Incidence and spectrum of primary respiratory disease throughout the course of rheumatoid arthritis: implications of structured repeated evaluation for detection and risk-factor analysis

Abstract

Objective

To determine the incidence and spectrum of lung involvement in early RA through a structured respiratory assessment and to identify its clinical predictors.

Methods

A retrospective study was conducted in a cohort of 204 early RA patients screened for lung involvement at RA onset and during follow-up. Cumulative incidence (CI) at four and eight years, incidence rate (IR), and frequency were calculated for the different manifestations identified. Cox regression was used to assess potential risk factors.

Results

Pulmonary involvement was identified in 89 of 204 patients (43.6%). The CI increased from 17.5% at four years to 25.2% at eight years. The IR was 42.0 per 1,000 person-years among patients without prior lung disease, rising to 50.2 when including pre-existing cases. The screening strategy proved effective, detecting asymptomatic lung involvement in one-quarter of patients (51/204).

Interstitial lung disease (ILD) (22.5%) and bronchiectasis (22.1%) were the most frequent manifestations, followed by follicular bronchiolitis (FB) (7.8%), pulmonary nodules (5.4%), pleural disease (3.4%), and obliterative bronchiolitis (OB) (1%). The IRs were 20.4 for bronchiectasis, 16.2 for ILD, 6.9 for FB, 4.9 for nodules, 2.0 for pleural disease, and 1.0 for OB.

Bronchiectasis showed the highest CI (8.8% at four years and 12.9% at eight years), followed by ILD (7.5% and 11.6%).

Age at RA onset (≥ 60 years) was independently associated with overall lung involvement (HR 2.22, 95% CI 1.20–4.11), ILD (HR 3.36, 95% CI 1.23–9.20), and bronchiectasis (HR 2.42, 95% CI 1.07–5.43). Male sex was associated with ILD (HR 5.11, 95% CI 1.52–17.13).

Conclusions

Proactive screening identified a high incidence of ILD and airway disease in early RA, supporting routine pulmonary evaluation to attempt to optimise early detection and patient outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13075-026-03736-7.

Introduction

Primary respiratory disease is the most frequent extra-articular manifestation of rheumatoid arthritis (RA). Despite its substantial morbidity and mortality, the true incidence and prevalence of RA-related pulmonary involvement remain difficult to ascertain reliably, with reported rates ranging from 30% to 80% [1–3]. This variability likely reflects differences in study populations (early vs. established RA), inclusion criteria (symptomatic vs. systematically screened patients), diagnostic approaches (clinical, radiological, pathological, or autopsy findings) and study designs.

Virtually all components of the respiratory system may be affected during RA. Pulmonary manifestations include pleural disease, rheumatoid nodules (including Caplan syndrome), interstitial lung disease (ILD), bronchiectasis, and bronchiolitis (follicular or obliterative) [1–14]. Less commonly, patients may develop primary pulmonary hypertension, apical bullous disease, thoracic cage immobility, secondary amyloidosis, thromboembolic events, or cricoarytenoid arthritis [1–14].

The occurrence of respiratory complications significantly affects clinical outcomes, contributing to persistent symptoms, impaired quality of life, increased treatment-related risks, and poorer prognosis. Given their frequency and impact, pulmonary manifestations should be assessed from diagnosis and monitored throughout follow-up. However, the lack of proactive screening in most clinical settings likely leads to under-recognition of subclinical disease and further limits the reliability of epidemiological estimates. Better epidemiological data are essential to estimate disease burden, evaluate its impact on healthcare systems, and support evidence-based planning.

This study aimed to describe the incidence, spectrum, and risk factors of primary respiratory disease throughout the disease course in an early RA cohort assessed with structured repeated respiratory evaluation within a real-life treat-to-target strategy.

Methods

This longitudinal study included 204 consecutive, unselected patients with early RA (symptom duration ≤ 12 months) diagnosed between January 2003 and December 2023 at a specialised clinic of a single tertiary hospital in Spain. RA was classified according to the 1987 ACR criteria [15] or the 2010 ACR/EULAR criteria [16], depending on the year of diagnosis. The study followed the 2007 STROBE statement for transparent reporting [17].

Following RA diagnosis, all patients underwent a screening protocol for the detection of interstitial lung disease (ILD), which consisted of a respiratory history, auscultation, chest X-ray (CXR) and pulmonary function tests (PFTs), including spirometry and haemoglobin-adjusted diffusing capacity for carbon monoxide (DLCO). These assessments were performed systematically for all patients and were to be completed within the first two years after diagnosis. The protocol was modified with the addition of routine plethysmography from 2010 onwards. High-resolution computed tomography (HRCT) was not performed universally; it was indicated in the presence of (1) respiratory symptoms, (2) abnormal findings on auscultation, or (3) abnormalities on CXR or PFTs.

Patients were managed according to a real-life treat-to-target (T2T) strategy. During follow-up, reassessment for lung involvement—including PFTs and HRCT—was performed when new respiratory symptoms or abnormal auscultation findings appeared. Otherwise, a second screening (mirroring the initial one) was systematically conducted in patients who remained in follow-up, to be completed within five years of the first screening. Subsequent screenings were performed at the discretion of the treating physician and were neither protocolised nor systematic.

Data collection and definitions

Clinical data were retrospectively reviewed, including age at diagnosis, sex, smoking history, disease activity (mean DAS28-ESR from diagnosis to screening), rheumatoid factor (RF) and anti-citrullinated protein antibody (ACPA) positivity and titre, erosive disease (by X-ray, ultrasonography, or MRI), subcutaneous nodules, respiratory symptoms at diagnosis (cough and/or dyspnoea > 3 months, except for acute events such as pleuritis), screening results, treatments, and date of pulmonary diagnosis. Although the analysis was retrospective, all data were prospectively recorded at diagnosis and during follow-up.

ILD was diagnosed based on HRCT findings assessed by expert thoracic radiologists and classified into three major ATS/ERS patterns [18]: usual interstitial pneumonia (UIP), non-specific interstitial pneumonia (NSIP), and other forms, including organising pneumonia (OP) and combined pulmonary fibrosis and emphysema (CPFE). Final diagnosis of RA-ILD was reached by a multidisciplinary team (pulmonologists and radiologists) after excluding alternative causes. No re-evaluation of HRCT images was specifically performed for this study.

Patients showing limited interstitial abnormalities of uncertain clinical significance, not meeting criteria for ILD but requiring follow-up, were considered separately. The term “interstitial lung abnormalities” (ILA) was avoided, as its current definition excludes abnormalities identified during screening of high-risk individuals, including those with RA [19]. Therefore, cases in whom subtle interstitial findings—such as very mild ground-glass opacities, reticular abnormalities, traction bronchiectasis, honeycombing, or lung distortion—were identified but were not classifiable as ILD after multidisciplinary evaluation, primarily because they were of very limited extent or very focal in nature, yet still warranted radiological follow-up to monitor for potential progression, were categorised, for the purpose of this article only, as “minimal interstitial changes” (MIC). If any progression occurred, patients were reclassified as ILD; thus, MIC denoted exclusively stable findings.

Bronchiectasis was diagnosed by HRCT. Follicular bronchiolitis (FB) was confirmed by an experienced pulmonologist based on clinical presentation, PFTs showing obstruction, and/or HRCT features (centrilobular nodules, hyperinflation, mosaic attenuation, air trapping). Obliterative bronchiolitis (OB) required histopathological confirmation.

Pulmonary nodulosis was identified on HRCT and confirmed by a multidisciplinary team, with or without histological confirmation. Pleural disease included episodes of pleurisy or evidence of pleural effusion or fibrosis on CXR or HRCT.

For the purpose of incidence analyses, the first pulmonary manifestation was defined as the earliest abnormality attributed to RA-related lung disease, whether clinically apparent or subclinical, identified through systematic screening. In cases where multiple abnormalities were detected simultaneously, the first documented finding in the medical record was considered the index event.

Statistical analysis

Continuous variables were expressed as mean and standard deviation (SD) when normally distributed, or as median and interquartile range (IQR, 25th–75th) otherwise. Categorical variables were summarised as frequencies and percentages.

Pulmonary manifestations occurring before or simultaneously with RA diagnosis were considered prevalent and excluded from incidence analyses for that manifestation, although such patients remained eligible for analyses of other manifestations. Lung disease detected after joint involvement through structured respiratory evaluation was considered incident.

Frequency was defined as the number and proportion of patients presenting a given pulmonary manifestation at any time during the disease course, including both prevalent and incident cases. The term “prevalence” was avoided because data did not refer to a single time point.

To account for differences in follow-up duration, incidence rate (IR) and cumulative incidence (CI) were estimated in separate but complementary populations. IRs were calculated for the whole cohort (n = 204) as the number of new cases divided by the total person-time at risk, defined from RA diagnosis to the pulmonary event, end of follow-up, or death, and expressed as cases per 1,000 person-years.

CI at four and eight years was estimated from patients with at least eight years of follow-up (i.e. diagnosed before December 2015, thus ensuring a period of eight years from diagnosis to present date; n = 151). The denominator included patients at risk at RA diagnosis, excluding prevalent cases for each manifestation; the numerator comprised new cases identified during follow-up. Patients lost to follow-up or deceased before completion were retained in the analysis to minimise selection bias.

As death precludes the development of pulmonary involvement and this effect cannot be captured using conventional measures such as IR and CI, a sensitivity analysis was performed by estimating the cumulative incidence function (CIF) for the different pulmonary manifestations in the full cohort, treating death as a competing event. For ILD specifically, Fine–Gray subdistribution hazard models were additionally fitted for age and sex separately to investigate the impact of each variable on the CIF, also treating death as a competing event.

In addition to individual manifestations, overall lung involvement was defined as the presence of at least one pulmonary manifestation, regardless of type, and reported as the number of affected patients. When the number of manifestations rather than patients was provided, this was explicitly stated.

Comparisons of numerical variables between groups were performed using either Student’s t-test (normal distribution) or Mann–Whitney U test (non-normal distribution). For categorical variables, chi-squared test or Fisher’s exact test were used as appropriate.

Analyses of predisposing risk factors were performed for all lung manifestations except OB due to its very low incidence. Variables assessed included age at RA diagnosis (< 60 vs. ≥60 years), sex, smoking status (never, ≤ 20 pack-years, > 20 pack-years), RF (considered positive in cases > 3 times above the upper limit of normal [ULN]), ACPA (categorised as negative, positive ≤ 3 times above the ULN, and positive > 3 times the ULN), mean DAS28-ESR from diagnosis to screening (≤ 3.2 vs. >3.2), and presence of subcutaneous rheumatoid nodules. Risk factor analysis was conducted using cause-specific Cox proportional hazards regression models, with RA diagnosis as baseline, excluding prevalent cases of that manifestation and censoring at death or last follow-up. Multivariable models including age, sex, smoking status, RF, ACPA, and disease activity were fitted for manifestations with a sufficient number of events (ILD, bronchiectasis, and overall lung involvement) to assess independent associations.

Incidence data and effect size measures are reported with 95% confidence intervals (95% CI). Statistical significance was set at p ≤ 0.05. Analyses were performed with IBM SPSS Statistics v25 and R v4.4.3.

Results

Clinical characteristics of the study population

Table 1 summarises the main clinical features, laboratory findings, and screening data of the 204 patients included in the study.

Table 1.

Clinical characteristics, laboratory data and screening findings of the study cohort

	N = 204
Age, years	52.9 ± 14
Women/Men	139 (68.1%)/65 (31.9%)
Smoking history (current or past)	110 (53.9%)
≤ 20 pack-years	49 (24%)
> 20 pack-years	61 (29.9%)
RA characteristics
RF-positive, missing data = 1	165 (81.3%)
≤ 3 times the ULN	30 (14.8%)
> 3 times the ULN	129 (63.5%)
Unknown titre	6 (3%)
ACPA-positive, missing data = 2	154 (76.2%)
≤ 3 times the ULN	7 (3.5%)
> 3 times the ULN	146 (72.3%)
Unknown titre	1 (0.5%)
Erosive	96 (47.1%)
Rheumatoid nodules	17 (8.3%)
Mean DAS28-ESR from diagnosis to pulmonary screening	4.3 ± 1.8
Patients with lung involvement preceding or coinciding with joint symptoms	23 (11.3%)
No pre-existing lung involvement	181 (88.7%)
Respiratory symptoms at RA diagnosis in patients without pre-existing lung involvement	24 (13.3%)
Pleuroparenchymal abnormalities on screening CXR in patients without pre-existing lung involvement	8 (4.4%)
Screening PFTs in patients without pre-existing lung involvement
%pFVC	104.4 ± 21.5
%pFEV1	96.2 ± 21.7
FEV1/FVC	0.76 ± 0.10
%pPEF	97.2 ± 22.6
%pMMEF	77.1 ± 38.1
%pTLC	104.1 ± 19.4
%pRV	117.7 ± 40.4
%pDLCO	83.2 ± 18.4
%pKCO	91.1 ± 17.7
Treatments
Glucocorticoids	191 (93.6%)
csDMARD	200 (98%)
Methotrexate	170 (83.3%)
Leflunomide	141 (69.1%)
Sulfasalazine	14 (6.9%)
Hydroxychloroquine	38 (18.6%)
bDMARD	107 (52.5%)
Adalimumab	28 (13.7%)
Etanercept	16 (7.8%)
Infliximab	24 (11.8%)
Golimumab	9 (4.4%)
Certolizumab pegol	15 (7.4%)
Tocilizumab	38 (18.6%)
Sarilumab	5 (2.5%)
Abatacept	44 (21.6%)
Rituximab	44 (21.6%)
tsDMARD	18 (8.8%)
Tofacitinib	5 (2.5%)
Baricitinib	14 (6.9%)
Upadacitinib	6 (2.9%)
Filgotinib	3 (1.5%)

Data are presented as mean ± standard deviation (SD) or as number and percentages

Abbreviations: ACPA anti-cyclic citrullinated peptide antibodies, bDMARD biologic disease-modifying antirheumatic drugs, csDMARD conventional synthetic disease-modifying antirheumatic drugs, CXR chest X-ray, DAS28-ESR 28-joint Disease Activity Score based on erythrocyte sedimentation rate, DLCO haemoglobin-corrected carbon monoxide diffusing capacity, FEV ₁ forced expiratory volume in one second, FVC forced vital capacity, KCO haemoglobin-corrected carbon monoxide diffusing capacity per unit of alveolar volume (DLCO/VA), MMEF maximum mid-expiratory flow, PEF peak expiratory flow, PFTs pulmonary function tests, RA rheumatoid arthritis, RF rheumatoid factor, RV residual volume, TLC total lung capacity, tsDMARD targeted synthetic disease-modifying antirheumatic drugs, ULN upper normal limit

Most patients were women (68%), with a mean age at RA diagnosis of 53 ± 14 years. Approximately half had a smoking history (54%), most of whom had accumulated more than 20 pack-years. Seropositivity rates were high, with 81% testing positive for RF and 76% for ACPA, most with titres > 3 times the ULN. The mean DAS28-ESR from diagnosis to screening was 4.3 ± 1.8. Nearly half of the patients (47%) developed erosions during follow-up, and 8.3% presented subcutaneous rheumatoid nodules.

Regarding treatments, most patients received oral glucocorticoids (93.6%) and conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) (98%), with methotrexate being the most prescribed (83.3%). In total, 53.4% received biologic or targeted synthetic DMARDs (b/tsDMARDs), mainly abatacept and rituximab; baricitinib was the most used tsDMARD.

Respiratory involvement: prevalent cases and cases detected through systematic screening

At RA diagnosis, 181 of 204 patients (88.7%) had no known respiratory involvement, while 23 (11.3%) had pre-existing or concurrent pulmonary disease (prevalent, extra-articular-onset cases).

Patients with a pulmonary onset of their rheumatoid disease constitute a distinct subgroup, as they had previously undergone lung assessments, and most were under pulmonology follow-up. Although these patients also underwent screening at diagnosis (as the presence of one pulmonary manifestation does not preclude the development of others), they were described separately from the rest of the cohort, as this represents a markedly different clinical scenario in which pre-existing lung disease alters both the interpretation of screening results and subsequent follow-up. In this group, ILD was the most frequent complication (16 cases), followed by bronchiectasis (8), pleural disease (3), FB (2), pulmonary nodules (1), and MIC (1). Seven patients presented two or more manifestations. Interestingly, most patients were male (52.2%), and the onset of articular symptoms tended to occur later (median age 66 years, IQR 57–73). The proportion of smokers (82.6%), particularly those with a high cumulative exposure, and the rate of RF and ACPA positivity (95.7% each), mostly at high titres, were remarkably elevated. These characteristics, except for RF, differed significantly from the remainder of the cohort (Supplementary Table S1).

Regarding serological evolution over time, at least one RF and ACPA measurement at the time of lung disease was available for 14 patients. Two of these patients presented with pleuritis before developing RA: one of them was negative for both RF and ACPA at the time of pleuritis and remained so during the subsequent articular disease; the other was positive for both antibodies already at the time of pleuritis. The remaining 12 patients had ILD (in some cases accompanied by bronchiectasis, FB, or pulmonary nodules). Nine were RF- and ACPA-positive already at ILD onset; in contrast, two patients were double-negative during the evaluation of the lung disease and developed both antibodies later with the onset of arthritis; finally, one patient was RF-positive and ACPA-negative at pulmonary onset and subsequently seroconverted to ACPA positivity when articular symptoms appeared.

Among patients without known pulmonary involvement (n = 181), initial systematic screening was completed a median of 6 months after RA diagnosis (IQR 2.2–21.6). Lung disease was detected in 38 patients (21%), including 18 ILD, 4 MIC, 16 bronchiectasis, 4 FB, 2 OB, 1 pleural disease, and 5 pulmonary nodulosis (some overlapping).

At diagnosis, 24 of these patients (13.3%) reported chronic respiratory symptoms, while pleuroparenchymal abnormalities were seen in 8 (4.4%) on chest radiograph. Baseline PFTs showed preserved volumes (mean predicted FVC 104.4%, TLC 104.1%) but mild gas transfer reduction (pDLCO 83.2%) and small airway dysfunction (predicted maximal mid-expiratory flow 77.1%, residual volume 117.7%), with normal FEV₁/FVC ratios (Table 1).

Of the 181 patients, 148 (81.7%) underwent first pulmonary screening within two years of diagnosis—126 (85.1%) in the first and 22 (14.9%) in the second year—detecting pulmonary involvement in 27 (18.2%). Screening was delayed beyond two years in 33 patients (18.3%) due to the COVID-19 pandemic (median 42.5 months, IQR 33–46), with lung disease identified in 11 (33.3%).

A flow diagram depicting the successive respiratory evaluation of these patients, along with the number of patients who underwent HRCT, is shown in Fig. 1. Among the 143 patients with a negative initial screening, ten developed symptoms thereafter, leading to the diagnosis of a RA-related lung manifestation in five. Of the remaining 133 asymptomatic patients, 98 (73.7%) underwent a second systematic screening after a median of 21.5 months (IQR 11–75) from the first, corresponding to 43.5 months (IQR 15–99) after diagnosis, with some delays also due to the pandemic. Subclinical involvement was detected in 12 (12.2%). Loss to follow-up, transfer to another hospital, late inclusion in the study, and death were the main reasons for the other 35 patients (19.3% of the 181) not being rescreened. Of the 86 patients with two negative screenings, 48 asymptomatic patients underwent a third evaluation after a median of 20.5 months (IQR 11–51) from the second (77 months from diagnosis, IQR 48–105), revealing subclinical disease in five (10.4%).

Fig. 1.

Flow diagram depicting the respiratory evaluation of patients without known lung involvement at RA onset through successive screenings. ^aMain reasons were loss to follow-up or transfer to another hospital, insufficient follow-up time since the previous screening (e.g., due to late inclusion in the study), or death. Abbreviations: RA = rheumatoid arthritis; HRCT = high-resolution computed tomography

Overall, 132 patients (64.7%) underwent at least one CT scan during follow-up. Indications for the first CT scan conducted were as follows: (1) the presence of respiratory symptoms or abnormal chest auscultation (regardless of CXR/PFT findings), in 48 patients (36.3%); (2) abnormalities detected in screening tests (CXR or PFTs) in patients without symptoms or sings of respiratory compromise, in 62 (47%); and (3) other reasons unrelated to RA, in 22 (16.7%). When comparing patients with and without CT imaging during follow-up, the subgroup with CT imaging had a higher proportion of men (37.9% vs. 20.8%, p = 0.001), presented with RA later in life (mean age 55 vs. 49 years, p = 0.013), and had higher smoking rates (61.4% vs. 40.3%, p = 0.004), particularly with high cumulative exposure (Supplementary Table S2). Serology, disease activity, and follow-up duration did not differ between groups.

Frequency and timing of RA-related pulmonary manifestations

Pulmonary involvement was identified in 89 of 204 patients (43.6%) [Table 2]. Among them, 42.7% were symptomatic at diagnosis. Fifty-two patients (25.5%) had a single pulmonary manifestation, 29 (14.2%) had two, and 8 (3.9%) had three or more, totalling 136 distinct manifestations. Pulmonary disease preceded or coincided with RA in 25.8% and developed subsequently in 74.2%. Regarding the 23 cases considered prevalent, the pulmonary diagnosis occurred a median of 16 months (IQR 10–30) before RA onset: 9 cases within the year preceding RA, 7 cases between one and two years before, and 7 cases evenly distributed between two and eight years prior.

Table 2.

Frequency and temporal pattern of pulmonary involvement in patients with rheumatoid arthritis

	Frequency N = 204	Symptomatic involvement	Time of diagnosis
			Before or concurrentb	Afterc
Lung involvement (overall)a	89 (43.6%)	38 (42.7%)	23 (25.8%)	66 (74.2%)
Interstitial lung disease	46 (22.5%)	27 (58.7%)	16 (34.8%)	30 (65.2%)
Minimal interstitial changes	9 (4.4%)	1 (11.1%)	1 (11.1%)	8 (88.9%)
Bronchiectasis	45 (22.1%)	19 (42.2%)	8 (17.8%)	37 (82.2%)
Follicular bronchiolitis	16 (7.8%)	7 (43.8%)	2 (12.5%)	14 (87.5%)
Obliterative bronchiolitis	2 (1%)	2 (100%)	0 (0%)	2 (100%)
Pleural disease	7 (3.4%)	7 (100%)	3 (42.9%)	4 (57.1%)
Pulmonary rheumatoid nodules	11 (5.4%)	3 (27.3%)	1 (9.1%)	10 (90.9%)

Values are expressed as number (percentage) of patients

Other relevant findings—not included in the overall count—were bronchial wall thickening in 29 patients (14.2%; 28 [96.6%] after the diagnosis of RA) and emphysema in 41 patients (20.1%; 35 [85.4%] after the diagnosis of RA)

ᵃData refer to the number of patients presenting with at least one pulmonary manifestation, irrespective of the specific type. For patients with multiple pulmonary findings, the information under ‘Symptomatic involvement’ and ‘Time of diagnosis’ refers to the first identified manifestation

^bPatients in whom the diagnosis of the pulmonary manifestation precedes or coincides with the diagnosis of rheumatoid arthritis (prevalent cases)

^cPatients in whom the diagnosis of the pulmonary manifestation is subsequent to the diagnosis of rheumatoid arthritis (incident cases detected through successive screenings)

ILD was the most frequent finding, affecting 46 patients (22.5%), 27 of whom (58.7%) were symptomatic. It was diagnosed after RA in 65.2% (median 47 months, IQR 19.5–96) and before or concurrently in 34.8% (median 16 months, IQR 4–46). Pre-existing ILD cases were distributed over time as follows: 6 were diagnosed within the year preceding RA onset, 5 between one and two years prior, and the remaining 5 evenly distributed between two and eight years before. The predominant HRCT pattern was UIP (n = 16, 35%), followed by NSIP (n = 12, 26%), OP or NSIP with OP overlap (n = 7, 15.2%), and CPFE (n = 4, 8.7%); 7 cases were classified as respiratory bronchiolitis–ILD or indeterminate. Most ILD (82.6%) showed fibrosing features.

Pulmonary hypertension (PH) of group 3 occurs frequently in the setting of CPFE and carries a poor prognosis. Two CPFE patients (50%) in our cohort developed this complication. The first patient was diagnosed with mild PH by right heart catheterisation (RHC) 13 years after the diagnosis of ILD. Repeat RHC eight months after diagnosis showed an increase in mean pulmonary arterial pressure (mPAP; from 24 to 34 mmHg) and pulmonary vascular resistance (from 4.2 to 6.4 WU), together with a fall in cardiac index (from 2.2 to 1.8 L/m²). This deterioration led to bilateral lung transplantation 11 months after the diagnosis of PH. In the second patient, PH developed only two years after the diagnosis of ILD and also worsened rapidly, with echocardiogram-estimated systolic PAP increasing from 40 mmHg to 108 mmHg in three years, resulting in right ventricular dilatation and dysfunction, and ultimately death later that same year.

MIC occurred in 9 patients (4.4%), mainly after RA onset (88.9%). Non-tractional bronchiectasis was found in 45 (22.1%), with 19 (42.2%) symptomatic; most were diagnosed after RA (82.2%). If we also take into account patients who showed bronchial wall thickening on CT scan (29 patients [14.2%], 28 of whom [96.6%] were diagnosed after RA onset), the total number of patients with features of bronchopathy, i.e. bronchiectasis and/or bronchial wall thickening, rises to 59 (28.9%). FB was observed in 16 (7.8%), 7 (43.8%) symptomatic and 2 (12.5%) prevalent, while OB was exceptional (2 patients, 1%), both symptomatic and post-RA onset.

Pleural disease occurred in 7 patients (3.4%), all symptomatic, diagnosed before or at RA onset in 42.9% and after in 57.1%. Pulmonary rheumatoid nodules were found in 11 (5.4%), mostly asymptomatic (72.7%) and post-RA onset (90.9%).

Regarding other relevant findings, 41 patients (20.1%) showed emphysema; this diagnosis occurred predominantly after the diagnosis of RA (35 patients, 85.4%) and was observed almost exclusively in smokers (40 patients, 97.6%).

Cumulative incidence and incidence rate

Cumulative incidence (CI) at four and eight years was calculated in the 151 patients with at least eight years of follow-up (Table 3). Among them, 143 (94.7%) had no prior lung involvement. Respiratory manifestations were observed in 25 patients within four years and in 36 after eight, yielding a CI of 17.5% (95% CI 11.6–24.7) and 25.2% (18.3–33.1), respectively.

Table 3.

Cumulative incidence and incidence rates of distinct pulmonary manifestations in rheumatoid arthritis

	Cumulative incidenceb					Incidence ratec
	Patients at risk	New cases within 4 years	CI at 4 years	New cases within 8 years	CI at 8 years	Patients at risk	Overall at-risk time	Total new cases	Incidence rate
			% (95% CI)		% (95% CI)		person-years		cases per 1,000 person-years (95% CI)
Lung involvement (overall)a	143	25	17.5 (11.6–24.7)	36	25.2 (18.3–33.1)	181	1570.1	66	42.0 (31.9–52.2)
Interstitial lung disease	147	11	7.5 (3.8–13.0)	17	11.6 (6.9–17.9)	188	1851.7	30	16.2 (10.4–22.0)
Minimal interstitial changes	151	3	2.0 (0.4–5.7)	5	3.3 (1.1–7.6)	203	2048.4	8	3.9 (1.2–6.6)
Bronchiectasis	147	13	8.8 (4.8–14.6)	19	12.9 (8.0–19.4)	196	1811.3	37	20.4 (13.8–27.0)
Follicular bronchiolitis	150	4	2.7 (0.7–6.7)	6	4.0 (1.5–8.5)	202	2019.7	14	6.9 (3.3–10.6)
Obliterative bronchiolitis	151	2	1.3 (0.2–4.7)	2	1.3 (0.2–4.7)	204	2065.9	2	1.0 (0.00–2.3)
Pleural disease	150	1	0.7 (0.02–3.7)	4	2.7 (0.7–6.7)	201	2045.5	4	2.0 (0.04–3.9)
Pulmonary rheumatoid nodules	151	2	1.3 (0.2–4.7)	5	3.3 (1.1–7.6)	203	2024.6	10	4.9 (1.9–8.0)

Values represent the number of patients, unless otherwise specified

Abbreviations: CI cumulative incidence, 95% CI 95% confidence interval

^aData refer to the number of patients who developed at least one pulmonary manifestation (regardless of type), rather than the total number of manifestations observed. Since the occurrence of a first pulmonary manifestation qualifies as a ‘new case’, any subsequent manifestations do not affect the incidence estimates presented here

^bCalculated based on the subgroup of 151 patients with 8 years of follow-up

^cCalculated based on the entire cohort of 204 patients

Across the entire cohort, 181 patients were at risk (no previous lung involvement), contributing 1,570.1 person-years (PY). Sixty-six developed a pulmonary complication, for an incidence rate (IR) of 42.0 per 1,000 PY (95% CI 31.9–52.2). Considering manifestations rather than patients, 105 pulmonary complications occurred after RA diagnosis (2,092.5 PY), corresponding to an IR of 50.2 per 1,000 PY (40.6–59.8); in the ≥ 8-year subcohort, 0.24 and 0.38 complications per patient were recorded at four and eight years, respectively.

When analysed separately, ILD showed a CI of 7.5% (95% CI 3.8–13.0) at four years and 11.6% (6.9–17.9) at eight, with an IR of 16.2 per 1,000 PY (10.4–22.0). MIC had a CI of 2.0% (0.4–5.7) and 3.3% (1.1–7.6) at the same time points, with an IR of 3.9 per 1,000 PY (1.2–6.6).

Regarding airway disease, non-tractional bronchiectasis (including bronchiolectasis) was most frequent (IR 20.4 per 1,000 PY; 95% CI 13.8–27.0), with a CI of 8.8% (4.8–14.6) at four years and 12.9% (8.0–19.4) at eight. FB was less common (CI 2.7% and 4.0%; IR 6.9 per 1,000 PY), while OB remained exceptional (two cases). Pleural disease and pulmonary nodules were infrequent, with IRs of 2.0 (0.04–3.9) and 4.9 (1.9–8.0) per 1,000 PY.

Given that not all patients underwent HRCT, IRs were recalculated including only those with at least one CT scan (Supplementary Table S3); however, this subanalysis likely overestimates incidence due to selection bias towards higher-risk patients.

As shown in Table 4, most first pulmonary events occurred within five years after RA diagnosis, about half within the first three, indicating an early incidence peak. A considerable proportion were subclinical, both early and later in follow-up.

Table 4.

Incident cases of lung involvement throughout the follow-up period after diagnosis of rheumatoid arthritis: all types and ILD-specific data

	Year of follow-up after RA diagnosis
	First	Second	Third	Fourth	Fifth	Sixth	Seventh	Eighth	Ninth	Tenth	> 10
Lung involvement (overall)a
Number of patients	19	5	8	3	8	2	2	5	4	4	6
Symptomatic	7 (36.8)	1 (20.0)	2 (25.0)	1 (33.3)	3 (37.5)	1 (50.0)	0 (0)	2 (40.0)	2 (50.0)	1 (25.0)	1 (16.7)
Subclinical	12 (63.2)	4 (80.0)	6 (75.0)	2 (66.7)	5 (62.5)	1 (50.0)	2 (100)	3 (60.0)	2 (50.0)	3 (75.0)	5 (83.3)
Interstitial lung disease
Number of patients	9	2	3	2	3	1	2	1	2	1	4
Symptomatic	3 (33.3)	1 (50.0)	1 (33.3)	0 (0)	2 (66.7)	1 (100)	1 (50.0)	0 (0)	2 (100)	1 (100)	2 (50.0)
Subclinical	6 (66.7)	1 (50.0)	2 (66.7)	2 (100)	1 (33.3)	0 (0)	1 (50.0)	1 (100)	0 (0)	0 (0)	2 (50.0)

Values represent the number (%) of patients

^aData refer to the number of patients who developed a first pulmonary manifestation, regardless of its type, rather than the total number of manifestations. Subsequent pulmonary events are not included

Overall, 17 patients died across the entire cohort, ten from respiratory causes, six of which were infectious. The cumulative incidence function (CIF) for the different pulmonary manifestations, accounting for this competing risk, showed a continued increase in incidence even beyond ten years, with no evidence of a plateau (Fig. 2). CI at four and eight years, estimated using this approach, was 19.9% and 30.8% for overall lung involvement, 8.8% and 13.3% for ILD, 1.6% and 3.4% for MIC, 9.0% and 14.0% for bronchiectasis, 3.6% and 6.1% for FB, 1.0% at both time points for OB, 0.5% and 2.5% for pleural disease, and 2.1% and 4.7% for lung nodules, respectively (Supplementary Table S4). For ILD, Fine–Gray models showed that both older age at RA onset (subdistribution hazard ratio [SHR] 2.24, 95% CI 1.09–4.57; p = 0.027) and male sex (SHR 3.19, 95% CI 1.56–6.50; p = 0.001) had a significant impact on the CIF (Fig. 3).

Fig. 2.

Cumulative incidence of lung manifestations over time in the full cohort, based on the cumulative incidence function (CIF), with death treated as a competing event

Fig. 3.

Cumulative incidence function (CIF) of interstitial lung disease stratified by age and sex, based on Fine-Gray subdistribution hazard models, with death treated as a competing event. A Age at RA diagnosis. B Sex

Risk factor analysis

The results for each predictor variable and specific lung manifestation are presented in Table 5. Age at RA diagnosis, male sex, and smoking were all associated with the presence of overall lung involvement in the univariable analysis; however, only age (≥ 60 years) remained an independent risk factor in the multivariable analysis, conferring approximately a twofold increase in risk compared with younger patients (HR 2.22, 95% CI 1.20–4.11; p = 0.011).

Table 5.

Analysis of risk factors for rheumatoid arthritis-related lung manifestations using Cox regression models

	Univariable		Multivariable
Risk factor	HR (95% CI)	p	HR (95% CI)	p
Lung involvement (overall)
Age	1.70 (1.02–2.83)	0.042	2.22 (1.20–4.11)	0.011
Sex	1.75 (1.06–2.88)	0.029	1.32 (0.66–2.61)	ns
Smoking	1.46 (1.11–1.91)	0.007	1.34 (0.94–1.90)	ns
RF	1.28 (0.92–1.80)	ns	1.37 (0.85–2.22)	ns
ACPA	1.13 (0.84–1.51)	ns	0.90 (0.58–1.38)	ns
Disease activity	0.99 (0.52–1.90)	ns	0.91 (0.47–1.76)	ns
Subcutaneous RN	1.32 (0.63–2.77)	ns
Interstitial lung disease
Age	2.29 (1.10–4.78)	0.028	3.36 (1.23–9.20)	0.018
Sex	3.26 (1.57–6.80)	0.002	5.11 (1.52–17.13)	0.008
Smoking	1.37 (0.91–2.05)	ns	0.80 (0.43–1.47)	ns
RF	1.30 (0.77–2.19)	ns	1.76 (0.77–4.01)	ns
ACPA	1.25 (0.79–1.98)	ns	0.67 (0.33–1.37)	ns
Disease activity	0.81 (0.29–2.30)	ns	0.95 (0.33–2.73)	ns
Subcutaneous RN	0.64 (0.25–2.70)	ns
Minimal interstitial changes
Age	1.59 (0.38–6.69)	ns
Sex	2.36 (0.59–9.43)	ns
Smoking	1.53 (0.70–3.33)	ns
RF	0.90 (0.40–2.05)	ns
ACPA	1.06 (0.47–2.41)	ns
Disease activity	-	ns
Subcutaneous RN	0.04 (0.00–852.36)	ns
Bronchiectasis
Age	1.92 (0.98–3.79)	ns	2.42 (1.07–5.43)	0.033
Sex	1.91 (0.98–3.70)	ns	1.27 (0.49–3.29)	ns
Smoking	1.47 (1.02–2.13)	0.040	1.20 (0.73–1.97)	ns
RF	1.17 (0.76–1.81)	ns	1.41 (0.75–2.63)	ns
ACPA	0.93 (0.64–1.34)	ns	0.71 (0.41–1.23)	ns
Disease activity	0.92 (0.39–2.18)	ns	0.89 (0.37–2.14)	ns
Subcutaneous RN	1.48 (0.58–3.82)	ns
Follicular bronchiolitis
Age	4.96 (1.65–14.95)	0.004
Sex	0.90 (0.28–2.88)	ns
Smoking	1.75 (0.96–3.20)	ns
RF	4.44 (0.82–24.14)	ns
ACPA	1.81 (0.76–4.34)	ns
Disease activity	0.49 (0.14–1.76)	ns
Subcutaneous RN	0.68 (0.09–5.20)	ns
Pleural disease
Age	2.56 (0.36–18.25)	ns
Sex	2.37 (0.33–16.82)	ns
Smoking	1.81 (0.59–5.59)	ns
RF	1.11 (0.32–3.86)	ns
ACPA	0.79 (0.28–2.22)	ns
Disease activity	0.17 (0.02–1.87)	ns
Subcutaneous RN	3.02 (0.31–29.08)	ns
Pulmonary rheumatoid nodules
Age	0.29 (0.04–2.27)	ns
Sex	1.54 (0.43–5.45)	ns
Smoking	1.47 (0.73–2.94)	ns
RF	2.20 (0.67–7.18)	ns
ACPA	1.50 (0.62–3.63)	ns
Disease activity	0.35 (0.09–1.40)	ns
Subcutaneous RN	6.48 (1.83–22.99)	0.004

“Age” refers to age at RA onset ≥ 60 years (reference: <60 years). “Sex” refers to male (reference: female). HRs for “smoking” and “ACPA” refer to the increase by one category on the ordinal scale (e.g., from never smoker to ≤ 20 pack-years, or from ACPA-positive ≤ 3 times the ULN to > 3 times the ULN). “Disease activity” refers to moderate-to-high activity, defined as a mean DAS28-ESR > 3.2 (reference: ≤3.2). “Subcutaneous RN” refers to presence (reference: absence)

Abbreviations: HR hazard ratio, 95% CI 95% confidence interval, RF rheumatoid factor, ACPA anti-citrullinated protein antibodies, RN rheumatoid nodules, ns not significant, ULN upper limit of normal

Regarding specific manifestations, older age at RA onset was also a strong predictor of ILD development, resulting in roughly a threefold increased risk after adjustment for other variables (HR 3.36, 95% CI 1.23–9.20; p = 0.018). Similarly, men had an adjusted hazard of incident ILD more than five times higher than that of women (HR 5.11, 95% CI 1.52–17.13; p = 0.008). Smoking, RF, ACPA, disease activity, and rheumatoid nodules showed no association with incident ILD in either the univariable or multivariable analyses. Likewise, no predictor variable demonstrated a significant association with the development of MIC.

With respect to airway disease, older age was an independent risk factor for the development of bronchiectasis in the multivariable analysis, conferring a little more than a twofold increase in risk (HR 2.42, 95% CI 1.07–5.43; p = 0.033). In the case of follicular bronchiolitis, the effect was even more pronounced, with a fivefold increase in risk (HR 4.96, 95% CI 1.65–14.95; p = 0.004), although based on an unadjusted model. Regarding the other variables, only smoking appeared to predict the development of bronchiectasis, although this association was no longer statistically significant in the multivariable analysis.

No association was observed between the variables studied and pleural disease. In contrast, the presence of subcutaneous rheumatoid nodules was a strong predictor of pulmonary nodule development, showing an unadjusted hazard more than six times higher than that of patients without subcutaneous nodules (HR 6.48, 95% CI 1.83–22.99; p = 0.004).

Discussion

In our study, pulmonary involvement was identified in 43.6% of patients with RA, underscoring the substantial burden of respiratory complications. Among those with at least eight years of follow-up, the cumulative incidence rose from 17.5% at four years to 25.2% at eight, and nearly half of first pulmonary events occurred within the first three years after RA onset. The incidence rate was 42 per 1,000 person-years among patients without prior lung disease, increasing to 50.2 when those with respiratory disease diagnosed before arthritis onset, that is, with extra-articular onset of rheumatoid disease, were included. These findings confirm early appearance and continuous accumulation of RA-related lung involvement, with a sustained rise beyond ten years. Notably, 57% of affected patients were asymptomatic at diagnosis, which supports systematic screening beyond symptom-based approaches. Baseline and follow-up screening detected preclinical pulmonary involvement in one quarter of patients (51 of 204).

As expected, ILD was the most frequent manifestation, observed in 22.5% of patients. This aligns with two recent meta-analyses that reported pooled prevalences of 18.7% [20] and 21.38% [21], both with substantial heterogeneity. At least three studies in early RA with disease duration under two years reported even higher rates, between 27.5% and 41.8% [9, 22, 23], suggesting that ILD may develop early.

The cumulative incidence of ILD in our cohort reached 7.5% at four years and 11.6% at eight (8.8% and 13.3%, respectively, by CIF estimates), closely mirroring the large United States cohort by Samhouri et al. [24], where CT-confirmed ILD occurred in 7.6% at five years, 11.0% at ten, and 15.3% at twenty after RA onset. In contrast, an earlier population-based study from Rochester, Minnesota, reported lower rates of 3.5%, 6.3%, and 7.7% at ten, twenty, and thirty years, respectively [6]. In that study, however, the risk of ILD among patients with RA was markedly higher than in the general population (hazard ratio 8.96), emphasising its clinical relevance despite methodological differences.

With an incidence rate of 16.2 cases per 1,000 person-years, our cohort shows the highest burden of RA-associated ILD reported to date. Much lower rates have been described in Asian cohorts: 1.06 per 1,000 in the Japanese IORRA cohort, where only NSIP and diffuse alveolar damage patterns were included [25]; 1.09 per 1,000 in a nationwide Taiwanese study [26]; and 0.7 to 6.1 per 1,000 in the South Korean KOBIO registry depending on treatment [27].

In Europe, a study based in the United Kingdom reported an annualised incidence rate of 4.1 per 1,000, although ILD screening was limited to symptomatic individuals [28]. In North America, the annual incidence rate in Canada ranged from 1.6 to 3.3 per 1,000 between 2000 and 2018 [29], while in the United States it was 0.03 to 0.04 per 1,000 [30]. In Oceania, prevalence estimates from New Zealand suggest a low burden, with 0.11 to 0.15 cases per 1,000 population [31].

The high incidence we observed likely reflects the yield of proactive and repeated screening, which improved early detection and uncovered otherwise unrecognised cases. ILD was asymptomatic in about 40% at diagnosis. In 34.8% of patients, ILD preceded joint symptoms by months or years or appeared simultaneously, and when ILD developed later, the median time from RA onset to diagnosis was about four years. These observations challenge the traditional view of ILD as a late extra-articular manifestation. Interestingly, most cases of pre-existing ILD were concentrated within the two years preceding arthritis onset, and when examining their serological data, we observed that the majority were already seropositive at the time of pulmonary onset. This finding reinforces the notion that ILD represents an early, related manifestation and suggests that particular attention should be paid to patients with new-onset ILD and RF- or ACPA-positive serology, as the onset of arthritis may occur shortly thereafter. Notably, the clinical profile of patients who presented with lung involvement prior to RA aligned with the well-established risk factors for respiratory compromise, some of which were corroborated by this study, such as male sex and late-onset RA.

The frequency of minimal interstitial changes was 4.4%. Comparisons with previous reports are limited by definitional differences, yet up to 86% of mild interstitial abnormalities in RA may progress, indicating that only a small proportion remain nonprogressive [32].

Bronchiectasis was the second most common complication after ILD (22.1%), with a frequency similar to that reported in pooled meta-analyses [33, 34]. However, the proportion of symptomatic patients in our cohort (42.2%) was substantially higher than in earlier reports (3%) [12], likely due to coexisting pulmonary manifestations. To date, only one study has assessed incidence longitudinally, reporting a 7% cumulative incidence at nine years and a twofold increased risk compared with matched controls without RA [35]. In contrast, our screening strategy detected a higher cumulative incidence at four and eight years. RA patients with bronchiectasis have higher Bronchiectasis Severity Index scores and mortality than those without RA [2, 3]. Chronic or recurrent bronchial infections may also limit the use of immunomodulatory therapy.

Regarding small airway disease, the incidence of FB in RA is not well quantified, and reported prevalence ranges widely from 6% to 65% depending on diagnostic definitions [14]. In our cohort, FB was the third most frequent manifestation after bronchiectasis, with a frequency of 7.8%, likely reflecting the application of strict diagnostic criteria. OB is recognised as a rare complication. In a retrospective study by Lin et al., 41 RA patients were diagnosed with OB over 15 years at a single centre [36]. In our series, OB was the least frequent manifestation (1.0%). To our knowledge, this is the first study to provide incidence data for small airway disease in RA.

Lung nodules were the next most frequent finding. Reported prevalence varies from 0.3% to 2.4% on chest X-ray but may reach 42% with systematic HRCT screening [13, 37, 38]. In our cohort, their frequency was moderate (5.4%) and probably underestimated since HRCT was not performed in all patients. No incidence data are available in the literature for comparison.

Pleural disease was identified in 3.4% of patients, exclusively as symptomatic pleuritis, consistent with the reported prevalence of 3% to 5% for clinical pleural involvement [3]. No asymptomatic effusions were detected. Two longitudinal studies reported annualised incidences of 0.2% to 1.5% and a cumulative incidence of 2.4% at 15 years [39, 40], findings broadly consistent with ours.

Given the strong association between ILD and increased morbidity and mortality in RA, multiple studies have examined its risk factors. The best-established clinical predictors include older age (both overall and at RA onset), male sex, smoking exposure, RF and ACPA positivity and titres, moderate-to-high disease activity, and longer RA duration [20, 41–43]. Additional biomarkers, still beyond routine clinical use, such as anti-carbamylated protein antibodies, members of the matrix metalloproteinase family, and genetic variants including the MUC5B rs35705950 risk allele and mutations in telomere-related genes, have also shown significant associations [41, 44, 45].

In our cohort, both older age at RA onset and male sex were strong independent predictors of ILD, conferring approximately threefold and fivefold increases in risk, respectively. Significantly, Fine-Gray models showed that both variables also had an impact on the cumulative incidence curve, after accounting for death as a competing event. Comparisons across studies are limited by methodological heterogeneity, including variable definitions of age and measures of association. Li et al. [46] reported a weaker effect (age > 60 years: OR 1.49, 95% CI 1.01–2.18), whereas Samhouri et al. [24] and Yu et al. [43] found HRs of 1.89 per 10-year and 1.05 per one-year increase, respectively. For male sex, Yu et al. [43] reported an HR of 2.15, smaller than in our study. Despite previous evidence, smoking, serological status, and disease activity were not associated with incident ILD in our cohort, nor was the presence of rheumatoid nodules, which had been identified as a risk factor in the meta-analyses by Yu et al. [43] and Wang et al. [20].

Several predictors have been linked to bronchiectasis in RA, including older age, male sex, RF, ACPA, and disease duration [12, 33, 47]. In the study by McDermott et al. [47], age at RA diagnosis increased the risk of bronchiectasis by 37% per 10-year increment. In our cohort, it was the only independent risk factor, with patients aged ≥ 60 years showing more than a twofold higher risk (HR 2.42, 95% CI 1.07–5.43). For follicular bronchiolitis, older age conferred a fivefold increase in risk, consistent with prior reports identifying advanced age, disease activity, and duration as contributing factors [48], though no effect size measures were available for comparison.

Subcutaneous rheumatoid nodules were a strong predictor of pulmonary nodulosis [49, 50], associated with an over sixfold increase in risk, although no other studies have quantified this effect. Conversely, none of the variables studied predicted pleural disease, despite prior reports implicating sex, RF, and nodules [12].

Overall, age at RA diagnosis emerged as the most influential clinical predictor of pulmonary involvement, affecting both interstitial and airway disease, while male sex represents an additional strong risk factor for ILD. These findings support heightened vigilance for respiratory manifestations, particularly in men diagnosed with RA at an older age.

When interpreting our study results, several limitations must be acknowledged, the foremost being the absence of systematic HRCT in all patients. This entails a risk of underdiagnosis and, consequently, of underestimating incidence values, particularly for manifestations with little clinical expression or easily missed by CXR and PFTs. However, 64.7% underwent a thoracic CT scan at some point during follow-up, partially mitigating this limitation, and combined screening strategies including CXR and PFTs have shown good sensitivity for ILD detection in previous studies [51]. Moreover, since the incidence for ILD observed in our cohort is among the highest reported in the literature, this potential underestimation further highlights the substantial burden of lung disease, highlighting the need for structured pulmonary evaluation during initial assessment and follow-up. Nonetheless, it should be noted that HRCT was performed selectively in patients presenting with symptoms, clinical signs, or abnormalities in screening tests suggestive of respiratory compromise, who were more frequently male, had later-onset RA, and exhibited greater tobacco exposure. Consequently, HRCT was conducted in a higher-risk subpopulation, which may not reflect the average RA patient. This selective, risk-based partial verification of cases represents a potential source of bias.

In addition, some methodological aspects should be considered when interpreting our incidence data. Crude CI estimates obtained by mathematical calculation were based on the subgroup of patients who could be followed for eight years rather than the full cohort, which may entail a certain risk of selection bias, and could not be adjusted for deaths. In this regard, the estimation of CIFs represents a methodological strength, as it is based on the entire cohort and accounts for deaths as competing risks, potentially providing more accurate estimates. Nevertheless, since the eight-year follow-up group comprised the majority of patients and the number of deaths was low, the differences between the two estimates are minimal and do not appear to be significant.

The relatively small sample size represents another limitation, particularly for the analysis of risk factors. The number of incident events, especially for manifestations such as pulmonary nodules, pleural disease, or bronchiolitis, was notably low, likely rendering the analyses underpowered and limiting our ability to perform multivariable analyses. This may have contributed to the absence of statistically significant associations with known predictors and, especially for the less frequent manifestations, necessitates cautious interpretation of the results, which should be considered exploratory.

Other limitations include the retrospective design, which may introduce information and detection bias; the lack of uniform follow-up assessments; and the use of non-standardised criteria for the diagnosis of FB and OB.

Despite these limitations, this study is based on a well-defined, unselected cohort of patients with early RA, all of whom underwent a predefined pulmonary assessment protocol from diagnosis and were managed with a real-life T2T strategy. The extended follow-up enabled a comprehensive evaluation of the incidence and timing of respiratory complications. Pulmonary involvement was carefully characterised through multidisciplinary assessment, and the study provides, for the first time, incidence data for under-recognised entities such as FB and OB. The inclusion of multiple types of pulmonary manifestations and the longitudinal design further strengthen the epidemiological value of our findings. Repeated evaluations during follow-up, even in asymptomatic patients, revealed a high burden of subclinical or late-onset manifestations. This highlights that patients with RA carry a substantial and heterogeneous burden of pulmonary involvement that remains underrepresented in the literature and insufficiently addressed in routine clinical practice. This is particularly relevant for ILD, since diagnostic delay has been shown to be an independent predictor of mortality [10]. Of note, and consistent with this notion, universal screening for ILD in patients with RA has recently been suggested by several scientific societies [52]. Future research to determine whether systematic screening improves patient outcomes is therefore strongly warranted.

Conclusion

Pulmonary disease is a major cause of morbidity and mortality in rheumatoid arthritis (RA). In this well-characterised early RA cohort with systematic pulmonary assessment and long-term follow-up, we provide the most detailed incidence data to date, including for under-recognised entities such as follicular and obliterative bronchiolitis. The results show a high burden of interstitial and airway involvement, often subclinical and particularly frequent in late-onset RA, and extend current knowledge by quantifying the timing and spectrum of lung disease from diagnosis. These findings support structured pulmonary evaluation in RA and call for further studies to assess whether systematic screening improves outcomes.

Abbreviations

ACPA

anti-citrullinated protein antibody

ACR

American College of Rheumatology

ATS/ERS

American Thoracic Society/European Respiratory Society

b/tsDMARDs

biologic or targeted synthetic disease-modifying antirheumatic drugs

CI

cumulative incidence

CPFE

combined pulmonary fibrosis and emphysema

csDMARDs

conventional synthetic disease-modifying antirheumatic drugs

CXR

chest X-ray

DAS28-ESR

Disease Activity Score in 28 joints based on erythrocyte sedimentation rate

DLCO

diffusing capacity for carbon monoxide

ESR

erythrocyte sedimentation rate

EULAR

European League Against Rheumatism

FB

follicular bronchiolitis

FEV₁/FVC

forced expiratory volume in 1 s/forced vital capacity ratio

FVC

forced vital capacity

HR

hazard ratio

HRCT

high-resolution computed tomography

ILA

interstitial lung abnormalities

ILD

interstitial lung disease

IQR

interquartile range

MIC

minimal interstitial changes

NSIP

non-specific interstitial pneumonia

OB

obliterative bronchiolitis

OP

organising pneumonia

OR

odds ratio

pDLCO

percent-predicted DLCO

PFTs

pulmonary function tests

PY

person-years

RA

rheumatoid arthritis

RF

rheumatoid factor

RV

residual volume

SD

standard deviation

TLC

total lung capacity

UIP

usual interstitial pneumonia

ULN

upper limit of normal

Authors’ contributions

MAC: 1b, 1c, 2 and 3. MMM: 1c, 2, 3. ARP: 1c, 2, 3. MRK: 1c, 2, 3. SB: 1c, 2, 3. BDR: 1c, 2, 3. JMN: 1c, 2, 3. JN: 1a, 1b, 1c, 2 and 3.

All authors had access to the data and it meets the Uniform Requirements for Manuscripts Submitted to Biomedical Journals Criteria for authorship.

1a. Substantial contributions to study conception and design.

1b. Substantial contributions to acquisition of data.

1c. Substantial contributions to analysis and interpretation of data.

2. Drafting the article or revising it critically for important intellectual content.

3. Final approval of the version of the article to be published.

Funding

None. This study is not a part of corporate-sponsored research effort.

Data availability

The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are included in the paper.

Declarations

Ethics approval and consent to participate

The study has been approved by of our institutional ethics committee (Clinical Research Ethics Committee of Bellvitge University Hospital-IDIBELL). The local ethics committee confirm that the findings in this report were based on normal clinical practice and were therefore suitable for dissemination. Informed consent was not obtained from the patients, but their clinical records and information were anonymised prior to analysis. Since our study was performed retrospectively, we were relieved by the ethics committee of our hospital from obtaining informed consent. This study was conducted in accordance with the principles of the Declaration of Helsinki and the International Conference for Harmonisation.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.