Rheumatoid Arthritis-Associated Interstitial Lung Disease: The Relevance of Histopathologic and Radiographic Pattern FREE TO VIEW

Rheumatoid arthritis (RA) is a systemic inflammatory disease that affects approximately 1% of the population,¹ and pulmonary involvement is common. Interstitial lung disease (ILD) is the primary pulmonary manifestation of RA, as it is for other connective tissue diseases (CTDs), such as scleroderma (systemic sclerosis), inflammatory myositis (polymyositis and dermatomyositis), Sjögren syndrome, and undifferentiated CTD.^2–7

RA-associated ILD (RA-ILD) is a source of substantial morbidity and mortality for affected patients. The histopathologic and radiographic appearance of RA-ILD is heterogeneous and primarily mimics the following two patterns seen in the idiopathic interstitial pneumonias (IIPs): usual interstitial pneumonia (UIP) and nonspecific interstitial pneumonia (NSIP) [Fig 1].

Distinguishing UIP from NSIP has important prognostic implications. Idiopathic UIP (ie, idiopathic pulmonary fibrosis [IPF]) has a uniformly poor prognosis and lacks an effective medical therapy; idiopathic NSIP has a better prognosis and often responds to anti-inflammatory therapy. Moreover, IPF, which is characterized by abnormal wound healing and abnormal epithelial cell function, is thought to have a distinct pathobiology from NSIP. Because of this, the diagnostic workup of patients with IIPs emphasizes the histopathologic and radiographic identification of UIP. In many cases, radiographic imaging is sufficient,⁸ and surgical lung biopsy is required only for those patients in whom radiographic imaging is nondiagnostic.⁹

In contrast to the IIPs, histopathologic and radiographic appearance is rarely emphasized in the assessment of patients with RA-ILD. There are two primary explanations for this approach. First, although it is generally recognized that RA-ILD has a higher proportion of UIP compared with other forms of CTD-associated ILD (CTD-ILD), the prognostic value of this pattern is unclear.^10,11 Second, the treatment of RA-ILD with anti-inflammatory agents, such as prednisone and immunomodulators (eg, azathioprine and cyclophosphamide), is generally recommended regardless of the underlying histopathologic pattern.

This review assesses the validity of these assumptions in RA-ILD patients by reviewing the data supporting or refuting each one. In particular, it focuses on the role that histopathologic and radiographic appearances play in determining the prognosis and management of RA-ILD. Further, it proposes a diagnostic approach to RA-ILD based on the available evidence. Finally, it identifies gaps in knowledge and suggests ways to address these gaps through future research.

The reported prevalence of ILD in patients with RA is highly variable and depends on the methods of detection (eg, high-resolution CT [HRCT] scan, chest radiograph, or pulmonary function testing) and the population selected for study (eg, symptomatic or asymptomatic, autopsy series). Studies^12–19 have reported a prevalence as low as 4% and as high as 68%. The majority of cases of RA-ILD occur in patients between the ages of 50 and 60 years. While several studies^{13,15–17,20} have reported smoking, male gender, and longstanding RA to be risk factors for the development of ILD, other reports^{13,14,16,17,19} have not shown any significant associations. In addition, genetic associations (human leukocyte antigen B40 and α₁-antitrypsin) have been described.^21,22

Though RA is often diagnosed before the detection of ILD due to the presence of articular disease, patients may present de novo with isolated pulmonary disease; in these cases, a high index of suspicion for RA-ILD is required to distinguish it from the IIPs. Similar to the IIPs, patients with RA and ILD most often present with chronic symptoms of dyspnea and cough. A physical examination may reveal inspiratory crackles, and pulmonary function test (PFT) results demonstrate restrictive physiology often with a reduced diffusing capacity. HRCT scanning is generally sufficient to confirm the diagnosis of ILD, although in a minority of cases, surgical lung biopsy may be required.

Although a broad range of histopathologic patterns has been associated with many of the CTD-ILDs, RA-ILD appears to be one of the most diverse (Fig 2). The predominant histopathologic pattern in patients with scleroderma is NSIP.¹⁰ Limited studies in patients with other forms of CTD-ILD (eg, polymyositis, Sjögren syndrome, and undifferentiated CTD) have also shown a predominance of NSIP pattern on surgical lung biopsy specimens.^5,23–25 In contrast, the currently available data show that, among RA-ILD patients, there is a higher proportion of patients with UIP pattern²⁶ when compared with patients with other CTDs. Lee et al²⁶ found UIP to be the most common histopathologic pattern in RA-ILD patients (56%). This was followed by NSIP (33%) and organizing pneumonia (11%).

The following four major radiographic patterns have been identified in patients with RA-ILD: a UIP-like pattern with bilateral subpleural reticulation with or without honeycombing; NSIP-like pattern with predominant ground-glass opacities; an inflammatory airway disease pattern with centrilobular branching lines with or without bronchial dilatation; and an organizing pneumonia-like pattern with patchy areas of consolidation.²⁷

There are data demonstrating excellent correlation between the histopathologic pattern found on surgical lung biopsy and the radiographic pattern found on HRCT scans in patients with IPF.^8,28 The correlation between histopathologic and radiographic patterns in RA-ILD patients has not been as rigorously studied, but it appears that the same correlation is present (Table 1). Akira et al²⁹ published an article showing HRCT scan findings for 29 patients with RA-ILD. Although these investigators did not specifically use the terms “UIP” or “NSIP” to describe the radiographic appearance, 19 patients had reticulation with and without honeycombing; these appearances likely represent UIP pattern. Lung biopsy specimens were available in 14 of these 19 patients, and all patients showed UIP pattern. Tanaka et al²⁷ demonstrated that 41% of a retrospective RA-ILD cohort had UIP pattern seen on HRCT scans, with 30% showing a fibrotic NSIP pattern. Of the 26 patients with UIP pattern seen on HRCT scan, only 4 patients underwent lung biopsies; 2 patients had UIP, and 2 patients had fibrotic NSIP. Of the eight patients with NSIP seen on HRCT scans who underwent lung biopsy, seven patients had NSIP on histopathology.

In 2005, Lee et al²⁶ examined data from 18 patients with RA-ILD who had undergone surgical lung biopsy. Ten patients had UIP on histopathology, and 6 patients had fibrotic NSIP. Among the patients with UIP pattern seen on the lung biopsy specimen, nine patients had HRCT scans with reticulation and honeycombing, and one patient demonstrated reticulation and ground-glass opacities. Of the patients with NSIP pattern seen on a lung biopsy specimen, three patients had reticulation and ground-glass opacities, and the other three patients had predominant ground-glass opacity seen on HRCT scan.

These data, although limited, suggest that HRCT scanning is reasonably specific for the presence of UIP pattern seen on surgical lung biopsy specimens in patients with RA-ILD. These findings need to be confirmed in larger studies.

Few studies have assessed whether histopathologic pattern suggests a distinct pathophysiology in RA-ILD patients. Gochuico et al¹⁷ compared patients with RA who had HRCT scan findings consistent with NSIP pattern (n = 21) with those who had findings consistent with UIP pattern (n = 10). They showed that patients with RA and NSIP pattern on HRCT scans had higher concentrations of platelet-derived growth factors AB and BB, interferon-γ, and transforming growth factor β₂ in BAL fluid compared with those patients with RA and UIP pattern. Patients with NSIP also had a longer duration of articular symptoms. Based on these findings, the authors suggest that the pathophysiology of NSIP and UIP pattern in patients with RA-ILD may be different. Biederer et al³⁰ reported that patients with RA-ILD with a predominantly ground-glass pattern seen on HRCT scans had a shorter mean (± SD) duration of RA than those patients with a reticular pattern (63 ± 38 months vs 133 ± 112 months, respectively; p < 0.01). Although the authors did not specifically distinguish between UIP and NSIP patterns, the ground-glass pattern described most likely represented NSIP.

Serologic and lymphocyte profiling have shown no clear evidence of distinction between patients with RA-ILD and UIP or NSIP pattern. Although one study³¹ did show a trend toward lower anticyclic citrullinated peptide antibody levels in patients with RA-ILD and UIP pattern, there have been no convincing data reported that citrullination is significantly associated with either pattern.³² Studies^33–35 looking at CD20+ B lymphocytes, CD4+ T lymphocytes, and mast cells in lung biopsy specimens from patients with RA-ILD did not show significant differences between UIP and NSIP patterns. Although several studies^36–38 have established the importance of genetic susceptibility in the development of RA, there are no data describing its association with UIP pattern.

The radiologic and histopathologic patterns may be influenced by cigarette smoking. Biederer et al³⁰ reported a trend toward a higher number of pack-years of smoking in those with pure reticular opacity vs those with predominant ground-glass opacity on HRCT scans, suggesting that smoking could be associated with UIP over NSIP. In addition, Lee et al²⁶ reported that all the patients with RA-ILD and histopathologic UIP pattern were smokers, while those with NSIP pattern were nonsmokers.

Thus, studies published to date have been limited and discordant. Until more data become available, the relevance of histopathologic pattern to pathogenesis in RA-ILD patients will remain unknown.

RA-ILD associated with UIP pattern appears to have an increased risk of disease progression and death (Table 2). A prospective study³⁹ of patients with RA-ILD followed up for 24 months demonstrated that two of the three patients with a predominant ground-glass pattern on HRCT scans had spontaneous improvement compared with none in the group with a reticular pattern. While this study did not characterize patients as having either UIP or NSIP pattern on HRCT scans, a predominant ground-glass pattern is more likely to be associated with NSIP. A study⁴⁰ of 28 patients with RA-ILD found a strong trend toward worse survival in patients with UIP compared with patients with NSIP (hazard ratio, 18.952; p = 0.08).

Acute worsening of disease has been reported in RA-ILD patients,⁴¹ and appears to occur primarily in those patients with UIP pattern.^29,39 In a study of a large cohort of 93 subjects with CTD-ILD, Park et al⁴² reported a 1-year incidence of acute worsening of 3.3% (four subjects). Three of these four patients had RA with proven UIP seen on a lung biopsy specimen.

There are limited data regarding the response to therapy based on histopathologic pattern in RA-ILD patients. In one recent narrative review,⁴³ the authors concluded that, based on their own experiences, patients with RA-ILD and UIP pattern are less likely to respond to therapy with corticosteroids than patients with RA-ILD and NSIP or organizing pneumonia pattern. It remains unclear, however, if the therapeutic implications of UIP pattern in patients with RA-ILD are similar to those of UIP pattern in patients with IPF.

There are limited data directly comparing patients with RA-ILD and UIP pattern with patients with IPF. The lung histopathology of UIP in patients with CTD has been shown⁴⁴ to have fewer fibroblast foci when compared with patients with IPF. In addition, the number of CD4+ lymphocytes appears to be increased in lung specimens from patients with RA-ILD and UIP pattern vs patients with IPF.³⁵ As a group, patients with RA-ILD are thought to have a better prognosis than those with IPF. In a prospective double cohort study of 18 patients with RA-ILD and 18 patients with IPF, Rajasekaran et al⁴⁵ reported an improved survival time in patients with RA-ILD (median survival rate: 60 months vs 27 months, respectively). This study, however, did not distinguish between UIP and NSIP patterns in its RA-ILD cohort.

The most relevant data come from Park et al⁴⁰ who compared survival times in 28 patients with RA-ILD with those in 203 patients with IPF (Fig 3). On univariate analysis, survival was worse in the group of patients with RA-ILD and UIP pattern (n = 10) when compared with the population of patients with CTD and NSIP (p = 0.04). Although this comparison lost statistical significance with multivariate adjustment (p = 0.07), the small size of the population of patients with RA-ILD and UIP pattern makes such an analysis underpowered. There was no significant difference in survival time between the patients with RA-ILD and UIP pattern and the patients with IPF.

Based on the evidence available to date, UIP pattern in RA-ILD patients appears to be common, identifiable by HRCT scanning, and predictive of poor survival. With this in mind, we propose the following approach to patients with RA-ILD (Fig 4). Given the possibility of subclinical disease, the clinician should have a low threshold for evaluating patients with RA for ILD. Although we do not recommend the routine screening of patients with RA with HRCT scans and PFTs, a future study of the cost-effectiveness of this approach is warranted. Each of these recommendations should be reviewed as additional data become available.

Some may continue to question whether the existing data are sufficient to make such recommendations. It can also be argued that if we are going to treat these patients with the same agents regardless of histopathologic and radiographic pattern, does prognostic information alone warrant the risk of surgical biopsy? These are legitimate questions that individual clinicians and patients must answer for themselves. We certainly agree that more information is needed, and this will come only from well-designed clinical studies.

Studying RA-ILD is challenging given the small numbers of affected patients. We believe that to adequately study this condition, a large, prospective, multicenter, longitudinal cohort of carefully described patients with RA-ILD must be developed. With such a collaborative effort, we could begin to definitively answer the following questions:

We believe the existing data from studies of RA-ILD suggest that UIP pattern is common and relevant to the management of these patients. Future research needs to focus closely on the questions raised above. Answering them will improve our understanding of RA-ILD, better inform our diagnostic approach, and provide insight into how to better the lives of these patients.

Financial/nonfinancial disclosures: The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Other contributions: We thank Kirk Jones, MD, for providing the histopathologic images in Figure 1.