To assess the efficacy, safety, and biologic activity of atacicept in tumor necrosis factor antagonist–naive patients with rheumatoid arthritis (RA) in whom the response to methotrexate treatment was inadequate.
To assess the efficacy, safety, and biologic activity of atacicept in tumor necrosis factor antagonist–naive patients with rheumatoid arthritis (RA) in whom the response to methotrexate treatment was inadequate.
In this phase II study, patients with active RA (n = 311) were randomized 1:1:1:1 to receive placebo, atacicept 150 mg weekly with or without a 4-week loading period (twice-weekly dosing), or open-label adalimumab 40 mg every other week, for 25 weeks. The primary end point was 20% improvement in disease severity according to the American College of Rheumatology criteria, assessed using the C-reactive protein level (ACR20-CRP), at week 26. Secondary end points included additional assessments of efficacy, biologic activity, and safety.
The proportion of patients meeting the primary end point (ACR20-CRP response) did not differ significantly in the atacicept groups and the placebo group (46% in the placebo group, 45% in the atacicept loading group, and 58% in the atacicept nonloading group). In contrast, an ACR20-CRP response was observed in 71% of patients in the adalimumab group (P < 0.001 versus placebo). ACR50-CRP response rates were significantly higher in all active-treatment groups compared with placebo, but ACR70-CRP response rates were superior only in the adalimumab group. Atacicept treatment reduced the levels of serum IgG, IgA, and IgM rheumatoid factor and the levels of circulating mature B cells and plasma cells. The effects of treatment were similar with and without loading. Immunoglobulin levels returned toward baseline values during the treatment-free followup period (week 38). The most frequent adverse events associated with atacicept represented common illnesses. No serious infections occurred among patients treated with atacicept.
The primary end point (ACR20-CRP response) was not met despite significant biologic effects of atacicept that were consistent with its proposed mechanism of action. Modest effects of atacicept were seen for some secondary efficacy end points. Treatment with atacicept raised no new safety concerns.
Rheumatoid arthritis (RA) is a chronic, inflammatory, autoimmune disease in which inflammation predominantly affects the peripheral joints (1). B cells are believed to have a role in the pathogenesis of RA. Autoreactive antibodies, including rheumatoid factor (RF) and anti–citrullinated protein antibodies (ACPAs), contribute to immune complexes found in joint fluid (2, 3), and B cells also produce cytokines and present self antigens to T cells. Targeting B cells, therefore, represents a rational approach to the treatment of RA (4). Indeed, the clinical efficacy of B cell–targeting therapy has been demonstrated with the B cell–depleting agent rituximab (anti-CD20) (5).
Atacicept is a soluble, fully human, recombinant fusion protein comprising the extracellular portion of the TACI receptor and the Fc portion of human IgG (6, 7). Atacicept neutralizes the B cell maturation/survival factors B lymphocyte stimulator (BLyS) and APRIL, a proliferation-inducing ligand (7, 8). BLyS and APRIL levels have been shown to be elevated in patients with RA (9–13). Binding of atacicept to BLyS and APRIL heterotrimers and homotrimers results in decreased numbers of mature B cells, plasma cells, and serum antibodies but spares B cell progenitors and memory B cells (4).
In contrast to rituximab, belimumab (anti-BLyS) has not shown a substantial clinical benefit in patients with RA (14). However, because atacicept also blocks APRIL (which can maintain plasma cells in the absence of BLyS ), and because the clinical response to rituximab correlates with the reduction in synovial plasma cell numbers (16, 17), it is reasonable to propose that atacicept would have significant clinical efficacy in RA.
Phase I studies of atacicept in healthy volunteers and in patients with systemic lupus erythematosus (SLE), patients with RA, or patients with B cell lymphoma indicated that atacicept was well tolerated over a range of doses, with no unexpected safety concerns (6–8, 18, 19). In the study of patients with RA, atacicept was associated with treatment-related decreases in immunoglobulins, including total immunoglobulin, RF, and ACPAs (6).
The objective of the Atacicept for Reduction of Signs and Symptoms in the Rheumatoid Arthritis Trial II (AUGUST II) reported here was to assess the efficacy, biologic effects, and safety of subcutaneously administered atacicept, with and without a dose-loading treatment period, in patients with active RA and an inadequate response to methotrexate (MTX). The AUGUST I study, which ran in parallel to AUGUST II, was a dose-finding study of atacicept in patients with RA and an inadequate response to tumor necrosis factor (TNF) antagonists. The results of AUGUST I are reported elsewhere (20, 21).
AUGUST II was a phase II, randomized, placebo-controlled, double-blind (atacicept versus placebo), parallel-arm, multicenter, prospective study (ClinicalTrials.gov identifier NCT00595413). Patients were randomized (1:1:1:1) to receive subcutaneously administered atacicept, 150 mg twice weekly for 4 weeks (loading) and then weekly for 21 weeks; subcutaneously administered atacicept, 150 mg weekly for 25 weeks (nonloading); placebo for 25 weeks; or open-label subcutaneously administered adalimumab, 40 mg every other week for 25 weeks. A final followup visit was scheduled 13 weeks after the last injection of study drug (week 38). To maintain blinding, all patients assigned to receive atacicept or placebo received 2 injections during the initial 4 weeks. Those assigned to receive atacicept without loading received alternate injections of placebo and atacicept.
The inclusion of an approved anti-TNFα therapy (adalimumab) was considered to be of value in order to obtain an impression of the relative efficacy of atacicept versus an established therapy, and as a positive control for treatment effects in this patient population. Descriptive statistics were produced for all treatment groups; odds ratios (ORs) were calculated for adalimumab versus placebo but not for adalimumab versus atacicept treatments. However, because no formal statistical comparisons between adalimumab and the atacicept treatment arms were planned, and because the study was not powered for such, blinding of the adalimumab arm was not deemed necessary.
The study was approved by the Institutional Review Board or Independent Ethics Committee of the participating institutions and was conducted in accordance with the Declaration of Helsinki (1996), the International Conference on Harmonisation Harmonised Tripartite Guideline for Good Clinical Practice, and all applicable local regulatory requirements. All patients gave written informed consent prior to any study assessment that did not constitute routine medical care. Additional contributors to the AUGUST II study are shown in Appendix A.
The study population comprised male and female patients at least 18 years of age who were attending outpatient rheumatology clinics. All patients had a diagnosis of RA according to the American College of Rheumatology (ACR) criteria (22), a disease duration of at least 6 months, had failed to experience an adequate response to treatment with MTX (15–25 mg/week for >3 months), and had never received treatment with TNF antagonists. Additional inclusion criteria were as follows: 1) active disease, defined as ≥8 swollen joints (of 66 joints counted), ≥8 tender joints (of 68 joints counted), plus a C-reactive protein (CRP) level of ≥10 mg/liter and/or an erythrocyte sedimentation rate (ESR) of ≥28 mm/hour; and 2) receipt of a stable MTX dose for ≥28 days prior to study day 1 (defined as the first day of study treatment).
The main exclusion criteria included inflammatory joint disease other than RA; previous or concurrent treatment with any approved or investigational biologic compound for RA; treatment with disease-modifying antirheumatic drugs other than MTX; prednisone dosage >10 mg/day (or equivalent) or change in steroid or nonsteroidal antiinflammatory drug dosing regimen ≤28 days before study day 1; live vaccine or immunoglobulin treatment ≤28 days before study day 1; history or presence of active or latent tuberculosis according to local and/or national recommendations in the past year; infection requiring hospitalization or intravenous treatment with antiinfective agents ≤28 days before study day 1; major concurrent illness or organ dysfunction; and serum IgG level <6 gm/liter.
Patients were randomized by the permuted-blocks method, stratified by RF status (positive or negative) and sex. The study was double-blinded except for the open-label active comparator (adalimumab). At the time of screening, patient information was entered into an electronic case report form, and a unique identification number was assigned. Central randomization was performed on study day 1, after written informed consent was received. Treatment kit numbers corresponding to the individual's assigned treatment were obtained through an interactive voice response system (IVRS). Treatments were provided in study medication kits labeled with a unique number obtained through an IVRS. Prefilled syringes were covered by opaque labels to hide any differences in the solutions from patients and study personnel. Atacicept and placebo were administered in the same volume according to the same schedule. To protect blinding, assessors of efficacy (joint counts and physician's global assessment) were not involved in IVRS contact, dispensing or administration of study medications, or collection of local tolerability data and could not access the electronic case report forms.
The primary end point was the proportion of patients with 20% improvement in disease severity according to the ACR criteria, as assessed using the CRP level (ACR20-CRP) (23). Secondary efficacy end points included the proportion of patients at week 26 with ACR50-CRP and/or ACR70-CRP responses, and changes in the Disease Activity Score in 28 joints (DAS28) (24) as assessed using the CRP level (DAS28-CRP), including the proportion of patients with good or moderate European League Against Rheumatism (EULAR) responses. Safety end points included the nature, incidence, and severity of adverse events (AEs) and local injection-site reactions and the results of clinical laboratory tests. Pharmacodynamic end points included changes over time in the levels of immunoglobulins, RF, ACPAs, and specific peripheral blood mononuclear cell (PBMC) subpopulations.
Assessment visits took place at the time of screening, at the baseline visit, on days 8 and 22 (weeks 2 and 4 in this study, respectively), at the beginning of weeks 8, 12, 16, 20, and 26, and at week 38 (followup visit). All laboratory assessments were performed centrally, except for the ESR. All measures were assessed at each visit, except for PBMC numbers (which were not assessed at weeks 2, 8, and 20). Demographic data, medical/disease history, and treatment history were recorded at the screening visit.
Efficacy was assessed using the ACR core disease activity measures (25); ACR20-CRP, ACR50-CRP, and ACR70-CRP responses and DAS28-CRP assessments were calculated based on these measures. AEs were assessed throughout the 38-week study. Treatment was discontinued if an IgG level of <3 gm/liter was recorded.
Pharmacodynamic assessments were made using standard clinical laboratory techniques. Flow cytometric analyses of circulating PBMCs were performed in a subset of patients identified by the central randomization service at selected centers (further methodologic details are available from the corresponding author). Absolute numbers of specific leukocyte populations were calculated from the resulting proportions, and white blood cell counts (cells/fL) were obtained using a Beckman Coulter LH750 hematology analyzer.
Four-color flow cytometry with a FACSCalibur was performed using 3 antibody cocktails with the following fluorophores: fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyll A protein (PerPC), and allophycocyanin (APC). The antibody cocktails were as follows: FITC-conjugated anti-CD8, PE-conjugated anti-CD4, PerPC-conjugated anti-CD45, APC-conjugated anti-CD3, FITC-conjugated anti-CD19, PE-conjugated anti-CD16/56, PerPC-conjugated anti-CD45, APC-conjugated anti-CD3, FITC-conjugated anti-IgD, PE-conjugated anti-CD27, PerPC-conjugated anti-CD19, and APC-conjugated anti-CD38. Leukocyte populations were gated on a CD45 side scatter plot, and cellular proportions were determined for total T cells (CD3+), cytotoxic T cells (CD3+CD8+), Th cells (CD3+CD4+), natural killer cells (CD3−CD16/56+), and total B cells (CD19+). The B cell subsets of mature B cells (CD19+IgD+CD27−) and memory B cells (CD19+CD27+CD38−) were gated on CD19+ cells within the lymphocytes and monocytes, as defined on the forward/side scatter plot. Plasma cells (CD19dimCD38bright) were gated on lymphocytes and monocytes defined on the forward/side scatter plot. Plasmablasts (CD19+CD27brightCD38bright) were defined as CD27+ cells within the plasma cell population.
Assuming a screening failure rate of 40%, 440 patients needed to be screened in order to achieve a sample size of 264 randomized patients (66 patients per treatment arm). This sample size (264 patients) was needed in order to detect a difference of ≥30% between either of the atacicept arms and placebo, with 90% power when Hochberg's procedure was applied to maintain the overall Type I error rate at 5%. An ACR20-CRP placebo response rate of 30% was assumed.
The intent-to-treat (ITT) population (used for the analysis of all efficacy end points) comprised all patients randomized and treated, analyzed as randomized. The safety population comprised all patients who received at least 1 treatment dose and for whom safety data were available, analyzed as treated.
The primary efficacy end point was assessed using a logistic regression model for each pairwise comparison, atacicept versus placebo, with the 2 stratification factors (RF status and sex) included. The interactions with treatment were tested. For the comparisons of the 2 atacicept treatment arms versus placebo, Hochberg's procedure was used to control the overall Type I error at the 5% level. Nonresponder imputation was used with binary responses for early terminations or incomplete data. Secondary and other end points were assessed using analysis of variance, analysis of covariance, or logistic regression, depending on the type of end point, adjusted for stratification factors. Safety data were summarized using descriptive statistics and AE summaries.
Recruitment took place between September 2007 and November 2008. Screening was stopped when enrollment was complete, and the study period ended after the last enrolled patient's last visit had occurred. A total of 512 patients from 104 sites in 19 countries were screened, and 311 were randomized (for placebo, n = 76 patients; for atacicept loading, n = 78 patients; for atacicept nonloading, n = 78 patients; for adalimumab, n = 79 patients). All randomized patients received at least 1 dose of study drug, and 267 patients (86%) completed treatment. Reasons for discontinuing treatment included AEs, lack of efficacy, and “other reasons” (Figure 1).
Baseline patient demographic and clinical characteristics, including disease activity, were similar across treatment groups (Table 1). Based on eligibility criteria and patient characteristics, it was estimated that almost all patients would have fulfilled the new ACR/EULAR criteria for RA (26).
|Characteristic||Placebo (n = 76)||Atacicept (loading) (n = 78)||Atacicept (nonloading) (n = 78)||Adalimumab (n = 79)|
|Age, years||54 ± 10.3||53 ± 11.3||53 ± 13.2||53 ± 11.5|
|Female sex, no. (%)||64 (84)||65 (83)||66 (85)||64 (81)|
|Weight, kg||75 ± 16.4||74 ± 14.3||72 ± 18.0||73 ± 16.7|
|Disease duration, years||8.4 ± 7.4||7.8 ± 7.3||7.3 ± 6.5||8.8 ± 7.4|
|Rheumatoid factor positive, no. (%)||63 (83)||64 (83)||64 (82)||63 (81)|
|Oral corticosteroid treatment, no. (%)||45 (59)||43 (55)||52 (67)||52 (66)|
|DAS28-CRP||5.8 ± 1.0||5.8 ± 0.9||5.8 ± 0.8||5.8 ± 1.0|
|SJC (66 joints assessed)||16.4 ± 8.5||17.1 ± 8.0||16.0 ± 8.6||16.2 ± 8.3|
|TJC (68 joints assessed)||24.3 ± 11.4||27.8 ± 15.5||27.5 ± 14.9||27.8 ± 14.9|
|CRP, mg/liter||16.5 ± 16.3||19.5 ± 24.1||23.8 ± 27.3||16.6 ± 18.1|
|ESR, mm/hour||39.3 ± 17.4||44.6 ± 21.6||42.1 ± 22.8||41.7 ± 18.1|
|HAQ disability index score||1.7 ± 0.6||1.7 ± 0.5||1.6 ± 0.6||1.6 ± 0.5|
The proportions of patients (ITT population) with an ACR20-CRP response at week 26 were as follows: 46% of patients in the placebo group, 45% of patients in the atacicept loading group, and 58% of patients in the atacicept nonloading group (Figure 2). For atacicept loading versus placebo, the OR was 0.96 (95% confidence interval [95% CI] 0.51–1.82, P = 0.91), and that for atacicept nonloading versus placebo was 1.61 (95% CI 0.78–3.35, P = 0.14); hence, the primary end point was not met. An ACR20-CRP response was achieved in 71% of the patients receiving open-label adalimumab (OR 2.95 [95% CI 1.51–5.76], P = 0.001 versus placebo). Regional variation in the ACR20-CRP response rate was seen but was balanced across treatment groups (Table 2).
|Predefined region||Proportion (%) of patients with response at week 26|
|Placebo (n = 76)||Atacicept (loading) (n = 78)||Atacicept (nonloading) (n = 78)||Adalimumab (n = 79)|
|Rest of World†||63||62||70||73|
The proportion of patients with an ACR50-CRP response was significantly greater with atacicept, with or without loading, and with adalimumab compared with placebo (Figure 2): 15% in the placebo group, 30% in the atacicept loading group (OR 2.5 [95% CI 1.1–5.6], P = 0.025), 33% in the atacicept nonloading group (OR 3.0 [95% CI 1.4–6.7], P = 0.007), and 38% in the adalimumab group (OR 3.7 [95% CI 1.7–8.2], P = 0.001). The time course of the ACR20-CRP and ACR50-CRP responses in patients receiving atacicept was similar with or without dose loading (data not shown). The proportions of patients with an ACR70-CRP response at week 26 were as follows: 5% in the placebo group, 13% in both atacicept groups, and 18% in the adalimumab group; only the difference between adalimumab and placebo was significant (P < 0.05) (Figure 2). Good/moderate EULAR responses (week 26) were seen in 59% of patients in the placebo group, 64% in the atacicept loading group, 68% in the atacicept nonloading group, and 81% in the adalimumab group (P < 0.05, adalimumab versus placebo).
Compared with placebo, atacicept was associated with greater reductions in IgM, IgA, and IgG levels. The time course and magnitude of the responses were similar with and without dose loading (Figure 3a). Between baseline and week 26, atacicept with and without loading decreased IgM levels by 66% (P < 0.001) and 72% (P < 0.001), respectively, IgA levels by 56% (P < 0.001) and 55% (P < 0.001), respectively, and IgG levels by 31% (P < 0.001) and 30% (P < 0.001), respectively, compared with reductions of 5% (IgM), 3% (IgA), and 2% (IgG) in the placebo group. Increases of 15% (IgM), 0.4% (IgA), and 2% (IgG) were seen in the adalimumab group. Median IgG levels remained above the lower limit of normal (7 gm/liter) throughout the study, and no patients discontinued the study because of IgG levels of <3 gm/liter.
Atacicept also reduced RF levels (Figure 3b). Between baseline and week 26, atacicept, with and without loading, reduced IgM-RF levels by 60% (P < 0.001) and 62% (P < 0.001), respectively, IgA-RF levels by 67% (both regimens; P < 0.001), and IgG-RF levels by 54% (P < 0.001) and 57% (P < 0.001), respectively, compared with reductions of 0% (IgM-RF and IgG-RF) and 14% (IgA-RF) in the placebo group. Adalimumab decreased IgM-RF levels by 4%, IgA-RF levels by 26%, and IgG-RF levels by 15%.
The decrease in immunoglobulin and RF levels occurred mostly within the first 12 weeks of treatment, with only small decreases seen thereafter. Immunoglobulin and RF levels returned toward baseline during the 12-week treatment-free followup period.
The effects of atacicept on ACPA titers did not reach statistical significance. The median percentage reduction from baseline to week 26 was 16% with placebo, 16% with atacicept loading, 30% with atacicept nonloading, and 5% with adalimumab.
The effects of treatment on PBMCs were assessed in a subset of patients. Compared with placebo, atacicept reduced the absolute levels of mature B cells; the median percentage change from baseline to week 26 was 8% with placebo (n = 14), −63% with atacicept loading (n = 17), and −57% with atacicept nonloading (n = 17) (P = 0.0045, pooled atacicept versus placebo). Atacicept also reduced the absolute levels of plasma cells; the median percentage change from baseline to week 26 was 25% with placebo (n = 13), −51% with atacicept loading (n = 16), and 0% with atacicept nonloading (n = 14) (P = 0.0236, pooled atacicept versus placebo) (Figure 3c). A transient increase in memory B cell numbers was seen during treatment in all groups except the placebo group (Figure 3c). Atacicept was also associated with a significant reduction in the proportion of mature B cells and plasma cells from baseline to week 26. There were no significant effects on the numbers of natural killer cells or total, cytotoxic, and helper T cells (data not shown).
A total of 582 AEs that started between baseline and week 38 (i.e., treatment-emergent AEs) were reported in 186 patients. The numbers of patients in each treatment group with at least 1 AE were as follows: 38 patients in the placebo group (50%), 49 patients in the atacicept loading group (63%), 49 patients in the atacicept nonloading group (63%), and 50 patients in the adalimumab group (63%) (Table 3). The frequency of AEs was numerically higher in atacicept-treated patients (combined group) than in those receiving placebo, in 5 system organ classes. The classes and frequencies of AEs in the placebo, atacicept (combined), and adalimumab groups were as follows: for infections and infestations, 29%, 35%, and 37%, respectively; for general disorders and injection-site conditions, 8%, 10%, and 11%, respectively; for skin and subcutaneous tissue disorders, 4%, 10%, and 5%, respectively; for psychiatric disorders, 1%, 3%, and 3%, respectively; and for neoplasms, 0%, 3%, and 0%, respectively. Of the 5 patients in whom a tumor developed, 4 had benign tumors (2 cases of uterine leiomyoma and 1 case each of lipoma and skin papilloma), and 1 had breast cancer, which was considered a serious AE (SAE). This SAE was diagnosed in a 75-year-old woman ∼2 months after the initiation of atacicept treatment.
|Preferred term||Placebo (n = 76)||Atacicept (loading) (n = 78)||Atacicept (nonloading) (n = 78)||Atacicept (combined) (n = 156)||Adalimumab (n = 79)|
|Any TEAE||38 (50)||49 (63)||49 (63)||98 (63)||50 (63)|
|Serious TEAEs||2 (3)||4 (5)||7 (9)||11 (7)||3 (4)|
|Disseminated tuberculosis||1 (1)|
|Dyspnea||1 (1)||1 (1)|
|Laryngeal edema||1 (1)||1 (1)|
|Obstructive airway disorder||1 (1)||1 (1)|
|Pulmonary embolism||1 (1)||1 (1)|
|Pulmonary hypertension||1 (1)||1 (1)|
|Pyrexia||1 (1)||1 (1)|
|Sudden cardiac death||1 (1)||1 (1)|
|Osteoarthritis||1 (1)||1 (1)|
|Rheumatoid arthritis||1 (1)||1 (1)|
|Spinal osteoarthritis||1 (1)||1 (1)|
|Tendon rupture||1 (1)|
|Breast cancer (female)||1 (1)||1 (1)|
|Knee arthroplasty||1 (1)||1 (1)|
|Hypertension||1 (1)||1 (1)|
|Spontaneous abortion||1 (1)||1 (1)|
|Pregnancy||1 (1)||1 (1)|
|TEAEs leading to discontinuation||2 (3)||3 (4)||6 (8)||9 (6)||2 (3)|
|Viral bronchitis||1 (1)||1 (1)|
|Injection-site reaction||1 (1)||1 (1)|
|Pyrexia||1 (1)||1 (1)|
|Sudden cardiac death||1 (1)||1 (1)|
|Drug hypersensitivity||1 (1)||1 (1)|
|Tendon rupture||1 (1)|
|Musculoskeletal pain||1 (1)||1 (1)|
|Breast cancer (females)||1 (1)||1 (1)|
|Laryngeal edema||1 (1)||1 (1)|
|Knee arthroplasty||1 (1)||1 (1)|
The most frequent AEs associated with atacicept treatment represented common illnesses including nasopharyngitis, bronchitis, headache, and upper respiratory tract and urinary tract infections, none of which occurred in ≥10% of patients with either active treatment (adalimumab group and pooled atacicept group).
The numbers of patients with SAEs were as follows: 2 patients in the placebo group (3%), 4 patients in the atacicept loading group (5%), 7 patients in the atacicept nonloading group (9%), and 3 patients in the adalimumab group (4%) (Table 3). No trends were seen in the nature of the SAEs reported in patients treated with atacicept. No serious infections were reported in patients treated with atacicept; 1 patient in the placebo group had an SAE (cellulitis), and 3 patients in the adalimumab group had SAEs (2 cases of pneumonia and 1 case of disseminated tuberculosis). There was 1 pregnancy during the posttreatment period, in a patient in the atacicept loading group, which terminated owing to spontaneous abortion. No relationship was observed between SAEs and the region where a patient was treated (6 SAEs in Eastern Europe, 6 in North America, 3 in Western Europe, and 6 in Rest of World).
Thirteen patients discontinued treatment due to AEs, which were variable in nature and did not suggest any trends (placebo, n = 2; atacicept loading, n = 3; atacicept nonloading, n = 6; adalimumab, n = 2). One patient in the atacicept nonloading group died of sudden cardiac arrest; this patient had a history of hypertension and diabetes.
In this phase II trial of atacicept in TNF antagonist–naive patients with RA and an inadequate response to MTX treatment, atacicept with or without dose loading did not significantly increase the proportion of patients with an ACR20-CRP response at week 26 compared with placebo; hence, the primary end point of the trial was not met. However, potent biologic activity consistent with the proposed mechanism of action (8) was demonstrated. Overall, dose loading had no effect on the clinical outcomes or pharmacodynamics. A separate phase II dose-finding study of atacicept in patients with RA and an inadequate response to TNF antagonists (the AUGUST I study) also showed no significant beneficial effect on clinical measures of disease, including the primary end point, the ACR20-CRP (20, 21). However, potent dose-dependent biologic activity was observed in that study (20, 21).
In the AUGUST II study reported here, potent biologic effects were evidenced by reductions in the levels of IgM, IgA, IgG, all 3 RF classes, mature B cells, and plasma cells. These parameters decreased up to week 16, but their response patterns differed thereafter. Immunoglobulin levels remained largely unchanged between week 16 and week 26 and recovered during the posttreatment period. Mature B cell levels remained unchanged from week 16 onward, whereas plasma cell numbers started to recover from week 16. The reason for these differential responses is not clear. One possible explanation is selective mobilization of plasma cells into the circulation during treatment (i.e., weeks 16–26). Another possibility is that this could result from rapid transition of mature B cells into plasma cells. However, such a transition would require a B cell development pathway that is independent of BLyS and APRIL.
ACPA levels were shown to be largely unaffected by atacicept. Although the reason ACPA levels were not reduced is unknown, it is possible that ACPAs are produced by longer-lived plasma cells that are relatively spared by atacicept.
In contrast to the effects on mature B cells and plasma cells, memory B cell numbers showed a biphasic pattern in response to atacicept treatment, initially increasing and subsequently returning toward baseline levels. A transient increase in circulating B cell numbers following administration of atacicept to patients with RA has previously been reported (6). The reason for the transient increase in the number of memory B cells is unclear, but possibilities include mobilization of cells from lymphoid tissues into the circulation, inhibition of their return to lymphoid tissues, transient proliferation, or preferential differentiation of mature B cells into memory cells. It would be interesting to define the mechanism behind the effects on B cell subsets through preclinical studies. As expected, atacicept had no significant effects on the numbers or proportions of natural killer cells or total, cytotoxic, and helper T cells.
The time course and magnitude of the immunoglobulin response were similar with atacicept 150 mg with or without dose loading in this study, and also with the 150-mg dose in the AUGUST I study (20, 21). These findings indicate that a dosage of 150 mg weekly, without loading, likely achieves the maximal biologic effect in patients with RA. Thus, a loading period need not be included in future studies of atacicept.
Potent biologic activity with only modest clinical efficacy was also reported in a phase II study of the anti-BLyS agent belimumab in patients with RA (14, 27). The notable differences in the magnitude of the biologic effects of atacicept and belimumab that have been observed may be attributable to the fact that atacicept blocks APRIL in addition to BlyS (28). Because APRIL can support plasma cells independently of BLyS (15), blockade of both factors may be expected to have a more profound effect on immunoglobulins. The percentage reduction in immunoglobulin levels with belimumab was more modest (14) than that with atacicept reported here, consistent with a significant decrease in plasma cell numbers with atacicept but not belimumab in the population with RA (14).
It is interesting that the reduction in immunoglobulin, particularly RF, levels and in mature B and plasma cell numbers achieved with atacicept did not translate into clinical efficacy in RA. In contrast, adalimumab had significant clinical benefit without reducing immunoglobulin and RF levels. Thus, BLyS, APRIL, their target cells (i.e., mature B cells and short-lived plasma cells), and autoantibodies may not be critical drivers of rheumatoid inflammation. Alternatively, more profound reductions in immunoglobulins and plasma cells than were achieved in this study may be required for an impact on clinical disease.
The B cell–targeting therapy rituximab (anti-CD20) has been shown to be effective in RA (5). Atacicept and rituximab target different B cell populations: rituximab targets all immature, mature, and memory B cells but does not directly affect antibody-producing plasma cells (29), whereas atacicept targets mature B cells and antibody-producing plasma cells (15, 30). It is possible that the depletion of B cell subsets that are not targeted by atacicept, such as immature cells, is necessary to achieve clinical effects in patients with established RA.
Another possible explanation for our findings is that the local synovial responses to atacicept were not as profound as those seen in the peripheral blood. Local B cell responses are thought to contribute to RA pathogenesis (31–33), and elevated synovial levels of BLyS and APRIL have been reported in a subset of patients (6, 11, 34, 35). In phase I trials, synovial examination in a small number of patients with RA treated with atacicept (6, 35) showed that atacicept does enter the synovium and form atacicept–BLyS complexes (6, 34). Indeed, synovial effects of rituximab may have predictive value (16, 33, 36–38). However, because synovial samples were not collected in the current study, the local effects of atacicept could not be verified. It is also possible that atacicept may benefit only an as-yet-undefined subpopulation of patients, such as those with elevated synovial BLyS and APRIL levels, making it difficult to detect clinical effects in an unselected patient population.
No new safety trends associated with exposure to atacicept were identified. The most frequent AEs represented common illnesses. Importantly, despite the clear effects of atacicept on B cells, the incidence of infectious AEs was only slightly higher among patients treated with atacicept than among those receiving placebo, and no serious infections were reported in either atacicept group. It is reassuring that after week 12, the median IgG level changed only slightly in the atacicept groups. This may suggest that long-lived plasma cells may be less dependent on BLyS and APRIL than are mature B cells. Consistent with this hypothesis, in the AUGUST I study of atacicept in RA, titers of specific antibodies conferring long-lived, protective immunity (against tetanus, pneumococcus, and diphtheria) reflected the decrease in total IgG (20, 21). Furthermore, little effect on protective immune status was observed (20, 21). It remains possible, however, that longer-term atacicept treatment could be associated with larger decreases in total and antigen-specific immunoglobulin levels. This will need to be evaluated in future clinical trials.
Tumors developed in 5 patients receiving atacicept, compared with none of the patients receiving placebo; 4 of these tumors were common and benign. Furthermore, no sign of an association between increased risk of cancer and atacicept has been detected in other studies to date.
Trial limitations must be considered when interpreting these results. The efficacy of B cell–targeting therapies may be limited to specific subgroups of patients, but no such patient profiling was undertaken in this study.
As in all trials with a negative outcome, it is possible that a true difference in the primary outcome was missed due to chance (Type II error). However, because this study was 90% powered, the risk of such an error was very low. It is possible that an ACR20-CRP response might not be a sufficiently sensitive measure to discriminate between patients with and those without a small response to treatment.
Regional differences in ACR20-CRP response rates were seen in all treatment arms, and the reason for this variation is unknown. Therefore, the regional variation seen may reflect differences in patient experiences prior to participation in the study, including access to care and adherence to background medication, which may be affected by practical, cultural, or economic differences. There may also be some variation in the application of measurement tools that have a subjective component.
The decision to include a placebo group is study specific. In this study, inclusion of a placebo group was intended to minimize the impact of confounding factors that may be associated with an active comparator group, such as difficulties in blinding and raised expectations for patient response without a placebo group. In the results of this study, regional differences in response were particularly highlighted among the placebo-treated patients. These regional differences may have important implications and should be considered when designing any future multinational trials in RA.
In this study, the primary end point (ACR20-CRP response) was not met. The confirmed biologic activity among those who completed the study, particularly with reference to immunoglobulin levels and circulating mature B cells and plasma cells, and the known safety profile may support further investigation of atacicept in other autoantibody-mediated diseases, such as SLE.
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. van Vollenhoven 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. Van Vollenhoven, Kinnman, Vincent, Wax.
Acquisition of data. Van Vollenhoven, Kinnman, Bathon.
Analysis and interpretation of data. Van Vollenhoven, Kinnman, Vincent, Bathon.
This trial was sponsored by Merck Serono SA – Geneva, which, in collaboration with ZymoGenetics Inc., is developing atacicept for various indications. Merck Serono SA – Geneva and ZymoGenetics Inc. facilitated the study design and the writing of the manuscript and reviewed and approved the manuscript prior to submission. The authors independently collected the data, interpreted the results, and had the final decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by the study sponsor.
We thank Andrea Plant, PhD (Caudex Medical; supported by Merck Serono SA – Geneva) for her assistance in the preparation of the manuscript. A. Di Cara and Y. Hyvert (Merck Serono SA – Geneva) contributed to the analysis and interpretation of the fluorescence-activated cell sorting data. G. de La Bourdonnaye (Merck Serono SA – Geneva) reviewed the manuscript for statistical accuracy. The manuscript was critically reviewed for accuracy by Eleni Stavridi, Anand Rajeswaran, Catherine Barbey, Claudia Pena Rossi, Corrine Barra, Henry Hess, and Valerie Meyrial (Merck Serono SA – Geneva) and Jennifer Johnson (ZymoGenetics Inc.).
Additional contributors to this study are as follows: M. A. Lazaro, J. Velasco, D. Fernández (Argentina); O. Zamani (Austria); M. Malaise (Belgium); A. Bookman, B. Nair, C. Thorne, R. Faraawi (Canada); C. Fuentealba, L. Massardo, P. Miranda, F. Radrigan, J. P. Riedemann (Chile); B. Curkovic, S. Grazio, T. Kehler, J. Morovic-Vergles, S. Milanovic, D. Rosic (Croatia); R. Becvar, P. Vavrincová, E. Dukoupilová (Czech Republic); P. Jarvinen, L. Paimela, M. Leirisalo-Repo (Finland); N. Zoghbi-Ziade (France); M. Bäuerle, J. Braun, M. Ronneberger, G. Bermester, H. Schulze-Koops, J. Wollenhaupt (Germany); B. Seriolo, G. Valesini, A. Doria (Italy); T. Arayssi, I. Uthman, H. Awada (Lebanon); A. Dudek, A. Filipowicz-Sosnowska, S. Jeka, J. Lacki, P. Sliwinska-Stanczyk, S. Sierakowski, L. Sczepanski (Poland); M. Viana de Queiroz (Portugal); C. Codranu, S. Rednic (Romania); O. Ershova, N. Lomareva, E. Nasonov, S. Yakushin (Russian Federation); N. Damjanov, A. Dimic (Serbia and Montenegro); K. Ooi Kong, J. Thumboo, K. Hon Yoon (Singapore); J. Ballina, J. J. Gómez-Reino, R. Sanmartí (Spain); S. C. Hsieh, J. C. Tseng, C. M. Huang (Province of Taiwan); Y. Nikirenkov, M. Stanislavchuk, S. Ter-Vartanyan (Ukraine); J. Fiechtner, G. Gladstein, M. Miniter, J. Taborn, R. Trapp, R. Ettlinger, M. Lowenstein, W. Shergy, R. White, J. Kaine, M. Sebai, W. Eider, K. Kolba, E. Fudman, E. Fung, J. Su, L. Moreland, R. Moskowitz, M. Stern, W. Edwards, A. Kivitz, P. Knibbe, J. Neal, A. Nussbaum, R. Rapaport, B. Samuels, R. Hymowitz, J. Starr, W. Muzutani (USA).