Biomarker profiles of endothelial activation and dysfunction in rare systemic autoimmune diseases: implications for cardiovascular risk

Abstract

Objectives

Vasculopathy is an important hallmark of systemic chronic inflammatory connective tissue diseases (CICTD) and is associated with increased cardiovascular risk. We investigated disease-specific biomarker profiles associated with endothelial dysfunction, angiogenic homeostasis and (tissue) inflammation, and their relation to disease activity in rare CICTD.

Methods

A total of 38 serum proteins associated with endothelial (dys)function and inflammation were measured by multiplex-immunoassay in treatment-naive patients with localized scleroderma (LoS, 30), eosinophilic fasciitis (EF, 8) or (juvenile) dermatomyositis (34), 119 (follow-up) samples during treatment, and 65 controls. Data were analysed by unsupervised clustering, Spearman correlations, non-parametric t test and ANOVA.

Results

The systemic CICTD, EF and dermatomyositis, had distinct biomarker profiles, with ‘signature’ markers galectin-9 (dermatomyositis) and CCL4, CCL18, CXCL9, fetuin, fibronectin, galectin-1 and TSP-1 (EF). In LoS, CCL18, CXCL9 and CXCL10 were subtly increased. Furthermore, dermatomyositis and EF shared upregulation of markers related to interferon (CCL2, CXCL10), endothelial activation (VCAM-1), inhibition of angiogenesis (angiopoietin-2, sVEGFR-1) and inflammation/leucocyte chemo-attraction (CCL19, CXCL13, IL-18, YKL-40), as well as disturbance of the Angiopoietin-Tie receptor system and VEGF-VEGFR system. These profiles were related to disease activity, and largely normalized during treatment. However, a subgroup of CICTD patients showed continued elevation of CXCL10, CXCL13, galectin-9, IL-18, TNFR2, VCAM-1, and/or YKL-40 during clinically inactive disease, possibly indicating subclinical interferon-driven inflammation and/or endothelial dysfunction.

Conclusion

CICTD-specific biomarker profiles revealed an anti-angiogenic, interferon-driven environment during active disease, with incomplete normalization under treatment. This warrants further investigation into monitoring of vascular biomarkers during clinical follow-up, or targeted interventions to minimize cardiovascular risk in the long term.

Introduction

Vasculopathy is an important hallmark of many chronic inflammatory connective tissue diseases (CICTD) affecting the skin. The vasculopathic component is well documented in (juvenile) dermatomyositis ((J)DM), SLE and scleroderma/ SSc [1–4]. Increasing evidence suggests that vasculopathic changes can also be present, although to a lesser extent, in rare scleroderma-spectrum disorders such as localized scleroderma (LoS), eosinophilic fasciitis (EF) and mixed connective tissue disease (MCTD) [5–7]. Endothelial dysfunction has important clinical implications in CICTD: it is associated with severe complications such as skin ulceration, renal, cardiac and pulmonary involvement [2–4, 8]. Moreover, especially in systemic CICTD vasculopathy contributes to morbidity and mortality through accelerated atherosclerosis, leading to an increased risk of cardiovascular events [2, 9, 10]. In LoS, a more localized CICTD with little systemic inflammation, the risk of cardiovascular disease is not increased [10]. For patients with EF, a disease characterized by severe systemic inflammation, long-term implications for cardiovascular risk are still unknown, but may be considerable.

Although multiple pathophysiologic events play a role in the development of vasculopathic changes, chronic immune activation has emerged as an important contributing factor. Deposition of complement and immune complexes and direct immune cell-mediated endothelial injury can cause endothelial loss [11]. Moreover, serum factors in CICTD can directly affect endothelial function: patient sera can induce adhesion molecule expression on cultured endothelial cells, while reducing angiogenesis, normal capillary morphogenesis, endothelial proliferation, migration and tube formation [12, 13]. Anti-endothelial antibodies present in these sera and a disturbed balance between angiogenic and angiostatic stimuli may contribute to vasculopathic changes [3, 4]. Interferon-driven inflammation in particular has been linked to the development of vasculopathic changes through direct and indirect angiostatic effects [2, 14, 15]. Notably, overexpression of type I IFN is observed in many CICTD [1, 16, 17]. A classic example of an angiogenic system that may be disturbed during inflammation is the Angiopoietin-Tie receptor system [18, 19]. The vascular endothelial growth factor (VEGF) system may also be affected in inflammation, and elevated VEGF is considered a biomarker for disturbed angiogenesis [3, 4]. Finally, soluble adhesion molecules (e.g. ICAM-1 and VCAM-1) are considered reliable and representative markers for endothelial activation [3, 4].

Thus, in many IFN-driven CICTD vasculopathy may play an important role in the disease pathology. However, it is yet unclear whether disturbances of circulating factors associated with endothelial dysfunction or activation show overlapping or distinct patterns between diseases, and whether they are related to disease activity. Here, we investigated disease-specific biomarker profiles associated with endothelial dysfunction, angiogenic homeostasis and (tissue) inflammation in a unique set of treatment-naive patients with rare systemic and localized CICTD involving skin and/or muscles (LoS, EF, (dermato)myositis and mixed connective tissue disease) as well as healthy controls and neuromuscular control patients. We relate these biomarker profiles to disease activity and persistent low-grade inflammation, which may have implications for the long-term cardiovascular risk.

Methods

Participants

Patients with (juvenile) dermatomyositis ((J)DM), mixed connective tissue disease with myositis (MCTD), localized scleroderma (LoS), eosinophilic fasciitis (EF), as well as healthy controls (HC) and neuromuscular control patients with hereditary proximal spinal muscular atrophy (SMA, a progressive, non-inflammatory neuromuscular disorder with a suspected vasculopathic component due to impaired endocytosis) [20], were recruited between January 2006 and June 2017 in tertiary referral centres in the Netherlands, Singapore and France. This study was approved by the institutional ethics committees of the involved centres [UMC Utrecht (NL47875.041.14, NL13046.091.06), UMC Utrecht-Radboud UMC (NL43660.041.13), AMC Amsterdam, SingHealth centralized IRB (CIRB2014/083/E), CHU Strasbourg] and conducted according to the Declaration of Helsinki. Age-appropriate written informed consent was obtained prior to study inclusion.

Disease classification and disease activity criteria

Patients with JDM were included if they met the Bohan and Peter criteria for definite or probable JDM [21, 22]. As clinical measures of muscle and global disease activity, the childhood myositis scale (CMAS; 0–52) and physician’s global assessment (PGA; 0–10) were recorded. Clinically inactive disease was defined according to the updated criteria for JDM [23, 24]; all other patients were considered active. Adult patients with dermatomyositis were classified according to the ENMC criteria [25]. Myositis was confirmed by biopsy unless typical skin manifestations of dermatomyositis were present. Patients with cancer-associated myositis were excluded. Disease activity was determined by combined evaluation of muscle strength with the medical research council scale, skin symptoms and muscle enzymes. Patients with myositis and presence of anti-RNP antibodies fulfilling the Kasukawa criteria were classified as having mixed connective tissue disease [26, 27]. Patients with localized scleroderma were diagnosed based on the typical clinical picture. Treatment-naive patients were considered active and had an mLoSSi and PGA > 5 [28]. In paired follow-up samples, inactive disease was defined as a modified LoS Skin Severity Index (mLoSSi) ≤5 out of 162. Adult patients with eosinophilic fasciitis were diagnosed based on the clinical picture and histopathological evaluation of skin biopsies containing the fascia. As the mLoSSi may stay high in these patients due to extensive irreversible sclerosis despite improved inflammatory symptoms, inactive disease was defined as a PGA for activity ≤5 out of 100 [28]. Paediatric and adult patients with hereditary proximal spinal muscular atrophy served as disease controls. Confirmation of a homozygous loss of function of the survival motor neuron 1 gene was obtained in all SMA patients [29]. Adult healthy volunteers were included as healthy controls.

Biomarker analysis

Blood was collected in serum tubes, according to the local study protocol. All samples were spun down within 4 h after collection and stored at −80°C until analysis. A total of 38 analytes were measured in 50 μL of serum by multiplex technology in all samples simultaneously, as described previously (xMAP; Luminex [30]): angiopoietin-1 (Ang-1), angiopoeitin-2 (Ang-2), angiopoietin-1 receptor (Tie-2), CXCL13, CCL2, CCL4, CCL17, CCL18, CCL19, CCL22, CCL27, CXCL9, CXCL10, CXCL12, endoglin, E-selectin (E-sel), fetuin, fibronectin, galectin-1 (Gal-1), galectin-3 (Gal-3), galectin-9 (Gal-9), IL-18, oncostatin M (OSM), periostin (OSF-2), placental growth factor (PlGF), plasminogen activator inhibitor (PAI-1), platelet-derived growth factor BB (PDGF-BB), P-selectin (P-sel), serum amyloid A1 (SAA-1), SPARC, soluble ICAM-1 (ICAM-1), soluble VCAM-1 (VCAM-1), soluble VEGF receptor 1 or Flt-1 (sVEGFR1), thrombomodulin (TM), thrombospondin-1 (TSP-1), TNF receptor-2 (TNFR2), TWEAK, vascular endothelial growth factor (VEGF) and YKL-40. Tie-1 was measured separately by ELISA assay (R&D, DY5907).

Statistical analysis

Statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, La Jolla, California, USA), SPSS Statistics 21 (IBM, Armonk, New York, USA) and R 3.5.1 (CRAN, Vienna, Austria). Multiplex values below the detection limit were imputed as 0.5× the lowest measured value. For correlations between biomarkers and clinical disease activity measures, assessed by Spearman rank, imputed biomarker values were excluded. To correct for multiple testing, Kruskal–Wallis tests with Dunn’s post hoc test were performed for each disease separately, stratified by activity (i.e. treatment-naive, active on medication, inactive and HC). Multiplicity adjusted P-values of these tests are reported in the tables and legends. For comparison between SMA and HC, the Mann–Whitney U test was used with correction for multiple testing by false discovery rate (FDR). For paired analyses, the Wilcoxon matched-pairs signed rank test with FDR correction was used. Multiplicity adjusted P-values or FDR <0.05 were considered statistically significant as indicated in the figure legends. For unsupervised clustering by principal component analysis (PCA) and heatmap analysis with hierarchical clustering by 1-Pearson correlation with average linkage, data were mean-centered and scaled per analyte. Analytes with >30% of measured values below the detection limit were excluded from the clustering analyses (OSM and CXCL9). K-means clustering was performed based on median analyte expression within the groups.

Results

Patient characteristics

We included 146 unique CICTD patients with (J)DM (72), MCTD with myositis (1), LoS (55) or EF (18), 43 control patients with SMA and 22 HC (Tables 1 and 2). Among CICTD patients, 72 samples were taken before start of treatment and 119 during treatment and/or inactive disease (details in Table 2). The age at sampling differed significantly between the treatment-naive disease groups (P <0.0001), as most myositis patients were juvenile, whereas the majority of other patients were adults. The gender distribution was similar between groups. Concordant with disease phenotypes, patients with EF had significantly higher VAS disease activity and mLoSSI scores than patients with LoS (P =0.0017 and P =0.0031). In 7/18 (39%) EF patients, clinical overlap with LoS was shown at any of the sampling time points. In 3/7 patients, a morphea en plaque phenotype was seen, in 3/7 a generalized pattern, and in one patient a generalized pattern at first and morphea en plaque later. Muscle enzymes were highest in the myositis group.

Unsupervised clustering reveals distinct biomarker profiles in treatment-naive myositis, EF and LoS

A panel of 38 markers for endothelial (dys)function and inflammation was measured in serum of treatment-naive patients and controls by multiplex immunoassay. Unsupervised PCA showed that patients with (J)DM, EF and LoS clustered separately based on their biomarker profiles (Fig. 1A), with (J)DM and EF patients being most distinct from HC. One patient with MCTD-associated myositis clustered close to the (J)DM cluster, suggesting a similar biomarker profile despite the distinct disease background. Patients with LoS and SMA had biomarker profiles similar to HC, indicating low/absent systemic inflammation. A total of 30 proteins were differentially expressed between at least one disease and HC (FDR < 0.05; Supplementary Table S1, available at Rheumatology online). Seven of these were increased in all three CICTD: Gal-9, CXCL10, TNFR2, IL-18, CXCL13, CCL19 and VCAM-1 (Fig. 1B). (J)DM and EF additionally shared upregulation of Ang-2, CCL2, ICAM-1, YKL-40 and sVEGFR1. CXCL9 and CCL18 were significantly increased in LoS and EF. Random forest analysis showed that Gal-9, CCL18 and CXCL10 were the most important analytes to distinguish the CICTD from HC and each other (Fig. 1C). To identify disease-specific analytes signatures, we performed K-means clustering, which revealed seven analyte clusters (Supplementary Fig. S1, available at Rheumatology online). Analyte cluster 1, with the highest expression in (J)DM, contained Gal-9, CCL2, PlGF and sVEGFR1 (Fig. 1D). Clusters 2 and 3, with similarly elevated levels in (J)DM and EF, consisted of CXCL10, VCAM-1, Ang-2, TNFR2, CCL19, OSF-2, IL-18, CXCL13, VEGF and YKL-40 (Fig. 1E and F). Analytes in cluster 4, CCL18, Fetuin, CXCL9, CCL4, ICAM-1 and Gal-1 were specifically high in EF (Fig. 1G). Clusters 5–7 showed mixed expression patterns across groups, with high Gal-3 in (J)DM, high fibronectin and TSP-1 in EF and low SPARC in (J)DM and EF.

These results suggest that patients with LoS, a localized CICTD, have a limited systemic biomarker signature, whereas patients with the systemic CICTD (J)DM and EF have clear and distinct, but also overlapping biomarker profiles. In both diseases, IFN-related and chemo-attractant proteins (CXCL10, TNFR2, IL-18, CXCL13, CCL19, CCL2, YKL-40), endothelial activation markers (ICAM-1, VCAM-1) and the anti-angiogenic Ang-2 and sVEGFR1 were highly expressed, whereas Gal-9 and CCL18, among others, were more disease-specific.

Relation of biomarker profiles with disease activity

To assess whether the biomarker profiles were related to disease activity and would normalize during treatment, we analysed HC and CICTD patients before start of treatment (treatment naive, TN), during active disease on medication (AM) and during clinically inactive disease by PCA. The biomarker profiles were most distinct from HC before treatment and became more similar to HC during treatment and subsequent inactive disease (Fig. 2A–C), suggesting that the biomarker profiles are highly associated with disease activity. In LoS, these differences were small, again indicating that the systemic effects are limited in this rather localized disease. To investigate which of the biomarkers were most related to disease activity, we performed three separate analyses in each CICTD: we analysed differential biomarker expression 1) between treatment-naive and inactive disease and 2) between longitudinal paired samples taken during active disease and after follow-up, and 3) assessed correlations with clinical measures of disease activity. Results of these separate analyses are shown in Supplementary Tables S2–S4, available at Rheumatology online, and summarized in the Venn diagrams in Fig. 2D. In (J)DM, 13 markers were related to disease activity in all three analyses: Gal-9, CXCL10, TNFR2, CCL2, ICAM-1, VCAM-1, OSF-2, YKL-40, sVEGFR1, CCL27, CCL19, Ang-2 and Gal-1 (Fig. 2D). In EF, eight markers were related to disease activity in all analyses, showing considerable overlap with the biomarkers in (J)DM: CCL18, TNFR2, CCL2, ICAM-1, OSF-2, CCL19, Gal-1 and Tie-1. In LoS, only CCL18 was identified in all three analyses. The heatmaps in Fig. 2E and paired samples in Fig. 2F–H (EF, (J)DM and LoS, respectively) show the considerable reduction of these core biomarkers from active to inactive disease.

In conclusion, many of the disease-specific analytes, including vasculopathy-associated markers, are related to disease activity and decrease to near-normal levels during treatment.

Homeostasis of angiogenic systems

To identify disturbances in the angiogenic homeostasis, we focused on the two important angiogenesis-regulating Ang-Tie and VEGF-PlGF-VEGFR systems. Under homeostatic conditions, the activation of the Tie-1/Tie-2 receptor complex by its ligand Ang-1 is key for endothelial proliferation and survival. Under inflammatory conditions, however, this interaction is antagonized on multiple levels by (i) cleavage of Tie-1, destabilizing the receptor complex; (ii) induction of the antagonistic alternative Tie receptor ligand Ang-2; and (iii) release of the soluble decoy receptor Tie-2. These inflammation-induced changes can compromise vascular integrity (Fig. 3A) [18, 19]. In active (J)DM and EF, Ang-2 levels were more than 2-fold increased compared with inactive disease, whereas Ang-1 levels were not significantly changed (Fig. 3B and C). The Ang-2/Ang-1 ratio, representing the balance of angiostatic over angiogenic signals, was significantly elevated in active (J)DM (P <0.0001); however, unaffected in EF and LoS (Fig. 3D). Soluble, cleaved Tie-1 was higher during active than inactive disease (P <0.05; Fig. 3B and C), especially in (J)DM and EF. These results indicate that in active (J)DM, and to a lesser extent in EF, homeostatic Ang-Tie signalling is compromised. The Ang-2/Ang-1 balance in these diseases is shifted towards an anti-angiogenic environment, which is further aggravated by destabilization of the Tie-receptor complex (Fig. 3A and B) [18, 19].

In the VEGF system, the VEGF receptor ligands VEGF and PlGF induce pro-angiogenic signalling, which may be antagonized by the soluble decoy receptor sVEGFR1. sVEGFR1 was significantly increased during active (J)DM and EF compared with controls (P <0.01; Fig. 3B and C). VEGF was increased during active EF (P =0.0323) and PlGF was increased during active (J)DM (P <0.0001). These disturbances normalized during inactive disease (Fig. 3B). In LoS, none of these components were affected. Taken together, in active (J)DM and EF the Ang-Tie system is shifted towards an anti-angiogenic balance and the VEGF-PlGF-VEGFR system is disturbed.

Biomarker normalization during inactive disease

In (J)DM and EF, biomarker profiles of patients with clinically inactive disease were more similar to HC than of patients with active disease (Fig. 2A and B). To examine whether biomarker profiles of patients completely normalized during clinically inactive disease, we compared clinically inactive patients to HC by PCA. Except for one JDM patient experiencing a flare after 2 months, all of these patients stayed in remission for >5 months after sampling. Patients with inactive disease were still distinct from HC (Fig. 4A). We observed remarkably elevated levels of the 12 analytes shared between (J)DM and EF in Fig. 1B: Gal-9, CXCL10, IL-18, TNFR2, CXCL13, CCL19 and VCAM-1 were still elevated in a subgroup (of up to 59%) of patients in both diseases (Fig. 4B). Ang-2 was rather decreased (Fig. 4B). In addition, patients with inactive (J)DM had significantly lower CCL27, Fetuin and P-sel than HC (Supplementary Table S5, available at Rheumatology online). A total of 50% of patients with inactive EF had increased fibronectin, and overall Gal-3 and SPARC were lower than in HC. This indicates that (a subgroup of) patients with clinically inactive CICTD can still have abnormal levels of markers indicating interferon-driven inflammation, leucocyte chemo-attraction, endothelial activation or angiogenic disturbance, possibly indicating continued subclinical inflammation and/or endothelial dysfunction. The observed differences were not due to medication effects, as we did not find differences between inactive patients on and off treatment (data not shown).

Discussion

Vasculopathy is an important hallmark of many CICTD affecting the skin. It is associated with chronic inflammation and can lead to severe complications, including an increased risk of cardiovascular events [2–4, 8, 9]. To date, it is unknown how biomarker profiles associated with vasculopathy compare between different CICTD. Here, we are the first to show that different interferon-associated, treatment-naive CICTD can be separated based on their biomarker signatures related to both inflammation and endothelial dysfunction/activation. (J)DM and EF, characterized by systemic inflammation, showed clear systemic biomarker signatures, with disease-specific markers such as Gal-9 and CCL18. In LoS, a more localized disease with little systemic inflammation, subtle biomarker changes were observed as well. Upregulated markers shared among diseases were related to interferon (CXCL10, CCL2), endothelial activation (VCAM-1), inhibition of angiogenesis (angiopoeitin-2, sVEGFR1) or inflammation/leucocyte chemo-attraction (CCL19, OSF-2, CXCL13, IL-18 and YKL-40). Remarkably, a subgroup of CICTD patients showed continued biomarker disturbances during clinically inactive disease, indicating that an anti-angiogenic interferon-dominated environment and endothelial activation may linger subclinically and possibly affect long-term outcome and cardiovascular risk.

Elevated levels of IFN-inducible chemokines CXCL9, CXCL10 and CCL2, which serve as potent chemo-attractants promoting leucocyte recruitment to inflamed tissues, were present in the three different CICTD, supporting previous studies [1, 17, 31]. We found higher circulating levels in (J)DM and EF compared with LoS, possibly reflecting the compartmentalization of the diseases. These chemokines can be produced within DM muscle and LoS skin, and their overexpression in DM muscle correlates with the severity of vasculopathy, such as loss of capillaries [17, 32]. Type I IFN can directly exert an angiostatic effect on endothelial cells [33], or indirectly via induction of e.g. CXCL9 and CXCL10 [32]. Moreover, the type I IFN signature has been associated with endothelial (progenitor cell) dysfunction and a higher risk of cardiovascular events in CICTD [2, 14, 15]. Gal-9 and CXCL10 have been recently validated as biomarkers for (J)DM and in the current study high levels of Gal-9 were found to be (J)DM specific [31]. Although the effects of Gal-9 on endothelial cells are largely context-dependent, at higher concentrations direct angiostatic effects have been demonstrated [34]. Thus, the observed presence of interferon-related biomarkers during active and even inactive disease may indicate that an anti-angiogenic environment is present in patients with CICTD for a prolonged period.

Also, IL-18, one of the core cytokines upregulated in all three CICTD, has been shown to induce endothelial progenitor cell dysfunction in SLE [35], correlates with the severity of coronary atherosclerosis in the general population [36], and is predictive of cardiovascular mortality [37]. Because we measured total IL-18, we cannot be certain that individuals with high IL-18 also had high free IL-18 (not bound to IL-18 binding protein). However, it has been shown that the pattern of total IL-18 levels and free IL-18 levels correlate strongly across various autoimmune diseases, implying that patients with high total IL-18 are likely to have high free IL-18 levels as well [38, 39]. In line with previous observations, soluble adhesion molecules (ICAM-1 and VCAM-1) were increased in active (and inactive) DM, EF and to a lesser extent in LoS [40, 41], reflecting increased surface expression on endothelial cells as described in DM muscle and skin of SSc patients [4, 42, 43]. Elevated adhesion molecule expression may enhance leucocyte migration into tissues, perpetuating and/or aggravating local (chronic) inflammation. Increased serum ICAM-1 and/or VCAM-1 are related to coronary artery calcification, and manifest or future cardiovascular disease and cardiovascular mortality in SLE [2], and an increased risk of clinical cardiovascular disease and atherosclerosis in the general population [44]. These observations imply that prolonged overexpression of soluble adhesion molecules in CICTD patients may reflect an increased cardiovascular risk in the long term.

We observed that the Angiopoietin-Tie receptor system was disturbed in active (J)DM and EF, with high anti-angiogenic Ang-2, disrupting homeostatic Ang/Tie signalling. Also, others have found high Ang-2 in (J)DM, which decreased during treatment [45]. In SSc, similarly increased Ang-2 levels were described, being highest in patients with advanced capillary damage [3, 4].

Levels of VEGF, PlGF and sVEGFR1 were increased during active (J)DM and EF, which normalized during treatment. High levels of sVEGFR1 are disruptive for angiogenesis [46]. High serum VEGF, normalizing during treatment, was previously reported in CICTD [2–4, 47, 48]. Although high levels of the pro-angiogenic molecule VEGF may seem beneficial for angiogenesis, higher VEGF levels have been associated with more severe vasculopathy: increased VEGF correlated with higher intima media thickness in SLE [2], with lower capillary density in SSc [3, 4], and with pulmonary arterial hypertension, acrosclerosis and myositis in MCTD [47]. In (J)DM, increased VEGF expression in inflamed muscle and vasculitic lesions correlated with the degree of angiopathy [48, 49]. Similarly, high PlGF was related to the development of digital ulcers in SSc [3]. Taken together, this suggests that in CICTD, vasculature is either unresponsive to increased VEGF or PlGF, that VEGF or PlGF upregulation may be an insufficient compensatory mechanism, or that vascular morphology becomes disturbed due to prolonged overexpression [3, 4, 32].

Together, our results suggest that the identified biomarker profiles reflect on multiple levels a systemic environment of disturbed endothelial function associated with vasculopathy and increased current and/or future cardiovascular risk, especially in CICTD with systemic inflammation. This may have important implications for EF, as in these patients an increased risk of cardiovascular events has not been investigated but may be considerable. In LoS, the more localized nature with little systemic disturbances may translate into the absence or only limited presence of an increased cardiovascular risk [10].

This study has several strengths. Most importantly, samples were taken before the start of treatment to capture disease-specific biomarker signatures free of medication effects. All markers and samples were measured simultaneously, enabling us to study their interrelation in the relevant angiopoietic systems. We were able to include relatively large numbers of treatment-naive patients, considering the rarity of the diseases, although for some (e.g. EF) the numbers were still small. Moreover, we collected (paired) follow-up samples to study the development of biomarker profiles during treatment and inactive disease. It would be interesting to follow up the relationship between prolonged elevation of interferon- and vasculopathy-related markers during clinically inactive disease and the cardiovascular risk profile, especially in patients with EF for which these implications are still unknown. The results of this study have to be interpreted keeping in mind the observational nature of the cohort, with differing sample numbers between diseases, and unmatched controls. Based on these biomarker data in serum, we can only speculate whether these biomarker profiles are similarly represented in the tissues, as within affected tissues, biomarker concentrations may significantly differ from circulating concentrations.

We observed inflammatory and anti-angiogenic biomarker disturbances not only in treatment-naive patients with (J)DM and EF, but also during treatment and, less pronounced, during clinically inactive disease. This suggests that even with well-controlled symptoms, these biomarkers may reflect subclinical inflammation or (mild) endothelial dysfunction. If so, there may be a window of opportunity to support future treat-to-target treatment strategies with biomarker profiling, to achieve not only clinical, but also ‘molecular remission’. Such a definition could be useful in long-term clinical follow-up of both inflammatory and vasculopathic disease aspects. Moreover, it may warrant the consideration of a targeted treatment to reduce the anti-angiogenic component in CICTD. For instance, vitamin D has been proposed to restore myeloid angiogenic cell function by reducing the expression of CXCL10 [50]. Direct targeting of the anti-angiogenic interferon signature by JAK-inhibition or anti-IFN monoclonal antibodies could also be considered [45, 51]. Although endothelial dysfunction can be reduced by control of inflammation [52], conventional and more general immunosuppressive therapy which is currently used, often combining prednisone, methotrexate, HCQ or other DMARDs, may not sufficiently target the anti-angiogenic IFN signature. Other considerations to reduce vasculopathy and concurrent cardiovascular risk in patients with CICTD may be to ‘hit hard’ at onset of disease, reducing the duration of active disease as the serum evidence of endothelial dysfunction is most pronounced in this period. Future studies will have to point out whether a treat-to-target approach for the vasculopathic component of CICTD may be beneficial in the long term, by monitoring endothelial dysfunction with established biomarkers for vasculopathy, also in patients with clinically inactive disease [2, 4]. Lastly, we identified several new biomarkers that highly correlate with clinical disease activity in (J)DM, LoS or EF, including Ang-2, ICAM-1, Gal-1, TNFR2, CCL2, CCL19, OSF-2 and CCL18 [53]. These may potentially serve as novel monitoring tools for disease activity in clinical follow-up.

In conclusion, we have identified disease-specific biomarker profiles in CICTD, which demonstrated an interferon-driven anti-angiogenic environment conducive to leucocyte recruitment to inflamed tissues. These biomarker profiles were related to disease activity, but did not completely normalize during clinically inactive disease. These findings warrant future studies into monitoring of biomarkers for inflammation and endothelial dysfunction during clinical follow-up of patients, possibly supporting a treat-to target approach to minimize cardiovascular risk in the long term.

Acknowledgements

We thank the Dutch juvenile myositis consortium, with Sylvia Kamphuis, Esther Hoppenreijs, Wineke Armbrust, Merlijn van den Berg, and Petra Hissink Muller, for their help and support with patient inclusion, sample collection and data collection. Dutch juvenile myositis consortium members: Sylvia S.M. Kamphuis Rotterdam, The Netherlands, Esther P.A.H. Hoppenreijs Nijmegen, The Netherlands Wineke Armbrust, Groningen, The Netherlands, J. Merlijn van den Berg, Amsterdam, The Netherlands, Petra C.E. Hissink Muller, Rotterdam, Leiden, The Netherlands. The Dutch JDM cohort was funded by Prinses Beatrix Spierfonds, The Bas Stichting, CureJM and Innovatiefonds Zorgverzekeraars. The Singaporean JDM cohort was supported by grants from the Singapore Ministry of Health’s National Medical Research Council (NMRC/CG/M003/2017, NMRC/STaR/020/2013, MOHIAFCAT2/2/08 and NMRC/TA/0059/2017). The funding sources had no role in the study design, collection, analysis, interpretation of data, the writing or decision to submit this publication. No payment was received for writing this manuscript. All researchers are independent from funders and all authors had full access to all of the data, and can take responsibility for the integrity of the data and the accuracy of the data analysis. We thank the paediatric rheumatology research group at KK Women’s and Children’s Hospital, Singapore, the Neurology department at the AMC, Amsterdam, the Rheumatology department at the University Hospital, Strasbourg, and the research groups within the UMC Utrecht for kindly providing the samples to perform this study. We thank the Dutch JDM network and especially Annette van Dijk-Hummelman and Ellen Schatorjé for their help and support in the patient inclusion and sample collection. We also thank Ester van Leeuwen for helping us with the logistics regarding samples of adult patients with myositis and the luminex core facility for performing all biomarker measurements. None of the data shown in this article have been published in conference abstracts. J.W. is the coordinator of the Dutch JDM cohort. J.W. and the contributors of the Dutch juvenile myositis consortium collected and selected Dutch JDM samples and clinical data. J.W. analysed data and wrote the manuscript. J.L., A.J.vdK., J.E.H. and A.M. collected and selected adult DM samples and clinical data. J.S.M., T.R.D.J.R. L.L.vdH., J.T., R.D.E.F.-S., H.G.O., E.M.G.J.dJ., M.M.B.S. and R.M.T. collected and selected EF and LoS samples and clinical data. C.A.W. and W.L.vdP. collected and selected SMA samples and clinical data. J.G.Y. and T.A. collected and selected JDM samples and clinical data. W.dJ. supervised and performed the luminex analysis and helped with data analysis. S.G. performed the Tie-1 ELISA and helped with data interpretation. A.vR.-K. and F.vW. supervised J.W. and were closely involved in setting up the study, data analysis, writing and editing of the manuscript. All authors critically reviewed the manuscript. The lead author (F.vW.) has had the final responsibility for submission for publication and affirms that this manuscript is an honest, accurate, and transparent account of the study being reported, that no important aspects of the study have been omitted, and that any discrepancies from the study as planned have been explained.

Funding: No specific funding was received from any funding bodies in the public, commercial or not-for-profit sectors to carry out the work described in this manuscript.

Disclosure statement: The authors have declared no conflicts of interest.

Supplementary data

Supplementary data are available at Rheumatology online.

References

Author notes