Rheumatoid arthritis (RA) is a systemic inflammatory disease characterised by chronic inflammation of the joints and their resulting destruction, deformity, and incapacity1, with extra-articular manifestations that contribute to morbidity and mortality2. The lungs are a commonly affected extra-articular location, with diffuse interstitial lung disease (ILD) representing the most frequent manifestation3. Up to 58% of patients with RA may experience lung involvement, which is associated with mortality in 10%–20% of cases4.

The prevalence of clinical RA with ILD (RA-ILD), its risk factors, and its presentation are of great interest. In Spain, there are very few epidemiological studies on the prevalence of RA-ILD, and these are limited5, and descriptive studies on lung involvement include few patients with RA6,7.

In 2007, a pioneering interdisciplinary consultation unit was set up at the La Paz University Hospital (HULP) in Madrid by the Rheumatology and Pulmonary services to jointly assess lung involvement in immune-mediated systemic diseases. The consultation unit was integrated with the centre’s ILD unit, accredited by the Spanish Society for Pulmonology and Thoracic Surgery (SEPAR) as a highly complex interdisciplinary unit. This made it possible to collect data in a structured manner, enabling all the dimensions of the disease to be studied within our area.

The objective of this study was to estimate the prevalence of symptomatic ILD, its radiological patterns in RA, and the characteristics of patients who develop respiratory symptoms within the health district of a tertiary referral hospital.

A prospective, longitudinal, observational study was designed between 2007 and 2018 in the HULP Rheumatology and Pulmonary interdisciplinary consultation unit. The protocol was approved by the HULP Research Ethics Committee.

A total of 2729 cases of RA were recorded in the area, which represents a prevalence of RA of 0.62%, which is to be expected. The symptomatic RA-ILD registry included 92 patients and 2 were eliminated (one case of juvenile idiopathic arthritis and one of pneumoconiosis). Of the 90 patients included, 69 belonged to the HULP district and 47 were alive as of the study date.

The population prevalence of symptomatic RA-ILD in the HULP health district was estimated to be 1.72% (95% CI: 1.26–2.29). By sexes, it was higher in males (3.11% vs. 1.23%), by age groups, the highest value was obtained in those younger than 55 years compared to those older than this age (2.20 95% CI: 1.28–3.52 vs. 1.53, 95% CI: 1.03–2.19). Although the CIs do not overlap for sex, which indicates the difference is statistically significant, this is not true for age and the observed difference could be due to chance (Table 1).

Table 2 shows the description of the 90 patients with RA and ILD in our series. Mean age at RA diagnosis was 57.3 ± 13.3 years and 49 (48.9%) were female. The majority, RF and positive CCPA, 16% smokers and 45% ex-smokers. Articular involvement was the initial symptomatology in 80 cases (90%), with an average time elapsed between onset of articular symptoms and respiratory symptoms of 9.9 years. At the baseline visits, 23% presented with cough; 58% with dyspnoea; 15.2% with cough and dyspnoea; 66% with crackles.

The most common radiological pattern was UIP (31%), followed by NSIP (22%), CPFE (13%), the nonspecific pattern (10%), and OP and LIP (6.7%) (Table 3). Regarding the distribution by sex, NSIP was more frequent in females (34% vs. 13%; p = 0.032) and CPFE only appeared in males (26%; p < 0.0001). Regarding age, a higher frequency of NSIP was observed in those <55 years (32% vs. 14%; p = 0.048) and CPFE in the oldest age group (20% vs. 5%; p = 0.03). UIP was also more common in those older than 55 years (39% vs. 22%; p < 0.08) but did not reach statistical significance (SS) (Table 3).

In the comparison by radiological groups, UIP + probable UIP versus NSIP (Table 4), the mean age at diagnosis in the group with UIP + probable UIP was 64.2 versus 52.3 (p = 0.039) and NSIP was more common in females (70% vs. 51.5%; p = 0.186) but did not reach SS (p = 0.18). When comparing the NSIP pattern with UIP + probable UIP, it was observed that the RF titre was greater in patients with NSIP (p = 0.020). Regarding DLCO and FVC, we did not find any significant differences, likely due to the sample size (Table 4).

In our study, a prevalence of ILD in RA lower than that described in other series was estimated17, thereby confirming that this disease is more frequent in smokers and patients with RF and positive CCPA and that UIP is the most frequent pattern.

Although the study is from a single centre, it involves a large representative area in which almost all RA patients are referred to Rheumatology (with an RA prevalence of 0.62 [95% CI: 0.60–0.64] according to own data, which matches that estimated for the general population). In addition, patients seen in the interdisciplinary consultation unit were referred by both the hospital’s Rheumatology service as well as specialised centres, meaning patients with RA across the entire spectrum of disease severity were included. Lastly, their clinical characteristics are very similar to those of other series, which supports their representation.

The prevalence of ILD in patients with RA varies from 1% to 67.3%17. These differences are not surprising due to the large number of definitions, diagnostic tests used18, criteria19, and even due to the statistical analysis used, which is not always reflected in articles19. The prevalence ranges from 1% to 6% if chest X-ray is used and from 5% to 67.3% if HRCT is performed20.

It must be considered that, while the proportion of patients with RA-ILD via HRCT is noteworthy, it is only considered clinically significant in 5%–10%21. This must be taken into account given that there is a significant gap between the large number of asymptomatic individuals with functional or radiological pulmonary changes and the number of patients who ultimately develop clinical ILD22. Kelly et al.23 conducted a study in symptomatic patients diagnosed via HRCT and obtained a result of 2%–3%. Likewise, in a study including patients with pleurisy or ILD, Bartels et al. published results of 2% in 1985 and 4% in 200618.

In Spain, the only available data on the prevalence of ILD in RA are those from the EMECAR cohort5, with a prevalence of 3.7% (95% CI: 2.4–5). In this study, the classification criteria consisted of the existence of a restrictive pattern or HRCT findings indicative of ILD but without any reference to the symptoms. Despite the fact that descriptive studies of RA-ILD in our country have recently been published6,7, we have not identified other studies that estimate its prevalence.

The prevalence of symptomatic RA-ILD in our area is 1.72% (95% CI: 1.26–2.29), comparable to the results of EMECAR5 and to the work of Kelly et al.23 Our study only includes patients with respiratory symptoms for whom we have confirmed ILD via HRCT. This means that our study could have underestimated the prevalence of RA-ILD. In addition, the fact that we did not select for only patients with severe RA could also explain this low prevalence.

In RA, respiratory involvement is the first manifestation in 10%–20% of patients3, data that coincide with our results, wherein 10 patients (10%) initially presented with respiratory symptoms. Onset of respiratory involvement in RA typically occurs around 50–60 years24 and age is considered a risk factor for developing ILD. On the other hand, age at diagnosis of RA is also related to ILD22. In our series, the mean age at diagnosis of RA was 57 years. It is noteworthy that the prevalence of ILD is higher in those younger than 55 years rather than those older, though these differences are not statistically significant.

In addition to age, smoking, RF and positive CCPA, particularly at high titres, family history of RA, and being male have been described as risk factors for developing ILD with RA23,25–27. In our series, 61% of the patients had a history of smoking and over 80% had RF and positive CCPA at high titres, which fits with the expectations.

In immune-mediated systemic diseases, the NSIP pattern is the most common with the exception of RA, for which UIP is the most frequent19. In our study, UIP is the most frequent (31.1%), followed by NSIP (22.2%), with a pattern similar to other publications, with numbers in the range of 40%–60% for UIP and 11%–30% for NSIP4. There are increasingly more data confirming that patients with RA and UIP pattern have a different phenotype than patients with non-UIP patterns. UIP is more common in males, older patients, and smokers3, while the NSIP pattern is almost 3 times more common in females28. As in other works, patients with UIP or probable UIP are older at diagnosis (p = 0.039), are more commonly men (48.5% vs. 30%), and more frequently have a history of smoking than the NSIP group (53.1% vs. 40%).

The majority of studies on ILD in auto-immune diseases have not taken in account the coexistence with emphysema, meaning the frequency of CPFE in these diseases is not known29. In our study, 13.3% of the patients showed a CPFE pattern. We were also able to confirm that these patients have a special phenotype; in our study, all were male, smokers, and over the age of 55.

The study presents some limitations. We used the age stratification provided by the primary care registries, which could represent a limitation given that the limit of 55 years is different from other series and does not allow for comparisons. Although the study was conducted at one centre, we think that the sample is sufficiently representative for the data to be extrapolated to other areas. The fundamental strength of the study is that the patients were recruited from an interdisciplinary unit which featured a diagnosis and follow-up protocol, allowing us to collect highly uniform data.

With this single-centre study, we have analysed the prevalence of symptomatic RA-ILD in our area, which is under 2%, this being far lower than that described in other works, likely in relation to the definitions used. We have also described that the UIP pattern is the most frequent in RA patients in our environment, followed by the NSIP pattern. Lastly, we analysed the prevalence of CPFE in RA for the first time, which reaches 13%.

This research did not receive any specific funding from agencies from the public sector, commercial sector, or not-for-profit organisations.

The authors declare that they do not have any conflicts of interest directly or indirectly related to the content of the manuscript.