Interstitial Lung Disease in Primary Biliary Cholangitis: A Cohort Prospective Study

Interstitial lung disease (ILD) has been recognized as an extrahepatic manifestation ofprimary biliary cholangitis (PBC), althoughlimited data are available on its prevalence and clinical significance. Therefore, we evaluated the occurrence and clinical features of ILD in a cohort of PBC patients. Ninety-three individuals without concomitant rheumatic diseases were enrolled in our prospective cohort study. All patients underwent chest high-resolution computed tomography (HRCT). Liver-related and lung-related survival wereassessed. A lung-related outcome was defined as death from ILD complications; a liver-related outcome was defined as liver transplantation or death from liver cirrhosis complications. HRCT findings suggestive ofILD were detected in 38 patients (40.9%). A sarcoid-like pattern of PBC-associated ILD was the most frequent, followed by subclinical ILD and organizing pneumonia. Patients with ILD were less likely to have liver cirrhosis and liver-related symptoms and presented with higher serum immunoglobulin M(IgM) and M2 subtype antimitochondrial antibodies (AMA-M2) positivity rates. In a multivariate analysis, the absence of liver disease symptoms at the disease presentation (OR 11.509; 95% CI 1.210–109.421; p = 0.033), the presence of hepatic non-necrotizing epithelioid cell granulomas (OR 17.754; 95% CI 1.805–174.631; p = 0.014), higher serum IgM (OR 1.535; 95% CI 1.067–2.208; p = 0.020) and higher blood leukocyte count (OR 2.356; 95% CI 1.170–4.747; p = 0.016) were independent risk factors associated with ILD in PBC. More than a third of patients with ILD showed no respiratory symptoms, and only one ILD-related death occurred during a follow-up of 29.0 months (IQR 11.5; 38.0). Patients with ILD had better liver transplant-free survival.ILD in PBC had a benign course and was associated with a lower liver disease severity. PBC-associated ILD should be included in a list of differential diagnoses of ILD.


Introduction
Primary biliary cholangitis (PBC), formerly known as primary biliary cirrhosis, is an autoimmune cholestatic liver disease resulting from immune-mediated destruction of small intrahepatic bile ducts and eventually progressing to liver cirrhosis. Chronic nonsuppurative destructive cholangitis is a histopathological hallmark of PBC. Poorly formed non-necrotizing epithelioid cell granulomas can also be found, particularly in the earlier stages of the disease [1]. Immunologically, PBC is characterized by the presence of circulating M2 subtype antimitochondrial antibodies (AMA-M2), antinuclear antibodies (ANAs), and elevated serum immunoglobulin M (IgM). PBC mostly affects middle-aged females and usually presents with fatigue, pruritus, cutaneous hyperpigmentation, and hepatosplenomegaly [1]. Ursodeoxycholic acid (UDCA) can be used in PBC patients

Study Population and Design
In a prospective single-centercohort study, we enrolled consecutive patients with PBC who were admitted to our clinic between January 2018 and January 2022. The study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the local ethics committee of Sechenov University. All patients signed an informed consent form prior toenrollment.
PBC was diagnosed according to the criteria of the American Association for the Study of Liver Diseases [2]. Primary biliary cholangitis-autoimmune hepatitis overlap syndrome (PBC-AIH overlap) was diagnosed using Paris diagnostic criteria [21]. UDCA treatment response was assessed by Paris-II criteria [22].
Patients with concomitant autoimmune diseases such as SjS, systemic sclerosis (SSc), systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), IgG4-related disease (IgG4-RD), or HCV infection were excluded from our study. Xerostomia solely or in combination with xerophthalmia was defined as sicca syndrome if the patient did not meet the classification criteria for SjS.

Data Collection and Outcomes
All patients underwent chest high-resolution computed tomography (HRCT) and spirometry. HRCT patterns of ILD were classified according to the American Thoracic Society and European Respiratory Society guidelines [23]. Criteria for subclinical ILD include ILD extent < 5%, preserved lung function (FVC > 80%), and absence of respiratory symptoms. Bilateral hilar adenopathy with multifocal peribronchovascular nodules was classified as a sarcoid-like HRCT pattern. Forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) were measured to evaluate pulmonary function [24]. Liver ANAs panel was assessed in 46 patients by enzyme-linked immunosorbent assay (ELISA) (EuroImmune, Germany). Liver stiffness was measured using transient elastography (Echosense, France).
The studied outcomes during follow-up included ILD-related death and liver transplantation or death from liver cirrhosis complications.
The study enrollment flow is illustrated in Figure 1. The study enrollment flow is illustrated in Figure 1.

Statistical Analysis
The Shapiro-Wilktest was used to check if the continuous variables follow a norm distribution. Continuous variables were presented as mean with standard deviation (SD or median with interquartile range (IQR), and categorical variables were expressed counts and percentages. The Student's -test was used to compare normally distribute continuous variables, whereas the Mann-Whitney U test was used when data were n normally distributed. Pearson`s χ2 test and Fisher`s exact test were used for comparin frequencies. Factors associated with ILD in PBC were evaluated by logistic regressio analysis. Survival analysis was performed using the Cox regression analysis, th Kaplan-Meier method, and the log-rank test. A two-sided p-value <0.05 was considere statistically significant. Statistical analyses were performed using IBM SPSS version (SPSS, Inc., Chicago, IL, USA).

Patient Characteristics
A total of 93 PBC patients were enrolled and followed up for a median of 29 months (IQR 11.5; 38.0). Most of them were females (95.7%). The mean age at disea presentation was 47.1 ± 1.0 years. Nearly half of the patients had liver cirrhosis. Eve sixth patient presented with PBC-AIH overlap syndrome. The diagnosis was confirme by ultrasound-guided percutaneous needle liver biopsy in 45 patients. All patients we treated with UDCA, whereas patients with PBC-AIH overlap also received glucocor coids and azathioprine. Circulating AMA-M2 wasdetected in 90.3% of the patients. Va iousANAs, including anti-gp210, anti-sp100, anti-SS-A, and anti-PML, were less prev lent (Table 1).

Statistical Analysis
The Shapiro-Wilktest was used to check if the continuous variables follow a normal distribution. Continuous variables were presented as mean with standard deviation (SD) or median with interquartile range (IQR), and categorical variables were expressed as counts and percentages. The Student's t-test was used to compare normally distributed continuous variables, whereas the Mann-Whitney U test was used when data were not normally distributed. Pearson's χ2 test and Fisher's exact test were used for comparing frequencies. Factors associated with ILD in PBC were evaluated by logistic regression analysis. Survival analysis was performed using the Cox regression analysis, the Kaplan-Meier method, and the log-rank test. A two-sided p-value < 0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS version 22 (SPSS, Inc., Chicago, IL, USA).

Patient Characteristics
A total of 93 PBC patients were enrolled and followed up for a median of 29.0 months (IQR 11.5; 38.0). Most of them were females (95.7%). The mean age at disease presentation was 47.1 ± 1.0 years. Nearly half of the patients had liver cirrhosis. Every sixth patient presented with PBC-AIH overlap syndrome. The diagnosis was confirmed by ultrasoundguided percutaneous needle liver biopsy in 45 patients. All patients were treated with UDCA, whereas patients with PBC-AIH overlap also received glucocorticoids and azathioprine. Circulating AMA-M2 wasdetected in 90.3% of the patients. VariousANAs, including anti-gp210, anti-sp100, anti-SS-A, and anti-PML, were less prevalent (Table 1).
Tissue histology was available in 9 of 14 patients with sarcoid-like patterns (lung biopsy via video-assisted thoracoscopic surgery in five, mediastinal lymph nodes in two, supraclavicular lymph nodes in two), and in all cases showed non-necrotizing epithelioid cell granulomas with giant cells. An open lung biopsy was performed on two patients with pulmonary consolidation showed granulation tissue buds in the alveoli and alveolar ductus consistent with the OP pattern. Histopathological findings in three patients with indeterminate radiological patterns included dense lymphocytic infiltration of the alveolar septa (n = 3), giant cells (n = 3), poorly formed non-necrotizing epithelioid cell granulomas (n = 2), foamy macrophages (n = 2), mild eosinophilic infiltrate (n = 2) and moderate interstitial fibrosis (n = 1). These findings were incompatible with radiological data, and we were unable to define a type of ILD. One patient with an unclassifiable radiological pattern with pulmonary consolidation showed granulation tissue buds in the alveoli and alveolar ductus consistent with the OP pattern. Histopathological findings in three patients with indeterminate radiological patterns included dense lymphocytic infiltration of the alve olar septa (n = 3), giant cells (n = 3), poorly formed non-necrotizing epithelioid cell gran ulomas (n = 2), foamy macrophages (n = 2), mild eosinophilic infiltrate (n = 2) and mod erate interstitial fibrosis (n = 1). These findings were incompatible with radiological data and we were unable to define a type of ILD. One patient with an unclassifiable radio logical pattern refused a lung biopsy. Representative HRCT patterns and tissue histology are shown in Figures 2-5.    ILD was an initial manifestation of PBC in nine (9.6%) patients. All of them had liver function tests abnormalities at the time of ILD detection. The most common clinical manifestations of ILD included chronic cough (50.0%), fever (44.7%), and dyspnea (42.1%). Respiratory failure was present in three (7.9%) of thirty-eightpatients. Fourteen (36.8%) of thirty-eightpatients with ILD had no respiratory symptoms during follow-up.   ILD was an initial manifestation of PBC in nine (9.6%) patients. All of them had liver function tests abnormalities at the time of ILD detection. The most common clinical manifestations of ILD included chronic cough (50.0%), fever (44.7%), and dyspnea (42.1%). Respiratory failure was present in three (7.9%) of thirty-eightpatients. Fourteen (36.8%) of thirty-eightpatients with ILD had no respiratory symptoms during follow-up.

Comparison of ILD and ILD-Free Patients
Patients with and without ILD were similar in age at the time of disease presentation, sex, smoking status, and the occurrence of overlap with AIH, whereas obesity was more common in patients with ILD. Liver cirrhosis was less prevalent in patients with ILD. ILDfree patients showed higher liver stiffness values. It is worth noting that patients with ILD more frequently did not present with liver disease symptoms at the disease presentation. FEV1 and FVC values did not differ significantly between the two groups (Table 1).
Hepatic non-necrotizing epithelioid cell granulomas were more common in patients with ILD compared to patients without ILD (44.4% vs. 11.1%, p = 0.016). Seven of eight patients with hepatic granulomas and ILD had a sarcoid-like pattern, and one patient had an OP pattern. Liver cirrhosis occurred less frequently among patients with hepatic granulomas than in patients without granulomatous liver disease (27.3% vs. 61.8%, p = 0.081).
Patients with ILD had higher blood leucocyte count, serum IgM levels

Factors Associated with ILD in PBC
Multivariate logistic regression analysis adjusted for age, gender, and s tus showed that the absence of liver disease symptoms at the disease presen ratio (OR) 11

Follow-Up
Repeated chest HRCT was conducted in 34 patients with ILD at an interval of 3-6 months. Immunosuppressive agents were started in case of radiological and/or clinical deterioration of ILD. A total of 16 patients were treated for ILD with azathioprine or methotrexate in combination with systemic glucocorticoid. Improvement in chest HRCT was shown in 22 patients. It was spontaneous in eight patients (sarcoid-like pattern in three, subclinical ILD in three, and OP in two) or induced by immunosuppression in 14 patients (sarcoid-like pattern in eight, OP in four, UnIP in two). ILD was stable in fivepatients who received no immunosuppression (subclinical ILD in fourand sarcoid-like pattern in one) and in two patients treated with immunosuppressive agents (sarcoid-like pattern and UIP). Radiological progression of ILD was found in five patients. Two patients with NSIP and UnIP died from liver cirrhosis complications before initiation of immunosuppressive treatment, whereas two patients with OP and UnIP had no respiratory symptoms and were clinically stable. One patient showed no ILD response to immunosuppressive agents. This was a 70-year-old female with NSIP ( Figure 2) who developed severe pulmonary hypertension and died from respiratory failure. The autopsy revealed extensive interstitial pulmonary fibrosis consistent with fibrosing NSIP. During follow-up, only this patient died from ILD complications.

Survival Analysis
At the end of the study, there were 12 deaths from complications of liver cirrhosis and two liver transplantations. Three deceased patients had ILD, but postmortem examination was not available. The 10-year liver transplant-free survival in ILD patients was higher than in patients without ILD (93.2% vs. 79.9%, p = 0.021, Figure 7).
In the univariate Cox regression analysis, ILD was associated with a lower risk of liver-related outcomes (hazard ratio (HR) 0.2; 95% CI: 0.044-0.911; p = 0.038), but this association lost statistical significance after adjusting for other covariates (Table 3). methotrexate in combination with systemic glucocorticoid. Improvement in chest HRCT was shown in 22 patients. It was spontaneous in eight patients (sarcoid-like pattern in three, subclinical ILD in three, and OP in two) or induced by immunosuppression in 14 patients (sarcoid-like pattern in eight, OP in four, UnIP in two). ILD was stable in fivepatients who received no immunosuppression (subclinical ILD in fourand sarcoid-like pattern in one) and in two patients treated with immunosuppressive agents (sarcoid-like pattern and UIP). Radiological progression of ILD was found in five patients. Two patients with NSIP and UnIP died from liver cirrhosis complications before initiation of immunosuppressive treatment, whereas two patients with OP and UnIP had no respiratory symptoms and were clinically stable. One patient showed no ILD response to immunosuppressive agents. This was a 70-year-old female with NSIP ( Figure 2) who developed severe pulmonary hypertension and died from respiratory failure. The autopsy revealed extensive interstitial pulmonary fibrosis consistent with fibrosing NSIP. During follow-up, only this patient died from ILD complications.

Survival Analysis
At the end of the study, there were 12 deaths from complications of liver cirrhosis and two liver transplantations. Three deceased patients had ILD, but postmortem examination was not available. The 10-year liver transplant-free survival in ILD patients was higher than in patients without ILD (93.2% vs. 79.9%, p = 0.021, Figure 7). In the univariate Cox regression analysis, ILD was associated with a lower risk of liver-related outcomes (hazard ratio (HR) 0.2; 95% CI: 0.044-0.911; p = 0.038), but this association lost statistical significance after adjusting for other covariates (Table 3).

Discussion
The results of our single-center cohort study showed that ILD is a common finding in PBC and can be detected on chest HRCT in up to 41% of patients without concomitant autoimmune diseases such as SjS, SSc, SLE, RA, and IgG4-RD. The most common patterns of ILD included sarcoid-like changes, subclinical ILD, and OP, whereas UnIP, NSIP, and UIP occurred less frequently. ILD was asymptomatic in one-third ofpatients. Of note, ILD manifesting by cough and/or dyspnea was the first presentation of the disease in a proportion of PBC patients (10%) who were referred to our department by pulmonologists due to elevated serum cholestatic enzymes. The absence of liver disease symptoms at the presentation of the disease was an independent risk factor for PBC-associated ILD. These data suggest that patients with unexplained ILD should be screened for asymptomatic PBC.
In the previous studies, the prevalence of ILD in PBC patients was lower. In two Chinese cohorts, ILD was detected only in 15.7% of 178 patients and 10.0% of 109 patients, respectively [18,25]. In both studies, ILD was frequently associated with connective tissue diseases, e.g., SjS. The reported prevalence of ILD in Caucasian PBC patients was even lower and varied from 2.2% to 5.4% [26][27][28]. It was probably underreported since a plain chest roentgenogram was used to detect lung abnormalities in these studies. Of note, all aforementioned studies included a proportion of patients with concomitant connective tissue diseases (CTD) who were excluded from our cohort. A higher prevalence of ILD in our PBC patients was probably related to the use of HRCT for ILD detection, including subclinical ILD.
Sarcoid-like lung involvement was the main pattern of ILD in our study. In our opinion, PBC-associated pulmonary granulomatosis should be considered as the extrahepatic manifestation of PBC rather than as a concomitant sarcoidosis. PBC and sarcoidosis share common features, and differential diagnosis may be challenging in some cases. Both hepatic and extrahepatic non-necrotizing granulomas can be found in PBC [29], particularly in the earlier stages of the disease [30,31]. In a previous necropsy series of 120 PBC patients, intrathoracic granulomas were detected less frequently than in our study (1.6% vs. 15.0%) [28]. The bronchoalveolar lavage CD4 + /CD8 + lymphocytes ratio in PBC patients without lung involvement was similar to that in pulmonary sarcoidosis [15]. Hepatic granulomas in PBC, in contrast to granulomas in hepatic sarcoidosis, are poorly formed, not confluent, and localize mostly within portal tracts [32]. Ductopenia is rarely seen in hepatic sarcoidosis [32,33]. As a rule, patients with genuine hepatic sarcoidosis do not have serum AMA-M2 and respond to immunosuppressing agents [34]. Lee et al. showed a better prognosis in PBC patients having hepatic granulomas [35]. You et al. suggested that hepatic granulomas in PBC may be an adaptive reaction to chronic bile duct injury and may reduce cholangitis activity [31]. In our study, liver cirrhosis was less common among patients with hepatic granulomas, and two-thirds of these patients presented witha sarcoid-like pattern on chest HRCT. Only one of 14 patients with a sarcoid-like type of ILD died from liver cirrhosis complications.
We identified subclinical ILD in nine (9.7%) of 93 PBC patients. Subclinical ILD was studied in patients with RA [36], SSc, antisynthetase syndrome, and mixed CTD [37] and was suggested to represent an early form of CTD-associated ILD [37][38][39]. The reported rate of radiological progression of subclinical ILD in patients with CTD varied from 34% to 57% over 1.5 to 4.5 years [37,40,41]. There was no progression of subclinical ILD during the 2.4-year follow-up in our study. Anyway, we believe that all patients with PBC should be screened with chest HRCT for subclinical ILD to improve early detection of PBCassociated ILD.
Similar to Lee and Shen, we found non-necrotizing epithelioid cell granuloma, OP, interstitial fibrosis, and alveolar septa lymphocytic infiltration to be the predominant histopathological features of PBC-associated ILD [25,42]. It has been speculated whether these findings constitute a specific pattern of ILD in some PBC patients [42].
Every third PBC patient with ILD in our study was obese. The association between obesity and pulmonary diseases, including ILD, can be related to TNF-α -, MCP-1 -, and TGF-β1-dependent pathways, insulin resistance, or altered lung microbiota [43][44][45]. Greater amounts of visceral adipose tissue predispose to a higher prevalence of interstitial lung abnormalities [46]. It was suggested that this association mightbe partially mediated by IL-6 and leptin [46].
In our study, blood leukocyte and neutrophil counts were higher in patients with ILD, although they did not exceed the upper limit of the normal range. Neutrophils have been recognized as an essential player in the pathogenesis of ILD [47,48] and promote interstitial pulmonary fibrosis via several pathways [47,49]. However, higher leucocyte counts in patients with ILD could be simply due to the less prominent hypersplenism related to portal hypertension, given a lower prevalence of liver cirrhosis and a trend to a higher platelet count in these patients.
It is well known that AMA-M2 is associated with a higher serum IgM in PBC [50][51][52]. It is unclear whether elevated serum IgM is an epiphenomenon or plays a role in PBC pathogenesis. Immune complexes containing IgM were found in the affected skin and kidneys of patients with PBC [53][54][55][56]. However, it is not known whether they can induce lung injury in these patients. Immune complexes containing IgG and complement C1q, but not IgM, were also detected in the pulmonary interstitium in a patient with PBC-associated ILD [57]. Moreover, immune complexes in PBC may be formed by AMAs and AMAantigens [58,59]. Further studies could shed light on the role of humoral immunity in PBC-associated ILD pathogenesis.
Data regarding the role of ANAs in the development of extrahepatic manifestations of PBC are scarce. Nickowitz et al. studied serum anti-gp210 and anti-Lamin B in patients with PBC and found no association with extrahepatic manifestations and associated CTD, except RA [60]. However, ILD was not mentioned in this study. Shen et al. found no significant difference in the seropositivity rates for all types of ANAs and anticardiolipin antibodies between groups of PBC patients with and without ILD [25]. Anti-gp210 was shown to be a potential predictor of adverse outcomes in PBC [61]. In our study, all PBCspecific ANAs, including anti-gp210, anti-sp100, anti-SS-A, and anti-PML, were found with similar frequency in patients with and without ILD. We suppose that these antibodies predominately target antigens of biliary epithelium and therefore are not involved in the pathogenesis of PBC-associated ILD.
The prognostic implication of ILD in PBC remains undetermined, although fatal cases were reported [62,63]. Only one of our 38 patients died from complications of progressive ILD, whereas most patients responded to glucocorticosteroids in combination with azathioprine or methotrexate and showed improvement or a stable course of lung disease. These findings suggest that ILD in PBC has a favorable prognosis if diagnosed and treated earlier.
Ten-year liver transplant-free survival in UDCA-treated PBC patients ranged from 63% to 94% [64]. Shen et al. did not find a correlation between ILD and the Mayo risk score (MRS) that is used to determine prognosis in PBC [25]. We did not assess MRS in our study. However, 10-year liver transplant-free survival in patients with ILD was higher than in patients without ILD (93.2% vs. 79.9%), although the predictive significance of lung disease was not confirmed by multivariate Cox regression analysis. Therefore, we cannot conclude that ILD was associated with better liver prognosis in PBC patients.
Our study has limitations. It was conducted in a single center, and the study sample was relatively small. The duration of follow-up was probably insufficient for adequate assessment of lung and liver-related outcomes. The number of lung-related outcomes was too small for survival analysis. Prospective multicentre cohort studies are needed to validate our results.

Conclusions
We present the results of the prospective single-center cohort study investigating the prevalence, clinical features, and risk factors for ILD in Russian patients with PBC. The diagnostic challenges of the diagnosis of asymptomatic PBC in unexplained ILD are highlighted. Apparently, PBC-associated ILD should be included in a list of differential diagnoses of ILD.
ILD was highly prevalent in our cohort (up to 41%). The most common HRCT patterns of ILD-PBC included sarcoid-like changes, followed by subclinical ILD and OP. The absence of liver disease symptoms at the disease presentation, the presence of hepatic nonnecrotizing epithelioid cell granulomas, higher serum IgM, and higher blood leukocyte count were independent risk factors for ILD. Most PBC patients showed radiological improvement of ILD after immunosuppressive therapy and a favorable short-term prognosis of ILD.
This study is the first to report the prognostic implication of ILD for the liver disease course in PBC. PBC-related ILD was associated with a lower liver disease severity, although it was not an independent factor for better liver-transplant-free survival. Informed Consent Statement: Informed consent was obtained from all subjects involved inthe study.

Data Availability Statement:
The datasets used during the current study areavailablefrom the corresponding author upon reasonable request.