A Randomized, Double‐Blind, Placebo‐Controlled Trial of Abatacept for the Treatment of Relapsing, Nonsevere Granulomatosis With Polyangiitis

Abstract

Objective

To compare the efficacy of abatacept to placebo for the treatment of relapsing, nonsevere granulomatosis with polyangiitis (GPA).

Methods

In this multicenter trial, eligible patients with relapsing, nonsevere GPA were randomized to receive abatacept 125 mg subcutaneously once a week or placebo, both together with prednisone 30 mg/day (or equivalent), tapered and discontinued at week 12. Patients already taking methotrexate, azathioprine, mycophenolate, or leflunomide continued this medication at a stable dose. Patients achieving remission remained on their randomized assignment until relapse, early termination, or the common close date 12 months after enrollment of the last patient. Those who had a nonsevere relapse, had nonsevere worsening, or were not in remission by month 6 had the option to receive open‐label abatacept. The primary end point was the rate of treatment failure, defined as relapse, disease worsening, or failure to achieve a Birmingham Vasculitis Activity Score for Wegener's Granulomatosis (BVAS/WG) of 0 or 1 by six months.

Results

Sixty‐five patients were randomized; 34 received abatacept and 31 received placebo. No statistical difference in the treatment failure rate was found between those who received abatacept and those who received placebo (P = 0.853). Treatment with abatacept did not demonstrate any statistical difference from placebo in key secondary end points, including time to full remission (BVAS/WG = 0), duration of glucocorticoid‐free remission, relapse severity, prevention of damage, and patient‐reported quality‐of‐life outcomes. There was no difference in the frequency or severity of adverse events between treatment arms, including infection.

Conclusion

In patients with relapsing, nonsevere GPA, abatacept did not reduce the risk of relapse, severe worsening, or failure to achieve remission.

INTRODUCTION

Granulomatosis with polyangiitis (GPA) is a systemic inflammatory disease characterized by necrotizing, granulomatous inflammation and vasculitis of small‐ to medium‐sized arteries. ¹ Although current therapies can induce disease remission, relapses occur in 60% to 70% of patients with GPA, ² , ³ which in many instances, consists of nonsevere disease features. ¹ , ⁴ Although nonsevere disease activity is not immediately life‐threatening, it is associated with morbidity and diminished quality of life due to disease‐related damage or drug‐induced toxicity. For this reason, there is a great unmet need to provide safe and effective management options for nonsevere disease relapses in GPA that allow sparing of glucocorticoids.

Abatacept (CTLA‐4 Ig) is composed of the ligand‐binding domain of CTLA‐4 plus human Ig. Through this construct, abatacept carries the potential to modulate the costimulatory signal required for T cell activation. Based on laboratory data and genetic studies supporting a potential role for activated CD4⁺ T cells in the pathogenesis of GPA, ⁵ , ⁶ , ⁷ , ⁸ an open‐label study was conducted to examine the role of abatacept in nonsevere, relapsing GPA. In 20 patients with GPA treated with abatacept, 90% improved, 80% achieved remission, and 11 of 15 patients (73%) were able to fully discontinue prednisone. ⁹ Based on these encouraging findings, a randomized, double‐blind, placebo‐controlled trial was conducted to determine the efficacy and safety of abatacept in nonsevere, relapsing GPA.

PATIENTS AND METHODS

Trial conduct overview

The protocol for this investigator‐initiated randomized trial was written by the first and last authors. The original study protocol and all amendments were approved by an independent Data and Safety Monitoring Board (DSMB) and the institutional review boards (IRB) that were overseeing the study for individual sites. After trial commencement, changes were made to the sample size and statistical plan, as discussed in the following sections. Bristol Myers Squibb provided funding and the study drug for this trial (see Role of the Study Sponsor). This research was conducted in compliance with the Helsinki Declaration and is consistent with Good Clinical Practice, and all patients provided written informed consent.

Setting and participants

The trial was conducted at 22 academic medical centers that enrolled patients with relapsing, nonsevere GPA aged 15 years or older (see full list of participating centers in Appendix A). The diagnosis of GPA was based on a history of meeting at least two of the five following modified 1990 American College of Rheumatology classification criteria for GPA ¹⁰ : (1) nasal or oral inflammation, defined as the development of painful or painless oral ulcers or purulent or bloody nasal discharge; (2) abnormal chest radiograph, defined as the presence of nodules, fixed infiltrates, or cavities; (3) active urinary sediment, defined as microscopic hematuria (more than five red blood cells per high‐power field) or red blood cell casts; (4) granulomatous inflammation on the biopsy specimen, defined as histologic changes showing granulomatous inflammation within the wall of an artery or in the perivascular or extravascular area (artery or arteriole); and (5) positive test result for antineutrophil cytoplasmic antibody (ANCA) specific for antibodies to proteinase 3 (PR3) or myeloperoxidase (MPO). Eligible patients were required to have a relapse of nonsevere disease within the 28 days before screening and could not have severe disease manifestations, defined as those that would be scored as a major element in the Birmingham Vasculitis Activity Score for Wegener's Granulomatosis (BVAS/WG) ¹¹ or any disease features that posed an immediate threat to either a critical individual organ or the patient's life. Key exclusion criteria included the following: active infection (including chronic infection, herpes zoster within the past two months, HIV infection, hepatitis C virus, hepatitis B virus, tuberculosis); pregnancy; breastfeeding; cytopenias; creatinine clearance of ≤20 mL/min; elevation of the transaminase level more than three times above the upper limit of normal; live vaccination within the prior three months; history of any malignant neoplasm, except adequately treated basal or squamous cell carcinoma of the skin or solid tumors treated with curative therapy and disease free for at least five years; other uncontrolled disease (comorbidity) that could prevent a patient from fulfilling the study requirements or that would substantially increase the risk of study procedures; and prior use of specified immunosuppressive therapies within established time parameters (see Supplementary Table 1).

Interventions

Eligible patients were randomized to receive either abatacept 125 mg by subcutaneous injection once a week or placebo (Figure 1). Both patients and investigators were blinded to the randomization assignment. All patients were treated with prednisone 30 mg/day (or an equivalent glucocorticoid), which was tapered according to a standard schedule to be discontinued by week 12 (See Supplementary Table 2). Patients receiving a maintenance immunosuppressive medication of methotrexate, azathioprine, mycophenolate mofetil or sodium, or leflunomide at the time of enrollment continued these agents without a dosage increase. Patients were permitted to continue standard prophylactic medications they were receiving at the time of enrollment if this was felt to be clinically indicated by the study physician. After randomization, patients were seen for study visits at months 1 and 3 and every 3 months thereafter, during which a medical history, physical examination, and laboratory assessments were conducted, and patients completed quality‐of‐life outcome questionnaires. Patients continued their randomized treatment assignment of abatacept or placebo until the common closing visit unless they had not achieved remission by month 6, they experienced a relapse, or they met criteria for early termination. The common closing date was 12 months after randomization of the final participant.

Figure 1.

Study diagram of a randomized trial of abatacept in relapsing, nonsevere granulomatosis with polyangiitis.

Patients who experienced a severe relapse or severe worsening discontinued the study drug and were treated according to best medical judgment. Severity of relapse was based on the same definition for severe disease used to determine trial eligibility, as described in the Setting and participants section. Patients who experienced nonsevere disease worsening, who were not in remission by month 6, or who experienced a nonsevere relapse before the common closing date were eligible to receive open‐label abatacept for up to one year.

After discontinuation of the study drug, all patients were asked to return for a three‐month posttreatment safety visit. Reasons for early study drug termination not related to disease activity included the following: pregnancy or breastfeeding; development of malignancy, with the exception of basal or squamous cell carcinoma of the skin that had been completely excised; grade 4 toxicity; hypersensitivity to abatacept; noncompliance with study procedures; decision by the patient to decline further study treatment; and when, in the medical judgment of the physician, discontinuation would be in the best interest of the patient.

Randomization and blinding

Randomization was computer generated in a 1:1 allocation to receive abatacept or placebo. The study was stratified by PR3‐ and MPO‐ANCA status. To ensure balance in treatment groups within clinical sites, random number lists with a fixed block size were generated. Patients and all study investigators were blinded to the randomized treatment assignment.

Assessment and outcome measures

The assessment of disease activity was obtained in a standardized manner throughout the study based on clinical history, physician examination, and laboratory tests obtained at each study visit. Study visits and data collection occurred at screening, baseline, month 1, month 3, and every 3 months thereafter; at early termination/common close; at the 3‐month posttreatment visit; and at unscheduled visits necessary for assessment of disease activity or adverse events.

The primary end point was the rate of treatment failure, defined as a relapse, disease worsening, or failure to achieve a BVAS/WG of 0 or 1 by six months. Remission was defined as a BVAS/WG of 0 or 1 with disease improvement and without disease worsening. Disease improvement was defined by a reduction in the BVAS/WG. Relapse was defined as the occurrence of any of the following elements after remission: an increase in the BVAS/WG, development of a new BVAS/WG item, or symptoms/signs of GPA that could not be attributed to any cause other than GPA and that required initiation or a dosage increase of prednisone. Disease worsening was the occurrence of any of these listed elements before remission.

Six secondary end points were examined: time to full remission (BVAS/WG = 0), duration of glucocorticoid‐free remission, relapse severity, prevention of damage, patient‐reported quality‐of‐life outcomes, and safety. All adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0.

Statistical analysis and sample size

The original statistical plan for this trial was based on a projected accrual of 150 patients to achieve 136 randomized participants. In 2018, the enrollment rate was slower than expected due to the marked increased use of rituximab for the treatment of GPA. After careful review, a new plan for the statistical analysis was introduced, which allowed the study question to be answered using a smaller sample size. In this new statistical analysis plan, no changes were made to the planned 80% power or 40% reduction in the expected rate of treatment failure; therefore, the potential for a Type II error was not different between the two analysis plans.

For the primary end point, 60 evaluable patients (65 randomized to allow for dropout) had 80% power at a 0.05 one‐sided significance level to detect a 40% reduction in the expected failure rate for the abatacept arm as compared to the placebo arm, based on assuming that 60% of the placebo group would experience a relapse by 12 months. The primary analysis was based on intent‐to‐treat, with the cumulative incidence of treatment failure over time since randomization within each arm being estimated using the Kaplan‐Meier method and treatment differences tested by the log‐rank test. The Cox proportional hazard (PH) model, adjusted for age at diagnosis, duration of disease, sex, ethnicity, race, and ANCA stratum, was used to assess the difference between arms in the cumulative incidence functions and the associated hazard functions.

Secondary end points were analyzed in the following manner:

1. Severity of relapses in those treated with abatacept versus placebo: the median relapse BVAS/WG was determined for each treatment and compared using the Wilcoxon test. Additionally, the frequency of the BVAS/WG was compared for treatment differences using Fisher's exact test.

2. Time to achievement of full remission: The time from randomization to achievement of full remission was calculated for each participant. Participants who failed to achieve full remission were censored as of their last visit. Kaplan‐Meier curves and the log‐rank test were used to test for treatment differences in the time to achievement of full remission. Additionally, Cox models adjusted for covariates were used to test for treatment differences.

3. Duration of glucocorticoid‐free periods: Kaplan‐Meier curves and the log‐rank test were used to test for treatment differences in duration of glucocorticoid‐free periods. Additionally, Cox models adjusted for covariates were used to test for treatment differences.

4. Health‐related quality of life in those treated with abatacept versus placebo: the average Short Form 36 (SF‐36), Patient‐Reported Outcomes Measurement Information System (PROMIS), patient global assessment, and Routine Assessment of Patient Index Data 3 (RAPID3) scores were calculated for each treatment arm at each visit and compared using mixed modeling to account for subject effects, time, and treatment.

5. Prevention of disease‐ or treatment‐related damage with abatacept versus placebo: the Cumulative Damage Assessment (CDA) instrument ¹² was calculated at each visit and compared across treatment arms using Poisson regression.

6. Safety of abatacept in GPA: the frequency of each type of severe adverse event was compared for treatment differences using Fisher's exact test. Additionally, the rate of severe adverse events in both treatment arms was calculated and compared using Poisson regression. The study was powered for the primary end point only, and no adjustment for multiple end points was made.

Safety monitoring and stopping guidelines

The study conduct was overseen by an independent DSMB, and each site received oversight from a local IRB or an IRB of record, based on national regulatory requirements. Rules for halting new patient accrual were established before trial initiation and consisted of any deaths that were possibly, probably, or definitely related to the study drug or any grade 4 toxicities that were probably or definitely related to the study drug.

RESULTS

Patient population

From April 2015 through May 2022, there were 77 patients with GPA who signed informed consent and underwent protocol screening, of whom 12 were not eligible (Figure 2). The remaining 65 patients underwent blinded randomization, with 34 receiving abatacept and 31 receiving placebo. The trial met the required sample size enrollment target and ended after all patients completed the required study visits. The baseline and demographic features of the 65 randomized patients are listed in Table 1. The diagnosis of GPA was biopsy proven in all patients who were ANCA‐negative as well as the two patients who tested positive for antibodies to both PR3 and MPO. Fifty‐one patients (24 abatacept, 27 placebo) continued to receive a maintenance immunosuppressive agent that they were taking at enrollment (Table 1). At enrollment, trimethoprim 160 mg/sulfamethoxazole 800 mg twice a day (high dose) was being received by one patient randomized to abatacept and one patient randomized to placebo. Trimethoprim 160 mg/sulfamethoxazole 800 mg three times a week or trimethoprim 80 mg/sulfamethoxazole 400 mg once a day (low dose) was being received by 10 patients randomized to abatacept and 4 patients randomized to placebo. There was no difference in baseline characteristics, ANCA status, or prior use of nonglucocorticoid immunosuppressive medications between patients in the two study groups (Table 1). Of those who discontinued randomized treatment, 32 went on to receive open‐label abatacept (Figure 2). The following results focus solely on the randomized treatment period, from which the primary and secondary end points were analyzed.

Figure 2.

Consolidated Standards of Reporting Trials (CONSORT) diagram outlining the course of the 77 participants who signed informed consent in this trial.

Table 1.

Baseline demographics of the 65 randomized patients^*

	Abatacept (n = 34)	Placebo (n = 31)	P value
Age at randomization, median (IQR), y	54.26 (45.05–60.07)	53.15 (45.95–61.34)	0.7877
Sex, % (n)
Female	61.8 (21)	54.8 (17)	0.5714
Male	38.2 (13)	45.2 (14)
Race, % (n)
American Indian/Alaska Native	2.9 (1)	0.0 (0)	0.4022
Asian	2.9 (1)	0.0 (0)
Unknown or not reported	0.0 (0)	3.2 (1)
White	94.1 (32)	96.8 (30)
Disease duration, median (IQR), y	6.48 (2.87–15.99)	4.59 (2.44–10.10)	0.4717
ANCA type, % (n)
Negative	14.7 (5)	6.5 (2)	0.6912
Anti‐MPO	17.6 (6)	16.1 (5)
Anti‐PR3	64.7 (22)	71.0 (22)
Both anti‐PR3 and anti‐MPO	2.9 (1)	3.2 (1)
Unknown	0.0 (0)	3.2 (1)
Characteristics of active disease at randomization, % (n)
Constitutional	61.8 (21)	56.7 (17)	0.6786
Musculoskeletal	44.1 (15)	36.7 (11)	0.5447
Skin	14.3 (1)	10.0 (1)	0.7872
Mucous membrane	2.9 (1)	0.0 (0)	0.3438
Eyes	2.9 (1)	16.7 (5)	0.0905
Ear, nose, and throat	79.4 (27)	72.4 (21)	0.5157
Lung	23.5 (8)	17.2 (5)	0.5387
Kidney (hematuria)	11.8 (4)	26.7 (8)	0.1275
Other	5.9 (2)	3.7 (1)	0.6959
BVAS/WG, median (IQR)	3.00 (2.00–4.00)	3.00 (2.00–4.00)	0.5820
Previous treatments, % (n)
Cyclophosphamide (oral)	36.4 (12)	35.5 (11)	0.9416
Cyclophosphamide (intravenous)	32.4 (11)	41.9 (13)	0.4240
Methotrexate	91.2 (31)	67.7 (21)	0.0183
Azathioprine	64.7 (22)	45.2 (14)	0.1134
Mycophenolate	27.3 (9)	25.8 (8)	0.8944
Leflunomide	7.1 (2)	3.3 (1)	0.5127
Cyclosporine	3.0 (1)	3.2 (1)	0.9642
Etanercept	0.0 (0)	3.2 (1)	0.4844
Infliximab	2.9 (1)	6.5 (2)	0.5005
Rituximab	50.0 (17)	36.7 (11)	0.2833
Concurrent treatment at randomization, % (n)
Methotrexate	44.1 (15)	56.7 (17)	0.3164
Azathioprine	23.5 (8)	19.4 (6)	0.6826
Mycophenolate	0.0 (0)	12.9 (4)	0.0495
Leflunomide	3.6 (1)	0.0 (0)	0.3045

^*ANCA, antineutrophil cytoplasmic autoantibody; BVAS/WG, Birmingham Vasculitis Activity Score for Wegener's Granulomatosis; IQR, interquartile range; MPO, myeloperoxidase; PR3, proteinase 3.

Primary end point assessment

Treatment failure occurred in 22 of 34 patients randomized to abatacept compared to 23 of 31 who received placebo. There was no statistical difference between abatacept and placebo by the log‐rank test (P = 0.255) (Figure 3). There was similarly no difference by Cox PH model: unadjusted (P = 0.308) or adjusted (P = 0.326). Of the treatment failures, relapse or worsening that consisted of predominantly nonsevere disease features occurred in 21 patients treated with abatacept and 21 patients who received placebo (Table 2, Supplementary Table 3, and Supplementary Figure 1).

Figure 3.

Treatment failure in 22 of 34 patients randomized to abatacept compared to 23 of 31 who received placebo.

Table 2.

Clinical features of the randomized patients who experienced relapse or worsening^*

	Abatacept (n = 21)	Placebo (n = 21)	P value
Characteristics of active disease at relapse/worsening, % (n)
Constitutional	70.0 (14)	52.4 (11)	0.2477
Musculoskeletal	50.0 (10)	57.1 (12)	0.6466
Skin	5.3 (1)	14.3 (3)	0.3422
Mucous membrane	5.0 (1)	9.5 (2)	0.5782
Eyes	10.0 (2)	19.0 (4)	0.4126
Ear, nose, and throat	75.0 (15)	81.0 (17)	0.6453
Lung	20.0 (4)	33.3 (7)	0.3355
Kidney	10.0 (2)	30.0 (6)	0.2351
Nervous system	10.0 (2)	0.0 (0)	0.2317
Other	5.3 (1)	0.0 (0)	0.3109
BVAS/WG at relapse, median (IQR)	3.00 (2.00–4.00)	3.00 (2.00–4.00)	0.9796
Severe relapse or worsening, % (n)	14.3 (3)	23.8 (5)	0.4319

^*BVAS/WG, Birmingham Vasculitis Activity Score for Wegener's Granulomatosis; IQR, interquartile range.

Secondary end point assessment

1. Severity of relapses

Severe relapse or worsening occurred in three patients treated with abatacept and five who received placebo. There was no statistical difference in the severity of relapse in those who received abatacept or placebo either by categorical comparison of whether the relapse was severe, based on BVAS/WG (Table 2), or by the Wilcoxon test (P = 0.980) or Fisher's exact test (P = 0.858), based on the comparison of the continuous values of the BVAS/WG.

2. Time to achievement of full remission

Full remission (BVAS/WG = 0) was achieved in 42 participants, of whom 22 were randomized to abatacept and 20 were randomized to placebo. The median time to full remission (BVAS/WG = 0) was 91 (interquartile range [IQR] 77–176) days in the placebo group and 91 (IQR 39–172) days in the abatacept group. There was no difference in the time to full remission in those who received abatacept or placebo (P = 0.944, log‐rank test; see Supplementary Figure 2). There was similarly no difference by Cox PH model: unadjusted (P = 0.945) or adjusted (P = 0.867).

3. Duration of glucocorticoid‐free periods

The median number of glucocorticoid‐free days was 42 (IQR 8–358, full range 0–1,016) days in the placebo group and 31.5 (IQR 0–280, full range 0–2,772) days in the abatacept group. There was no difference in the duration of glucocorticoid‐free periods in those who received abatacept or placebo (P = 0.515, log‐rank test; see Supplementary Figures 3 and 4). There was similarly no difference by Cox PH model: unadjusted (P = 0.545) or adjusted (P = 0.549). Of the 45 patients who experienced treatment failure, there were 20 patients (8 abatacept, 12 placebo; P = 0.1854) who were able to discontinue glucocorticoids before treatment failure occurred. Of those who did not experience treatment failure, all except one patient discontinued glucocorticoids.

4. Health‐related quality of life

A total of 21 measures examining patient quality of life were collected during this trial using the following instruments: SF‐36 (physical component T score, mental component T score, physical functioning, role physical, bodily pain, general health results vitality, social functioning, role emotional, mental health), PROMIS (anxiety, fatigue, depression, sleep disturbance, satisfaction with social roles, pain interference, physical function), patient global assessment (today), patient global assessment (past 28 days; see Supplementary Figure 5), RAPID3 cumulative score, and RAPID3 scale. No differences between abatacept and placebo were seen in any measure. Additionally, no trends were seen in any of these measures to suggest a benefit for abatacept in any of these domains.

5. Prevention of disease‐ or treatment‐related damage

There was no difference in the prevention of disease‐ or treatment‐related damage as assessed by the CDA in those who received abatacept or placebo (P = 0.238).

6. Safety/Adverse events

During the randomized period of the trial, 71 adverse events occurred in 36 patients. This included 31 serious adverse events (SAEs) in 20 patients (Table 3). There was no difference in the frequency or severity of adverse events between treatment arms, including the rate of infection or the rate of SAEs. No deaths occurred while on study.

Table 3.

Description of the 31 serious adverse events that occurred during the randomized treatment period

Serious adverse event	Abatacept (n = 31)	Placebo (n = 34)	P value
Anemia	0	1	0.476
Atrial fibrillation	1	0	1.00
Eye disorder	1	0	1.00
Flu‐like symptoms	0	1	0.476
Allergic reaction	0	1	0.476
Autoimmune disorder	0	1	0.476
Malignancy	0	2	0.223
Stroke	1	0	1.00
Bladder perforation	1 (2 events)	0	1.00
Bronchial obstruction	0	1	0.476
Bronchopulmonary hemorrhage	0	1	0.476
Dyspnea	1	0	1.00
Surgical procedure	2	0	0.493
Thromboembolic event	0	1 (2 events)	0.476
Vasculitis	1	2	0.601
Infections
Upper respiratory	1	0	1.00
Bronchial	0	1	0.476
Lung	0	1 (2 events)	0.476
Skin	1	0	1.00
Urinary catheter related	1	0	1.00
Other infection	2	2	1.00

During the trial, 2 malignancies (0 abatacept, 2 placebo) and 46 infections (25 abatacept, 21 placebo) were reported, which included 2 cases of herpes zoster (1 abatacept, 1 placebo). COVID‐19 infection occurred in five patients (three abatacept, two placebo). Study sites were asked to report all episodes of a proven COVID‐19 infection as an SAE even when the patient was managed outside the hospital and did not meet any other SAE criteria. Although this likely increased the overall number of SAEs, there was no difference seen between abatacept or placebo in terms of infections or SAEs. Administration of the blinded study drug abatacept or placebo was held upon a confirmed diagnosis of COVID‐19 and was restarted following clinical improvement; no patients experienced disease relapse or worsening immediately following this infection and the brief holding of study medication.

DISCUSSION

For patients with GPA who have nonsevere, relapsing disease, effective management that allows discontinuation of glucocorticoids and reduces future relapses remains an important unmet need. A number of medications have been studied in nonsevere disease, yielding valuable information. Trimethoprim/sulfamethoxazole was examined in several studies conducted during the 1980s to 1990s. These investigations differed from the current trial in focusing on disease isolated to the upper and or lower respiratory tract, ¹³ being open‐label observational cohorts, ¹⁴ and examining the rate of relapse in those who received trimethoprim/sulfamethoxazole compared to placebo following induction of remission with glucocorticoids and cyclophosphamide. ¹⁵ Although these studies suggested benefit in some patients, severe and nonsevere relapses continued to occur in a significant number of patients. Methotrexate was explored as an alternative to cyclophosphamide for patients with active nonsevere disease. ¹⁶ , ¹⁷ Mycophenolate mofetil was similarly compared to cyclophosphamide in patients with mild to moderate disease activity. ¹⁸ Although both of these drugs were found to be noninferior to cyclophosphamide, relapses continued to occur as they have following induction of remission with cyclophosphamide or rituximab. ¹ , ² It is these occurrences of relapsing, nonsevere disease, often requiring long‐term use of glucocorticoids, that highlights the need for novel approaches in nonsevere disease aimed at preventing relapse and allowing discontinuation of glucocorticoids.

This study found that in patients with relapsing, nonsevere GPA, the addition of abatacept to glucocorticoids did not reduce the risk of relapse, severe worsening, or failure to achieve remission. This study additionally found that treatment with abatacept did not demonstrate any statistical difference from placebo in key secondary end points, including relapse severity, time to full remission, duration of glucocorticoid‐free remission, prevention of damage, or patient‐reported quality of life. Treatment with abatacept was not associated with increased toxicity beyond glucocorticoids.

The findings from this trial were contrary to the positive impressions seen in the open‐label study that was the basis for conducting this new trial. It is possible for individual patients to have prolonged periods of remission, which in small numbers can suggest therapeutic efficacy. These differing results illustrate why even the most promising results found in open‐label studies need to be examined in a randomized trial to determine efficacy. Although the current findings do not support the use of abatacept in GPA, these findings should not deter exploration of the role of T cells and T cell activation in the pathophysiology of GPA and investigation of novel therapies for which there is a supportive safety and efficacy rationale.

There were a number of important strengths of this trial. This interventional study was conducted at 22 expert international centers. The trial was double‐blinded to the investigational agent and focused on the treatment of relapsing, nonsevere GPA, a subgroup that has rarely been specifically examined in randomized trials. The enrollment of 65 eligible patients represents a moderately large sample size for such a subgroup of patients with a rare disease. Additionally, this trial included patients with high unmet need who had experienced one or more relapses despite other remission maintenance treatments.

This study has certain limitations. It is possible that this study failed to detect a difference between treatment arms that actually exists, which could have been impacted by a number of factors. This study sought to find a large difference that would be clinically meaningful such that a smaller difference may not have been detected. Although this would require a much larger sample size, this also raises the question of whether a substantively smaller effect of a biologic agent would be clinically meaningful. The long disease duration also raises the question as to whether the clinical features present at enrollment reflected damage rather than active disease, which could have affected the trial results. Arguing against this is that only 3 patients did not achieve BVAS = 0 or 1 within 6 months with all remaining patients having disease remission that persisted, or experiencing worsening or relapse of their disease. Because damage stays fixed in the face of treatment, these changes in the level of disease activity indicate that the manifestations present at enrollment represented active inflammation rather than damage.

In conducting this trial, much was learned about the challenges of conducting a randomized trial in patients with GPA with nonsevere disease. Although the evaluation of active nonsevere GPA used in this trial reflected standard clinical practice, nonsevere manifestations can be difficult to assess in terms of disease activity. Differentiating active sinonasal disease from damage or infection and assuring that musculoskeletal symptoms and fatigue reflect active disease rather than other causes, particularly glucocorticoid withdrawal, represent some of the greatest difficulties often faced by clinicians. No patient solely had fatigue or other constitutional manifestations as the only feature of relapse. In every instance, fatigue was accompanied by active disease in another organ system, which provided further support that this symptom was related to relapse. This experience provides further evidence of the need to continue the investigation of clinical, laboratory, and imaging biomarkers as well as patient‐reported outcomes to better assess nonsevere disease activity in GPA in conjunction with the pursuit of novel treatment options.

In conclusion, the results of this trial found that the addition of abatacept to prednisone did not reduce the rate of treatment failure in patients with relapsing, nonsevere GPA compared to placebo, but it was not associated with any increased risk of toxicity.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr Langford confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

ROLE OF THE STUDY SPONSOR

Bristol Myers Squibb reviewed the protocol and results but had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Bristol Myers Squibb.

Supporting information

Disclosure Form:

Data S1 Supporting Information

APPENDIX A. VASCULITIS CLINICAL RESEARCH CONSORTIUM AND EUROPEAN VASCULITIS SOCIETY MEMBERS

The additional study collaborators at each site are as follows: Boston University: Eric Stratton, Marcin Trojanowski, Chris Zammitti, and Jessica Ziemek; Cedars‐Sinai Medical Center: Javier Ache, Marianne Bernardo, Bryan Gonzalez, Mariko Ishimori, Bonnie Paul, Mahbuba Tusty, Swamy Venuturupalli, Daniel Wallace, and Michael Weisman; Cleveland Clinic: Debora Bork, Tiffany Clark, Sonya Crook, Kathleen Gartner, Lori Strozniak, Elizabeth Kisela, Marcia Leon, Loretta Williams, Jeff Negrey, Hannah Thome, and Andrew Wiecek; Hospital for Special Surgery: Beemnet Amdemicael, Emily Bakaj, Aliza Bloostein, Annel Fernandez, Chris Hatzis, and Lindsay Lally; Mayo Clinic: Misbah Baqir, Cynthia Beinhorn, Shannon Daley, Sue Donlinger, Samantha Fatis, Yeoniee Kim, Matthew Koslow, Matthew J. Koster, Kathleen Mieras, Tobias Peikert, Michael Stachowitz, and Gwen E. Thompson; Medius Klinik Kirchheim, University of Tuebingen: Nicole Hollinger, Nina Kempiners, Hartmut Mahrhofer, Anke Reichelt de Tenorio, and Bastian Walz; Mount Sinai Hospital and University of Toronto: Banita Aggarwal, Simon Carette, Nazrana Haq, Samyukta Jagadeesh, Suneet Khurana, and Judy Vendramini; Nottingham University Hospital NHS Trust: Amanda Butler, Mithun Chakravorty, Paige Draper, Cattleya Godsave, Marie‐Josèphe Pradère, Megan Rutter, and Alan Thomas; Oregon Health and Science University: Rebekah Alexander, Cong‐Qiu Chu, Daniela Ghetie, Amy Jeon, Michelle Jin, Kim Phung Nguyen, Brenda Olmos, Kelly Peel, Rosy Quinn, and Cailin Sibley; Royal Berkshire Hospital: Mayamol Joseph and Deepa Thapa; St Joseph's Healthcare and McMaster University: Sandra Messier and Diane Robins; St Paul's Hospital and University of British Columbia: Jayamarx Jayaraman and Kam Shojania; St Vincents University Hospital: Richard Conway, Kathy Brickell, and Phil Gallagher; University of Alberta: Alison Clifford, Sylvia Gaucher, Ainslie Hildebrand, Karen Martins, and Breanne Stewart; University of Calgary: Stephen Chaudary, Namneet Sandhu, and Stephen Williams; University of Cambridge: Karen Dahlsveen, Seerapani Gopaluni, Cecilia Matara, Mark McClure, and Rona Smith; University of Kansas Medical Center: Amrita Bath, Leslie Glasco, Theresa Howard, Caitlin McMillian, Na Yu, and Mingcai Zhang; University of Michigan: Ruba Kado, Emily Lewis, W. Joseph McCune, Emily Siegwald, and Jennifer Sylvester; University of Pennsylvania: Naomi Amudala, Kathryn Doyle, Matthew MacDonald, Leah Madden, Brian Rice, Antoine Sreih, and Sally Thompson; University of Pittsburgh: Laurie Hope, Kimberly Liang, Douglas Lienesch, Niveditha Mohan, and Kelly Reckley; University of South Florida: John Carter, Gail Lewis, and Michelle Orzechowski; University of Utah: Martha Finco, Jennifer Godina, Jessica Gonzalez, Julieanne Hall, Nereida Ortez, and Gopi Penmetsa.