Assessment of right ventricular reserve utilizing exercise provocation in systemic sclerosis

Right ventricular (RV) capacity to adapt to increased afterload is the main determinant of outcome in pulmonary hypertension, a common morbidity seen in systemic sclerosis (SSc). We hypothesized that supine bicycle echocardiography (SBE), coupled with RV longitudinal systolic strain (RVLSS), improves detection of limitations in RV reserve in SSc. 56 SSc patients were prospectively studied during SBE with RV functional parameters compared at rest and peak stress. We further dichotomized patients based on resting RV systolic pressure (RVSP) to determine the effects of load on contractile response. Our pooled cohort analysis revealed reduced global RVLSS at rest (−16.2 ± 3.9%) with normal basal contractility (−25.6 ± 7.7%) and relative hypokinesis of the midventricular (−14.1 ± 6.0%) and apical (−8.9 ± 5.1%) segments. With exercise, global RVLSS increased significantly (p = 0.0005), however despite normal basal contractility at rest, there was no further augmentation with exercise. Mid and apical RVLSS increased with exercise suggestive of RV contractile reserve. In patients with resting RVSP < 35 mmHg, global and segmental RVLSS increased with exercise. In patients with resting RVSP ≥ 35 mmHg, global and segmental RVLSS did not increase with exercise and there was evidence of exertional RV dilation. Exercise provocation in conjunction with RVLSS identified differential regional contractile response to exercise in SSc patients. We further demonstrate the effect of increased loading conditions on RV contractile response exercise. These findings suggest subclinical impairments in RV reserve in SSc that may be missed by resting noninvasive 2DE-based assessments alone.


Introduction
Cardiac involvement in systemic sclerosis (SSc) is associated with increased morbidity and mortality, primarily due to the development of right ventricular (RV) dysfunction and associated pulmonary arterial hypertension (PAH) [1][2][3]. SSc patients with PAH (SSc-PAH) suffer disproportionately poor outcomes, with diminished response to treatment, worsened functional status, and increased rates of mortality in comparison to other PAH etiologies [3,4]. Although 2-dimensional echocardiography (2DE) is a useful screening tool in PAH given its high specificity and high positive predictive value [5,6], RV dysfunction and emerging PAH are often undetected or underestimated until late in the disease course [7]. Several clinically relevant 2DE-based metrics that have been frequently used in this population are right ventricular systolic pressure (RVSP), which allows for the noninvasive estimation of pulmonary arterial pressure (PAP) [8], the rate of change which has been previously associated with risk of developing PAH and outcomes in SSc [9], as well as tricuspid annular plane systolic excursion (TAPSE) [10,11], tissue doppler of the tricuspid annulus velocity [12], and fractional area change (FAC) [13,14]. A newer imaging modality, speckle-tracking echocardiography (STE), has several advantages over 2DE alone to allow for optimized RV imaging and assessment of both global and regional contractility [15,16]. STE can be used in conjunction with standard 2DE imaging, and by utilization of a softwarebased algorithm, is not limited by Doppler beam alignment angle or dependent on user technique [17].
In a prior cross-sectional study of SSc patients with and without PAH, we demonstrated both global and regional abnormalities in resting RV contractility in SSc patients that were independent of PAP and were not detectable by conventional 2DE measures alone [18]. These findings support the hypothesis that STE can detect occult abnormalities in RV myocardial function, and that abnormalities in RV contractility can develop in SSc patients regardless of PAP. In several recent studies from our group using invasive pressure-volume hemodynamics, we have demonstrated that at similar afterloads, SSc-PAH patients have depressed RV contractility [19] and diminished contractile response [20] to exercise provocation when compared with idiopathic PAH, due to an underlying sarcomeric defect in calcium-handling [21]. Taken together, these findings suggest that the RV is inherently diseased in SSc patients due to underlying sarcomere dysfunction, and is therefore unable to effectively adapt to the increased pressure load that occurs with the development of PAH [22].
Although there is increased attention on cardiopulmonary screening in SSc [23], risk prediction of cardiac involvement and PAH in SSc remains poor [24,25]. Existing screening recommendations may fail to detect early changes in RV contractility that signal impending PAH as standard 2DE metrics fail to identify occult contractile deficits [18], and the invasive nature of pressure-volume hemodynamics along with its expense and requirement of specialized technical expertise make it unfeasible as a general screening technique. Additionally, current methods to evaluate cardiovascular disease risk in SSc are mostly based on assessments made in the resting state; and as SSc patients frequently complain of dyspnea with exertion, the systemic response of the RV to exercise may carry key prognostic information regarding RV contractility, reserve capacity, and presence of early pulmonary vascular disease.
In the present study, we investigated whether RV longitudinal systolic strain (RVLSS), a novel noninvasive metric of regional and global RV contractile function derived by STE and measured during supine bicycle exercise stress, provides an improved detection method over conventional measures alone to unmask RV contractile defects in at-risk SSc patients. We hypothesized that supine bicycle exercise echocardiography (SBE) coupled with RVLSS is an improved noninvasive tool to detect impaired RV reserve in SSc.

Patient population
Our study was approved by the Johns Hopkins Medicine Institutional Review Board. Consenting SSc patients seen at the Johns Hopkins Scleroderma Center for routine clinical care between Jan 2009 and Dec 2018 were eligible to participate in this single center observational study. All participants met 1980 or 2013 American College of Rheumatology classification criteria for SSc [26,27], or had at least 3 of 5 CREST (calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasia) criteria. Our center's standard practice is to perform annual pulmonary function and resting echocardiography testing in SSc patients to screen for cardiopulmonary complications [6].
We included SSc patients deemed at risk for PAH as identified by: resting RVSP ≥ 40 mmHg on a routine screening echocardiogram with associated dyspnea, an RVSP ≥ 45 mmHg on routine screening echocardiogram regardless of symptoms, an isolated decline in diffusing capacity (DLCO) ≥ 10% predicted from baseline, or new onset unexplained dyspnea. Unexplained dyspnea was defined as dyspnea without significant anemia (hematocrit < 28%), symptomatic interstitial lung disease, significant chronic obstructive pulmonary disease (FEV1/FVC 1 3 ratio < 0.7 with history of smoking), or a left ventricular (LV) ejection fraction < 50%. SSc patients were enrolled if deemed able to exercise by their referring provider. Patients with technically adequate echocardiographic image quality at baseline and peak stress were included for analysis.
Classification of SSc cutaneous subtype was defined by established criteria [28], and SSc disease duration was calculated as the time interval between the first scleroderma symptom (either Raynaud's or non-Raynaud's) and the exercise echocardiogram date. Measurements of forced vital capacity and diffusing capacity were standardized by age and gender [29,30].

Exercise stress protocol
Supine bicycle stress echocardiograms were performed utilizing a single supine ergometer (Medical Positioning Inc, Kansas City, MO) at a single clinical site, Johns Hopkins Bayview Medical Center. Patients initiated exercise at 25 Watts and increased by 25 Watts every 3 min until achieving 85% of their age-predicted maximum heart rate or until limited by symptoms. Continuous electrocardiographic monitoring was performed throughout study duration with heart rate and blood pressure obtained at each stage of exercise. Cardiac output was derived utilizing the following formula: (pulse pressure × heart rate) * 0.002 [31]. An exercise physiologist, cardiac sonographer, and physician's assistant specialized in Cardiology were present throughout the study duration. 2D echocardiographic images and measurements of right and left ventricular chamber size and function were obtained at baseline and within 1 min of peak exertion.

Echocardiographic acquisition and measurements
Echocardiograms were performed using Phillips ie33 or Epiq seven ultrasound machines (Phillips Healthcare, Andover, MA) with subjects in the left lateral decubitus position during image acquisition at 70-90 frames per second at end-expiration. 2D-directed methods were taken at enddiastole to obtain linear and volumetric measurements of the RV chamber in accordance with American Society of Echocardiography (ASE) guidelines [8,32]. Right atrial area (RAA) was estimated using volumetric area from the apical 4-chamber view. Right ventricle (RV) function was assessed using TAPSE and FAC, with abnormal defined as < 16 mm and < 35% respectively. Tricuspid regurgitant (TR) velocity was used to estimate RVSP using the modified Bernoulli equation and adding estimated RA pressure based on inferior vena cava dimension and collapsibility with sniff [33,34]. Left ventricle (LV) systolic function was estimated using Simpson's biplane method of discs to calculate ejection fraction (EF). Left ventricle (LV) diastolic parameters including mitral inflow with early diastolic (E) and late diastolic (A) velocities, and tissue doppler medial e' velocities were obtained [35].
STE-based strain analysis was performed of the RV using commercially available strain software (Epsilon EchoInsight Version 3.1.0.3358, Milwaukee, WI). From the 4-chamber apical view, peak systolic longitudinal strain of the RV free wall segments was obtained by tracing the RV chamber endocardial borders in end-systolic still frames [15,16,36]. In post-processing, automated tracking was visually verified and manually adjusted to ensure adequate border delineation and segment thickness. Peak longitudinal systolic strain was defined as the difference in shortening from the region of interest relative to original length, and by convention, is expressed as a negative percentage with decreased strain refers to a less negative number (i.e. a lower absolute value) than expected, while increased strain refers to a more negative number (i.e. a higher absolute value [37]. Global RVLSS was defined as the average of regional strain from the basal, midventricular, and apical RV free wall segments, and compared to published standard reference values [37,38]. RV reserve was defined as a statistically significant increase in strain and 2DE-based functional measures from rest to stress [39,40]. Two board certified echocardiologists (MM, VM) who were blinded to clinical variables performed conventional and speckle-based strain analysis to determine intra-and inter-observer variability.

Statistical analysis
Echocardiographic and hemodynamic parameters were compared at rest and at peak stress using a paired t test for parametric data, and Wilcoxon-Mann-Whitney test for nonparametric data. Statistical analyses were performed using STATA version 15.0 (College Station, TX, USA). Statistical significance was defined by a two-sided p value < 0.05. Intra-observer and inter-observer variability were assessed by intraclass correlation coefficient.

Baseline patient characteristics
Fifty seven patients were recruited, and 1 patient was not analyzed due to poor image quality at peak exertion limiting strain analysis. The final study population consisted of 56 SSc patients with technically adequate stress bicycle echocardiograms that were predominantly female (87.5%) and white (74.1%), with a mean age of 55.9 ± 12.1 years and SSc disease duration of 16.0 ± 10.9 years. The majority of our patients had limited cutaneous disease (39 patients, 69.6%) while 30.4% of SSc patients had diffuse disease. Additional characteristics of participants are detailed in Table 1. Based on our inclusion criteria, 9 patients (16.1%) were referred for RVSP ≥ 40 mmHg on a routine screening echocardiogram with associated dyspnea, 4 patients (7.1%) for RVSP ≥ 45 mmHg regardless of symptoms, and 12 patients (21.4%) for an isolated decline in DLCO ≥ 10% predicted from baseline. The majority of the patients (31 patients, 55.4%) were referred for new onset unexplained dyspnea. Baseline echocardiographic images and medical records were reviewed and no patients in our cohort had evidence of hemodynamically significant valvular disease (any stenosis and/or regurgitation greater than mild in severity), ischemic, dilated or hypertrophic cardiomyopathy, and/ or evidence of congenital heart disease.

Resting 2D echocardiographic and speckle-based strain characteristics
During rest, all SSc patients had normal LV ejection fraction and normal left atrial areas as shown in Table 2. The mean mitral inflow pattern ratio of early (E) and late (A) diastolic velocities was 1.07 ± 0.30 with normal resting LV filling pressures (septal E/e' 10.8 ± 3.7, n = 46). Three SSc patients (6.5%) had septal E/e' ≥ 15 at baseline suggestive of elevated LV end-diastolic filling pressures at rest [38]. Despite having elevated E/e' ratios at rest, none of these patients met all criteria for the definition of diastolic dysfunction, rendering their classification as "indeterminate" diastolic function.
RAA was normal in size as were linear measures of RV chamber size including mid and distal RV outflow tract and basal, mid, and longitudinal dimensions. Resting RV enddiastolic area (RVEDA) and end-systolic areas (RVESA) were top normal. Specifically, 22 [38]. In terms of regional strain, RVLSS was normal at rest in the basal RV free wall segments (−25.6 ± 7.7%) relative to the midventricular (−14.1 ± 6.0%) and apical (−8.9 ± 5.1%) segments [18].

Exercise stress hemodynamics and echocardiographic findings
SSc patients were exercised to 85% MPHR, and on average achieved 81 ± 28 Watts consistent with functional class II level of exertion [41]. With stress, LV ejection fraction appropriately increased to 73 ± 6% as shown in Table 3. Cardiac output at peak exertion was normal at 18.8 ± 6.1 L/ min. There was no difference in resting and peak septal E/e' values in the 23 patients for which both values were available, p = 0.23. With stress, there were increases in TAPSE (p < 0.0001) and FAC (p = 0.0001). We noted a statistically significant increase in RVSP with exertion, from resting value of 30.6 ± 8.9 mmHg to peak 49.9 ± 12.7 mmHg (p < 0.0001). With exercise, global RVLSS increased significantly from a diminished baseline of −16.2 ± 3.9% to −18.9 ± 4.2% (p = 0.0005). Despite the basal RVLSS being normal at rest, there was no significant augmentation of basal contractility with exertion as shown in Fig. 1. On the other hand, RVLSS of the mid and apical RV segments was diminished at baseline and increased with exercise (p = 0.02 and p = 0.001, respectively).
As the majority of our patients were referred for unexplained dyspnea, we sought to understand whether the In the 38 patients with a resting RVSP < 35 mmHg, 35 patients had paired data available for analysis of both resting and peak strain parameters. Global RVLSS was diminished at rest and increased significantly with exercise (rest −15.3 ± 3.5% vs. peak −19.0 ± 4.5%, p = 0.0001) with augmented contractility across all three RV free wall segments: base (p = 0.05), midventricular (p = 0.002), and apical RVLSS (p = 0.0003). RVEDA and RVESA areas did not change significantly from rest to stress as shown in Table 4.
When comparing the low load (RVSP < 35 mmHg) and high load (resting RVSP ≥ 35 mmHg) groups (Table 4), we observe that the changes in midventricular, apical and global RVLSS from rest to peak stress are statistically significantly   Tables 2 and 3 are due to the use of paired data samples in Table 3 Parameter higher in the low load group than the high load group. Of note, the mean resting midventricular and apical RVLSS in the high load group is roughly equivalent to the strain values at peak exertion in the low load group, suggesting that these segments are contracting maximally at rest in the setting of a high afterload. The three patients with elevated E/e' ratio at baseline, a noninvasive marker of elevated LV end-diastolic pressure, also had elevated resting RVSP and were among the 18 patients in the high afterload RVSP group. As these patients did not meet all criteria for the classification of "abnormal" diastolic function, a single parameter (elevated E/e' ≥ 15) was used to determine if LV diastolic function affected RV contractile reserve. Resting septal E/e' (as a single parameter) was not associated with change in basal (p = 0.46), midventricular (p = 0.46), apical (p = 0.32), or global RVLSS (p = 0.14) from rest to peak stress. Bold values signify statistical significance, p-value < 0.05 Absolute and percent changes for each parameter within group is demonstrated with (*) signifying statistical significance of the rest-peak difference Comparisons for each parameter between groups are shown with bold print signifies statistical significance, p value ≤ 0.05. Absolute change was calculated as absolute value (stress-rest difference) and percentage change of that difference from rest Statistically significant differences within groups are shown with delta and percent change in parameter between rest-stress: * p ≤ 0.05, ** p ≤ 0.01,

Discussion
In the present study, we sought to investigate whether RVLSS, a novel noninvasive metric of regional and global RV contractile function derived by STE, measured during supine bicycle exercise, provides an improved detection method over conventional 2DE measures alone to identify RV contractile dysfunction in at-risk SSc patients. SSc patients in our study were primarily referred for SBE based on the indication of unexplained dyspnea. At rest, mean RVSP was normal with normal conventional 2DE derived measures of RV contractility (TAPSE, FAC). With exertion, there were increases in these conventional measures suggesting normal RV contractile response to exercise provocation. However, the innovative application of STE-based metrics revealed both resting and exercise abnormalities in RV contractile function. At rest, global RVLSS was diminished with evidence of regional heterogeneity in which there was normal contractility of the basal RV free wall segment with relative hypokinesis of midventricular and apical segments. With exercise, there were significant increases in global RVLSS, largely due to regional increases in midventricular and apical contractility, suggestive of myocardial reserve in these segments. Interestingly, despite normal resting basal contractility, there was no further augmentation of basal segment contractility with exercise. These findings suggest that at rest, the basal segment is already contracting maximally and is unable to augment further with exercise provocation.
Overall, these findings demonstrate that SSc patients have a differential RV regional contractile response to exercise, suggesting limitations in RV reserve that are missed by conventional 2DE. We also examined the effects of loading conditions on RV contractility and found differential contractile response based on resting loading conditions. In patients with low load, defined as resting RVSP < 35 mmHg, global RVLSS increased with exertion due to augmented contractility across all three RV free wall segments, including the initially hypokinetic midventricular and apical segments. In patients with high load, defined as RVSP ≥ 35 mmHg, global RVLSS did not augment significantly from baseline and there were no regional increases in RVLSS of the basal, midventricular, or apical segments. These findings suggest that the RV in high loading conditions may already contracting at its maximal capacity at rest and is unable to augment further with exercise.
Additionally, we found significant increases in RV enddiastolic areas with exercise in the high load group; this is in contrast to the expected normal contractile response to exercise in healthy adults, in which end-systolic and enddiastolic areas decrease with exertion. Our findings are consistent with several prior studies from our group using invasive pressure-volume assessments showing that SSc-PAH patients have depressed RV contractility compared with patients with idiopathic PAH (IPAH) at similar levels of afterload [19], and maintain stroke volume and augment cardiac output by dilatation of the RV chamber [20]. We also found that there were minimal differences in peak RVSP between the low and high resting load groups suggesting that contractile differences are not entirely due to a primary load phenomenon at the time of exercise and rather, represent a complex interplay between afterload and maladaptive RV response to changes in afterload seen with exercise. In fact, we have previously demonstrated differential myofilament contractility and calcium sensitivity in SSc [42] that may account for intrinsic contractile abnormalities at rest in addition to changes to the pulmonary vascular bed inherent to the disease process that are further unmasked with exercise provocation. To our knowledge, our study is the first to demonstrate the utility of novel noninvasive STE-based metrics that establish the failure of the scleroderma RV to augment contractility with high loading conditions, resulting in RV chamber dilatation. STE in conjunction with exercise provocation may provide crucial clinical information by identifying SSc patients with limitations in RV reserve and emerging pulmonary vascular disease.
There were several limitations to our study. First, this was a single-center prospective study with a relatively small sample size as exercise studies were performed based on clinical indications. There was a lack of correlation with simultaneous invasive hemodynamics and absence of a control group. Moreover, there are currently no accepted reference values for conventional and STE-based contractile measures with exercise nor an established definition of normal RV reserve utilizing echocardiographic parameters, representing an important priority for future studies. Recently, there has been a notable change in the invasive hemodynamic definition of PAH [43], with mean PAP ≥ 21 to < 25 mmHg representing an intermediate form of disease that is at high risk for emerging PAH. However, there has not been a corresponding change to the definition of pulmonary hypertension by echocardiography. While understanding that this is a continuum of disease, we chose to dichotomize our SSc patients based on the current ASE recommendations for the noninvasive cut-off of RVSP ≥ 35 mmHg for the purposes of 1 3 this study [8]. The definition, natural history, and long term clinical significance of exercise-induced PH remain unclear, and future studies are sorely needed to validate a non-invasive screening algorithm to identify patients with emerging pulmonary vascular disease in this at-risk population. Further, while we have demonstrated that exercise provocation may have a role in unmasking abnormal contractile response in at-risk SSc patients, we cannot assess temporality in this cross-sectional study. In a recent study using invasive hemodynamics, several parameters such as peak pulmonary vascular resistance, transpulmonary gradient, and mPAP/cardiac output ratio were found to be predictors of age-adjusted long-term mortality in SSc patients with mild PH [44]. However, prospective longitudinal studies utilizing noninvasive metrics that incorporate contractility and load are needed to determine if exertional changes in contractility precede resting elevations in pulmonary pressures thereby identifying SSc patients at risk for resting PAH. Another important area of future investigation is whether strain metrics associate with gold standard invasive measures of contractility and can serve as a target for disease-modifying therapies as a noninvasive surrogate of RV contractility. Lastly, given the well-described vendor-specific variability in strain measures, there may be inherent differences that limit the generalizability of our findings [45].
In this prospective study of at-risk SSc patients, we demonstrated STE-based abnormalities in RV contractile response to exercise that was not detectable by conventional measures alone and may identify SSc patients with limitations in RV reserve and poor clinical outcomes. We further demonstrated the effect of resting loading conditions on RV contractile response to exercise in which the RV chamber differentially dilates with no further augmentation in contractility in those with elevated RVSP at rest. These findings suggest a maladaptive contractile response of the RV when resting loading conditions are elevated. The significance of our study lies in the use of exercise provocation in conjunction with STE measures, an important noninvasive tool, in assessing the degree of RV contractile reserve in SSc patients at risk for PAH. Our findings of impairments in RV reserve may have clinical implications for the use of exercise provocation in screening for emerging pulmonary vascular disease in at-risk SSc patients.