Transthoracic Dopplerechocardiography (TTE) can estimate mean pulmonary arterial pressure (MPAP) and pulmonary capillary wedge pressure (PCWP) reliably, and cardiac magnetic resonance (CMR) is the best modality for noninvasive measurement of cardiac output (CO). We speculated that the combined use of TTE and CMR could provide a feasible method for noninvasive measurement of pulmonary vascular resistance (PVR) in pulmonary arterial hypertension (PAH).
Right heart catheterization (RHC) was undertaken in 77 patients (17M/60F) with PAH, and simultaneous TTE was carried out to evaluate MPAP, PCWP and CO. Within 2 days, CO was measured again with CMR in similar physiological status. Then, PVR was calculated with the integrated noninvasive method: TTEderived (MPAP–PCWP)/CMRderived CO and the isolated TTE method: TTEderived (MPAP–PCWP)/TTEderived CO, respectively. The PVR calculated with integrated noninvasive method correlated well with RHCcalculated PVR (r = 0.931, 95% confidence interval 0.893 to 0.956). Between the integrated noninvasive PVR and RHCcalculated PVR, the Bland–Altman analysis showed the satisfactory limits of agreement (mean value: − 0.89 ± 2.59). In comparison, the limits of agreement were less satisfactory between TTEcalculated PVR and RHCcalculated PVR (mean value: − 1.80 ± 3.33). Furthermore, there were excellent intra and interobserver correlations for the measurements of TTE and CMR (P < 0.001 for all).
The combined use of TTE and CMR provides a clinically reliable method to determine PVR noninvasively. In comparison with RHC, the integrated method shows good accuracy and repeatability, which suggests the potential for the evaluation and serial followup in patients with PAH.
In PAH, the noninvasive measurement of PVR is very important in clinical practice. Up to now, however, the widely accepted noninvasive method is still unavailable. Since TTE can estimate (MPAP–PCWP) reliably and CMR is the best image modality for the measurement of CO, the combined use of two modalities has the potential to determine PVR noninvasively. In this research, the integrated noninvasive method showed good diagnostic accuracy and repeatability compared with RHC. Therefore, it might be a feasible method for noninvasive measurement of PVR in patients with PAH.
Pulmonary arterial hypertension;Pulmonary vascular resistance;Doppler echocardiography;Cardiac magnetic resonance
The measurement of pulmonary vascular resistance (PVR) is crucial in the diagnosis and management of pulmonary arterial hypertension (PAH). Right heart catheterization (RHC) is widely considered the “gold standard” to determine PVR, however, it is limited by the associated disadvantages [1]. Therefore, it is necessary to develop a feasible method for the noninvasive measurement of PVR. Recent reports have investigated the applicability of transthoracic Dopplerechocardiography (TTE) or cardiac magnetic resonance (CMR), and some novel methods have been proposed [2]; [3]; [4]; [5]; [6]; [7]; [8]; [9]; [10]; [11]; [12] ; [13]. However, the intrinsic limitations of each image modality compromised the accuracy and reproducibility in these methods. Up to now, with the isolated use of TTE or CMR, there was no widely accepted method for noninvasive evaluation of PVR.
Previous studies have demonstrated that TTE and CMR had complementary advantages. TTE had the potential to estimate mean pulmonary arterial pressure (MPAP) and pulmonary capillary wedge pressure (PCWP) reliably [14]; [15]; [16]; [17]; [18]; [19]; [20]; [21]; [22] ; [23]. In the noninvasive measurement of cardiac output (CO), however, CMR had more accuracy and reproducibility than TTE [24]; [25] ; [26]. Based on the formula (MPAP–PCWP)/CO, it became possible to calculate PVR directly with TTEderived (MPAP–PCWP) and CMRderived CO. Therefore, we hypothesized that the combined use of TTE and CMR might have the potential to estimate PVR noninvasively. This study was performed to determine whether the integrated modality formed a reliable noninvasive method to measure PVR in PAH.
From January 2009 to July 2014, a total of 77 patients (age 32.42 ± 11.01 years, 17M/60F) with PAH were enrolled. Among these patients, there were 70 patients with idiopathic PAH and 7 patients with postoperative persistent PAH after surgical closure of ventricular septal defect (n = 4), patent ductus arteriosus (n = 2) and atrial septal defect (n = 1). All patients were in sinus rhythm. Patients (n = 16) with unstable clinical condition, inadequate image quality, arrhythmia (such as atrial fibrillation, frequent premature beats and so on), significant mitral or aortic regurgitation, variations of heart rate and blood pressure ≥ 10% (between RHC/TTE and CMR), and contraindications of CMR examination were all excluded from the study. For the patients aged ≥ 45 years (n = 11), coronary artery disease was excluded with selective coronary angiography (defined as ≥ 50% reduction in lumen diameter). The study was approved by our hospital research ethics committee, and informed consent was obtained from each patient.
In this study, PAH was defined as MPAP > 25 mm Hg and PCWP ≤ 15 mm Hg at rest (RHC) [27]. During RHC, simultaneous TTE examination was performed in patients with PAH. The invasive hemodynamic data, TTEderived (MPAP–PCWP) and TTEderived CO were obtained at rest. Within 2 days after RHC/TTE, CMR was performed without intervening therapy and the resting CO was calculated noninvasively in similar physiological status (variations of heart rate and blood pressure < 10%). According to the formula TTEderived (MPAP–PCWP)/CMRderived CO and TTEderived (MPAP–PCWP)/TTEderived CO, the noninvasive PVR was calculated, respectively. Furthermore, the results were compared with RHCcalculated PVR using the Bland–Altman analysis. Individuals in whom RHC, TTE and CMR variables were obtained were blinded to each others calculations. Furthermore, thirtyeight patients were randomly selected to validate the intraobserver and interobserver variability, respectively. For intraobserver reproducibility, the examination was repeated twice by the same observer in a consecutive manner. For interobserver reproducibility, CMR was validated by two independent observers based on the same imaging result (without any communication). And for TEE, two independent observers conducted the examination in turn also without communication.
Our procedure for cardiac catheterization has been described previously [28]. In brief, all patients underwent routine RHC (SwanGanz, Edwards Lifesciences) under local anesthesia. The complete hemodynamic data and blood samples were obtained at rest. The measurements included mean right atrial pressure, right ventricular pressure, systolic pulmonary arterial pressure (SPAP), diastolic pulmonary arterial pressure (DPAP), MPAP and PCWP. According to the oxymetric principle of Fick, CO and PVR [PVR = (MPAP–PCWP) / CO] were calculated.
In patients undergoing RHC, simultaneous TTE (Philips IE33, instrument equipped with a 3–5 MHz transducer) was performed. TTEderived MPAP was calculated as TTEderived SPAP × 0.61 + 2 mm Hg, according to Chemla et al. [16]. SPAP was estimated by TTE from the systolic right ventriculartoright atrial pressure gradient using the modified Bernoulli equation, and the assessment of right atrial pressure was performed in accordance with previously described methods. Furthermore, isovolumic relaxation time (IVRT) and color Mmode Doppler flow propagation velocity (FPV) were measured. According to the equation: 4.5 (10^{3} / [2 · VRT] + FPV) − 9, PCWP was estimated noninvasively [18]. Based on TTEderived MPAP and PCWP, the transpulmonary pressure gradient (MPAP–PCWP) was calculated noninvasively. According to the recommendations in guideline, the measurements of stroke volume (SV) and CO (SV × heart rate) were made at the level of the left ventricular (LV) outflow tract [29].
All examinations were performed on a 1.5 Tesla MR scanner (Magnetom Avanto, Siemens Medical Solutions, Erlangen, Germany) at rest, and the protocol has been described previously [30] ; [31]. In brief, scout images acquired by using halfFourier acquisition singleshot turbo spinecho (HASTE) sequence were used to analyze the morphology and structure of the heart. Retrospective electrocardiographicgating cine images were acquired using a true fast imaging with steadystate precession (TrueFisp) sequence. For the volumetric and functional measurements, contiguous shortaxis images through the entire ventricle (from base to apex, no gap) were obtained. The following parameters were used: FOV, 250 mm × 188 mm; slice thickness, 8 mm; matrix, 156 mm × 192 mm; TR, 2.8 ms; TE, 1.39 ms; flip angle, 70°; number of signal averages, 2–4. LV enddiastolic and endsystolic volumes, LV ejection fraction (LVEF), SV, and CO (SV × heart rate) were calculated. In addition, right ventricular (RV) enddiastolic and endsystolic volumes, RVEF were also calculated (the endocardial borders of all shortaxis images at enddiastole and endsystole were manually traced with the inclusion of RV outflow to the pulmonary valve and the trabeculae in the RV volume).
Categorical variables were presented as counts with percentages and compared by Chisquare test or Fishers exact test. Continuous variables were presented as mean ± SD or median with IQR and compared by grouped ttest or Wilcoxon rank sum test. Pearson or spearman correlation coefficients were calculated first and linear regression models were constructed. Furthermore, Bland–Altman analysis was carried out for agreement assessments, the lower and upper limits of agreement were estimated, as the mean ± 2SDs with 95% confidence interval (CI). Pearsons or Spearmans correlation and Bland–Altman analysis were also used to assess the intraobserver and interobserver reproducibility. A twosided P < 0.05 was considered statistically significant. Statistical software used in this study was SPSS 16.0 and MedCalc 9.5.
Clinical and demographic characteristics of the patients were listed in Table 1. The TTEderived (MPAP–PCWP) and RHCderived (MPAP–PCWP) were comparable (55.06 ± 17.67 mm Hg vs 58.45 ± 18.76 mm Hg, P = 0.25). The linear regression analysis revealed a good correlation (r = 0.934, 95% CI: 0.897–0.957) for all patients (Supplemental Fig. 1). Using the Bland–Altman analysis (Supplemental Fig. 2), TTEderived (MPAP–PCWP) showed satisfactory limits of agreement with RHCderived (MPAP–PCWP). The mean value was 3.39 ± 6.73 (SD), and TTEderived (MPAP–PCWP) and RHCderived (MPAP–PCWP) values were well within one standard deviation (17.67 and 18.76, respectively).
Sex, M/F  17/60 
Age, years  32.42 ± 11.01 
BSA, m^{2}  1.48 ± 0.14 
Diagnosis, n  77 
Idiopathic PAH, n (%)  70 (90.9%) 
Postoperative persistent PAH, n (%)  7 (9.1%) 
RHC  
SPAP, mm Hg  103.64 ± 31.38 
DPAP, mm Hg  39.97 ± 16.31 
MPAP, mm Hg  68.00 ± 19.28 
PCWP, mm Hg  9.55 ± 1.96 
MPAP–PCWP, mm Hg  58.45 ± 18.76 
CO, L/min  4.83 ± 1.91 
Heart rate, bpm  79.23 ± 10.05 
Invasive PVR, Wood  13.86 ± 7.05 
TTE  
MPAP, mm Hg  64.56 ± 18.05 
PCWP, mm Hg  9.49 ± 1.99 
MPAP–PCWP, mm Hg  55.06 ± 17.67 
CO, L/min  3.76 ± 1.45 
TTEcalculated PVR, Wood  16.55 ± 8.27 
Mild tricuspid regurgitation, n (%)  42(54.5%) 
Moderate tricuspid regurgitation, n (%)  31(40.3%) 
Severe tricuspid regurgitation, n (%)  4(5.2%) 
CMR  
LVEDV, ml  88.90 ± 50.78 
LVESV, ml  40.65 ± 32.09 
LVEF, %  55.92 ± 9.91 
Heart rate, bpm  77.97 ± 11.64 
CO, L/min  4.18 ± 1.72 
RVEF, %  25.15 ± 9.6 
Integrated noninvasive PVR, Wood  14.75 ± 6.83 
Notes: BSA, body surface area; Invasive PVR, PVR calculated with RHC; Integrated noninvasive PVR, PVR calculated with TTEderived (MPAP–PCWP)/CMRderived CO.
The CO from CMR and RHC were 4.18 ± 1.72 L/min and 4.83 ± 1.91 L/min, respectively. The linear regression analysis also showed a good correlation (r = 0.805, 95% CI: 0.709–0.872) (Supplemental Fig. 3). Using the Bland–Altman analysis (Supplemental Fig. 4), CMRderived CO demonstrated the good limits of agreement with RHCderived CO. The mean value was − 0.65 ± 0.85 (SD), and CMRderived CO and RHCderived CO values were well within one standard deviation (1.73 and 1.92, respectively).
The heart rates were comparable between RHC/TTE and CMR (79.23 ± 10.05 bmp vs 77.97 ± 11.64 bmp, P = 0.474), and variation of heart rate was 1.26 ± 4.61. The integrated noninvasive PVR was 14.75 ± 6.83 Woods and invasive PVR was 13.86 ± 7.05 Woods. The good correlation (r = 0.931, 95% CI: 0.893–0.956) was confirmed with linear regression analysis (Fig. 1A). The Bland–Altman analysis showed the satisfactory limits of agreement between integrated noninvasive PVR and invasive PVR (Fig. 1B). The mean value was − 0.89 ± 2.59 (SD), and integrated noninvasive PVR and invasive PVR values were well within one standard deviation (6.84 and 7.06, respectively). While between TTEcalculated PVR and invasive PVR, linear regression analysis (Supplemental Fig. 5) showed the good correlation (r = 0.912, 95% CI: 0.865–0.943), and the Bland–Altman analysis (Supplemental Fig. 6) showed the less satisfactory limits of agreement (mean value: − 1.80 ± 3.33).

Fig. 1. A, Scatter diagram and regression line between integrated noninvasive PVR and invasive PVR (r = 0.931, 95% CI: 0.893–0.956) in all patients. B, Bland–Altman plot of the difference between integrated noninvasive PVR and invasive PVR against the mean of the noninvasive PVR and invasive PVR.

Furthermore, the difference between TTEcalculated PVR and invasive PVR was significantly larger than the difference between integrated noninvasive PVR and invasive PVR (Fig. 2, t = − 4.78, P < 0.001 by the paired ttest analysis). The mean value of the difference between them was − 1.79 ± 3.33 (SD). Based on the formula ABS ((X PVR  invasive PVR)/invasive PVR) for measuring the relative error, sd1 and sd2 were produced by noninvasive PVR and TTEcalculated PVR, respectively. And they all were divided into 4 categories by 0 to 0.1, 0.1 to 0.3, 0.3 to 0.5, and 0.5 to 1 (Table 2), respectively. Statistically significant difference was found between sd1 and sd2 for these categories by McNemar test (χ^{2} = − 19.371, P = 0.004).

Fig. 2. Scatter of the difference between integrated noninvasive PVR and invasive PVR and the difference between TTE–PVR and invasive PVR against the invasive PVR. The latter was significantly larger than the former (t = − 4.78, P < 0.001) by the paired ttest analysis.

Difference comparing to RHC (integrated noninvasive)  Difference comparing to RHC (TTE only)  Total  

0% to 10%  10% to 30%  30% to 50%  50% to 100%  
0% to 10%  8(10.39)  16(20.78)  3(3.90)  1(1.30)  28(36.36) 
10% to 30%  5(6.49)  15(19.48)  12(15.58)  4(5.19)  36(46.75) 
30% to 50%  0(0.00)  2(2.60)  3(3.90)  4(5.19)  9(11.69) 
50% to 100%  0(0.00)  2(2.60)  1(1.30)  1(1.30)  4(5.19) 
Total  13(16.88)  35(45.45)  19(24.68)  10(12.99)  77(100.00) 
For TTEderived (MPAP–PCWP), the intraobserver and interobserver reproducibility was high, and the average differences were − 1.13 and − 1.87, respectively. The Pearson correlation coefficient was 0.977 and 0.955, respectively (P < 0.001 for both). For CMRderived CO, the intraobserver and interobserver variability were − 0.076 and − 0.186, respectively. The Pearson correlation coefficient was 0.995 and 0.993, respectively (P < 0.001 for both). Both the intraobserver (Supplemental Fig. 7) and interobserver (Supplemental Fig. 8) have the respectively satisfactory reproducibility for TTEderived (MPAP–PCWP) by using the Bland–Altman analysis, and the intraobserver (Supplemental Fig. 9) and interobserver (Supplemental Fig. 10) for CMRderived CO do the same as it.
In this research, PVR was calculated directly with the combined use of TTEderived (MPAP–PCWP) and CMRderived CO. In comparison to RHC, PVR from this integrated noninvasive method showed the good accuracy and reproducibility in patients with PAH. To our knowledge, it is the first report to determine PVR with the combined use of TTE and CMR, and our findings suggested that this integrated method was clinically feasible for noninvasive calculation of PVR in PAH.
In patients with PAH, PVR is an important prognostic factor and is frequently repeated for serial followup. Though RHC is the “gold standard” for the evaluation of PVR, its invasive nature makes it unsuitable for frequent and repeated use [1] ; [27]. In addition, the cost and ionizing radiation also become the important disadvantages of RHC. Therefore, it is urgent to develop the reliable and accurate methods for noninvasive estimation of PVR. In preliminary studies, the significant relationships have been identified between some parameters of TTE/CMR and PVR, and several composite numerical models have been established to evaluate PVR noninvasively [2]; [3]; [4]; [5]; [6]; [7]; [8]; [9]; [10]; [11]; [12] ; [13]. Limited by the intrinsic nature of each image modality, these methods were usually indirect, and the widely accepted method was still unavailable.
Based on the complementary advantages of TTE and CMR, a new method was proposed for the noninvasive measurement of PVR in this study. Since PVR is directly related to transpulmonary pressure gradient (MPAP–PCWP) and inversely related to CO, the noninvasive PVR can be achieved by noninvasive estimation of (MPAP–PCWP) and CO, respectively. It has been previously demonstrated that TTE estimated MPAP and PCWP reliably, but its ability to calculate CO was not as accurate and reproductive as CMR [25] ; [26]. In this study, our findings also proved the flaw of 2DTTE in the measurement of CO. On the contrary, CMR was the best noninvasive modality to measure CO but its ability to determine MPAP/PCWP was limited [13] ; [32]. Therefore, TTE and CMR can complement each other, and the combined use of two image modalities made it possible to determine PVR noninvasively. In this study, our findings confirmed the potential of this integrated method. Furthermore, the attempt was also made to estimate PVR directly with the isolated use of 2DTTE in this study. In comparison with RHC, however, TTEcalculated PVR was less satisfactory, which disagreed with recent research [33].
Compared with RHC, TTE can estimate transpulmonary pressure gradient (MPAP–PCWP) reliably in this study. In the estimation of MPAP, several studies have identified a strong relationship between TTE and RHC [14]; [15] ; [16]. The incremental value of TTE for noninvasive measurement of MPAP has been proven in routine clinical practice [34]. Several TTE methods have been proposed to estimate MPAP, and the simple Chemla' formula (MPAP = 0.61 × SPAP + 2) was adopted in this study [16]. Furthermore, it has previously been demonstrated that PCWP can be estimated reliably with different TTE methods [17]; [18]; [19]; [20]; [21]; [22] ; [23]. In this study, PCWP was determined noninvasively based on the combined use of pulsed and color Mmode Doppler echocardiography [18]. Therefore, it became possible to estimate transpulmonary pressure gradient (MPAP–PCWP) noninvasively with TTE, which was confirmed in this study.
In this research, CO was measured noninvasively with volumetric cine CMR and agreed closely with that from RHC in similar physiological status. There were two methods available in CMRcalculated CO (volumetric cine and velocityencoded cine), and they had excellent agreement [26]. CMR can overcome the limitations of TTE and provides the best noninvasive method for measurements of SV and CO. Previous reports have showed that CMRderived CO had good accuracy and reproducibility compared with RHCderived CO [24]; [25] ; [26]. In this study, our findings also showed the reasonable agreement between CMR and RHC, and further corroborated the previous studies.
In patients with PAH, both physiologic and imaging factors were contributors to variations in the measurement of PVR. To minimize these influences, the related factors were controlled strictly between RHC/TTE and CMR. In addition, PAH was severe and PVR was high in this study. Therefore, the patients were selected highly in this study. Given the difficulty in the estimation of PAP, the normal controls were absent in this research. CO was measured noninvasively with volumetric cine CMR in this study. Though there was excellent agreement between volumetric and velocityencoded cine CMR, the latter was used more frequently and might permit inclusion of more patients. In this research, 30% or greater difference was observed in more than 10% of cases, which added more insight to the influence if the integrated method was used in clinical practice. Compared with the benefit of noninvasive procedure, however, the diagnostic value of integrated method might be acceptable. Furthermore, TTEderived CO was based on conventional 2DTTE, not 3DTTE, in this research. Recent reports suggested that 3DTTE might be more accurate than 2DTTE in the measurement of SV [35] ; [36]. Therefore, further research was required to evaluate the possibility of noninvasive PVR based on 3DTTE.
In this study, our findings suggest that PVR can be calculated directly using TTEderived (MPAP–PCWP) and CMRderived CO. Furthermore, the results from this integrated noninvasive method showed good diagnostic accuracy and repeatability compared with RHC. Therefore, the combined use of TTE and CMR can provide a clinically feasible method for noninvasive measurement of PVR in patients with PAH.
This work was supported by the Peking Union Medical College Youth Fund (33320140038) and National Natural Science Foundation of China (81341045).
None.
Supplementary figures
Published on 19/05/17
Submitted on 19/05/17
Licence: Other
Are you one of the authors of this document?