4.7||5.6||4.8|
Of the 1,039 patients who received ≥1 infusion of rituximab, 96 patients (9.2%) had ≥1 finding of HACA positivity, corresponding to a rate of 5.7 events of HACA positivity per 100 patient-years. Of these, 79 patients (82%) were HACA positive after course 1, while the remaining 17 patients developed HACA positivity after course 2. Therefore, the proportion of patients who were HACA positive was 7.6% after course 1 (79 of 1,039 patients) and 3% after course 2 (17 of 570 patients). One HACA-positive patient developed a nonserious pruritic rash after the first infusion of course 1 and bronchospasm during the second infusion of course 2, and did not continue in the study. In another HACA-positive patient, B cell depletion failed to occur following a repeated course of treatment.
The safety and efficacy of additional courses of treatment with rituximab were evaluated in 2 open-label extension studies. Cumulative data derived from more than 1,600 patient-years of followup demonstrated the safety and efficacy of repeated exposure to rituximab. Overall, clinical efficacy was maintained with subsequent courses, and efficacy was similar following the second course, regardless of prior exposure to TNF inhibitors. Relative to the original baseline, ACR and EULAR responses following a second course of rituximab were at least as good as those reported from the first course. Since mean disease activity was lower prior to the second course than the original pretreatment baseline, the observed ACR and EULAR responses suggest that additional courses of rituximab may not only maintain the efficacy response relative to the original baseline, but may further decrease disease activity. The sustained outcomes were observed in all patient populations studied, including those who had never taken TNF inhibitors and those previously treated with ≥1 TNF inhibitor.
Repeated courses of rituximab were well tolerated, and safety data were consistent with the safety profile reported in previous controlled rituximab trials, with no additional safety concerns following further exposure (17, 18, 20, 30). In the all-exposure population, rates of AEs were highest in the first 3 months and declined thereafter. This is consistent with the observation that the majority of AEs are associated with infusions of rituximab (given on days 1 and 15). Infections were also commonly reported during the first 3 months and possibly associated with concomitant use of glucocorticoids during the infusion periods. For each rituximab treatment course, the majority of patients in the present study received premedication with glucocorticoids, as well as oral glucocorticoids on the days between rituximab infusions. While the use of glucocorticoids as a premedication is recommended to reduce the incidence and severity of infusion reactions, glucocorticoids have not been shown to affect clinical response (19).
The incidence of acute infusion-related events also decreased with additional treatment courses, with fewer events reported with subsequent courses, as well as with the second infusion of each specific course, a pattern also found in an earlier analysis (30). This may be because additional courses of rituximab were administered to patients who had lower peripheral B cell counts than prior to their first exposure so that known events, including fever, chills, nausea, vomiting, hypotension, and dyspnea, which may be associated with cytokine release because of B cell lysis, may be less frequent (31).
In the present study, the observed incidence of malignancies following exposure to rituximab was 1.6 per 100 patient-years, which was not atypical for this population and was within the range expected for a population of patients age 60 years or older, with a sex distribution similar to that of the population examined in this study and who had previously received immunosuppressive therapies and were treated with concomitant MTX (32, 33). The mortality rate observed in this study was 0.66 per 100 patient-years, which is consistent with mortality rates of 0.88 per 100 patient-years and 0.7 per 100 patient-years demonstrated in long-term studies of other biologic therapies for RA (34, 35).
The rates of infection, including serious infections, remained stable with each subsequent course of rituximab. The infection rate for course 1 in the current study (82.8 per 100 patient-years) was lower than that reported in rituximab-treated patients (138.2 per 100 patient-years) and placebo-treated patients (154.6 per 100 patient-years) at 24 weeks in the phase III double-blind trial (20, 36). The type and clinical course of the infections observed in these patients were also similar (20, 36). No cases of opportunistic infections or tuberculosis were observed.
The serious infection rate in the all-exposure population in this analysis (5.0 per 100 patient-years) is consistent with published serious infection rates from previous rituximab trials and from additional trials of anti-TNF biologic agents in RA (17). In rituximab-treated patients, serious infection rates at the primary end point were 5.3 per 100 patient-years in the phase III trial (20) and 4.7 per 100 patient-years in the phase IIb trial (17). These results are consistent with those of a long-term, open-label extension analysis of etanercept in patients with RA, which has shown a serious infection rate of 4.8 per 100 patient-years (37). A recent prospective observational study of TNF inhibitors (etanercept, infliximab, and adalimumab) demonstrated a serious infection rate of 5.3 per 100 patient-years (525 infections occurred in 9,868 patient-years [5.3%]) (38). Considering the overlapping CIs, these rates of infection appear consistent with the rates observed with rituximab.
B cell depletion was associated with reductions in circulating IgM and IgG concentrations. In the patients who had IgM or IgG concentrations below the LLN at any time after administration of rituximab, infection rates and serious infection rates were comparable with the rate of infection in the all-exposure population, as well as the rate in patients who had IgM and IgG levels within the normal range. No patient received IV immunoglobulin replacement. Due to the relatively low numbers of patients receiving ≥3 courses and the limited duration of followup, these data should be considered preliminary and are insufficient for definitive conclusions. Further followup is required as more patients receive additional courses; however, it is encouraging that infection rates (including serious infection rates) are consistent with those observed with other biologic therapies for RA and remain comparable both between treatment courses and over time, despite prolonged B cell depletion, and, in some patients, when IgM and/or IgG concentrations decrease below the LLN (17, 30, 36, 39, 40).
The proportion of patients with HACA positivity was 9.2%. There was no clear evidence that the presence of HACA interfered with the safety or efficacy of additional courses of rituximab (20, 30, 36, 41). A subgroup analysis of the AE data from the 96 HACA-positive patients did not reveal increased safety concerns compared with HACA-negative patients. No clear association between HACA levels and decreased clinical responses in patients receiving >1 course was seen. While the actual prevalence of HACAs may be greater than observed in this study because of the potential interference of circulating rituximab, the current data on HACA-positive patients do not show a correlation between development of HACAs and increased safety risks or tachyphylaxis, and HACA-positivity does not appear to be a significant concern in determining whether a patient should receive additional rituximab courses.
Within-patient comparisons of patients who received 2 courses of rituximab showed a similar pattern of peripheral B cell depletion and repletion across treatment courses, suggesting that rituximab has no cumulative effect on peripheral B cell response over time when additional courses are administered. No apparent correlation was determined between B cell depletion and response. B cell depletion occurred in all patients following treatment with rituximab, but not all patients responded to treatment (42). These findings suggest that that is no consistently clear relationship between absolute B cell number and return of disease activity in an individual patient. Accordingly, the decision to provide additional courses of rituximab must include clinical assessment of disease activity apart from the patient's B cell count (43). In general, patients received additional courses before their disease activity had returned to near original baseline levels, and received subsequent courses an average of every 6–9 months.
While such open-label extension studies are potentially biased by enrichment of a population of patients who respond to treatment, within-patient comparisons suggest that sustained and comparable responses may still be achieved following subsequent courses of rituximab. It is notable, however, that patients receiving the first course participated in placebo-controlled randomized clinical trials, while patients receiving the second course were evaluated in an open-label analysis. To date, the data also highlight the number of patients who remain in the clinical program and who have yet to receive additional courses of rituximab due to sustained clinical efficacy following an initial course.
In conclusion, observational data from a cohort of 1,039 patients with over 1,600 patient-years of exposure suggest that the safety and efficacy of rituximab are consistent with that reported in earlier randomized controlled clinical trials. Repeated courses of rituximab appeared to produce consistent and sustained efficacy relative to the original baseline, with no new adverse events in patients who had never taken TNF inhibitors or in patients who previously had an inadequate response to these agents.
Dr. Keystone 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 design. Fleischmann, Emery, Furst, Dougados, Agarwal, Magrini.
Acquisition of data. Fleischmann, Emery, van Vollenhoven, Bathon, Dougados, Baldassare, Chubick, Udell.
Analysis and interpretation of data. Keystone, Fleischmann, Emery, Furst, van Vollenhoven, Bathon, Dougados, Ferraccioli, Udell, Cravets, Agarwal, Cooper, Magrini.
Manuscript preparation. Keystone, Fleischmann, Emery, Furst, van Vollenhoven, Bathon, Dougados, Baldassare, Ferraccioli, Chubick, Udell, Agarwal, Cooper, Magrini, and Keith Del Villar (nonauthor; Genentech).
Statistical analysis. Cravets.
F. Hoffman La Roche was responsible for data collection. Statistical analysis was conducted by suitably qualified statisticians who were the employees of the sponsors. The trial protocol was designed jointly by F. Hoffman La Roche and the lead clinical investigators. All authors had access to the data and were involved in the interpretation of the data, and all authors had input into and control over the content of this manuscript, supervised by Dr. E. Keystone. Keith Del Villar, PhD (Genentech, Inc.) provided writing assistance.
The authors wish to thank Karen Rowe, Penny Ward, Eva Hessey, Bhupendra Mistry, Laura Burke, Anne-Marie White, Simon Safa-Leathers, Patricia Lehane, and Tim Shaw (Roche Products Ltd). The authors also wish to thank Alan Donohoe, Ahmed Tanveer, Jenny Moore, Cathy Ofori-Atta, Chinnie Nwandu, Sumu Sethi, Marcia Boakye, and Sunny Patel for management of the clinical studies. The authors also thank the investigators involved in both extension studies.