Pulmonary arterial wave reflection can be assessed based on the concept of wave separation analysis. This analysis determines the origin, type and timing of traveling waves in a circulation from combined P and U measurements and allows wave separation into forward-traveling and backward-traveling components. Theoretical background and practical applications have been described elsewhere (Hollander et al., 2001; Yoshida et al., 2021). Briefly, wave speed (WS), which represents the elastic properties of the local artery, can be determined by P-U loop method (Khir et al., 2001); it takes advantage of the water hammer equation relating P and U on the condition that there is no wave reflection.

where dP and dU are the changes in P and U, ρ is the density of blood (1050 kg/m³) and c is WS. Pulse pressure can be separated into those attributed to forward-traveling (Pf) and backward-traveling (Pb) waves using equations 2 and 3.

The new noninvasive method we propose herein uses echo-Doppler derived P and U waveforms, instead of the direct measurements. A pulsed-wave Doppler tracing of RVOT flow is used as a surrogate for U waveform. On the other hand, P waveform is estimated by applying the simplified Bernoulli equation to an instantaneous Doppler velocities of TR jet and adding a term of right atrial pressure as below. This clinical equation has been derived from the more complex Bernoulli equation by assuming that viscous losses and acceleration effects are negligible and by using an approximation for the constant that relates to the mass density of blood, a conversion factor for measurement units (Baumgartner et al., 2009).

where it is time, TRV is TR velocity and RAP is right atrial pressure which we assume is constant throughout the cardiac cycle.

End diastolic P is determined as P-value at the beginning of ejection (shown as t0) identified from the U waveform.

Subtracting end-diastolic P from P waveform yields pulse pressure waveform; this subtraction eliminates the term of right atrial pressure.

Wave separation analysis is performed on the pulse pressure and flow velocity waveforms by synchronizing waveforms of TR flow and RVOT flow using electrocardiogram. The pulse pressure waveform will then be separated into Pf and Pb. This analysis yields 1 arterial stiffness and 3 wave reflection indices: WS, peak Pb, peak Pf, and reflection coefficient (RC) calculated as the ratio of peak Pb to peak Pf.

The present study was approved by the Animal Experimental Committee of Tokyo University of Agriculture and Technology (approval number: 30-146). All process of the study was carried out in compliance with the ARRIVE (Animal Research: Reporting of in vivo Experiments) guidelines and the regulations on animal experiments and the guide for the care and use of laboratory animals of Tokyo University of Agriculture and Technology (Percie du Sert et al., 2020). Eight healthy Beagle dogs (Kitayama Labes, Nagano, Japan) were used in the present study (all females, aged 4–5 years old, weighted 10–13 kg). All dogs were sedated with butorphanol (0.2 mg/kg), midazolam hydrochloride (0.2 mg/kg) and meloxicam (0.2 mg/kg) and initially anesthetized with propofol (4 mg/kg) before intubated and mechanically ventilated. Complete anesthesia was maintained by inhalation of isoflurane (end-tidal concentration 1.5 ± 0.1%). The dog was then placed in right lateral recumbency. Heparin sodium (100 IU/kg) was administered for preventing thrombosis.

A 4-Fr catheter (Atom nutrition catheter, Atom Medical, Tokyo, Japan) was inserted into the right femoral artery to monitor systemic arterial pressure. A 4.2-Fr multipurpose angiographic catheter (Goodtec angiographic catheter, GOODMAN, Aichi, Japan) was advanced retrograde into the left atrium through the left carotid artery to monitor left atrial pressure. Another 4.2-Fr multipurpose angiographic catheter was placed in the right atrium through the left jugular vein to monitor right atrial pressure. These fluid-filled catheters were connected to pressure transducers (Life kit DX-360, Nihon Kohden, Tokyo, Japan) and pressure waveforms were displayed using a multi-channel monitor (Life Scope BSM-5192; Nihon Kohden, Tokyo, Japan). These two catheters were also used to sample blood from the left and right atriums to calculate cardiac output.

To obtain P and U waveforms as a reference for validating the new echo-Doppler method, a dual sensor-tipped pressure and flow wire (Combowire, Royal Philips, Amsterdam, Netherlands) was advanced to approximately 1 cm beyond the pulmonary valve through another 4.2-Fr multipurpose angiographic catheter inserted from the left jugular vein. Careful catheter and wire manipulation ensured that signals were steadily obtained.

The study design is outlined in Figure 1. To assess whether the new echo-Doppler method can detect the alterations in pulmonary arterial wave reflection associated with the development of PH, each dog was given repeated injections of dextran microsphere cross-linked with epichlorohydrine (Sephadex G-50, GE healthcare, diameter 300 μm) until mean PAP increased above 30 mmHg. To investigate the influences of hemodynamic alterations on wave reflection, each dog received separate continuous drip infusions of lactated Ringer’s solution until mean right atrial pressure increased by 2 mmHg (fluid loading) and dobutamine (inotropic stimulation, 3 μg/min/kg), both at baseline and after the induction of PH. Every time a single intervention of either fluid or dobutamine challenge was finished, we waited until pulmonary arterial hemodynamics became back to the baseline before moving onto the next state-load condition. As a result, each stage took at least 30 min. At each state-load condition, the direct P and U measurements and echocardiographic image acquisitions as well as pulmonary hemodynamic assessment were conducted, as described later in this section.

Cardiac output was calculated by Fick method from oxygen content of blood sampled from the left and right atriums and oxygen consumption estimated using Sykes’ formula (Percie du Sert et al., 2020). Stroke volume was calculated as cardiac output / heart rate. PVR was calculated as (mean PAP - mean left atrial pressure) / cardiac output. Pulmonary artery compliance (PAC) was approximated as stroke volume / (systolic PAP – diastolic PAP).

P and U data were acquired simultaneously for 30 to 60 s during end-expiration apnea, using the dual sensor-tipped catheter. The sampling rate at the acquisition was 200 Hz (5-msec temporal resolution). The data were then exported digitally for subsequent wave separation analysis.

Comprehensive echocardiographic examination was performed, using a ProSound F75 premier CV with a 5-MHz transducer (Hitachi, Tokyo, Japan). Left ventricular ejection fraction were measured by the disk summation method. To assess right ventricular systolic function, TAPSE was measured in an apical 4-chamber view. A pulsed-wave Doppler tracing of RVOT flow and a continuous-wave Doppler tracing of TR jet were recorded at a sweep speed of 300 mm/s to assess pulmonary hemodynamics. Careful attention was paid to obtain clear spectral Doppler envelopes. TR jet was able to be detected in all cases, because angiographic catheters passed across the tricuspid valve may create TR. To assess the severity of PH, peak TR velocity was measured. The time from the onset of ejection to peak TR velocity was also measured to evaluate peak pattern of TR velocity profile. From the RVOT flow profile, acceleration time was measured as the time from the onset of ejection to the peak velocity. Mid-systolic notching was considered present, if there was a distinct notch in the mid portion of the RVOT flow profile, dividing the flow profile into two different peaks. These Doppler echocardiographic images were stored in a digital format for subsequent echo-Doppler based wave separation analysis.

The above echocardiographic images were processed using in-house program code written in MATLAB (MathWorks 2019b, MA, United States). The envelope of RVOT flow Doppler signal was traced semi-automatically to obtain U waveform. This waveform was smoothed using a Savitzky–Golay filter and then ensemble averaged over 3 cardiac cycles with reference to the R wave on electrocardiogram. The timing of beginning and end of ejection were identified from the U waveform. P waveform was estimated by applying the simplified Bernoulli equation to a semi-automatic tracing of the envelope of TR Doppler signal. This waveform was smoothed using a Savitzky–Golay filter and then ensemble averaged over 3 cardiac cycles with reference to the R wave on electrocardiogram. Pulse pressure waveform was obtained by subtracting end-diastolic P from the P waveform [equation (8)]. The temporal resolution for both waveforms was approximately 2.0 msec for all acquisitions that resulted from the setting of sweep speed at the acquisition (300 mm/s) and a pixel size of the echocardiographic images. The velocity resolution was dependent on the Nyquist limit and a pixel size of the echocardiographic images; this yielded a velocity resolution of 0.5 ± 0.7 cm/s for RVOT flow and 1.4 ± 0.4 cm/s for TR jet. The pressure resolution varied depending on the level of pressure, since pressure was estimated proportional to the square of velocity. The waveforms were then interpolated using cubic spline and resampled to 1 msec temporal resolution (1,000 Hz).

The reference P and U waveforms measured directly with the catheter were processed similarly as were the echo-Doppler waveforms. The waveforms were smoothed using a Savitzky–Golay filter and then ensemble averaged over 3 cardiac cycles with reference to the R wave on electrocardiogram. These waveforms were then interpolated using cubic spline and resampled to 1 msec temporal resolution (1,000 Hz). The timings of the beginning and end of ejection were identified from the U waveform. End-diastolic P was subtracted from the P waveform to obtain pulse pressure waveform before wave separation analysis.

Wave separation analysis was performed separately on the echo-Doppler derived waveforms and reference waveforms, using in-house program code written in MATLAB. To determine WS, pulse pressure was plotted against velocity over the ejection period. At the beginning of ejection, a linear portion was always observed on the P-U loop, suggesting that a wavefront during this period was only that generated by the contraction of right ventricle. The slope of the linear portion was calculated from the loop and divided by the density of blood to obtain WS. Pulse pressure waveform was then separated into Pf and Pb waveforms, as described earlier in this section. In this study, Pf and Pb were presented as the peak value and these peak values were used for calculating RC.

Categorical variables were expressed as number and percentage and were compared using Fisher’s exact test. Continuous variables were expressed as mean ± standard deviation. After an assessment of the differences in the before challenge, fluid challenge, and dobutamine challenge results using two-way repeated-measures ANOVA, Tukey’s honestly significant difference post hoc test was used to compare variables between each result. Linear regression analysis was used to compare echo-Doppler derived wave reflection and arterial stiffness indices against catheter derived indices. Bland-Altman plot was used to visualize random and systematic errors. Linear regression analyses were also performed to identify the hemodynamic determinants of wave reflection and arterial stiffness indices. Explanatory variables included PVR, pulmonary arterial compliance, cardiac output and heart rate. Analysis of covariance was used to determine the effect of fluid loading and inotropic stimulation on the relationship between wave reflection and arterial stiffness indices and right ventricular systolic function. P-value < 0.05 was considered as statistically significant. All statistical analyses were performed using IBM SPSS statistics version 19 (IBM, IL, United States) and GraphPad Prism version 8.0 (GraphPad, California, CA, United States).

The changes in pulmonary hemodynamics and right ventricular function induced by pulmonary arterial microembolization and hemodynamic manipulations are summarized in Table 1. The injection of dextran microsphere significantly increased mean PAP from 14 ± 3 to 34 ± 7 mmHg (p = 0.008) and PVR from 2.9 ± 1.0 to 13.7 ± 4.9 Woods unit (p = 0.011). Cardiac output decreased (p = 0.04). Right ventricular systolic function was also reduced, as demonstrated by a significant decrease in tricuspid annular plane systolic excursion (TAPSE) (p = 0.015). There were alterations in RVOT flow profile such as the presence of mid-systolic notching; TR jet profile exhibited high and late peaking pattern as shown by higher TR velocity (p = 0.006) and longer time to peak velocity (p = 0.01). Fluid loading did not change PVR (at baseline, p = 0.99; at PH state, p = 0.63); it did not increase cardiac output at both baseline (p = 0.46) or not at all at PH state (p = 0.61). Inotropic stimulation did not change PVR (at baseline, p = 0.93; at PH state, p > 0.99), although it significantly increased cardiac output at baseline (p < 0.001), but did not increase cardiac output at PH state (p = 0.36).

Figure 2 compares sample separated waveforms obtained from direct P and U measurements and those obtained from echo-Doppler derived P and U measurements. At baseline, almost no wave reflection was observed. After the induction of PH, pulse pressure by Doppler dramatically increased from 7.7 to 29.2 mmHg with late-peaking pattern; Pb by Doppler started to rise 52 msec later than the onset of ejection and peaked at 12.1 mmHg. This comparison demonstrated that the new echo-Doppler method produced comparable Pb and Pf waveforms with those derived from direct measurements. Overall, Pb by Doppler was not zero, but minimal (1.4 ± 1.3 mmHg) at baseline and increased 7.5 times, up to 10.4 ± 3.9 mmHg, associated with the development of PH, even though mean PAP increased only 2.4 times, up to 34 ± 7 mmHg; its increase was substantially larger, compared to 1.9 times increase in Pf by Doppler (from 8.6 ± 2.5 to 16.1 ± 5.1 mmHg), resulting in a significant rise in RC by Doppler (from 0.16 ± 0.10 to 0.65 ± 0.13). WS by Doppler also significantly increased from 1.2 ± 0.3 to 3.3 ± 1.0 m/s.

When the data for all load-state conditions were pooled, Pb, Pf, and RC by Doppler correlated linearly with the corresponding indices by catheter with acceptable variabilities (Figure 3). Lower part of Figure 3 also shows the result of a Bland-Altman analysis between the Pf (Mean difference, 0.3; SD of mean difference, 3.94; the limits of agreement, −7.4 to 8.1), Pb (Mean difference, 0.4; SD of mean difference, 2.36; the limits of agreement, −4.3 to 5.0), RC (Mean difference, 0.003; SD of mean difference, 0.13; the limits of agreement, −0.26 to 0.26), WS (Mean difference, 0.31; SD of mean difference, 0.78; the limits of agreement, −1.22 to 1.85) evaluated by the catheter and Doppler methods. Almost of all dot plots did not exceed the limits of agreement. WS by Doppler also significantly correlated with that by catheter; the regression line in the scatterplot was deviated slightly upward from the identical line, suggesting that the echo-Doppler method overestimated WS. When looking at individual change in wave reflection and arterial stiffness indices for each dog, the induction of PH increased the indices by Doppler similarly as did those by catheter (Figure 4). This indicated that the echo-Doppler method was able to track the alterations in pulmonary arterial wave reflection associated with the development of PH, with reasonable accuracy. On the other hand, Bland-Altman analysis also showed that the error tended to increase as the wave reflection increased.

Table 2 summarizes the changes in wave reflection and arterial stiffness indices by Doppler induced by preload loading and inotropic stimulation. An increase in preload by fluid loading, which caused a significant, but small increase in cardiac output (Table 1, Baseline Fluid challenge), did not alter the wave reflection and arterial stiffness indices. In Table 2, the result of fluid challenge is following: Pb in Fluid challenge Baseline before vs. after, p = 0.4; Pb in Fluid challenge Pulmonary hypertension before vs. after, p = 0.57; Pf in Fluid challenge Baseline before vs. after, p = 0.4; Pf in Fluid challenge Pulmonary hypertension before vs. after, p = 0.89; RC in Fluid challenge Baseline before vs. after, p = 0.44; RC in Fluid challenge Pulmonary hypertension before vs. after, p = 0.16; WS in Fluid challenge Baseline before vs. after, p = 0.23; WS in Fluid challenge Pulmonary hypertension before vs. after, p = 0.84. Inotropic stimulation produced a larger increase in cardiac output than fluid loading (Table 1, Baseline DOB challenge); it increased Pb as well as Pf, but did not alter RC. In table 2, the result of dobutamine challenge was following: Pb in dobutamine challenge baseline before vs. after, p = 0.08; Pb in dobutamine challenge pulmonary hypertension before vs. after, p = 0.14; Pf in dobutamine challenge baseline before vs. after, p = 0.17; Pf in dobutamine challenge pulmonary hypertension before vs. after, p = 0.03; RC in dobutamine challenge baseline before vs. after, p = 0.11; RC in dobutamine challenge pulmonary hypertension before vs. after, p = 0.08; WS in dobutamine challenge baseline before vs after, p = 0.23; WS in dobutamine challenge pulmonary hypertension before vs. after, p = 0.34. These results suggested that both Pb and Pf were flow-dependent and the effect of flow status on Pb can be cancelled out by taking a ratio of it to Pf. Inotropic stimulation also did not alter WS. These trends were also confirmed for catheter-derived wave reflection and arterial stiffness indices (Supplementary Table 1). In Supplementary Table 1, the result of Fluid challenge was following: Pb in Fluid challenge Baseline before vs. after, p = 0.18; Pb in Fluid challenge Pulmonary hypertension before vs. after, p = 0.54; Pf in Fluid challenge Baseline before vs. after, p = 0.11; Pf in Fluid challenge Pulmonary hypertension before vs. after, p = 0.51; RC in Fluid challenge Baseline before vs. after, p = 0.53; RC in Fluid challenge Pulmonary hypertension before vs. after, p = 0.8; WS in Fluid challenge Baseline before vs. after, p = 0.3; WS in Fluid challenge Pulmonary hypertension before vs. after, p = 0.11 (Inotropic stimulation increased Pb and Pf, but did not change RC). In Supplementary Table 1, the result of dobutamine challenge was following: Pb in Dobutamine challenge Baseline before vs. after, p = 0.03; Pb in Dobutamine challenge Pulmonary hypertension before vs. after, p = 0.01; Pf in Dobutamine challenge Baseline before vs. after, p = 0.03; Pf in Dobutamine challenge Pulmonary hypertension before vs. after, p = 0.04; RC in Dobutamine challenge Baseline before vs. after, p = 0.62; RC in Dobutamine challenge Pulmonary hypertension before vs. after, p = 0.78; WS in Dobutamine challenge Baseline before vs. after, p = 0.07; WS in Dobutamine challenge Pulmonary hypertension before vs. after, p = 0.07.

The results of linear regression analyses identifying the determinants of echo-Doppler derived wave reflection and arterial stiffness indices are shown in Table 3. RC was correlated strongly with PAC, moderately with PVR, and weakly with cardiac output. WS was also correlated strongly with PAC and PVR, moderately with cardiac output. The magnitude of the associations with both RC and WS was the highest for PAC (RC by Doppler, r = −0.85; WS by Doppler, r = −0.71). Stepwise multiple linear regression analysis revealed that PAC and cardiac output were independent determinants of RC, and PAC was an independent determinant of WS. PAC was also identified as the strongest hemodynamic determinant for catheter-derived RC and WS (RC, r = −0.70; WS, r = −0.79, Supplementary Table 2).

Figure 5 relates right ventricular systolic function measured by TAPSE with echo-Doppler derived wave reflection and arterial stiffness indices. Overall, TAPSE was negatively correlated with both RC (r = −0.40, p = 0.005) and WS (r = −0.49, p = 0.001). When analyzed separately for loading conditions, the relationships of TAPSE with RC and WS did not change by fluid loading (RC, p = 0.48; WS, p = 0.28). However, the relationships were shifted upward by inotropic stimulation (RC, p = 0.06; WS, p = 0.008). When catheter-derived RC and WS were used instead for the above analysis, similar upward shifts by inotropic stimulation were observed (Supplementary Figure 1).

To infer a diagnosis of PH, the effects of PH on the right heart are imaged with echocardiography in clinical practice. Some of such echocardiographic signs are related to the presence of pulmonary arterial wave reflection. For example, a late-peaking pattern on TR velocity profile is caused by PAP augmentation by early arrival of reflected wave (Bech-Hanssen et al., 2010). A mid-systolic notching on RVOT flow profile represents an abrupt reduction in right ventricular ejection flow secondary to wave reflection (López-Candales and Edelman, 2012). Although these signs are reasonably specific, the diagnostic sensitivity is generally low and these signs are absent in some patients (Arkles et al., 2011). The signs are also not sensitive enough to be used for monitoring the treatment course of patients.

To provide a more accurate assessment of wave reflection, wave separation analysis is performed on pulmonary arterial pressure and flow waveforms into their forward and backward components. This analysis is classically undertaken in the frequency domain, but it is fairly complicated and needs sophisticated software (Segers et al., 2007). An alternative method, called wave separation analysis, has been particularly attracting attention, because it is a time-domain based approach and therefore can be more easily applied to study the timing and magnitude of wave reflection (Laxminarayan, 1979; Li, 1986). The results are easier to interpret in physiological terms.

Most previous studies of pulmonary arterial wave separation analysis measured pressure and flow waveforms directly with catheters, which hinders its use in the clinical setting (Hollander et al., 2001). A recent study has attempted to obtain these waveforms noninvasively, using magnetic resonance imaging, where It measured the temporal changes in cross-sectional area of pulmonary artery for a surrogate for PAP (Quail et al., 2015). The noninvasive method we introduce herein uses Doppler echocardiography to obtain pulmonary arterial pressure and flow waveforms from TR and RVOT flow profiles. These measurements are routinely performed for diagnosing PH or monitoring the treatment course, so this echo-Doppler method can be easily applied in the clinical setting.

The magnitude of wave reflection was usually quantified by RC that was defined as a ratio of backward to forward component of either energy of waves or pressure. This study has reported that RC was not zero, but small at 0.16 ± 0.10 at baseline in dogs. This may be because dogs have inherently a mild degree of PH with a PVR of 2.9 ± 1.0 Woods unit. This finding suggested that wave separation analysis may be sensitive for detecting a mild degree of PH. This study also found that RC correlated negatively with right ventricular systolic function. This negative correlation suggested that RC was able to reasonably capture right ventricular – arterial coupling in the acute setting. The experimental model in this study was different from typical clinical scenarios, i.e., a long-term adaptation or maladaptation of pulmonary vasculature and right ventricular myocardium in response to chronic pressure loading. Thus, it needs to be tested whether RC was able to describe the coupling in the clinical setting.

This echo-Doppler method involves several assumptions. Firstly, right atrial pressure is assumed to be constant during systole. The fluctuation in right atrial pressure during this period is usually small enough to be negligible, when compared with PAP. This assumption seemed valid from the results of this validation study demonstrating that the echo-Doppler method accurately quantified wave reflection. It may be because all dogs used in this study presented with only mild degree of TR. Once severe, it causes a significant rise in right atrial pressure in late systole; thus, we assume that the echo-Doppler method may not be applicable to those with severe TR. Secondly, a pulsed-wave Doppler tracing of RVOT flow was used as a surrogate for PA flow velocity waveform. The velocity should be higher when measured at RVOT than when measured at pulmonary arterial trunk, which should result in higher characteristic impedance, i.e., wave speed when assessed by the echo-Doppler method. This overestimation was confirmed by the present study showing that wave speed measured by the echo-Doppler method was higher, by a mean of 0.42 m/s, than that by direct measurement. Thirdly, RR interval on electrocardiogram is assumed to be constant throughout image acquisitions. In this method, pressure and flow waveforms are obtained from echo-Doppler images acquired at different cardiac cycle and the both waveforms are aligned in time with reference to the R wave on electrocardiogram. Therefore, this method cannot be applied to patients with large heart rate variability such as atrial fibrillation. Fourthly, Hollander et al. reported the existence of negative reflections which occurs in early systole in dogs, indicating which casts doubt on the validity of early systole being reflection-free which is a necessary assumption for use of the Water Hammer equation. However, we think that the impact of this negative reflections in systole are small.

Despite the above assumptions, the present study demonstrated that the echo-Doppler method can quantify pulmonary arterial wave reflection with reasonable accuracy and it is sensitive for detecting the changes in pulmonary arterial wave reflection associated with the development of PH.

Pulmonary arterial hypertension remains a fatal disease with a 1-year mortality of approximately 20%, despite advances in disease-specific therapies (Benza et al., 2012). This may be because the disease is often far established at diagnosis as demonstrated by contemporary disease registries reporting that mean PVR at diagnosis ranged between 8 and 10 Wood units (Thenappan et al., 2007). Early initiation of the therapies may be effective, as suggested by a previous randomized control study including patients with mildly symptomatic pulmonary arterial hypertension (Galiè et al., 2008). At the early stage of disease, PAC falls considerably while PVR increases only slightly. Thus, the assessment of pulsatile load may allow an early diagnosis even when PVR stays within normal limits. In the present study, wave reflection, an important component of the pulsatile load, rose substantially despite a slight increase in mean PAP induced by pulmonary arterial microembolization. This finding is consistent with another study of wave separation analysis, evidencing that a significant wave reflection was present even in PH patients with mildly elevated PAP (Su et al., 2017). Therefore, this echo-Doppler assessment of wave reflection may be able to offer an early diagnosis of PH.

PH-specific treatments do not always produce desirable clinical outcomes. For example, some patients with chronic thromboembolic pulmonary hypertension exhibit exercise intolerance even after pulmonary endarterectomy, often despite normalization of pulmonary arterial hemodynamics. This may partly be attributable to persistent pulmonary arterial wave reflection after the surgery. A previous study demonstrated using wave separation analysis that there remained a large reflected pressure wave 3 months after the surgery, even in patients without residual PH (Su et al., 2019). Pulmonary arterial wave reflection could be an attractive target for the treatment of PH (Su et al., 2018). However, it remained to be tested whether a reduction in pulmonary arterial wave reflection will lead to a significant improvement in clinical outcomes.

Wave reflection is one component of pulsatile load, so it is inherently included in arterial compliance. This is why RC correlated with pulmonary arterial compliance. Pulse pressure, a major component of arterial compliance, is a sum of forward pressure representing stiffness of proximal artery and backward pressure representing pathology of more peripheral arteries. When pulse pressure is elevated, we are not sure which of the two pressures is elevated. Therefore, assessing wave reflection may offer benefits over pulmonary arterial compliance for discriminating diseases with different sites of arteries affected, for example, primary pulmonary arterial hypertension and chronic thromboembolic pulmonary hypertension. While the clinical implication of wave reflection in the systemic circulation is becoming more apparent, only a few studies have applied wave reflection in the pulmonary circulation and these have mainly been in animal models (Hollander et al., 2001; Nie et al., 2001). Therefore, since the current concept of wave intensity analysis in the pulmonary circulation has been established by animal studies, the results of this study may be applicable in human clinical practice. The present study demonstrated that a noninvasive measurement of PA wave reflection is feasible using Doppler echocardiography. The methodology developed in our study may be potentially helpful in future clinical applications involving pathological changes in the distensibility of the pulmonary artery. Recent advances in imaging technologies should facilitate future use of pulmonary wave intensity analysis in a clinical setting. We expect PA wave reflection provides additional information about hemodynamic of pulmonary circulation.