Article Text
PDF
Abstract
Background RV performance is an important determinant of outcomes in children with pulmonary arterial hypertension (PAH). RV stroke work (RVSW), the product of mean pulmonary artery pressure and stroke volume, integrates contractility, afterload and ventricular-vascular coupling. RVSW has not been evaluated in children with PAH. We tested the hypothesis that RVSW would be a predictor of outcomes in children with PAH.
Methods Patients in the Children's Hospital Colorado PAH database were evaluated retrospectively, and those with idiopathic PAH and those with minor or repaired congenital heart disease were included. Haemodynamic data were obtained by catheterisation and echocardiography, performed within 3 months. RVSW was calculated: mean pulmonary arterial pressure × stroke volume, and indexed to body surface area. Statistics included Kruskal–Wallis, Wilcoxon rank sum, and Spearman correlation.
Results Fifty patients were included. Median age of the cohort was 9.5 (6.0, 15.7) years, with a median indexed pulmonary vascular resistance (PVRi) of 6.5 (3.7, 11.6) WU m2. RVSW had a significant association with PVRi (r=0.6, p<0.0001), tricuspid annular systolic plane excursion (r=0.55, p=0.0001), and RV fractional area change (r=−0.4, p=0.005). Grouped by WHO class, there was a significant difference in RVSW (p=0.04). Need for atrial septostomy and death were associated with higher RVSW (p=0.04 and p=0.03, respectively).
Conclusions RVSW can be estimated in children with PAH, and is significantly associated with abnormal WHO class, the need for septostomy, as well as mortality. Indices accounting for RV performance as well as ventricular-vascular coupling may be useful in the prognosis and, hence, management of children with PAH.
Statistics from Altmetric.com
Introduction
RV performance is an important determinant of morbidity and mortality of children with pulmonary arterial hypertension (PAH).1–4 The assessment of RV function and prediction of outcomes related to pulmonary haemodynamics remains challenging.5 Invasive evaluation of RV performance in children is limited to subjective assessment during RV angiogram and inference based on haemodynamic parameters, such as pulmonary vascular resistance (PVR) and RV diastolic pressure.
Cardiac output and stroke volume are related to outcomes in PAH,6 and are determined by ventricular function and by ventricular-vascular (VV) coupling.7 Furthermore, pulmonary arterial (PA) capacitance, which can be estimated as the quotient of stroke volume over pulse pressure, is a strong determinant of RV work.8 We have previously found that PA capacitance is associated with clinical outcomes in children with PAH.9 Pressure-volume loops are the gold standard methodology for assessment of ventricular function and for determination of ventricular stroke work, but are difficult to apply routinely in clinical practice. Alternatively, ventricular stroke work, defined as the area bounded by the pressure-volume loop, can be estimated as the product of the mean pulmonary artery pressure and the stroke volume. Ventricular stroke work, although dependent on preload,10 integrates contractility, afterload, and VV coupling, and therefore may be an important determinant of RV performance, and ultimately clinical outcomes, in PAH. RV stroke work (RVSW) has been studied in adult subjects undergoing LV assist device placement, as a means of predicting whether the patient will require biventricular mechanical support11–14; however, to our knowledge, RVSW has not been studied in children with PAH. The aim of this study is to evaluate invasively derived RVSW in a population of children with PAH, and to investigate its association with clinical outcomes. We hypothesised that RVSW would be associated with RV systolic performance and clinical outcomes in children with PAH.
Methods
Subjects
A cross-sectional retrospective chart review was conducted. Subjects in the Children's Hospital Colorado PAH database were reviewed for inclusion in the study. Patients were included if they met diagnostic criteria for PAH15 (mean pulmonary artery pressure of greater than 25 mm Hg, pulmonary capillary wedge pressure of less than 15 mm Hg, and indexed PVR (PVRi) of greater than 3 Wood units m2). The time of diagnosis of PAH was the time at which these criteria were met; time since diagnosis was defined as the number of days between this initial catheterisation and the catheterisation at which the study data was collected. Exclusion criteria from the database include age greater than 22 years. Subjects in whom PAH was related to pulmonary venous or left heart obstructive lesions were also excluded. Subjects were included in the present analysis based on the availability of cardiac catheterisation data and echocardiographic data within 90 days. Demographic and clinical data were obtained from medical records, including cardiac catheterisation and echocardiogram reports. Medications used to treat pulmonary hypertension were also reviewed and tallied.
Subjects were grouped into one of four categories, based on the aetiology of their PAH: (1) idiopathic PAH, (2) unrepaired congenital heart disease (CHD), (3) repaired CHD and (4) haemodynamically insignificant CHD. Patients in the fourth group included those with small patent foramen ovale, or tiny muscular ventricular septal defects (VSD), which were not felt to be causative of PAH; this determination was confirmed by the absence of significant shunting by oximetry. Approval for the study was obtained from the Colorado Multiple Institutional Review Board (COMIRB).
Cardiac catheterisation
Right heart catheterisation was performed during respiration of 21% oxygen and under either general anaesthesia, or conscious sedation in rare cases, based on clinician preference. A fluid-filled, end-hole catheter was used to obtain right atrial (RA), RV, PA, and pulmonary artery occlusion pressures. Cardiac output was measured by thermodilution in those without intracardiac shunts, and by oximetry (Fick method) when felt to be appropriate by the cardiologist performing the catheterisation. PVR was calculated by dividing the trans-pulmonary gradient (mean PA pressure/pulmonary capillary wedge pressure) by pulmonary blood flow; PVR was then indexed to body surface area (PVRi, Wood units m2). Cardiac catheterisation reports were reviewed in order to obtain anthropometric data, right heart pressures, and cardiac output.
Echocardiography
Echocardiography was performed using a Vivid 7 (GE Healthcare, Milwaukee, Wisconsin, USA) or iE33 (Philips Medical Systems, Andover, Massachusetts, USA) platform, with probe frequencies appropriate for body habitus. Studies were also performed with subjects breathing room air during quiet respiration. Echocardiogram reports were reviewed to obtain anthropometric data, and echocardiographic images were reviewed offline, using commercially available software (AGFA Healthcare Cardiovascular Review Station V.2.14.03, Mortsel, Belgium). Spectral Doppler patterns of tricuspid regurgitation were analysed to determine peak velocity of the regurgitant jet, when present. The duration of the regurgitant jet was also noted; the RV ejection time was obtained from the spectral Doppler pattern of interrogation of the RV outflow tract from the parasternal window, for the purposes of calculating the myocardial performance index (MPI).16 ,17 Tricuspid annular systolic plane excursion (TAPSE) was measured as previously described.18 Similarly, the RV end-diastolic area and end-systolic area were obtained by planimetry, in order to calculate RV fractional area change (fractional area change=end-diastolic area–end-systolic area/end-diastolic area×100).
Calculation of RVSW
Stroke volume (mL/beat) was determined by dividing pulmonary blood flow (mL/min) by the heart rate (beats/min). RVSW was calculated by multiplying mean pulmonary artery pressure by stroke volume. RA pressure was not subtracted from mean pulmonary artery pressure because its impact on the product of stroke volume and mean arterial pressure is small. RVSW was then indexed to body surface area, resulting in a metric with the units: (mm Hg×mL/m2).
Statistical analysis
Continuous data are presented as median (IQR), and categorical data are presented as proportions (%). The Shapiro–Wilk test was used to determine normality. Comparison of median values between two groups was performed using the Wilcoxon rank sum test; comparison of median values between greater than two groups was performed using the Kruskal–Wallis test. Fisher's exact test was used to evaluate the association between two categorical variables. Correlation between RVSW and other continuous variables was performed by calculating Spearman correlation coefficients. Univariate logistic regression was performed to assess for associations between predictor variables and a dichotomous outcome. Cox regression (proportional hazard modelling) was performed to assess for the contribution of different follow-up periods to the analysis. A p value of <0.05 was determined to be statistically significant. STATA, V.12 (StataCorp. 2011. Stata Statistical Software: Release 12, College Station, TX: StataCorp LP), was used to perform the analysis.
Results
Of the 135 patients for whom cardiac catheterisation data was available, 50 patients had echocardiographic data available within the acceptable time frame. RVSW could not be calculated for one subject due to incomplete haemodynamic data.
Baseline characteristics and haemodynamic data are presented in table 1 for the entire cohort, as well as a comparison between patients who had adverse outcomes (death, need for atrial septostomy, and WHO class IV) versus those who are living, with WHO class I–III. The median age of the cohort was 9.6 (6.0, 15.7) years, and there was a slight female preponderance (60%). Twenty-one subjects had idiopathic PH (43%), 13 had repaired CHD (26%), 11 (23%) had haemodynamically insignificant CHD, and four patients (8%) had unrepaired CHD. The diagnoses of the subjects with unrepaired CHD were: muscular VSD, perimembraneous VSD, patent ductus arteriosus, and ostium secundum atrial septal defect. The median number of days between cath and echo is 1 (1, 23) days. There was no statistically significant difference in baseline characteristics between the patients in the dichotomous outcome groups.
Baseline patient characteristics and haemodynamic data
Median PA pressure for the entire cohort was 33.5 (24, 54) mm Hg, and the median PVRi was 6.5 (3.7, 11.6) WU m2. Median-indexed RVSW for the entire cohort was 1666 (1120, 2124) mm Hg mL/m2. The adverse outcome group had higher PA pressure and PVRi, with lower stroke volume and cardiac index. RV MPI was significantly higher in the adverse outcome group, indicating poorer RV performance. There was no significant difference in TAPSE, RV fractional area change, or RVSW; however, a trend toward poorer measured RV function was present for TAPSE, RV FAC and RVSW in the adverse outcome group.
Haemodynamic characteristics of patients within each diagnostic group, with a comparison of median values by the Kruskal–Wallis test, are shown in online supplementary table S1. There was no difference between the diagnostic groups in terms of age, body surface area, heart rate, or cardiac index. There was a statistically significant difference in PVRi (p=0.02) and PA pressure (p=0.04); however, no difference in stroke volume or indexed RVSW was noted.
Figure 1 is a box plot demonstrating the significant difference in indexed RVSW by WHO class (p=0.04). WHO class was assessed as a dichotomous variable, comparing class I versus classes II–IV; subjects in WHO class I had a significantly lower indexed RVSW (p=0.02) (see online supplementary figure S1). Subjects were grouped by how many medications used to treat pulmonary hypertension they were prescribed, and RVSW was compared in figure 2, demonstrating that RVSW increased incrementally with each additional medication (p=0.04).
Indexed RV stroke work (RVSW) by WHO class. When grouped by WHO heart failure class, there was a significant difference in indexed RVSW (p=0.04). RVSW appeared to trend upward in WHO classes 1, 2, and 3, but showed a decline in WHO class 4. Box plots presented here feature the median as a horizontal line within the box, while the box itself represents the IQR (25–75th); the ‘whiskers’ represent the bounds of the extreme values in which Q3+1.5(Q3-Q1) and Q1–1.5*(Q3-Q1), respectively, and single points are, therefore, outliers.
Indexed RV stroke work (RVSW) by number of PH medications. There is a significant difference in indexed RVSW by number of antipulmonary hypertensive medications (p=0.04).
The need for atrial septostomy was interpreted as an indicator of disease severity. Figure 3A is a box plot depicting indexed RVSW by the need for atrial septostomy. Subjects who underwent atrial septostomy had significantly higher indexed RVSW (p=0.04). Similarly, those subjects who were deceased had a significantly higher RVSW (p=0.03), as shown in figure 3B.
(A and B) Indexed RV stroke work (RVSW) by septostomy and mortality. Patients who were felt to require atrial septostomy had a higher indexed RV stoke work (p=0.04). Patients who died had a higher indexed RVSW (p=0.03).
There is a statistically significant monotone increasing association between indexed RVSW and PVRi (R=0.57, p<0.0001) (see online supplementary figure S2). RV MPI had a significant monotone increasing association as well (R=0.4, p=0.005). TAPSE was plotted against unindexed RVSW, and also had a significant monotone increasing association (R=0.55, p=0.0001). RV fractional area change had an insignificant association with indexed RVSW (R=−0.24, p=0.11). Brain-type natriuretic peptide (BNP) was not associated with RVSW (R=0.12, p=0.4), nor was NT-proBNP (R=0.09, p=0.6). Six-minute walk distance was also not associated with RVSW (R=0.01, p=0.93).
Subjects with a high PVRi (greater than the 50th percentile) and a low RVSW (less than the 50th percentile) were grouped together, for the purposes of evaluating the impact of VV coupling on RV performance. The rationale behind this categorisation was that patients who have high RV afterload (elevated PVRi) should have high RVSW if VV coupling is preserved. Those with high PVRi and low RVSW were hypothesised to have poor VV coupling. Figure 4 is a comparison of RV MPI between subjects with preserved coupling and those who were uncoupled, by the definition provided above. There was a significantly increased RV MPI in the uncoupled group (p=0.02), indicating poorer RV performance.
RV Myocardial Performance Index by coupling status. In subjects with high-indexed pulmonary vascular resistance and low-indexed RV stroke work, defined here as having ventricular-vascular uncoupling, there was a significant difference in RV Myocardial Performance Index (p=0.02).
Subjects were grouped by quartile with regard to indexed RVSW (see online supplementary table S2). There is a significant difference in PVRi (p=0.0001) between the quartiles. This upward trending PVRi with ascending quartiles is also shown in figure 5. No difference in TAPSE or fractional area change was noted between quartiles, but a significant difference in median RV MPI was present (p=0.02).
Pulmonary vascular resistance (PVR) by Indexed RV stroke work (RVSW) quartile. PVR for each RVSW quartile. There was a significant difference in PVR between the groups (p=0.0001).
Univariate logistic regression was performed to assess for associations between indices of RV performance and the dichotomous outcome defined in table 1 (see online supplementary table S3). There were significant associations between patients who had died, had an atrial septostomy, or were in WHO class IV and mean PA pressure, PVRi, and RV MPI. Multivariate logistic regression was performed, and indexed RVSW was not a better predictor than PVRi alone. Kaplan–Meier survival analysis is depicted in figure 6, stratified by RVSW quartiles. The curves show a clear trend toward worse survival in those subjects with ascending RVSW quartiles, without reaching statistical significance by Cox regression analysis (p=0.1), likely due to the relatively small sample size in each group.
Kaplan–Meier Survival Analysis stratified by indexed RV stroke work (RVSW) quartile. When stratified by indexed RVSW quartile, there was a trend toward poorer outcomes in subjects with higher RVSW. This trend did not reach statistical significance, likely due to the relatively small sample size in each group.
Discussion
The ability of the RV to maintain cardiac output in the face of increased afterload is an important determinant of outcomes in PAH. We present the novel use of estimated RVSW as a means of evaluating RV performance in children with PAH. This method approximates the area bounded by the pressure-volume loop and thereby incorporates elements of RV contractile performance, RV afterload and VV coupling. More specifically, RVSW informs us as to the ability of the RV to generate stroke volume in the face of increased afterload. In this cohort, our findings reveal an association between elevated indexed RVSW and adverse outcomes, specifically functional decline (WHO class), need for septostomy (a marker of disease severity) and death. We also showed that indexed RVSW has linear associations with echocardiographic metrics of RV performance including TAPSE and the RV MPI.
The strong correlation we discovered between PVR and RVSW should not be surprising, in light of the fact that resistance in the pulmonary vascular bed is a prime determinant of PA pressure. Though PA pressure is common to PVR and RVSW, by incorporating stroke volume, RVSW does not neglect the pulsatile nature of RV performance. Multivariate logistic regression was performed to investigate whether RVSW provided additional prognostic value in determining which patients may experience adverse outcomes. RVSW did not appear to be a better predictive index than PVRi alone; however, RVSW remains a valid measure of RV performance and VV coupling. The fact that PVRi and RVSW have common determinants may make it difficult to ultimately distinguish which will be a superior predictor of adverse events in larger, prospective studies.
We noted a positive linear relationship between RV MPI and RVSW. In an animal model of pulmonary hypertension, Guihaire et al19 described a relationship between RV MPI and VV coupling; specifically, as MPI increased (indicating worsening function), there was a greater degree of uncoupling, as determined by an end-systolic RV elastance (Ees) over PA elastance (Ea) of less than 1. Our analysis of RV MPI in subjects with elevated PVRi and low RVSW, whom we hypothesise have poor VV coupling, revealed a significantly elevated MPI. We further hypothesise that the worse MPI may be driven by the relationship between isovolumic contraction time and PA pressures. As the force generated by the RV becomes more of a square wave under higher distal resistance, the VV coupling and MPI would both become more abnormal. Further study will be required to define the relationship between RVSW and Ees/Ea in this population.
The trajectory of RVSW over time in children with PH is also of interest. Our hypothesis is that as RV afterload increases, there will be an increase in RVSW; as mean PA pressure remains chronically elevated, we hypothesise that RVSW will either plateau if stroke volume remains constant, or begin to decline as the ability of the RV to eject fails. This would be an example of how RVSW might inform the clinician of VV uncoupling, where PVR may fail to do so. We speculate that monitoring RVSW in this setting may also alert the clinician to impending RV failure. This may be reflected in our data, where subjects categorised as having WHO classes 1 through 3 show increasing RVSW, while individuals who are WHO class 4 appear to have lower indexed RVSW.
Other factors, such as proximal PA compliance, also influence RVSW. In a study by Kuehne et al,7 simultaneous quantification of RV pressure and volume was performed in children with PAH by making invasive pressure measurements during cardiac MRI. These subjects with PAH had significantly increased load-independent measures of RV contractility. Kuehne et al also evaluated PA elastance in their small cohort of subjects with PAH, and found a significant increase compared to controls. Importantly, in conditions where PVR is elevated, proximal pulmonary artery properties may be important determinates of pulmonary artery capacitance and of RVSW.
In our cohort, there was an association between higher RVSW and the number of vasoactive medications aimed at decreasing PA pressures, shown in figure 3. This association may be driven by the fact that the treating physicians were more aggressive in patients with more severe disease. The interesting aspect of this association is that a dose-response relationship was not seen; in other words, the effect of a higher number of medications was not a decrease in RVSW. This may indicate that treatment of patients with the most severe pulmonary hypertension remains inadequate, and that further study is necessary to determine which medications are most helpful in this setting.
Study limitations
This is a retrospective study, and the available data are extracted from medical records, which make them vulnerable to measurement bias. Also, ours is a tertiary referral centre where children with more severe PAH are referred, so there is potential selection bias. The relatively small sample size, particularly of the group with unrepaired CHD, makes the data vulnerable to outliers; however, we feel that trends in the data of combined diagnostic categories make the statistically significant relationships between RVSW and clinical outcomes more generalisable, given that the relationships appear to hold true for a heterogeneous study cohort. Multivariate logistic regression was hindered by rare mortality.
Conclusions
In summary, this study shows that indexed RVSW can be calculated from haemodynamic data derived from cardiac catheterisation in children with PAH, and that these data are associated with adverse outcomes, including mortality. Additionally, through the common haemodynamic components of pressure and flow, RVSW is related to PVRi, as well as other non-invasive indices of RV function. Further study is required to delineate the value of measurement of RVSW in the longitudinal management of children with PAH and its ability to detect VV ‘uncoupling’. In this regard, non-invasive measurement of RVSW would hold great appeal, as we have shown that this is possible for measurement of pulmonary artery capacitance.9 The addition of more comprehensive measures of RV performance, which account for RV function as well as VV coupling, may be useful in the prediction of RV failure in this high-risk population.
Key messages
What is already known on this subject?
-
RV performance is an important determinant of outcomes in children with pulmonary arterial hypertension (PAH). RV stroke work (RVSW), the product of mean pulmonary artery pressure and stroke volume, integrates contractility, afterload and ventricular-vascular coupling; however, this has not been evaluated in children with PAH.
What this study adds?
-
We present the novel use of estimated RVSW as a means of evaluating RV performance in children with PAH. Our findings reveal an association between elevated indexed RVSW and adverse outcomes, specifically functional decline (WHO class), need for septostomy (a marker of disease severity) and death. We also showed that indexed RVSW has linear associations with echocardiographic metrics of RV performance including tricuspid annular systolic plane excursion and the RV myocardial performance index.
How might this impact on clinical practice?
-
RVSW may prove to be a more comprehensive measure of RV performance, which accounts for RV function as well as ventricular-vascular coupling; furthermore, it may be useful in the prediction of RV failure in this high-risk population.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
- Data supplement 2 - Online figures
Footnotes
-
KSH and MKF are co-senior authors.
-
Contributors The seven authors are justifiably credited with authorship, according to the authorship criteria. MVDM: conception, design, analysis and interpretation of data, drafting of the manuscript, final approval given. AY: acquisition of data, analysis and interpretation of data, final approval given. LM: conception, design, analysis and interpretation of data, critical revision of manuscript, final approval given. BL: conception, design, analysis and interpretation of data, critical revision of manuscript, final approval given. DI: critical revision of manuscript, final approval given. KH: Conception, design, analysis and interpretation of data, drafting of the manuscript, final approval given. MKF: Conception, design, analysis and interpretation of data, drafting of the manuscript, final approval given.
-
Funding DI is supported in part by a grant from the National Institutes of Health and the National Center for Advancing Translational Sciences, UL1 TR000154. KH is supported, in part, by the National Institutes of Health K25 grant, HL094749, and the R01 grant, HL114753, from the National Heart, Lung and Blood Institute.
-
Competing interests None.
-
Ethics approval Colorado Multi-insitutional Review Board (COMIRB).
-
Provenance and peer review Not commissioned; externally peer reviewed.
-
Data sharing statement All relevant data is presented in the manuscript.