Echocardiographic Prediction of Hemodynamics During Acute Vasoreactivity Testing in Pediatric Pulmonary Arterial Hypertension

Abstract

Purpose:

In pediatric pulmonary arterial hypertension (PAH), echocardiography is frequently used to assess response to acute vasodilator therapy. We sought to evaluate whether echocardiography is sensitive enough to detect significant changes in hemodynamics using two commonly used thresholds (Barst and Sitbon criteria) as benchmarks to establish a positive response.

Methods:

Simultaneous echocardiography and invasive hemodynamics were obtained in 71 children with PAH at their baseline and maximal vasodilatory conditions. The change in echocardiographic parameters was used to predict AVT responder status as determined by Barst and Sitbon criteria.

Results:

There were strong positive correlations between invasive hemodynamics and echocardiographic markers of pulmonary hypertension. Between the baseline and AVT conditions, there were significant reductions in the tricuspid valve regurgitation velocity (TR Vmax), right ventricular systolic-to-diastolic ratio (S/D ratio), and systolic and maximal eccentricity index (EIs and EIm). Barst criteria responders had more pronounced reductions in S/D ratio and EIm. A 22.5% decrease in TR Vmax, 48.2% in S/D ratio, and 29.0% in EIm were associated with positive Barst responder status. The degree of change in echocardiographic variables was not strongly associated with clinical outcomes.

Conclusions:

Using simultaneously obtained invasive hemodynamics and echocardiography, this is the first study to demonstrate that there are appreciable changes in echocardiographic markers of pulmonary hypertension during acute vasoreactivity testing (AVT) in children. The degree of change in three of these parameters (TR Vmax, S/D ratio, and EIm) was the most useful in identifying those with a more pronounced response to pulmonary vasodilatory therapy. Future studies are needed to validate these findings and establish a prognostic significance.

1 |. Introduction

Left untreated, the unrelenting progression of pulmonary arterial hypertension (PAH) in children and adolescents carries a very poor prognosis [1]. Although advances in evaluation and management strategies have improved outcomes, mortality remains high for pediatric patients with PAH [2, 3]. Accurate and timely diagnosis is therefore critical for understanding prognosis and guiding therapeutic decision-making for these patients.

Invasive hemodynamic testing with acute vasoreactivity testing (AVT) during cardiac catheterization remains the gold standard for the diagnosis of PAH in children and adolescents [4–6]. AVT is typically performed by administering either oxygen and/or inhaled nitric oxide (iNO) after the baseline set of hemodynamics is obtained [4, 7]. It is particularly useful to help identify patients who have a robust hemodynamic response to pulmonary vasodilators and more favorable survival after treatment with calcium channel blockers (CCB) [8, 9]. A positive response to AVT is based on well-established hemodynamic thresholds, typically either the Barst or Sitbon criteria [10–13]. Secondarily, AVT performed at the time of diagnosis and serially across the time-course of the disease aids in prognostication for pediatric patients with PAH [14].

It is important to note, however, that this type of testing in children and adolescents is not a simple undertaking. Invasive cardiac catheterization in this patient population carries significant risks, often associated with the need for anesthesia [15–17]. This risk and financial cost is only amplified when you consider that pediatric patients with PAH require serial evaluations over the course of their lives [18]. As a result, in children, it is not uncommon that invasive hemodynamic testing and AVT are either deferred or not performed under “optimal” circumstances (such as in patients who are started on pulmonary vasodilator therapies before testing) [19]. Interest has therefore grown in developing more reliable, non-invasive alternatives for hemodynamic testing, which would allow for more frequent, lower-risk evaluations.

Transthoracic echocardiography has long been employed to estimate pulmonary pressures and evaluate ventricular function in pediatric PAH [20]. However, the current literature lacks objective data that describe what happens acutely to echocardiographic parameters after the initiation of pulmonary vasodilation [21].We therefore sought to evaluate if standard and speckle-tracking derived echocardiographic estimates of ventricular function and pulmonary pressures changed during AVT, and if such changes relate to simultaneous invasive hemodynamics. Furthermore, using the Barst and Sitbon criteria as a hemodynamic benchmarks for what is generally considered a significant response to acute pulmonary vasodilation, we sought to evaluate if changes in echocardiographic metrics could be used to predict AVT responder status.

2 |. Materials and Methods

2.1 |. Study Demographics

Children and adolescents with known or suspected PAH were prospectively enrolled and underwent their clinically indicated RHC for hemodynamic assessments between November 2008 and October 2021 at Children’s Hospital Colorado. Simultaneous two-dimensional transthoracic echocardiograms were obtained during the RHC. A total of 95 patients with suspected or previously diagnosed pulmonary hypertension were enrolled and had simultaneous echocardiography during RHC. For the purposes of this study, data were analyzed retrospectively. We included patients who either previously had or currently had resting mean pulmonary artery pressures of ≥20 mmHg and pulmonary capillary wedge pressure of ≤15 mmHg as measured by invasive RHC. As echocardiographic metrics were to be analyzed with a within-subject design, we elected to include patients with mPAP <20 mmHg on the current assessment as long as they previously had met diagnostic criteria for PAH by catheterization. This was also an effort to better reflect the heterogeneity of this population, especially with repeat hemodynamic assessments. Similarly, patients with open intracardiac shunts were also included in this analysis. Patients who did not undergo AVT, those who underwent hypoxia challenges, and those with incomplete imaging in either condition were excluded, which left 71 patients in total. The study was approved by the Institutional Review Board, and written informed consent was obtained from the patient or legal guardian.

2.2 |. Right-Heart Catheterization

All patients underwent clinically indicated RHC under general anesthesia. Operators were blinded to echocardiographic measures. Results of RHC were obtained by chart review. If intracardiac shunts were present, the modified Fick equation was used for calculations of cardiac index. In the absence of shunting, thermodilution was used. Standard techniques were used to measure pressures in the right atrium, right ventricle, pulmonary artery, femoral artery, and either pulmonary capillary wedge pressure (PCWP) or left atrium. Systemic (Qs) and pulmonary (Qp) blood flow were documented. Transpulmonary gradient (using either PCWP or left atrial pressure) and pulmonary vascular resistance indexed to body surface area (PVRi) were calculated. Operators waited at least 10 min after initiation of pulmonary vasodilation before measuring hemodynamics during AVT.

Ideally, patients underwent AVT under three conditions: 21% FiO2, 100% FiO2, and 100% FiO2 with 40 ppm of iNO. Given that these were clinically indicated cardiac catheterizations, some patients’ baseline condition was already breathing 100% oxygen (typically in those who use supplemental oxygen at baseline), and they only underwent one additional condition for AVT with the addition of iNO. Data were analyzed at the patient’s baseline condition and compared to the condition in which they received the maximal amount of pulmonary vasodilator medications. A patient was considered a positive AVT responder by Sitbon criteria if mPAP declined by ≥10 mmHg, reaching a level less than 40 mmHg with an increase or no change in cardiac output. Barst responders were defined as those who had ≥20% decrease in mPAP, with an increase or no change in cardiac index (CI).

2.3 |. Echocardiography

Transthoracic two-dimensional echocardiograms were obtained simultaneously with invasive hemodynamics (with the catheter in the main pulmonary artery) in both conditions. Standard images were obtained in the apical 4-chamber and 2-chambers views using General Electric Vivid 7 or E9, or E95 system using either the M5Sc-D, 6S-D, 12S-D phased array transducers (General Electric Healthcare). Individuals making echo measurements were blinded to the invasive hemodynamics and clinical outcomes. Measurements were made on 3–5 beats to account for beat-to-beat variability, and then averaged. The maximum velocity of the tricuspid regurgitation jet (TR Vmax) was measured only in patients with complete Doppler envelopes. Tricuspid Annular Plane Systolic Excursion (TAPSE) was measured using M-mode tracing from the apical 4-chamber. The right ventricular systolic-to-diastolic ration (S/D ratio) was calculated using ratio of the duration of tricuspid regurgitation (RV systole) to the duration of tricuspid valve inflow (RV diastole) as measured on continuous wave Doppler from the apical 4-chamber view [22]. LV and RV volumes were measured in the apical 4-chamber and an FAC was obtained for the RV, and ejection fraction calculated for the LV using Simpson’s Biplane (rule of disc’s). Systolic eccentricity index (EIs) was measured from the parasternal short axis at the level of the papillary muscles as the ratio of the diameters parallel and perpendicular to the interventricular septum at end-systole. Maximal eccentricity index (EIm) was similarly measured, however, performed at the maximal point of septal flattening that occurs when the pressure loaded right ventricle continues to contract during the left ventricular isovolumetric relaxation phase [23].

Speckle-tracking echocardiography was performed on the apical 4-chamber images offline using EchoPac software (Version 113; GE Healthcare). 3–5 beats were again measured and averaged to account for beat-to-beat variability. The endomyocardial border was traced to establish a region of interest. If >4 segments appeared to track well by both visual inspection and by EchoPac assessment, the curve was accepted. Curves that did not track well despite manual adjustments of the region of interest were not used. Left ventricular end-systole was defined by the aortic valve closure time that had been calculated with pulse-wave Doppler.

2.4 |. Outcome Measures

Patient’s electronic medical records were reviewed between the onset of their initial study and February 2022. Patient outcomes, including death, heart and/or lung transplant, and need for extra corporeal membrane oxygenation (ECMO), were recorded.

2.5 |. Statistical Analysis

Statistical analysis was performed using SAS software, Version 8 (SAS Institute Inc., Cary, NC, USA). Data were assessed for normality using the Shapiro-Wilk test. The majority of the data were not normally distributed, so continuous variables are presented as a median with a range unless otherwise indicated. For categorical variables, data are presented as frequencies and percentages. Wilcoxon signed-rank tests were used to compare the distribution of medians in matched data. Chi-squared testing was used for categorical variables. Exact methods were used when appropriate. Spearman correlation coefficients were used to assess relationships between continuous variables. Univariate logistic regression analysis and receiver operating characteristic (ROC) curves were utilized to assess the ability of the percent change in an echocardiographic variable to predict either Barst or Sitbon responder status, as well as the combined clinical outcome of death or transplant. Optimal cut-off values for echocardiographic predictors of responder status were determined by Youden’s index. Multivariate Cox proportional models were attempted but were significantly limited by overfitting given the small sample size, so they were not utilized in this analysis.

3 |. Results

3.1 |. Patient Characteristics

Full patient demographic data are available in Table 1. For this analysis, 71 patients were identified who had full echocardiographic assessments in at least two conditions with AVT. The median age was 9 years. The youngest patient was 13 months of age, and the eldest was 23 years old. The majority of patients were female (61.4%). This cohort encompassed a wide variety of PAH etiologies, and one quarter (18% or 25.7%) were studied during their initial diagnostic catheterization, meaning the majority of patients had previously undergone diagnostic RHC to confirm the diagnosis of PAH. Of the patients who had undergone prior catheterization, 17 (25.4%) had previously been categorized as a CCB-eligible AVT responder.

TABLE 1 |.

Patient demographic and clinical characteristics.

Patient demographics	All patients (N = 71)
Age, y	9.0 (1–23)
Age, mo	114 (13–1274)
Female	43 (61.4%)
Weight, kg	33.2 (20.0–57.0)
Etiology of pulmonary hypertension
Group 1 (idiopathic, heritable, CHD)	49 (69%)
Group 3 (lung/hypoxia related)	12 (17.1%)
Group 4 (chronic thromboembolic)	1 (1.4%)
Group 5 (multifactorial)	7 (11.4%)
Initial diagnostic catheterization	18 (25.7%)
Time since initial diagnostic RHC, y	2.1 (0–17.18)
History of repaired CHD	9 (12.9%)
Right-to-left shunt at the time of Cath	5 (8.1%)
History of prior positive response to	17 (25.4%)
vasoreactivity testing
Baseline oxygen use	18 (25.3%)
PAH medications
None	12 (17.0%)
Single therapy	20 (28.1%)
Dual therapy	20 (28.1%)
Triple therapy	16 (22.5%)
CCB monotherapy	3 (4.2%)
Medication classes
PDE-5-inhibitors	55 (78.6%)
Endothelin receptor antagonists	32 (45.7%)
Prostacyclin analogs	21 (30.0%)
Calcium channel blockers	13 (18.6%)
WHO class
I	21 (32.8%)
II	33 (51.6%)
III	10 (15.6%)
Need for transplant (heart and/or lung)	2 (3.0%)
Required ECMO	2 (2.9%)
Death	6 (8.7%)

Note: Values are given as median (range) or n (%).

Abbreviations: CCB, calcium channel blocker; CHD, congenital heart disease; PDE-5, phosphodiesterase-5; RHC, right heart catheterization; WHO, World Health Organization.

Most patients in this cohort were being treated with targeted pulmonary vasodilator therapies. Just 12 (17.0%) were not on targeted PAH medications. Within the study period, there were two patients who required heart or lung transplantation, two who required ECMO therapy, and six who died. Both patients who underwent transplant also died.

3.2 |. AVT

3.2.1 |. Invasive Hemodynamics

Hemodynamic measurements at baseline and maximal AVT conditions are presented in Table 2. Between conditions, there were significant declines in the measured RV systolic pressure (RVSP; 56.0 mmHg [24.0–104.0] to 49.0 mmHg [22.0–107.0], p < 0.001), systolic pulmonary artery pressure (sPAP; 52.0 mmHg [20.0–112.0] to 46.0 mmHg [20.0–107.0], p < 0.001), mean pulmonary arterial pressures (mPAP; 32.0 mmHg [13.0–71.0] to 27.0 mmHg [13.00–67.00], p < 0.001), transpulmonary gradient (TPG; 23.0 mmHg [7.0–65.0] to 17.0 mmHg [6.00–59.00], p < 0.001), and pulmonary vascular resistance (PVRi; 6.71 Wu*m2 [1.48–31.25] to 4.51 Wu*m2 [0.47–35.33], p < 0.001). There were no significant changes in the cardiac output. The systemic vascular resistance (SVR) increased between conditions (15.54 Wu*m2 [6.93–33.96] to 16.22 Wu*m2 [7.26–33.96], p < 0.001) and there was an associated decline in the PVR:SVR ratio (0.42 [0.09–1.33] to 0.31 [0.04–1.31], p < 0.001). There were no significant changes in the right atrial or left atrial pressures, nor in the RV end-diastolic pressure.

TABLE 2 |.

Results of invasive hemodynamic measurements and echocardiographic-derived parameters for the study cohort at the baseline condition and after acute vasoreactivity testing, as well as the median change between the two conditions.

Variable	n	Baseline median (range)	Vasoreactivity testing median (range)	Median change	p value
Invasive hemodynamics
RAP (mmHg)	68	6.00 (3.00–14.00)	7.00 (3.00–14.00)	0.00 (−5.00 to 4.00)	0.19
RVSP (mmHg)	49	56.00 (24.00–104.00)	49.00 (22.00–107.00)	5.00 (−8.00 to 24.00)	<0.001**
RVEDP (mmHg)	49	8.00 (3.00–14.00)	8.00 (4.00–14.00)	0.00 (−5.00 to 4.0)	0.90
sPAP (mmHg)	71	52.00 (20.00–112.00)	46.00 (20.00–107.00)	5.00 (−9.00 to 25.00)	<0.001**
mPAP (mmHg)	71	32.00 (13.00–71.00)	27.00 (13.00–67.00)	4.00 (−11.00 to 21.00)	<0.001**
PCWP (mmHg)	71	9.00 (4.00–15.00)	9.00 (3.00–15.00)	0.00 (−4.00 to 5.00)	0.09
TPG (mmHg)	71	23.00 (7.00–65.00)	17.00 (6.00–59.00)	5.00 (−10.00 to 22.0)	<0.001**
PVRi (Wu*m2)	71	6.71 (1.48–31.25)	4.51 (0.47–35.33)	1.56 (−5.16 to 10.07)	<0.001**
MAP (mmHg)	70	59.00 (45.00–78.00)	60.00 (47.00–101.00)	−2.00 (−25.00 to 11.00)	<0.001**
QP	71	3.44 (1.92–10.83)	3.50 (2.23–6.64)	0.12 (−3.19 to 10.36)	0.08
QS	71	3.39 (2.23–6.64)	3.41 (1.74–6.64)	0.06 (−2.12 to 3.19)	0.93
QP:QS	71	1.00 (0.84–2.40)	1.00 (0.84–4.86)	0.00 (−2.46 to 0.15)	<0.001**
SVR	70	15.54 (6.93–33.96)	16.22 (7.26–33.96)	−0.31 (−16.90 to 8.85)	0.04*
PVR:SVR	70	0.42 (0.09–1.33)	0.31 (0.04–1.31)	0.12 (−0.12 to 0.60)	<0.001**
Echocardiographic parameters
TR Vmax (m/s)	34	3.32 (2.11–5.00)	3.04 (2.04–4.53)	0.15 (−0.28 to 1.22)	<0.001**
S/D ratio	53	1.38 (0.68–2.06)	1.09 (0.66–2.96)	0.13 (−1.05 to 1.17)	0.002*
EIs	62	1.29 (1.00–2.38)	1.17 (0.90–2.24)	0.07 (−0.16 to 0.59)	<0.001**
EIm	62	1.48 (1.06–3.18)	1.32 (0.94–2.90)	0.10 (−0.31 to −0.77)	<0.001**
TAPSE (cm)	54	1.70 (0.60–2.80)	1.90 (1.00–2.90)	−0.20 (−1.00 to 0.60)	<0.001**
RV FAC (%)	45	34.46 (9.00–55.00)	29.30 (7.04–54.12)	3.48 (−20.79 to 24.41)	0.026*
RVFW-LS (%)	45	−20.85 (−34.10 to −8.20)	−19.95 (−29.17 to −9.25)	−0.36 (−14.94–92.8)	0.24
RVFW-bLS (%)	45	−22.75 (−42.70 to −8.63)	−23.77 (−36.90 to −8.28)	−0.87 (−17.01 to 15.50)	0.31
RVFW-mLS (%)	45	−23.40 (−37.78 to −7.30)	−22.19 (−33.67 to −8.74)	0.44 (−15.97–11.34)	0.97
RVFW-aLS (%)	45	−17.63 (−36.30–1.24)	−15.91 (−31.62–2.75)	−0.15 (−17.30–15.83)	0.48
RVFWSR (/s)	39	−1.32 (−2.51 to −0.43)	−1.40 (−2.35 to −0.81)	0.13 (−0.66–1.04)	0.01*
LV EF (%)	39	55.00 (43.00–75.00)	53.00 (44.00–68.00)	2.00 (−18.00 to 21.00)	0.13
LV GLS (%)	50	−18.20 (−26.09 to −12.56)	−16.46 (−26.51 to −11.04)	−1.38 (−7.67–2.42)	<0.001**
LV GLSR (/s)	45	−1.02 (−1.83 to −0.69)	−0.95 (−85.00–0.77)	−0.11 (−1.46 to 84.09)	<0.001**
LV DLSR (/s)	50	2.04 (0.42–3.99)	0.99 (0.60–3.13)	0.81 (−0.42 to 3.04)	<0.001**

Note: All values are presented as median (range).

Abbreviations: EIm, maximal eccentricity index; EIs, systolic eccentricity index; LV DLSR, left ventricular global longitudinal diastolic strain rate; LV EF, left ventricular ejection fraction; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; MAP, mean systemic arterial pressure; mPAP, mean pulmonary artery pressure; PCWP, pulmonary capillary wedge pressure; PVRi, indexed pulmonary vascular resistance; QP, pulmonary blood flow; QS, systemic blood flow; RAP, right atrial pressure; RVEDP, right ventricular end diastolic pressure; RV FAC, right ventricular fractional area change; RVFW-aLS, right ventricular free wall longitudinal strain apical segment; RVFW-bLS, right ventricular free wall longitudinal strain basilar segment; RVFW-LS, right ventricular free wall longitudinal strain; RVFW-mLS, right ventricular free wall longitudinal strain middle segment; RVFWSR, right ventricular free wall strain rate; RVSP, right ventricular systolic pressure; S/D ratio, right ventricular systolic-to-diastolic ratio; sPAP, systolic pulmonary artery pressure; SVR, systemic vascular resistance; TAPSE, tricuspid annular plane systolic excursion; TPG, transpulmonary gradient; TR Vmax, maximal velocity of the tricuspid valve regurgitation jet.

la

p ≤ 0.05.

^**p ≤ 0.001

3.2.2 |. Echocardiographic Parameters

Echocardiographic measurements at baseline and maximal AVT conditions are presented in Table 2. The following variables all declined between conditions; TR Vmax (3.32 m/s [2.11–5.00] to 3.04 m/s [2.04–4.53], p < 0.001), S/D ratio (1.38 [0.68–2.06] to 1.09 [0.66–2.96], p = 0.002), EIs (1.29 [1.00–2.38] to 1.17 [0.90–2.24], p < 0.001) and EIm (1.48 [1.06–3.18] to 1.32 [0.94–2.90], p < 0.001). TAPSE increased (1.70 cm [0.60–2.80] to 1.90 cm [1.00–2.90], p < 0.001) significantly between conditions, whereas RV FAC (right ventricular fractional area change) decreased (34.46% [9.00–55.00] to 29.30% [7.04–54.12], p = 0.03). There was a significant increase in the RV free-wall strain rate (RVFWSR) with AVT (−1.32/s [–2.51 to 0.43] to −1.40/s [−2.35 to −0.81], p < 0.01), but there were no other changes noted in the remaining RV strain parameters.

The left ventricular global longitudinal systolic strain (LV GLS; −18.2% [−26.09 to −12.56] to −16.4% [−26.51 to −11.04], p < 0.001), left ventricular global longitudinal strain rate (LV GLSR; −1.02 /s [−1.83 to −0.69] to −0.95 /s [−85.00 to 0.77], p < 0.001), and diastolic strain rate (LV DLSR; 2.04 /s [0.42–3.99] to 0.99 /s [0.60–3.13], p < 0.001) all declined significantly between conditions. There was no significant change in left ventricular ejection fraction (LV EF).

3.3 |. AVT Responder Status

Based on invasive hemodynamics measured during RHC, 26 (36.6%) patientswere found to beAVT responders by Barst criteria and 13 (18.3%) by Sitbon Criteria. There was moderate agreement between the two criteria (k = 0.49, 95% CI 0.29–.69), with all but one patient who met Sitbon criteria also meeting Barst criteria.

3.4 |. Demographic and Baseline Hemodynamics

There were relatively few significant differences in the study cohort by responder status (Table S1). Sitbon responders tended to be younger compared to non-responders (6.0 yrs [1.0–17.0] vs. 11.0 yrs [1.0–23.0], p = 0.05). 6-minute walk distance (6MWD) was significantly higher in Barst responders (570.0 m [330.0–690.0] vs. 483.0 m [174.0–693.0], p = 0.03). Notably, none of the responders by either criterion died, required transplant, or needed ECMO during the study period. Baseline hemodynamics were not significantly different between Barst responders and non-responders.

3.5 |. Echocardiographic Parameters

Echocardiographic parameters were compared between responders and non-responders in both criteria at baseline, the AVT condition, and the percent change between the two conditions (Tables 3 and S2). In Barst responders, EIs and EIm were both lower at baseline and AVT conditions compared to non-responders. The percent changes in S/D ratio (17.5 [−36.8 to 58.7] vs. 3.3 [−6.2 to 45.1], p = 0.04) and EIm (12.4 [−14.0 to 36.5]) vs. 3.6 [−22.8 to 32.2], p = 0.04) were significantly greater in Barst responders compared to non-responders.

TABLE 3 |.

Differences in the percent changes of echo parameters between responders and non-responders as defined by either Barst or Sitbon criteria.

Echo parameters (% change)
	Barst criteria			Sitbon criteria
Variable	Non-responders (n = 45) median (range or %)	Responders (n = 26) median (range or %)	p value	Non-responders (n = 58) median (range or %)	Responders (n = 13) median (range or %)	p value
TR Vmax (%change)	3.3 (−10.0 to 24.3)	9.9 (0.6–37.4)	0.06	4.3 (−10.0 to 24.3)	14.4 (0.6–37.4)	0.25
S/D ratio (% change)	3.3 (−6.2 to 45.1)	17.5 (−36.8 to 58.7)	0.04 *	8.6 (−65.2 to 59.7)	12.1 (−36.8 to 48.4)	0.42
EIs (% change)	5.2 (−15.1 to 38.7)	11.0 (−15.1 to 34.8)	0.21	4.9 (−15.1 to 38.7)	11.1 (−1.9 to 34.8)	0.30
EIm (% change)	3.6 (−22.8 to 32.2)	12.4 (−14.0 to 36.5)	0.04 *	5.6 (−22.8 to 32.2)	12.9 (−4.7 to 36.5)	0.13
TAPSE (% change)	−9.2 (−142.9 to 21.4)	−7.8 (−54.5 to 16)	0.66	−7.8 (−142.9 to 21.4)	−23.3 (−50.0 to 16.0)	0.31
RV FAC (% change)	10.8 (−80.1 to 49.8)	12.8 (−62.4 to 45.5)	0.95	11.4 (−80.1 to 49.8)	8.1 (−60.0 to 27.8)	0.58
RVFW-LS (% change)	4.3 (−50.3 to 43.8)	−1.4 (−24.1 to 29.4)	0.37	3.0 (−50.3 to 43.8)	1.1 (−23.1 to 29.4)	0.91
RVFWSR (% change)	−9.6 (−111.6 to 31.0)	−7.4 (−79.4 to 41.8)	0.50	−9.6 (−111.6 to 31.0)	−7.4 (−79.4 to 41.8)	0.45
LV EF (% change)	3.9 (−36.0 to 31.3)	0.0 (−26.7 to 23.8)	0.97	5.0 (−36.0 to 31.3)	0.0 (−26.7 to 16.4)	0.53
LV GLS (% change)	8.8 (−14.5 to 36.2)	6.2 (−12.1 to 27.2)	0.96	8.0 (−14.5 to 36.2)	4.9 (−12.1 to 27.2)	0.60
LV GLSR (% change)	10.0 (−9240.7 to 211.6)	12.5 (−39.2 to 32.5)	0.37	10.5 (−9240.7 to 211.6)	7.2 (−16.2 to 28.7)	0.38

Abbreviations: EIm, maximal eccentricity index; EIs, systolic eccentricity index; LV DLSR, left ventricular global longitudinal diastolic strain rate; LV EF, left ventricular ejection fraction; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; RV FAC, right ventricular fractional area change; RVFW-LS, right ventricular free wall longitudinal strain; RVFWSR, right ventricular free wall strain rate; S/D ratio, right ventricular systolic-to-diastolic ratio; TAPSE, tricuspid annular plane systolic excursion; TR Vmax, maximal velocity of the tricuspid valve regurgitation jet.

^*p ≤ 0.05.

3.6 |. Relationships Between Echo and Invasive Hemodynamics

Associations between echocardiographic and invasive hemodynamic parameters are presented in Table S3. In both the baseline and AVT conditions, moderate positive correlations were noted between TR Vmax, EIs, and EIm with the hemodynamic measures. S/D ratio was not significantly correlated with any of the measured hemodynamics in either condition. RV FAC demonstrated relatively weak negative correlations with hemodynamics in both conditions. RVFW-LS (right ventricular free wall longitudinal strain) and its segments were more consistently positively correlated with hemodynamics in the AVT condition. The left ventricular function indices did not demonstrate consistent associations with hemodynamics in either condition.

To assess if the percent change in echocardiographic parameters was similar to that seen in hemodynamic parameters, the percent change in hemodynamics and echocardiographic variables was then compared, and the results are in Table S4. The percent change in TR Vmax, S/D ratio, and EIm demonstrated moderate positive correlations with the percent change in sPAP, Mpap, and TPG. The free-wall strain of the apical segment of the RV (RVFW-aLS) demonstrated a weak negative correlation with the percent change in PVRi.

3.7 |. Univariate Prediction Model: Responder Status

A univariate logistic regression analysis was then performed on the ability of the percent change in echocardiographic variables between conditions to predict either Barst or Sitbon responder status (Table 4). The percent change in TR Vmax (AUC 0.713, p = 0.04), S/D Ratio (AUC 0.684, p = 0.04), and EIm (0.655, p = 0.04) all predicted Barst responder status. Whereas only the change in TR Vmax (AUC 0.669, p = 0.04) and the LV diastolic strain rate (AUC 0.66, p = 0.03) predicted Sitbon responder status.

TABLE 4 |.

Univariate analysis of the ability of the percent change in echo variables to predict either Barst or Sitbon responder status.

% Change in Echo variable	Barst criteria			Sitbon criteria
	AUC	OR (CI)	p value	AUC	OR (CI)	p value
TR Vmax (m/s)	0.713	1.09 (1.00–1.18)	0.04 *	0.669	1.10 (1.00–1.21)	0.04 *
S/D ratio	0.684	1.03 (1.00–1.06)	0.04 *	0.580	1.01 (0.98–1.04)	0.46
EIs	0.596	1.03 (0.985–1.08)	0.19	0.595	1.03 (0.977–1.08)	0.26
EIm	0.655	1.05 (1.00–1.10)	0.04 *	0.639	1.05 (0.99–1.10)	0.08
TAPSE	0.539	1.00 (0.97–1.02)	0.98	0.601	0.99 (0.97–1.02)	0.54
RV FAC	0.494	0.99 (0.97–1.01)	0.65	0.564	0.98 (0.96–1.01)	0.33
RVFW-LS	0.585	0.98 (0.95–1.02)	0.39	0.52	1.00 (0.96–1.05)	0.85
RVFW-bLS	0.581	0.98 (0.97–1.00)	0.58	0.579	1.00 (0.98–1.03)	0.44
RVFW-mLS	0.576	1.01 (0.98–1.03)	0.57	0.490	0.99 (0.97–1.03)	0.94
RVFW-aLS	0.678	0.98 (0.97–1.00)	0.67	0.653	0.99 (0.99–1.00)	0.61
RVFWSR	0.568	1.00 (0.98–1.03)	0.56	0.594	1.01 (0.95–1.03)	0.59
LV EF	0.504	1.01 (0.96–1.05)	0.70	0.572	0.99 (0.94–1.04)	0.70
LV GLS	0.505	0.99 (0.95–1.04)	0.89	0.555	0.98 (0.93–1.04)	0.59
LV GLSR	0.485	1.00 (0.99–1.00)	0.67	0.399	1.00 (0.99–1.00)	0.76
LV DLSR	0.578	1.00 (0.98–1.02)	0.72	0.660	0.97 (0.95–0.99)	0.03 *

Abbreviations: AUC, area under the curve; EIm, maximal eccentricity index; EIs, systolic eccentricity index; LV EF, left ventricular ejection fraction; LV DLSR, left ventricular global longitudinal diastolic strain rate; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; RV FAC, right ventricular fractional area change; RVFW-LS, right ventricular free wall longitudinal strain; RVFWSR, right ventricular free wall strain rate; S/D ratio, right ventricular systolic-to-diastolic ratio; TAPSE, tricuspid annular plane systolic excursion; TR Vmax, maximal velocity of the tricuspid valve regurgitation jet.

^*p < 0.05.

^**p < 0.001.

Utilizing receiver operating characteristic (ROC) curves and Youden’s index, cut-points were established for the three variables noted to predict Barst responder status. For TR Vmax, ≥22.5% change between baseline and maximal vasodilation was found to optimize sensitivity and specificity (positive predictive value [PPV] = 0.47, negative predictive value [NPV] = 0.88). The optimal cut-point for the change in S/D ratio was ≥48.2% (PPV = 0.47, NPV = 0.89) and ≥29.0% for EIm (PPV = 0.51, NPV = 0.78).

3.8 |. Outcome Prediction: Death/Transplant/ECMO

A univariate logistic regression analysis was performed on the ability of the percent change in echocardiographic variables between conditions to predict the combined outcome of Death, Transplant, or need for ECMO (Table S5). There were no significant predictors from any of the echo variables in predicting the adverse outcome.

4 |. Discussion

Here, for the first time, we demonstrate evidence of significant changes in echocardiographic markers of PAH during acute vasodilatory testing in pediatric patients. The changes seen in these echocardiographic markers directly correlated with changes seen in simultaneously obtained invasive hemodynamics. Furthermore, using well-established hemodynamic benchmarks for what would be considered a significant response to vasodilatory testing, we identified several commonly used echocardiographic parameters that could be utilized to predict this response.

4.1 |. Changes in Echocardiographic Parameters and Hemodynamics Across Conditions

In this population of pediatric patients with PAH due to a variety of etiologies, we found that several commonly used estimates of right-sided pressure and function (TR Vmax, S/D Ratio, EIs, and EIm) decreased significantly with the addition of pulmonary vasodilators. These changes directly mirrored the changes observed in the invasively measured hemodynamics, particularly when using TR Vmax, S/D Ratio, and EIm. To our knowledge, these are the first data to confirm that these commonly used echocardiographic parameters are truly sensitive enough to change in the acute setting. These results have immediate clinical relevance as there are many clinical situations wherein pediatric patients are unable to go for invasive hemodynamic monitoring. This leaves clinicians to try and compare echocardiographic parameters at the bedside before and after the addition of pulmonary vasodilators without clear evidence that the observed changes are truly reflective of changes in pulmonary hemodynamics.

Changes in markers of RV function were less consistent after AVT. TAPSE was noted to increase, whereas RV FAC decreased slightly. There were no significant changes to RVFW-LS (or in any of the individual segments), but the RVFW strain rate did improve somewhat. The reason for these somewhat discrepant findings is unclear. Right ventricular systolic function is notoriously challenging to measure given its complex shape–and TAPSE, particularly in children, may reflect more of the basilar rocking motion of the heart as opposed to actual intracavitary change [24]. What is most notable is that the absolute degree of change in all variables is relatively small and likely would not be clinically significant. Ultimately, right ventricular function is likely not significantly altered in this acute setting, and the subtle changes noted on echocardiography may be impacted by measurement error or more subtle changes to other factors that impact cardiac output, such as heart rate and preload that were not captured by our hemodynamic testing. Supporting this, the invasively estimated cardiac output from the RV (Qp) did not change significantly after AVT. Responder status also did not seem to impact the changes in RV function. This differs somewhat from a small study in adult patients with connective tissue disease, where RV function was noted to be lower in non-responders. This study also found that RV S’ (a tissue-Doppler derived marker of RV function) only increased significantly in non-responders [25]. In this study, TAPSE and RV FAC did not change significantly in any of the patients. Future work focused on the RV myocardial response to acute changes in loading conditions will be important to help clarify these findings.

Interestingly, there was a more consistent signal toward worsening left ventricular function after AVT. There was a non-significant trend towards worsening of the LV EF, and significant worsening of the LV global longitudinal strain, strain rate, and diastolic strain rate. This is in the context of no significant changes in the invasively measured cardiac index and was not modified by responder status. Our group has previously identified patients with PAH to have impaired left ventricular systolic and diastolic function compared to healthy controls [26, 27]. It would be reasonable to think that reducing the pulmonary vascular resistance, increasing preload, and increasing systemic afterload to a stiff left ventricle could, at least temporarily, worsen echocardiographic estimates of systolic function. Again, these were relatively small absolute changes between variables that are likely of minimal clinical significance, and it remains to be seen if these changes are sustained or just present in the acute period.

4.2 |. Ability to Predict Responder Status

For the purposes of this study, we elected to evaluate the degree of change in echocardiographic estimates of right-sided pressures against well-established hemodynamic criteria to determine if these non-invasive metrics could be used to acutely identify patients who more robustly respond to additional pulmonary vasodilatory treatment. Providers caring for pediatric patients with PAH will no doubt recognize numerous circumstances when invasive hemodynamic testing is either deferred or performed under non-traditional conditions. These include parental refusal for invasive testing when presented with the risks of anesthesia, children who no longer have reliable vascular access after repeat instrumentation, patients who are deemed too ill to undergo anesthesia and have already been started on medical therapy, and neonates with multifactorial disease who have fluctuating needs and response to pulmonary vasodilatory therapy. Although echocardiography is unlikely to replace hemodynamic testing and AVT to determine who is eligible for CCB therapy, it does at least appear to be sensitive enough to identify a clinically significant response to acute pulmonary vasodilation.

In this cohort, we noted a relatively modest agreement between Sitbon and Barst criteria as defined by cardiac catheterization (k = 0.49), which is slightly lower than previously published works in the pediatric population (k value of 0.67) [28]. Controversy remains in the literature as to the validity of Sitbon criteria in patients with baseline mPAP <40 mmHg, which is often the case in the pediatric population. Prior pediatric cohort studies have nonetheless found it useful even in this setting, so we elected to include it in this analysis despite this limitation [29]. As a whole, we had relatively few Sitbon responders, and postulate that patients with a baseline mPAP <40 mmHg may be more likely to meet Barst criteria, which is based on a relative change in hemodynamics as opposed to an absolute drop of 10 mmHg. Similarly, we noted Barst criteria responders had significantly more pronounced changes in S/D Ratio and EIm compared to non-responders, which did not rise to statistical significance in Sitbon responders. This may suggest that monitoring for relative changes in echocardiographic parameters may prove to be a more nimble way to assess for significant alterations in hemodynamics.

We therefore, performed univariate cox-proportional-hazards regression models which identified the percentage change in TR Vmax, S/D Ratio, and EIm as significant predictors for Barst criteria responders. These results conceptually make sense seeing as the percent change in TR Vmax, S/D Ratio, and EIm all correlated with the percent change in the invasively measured mPAP which is primarily what the Barst and Sitbon criteria are based on. Again, likely related to a limited number of Sitbon responders, only TR Vmax and the LV DLSR were significant predictors for this group. The role of the LV DLSR in predicting response to vasoreactivity testing remains unclear and requires further investigation.

Our goal with these results is to provide a framework for clinicians to interpret the results of vasodilatory challenges with the aid of echocardiography under real-world conditions. The advantage of TR-Vmax, EIm, and S/D Ratio is that they are easily obtainable in the majority of echo labs and do not require additional post-processing software to measure. Future studies with validation cohorts could consider creating a multiparametric score using these four variables to strengthen their predictive value.

4.3 |. Associations With Clinical Outcomes

Univariate modeling was also performed to assess the degree of change in each echo parameter to predict a combined outcome of death, transplant, or ECMO. None of the studied variables were found to be significant predictors of this endpoint. This is likely attributable to the relatively few patients who died (n = 6), underwent transplant (n = 2), or needed ECMO (n = 2) during the study period. The combined outcome had just 8 patients, as both patients who underwent transplant later died. Further studies in larger, longitudinal cohorts with additional “softer” end-points would likely be needed to better understand the prognostic capability of the change in echo variables during this type of testing.

4.4 |. Limitations

This study is limited by its relatively small sample size, which impacts the ability to predict relatively rare outcome events. Furthermore, all echocardiograms were performed under general anesthesia in this study. Although this strengthens our associations with simultaneously obtained invasive hemodynamics, it could potentially limit generalizability to an awake patient. Variations in the exact anesthetic protocol that were used could also have impacted the results, though dictating the anesthetic regimen during clinical catheterization was beyond the scope of this study.

The patient selection for this study may also limit its generalizability, but it also more accurately reflects the heterogeneity of the pediatric PAH population. For instance, the validity of EI to predict hemodynamics in patients with open intracardiac shunts has not yet been established, but remains an active area of research. Furthermore, this was not the first hemodynamic study for the majority of patients, and many were already being treated for PAH. As a result, there were seven patients with normal mean pulmonary artery pressures at the time of cardiac catheterization; however, all these patients had previously met diagnostic criteria and had been started on treatment. Only two Barst responders and none of the Sitbon responders fell into this category, so it likely had little impact on the analysis. These patients were still included as normalization of hemodynamics with treatment does not mean their PAH has been cured. Furthermore, this better reflects the growing literature that exists in the pediatric population, highlighting both the fluidity and prognostic importance of responder status across serial assessments [14].

5 |. Conclusions

Our results verify that there are appreciable changes in commonly used echocardiographic measures of pulmonary hypertension that directly parallel simultaneously measured invasive hemodynamics during AVT. Furthermore, here we have identified a set of commonly used and easily reproducible echocardiographic metrics with adequate sensitivity to identify patients who have a more robust response to AVT. These results would need to be confirmed with a validation cohort and tested prospectively with a longitudinal study to establish long-term prognostic value.

Supplementary Material

Supplemental tables

Supplemental Table 1: Differences in demographic and baseline invasive hemodynamics between responders and non-responders as defined by either Barst or Sitbon criteria. Values are presented as Median (range) or n(%). 6MWD indicates six-minute walk distance; ECMO, extracorporeal membrane oxygenation; RVSP, right ventricular systolic pressure; sPAP, systolic pulmonary artery pressure; mPAP, mean pulmonary artery pressure; PVRi, indexed pulmonary vascular resistance; TPG, transpulmonary gradient. *= p<0.05.

Supplemental Table 2: Differences in Echo parameters at baseline, AVT and the percent changes between responders and non-responders as defined by either Barst or Sitbon criteria.; TR Vmax, maximal velocity of the tricuspid valve regurgitation jet; S/D ratio,), right ventricular systolic-to-diastolic ratio; EIs, systolic eccentricity index; EIm, maximal eccentricity index; TAPSE, tricuspid annular plane systolic excursion; RV FAC, right ventricular fractional area change; RVFW-LS, right ventricular free wall longitudinal strain; RVFWSR, right ventricular free wall strain rate; LV EF, left ventricular ejection fraction; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; LV DLSR, left ventricular global longitudinal diastolic strain rate **= p<0.001; *= p<0.05.

Supplemental Table 3: Correlation coefficients between echocardiographic parameters and selected invasive hemodynamics at baseline and maximal acute pulmonary vasodilation (AVT). r indicates Spearman’s correlation coefficient. TR Vmax, maximal velocity of the tricuspid valve regurgitation jet; S/D ratio,), right ventricular systolic-to-diastolic ratio; EIs, systolic eccentricity index; EIm, maximal eccentricity index; TAPSE, tricuspid annular plane systolic excursion; RV FAC, right ventricular fractional area change; RVFW-LS, right ventricular free wall longitudinal strain; RVFWSR, right ventricular free wall strain rate; LV EF, left ventricular ejection fraction; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; LV DLSR, left ventricular global longitudinal diastolic strain rate; RVSP, right ventricular systolic pressure; sPAP, systolic pulmonary artery pressure; mPAP, mean pulmonary artery pressure; PVRi, indexed pulmonary vascular resistance; TPG, transpulmonary gradient. **= p<0.001; *= p<0.05.

Supplemental Table 4: Correlation coefficients between the percent change in echocardiographic parameters and selected invasive hemodynamics. r indicates Spearman’s correlation coefficient. TR Vmax, maximal velocity of the tricuspid valve regurgitation jet; S/D ratio,), right ventricular systolic-to-diastolic ratio; EIs, systolic eccentricity index; EIm, maximal eccentricity index; TAPSE, tricuspid annular plane systolic excursion; RV FAC, right ventricular fractional area change; RVFW-LS, right ventricular free wall longitudinal strain; RVFWSR, right ventricular free wall strain rate; LV EF, left ventricular ejection fraction; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; LV DLSR, left ventricular global longitudinal diastolic strain rate **= p<0.001; *= p<0.05.

Supplemental Table 5: Univariate logistic regression analysis of the percent change in echocardiographic parameters to predict a combined outcome of death, transplant or need for ECMO within the study period. TR Vmax, maximal velocity of the tricuspid valve regurgitation jet; S/D ratio,), right ventricular systolic-to-diastolic ratio; EIs, systolic eccentricity index; EIm, maximal eccentricity index; TAPSE, tricuspid annular plane systolic excursion; RV FAC, right ventricular fractional area change; RVFW-LS, right ventricular free wall longitudinal strain; RVFWSR, right ventricular free wall strain rate; LV EF, left ventricular ejection fraction; LV GLS, left ventricular global longitudinal systolic strain; LV GLSR, left ventricular global longitudinal systolic strain rate; LV DLSR, left ventricular global longitudinal diastolic strain rate.

Supporting Information

Additional supporting information can be found online in the Supporting Information section.

Abbreviations:

AVT

acute vasoreactivity testing

ECMO

extra corporeal membrane oxygenation

EIm

maximal eccentricity index

EIs

systolic eccentricity index

LV EF

left ventricular ejection fraction

LV GLS

left ventricular global longitudinal systolic strain

LV GLSR

left ventricular global longitudinal systolic strain rate

PAH

pulmonary arterial hypertension

RV FAC

right ventricular fractional area change

RVFW-LS

right ventricular free wall longitudinal strain

RVFWSR

right ventricular free wall strain rate

S/D ratio

right ventricular systolic-to-diastolic ratio

TAPSE

tricuspid annular plane systolic excursion

TR Vmax

maximal velocity of the tricuspid valve regurgitation jet

WHO

World Health Organization