CXCL9, CXCL10, and CXCL11; biomarkers of pulmonary inflammation associated with autoimmunity in patients with collagen vascular diseases–associated interstitial lung disease and interstitial pneumonia with autoimmune features

Introduction Interstitial lung disease (ILD) is a heterogeneous group of diseases characterized by varying degrees of lung inflammation and/or fibrosis. We investigated biomarkers to infer whether patients with collagen vascular diseases associated ILD (CVD–ILD) and interstitial pneumonia with autoimmune features (IPAF) benefit from immunosuppressive therapy. Materials and methods We retrospectively investigated patients with CVD–ILD, IPAF, and idiopathic pulmonary fibrosis (IPF) between June 2013 and May 2017 at our department. First, we assessed differences in serum and bronchoalveolar lavage fluid (BALF) levels of cytokines between groups. Second, we assessed the associations of patient’s clinical variables with serum and BALF levels of those cytokines that were different between groups. Finally, we assessed the associations of diagnosis and response to immunosuppressive therapy with serum levels of those cytokines that were different between groups. Results We included 102 patients (51 with IPF, 35 with IPAF, and 16 with CVD–ILD). Serum and BALF levels of CXCL9, CXCL10, and CXCL11 were significantly elevated in patients with IPAF or CVD–ILD compared with those in patients with IPF. BALF levels of CXCL9 and CXCL10 were correlated with the percentages of lymphocytes and macrophages in BALF. Serum levels of CXCL9 and CXCL10 were correlated with BALF levels. Serum levels of CXCL9, CXCL10, and CXCL11 were correlated C-reactive protein, percent predicted forced vital capacity, alveolar-arterial oxygen difference, and the percentages of lymphocytes and macrophages in BALF. Serum levels of CXCL9, CXCL10, and CXCL11 showed moderate accuracy to distinguish patients with CVD–ILD from those with IPAF and IPF. Pre-treatment serum levels of CXCL9 and CXCL11 showed strong positive correlations with the annual forced vital capacity changes in patients with IPAF and CVD–ILD treated with immunosuppressive drugs. Conclusions Serum CXCL9, CXCL10, and CXCL11 are potential biomarkers for autoimmune inflammation and predictors of the immunosuppressive therapy responses in ILD with background autoimmunity.


Introduction
Interstitial lung disease (ILD) is a heterogeneous group of lung diseases characterized by a combination of inflammation and fibrosis of the diffuse lung parenchyma. The clinical course and immunosuppressive therapeutic response vary substantially among the different ILD types. Idiopathic pulmonary fibrosis (IPF) is the most common ILD characterized by chronic progressive fibrosis and a poor prognosis [1]. In contrast, collagen vascular diseases (CVDs) are heterogeneous diseases characterized by systemic autoimmunity and varying degrees of inflammation and immune-mediated organ damage. Patients with CVD associated ILD (CVD-ILD) have a more favorable clinical course than those with IPF. Therefore, to distinguish IPF and CVD-ILD is important in clinical practice [1]. However, some patients with ILD have features of CVD, that do not meet the classification criteria for a defined CVD. Consequently, the European Respiratory Society/American Thoracic Society (ERS/ATS) proposed a new classification, interstitial pneumonia with autoimmune features (IPAF), to provide an initial framework for evaluating these patients and detecting those who may benefit from immunosuppressive therapy [2]. In clinical practice, the differential diagnosis of ILDs with background autoimmunity and other ILDs remains difficult, because it requires multidisciplinary discussion. Most CVDs can respond to immunosuppressive therapy, suggesting inflammation is a pathological mechanism of this disease. Cytokines are important for CVD pathogenesis, especially C-X-C motif chemokine (CXCL)9, CXCL10, and CXCL11 create local amplification loops responsible for sustaining inflammation in target organs [3]. These three chemokines are ligands for CXCR3 and recruit CD4+ Th1 cells and CD8+ T cells to the site of tissue damage and inhibit angiogenesis during pulmonary fibrosis [4]. Studies have shown that these three chemokines play roles in the inflammatory pathophysiology of CVD and ILD [3,[5][6][7]. However, the roles of autoimmunity and inflammation in the pathogenesis of CVD-ILD and IPAF remain unclear, and consensus guidelines for their diagnosis or treatment are not available. Therefore, biomarkers that can evaluate autoimmune inflammation and be useful for ILD diagnosis and management are needed [8][9][10].
We conducted a verifiable analysis based on serum and bronchoalveolar lavage fluid (BALF) CXCL9, CXCL10, CXCL11, and other cytokines in patients with CVD-ILD, IPAF, and IPF. We compared cytokines levels among patients with CVD-ILD, IPAF, and IPF. We then assessed the associations of CXCL9, CXCL10, and CXCL11, that were different between groups, with clinical characteristics, diagnosis, and treatment responsiveness to identify biomarkers of inflammation in ILD with background autoimmunity. examination, BGA, PFTs, and HRCTs 12 ± 3 months after the treatment. The Institutional Review Board of Sapporo Medical University Hospital approved this study (No. 282-160, approved on 12/13/2016).

Aim
The aim of our study was to identify biomarkers of inflammation in ILD with background autoimmunity for the diagnosis and treatment.

BAL processing
A flexible bronchoscope, wedged into a subsegmental bronchus of the right middle lobe or the left lingula, was used to infuse 50 ml of 0.9% saline at body temperature and to collect a lavage specimen by applying gentle suction. This process was repeated three times, and the BAL specimens were pooled together.

Statistical analysis
To verify biomarkers reflect inflammation in ILD with background autoimmunity, we assessed differences in characteristics and cytokines levels between the CVD-ILD, IPAF, and IPF groups using the Kruskal-Wallis test with the Steel-Dwass post-hoc test, or Fisher's exact test with the post-hoc Benjamini-Hochberg test. We expressed continuous variables as medians with interquartile ranges and categorical variables as numbers or percentages. To assess the associations of CXCL9, CXCL10, and CXCL11 with the clinical characteristics of ILD, we calculated the correlations between serum and bronchoalveolar lavage fluid (BALF) levels of these chemokines and patient's clinical variables using the Spearman's rank correlation coefficient. To evaluate the value of serum CXCL9, CXCL10 and CXCL11 as diagnostic biomarkers for ILD with background autoimmunity, we constructed receiver operating characteristic (ROC) curves. To evaluate the value of serum CXCL9, CXCL10 and CXCL11 as predictive biomarkers of treatment responsiveness of ILD with background autoimmunity, we calculated correlations between pre-treatment serum levels of these chemokines and the annual forced vital capacity (FVC) changes using the Spearman's rank correlation coefficient. We performed the Benjamini-Hochberg test using the R software (R Development Core Team, Vienna, Austria) and other analyses using the JMP version 10.0 software (SAS Institute, Cary, NC, USA). We set statistical significance at p < 0.05.

Associations between serum and BALF cytokines levels
S2      Table 5 shows the results of the ROC curve to assess the diagnostic value of serum CXCL9, CXCL10, and CXCL11 for the differential diagnosis of the CVD-ILD, IPAF, and IPF groups. To distinguish CVD-ILD from IPAF and IPF groups, serum CXCL9, CXCL10, and CXCL11 showed moderate accuracy with AUC ranged 0.72-0.90. To distinguish IPAF from IPF groups, serum CXCL10 and CXCL11 levels showed low accuracy with AUC ranged 0.67-0.68.

Discussion
This study demonstrated that in patients with CVD-ILD and IPAF with background autoimmunity, serum and BALF levels of CXCL9, CXCL10, and CXCL11 were significantly higher than those in patients with IPF. The serum levels of CXCL9 and CXCL10 were significantly associated with the BALF levels. The serum levels of CXCL9, CXCL10, and CXCL11 were correlated with increases in BALF lymphocytes%, CRP levels, and A-aDO 2 . In addition, BALF levels of CXCL9 and CXCL10 were correlated with increases in A-aDO 2 and BALF lympho-cytes%. Serum CXCL9, CXCL10, and CXCL11 showed moderate accuracy to distinguish CVD-ILD from IPAF and IPF. Pre-treatment serum levels of CXCL9 and CXCL11 showed strong positive correlations with the annual FVC changes among patients treated with immunosuppressive drugs. These findings suggest that CXCL9, CXCL10, and CXCL11 may be biomarkers of autoimmune inflammation in patients with CVD-ILD and IPAF. We showed that the serum levels of CXCL9, CXCL10, and CXCL11 and the BALF levels of CXCL9 and CXCL10 were higher in patients with CVD-ILD and IPAF than in those with IPF. CXCL9, CXCL10, and CXCL11 are involved in the pathogenesis of CVD-ILD and ILD as follows: In patients with PM/DM and SSc, the serum CXCL10 levels are higher in patients with ILD than in those without ILD [19,20]. In the case of PM/DM-ILD, patients with anti-Jo-1 antibody showed higher serum CXCL9 and CXCL10 levels than patients with IPF [21]. In terms of RA-ILD, serum CXCL9, CXCL10, and CXCL11 are increased and induce CXCR3+ T cells in the lung [22,23]. In ILD, CXCL9, CXCL10, and CXCL11 levels are involved in inflammation through the induction of CXCR3+ T cells [5,6]. In IPAF, diagnosis requires patients presenting ILD, not meeting the CVD classification criteria, and having at least one feature from 2 of the 3 domains (clinical, serologic, and morphologic) [2]. IPAF has been shown to manifest with clinical extrathoracic autoimmune features in 47.3% to 62.5% of patients [8]. These results suggest that CXCR3+ T cells related inflammation is activated in the lungs and systemically in patients with CVD-ILD and IPAF, respectively.
This study showed that in patients with CVD-ILD, IPAF, and IPF, the BALF levels of CXCL9 and CXCL10 were correlated with increases in A-aDO 2 and BALF lymphocytes% and that serum levels of CXCL9, CXCL10, and CXCL11 were correlated with increases in BALF lymphocytes%, CRP levels, and A-aDO 2 . We consider that these three chemokines may induce pulmonary inflammation through lymphocyte induction that lead to alveolitis resulting diffusion impairment in patients with CVD-ILD and IPAF. In particular, the serum CXCL9 and CXCL10 levels in our patients showed correlations with BALF levels so that these serum chemokines may reflect not only systemic inflammation but also local pulmonary inflammation. On the other hand, BALF CXCL11 levels, unlike CXCL9 and CXCL10, were similar among CVD-ILD, IPAF, and IPF patients and had no correlations with BALF or serum levels. This may be due to the insufficient measurement sensitivity for CXCL11 because the BALF levels of CXCL11 were below the detection limit in many patients. As the levels of CXCL9, CXCL10, and CXCL11 differed between the CVD-ILD, IPAF, and IPF groups, we expected the correlations between these chemokines and characteristics may differ between groups. However, because of retrospective study, we were unable to adjust pulmonary function and retrieved rates of BALF between the groups and could not assess differences in the correlations caused by background autoimmunity in a subgroup analysis (S3 Table).
Our AUCs for serum CXCL9, CXCL10 and CXCL11 in the differential diagnosis of CVD-ILD from IPAF and IPF suggest that these serum chemokines may be diagnostic biomarkers of ILD with autoimmunity. In addition, we showed that patients with CVD-ILD and IPAF treated with immunosuppressants had higher pre-treatment serum levels of CXCL9 and CXCL11 and greater improvements in annual FVC after treatment. Studies have also suggested associations between the disease activity of autoimmune diseases and serum or BALF CXCL9, CXCL10, and CXCL11 levels in patients with SLE, DM, or SScs [7,20,24,25]. However, the clinical management of CVD-ILD and IPAF remains difficult due to a lack of accurate diagnosis, disease activity, and therapeutic response markers [2,8]. Our results suggest that high serum CXCL9, CXCL10, and CXCL11 levels may reflect reversible inflammation and that these chemokines are predictive biomarkers of the response to immunosuppressive therapy in ILD with background autoimmunity. Especially, serum CXCL9 may be a novel independent biomarker because it does not correlate with clinical ILD biomarkers, surfactant protein (SP)-A, SP-D, or KL-6 (S4 Table) [26]. In the future, monitoring of these serum biomarkers from the time of diagnosis could be useful in determining whether immunosuppressive therapy is initiated without BAL. In particular, when ILD worsens during follow-up, these biomarkers may enable clinicians to choose anti-inflammatory drugs or anti-fibrotic drugs.
We found that the cytokines profile of IPAF was similar to that of CVD-ILD. However, we found a difference in terms of the serum levels, especially serum CXCL9, CXCL10, and CXCL11 levels in the IPAF group were intermediate between those in the IPF and the CVD-ILD group. Autoimmune inflammation in CVD-ILD occurs systemically. In comparison, in IPAF, pulmonary inflammation and some extrathoracic autoimmune features that are not enough for a CVD diagnosis occurs [8,10]. Thus, serum CXCL9, CXCL10, and CXCL11 levels may reflect the activities of autoimmune inflammation and the existence of autoimmune features.
We are aware of the limitations of our study. First, we studied only a small number of patients at a single center. The number of CVD-ILD patients was relatively small, and CVD included different diseases. Second, many patients in the IPF group had mild diseases with small decrease in FVCs. Japanese clinicians actively perform HRCTs in patients suspected to have ILDs, because anti-fibrotic therapies for IPF are effective during the early stages of the disease. Consequently, our cohort included many early stage IPF patients. Furthermore, these 3 chemokines showed no differences depending on the severity of IPF (S5 Table). Third, we evaluated only a few treated patients. Twenty-seven of the 41 patients in the CVD-ILD and IPAF groups were treated with immunosuppressive drug, but only 10 of them had pre-treatment serum samples and respiratory function test data one year after treatment due to the retrospective study nature. Further studies are needed to evaluate the utility of CXCL9, CXCL10, and CXCL11 as biomarkers of treatment responsiveness. Fourth, we could not confirm the presence of CXCR3+ T cells in lung tissues, so our results are based on indirect proof. Ideal biomarkers should be easily and repeatedly collected.

Supporting information
Takatsuki, Japan) for support of statistical analysis, and Enago (www.enago.jp) for the English language review.