Abstract
Background
We tested the hypothesis that non-invasive pulse wave analysis (PWA)-derived systemic circulation variables can predict invasive hemodynamics of pulmonary circulation and the indicator of right heart function, N-terminal pro-brain natriuretic peptide (NT-proBNP), in patients with precapillary pulmonary hypertension (PH).
Methods
This prospective study enrolled patients with group 1 and 4 PH who had complete PWA, NT-proBNP, and hemodynamics data. Risk assessment-based "hemodynamic score (HS)" and principal component analysis-based PWA variable grouping were determined/performed. Models of hierarchical multiple linear regression (HMLR) and receiver operating characteristic (ROC) curves were used to determine the relationships of PWA variables with HS and NT-proBNP and to predict the latter parameters.
Results
Fifty-three PWAs were included. PWA variables were classified into 4 eigenvalue principal components (representing 90% configuration). Univariate analysis showed that left ventricular ejection time (LVET) was significantly negatively associated with HS and NT-proBNP levels. HMLR analysis showed that LVET was still significantly, negatively, and independently associated with HS (B = -0.006 [-0.010~-0.001]) and NT-proBNP (B = -13.47 [-21.20~-5.73]). ROC curve analysis showed that LVET > 306.9 msec and > 313.2 msec predicted the low-risk group of HS (AUC: 0.802; p = 0.001; sensitivity: 100%; and specificity: 59%) and low-to-intermediate risk levels of NT-proBNP (AUC: 0.831; p < 0.001; sensitivity: 100%; and specificity: 59%).
Conclusions
The non-invasive PWA parameter, LVET, is an independent predictor of invasive right heart HS and NT-proBNP levels; it may serve as a novel biomarker of right ventricular function in patients with pre-capillary PH.
Keywords: Chronic thromboembolic pulmonary hypertension, Left ventricular ejection time, Pulmonary arterial hypertension, Pulse wave analysis
Abbreviations
AI, Augmentation index
AUC, Area under curve
BP, Blood pressure
BMI, Body mass index
CI, Cardiac index
CTEPH, Chronic thromboembolic pulmonary hypertension
DBP, Diastolic blood pressure
FA, Factor analysis
FC, Functional class
HMLR, Hierarchical multiple linear regression
HS, Hemodynamic score
LV, Left ventricle
LVET, Left ventricular ejection time
mPAP, Mean pulmonary arterial pressure
NT-proBNP, N-terminal-pro-B type natriuretic peptide
PH, Pulmonary hypertension
PAH, Pulmonary arterial hypertension
PP, Pulse pressure
PVR, Pulmonary vascular resistance
PWA, Pulse wave analysis
RHC, Right heart catheterization
ROC, Receiver operating characteristic
RV, Right ventricle
SBP, Systolic blood pressure
SEVR, Subendocardial viability ratio
SvO2, Mixed central venous oxygen saturation
WHO, World Health Organization
6MWD, Six-minute walking disease
INTRODUCTION
Pulmonary hypertension (PH) is a life-threatening disease despite recent advances in targeted therapy and potential novel treatments1 for pulmonary arterial hypertension (PAH, group 1 PH) and chronic thromboembolic pulmonary hypertension (CTEPH, group 4 PH). A multidimensional approach which includes clinical presentation and monitoring of exercise tolerance, imaging findings, level of N-terminal pro-brain natriuretic peptide (NT-proBNP), and pulmonary hemodynamics, is crucial for risk assessment and goal-orientated therapy.2,3 Currently, parameters obtained from invasive right heart catheterization (RHC) are still the gold standard for PAH diagnosis and the major components for subsequent risk assessment.4 Although it has been reported that non-invasive echocardiographic parameters collected by an experienced echo technician are highly consistent with invasive hemodynamic data,5 substantial inaccuracy remains a great concern,6 particularly for patients without an ideal echocardiographic window. Therefore, only right atrial (RA) area and presence of pericardial effusion, which demand less technical skill to acquire, remain useful in risk assessment.2 It is widely acknowledged that echocardiography is still insufficient to replace RHC in many aspects.7
Pulse wave analysis (PWA) is a non-invasive technique that is used to predict the central aortic pressure waveform from peripheral pressure pulse by applying applanation tonometry. It measures the left ventricular ejection time (LVET)8 and other tonometry-derived variables, such as central blood pressure (BP), augmentation index (AI), subendocardial viability ratio (SEVR), and reflection magnitude. Unlike echocardiographic parameters, PWA data can be easily and reproducibly obtained with a newer generation of PWA device, for which skill requirement is low. PWA-related parameters have been found to be associated with cardiovascular disease.9-11 However, most studies on PWA have focused on left heart disease and systemic vasculopathy, and few studies have explored its role in the evaluation of right heart function and pulmonary circulation.12
The left ventricle (LV) chamber size and LV stroke volume change when the right ventricle (RV) is overloaded.13 This prompted us to postulate that non-invasive and outpatient-based PWA-derived systemic parameters may potentially serve as useful surrogates for RV function, which is traditionally evaluated by invasive and inpatient-based RHC. In this study, we analyzed patients with precapillary PH, PAH and CTEPH, which were each confirmed by RHC for PAH diagnosis and RV function stratification. We aimed to investigate whether PWA parameters are correlated with RHC score and the major RV functional biomarker, NT-ProBNP, in patients with precapillary PH.
METHODS
Study participants, data collection, and PWA
This prospective single-center study enrolled consecutive patients (age, 20-80 years) with groups 1 and 4 PH who were referred to the PH clinic from December 2017 to March 2020. The diagnosis and classification of PH was confirmed by RHC and a diagnostic algorithm.2 This study was approved by the MacKay Memorial Hospital Institutional Review Board (IRB number 19MMHIS 252e). All participants signed informed consent forms.
In the outpatient clinic, each patient underwent PWA performed using a standard brachial cuff equipped to a SphygmoCor Xcel device (AtCor Medical, Sydney Australia), which performed oscillometry. The brachial waveform was then analyzed to generate a central aortic waveform and provide the central waveform variables. Three consecutive recordings of arterial waveforms with an operation index of > 90% were made.
Data on characteristics including age, sex, body mass index (BMI), levels of serum hemoglobin, sodium, creatinine, total bilirubin, and NT-proBNP, and World Health Organization (WHO) functional class (FC) were collected on the date of the RHC. Six-minute walking distance (6MWD) and echocardiographic profiles closest to the date of the RHC (< 30 days apart) were also collected. The following PWA variables were also obtained: central/brachial systolic blood pressure (SBP), diastolic blood pressure (DBP), mean arterial pressure and pulse pressure (PP); LVET and LVET%, defined as the percentage of LVET over cardiac cycle length; SEVR; reflected magnitude; central augmentation pressure, AI at a heart reate of 75 beats/min (AIx@HR75), P1 height, and aortic AIx P2/P1 ratio. P1 was defined as the inflection point of the reflected wave, and P2 was defined as the maximum point of systolic pressure determined by the return of the reflected wave. Hemodynamic parameters including RA pressure, mean pulmonary artery pressure (mPAP), pulmonary vascular resistance (PVR), cardiac index (CI), and mixed central venous oxygen saturation (SvO2) were measured. All other data obtained closest (< 30 days apart) to the date of hemodynamic procedure were selected for the analysis.
PAH risk assessment and the hemodynamic score derived from invasive procedure
According to the European Society of Cardiology guidelines, the risk of PAH is classified as low, intermediate, and high risk,2 with hemodynamic determinants of RA pressure, CI, and SvO2. According to the guidelines, our "hemodynamic score (HS)" was defined as follows: 1, low risk; 2, intermediate risk; and 3, high risk. Each of the 3 determinants had the same weight in scoring. The average score of the 3 determinants was defined as the final HS. Thus, the maximum HS was 3 and the minimum score was 1.
Statistical analysis
All data are presented as mean ± standard deviation (mean ± SD) or frequency (percentage) depending on whether they are continuous or categorical variables. All reported p values were based on two-sided tests and were considered statistically significant if they were less than 0.05. Data were analyzed using IBM SPSS release 24.0 (IBM, Armonk, New York).
Factor analysis
Factor analysis (FA) of extracted indicators such as arterial waveform was performed with principal component extraction and varimax using Kaiser normalization rotation. The purpose of FA is to describe the relationship among many indicators with a small number of principal components. Factors with eigenvalues greater than one were retained and selected to represent over 90% of configuration. While factor loadings above 0.40 were selected, the indicators were considered as the principal components.14,15
Hierarchical multiple regression analysis
The association among right heart function indicators (HS and NT-proBNP), sex, age, BMI, hemoglobin level, creatinine level, LVET, and 4 PWA-derived components from FA were determined using Pearson’s correlation analysis. The joint effects of the independent variables were assessed with hierarchical multiple linear regression analysis. Thus, the above parameters related to HS were included in the hierarchical multiple linear regression analyses in 3 models. The variables in model 1 were sex, age, BMI, hemoglobin level, and creatinine level. The variables in model 2 were all of the variables in model 1 along with LVET. The variables in model 3 were all of the variables in model 2 along with the 4 components.
Receiver operating characteristic curve (ROC)
ROC and area under the ROC curve (AUC) analyses were used to differentiate high from low risk status according to HS or NT-proBNP levels by LVET. The sensitivity and specificity values were generated from the ROC curve. Youden’s index was used to determine the optimal cut-off point of LVET in the ROC curve to maximize the sensitivity and specificity.
RESULTS
Baseline characteristics
A total of 53 PWAs with corresponding hemodynamics reports from 30 patients (49.0 ± 15.5 years, 22 females) obtained less than 1 month apart were included. The demographic and baseline characteristics are presented in Table 1. Fourteen of the patients were newly diagnosed with PH, and 16 patients had had PH for an average of 32.0 ± 25.6 months. Twenty-three and 5 patients were diagnosed with group 1 and group 4 PH, respectively. Two patients had PAH but with localized thromboembolism. The WHO FCs of the patients were as follows: FC-1, 2 patients; FC-2, 25 patients; and FC-3, 3 patients. The average 6MWD and serum NT-proBNP level were 418.3 ± 129.9 m and 1077.2 ± 1079.0 pg/mL, respectively. The hemodynamic parameters from RHC were as follows: RAP, 5.6 ± 4.2 mmHg; mPAP, 45.4 ± 14.5 mmHg; PVR, 9.0 ± 4.8 WU; cardiac index, 2.8 ± 0.9 L/ min/m2; and SvO2, 58.7 ± 11.8%. The average HS was 1.7 ± 0.6, indicating an intermediate risk for the enrolled patients. The average RA area was 20.8 ± 8.5 cm2, and 2 patients had pericardial effusion. The variables generated from PWA yielded an average LVET of 289.8 ± 38.0 msec, SEVR of 136 ± 33.8%, AIx@HR75 of 17.0 ± 15.5%, and reflection magnitude of 55.6 ± 8.4%.
Table 1. Demographics and baseline characteristics.
| Characteristic | |
| Age, y | 49.0 ± 15.5 |
| Sex female (male) | 22 (8) |
| Height, cm | 159.9 ± 11.9 |
| Weight, kg | 61.3 ± 14.2 |
| BMI, kg/m2 | 23.8 ± 3.7 |
| Type of PH | |
| Group 1 | 23 |
| Group 4 | 5 |
| Combined group 1 and 4 | 2 |
| WHO functional class | |
| I | 2 |
| II | 25 |
| III | 3 |
| Heart rate, bpm | 82.6 ± 19.0 |
| Peripheral SBP, mmHg | 121.3 ± 16.8 |
| Peripheral DBP, mmHg | 77.1 ± 14.4 |
| Peripheral PP, mmHg | 44.2 ± 9.3 |
| SpO2, % | 94.2 ± 5.3 |
| 6MWD, m | 418.3 ± 129.9 |
| Hemodynamics from RHC | |
| RAP, mmHg | 5.6 ± 4.2 |
| Mean PAP, mmHg | 45.4 ± 14.5 |
| PVR, WU | 9.0 ± 4.8 |
| Cardiac index, L/min/m2 | 2.8 ± 0.9 |
| SvO2, % | 58.7 ± 11.8 |
| Hemodynamic score | 1.7 ± 0.6 |
| Laboratory data | |
| Serum hemoglobin level, g/dL | 13.1 ± 1.7 |
| Serum MCV level, fL | 84.8 ± 11.1 |
| Serum creatinine level, mg/dL | 0.8 ± 0.2 |
| Serum total bilirubin level, mg/dL | 1.1 ± 0.7 |
| Serum sodium level, mEq/L | 139.1 ± 2.5 |
| Serum NT-proBNP level, pg/mL | 1077.2 ± 1079.0 |
| Echocardiography | |
| RA area, cm2 | 20.8 ± 8.5 |
| RVEF, % | 44.4 ± 15.6 |
| TAPSE, mm | 2.2 ± 0.7 |
| LVEF, % | 63.5 ± 8.2 |
| Pericardial effusion | 2 |
| Pulse wave analysis | |
| Central SBP, mmHg | 109.7 ± 16.0 |
| Central DBP, mmHg | 77.1 ± 14.4 |
| Central MBP, mmHg | 91.9 ± 15.0 |
| Central PP, mmHg | 31.0 ± 7.3 |
| LVET, msec | 289.8 ± 38.0 |
| SEVR, % | 136.0 ± 33.8 |
| AIx@HR75, % | 17.0 ± 15.5 |
| Reflection magnitude, % | 55.6 ± 8.4 |
AIx@HR75, augmentation Index standardized for HR at 75 bpm; BMI, body mass index; DBP, diastolic blood pressure; LVEF, left ventricular ejection fraction; LVET, left ventricular ejection time; MBP, mean blood pressure; MCV, mean corpuscular volume; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; PAP, pulmonary artery pressure; PH, pulmonary hypertension; PP, pulse pressure; PVR, pulmonary vascular resistance; RA, right atrial; RAP, right atrium pressure; RVEF, right ventricular ejection fraction; SBP, systolic blood pressure; SEVR, subendocardial viability ratio; SpO2, pulse oximetry; SvO2, mixed venous oxygen saturation; TAPSE, tricuspid annular plane systolic excursion; WHO, World Health Organization; 6MWD, 6-minute walking distance. WU, Wood units.
FA for PWA variables
PWA variables were divided into the 4 components based on eigenvalues and the variable characteristics (Table 2 and Figure 1). All of the factor-pattern coefficients were above 0.75, indicating that all of the items were salient (Table 2). Each of the components was associated with the following: component A, BP (including SBP/DBP/MBP and peripheral SBP/DBP); component B, aortic augmentation by reflected wave (including central AIx@HR7, central AP, aortic P2/P1 ratio, and T2); component C, PP (including central PP, peripheral PP, and P1 height); and component D, cardiac cycle linked variables [including LVET% (i.e., LVET/cardiac cycle duration) and SEVR]. These variables produced by PWA were highly associated with their components.
Table 2. Four components of PWA variables: individual item loadings for factor analysis.
| Components | ||||
| A | B | C | D | |
| Central MBP | 0.995 | |||
| Central DBP | 0.968 | |||
| Peripheral DBP | 0.963 | |||
| Central SBP | 0.930 | |||
| Peripheral SBP | 0.886 | |||
| Aortic P2/P1 ratio | 0.971 | |||
| Central AIx@HR75 | 0.955 | |||
| Central augmentation pressue | 0.898 | |||
| Aortic T2 | 0.812 | |||
| Reflection magnitude | 0.851 | |||
| P1 height | 0.992 | |||
| Peripheral PP | 0.985 | |||
| Central PP | 0.897 | |||
| LVET (%) | 0.885 | |||
| Heart rate | 0.751 | |||
| Inverse SEVR | 0.786 |
Extraction method: factor analysis.
Rotation method: Varimax with Kaiser normalization.
PWA, pulse wave analysis. Other abbreviations are the same as in Table 1.
Figure 1.
Scree plot of the arterial waveform: four components presented with an eigenvalue of 1.05, accounting for 95.59% of variance.
LVET and aortic T2 from PWA significantly predicted invasively derived HS in the univariate analysis
The major aim of this study was to test whether the non-invasive "left-sided" tool, PWA, can independently predict invasive "right-sided" hemodynamics in patients with PAH and CTEPH. To this end, Pearson correlation analysis was first performed (Table 3), which showed that LVET was significantly (p = 0.001) and negatively associated with HS. In addition, the well-known PAH risk factor, creatinine level (p = 0.026), was significantly associated with HS, with component B showing a borderline significance in this association (p = 0.075). Univariate analysis showed that (Table 4) LVET was highly associated with HS (p = 0.001). Among individual factors in component B, aortic T2 seemed to be most closely associated with HS (p = 0.018).
Table 3. Models of hierarchical multiple linear regression analyses examining for NT-proBNP.
| Model 1 | Model 2 | Model 3 | |||||||
| B | 95% CI | p value | B | 95% CI | p value | B | 95% CI | p value | |
| Sex (male) | -216.84 | -928.33~494.65 | 0.542 | 4.25 | -638.33~646.83 | 0.989 | -227.89 | -891.09~435.30 | 0.490 |
| Age | -0.60 | -21.22~20.02 | 0.953 | 1.63 | -16.67~19.93 | 0.858 | -1.64 | -29.81~26.54 | 0.907 |
| BMI | -58.33 | -127.76~11.09 | 0.097 | -35.21 | -98.09~27.68 | 0.265 | -42.73 | -104.43~18.96 | 0.169 |
| Hemoglobin level | 45.76 | -155.23~246.74 | 0.648 | -39.59 | -224.15~144.98 | 0.667 | -102.13 | -284.48~80.22 | 0.263 |
| Creatinine level | 509.18 | -807.67~1826.04 | 0.439 | -24.32 | -1229.77~1181.13 | 0.968 | -259.64 | -1424.74~905.46 | 0.654 |
| LVET | -13.47 | -21.20~-5.73 | 0.001 | -18.57 | -30.53~-6.61 | 0.003 | |||
| Component A | 8.53 | -14.99~32.05 | 0.467 | ||||||
| Component B | -20.34 | -57.35~16.67 | 0.272 | ||||||
| Component C | 4.78 | -54.82~64.39 | 0.872 | ||||||
| Component D | -83.70 | -138.57~-28.82 | 0.004 |
CI, confidence interval. Other abbreviations are the same as in Table 1.
Table 4. Simple linear regression analysis for hemodynamic score.
| B | 95.0% confidence interval for B | p value | |
| Component A | |||
| Central mean pressure | 0.003 | -0.008~0.014 | 0.620 |
| Central diastolic pressure | 0.004 | -0.007~0.015 | 0.435 |
| Brachial diastolic pressure | 0.005 | -0.006~0.017 | 0.344 |
| Central systolic pressure | 0.001 | -0.010~0.011 | 0.891 |
| Brachial systolic pressure | 0.001 | -0.009~0.011 | 0.821 |
| Component B | |||
| Aortic Aix P2P1 | -0.010 | -0.023~0.003 | 0.140 |
| Central augmentation index | -0.009 | -0.019~0.001 | 0.076 |
| Central augmentation pressure | -0.019 | -0.045~0.007 | 0.145 |
| Aortic T2 | -0.009 | -0.016~-0.002 | 0.018 |
| Reflection magnitude | -0.003 | -0.018~0.012 | 0.696 |
| Component C | |||
| P1 Height | -0.009 | -0.034~0.016 | 0.478 |
| Brachial pulse pressure | -0.007 | -0.022~0.008 | 0.350 |
| Central pulse pressure | -0.011 | -0.030~0.008 | 0.263 |
| Component D | |||
| Ejection duration (%) | -0.011 | -0.037~0.015 | 0.411 |
| Heart rate | 0.006 | -0.003~0.016 | 0.170 |
| SEVR | 0.003 | -0.002~0.008 | 0.191 |
| Component A mean | 0.003 | -0.008~0.014 | 0.596 |
| Component B mean | -0.012 | -0.024~0.001 | 0.075 |
| Component C mean | -0.009 | -0.029~0.010 | 0.336 |
| Component D mean | 0.009 | -0.013~0.031 | 0.404 |
| LVET (msec) | -0.006 | -0.010~-0.003 | 0.001 |
Abbreviations are the same as in Table 1.
LVET from PWA independently predicted the invasively derived HS in the multivariate analysis
To further elucidate whether a PWA parameter, LVET in particular, was an independent predictor of HS in multivariate regression analysis, we used 3 prediction models. Model 1 showed that only creatinine level was significantly associated with HS (p = 0.04). Importantly, model 2 revealed that only LVET was significantly associated with HS (B = -0.006, p = 0.011). Model 3 showed that both components B and D tended to be associated with HS (B = -0.022, p = 0.062 and B = -0.032, p = 0.066, respectively, Table 5), however LVET lost its significant association in this model, implying that both components partially explained the association between LVET and HS.
Table 5. Models of hierarchical multiple linear regression analyses for determining the predictors of hemodynamic score.
| Model 1 | Model 2 | Model 3 | |||||||
| B | 95% CI | p value | B | 95% CI | p value | B | 95% CI | p value | |
| Sex (male) | 0.032 | -0.362~0.426 | 0.872 | 0.129 | -0.246~0.505 | 0.490 | -0.046 | -0.464~0.371 | 0.824 |
| Age | 0.003 | -0.008~0.014 | 0.594 | 0.004 | -0.007~0.015 | 0.451 | 0.013 | -0.005~0.030 | 0.161 |
| BMI | -0.009 | -0.048~0.029 | 0.625 | 0.001 | -0.036~0.038 | 0.962 | 0.010 | -0.029~0.049 | 0.604 |
| Hb | -0.001 | -0.112~0.110 | 0.988 | -0.039 | -0.146~0.069 | 0.473 | -0.007 | -0.122~0.108 | 0.901 |
| Cr | 0.768 | 0.038~1.497 | 0.040 | 0.531 | -0.173~1.235 | 0.135 | 0.417 | -0.316~1.151 | 0.256 |
| LVET | -0.006 | -0.010~-0.001 | 0.011 | -0.006 | -0.013~0.002 | 0.131 | |||
| Component A | -0.005 | -0.020~0.010 | 0.496 | ||||||
| Component B | -0.022 | -0.045~0.001 | 0.062 | ||||||
| Component C | -0.014 | -0.051~0.024 | 0.467 | ||||||
| Component D | -0.032 | -0.067~0.002 | 0.066 |
CI, confidence interval; Cr, creatinine; Hb, hemoglobin. Other abbreviations are the same as in Table 1.
Model 1, adjusted for all sociodemographic variables and laboratory data (Hb, Cr). Model 2, adjusted for model 1 and ejection duration. Model 3, adjusted for model 2 and component A, B, C, and D.
Cut-off value of LVET to differentiate low-risk from intermediate-to-high risk patients
Treating patients with low-risk status is highly recommended by the current PH guidelines. To make non-invasive LVET more clinically useful, we sought to identify a cut-off value of LVET for intermediate-to high-risk hemodynamic status using ROC curve analysis. The cuff-off value with the maximum Youden’s index of LVET was 306.9 msec with a sensitivity of 100% and a specificity of 59%, which indicated that patients with LVET > 306.9 msec were certainly in the low-risk group. LVET was demonstrated to be an ideal tool to confirm the low-risk hemodynamic status in patients with PAH and CTEPH (AUC, 0.802, p = 0.001; Table 6 and Figure 2).
Table 6. Recommended LVET cut-off value derived from the ROC curve analysis by the comparison between low- and intermediate-to-high risk patients in terms of hemodynamic scores.
| Hemodynamics score (“1” vs. “2 + 3”) | AUC | 95% CI | p value | Cut-off value | Sensitivity | Specificity | Youden’s index |
| LVET | 0.802 | 0.683-0.921 | 0.001 | 307 | 1.00 | 0.59 | 0.60 |
AUC, area under curve; CI, confidence interval; ROC, receiver operating characteristic. Other abbreviations are the same as in Table 1.
Scores of “1” and “2 + 3” denote low-risk status and intermediate-to-high risk status, respectively.
Figure 2.
ROC curve for hemodynamics scores of low-risk vs. intermediate-to-high risk according to LVET. LVET, left ventricular ejection time; ROC, receiver operating characteristic.
LVET was also an independent predictor for NT-proBNP level, another RV functional marker
Apart from invasively derived RHC data, NT-ProBNP is another highly accurate biomarker of RV failure in patients with PH. To confirm the concept that PWA parameters are a useful proxy for RV dysfunction, we performed Pearson correlation analysis again. The results showed that LVET and component A were significantly associated with NT-proBNP level (p < 0.001 and 0.017, respectively; Supplementary Table 1). Univariate analysis, as expected, revealed that LVET was highly associated with NT-ProBNP level; LVET% and SEVR (both are closely related to LVET) also showed significant associations. In addition, several factors in component A including central mean pressure, central DBP, and brachial DBP were positively correlated to NT-proBNP level (Supplementary Table 2). We also used 3 models to analyze the independent predictors of NT-proBNP with hierarchical multiple linear regression, which revealed the following: model 1, no significant relationship between NT-proBNP level and main demographic variables; model 2, significant negative association between LVET and NT-proBNP level (B = -13.47, p = 0.001); model 3, significant correlation of LVET and component D with NT-proBNP level [B = -18.57 (p = 0.003) and B = -83.70 (p = 0.004), respectively; Supplementary Table 3). ROC curve analysis showed that LVET could differentiate low-to-intermediate from high risk levels of NT-proBNP (AUC = 0.831, p < 0.001; Supplementary Table 4 and Supplementary Figure 1). The cut-off value with maximum Youden’s index of LVET was 313.2 msec with a sensitivity of 100% and a specificity of 59%, which indicated that the patients with LVET > 313.2 msec did not have a high-risk level of NT-ProBNP (> 1400 pg/ml).
Supplementary Table 1. Pearson’s correlation analysis between noninvasively derived variables and NT-proBNP level.
| NT-ProBNP | ||
| r | p value | |
| Sex | -0.107 | 0.461 |
| Age | 0.052 | 0.721 |
| BMI | -0.248 | 0.083 |
| Hemoglobin level | 0.123 | 0.396 |
| Creatinine level | 0.105 | 0.483 |
| LVET | -0.547 | < 0.001 |
| Component A | 0.336 | 0.017 |
| Component B | -0.071 | 0.625 |
| Component C | -0.201 | 0.161 |
| Component D | -0.054 | 0.708 |
Abbreviations are the same as in Table 1.
Supplementary Table 2. Simple linear regression analysis for NT-proBNP.
| B | 95.0% confidence interval for B | p value | |
| Component A | |||
| Central mean pressure | 24.044 | 5.273~42.815 | 0.013 |
| Central diastolic pressure | 26.663 | 7.926~45.401 | 0.006 |
| Brachial diastolic pressure | 28.900 | 10.091~47.709 | 0.003 |
| Central systolic pressure | 17.686 | -0.762~36.135 | 0.060 |
| Brachial systolic pressure | 13.428 | -4.725~31.581 | 0.143 |
| Component B | |||
| Aortic AIx P2P1 | -7.357 | -31.959~17.245 | 0.550 |
| Central augmentation index | -8.706 | -27.261~9.850 | 0.350 |
| Central augmentation pressure | -17.040 | -64.558~30.479 | 0.474 |
| Aortic T2 | -4.068 | -17.930~9.793 | 0.558 |
| Reflection magnitude | 13.018 | -16.110~42.146 | 0.373 |
| Component C | |||
| P1 height | -26.728 | -70.863~17.407 | 0.229 |
| Brachial pulse pressure | -21.649 | -48.148~4.851 | 0.107 |
| Central pulse pressure | -20.597 | -55.264~14.071 | 0.238 |
| Component D | |||
| LVET % | -62.660 | -106.868~-18.453 | 0.006 |
| Heart rate | 4.102 | -13.214~21.419 | 0.636 |
| SEVR | 14.809 | 7.319~22.298 | < 0.001 |
| Component A mean | 23.907 | 4.451~43.363 | 0.017 |
| Component B mean | -5.908 | -30.085~18.269 | 0.625 |
| Component C mean | -24.387 | -58.838~10.065 | 0.161 |
| Component D mean | -7.581 | -48.091~32.929 | 0.708 |
| LVET (msec) | -13.815 | -19.954~-7.677 | < 0.001 |
Abbreviations are the same as in Table 1.
Supplementary Table 3. Models of hierarchical multiple linear regression analyses examining for NT-proBNP.
| Model 1 | Model 2 | Model 3 | |||||||
| B | 95% CI | p value | B | 95% CI | p value | B | 95% CI | p value | |
| Sex (male) | -216.84 | -928.33~494.65 | 0.542 | 4.25 | -638.33~646.83 | 0.989 | -227.89 | -891.09~435.30 | 0.490 |
| Age | -0.60 | -21.22~20.02 | 0.953 | 1.63 | -16.67~19.93 | 0.858 | -1.64 | -29.81~26.54 | 0.907 |
| BMI | -58.33 | -127.76~11.09 | 0.097 | -35.21 | -98.09~27.68 | 0.265 | -42.73 | -104.43~18.96 | 0.169 |
| Hemoglobin level | 45.76 | -155.23~246.74 | 0.648 | -39.59 | -224.15~144.98 | 0.667 | -102.13 | -284.48~80.22 | 0.263 |
| Creatinine level | 509.18 | -807.67~1826.04 | 0.439 | -24.32 | -1229.77~1181.13 | 0.968 | -259.64 | -1424.74~905.46 | 0.654 |
| LVET | -13.47 | -21.20~-5.73 | 0.001 | -18.57 | -30.53~-6.61 | 0.003 | |||
| Component A | 8.53 | -14.99~32.05 | 0.467 | ||||||
| Component B | -20.34 | -57.35~16.67 | 0.272 | ||||||
| Component C | 4.78 | -54.82~64.39 | 0.872 | ||||||
| Component D | -83.70 | -138.57~-28.82 | 0.004 |
CI, confidence interval. Other abbreviations are the same as in Table 1.
Supplementary Table 4. Cut-off score estimated using the ROC curve by comparing patients with low-to-intermediate and high levels of NT-proBNP.
| NT-proBNP (“≤ 1400” vs. “> 1400”) | AUC | 95% CI | p value | Cut-off value | Sensitivity | Specificity | Youden’s index |
| LVET | 0.831 | 0.719-0.943 | < 0.001 | 313.2 | 1.0 | 0.59 | 0.59 |
AUC, area under curve; CI, confidence interval; LVET, left ventricular ejection time; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; ROC, receiver operating characteristic.
Supplementary Figure 1.
ROC curve for NT-ProBNP (cut-off level, 1400 pg/mL) according to LVET. LVET, left ventricular ejection time; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; ROC, receiver operating characteristic.
DISCUSSION
The results of this study clearly demonstrated that LVET could independently predict invasively derived HS and differentiate low from intermediate-high risk patients with PAH and CTEPH. In line with this finding, the ability of LVET to independently predict levels of NT-proBNP — a sensitive marker of RV function in patients with PH — was confirmed.16 The main aim of this study was to investigate whether the non-invasive and left heart circulation-based PWA could predict invasive and right heart circulation-based hemodynamics through the concept of RV-LV interaction.
Previous studies have referred to LVET as an index of LV systolic function and emphasized its unique role in the prediction of heart failure and survival of patients with PH in an intensive care unit.17-19 Other factors that shorten LVET are decreased LV preload and mitral regurgitation, whereas factors that prolong LVET are increased afterload, diastolic dysfunction, and certain medications.8,20,21 The pathophysiology of RV-LV interaction in patients with PH emphasizes two major concepts: first, an increase in RV afterload contributes to the limitation of RV output, which facilitates right-sided filling pressures and underfilling of the LV;22 second, PH-mediated ventricular interdependence leads to bowing of the septum towards the left ventricle and impairment of LV filling.23,24 Consequently, LV preload and a decrease in stroke volume shorten LVET. In our study, all of the patients had preserved LV function, and LVET could indirectly but consistently reflect RV hemodynamics as well as NT-proBNP levels. In addition, LVET may reflect RV performance even more directly through the phenomenon of "ventricular interdependence".25
More than 20 parameters have been generated from the pulse wave. For analyzing pulse wave conformation, FA can be a useful tool to reduce complex parameters to a small number of principal components. In our study, we categorized pulse wave parameters into 4 principal components. Although LVET was originally grouped into component D, it was selected as an independent factor in our analysis because it has been reported to be an independent predictor of LV heart failure.17,18 Our intention was to emphasize that LVET among all PWA parameters may be particularly useful in the prediction of HS through the mechanism of ventricular interdependence, not to indicate that LVET is independent of other PWA parameters. Indeed, hierarchical multiple linear regression analysis revealed that LVET was independently associated with HS in model 2. However, after including PWA components into model 3, the significant relationship between LVET and HS was lost. Components B (aortic augmentation by reflected wave) and D (heart rate-linked variables) were borderline associated with HS (p = 0.062 and 0.066, respectively). Clearly, the presence of components B and D substantially diminished the association of LVET with HS. With fixed vascular elasticity, component B may be positively related to stroke volume; furthermore, heart rate-related component D can be considered to be a related function of LVET. Owing to this mutual dependence of these pulse wave-derived variables on hemodynamic responses, the independent impact of LVET on HS was thus attenuated.
Research on pulsatile hemodynamics and NT-proBNP level has focused on left heart failure.26-28 However, the relationship between these parameters and NT-proBNP level in patients with PH-related RV failure is still elusive. In our results, LVET and component D were significantly and negatively correlated to NT-proBNP level as shown in hierarchical multiple linear regression analysis. Although the change in component D (heart rate-linked variables) can be partially explained by the change in LVET, component D may reflect something beyond LVET. Heart rate-related information actually reveals overall sympathetic activation and general cardiac burden, making component D an independent predictor of NT-ProBNP in our model.
Investigations have revealed that NT-proBNP level is highly associated with the hemodynamics derived from RHC;29,30 our results also confirmed this finding. Beyond basic characteristics, hemoglobin level, biochemistry, and PWA parameters, NT-proBNP level remained the most important and independent predictor of invasively derived HS. Although CO, RA pressure,31 and PVR5,32 can also be estimated using non-invasive Doppler echocardiography, this method is limited by the requirement for relatively high skill, a good echocardiographic window, and by a substantially higher rate of inaccuracy than invasive methods.6 All of these limitations make echocardiography a better tool for screening than for confirming pulmonary hemodynamics. For the first time, we showed that LVET, based on the mechanism of RV-LV interaction, derived from non-invasive PWA, which is highly reproducible with low technical requirement, could predict the severity of invasive HS in patients with precapillary PH.
Unlike LV ejection fraction, LVET does not have a clearly established normal range. A previous study suggested that LVET < 355 msec could potentially lead to heart failure if left untreated in a community-based population,17 and another study reported that the cut-off for predicting mortality in patients with heart failure with reduced ejection fraction (HFrEF) was 248 msec.19 In our study, the best cut-off values of LVET to differentiate low from intermediate-to-high risk of high HS and low-to-intermediate from high risk of high NT-proBNP level were 306.9 and 313.2 msec, respectively, in patients with PH. These two values were between the cut-off values of the general population and patients with HFrEF. RV failure may have an indirect, and thus, relatively smaller impact on LVET than LV failure, which has a direct influence on LVET. In our study, both cut-off values of LVET had 100% sensitivity with a moderate specificity in the prediction of high-risk levels of HS and NT-proBNP levels. This means that patients with PH with an LVET value above these cut-off values can be confidently allocated to the low-risk group.
This study was limited by its relatively small sample size, and thus, potential bias may exist. Furthermore, the PWA was done on an outpatient basis. Although the PWA closest to the date of RHC (reference standard) was chosen as the index test (all durations between the two examinations < 30 days), potential PWA and hemodynamic variations during this period may still have occurred. Nevertheless, the outpatient-based enrollment implies a relatively stable status of the enrolled patients, and the recommended follow-up interval for these patients is 3-6 months,1 suggesting that a significant variation in the period of < 30 days is unlikely in our study. Furthermore, the robust association of LVET with right heart hemodynamics was doubly confirmed by its association with HS and NT-proBNP levels, both of which are highly reliable indicators of right heart function. Although the role of PWA in pre-capillary PH is ideally tested in a case-control design, invasive RHC in control subjects without suspected PH is generally prohibited due to ethical considerations. Nevertheless, we further compared LVET in the high-risk pre-capillary PH patients with that in healthy young males (n = 32, unpublished data) and found that LVET was significantly shorter in the high-risk pre-capillary PH patients (271 ± 22 vs. 309 ± 17 ms, p = 0.0034). However, LVET in the low-to-intermediate-risk pre-capillary PH patients was not significantly different from that in the healthy young subjects (305 ± 39 vs. 309 ± 17 ms, p = 0.646). These data further strengthen the usefulness of LVET in identifying different risk levels in patients with pre-capillary PH.
CONCLUSIONS
In conclusion, non-invasive PWA-derived LVET demonstrated a relatively high predictive value for both invasively derived HS and NT-proBNP level. LVET may potentially serve as a non-invasive, easily available, and reliable tool to evaluate right heart function during regular risk assessments in patients with precapillary PH.
Acknowledgments
This work was supported by the Ministry of Science and Technology (MOST 108-2314-B-715-008-MY3), MacKay Memorial Hospital (MMH-E-111-03), and MacKay Medical College (MMC-RD-110-1B-P013), Taipei, Taiwan. We thank Miss Wei-Lun Chang and Miss Ya-Ting Chang for the PWA measurement and data collection.
DECLARATION OF CONFLICT OF INTEREST
All the authors declare no conflict of interest.
REFERENCES
- 1.Hsu CH, Huang WC, Chang WT. Future perspectives of pulmonary hypertension treatment. Acta Cardiol Sin. 2022;38:435–442. doi: 10.6515/ACS.202207_38(4).20220331A. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Galie N, Humbert M, Vachiery JL, et al. 2015 ESC/ERS Guidelines for the diagnosis and treatment of pulmonary hypertension: The Joint Task Force for the diagnosis and treatment of pulmonary hypertension of the European Society of Cardiology (ESC) and the European Respiratory Society (ERS): endorsed by: Association for European Paediatric and Congenital Cardiology (AEPC), International Society for Heart and Lung Transplantation (ISHLT). Eur Heart J. 2016;37:67–119. doi: 10.1093/eurheartj/ehv317. [DOI] [PubMed] [Google Scholar]
- 3.Wu SH, Wu YJ. Regular risk assessment in pulmonary arterial hypertension - a whistleblower for hidden disease progression. Acta Cardiol Sin. 2022;38:113–123. doi: 10.6515/ACS.202203_38(2).20211005A. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Humbert M, Weatherald J. Right heart catheterisation is still a fundamental part of the follow-up assessment of pulmonary arterial hypertension. Eur Respir J. 2018;52:1800738. doi: 10.1183/13993003.00738-2018. [DOI] [PubMed] [Google Scholar]
- 5.Abbas AE, Franey LM, Marwick T, et al. Noninvasive assessment of pulmonary vascular resistance by Doppler echocardiography. J Am Soc Echocardiogr. 2013;26:1170–1177. doi: 10.1016/j.echo.2013.06.003. [DOI] [PubMed] [Google Scholar]
- 6.Parent F, Bachir D, Inamo J, et al. A hemodynamic study of pulmonary hypertension in sickle cell disease. New Engl J Med. 2011;365:44–53. doi: 10.1056/NEJMoa1005565. [DOI] [PubMed] [Google Scholar]
- 7.Farber HW, Foreman AJ, Miller DP, et al. REVEAL Registry: correlation of right heart catheterization and echocardiography in patients with pulmonary arterial hypertension. Congest Heart Fail. 2011;17:56–63. doi: 10.1111/j.1751-7133.2010.00202.x. [DOI] [PubMed] [Google Scholar]
- 8.O'Rourke MF, Pauca A, Jiang XJ. Pulse wave analysis. Br J Clin Pharmacol. 2001;51:507–522. doi: 10.1046/j.0306-5251.2001.01400.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pini R, Cavallini MC, Palmieri V, et al. Central but not brachial blood pressure predicts cardiovascular events in an unselected geriatric population: The ICARe Dicomano Study. J Am Coll Cardiol. 2008;51:2432–2439. doi: 10.1016/j.jacc.2008.03.031. [DOI] [PubMed] [Google Scholar]
- 10.Nürnberger J, Keflioglu-Scheiber A, Opazo Saez AM, et al. Augmentation index is associated with cardiovascular risk. J Hypertens. 2002;20:2407–2414. doi: 10.1097/00004872-200212000-00020. [DOI] [PubMed] [Google Scholar]
- 11.Tsiachris D, Tsioufis C, Syrseloudis D, et al. Subendocardial viability ratio as an index of impaired coronary flow reserve in hypertensives without significant coronary artery stenoses. J Hum Hypertens. 2012;26:64–70. doi: 10.1038/jhh.2010.127. [DOI] [PubMed] [Google Scholar]
- 12.Harjola VP, Mebazaa A, Čelutkienė J, et al. Contemporary management of acute right ventricular failure: a statement from the Heart Failure Association and the Working Group on Pulmonary Circulation and Right Ventricular Function of the European Society of Cardiology. Eur J Heart Fail. 2016;18:226–241. doi: 10.1002/ejhf.478. [DOI] [PubMed] [Google Scholar]
- 13.Ren W, Guo JJ, Yang F, et al. Indication of the prognosis of pulmonary hypertension by using CMR function parameters. Eur Radiol. 2021;31:7121–7131. doi: 10.1007/s00330-021-07835-8. [DOI] [PubMed] [Google Scholar]
- 14.Thompson B. Exploratory and confirmatory factor analysis: understanding concepts and applications. Washington, DC: American Psychological Association; 2004. [Google Scholar]
- 15.Tabachnick BG, Fidell LS. Using Multivariate Statistics. 3rd ed. New York: Harper Collins; 1996. [Google Scholar]
- 16.Blyth KG, Groenning BA, Mark PB, et al. NT-proBNP can be used to detect right ventricular systolic dysfunction in pulmonary hypertension. Eur Respir J. 2007;29:737–744. doi: 10.1183/09031936.00095606. [DOI] [PubMed] [Google Scholar]
- 17.Biering-Sorensen T, Querejeta Roca G, Hegde SM, et al. Left ventricular ejection time is an independent predictor of incident heart failure in a community-based cohort. Eur J Heart Fail. 2018;20:1106–1114. doi: 10.1002/ejhf.928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sztrymf B, Gunther S, Artaud-Macari E, et al. Left ventricular ejection time in acute heart failure complicating precapillary pulmonary hypertension. Chest. 2013;144:1512–1520. doi: 10.1378/chest.12-2659. [DOI] [PubMed] [Google Scholar]
- 19.Alhakak AS, Sengeløv M, Jørgensen PG, et al. Left ventricular systolic ejection time is an independent predictor of all-cause mortality in heart failure with reduced ejection fraction. Eur J Heart Fail. 2021;23:240–249. doi: 10.1002/ejhf.2022. [DOI] [PubMed] [Google Scholar]
- 20.Kolev N, Zimpfer M. Left ventricular ejection time and end-systolic pressure revisited. Anesth Analg. 1995;81:889–890. doi: 10.1097/00000539-199510000-00055. [DOI] [PubMed] [Google Scholar]
- 21.Singer M, Allen MJ, Webb AR, et al. Effects of alterations in left ventricular filling, contractility, and systemic vascular resistance on the ascending aortic blood velocity waveform of normal subjects. Crit Care Med. 1991;19:1138–1145. doi: 10.1097/00003246-199109000-00008. [DOI] [PubMed] [Google Scholar]
- 22.Naeije R, Manes A. The right ventricle in pulmonary arterial hypertension. Eur Respir Rev. 2014;23:476–487. doi: 10.1183/09059180.00007414. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gorter TM, Willems TP, van Melle JP. Ventricular interdependence in pulmonary arterial hypertension: providing small pieces of a complex puzzle. Eur J Heart Fail. 2015;17:1–2. doi: 10.1002/ejhf.195. [DOI] [PubMed] [Google Scholar]
- 24.Gan C, Lankhaar JW, Marcus JT, et al. Impaired left ventricular filling due to right-to-left ventricular interaction in patients with pulmonary arterial hypertension. Am J Physiol Heart Circ Physiol. 2006;290:H1528–H1533. doi: 10.1152/ajpheart.01031.2005. [DOI] [PubMed] [Google Scholar]
- 25.Segers VF, Brutsaert DL, De Keulenaer GW. Pulmonary hypertension and right heart failure in heart failure with preserved left ventricular ejection fraction: pathophysiology and natural history. Curr Opin Cardiol. 2012;27:273–280. doi: 10.1097/HCO.0b013e3283512035. [DOI] [PubMed] [Google Scholar]
- 26.Sung SH, Yu WC, Cheng HM, et al. Pulsatile hemodynamics and clinical outcomes in acute heart failure. Am J of Hypertens. 2011;24:775–782. doi: 10.1038/ajh.2011.26. [DOI] [PubMed] [Google Scholar]
- 27.Sung SH, Yu WC, Cheng HM, et al. Excessive wave reflections on admission predict post-discharge events in patients hospitalized due to acute heart failure. Eur J Heart Fail. 2012;14:1348–1355. doi: 10.1093/eurjhf/hfs124. [DOI] [PubMed] [Google Scholar]
- 28.Feola M, Testa M, Ferreri C, et al. The analysis of arterial stiffness in heart failure patients in comparison with healthy subjects and patients with cardiovascular risk factors. J Clin Med. 2019;8:1721. doi: 10.3390/jcm8101721. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Gan CT, McCann GP, Marcus JT, et al. NT-proBNP reflects right ventricular structure and function in pulmonary hypertension. Eur Respir J. 2006;28:1190–1194. doi: 10.1183/09031936.00016006. [DOI] [PubMed] [Google Scholar]
- 30.Souza R, Jardim C, Julio Cesar Fernandes C, et al. NT-proBNP as a tool to stratify disease severity in pulmonary arterial hypertension. Respir Med. 2007;101:69–75. doi: 10.1016/j.rmed.2006.04.014. [DOI] [PubMed] [Google Scholar]
- 31.Rudski LG, Lai WW, Afilalo J, et al. Guidelines for the echocardiographic assessment of the right heart in adults: a report from the American Society of Echocardiography endorsed by the European Association of Echocardiography, a registered branch of the European Society of Cardiology, and the Canadian Society of Echocardiography. J Am Soc Echocardiogr. 2010;23:685–713. doi: 10.1016/j.echo.2010.05.010. [DOI] [PubMed] [Google Scholar]
- 32.Abbas AE, Fortuin FD, Schiller NB, et al. A simple method for noninvasive estimation of pulmonary vascular resistance. J Am Coll Cardiol. 2003;41:1021–1027. doi: 10.1016/s0735-1097(02)02973-x. [DOI] [PubMed] [Google Scholar]