Objective To optimise an echocardiographic estimation of pulmonary vascular resistance (PVRe) for diagnosis and follow-up of pulmonary hypertension (PHT).
Design Cross-sectional study.
Setting Tertiary referral centre.
Patients Patients undergoing right heart catheterisation and echocardiography for assessment of suspected PHT.
Methods PVRe ([tricuspid regurgitation velocity ×10/(right ventricular outflow tract velocity-time integral+0.16) and invasive PVRi ((mean pulmonary artery systolic pressure-wedge pressure)/cardiac output) were compared in 72 patients. Other echo data included right ventricular systolic pressure (RVSP), estimated right atrial pressure, and E/e' ratio. Difference between PVRe and PVRi at various levels of PVR was sought using Bland–Altman analysis. Corrected PVRc ((RVSP−E/e')/RVOTVTI) (RVOT, RV outflow time; VTI, velocity time integral) was developed in the training group and tested in a separate validation group of 42 patients with established PHT.
Results PVRe>2.0 had high sensitivity (93%) and specificity (91%) for recognition of PVRi>2.0, and PVRc provided similar sensitivities and specificities. PVRe and PVRi correlated well (r=0.77, p<0.01), but PVRe underestimated marked elevation of PVRi—a trend avoided by PVRc. PVRc and PVRe were tested against PVRi in a separate validation group (n=42). The mean difference between PVRe and PVRi exceeded that between PVRc and PVRi (2.8±2.7 vs 0.8±3.0 Wood units; p<0.001). A drop in PVRi by at least one SD occurred in 10 patients over 6 months; this was detected in one patient by PVRe and eight patients by PVRc (p=0.002).
Conclusion PVRe distinguishes normal from abnormal PVRi but underestimates high PVRi. PVRc identifies the severity of PHT and may be used to assess treatment response.
- Pulmonary vascular resistance
- pulmonary artery pressure
- pulmonary hypertension
- pulmonary arterial hypertension (PAH)
Statistics from Altmetric.com
- Pulmonary vascular resistance
- pulmonary artery pressure
- pulmonary hypertension
- pulmonary arterial hypertension (PAH)
Pulmonary vascular resistance (PVR) is an important haemodynamic measure in patients with pulmonary hypertension (PHT), as it provides vital prognostic information and is crucial in assessing response to available treatments.1–7 The American College of Cardiology Foundation/American Heart Association (ACCF/AHA) expert consensus document on PHT recommends PVR as a more robust diagnostic criterion for PHT because it reflects the influence of transpulmonary gradient and cardiac output and is only elevated if the vascular obstruction occurs within the precapillary pulmonary circulation. The consensus proposed that it should be routinely applied to patients with raised mean pulmonary arterial pressure.8 Right heart catheterisation has been the traditional means of obtaining this important information, but this carries the risks of invasive assessment (PVRi). Even though the risk of invasive testing with modern techniques is low, it is not negligible, with one study reporting an adverse event rate of 1.1% and mortality of the order of 0.05%.9 A simple non-invasive method for assessing PVR using Doppler echocardiography (PVRe) is based on the Doppler-based PVR equation: TRV×10/TVIRVOT (where TRV is tricuspid regurgitation velocity and TVIRVOT is right ventricular (RV) outflow time–velocity integral).10 However, this does not account for left ventricular (LV) filling pressure and was validated against invasive measures in only 44 patients, many of whom had normal or mildly elevated PVR.
The wider use of pulmonary vasodilator therapy5 7 11 has brought about the need to identify PHT and also to quantify its degree of elevation and track it over time. In this study, we sought to examine (1) the reliability of PVRe for the diagnosis of PHT, including its ability to predict high levels of PVR; (2) the merits of correcting the PVR for LV filling pressure (PVRc) and (3) the reliability of sequential invasive- and echo-PVR in patients with PHT undergoing serial right heart catheterisation.
Patients and methods
The initial study group comprised 72 patients with moderate or less tricuspid regurgitation who had an echocardiogram within 24 h of right heart catheterisation. In all cases right heart catheterisation was performed for suspected PHT in patients with dyspnoea—the largest groups being patients with scleroderma and patients undergoing investigation before a liver transplant. A subsequent group of 42 patients was used to validate a new prediction equation based on the initial group. Medical records of each patient were reviewed for clinical and demographic details (table 1).
Right heart catheterisation was performed using a standard protocol with a Swan–Ganz catheter. Cardiac output was measured using a thermodilution method as a mean of three readings with <10% variability. The following pressures were obtained: right atrial (RA), pulmonary artery systolic pressure (PASP), mean pulmonary artery pressure (MPAP) and pulmonary capillary wedge pressure (PCWP). Invasive PVR was calculated by the equation: PVRi = MPAP−PCWP/cardiac output.
Studies were performed using standard commercially available equipment (ie, 33, Philips Medical Systems, Andover, Massachusetts, USA; Vivid 7, General Electric Healthcare, Milwaukee, Wisconsin, USA; Sequoia, Siemens, Erlangen, Germany). Comprehensive echocardiographic studies were performed by a sonographer who was blinded to the results of cardiac catheterisation. TVIRVOT was obtained with the sample volume in the outflow tract in the parasternal short-axis view, optimising the signal to obtain the closing but not the opening valve click of the pulmonary valve. TRV was measured from the highest peak velocity gathered with continuous-wave Doppler through the tricuspid valve, using multiple views. Inadequate TR signals were enhanced with agitated saline contrast.12 13 RV systolic pressure (RVSP) was calculated using the modified Bernoulli equation from the TRV, plus the RA pressure, estimated by inspection of the inferior vena cava.14 RV function was assessed using tissue Doppler S', with the sample volume placed at the lateral tricuspid annulus.15 Other variables measured in our echo studies included left ventricular end-diastolic and systolic diameter, left ventricular ejection fraction, left atrial volume index (measured by biplane method and indexed to body surface area) and deceleration time of early (E) filling wave obtained from a transmitral flow profile (table 1).
An index of LV filling pressure as a surrogate for invasive PCWP was measured as the ratio of early (E) filling wave to early diastolic myocardial velocity measured at the medial mitral annulus using pulsed-wave tissue Doppler imaging.16 PVRc was derived as (RVSP−E/e')/TVIRVOT, where (RVSP−E/e') and RVOT VTI are surrogates for transpulmonary gradients and cardiac output, respectively. This was developed in a training group and tested in a separate validation group primarily composed of patients with pulmonary arterial hypertension (table 1).
Echocardiograms were stored in digital format and Doppler measurements were made offline.
Conventional measures (sensitivity and specificity, using a 2×2 contingency table) were used to describe the diagnostic accuracy of PVRe relative to PVRi, using 2.0 Wood units as the definition of normal PVRi. The correlation of PVRi and PVRe was expressed by Pearson's correlation coefficient. Regression models were constructed for PVRi and PVRe. Variables included RVSP, RV function measured as S' and filling pressures measured as E/e'. Univariate analysis was performed to assess the association of these factors with underestimation of PVRi. Differences between the two modalities were also expressed using Bland–Altman analysis. In 10 patients with follow-up right heart catheterisation, changes after the first 6 months of treatment were compared between two modalities (standard PVRe vs PVRi).
Accuracy of PVRe
To confirm the reliability of echocardiographic estimation of PHT, patients were divided into two groups based on PVRi: patients with elevated PVR and those with normal PVR. Elevated PVRe (>2.0 Wood units) had very high sensitivity (93%) in 50 patients with elevated PVR by cardiac catheterisation (PVRi>2.0 Wood units). In 22 patients with normal PVRi (PVR <2.0 Wood units), PVRe had good specificity (91%). There was a positive correlation between PVRe and PVRi (r=0.77, p<0.01), but echo underestimated PVRi when this was markedly elevated (figure 1A). On Bland–Altman analysis, the agreement between two modalities was greater at normal and mildly elevated PVR and there was systematic underestimation of PVRi by PVRe in patients with moderate or more elevation of PVRi (figure 1B). Underestimation of PVRi by PVRe was independently associated with RV function measured as RVs' (β=−0.38, 95% CI −0.08 to −0.56, p=0.011) and the difference between echo- and invasive-PASP (β=−0.30, 95% CI −0.14 to 0.17, p=0.02), but was unrelated to gender, left ventricular ejection fraction and filling pressure.
Contribution of estimated LV filling pressure
PVRc (derived from RVSP, E/e' and TVIRVOT) had a sensitivity (91%) and specificity (90%) comparable to PVRe. However, PVRc had improved correlation even at markedly elevated levels of invasive PVR (figure 2A,B). The mean difference between PVRi and PVRc was estimated to be 2.8±2.7 Wood units (p<0.001). This mean difference between two modalities dropped to 0.8±3.0 Wood units by using PVRc. Figure 3 emphasises the difference between PVRe and PVRc at high levels of PVR in separate definition and validation groups. Correlation between PVRc and PVRi remained more robust than the one between PVRe and PVRi among the two subgroups (PCWP ≤15 or PCWP>15 mm Hg) of patients with PHT (MPAP >25 mm Hg) (r=0.81 vs r=0.69 and r=0.88 vs r=0.68 respectively) (figure 4).
Use in follow-up studies
In 10 patients with follow-up studies, the change between before treatment and 6 months after treatment was significantly different between two modalities when PVRe rather than PVRc was used to estimate non-invasive PVR (−3.9±4.4 vs +0.18±0.80, p=0.03). In follow-up, PVRc improved upon underestimation of PVRi by traditional PVRe and also was able to track changes in invasive PVRi during 12 months' follow-up treatment (figure 5). PVRe accurately detected a drop in PVR by at least one SD in only one out of 10 patients whose PVRi fell after first 6 months, while PVRc accurately predicted a drop in PVRi in eight out of 10 patients (p=0.002).
PVR is a flow-independent parameter used for diagnosis and surveillance in patients with PHT. Echocardiography has an important role in the non-invasive assessment of patients with PHT, but its role in assessment of PVR is less well defined. PVR is traditionally measured invasively as the ratio of transpulmonary gradient and cardiac output.10 Estimation of PVR using Doppler echocardiography is based on tricuspid velocity and RVOT velocity time integral, as surrogates for transpulmonary gradient and cardiac output, respectively.5 Our study confirms the value of PVRe as a stroke-volume corrected screening tool for PHT. However, in 10 patients undergoing sequential right heart catheterisation to assess response to vasodilator treatment, traditional PVRe accurately predicted a favourable response in only one out of 10 patients at 6 months after inclusion into the study. This observation reflects the limitations of standard PVRe as a standalone technique to follow response to treatment in patients with PHT. A more reliable non-invasive methodology is required to reduce the requirement of invasive procedures in this often debilitated patient population.
Non-invasive evaluation of PVR
Several other methods have been postulated to estimate PVR non-invasively. These methods use multiple variables such as calculating pulmonary artery diastolic pressure by superimposing spectral Doppler from the pulmonary valve onto the tricuspid regurgitation signal,17 or using RVOT pre-ejection and ejection time.18 These methods depend on multiple variables which increase the potential for error and reduce the robustness of their data, whereas the method validated by Abbas et al10 is simpler. Vlahos et al19 found a linear correlation between invasive PVR and TRV/RVOTVTI in 12 patients (including six children) with extreme values of invasive PVR. In a study of 52 patients from a PHT clinic, Rajagopalan et al,20 showed a poor correlation between TRV/RVOTVTI when PVR was >8 Wood units. However, their study cohort was limited to patients with established PHT, and there was a lack of follow-up data. Even though there are no specific guidelines for following up these patients, it is recommended that they should be reviewed regularly. The ACCF/AHA expert consensus document considers that echo-derived PASP is only an estimate and strongly advises against using this as the sole parameter in making a treatment decision.8 Not infrequently patients are subjected to invasive testing during follow-up to obtain important haemodynamic data including PVR. None of the non-invasive techniques of PVR estimation have been tested in follow-up of patients with PHT undergoing treatment.
RV dysfunction and remodelling secondary to persistently elevated PVR is likely to influence both TRV and LV filling. The effect on TRV would explain the relationship between RV dysfunction and underestimation of PVR as noted in our study. This is further supported by poor correlation (r=0.49, p<0.01) between MPAP and TRV in subgroup of patients with markedly elevated PVR (PVRi >5 Wood units). Structural changes in the right ventricle and RVOT in patients with PHT may also have unpredictable affect on the ability to use RVOTVTI as a surrogate for cardiac output. The PVRe equation also does not take into account the negative effect of markedly elevated RA pressures on the ability of TRV to act as a surrogate for transpulmonary gradient. The combination of these factors probably accounts for the underestimation of PVRe in patients with severe PHT and adversely affects the ability to track PVRi in follow-up. This limitation of PVRe was highlighted in our study as PVRe accurately detected a drop in PVR by at least one SD in only one out of 10 patients whose PVRi fell after the first 6 months.
Role of LV diastolic pressure
The effect of RV changes on LV filling is more controversial. Delayed relaxation is commonly seen, and may in part relate to underfilling of the left ventricle and changes in LV geometry.21 However, there is also evidence of increased LV filling pressures due to PHT and RV hypertrophy.22 In the new PVR equation PVRc = (RVSP−E/e')/RVOTVTI, RVSP, E/e', RVOTVTI were assumed to be surrogates for MPAP, PCWP23 and cardiac output, respectively. Contrary to the standard PVRe equation which uses only TRV as surrogate for transpulmonary gradient, PVRc uses RVSP, which accounts for RA pressure. We propose that difference between RVSP and E/e'(RVSP−E/e') is a more robust surrogate of transpulmonary gradient.
The measurement of E/e' provides an imperfect approximation of PCWP. Nonetheless, the incorporation of a marker of LV filling pressure may be an important contributor to PVR estimation when septal displacement due to RV enlargement compromises LV filling in severe PHT. This new equation reduces the non-invasive underestimation of PVR and may be useful in following up patients with elevated PVR. Furthermore, the use of this new equation during exercise echocardiography may assist in differentiating exercise-induced pulmonary arterial hypertension from PHT secondary to left heart disease in a patient being investigated for exertional dyspnoea. This could be advantageous in screening patients for exercise-induced PHT as exercise right heart catheterisation is technically demanding.
We limited this study to patients with moderate or less tricuspid regurgitation, as severe tricuspid regurgitation is associated with RA pressure elevation (which may lead to underestimation of pulmonary artery pressure) and the thermodilution technique used to measure cardiac output in our study is considered less accurate in patients with severe tricuspid regurgitation.24 25 However, this avoidance of severe tricuspid regurgitation may compromise the applicability of this approach. Similarly, the study did not include patients with significant mitral valve disease or mitral valve prostheses, in whom E/e' is problematic. The new equation may not apply to these groups of patients. We only used early diastolic myocardial velocity measured at the medial mitral annulus using pulsed-wave tissue Doppler imaging which could be influenced by impaired right ventricular function in patient with severe PHT. Lateral mitral annulus tissue imaging can be used deal with this problem but was not used in our study.
In patients with suspected PHT, either PVRc or PVRe may be used as the first line of investigation because of their high sensitivity and specificity to detect abnormal PVR. Although traditional PVRe is effective in distinguishing normal from abnormal PVR it underestimates high levels of PVR.
The authors express their gratitude to David J Holland BS and Harry Gibbs MD, for assistance with data collection and analysis.
Competing interests None.
Ethics approval This study was conducted with the approval of the Princess Alexandra Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.