Abatacept for the treatment of myositis-associated interstitial lung disease

Abstract

Objectives

This randomized, placebo-controlled pilot trial evaluated the efficacy and safety of abatacept in patients with anti-synthetase syndrome-associated interstitial lung disease (ASyS-ILD).

Methods

Participants with active ASyS-ILD were randomized to receive abatacept (n = 9) or placebo (n = 11) for 24 weeks, followed by a 24-week open-label extension with abatacept for all participants. The primary endpoint was a change in % predicted forced vital capacity (%FVC) from baseline to week 24. Secondary endpoints included changes in the FVC (ml), % predicted diffusing capacity of the lung for carbon monoxide (%DLCO), shortness of breath questionnaire (SOBQ) and pulmonary disease activity on a visual analogue scale (VAS) at weeks 24 and 48. Pre-post baseline analysis of FVC and quantitative image analysis (QIA) of high-resolution computed tomographic scans were performed. Data were analysed using a generalized linear mixed model. The study was not powered for primary or secondary endpoints.

Results

At week 24, there was no significant difference in the primary endpoint of %FVC change between abatacept and placebo (between treatment difference of −0.35, 95%CI −6.91 to 6.21, P = 0.914) and in all secondary endpoints. However, by week 48, trends favouring abatacept in %FVC, FVC (ml), %DLCO and SOBQ were observed without statistical significance. There was a significant improvement in pulmonary disease activity VAS and pre-post baseline slopes of %FVC and QIA scores in the abatacept arm. Abatacept was generally well tolerated.

Conclusion

Abatacept did not significantly improve %FVC at 24 weeks. However, trends at 48 weeks suggest potential benefits, supporting the need for a larger, long-term randomized controlled trial.

Clinical trial registration

clinicaltrials.gov; NCT03215927

Rheumatology key messages.

• Abatacept did not significantly improve lung function in ASyS-ILD at 24 weeks compared to placebo.

• Abatacept showed favourable trends in some pulmonary parameters and patient-reported outcomes over 48 weeks.

• The study supports the need for larger, long-term trials to evaluate abatacept’s role in ASyS-ILD.

Introduction

Idiopathic inflammatory myopathies (IIMs) are a rare, heterogeneous group of systemic autoimmune rheumatic diseases (SARDs) characterized by muscle weakness and systemic organ involvement. Anti-synthetase syndrome (ASyS) is a distinct subtype of IIMs, defined by autoantibodies against aminoacyl-tRNA synthetases and a triad of interstitial lung disease (ILD), myositis and inflammatory polyarthritis [1]. Several definitions of ASyS exist in the literature [2–5]. ILD is a predominant clinical feature in ASyS with an estimated prevalence of 67–100% and is associated with significant mortality [6–8].

For the treatment of IIM-associated ILD (IIM-ILD), immunosuppression with glucocorticoid and various steroid-sparing agents remains the mainstay of therapy [9, 10]. However, the evidence for treatment is lacking due to the paucity of prospective studies and randomized controlled trials (RCTs).

T cells play a key role in the pathogenesis of IIMs [11], and T cell-mediated pathogenesis has been implicated in IIM-ILD, as suggested by study showing activated T cells in bronchoalveolar lavage (BAL) samples from affected patients [12]. In a murine model of hypersensitivity pneumonitis, characterized by activated T cells in the lung parenchyma similar to IIM-ILD, blockade of T cell co-stimulation by CTLA-4 immunoglobulin (Ig) ameliorated lung inflammation [13]. Moreover, treatment with CTLA-4 Ig led to a decrease in lung damage and fewer inflammatory cells in the BAL fluid of another murine hypersensitivity pneumonitis model [14].

Prior T cell-targeted therapies, including ciclosporin and tacrolimus, have demonstrated efficacy in treating IIM-ILD, especially in ASyS patients [15–20]. Use of these therapies remains limited due to toxicity, therapeutic drug monitoring requirement and patient intolerance [21].

Abatacept, a soluble fusion protein comprising CTLA-4 and the Fc portion of Ig, inhibits T cell activation. Its effectiveness and safety profile have been established in various SARDs and is FDA-approved for rheumatoid arthritis (RA), psoriatic arthritis and juvenile idiopathic arthritis [22]. For SARD-ILD, the most evidence for abatacept relates to RA, where reports suggest no worsening of RA-associated ILD (RA-ILD) with abatacept although its efficacy is uncertain [23–26]. Abatacept has shown promise in refractory adult IIM, although pulmonary outcomes were not assessed in these studies [27–32].

The aim of this study was to evaluate the efficacy, safety and tolerability of weekly subcutaneous abatacept in patients with ASyS-ILD via a multicentre, double-blind, randomized, placebo-controlled pilot trial. We anticipate our findings will serve as proof-of-concept for future clinical trials of abatacept and novel therapeutics in IIM-ILD.

Methods

Trial design and oversight

The ATtackMy-ILD trial was a 24-week RCT of abatacept or placebo in ASyS-ILD patients followed by a 24-week open-label extension (OLE) phase at six clinical sites in the USA (ClinicalTrials.gov: NCT03215927): University of Pittsburgh, Northwell Health, Cedars-Sinai Medical Center, Brigham and Women’s Hospital, John Hopkins Medical Center and University of Colorado (Fig. 1A). The trial was approved by University of Pittsburgh and each centre’s Institutional Review Board (IRB). All subjects provided written informed consent prior to trial entry.

Figure 1.

Schematic diagram of the Abatacept for the Treatment of Myositis-associated Interstitial Lung Disease (ATtackMy-ILD) trial. (A) Design of the ATtackMy-ILD trial. (B) Participant flow diagram. SOC: standard of care treatment

Participants

Recruitment began in June 2017 and ended in May 2021. Eligible participants were at least 18 years of age and had ASyS defined as possessing one ASyS antibody in the presence of ILD, confirmed by a high-resolution computed tomographic (HRCT) scan within 12 months, showing reticulation or honeycombing or ground glass opacities (GGO) without another plausible explanation. Subjects had to have active ILD and previously failed at least one standard of care (SOC) immunosuppressive (IS) treatment. Active ILD was defined as new-onset ILD within 3 months or chronic worsening ILD necessitating a treatment change with documented worsening of at least two of three parameters: patient-reported worsening dyspnoea, worsening of chest HRCT or a relative decline in the forced vital capacity (FVC) as a percentage of the predicted normal value (%FVC) of >10%. At the time of enrolment, subjects had to have a minimal threshold of ILD severity defined as %FVC of <80%, or %FVC of 80% to100% combined with a relative decline in the %FVC of >10% in the last 12 months. Active myositis was not required.

Allowable SOC treatments included glucocorticoids (≤60 mg/day of prednisone equivalent), or one of the allowed IS agents (either mycophenolate ≤3 g/day or azathioprine ≤200 mg/day), or a combination of glucocorticoid and one allowed IS agent. No other concomitant IS agents, IVIg or biologic agents were allowed.

Patients who had severe end-stage lung disease, active infections, previous treatment with abatacept and concomitant interfering pulmonary disease were excluded from the trial. Further details of key inclusion and exclusion criteria are listed in Supplementary Data S1, available at Rheumatology online.

Randomization and intervention

Participants were randomly assigned in a 1:1 ratio to receive subcutaneous abatacept at a dose of 125 mg weekly or placebo with their SOC treatments for 24 weeks with a targeted enrolment of 20 subjects. Active muscle disease was treated as a stratification variable, and a ‘balanced coin’ approach was employed to control treatment assignment within enrolment sites. A 24-week OLE phase was conducted to assess the durability of the response to the study drug and its safety.

A forced glucocorticoid taper began within 1 week after beginning the study drug. The goal was to taper glucocorticoid to 5 mg or lower dose within 8 weeks. The recommended regimen is provided in Supplementary Data S2, available at Rheumatology online.

If there were flare of disease meeting the myositis or ILD worsening criteria (Supplementary Data S3 and S4, available at Rheumatology online), an increase in therapy, ≤20 mg/day of prednisone or equivalent, was allowed as rescue therapy. The addition of a new IS agent or an increase of the dose of IS therapy or glucocorticoid above baseline after the initial 8 weeks of the study drug was considered a treatment failure.

Clinical evaluation and safety laboratory testing were performed at baseline, weeks 12, 24, 36 and 48. Pulmonary function tests (PFTs) were performed at baseline, weeks 12, 24 and 48. FVC and diffusing capacity of the lung for carbon monoxide (DLCO) were assessed in accordance with the European Respiratory Society (ERS)/American Thoracic Society (ATS) standards [33–35]. The percentage of the predicted normal value in diffusing capacity of the lung (%DLCO) was calculated centrally using age, sex, height and haemoglobin collected [36]. HRCT scans were performed at baseline, weeks 24 and 48, using a non-contrast thin section protocol at each institution. Pre-baseline HRCT scans and PFTs in the last 1 year were also obtained. HRCT images were assessed by quantitative image analysis (QIA) using previously published methods [37–39]. The QIA scores were assigned as a percentage of the extent of ground glass (QGG), fibrotic patterns (QLF), honeycombing (QHC) and consolidations (QCON). The quantitative ILD score (QILD) was the sum of QGG, QLF, QHC and QCON. Safety was assessed based on adverse events reported, irrespective of causality, over 48 weeks using Common Terminology Criteria for Adverse Events (CTCAE) version 4.0.

Endpoints

The primary efficacy endpoint was the absolute change in the %FVC from the baseline to week 24. Secondary endpoints were absolute change from baseline to week 24 in (1) other PFT parameters, including FVC in milliliters (ml) and %DLCO; (2) patient-reported outcomes, including the total score on the University of California San Diego Shortness of Breath Questionnaire (SOBQ) [40], Short Form 36 (SF-36) [41] and HAQ-Disability Index (HAQ-DI); (3) 6-min walk distance (6MWD); (4) pulmonary disease activity on a 0–10 cm visual analogue scale (VAS); (5) HRCT QIA scores; (6) glucocorticoid dose; (7) proportion of participants with treatment failure, participants with disease progression or treatment failure; (8) proportion of participants with 10%FVC changes; (9) proportion of participants who met SOBQ minimal clinically important difference (MCID) without treatment failure by week 24 [42]. All endpoints were also assessed at the end of the OLE phase (week 48). Details of key measures are provided in Supplementary Data S5, available at Rheumatology online.

The progression definition was the first occurrence of any of the following: death, lung transplantation, decline in %FVC ≥10% or decline in %FVC ≥5% with a decline in %DLCO ≥15%. The 10% FVC change was defined as a participant who had stable (within 10% increase or decrease from baseline) or improved >10% from baseline without treatment failure. We also evaluated time to progression-free survival, and time to improvement in %FVC ≥10% over the 48-week period.

Statistical analysis

Analysis was based on the intention-to-treat principle. The PFT parameters, patient-reported outcomes, 6MWD, pulmonary VAS score, QIA scores and glucocorticoid dose, were analysed with the use of a generalized linear mixed model for the treatment-by-time interaction effect, adjusting the effects of baseline FVC, age and sex. The SOBQ score was additionally adjusted for baseline SOBQ. Sensitivity analyses further adjusting for treatment failure, Manual Muscle Testing 8 (MMT8) and BMI were done for FVC and DLCO parameters, and further adjusting for MMT8 were done for 6MWD. For time to progression-free survival and time to improvement in %FVC ≥10%, we generated Kaplan–Meier survival curves for each treatment arm up to 48 weeks. Longitudinal trajectories of pre-baseline and post-baseline FVC, DLCO and QIA scores with projected values based on pre-baseline and post-baseline slopes for FVC and DLCO at 48 weeks were estimated using a generalized linear mixed model. The sample size of 20 subjects was justified based on this trial being a proof-of-concept pilot study.

The data were collected and analysed at the data coordinating centre. The statistical analyses were 2-sided with a significance threshold of 0.05. All analyses were performed using SAS version 9.4 (RTP, NC).

Results

Baseline characteristics

A total of 20 participants were randomized (9 in abatacept arm, 11 in placebo arm). All participants completed the 24-week RCT phase (Fig. 1B). In the OLE phase, one participant died from respiratory failure. Most subjects (n = 12) were enrolled at the coordinating centre (University of Pittsburgh), followed by Cedar-Sinai Medical Center (n = 3), Brigham and Women’s Hospital (n = 3), and Northwell Health (n = 2). There were some non-statistical significant baseline imbalances for age, sex, disease duration, some PFT parameters and prior medication use (Table 1).

Table 1.

Baseline characteristics of the participants

Clinical variable n (%) or median (IQR)	Abatacept (n = 9)	Placebo (n = 11)
Age (years)	49.6 (45.6–59.3)	57.6 (47.8–60.8)
Female	5 (56)	4 (36)
Ethnic origin
White	7 (78)	10 (91)
African American	1 (11)	1 (9)
Asian	1 (11)	0 0
ILD disease duration (months)	39.0 (6.0–67.0)	17.0 (3.0–52.0)
ILD status
New onset	2 (22)	3 (27)
Chronic worsening	7 (78)	8 (73)
ILD pattern on HRCT (n = 17)	(n = 7)	(n = 10)
Nonspecific interstitial pneumonia	2 (29)	6 (60)
Organizing pneumonia	2 (29)	2 (20)
Indeterminate for UIP	2 (29)	2 (20)
Other	1 (14)	0 0
HRCT scores with quantitative image analysis
Whole lung QIA, %
QILD	43.7 (26.3–61.8)	36.3 (28.3–45.0)
OGG	19.6 (13.2–29.3)	18.9 (14.9–22.5)
QLF	14.3 (10.3–38.4)	17.0 (13.4–22.4)
QHC	0.0 (0.0–0.4)	0.2 (0.0–0.9)
QCON	0.2 (0.1–0.8)	0.2 (0.1–0.4)
Myositis status
Active	5 (56)	4 (37)
Inactive	4 (44)	7 (64)
Manual muscle test	150.0 (144.5–150.0)	148.0 (132.0–150.0)
Oxygen requirement	3 (33)	3 (27)
Anti-synthetase antibody
Jo-1	4 (44)	7 (64)
Non-Jo-1	5 (56)	4 (36)
PL-12	1 (11)	1 (9)
PL-7	2 (22)	1 (9)
KS	0 0	0 0
OJ	1 (11)	1 (9)
EJ	1 (11)	0 0
Zo	0 0	1 (9)
Forced vital capacity
Value (ml)	2160.0 (1995.0–2835.0)	2630.0 (2160.0–3510.0)
Percent of predicted value	66.0 (49.0–68.5)	61.0 (53.0–80.0)
Diffusing capacity of lungs for carbon monoxide
Percent of predicted value	52.3 (40.9–58.4)	48.5 (41.4–61.4)
6-min walk distance (meter)	488.0 (324.5–543.5)	336.0 (320.0–466.0)
Pulmonary disease activity on VAS	5.3 (3.0–6.5)	3.5 (2.6–5.0)
Patient-reported outcome measure scores
SOBQ	29.0 (18.5–54.0)	33.0 (24.0–54.0)
Short form-36 physical component	29.2 (23.7–39.6)	31.5 (27.5–37.7)
Short form-36 mental component	50.1 (44.0–55.2)	52.4 (43.4–56.9)
HAQ-DI	0.4 (0.2–0.7)	0.5 (0.3–0.9)
Previously failed medications
Prednisone	9 (100)	11 (100)
Mycophenolate	4 (44)	9 (82)
Rituximab	1 (11)	2 (18)
IVIg	1 (11)	1 (9)
Pirfenidone	0 0	0 0
Methotrexate	1 (11)	1 (9)
Azathioprine	2 (22)	2 (18)
Tacrolimus	2 (22)	2 (18)
Cyclophosphamide	1 (11)	0 0
Hydroxychloroquine	1 (11)	0 0
Number of previously failed medications	2.0 (1.0–3.5)	2.0 (2.0–3.0)
Concomitant immunosuppressive drugs
Mycophenolate	7 (78)	8 (73)
Azathioprine	1 (11)	0 0
Baseline prednisone dose (mg/day)	20.0 (7.5–30.0)	10.0 (10.0–30.0)

HAQ-DI: HAQ-Disability Index; HRCT: high-resolution computed tomography; ILD: interstitial lung disease; IQR: interquartile range; mg: milligram; ml: milliliter; QCON: quantitative score of consolidations; QGG: quantitative score of ground glass; QHC: quantitative score of honeycombing; QIA: quantitative image analysis; QILD: quantitative interstitial lung disease score; QLF: quantitative score of fibrotic patterns; SOBQ: University of California San Diego Shortness of Breath Questionnaire; UIP: usual interstitial pneumonia; VAS: visual analogue scale.

Subjects were predominantly White (85%) and 45% female, with a median (Q1–Q3) age of 56.5 (47.1–60.3) years. All satisfied ASyS criteria by Connor et al. [2] and Lega et al. [4] Most had chronic worsening ILD (75%), and 45% had concomitant active myositis. Anti-Jo-1 was the most prevalent anti-synthetase antibody, detected in 55% of subjects. Nonspecific interstitial pneumonia (NSIP) was the most common predominant ILD pattern on HRCT (47%). The median number of previously failed SOC medications was 2.0 (1.0–3.0). Mycophenolate was the most common failed medication (65%) and the most common concomitant medication (75%).

Primary endpoint

There was no significant difference in the absolute change from baseline in %FVC at week 24 between the abatacept arm (−2.05, 95%CI −6.97 to 2.87) and placebo arm (−1.70, 95%CI −5.90 to 2.50), with a difference between treatment arms of −0.35, (95%CI −6.91 to 6.21, P = 0.914). By week 48 in the OLE phase, the absolute change from baseline was −0.20 (95%CI −5.52 to 5.12) for the abatacept arm and −1.36 (95%CI −5.57 to 2.85) for the placebo arm, resulting in a difference between treatment arms of 1.16, (95%CI −5.76 to 8.06, P = 0.736) (Table 2, Fig. 2A, Supplementary Fig. S1, available at Rheumatology online).

Table 2.

Efficacy endpoints from Generalized Linear Mixed Model^a

Endpoints	Abatacept (n = 9)	Placebo (n = 11)	Difference (Abatacept—Placebo)	P-value
	LSM (95% CI)	LSM (95% CI)	LSM (95% CI)
Primary endpoint: absolute change from baseline
%FVC at week 24	−2.05 (−6.97 to 2.87)	−1.70 (−5.90 to 2.50)	−0.35 (−6.91 to 6.21)	0.914
%FVC at week 48	−0.20 (−5.52 to 5.12)	−1.36 (−5.57 to 2.85)	1.16 (−5.76 to 8.06)	0.736
Secondary endpoints: absolute change from baseline
FVC (ml) at week 24	−91 (−314 to 131)	−60 (−249 to 129)	−31 (−327 to 264)	0.830
FVC (ml) at week 48	−11 (−252 to 229)	−51 (−241 to 138)	40 (−272 to 352)	0.794
%DLCO at week 24	2.90 (−5.72 to 11.51)	−2.20 (−8.99 to 4.59)	5.09 (−6.13 to 16.32)	0.361
%DLCO at week 48	7.39 (−1.20 to 15.97)	−0.99 (−8.11 to 6.13)	8.38 (−3.00 to 19.76)	0.143
SOBQ score at week 24	3.20 (−8.86 to 15.26)	−4.59 (−15.54 to 6.36)	7.79 (−8.67 to 24.24)	0.347
SOBQ score at week 48	−9.58 (−22.45 to 3.30)	−4.13 (−15.10 to 6.84)	−5.45 (−22.63 to 11.73)	0.527
SF-36 Physical component (PCS) at week 24	5.45 (−0.09 to 11.00)	3.09 (−1.62 to 7.79)	2.37 (−5.00 to 9.74)	0.521
SF-36 Physical component (PCS) at week 48	4.04 (−1.89 to 9.98)	1.30 (−3.42 to 6.02)	2.75 (−4.95 to 10.44)	0.477
SF-36 Mental component (MCS) at week 24	5.43 (−0.41 to 11.27)	−0.50 (−5.45 to 4.45)	5.93 (−1.82 to 13.68)	0.131
SF-36 Mental component (MCS) at week 48	4.45 (−1.80 to 10.69)	3.78 (−1.19 to 8.74)	0.67 (−7.43 to 8.77)	0.868
HAQ-DI at week 24	−0.03 (−0.35 to 0.29)	0.06 (−0.23 to 0.35)	−0.09 (−0.53 to 0.34)	0.673
HAQ-DI at week 48	−0.17 (−0.53 to 0.18)	0.14 (−0.15 to 0.43)	−0.31 (−0.78 to 0.15)	0.179
Mean 6MWD at week 24	−29 (−94 to 37)	51 (−11 to 114)	−80 (−171 to 11)	0.085
Mean 6MWD at week 48	16 (−59 to 91)	62 (−1 to 124)	−46 (−145 to 53)	0.358
Pulmonary disease activity VAS at week 24	−1.06 (−2.55 to 0.42)	−0.49 (−1.84 to 0.85)	−0.57 (−2.60 to 1.45)	0.574
Pulmonary disease activity VAS at week 48	−2.67 (−4.25 to −1.09)	−0.50 (−1.84 to 0.85)	−2.17 (−4.28 to −0.06)	0.044
QILD-whole lung at week 24	−7.92 (−24.33 to 8.50)	−10.07 (−24.73 to 4.59)	2.15 (−19.84 to 24.14)	0.837
QILD-whole lung at week 48	−11.01 (−24.30 to 2.29)	−0.03 (−11.72 to 11.65)	−10.97 (−28.64 to 6.69)	0.204
QGG-whole lung at week 24	0.48 (−5.90 to 6.86)	−4.64 (−10.29 to 1.02)	5.12 (−3.40 to 13.63)	0.218
QGG-whole lung at week 48	−3.93 (−9.07 to 1.21)	−0.83 (−5.38 to 3.72)	−3.10 (−9.96 to 3.75)	0.349
QLF-whole lung at week 24	−7.49 (−20.06 to 5.08)	−4.27 (−15.51 to 6.98)	−3.22 (−20.08 to 13.64)	0.688
QLF-whole lung at week 48	−6.25 (−16.45 to 3.95)	−0.12 (−9.07 to 8.84)	−6.13 (−19.67 to 7.41)	0.348
QHC-whole lung at week 24	0.27 (−1.09 to 1.62)	−1.29 (−2.49 to −0.08)	1.55 (−0.26 to 3.36)	0.088
QHC-whole lung at week 48	−0.04 (−1.14 to 1.05)	0.90 (−0.07 to 1.86)	−0.94 (−2.40 to 0.52)	0.188
QCON-whole lung at week 24	−0.55 (−1.28 to 0.18)	−0.12 (−0.78 to 0.54)	−0.44 (−1.42 to 0.55)	0.360
QCON-whole lung at week 48	−0.42 (−1.02 to 0.18)	−0.13 (−0.65 to 0.39)	−0.29 (−1.08 to 0.50)	0.443
Glucocorticoids (prednisone or equivalence)
Absolute change from baseline at week 24 (mg/day)	−15.1 (−27.2 to −3.0)	−14.2 (−25.2 to −3.2)	−0.9 (−17.6 to 15.8)	0.909
Absolute change from baseline at week 48 (mg/day)	−17.9 (−30.8 to −5.0)	−14.0 (−25.1 to −3.0)	−3.9 (−21.3 to 13.6)	0.645

Bold font highlights significant results.

^aModels adjusted for age, sex and baseline %FVC. SOBQ score was additionally adjusted for baseline SOBQ. Higher PCS and MCS and lower SOBQ and HAQ-DI scores indicate better outcomes.

6MWD: 6-min walk distance; DLCO: diffusing capacity of the lung for carbon monoxide; FVC: forced vital capacity; HAQ-DI: HAQ-Disability Index; HRCT: high-resolution computed tomography; ILD: interstitial lung disease; IQR: interquartile range; LSM: least squares means; mg: milligram; ml: milliliter; QCON: quantitative score of consolidations; QGG: quantitative score of ground glass; QHC: quantitative score of honeycombing; QLF: quantitative score of fibrotic patterns; SOBQ: University of California San Diego Shortness of Breath Questionnaire; UIP: usual interstitial pneumonia; VAS: visual analogue scale.

Figure 2.

Absolute change from baseline in the primary endpoint and selected secondary endpoints. (A) Absolute change in percent predicted of forced vital capacity (%FVC) from the baseline. (B) Absolute change from baseline in percent predicted diffuse capacity of the lung for carbon monoxide (%DLCO). (C) Absolute change from baseline in the total score on the University of California San Diego Shortness of Breath Questionnaire (SOBQ)

Secondary endpoints

Pulmonary function parameters

There was no significant difference in the absolute change from baseline in FVC (ml) at week 24 between the abatacept arm and placebo arm, with a difference between treatment arms of −31, (95%CI −327 to 264, P = 0.830). By week 48, the difference between treatment arms was 40 (95%CI −272 to 352, P = 0.794).

There was no significant difference in the absolute change from baseline in %DLCO at week 24 between the abatacept arm and placebo arm, with a difference between treatment arms of 5.09, (95%CI −6.13 to 16.32, P = 0.361). By week 48, the difference between treatment arms was 8.38 (95%CI −3.00 to 19.76, P = 0.143) (Fig. 2B).

Patient-reported outcomes

In terms of patient-reported dyspnoea, there was no significant difference in the absolute change from baseline in the SOBQ score at weeks 24 and 48 (Table 2, Fig. 2C) between the treatment arms, with week 24 favouring the placebo arm and week 48 favouring abatacept arm. Similar results were the proportion of patients with SOBQ MCID in dyspnoea without treatment failure, with 75% of patients improved in the abatacept arm as compared with 45% in the placebo arm at week 48 (P = 0.352).

In terms of quality of life, there was no significant difference in the absolute change from baseline in the SF-36 PCS at weeks 24 and 48 between the treatment arms, with weeks 24 and week 48 favouring abatacept. Similar results were observed in SF-36 MCS (Table 2).

In terms of health-related disability, there was no significant difference in the absolute change from baseline in the HAQ-DI at weeks 24 or 48 between the treatment arms, with weeks 24 and 48 favouring abatacept (Table 2).

6-min walk distance

There was no significant difference in the absolute change from baseline in the 6MWD at weeks 24 or 48 between the treatment arms (Table 2).

Pulmonary disease activity

There was no significant difference in the absolute change from baseline in pulmonary disease activity VAS at week 24 between the abatacept arm and placebo arm. However, by week 48, the difference between treatment arms was −2.17 (95%CI −4.28 to −0.06, P = 0.044), statistically significant in favouring the abatacept.

Quantitative image analysis

There was no significant difference in the absolute change from baseline in the QILD score at week 24 and week 48 between the abatacept arm and placebo arm.

Glucocorticoid use

There was no significant difference in the absolute change from baseline in the glucocorticoid dose at week 24 and week 48 between the abatacept arm and placebo arm.

Sensitivity analyses, which included further adjustments for primary and secondary outcome parameters, yielded results consistent with those obtained without additional adjustments (Supplementary Table S1, available at Rheumatology online). Additional data from simple summary statistics were provided in Supplementary Table S2, available at Rheumatology online.

Treatment failure and disease progression

Over the course of 24 and 48 weeks, 11% (1/9) and 13% (1/8) of participants receiving abatacept experienced treatment failure, as compared with 27% (3/11) and 36% (4/11) in the placebo arm, respectively. Similarly, 33% (3/9) and 38% (3/8) of abatacept-treated participants experienced either disease progression or treatment failure, compared with 36% (4/11) and 64% (7/11) in the placebo arm at the same time points. In terms of PFT parameters, 67% (6/9) and 63% (5/8) of participants in the abatacept arm achieved 10% FVC change, compared with 64% (7/11) and 45% (5/11) in the placebo arm at week 24 and 48, respectively. None of these comparisons reached statistical significance (Supplementary Table S3, available at Rheumatology online).

Time to progression-free survival and time to improvement

Due to the limited number of participants experiencing events of interest, we presented the results using survival curves rather than the median time to events (Supplementary Tables S4 and S5, Figs S2 and S3, available at Rheumatology online). There were no significant differences between the treatment arms in weeks 24 or 48.

Pre-post-baseline parameters

There was a statistically significant difference between pre-baseline (worsening) and post-baseline (improving) in the %FVC slopes (trajectory) in the abatacept arm (P = 0.025) but not in the placebo arm (P = 0.318). Similar trends were found in pre-baseline and post-baseline FVC (ml) and %DLCO slopes favouring abatacept, although they did not reach statistical significance (Fig. 3A, Supplementary Table S6, available at Rheumatology online). In abatacept arm, the projected %FVC and FVC (ml) at week 48 demonstrated statistically significant differences between the actual observed values post-baseline as compared with projected values based on pre-baseline trends [%FVC 65.90 vs 44.19 and FVC (ml) 2447 vs 1429, respectively, P< 0.001 for both]. However, in the placebo arm, significant improvement was observed for %FVC to a lesser extent (64.56 vs 55.75, P = 0.039), and not for the FVC (ml) (Fig. 3B, Supplementary Table S7, available at Rheumatology online). Similarly, for longitudinal pre-baseline and post-baseline HRCT QIA scores, there was a statistically significant difference between pre-baseline (worsening) and post-baseline (improving) in the QILD and QLF slopes in the abatacept arm (P = 0.011 and P = 0.006, respectively), but not in the placebo arm (P = 0.140 and P = 0.083, respectively) (Fig. 3C).

Figure 3.

Pre-post-baseline parameters. (A) Longitudinal trajectories in pre-baseline and post-baseline pulmonary function test parameters. For the %FVC, the estimated difference between pre-baseline slope versus post-baseline slopes (slope estimate$\pm$SE) was 0.45$\pm$0.20, P = 0.025 in the abatacept arm and 0.19$\pm$0.19, P = 0.318 in the placebo arm. Similar trends were found in pre-baseline and post-baseline FVC (ml) and %DLCO slopes favouring abatacept, although they did not reach statistical significance. (B) Post-baseline spline modelling expected value at week 48 vs projected value at week 48 based on pre-baseline spline slope for pulmonary function test parameters. In the abatacept arm, the projected %FVC and FVC (ml) at week 48 demonstrated statistically significant differences between the actual observed values post-baseline as compared to projected values based on pre-baseline trends [%FVC 65.90 vs 44.19, P<0.001; FVC (ml) 2447 vs 1429, P<0.001; %DLCO 58.43 vs 43.83, P = 0.005]. However, in the placebo arm, significant improvement was observed for %FVC to a lesser extent (64.56 vs 55.75, P = 0.039), and not for the FVC (ml) and %DLCO. (C) Longitudinal trajectories in pre-baseline and post-baseline quantitative HRCT scores (n = 11). Estimated trajectories of high-resolution chest computed tomography quantitative image analysis (QILD, QLF, QGG) % whole lung scores modelled using piecewise linear spline by treatment arm. The estimated difference between pre-baseline slope versus post-baseline slopes (slope estimate$\pm$SE) was 3.37$\pm$1.21, P = 0.01* for QILD, 2.84$\pm$0.93, P = 0.006* for QLF, 0.31$\pm$0.59, P = 0.60 for QGG in abatacept arm; 1.46$\pm$0.95, P = 0.14 for QILD, 0.86$\pm$0.47, P = 0.08 for QLF, 0.65$\pm$0.44, P = 0.16 for QGG in the placebo arm. FVC: forced vital capacity; ml: milliliters; HRCT: high-resolution computed tomographic scan; QGG: quantitation of ground glass; QILD: quantitation of interstitial lung disease; QLF: quantitation of lung fibrosis; SE: standard error; %DLCO: percent predicted diffusing capacity of the lung for carbon monoxide; %FVC: percent predicted forced vital capacity

Adverse events

Overall, abatacept was well tolerated. During the 24-week RCT phase, one participant in the abatacept arm experienced a serious adverse event (SAE) that led to treatment discontinuation at week 12 due to worsening disease. Another participant in the abatacept arm died from an SAE related to disease progression just after finishing the RCT phase but before receiving the first dose of OLE medication at week 25 (Fig. 1B). Both events were unrelated to treatment.

In the OLE phase, 2 of 19 participants (10.5%) discontinued treatment due to adverse events: one due to worsening disease and one due to increasing cough, both occurring at week 36. Additionally, one participant experienced hypoxia (SAE) at week 33, which was the same participant who had discontinued treatment at week 12 due to a previous SAE.

The most frequent adverse event among abatacept-treated participants was upper respiratory tract infection, which affected one participant (11.1%) during the randomized phase and three participants (15.8%) during the OLE phase (Table 3).

Table 3.

Adverse events

Adverse events	Randomized phase		Open-label phase
	Abatacept (n = 9)	Placebo (n = 11)	Abatacept (n = 19)
Patients with any AE, n (%)	7 (77.8)	5 (45.6)	14/20a (70.0)
Patients with AE leading to discontinuation of the intervention, n (%)	1 (11.1)	0 (0.0)	3/20a (15.0)
Patients with serious adverse events,b  n (%)	1 (11.1)	0 (0.0)	2/20a (10.0)
Number of events	15	19	27
Events that are possibly related to treatment	3	6	12
AE type
Infection
Upper respiratory tract infection	1	2	3
Herpes zoster	0	1	0
Oral thrush	1	0	0
Eye infection	1	0	0
Lip infection	1	0	0
Respiratory tract disorders
Cough	1	2	2
Voice alteration	0	1	1
Dyspnoea	1b	1	1
Hypoxia	0	0	1b
Respiratory failure	1b	0	0
Other	0	0	1
Cardiovascular disorders
Chest pain	0	0	2
Sinus tachycardia	0	2	0
Hypertension	1	0	0
Gastrointestinal disorders
Abdominal pain	1	0	0
Nausea/vomiting	0	2	1
Diarrhea	0	0	3
Colitis	0	0	1
Oral mucositis	0	0	1
Musculoskeletal disorders
Arthralgia	0	1	0
Back pain	1	0	0
Shoulder pain	0	0	1
Extremity pain	0	0	1
Central nervous system disorders
Headache	1	1	0
Anxiety	1	0	1
Insomnia	1	0	0
Depression	1	0	0
Laboratory disorders
Decrease lymphocyte count	1	0	1
Transaminitis	0	1	0
Others
Fever	0	1	0
Fatigue	0	1	0
Anorexia	0	0	1
Rashes	0	0	1
Dry skin	0	0	1
Bruising	0	2	1
Renal calculi	0	0	2

^aOne patient in the abatacept arm died from an SAE related to disease progression just after finishing the randomized placebo-controlled phase but before receiving the first dose of OLE medication at week 25.

^bSerious adverse event: an event that resulted in death, was life-threatening, resulted in hospitalization or prolongation of hospitalization, resulted in persistent or clinically significant disability or incapacity, or was deemed to be serious for any other reason.

AE: adverse event.

Discussion

In this randomized, placebo-controlled pilot trial, abatacept did not achieve a statistically significant difference in the primary endpoint of absolute change from baseline in %FVC at week 24 compared with placebo. However, by week 48 in the OLE phase, the abatacept arm showed a favourable, non-statistically significant trend. This pattern, observed at week 48 but not at week 24, was also evident in the FVC (ml) parameter. It is important to note that the study was not powered to detect significant differences and was designed to evaluate trends and effect sizes and generate hypotheses for future studies.

Several secondary endpoints also showed non-statistically significant results favouring abatacept at 48 weeks. However, these trends were absent in week 24. These parameters included improvements in %DLCO, patient-reported outcomes including dyspnoea scores (SOBQ), SF-36 (PCS and MCS), HAQ-DI, rate of progression or treatment failure and reductions in glucocorticoid dose. It should be acknowledged that patients in the placebo group transitioned to abatacept treatment from weeks 24–48 during the OLE. Nevertheless, these trends raise the hypothesis that abatacept may have a delayed response in ILD.

The potential long-term benefit of abatacept was further supported by a statistically significant difference between pre-baseline and post-baseline slopes for %FVC, and the QILD and QLF scores on HRCT, suggesting a significant improvement in the abatacept arm but not in the placebo arm at week 48. Additionally, a statistically significant difference in the absolute change from baseline of pulmonary disease activity VAS between treatment arms at 48 weeks also supported these findings. Notably, 75% of patients in the abatacept arm reported improved dyspnoea symptoms (SOBQ) compared with 45% in the placebo arm at week 48. Further exploration of functional imaging techniques could help identify subgroups of patients who may benefit from abatacept.

While the abatacept arm showed a numerical improvement in the %DLCO, the FVC outcomes had a numerical worsening at week 24. Week 48 results of FVC and %DLCO outcomes more consistently showed trends in favour of abatacept. Several factors could contribute to this discrepancy in week 24. First, concurrent respiratory muscle weakness associated with ASyS may result in a reduced FVC, with a lesser impact on DLCO [43, 44]. Another complicating factor is obesity, which could result in restrictive ventilatory defect. However, sensitivity analyses adjusting for MMT8 and BMI did not alter the direction of FVC and DLCO. Other studies in ASyS-ILD also found different directions for these two parameters [45, 46], and one study proposed that the improvement rate of DLCO with abatacept appeared to be higher than that of FVC [47]. Lastly, pulmonary hypertension, which was not assessed or followed in this trial, might have contributed to this discrepancy. Thus, to understand these dynamics comprehensively, long-term follow-up of the patients with data on pulmonary arterial pressure is warranted.

The overall trend towards improvement with abatacept aligns with the previous case series of eight ASyS-ILD patients, where improvements in the median %DLCO and symptom alleviation were noted after abatacept therapy [47].

This study had several limitations, the most significant being the small sample size, which reduced statistical power to detect significant meaningful differences in primary or secondary outcomes. Most of the observed differences were neither statistically nor clinically significant and fell within the noise of the testing. Additionally, baseline differences between treatment arms, even after adjusting for age, sex and %FVC, may have influenced results. Notable differences included longer ILD disease duration (39.0 months vs 17.0 months), lesser NSIP pattern on HRCT (29% vs 60%) and higher QILD score (43.7 vs 36.3) in the abatacept arm compared with the placebo arm. Therefore, the findings should be interpreted with caution.

The safety profile of abatacept observed in this study was consistent with its established use in other autoimmune conditions [48, 49]. No major safety signals were identified. All three serious adverse events were not related to treatment but rather progressive disease. Overall, abatacept is generally well-tolerated in this patient population.

In conclusion, this study did not demonstrate statistically significant efficacy for abatacept in ASyS-ILD. However, the observed trends towards improvement with abatacept at week 48 and the favourable safety profile support the rationale for conducting larger RCTs with extended follow-up periods. This study serves as a valuable foundation for future IIM-ILD clinical trials aimed at optimizing treatment strategies for this challenging condition.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The data that support the findings of this study are available upon request from the corresponding author.

Funding

Grant support for Investigator Initiated Study (IIS) by Bristol Myers Squibb (BMS): Abatacept for the Treatment of Myositis-associated Interstitial Lung Disease (Attack My-ILD).

Disclosure statement: R.A. has received grants/research funding from Boehringer Ingelheim, Bristol Myers-Squibb, EMD Serono, Janssen, Pfizer and Priovant. He has served as a consultant for Alexion, ANI Pharmaceuticals, Argenx, Artasome, AstraZeneca, Boehringer-Ingelheim, Bristol Myers-Squibb, Cabaletta Bio, Capella Bioscience, Capstanx, Corbus, CSL Behring, EMD Serono, Galapagos, Horizon Therapeutics, I-Cell, Immunovant, Janssen, Kezar, Kyverna, Lilly, Manta Medicines Corporation, Novartis, Nuvig Therapeutics, Nkarta, Octapharma, Pfizer, Priovant, Teva, Tourmaline Bio and Verismo Therapeutics. D.I.S. has received grants from Boehringer Ingelheim (MINT Trial) and AstraZeneca (D5985C00006 trial). He has received honoraria for lectures from the Ohio State University Rheumatology Winter Symposium and the West Virginia Immunology Summit. S.S.B. is supported by the NIAMS K23 AR081423 grant. D.S. has received Rheumatology Research Foundation Scientist Development Award, Rush to Progress Pilot Award, and Rukel Funding. S.V. has received grants from Navidea Biopharmaceuticals, Inc., ARGENX and Janssen, with payments made to his institution. P.F.D. is an editor for UpToDate. He serves as an advisory member for the FDA. S.K.D. has received grants from Boehringer-Ingelheim, United Therapeutics, and Bristol Myers Squibb. She has served on advisory boards for BMS, CSL Behring, and AbbVie, with payments made to her personally. She has also received honoraria from the France Foundation. She has participated in data safety monitoring and advisory boards for Avalyn and BMS. She serves as a Senior Medical Advisor for the Pulmonary Fibrosis Foundation (payments to her institution) and as a Medical Advisory Board member for The Myositis Association. She has received medical writing support from Boehringer-Ingelheim and Galapagos. T.D. has received grants from NIH, Sanofi, Bayer, and Genentech. G.M.H. is supported by NIH grants R01 HL111024, R01 HL130974 and R01 HL135142. He has received consulting fees from Boehringer Ingelheim and the Gerson Lehrman Group. J.S.L. has received grants from Boehringer Ingelheim and NIH, consulting fees from Blade, Avalyn, Boehringer Ingelheim, United Therapeutics, AstraZeneca, Elima, Gatehouse Bio, Mannkind, Syndax and Eleven P15. She has served on advisory boards for United Therapeutics (TETON trial) and Avalyn Pharma (ATLAS trial) and as a Senior Medical Advisor for the Pulmonary Fibrosis Foundation. She has received research gifts from Pliant Therapeutics. A.F. is a full-time employee of Bristol Myers Squibb and holds stock in the company. J.F. serves as the Steering Committee President for the Trudeau Society.