A systematic review and meta-analysis of the endothelial-immune candidate biomarker endoglin in rheumatic diseases

Abstract

Existing challenges in accurately diagnosing various rheumatic diseases (RDs) have stimulated the search for novel biomarkers to aid clinical evaluation and monitoring. We conducted a systematic review and meta-analysis of studies investigating the candidate biomarker endoglin (CD105), a transmembrane glycoprotein expressed in endothelial, myeloid, and lymphoid cells, in RD patients and healthy controls. We searched PubMed, Scopus, and Web of Science from inception to 10 August 2024 to identify relevant studies. We evaluated the risk of bias using the JBI Critical Appraisal Checklist and the certainty of evidence using GRADE (PROSPERO registration number: CRD42023581008). Overall, circulating endoglin concentrations were significantly higher in RD patients compared to controls (13 studies; standard mean difference, SMD = 0.64, 95% CI 0.13 to 1.14, p = 0.014; low certainty of evidence). The effect size of the between-group differences in endoglin concentrations was not significantly associated with age, male-to-female ratio, year of publication, number of participants, or mean RD duration. By contrast, the effect size was statistically significant in studies conducted in the European region (p = 0.033), involving patients with systemic sclerosis (p = 0.032), and measuring serum (p = 0.019). The results of this systematic review and meta-analysis suggest the potential pathophysiological role of endoglin in RDs. This, however, requires further investigation in prospective studies, particularly in patients with systemic sclerosis.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10238-024-01519-5.

Introduction

Clinicians face significant challenges with diagnosing various rheumatic diseases (RDs), particularly in patients with non-specific signs and symptoms and uncertain imaging and laboratory findings [1–3]. This issue has significant implications as an accurate diagnosis facilitates early treatment with appropriate immunomodulatory and anti-inflammatory drugs, ultimately improving this group’s quality of life and long-term outcomes [4–7].

Endothelial dysfunction is a common feature of RDs with various degrees of autoinflammation and autoimmunity [8–12]. Endothelial cells in an activated state play a critical role in inflammatory states by regulating immune cell trafficking, activation, and extravasation into the parenchyma of specific tissues and organs [13–15]. Therefore, the availability of biomarkers of endothelial activation and dysfunction might help assess patients with suspected RDs and complement the information provided by clinical assessment and specific diagnostic criteria.

A relevant potential candidate biomarker is endoglin, also known as CD105, a membrane glycoprotein highly expressed in endothelial cells and various other myeloid and lymphoid cells, including B cells and T cells [16–21]. Although endoglin has been primarily studied as an accessory receptor for members of the transforming growth factor (TGF)-β family, increasing evidence suggests its role not only in angiogenesis-related pathologies (e.g. cancer) but also in other conditions characterized by dysregulated immunity and inflammation (e.g. atherosclerosis and cardiometabolic diseases) [22–24]. Notably, a soluble form of endoglin, measurable in serum or plasma, has been shown to be associated with several disorders, including preeclampsia [25], cerebrovascular disease [26], and non-alcoholic steatohepatitis [27]. However, the role of circulating endoglin as a biomarker of RDs has not been comprehensively appraised.

We sought to address this issue by conducting a systematic review and meta-analysis of studies reporting circulating endoglin concentrations in RD patients and healthy controls. Where possible, we also assessed possible associations between the effect size of the between-group differences in circulating endoglin and several study and patient characteristics, including the specific type of RD and the mean disease duration.

Methods

Literature search and study screening and selection

We systematically searched the electronic databases PubMed, Web of Science, and Scopus from inception to 10 August 2024 using the following terms (full details of the search in individual databases are provided in Supplementary Table 1): “endoglin” OR “sCD105″ AND “rheumatic diseases” OR “rheumatoid arthritis” OR “psoriatic arthritis” OR “reactive arthritis” OR “ankylosing spondylitis” OR “systemic lupus erythematosus” OR “systemic sclerosis” OR “scleroderma” OR “Sjogren’s syndrome” OR “connective tissue diseases” OR “vasculitis” OR “Behçet’s disease” OR “idiopathic inflammatory myositis” OR “polymyositis” OR “dermatomyositis” OR “gout” OR “pseudogout” OR”systemic vasculitis” OR “ANCA-associated vasculitis” OR “Takayasu arteritis” OR “polyarteritis nodosa” OR “osteoarthritis” OR “fibromyalgia” OR “granulomatous polyangiitis” OR”Henoch-Schonlein purpura” OR “Wegener’s granulomatosis” OR “familial Mediterranean fever” OR “polymyalgia rheumatica”.

Two investigators independently screened each abstract and, if relevant, the full text of the article. The inclusion criteria were: (1) the measurement of circulating endoglin, (2) the comparison between RD patients and healthy controls (case–control design), (3) the use of the English language, (4) the inclusion of at least ten RD patients and/or controls, and (5) the availability of the full-text of the publication. Exclusion criteria were: (1) cellular or animal studies, (2) the lack of a healthy control group, and (3) the inclusion of participants under 18 years. All articles’ references were hand-searched to identify additional studies.

Two investigators independently extracted the following data in separate electronic sheets: year of publication, first author, study country and continent, type of RD, mean RD duration, sample size, age, male-to-female ratio, endoglin concentrations, sample matrix (sample or plasma), and analytical method used. Any disagreement was resolved by a third investigator.

We assessed the risk of bias in each study using the Joanna Briggs Institute Critical Appraisal Checklist for analytical studies [28]. We assessed the initial and overall level of the certainty of evidence using the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) Working Group system, which considers the study design, the risk of bias, the presence of unexplained heterogeneity, the indirectness of evidence, the imprecision of results, the effect size, and the presence of publication bias [29]. We followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 statement (Supplementary Table 1) [30]. We registered the study protocol in an international registry (PROSPERO registration number: CRD42023581008).

Statistical analysis

We generated forest plots of standardized mean differences (SMDs) and 95% confidence intervals of endoglin concentrations between RD patients and healthy controls (a p-value < 0.05 was considered statistically significant). The medians and interquartile ranges were extracted from graphs using the Graph Data Extractor software (San Diego, CA, USA). We extrapolated the means and standard deviations from the medians and interquartile or full ranges [31]. We used the Q statistic (a p-value < 0.10 was considered statistically significant) to assess the heterogeneity of SMD across studies. I² values of ≤ 25%, > 25% and < 75%, and ≥ 75% indicated low, moderate, and high heterogeneity, respectively [32, 33]. We used a random-effect model based on the inverse-variance method in the presence of high heterogeneity [34]. We conducted sensitivity analyses to test the stability of the meta-analysis results and assessed publication bias using standard methods (a p-value < 0.05 was considered statistically significant) [35–38]. We conducted meta-regression and subgroup analyses to investigate associations between the effect size and year of publication, study continent, RD type, mean RD duration, sample size, age, male-to-female ratio, sample matrix assessed, and analytical method used. All statistical analyses were performed using Stata 14 (Stata Corp., College Station, TX, USA).

Results

Figure 1 describes the flow chart of study screening and selection. After initially identifying 780 articles, we excluded 762 because they reported duplicate or irrelevant information. After reviewing the full text of the remaining 18 articles, two were excluded because of missing data, two because they did not have a case–control design, and one because of duplicate data. The selected 13 studies, including 14 group comparators, investigated circulating endoglin in 1,080 RD patients (mean age 52.2 years, 87.0% females) and 386 healthy controls (mean age 50.5 years, 84.1% females; Table 1) [39–51].

Fig. 1.

PRISMA 2020 flow diagram of study screening and selection

Table 1.

Characteristics of the studies investigating endoglin patients with rheumatic diseases and healthy controls

Study	Controls				Patients with rheumatic diseases				Disease type	MDD (Years)
	n	Age (Years)	M/F	Endoglin (Mean ± SD)	n	Age (Years)	M/F	Endoglin (Mean ± SD)
Fujimoto et al. Japan [39]	20	matched	matched	6.27 ± 1.72	70	45	4/66	7.04 ± 2.02	SSc	6.3
Fujimoto et al. Japan [39]	20	matched	matched	6.27 ± 1.72	20	matched	matched	5.94 ± 1.65	SLE	NR
Wipff et al. France [40]	48	59.4	8/40	5.3 ± 0.9	187	55.9	30/157	6 ± 1.6	SSc	NR
Coral-Alvarado et al. Colombia [41]	20	NR	matched	0.76 ± 0.09	20	54.4	6/14	0.95 ± 0.24	SSc	9.35
Honsawek et al. India [42]	15	65.5	3/12	4.43 ± 1.16	39	67.8	5/34	5.16 ± 1.37	OA	NR
Robak et al. Poland [43]	20	38	2/18	5.43 ± 1.31	61	39	5/56	5.02 ± 1.08	SLE	5.33
Bassyouni et al. Egypt [44]	36	29.4	4/32	325.38 ± 66.59	87	30.1	9/78	324.58 ± 112.25	SLE	6.19
Avouac et al. France [45]	20	matched	matched	3.8 ± 0.48	60	54	14/46	3.73 ± 0.8	SSc	17.25
Ciurzyński et al. Poland [46]	21	49.3	3/18	3.95 ± 1.5	111	54.2	10/101	3.43 ± 0.97	SSc	9.4
Silva et al. Portugal [47]	34	47.1	5/29	0.38 ± 0.43	77	53	5/72	2.9 ± 3.06	SSc	12.5
Cossu et al. Italy [48]	43	matched	matched	686.9 ± 309	95	57.55	3/92	821.4 ± 330.7	SSc	4.5
Sodergren et al. Sweden [49]	40	48.1	8/32	11.6 ± 13.2	71	51.5	10/61	93.6 ± 12.5	RA	6.34
Lammi et al. USA [50]	18	NR	NR	364 ± 81	128	57.5	27/101	381 ± 101	SSc	NR
Hofstedt et al. Sweden [51]	31	60.1	6/25	2.86 ± 0.94	54	58.1	7/47	4.48 ± 4.1	RA	11

MDD, mean disease duration; M/F, male-to-female ratio; NR, not reported; OA, osteoarthritis; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; SSc, systemic sclerosis.

Eight studies were conducted in Europe [40, 43, 45–49, 51], two in Asia [39, 42], two in America [41, 50], and one in Africa [44]. Eight study groups included patients with systemic sclerosis (SSc) [39–41, 45–48, 50], three with systemic lupus erythematosus (SLE) [39, 43, 44], two with rheumatoid arthritis (RA) [49, 51], and one with osteoarthritis (OA) [42]. All studies measured endoglin in serum except two, which measured plasma [42, 44]. An enzyme-linked immunosorbent assay (ELISA) was used in all studies expect two, which used a platform for multi-analyte profiling [48, 51]. The mean RD duration, reported in ten studies [39, 41, 43–49, 51], ranged between 4.5 and 17.25 years.

We ranked the risk of bias as low in all studies except two, which had moderate risk [40, 48] (Supplementary Table 2). We ranked the initial level of the certainty of evidence as low (level 2) as all studies were cross-sectional.

The forest plot (Fig. 2) showed that RD patients had significantly higher circulating endoglin concentrations when compared to controls (SMD = 0.64, 95% CI 0.13 to 1.14, p = 0.014; I² = 93.7%, p < 0.001). The meta-analysis results were stable in sensitivity analysis, with pooled SMD values ranging between 0.26 and 0.70 (Fig. 3). However, the study by Sodergren A et al. had a distortive effect [49], as also shown in the funnel plot (Fig. 4). Its removal reduced the effect size, which, however, remained significant with a relatively lower heterogeneity (SMD = 0.26, 95% CI 0.02 to 0.50, p = 0.037, I² = 71.7%, p < 0.001).

Fig. 2.

Forest plot of studies investigating endoglin in patients with rheumatic diseases and healthy controls

Fig. 3.

Sensitivity analysis of the association between endoglin and rheumatic diseases

Fig. 4.

Funnel plot of studies investigating the association between endoglin and rheumatic diseases

We did not observe any significant publication bias according to Begg’s (p = 0.74) and Egger’s (p = 0.20) test. The “trim-and-fill” method did not identify any missing study required to ensure symmetry (Fig. 5).

Fig. 5.

Funnel plot of studies investigating the association between endoglin and rheumatic diseases using the “trimming-and-filling” procedure. Dummy studies and genuine studies are represented by enclosed circles and free circles, respectively

In univariate meta-regression analysis, we observed no significant associations between the effect size and age (t = 0.86, p = 0.41), male-to-female ratio (t =  − 0.86, p = 0.41), year of publication (t = 1.05, p = 0.32), number of participants (t = 0.08, p = 0.94), or mean RD duration (t =  − 0.52, p = 0.62). In subgroup analysis, the pooled SMD was statistically significant in European (SMD = 0.91, 95% CI 0.07 to 1.74, p = 0.033; I² = 96.4%, p < 0.001) but not Asian (SMD = 0.27, 95% CI − 0.15 to 0.69, p = 0.21; I² = 38.2%, p = 0.198) or American studies (SMD = 0.58, 95% CI − 0.28 to 1.44, p = 0.184; I² = 76.8%, p < 0.0010), with a lower heterogeneity in the Asian subgroup (Fig. 6). The pooled SMD was also significant in studies in SSc (SMD = 0.35, 95% CI 0.03 to 0.68, p = 0.032; I² = 75.9%, p < 0.001) but not SLE (SMD = -0.15, 95% CI − 0.43 to 0.13, p = 0.29; I² = 0.0%, p = 0.552) or RA patients (SMD = 3.44, 95% CI − 2.38 to 9.27, p = 0.25; I² = 99.2%, p < 0.001), with a virtually absent heterogeneity in the SLE subgroup (Fig. 7). In addition, the pooled SMD was significant in studies measuring serum (SMD = 0.71, 95% CI 0.12 to 1.30, p = 0.019; I² = 94.5%, p < 0.001) but not plasma (SMD = 0.22, 95% CI − 0.32 to 0.77, p = 0.42; I² = 57.4%, p = 0.125), with a lower heterogeneity in the plasma subgroup (Fig. 8). Finally, the pooled SMD was significant both in studies using ELISA (SMD = 0.68, 95% CI 0.07 to 1.30, p = 0.03; I² = 94.6%, p < 0.001) and a platform for multi-analyte profiling (SMD = 0.44, 95% CI 0.16 to 0.73, p = 0.002; I² = 0.0%, p = 0.806), with a virtually absent heterogeneity in the latter subgroup (Fig. 9).

Fig. 6.

Forest plot of studies investigating endoglin in patients with rheumatic diseases and healthy controls according to geographical area

Fig. 7.

Forest plot of studies investigating endoglin in patients with rheumatic diseases and healthy controls according to the type of rheumatic disease

Fig. 8.

Forest plot of studies investigating endoglin in patients with rheumatic diseases and healthy controls according to biological matrix used for measurement (serum or plasma)

Fig. 9.

Forest plot of studies investigating endoglin in patients with rheumatic diseases and healthy controls according to the analytical method used

The overall level of the certainty of evidence remained low (level 2) because of the low-moderate risk of bias in all studies (no change), the extreme but partially explainable heterogeneity (no change), the lack of indirectness (no change), the moderate effect size (SMD = 0.64, no change) [52], and the absence of publication bias (downgrade one level).

Discussion

In this systematic review and meta-analysis, we reported that RD patients have significantly higher circulating endoglin concentrations when compared to healthy controls. The effect size remained significant after removing a study exerting a distortive effect [49]. In meta-regression and subgroup analysis, the effect size of the between-group differences in circulating endoglin was not significantly associated with age, male-to-female ratio, year of publication, number of participants, or mean RD duration. By contrast, the effect size was significant in studies conducted in the European region, involving patients with SSc, and measuring serum endoglin. Further prospective studies are required to confirm these findings and evaluate the potential role of circulating endoglin as a candidate biomarker in SSc and a broad range of autoinflammatory and autoimmune RDs.

The initial studies investigating endoglin primarily focussed on its expression in activated endothelial cells in the context of physiological (e.g. developmental) and pathological (e.g. cancer-related) angiogenesis [53–55]. However, the discovery that this glycoprotein is also expressed in other cell populations has led to further studies investigating additional effects. For example, endoglin has been shown to play a role in the differentiation of monocytes into M1 macrophages, exerting pro-inflammatory effects through the release of specific cytokines, and M2 macrophages, exerting suppressing effects on inflammatory pathways. Studies have reported that the endoglin expression in M2 macrophages can downregulate c-myc, potentially transforming these cells into a M1 phenotype [56, 57]. More recently, studies have also investigated the role of endoglin in cells of the adaptative immune system. For example, endoglin expression has been reported in lymphocytes, primarily CD4 + T-cells, which are critical in favouring the onset and progression of autoimmune and autoinflammatory disorders [58–61]. Notably, the cross-linking of endoglin has been shown to enhance CD4 + T-cell proliferation and counteract the suppressive effect exerted by TGF-β [16]. The pathophysiological role played by endoglin in regulating angiogenesis, immunity, and inflammation can account, at least in part, for the observed elevations in patients SSc, an autoimmune disease state that is characterized by vascular dysfunction, ineffective angiogenesis, and systemic inflammation [62–64]. Further research is warranted to determine whether similar elevations occur in other RDs characterized by various combinations of vascular dysfunction, dysregulated angiogenesis, autoimmunity, and autoinflammation [65–69].

Another interesting finding in our subgroup analysis was the presence of significant between-group differences in circulating endoglin in studies conducted in the European region but not in the American or Asian regions, suggesting the potential influence of ethnicity or other geographical factors on endoglin concentrations. This issue has been investigated in other studies. For example, in pregnancy, black women had significantly higher endoglin concentrations than white women but lower concentrations when compared to Hispanic women [70]. Other studies have reported significant differences in the endoglin intron 7 insertion polymorphism between Japanese and white subjects [71]. Therefore, the role of geographical variations in modulating the association between endoglin and RDs warrants further investigation.

Our systematic review and meta-analysis has several strengths, including the comprehensive assessment of circulating endothelin in various types of RDs, the evaluation of associations between the effect size and various study and patient characteristics, and the rigorous assessment of the level of the certainty of evidence. Important limitations include the relatively small number of identified studies, particularly in non-SSc RD patients, and their cross-sectional design, which prevented assessing a cause-effect relationship between endoglin concentrations, clinical manifestations, and disease activity. Another limitation is related to the high heterogeneity observed in our primary analysis. However, several sources of heterogeneity were identified in subgroup analysis (i.e. geographical location, type of RD, biological matrix assessed, and analytical method used to measure endoglin).

In conclusion, our study suggests that endoglin is worthy of further investigation as a candidate biomarker of RDs, particularly in patients with SSc. Further prospective studies should investigate the diagnostic and prognostic accuracy of endoglin in a broad range of RDs, the potential influence of geographical location, biological matrix assessed, and analytical method used, and whether the information provided its measurement effectively complements that provided by clinical evaluation, imaging studies, and available biomarkers. Only then can the potential clinical use of measuring endoglin in RDs be justified.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Study conception: AZ, AAM; Data collection and analysis: AZ; Data interpretation: AZ, AAM; Writing—first draft: AAM; Writing—Review & Editing, AZ, AAM.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

The data supporting the findings of this systematic review and meta-analysis are available from AZ upon reasonable request.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethical approval

Ethics approval was not required as this was a systematic review of published studies.

Informed consent

N/A