Characteristics and outcomes of anti-Ku antibody–associated interstitial lung disease: a retrospective study of 60 patients

Abstract

Objective

Anti-Ku antibodies (Abs) are rare and detected in various autoimmune diseases (AIDs), presenting in various phenotypes, potentially affecting different organs including the lungs. This study aimed to describe the characteristics and evolution of interstitial lung disease (ILD) in anti-Ku-positive patients.

Methods

An observational, multicentre, retrospective study was conducted across 10 French University Hospitals between January 2010 and June 2025, including patients with anti-Ku Abs with ILD. Clinical data and all pulmonary function tests and chest CT scans available were reviewed for this study. ILD progression was defined using criteria inspired by the American Thoracic Society/European Respiratory Society/Japenese Respiratory Society/Asociación Latinoamericana de Tórax Clinical Practice Guidelines, applied across the entire follow-up period rather than within the 12-month timeframe, to capture all clinically meaningful progression events. To account for event timing, time to ILD progression was analysed as a time-to-event outcome using Cox proportional hazards models.

Results

Among 154 anti-Ku-positive patients (51 with idiopathic inflammatory myopathy, 46 with systemic lupus erythematosus, 30 with Sjögren’s disease, 27 with systemic sclerosis), 60 (39%) had ILD (68% women, median age 56 years). The predominant ILD pattern was fibrotic non-specific interstitial pneumonia (27%, n=16). ILD was already present at AID diagnosis in 48 patients (80%). Forty-five (75%) patients progressed after a median time of 4 (2–11) years. Male sex (adjusted HR (aHR) 2.7; 95% CI 1.4 to 5.2) was associated with ILD progression. When the 12-month cut-off was strictly applied, only six patients fulfilled the progressive fibrosing ILD definition. Baseline pulmonary fibrosis was present in 34 (57%) patients and was associated with a reduced survival when adjusting for age at ILD diagnosis and cardiac involvement (aHR 6.5, 95% CI 1.5 to 28.2).

Conclusion

ILD occurred in 39% of anti-Ku-positive patients and progressed in 75% of them, underscoring the importance of systematic ILD screening and monitoring in these patients.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• One study including 19 anti-Ku-positive patients described the characteristics of interstitial lung disease (ILD), reporting an overall survival rate of 67% at 10 years and suggesting that ILD impacts disease prognosis; however, this conclusion is limited by the absence of a control group.

• The development or progression of ILD and the occurrence of pulmonary fibrosis have never been assessed.

WHAT THIS STUDY ADDS

• Among a large cohort of 154 anti-Ku-positive patients, we showed that ILD occurred in 39% of patients and that 75% of them met criteria inspired by the ATS/ERS/JRS/ALAT Clinical Practice Guidelines for ILD progression, applied across the entire follow-up period rather than within the 12-month timeframe.

• Male sex was associated with a shorter time to ILD progression.

• Radiological progression most often manifested as the appearance of new fibrotic lesions or as the extension of pre-existing fibrosis.

• Lung fibrosis at ILD onset was associated with a reduced survival.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our results underscore the importance of systematic ILD screening and monitoring in anti-Ku-positive patients and highlight the need to evaluate antifibrotic agents in this setting.

Introduction

Anti-Ku antibodies (Abs) are directed against heterodimeric DNA-binding proteins and detected in several systemic autoimmune diseases (AIDs), most often in systemic sclerosis (SSc), idiopathic inflammatory myopathies (IIMs), Sjögren’s disease (SjD) and systemic lupus erythematosus (SLE).^{1 2} About one-third of anti-Ku-positive patients develop interstitial lung disease (ILD), which usually occurs with muscle involvement³; either ILD or muscle involvement can appear first.² Little is known about the characteristics of ILD associated with anti-Ku Abs, as the only data come from a case series of 19 patients.⁴ This study showed that non-specific interstitial pneumonia (NSIP) pattern was observed in 53% of patients. Pulmonary function tests (PFTs) revealed that a similar proportion of patients displayed a restrictive ventilatory pattern and 89% exhibited diffusion impairment. The overall survival rate was 67% at 10 years, suggesting that ILD impacts the prognosis of the disease, but this conclusion is limited by the absence of a control group.⁴ In addition, the development or the progression of ILD and the occurrence of pulmonary fibrosis have never been assessed and remain to be characterised. This question also deserves renewed attention in the light of the efficacy of antifibrotic therapies in AID-related ILD.⁵ In order to better delineate the spectrum of anti-Ku-positive patients, we conducted this retrospective longitudinal study to better characterise ILD observed in anti-Ku-positive patients with systemic AIDs, as well as its evolution over time.

Methods

Study population

A previously reported observational, multicentric and retrospective study was conducted in 10 University Hospitals located in Île-de-France region (Assistance Publique-Hôpitaux de Paris, France) between January 2010 and June 2025.² Patients were selected for the study through an exhaustive screening of the database of four immunology departments (Le Kremlin-Bicêtre, La Pitié-Salpêtrière, Bichat and Cochin Hospitals, Paris, France), which performed the anti-Ku assay for all the 10 included centres. Inclusion criteria were (i) the positivity of anti-Ku Abs and (ii) a suspected systemic autoimmune or inflammatory disease. For the present study, we only included patients with a diagnosis of ILD at diagnosis or during the follow-up. Anti-Ku-positive patients without ILD from the same study were used as controls.

Data collection

Data were collected from medical records both at diagnosis and regularly updated during follow-up using a standardised and anonymous case report form. Organ involvement was identified using clinical symptoms, laboratory exams, imaging and/or electrophysiological explorations, both at diagnosis and the end of the follow-up. Details of the variables for each organ were previously published² and are available in online supplemental appendix 1. Disease diagnoses were retrospectively reviewed and eventually reclassified in accordance with the current classification criteria: SLE,⁶ SjD,⁷ SSc,⁸ IIM,⁹ rheumatoid arthritis,¹⁰ mixed connective tissue disease.¹¹ Antisynthetase syndrome (ASyS) diagnosis was retained in the presence of ASyS-associated Abs (ie, anti-Jo1, anti-PL7, anti-PL12, anti-EJ, anti-OJ) and compatible clinical features.¹² Patients were classified as having undifferentiated connective tissue disease (UCTD) when symptoms did not match any other available criteria, or as having interstitial pulmonary fibrosis with autoimmune features when ILD was present. Each treatment line was recorded, with non-response considered an indication for treatment change.

Pulmonary assessment

Dyspnoea was assessed by the physician using modified Medical Research Council or New York Heart Association scales. Data from PFTs were recorded from diagnosis to the last visit, including forced vital capacity (FVC, % predicted and mL) and diffusing capacity of the lung for carbon monoxide (DLCO, % predicted, corrected for haemoglobin concentration (DLCOc)). Chest CT scans were analysed in a double-blind manner by two expert radiologists (M-PD and P-YB), disagreements were resolved by consensus and inter-reader agreement was not formally assessed. For all patients, at least the first and the last CT scans were reviewed. For each chest CT, the pattern (ie, usual interstitial pneumonia (UIP) definite, UIP probable, NSIP, NSIP/organising pneumonia (OP), minimal ILD, other/unclassifiable) was determined according to the 2018 ATS/ERS/JRS/ALAT guidelines,¹³ and ILD extent was visually determined.¹⁴ Based on this double-blind review, ILDs were then divided in two groups according to the presence or absence of predominant pulmonary fibrosis.¹⁵ Fibrosing ILD included patterns classified as definite UIP, probable UIP, indeterminate for UIP and fibrotic NSIP, whereas cellular NSIP, NSIP/OP and minimal ILD were considered non-fibrosing ILD. Baseline fibrosis was defined by the presence of a fibrosing ILD at the time of ILD diagnosis. In case of worsening of the ILD over time or if the patient had presented worsening of symptoms or of PFTs throughout the follow-up,¹⁶ all CT scans were reviewed.

Detection of antinuclear and anti-Ku, anti-extractable nuclear antigen antibodies and other autoantibodies

Details regarding the detection antinuclear Abs, anti-Ku Abs and of other autoantibodies were already described.² Four hospital’s immunology department performed the detection of autoantibodies for the whole cohort following the same method. Briefly, indirect immunofluorescence on Hep2 cells was used to detect antinuclear Abs.¹⁷ All patients were positive for ANA detection with titres ≥1/160. Screening for anti-Ku Abs was performed in two situations: (i) when the immunofluorescence pattern was suggestive of anti-Ku Abs, corresponding to a fine dense granular pattern with nucleolar and perimitotic staining or (ii) at the request of clinicians in patients who were ANA-positive and had clinical manifestations of SSc or IIM. Anti-Ku specificity was soaked using dot-blot immunoassays from two different suppliers (Polymyositis/Scleroderma 8, Dtek, Mons, Belgium and Euroline Systemic Sclerosis Profile, Euroimmun, Lübeck, Germany). Screening of other autoantibodies is associated with the search for anti-Ku positivity.² Autoantibody positivity was established during the whole follow-up.

Outcome definition

The primary outcome was the time to progression of ILD, defined as cumulative ILD progression over time using modified criteria inspired by the ATS/ERS/JRS/ALAT Clinical Practice Guidelines, without the 12-month timeframe,¹⁶ and including at least two of the following three criteria: (i) worsening respiratory symptoms, (ii) physiological evidence of disease progression (either absolute decline in FVC ≥5% predicted or absolute decline in DLCOc ≥10% predicted) and (iii) radiological evidence of disease progression. Radiological signs of progression included one or more of the following items: increased extent or severity of traction bronchiectasis and bronchiolectasis, new ground-glass opacity with traction bronchiectasis, new fine reticulation, increased extent or increased coarseness of reticular abnormality, new or increased honeycombing, increased lobar volume loss. In this retrospective cohort, follow-up intervals and timing of assessments were not uniform; therefore, applying the strict 12-month timeframe was not feasible. To avoid excluding clinically meaningful but later progression events, we therefore applied these criteria throughout the entire observation period to identify progression events. In a sensitivity analysis, baseline characteristics of anti-Ku-positive patients were compared according to ILD progression and its timing (early ≤12 months vs late >12 months). Another sensitivity analysis was conducted comparing baseline characteristics of patients according to the fibrotic evolution of the ILD (no or non-fibrotic progression vs fibrotic progression).

Statistical analysis

Categorical variables are expressed as numbers (percentages) and continuous variables as medians (IQRs). Fisher’s exact test and Pearson’s χ² test were used for qualitative data, and Mann-Whitney U test was used to compare continuous variables between groups. For the overall cohort, baseline characteristics of anti-Ku-positive patients were compared according to the presence of ILD. Overall survival was analysed using log-rank tests and multivariable Cox proportional hazard regression model adjusted for age at AID diagnosis and for cardiac involvement, which was associated with mortality (see below) and represents a characteristic feature of anti-Ku-positive patients.^{2 18} Patients contributed person-time from the date of AID’s diagnosis until the date of death or loss to follow-up, whichever occurred first. Then, among patients with ILD, factors associated with time to ILD progression were evaluated using Cox proportional hazard regression models, and results were expressed as HRs with 95% CIs. Time to ILD progression was modelled as a time-to-event outcome, measured from ILD diagnosis (time 0) to the first documented progression or censored at death or last follow-up, whichever occurred first. Clinically relevant variables identified through univariable analyses were considered for the multivariate Cox model. Kaplan-Meier curves were generated. In a sensitivity analysis, the effect of treatment on ILD was evaluated through the necessity to intensify the treatment or not according to the physician in charge of the patient. These decisions were discussed in multidisciplinary meetings, based on clinical manifestations, PFTs and radiographic changes.

All analyses were performed with R V.4.4.0 (R Foundation for statistical computing, Vienna, Austria) and two-tailed p≤0.05 was considered statistically significant.

Results

Characteristics of anti-Ku-positive patients with AID according to ILD status

One hundred and fifty-four patients with AIDs (51 with IIM, 46 with SLE, 30 with SjD, 27 with SSc) and anti-Ku Abs were studied. Among them, 128 (83%) underwent a chest CT scan. Sixty (39%) patients with ILD were identified among the whole cohort, 41 (68%) of whom were female. Baseline patients characteristics are reported in table 1. Ethnicity was comparable in patients with and without ILD, as well as smoking status. The median (IQR) age at AID diagnosis was significantly older in the ILD group as compared with patients without ILD (median (IQR): 56 (44–66) years vs 40 (24–56) years; p<0.001). Of note, there was no difference between patients without ILD who underwent a chest CT scan and those who did not, except the proportion of death, which was significantly higher in patients who had a chest CT scan (online supplemental table S1). Patients with ILD more frequently exhibited Raynaud’s phenomenon, gastrointestinal involvement and dysphagia but displayed fewer rheumatological manifestations compared with patients without ILD (p<0.05) (table 1). In addition, diagnoses of SSc (27% vs 12%, p=0.017) and UCTD (27% vs 11%, p=0.010) were more common among patients with ILD whereas SLE was less frequent (10% vs 43%, p<0.001). In a sensitivity analysis, we excluded patients from the ILD-negative group who did not have chest CT scan (online supplemental table S2). As compared with the previous analysis (table 1), we still observed that UCTD was more common (27% vs 8.8%, p<0.01) among patients with ILD whereas SLE was less frequent (27% vs 6%, p<0.001) (online supplemental table S2). The observed difference of cardiac involvement between the two groups relied on the high proportion of pericarditis in the group without ILD (n=13/19 patients).

Table 1. Main characteristics of anti-Ku-positive patients according to ILD.

Characteristic	N	Overall N=154*	No ILD N=94*	ILD N=60*	P value
Demographics
Sex, female, n (%)	154	118 (77)	77 (82)	41 (68)	0.052†
Black ethnicity, n (%)	154	71 (46)	39 (41)	32 (53)	0.2†
Smoking status, n (%)	151				0.3‡
People who have never smoked		114 (75)	70 (77)	44 (73)
People who previously smoked		25 (17)	12 (13)	13 (22)
People who smoke		12 (8)	9 (10)	3 (5)
Age at AID diagnosis, median (IQR), years	154	48 (29–62)	40 (24–56)	56 (44–66)	<0.001§
Length of FU, median (IQR), years	154	7 (4–14)	7 (3–14)	6 (4–12)	0.8§
Death, n (%)	154	27 (18)	11 (12)	16 (27)	0.017†
Organ involvements (baseline), n (%)
Muscle involvement	154	62 (40)	35 (37)	27 (45)	0.3†
Dysphagia	150	16 (11)	6 (7)	10 (17)	0.045†
Serositis	152	7 (5)	6 (7)	1 (2)	0.2‡
Pulmonary hypertension	83	6 (7.2)	2 (5)	4 (10)	0.7‡
Renal involvement	154	9 (6)	6 (6)	3 (5)	>0.9‡
Peripheral and central neurological involvement	154	20 (13)	16 (17)	4 (7)	0.062†
Cardiac involvement	153	29 (19)	21 (23)	8 (13)	0.2†
Skin/Mucosal involvement	153	109 (71)	67 (72)	42 (70)	0.8†
Raynaud’s phenomenon	150	64 (43)	32 (36)	32 (53)	0.031†
Gastrointestinal involvement	154	56 (36)	24 (26)	32 (53)	<0.001†
Rheumatological involvement	153	108 (71)	72 (77)	36 (60)	0.021†
Laboratory findings (ever), n (%)
Elevated gammaglobulins	143	88 (62)	50 (57)	38 (68)	0.20†
Hypocomplementaemia	115	49 (43)	38 (50)	11 (28)	0.025†
Autoanti-SSA60 antibodies	152	50 (33)	36 (39)	14 (23)	0.043†
Autoanti-SSA52 antibodies	152	22 (14)	12 (13)	10 (17)	0.5†
Autoanti-dsDNA antibodies	152	45 (30)	38 (41)	7 (12)	<0.001†
Autoanti-RNP antibodies	151	22 (15)	15 (16)	7 (12)	0.4†
Antismooth muscle antibodies	149	16 (11)	5 (6)	11 (18)	0.014†
Autoanti-Scl70 antibodies	151	9 (6)	3 (3)	6 (10)	0.2‡
Autoanticentromere antibodies	151	8 (5)	5 (6)	3 (5)	>0.9‡
MSA	149	6 (4)	2 (2)	4 (7)	0.2‡
Rheumatoid factor	134	30 (22)	12 (15)	18 (33)	0.017†
Isolated anti-Ku positivity	154	42 (27)	29 (31)	13 (22)	0.2†
AID diagnosed, n (%)
IIM	154	51 (33)	27 (29)	24 (40)	0.15†
SLE	154	46 (30)	40 (43)	6 (10)	<0.001†
SjD	154	30 (19)	15 (16)	15 (25)	0.2†
SSc	154	27 (18)	11 (12)	16 (27)	0.017†
RA	154	12 (8)	6 (6)	6 (10)	0.5‡
MCTD	154	1 (1)	1 (1)	0	>0.9‡
UCTD	154	26 (17)	10 (11)	16 (27)	0.010†
Number of diagnoses	154				0.9‡
1		116 (75)	71 (76)	45 (75)
2		32 (21)	20 (21)	12 (20)
3		6 (3.9)	3 (3.2)	3 (5.0)

^*n/N (%); median (IQR).

^†Pearson’s χ2 test.

^‡Fisher’s exact test.

^§Wilcoxon rank-sum test.

AID, autoimmune disease; dsDNA, double-stranded DNA; FU, follow-up; IIM, inflammatory myopathies; ILD, interstitial lung disease; MCTD, mixed connective tissue disease; MSA, myositis-specific or myositis-associated antibodies; RA, rheumatoid arthritis; SjD, Sjögren’s disease; SLE, systemic lupus erythematosus; SSc, systemic sclerosis; UCTD, undifferentiated connective tissue disease.

We then analysed overall survival according to ILD status at diagnosis. Patients with inaugural ILD had a worse overall survival compared with those without (log-rank test, p=0.0035, figure 1) (univariable HR 3.1, 95% CI 1.4 to 6.8). After adjustment for age and the presence of heart involvement at AID diagnosis, this association between inaugural ILD and poorer survival was still statistically significant (adjusted HR (aHR) 3.1, 95% CI 1.3 to 7.3). Age (aHR 1.0, 95% CI 1.0 to 1.1) and cardiac involvement at AID diagnosis (aHR 3.9, 95% CI 1.7 to 9.3) were also associated with worse survival in this multivariable analysis.

Figure 1. Kaplan-Meier plot estimating the survival probability according to ILD status at baseline. T0 corresponded to the age at the diagnosis of the AID. P value was calculated by log-rank test. AID, autoimmune disease; ILD, interstitial lung disease.

Pulmonary features at ILD-onset and at last visit

Among the cohort of 60 anti-Ku-positive patients with ILD, ILD appeared at a median (IQR) age of 58 (44–67) years and ILD was already present at the time of AID diagnosis in 48/60 (80%). Demographic characteristics of patients, PFTs and CT scan patterns at baseline are represented according to ILD progression in table 2. At ILD baseline, the most common chest CT scan pattern was fibrotic NSIP in 16/60 (27%) patients, followed by indeterminate for UIP and OP/NSIP-OP in 15/60 patients (25%). Thirty-four (57%) patients already displayed a pattern with a predominance of fibrosis. Bronchoalveolar lavage was performed in 22 individuals. The median (IQR) total cell count was 163 (80–376) cells/mm³. The median proportions of macrophage, neutrophils, eosinophils and lymphocytes were 71% (49–88), 13% (5–31), 3% (0–5) and 6% (3–16), respectively. No patients were biopsied. During a median (IQR) follow-up of 6 (4–10) years, 45/60 (75%) patients met ATS/ERS/JRS/ALAT Clinical Practice Guideline’s criteria for ILD progression and were considered as progressors.¹⁶ According to these criteria, 36 (60%) out of 60 patients experienced a single episode of progression and nine (15%) had two or more (n=2 for seven patients, n=3 for two patients). Interestingly, 29 of the 34 patients who already displayed a predominantly fibrotic pattern at diagnosis experienced further disease progression. By the last visit, progressors displayed significantly lower FVC and DLCOc values, and fibrotic changes were present in 42 (93%) of them vs four (27%) non-progressors (p<0.001, Fisher’s exact test). Among the 37/45 progressors who had both serial PFTs and CT scan available, PFTs decline came first in 20 (54%) of those 37 patients, occurred concomitantly with radiological change in 12 (32%) patients and followed CT deterioration in five (14%) patients (table 2). Radiological progression of ILD-associated anti-Ku Abs most often manifested as the appearance of new fibrotic lesions (figure 2A,B) or as the extension of pre-existing fibrosis (figure 2C,D). The most common patterns at last visit were fibrotic NSIP in progressors (n=23, 51%) and minimal ILD in non-progressors (n=8, 53%) (table 2).

Table 2. Main characteristics of anti-Ku patients with ILD according to the occurrence of progression during follow-up.

Characteristic	N	No progression N=15*	Progression N=45*
Demographics
Sex, female, n (%)	60	11 (73)	30 (67)
Black ethnicity, n (%)	60	5 (33)	27 (60)
Tobacco (ever), n (%)	60	6 (40)	10 (22)
Age at AID diagnosis, median (IQR), years	60	61 (50–75)	53 (43–64)
Age at ILD diagnosis, median (IQR), years	60	61 (51–78)	56 (43–65)
Length of FU, median (IQR), years	60	4 (1–8)	7 (5–13)
Death, n (%)	60	2 (13)	14 (31)
AID diagnosed, n (%)
IIM	60	10 (67)	14 (31)
SLE	60	1 (7)	4 (9)
SjD	60	4 (27)	12 (27)
SSc	60	3 (20)	13 (29)
RA	60	1 (7)	4 (9)
UCTD	60	2 (13)	12 (27)
Pulmonary features at baseline
Initial lung involvement, n (%)	60	12 (80)	36 (80)
FVC, median (IQR), mL	47	2400 (1940–2840)	2200 (1870–2750)
FVC, median (IQR), %	55	88 (62–111)	71 (61–87)
DLCOc, median (IQR), %	48	48 (43–65)	45 (36–56)
Chest CT scan, n (%)	60
Patterns
UIP		0	3 (7)
Probable UIP		0	0
Indeterminate for UIP		3 (20)	12 (27)
OP or NSIP/OP		6 (40)	9 (20)
Cellular NSIP		1 (7)	3 (7)
Fibrotic NSIP		2 (13)	14 (31)
Minimal ILD		3 (20)	4 (9)
Extent	58	15 (8–20)	20 (12–25)
Pattern according to fibrosis, n (%)
Fibrotic pattern		5 (33)	29 (64)
Minimal or no fibrosis		10 (67)	16 (45)
Bronchoalveolar lavage	22
Cell count, median (IQR), /mm3	18	250 (126–390)	140 (50–376)
Macrophages, median (IQR), %	21	51 (10–76)	74 (51–90)
Neutrophils, median (IQR), %	21	14 (11–35)	9 (4–31)
Eosinophils, median (IQR), %	21	9 (5–21)	2 (0–5)
Lymphocytes, median (IQR), %	22	32 (11–53)	4 (2–7)
Pulmonary features at last visit
FVC, median (IQR), mL	53	2440 (2000–3490)	1870 (1450–2610)
FVC, median (IQR), %	58	98 (78–105)	65 (55–78)
DLCOc, median (IQR), %	57	61 (48–70)	38 (20–55)
Chest CT scan	60
Patterns, n (%)
UIP		0	7 (16)
Probable UIP		0	0
Indeterminate for UIP		1 (7)	12 (27)
OP or NSIP/OP		1 (7)	2 (4)
Cellular NSIP		1 (7)	0
Fibrotic NSIP		3 (20)	23 (51)
Minimal ILD		8 (53)	1 (2)
Normalisation		1 (7)	0
Extent, median (IQR), %		4 (3–12)	25 (15–35)
Pattern according to fibrosis, n (%)
Fibrotic pattern		4 (27)	42 (93)
Minimal or no fibrosis		11 (73)	3 (7)

FVC and DLCOc are reported as percentages of predicted values.

^*n/N (%); median (IQR).

AID, systemic autoimmune disease; DLCOc, diffusion capacity of the lung for carbon monoxide, corrected for haemoglobin; FU, follow-up; FVC, forced vital capacity; IIM, inflammatory myopathies; ILD, interstitial lung disease; MSA, myositis-specific or myositis-associated antibodies; NSIP, non-specific interstitial pneumonia; OP, organising pneumonia; RA, rheumatoid arthritis; SjD, Sjögren’s disease; SLE, systemic lupus erythematosus; SSc, systemic sclerosis; UCTD, undifferentiated connective tissue disease; UIP, usual interstitial pneumonia.

Figure 2. Patterns of interstitial lung disease (ILD)-associated anti-Ku antibodies at CT. Representative pulmonary CT scans of ILD progression over time, including the development of fibrosis ((A) initial CT scan; (B) after 5 years of follow-up) and worsening of fibrosis ((C) initial CT scan; (D) after 4 years of follow-up).

Primary outcome

The primary end point, that is, cumulative ILD progression, was observed in 45/60 (75%) patients during follow-up. The median (IQR) time between the onset of ILD and the first progression episode was 4 (2–11) years (figure 3A). In univariable analyses, older age at ILD diagnosis and male sex were significantly associated with ILD progression (table 3). In the adjusted Cox model, male sex (aHR 2.7; 95% CI 1.4 to 5.2) remained significantly associated with ILD progression, unlike age at ILD diagnosis (aHR 1.0; 95% CI 1.0 to 1.0) (table 3). No other significant associations were pinpointed between progression and smoking history, baseline PFTs, initial chest CT scan pattern, other organ involvement or the specific underlying diagnosis. As age was one of the most predominant factors, progression-free survival was stratified by age at ILD diagnosis: patients diagnosed after the age of 60 years progressed significantly faster than those diagnosed before the age of 60 years (log-rank test, p=0.0053) (figure 3B). Kaplan-Meier curves revealed earlier progression in men (log-rank test, p=0.002) (figure 3C). When restricted to the first 12 months after ILD onset, only six patients fulfilled the progressive pulmonary fibrosis definition; descriptive characteristics of these early events are provided in online supplemental table S3. In a sensitivity analysis, patients were stratified according to fibrotic evolution on serial CT scans. Forty-two (70%) patients developed or worsened lung fibrosis (‘fibrotic progression’), whereas 18 (30 %) had no or non-fibrotic progression, including both stable and non-fibrotic progressive cases (online supplemental table S4). Patients with no or non-fibrotic progression had a higher prevalence of IIM compared with those who experienced fibrotic progressors (p=0.029).

Figure 3. Estimates of the risk of ILD progression in overall population and subgroups. Kaplan-Meier curves estimating the risk of ILD progression in the whole cohort (A), stratified by (B) age at ILD diagnosis and (C) sex. T0 corresponded to the age at the diagnosis of the ILD. P values were calculated by log-rank test. ILD, interstitial lung disease.

Table 3. Estimates of factors associated with the risk of ILD progression.

Variables		N (%) events or mean (SD)	Univariable analysis HR (95% CI)	Multivariable analysis adjusted HR (95% CI)
Age at ILD diagnosis, years		56 (17)	1.0 (1.0 to 1.0)	1.0 (1.0 to 1.0)
Sex	Female	41 (68)	Ref	Ref
	Male	19 (32)	2.8 (1.4 to 5.4)	2.7 (1.4 to 5.2)
Smoking status	Never	44 (73)	Ref	–
	Ever	16 (27)	1.2 (0.6 to 2.4)	–
Initial PFTs	FVC, %	77 (25)	1.0 (1.0 to 1.0)	–
	DLCOc, %	49 (18)	1.0 (1.0 to 1.0)	–
Initial chest CT scan pattern	Minimal or no fibrosis	26 (43)	Ref	–
	Fibrosis	34 (57)	1.2 (0.6 to 2.2)	–
Baseline organ involvements	Neurological	4 (7)	2.4 (0.8 to 7.0)	–
	PH	17 (28)	0.8 (0.4 to 1.5)	–
	Muscle	27 (45)	0.7 (0.4 to 1.2)	–
	Renal	2 (3)	0.7 (0.1 to 5.4)	–
	Cardiac	8 (13)	1.9 (0.7 to 5.1)	–
	Skin/Mucocutaneous	43 (72)	0.5 (0.3 to 1.1)	–
	Gastrointestinal	32 (53)	0.8 (0.4 to 1.4)	–
	Rheumatic	36 (60)	1.0 (0.6 to 1.9)	–
Diagnosis	UCTD	14 (23)	1.6 (0.8 to 3.2)	–
	IIM	24 (40)	0.7 (0.4 to 1.3)	–
	SSc	16 (27)	0.9 (0.5 to 1.8)	–
	SjD	16 (27)	0.9 (0.5 to 1.7)	–
	RA	5 (8)	1.3 (0.5 to 3.7)	–
	SLE	5 (8)	0.7 (0.3 to 2.1)	–

HRs (95% CIs) for the risk of ILD progression over time in the whole cohort (n=60) were estimated using Cox models.

Multivariable mode included age at ILD diagnosis, sex, ethnicity, UCTD diagnosis.

DLCOc, diffusion capacity of the lung for carbon monoxide, corrected for haemoglobin; FVC, forced vital capacity; IIM, inflammatory myopathies; ILD, interstitial lung disease; PFT, pulmonary function test; PH, pulmonary hypertension; RA, rheumatoid arthritis; Ref, reference; SjD, Sjögren’s disease; SLE, systemic lupus erythematosus; SSc, systemic sclerosis; UCTD, undifferentiated connective tissue disease.

Survival

Over a median (IQR) follow-up of 6 (4–10) years, 16 patients with ILD died. Causes of death were infections (n=7), respiratory insufficiency (n=4), cardiogenic shock (n=2), cancer (n=1) and one was not retrieved. When overall survival was analysed according to ILD course, adjustment for age at ILD diagnosis and cardiac involvement showed no significant difference between patients who progressed and those who remained stable (aHR 1.6, 95% CI 0.3 to 7.5). In contrast, the presence of lung fibrosis at ILD diagnosis was independently associated with poorer survival after adjustment for age at ILD diagnosis and cardiac involvement (aHR 6.5, 95% CI 1.5 to 28.2).

Treatments

In a sensitivity analysis, we wanted to describe the ILD evolution under treatment. Therapeutic management of ILD in anti-Ku-positive patients mainly relied on corticosteroids (CS) (n=46, 77%) and immunosuppressives (IS) (n=54, 90%), including biologics (n=26, 43%: rituximab (RTX) (n=22), tumour necrosis factor inhibitors (n=3) and anti-interleukin-6 receptor (n=1)) (online supplemental table S5). Of note, 11 patients received intravenous immunoglobulins but mostly for muscle involvement. Only six (10%) patients remained untreated, all of them experienced ILD progression and two eventually died from respiratory failure. During the whole follow-up, each patient was exposed to a median (IQR) of 3 (2–4) distinct immunomodulating or IS treatments. Thirty-four (57%) patients received ≥20 mg/day of oral CS, which were preceded by intravenous pulse therapy in 15 patients. Mycophenolate mofetil (MMF) was given to 28 (47%) patients for a median time of 43 (22–75) months; in seven of these cases, MMF followed an induction course of intravenous cyclophosphamide (CYC) pulse therapy. The ILD of five of the MMF-treated patients remained uncontrolled despite therapy. RTX was administered to 22 (37%) patients for a median of 3 (1–7) treatment cycles. ILD control was not achieved in nine (41%) of these patients. CYC was used in 16 (27%) patients with a median number of six infusions, and the ILD of four (25%) patients remained uncontrolled. MTX was given to 14 (23%) patients for a median time of 60 (11–112) months, with persistent ILD activity in half of them. Azathioprine (AZA) was prescribed to nine (15%) patients, following intravenous CYC pulse therapy in four patients. Five patients received MMF together with RTX; this combination did not control the disease in one patient. TCZ was added to MMF in one refractory patient, ILD was still progressing at last visit. A single patient received both AZA and MTX for <3 months because of hepatotoxicity. RTX and MTX were associated in three patients, a combination not efficient in one patient. Drug-related adverse events for each therapy are detailed in online supplemental table S5.

Discussion

This study describes, to our knowledge, the largest cohort of anti-Ku-positive patients with ILD. Among the whole population of individuals who carried anti-Ku Abs, nearly half of them eventually developed ILD, usually at the time their AID was diagnosed. The presence of ILD reduced overall survival. During a median follow-up of 6 years, three-quarters of all anti-Ku-positive patients with ILD experienced at least one episode of progression. Male sex was associated with higher risk of worsening of lung disease. Serial chest CT scans revealed new or worsening fibrotic changes in almost 90% of cases; the presence of pulmonary fibrosis at baseline on CT was linked to a significant higher mortality risk. The overwhelming majority of patients were treated with CS and IS agents.

Our study reports for the first time that ILD affects approximately 40% of patients who carry anti-Ku Abs. Similar to anti-U1RNP Abs, anti-Ku are associated with systemic AIDs and overlap syndromes, but the frequency of ILD is rather higher among anti-Ku patients when comparing with anti-U1RNP patients (ie, 26%).^{3 19} Knowing the frequency of ILD in anti-Ku-positive patients, a systematic search for ILD should be performed in every anti-Ku-positive patient both at disease onset and in the first years of follow-up, as suggested by previous studies.⁴ Whereas a previous series of 19 anti-Ku patients reported a mean survival (±SD) of 6.7 (±5.5) years,⁴ our larger cohort confirms the association between ILD and decreased survival in univariate analysis.²⁰ Deaths were either directly (ie, respiratory failure) or indirectly (ie, infections) related to lung involvement or its treatment. In addition, the age at ILD diagnosis and cardiac involvement were also associated with a reduced survival, which highlights the need for systematic cardiac involvement screening in anti-Ku-positive patients.^{2 18}

Next, we wanted to assess cumulative ILD progression over time. Remarkably, three-quarters of the patients experienced at least one episode of ILD progression during follow-up according to modified criteria inspired by the ATS/ERS/JRS/ALAT Clinical Practice Guidelines, without the 12-month timeframe.¹⁶ Consistent with published data,²¹ progression was more frequent in men and in older individuals. In our study, the median interval between ILD onset and progression was several years, indicating a slower course than that observed in some forms of myositis.²² In most cases, ILD evolution was detected by changes in PFTs. These findings underscore the value of regular PFT monitoring.²³

A previous case series described NSIP as the dominant CT pattern at ILD onset in anti-Ku-positive patients and noted that 68% later exhibited radiological progression, although the ultimate pattern was not specified.⁴ We found similar results in our cohort, but we divided NSIP into cellular and fibrotic subtypes and observed that the fibrotic pattern was much more frequent. We also described chest CT scans according to the presence or absence of predominant fibrosis. The proportions of fibrotic and non-fibrotic ILD were rather similar at baseline, whereas by the last visit, nearly all patients exhibited a fibrotic pattern. Almost all patients of our cohort received CS together with IS, in line with current recommendations for the initial treatment of ILD in AIDs other than SSc.²⁴ Those guidelines favour short-term CS plus MMF, AZA or RTX as first-line therapy.²⁴ Recent data suggest that RTX should be considered as a therapeutic alternative of CYC in patients with AID-ILD, as it is associated with fewer side effects.²⁵ In addition, the combination of RTX and MMF appeared more effective than MMF alone in NSIP pattern.¹⁴ In the present study, five patients were treated with this dual regimen and its effectiveness seems consistent with these results.

The present study predates the availability of antifibrotic agents and only three patients received these drugs. Here, pulmonary fibrosis was an independent predictor of death. In almost 90% of the patients with ILD progression, the radiological evolution was characterised by new-onset or worsening fibrotic lesions. These observations indicate that current treatments can at best slow—rather than stop—fibrosis progression. Antifibrotic treatments, such as nintedanib, pirfenidone and nerandomilast, have been shown to slow lung function decline across several fibrosing ILDs,^26–29 even when these agents were added to IS regimens,^{28 30 31} usually required in anti-Ku-positive patients. Their potential role in anti-Ku-associated ILD should now be systematically evaluated to determine the most suitable molecule, the timing of initiation, initiation criteria and the optimal treatment duration.

Our study has some limitations. First, it is a retrospective study with inherent bias that limits our conclusions. Anti-Ku Abs are very rare,³² making the formation of a prospective cohort difficult and the rarity of the anti-Ku-positive patients hampered the power of statistical analysis. It is undeniable that a standardised monitoring approach would have improved the evaluation of ILD at diagnosis, its progression and treatment effectiveness. Second, since not all anti-Ku-positive patients (ie, 83%) underwent chest CT imaging to screen for ILD, some individuals may have been incorrectly classified leading to underestimate ILD prevalence in this population. However, comparisons of ILD-negative patients with and without chest CT scan showed no major differences in baseline characteristics suggesting limited bias in ILD versus non-ILD comparisons. The proportion of patients wth SLE was higher in patients who did not have a chest CT scan, which is not so surprising as ILD is very rare in SLE without overlapping syndrome.³³ In addition, patients who did not undergo chest CT imaging were followed for 5 (2–15) years without developing any respiratory symptoms, suggesting that the presence of ILD was unlikely. Third, criteria inspired by the ATS/ERS/JRS/ALAT Clinical Practice Guidelines were applied throughout the entire follow-up period rather than within a strict 12-month timeframe,¹⁶ because regular annual evaluations were not available in this retrospective cohort and a fixed 12-month window could not be systematically applied. We therefore studied cumulative ILD progression over time using these modified criteria without a predefined time restriction, but by modelling time to progression to capture clinically meaningful progression that might occur later in the disease course. Although this approach precludes direct comparisons with studies that strictly adhered to the original 12-month definition, it allows us to integrate the notion of time in the outcome definition. Our approach increased the likelihood of classifying patients as progressors compared with other studies. Fourth, our study predates the new classification of interstitial pneumonias.^34–36 Fifth, treatment exposure represented a major source of heterogeneity in this long-term, real-life cohort. Because detailed longitudinal data on treatment timing and duration were unavailable, treatment variables could not be reliably included as time-dependent covariates. Moreover, extra lung manifestations may have influenced therapeutic decisions and disease progression, adding potential confounding. To limit indication bias, black ethnicity—possibly associated with differences in treatment allocation or healthcare access—was excluded from multivariable models. Finally, age was a major confounder of both overall survival and disease progression in our cohort. To account for this, all Cox models were adjusted for age at ILD diagnosis to minimise its confounding effect, as well as with cardiac involvement for the overall survival. Strengths of our study included the unbiased recruitment permitted by the multicentric design with departments from various specialties. In addition, anti-Ku Abs testing was done using two different methods to avoid false-positive results.³⁷ Blinded dual readings of chest CT scans by two expert radiologists enabled a consistent assessment of ILD at baseline and throughout follow-up.

In conclusion, ILD is frequent and usually inaugural in anti-Ku-positive patients with AID. The overwhelming majority of patients undergo at least one episode of progression and ultimately develops pulmonary fibrosis over time despite IS treatments. These findings underscore the need to evaluate antifibrotic agents in this setting.

Data availability statement

Data are available on reasonable request.