Clinical Characteristics and Predictors of the Recurrence of Organising Pneumonia Associated with Rheumatoid Arthritis

Background: Organizing pneumonia (OP) accounts for ~10% of the interstitial lung diseases in patients diagnosed with rheumatoid arthritis (RA). There are only a few published studies that have described RA-associated OP (RA–OP), each with comparatively few study participants. Furthermore, despite the high frequency of secondary �areups, no studies to date have identi�ed factors that predict RA–OP recurrence. This study aimed to clarify the clinical characteristics of RA–OP, to determine the relationship between OP and RA exacerbations, and to identify predictors of RA–OP disease recurrence. Methods: The data of 33 patients with RA–OP admitted to our hospital between 2006 and 2016 were retrospectively analysed. Results: RA onset preceded OP in 82% of patients; OP preceded or co-occurred with RA in 9% each. The median [Q1, Q3] age at �rst OP onset was 64.0 [55.0, 68.0] years and 5.5 [1.0, 19.3] years after RA onset. At OP onset, 42% of events exhibited unilateral shadows and 76% had normal KL-6; RA disease control remained good in 52% and was exacerbate d in only 18%. Ten (30%) patients experienced OP recurrence with an interval of 13.0 [7.5, 22.5] months between events, with a rate of �rst recurrence of 127 per 1,000 person-years. Compared with 14 non-recurrent cases, 10 recurrent cases showed lower age at �rst OP (59.5 years vs. 67.1 years; p = 0.043) and shorter period from RA to �rst OP (6.4 vs. 14.2 years; p = 0.047) and included more OP-preceding patients (30% vs. 0%; p = 0.029) and more ever-smokers (80% vs. 36%; p = 0.032). OP-preceding patients showed shorter median recurrence-free survival time (15 vs. 136 months; p = 0.009) and higher recurrence risk (hazard ratio, 5.5; p = 0.021). Conclusions: RA–

Conclusions: RA-OP showed a high recurrence rate but no association with RA exacerbation. Four predictors of RA-OP recurrence were identi ed.

Background
Rheumatoid arthritis (RA) is a systemic autoimmune disease with extra-articular involvement, with the lung being one of the most frequently affected organs. Several types of interstitial lung diseases (ILDs), reported in up to 58% of patients with RA [1][2][3][4], generally develop several years after the onset of RA and occasionally develop before or concomitantly with RA, signi cantly affecting the life prognoses of these patients and restricting therapeutic options for their arthritis [1-3, 5, 6]. The two main histopathological patterns of ILDs in RA are usual interstitial pneumonia and nonspeci c interstitial pneumonia [7,8]; however, organising pneumonia (OP) is also a substantial subtype, accounting for approximately 10% of ILDs in patients with RA [9][10][11][12][13]. In Japan, the estimated prevalence of OP in patients with RA (RA-OP) has been reported to be 1.9-4.8% [14].
OP, histologically characterised by intraluminal buds of granulation tissue within alveolar ducts and alveoli [15], starts usually with u-like symptoms, such as fever, cough, fatigue, anorexia, and progressively mild dyspnoea [16]. Its radiographic manifestations are typically distinctive with bilateral, patchy or diffuse consolidation with or without ground-glass opacities [17]. OP is classi ed as cryptogenic or secondary, the underlying causes of which include infections, drugs, radiation, and connective tissue diseases (CTDs), such as RA.
Most patients with OP recover fully either after glucocorticoid therapy or spontaneously; therefore, the prognosis of OP has been considered to be better than that of other types of ILDs [12,18]. However, several studies have reported that the prognosis of OP associated with CTDs (CTD-OP), including RA-OP, is poorer than that of cryptogenic OP [19][20][21] and that while OP sometimes recurs after reduced glucocorticoid dosage, its recurrence rate is higher in CTD-OP than in cryptogenic OP [20,21], suggesting the particularity of CTD-OP and indicating the importance of further research focussing speci cally on CTD-OP.
Only a few studies have reported RA-OP's clinical characteristics [14,22], and further studies are required to understand the relationship between OP development and RA exacerbation. In addition, these studies did not exclude drug-induced and post-infectious OP in patients with RA, which may mask the true characteristics of RA-OP; therefore, they emphasise the need for studies focussing exclusively on RA-OP.
This retrospective cohort study aimed to investigate the clinical characteristics of RA-OP patients and RA-OP events, to examine the disease activity changes of RA at the development of OP, and speci cally to clarify which characteristics at the rst OP onset are predictors of the later OP recurrence.

Study population
Using either paper or electronic medical records, we retrospectively reviewed data of patients who were admitted to the Division of Rheumatic Diseases, National Center for Global Health and Medicine, Japan, between August 2006 and August 2016. RA-OP patients who ful lled the following inclusion criteria were enrolled in this study: (1) diagnosis of RA according to the 1987 American College of Rheumatology (ACR) criteria [23] or the 2010 ACR/European League Against Rheumatism criteria [24]; (2) diagnosis of OP based on the criteria described in the section, Diagnosis of OP; and (3) clinical courses where the patient's OP could not be explained by any underlying causes other than RA itself, such as causative infectious agents or drug usage. In total, 33 RA-OP patients were included in the study.

Diagnosis of OP
The diagnosis of OP was made according to the previously described criteria [14]. OP was diagnosed either when the pathological ndings of lung biopsy samples were compatible with OP or when the patients ful lled all of the following criteria: (a) the typical image ndings on chest computed tomography (CT), determined to be OP by at least one rheumatologist and one radiologist; (b) no response to broad-spectrum antibiotics; and (c) substantial response to glucocorticoid therapy, demonstrated by improvement in systemic and respiratory symptoms and chest in ltrates within two weeks.

Data collection
Clinical information, laboratory results, medication, lung biopsy ndings, and CT images were collected from the patients' medical records. The follow-up data were abstracted until the time of death, dropout, or 30 October 2016. Rheumatoid factor and anti-citrullinated protein antibody were regarded as positive if positive ndings were reported at least once during the follow-up.
RA disease activity at the time of and several months before the onset of OP were assessed according to the Clinical Disease Activity Index (CDAI) [25], and were divided into two categories: 'Low' if the CDAI score was ≤10 (i.e. low disease activity or remission) or 'High' if the CDAI score was >10 (i.e. high or moderate disease activity). To analyse the relationship between OP development and RA exacerbation, each patient was categorised into one of following four groups based on the temporal changes of the RA disease activity along with OP development: 'Good Control,' where disease activity remained 'Low' both before and at the onset of OP; 'Poor Control,' where disease activity remained 'High' both before and at the onset of OP; 'Exacerbation,' where disease activity had been 'Low' before but became 'High' at the onset of OP; or 'Improvement,' where disease activity had been 'High' before but became 'Low' at the onset of OP.
For the examinations of the predictors of RA-OP recurrence, the OP recurrence was de ned as the appearance of new in ltrates on chest imaging with compatible clinical features during or after the course of OP treatment, which required consideration of hospital treatment and a signi cant increase in the glucocorticoid dosage. In this analysis, the recurrence group included all of the patients who presented with recurrence at least once during follow-up, and the non-recurrence group included the patients who experienced no recurrence during follow-up for at least six months after the onset of OP. The cutoff value of the six-month follow-up was set before performing any statistical analysis and was chosen for two reasons. First, according to the standardised protocol of the OP treatment used in our division, the glucocorticoid dosage is increased at the onset of OP and is usually tapered to its maintenance dosage approximately six months later; therefore, the cutoff ensured that the patients in the non-recurrence group had completed the tapering period without recurrence. Second, some patients changed hospital or died from other causes within a few months after OP onset; therefore, the cutoff helped to exclude patients not fully followed up from the non-recurrence group.

Statistical analysis
The data were summarised using descriptive statistics, including medians (with rst and third quartiles [Q1, Q3]), percentages, means, and con dence intervals (CIs). Continuous variables were compared using the Mann-Whitney U test, and categorical variables were compared using Pearson's chi-square test or Fisher's exact test. Recurrence-free survival times were calculated using the Kaplan-Meier method and were compared using the log-rank test. Cox proportional hazards regression analysis was performed to examine the associations between patients' characteristics and recurrence. All of the analyses were twosided, and the level of signi cance was set at p < 0.05.

Clinical characteristics of patients
The clinical features of the 33 RA-OP patients are summarised in Table 1 Of the 33 patients, 10 (30%) patients experienced OP recurrence, with a total of 27 recurrence events; the median interval between two consecutive OP events was 13.0 [7.5, 22.5] months. The rate of the rst recurrence of OP was 127 per 1,000 person-years.

Clinical characteristics of RA-OP events
We analysed a total of 60 RA-OP events, comprising 33 rst OP events in each patient and 27 recurrences observed in 10 patients ( Table 2). The most likely symptoms were coughing and fever, and dyspnoea was experienced during 39% of the events. Although the typical radiographic manifestation of OP is considered bilateral shadows, 42% of the events showed unilateral shadows. At the OP onset, the patients showed raised C-reactive protein (CRP) levels with a median of 10.2 [5.9, 16.9] mg/dL, and elevated erythrocyte sedimentation rate (ESR) levels of 109.0 [99.5, 124.5] mm/h. However, the median KL-6 titre, a serum marker for interstitial pneumonia, was 297.5 [233.3, 438.0] U/mL and remained within normal limits (<500 U/mL) in no fewer than 76% of the events. Nearly all of the events were treated with prednisolone (PSL), and steroid pulse therapy was performed concomitantly for 20% of the events.
Focussing on the 36 RA-OP events, where OP occurred even with PSL treatment, the median PSL dosage at the OP onset was 6.5 [5.0, 10.3] mg/day.

Relationship between OP development and RA exacerbation
Changes in RA disease activity, along with OP development, were examined for each RA-OP event ( Table  2). At the OP onset, the RA disease activity remained under good control in 52% of the events and was exacerbated in only 18%, suggesting that OP development is not accompanied by RA exacerbation. Consistently, the CRP and the ESR levels related to RA were not so high during a few months before the OP onset; the median CRP level was 0.5 [0.2, 1.4] mg/dL, and the ESR level was 24.0 [16.0, 52.0] mm/h.

Predictors of RA-OP recurrence
The predictors of RA-OP recurrence were determined among the patients' characteristics at the rst OP onset, using analyses comparing the recurrence group (n = 10) and the non-recurrence group (n = 14), as de ned in the 2.3 Data collection. The age at the rst OP onset was signi cantly lower in the recurrence group than in the non-recurrence group (59.5 years vs. 67.1 years; p = 0.043), with a signi cantly shorter period from the RA onset to the rst OP onset (6.4 years vs. 14.2 years; p = 0.047) ( Table 3). The receiver operating characteristic curves for these two predictors are shown in Fig. 1. A cutoff value of ≤66 years for age at the rst OP onset predicted OP recurrence with a speci city of 57% and sensitivity of 90% (Fig.  1A). Similarly, a cutoff value of ≤4 years for the period from the RA onset to the rst OP onset predicted the OP recurrence with a speci city of 69% and sensitivity of 80% (Fig. 1B).
The recurrence group included signi cantly more patients whose OP onset preceded the RA onset (the OP-preceding) than the non-recurrence group (30% vs. 0%; p = 0.029), and signi cantly more patients with a smoking history (80% vs. 36%; p = 0.032) ( Table 3). Bilateral lung shadows at the rst OP were more frequently observed in the recurrence group, although the difference was not statistically signi cant (p = 0.056). To estimate the recurrence-free survival time, the Kaplan-Meier method was applied to these two signi cant factors, the OP-preceding patients and the ever-smokers, for all of the 33 RA-OP patients (Fig. 2). The OP-preceding patients showed signi cantly shorter median recurrence-free survival time than the other: 15 (95% CI, 2-not available (NA)) vs. 136 (20-NA) months, p = 0.009, and showed a signi cantly higher risk for recurrence with a hazard ratio of 5.5 (1.3-23.0), p = 0.021 ( Fig. 2A). The eversmokers also exhibited shorter median recurrence-free survival time and higher risk for recurrence than the never-smokers, although these differences did not reach statistical signi cance: 34 (13-NA) vs. NA (13-NA) months, p = 0.15, with a hazard ratio of 3.0 (0.6-14.1), p = 0.17 (Fig. 2B).
There were no signi cant differences between the recurrence and the non-recurrence groups in the laboratory data at the rst OP onset, such as CRP levels, rheumatoid factor titres, or KL-6 titres (Table 3). Importantly, the initial treatment for the rst OP was similar in the two groups, including the initial PSL dosage and the concomitant use of steroid pulse therapy, and there was no signi cant difference in the minimum PSL dose without recurrence during the tapering process (4.7 mg/day vs. 5.4 mg/day, p = 0.37), which ensures the equivalent treatment in the two groups. Moreover, the mean follow-up period after the rst OP did not differ signi cantly between the two groups (81.3 months vs. 47.4 months, p = 0.11), and this follow-up period after the rst OP in the non-recurrence group (47.4 months) was, on the contrary, longer than the mean recurrence-free follow-up period after the rst OP in the recurrence group (27.1 months), indicating that the follow-up period was long enough to detect recurrence in the non-recurrence group.

Discussion
In this study, we clari ed the clinical characteristics of RA-OP patients and events, showed that OP development is not associated with RA exacerbation, revealed the high rate of the rst recurrence of RA-OP, and identi ed four predictive patient characteristics at the rst OP onset for later OP recurrence: lower age at the rst OP onset, shorter period from the RA onset to the rst OP onset, OP-preceding, and history of smoking. This study is the third retrospective cohort study for RA-OP [14,22] and the rst to establish the predictors of RA-OP recurrence.
This study contributes to enhance our understanding and knowledge of the clinical characteristics of RA-OP. In our cohort, 9% of cases were the OP-preceding, who exhibited a higher risk for OP recurrence.
This proportion of cases in our study is similar to those reported in previous studies: 11% [14] or 14% [22] of RA-OP patients. Meanwhile, the di culty in differentiating RA-OP from bacterial pneumonia in clinical practice may be, in part, explained by our ndings: 42% of the OP events showed unilateral shadows, 76% had normal KL-6 titres, and most patients presented with u-like symptoms, all of which are the typical characteristics of bacterial pneumonia. In particular, this proportion of the unilaterality of lung shadows seems similar to 58% reported on RA-OP [14] and might be higher than 30% [26] or 27% [27] reported on cryptogenic OP.
Importantly, we revealed that OP development is not associated with RA exacerbation. In this study, the RA disease activity remained good in 52% of the patients and was exacerbated in only 18% at the OP onset, similar to those reported in the previous study, 57% and 14%, respectively [14], increasing the reliability of our results. That RA-OP develops even under good control of RA indicates the di culty in foreseeing or preventing the occurrence of OP in patients with RA. This di culty will be highlighted, considering our nding that the period from the RA onset to the rst OP onset varied widely from −1 to 45 years.
Our study clari ed that 30% of the patients with RA-OP experienced OP recurrence even with proper glucocorticoid treatment. This proportion corresponds to those in previous reports on RA-OP: 21% [14] or 10% [22]. These proportions of RA-OP might be lower than in those of cryptogenic OP: 58% [28], 56% [29], 30% [26], or 27% [27]. These lower proportions in RA-OP patients may re ect the effect of other immunosuppressive agents concomitantly administered for arthritis during the tapering process of glucocorticoid. This study also demonstrated the high rate of the rst recurrence of RA-OP, 127 per 1,000 person-years. This rate is much higher than the reported rate of the total OP occurrence in RA patients, 2.72 per 1,000 person-years [14], indicating the need for the follow-up of RA-OP patients with careful attention to the likelihood of recurrence and underscoring the importance of this study, in which the predictors of RA-OP recurrence were identi ed.
OP, in general, relapses frequently and, therefore, there have been several studies on the predictors for its recurrence. However, the majority of the studies have focussed on cryptogenic OP, and until this study, no study has analysed the RA-OP recurrence. The identi ed predictors of OP recurrence in previous reports and this study are summarised in Table 4 [26 -32]. This comparison emphasises the value of our data exclusively on RA-OP, and suggests the particularity of RA-OP compared with other forms of OP.
We identi ed four predictors of RA-OP recurrence. Three of them are related to the timing of OP onset: lower age at the rst OP onset (especially ≤66 years), shorter period from the RA onset to the rst OP onset (especially ≤4 years), and OP-preceding. However, the age at the rst OP onset has not been identi ed as a predictor of recurrence in previous studies focussing mainly on cryptogenic OP (Table 4) [26 -32]. This discrepancy may indicate the specialty of RA-OP and re ect the difference of underlying immunological mechanisms between RA-OP and cryptogenic OP.
There have been several case series on RA-OP, including recurrent cases. In reviewing the literature, we found that many of these reported cases ful lled the identi ed predictors related to the timing. A retrospective cohort [22] included two patients with recurrence: a case whose period from the RA onset to the rst OP onset was 0 year, ful lling ≤4 years, and a patient whose rst OP onset was at the age of 61 years, ful lling ≤66 years. Another case series included a case whose rst OP onset preceded the RA onset [33], ful lling the OP-preceding, and the others included cases whose rst OP onset was at age 21 years [19], 55 years [34], or 57 years [35], all ful lling ≤66 years. This consistency may prove the validity of our predictors.
A history of smoking was the other predictor of RA-OP recurrence identi ed in this study. However, this characteristic has been previously reported as irrelevant to the recurrence of the other types of OP: cryptogenic OP [26, 28] or cryptogenic and secondary OP [32]. It is well established that smoking is a strong risk factor for RA [36] and even for ILDs in patients with RA [37,38], and that smoking in uences the severity of RA [39,40]. Therefore, smoking may have some speci c effects on the immunological status of RA-OP patients, resulting in its recurrence. This discrepancy again highlights the particularity of RA-OP, indicating the need for further studies that focus speci cally on RA-OP, excluding cryptogenic OP or other types of secondary OP.
The identi ed predictors of RA-OP recurrence may help to improve the treatment strategy for RA-OP patients in the clinical practice. For patients with these predictors, the close follow-up with careful attention to their symptoms and the glucocorticoid tapering at a cautious pace can be considered. Given that the median PSL dosage at the OP onset under PSL administration was 6.5 mg/day in this study, RA-OP patients treated with such doses should be followed up with careful attention to the likelihood of recurrence. For patients with the predictors, the usage of azathioprine as a glucocorticoid-sparing agent, in addition to PSL, could also be considered, although the evidence supporting its effectiveness has been limited to a few case reports [41][42][43].
This study had several limitations. First, this study was single-centred and retrospective, indicating the need for a prospective multicentre cohort study, where patients with RA at the rst OP onset or OPpreceding patients at the RA onset are registered and followed up. However, because this study is singlecentred, it has advantages in some aspects; the starting dosage and the tapering rate of glucocorticoid were almost uniform for all of the RA-OP patients according to the standardised protocol in our division, which reduced the likelihood of recurrence due to the inadequate remedy or too rapid glucocorticoid tapering. Second, the number of subjects was so small that we could not conduct multivariate analyses to exclude confounding factors, even though our cohort was larger than in those of the previous reports on RA-OP [14,22].

Conclusions
This study clari ed the clinical characteristics of RA-OP and showed the absence of association between OP development and RA exacerbation. RA-OP showed a high recurrence rate, especially in patients with four predictors newly identi ed. These ndings may help to improve the therapeutic strategy for RA-OP patients. Considering the specialty of RA-OP clari ed in this study, additional studies exclusively focussing on RA-OP will be needed based on much larger cohorts to con rm our ndings. While this study determined the predictors of recurrence, further research on the predictors of the rst occurrence of OP in patients with RA is warranted.

Declarations
Ethics approval and consent to participate This study was performed in compliance with the Declaration of Helsinki and was approved by the ethics committee of National Center for Global Health and Medicine (NCGM-G-003354-00). No additional consent form was needed for this retrospective non-interventional study.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors.
Authors' contributions RK, HY, YT, and HK designed the study. RK acquired, analysed, and interpreted the patient data. RK and HY wrote the manuscript. All authors read and approved the nal manuscript.