The Role of Pirfenidone in the Treatment of Interstitial Pneumonia With Autoimmune Features

Rationale: No approved pharmacotherapies are available for patients with interstitial pneumonia with autoimmune features (IPAF). Objective: In the present work, we aimed to evaluate the ecacy and safety of pirfenidone for the treatment of IPAF. Methods: A retrospective cohort study consisting of patients who met diagnostic criteria for IPAF was performed after a multidisciplinary review, and the patients receiving pirfenidone were compared with those in the non-pirfenidone group. The baseline data and diagnostic characteristics of patients were assessed. Pulmonary function and prednisone dose were analyzed by a mix-effects model. Results: A total of 184 patients, who met the diagnostic criteria of IPAF, were divided into two groups: pirfenidone group (n=81) and non-pirfenidone group (n=103). Patients in the pirfenidone group had a lower forced vital capacity (FVC%, P< 0.001) and a lower diffusion capacity for carbon monoxide (DLCO%, P=0.003). The pirfenidone group exhibited a greater increase of FVC% at 6 (P=0.003), 12 (P=0.013), and 24 (P=0.003) months. After adjustment for sex, age, UIP pattern, baseline FVC% and DLCO%, patients in the pirfenidone group continued to show a greater improvement in FVC% (χ2 (1) =4.59, P=0.032). Subgroup analysis identied superior therapeutic effects of pirfenidone in patients with dosage > 600 mg/day (P = 0.010) and medication course > 12 months (P = 0.007). Besides, the pirfenidone group had a lower prednisone dose than the non-pirfenidone group after 12 months of treatment (P=0.002). Moreover, 17 patients (19.32%) experienced side effects after taking pirfenidone, including one case of anaphylactic shock. Conclusions: with an acceptable safety and tolerability prole in IPAF patients.


Introduction
As a heterogeneous collection of uncommon disorders, interstitial lung disease (ILD) is characterized by interstitial brosis and progressive decline in lung function. A signi cant proportion of ILD patients demonstrate clinical features suggestive of a connective tissue disease (CTD) but fail to meet established CTD diagnostic criteria. Interstitial pneumonia with autoimmune features (IPAF) is used to label these patients according to a European Respiratory Society/American Thoracic Society research statement [1][2] . This new classi cation system combines clinical, serological, and morphological domains, with an IPAF diagnosis requiring at least two of the three domains. Importantly, IPAF criteria are not diagnostic but standards for classi cation, which are used to interpret study ndings and compare results between studies [3] .
As IPAF patients do not have de ned CTD, treatment may be similar to CTD-ILD for some IPAF patients [1] . The INBUILD study has shown that nintedanib is bene cial to progressive brosing ILD from a variety of CTDs [11] . Besides, nintedanib can slow down the annual rate of FVC decline in patients with systemic sclerosis-associated ILD [12] . On the other hand, pirfenidone also shows the potential treatment effects for IPAF. A multi-center clinical trial has demonstrated that pirfenidone can prevent the decline of FVC in patients with progressive brosing unclassi able ILD (PF-ILD) [13] , including IPAF patients. Li T et al. have reported that pirfenidone can improve the prognosis of patients with amyopathic dermatomyositis [14] . Taken together, we postulated that pirfenidone was associated with the improvement of pulmonary function in IPAF patients. To verify such a hypothesis, we explored the e cacy and safety of pirfenidone capsules for the treatment of IPAF, and it was registered in the Chinese Clinical Trial Registry (ChiCTR-IPR-17010813).
Patients And Methods

Screening process of patients
A total of 1,070 ILD patients diagnosed at Shanghai Pulmonary Hospital (Shanghai, China) from January 2014 to January 2019 were enrolled in this cohort. The screening process is illustrated in Figure 1. Finally, 242 patients met the diagnostic criteria of IPAF [2] . Among these patients, there were 172 cases with UCTD-ILD, and 70 cases were diagnosed with idiopathic interstitial pneumonia (IIP), including four with biopsy-proven cryptogenic organic pneumonia (COP), eight with IPF, and 58 with unclassi able IIP.
Exclusion criteria were set as follows: (1) patients without follow-up data (n=30); (2) patients with other complications (n=15 including any active infection, heart or hepatic or renal impairment); (3) the duration of pirfenidone treatment was less than 3 months (n = 7); and (4) the follow-up interval was more than 40 months (n = 6). This study was approved by the Ethics Committee of Shanghai Pulmonary Hospital (Approval No. K17-H1).

Data collection
Clinical data were collected from patient-visit records, including demographic characteristics, body mass index (BMI), smoking history, RFs and autoantibodies (ANA, anti-CCP, anti-double-stranded DNA, anti-SSA, anti-SSB, anti-RNP, anti-smith, anti-Scl-70, anti-tRNA synthetase), arterial oxygen saturation, and pulmonary function test (PFT). Medication history included glucocorticoids, immunosuppressive agents, and pirfenidone (dosage and duration of therapy). Baseline data were recorded at the time when the patient started pirfenidone or corticoid therapy (allowable range was 0-3 months to permit the inclusion of patients). The time table began with the time of baseline for all analyses. PFT was recorded at baseline and after 3 months of pirfenidone treatment, and then it was performed every 6 months as clinically indicated.

Pirfenidone treatment
Patients with the following situations were recommended to pirfenidone treatment: 1) patients exhibited more than 10% brosis on high-resolution computed tomography (HRCT); 2) patients had a more than 5% absolute decline in percent predicted FVC within the previous 6 months. All the patients started the pirfenidone therapy with a dose of 600 mg/day, and such a dose was increased to 1,800 mg/day in 6 months unless the patients experienced serious side effects. The nal dose (1,800 mg/day) was decided based on the clinical trial of pirfenidone [15] . A severe side effect was de ned as an event that caused an inability to work or perform daily activity.

Treatment of prednisone and immunosuppressants
The dose of prednisone was adjusted according to disease severity and body weight. A su cient dose of prednisone was administered at the beginning, and then it was gradually reduced. Unless the patients experienced an exacerbation, the dose of prednisone would be maintained at a relatively low level. All the immunosuppressants were administered by rheumatologists.

Diagnostic criteria
The nal diagnosis was made by a multidisciplinary discussion (MTD) (three experienced pulmonologists, two rheumatologists, two chest radiologists, and two pathologists). The diagnosis of ILD was made according to the diagnostic criteria described previously [16,17] . Diagnosis of IPAF was made based on the evaluation of three diagnostic domains (clinical, serological, and morphological domains) [2] . The morphological domain referred to HRCT or in combination with pathological results when lung biopsies were performed. All patients with CTD-ILD or UCTD-ILD were con rmed by rheumatologists. The diagnostic criteria for CTD in this study followed the recommendations by the American Rheumatism Association and the American College of Rheumatology [18][19][20][21][22][23] . UCTD was de ned as patients who showed systemic autoimmune features but did not meet de nite classi cation criteria [1] .

Chest HRCT evaluation
HRCT patterns were blindly reviewed and interpreted by two dedicated chest radiologists. HRCT diagnosis referred to proposed criteria for IPAF by ERS/ATS guidelines [2] , including NSIP, organic pneumonia (OP), NSIP in combination with OP, and UIP ( Figure S1). NSIP pattern was de ned as basal predominant reticular abnormalities with traction bronchiectasis, which was frequently associated with ground-glass attenuation. OP pattern was de ned as bilateral patchy areas of consolidation with a subpleural and lower lung zone predominance or peri-bronchovascular distribution. NSIP in combination with OP was de ned as basal predominant consolidation, which was associated with features of brosis. UIP pattern was de ned as basal and subpleural predominant honeycombing opacities associated with traction bronchiectasis. No lymphoid interstitial pneumonia (LIP) HRCT pattern was found in this cohort.

Data processing
Continuous variables were presented as mean (standard deviation) and compared by two-tailed Student's t-test. Categorized variables were expressed as frequency (percentage) and compared using the Chisquare test or Wilcoxon rank-sum test. All analyses were performed using GraphPad Prism 6 and SPSS 24 software (IBM, Armonk, NY, USA).
The PFT results were recorded at baseline and follow-up visits. The differences between the follow-up value and baseline value were calculated (change = follow-up value -baseline value), and then the changes in FVC absolute value, FVC%, and DLCO% were compared using a mixed-effects model. Fixed effects included gender, age, UIP pattern, baseline FVC%, and DLCO%. The mixed-effects model has been proved reliable in other retrospective studies [24][25][26] . The prednisone doses were compared by the same method. These analyses were carried out by R software. Table 1 shows that 184 patients were nally included in the analysis, including 81 (44.0%) patients in the pirfenidone group, and 103 (56.0%) patients in the non-pirfenidone group. The mean age of the cohort was 59.4 years old, 54.3% were females, and 53 (28.8%) patients had a history of smoking. There were no differences in gender, smoking history and UIP pattern. However, both FVC% and DLCO% were lower in the pirfenidone group compared with the non-pirfenidone group (FVC%, P < 0.001; DLCO%, P = 0.003). As for the treatment, the baseline data of glucocorticoid and immunosuppressant treatment were not different between the two groups. Generally speaking, 151 (82.1%) patients received oral glucocorticoid, and 13 (7.1%) patients received immunosuppressants. The duration of prednisone treatment was 2.25-40 months, with an average of 28.8 months. The mean duration of pirfenidone treatment was 14.4 months, and the dose of pirfenidone ranged from 600 to 1,800 mg/day, with an average of 1,492 mg/day. Table 2 shows the diagnostic characteristics. Overall, 66 (35.9%) patients met the diagnostic criteria of IPAF using a combination of serological and morphological domains, 53 (28.8%) patients met the diagnostic criteria of IPAF using clinical and morphological domains, 34 (18.5%) patients met the diagnostic criteria of IPAF using clinical and serological domains, and 31 (16.8%) patients met the diagnostic criteria of IPAF using all the three domains.

Diagnostic characteristics of IPAF patients
A breakdown of features into each IPAF domain showed that the most common clinical ndings were Raynaud's phenomenon (49, 26.6%) and in ammatory arthritis or polyarticular morning joint stiffness lasting ≥60 min (45, 24.5%). Moreover, 131 patients had positive serum autoantibody (71.2%), and 51 cases had two or more positive antibodies. An ANA ≥ 1:320 (or nucleolar or centromere pattern of any titer) was the most common serological nding (81, 44.0%). Within the morphological domain (150, 81.5%), the NSIP pattern by HRCT was found in 62.0% (114) of patients, while the OP pattern was found in 14.1% (26) patients. There were no differences in the diagnostic characteristics between the pirfenidone group and the non-pirfenidone group.

Changes in pulmonary function
The changes in FVC% ( Figure 2A) and DLCO% ( Figure 2B) between the two groups were compared at the time points of 3, 6, 12, 18, and 24 months. After 12 months of treatment, FVC% in the pirfenidone group was increased by 10.44%, while such value was decreased by 1.18% in the non-pirfenidone group (P=0.013). Besides, a greater increase of FVC% was observed in the pirfenidone group after 6 (P=0.003) and 24 months (P=0.003). A greater improvement of DLCO% was also observed in the pirfenidone group after 6 months (P=0.043).

Subgroup analysis of the pulmonary function
To further explore the effect of pirfenidone in different subgroups, subgroup analysis was performed. Table 3 shows the average annual change in FVC absolute value. The volume of FVC (liters) was increased by 0.0390 L/year in the pirfenidone group, while such value was decreased by 0.0769 L/year in the non-pirfenidone group (P=0.038). The association between pirfenidone use and greater improvement in FVC showed a qualitatively same trend in patients with FVC < 70% (P=0.021), with pirfenidone > 600 mg/day (P=0.010), and with total medication time > 12 months (P=0.007). Moreover, pirfenidone also showed superior effects in patients diagnosed by morphological and serological domains (P=0.033). Consequently, pirfenidone treatment had superior effects on FVC improvement when dose>600 mg/day and treatment time>12 months.

IPAF patients can reduce the dose of prednisone after 12 months
In our cohort, the prednisone dose ranged from 2.5 to 50 mg/day, with an average of 14.4 mg/day. The total dose ( Figure 3A) and daily dose ( Figure 3B) had no difference between the two groups when assessing the full duration of 40 months. However, when we separated the period into the initial 12 months and the remaining 12-40 months, both the total dose and daily dose of prednisone were signi cantly lower in the pirfenidone group (total dose, P=0.012; daily dose, P=0.032) during 12-40 months. After adjustment for potential confounders (sex, age, UIP pattern, baseline FVC%, and baseline DLCO%) in the mixed-effects model, patients in the pirfenidone group continued to show a reduced dose of prednisone by 6.27 mg per day ( Figure 3C, D, E χ2 (1) =9.8385, P=0.002, pirfenidone n=34, nonpirfenidone n= 27).

Side effects of pirfenidone
In the present study, 17 (19.32%) patients had side effects after taking pirfenidone ( Figure 4A) with seven (7.95%) cases of severe side effects (one case of anaphylactic shock, one case of arthritis, one case of liver injury, one case of photosensitivity, and three cases of skin rash) who stopped the medication. Skin rash (10.23%) and liver injury (5.68%) were the most common side effects, which were similar to those of IPF patients [12] . Moreover, 14 (14/17, 82.35%) patients experienced side effects at the initial dose (600 mg), and three (17.65%) patients experienced side effects after the dose of pirfenidone was increased ( Figure 4B).

Discussion
Several clinical trials have con rmed the e cacy of pirfenidone in IPF, demonstrating that pirfenidone can delay the decline of FVC and increase the progression-free survival rates [15,[27][28] . However, no study has explored the effects of pirfenidone in IPAF patients. Our observational study identi ed that the use of pirfenidone was associated with the improvement of FVC and the reduction of prednisone dose. The strengths of the study included the longitudinal data of PFT and prednisone dosage throughout 40 months as well as subgroup analyses of lung function.
The pathological features of IPAF are autoimmune in ammatory exudation and interstitial brosis. Therefore, the treatment for IPAF would cover both of the two sides. Wiertz et al. [29] have reported that IPAF patients may bene t from cyclophosphamide treatment. Besides, McCoy et al. [30] have shown that mycophenolate therapy can attenuate disease progression in IPAF patients. Nevertheless, all these published studies are designed to explore the effect of immunosuppressive therapy. No studies have yet explored the effect of anti-brosis treatment in IPAF patients. In the present study, we, for the rst time, reported that the anti-brosis treatment of pirfenidone could improve the pulmonary function of IPAF patients.
The average dosage of pirfenidone was 1,492 mg/day, suggesting that the dosage of pirfenidone for IPAF was not necessarily as high as that for IPF. Reasons might be as follows: 1) IPAF patients are relatively younger than IPF patients, as the mean age is 57-68 for IPAF [4][5][6][7][8][9] and 68-79 for IPF [17,[27][28] at diagnosis; 2) IPAF patients have more in ammatory exudative lesions on chest CT scans (i.g. NSIP and OP); 3) pirfenidone is mostly prescribed in combination with glucocorticoids; and 4) effective dose for East-Asian patients may be lower than Caucasian. In a phase-III clinical trial in Japan [28] , the effective dose of pirfenidone is 1,800 mg/day or 1,200 mg/day for IPF patients, which is lower than that in clinical trials (CAPACITY and ASCEND) in Caucasians (2,400 mg/day) [27,28] . Besides, we began with a low dose (600 mg/day) for the following considerations. 1) We observed that a low dose could achieve a certain effect on IPAF patients. 2) Low dose could help prevent side effects. 3) There is a heavier nancial burden for some patients in China if they take a high dose of pirfenidone. The duration of pirfenidone treatment was similar between our study and the IPF clinical trials [15,27,28] . Both indicated that the change of FVC was noted when the medication course was longer than 12 months.
The overall incidence of side effects was lower (19.32%) in the present study compared with other IPF clinical trials [27,28] . Only 10.23% of the patients had skin rash in our study, while such proportion is 28.1-32% in other IPF clinical trials [27,28] , which could be explained by the lower dose of pirfenidone (average 1,492 mg/day) in our study. The side effects of pirfenidone were dose-related in this study. Three (3.4%) patients experienced skin rash and liver damage when the pirfenidone dose was increased. These results further demonstrated the bene ts of lower-dose pirfenidone for IPAF patients.
Corticosteroids are widely used in IPAF patients. In the present study, the steroid dose was signi cantly reduced when pirfenidone was used for initial steroid-sparing therapy. Speci cally, the dose of prednisone was reduced by 6.27 mg per day in the pirfenidone group after 12 months of pirfenidone treatment. Consistent with this, J. A. Huapaya et al. have reported the use of immunosuppressants (azathioprine and mycophenolate) in 110 patients with myositis-related ILD (M-ILD) is associated with the reduction of prednisone dose [31] . The reduction of prednisone dose prevents the side effects of corticoids, therefore improving the medication compliance and treatment outcomes in IPAF patients.
Our study has several limitations. First, the study is limited to reporting associations, but unable to identify causal relationships due to the retrospective, single-center, and observational nature. Second, patients were not randomized to pirfenidone treatment. Therefore, pirfenidone exposure might cause an indication bias. Patients receiving pirfenidone were more likely to have a progressive brosing ILD. However, the subgroup analysis identi ed the same effect of pirfenidone in patients with FVC%<70%. Besides, limited follow-up of subjects over time might lead to misleading estimates of bene cial drug effects. Nevertheless, the follow-up bias could be weakened by adding the time interval as a random effect into the mixed-effects model. Last, although the analysis was adjusted by the mixed-effects model, system differences in the cohorts could not be ignored. Therefore, our current ndings need to be con rmed by prospective studies. However, multi-center clinical trials cannot be accomplished in a short time. Therefore, in the meantime, our study might help provide suggestions for therapy in IPAF patients.

Conclusions
Collectively, our ndings indicated that low-dose pirfenidone (1,492 mg/day) might help improve FVC with an acceptable safety and tolerability pro le in IPAF patients.
Abbreviations BMI= body mass index; COP= cryptogenic organic pneumonia; CTD-ILD= Connective tissue diseaseassociated ILD; DLCO= diffusion capacity of the lung for carbon monoxide; FVC= Forced vital capacity; HRCT=high-resolution computed tomography; IIP= idiopathic interstitial pneumonia; ILD= Interstitial lung disease; IPAF=Interstitial pneumonia with autoimmune features; IPF= idiopathic pulmonary brosis; MTD= multidisciplinary discussion; NSIP= Non-speci c interstitial pneumonia; Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of Shanghai Pulmonary Hospital (Approval No. K17-H1). Informed consent was obtained from all patients before. enrollment in this study.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this article.

Competing interests
The authors declare that they have no competing interests.  OP, organizing pneumonia. ※ , , , respectively represent one, two, three or four different kinds of autoantibodies are positive with the patients. ♯ ANA ⩾ 1:320 titer, diffuse, speckled, homogeneous patterns or a. ANA nucleolar pattern (any titer) or b. ANA centromere pattern (any titer). * P<0.05 Table 3.   and DLCO% (D) in the two groups of patients. The adjustment was performed by mix-effect model for gender, age, UIP pattern, baseline FVC% and DLCO%.  Side effects of pirfenidone. A. 17 (19.32%) patients in the pirfenidone group suffered from side effects, and 7 (7.95%) patients stopped the pirfenidone therapy due to side effects. B. Side effects occurred in 14