INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

IN CONTRAST TO THE SYSTEMIC arterial system, which can be described as a single long vessel (the aorta) with side branches, the pulmonary artery (PA) branches immediately and rapidly within a few centimeters of the heart (11, 12, 30). Reflected waves observed in the aorta have generally been of the closed-end type, where the forward-going compression waves responsible for accelerating the blood out of the ventricle are returned as compression waves that oppose the flow of blood from the ventricle (23-25). This is most clearly indicated in the type A waveforms of older patients (21, 23). In the PA, reflected waves have not been so clearly identified, except in the presence of a vasoconstrictor, i.e., serotonin (2, 29). In the control state, it is generally considered that the PA circulation does not create reflections of significant magnitude and, therefore, that it is optimally coupled with the right ventricle (30). However, having assumed that no energy is added to the system (i.e., that the PA vasculature is passive), Attinger (1) concluded that the only possible explanation for all of the harmonics increasing in magnitude is that incident and reflected waves are superimposed.

To characterize wave reflection in the PA system, we used wave-intensity analysis (WIA) (25, 26) a method that can distinguish between forward- and backward-going waves and identify whether they are compression or expansion waves. As recently demonstrated (33) for the coronary arterial circulation, if wave speed can be determined, the respective intensities of simultaneous forward- and backward-going waves can also be calculated.

The overall pattern of waves observed in the PA is similar to that seen in the systemic arteries (Fig. 1). There is a large forward-going compression wave at the start of systole generated by the contraction of the ventricle, a period in midsystole when there is relatively little wave motion, and a complex of waves starting near the end of systole and continuing into early diastole. In this study, we will concentrate on the waves in early systole, primarily because their interpretation is easier. All of the waves from the previous heartbeat have dissipated during diastole and so the initial (forward) compression wave propagates into a nearly quiescent PA. Thus any backward-going waves are almost certainly reflections of this initial wave.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Proximal pulmonary arterial wave-intensity analysis at low (LoV) and high volume (HiV) and the pressure (PPA) and velocity (UPA) waveforms (top) and the corresponding wave-intensity analysis (bottom). By convention, intensities of forward-going waves (dIw+) are plotted as positive and intensities of backward-going waves are negative (dIW); compression waves are denoted by c and expansion waves by e. Note the forward-going compression wave associated with acceleration of blood out of the ventricle and the backward-going expansion wave that followed. At HiV, note the increased magnitude of the backward-going expansion wave (e) relative to the forward-going compression wave (c).

To generate well-defined waves that we could control and manipulate, we introduced a counter-pulsation balloon into the left atrium (LA) that, by direct transmission through the heart, produced in the PA a forward-going compression wave on inflation and a forward-going expansion wave on deflation (15). We analyzed the behavior of the artificially generated waves under a variety of conditions; i.e., low- and high-blood-volume states, hypoxia, and positive end-expiratory pressure (PEEP), to determine their effects on PA wave reflection. Under all of these conditions, we analyzed the waves that were reflected from the PA vasculature to evaluate the hypothesis that there may be sites in the PA vasculature that have a negative reflection coefficient. It was anticipated that due to its marked and immediate increase in cross-sectional area, the PA circulation might behave in a manner consistent with an open-end reflector, as opposed to the closed-end types of reflections characteristic of the normal systemic circulation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Animal preparation. The studies were performed on 11 dogs (20-27 kg wt) of either gender. Dogs were anesthetized with thiopental sodium (30 µg/kg) and then with 30 mg · kg¹ · h¹ of fentanyl citrate, and ventilated with a 2:1 NO₂-O₂ mixture using a constant-volume ventilator with a tidal volume of 15 ml · kg¹ · min¹ (model 607, Harvard Apparatus; Natick, MA). Temperature was maintained at 37°C by a circulating-water warming blanket. A midline sternotomy was performed with the dog in the supine position. The pericardium was opened with an incision along the atrioventricular groove so that the heart could be instrumented. Right and left ventricular pressures (P_RV and P_LV) were measured using 8-Fr micromanometer-tipped catheters with reference lumens (model PC-480, Millar Instruments; Houston, TX) introduced through a branch of the external jugular vein and carotid artery, respectively. PA and LA pressures (P_PA and P_LA) were measured with the use of a 3-Fr micromanometer-tipped catheters (model SPR-524, Millar) introduced in a retrograde fashion through small pulmonary arterial and venous branches, respectively. As ascertained by palpation, the PA catheter was advanced within 1 cm of the flowmeter and the LA catheter was advanced just into the LA. Aortic pressure was measured by using an 8-Fr fluid-filled catheter introduced through a femoral artery and was attached to a transducer (model P23ID, Statham-Gould). PA flow was measured using an ultrasonic flowmeter (Transonic Systems; Ithaca, NY) placed on the common PA within 2 cm of the pulmonic valve and converted to velocity (U_PA) using a value of cross-sectional area estimated from the size of the flow probe. The time delay in the response of the flowmeter, measured in vitro by comparing the time derivative of the displacement of a syringe pump (measured using a linear potentiometer) to the output of the flowmeter, was 3.5 ± 0.5 ms for the conditions used in the experiments (14, 33). A 35-ml special-order spherical balloon was inserted in to the LA through the appendage and inflated using an aortic counter-pulsation pump (Datascope; Paramus, NJ). The balloon remained deflated until required. The heart was repositioned within the pericardial sac and its margins were loosely reapproximated. A large-bore catheter in the external jugular vein was used to infuse a 2% albumin-saline solution or remove blood for adjustment of the intravascular volume. A single electrocardiographic lead was also recorded. Conditioned signals (model VR16, Electronics for Medicine, Honeywell; Pleasantville, NY) were recorded by a means of a computer using acquisition software (Sonometrics; London, Ontario, Canada). The analog signals were passed through anti-aliasing low-pass filters with a cutoff frequency of 100 Hz and were then sampled at a frequency of 200 Hz. The digitized data were subsequently analyzed with specialized software (CVSOFT, Odessa Computer Systems; Calgary, Alberta, Canada).

Experimental protocol. Of the 11 dogs, the low volume (LoV) and high volume (HiV) protocol and the control balloon inflation protocol were performed in 9 dogs. Of these nine, the PEEP protocol was also performed in six dogs and, of these six, the hypoxia protocol was performed in three dogs. In addition, the hypoxia protocol was performed in two other dogs.

Data analysis. WIA was used to evaluate the direction, intensity, and type of traveling waves (25). Unlike impedance analysis, which considers the measured pressure and flow waves as the summation of sinusoidal wave trains of different frequencies, this analysis deals with the propagation of infinitesimal wave fronts defined for convenience by the changes in pressure and velocity during each sampling interval. Thus the measured pressure and velocity waveforms are thought of as the summation of successive incremental wave fronts, denoted here simply as "waves." These waves must be considered as pressure-velocity waves because the change in pressure (potential energy) across the wave is related to the change in velocity (kinetic energy).

W

W

W

W

Statistical analysis. Comparisons between beats were made using the analysis of variance with a Bonferroni correction for multiple comparisons; P < 0.05 was considered significant. All data are expressed as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

The intensities of the forward- and backward-going PA waves are shown on the left side of Fig. 1, indicating that the forward-going compression wave associated with the acceleration of blood out of the RV was followed by a backward-going (reflected) expansion wave. At LoV, the backward-going wave arrived 24 ± 2 ms after the forward-going wave, indicating reflection occurred from a site 3.3 ± 0.5 cm downstream. As the sign of dP for the reflected wave was opposite to that of dP₊ for that of the forward-going wave, the wave reflection was negative (i.e., from an open-end type of reflection site); analysis of the ratios of the wave energies indicated that R = 0.07 ± 0.02 (Table 1). Increasing blood volume (HiV) increased R to 0.20 ± 0.04 (P < 0.01); it was calculated that the reflected backward-going expansion wave originated from a similar location 2.6 ± 0.3 cm downstream. The increased magnitude of negative reflection with volume loading indicates that the properties of the reflection site have been altered, thus causing a larger proportion of the incident wave to be reflected.

The addition of 20 cmH₂O PEEP at HiV increased R from 0.20 ± 0.03 to 0.26 ± 0.02 (P < 0.01, see Fig. 2). Therefore, at HiV, the addition of PEEP increased the magnitude of the open-end type of reflection, thus tending to augment the flow out of the RV somewhat further.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Proximal pulmonary arterial wave-intensity analysis at 0 and 20 cmH2O positive end-expiratory pressure (PEEP). Note the larger backward-going expansion wave (e) with PEEP (see Fig. 1).

When early-systolic, forward-going, compression waves in the PA were augmented by LA balloon inflation as shown in the right panel of Fig. 3, the energies of the forward- and backward-going waves were increased in proportion, resulting in no significant change in R (Table 1). The character of the reflection from this site is dramatically underscored by the fact that balloon deflation generated a new forward-going expansion wave, which was also reflected negatively as a well-defined backward-going compression wave.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Proximal pulmonary arterial wave-intensity analysis at HiV showing the effect of an early systolic artificially generated wave. Control conditions (left) and effects of early systolic balloon inflation and deflation (right). Note that, on balloon deflation, a forward-going expansion wave appeared, which was then reflected as a midsystolic, backward-going compression wave (bottom right) (see Fig. 1).

Figure 4 shows one beat recorded under normoxic conditions and one beat recorded under hypoxic conditions. Hypoxia increased peak P_PA from 30 ± 3 mmHg to 43 ± 2 mmHg (P < 0.01) and increased the slopes of both pressure and velocity waveforms (i.e., dP and dU were increased). As evaluated in terms of the early-systolic backward-going expansion wave, negative reflection was maintained in the presence of hypoxia and R was unchanged in magnitude (control, R = 0.10 ± 0.05, and hypoxia, R = 0.08 ± 0.03; P = 0.6). In addition; however, the vasoconstricting effect of hypoxia was manifested as a reflected compression wave that arrived in the proximal PA in midsystole, 77 ± 8 ms after the forward-going wave, suggesting a closed-end reflection site 9.0 ± 0.1 cm downstream. This backward-going compression wave increased P_PA and decreased U_PA, thus tending to impede flow from the RV. The energy of the backward-going compression wave was similar to that of the backward-going expansion wave (Table 1) in that R for the backward-going compression wave was +0.09 ± 0.03. Thus, in the presence of hypoxic conditions, the negative, open-end type reflection was maintained, whereas a positive, closed-end type reflection was created further downstream.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Proximal pulmonary arterial wave-intensity analysis during hypoxia. Note that there were no changes in the magnitude of the backward-going expansion wave relative to the preceding forward-going compression wave. During hypoxia, there was a backward-going compression wave representing positive wave reflection from constricted arterioles (see Fig. 1).

Because the measured time delay for the flowmeter using the settings employed in these experiments (3.5 ± 0.5 ms) was <1 sampling period, no time shift was used in the calculation of the data presented here. However, because WIA as a time-domain method is particularly sensitive to time delay in the measurements, similar calculations were carried out shifting the velocity measurements relative to the pressure measurements by 5 ms (one sampling period). The resultant changes in the magnitudes of wave intensity and wave energy were minimal and none of the qualitative results of the study were affected by this correction (14).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Our major finding is that wave reflections in the proximal PA are negative, indicating that they arise from an open-end type of reflection site located in the proximal portion of the pulmonary arterial tree. The compression wave that is associated with early-systolic acceleration of blood out of the RV was reflected as a backward-going expansion wave, which tended also to accelerate the flow of blood. This backward-going expansion wave has a "pulling" type of action that tends to draw the blood forward, thus supplementing the action of the forward-going compression wave directly generated by the RV. The magnitude of the negative reflection and the further acceleration it provided was augmented by increasing blood volume and by PEEP. Hypoxia, a known pulmonary vasoconstricting stimulus with effects similar to serotonin (2, 29), did not affect the negative reflection site in the proximal portion of the PA (~3 cm from the site of measurement in the main PA), but did create an additional positive reflection site farther downstream (~9 cm from the site of measurement), near the third or fourth generation of arterial branches (1, 27, 29).

These estimates of the locations of the reflection sites are, of course, approximate because their exact determination would require knowledge of the local wave speeds throughout the pulmonary circulation under each experimental condition. Because we can only determine the wave speed in the PA where the pressure and velocity measurements were made, this wave speed was used to estimate the distances of the main reflection sites from the measured delay between the initial FCW and the principal reflections indicated by the peaks in dI_W. As other studies suggest that wave speeds tend to increase peripherally in the pulmonary circulation (1, 27), these estimates probably underestimate the true distances. They do, however, give some indication of the location of the putative reflection sites.

In the pulmonary arterial system, the large increase in cross-sectional area over a short distance (11, 12, 32) is likely to be the primary factor responsible for creating the open-end type of reflections that we measured. The concept of an open-end type of reflection site was described by McDonald (19) and is illustrated in the APPENDIX: "If the termination is closed it will be returned as a positive pressure wave; if the end opens into a large reservoir (an "open" end) it will be returned as a negative wave. This second concept is often found difficult to understand but it is clear that once the fluid flow reaches a reservoir, which is large compared with the volume injected, there will be no pressure variation in that reservoir. The energy of the transmitted wave cannot, by the law of conservation of energy, disappear; thus a wave equal in amplitude but opposite in sign is created."

We have calculated R as the (signed) ratio of the energy of the reflected (backward) wave to that of the incident (forward) wave. Womersley (35) showed that reflection depends on both the area and elastic properties of the vessel such that no reflection will occur if the characteristic impedances on either side of a junction are equal (i.e., the impedances are "matched"). More specifically, with respect to reflection from a given bifurcation, the sign and magnitude of R depend on the characteristic impedances, Z_n = c_n/A_n, of the parent and daughter vessels, where A_n and c_n are the area and wave speed of vessel n.
where 0 refers to the parent vessel and 1 and 2 refer to the daughter vessels. Because the value of R depends on the areas and wave speeds in the parent and daughter vessels and these parameters will change with changing experimental conditions, we expect that the value of R for every bifurcation would have changed significantly in the course of our experiments (19, 27). We will therefore discuss our observations in terms of changes in R, the quantity actually measured, and infer the changes in vessel area and distensibility necessary to produce the observed reflections.

For a well-matched symmetrical bifurcation (i.e., a bifurcation at which the characteristic impedances are matched and from which no reflections will emanate), the area ratio of daughter-to-parent areas,  = (A₁ + A₂)/A₀, ranges from 1.1 to 1.2, depending on the relationship between area and wave speed for the vessels (19, 20, 22, 28, 34). When  decreases below 1.1, the magnitude of positive, closed-end type reflection will increase; as  increases above 1.2, the magnitude of negative, open-end type reflection will increase. In the PA, values of  from 1.2 to 1.3 have been measured (3, 5, 27). Our observation of negative reflection from a site ~3 cm from the valve implies that  exceeded 1.2 at that site. During hypoxia, our observation of positive reflection from a site ~9 cm from the valve implies that  was <1.1 there.

When we studied the effects of volume loading, the increase in R that we observed (from 0.07 ± 0.02 to 0.20 ± 0.04) suggests that  had increased further at that same reflection site. When intravascular volume was increased, the greater distensibility of the daughter vessels may have led to a greater increase in diameter compared with their parent vessel, and therefore a greater  (9, 11, 18).

When we studied the effects of PEEP, the increase in R that we observed (from 0.20 ± 0.03 [HiV] to 0.26 ± 0.02) suggests that  had increased even further. This increase in the magnitude of R with the addition of PEEP may have been related to a further increase in  caused by a tethering effect of the PEEP-induced expansion of the lung (4, 6, 16, 17). Although PEEP probably increased the diameter of large pulmonary arteries, in our model it may have increased the diameter of the daughter vessels even more. One might expect that PEEP would decrease vessel diameters by compression within the parenchyma of the lung but because we used an open-chest preparation, 20 cmH₂O of PEEP expanded the lungs substantially so that they protruded outward from the cavity and possibly increased vessel diameters. This speculation remains to be confirmed by further studies.

The dramatic increases in R that followed volume loading and PEEP suggest that anything that affects the diameter of the large pulmonary arteries may have substantial implications for wave reflection. The hypoxia data suggest that despite significant vasoconstriction, the negative reflection, which tends to promote flow out of the RV, is maintained, although positive reflection from more distal vessels is also created. The data presented here are compatible with the conclusion that both negative and positive wave reflections are primarily functions of  at given sites, which can be altered independently.

We expressed the energy of the reflected wave as a fraction, R, of the energy of the incident wave. At LoV, the magnitude of the reflected wave was ~8% of the incident wave. Whether this pattern of reflection is "optimal" is not clear. When blood volume was increased (HiV), the magnitude of the backward-going expansion wave increased to ~20% of the incident wave. This would appear to facilitate forward flow substantially, and without any "cost" to the RV. The increased negative reflection with volume loading may have implications in high-cardiac-output states in which RV output can be increased without a proportional increase in RV stroke work. Even though the afterload of the RV is much less than that of the LV, it has seemed remarkable that the thin-walled RV is able to eject the same volume as the LV, an output that may approach 20 l/min in a stressed and trained subject. Negative wave reflection may provide a partial answer to that apparent paradox.

The data of Ha et al. (13), who observed a decrease in the backward pressure and an increase in the backward flow waveform early during ejection in chronically instrumented control dogs (see Fig. 6, top, of Ref. 13), are compatible with our conclusions. These changes disappeared and evidence for a prominent backward-going compression wave developed in their model of pulmonary hypertension (monocrotaline pyrrole).

For the first time, we have demonstrated an unsuspected feature of the pulmonary circulation that might have very important consequences. However, our experiments were performed in anesthetized open-chest dogs and the magnitude of this negative reflection was not great. Our results are statistically significant but the true significance of our findings cannot be determined until confirmed in people. Fortunately, because WIA only requires simultaneous measurements of pressure and velocity and because high-fidelity measurements of pressure and Doppler measurements of velocity are feasible and convenient in most catheterization laboratories, the results of such clinical studies soon may be forthcoming.

The hemodynamic burden of pulmonary hypertension has never been fully explained by analysis of pulmonary vascular resistance. From these observations, we speculate that pulmonary hypertension might change vascular stiffness or geometry in a way that minimizes or eliminates the negative reflection on which the RV normally depends (10, 31). Alternatively, insofar as clinical pulmonary hypertension resembles the HiV, increased-PA-pressure condition studied here, these patients may already have the benefit of increased negative reflection (although the benefits might be minimized by decreased compliance). In the absence of this "compensating" mechanism, RV ejection might be even further impaired. Clinical studies to test these hypotheses should be undertaken.

In conclusion, by using WIA, we have demonstrated that the normal pulmonary arterial circulation is characterized by negative wave reflections and that the degree of these negative reflections is increased with increasing blood volume and PEEP. This behavior, so fundamentally different from that of the systemic circulation, may help to explain the marked differences between the hemodynamics of the RV and the LV.