Subsequent Identication of a Connective Tissue Disease Amongst Patients with Idiopathic Pulmonary Fibrosis : A Long-term Observational Study in 527 Patients

Background: Connective tissue disease (CTD) might occur during the course of idiopathic pulmonary brosis (IPF). Clinical factors associated with CTD development in IPF patients have still not been identied. We investigated which antibodies have a signicant association with the development of CTD during the clinical course of IPF. Methods: We retrospectively reviewed the records of 527 patients with a rst diagnosis of IPF between January 2007 and March 2014 and investigated the time to CTD development after IPF diagnosis in these patients. Results: CTD developed in 15 patients at a median of 2.1 years (range 1.2–4.8) after IPF diagnosis. All patients had anti-neutrophil cytoplasmic antibodies (ANCA) or autoantibodies that met the serology criteria for interstitial pneumonitis with autoimmune features. Survival duration for IPF patients with progression to CTD was 5.3 (3.8, 6.7) years, which was signicantly longer than for IPF patients without progression to CTD [2.9 (1.7, 4.8), p = 0.001]. Independent risk factors for CTD development in IPF patients included female gender [adjusted hazard ratio (HR) 5.319, p = 0.0082], titer of rheumatoid factor (RF; adjusted HR, 1.006; p = 0.022), titer of anti-citrullinated protein antibody (ACPA; adjusted HR, 1.009; p = 0.0011), and titer of myeloperoxidase (MPO)-ANCA (adjusted HR, 1.02; p < 0.0001). Conclusions: Progression to CTD is uncommon in IPF patients. However, a signicant number of IPF patients with high titers of RF, ACPA, or MPO-ANCA progressed to CTD. RF, ACPA, and MPO-ANCA might be signicantly associated with CTD development in IPF patients. by the Student’s t test or the Mann–Whitney U test for continuous variables, and by the χ2 or Fisher’s exact test for categorical variables. Multivariable Cox proportional-hazards models with backward elimination were used to investigate the risk factors for the development of CTD. The proportional-hazards assumptions were assessed based on Schoenfeld residuals and log-log plots. Statistical signicance was set at p < 0.05. Statistical analysis was performed with R software (R Foundation for Statistical Computing, Vienna, Austria). percentages in and lymphoid aggregates with germinal centers in lung biopsy. pneumonitis autoimmune IPF: polyangiitis; MPO: myeloperoxidase; proteinase-3; SJS:


Background
Idiopathic pulmonary brosis (IPF) is a chronic disease characterized by pulmonary brosis and parenchymal destruction.(1, 2) Usual interstitial pneumonia (UIP) is the typical radiological and histopathological pattern of IPF. (2,3) IPF is, by de nition, UIP of unknown cause, and UIP is also observed in patients with connective tissue disease (CTD). Generally, CTD/UIP is diagnosed following investigation of patients with suspected UIP.(1) However, we sometimes encounter patients who progress to CTD during the clinical course of IPF/UIP and whose diagnosis changes to CTD/UIP.(4-7) Because CTD/UIP differs from IPF/UIP in treatment and prognosis, (8)(9)(10)(11)(12) it is important to predict the development of CTD during the clinical course of IPF/UIP. However, the cumulative incidence and predictive factors associated with the development of CTD during the clinical course of IPF remain unclear. These factors might be valuable clues for determining the pathogenesis of pulmonary brosis in patients with CTD.
Previously, we identi ed that the autoantibody-positive IPF patients were associated with a good prognosis; moreover, the use of immunosuppressants was associated with a better prognosis in these patients as opposed to the autoantibody-negative patients. (13) At present, autoantibody-positive IPF cases have not been classi ed as interstitial pneumonitis with autoimmune features (IPAF), and thus, the use of immunosuppressants is not recommended. (14) The clinical characteristics of autoantibody-positive IPF are similar to those of CTD/UIP patients.
We hypothesize that some IPF patients have a clinically signi cant association with autoimmunity, and that autoantibodies are important biomarkers for identifying these patients. Based on this hypothesis, we investigated whether the serology criteria presented above were associated with the development of CTD during the clinical course of IPF in the patients from our previous study, with a particular focus on which autoantibodies have a signi cant association with the development of CTD.

Methods
We retrospectively reviewed the records of 527 patients diagnosed with IPF by a multidisciplinary team (pulmonologist, radiologist, rheumatologist, and pathologist) assessing the results of high-resolution computed tomography or lung biopsy in a tertiary hospital from January 2007 through March 2014. The patients were followed up until April 2016 (for 2.4 [IQR, 1.1; 4.2] years). Information regarding demographics, medication, laboratory tests, bronchoscopy, and pulmonary function tests (PFT) was retrospectively collected from the institutional electronic medical records and entered into a database for analysis.
CTD-related symptoms were de ned by the IPAF clinical domain. (15) IPF patients with suspected CTD-related symptoms or autoantibodies were evaluated by experienced rheumatologists, and those with CTD-related symptoms con rmed by a rheumatologist at the rst visit to the clinic for IPF diagnosis (baseline) were excluded from this study. No clinical or morphologic manifestation suitable for the IPAF, were identi ed in the patients included in our study. We investigated the length of time from baseline to CTD diagnosis by an expert rheumatologist in IPF patients. Diagnosis of the speci c type of CTD was made in accordance with the criteria from their corresponding societies.(16-20) We de ned the survival duration of patients with IPF from their rst visit to the clinic for IPF diagnosis to death, and we de ned the baseline as the date of the rst visit to the clinic for IPF diagnosis. Death was de ned as cessation of national health insurance cover. Glucocorticoid (not for management of acute exacerbation) and immunosuppressants for immunomodulation were administered on an individual basis. The present study protocol was approved by the Institutional Review Board of Asan Medical Center (IRB number: 2016 − 0222).

Tests for serologic autoantibodies
The following antibodies were tested; this consisted of tests for anti-citrullinated peptide antibody (ACPA), antinuclear antibody (ANA), rheumatoid factor (RF), anti-dsDNA antibody, anti-Sm antibody, anti-RNP antibody, anti-Ro (SSA) antibody, anti-La (SSB) antibody, anti-Jo-1 antibody, anti-Scl-70 antibody, myeloperoxidase (MPO) anti-neutrophil cytoplasmic antibody (ANCA), and proteinase-3 (PR3) ANCA. Anti-synthetase syndrome Ab (except Jo-1), anti PM-Scl, anti-MDA-5 in the IPAF serologic domain were not tested because that was incompatible with current practice in ASAN Medical Center. Each untested autoantibody was considered autoantibody negative. Autoantibody tests were performed at the time of initial IPF diagnosis in most patients. These autoantibodies were repeatedly tested, usually 1-2 years after the rst test in some patients.

Statistical analyses
Continuous variables are reported as means ± standard deviations or medians and interquartile ranges (IQR), and categorical variables are presented as percentages. Differences between groups were evaluated by the Student's t test or the Mann-Whitney U test for continuous variables, and by the χ2 or Fisher's exact test for categorical variables. Multivariable Cox proportional-hazards models with backward elimination were used to investigate the risk factors for the development of CTD. The proportional-hazards assumptions were assessed based on Schoenfeld residuals and log-log plots. Statistical signi cance was set at p < 0.05. Statistical analysis was performed with R software (R Foundation for Statistical Computing, Vienna, Austria).

Results
Of the 527 IPF patients in the study, 42% (n = 221/527) had autoantibodies, and 29% (n = 153/527) had ANCA or autoantibodies that met the IPAF serology criteria. (Fig. 1, Supplementary table 1). CTD developed in 15 (2.8%) IPF patients, at a median of 2.1 [IQR, 1.2-4.8] years after the initial diagnosis of IPF [seven patients with rheumatoid arthritis (RA), one undifferentiated connective tissue disease, two Sjögren syndrome (SJS), one polyarteritis nodosa, and four microscopic polyangiitis (MPA), Tables 1 and 2]. All of these patients who progressed to CTD had either ANCA or any other tested autoantibodies that satis ed the IPAF serology criteria in the clinical course. Two IPF patients who progressed to RA had MPO-ANCA. A signi cant number of IPF patients with high titers of RF, ACPA, or MPO-ANCA tested at rst visit to the clinic progressed to CTD (Fig. 2). Furthermore, four patients (RA case 1, 3, 6, and MPA case 12, Table 2), who initially had either a low titer or absence of RF, ACPA, or MPO-ANCA, were found to have high titers of these autoantibodies following progression to CTD.     Anti-Scl antibody on f/u 174 192     In a Japanese IPF cohort, CTD developed in 9% of the patients (10/111 patients − 4 RA, 4 MPA, 1 SSc, 1 SSc/SJS) following an initial IPF diagnosis. (35) As in the Japanese IPF cohort, our study showed progression to RA and MPA, rather than the other CTDs, to be more common, with CTD progression more greatly observed among the lower age patients and females. However, they did not nd autoantibodies to be signi cant risk factors for the development of CTD as ACPA was not tested and autoantibodies, with the exception of RF, ANA, and anti-dsDNA antibodies, were tested in only approximately half of the Japanese IPF patients; hence, their study could have a greater number of patients with false negatives compared to our study. Additionally, the Japanese study found that lymphoid aggregates with germinal centers were signi cant risk factors for development of CTD, which were not collected in our study. Notably, we observed a signi cantly higher lymphocyte percentage in BAL in IPF patients with CTD progression than in those without progression to CTD. However, we did not include lymphocyte percentage in BAL, which was assessed in approximately half of the patients, in the multivariable analysis. Hence, the factors that may be associated with autoimmunity in the diagnosis of suspected IPF patients include, relatively young age, sex (female), presence of RF, ACPA, or MPO-ANCA, lymphocyte percentages in BAL, and lymphoid aggregates with germinal centers in lung biopsy.
This study has several limitations in addition to those associated with being retrospective. Some patients had not been continuously visiting the hospital before death, and as a result, we may not know about the development of CTD in some cases. Nevertheless, the mean ratio of individual clinic follow-up duration to survival duration was 0.84. The IPF patients with development of CTD had a longer survival, and some patients who had stopped visiting the hospital came back and were diagnosed with CTD. In considering this, it is likely that there were not many cases of unknown CTD.

Conclusions
Autoantibodies have been identi ed in many IPF patients in previous studies, and some of these patients have been thought to be closely associated with autoimmunity. We observed development of CTD in IPF patients with ANCA or autoantibodies that met the IPAF serology criteria. Among these autoantibodies, RF, ACPA, and MPO-ANCA were signi cantly associated with the development of Flowchart representing the differentiation of IPF patients into distinct categories. ANCA = anti-neutrophil cytoplasmic antibodies; CTD = connective tissue disease; IPAF = interstitial pneumonitis with autoimmune features; IPF = idiopathic pulmonary brosis;