Introduction

The immune system is complicated, far-reaching, and difficult to predict. Multiple interconnected systems join together to maintain a perfect balance between physiological response and pathological autoimmunity. If not, the autoimmune disorders and consequent organ damage will follow. Hitherto, the exact causes of approximately 50% autoimmune diseases are still unknown [1]. Systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA) are both typical chronic inflammatory autoimmune diseases. SLE or lupus is a progressive and lifelong autoimmune disorder, gradually damaging multiple organs including bones, muscles, heart, kidney, nerves, and blood [2,3,4,5] with a broad range of clinical manifestations, such as fatigue, fever, anorexia, dermopathy, systemic infection, glomerulonephritis or nephrotic syndrome symptoms, myalgia or myasthenia, pericarditis or pericardial effusion, and neuronal damage [5]. RA is a long-lasting inflammatory joint disease. If failure to control the disorder progression, the cartilage and bone will suffer heavy damage and eventually patients in RA even become disabled. At the outset, RA involves the joints. Subsequently, extra syndromes will occur, manifestations including rheumatoid nodules, pulmonary vasculitis, and the like. For now, because of the complicated pathogenesis of SLE and RA, it is difficult to identify and clarify accurate etiology, especially in the case of unhealable infection and electrolyte disturbance [6].

The etiology of autoimmune diseases is very complex, but the activation of autoimmune T cells and B cells is an inevitable part. Cytokines, including chemotactic factor synthesized and secreted by various immune or non-immune cells, are involved in a variety of immune responses. CXC chemotactic cytokine subfamily mainly acts on neutrophils and has a chemotactic effect on lymphocytes; CXC ligand 13 (CXCL13) is one of which, also known as B cell chemotactic factor (BLC). CXCL13 promotes the migration and aggregation of B lymphocytes by interacting with its receptor CXCR5, leading to the formation of germinal centers. Meanwhile, the germinal centers promote the further differentiation of B lymphocytes into plasma cells to produce corresponding antibody mountains. Thus, CXCL13 and CXCR5 play a key role in the differentiation of B lymphocytes into plasma cells to generate antibodies. The occurrence of SLE and RA is likely to be related to the over expression of CXCL13 and CXCR5 by generating an over-strong immune auxiliary effect. These discoveries provide strong evidence that CXCL13 will become an effective target for the treatment of SLE and RA [7, 8]. In the previous several decades, several studies have contributed to studying the CXCL13 expression level in many autoimmune diseases, especially in SLE and RA. Increased circulating CXCL13 levels in RA and SLE patients have been reported. Nevertheless, these reported results were individualized. Although there were three reviews having reported the association of circulating CXCL13 level with SLE [9,10,11], there is no quantitative analysis like a meta-analysis. In order to systematically quantify and evaluate the relationship between the circulating CXCL13 levels in SLE and RA, this article will employ a meta-analysis to acquire a more accurate and convincing estimation of the contact between circulating CXCL13 levels and SLE and RA.

Materials and methods

Search strategy

We searched PubMed, Web of Science, and China National Knowledge Infrastructure respectively to derive relevant articles up to 10 April 2019. “B lymphocyte chemoattractant,” “CXC ligand 13,” “CXCL13,” “BLC,” “BCA-1,” “Systemic Lupus Erythematosus,” “SLE,” “lupus,” “lupus nephritis,” “Rheumatoid Arthritis,” and “RA” were all used as keys terms to look for related and eligible articles, increasing the opportunity for searching out more articles. For fear that we might miss some suitable articles, the references of the selected articles were screened carefully to ensure the reliability and validity of this meta-analysis.

Inclusion criteria and exclusion criteria

Articles included in this meta-analysis were bound to meeting the following requirements: (a) case-control or cross-sectional studies; (b) studies based on adults; (c) SLE and RA patients must conform to the American College of Rheumatology (ACR) criteria or European League Against Rheumatism (EULAR) or Systemic Lupus International Collaborating Clinics (SLICC); (d) complete serum or plasma CXCL13 concentrations of both patients and controls were available; e) if one article has published more than one time, we only used the latest edition.

Articles excluded in this meta-analysis are as follows: (a) reviews; (b) subjects are not humans; (c) lack of healthy controls. Detailed flow chart of article inclusion and exclusion process was presented in Fig. 1.

Fig. 1
figure 1

Flowchart of selected articles

Full size image

Data extraction

From the eligible articles, the information was presented and listed as below: first author’s name and year of publication, region and ethnicity, kind of disease, language, original indicators, number of subjects, the mean age (± standard deviation), sex ratio, SLEDAI or disease activity score 28 (DAS28) score, duration years, assay method, study type, criteria for the classification of SLE or RA, and the mean CXCL13 concentrations (± standard deviation). It is worth noting that when the original full text only included median and inter-quartile range or median and range, we transformed the given data to figure out useful values [12]. On the basis of Newcastle-Ottawa scale (NOS), we assessed the methodological quality of all the selected articles. When divergence appeared, we strived to reach an agreement after panel discussion.

Statistical analysis

For each article, the standardized mean difference (SMD) and 95% confidence intervals (95% CI) were elaborately shown by forest plot. The formula I2 = ([Q−df]) × 100% was used to quantify the effect of heterogeneity once more. Assuming that the heterogeneity is significant (P < 0.05), the random effects model was adopted to pool the SMD value; or else, the fixed-effect model was adopted. In order to evaluate the publication bias, Egger’s and Begg’s tests were used. Stata 12.0 software was used in this meta-analysis to perform all statistical analysis.

Results

Study characteristics

A total of 461 articles were retrieved from PubMed, Web of Science, and China National Knowledge Infrastructure (CNKI); 435 articles were filtered out for the following reasons: duplicates, reviews, non-human subjects, and not SLE or RA. After carefully examining each article’s title, abstract, and its full text, 15 eligible studies were included from the remaining 26 articles. Eventually, 981 SLE patients and 380 healthy controls as well as 332 RA patients and 147 healthy controls were chosen in this meta-analysis (Fig. 1). All of the 15 studies included were case-control studies; all of the diagnostic criteria for SLE and RA was ACR. Some studies also adopted EULAR or SLICC as criteria. The basic features of articles needed in this meta-analysis were shown in Table 1. Extracted data on circulating CXCL13 levels of 15 studies in this meta-analysis was presented in Table 2.

Table 1 Characteristics of included studies
Full size table
Table 2 Extracted data on circulating CXCL13 levels of 15 studies in this meta-analysis
Full size table

Quality evaluation of included articles

A total of 15 articles were included in this meta-analysis, all of which were case-control studies. Of the 15 articles, 9 were SLE cases and 6 were RA cases. In these studies, the serum or plasma CXCL13 concentrations were all tested by ELISA. Studies included were almost gender and age matched. NOS scores of 15 studies are presented in Table 1.

Meta-analysis results

All the meta-analysis results of SLE and RA were respectively presented in Table 3. The forest plots visually showed the significant heterogeneity (Figs. 2 and 3).

Table 3 Subgroup analysis of CXCL13 levels in SLE and RA
Full size table
Fig. 2
figure 2

Forest plot of 9 studies in circulating CXCL13 levels for SLE patients versus healthy controls, based on random effects model

Full size image
Fig. 3
figure 3

Forest plot of 6 studies in circulating CXCL13 levels for RA patients versus healthy controls, based on random effects model

Full size image

Analysis in SLE

Among the included 9 studies, a significant statistical heterogeneity was discovered, which cannot be neglected (P < 0.001, I2 = 98.4%) (Table 3). Consequently, the random effects model was adopted for data analyzing, demonstrating that circulating CXCL13 concentrations in patients were significantly higher than that in control groups (SMD = 1.851, 95% CI 0.604–3.098) (Fig. 2). For the sake of exploring the exact sources of heterogeneity, subgroup analyses were implemented. When classified by language and SLEDAI score, the Chinese group and SLEDAI score ≥ 6 group all had higher circulating CXCL13 levels respectively (SMD = 2.182, 95% CI 0.135–4.229; SMD = 0.767, 95% CI 0.503–1.030). When stratified by ethnicity, sample size, and original indicators, all patients in SLE showed increased circulating CXCL13 levels (SMD = 2.175, 95% CI 0.540–3.809; SMD = 0.746, 95% CI 0.451–1.041; SMD = 2.336, 95% CI 0.544–4.128; SMD = 0.882, 95% CI 0.576–1.187; SMD = 3.090, 95% CI 0.437–5.743; SMD = 0.842, 95% CI 0.619–1.066). However, there were no significant differences in the English group and SLEDAI score < 6 group (Table 3). In order to investigate the publication bias, Egger’s test and Begg’s test were both performed simultaneously. (Egger’s t = 0.34, P = 0.744; Begg’s z = 1.56, P = 0.118), indicating there was no publication bias.

Analysis in RA

Obviously, the heterogeneity among the 6 studies was found to be of great difference (P < 0.001, I2 = 86.5%) (Table 3). Therefore, random effects model was chosen, ensuing that similar results were acquired exactly as that analyzed in RA: higher circulating CXCL13 concentrations in RA patients (SMD = 1.801, 95% CI = 1.145–2.457) (Fig. 3). Accordingly, subgroup analyses, stratified by ethnicity, sample size, and original indicator, were also performed to explore the latent confounders. Results showed that RA patients in all groups had higher circulating CXCL13 concentrations (SMD = 1.105, 95% CI 0.841–1.369; SMD = 2.645, 95% CI 1.503–3.788; SMD = 1.384, 95% CI 0.838–1.929; SMD = 2.355, 95% CI 0.685–4.026; SMD = 1.962, 95% CI 1.175–2.749; SMD = 1.105, 95% CI 0.499–1.712) (Table 3). In order to ensure the accuracy, we used Egger’s test and Begg’s test to examine publication bias. Egger’s t = 2.23 P = 0.090; Begg’s z = 1.50, P = 0.133 showing no publication bias.

Discussion

CXCL13 is a B cell chemokine, which was first discovered in 1998. It is one of the members of the chemokine CXC family. Current studies at home and abroad have shown that CXCL13 plays an important role in autoimmune diseases, especially in SLE and RA [28,29,30]. In 2001, Japanese scholars Ishikawa et al. [31] found that CXCL13 levels in the kidney of lupus BWF1 mice in the elderly group were higher than that in the young group. It was the first time connecting CXCL13 with SLE. In 2009, some German scholars Schiffer et al. [13] respectively detected the serum concentrations of CXCL13 in 91 SLE patients and 40 normal controls with the results showing that the serum CXCL13 concentrations in SLE patients were significantly higher, especially patients with LN. Moreover, CXCL13 level had a positive correlation with SLEDAI score and level of antibody against double-stranded (ds) DNA, which were statistically verified [9]. In 2015, the experiment of Wu et al. [32] in vivo found the decreased urinary protein, serum creatinine, and anti-dsDNA as well as the decreased deposition of kidney immune complex and secretion of inflammatory factor (IL-1, IL-6, IL-17, IL-33) in MRL/lpr mice after vaccinating, which fully indicating CXCL13 participated in the pathological process of SLE, and the application of anti-CXCL13 antibody might have certain therapeutic effect [29]. Based on these studies, CXCL13 is expected to become a new biomarker for SLE, even may reflect the degree of disease activity and renal involvement [32]. Edwards et al. [33] have proven that B cell–targeted therapy was highly effective to prevent disease progression in the treatment of RA. Rituximab was a successful example, contributing to managing 80% RA patients safely. CXCL13 is a critical B lymphocyte chemoattractant, playing vital role in the formation of B cell follicles in the lymphoid tissues. CXCL13 may serve as a novel biomarker to predict diseases severity. Jones et al. [34,35,36,37] also illustrated a similar perspective.

The pathogenic mechanism of SLE and RA is complex and influenced by multiple etiology, such as environment, genetics, epigenetics, endocrine, hormone, and individual susceptibility [1, 38, 39]. At present, 40% RA patients still have not received regular and reasonable treatment in China [40]. Various studies have reported that chemokine CXC family and corresponding receptors played a key role in the disease manifestation and progression. As it is well-known, SLE and RA are featured by the elevated production of auto-antibodies, which were generated long before clinical symptoms, resulting in more chronic and serious consequences. During the last decades, more than 150 auto-antibodies have been found in SLE, especially anti-dsDNA antibody, a sensitive biomarker to detect disease activity in SLE [9]. Different studies have indicated that Sjogren’s syndrome (SS), adult-onset still’s disease [41], SLE, and RA have association with members of CXC chemokines family and related receptors. Although CXCL13 has not been used in clinical practice, it cannot be denied that it is a very potential and promising biomarker. If ruling out serious infection and other diseases, CXCL13 is of great value for the diagnosis and prognosis of LN patients. In addition, anti-CXCL13 antibody can block CXCL13 and CXCR5 interaction and signaling pathway, which may provide a new therapeutic target for LN patients [10]. Synovitis is a common clinical manifestation in RA patients, characteristic of the formation of lymphoid follicles. In this process, CXCL13 plays a critical role by migrating B cells and T follicular helper cells to the follicle; thus, it can be shown that CXCL13 is expected to become a new biomarker for the early RA [34]. But without ignorance, CXCL13 plays a key role in the differentiation of B lymphocytes into plasma cells to generate antibodies. When generating an abnormal immune auxiliary effect, autoimmune diseases will occur consequently. Although this study proved that circulating CXCL13 levels may be of great value to the diagnosis of RA and SLE, we still need to pay attention to the differential diagnosis of autoimmune diseases.

In the meta-analysis, there were 15 articles selected in this study eventually, including 9 articles related to SLE and 6 articles related to RA. We estimated the association of circulating CXCL13 level with SLE and RA. The final results turned out to correspond with the original assumptions that there was correlation between CXCL13 levels and SLE (SMD = 1.851, 95% CI 0.604–3.098) as well as RA (SMD = 1.801, 95% CI = 1.145–2.457), suggesting that SLE and RA patients were significantly increased in circulating CXCL13 levels. Although Schiffer et al. [9], Li et al. [10], and Worthmann et al. [11] have never discussed this topic before, they just generalized and summarized rather than providing accurate quantitative description. In this study, the potential sources of heterogeneity were further explored due to strong heterogeneity. Therefore, subgroup analyses were performed to evaluate the exact sources of heterogeneity, and to minimize the potential confounding factors. For patients with SLE, we found that when classified by language and SLEDAI score, both the Chinese group and SLEDAI score ≥ 6 group had higher circulating CXCL13 levels. However, there were no significant differences in the English group and SLEDAI score < 6 group. This finding suggested that ethnicity and SLEDAI score may be associated with circulating CXCL13 concentrations in SLE. For patients with RA, subgroup analyses, stratified by ethnicity, sample size, and original indicator, suggested that patients with RA in all subgroups had higher circulating CXCL13 concentrations. Furthermore, due to the limited data included in the selected studies, only five studies related to SLE were stratified by SLEDAI and only one study with the original indicator of mean (95% confidence interval), which may have a negative impact on the final results. Considering renal involvement is a common clinical manifestation of SLE, lupus nephritis was also included in this study. It is worth noting that both Egger’s test and Begg’s test were implemented to prove no publication bias.

Admittedly, this article had some unavoidable limitations. First, we only retrieved three databases so the literature included might be not complete enough. Second, patients with LN or without LN might have a significant difference in CXCL13 concentrations; we could not guarantee that all articles included paid attention to this key point. Third, due to the lack of explicit data, we were incapable of analyzing the relationship DAS28 disease duration, some clinical symptoms and environmental characteristics with circulating CXCL13 levels of patients with SLE and RA, which might exactly be the sources of heterogeneity. Fourth, part of the data used in meta-analysis was approximate, resulting in a deviation. Fifth, the results of the pooled SMD in SLE and RA patients could rather be conservative because of using the random effects model.

In spite of these disadvantages above, this article also had advantages. First, it was the first comprehensive meta-analysis to clarify the relationship of CXCL13 concentrations with SLE and RA. Second, we included diverse literature from different countries, such as Egypt, America, Germany, and China. Third, no publication bias has been found, suggesting that there was no bias in selected articles.

Conclusions

In summary, circulating CXCL13 level is higher in patients with SLE or RA, which may be prospective to become a novel auxiliary biomarker. More importantly, a new therapeutic target inspiring from CXCL13 nosogenesis can be detected and put into clinical practice.