INTRODUCTION Periostin is an extracellular matrix protein that belongs to the Fasciclin family because of its homology to Fasciclin 1. Periostin is produced in mesenchymal lineage cells such as osteoblasts, periodontal ligament, and periosteum, and it is secreted into the extracellular space [1]. Periostin plays a role in forming and maintaining cellular structures by interacting with structural matrix proteins, including collagen [2]. Conversely, as a matricellular protein, it may also play other roles in regulating and modulating cellular functions through cross-talk with cell-surface receptors, proteases, and hormones [3,4]. The clinical utility of serum periostin was first reported in allergic diseases because of the induction of periostin gene expression by interleukin (IL)-4 and IL-13 [2,5]. Subsequently, many studies have reported the utility of periostin as a serum biomarker and therapeutic target in various chronic inflammatory and fibrotic diseases, such as liver and lung diseases [6]. In particular, periostin is involved in signalling pathways via nuclear factor kappa-light-chain-enhancer of activated B cells, IL-8, extracellular signal-regulated kinase, and mitogen-activated protein kinase in the pathophysiology of chronic kidney diseases and inflammatory bowel disease [7,8]. The possibility of the clinical utility of periostin in antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) with a similar pathological mechanism has been proposed [9,10]. To date, only one study that included patients with eosinophilic granulomatosis with polyangiitis (EGPA) has reported the clinical utility of serum periostin in AAV [11]. However, since the pathogenesis of EGPA is also partially based on a well-known allergic immunological mechanism, the previous study could not represent all cases of AAV with vasculitic immunological mechanisms and thus; may not provide new and additional information on pre-existing concepts [12]. To determine the clinical role of serum periostin in AAV, patients with three subtypes of AAV: microscopic polyangiitis (MPA), granulomatosis with polyangiitis (GPA), and EGPA were included in this study and the clinical utility of serum periostin in patients with AAV was evaluated. METHODS Patients One hundred patients with AAV were randomly selected from the Severance Hospital ANCA-associated VasculitidEs (SHAVE) cohort, a prospective and observational cohort of Korean patients with AAV, and retrospectively screened by reviewing their medical records. Among the 100 patients, 76 were included in this study based on the inclusion criteria: (1) fulfilment of the following three criteria for or definitions of AAV: the 2007 European Medicines Agency algorithms for AAV, the 2012 revised International Chapel Hill Consensus Conference Nomenclature of Vasculitides, and the 2022 American College of Rheumatology (ACR)/European Alliance of Associations for Rheumatology (EULAR) classification criteria for AAV [13–18]; (2) first classification of AAV at the Division of Rheumatology, Department of Internal Medicine, Yonsei University College of Medicine, and Severance Hospital, from November 2015 to June 2023; (3) completion of sufficiently well-written medical records to collect clinical, laboratory, radiological, and histopathological data for AAV classification at diagnosis and to obtain data for further assessment of the progression to all-cause mortality during follow-up [15–17,19]; (4) a minimum follow- up of six months; (5) completion of written informed consent at diagnosis; (6) availability of sera stored at diagnosis; (7) no concomitant serious medical conditions such as malignancies and severe infectious diseases at the time of AAV diagnosis [18]; (8) no exposure to immunosuppressive drugs within four weeks before AAV diagnosis.

This study was approved by the Institutional Review Board (IRB) of Severance Hospital, Republic of Korea (IRB No. 4-2016-0901). All patients in this study provided written informed consent at the time of being included in the SHAVE cohort (at the time of both AAV diagnosis and blood sampling). The IRB waived the need for additional written informed consent when it had been previously obtained at entry into the SHAVE cohort.

Clinical information, blood samples, and measurement of serum periostin Clinical data as described in Table 1 were collected. Explaining several important items, first, perinuclear (P)-ANCA and cytoplasmic (C)-ANCA were accepted as ANCA results in addition to myeloperoxidase (MPO)-ANCA and proteinase 3 (PR3)-ANCA according to the 2022 ACR/EULAR criteria for AAV [15–17]. Second, AAV-specific indices included the Birmingham Vasculitis Activity Score (BVAS), the Five-Factor Score (FFS), the 36-item short form survey physical and mental component summary (SF-36 PCS and SF-36 MCS), and the Vasculitis Damage Index (VDI) [20–23]. Third, type 2 diabetes mellitus, hypertension, and dyslipidaemia were recorded as comorbidities and as a part of the traditional risk factors for mortality [24]. Fourth, we investigated all-cause mortality as a poor outcome in AAV cases [25]. Although almost all patients died of two causes such as disease progression and infection, because it was impossible to clearly distinguish one from the other, we used the term, “all-cause mortality” in this study. We defined the follow-up duration based on all-cause mortality as the period from AAV diagnosis to death for deceased patients and as that from AAV diagnosis to the last visit for surviving patients. Finally, the number of patients who had ever received each medication during follow-up was counted.

Whole blood was obtained from patients with AAV on the day of the completion of written informed consent (the same day of AAV diagnosis). Sera were immediately isolated from whole blood and stored at −80°C. The concentration of serum periostin was measured using enzyme-linked immunosorbent assay kits (R&D Systems, Minneapolis, MN, USA) from collected and stored sera at diagnosis.

Statistical analyses

All statistical analyses were performed using SPSS Statistics for Windows, version 26 (IBM Corp., Armonk, NY, USA). Continuous and categorical variables were expressed as medians (25–75 percentiles), and numbers (percentages). Correlation coefficients (r) between the two variables were obtained using either Pearson correlation analysis or univariable linear regression analysis. The standardised correlation coefficient (β) was obtained by multivariable linear regression analysis using variables with statistical significance in univariable analysis. Significant differences between the two categorical variables were analysed using chi-square and Fisher’s exact tests. Significant differences between two continuous variables were compared using Mann–Whitney U test. A multivariable Cox proportional hazard model using variables with p < 0.1 in a univariable Cox analysis was performed to obtain a hazard ratio (HR) during follow-up. The significant area under the curve (AUC) was confirmed by performing a receiver operator characteristic (ROC) curve analysis. The optimal cut-off was extrapolated by performing ROC curve analysis and selected as one with the maximum sum of sensitivity and specificity the relative risk (RR) of the cut-off for all-cause mortality was analysed using contingency tables and chi-square test. A comparison of the cumulative survival rates between the two groups was performed using Kaplan–Meier survival analysis with the log-rank test. p values < 0.05 were considered statistically significant.

DISCUSSION

This study investigated the clinical utility of serum periostin in AAV and there were several notable findings. First, serum periostin at diagnosis was significantly correlated with cross-sectional AAV activity and acute-phase reactants. Additionally, serum periostin at diagnosis exhibited the potential as a predictor of all-cause mortality during follow-up in patients with AAV. In particular, clinical implication of this study is that this is the first to elucidate the clinical roles of serum periostin at diagnosis during the disease course of AAV.

We speculated that the mechanistic background enables serum periostin to play a crucial clinical role in patients with AAV. IL-4 and IL-13 have been reported to enhance gene expression and production of periostin, revealing the immunological mechanisms involved in the pathogenesis of asthma [2,5]. In addition, a previous study reported the clinical role of serum periostin in patients with EGPA, including an allergic component [11]. Therefore, based on these prior studies, we divided the patients into two groups; patients with MPA and GPA, and patients with EGPA, and predictions were made by comparing the variables between the two groups. First, the count of eosinophils at diagnosis may be higher in patients with EGPA than in those with MPA and GPA. As expected, patients with EGPA exhibited a higher median eosinophil count than those with MPA and GPA (280.0/mm3 vs. 90.0/mm3, p = 0.003). Second, serum periostin may be higher in patients with EGPA than in those with MPA and GPA. This is because periostin production is influenced by the eosinophil-specific cytokines, IL-4, and IL-13. However, in contrast to our expectations, patients with EGPA had a significantly lower median serum periostin than those with MPA and GPA (9.3 ng/mL vs. 11.7 ng/mL, pp = 0.040). Therefore, based on these results, it can be reasonably concluded that serum periostin in AAV, including EGPA, may be affected by signalling pathways other than those involving IL-4 or IL-13 [2,5,7,8]. Ideally, this issue should be clarified by investigating the intracellular signalling pathways involved in the crosslink between serum periostin and cross-sectional BVAS. However, because the cells that produce and secrete periostin or the tissues with these cells were no obtained from the patients included in this study, this proved impractical. Nevertheless, several inferences are made based on the results of multivariable linear regression analysis of the variables at diagnosis (Table 3). In multivariable linear regression analysis, the ability of serum periostin to independently reflect the current activity of AAV was proved to be comparable to that of VDI. Therefore, the first inference is that serum periostin may reflect cross-sectional BVAS by participating in intracellular signals related to the pathogenesis of AAV, that could induce damage in various major organs [23,26,27]. In addition, multivariable analysis revealed that the potential of serum periostin to independently estimate cross-sectional BVAS was not inferior to white blood cell count and serum albumin. Therefore, the second inference is that serum periostin may indirectly estimate cross-sectional BVAS by facilitating intracellular signals related to general inflammatory reactions [28]. These findings highlight that serum periostin is linked to intracellular signalling pathways directly and indirectly related to AAV, which we believe may represent a great advantage as a biomarker. In the present study, we found that serum periostin at diagnosis was significantly and independently correlated with cross-sectional BVAS in patients with AAV. We further investigated which of the nine systemic items of BVAS contributed to the observed correlation with serum periostin [20]. Among the items of BVAS, serum periostin was significantly correlated with general (r = 0.280, p = 0.014), pulmonary (r = 0.237, p = 0.039), renal (r = 0.530, pp = 0.033) manifestations (Supplementary Table 1). Additionally, we identified more detailed correlations between serum periostin and the subitems of each systemic item of BVAS as follows: among the subitems of general manifestations, serum periostin was significantly correlated with arthralgia/arthritis (r = 0.278, p = 0.015) and high fever (r = 0.276, p = 0.016). Among the subitems of pulmonary manifestations, serum periostin was significantly correlated with diffuse alveolar haemorrhage (r = 0.328, p = 0.004). Among the subitems of renal manifestations, serum periostin was significantly correlated with proteinuria > 1+ (r = 0.501, ppp = 0.006). However, among the subitems of neurological systemic manifestations, no correlation was observed between serum periostin and the subitems. Although limited information prevented further analysis, given that previous studies have reported an association between periostin and central neurological events, lung and kidney diseases, and arthritis [7,8], we believe that this result may be inferred to have some validity and may support the clinical utility of serum periostin in patients with AAV. The present study also investigated whether serum periostin at diagnosis has a predictive potential for all-cause mortality during follow-up in patients with AAV. We have provided a method to obtain the cut-off of serum periostin for all-cause mortality and demonstrated that patients with serum periostin exceeding the cut-off had a significantly increased risk of death and a decreased cumulative survival rate compared to those without (Fig. 1). However, we failed to demonstrate the independent ability of serum periostin at diagnosis for predicting all-cause mortality in patients with AAV in multivariable Cox proportional hazard analysis (Table 5). Nonetheless, since we found the clinical potential of serum periostin for mortality, we inferred how periostin could predict all-cause mortality through the results that serum periostin, along with the frequency of dyslipidaemia and the levels of VDI and CRP, was significantly higher in deceased patients than in surviving patients described in Table 4. First, in terms of dyslipidaemia as a conventional risk for mortality, in this study, patients having dyslipidaemia had a significantly higher serum periostin than those without (14.0 ng/mL vs. 10.7 ng/mL, p = 0.044). Therefore, it is inferred that serum periostin at diagnosis might have the predictive ability for all-cause mortality by interacting with the presence of dyslipidaemia [24]. Second, in terms of VDI as an AAV-specific risk for mortality, serum periostin exhibited a highly close correlation with cross-sectional VDI (Table 2). In the present study, VDI at diagnosis was defined as the first VDI assessing the items lasting for at least 3 months after the first clinical manifestation related to AAV. A recently published study demonstrated the independent predictive potential of the earliest VDI for all-cause mortality in patients with AAV [29]. Therefore, it is also inferred that serum periostin at diagnosis might have the predictive ability for death by borrowing the earliest VDI’s ability to predict all-cause mortality during follow-up in patients with AAV. Third, in terms of CRP as an inflammation-related risk for mortality, serum periostin was also significantly correlated with cross-sectional CRP levels (Table 2). Therefore, it is inferred that serum periostin at diagnosis might have the predictive ability for all-cause mortality during follow-up by being affected by the inflammatory burden at diagnosis [30].

The advantage of the present study is that this is the first to investigate the clinical perspectives involving serum periostin in patients with AAV and to demonstrate further that serum periostin at diagnosis could not only reflect cross-sectional AAV activity but also help to foresee all-cause mortality during follow-up. Therefore, as a pilot study, this study is believed to provide valuable information surrounding the clinical significance of serum periostin as a biomarker for AAV activity and prognosis.

The present study had certain limitations. First, although all study subjects were selected from the prospective and observational cohort of AAV patients, their clinical data were analysed retrospectively, and thus, posed difficulties in further analysis of several variables not included in this study. Owing to the characteristics of a pilot study, the number of enrolled patients was insufficient to generalise the results of this study and apply them to real-world clinical practice immediately. The most critical issue regarding this study might be the absence of mechanistic research and analysis of the intracellular signalling pathways linking serum periostin and both AAV activity at diagnosis and AAV-associated mortality during follow-up. Cross-sectional measurement of serum periostin at diagnosis might also be another limitation. We believe that a prospective future study that includes more patients and serially measures serum periostin will provide more reliable and dynamic information concerning the clinical perspective of serum periostin in patients with AAV not only at diagnosis but also during monitoring and follow-up periods.

In conclusion, this study is the first to demonstrate that serum periostin measured at diagnosis could independently reflect cross-sectional vasculitis activity at diagnosis and further contribute to the prediction of all-cause mortality during follow-up in patients with AAV. Additionally, this study also suggested that mechanisms underpinning the clinical roles of serum periostin might be linked to both intracellular signalling pathways directly and indirectly related to AAV, which may represent a great advantage as a biomarker.