Article Text

Download PDFPDF

Extended report
In juvenile dermatomyositis, cardiac systolic dysfunction is present after long-term follow-up and is predicted by sustained early skin activity
  1. Thomas Schwartz1,2,
  2. Helga Sanner3,4,
  3. Ola Gjesdal5,
  4. Berit Flatø2,3,
  5. Ivar Sjaastad1,5
  1. 1Institute for Experimental Medical Research, Oslo University Hospital-Ullevål and University of Oslo, Oslo, Norway
  2. 2Medical Faculty, Institute for Clinical Medicine, University of Oslo, Oslo, Norway
  3. 3Section of Rheumatology, Oslo University Hospital-Rikshospitalet, Oslo, Norway
  4. 4Norwegian Competence Centre of Pediatric and Adolescent Rheumatology, Oslo University Hospital-Rikshospitalet, Oslo, Norway
  5. 5Department of Cardiology, Oslo University Hospital-Ullevål, Oslo, Norway
  1. Correspondence to Dr Thomas Schwartz, Institute for Experimental Medical Research, Oslo University Hospital-Ullevål and University of Oslo, Oslo 0407, Norway; thomas.schwartz{at}


Objective To compare systolic cardiac function in patients with juvenile dermatomyositis (JDM) with matched controls and examine associations between systolic and diastolic cardiac function and disease variables.

Methods Fifty-nine patients, examined at follow-up, median 16.8 years (2–38 years) after disease onset, were compared with 59 age-matched and sex-matched controls. Echocardiography was performed and analysed blinded to patient information. We used mitral annulus displacement to assess the relative long-axis shortening of the left ventricle (long-axis strain) and early diastolic tissue velocity (e′), as markers for systolic and diastolic function, respectively. Disease activity and organ damage were assessed at follow-up by clinical examination and retrospectively by chart review.

Results Long-axis strain was reduced in patients compared with controls (16.6% (2.5) vs 17.7% (2.0), mean (SD), p=0.001), whereas no difference was seen between patients with active and inactive disease. Disease duration correlated with systolic and diastolic function (rsp=−0.50 and rsp=−0.73, both p<0.001) and so did Myositis Damage Index (MDI) 1 year (rsp=−0.36 and rsp=−0.46) and MDI at follow-up (rsp=−0.33 and rsp=−0.60), all p<0.01. High early disease activity score (DAS) in skin (DAS skin 1 year), but not in muscle, predicted systolic (standardised β=−0.28, p=0.011, R2=48%) and diastolic dysfunction (β=−0.36, p<0.001, R2=72%) at follow-up.

Conclusions Long-axis strain was reduced in JDM patients compared with controls, suggesting systolic dysfunction. Impaired systolic and diastolic function was predicted by DAS skin 1 year. This indicates a common pathway to two different cardiac manifestations in JDM, perhaps with similar pathogenesis as skin affection.

  • Dermatomyositis
  • Disease Activity
  • Cardiovascular Disease
  • Inflammation

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Juvenile dermatomyositis (JDM) is a systemic disease with skeletal muscle and skin as primary target organs, but may also affect other organ systems. Interest has been paid to whether inflammation similar to what is seen in skeletal muscle can also occur in the heart. Cardiac affection in other rheumatic diseases, such as rheumatoid arthritis and systemic lupus erythematosus has been extensively documented.1 ,2 We have previously shown that diastolic dysfunction can be a long-term complication in JDM (median 16.8 (2–38) years after diagnosis).3 However, systolic dysfunction has not yet been described in JDM.

The most commonly used parameter of systolic function, left ventricular (LV) ejection fraction (EF) assessed by echocardiography, has limitations due to difficulties in data acquisition and assumptions of LV geometry.4 ,5 Other modalities, such as contrast ventriculography,6 cardiac MRI (CMR)7 and strain echocardiography,8 ,9 are accurate methods but face challenges due to either risk, availability, cost or time consumption.

Since the early era of echocardiography, displacement of mitral annulus (MA) plane has been measured as an index of systolic LV function.10 ,11 Recently, it has regained interest as an accurate, easy and sensitive way of assessing systolic function. The MA descent from diastole to systole visualises LV movement in the longitudinal axis, and hence, LV contractility. A relative estimate of this (eg, long-axis strain) is useful since the length of LV increases with body size. This method's sensitivity to systolic dysfunction is probably higher than movement in the transverse (short) axis and EF, and it is comparable with global longitudinal strain.12 ,13

The primary aim of the present study was to assess LV systolic function in JDM patients compared with controls by using normalised MA displacement (long-axis strain) and to investigate if systolic dysfunction could be predicted by early disease parameters. Having previously evaluated diastolic dysfunction in JDM assessed by the ratio E/e′3 (E, early diastolic transmitral flow; e′, early diastolic tissue velocity), our secondary aims were to investigate how diastolic function could be assessed by e′ alone and how e′ was associated with patient characteristics and disease parameters in JDM. We studied an age-matched and sex-matched cohort of JDM patients, which offers a unique opportunity to detect subtle signs of organ damage like cardiac dysfunction.

Patients and methods

Patients and controls

Patients with a probable or definitive diagnosis of dermatomyositis14 were included if disease onset occurred before the age of 18 years, disease duration was ≥24 months and age was ≥6 years at follow-up. After a manual and electronic search in medical records, a retrospective inception cohort of JDM patients, diagnosed between January 1970 and June 2006 in Norway was identified and tracked through the Norwegian population register.15 ,16 Of 66 patients included, 4 were deceased and 3 did not want to participate; the remaining 59 (95%) represent our study population. Since Oslo University Hospital (OUS) is the major referral centre for JDM patients in Norway, we believe this represents the vast majority of JDM patients diagnosed during this period. One sex-matched and age-matched control per patient was randomly drawn from the National Population Register. Controls with lung or heart diseases (one with atrial fibrillation) were excluded. Informed consent was obtained from all participants (and their parents if aged <16 years), according to the Declaration of Helsinki.

Data collection and clinical measures

Clinical examination (defined as follow-up examination) of all patients and matched controls was performed by a single physician (HS) at OUS from September 2005 to May 2009. Disease activity was measured by Disease Activity Score (DAS) for JDM (range 0–20, 0 means no activity),17 composed of subscores for skin (DAS skin, 0–9) and muscle (DAS muscle, 0–11). Cumulative organ damage was measured by Myositis Damage Index (MDI, range 0–35/40)18; DAS and MDI were also scored retrospectively from the first year postdiagnosis. If an item was not mentioned in a chart, a score of 0=absent was given. Data on disease onset, course, comorbidity and use of medication including cumulative prednisolone dose were obtained retrospectively from the medical records at OUS or, when necessary, from other hospitals.19

Inactive disease was defined by the proposed Pediatric Rheumatology International Trials Organisation (PRINTO) criteria of 2012, including at least three of the four following: manual muscle test (MMT-8) ≥78 (0–80),16 physician global assessment of muscle activity (phyGloVAS) ≤0.2, Childhood Myositis Assessment Scale ≥48 and creatine kinase ≤150.20 JDM inactive patients off medication are abbreviated JDM inactive; the remaining patients are abbreviated JDM active. Physical health was measured by the Short Form-36 (SF-36) physical component summary (PCS).21 Health Assessment Questionnaire (HAQ)22 and the Child HAQ23 were used to measure physical function in patients aged ≥18 years (n=39) and <18 years (n=20), respectively. Disease onset was defined as time of the first muscle or skin symptom and disease duration as the time from disease onset to the follow-up examination. The serum samples were collected from non-fasting patients and controls.


Two-dimensional, M-mode and Doppler echocardiography were performed at the time of follow-up examination, except the 19 first patients, who were examined median 4 months later. No systematic echocardiographic assessment had been done prior to this. We used a Vivid 7 ultrasound scanner (GE—Vingmed Ultrasound, Horten, Norway).4 ,24 A minimum of three cardiac cycles were recorded, analysed and averaged.

MA displacement was assessed by anatomical M-mode in lateral and septal position using the four-chamber apical view. LV length was assessed as distance from the endocardial border of apex to the midportion of the MA plane. Long-axis strain was calculated as MA displacement expressed as percentage of LV end-diastolic length.

Colour-coded tissue Doppler was performed in 58 patients and 57 controls, with a frame rate of ∼180/s. We used sample volumes from MA—lateral and septal position in four-chamber view and two corresponding positions in two-chamber view.

Early diastolic transmitral flow (E) and early diastolic tissue velocity (e′) were recorded and E/e′ ratio calculated. Both E/e′25 and e′ are commonly used parameters reflecting LV diastolic function. However, recent publications and clinical experience suggest that e′ alone is simpler and equally reliable to E/e′.26 Therefore, in the present paper we have used e′ alone as parameter of diastolic function.

Analyses were performed blinded to clinical information and to patient/control identity (TS). To assess interobserver reliability, echocardiographic data from a subset of 10 patients were analysed by TS and IS under the same physical conditions. To assess intraobserver reliability, TS analysed the data a second time, 2 weeks later.

Statistical analysis

Differences between patients and matched controls and between scorers were tested by the paired sample t test. Differences between patient groups were tested by the independent sample t test (normally and continuous distributed variables). Two tailed tests were used for all calculations and p<0.05 was considered significant. Correlations were determined by the Spearman correlation coefficient (rsp). Associations between e′ and long-axis strain at follow-up (dependant variables) and DAS skin, DAS muscle and MDI at 1 year after diagnosis and cumulative prednisolone dose were controlled for male gender and disease duration in a linear regression model with forward deletion of possible predictors. Interscorer and intrascorer reliability was expressed as the average difference between the scorers ±2SD, corresponding to the mean and ‘limits of agreement’ in the Bland–Altman analysis.27 Average absolute difference between the scorers was expressed as mean±SD. SPSS V.19 (SPSS, Chicago, Illinois, USA) was used for statistical analyses.


Clinical characteristics in JDM patients and controls

Table 1 shows clinical characteristics in the 59 JDM patients and 59 sex-matched and age-matched controls. Of the patients, one had coronary artery disease and seven had hypertension whereas none of the controls. Patients had lower cholesterol levels (total, high-density lipoprotein (HDL) and low-density lipoprotein (LDL); p=0.007, p<0.001 and p=0.003) than controls, without any difference in HDL/LDL ratio (p=0.09).

Table 1

Characteristics in patients with JDM controls

Cardiovascular parameters in JDM patients versus controls

LV end-systolic diameter was slightly lower in patients compared with controls (p=0.047, table 2). Except for this, no differences in LV dimensions or EF between patients and controls were seen. However, long-axis strain was lower in patients than in controls (p=0.001, table 2 and figure 1). Patients also had lower e′ and higher E/e′ than controls, both parameters of diastolic dysfunction.

Table 2

Cardiovascular parameters in patients with JDM median 16.8 years after disease onset with active and inactive disease and controls

Figure 1

Systolic function (long-axis strain) in 59 patients with juvenile dermatomyositis (JDM) compared with 59 controls, and in patients with inactive (26) compared with active disease (33). Data are presented as box plots, values are median and the boxes represent the 25th–75th percentiles.

No differences were seen between JDM active and JDM inactive for e′ (p=0.64), E/e′ (p=0.09), long-axis strain (p=0.25, figure 1), systolic (p=0.34) or diastolic (p=0.36) blood pressure or cholesterol levels (data not shown). JDM inactive had larger LV diameters than JDM active, but no difference was seen when normalised to patient height (LV diastole/height=2.94 (0.24) vs 2.89 (0.21), p=0.40).

Correlations between cardiac function, age and disease duration in JDM patients and controls at follow-up

Long-axis strain and e′ both correlated with age (rsp=−0.57 and rsp=−0.71, both p<0.001) and disease duration (rsp=−0.50 and rsp=−0.73, both p<0.001) in patients (figure 2A,D), indicating possible increasing systolic and diastolic dysfunction with age and disease duration. To compare, the correlation with age was weaker in controls (rsp=−0.30, p=0.017 and rsp=−0.43, p=0.001, respectively). Long-axis strain and e′ intercorrelated stronger in patients than in controls (rsp=0.63, p<0.001 vs rsp=0.31, p=0.023).

Figure 2

Correlations between systolic (long-axis strain, A–C) and diastolic function (e′, early diastolic tissue velocity, D–F) and disease duration, organ damage median 16.8 years after diagnosis (Myositis Damage Index (MDI) at follow-up) and disease activity in skin 1 year after diagnosis (DAS skin 1 year) in 59 patients with juvenile dermatomyositis (JDM). rs, Spearman correlation coefficient.

Associations between disease parameters and cardiac function in JDM patients at follow-up and during disease course

Long-axis strain and e′, but not EF, correlated with blood pressure and organ damage (MDI, table 3 and figure 2B and E). Neither for e′, long-axis strain nor EF, any correlations with smoking, DAS or CHAQ/HAQ were present. e′ correlated with cumulative prednisolone dose.

Table 3

Correlations between parameters of systolic and diastolic function and disease variables in 59 patients with JDM during disease course, at follow-up and 1 year postdiagnosis

Predictors for systolic and diastolic dysfunction in JDM patients

Organ damage (MDI) and disease activity (particularly DAS skin) at 1 year, correlated well with e′ at follow-up. Also, long-axis strain correlated with organ damage and early disease activity, except DAS muscle (table 3 and figure 2C,F). EF only correlated weakly with DAS skin.

In linear regression models, DAS skin 1 year (standardised β=−0.36, p<0.001) and cumulative prednisolone dose (standardised β=−0.24, p=0.006) predicted low e′, with significant contributions from both control measures (gender, standardised β=−0.29, p<0.001; disease duration, standardised β=−0.51, p<0.001; R2 final model=72%). Furthermore, DAS skin 1 year also predicted low long-axis strain (standardised β=−0.28, p=0.011; gender, standardised β=−0.45, p<0.001; disease duration, standardised β=−0.28, p=0.015; R2 final model=48%). We were not able to predict cardiac outcome with variables assessed at time of diagnosis.


Interobserver and intraobserver reliability (table 4) were acceptable with mean difference <10% and without significant differences. Long-axis strain was successfully assessed in all, whereas EF was not obtainable due to acquisition challenges in 13 of 118 individuals.

Table 4

Reproducibility of echocardiographic measurements


In JDM patients seen 16.8 years after disease onset, we found signs of impaired cardiac systolic function with reduced long-axis strain, compared with controls. A strong correlation between systolic (long-axis strain) and diastolic (e′) function was present in patients but not in controls. Systolic dysfunction was linked to high long-term organ damage and high early (1 year) disease activity; diastolic dysfunction was linked to the same parameters even stronger. High score of DAS skin 1 year, but not DAS muscle, predicted impaired systolic and diastolic parameters. The findings implicate that JDM patients may develop both systolic and diastolic dysfunction. Both are associated with long disease duration and may be initiated by the same mechanisms that cause sustained skin inflammation during the first year of the disease.

The representativeness of our inception cohort has been described earlier; we believe it covers most of the JDM cases in Norway diagnosed from 1970 to 2006. Predominance of women, age at diagnosis and symptom duration (from disease onset to diagnosis) correspond with other hospital-based or registry-based cohorts.28 ,29

The random selection of age-matched and sex-matched controls from the National Population Register represents a strength of our study.

Our data suggest systolic dysfunction in JDM by impaired long-axis LV shortening (long-axis strain), but interestingly not by EF, the more commonly used systolic parameter. As we will highlight below, in our analyses with disease parameter and cardiac parameter (DAS, MDI, disease duration, blood pressure and e′), long-axis strain showed much more consistent correlations than EF; hence we believe this is a more robust parameter assessing systolic function30

In patients, both systolic and diastolic function correlate strongly with disease duration, and therefore, necessarily with age. Compared to the consistent but weaker correlation with age in the control group, it seems evident that disease duration, not only age, contributes substantially to the cardiac affection in JDM. It was not surprising that diastolic and, to some degree, systolic dysfunction correlated with organ damage. In particular, diastolic dysfunction correlated with MDI assessed at follow-up, indicating that myocardial injury is a long-term process. The significance of inflammation in cardiac diseases is widely recognised.31 ,32 It is reasonable to believe that cardiac affection in JDM is a chronic low-grade inflammation, either organ-specific or as part of general organ damage. This fits nicely with the observed correlation between cardiac function and disease duration. Possible mechanisms for the duration-dependent decrease in cardiac function can involve proinflammatory mediators that cause cardiac remodelling, leading subsequently to cardiac fibrosis.33

Between JDM active and JDM inactive, no differences were seen in either diastolic or systolic function. Not included in the PRINTO criteria, DAS 1 year correlated well with systolic and diastolic function. Furthermore, in linear regression models that explained 48% and 72% of the variance, DAS skin 1 year predicted both systolic and diastolic dysfunction, whereas DAS muscle did not. Several studies suggest there are distinct pathophysiological processes leading to skin versus musculoskeletal affection in JDM. Smith et al34 found that nailfold capillaroscopy (NFC) correlated with DAS skin but not with DAS muscle. Stringer et al35 found that active skin affection (in particular Gottron's papules) and NFC abnormalities, 6 months after diagnosis, were associated with a long phase of untreated disease. As NFC reflects systemic vasculopathy, there is reason to believe that vasculopathy plays a central role in skin involvement in JDM. It is possible that a similar vasculopathic process is responsible for disease activity in skin and myocardium, different from the inflammatory process involved in skeletal muscle. In the same regression models, cumulative prednisolone dose was another predictor of diastolic, but not systolic dysfunction. A likely explanation is that cumulative prednisolone is reflecting increased disease activity over time, since we have no indication that prednisolone per se can induce such changes in the myocardium.36

One possible mechanism of cardiac dysfunction is through dyslipidaemia, found by other investigators in JDM.37 Based on our data, however, the cardiovascular affection in JDM does not seem related to lipid status. Long-axis strain reflects the myocardium's ability to contract (systolic function),10 ,13 whereas e′ reflects the stiffness of the myocardium and its ability to relax and recoil (diastolic function). In our data, systolic and diastolic function was found to correlate with the same parameters such as blood pressure, DAS 1 year, MDI and disease duration. Systolic and diastolic dysfunction could represent two different aspects of cardiac damage occurring simultaneously in JDM, caused by the same underlying aetiology. Or perhaps, more likely, they could both reflect the same pathophysiological phenomenon in the myocardium leading to systolic as well as diastolic dysfunction in JDM. In the literature, some authors claim that diastolic and systolic heart failure are two different entities38; others see this as a continuum of cardiac damage.39 As for EF, the lack of correlation with most diseases and cardiovascular parameters supports, in our opinion, the limited value of EF previously reported to detect incipient heart failure.4 ,5

In our study, measurement of long-axis strain was highly feasible. In systole, the MA plane moves towards the almost stationary apex of the heart.40 The subendocardium is the most vulnerable layer of the LV wall, and myocardial fibres in this layer are organised in a longitudinal orientation. Hence long-axis strain may reflect early systolic dysfunction more accurately than EF.12 Also, the parameter is easily obtainable, even in cases where technical recording quality is poor. For instance, in our material, we were unable to measure EF in 11% of the participants due to limited acoustic access, whereas long-axis strain could be measured in all. The parameter was also robust with low interobserver and intraobserver variability. In our opinion, this shows that long-axis strain can be a more suitable parameter when assessing systolic function than EF, allowing inclusion of all study subjects.

A limitation is the cross-sectional nature of the study; a prospective study could have clarified the importance of time further. Also, in JDM, DAS and MDI were validated recently (2003 and 2004 respectively).17 ,18 The retrospective scoring and application of diagnostic criteria might have been based on incomplete charts, and hence, represents a limitation of the study.

In conclusion, we have shown that JDM patients have impaired systolic function compared with controls, whereas no difference was found between patients with active and inactive disease. Both systolic and diastolic dysfunction were associated with high score of long-term organ damage (MDI, follow-up) and high early disease activity in the skin (DAS skin 1 year), but not in muscle. This could suggest a similar vasculopathy in the myocardium as found in the skin. The association with disease duration indicates that myocardial remodelling in JDM is an ongoing long-term process. High initial disease activity in skin might be a particular risk factor; thorough cardiological follow-up should be performed to identify cardiac dysfunction at an early stage in these individuals.


TS and IS are supported by KG Jebsen Cardiac Research Center and Center for Heart Failure Research, University of Oslo, Oslo, Norway.



  • Handling editor Tore K Kvien

  • Contributors TS, HF, BF and IS designed the study. TS and IS analysed the data and drafted the paper. HS, IS and BF collected the data and participated in drafting the paper. OG participated in designing the study and drafting of the paper. All authors contributed with critical review of the paper, read and approved the final manuscript.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was carried out in compliance with the declaration of Helsinki and was approved by the Regional Ethics Committee (Helse Sør-Øst).

  • Provenance and peer review Not commissioned; externally peer reviewed.