Rheumatoid arthritis (RA) is a prevalent chronic rheumatic disease that can cause progressive articular damage and extra-articular manifestations, including pulmonary involvement, vasculitis, rheumatoid nodules, and systemic comorbidities (1). Interstitial lung disease (ILD) is the most common pulmonary manifestation, occurring in 10% of patients with RA (2). ILD is strongly associated with poor prognosis and is the common cause of RA mortality, secondary only to cardiovascular diseases (3, 4). A recent study shows that the 1- and 5-year mortality rates of rheumatoid arthritis-associated ILD (RA-ILD) were 13.9% and 39.0%, while these rates were 3.8% and 18.2% in RA patients without ILD (RA-nILD) (5). Moreover, despite decreasing mortality rates of RA, there has been an increase in deaths from RA-ILD (2), thus prompting a great effort to achieve an earlier diagnosis and improve the prognosis.

High-resolution chest computed tomography (HRCT) is accepted as the gold-standard approach for the diagnosis of ILD (6). Although showing abnormalities in 19% of patients with RA (6), HRCT of the lungs is not performed for routine RA diagnoses in consideration of radiation exposure and high cost (7), which is not conducive to early detection of RA-ILD. Early recognition and monitoring are crucial to provide better treatment and improve the prognosis. Given the impressive advances in the development of molecular detection techniques, the discovery and validation of disease-specific biomarkers could be of great clinical importance (8).

Antibodies and lung epithelium-specific biomarkers play a crucial role in predicting RA-ILD (9). Rheumatoid factor (RF) and anti-citrullinated peptide antibodies (ACPA), two serologic biomarkers for RA included in the European League Against Rheumatism (EULAR)/American College of Rheumatology (ACR) 2010 criteria (10), have been demonstrated to act as biomarkers of RA-ILD (11, 12). Other antibodies have also been proven to have strong associations with RA-ILD, such as anti-malondialdehyde-acetaldehyde (anti-MAA) (13) and anti-carbamylated protein (anti-CarP) (14). In addition, inflammatory mechanisms may cause repetitive injuries to alveolar epithelial cells (AECs), such as the apoptosis of ACE type I (ACE I) and the hyperplastic of AEC type II (ACE II). The regenerating ACE II could produce and recruit a series of cytokines and growth factors, and release surfactant proteins (15), which have potential predictive value. Factors like Krebs von den Lungen-6 (KL-6) (16), matrix metalloproteinase (MMP)-7 (17), and C-X-C motif chemokine 10 (CXCL10) (17) have been demonstrated to aid RA-ILD identification.

Chitinase-3 like-protein-1 (CHI3L1) plays a vital role in tissue repair, inflammation, and remodeling responses. It is synthesized and secreted by a multitude of cells including macrophages, neutrophils, synoviocytes, chondrocytes, fibroblast-like cells, and tumor cells (18–21). CHI3L1 has been regarded as a promising biomarker for predicting and evaluating the severity of RA (21, 22). In addition, serum CHI3L1 levels are significantly higher in ILD patients (23). It plays a profibrotic role in lung fibrosis repair phase by augmenting alternative macrophage activation, fibroblast proliferation, and matrix deposition (24). However, the associations between CHI3L1 and RA-ILD remain unknown.

The purpose of this study is to evaluate whether CHI3L1 can identify the presence of RA-ILD in RA patients. Additionally, we will assess its diagnostic capacity for ILD in patients with different disease activities of RA.

The demographic, clinical, and laboratory features of the 239 RA patients are summarized in Table 1 . Among all enrolled RA patients, 60 (25.1%) showed definite ILD features on chest HRCT. Compared with patients without ILD features, those with ILD were more likely to be older (median [IQR], 68.00 [62.00-71.75] vs 48.00 [37.00-57.00], p<0.001), have a higher proportion of males (21 (35.0%) vs 30 (16.8%), p=0.003), and have high titer positivity of RF or/and anti-CCP (55 [91.7%] vs 137 [76.5%]). Furthermore, the age of RA patients and CHI3L1 levels were moderately correlated, and the relationship was statistically significant revealed by Pearson Correlation Coefficient (r(237) = 0.323, p < 0.01). Different age groups were divided and patients older than 65 were more likely to have elevated CHI3L1 than 50-55 years old patients, as same as 60-65 and 30-40 years old patients ( Supplementary Figure 1 ). However, there was no statistical difference between gender and CHI3L1 levels ( Supplementary Figure 2 ). The presence of RA-ILD and DAS28 scores might have no relationship, neither DAS28-ESR (p=0.732) nor DAS28-CRP (p=0.202). In addition, the disease duration was similar in RA patients with and without ILD (median [IQR], 5.56 [1.06-18.31] vs 5.10 [1.68-10.16], p=0.548). Among the baseline therapies, patients with ILD were more likely to be treated with glucocorticoids (9 (90%) vs 85 (47.5%), p=0.010).

To our knowledge, this is the first study to analyze the association between CHI3L1 and RA-ILD. We found that higher levels of serum CHI3L1 were measured among patients with radiographic evidence of ILD features compared with RA patients without lung involvement. Additionally, older age and male sex were associated with RA-ILD. In particular, CHI3L1 exhibited the valuable diagnostic ability to identify the presence of ILD in patients with different disease activities of RA.

Survival of RA has been improved in the last few decades, along with the changes in treatment paradigms, early intervention, and treat-to-target approaches (29). The declined mortality rates are mostly due to a decrease in cardiovascular mortality while deaths from ILD remain significant (30). Considering that respiratory symptoms are not always apparent in RA-ILD (31), early detection and management are crucial to improving the prognosis and survival. The relationship between CHI3L1 and RA or ILD has been addressed in different studies (21, 22, 24, 32), while no association between CHI3L1 and RA-ILD has been confirmed.

Consistent with multiple studies (33, 34), the association of RA-ILD with older age and male sex is demonstrated in this study. Older age is associated with RA-ILD suggesting that immunosenescence and telomere shortening may be linked with RA-ILD risk (35). Notably, it seems somewhat paradoxical that RA is more prevalent in females than males, with a female-to-male sex ratio ranging from 4:1 to less than 2:1 in younger and older patients respectively (36). The distinctions might be explained partly by different exposure to environmental risk factors like smoking (37), which is involved in chronic mucosal inflammation of lung and plays a positive role in the development of RA-ILD (38). Some studies have demonstrated that smoking might induce citrullination of peptide antigens in the lungs via peptidyl arginine deiminase-2 (PAD2) (39, 40). Moreover, increased PAD2 was observed in the broncho-alveolar lavage (BAL) fluid of smokers compared to non-smokers (41). In addition, higher level of ACPA has been found among RA patients with ILD (12). It seemed that smoking might partly explain the higher proportion of male RA-ILD patients. Furthermore, several studies suggest that gonadal hormones (36), genetic background (42–44), and other confounding factors (44) also contribute to RA-ILD. The role of aging and gender in RA-ILD remains to be clarified. Traditional disease-modifying antirheumatic drugs (DMARDs) like methotrexate are the basis of therapeutic intervention in most patients with RA, while the effects on the presence of RA-ILD remain controversial (45). Treating with methotrexate seemed not to affect the arise of ILD but glucocorticoids did in this study.

The disease activity of RA determined by DAS28 appears to have no influence on the presence of RA-ILD, indicating the lack of association between the disease activity of RA and RA-ILD. In contrast, a cross-sectional study identified mean DAS28 as an independent predictor for preclinical RA-ILD (odds ratio (OR)=2.0, 95% CI [1.2-3.4]) (46). However, the sample size of this study is smaller than ours, which might contribute to the variance. A previous prospective cohort study conducted in the UK with 1460 RA patients involved shared similar results with ours that the DAS28 score at baseline had no association with RA-ILD (hazard ratio (HR)=1.11, 95% CI [0.94-1.31]) (47). The difference among these studies may come from differences in patient population, exclusions of confounding factors, and disease definition.

Mann-Whitney U test and AUC analysis showed a robust relationship between CHI3L1 level and the presence of RA-ILD, suggesting the potential role of CHI3L1 as a biomarker and its immune response on RA-ILD development. So far, no studies have evaluated CHI3L1 levels in RA-ILD patients while CHI3L1 have been detected in RA (22) patients and ILD patients (32). Several serum biomarkers have been proposed for RA-ILD diagnosis, including antibodies (ie. RF and ACPA) (11, 12) and a range of cytokines and growth factors released by active specialized macrophages and stromal cells (ie. MMP-7, surface protein D, and KL-6) (16, 17). Given the similarities in clinical, radiographic, and genetic characteristics shared by RA-ILD and idiopathic pulmonary fibrosis (IPF), several studies have noted that the biomarkers might repurpose from one to the other (48). For instance, serum MMP-7 and ACPA levels which are significantly higher in IPF than controls are also elevated in RA-ILD compared to those with RA without ILD (17, 49–51). Interestingly, the serum levels of CHI3L1 could distinguish between IPF and controls (245.8 ± 180.2 ng/mL vs 116.0 ± 58.3 ng/mL, p<0.001) and predict survival (52). Considering the biological activities of CHI3L1 determined in RA and hepatic fibrosis such as stimulating the growth of fibroblasts (19), driving macrophage activation and differentiation (53), and regulating the ECM (54), it may have a significant impact on the potential pathobiology of both IPF and RA-ILD. In addition, the data from our study suggested that CHI3L1 was valuable for the diagnosis of RA-ILD in all spectrums of disease activity based on the DAS28-ESR score, and in moderate and low disease activity groups based on DAS28-CRP. It seems that applying DAS28-ESR as the scoring standard may increase the accuracy of RA-ILD assessment.

Our study has some limitations. First, patients’ smoking history and specific autoantibodies concentration are not included; thus, these confounders may introduce bias to our results. Cigarette smoking could increase the risk of RA-ILD with a dose effect relationship (55). Second, this is a single-center study focused on RA patients, which limits the representativeness of the results. In the future, large sample and multi-center studies are needed to evaluate the link between CHI3L1 and RA-ILD. Third, several studies have shown that the efficiency of biomarkers varies in different disease statuses of RA-ILD (17, 56), while our study did not assess the severity of ILD. Moreover, the inclusion of RF, ACPA, and other potential biomarkers determined by previous studies tends to improve the accuracy of the prediction model (16, 17, 56). Additionally, comparing and contrasting the diagnostic ability of CHI3L1 with validated indicators such as KL-6 and HRCT scores might help verify its potency as a biomarker. Finally, there were no hold-out test sets or external validation cohorts.

In conclusion, an association between CHI3L1 and RA-ILD was found. It poses the debate of whether CHI3L1 could contribute to the development of this devastating extra-articular manifestation. Future studies will help define the potential underlying pathological mechanism of this connection.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Medical Ethics Committee of Peking Union Medical College Hospital (ethics number I-22PJ457). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

All authors were involved in the conception and design of this study. RYu, XL, XD, SL, and YW contributed to the performance of experiments. QW and XT assisted in the collection and preservation of blood samples. YZ, DK, and RYan collected and evaluated patient data. QW, XT, and ML assisted in the recruitment of patients. RYu and XL analyzed the data and wrote the first draft of the manuscript. XD, SL, and YW contributed to validation. YZ, DK, RYan, and ML assisted in supervision of general work. All authors contributed to manuscript revision, read, and approved the submitted version.