Right ventricular (RV) free wall strain (RVFWS) is a feasible method for quantitation and follow-up of RV function and may have benefits over traditional markers such as fractional area change. However, like all ejection phase parameters, RVFWS is difficult to assess in the presence of changing afterload. The aim of this study was to compare RVFWS and traditional RV function parameters for tracking progress of RV function in patients with pulmonary arterial hypertension (PAH) over a range of pulmonary artery systolic pressure (PASPs).
Sequential echocardiograms were collected retrospectively at two time points between 2005 and 2015 in 187 patients (71% women; mean age, 63 ± 14 years) undergoing pulmonary vasodilator therapy for group 1 PAH. Patients were either studied during PAH therapy ( n = 111) or before and after treatment initiation ( n = 76). Standard measurements of RV and left ventricular function and PASP were performed, and speckle-tracking strain was used to calculate RVFWS. The linear response of RVFWS to afterload (PASP) was assessed using a standard regression equation. Because it is unclear if the response might be nonlinear, a quadratic association (PASP squared) was also used in the regression model.
At visit 1, patients with PAH showed impaired functional capacity (mean 6-min walk distance, 371 ± 131 m), increased PASP (mean, 54 ± 26 mm Hg), and borderline RVFWS (mean, 18 ± 6%). Patients before PAH therapy showed more pronounced reduction in 6-min walk distance (mean, 302 ± 136 m) and RVFWS (mean, 16 ± 5%). RVFWS at baseline was associated with PASP ( R 2 = 0.25, P = .001), RV end-diastolic area ( R 2 = 0.36, P < .001), and fractional area change ( R 2 = 0.21, P < .001). Change in RVFWS was more strongly associated with ΔPASP (std β = −0.20, P = .02) than ΔPASP squared (std β = 0.11, P = .20). RVFWS showed strength over fractional area change for sequential RV assessment over a range of PASP changes.
Afterload changes should be taken into account in the evaluation of RVFWS during PAH follow-up, with the relationship to PASP likely to be linear.
There is a significant linear correlation between changes in RVFWS and changes in PASP during follow-up of PAH.
RVFWS was significantly better than FAC at tracking changes in function over varying ranges of ΔPASP.
RVFWS should be a used to track RV function in patients receiving PAH therapy.
Following the invasive diagnosis of pulmonary arterial hypertension (PAH) (mean pulmonary arterial pressure > 25 mm Hg, with a pulmonary artery wedge pressure < 15 mm Hg), efficacy of pulmonary vasodilator treatment is assessed by echocardiographic estimation of pulmonary artery systolic pressure (PASP). Echocardiographically derived PASP has shown adequate correlation with right-heart catheterization, although some have questioned its reliability. Echocardiography continues to be recommended to assess treatment efficacy at 6- to 12-month intervals over long-term follow-up of treated patients. However, although much attention has been paid to following PASP, the strongest echocardiographic markers of outcome in patients with PAH are right atrial size and pericardial effusion, both of which are related to late-stage disease.
There is increasing evidence of the importance of right ventricular (RV) systolic markers in PAH. The measurement of RV strain using speckle-tracking echocardiography is feasible for the quantification of RV function in PAH, with RV free wall strain (RVFWS) shown to predict clinical deterioration and mortality. However, despite the long-standing appreciation of the role of afterload in the assessment of RV function and the load dependence of traditional RV functional indices in PAH, little attention has been paid to the importance of afterload on RV strain. The interpretation of RV strain changes is therefore difficult to interpret in the follow-up of treated patients with PAH because of fluctuations in pulmonary vascular resistance. In animal models, PAH has been artificially created by banding of the pulmonary artery. In a clinical cohort, follow-up of patients receiving pulmonary vasodilator therapy is a proposed method to track afterload changes over time. It is assumed that RV afterload to RVFWS response is linear. Invasive animal studies have found mixed results, with a linear association found but only when the “severe” category has been removed. This may indicate a “quadratic” relationship, with PASP having differing effects on RV strain at increasing afterload levels. This has implications about how we adjust for afterload. Statistically, this might be best represented by a quadratic equation, which incorporates the dependent variable (PASP) squared. Our hypothesis was that RVFWS would be a strong echocardiographic marker to track progress of RV function in PAH over a range of ΔPASP, with changes in RVFWS showing a linear relationship to ΔPASP.
Patients with idiopathic PAH and that due to connective tissue disease were recruited to this study from the Tasmanian Pulmonary Hypertension Registry (Hobart, Australia) and the Princess Alexandra Hospital (Brisbane, Australia). After the initial search, 206 patients were identified, but three patients did not have adequate tricuspid regurgitation traces at either baseline or follow-up. A further 14 patients had inadequate RV strain images at baseline or follow-up, precluding the calculation of changes in the relevant parameters. Two patients had neither RV strain nor tricuspid regurgitation jets. Data were retrospectively collected between 2005 and 2015 in 187 patients (71% women; mean age, 63 ± 14 years) at two time points (median time between scans, 6 months) during treatment with endothelin receptor antagonists, phosphodiesterase-5 inhibitors, intravenous or inhaled prostacyclin analogs, or their combination, according to current guidelines.
There were two groups of patients, the first being treatment naive (in whom the baseline echocardiogram was available before treatment and repeated on therapy; n = 76) and the second on therapy, who underwent echocardiography to assess treatment response ( n = 111). This design was selected to provide a spectrum of change in PASP: our interest was in the relationship of RV function to pulmonary artery pressure rather than the response to particular treatments. Ethics approval was obtained from the Human Research Ethics Committee (Tasmania) Network (approval number H0013333) and from the Metro South Human Research Ethics Committee (HREC 16/QPAH/008).
Echocardiography was performed using standard commercial equipment (Vivid 7, Vivid i, and Vivid e9 [GE Vingmed Ultrasound AS, Horten, Norway]; ie33 [Philips, Bothell, WA]). RV-focused views were obtained in addition to a standard imaging protocol. All measurements were performed by a single reader, using standard views.
PASP was measured from the peak tricuspid regurgitation velocity using the modified Bernoulli equation. Right atrial pressure was derived from the inferior vena cava dimension and distensibility from the subcostal view.
Left ventricular (LV) and RV measurements were performed according to American Society of Echocardiography guidelines by a single reader for both studies. LV ejection fraction was measured using the Simpson biplane method from the apical four- and two-chamber views. RV end-diastolic area and RV end-systolic area were calculated from an apical RV-focused view. Fractional area change (FAC) was calculated as the percentage change between the RV end-diastolic area and RVESA. Peak velocities of the early (E) and late (A) diastolic filling were derived from the transmitral inflow pattern. Doppler tissue imaging using the color Doppler method was used to determine the peak diastolic early velocity (e′) of the lateral and septal mitral annulus from the apical four-chamber view.
Tricuspid annular plane systolic excursion (TAPSE) was measured as the displacement (centimeters) of the tricuspid lateral annulus toward the RV apex in the RV four-chamber view. In a smaller group of patients, RV tricuspid valve systolic motion was used with Doppler tissue imaging, with the pulse-wave sample volume placed at the tricuspid valve annulus.
Speckle-Tracking Strain Analysis
Wall motion tracking software (Image Arena; TomTec, Unterschleissheim, Germany) was used to analyze two-dimensional echocardiographic images. The optimal image in which the whole RV cavity was visualized was selected from each study. The RV end-diastolic endocardial border was manually traced along the RV septal and RV free wall. These borders were then tracked automatically frame by frame throughout the cardiac cycle. Manual adjustments were performed when necessary and regions excluded if excessive noise or inadequate tracking were present. RVFWS was calculated from an average of three segmental RVFWS traces.
Right-heart catheterization studies were undertaken after premedication and local anesthesia. A 4-lumen, 110-cm, 7-Fr Swan-Ganz catheter (Edwards Lifesciences, Irvine, CA) was floated to the right heart, and resting measurements of right atrial, RV, pulmonary arterial, and pulmonary capillary wedge pressures were made at end-expiration using a pressure transducer (21BB; ITL Healthcare, Chelsea Heights, Australia). This was calibrated to atmospheric pressure at the level of the right atrium and rechecked at intervals to avoid zero drift. Cardiac output was determined by thermodilution, using an average of four consecutive values that varied <10%. Electrocardiographic leads were connected to both arms and the left leg, allowing three electrocardiographic channels for timing of signals. All hemodynamic monitoring was recorded using a Horizon SE Hemodynamic System (Mennen Medical, Yavne, Israel) and subsequently analyzed offline.
A 6-min walk test was performed at the time of the echocardiographic examination by a research nurse. This was done in a quiet hospital corridor and followed a standard protocol.
Statistical analysis was performed using standard software (SPSS version 20.0; IBM, Chicago, IL) with statistical significance set to P < .05. To calculate the delta values, final-visit values were subtracted from visit 1. The change between baseline and follow-up was than plotted using linear and quadratic regression models. To understand which model showed the strongest relationship between variables, ΔPASP squared (for the quadratic model) was added using a sequential linear regression model to assess whether there was a significant reduction in the mean of the squared errors. A relative change in LV free wall strain of 10% has been defined as being clinically relevant on the basis of the reproducibility of strain. We therefore used cutoffs of relative change of 10% for FAC and RVFWS to evaluate the relationship between ΔPASP and ΔRVFWS. One-way analysis of variance with the Tukey honestly significantly different test was used to compare groups. Actual ΔRVFWS was compared with predicted ΔRVFWS from linear and quadratic models. Twenty patients were randomly selected for inter- and intrareader variability analysis of RVFWS, 2 years after initial readings. Variability was assessed in preselected images, with measurement of single beats by readers blinded to original measurement results. Bland-Altman plots were used to assess the level of agreement between measures, with the intraclass correlation coefficient used to determine reliability of measures.
Baseline demographic details ( Table 1 ) were similar to other PAH cohorts (mean age, 63 ± 14 years; 134 women [72%]), with most patients having idiopathic PAH (52%), followed by connective tissue disease (36%). Patients were mostly treated with endothelin receptor antagonists (57%) or phosphodiesterase-5 inhibitors (28%), and some participants were receiving multiple therapies (14%).
|Age (y)||63 ± 14|
|HR (beats/min)||76 ± 14|
|BMI (kg/m 2 )||28.1 ± 7.0|
|Women (%)||134 (72%)|
|Connective tissue disease||66 (35%)|
|Idiopathic PAH||97 (52%)|
|Any ERA||139 (74%)|
|PDE-5 inhibitor||70 (37%)|
|Multiple therapies||26 (14%)|
Associations of Baseline Measurements
Baseline PASP was significantly related to 6-min walk distance (6MWD) ( r = −0.37, P < .001) and RV function. Six-minute walk distance had significant relationships with RV function and LV diastolic markers but not LV ejection fraction ( Table 2 ).
Baseline versus Follow-Up
At visit 1, 76 patients had echocardiograms available for analysis before initiation of medication, and another 111 had sequential echocardiograms obtained on PAH therapy. Between baseline and follow-up ( Table 3 ), there was a decrease in PASP and an improvement in RV strain, with no significant difference in 6MWD, LV systolic function, or estimated LV filling pressures.
|Variable||Visit 1||Visit 2||Δ||P|
|6MWD (m)||371 ± 130.8||380 ± 130.6||9.7 ± 71.2||.12|
|Heart rate (beats/min)||76 ± 13.8||75 ± 14.2||−0.7 ± 12.0||.42|
|Ejection fraction (%)||62 ± 9.1||63 ± 8.0||0.65 ± 9.6||.40|
|LVEDV (mL)||72.3 ± 24.8||73.9 ± 24.2||1.6 ± 17.1||.22|
|LVESV (mL)||27.9 ± 13.2||28.2 ± 12.3||0.35 ± 10.2||.64|
|LV diastolic dimension (cm 2 )||4.6 ± 0.70||4.6 ± 0.73||0.02 ± 0.70||.65|
|PW (cm)||0.96 ± 0.34||0.94 ± 0.16||−0.02 ± 0.32||.42|
|e′ septal (cm/sec)||6.2 ± 2.3||6.2 ± 2.2||−0.04 ± 1.8||.83|
|E/e′ average||10.8 ± 4.8||11.6 ± 6.2||0.8 ± 5.1||.07|
|E/e′ lateral||9.3 ± 4.5||9.0 ± 5.0||−0.22 ± 3.6||.50|
|E/e′ septal||12.4 ± 6.2||13.8 ± 7.0||0.04 ± 1.8||.01|
|Estimated PASP (RVSP + RAP, mm Hg)||53.8 ± 25.8||50.2 ± 24.7||−3.6 ± 15.8||.001|
|End-diastolic area (cm 2 )||21.4 ± 8.4||21.6 ± 8.3||−0.02 ± 0.32||.71|
|FAC (%)||33.3 ± 12.0||34.2 ± 11.2||0.92 ± 11.0||.24|
|TAPSE (cm)||1.9 ± 0.53||2.0 ± 0.55||0.10 ± 0.48||.006|
|RV S′ (cm) ( n = 75)||9.7 ± 2.9||9.5 ± 3.2||−0.16 ± 2.6||.59|
|RVFWS (%)||18.0 ± 5.7||18.9 ± 5.4||0.94 ± 3.9||.001|
|IVC (cm)||1.7 ± 0.81||1.5 ± 0.85||−0.2 ± 0.93||.002|
|RVFWS/PASP||0.44 ± 0.30||0.49 ± 0.31||−0.05 ± 0.19||<.001|
|PASP ( n = 58)||79.9 ± 20.5||66.2 ± 19.9||−3.7 ± 14.7||.06|
|mPAP ( n = 64)||43.8 ± 13.1||40.2 ± 13.2||−3.5 ± 10.3||.008|
|PCWP ( n = 58)||12.3 ± 6.0||11.8 ± 4.6||−0.57 ± 6.1||.48|
In the subgroup studied before and after initiation of pulmonary vasodilators ( Supplemental Table 1 , available at www.onlinejase.com ), there were no significant improvements in 6MWD, but the intervention group displayed borderline improvements in RVFWS ( P = .06) and PASP ( P = .02).
A subset of patients ( n = 64) had invasive measurements performed at two time points ( Table 4 ). Pulmonary artery pressure showed a significant reduction with therapy, but there was no significant change in pulmonary capillary wedge pressure. At visit 1, the time difference between right-heart catheterization and echocardiography was at a median interval of 0 days (interquartile range, 167 days), with invasive PASP and Doppler-derived PASP showing a moderate significant correlation ( r = 0.7171, P < .001). The interclass correlation coefficient at visit 1 was 0.71 (95% confidence interval [CI], 0.59–0.79; P < .001). At visit 2 (median interval, 0 days), t (mean, 11 ± 101.2), the interclass correlation coefficient was 0.76 (95% CI, 0.50–0.78; P < .001), with a moderate correlation ( r = 0.76, P < .001). Similar to echocardiographic measures, there were modest significant correlations with invasive markers and echocardiographic variables.
|Invasive PASP baseline visit 1||Invasive PASP visit 2|
|PASP (mm Hg)||0.71||<.001||0.67||<.001|
Afterload Dependence and RV Function Markers
The decrease in PASP over follow-up in the group as a whole was associated with changes in 6MWD ( r = −0.26, P = .003), RV function with RVFWS ( r = −0.25, P = .001), but not FAC ( r = −0.12, P = .10), and heart rate ( r = 0.16, P = .03). A regression equation was then developed to predicted ΔRVFWS from the linear regression model (ΔRVFWS = 0.75 + −0.06 × ΔPASP). The quadratic regression was ΔRVFWS = 0.60 + (−0.047 × ΔPASP) + (0.0007 × ΔPASP ×ΔPASP). Both of these regression lines are shown in Figure 1 . Figure 2 A shows the actual versus predicted ΔRVFWS (derived from the regression equation; ΔRVFWS = 0.99 + [−0.11 × ΔPASP] + [0.0006 × ΔPASP × ΔPASP]) for the entire group ( r = 0.26, P = .001) and for those restricted to the treatment intervention group ( r = 0.44, P < .001). Figure 2 C shows the linear and actual plotted against each other for the whole group ( r = 0.25, P = .001) and for the treatment intervention group (ΔRVFWS = 1.12 + [−0.12 × ΔPASP]) ( r = 0.44, P < .001). When models based on ΔPASP and ΔPASP squared were compared, this showed the linear model to show the stronger association (std β = −0.20 [ P = .02] vs std β = 0.11 [ P = .20]). ΔFAC showed no significant association for the linear or quadratic model (std β = 0.12 [ P = .10] vs std β = 0.14 [ P = .34]) ( Figure 3 C). This was similar in a group restricted to pre- and posttreatment only (std β = −0.24 [ P = .09] vs std β = −0.10 [ P = .49]) ( Figure 3 D).