Progressive Preclinical Interstitial Lung Disease in Rheumatoid Arthritis

Background  Early detection and treatment for interstitial lung disease (ILD) in patients with rheumatoid arthritis (RA) may ameliorate disease progression. The objective of this study was to identify asymptomatic lung disease and potential therapeutic targets in patients having RA and preclinical ILD (RA-ILD).

Methods  Sixty-four adults with RA and 10 adults with RA and pulmonary fibrosis (RAPF) were referred to the National Institutes of Health, Bethesda, Maryland, and underwent high-resolution computed tomography (HRCT) and pulmonary physiology testing. Proteins capable of modulating fibrosis were quantified in alveolar fluid.

Results  Twenty-one of 64 patients (33%) having RA without dyspnea or cough had preclinical ILD identified by HRCT. Compared with patients without lung disease, patients with RA-ILD had statistically significantly longer histories of cigarette smoking (P < .001), increased frequencies of crackles (P = .02), higher alveolar-arterial oxygen gradients (P = .004), and higher HRCT scores (P < .001). The HRCT abnormalities progressed in 12 of 21 patients (57%) with RA-ILD. The alveolar concentrations of platelet-derived growth factor–AB and platelet-derived growth factor–BB were statistically significantly higher in patients having RA-ILD (mean [SE], 497.3 [78.6] and 1473 [264] pg/mL, respectively) than in patients having RA without ILD (mean [SE], 24.9 [42.4] and 792.7 [195.0] pg/mL, respectively) (P < .001 and P =.047, respectively). The concentrations of interferon gamma and transforming growth factor β₂ were statistically significantly lower in patients having RAPF (mean [SE], 5.59 [1.11] pg/mL and 0.94 [0.46] ng/mL, respectively) than in patients having RA without ILD (mean [SE], 14.1 [1.9] pg/mL and 2.30 [0.39] ng/mL, respectively) (P =.001 and P =.006, respectively) or with preclinical ILD (mean [SD], 11.4 [2.6] pg/mL and 3.63 [0.66] ng/mL, respectively) (P =.04 and P =.007, respectively). Compared with patients having stable RA-ILD, patients having progressive RA-ILD had statistically significantly higher frequencies of treatment using methotrexate and higher alveolar concentrations of interferon gamma and transforming growth factor β₁ (P =.046, P =.04, and P =.04, respectively).

Conclusions  Asymptomatic preclinical ILD, which is detectable by HRCT, may be prevalent and progressive among patients having RA. Cigarette smoking seems to be associated with preclinical ILD in patients having RA, and treatment using methotrexate may be a risk factor for progression of preclinical ILD. Quantification of alveolar proteins indicates that potential pathogenic mechanisms seem to differ in patients having RA-ILD and symptomatic RAPF.

Interstitial lung disease (ILD) occurs in some individuals with rheumatoid arthritis (RA).^1-4 The diagnosis of ILD in RA is clinically relevant because it can be associated with a progressive natural history of disease.^2,3 Although it has been reported that 19% of outpatients with RA had radiographic findings of pulmonary fibrosis (a subtype of ILD), the prevalence of asymptomatic preclinical ILD among individuals with RA is unknown.⁴

Pathogenic mechanisms of ILD are unknown, which may contribute to limited treatment options of variable efficacy. However, recent data suggest that early institution of antifibrotic drugs to treat pulmonary fibrosis, the most common ILD, may ameliorate disease progression. For example, early (but not late) administration of pirfenidone was associated with less severe fibrosis in an animal model of pulmonary fibrosis.⁵ In addition, treatment of patients having pulmonary fibrosis and Hermansky-Pudlak syndrome with pirfenidone slowed the progressive decline in lung function more effectively in patients with mild or moderate (but not severe) lung disease.⁶ Therefore, early identification of disease and institution of therapy may improve clinical outcomes of individuals with ILD.

Given this information, we sought to develop a strategy to improve the outcomes of individuals with RA and ILD. We hypothesized that screening individuals having RA could identify early ILD before the development of symptoms. We also hypothesized that preclinical ILD, like symptomatic pulmonary fibrosis, would be progressive. To test these hypotheses, individuals having RA underwent baseline and longitudinal evaluations. In addition, to improve the understanding of the pathogenesis of early ILD and to identify potential therapeutic targets, the concentrations of alveolar proteins were quantified.

Sixty-four consecutive eligible individuals at least 21 years of age diagnosed as having definitive RA according to American College of Rheumatology criteria⁷ and 10 individuals diagnosed as having RA and pulmonary fibrosis (RAPF) were enrolled in protocol 99-H-0056, which was approved by the institutional review board of the National Heart, Lung, and Blood Institute, Bethesda, Maryland. Subjects having RA or RAPF were recruited using advertisements and were referred by their physicians. Lung biopsy specimens from 10 subjects having RAPF demonstrated usual interstitial pneumonia. Subjects having RA without symptoms of dyspnea or cough who had ILD identified by high-resolution computed tomography (HRCT) at baseline or during follow-up evaluations were classified as having RA and preclinical ILD (RA-ILD). Written informed consent was obtained, and subjects underwent evaluations at the Clinical Center of the National Institutes of Health, Bethesda. Subjects were excluded if they had smoked within 2 years or if they had other collagen vascular disorders, had chronic pulmonary disorders other than ILD, or had a history of exposure to fibrogenic substances.

Conventional and HRCT of the chest without intravenous contrast medium was performed during end inspiration with the patient in the prone position using 1-mm collimation at 10-mm intervals (General Electric Medical Systems, Milwaukee, Wisconsin). The HRCT features of ILD included septal lines, reticulation, traction bronchiectasis, cyst formation, and ground glass attenuation.^8,9 Images were read independently by 2 blinded radiologists (N.A.A. and C.K.C.). Discrepant readings were rereviewed to determine consensus findings. Using a modification of a previously described quantitative scale,^8,9 HRCT findings were assigned scores, defined as 0 (normal), 1 (minimal disease [ie, 3-4 septal lines]), 2 (mild [ie, ≥5 septal lines, reticulations, subpleural cysts, and ground glass opacities]), 3 (moderate disease [ie, grade 2 findings and traction bronchiectasis, peribronchovascular thickening, or tracheal retraction with one-third to two-thirds lung involvement]), or 4 (severe [ie, grade 2 or 3 findings with more than two-thirds lung involvement]). Change in HRCT scores was assigned as 0 (no change or improvement), 1 (worsening by 1 score), or 2 (worsening by ≥2 scores).

Pulmonary function measurements were made according to American Thoracic Society recommendations.¹⁰ Forced expiratory volume in 1 second (FEV₁), forced vital capacity (FVC), total lung capacity (TLC), and diffusion capacity of carbon monoxide (DLCO) values were expressed as the percentages of predicted values (SensorMedics, Yorba Linda, California). Arterial blood gas sampling was performed while subjects were resting and breathing ambient air.

Fifty-one subjects without medical contraindications who volunteered to undergo elective bronchoscopy received topical 1% lidocaine, intravenous midazolam hydrochloride with or without fentanyl citrate, and supplemental oxygen. A bronchoscope (Olympus America, Melville, New York) was used to instill up to four 30-mL portions of 0.9% sterile saline solution in up to 3 lung segments in the middle or upper lobes. Bronchoalveolar lavage fluid was collected, filtered through gauze, and centrifuged at 1000g for 15 minutes at 4°C. Supernatant was stored in polypropylene tubes at −80°C and thawed immediately before analysis. Following bronchoscopy, a blood sample was obtained, and urea concentrations in bronchoalveolar lavage fluid and plasma were measured to calculate the dilution of bronchoalveolar lavage fluid (Sigma Diagnostics, St Louis, Missouri).¹¹

The concentrations of interferon gamma, transforming growth factor (TGF) β₁, TGF-β₂, platelet-derived growth factor (PDGF)–AA, PDGF-AB, and PDGF-BB in bronchoalveolar lavage fluid were measured using high-sensitivity enzyme-linked immunosorbent assay kits in accord with the manufacturers' instructions (R&D Systems, Minneapolis, Minnesota; and Amersham Biosciences, Piscataway, New Jersey). The final alveolar epithelial lining fluid concentrations were determined by correcting for the bronchoalveolar lavage fluid dilution.

A global test for equality of 3 groups (patients having RA without lung disease [n = 43], patients having RA-ILD [n = 21], and patients having symptomatic RAPF confirmed by lung biopsy [n = 10]) was performed using the Jonckhere-Terpstra test for HRCT data and using analysis of variance for pulmonary function test data.¹² If the global test was statistically significant at α = .05 using the Wilcoxon rank sum test for HRCT data and using the t test for pulmonary function test data, 3 pairwise group comparisons were performed. Tests were 2-sided.

All subjects underwent baseline and follow-up pulmonary function tests and HRCT. The baseline bronchoalveolar lavage fluid concentrations were measured in 51 patients having RA (27 without lung disease, 16 with preclinical ILD, and 8 with pulmonary fibrosis). We compared HRCT score change, medical history variables, baseline pulmonary function test results, bronchoalveolar lavage fluid concentration measurements, and rate of change in pulmonary function test results (derived by an estimate of the slope of the linear regression model using each subject's percentage-predicted pulmonary function test result as the response variable and time of visit as the independent variable). Follow-up pulmonary function test results and HRCT scores were compared with those at baseline. Agreement of radiologists' readings was tested using a weighted κ statistic.

Radiographic findings of ILD were detected in 31 of 74 subjects having RA; there was statistically significant agreement among radiologists' independent HRCT readings (κ = 0.972, P < .001). Ten subjects reported symptoms of dyspnea or chronic cough for approximately 2.3 years and were previously diagnosed as having pulmonary fibrosis (ie, usual interstitial pneumonia) that was confirmed by lung biopsy approximately 1.6 years before enrollment. Of 64 patients having RA without dyspnea or cough, 21 (33%) had preclinical ILD identified by HRCT (Figure, A and B). A subject with preclinical ILD underwent thoracoscopic lung biopsy and was diagnosed as having nonspecific interstitial pneumonia (Figure, C). Therefore, asymptomatic preclinical ILD was prevalent among this cohort of subjects with RA who were referred to the National Institutes of Health.

Subject characteristics are summarized in Table 1. There were no statistically significant age differences between groups. The mean durations of articular disease in subjects having RA without lung disease, with preclinical ILD, and with pulmonary fibrosis were 11.3, 13.7, and 2.7 years, respectively. There was no statistically significant difference in the mean duration of articular disease between subjects having RA with vs without preclinical ILD. However, subjects having RAPF had a statistically significantly shorter mean duration of articular disease than subjects having RA-ILD (P =.001).

In this cohort of subjects with RA, no subjects were current smokers, and none had smoked within 2 years of entering this study. Our data demonstrated that 23% (10 of 43 subjects) without lung disease, 71% (15 of 21 subjects) with preclinical ILD, and 60% (6 of 10 subjects) with symptomatic pulmonary fibrosis were former smokers. The mean number of pack-years smoked was statistically significantly higher in subjects having RA with preclinical and symptomatic lung disease than in those without lung disease (P < .001).

Among physical examination findings, auscultation of crackles identified some subjects having RA-ILD. Although statistically significant differences in the percentages of subjects having audible crackles were found between groups, no statistically significant differences in joint deformities or synovitis were demonstrated.

The history of treatment using some clinically relevant medications is summarized in Table 1. In this cohort, a statistically significantly higher percentage of subjects having RAPF was treated using prednisone compared with having RA without lung disease (P =.009). In contrast, statistically significantly lower percentages of subjects having RAPF were treated using methotrexate compared with those having RA with and without preclinical ILD (P =.007 and P =.004, respectively). Furthermore, the mean duration of methotrexate use, which did not differ statistically significantly between subjects having RA with and without preclinical ILD, was statistically significantly shorter among subjects having RAPF than among those having RA with and without preclinical ILD (P =.002 and P < .001, respectively). No statistically significant differences were found among groups treated using leflunomide, hydroxychloroquine sulfate, tumor necrosis factor α inhibitors, or nonsteroidal anti-inflammatory drugs.

Baseline HRCT scores, pulmonary function test results, and arterial blood gas measurements were compared (Table 2). At the time of study entry, subjects having RA-ILD had statistically significantly higher HRCT scores than those having RA without lung disease and statistically significantly lower scores than those having RAPF (0.93, 0, and 2.7, respectively) (P < .001). Airflow and lung volumes (ie, the percentages of predicted FEV₁, FVC, and TLC) were normal in subjects having RA-ILD at baseline and were not statistically significantly different from those of subjects having RA without lung disease. However, the percentages of predicted FEV₁, FVC, TLC, and DLCO were statistically significantly lower among subjects having RAPF than among those having RA without lung disease (P =.002, P =.002, P < .001, and P < .001, respectively). Although DLCO was normal, the percentage of predicted DLCO was statistically significantly lower and the alveolar-arterial oxygen gradient was statistically significantly higher among subjects having RA-ILD than among subjects having RA without lung disease (P =.01 and P =.004, respectively). Therefore, impaired gas exchange is a feature of early, asymptomatic, preclinical ILD in subjects with RA.

Given the natural history of pulmonary fibrosis, we hypothesized that early ILD in subjects having RA would progress. Therefore, we quantified and compared serial HRCT findings and pulmonary function test measurements between groups (Table 3). The mean length of follow-up was at least 1.5 years in subject subcohorts and did not differ statistically significantly.

Abnormalities on HRCT progressed in 12 of 21 subjects (57%) with RA-ILD (Figure, B). Of 12 subjects with progressive ILD, 7 had preclinical ILD identified at baseline, and 5 had disease identified during longitudinal evaluation (data not shown). In addition, 6 of 10 subjects (60%) having RAPF had progressive lung disease detected by HRCT. Interval changes in HRCT scores were statistically significantly different among subjects having RA-ILD and among those having RAPF compared with subjects having RA without lung disease (P < .001 for all). However, there was no statistically significant difference in the interval changes in HRCT scores between subjects having RA-ILD vs RAPF. There were no statistically significant differences among subcohorts in annual rates of change in the percentages of predicted FEV₁, FVC, TLC, or DLCO measurements. Therefore, this study demonstrates that, in this cohort of patients with RA, preclinical ILD is prevalent and progressive, the natural history of disease resembles that of pulmonary fibrosis, and HRCT is more sensitive than pulmonary function test measurements in detecting early progression of preclinical ILD.

To improve our understanding of the pathogenesis of preclinical ILD in RA and to identify potential therapeutic targets, the concentrations of cytokines and growth factors were quantified in alveolar epithelial lining fluid procured by bronchoalveolar lavage. The concentrations of PDGF-AB and PDGF-BB were statistically significantly higher among subjects having RA-ILD than among those having RA with (P =.02 and P =.03, respectively) or without (P < .001 and P =.002, respectively) pulmonary fibrosis (Table 4). In addition, the concentrations of interferon gamma and TGF-β₂ were statistically significantly lower in subjects having RAPF than in those having RA with (P =.04 and P =.007, respectively) or without (P =.001 and P =.006, respectively) preclinical ILD. There were no statistically significant differences in the concentrations of interferon gamma or TGF-β₂ between subjects having RA-ILD and subjects having RA. No statistically significant differences in PDGF-AA or TGF-β₁ concentrations were detected between groups. Hence, the alveolar microenvironment of subjects having RA-ILD is presumably favorable for the development of fibrosis and differs from that of subjects having RAPF. In addition, PDGF is a potential therapeutic target in individuals with RA-ILD.

To identify potential risk factors for progressive preclinical ILD in subjects with RA, data from subjects without RA with stable vs progressive preclinical ILD were compared. No statistically significant differences in age, duration of joint disease, tobacco use history, or the presence of crackles, synovitis, or joint deformities were found between subjects having RA with stable vs subjects having progressive preclinical ILD (Table 5). Although there were no statistically significant differences between these 2 subgroups in the numbers of years of treatment using prednisone, leflunomide, methotrexate, hydroxychloroquine, tumor necrosis factor α inhibitors, or nonsteroidal anti-inflammatory drugs, a statistically significantly higher percentage of subjects having RA with progressive preclinical ILD was treated using methotrexate compared with subjects having RA with stable preclinical ILD (P =.046).

Analyses of baseline and longitudinal measurements demonstrated no statistically significant differences between these 2 subgroups in baseline HRCT scores, the mean duration of follow-up, or baseline or annual rates of change in pulmonary physiology values (Table 5). However, the annual rate of change in HRCT scores was statistically significantly lower among subjects having RA with progressive preclinical ILD compared with those having RA with stable lung disease (P < .001).

Analyses of alveolar epithelial lining fluid demonstrated differences in the concentrations of some proteins between subjects having RA with stable vs progressive preclinical ILD (Table 5). The concentrations of interferon gamma and TGF-β₁ were statistically significantly higher in subjects having RA with progressive preclinical ILD compared with those having RA with stable lung disease (P =.044 and P =.038, respectively). There was also a trend for higher concentrations of PDGF-BB in subjects having RA with progressive preclinical ILD compared with those having RA with stable lung disease (P =.06).

This study demonstrates that in this cohort of adult patients with RA, preclinical ILD is a prevalent progressive disease. Furthermore, pathogenic mechanisms and potential therapeutic targets seem to differ in preclinical and symptomatic ILD. Using HRCT to screen patients, we identified preclinical ILD in 21 of 64 patients (33%) having RA without symptoms of lung disease. The mean HRCT score of 0.93 indicates that patients having RA-ILD had minimal findings of ILD on HRCT at baseline. The mean annual rates of change in HRCT scores of −0.30 in patients having RA-ILD, −0.52 in a subset of patients having RA with progressive preclinical ILD, and −0.42 in patients having RAPF indicate that HRCT findings of ILD progress slightly and at similar rates in patients having RA-ILD and RAPF. Consistent with HRCT findings of ILD, gas exchange was impaired, as indicated by an increased alveolar-arterial oxygen gradient. A history of smoking seems to be a potential risk factor for the development of ILD, and treatment using methotrexate may be associated with progressive preclinical ILD in patients having RA. However, there was no statistically significant difference in the duration of treatment using methotrexate between patients with stable vs progressive preclinical ILD associated with RA. Physical examination and pulmonary function test measurements were not sensitive methods of detecting preclinical ILD, but auscultation of crackles may identify some patients with preclinical ILD. Notably, longitudinal evaluations demonstrated radiographic progression in 57% (12 of 21) of patients having RA-ILD.

The mean duration of articular disease in patients having RAPF was statistically significantly less than that of patients having RA-ILD. This result is interesting and may indicate that, in patients with RA, there is a bimodal distribution of ILD that occurs in close proximity to or several years following the onset of articular disease. These data may also indicate that different pathogenic mechanisms may contribute to the development and progression of these 2 types of ILD in RA.

The pathogenesis of ILD in patients with or without RA is unknown, which may contribute to limitations in treatment options. In this regard, it is relevant that our data demonstrated that the alveolar microenvironment in patients having RA-ILD is “profibrotic.” Specifically, we found that the concentrations PDGF-AB and PDGF-BB were statistically significantly higher in patients having RA-ILD than those in patients having RA with and without symptomatic ILD. Although it is possible that differences in the concentrations of PDGF between patients with preclinical ILD and those with pulmonary fibrosis may be a consequence of sampling areas with distorted lung architecture, these novel findings support the possibility that pathogenic processes in preclinical and symptomatic ILD differ. Our findings of statistically significantly different concentrations of interferon gamma and TGF-β₂ between patients having RAPF and those having RA with and without ILD further support this possibility. Therefore, it seems that PDGF-AB and PDGF-BB may contribute to the development of early asymptomatic ILD associated with RA and that low concentrations of interferon gamma may contribute to pulmonary fibrosis in RA.

Platelet-derived growth factor, a potent fibroblast mitogen, contributes to the development of pulmonary fibrosis. Studies^13-17 in animal models demonstrated that intratracheal administration or overexpression of PDGF in the lung leads to fibrosis and that treatment using inhibitors of PDGF bioactivity ameliorated fibrotic lung disease. In addition, high concentrations of PDGF in the bronchoalveolar lavage fluid among patients having Hermansky-Pudlak syndrome with early pulmonary fibrosis were reported.¹⁸ Although previous studies^19,20 have shown that PDGF is overexpressed in vitro by alveolar macrophages and in vivo by epithelial cells and macrophages in patients with idiopathic pulmonary fibrosis, there are no reports of high concentrations of PDGF in bronchoalveolar lavage fluid among patients with pulmonary fibrosis (to our knowledge). Together, these studies are consistent with our findings of PDGF in patients having RA-ILD or RAPF.

Our findings have potentially important therapeutic implications. For example, PDGF receptor α and PDGF receptor β, 2 cell surface receptors with intracellular tyrosine kinase domains, bind PDGF and are expressed on lung fibroblasts from patients with pulmonary fibrosis.^21-23 Therefore, drugs capable of modulating PDGF or inhibiting tyrosine kinase are candidate medications for the treatment of preclinical ILD in patients with RA. Our data also demonstrated that alveolar epithelial lining fluid concentrations of interferon gamma are low in patients with symptomatic RAPF. Although treatment using interferon gamma has not been shown to be effective in patients with idiopathic pulmonary fibrosis, our data suggest that interferon gamma could be of potential benefit to patients with RAPF. Furthermore, our analyses among subgroups of subjects having RA with stable vs progressive preclinical ILD demonstrated that alveolar concentrations of TGF-β₁ are high among those with progressive preclinical ILD. Therefore, medications capable of down-regulating the bioactivity of TGF-β₁ have potential therapeutic benefit for patients having RA with progressive preclinical ILD. Clinical trials are necessary to determine whether such medications are effective in the treatment of these lung diseases in patients with RA.

A potential limitation of this study is the possibility of referral bias. Subjects having RA with or without pulmonary fibrosis were recruited, and it is possible that patients with suspected lung disease were referred to this study. Therefore, the prevalence of preclinical ILD among this cohort may be overestimated. However, referral bias would not affect other data generated by this study, including our findings of progression of preclinical ILD and of differences in potential pathogenic mechanisms of preclinical ILD and pulmonary fibrosis in patients with RA.

In summary, the results of this study indicate that preclinical ILD in this cohort of adult patients with RA is prevalent and progressive. Based on these findings, smoking cessation should be recommended to patients with RA. In addition, HRCT may be indicated to identify early preclinical ILD in patients with RA whose physical examinations reveal crackles. Avoidance of methotrexate therapy should be considered for individuals with RA-ILD. Furthermore, our results suggest that potential pathogenic mechanisms and therapeutic targets in patients having RA-ILD vs RAPF are different and may warrant additional studies.

Correspondence: Bernadette R. Gochuico, MD, Pulmonary–Critical Care Medicine Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, 10 Center Dr, MSC 1590, Bethesda, MD 20892-1590 (gochuicb@mail.nih.gov).

Accepted for Publication: August 29, 2007.

Author Contributions: Dr Gochuico had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Gochuico and Rosas. Acquisition of data: Gochuico, Avila, Chow, Novero, Wu, MacDonald, Travis, and Rosas. Analysis and interpretation of data: Gochuico, Ren, Travis, Stylianou, and Rosas. Drafting of the manuscript: Gochuico, Chow, MacDonald, Stylianou, and Rosas. Critical revision of the manuscript for important intellectual content: Gochuico, Avila, Novero, Wu, Ren, Travis, Stylianou, and Rosas. Statistical analysis: Gochuico, Ren, and Stylianou. Obtained funding: Gochuico. Administrative, technical, and material support: Gochuico, Avila, Novero, Wu, MacDonald, and Rosas. Study supervision: Gochuico and Rosas.

Financial Disclosure: None reported.

Funding Support: This study was sponsored by the Division of Intramural Research, National Heart, Lung, and Blood Institute, National Institutes of Health.

Additional Contributions: Joel Moss, MD, PhD; Martha Vaughan, MD; and Vincent Manganiello, MD, PhD, provided insights and critical review of the manuscript.