The Vasculopathy of Juvenile Dermatomyositis: Endothelial Injury, Hypercoagulability, and Increased Arterial Stiffness

Vasculopathy is considered central to the pathogenesis of juvenile dermatomyositis (DM) and is associated with severe extramuscular manifestations. We undertook this study to investigate the hypothesis that the vasculopathy of juvenile DM can be noninvasively tracked by examining biomarkers of endothelial injury, subclinical inflammation, hypercoagulability, and vascular arterial stiffness.


INTRODUCTION
Vasculopathy is considered central to the pathogenesis of juvenile dermatomyositis (DM) and is associated with severe extramuscular manifestations (1)(2)(3)(4)(5). The nature of this vasculopathy is complex, with evidence of both a true inflammatory small vessel vasculitis during active phases of the disease (2,6) and a noninflammatory occlusive vasculopathy with capillary dropout later in the disease process (4,7). Notably, previous studies have indicated that the presence of severe vascular changes on muscle biopsy was predictive of a chronic disease course (8,9), suggesting that persistent vasculopathy is a poor prognostic factor and determinant of adverse outcome in juvenile DM (5). Moreover, in the longer term there may also be a systemic vasculopathy affecting larger arteries, potentially leading to accelerated atherosclerosis and premature cardiovascular morbidity later in adulthood (10,11).
A major hurdle to the study of the vasculopathy of juvenile DM has been a lack of noninvasively measurable biomarkers that reliably capture the full spectrum of the proposed pathogenesis (12,13). Therefore, defining disease activity trajectories related to persistent endothelial injury in juvenile DM historically has been challenging. We and others have previously described 2 methods for detecting endothelial cell components in blood that allow noninvasive assessment of vascular injury in systemic vasculitides: circulating endothelial cells (CECs) and endothelialderived microparticles (EMPs) (14)(15)(16)(17)(18)(19)(20)(21)(22). We hypothesized that these noninvasively measured biomarkers of endothelial injury could be used to detect chronic vasculopathic injury and a putative prothrombotic state in juvenile DM. The present study was undertaken to examine biomarkers of endothelial injury, subclinical inflammation, hypercoagulability, and arterial stiffness in a UK cohort of patients with juvenile DM compared to age-similar healthy controls.

PATIENTS AND METHODS
Study design, subjects, and data collection. This was an observational comparative study, with ethical approval (MREC 1/3/022). The legal guardians of all subjects (or the subjects themselves if of legal age) provided written informed consent.
Patients with juvenile DM. For study inclusion, patients had to be age 2-19 years and have a diagnosis of juvenile DM (23). Patients were excluded from enrollment if they had any significant acute or chronic comorbidity that could cause acute endothelial injury, including intercurrent infection. Patients with juvenile DM were recruited from Great Ormond Street Hospital NHS Foundation Trust through the Juvenile Dermatomyositis Cohort and Biomarker Study (3,24) between September 2015 and January 2018 and were studied cross-sectionally. A subgroup of the patients studied cross-sectionally were also evaluated prospectively.
Definition of inactive juvenile DM. Patients were classified as having clinically inactive juvenile DM based on a modification of the Paediatric Rheumatology International Trials Organisation (PRINTO) criteria (25), as follows: absence of skin disease at the time of assessment, and at least 3 of the following 4 criteria: 1) creatine kinase (CK) ≤150 units/liter, 2) Childhood Myositis Assessment Scale (CMAS) score (26,27) ≥48/52, 3) Manual Muscle Testing 8 (MMT-8) score (28) ≥78/80, and 4) physician global assessment ≤0.2 (of a possible 10). Juvenile DM disease activity was ascertained by independent scrutiny of patients' medical records by 2 senior clinicians (MAO and DE); any discrepant cases were discussed to achieve consensus. All clinical and laboratory assessments were performed by one of the authors (CP), with blinding with regard to study group (healthy control or juvenile DM case), and juvenile DM disease status (active or inactive).
Healthy controls. Age-similar and sex-matched children who had no acute or chronic illnesses at the time of recruitment and were not regularly taking any medication at time of sampling were recruited as controls, with ethical approval (REC 11/ LO/0330). These children were either healthy unaffected siblings of patients with Kawasaki disease recruited for another major study our group has undertaken in the past (29) or were recruited through the Versus Arthritis Centre for adolescent rheumatology young scientist days, where healthy adolescents were invited to spend a day in the laboratory and donate blood with written informed consent.
Demographic, clinical, and laboratory data. Data collected included: age, sex, age at disease onset, disease duration, clinical features at initial presentation, routine echocardiography results, histopathologic severity scores on muscle biopsy, presence and typing of myositis-specific antibodies (MSAs) (30), and treatments (past and current). Validated clinical tools and indices were used to capture the full extent of disease activity in a systematic manner, i.e., the CMAS, MMT-8, physician global assessment of disease activity using a 10-cm visual analog scale (VAS) (31), functional ability according to the Childhood Health Assessment Questionnaire (32), parent global assessment of the patient's overall well-being on a 10-cm VAS, and parent global assessment of the patient's pain on a 10-cm VAS. The following laboratory test results were also collected: complete blood cell count, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), CK, alanine aminotransferase, and lactate dehydrogenase (LDH) levels, and antinuclear antibody status (33). Nailfold capillaroscopy was performed at the bedside with the use of a light, a 10× magnifying glass (otoscope), and a water-soluble gel (34) placed on the nailfold bed (of each of 8 fingers, excluding thumbs) to increase resolution; the result was considered abnormal in the presence of capillary loss with irregular capillary distribution, enlargement of capillary loops, changes in the capillary shape, or areas of hemorrhage (35).

| 1255
Conventional cardiovascular risk factors. Age, height, weight, body mass index, and smoking status were recorded before vascular studies were performed. Echocardiography was performed in patients with juvenile DM as part of routine clinical practice at the time of recruitment. Resting (minimum 15 minutes) blood pressure and heart rate were measured at the brachial artery using an oscillometric manual sphygmomanometer (Greenlight 300; Accoson). Nonfasting total cholesterol and triglycerides were also measured.
Plasma thrombin generation assay. To assess the prothrombotic tendency of plasma, a thrombin generation assay was performed in recalcified citrated platelet-poor plasma (PPP), as previously described (16,38). PPP (40 µl) was incubated with 50 µl fluorogenic substrate (0.5 mM Z-G-G-R-AMC/7.5 mM Ca 2+ ) and the reaction monitored by excitation/emission (360/460 nm) at 1-minute intervals for 90 minutes with an Optima Fluorescence plate reader (BMG Labtech). Lag time, peak thrombin (nM), peak time, velocity index, and endogenous thrombin potential (ETP) were quantified using a Technothrombin kit according to the protocol recommended by the manufacturer (DiaPharma).
Assessment of arterial stiffness. Carotid-femoral pulse wave velocity (PWV) and carotid-radial PWV were used as markers of arterial stiffness, measured by oscillometry using a Vicorder device (Skidmore Medical) in accordance with American Heart Association recommendations (39).

Statistical analysis.
Descriptive statistics were reported as the median and interquartile range (IQR) for continuous variables and as the absolute frequency and percentage for categorical variables. The significance of the differences between groups was assessed by Mann-Whitney U test (for 2 groups) or Kruskal-Wallis test (for multiple groups), and correlations between variables were assessed using Spearman's rank correlation coefficient. Categorical data were compared by chi-square test, or by Fisher's exact test in           the case of expected frequencies of <5. Differences between medians with 95% confidence intervals (95% CIs) of the differences were calculated. The Wilcoxon matched pairs signed rank test was used to compare variables at initial presentation and at latest follow-up for patients who were studied prospectively. Analysis of covariance was used to compare the slope of blood pressure versus age and PWV versus age between groups, using linear regression. P values less than 0.05 were considered significant (2-sided for CEC analyses; analysis of all the other indices was considered exploratory, and therefore no adjustments were made for multiple comparisons). Tibco Statistica, release 13.3 (StatSoft) and GraphPad Prism version 4.0 were used for data analyses.

RESULTS
Demographic characteristics of the study subjects. Ninety-patients with juvenile DM (median age 10.21 years [IQR 6.68-13.40]) were studied cross-sectionally. Fifty-seven (63.3%) were female. Seventy-nine healthy control children and adolescents were included in the final analysis; 3 additional control subjects had been enrolled but were subsequently excluded (due to severe eczema, upper respiratory tract infection, and ongoing medication treatment, respectively). The median age of the healthy controls was 16.7 years (IQR 10.7-17.4). There was no significant difference between the juvenile DM and healthy control groups in demographic characteristics, body mass index, or blood pressure (Table 1). In addition to the cross-sectional study, 25  Of the 49 patients tested, 33 (67.3%) were positive for MSAs, with nuclear matrix protein 2 (NXP-2) being the predominant type (n = 10). Most of the children with active disease were female (P = 0.0003 versus those with inactive disease), and the group with active disease was older at disease onset (P = 0.0339) and had a shorter disease duration (P = 0.0005) compared to children with inactive disease. They also had higher ESR (P = 0.0330) and LDH levels (P = 0.0008) compared to the group with inactive disease, whereas CK and CRP levels did not differ. Echocardiography was performed in 66 of the patients with juvenile DM. Results were normal in 58 patients, and a small pericardial effusion was detected in 3. The remaining 5 patients had tricuspid regurgitation, aortic regurgitation, mild concentric left ventricular hypertrophy, mildly reduced right ventricular systolic function, and patent foramen ovale (1 patient each).
Of the 25 patients with juvenile DM studied prospectively, 17 had active disease at the time of recruitment, and 8 had inactive disease. Three of the patients initially classified as having inactive disease had a disease flare (mainly affecting the skin) at the last follow-up visit.
Endothelial injury. CECs. CEC numbers were higher in patients with juvenile DM (median 96 cells/ml [IQR ) compared to healthy controls (median 12 ) (difference −84 [95% CI −100.0, −56.00]; P < 0.0001) ( Figure 1A). Patients with active juvenile DM had higher numbers of CECs than those with inactive juvenile DM (difference −82 [95% CI −40.00, −128.00]; P < 0.0001). Previous studies have suggested that vasculopathy may play a role in the pathogenesis of calcinosis in juvenile DM (1,5), and we noted higher numbers of CECs in juvenile DM patients with calcinosis compared to healthy controls (difference −54 [95% CI −100.00, −28.00]; P < 0.0001) ( Figure 1A), but no significant difference between the patients with and those without calcinosis (P = 0.5). We also observed that 10 of the 12 patients with calcinosis had active juvenile DM. Further analysis of specific disease features pertinent to the vasculopathy of juvenile DM showed that CEC numbers were higher among patients who  Figure 1D). Seventeen of these patients (68%) had active juvenile DM at the time of recruitment and 14 (56%) had active juvenile DM at the last follow-up. There was a decrease in CEC numbers among patients whose disease status changed from active at baseline to inactive at the last follow-up (n = 6) (median difference −32 [95% CI −504, −12]; P = 0.03) and an increase among patients who had inactive disease at baseline and active disease at the last follow-up (n = 3) (median difference 280 [95% CI 60, 360]; P = 0.25) (Figures 1E and F).
Circulating levels of inflammation markers. Overall, there was a significant difference in circulating levels of inflammation markers between patients with juvenile DM and controls, and between patients with active juvenile DM and those with inactive juvenile DM | 1261    * Values are the median (interquartile range). DM = dermatomyositis; 95% CI = 95% confidence interval; AnxV = annexin V; MPs; = microparticles; EMPs = endothelial-derived MPs; PMPs = platelet-derived MPs; TF = tissue factor; ETP = endogenous thrombin potential (corresponding to the area under the curve for thrombin generation assay curves). † P values were determined by Kruskal-Wallis test followed by Mann-Whitney U test.
( Table 3). Patients with active disease had higher levels of IL-10, IL-6, IFNλ1, MCP-1, IP-10, and galectin-9 compared to patients with inactive disease or healthy controls. As endothelial cells are the main source of galectin-9 (40), we then examined the correlation between levels of galectin-9 and CECs; a strong correlation was identified (r = 0.48, P < 0.0001 Circulating MPs and plasma thrombin generation. Total AnxV+ MP numbers were significantly increased among patients with juvenile DM compared to patients with inactive juvenile DM and healthy controls (both P < 0.0001) ( Table 4). MPs were mainly of platelet and endothelial origin. B cell-derived MPs were the third most common MP population. Total AnxV+ MP numbers correlated with CEC numbers (r = 0.42, P < 0.0001) and with galectin-9 levels (r = 031, P = 0.01). (Figures 2A and  B). CD62E+ MP counts also correlated strongly with CEC counts (r = 0.20, P = 0.027) ( Figure 2C).
Arterial stiffness. We confirmed a strong positive association between age and carotid-femoral and carotid-radial PWV (both P < 0.0001) (Figures 3A and B). The slope for carotid-radial PWV in relation to age among patients with juvenile DM (0.44 m/second/year, y-intercept 3.414) differed significantly from that among healthy controls slope (0.12 m/second/year, y-intercept 5.903) (P = 0.003) ( Figure 3C), indicating significantly increased arterial stiffness among patients with juvenile DM. No significant difference in carotid-femoral PWV was observed (P = 0.12) ( Figure 3D).

DISCUSSION
We conducted a large cross-sectional study of patients with juvenile DM and explored biomarkers to monitor the vasculopathy of this disease. Our data provide evidence of increased endothelial injury in children with active juvenile DM, associated with proinflammatory cytokines, high levels of circulating MPs with a propensity to drive thrombin generation and potentially increase occlusive vasculopathy, and increased arterial stiffness in patients with juvenile DM compared to controls. These noninvasively measured vascular indices provide unique insight into the pathogenesis of vascular injury in this disease and could be used for clinical monitoring of the vasculopathy of juvenile DM.  CECs are mature cells that have detached from the vessel wall in response to endothelial injury (17,22,42,43). We demonstrated increased levels of CECs in juvenile DM, in accordance with another recent study that also demonstrated increased CEC counts in juvenile DM, despite the use of a different method (flow cytometry) for cell enumeration (44). CEC numbers also strongly correlated with other biomarkers of endothelial injury such as galectin-9 levels and EMP counts, thus supporting the robustness of these endothelial injury indices. CEC counts were also found to be elevated in patients with juvenile DM whose disease was considered to be clinically inactive. This raises the possibility that in some patients with juvenile DM there is ongoing subclinical endothelial injury and disease activity that is not captured by laboratory parameters and disease activity measures that are currently routinely used. In addition, the demonstration of elevated CEC counts in patients with abnormal nailfold capillaries supports the notion that this finding is indeed a clinical sign of active juvenile DM vasculopathy.
We did not demonstrate any differences in traditional markers of systemic inflammation as assessed by hsCRP or SAA or any differences in routine cardiovascular risk factors (45) to account for the elevated CEC or MP counts we observed (Tables 1 and  3). We did, however, detect consistently higher levels of endothelial activation-related cell adhesion molecules, cytokines, and chemokines in patients with juvenile DM compared to healthy controls. This observation likely indicates a chronic disturbance in endothelial cell homoeostasis in patients with juvenile DM, including in some patients with apparently quiescent clinical disease activity as assessed using routine clinical tools. Juvenile DM is considered an interferonopathy, and therefore, not surprisingly, we detected high levels of IFN-driven cytokines/chemokines (IFNα, IFNλ1, MCP-1, IP-10) (46,47) in all patients with juvenile DM compared to controls, especially in patients with active disease, though we do note that other cytokines also correlated with active disease.
Additionally, we detected elevated levels of circulating endothelial, platelet, monocyte, and B cell-derived MPs that are highly prothrombotic (48) in patients with juvenile DM. We have previously demonstrated enhanced MP-mediated thrombin generation in children with active vasculitis (16), potentially explaining some of the excess thrombotic risk associated with vasculitis. Similarly, we detected elevated levels of MPs, including highly prothrombotic TF+ MPs, and enhanced plasma thrombin generation in patients with active juvenile DM. This increased prothrombotic propensity, mediated by MPs among other prothrombotic factors, might contribute to occlusive vasculopathy and organ injury in juvenile DM. The exact mechanism by which different types of MPs may promote endothelial dysfunction and thrombogenicity in juvenile DM remain to be established.
We also showed that children with juvenile DM have enhanced carotid-radial PWV, consistent with increased arterial stiffness compared to healthy children. This increased PWV may suggest a generalized secondary systemic vasculopathy, ultimately leading to accelerated atherosclerosis. Other factors such as sedentary lifestyle, long-term treatment with glucocorticoids, and ongoing systemic inflammation may Figure 3. Carotid-radial and carotid-femoral pulse wave velocity (PWV) in patients with juvenile dermatomyositis (JDM) and healthy controls. A and B, Both carotid-radial PWV (A) and carotid-femoral PWV (B) correlated with age in the group of all subjects combined (patients with juvenile DM and controls). C and D, The slope for carotid-radial PWV in relation to age in patients with juvenile DM differed significantly from the slope in healthy controls (C), whereas a difference was not observed for carotid-femoral PWV (D), by analysis of covariance. also contribute to this finding. It is not yet known whether premature cardiovascular morbidity occurs later in adulthood in patients with juvenile DM, but our data strongly suggest that this could be a future concern, and indeed has been observed in adults with juvenile DM (49,50). Prospective studies to evaluate changes of PWV over time in patients with juvenile DM are needed.
No differences in carotid-femoral PWV between patients with juvenile DM and controls were demonstrated in this study. Carotidradial PWV mainly reflects the peripheral arterial stiffness of upper limb muscular arteries (branchial and radial arteries), while carotidfemoral PWV is a marker of central arterial (aortic, i.e., elastic artery) stiffness (51,52). Previous studies have suggested that the variation in elastin-collagen smooth muscle proportions within different arterial segments determines the observed arterial stiffness in response to various cardiovascular risk factors (53). It is therefore perhaps not surprising that inflammatory processes such as juvenile DM may also have a different effect on the arterial stiffness of separate parts of the arterial tree. In addition, other studies have suggested that carotid-radial PWV mainly reflects microvascular endothelial dysfunction (54,55), and therefore it could be the case that juvenile DM induces such microvascular changes rather than larger structural arterial changes.
The present findings have multiple potential implications with regard to therapy. In patients with ongoing vasculitic endothelial injury, prolonged immunosuppressive treatment and/or consideration of novel directed therapeutic strategies may be needed to target the vasculopathy of juvenile DM. We observed an upregulation of several proinflammatory cytokines/chemokines that could potentially provide a therapeutic target. Of particular interest is the up-regulation of IFN-driven cytokines/chemokines that could be contributing to driving endothelial injury. Notably, several recent transcriptomic studies (56,57) have shown up-regulation of IFN-stimulated genes within the capillaries of the muscle and disruption of vascular network organization upon exposure of endothelial cells to IFN, highlighting the involvement of this pathway in the vasculopathy of myositis (58). Based on these observations, targeting IFN-related endothelial injury with JAK inhibition has therefore emerged as a novel therapeutic strategy for myositis (59,60). In that context, we have reported the use of CECs to monitor the rapid response of endothelial injury to JAK inhibition in a patient with juvenile DM (60). MP profiling and thrombin generation assays could provide a novel means of assessing prothrombotic risk in patients with juvenile DM, allowing improved risk stratification and potential targeting of primary thrombosis prevention (61). Finally, if larger prospective studies confirm increased arterial stiffness in children with juvenile DM, formal therapeutic lifestyle interventions may be considered, in order to reduce this risk of accelerated cardiovascular morbidity.
Our study has several limitations. It was a single-center study of a heterogeneous cohort of patients with juvenile DM. At the time of recruitment, patients were receiving a variety of treatments, although they were treated in accordance with published clinical guidelines (62). MSA testing to better understand the potential relevance of these antibodies to vascular phenotype and influence on circulating IFN and galectin-9 levels was not available in all patients. Our control population was age similar, but not exactly age matched, due to of lack of availability of control samples from very young healthy children.
In conclusion, we have demonstrated dynamic changes in biomarkers of endothelial injury (MPs and CECs) in children with juvenile DM. Future studies could also explore these indices in the context of clinical trials, to better understand the use of more targeted therapeutic strategies on vascular phenotype in juvenile DM.