Echocardiographic strain analysis reflects impaired ventricular function in youth with pediatric-onset systemic lupus erythematosus

Abstract

Background:

Strain analysis with speckle-tracking echocardiography shows promise as a screening tool for silent myocardial dysfunction in pediatric-onset systemic lupus erythematosus (pSLE). We compared left ventricular (LV) systolic deformation (measured by strain) in children and adolescents with pSLE to controls, and assessed the relationship between strain, disease activity, and other non-invasive measures of cardiovascular health.

Methods:

20 pSLE subjects ages 9–21 underwent comprehensive cardiovascular testing, including 2D speckle-tracking echocardiography, ambulatory blood pressure monitoring (ABPM), peripheral endothelial function testing, pulse wave velocity and analysis, and carotid ultrasound. Longitudinal apical-4 chamber (LS_A4C) and midpoint circumferential strain (CS_mid) were compared to that of 70 healthy controls using multivariable linear regression. Among pSLE subjects, Pearson correlation coefficients were calculated to evaluate relationships between global longitudinal or circumferential strain and other measures of cardiovascular health.

Results:

Average SLE disease duration was 3.2 years (standard deviation (SD) 2.1). 2/20 pSLE subjects had persistent disease activity, and only one met criteria for hypertension by ABPM. LS_A4C was significantly reduced in pSLE subjects compared to controls (mean −18.3 (SD 3.2) vs. −21.8% (SD 2.2), p-value <0.001). There was no significant difference in CS_mid (−24.8 (SD 3.7) vs. −25.7% (SD 3.4), p = 0.29). Among pSLE subjects, decreased nocturnal blood pressure dipping on ABPM was associated with reduced global circumferential strain (𝒓 −0.59, p=0.01).

Conclusions:

Longitudinal myocardial deformation is impaired in pSLE patients despite clinical remission and may represent early myocardial damage. Strain analysis should be considered in addition to standard echocardiographic assessment during follow-up of patients with pSLE.

Introduction

Systemic lupus erythematosus (SLE) is a chronic, multi-system autoimmune disease with well-described cardiovascular comorbidities due to chronic inflammation and direct cardiovascular involvement(1–4). There is a significantly increased risk of heart failure and premature atherosclerotic disease that is most pronounced in younger SLE patients, including a 50-fold increased relative risk of heart failure in young adults with SLE ages 20–24 compared to the general population (5). Pediatric-onset SLE (pSLE) is associated with greater cumulative organ damage and mortality (6,7), and cardiovascular complications occur at a younger age (8). As a result, methods to detect cardiovascular disease during the early subclinical stages are essential for developing adequate screening and prevention measures.

There has been increasing interest in the use of myocardial deformation (measured as strain) by echocardiography to assess ventricular function, since strain has been shown to detect silent myocardial dysfunction across a variety of conditions, including SLE. Strain describes myocardial deformation, that is, the fractional change in length of a myocardial segment (9). It is a more sensitive measure of ventricular function and contractility than other conventional echocardiographic parameters (for example, shortening fraction) and can be used to quantify both regional and global myocardial deformation (10). Longitudinal and circumferential deformation during systole are represented as negative values, with less negative values representing decreased function. Two-dimensional speckle-tracking echocardiography is an automated analytic technique that utilizes the trajectory of tissue speckles (acoustic markers) from frame-to-frame to derive strain (11,12). It has excellent reproducibility and minimal intra and inter-observer variability, and unlike Doppler-derived strain measures, there is no angle dependency (13).

Several studies in adults with SLE without overt cardiovascular disease have detected reduced systolic function by left ventricular (LV) strain analysis using speckle-tracking echocardiography (14–17). The subclinical myocardial dysfunction has been hypothesized to represent damage from inflammation and hypoperfusion (15,18,19) or ongoing subclinical myocarditis in association with disease activity (14). In SLE patients with clinical lupus myocarditis, reduced strain is also prognostic of persistent systolic dysfunction following treatment (20). There are two pediatric studies suggesting that impaired LV systolic strain is present in children with pSLE (21,22). However, the majority of these patients were still using glucocorticoids, which can influence afterload. In addition, there have been no studies to date evaluating strain in relation to blood pressure abnormalities and other measures of subclinical atherosclerosis in pSLE. As a result, the clinical value of strain measurements in pSLE has not been fully established. Further, it is also unknown whether strain can function as a marker of early atherosclerotic disease, active inflammation, or cumulative damage in this population.

The objectives of this study were to apply two-dimensional speckle-tracking echocardiography to 1) compare left ventricular (LV) systolic myocardial deformation in children and adolescents with pSLE to healthy controls, 2) determine the association between disease activity and systolic strain measurements in pSLE, and 3) identify relationships between LV strain and other markers of subclinical cardiovascular disease.

Methods

Study Design:

We compared LV systolic strain measured prospectively by 2D speckle-tracking echocardiography in youth with pSLE to retrospective echocardiographic data from healthy controls.

Study Population:

Twenty pSLE subjects ages 9–21 years inclusive, meeting American College of Rheumatology or Systemic Lupus International Collaborating Clinics classification criteria by age 18, were prospectively recruited at a tertiary academic center between October 2018 – June 2019 to undergo comprehensive cardiovascular testing during a single study visit. We excluded subjects with a history of chronic kidney disease stage 3 or greater, dialysis, kidney transplantation, or obstructive sleep apnea. We also excluded subjects with known hypertension at enrollment, but patients with a history of resolved hypertension diagnosis were eligible for inclusion.

As a comparator group, we included 70 healthy controls within the same age range from a retrospective cohort. The controls were originally sampled randomly from patients with structurally normal hearts who had undergone echocardiograms in the pediatric cardiology clinic for the primary indication of benign murmur or chest pain, and frequency matched by age and sex to a historical pSLE cohort, as previously described (23).

Study Procedures:

Echocardiography:

pSLE subjects underwent comprehensive transthoracic echocardiography on a Philips GE machine by one experienced cardiac sonographer (A.F.) using a standardized protocol, including 2D, M-mode, pulse Doppler, myocardial tissue velocities (tissue Doppler), and grayscale images for speckle-tracking analysis. LV myocardial deformation was determined by global longitudinal strain (GLS) averaged over apical 2-, 3- and 4-chamber views; longitudinal strain measured from a single apical 4-chamber view (LS_A4C); global circumferential strain (GCS) averaged over basal, midpoint and apical levels of the short axis view; and midpoint circumferential strain from the short axis view at the level of the papillary muscles (CS_mid). Only the single view assessments of strain (LS_A4C and CS_mid) were analyzed for healthy controls, as additional levels of the apical and short axis views were not routinely performed for clinical echocardiograms at the time they were obtained. While LS_A4C and CS_mid are more limited components of strain, published pediatric reference values for calculation of standard deviation scores (SDS) by body surface area (BSA) are available (24), and there is also data to suggest they are good approximations of global measures based on averages of three views (25). Strain measurements were analyzed off-line using Image Arena Version 4.6 (TomTec Imaging Systems, Munich, Germany) by an experienced research sonographer (Y.W.) with proven intra-observer reliability (26,27).

Standard measures of LV systolic function were assessed, including ejection fraction (LVEF) and shortening fraction (LVSF). We also assessed measures of LV diastolic function, including mitral E/A ratio (peak early over peak late diastolic filling velocity), septal and lateral e’ (early diastolic mitral annular tissue velocity), E/e’ ratio (peak early diastolic filling velocity over early diastolic mitral annular tissue velocity), isovolumetric relaxation time (IVRT), deceleration time (DT), and left atrial volume index (LAVI). All measurements were obtained as per American Society of Echocardiography guidelines.(28) IVRT was corrected for age.(29) LV mass indexed to BSA (LVMI) was obtained as a measure of LV remodeling. LVMI was considered abnormal if > 95^th%ile for sex (>45 and > 40 for males and females, respectively).(30)

Assessment of Vascular Health:

At the same study visit, SLE subjects underwent comprehensive cardiovascular testing, including manual auscultatory blood pressure (BP) measurement, 24-hour ambulatory blood pressure monitoring (ABPM), peripheral endothelial function testing (EndoPAT), pulse wave velocity and analysis, and carotid ultrasound for intima-media thickness. These study procedures have previously been described in detail (31). In short, 24-hour ABPM was performed using oscillometric Spacelab 90217 monitors (Spacelabs Medical, CA) according to American Heart Association guidelines for standard ambulatory assessment (32). Endothelial vasomotor function, as quantified by the digital reactive hyperemia index (lnRHI), was assessed in the fasting state using the EndoPAT device (Endo-PAT2000, Itamar-Medical, Caesarea, Israel), as previously described (33). To quantify aortic stiffness, carotid-femoral pulse wave velocity (PWV) and aortic augmentation index by pulse wave analysis (PWA) were measured using the SphygmoCor Vx system (AtCor Medical Pty Ltd, Australia) (34). High-resolution, real-time B-mode carotid ultrasounds were performed by a single experienced sonographer to obtain intima-media thickness (IMT) measurements in the far wall of the bilateral distal common carotid arteries, internal carotid arteries, and carotid bulbs. The mean of the bilateral common carotid measurements (CCA-IMT) was used as the primary measure of intima-media thickness (35,36). SDS by age and sex for CCA-IMT were calculated by the LMS method using published reference norms (37).

Study Measures:

The primary analysis was to compare pSLE subjects versus healthy controls, and the primary outcomes of interest included longitudinal strain, as measured by LS_A4C, and circumferential strain, as measured by CS_mid. By convention, both longitudinal and circumferential strain values are expressed as negative numbers, with a higher magnitude of the absolute value indicating better systolic function. Hence, less negative values are referred to throughout the manuscript as a reduction in strain, corresponding to worse systolic function. Additional covariates included age, sex, race, ethnicity, body mass index percentile (BMI%ile) for age and sex, BSA, heart rate, and mean arterial pressure (MAP) calculated from manual BP at the time of echocardiographic assessment.

The secondary analysis was cross-sectional comparisons using pSLE subjects only, and the outcomes included LS_A4C, CS_mid, GLS and GCS, since pSLE subjects had all strain measurements. Exposures of interest included a) persistent disease activity (SLE disease activity index (SLEDAI-2K) > 4), b) high cumulative disease burden (<50% of disease duration in a lupus low disease activity state (LLDAS) with a daily oral prednisone dose ≤ 7.5 mg), and c) measures of vascular health, including the magnitude of nocturnal BP dipping, lnRHI, PWV, augmentation index, and CCA-IMT SDS for age and sex.

Statistical Analysis:

Baseline demographics, clinical characteristics and echocardiographic measures were summarized using standard descriptive statistics and compared using Student’s t-test for continuous variables and chi-square tests for categorical variables. Shapiro Wilk tests were used to assess normality of continuous variables. To determine how well LS_A4C and CS_mid approximate GLS and GCS, respectively, we used intraclass correlation coefficients (ICC) for agreement.

We used separate multivariable linear regression models with stepwise selection to estimate the difference in LS_A4C or CS_mid strain values between pSLE subjects and controls, adjusted for demographic characteristics and body size. Additional covariates determined a priori to be potential confounders (MAP, heart rate) were forced into the models and retained if the coefficient for SLE changed by more than 10%. We also performed sensitivity analyses limited to pSLE subjects without hypertension or current glucocorticoid use.

In the cross-sectional analysis using pSLE subjects only, we used 2-sample t-tests to determine whether strain values differed by a) the presence of persistent disease activity, or b) high cumulative disease burden. We used Pearson correlation coefficients to assess the relationship between strain and other measures of vascular health, including peripheral endothelial function, nocturnal BP dipping, arterial stiffness, and carotid intima-media thickness. Coefficients between 0.5 – 0.7 were considered moderate correlations, while coefficients > 0.7 were considered strong. To account for multiple testing in the correlation analysis, we used the Benjamini-Hochberg procedure to control the false discovery rate (FDR). An FDR of 10% was chosen due to the small sample size in this pilot study. A pre-specified significance level of 0.05 (two-sided) was used for all other analyses. Statistical analyses were performed using STATA 15.0 (College Station, TX).

Results

Baseline Characteristics

Demographic and clinical characteristics of pSLE subjects compared to healthy controls are shown in Table 1. SLE subjects were on average one year older than healthy controls (mean age 16.4 vs. 15.2, p=0.04) (Table 1). They were also more often of non-white race or of Hispanic ethnicity, though the difference was not statistically significant (p=0.18).

Table 1.

Baseline Subject Characteristics

	SLE n = 20	Control n = 70	p-value*
Age, mean (SD)	16.4 (2.7)	15.2 (2.2)	0.04
Female, n (%)	17 (85%)	57 (81%)	0.71
Race
White	7 (35%)	40 (57%)	0.18
Black/African American	8 (40%)	14 (20%)
Asian	3 (15%)	6 (9%)
Other/Mixed	2 (10%)	10 (14%)
Hispanic ethnicity	3 (15%)	5 (7%)	0.30
BMI percentile for age-sex	68 (29)	57 (30)	0.13
Body surface area, m2	1.7 (0.2)	1.6 (0.3)	0.04
Heart rate, bpm	73 (11)	66 (13)	0.02
Mean arterial pressure,^ mmHg	83 (10)	81 (8)	0.28

^*Fisher’s exact or Student’s t-test as appropriate

^{^}Calculated from office-measured systolic and diastolic BP

Average SLE disease duration was 3.2 years (SD 2.1). One quarter (5/20) of pSLE subjects had a history of lupus nephritis, one subject had a history of serositis, and one had a history of neuropsychiatric SLE. Average disease activity at the initial onset of SLE was high (mean baseline SLEDAI of 12.4 (SD 7.3)). At the time of assessment, only 2/20 pSLE subjects had persistent disease activity (SLEDAI score > 4), and the mean SLEDAI was 2.9 (SD 4.4). A quarter of pSLE subjects were still using glucocorticoids (range 0.1 – 0.5 mg/kg/day of prednisone). On average, subjects spent 57% of disease duration in LLDAS (SD 32%), and 5/20 met criteria for higher cumulative disease burden (<50% of time in LLDAS). Additional disease characteristics of these subjects have been reported elsewhere (31).

Three pSLE subjects had a history of hypertension that had resolved, only one of which was still being treated with a renin-angiotensin system (RAS) blocker for nephritis. In total, 4/5 subjects with nephritis were being treated with RAS blockers at the time of assessment, and 2/15 subjects without nephritis were being treated with calcium-channel blockers for a history of Raynaud’s phenomenon.

Among the pSLE subjects, standard measures of systolic function (LVEF and LVSF) were within the normal range and similar to controls (Table 2). IVRT was longer in pSLE subjects compared to controls (p < 0.01), but none met ASE criteria for diastolic dysfunction. In contrast, 7/20 (35%) pSLE subjects had abnormal GLS values less negative than −16%. Accounting for BSA, 8/20 (40%) and 2/20 (10%) pSLE subjects had abnormal (SDS > 2.0) LS_A4C and CS_mid, respectively, compared to none of the controls. Agreement between GLS and LS_A4C was excellent (ICC 0.81, 95% CI [0.59 – 0.92]). There was also good agreement between GCS and CS_mid (ICC 0.73, 95% CI [0.43 – 0.89]).

Table 2.

Left ventricular function in pSLE subjects and controls

	SLE	Controls	p-valuea
E/A ratio, mean (SD)	2.3 (0.6)	2.2 (0.6)	0.62
e’ lateral	17.5 (2.6)	16.8 (3.2)	0.36
e’ septal	13.1 (1.9)	12.4 (1.6)	0.11
E/e’ lateral	5.6 (1.7)	5.5 (1.2)	0.72
E/e’ septal	7.4 (2.1)	7.4 (1.4)	0.91
E/e’ average	6.5 (1.8)	6.4 (1.2)	0.83
cIVRTb (ms)	72.6 (12.4)	54.4 (11.7)	<0.01
Deceleration time (ms)	199.0 (32.6)	-
Left Atrial Volume Index	25.1 (7.4)	-
LV Shortening Fraction	37.2 (4.9)	36.0 (3.8)	0.25
LV Ejection Fraction	63.2 (2.7)	64.1 (5.4)	0.50
Left Ventricular Mass Index	33.0 (7.0)	-
Global Longitudinal Strain	−17.2 (2.8)	-
AP4 Longitudinal Strain	−18.3 (3.2)	−21.8 (2.2)	<0.01
AP4 Longitudinal SDSc	1.7 (1.3)	0.5 (0.9)	<0.01
AP3 Longitudinal Strain	−15.3 (4.8)	-
AP2 Longitudinal Strain	−17.9 (2.9)	-
Global Circumferential Strain	−24.3 (3.8)	-
Base Circumferential Strain	−21.8 (5.0)	-
Midpoint Circumferential Strain	−24.8 (3.7)	−25.7 (3.4)	0.29
Mid-circumferential SDSc	0.4 (1.1)	0.1 (1.0)	0.25
Apical Circumferential Strain	−25.6 (5.8)	-

^aStuden’t t-test

^bIsovolumetric relaxation time corrected for age

^cStandard deviation score for body surface area (Adar et al. Am J Cardiol, 2019)

Strain in pSLE subjects compared to controls

LS_A4C was significantly reduced in pSLE subjects compared to controls (mean −18.3 (SD 3.2) vs. −21.8 (SD 2.2), p-value <0.001) (Figure 1). This was still true after adjusting for age, BSA, and BMI percentile in a linear regression model (β 3.6, 95% CI [2.3, 4.8], p < 0.001). Sensitivity analyses excluding the single pSLE subject with masked hypertension by 24-hour ABPM or pSLE subjects still using glucocorticoids resulted in similar estimates (β 3.6 and 3.2, respectively, p < 0.001). There was no significant difference in CS_mid between pSLE subjects and controls, either by unadjusted analysis (−24.8 (SD 3.7) vs. −25.7% (SD 3.4), p = 0.29) or after adjustment for BSA and heart rate (β 0.6, 95% CI [−1.2, 2.3], p = 0.51).

Figure 1.

Dot plot illustrating differences in the distribution of A) longitudinal apical 4-chamber and B) circumferential midpoint strain values between pSLE subjects and non-SLE healthy controls.

SLE disease activity and strain

There were no significant differences observed in GLS or GCS in pSLE subjects who had persistent active disease compared to those in remission (mean −16.0 (SD 3.5) vs. −17.3% (SD 2.8), p = 0.52 and −25.8 (SD 3.0) vs. −24.1% (SD 3.9), p = 0.57, respectively), albeit only 2 pSLE subjects met criteria for persistent active disease. Similarly, there were no significant differences by greater cumulative disease burden (GLS −16.7 (SD 2.4) vs. −17.4% (SD 3.0), p = 0.63 and GCS −25.8 (SD 2.2) vs. −23.7% (SD 4.1), p = 0.31).

Strain and other measures of cardiovascular health

Comprehensive cardiovascular testing demonstrated the presence of nocturnal BP non-dipping in 9/18 (50%) pSLE subjects despite otherwise normal 24-hour BP load. Peripheral endothelial function testing was abnormal (lnRHI ≤ 0.51) in 22% of 18 subjects with interpretable waveforms. PWV was increased in only 1/20 (5%), while CCA-IMT SDS for age/sex was greater than 2.0 in 12/20 (60%) of subjects (mean 2.4, SD 1.3).

There was a moderate correlation between reductions in the magnitude of nocturnal BP dipping and reduced GCS (Table 4). In contrast, there was no significant correlation between longitudinal strain and nocturnal BP dipping, or between strain values and arterial stiffness or carotid intima media thickness. There was only one subject with masked hypertension by 24-hour ABPM, in whom GCS, GLS and nocturnal BP dipping pattern were all normal. In a post-hoc analysis, clinical characteristics that were previously identified to be associated with nocturnal SBP non-dipping in this cohort were adjusted for in multivariable linear regression models of GCS, which did not change the estimated association between nocturnal SBP dipping and GCS (β −0.59, p = 0.02 and β −0.71, p<0.01 with adjustment for BMI percentile and high-density lipoprotein levels, respectively).

Table 4.

Correlation between strain and measures of vascular health

		GLS		LSA4C		GCS		CSmid
	N	𝒓	p-value	𝒓	p-value	𝒓	p-value	𝒓	p-value
lnRHI	18	0.45	0.06	0.16	0.52	−0.38	0.12	−0.25	0.32
%SBP dip	18	0.33	0.17	0.36	0.14	−0.59	0.01*	−0.49	0.04*
%DBP dip	18	0.10	0.71	0.03	0.92	−0.43	0.07	−0.52	0.03*
PWV SDS	20	0.01	0.95	0.05	0.82	−0.25	0.29	−0.01	0.97
CCA-IMT SDS	20	−0.28	0.22	−0.27	0.24	0.31	0.18	0.07	0.76

Pearson correlation coefficients describing the relationship between left ventricular global longitudinal, apical 4-chamber longitudinal, global circumferential, midpoint circumferential strain values and measures of vascular health.

lnRHI = natural log transformation of the reactive hyperemia index; %SBP dip = nocturnal decline in systolic blood pressure; %DBP dip = nocturnal decline in diastolic blood pressure; PWV = pulse wave velocity standard deviation score for age and sex; CCA-IMT SDS = mean distal common carotid artery intima-media thickness standard deviation score for age and sex

^*Statistically significant based on the Benjamini-Hochberg critical value for each strain parameter

Discussion

In our study, LV longitudinal strain was significantly decreased among pSLE subjects compared to controls, despite normal standard echocardiographic measures of systolic function. The magnitude of impairment observed was comparable to the 3.2 point difference reported previously by Dedeoglu et al. (22), and would be considered a clinically significant difference. Unlike the two previous studies of strain analysis conducted in pSLE, the majority of our subjects had achieved clinically inactive disease and were no longer taking glucocorticoids. As a result, the observed reductions in strain are unlikely due to acute secondary effects of active inflammation or concurrent glucocorticoid use. Furthermore, our findings are not explained by comorbid obesity or hypertension. In a recent meta-analysis of studies assessing strain in SLE, a higher prevalence of hypertension was thought to attenuate differences in strain estimates between cases and controls (38). However, all pSLE subjects in our study underwent 24-hour ABPM, the gold standard for hypertension assessment, and exclusion of the single subject with masked hypertension did not alter the results. Therefore, strain analysis may be a sensitive, clinically useful measure of myocardial impairment in pSLE patients even if they do not have comorbid cardiovascular risk factors.

In contrast to longitudinal strain, the magnitude of circumferential strain in pSLE subjects was comparable to controls. The lack of a detectable mean difference in circumferential strain was also noted in the pSLE population described by Leal et al (21). One of the proposed explanations is that mid-myocardial layers are last to be affected in progression of myocardial disease. In other subclinical cardiovascular disease states, such as adults with traditional cardiovascular risk factors, longitudinal deformation is also the first to be impaired, and is compensated for by increased circumferential shortening (39). As such, LVEF may be preserved despite evidence of systolic dysfunction by strain analysis (40). Reduced circumferential strain has been noted in several studies of adult-onset SLE (16,17) and adolescents with systemic hypertension (41). Therefore, adolescents with pSLE who have impaired longitudinal strain may need to be monitored more carefully for the development of reduced circumferential strain over time, particularly as they enter adulthood.

We did not observe a significant association between concurrent disease activity and strain, however this is likely due to the high proportion of subjects in complete remission. Previous studies in pSLE subjects with greater disease activity have demonstrated associations between persistent active disease and lower strain values in some LV segments (21,22). Our study emphasizes that significant reductions in longitudinal strain are still detected in children and adolescents with pSLE even in the setting of complete clinical remission without glucocorticoid use. This supports the hypothesis that strain does in part reflect damage, or cumulative effects of disease, and not merely a transient response to active systemic inflammation. Longitudinal strain has been shown to be an independent predictor of adverse cardiovascular outcomes in other disease states as well as adults in the general population with cardiovascular risk factors (42,43). As a result, strain measures may be useful to follow over time in patients with pSLE. Longitudinal studies are needed to determine whether subclinical impairments in strain are prognostic of SLE-related heart failure or coronary artery disease. It will also be important to establish which disease activity measures are more important determinants of strain, which will facilitate risk stratification for cardiac screening.

Lastly, although circumferential strain was on average comparable to controls in pSLE subjects, we observed a significant association between decreased circumferential strain and reduced nocturnal BP dipping. Attenuated nocturnal BP dipping is the loss of normal diurnal variation in BP from daytime to nighttime. In our previous work in this pSLE cohort, we observed that loss of nocturnal BP dipping was associated with increased cIMT and decreased peripheral endothelial function (31). It has also been shown to be an independent predictor of cardiovascular events and mortality in adults in the general population (44,45), and has been hypothesized to be a consequence of endothelial dysfunction (46). While we did not observe a statistically significant concurrent association between digital reactive hyperemia and strain, this may have been due to our sample size limitations and the greater biologic variability in peripheral endothelial function measures compared to nocturnal BP dipping. Relationships between myocardial strain and endothelial function have been observed in adults with systemic sclerosis (47,48), rheumatoid arthritis(49), and chronic kidney disease, which are thought to be mediated by coronary microvascular dysfunction (50). More importantly, both endothelial function and cardiac strain improve after interleukin-1 inhibition in adults with rheumatoid arthritis,(49) demonstrating reversibility of these measures and potential opportunities for intervention. As a result, the relationships between cardiac strain, endothelial function and nocturnal BP dipping, warrant further study in adolescents with pSLE.

To our knowledge, this is the first study to explore the relationship between LV strain and other measures of subclinical cardiovascular disease in pSLE, as well as the first study to evaluate strain in a racially diverse, North American cohort of pSLE patients. In addition, our study contributes to the understanding of the mechanisms of impaired myocardial function in this population. We also acknowledge several limitations of this study. First, we were unable to assess all strain components in the historical controls, as these were clinically obtained images that were not prospectively standardized for strain analysis. However, the standard views used for comparison are routinely obtained, and our estimates were reassuringly consistent with previously published studies. Secondly, the small sample size of pSLE subjects precluded multivariable adjustment for all covariates that could potentially confound the associations between strain and other measures of vascular health among pSLE subjects, and therefore our findings need to be reproduced in a larger study. Lastly, due to the cross-sectional nature of the cardiovascular assessment in pSLE subjects, we are unable to draw any conclusions regarding changes in strain values over time. Prospective longitudinal studies will be needed to determine if and when alterations in strain progress to overt myocardial dysfunction in this population.

In summary, strain analysis is feasible and may represent a more sensitive measure of systolic impairment in children and adolescents with pSLE than standard echocardiographic measures. Alterations in longitudinal myocardial deformation are present even in pSLE patients who are in clinical remission, and they are not explained by concurrent glucocorticoid use or hypertension. Therefore, we suggest considering the analysis and report of strain in patients with pSLE who are undergoing echocardiograms. There are several remaining knowledge gaps that would inform screening practices and interpretation of strain analysis, including the prognostic capability of strain in pSLE. In addition, the mechanisms of myocardial dysfunction and potential compensatory relationships between longitudinal and circumferential strain need to be explored further in the context of pediatric rheumatic diseases.

Table 3.

Left ventricular strain in pSLE subjects compared to controls

	Longitudinal strain			Circumferential strain
	𝛃	95% CI	p-value	𝛃	95% CI	p-value
SLE	3.59	[2.31, 4.87]	<0.01	0.39	[−1.43, 2.21]	0.67
Age	−0.24	[−0.55, 0.07]	0.13	−
BSA	3.02	[−0.18, 6.21]	0.06	2.68	[−0.19, 5.55]	0.07
BMI%ile	−0.02	[−0.04, 0.01]	0.15	−
Heart rate	−			0.02	[−0.03, 0.08]	0.45

Final multivariable linear regression models estimating the mean difference in longitudinal strain (apical-4 chamber) and midpoint circumferential strain between SLE subjects (n=20) and controls (n=70), adjusted for demographic characteristics and body size. BSA = body surface area; BMI%ile = body mass index percentile for age and sex