Introduction

RA is a chronic systemic autoimmune disease, with a prevalence of 1% in the Black adult population [1]. It is characterized primarily by proliferating synovitis with systemic inflammation and increasing morbidity and mortality.

Among the extra-articular manifestations of rheumatoid arthritis (RA), interstitial lung disease (ILD) warrants special attention, given the fact that up to 10% of RA patients will eventually develop clinically significant ILD [2,3,4,5]. ILD was first described in 1948 when findings ranging from fine to widespread heavy reticulation were noted in chest radiographs (CXR) of patients of RA [6]. Recent studies have found that the prevalence of RA-ILD is increasing, while the incidence is stable over time, perhaps reflecting the prolonged survival among this population [7].

RA-ILD is associated with a higher mortality rate in comparison with RA without ILD likely due to the higher frequency of comorbid conditions including ischemic heart disease, congestive heart failure, and diabetes. Lack of definitive prognostic indicators or gold standard treatment therapies for RA-ILD is a likely explanation for the higher mortality rates observed among these patients [8, 9].

A large population study from Mayo Clinic published in 2010 assessed the incidence, prevalence, and mortality of ILD in RA. This study indicated that the disease typically affects middle-aged men with seropositive RA of about 9 years duration with long-term smoking history [10,11,12]. The population included in this study was largely Whites [13]. In contrast, a recently published report from December 2018 indicated the RA-ILD occurs mainly among female patients with common co-morbid conditions. Information on the racial classification of the study cohort was not reported since that study was based on claims data [7]. These two conflicting reports suggest racial and gender differences in the prevalence and associations of RA-ILD among different populations. Our study aimed at assessing the prevalence and clinical characteristics as well as co-morbidities associated with RA-ILD in a predominantly Black population with RA from two large urban academic medical centers.

Methods

This is a cross-sectional analysis of data obtained from records of RA patients utilizing the International Classification of Diseases Ninth Revision, Clinical Modification and the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-9 CM 714.0 and ICD-10 M06.00 thru M06.09). We identified RA as principal or secondary diagnosis including all inpatient discharges between January 1, 2010, to May 30, 2017, that took place at State University of New York (SUNY) Downstate Medical Center and New York City (NYC) Health + Hospitals/Kings County which serve the population of Central Brooklyn, NY.

Prior to the initiation of the study, the protocol was reviewed and approved by the SUNY Downstate Institutional Review Board Study [1080808] and the Office of Research Administration for implementation at NYC Health + Hospitals/Kings County [001252].

We identified the hospital discharges with RA and proceeded to select patients 18 years or older by Jan 1, 2010, to be included in the study. Data abstraction was performed for the selected cases, utilizing a uniform study data collection sheet. We collected demographics and clinical data including smoking history, year of diagnosis, comorbidities, laboratory values, hand imaging, and treatment regimens. Patients who received care at both hospitals were counted once, and data for the individual patient were merged for the purpose of this analysis.

Two investigators (IMM, SYZ) independently reviewed the cases identified by ICD codes to confirm RA diagnosis by the 2010 American College of Rheumatology criteria [14]. We used physician entries (inpatient/outpatient notes and consultations) and presence of disease-modifying anti-rheumatic drugs (DMARDs) in the medication reconciliation or DMARDs prescriptions.

RA patients with chronic respiratory symptoms of dyspnea and cough who had been evaluated for RA-related interstitial lung disease with pulmonary function tests (PFTs) by spirometry and/or plethysmography and chest computed tomography were further screened to exclude those who had the diagnosis of ILD prior to the establishment of RA diagnosis.

We also excluded those with drug-induced ILD, sarcoidosis, vasculitis, and pneumoconiosis.

A chest radiologist (SAW), a pulmonary specialist (AB), and a rheumatologist (IMM) reviewed the clinical course, management, pulmonary function tests, and chest computed tomography (CT) images for each patient during the multidisciplinary reviews. When UIP was suspected, it was categorized using the Fleischner Society idiopathic pulmonary fibrosis diagnostic criteria [15]. The CT features considered for a diagnosis of usual interstitial pneumonia (UIP) are honeycombing, reticular pattern, traction bronchiectasis, and the pattern of distribution (heterogenous and predominantly basal or sub-pleural).

Three diagnostic categories of UIP recognized by the Fleischner Society Criteria are:

  • The typical UIP CT pattern features honeycombing, reticular pattern with peripheral traction bronchiectasis or bronchiolectasis in a predominantly basal (occasionally diffuse) and sub-pleural distribution that is often heterogeneous.

  • Probable UIP CT pattern consists of a reticular pattern with peripheral traction bronchiectasis or bronchiolectasis in a predominantly basal and sub-pleural distribution that is often heterogeneous. Honeycombing is absent.

  • Indeterminate UIP CT pattern demonstrates evidence of fibrosis with some inconspicuous features suggestive of non-UIP pattern in variable or diffuse distribution [15].

Chest CT images demonstrating a homogenous and bilateral pattern of ground-glass opacities associated with fine reticulations, pulmonary volume loss, traction bronchiectasis, and sparing of the sub-pleural space were classified as non-specific interstitial pneumonia (NSIP) [16].

All available PFTs were reviewed during the multidisciplinary meetings. Forced vital capacity (FVC), forced expiratory volume in first second (FEV1), FEV1/FVC ratio, total lung capacity (TLC), and diffusing lung capacity for carbon monoxide (DLCO) values in the most recent PFTs were recorded. A reduction in TLC, FVC, and FEV1 along with a reduction in DLCO was used as supporting evidence suggestive of ILD [17].

Cases deemed during the multidisciplinary reviews to be non-interstitial lung disease were further categorized into airway disease, pleural disease, vascular, and miscellaneous [18].

A musculoskeletal radiologist (SK), who was blinded to the serology results for the cases, reviewed the most recent hand radiographs. Periarticular osteopenia, number of erosions, and joint space narrowing were evaluated utilizing the Simple Erosion Narrowing Score (SENS) [19].

Descriptive statistics using SPSS® version 23 was applied. We used measures of central tendencies and dispersion for continuous variables and frequency distribution for categorical variables. Data is presented as the mean ± standard error of the mean (± SEM). We used t test to compare between two groups for continuous variables and ANOVA for multiple comparisons. Chi-square analysis was used for categorical variables.

Results

RA diagnosis was ascertained in 503 subjects from a total number of 1142 records identified by ICD-9 and ICD-10 codes. Figure 1 illustrates the process of selection of the RA patients, with 529 (281 + 241) cases being excluded due to either insufficient data to confirm RA or alternative diagnosis found upon review. Women represented 88% of the study cohort with Blacks accounting for 88.5%, and Hispanics 9.2%. Body mass index (BMI) was mean age was 65 ± 0.67 for women vs. 61 ± 2.19 for men (p < 0.04 (Table 1)) 28.9 ± 0.36 (± SEM).

Table 1 Total study population characteristics
Full size table
Fig. 1
figure 1

Selection process for RA cases. ACOS asthma COPD Overlap Syndrome, COPD chronic obstructive pulmonary disease, DMARDs disease-modifying anti-rheumatic drugs, PFTs pulmonary function tests RB-ILD respiratory bronchiolitis interstitial lung disease, NSIP non-specific interstitial pneumonia, UIP usual interstitial pneumonia

Full size image

In our cohort, we found a high prevalence of traditional cardiovascular risk factors including hypertension (66.6%), dyslipidemia (41.3%), diabetes (28.5%), and smoking (29.4%) (Table 1) as we previously reported [20]. In addition, among the cardiovascular outcomes examined, myocardial infarction (MI) or known coronary artery disease (CAD) was reported in 19.8%, congestive heart failure in 14.8%. Stroke or transient ischemic attack was reported in 10.1%. Gastroesophageal reflux disease (GERD) was encountered in 20.7%, asthma in 16.5%, chronic obstructive pulmonary disease in 12%, and venous thromboembolic disease in 12.9% of the entire cohort (Table 1).

Of the 503 RA patients, eighty-six had chronic respiratory symptoms of cough and dyspnea that developed after the RA diagnosis and had clinical note entries of suspected RA-related lung disease; pulmonary referral or work up ordered for evaluation of the chronic respiratory symptoms. The multidisciplinary team of this study reviewed the clinical history including date of RA diagnosis, disease course, comorbidities, serologies, therapeutic interventions, chest CT, and serial PFTs (Fig. 1).

Thirty-two cases had PFTs and CT features compatible with ILD [2, 13, 21,22,23]. The categories of ILD were as follows: typical usual interstitial pneumonia (UIP), 16 cases; probable UIP, 6 cases; and non-specific interstitial pneumonia (NSIP), 8 cases; five cases had lung biopsies confirming ILD diagnosis (one case was respiratory bronchiolitis-ILD). None of the cases was an indeterminate pattern of UIP.

Therefore, the prevalence of ILD in our RA population was (32/503) 6.36% (Table 2).

Table 2 RA cases assessed by the multidisciplinary investigators*
Full size table

During the multidisciplinary meetings, the investigators confirmed that the patients under review did not have exposure to asbestos, previous administration of bleomycin, amiodarone, history of hypersensitivity pneumonitis, sarcoidosis, or vasculitis. The CT features more frequently found among the RA-ILD cases were architectural distortion, traction bronchiectasis, low lung volume, peripheral and lower lobe predilection, reticular pattern honeycombing, and ground-glass opacities (Table 3). GERD was more frequent among UIP cases, compared with NSIP population (39% vs. 12.5%) (Figs. 2, 3, and 4).

Table 3 Chest Computed Tomography Findings
Full size table
Fig. 2
figure 2

Typical usual interstitial pneumonia. CT image demonstrating peripheral and basilar severe honeycombing, with architectural distortion and traction bronchiectasis

Full size image
Fig. 3
figure 3

Probable usual interstitial pneumonia (a, b). CT demonstrates basal, sub-pleural, and peripheral distribution of reticulation, with peripheral traction bronchiectasis, distorted airways, and relatively minimal honeycombing

Full size image
Fig. 4
figure 4

Non-specific interstitial pneumonia (NSIP). CT images demonstrate predominantly peripheral ground-glass opacities with no evidence of fibrosis. Subpleural sparing in the lingula is appreciated. Findings consistent with cellular NSIP

Full size image

Two of the eight RA-NSIP cases presented within the first year of RA diagnosis; five of the 32 RA-ILD patients had RA for 7–11 years, and nine patients for more than 20 years.

Nine of the 32 (37.5%) UIP patients were seropositive for either rheumatoid factor (RF) or anti-citrullinated peptide antibodies (ACPA) and 33.3% had positive anti-nuclear antibodies. A second autoimmune disease was found in 31.2% (10/32) of the RA-ILD cases and in 50% of the NSIP patients (Table 4).

Table 4 Comparison between the two ILD groups
Full size table

PFTs had been obtained in 46.5% (40/86) of the patients who were reviewed by the multidisciplinary investigators with 40.62% (13/32) of the patients in the RA-ILD group had PFTs recorded. The values for FVC, TLC, and DLCO were lower in RA-ILD vs. RA with other lung disease groups; however, only the FEV1/FVC ratio between the groups reached statistical significance (Table 5).

Table 5 Comparison of PFTs results among the RA lung disease groups
Full size table

Cases not meeting RA-ILD inclusion criteria by CT features were classified as follows: airway disease which included chronic obstructive pulmonary disease (COPD), 19 cases; restrictive lung disease, 6 cases; asthma, 3 cases; chronic atelectasis with scarring, 3 cases, with 2 cases reported for each of the following categories: asthma-COPD overlap syndrome, obliterative bronchiolitis, and bronchiectasis. Pleural disease group includes two cases of pleural plaques and one case of pleural effusion; vascular pathology, 4 cases of pulmonary hypertension (PAH); and mixed pathologies, ten other cases including pneumonia (2 cases), COPD-adenocarcinoma, infection-metastasis, PAH-bronchiectasis, PAH-chronic venous thromboembolism, scarring-bronchiectasis, fibrosis/cystic air disease, undetermined pathology, and a normal chest.

In Table 6, we present a comparison of the cohorts of RA-related ILD, other RA lung disease, and the RA only population. GERD was encountered at a rate of 32.3% and 34% among the patients with ILD and other lung diseases respectively versus 18% for RA only patients. Sepsis occurred at a higher rate among patients with other lung diseases. Cancer rate was 11.9% for the entire cohort of RA patients with 12.5% being breast cancer among RA-ILD population. Among those with other RA lung disease, the cancer rate was 9.25%. There were no statistically significant differences among the groups in regard to other clinical parameters, inflammatory markers, presence of RF or ACPA positivity, or SENS scores.

Table 6 Demographic and clinical characteristics of the study populations*
Full size table

We recorded glucocorticoid use in 68% of the RA patients with other lung diseases versus 50% among patients with RA-related ILD. Methotrexate, biologics, and the combination of steroids, disease-modifying anti-rheumatic drugs (DMARDs), and biologics use were the lowest among RA-ILD (16%, 12%, and 21.9% respectively). Only one patient in the RA-ILD category had received rituximab. DMARDS and biologics were used with low frequency among RA patients with other lung diseases (see Table 7).

Table 7 Therapeutic regimens
Full size table

Seventeen of the 32 patients with RA-ILD continued to receive care at our institutions by the time this cross-sectional analysis was completed. Four patients expired, three belonging to the UIP category with death occurring at 5 years and 19 years after ILD diagnosis. The NSIP patient expired within the first year of RA-ILD diagnosis. The other 13 patients were alive as per clinic/progress notes.

Given the current interest in COPD, asthma, and RA, we examined further our RA patients with a clinical diagnosis of COPD. Smoking rates were as follows: 55.5% were former smokers while only 22.2% were current smokers among our COPD patients. Seropositivity (either RF or ACPA) was present in 38.6% of the COPD patients with 28% of them being double seropositive.

For asthma, 12% of the patients were reported as former smokers and 25% as current smokers. Seropositive RA (either RF or ACPA) was recorded in 89.7% and double seropositivity in 47.2%. The SENS for the hand radiographs (maximum = 62), which adds the total number erosions (up to 32) and total number of joint space narrowing (up to 30), for the COPD group had a mean score of 32.71 ± 6.5 (± SEM), and for asthma, the SENS was 25.42 ± 7.4 (± SEM).

Discussion

In our predominantly Black RA population, ILD was found with a prevalence of 6.36%, which is consistent with previously published reports; given that the patterns of ILD in connective tissue disease resemble those seen in idiopathic pulmonary fibrosis, we were able to apply the Fleischner Society Criteria for UIP [7, 13, 15]. Consistent with the current literature, the UIP was the most CT pattern encountered. Our RA-ILD population consisted mainly of females (93.75%) with a mean age of 62.65 years. This is contrary to previous reports indicating middle-aged men as the predominant population affected by ILD [13]. Smoking history reported among our population was 31%, less than half of that reported in the Mayo Clinic study (64%) [13].

The difference in the rate of smoking in addition to the fact that males are heavier smokers (higher serum cotinine levels among male smokers) compared with female smokers could explain the gender discrepancy encountered between our RA-ILD population predominantly composed of women compared with the RA-ILD population from Mayo Clinic [24].

Half of our RA-ILD patients had double seropositivity [7, 22]. The length of RA disease was more than 10 years in about 59% of the RA-ILD cases, finding consistent with the current knowledge that the risk of RA-ILD increases over time [13].

The fact that only 35% (6/17) of the RA-ILD cases had RA for more than 20 years of disease could be explained by the high mortality rate, in which only a smaller number of ILD cases live past 20 years of disease. NSIP developing within the first year of RA diagnosis was found in a quarter of the patients; it has been described that cases with NSIP pattern on chest CT will be read as UIP pattern on lung biopsy which is known to be linked to increased mortality [25].

Our RA-ILD population had a higher prevalence of a second auto-immune disease (31.25%), in contrast with recent reports on the co-existence of RA and other autoimmune diseases in 23.4% of the patients which appears to be an important characteristic among our patients [26, 27].

Four cases of breast cancer were encountered (12.5%) among the UIP RA-ILD patients with no lung cancer being detected in this group.

PFTs were available in about 40.6% (13/32) of the RA-ILD cases, revealing parameters consistent with loss of pulmonary elasticity with restrictive lung deficit and low diffusion capacity.

These PFT results reflect the common pattern of physiological abnormalities seen in interstitial lung diseases characterized by a restrictive ventilatory defect and reduced diffusing capacity (DLCO). The statistically significant FEV1/FVC ratio, found when comparing RA-ILD vs. RA with other lung diseases, also underlines that the ILD process favors more FVC reduction than reduction in FEV1. Additionally in ILD, DLCO is typically reduced to a greater extent than can be expected by the reduction in lung volumes only [28].

The most commonly seen chest CT features, in descending order, were architectural distortion, traction bronchiectasis, low lung volume, and the typical predilection for the lung periphery and the lower lobes. Reticular pattern, honeycombing, and ground-glass opacities were also found in two-thirds to about half of the patients. Our RA-ILD patients had a mean SENS slightly higher than the rest of the population, likely representing aggressive disease.

The 54 cases that underwent multidisciplinary review and deemed to be non RA-ILD cases were studied further and classified into categories of RA with other lung diseases [18]. Sixty-eight percent of the patients had airway disease. COPD was the most common, followed by restrictive lung disease. Our data suggests higher airway disease compared with previous reports where airway disease prevalence was between 31 and 60% of the RA populations [18, 29, 30]. Pulmonary hypertension was found in 7.4% of the RA with other lung disease group.

The eleven cases of RA with other lung diseases in the miscellaneous category contain pathology ranging from COPD and lung cancer, infection, pulmonary hypertension with bronchiectasis, fibrosis/cystic changes to undetermined pathology which would explain the chronic respiratory symptoms of the patients in this group.

The reassignment of cases that were once considered “RA-related lung disease,” by clinician’s notes highlights the importance of pursuing a multidisciplinary approach in the evaluation of suspected cases. A multidisciplinary review is already a proven strategy in the diagnosis of ILD that leads to early diagnosis and thorough discussion of treatment plans and monitoring plans that take into account the patient’s other comorbid conditions [31].

When we consider our entire RA population, the clinical diagnosis of COPD was found at a rate higher than previously reported (12% vs. 8.1%) [32]. It has been postulated that RA is associated with an increased risk of COPD with a hazard ratio of 1.52 as per the results of a matched prospective cohort within the Nurses’ Health Study. Smoking did not explain fully the association between COPD and RA. Other risk factors such as seropositivity were thought to be implicated in the pathogenesis of COPD associated with RA [32].

However, in our RA-COPD population, we found high rates of smoking history (46/60) (77%) with only 38.6% of the patients being seropositive for either RF or ACPA, making smoking an attractive common soil hypothesis for the comorbidity of COPD and RA in our predominately Black female population.

Another important finding in our study is the nearly double rate of asthma among our population compared with previously published reports (12% vs. 6.6%) [33, 34].

A possible explanation to the higher rate of asthma among our patients is the relatively higher smoking rate compared with that of the entire RA population (37.3% vs. 29.54%) for RA-asthma and RA-only group respectively. 89.7% of the RA asthma patients were seropositive with almost half of them being double seropositive highlighting the possible common immunologic mechanisms for these 2 autoimmune disorders [35, 36]. However, the number of erosions and joint space narrowing as calculated by SENS did not point to a more aggressive disease among asthmatics when compared with the entire RA cohort (25.42 ± 7.4 vs. 35.53 ± 2.05).

The prevalence rate of cancers encountered in our RA population was 11.9%, which falls at a higher rate than previously reported of 6.3% and most recently 4.16% when non melanoma skin cancer was excluded [37, 38].

Interestingly, the malignancies found among the 4 UIP patients were breast cancer that is the least reported in published series. Another interesting finding is the absence of lung cancer in our populations, which is the most frequently reported cancer among ILD patients. The reasons for this discrepancy are not clear, except that our population is predominately women where the incidence of breast cancer is much higher than that reported in men.

Further studies are needed to understand the higher overall risk of malignancy among our RA population. Possible risk factors include higher rates of obesity, smoking, advanced age, and use of steroids and immunosuppressive drugs, as well as high levels of inflammatory markers among our patients.

In regard to the therapeutic patterns, the lower use of MTX and biologics found among the RA-ILD cases is expected given their possible association with ILD and shorten survival reported with the use of tumor necrosis factor inhibitors (TNFi), before Rituximab began to be more frequently used.

Finally, our study is limited by the retrospective nature of the analysis, lack of available RA-specific disease activity measurements and survival outcomes for all of the cases. Alive status was only ascertained in 17 cases of the 32 cases with RA-ILD. Inaccuracy in coding explains the number of cases in which RA diagnosis was not found in the clinical documentation. We plan a prospective study to ascertain the survival predictors among our RA population with a special focus on RA-related ILD.

Most of the examinations on our study were routine supine chest CTs limiting our ability to diagnose subtle or early ILD. Dependent densities are common in the posterior subpleural aspects of the lungs on CT scans. This finding is usually attributable to micro-atelectasis; however, it can represent subtle early interstitial lung disease, without examining prone imaging routinely performed on high-resolution computed tomography.

Although our study populations had PFT available for 40.62% of the RA-ILD and 50% for the non RA-ILD group, having PFT’s for all cases would have been ideal. PFTs support the diagnosis of ILD although the pattern is non-specific. In addition, in 15% (6/40) of our study cases, baseline PFTs were available. Serial PFTs provide valuable information determining the disease progression and response to therapy or lack of [17]. Having an exercise-induced gas exchange difference would have further enhanced the diagnostic and monitoring value of the PFTs; however, 6-min walk tests were not available in the cases with low DLCO. There is also data missing at random, which we believe is harmless and does not represent a systematic bias.

Conclusion

This is the first study to evaluate lung disease in our predominantly Black RA population assessed by group multidisciplinary experts in rheumatology, pulmonary, and chest radiology. In contrast to previous report from Mayo Clinic indicating that the RA-ILD is typically a disease of White men (58.7%) with high rate of smoking (64%), our study suggests that RA-ILD among Blacks is predominantly a disease of females (93.7%) with high rates of RA seropositivity, GERD, and CVD risk factors. In support of our results, a recent report derived from the US claims data indicates that RA-ILD is predominant among female RA patients (70%).

Smoking history rate among our population was 31%, which is less than half of that reported in the Mayo Clinic study (64%). This together with the fact that men tend to smoke much heavier than women provide at least a partial explanation for the gender differences reported in our study, compared with the Mayo Clinic cohort. Nevertheless, further studies are necessary to identify the pathogenetic mechanisms accounting for the gender differences in the distribution of RA-ILD among Black and White populations.

Finally, among our cohort of RA patients with other lung disease, airway diseases such as COPD, restrictive lung disease, and asthma were the most frequently reported diagnoses. These findings are consistent with the established literature in this area.