Circulatory Mechanics in the Toad Bufo Marinus: II. Haemodynamics of the Arterial Windkessel

The viscoelasticity of the arterial tree is a major haemodynamic determinant that influences the pulse-wave velocity, the hydraulic input impedance and the shape of blood pressure and flow waveforms. Wave propagation effects, such as pulse amplification, distortion and reflections evident as secondary peaks are seen in mammalian circulatory systems, partly because of the ‘elastic taper’ of the aorta, but primarily as a result of the relatively high heart rates (Taylor, 1964, 1966b). These features are not evident in poikilothermic vertebrates or invertebrates that have substantially lower heart rates than comparable-sized mammals (Shelton and Jones, 1965, 1968; Shelton, 1970; Burggren, 1977b; Langille and Jones, 1977; West and Burggren, 1984; Shadwick et al. 1987). Thus, the mammalian aorta is appropriately modelled as a wave-transmission system (McDonald,1974), while the arterial tree of a poikilotherm can be described as a simple Windkessel.

Briefly, the Windkessel model treats the aorta and major arteries as a single compliant chamber linked in series with a single peripheral resistor. Pulse waves are transmitted instantaneously, such that the whole system is inflated in synchrony and no propagation effects occur. Flow into the Windkessel is pulsatile, but the outflow through the peripheral resistor is relatively steady because the pulse is smoothed by the expansion and recoil of the elastic chamber. These conditions hold true for an arterial system where the transit time of the pressure wave is a negligible portion of the time of one cardiac cycle. In mammals this is not the case but in poikilotherms, such as small reptiles and amphibians (Shelton and Jones, 1968; Jones and Shelton 1972; Langille and Jones, 1977; Burggren 1977a, Gibbons and Shadwick, 1989), the heart rate is low and the arterial tree is relatively short compared to the pressure wavelengths. Wave propagation effects are insignificant and, thus, the Windkessel model provides an adequate description of the arterial haemodynamics.

Absent from the literature on cardiovascular mechanics of lower vertebrates are any in vivo measurements of arterial elastic properties. In the previous paper (Gibbons and Shadwick, 1991) we showed by in vitro mechanical tests that the aorta of the toad is a resilient elastic vessel that has the same functional properties as the aorta in mammals, i.e. at their respective blood pressures these aortas have similar elastic moduli. In the current study we have determined the dynamic elastic modulus of the aorta from in vivo measurements of pressure and diameter and predicted the wave velocity and characteristic impedance. In addition, we demonstrate the Windkessel characteristics of the system by an analysis of pressure and flow waveforms obtained at different locations within the arterial tree.

Experiments were carried out on nine healthy toads (Bufo marinus L.) weighing between 300 and 450 g. The animals were anaesthetized by intraperitoneal injection of MS-222 (Sandoz, 1:1000, 0.01 mlg⁻¹ body mass). The level of anaesthesia was maintained so that the animal was continually unconscious but breathing regularly. All experiments were acute and terminal, the animals being killed by an excess of MS-222.

Aortic pressure was measured from a cannula inserted dorsally into the subclavian artery up to its junction with the aortic arch. Simultaneously, pressure was measured from a catheter inserted dorsally into the sciatic artery of the leg. Both catheters were less than 15 cm in length. Pressures were measured by two matched Gould P23Db transducers. These were calibrated both dynamically, by the ‘pop test’ technique (Milnor, 1982), and statically, against a water column. Wave velocity was determined from the time delay between the two pressure signals. The positions of the pressure transducers were then reversed to see if any phase errors were present. Following this, a mid-line ventral incision was made to expose the aortic arches and dorsal aorta.

Flow was measured by a pulsed Doppler flowmeter (model 545C-4, University of Iowa Bioengineering). The flow probe consisted of a piezoelectric crystal mounted in epoxy resin on a fabric backing which, when placed against the vessel wall, held the crystal face at 45° to the direction of blood flow (for a description of such an instrument, see Milnor, 1982). The range was set to give the maximum signal, indicating that the probe was recording from the midstream of the flow. This instrument has a fixed output of 0.5 volts per kilohertz of frequency shift. From this, a calibration equation can be derived: Q(1min⁻¹)=(XD²)/ (271.2cosA), where Q is flow, X is the voltage of the flow signal, D is the vessel diameter and A is the angle between the crystal face and the vessel long axis (45°). Flow measurements were taken sequentially at the five sites indicated in Fig. 1. At site 5, pressure and flow were measured in contralateral vessels.

Pulsatile changes in aortic diameter were also measured on several animals, using a video dimension analyzer (VDA, Instrumentation for Physiology and Medicine, model 303) as described by Fung (1981). The VDA has a 15 Hz low-pass filter that introduces significant phase shift and attenuation. The frequency response of the VDA was quantified from 0.05 to 15 Hz, using a vibrator and a transfer function analyzer (series SM272, SE Laboratories), as described previously (Shadwick and Gosline, 1985). Appropriate corrections were made to the phase and amplitude for each harmonic component of the diameter waveforms.

Pressure, flow and diameter were recorded on an FM tape recorder. Subsequent analogue-to-digital conversion was performed by a PDP-11/23 laboratory computer (Digital Equipment Corporation) at a sampling rate of 100 s⁻¹. Signalaveraged pulses were subjected to Fourier analysis. The amplitude and phase were calculated for nine harmonics. The application of this technique to haemodynamics is described by McDonald (1974) and Milnor (1982).

The mean peripheral resistance, R, can be calculated from the mean pressure (P₀) and the mean flow (Q₀) as R=P₀/Q₀. The input impedance is a dynamic analogue of peripheral resistance. Therefore, impedance amplitude (Z) is calculated for each harmonic frequency as Z_n=P_n/Q_n, where P_n and Q_nare the respective amplitudes of the nth harmonics of pressure and flow waves. The phase (ϕ) is calculated from ϕ_n⁼β_n–α_n, where α_n and β_n are, respectively, the phase shifts of the nth harmonic of pressure and flow.

Fig. 2 shows examples of blood pressure waves recorded simultaneously at the aortic arch and 11 cm distally at the top of the sciatic artery. These waveforms are typical of those recorded previously in this species and from other poikilotherms (Shelton and Jones, 1968; Shelton and Burggren, 1976; Langille and Jones, 1977; Burggren, 1977b; West and Burggren, 1984; Shadwick et al. 1987). Pressure falls smoothly during diastole and no secondary peaks are evident. The onset of diastole is not marked by an incisura but only by an inflection in the slope of the descending wave. There is essentially no change in the shape of the pressure wave between the aortic arch and the sciatic artery, and only a small degree of attenuation.

The mean arterial pressure generally ranged from 2.5 to 3 kPa and the pulse pressure ranged from 1.7 to 2.0 kPa. These are similar to values obtained in previous studies on anaesthetized toads (Shelton and Jones, 1968; Kirby and Bumstock, 1969) but lower than those recorded in pithed toads by Van Vliet and West (1987a,b). In the latter studies, blood pressures of 4.5–6.5 kPa were reported, while in conscious and anaesthetized toads aortic pressures range from 3 to 4.5 kPa (Van Vliet and West, 1986; West and Burrgren, 1984).

Pressure pulses from the aortic arch and the sciatic artery were separated by 40–45 ms, as measured by the position of the wave front (Fig. 2). When the catheters were interchanged to feed into the opposite transducers, the time delay remained, indicating that no phase errors were introduced by the transducers.

The pulse velocity calculated from the time delay of the distal pressure wave was 2.4–2.75 m s⁻¹. Heart rate averaged 37 min⁻¹ (0.62Hz), giving a wavelength for the fundamental frequency of about 4.2 m. Fourier analysis of signal-averaged waveforms showed a phase shift of 9–11° in the fundamental harmonics. This yields a pressure wave velocity of 2.25–2.75ms⁻¹ (see Table 1).

Fig. 3 shows examples of blood flow pulses recorded at different levels in the arterial tree of one toad. Peak flows in all our specimens ranged from 0.5 to 0.9 ml s⁻¹ in the aortic arch (site 1), from 0.45 to 0.75 ml s⁻¹ in the dorsal aorta (site 4) and from 0.2 to 0.35 ml s⁻¹ in the sciatic artery (site 5). The end of systole is marked by a slight flow reversal in recordings from the top of the aortic arch only (Fig. 3), presumably indicating closure of the ventricular valves (Shelton, 1970; Langille and Jones, 1977). The positive diastolic flow immediately following this reversal may result from contraction of the conus arteriosus (Langille and Jones, 1977), but this was not always observed in our experiments. Negative flow does not occur at the other arterial locations (Fig. 3). In the lower portion of the aortic arch (site 2) the flow drops rapidly in diastole and remains near zero for over half of the diastolic period. In contrast, flow in the sciatic artery decreases more gradually during diastole, not reaching zero until just before the beginning of the next cycle. These changes demonstrate that the blood flow pulse is being smoothed as it travels through the elastic aorta. In contrast, negative flow and secondary peaks in pressure and flow occur in the mammalian system at sites as peripheral as the femoral artery (McDonald, 1974), owing to the interaction of reflected waves. Blood flow in the distal part of the toad aorta is never negative and falls to zero only at the end of each pulse (Fig. 3)

Fig. 4 shows trains of pressure and flow pulses in the lower aortic arch (site 2) that have been digitized, signal-averaged and subjected to Fourier analysis. The amplitudes of the first nine harmonics of pressure and flow are shown in Fig. 5. The mean pressure and flow are the zero harmonics. Heart rate determines the frequency of the first harmonic, and higher harmonics are multiples of this fundamental frequency. The relatively small amplitude of harmonics above the third results from the smooth shapes of pressure and flow waveforms and the absence of any secondary peaks. In this example, 97% of the pulsatile hydraulic power (see Milnor, 1982) is represented by the first three harmonic pairs.

Vascular impedance spectra for three animals are shown in Fig. 6. Impedance amplitude was normalized to R (the zero frequency value) to account for slight differences in animal size. This ratio falls sharply with the first harmonic and remains relatively constant up to 5 Hz. The final values are less than 10% of the peripheral resistance but remain above the characteristic impedance Z_c. Impedance phase was negative at all frequencies, indicating that flow peaks ahead of pressure by about 80° on average. This is close to the limiting value of 90° expected in a Windkessel model (see Fig. 8; Milnor, 1982).

Fig. 7 shows impedance curves for sites 2–5 in the arterial tree of one animal. Since pressures were not recorded at sites 3 and 4, the pressure pulse measured simultaneously at site 2 was substituted for these impedance calculations. This required a slight phase shift in the pressure pulse so that its wave front coincided with that of the flow pulse. With progression towards the periphery the impedance curves are shifted upwards and to the right because the flow becomes less pulsatile and smaller, but the pressure does not.

Fig. 8 shows the close agreement between impedance data from one animal and a curve for the Windkessel model, calculated from the aortic time constant according to equation 2.

Fig. 9 shows an example of pressure and diameter pulses recorded simultaneously at site 4. The shapes of the pressure and diameter waves are very similar because the radial displacement of the artery wall is correlated with pressure changes rather than flow. Peak aortic diameter at this location ranged from 2.0 to 2.3mm, with diastolic values from 1.8 to 2.1mm at mean blood pressure of 2.9 kPa. Recordings were also made at pressure levels that occasionally ranged as high as 4.0 kPa and as low as 1.3 kPa. In Fig. 10 the dynamic elastic modulus is plotted as a function of frequency at several mean pressures. The curves derived from in vivo pulses and those calculated from forced sinusoidal oscillations in vitro show similar trends, i.e. E_dyn increases with pressure, but is relatively insensitive to increasing frequency. Results from the in vivo and in vitro tests are in close agreement up to about 2 kPa, but diverge somewhat at higher pressures. These differences probably arise from the different methodologies employed in the two types of test, and we believe that the results in Fig. 10 support the premise that the mechanical properties of the isolated aortic segment are essentially the same as those of the intact aorta in vivo. Phase differences between pressure and diameter harmonics (after correction for the VDA response) were typically from 5 to 15°, but showed no apparent frequency or pressure dependence. Values of c, calculated from both in vivo and in vitro determinations of E_dyn, ranged from 2.08 to 2.46ms⁻¹ and from 2.23 to 2.95ms ⁻¹, respectively (Table 1).

This study demonstrates the direct effects of aortic elasticity on haemodynamic relationships in the arterial system of a typical poikilothermic vertebrate. The size and shape of the pressure pulse, the pressure wave velocity, the smoothing of the flow pulse as it travels distally and the hydraulic impedance of the arterial tree are all strongly influenced by the dynamic elastic modulus of the arterial wall. The combination of these elasticity-based properties with the relatively low heart rate and short arterial length results in wave transmission characteristics that are quite different from those of the mammalian circulation and, consequently, can be described by the simple Windkessel model. Thus, the aorta acts essentially as a single capacitance element with uniform elastic properties, and inflation by the heart occurs almost simultaneously throughout. In general, these characteristics should be found in any circulatory system where the transit time of the pressure pulse through the arterial tree is a negligible fraction of the cardiac cycle (Langille and Jones, 1977). In this regard, the critical haemodynamic parameters are the pressure-wave velocity, the length of the arterial tree and the frequency of the heartbeat.

Pressure-wave velocity can be calculated from the physical properties of the vessel and the fluid in it (equation 4), but it is often determined empirically from the time difference between proximal and distal pressure waves in vivo (McDonald, 1974). We compared these methods and found that the values of c calculated from in vivo and in vitro measurements of E_dyn were very similar to the values of c′ determined from proximal and distal pressure pulses (Table 1). This is an important result because it demonstrates that the properties of isolated segments measured in vitro do represent the in vivo elastic behaviour of the aorta and, furthermore, that the elasticity of the vessel wall indeed determines the wave velocity, according to the theoretical relationship of equation 1. Our measurements of E_dyn from in vivo pressure and diameter pulses are the only ones reported for any non-mammalian vertebrate. Arterial pressure wave velocities in the range of 2–3 m s⁻¹ (Table 1) are also typical of other poikilothermic vertebrates (Shelton and Jones, 1968; Langille and Jones, 1977; Gibbons and Shadwick, 1989), but somewhat lower than that observed in the mammahan aorta, particularly at its more distal portions. The pulse velocity for mammalian species ranges from 4.10 to 4.70 ms⁻¹ in the thoracic aorta and increases distally due to the ‘elastic taper’ that is characteristic of the mammalian aorta (Taylor, 1973; Avolio et al. 1976; Cox, 1975).

The average length of the arterial tree, from the aortic arch to the sciatic artery, was 11 cm in a 350 g toad. Assuming that the major reflection site is the abdominal bifurcation, the transit time of a pulse through the aorta would be only 0.05 s, or approximately 3% of the cardiac cycle (Table 2). This means there is time for up to 30 reflections of a pulse within one heartbeat period. The pressure amplitude will diminish rapidly as the wave travels because energy is lost to blood and artery wall viscosity, and because the reflection is incomplete, such that each wave is essentially attenuated before the next begins. Therefore, wave reflections from one pulse do not significantly affect the size or shape of any successive pulses, and wave transmission effects such as peripheral amplification, distortion or secondary peaks in diastole are not evident (Fig. 2). This results in the pressure waveforms at proximal and distal sites being nearly identical. This conclusion is in agreement with those of previous haemodynamic studies on other poikilothermic vertebrates (Shelton and Jones, 1968; Shelton, 1970; Shelton and Burggren, 1976; Burggren 1977a; Langille and Jones, 1977; Avolio et al. 1983; Gibbons and Shadwick, 1989) and invertebrates (Shadwick et al. 1987, 1990). It is interesting to note that viscous damping in the aorta (due to wall and blood viscosity), reduces the elastic efficiency of the system, but is an important feature of the Windkessel because it helps attenuate the reflected pressure waves, thus preventing resonance from occurring.

The lack of wave propagation effects in a Windkessel results in an impedance spectrum that is a continuously decreasing function of frequency (Fig. 8), where Z remains above Z_c at the fundamental heart frequency (Fig. 6). In contrast, a transmission-line system generates oscillations in Z above and below Z_c, with the first minimum coming at a frequency where the distance to the major reflecting site (L) is one-quarter of the wavelength (λ) (Fig. 11). This occurs in mammals at about the fundamental frequency (Milnor, 1982), but in poikilotherms the L/λ ratio at the fundamental frequency is very much less than 0.25 (Table 2; Gibbons and Shadwick, 1989), primarily because of their comparatively low heart rates. In the toad, for example, the quarter-wavelength frequency would occur at about 6 Hz, or the eighth harmonic, and this represents an insignificant component of the pressure pulse (Fig. 5). In comparison, the quarter-wavelength frequency for a 350 g rat would occur at about 7 Hz, which is nearly its resting heart rate. Although specific differences exist between the physical properties of the aorta in the toad and mammals, namely the presence of a geometric and elastic taper in the latter (Gibbons and Shadwick, 1991), these alone do not determine whether their arterial systems function as Windkessels or transmission Unes. The most important haemodymanic difference between the toad and similar-sized mammals that influences L/λ, and thus determines whether the aorta acts as a Windkessel or a transmission line, is the heart rate.

Taylor (1964) proposed that the elastic taper of the mammalian aorta acts like a series of wave reflection sites distributed along the length of the vessel. The effect of this feature in a transmission line system is to amplify incident waves, attenuate reflected waves and increase the impedance oscillations which, in turn, reduce the work of the heart (Taylor, 1965,1966a,b; O’Rourke and Taylor, 1967). In the toad circulation, where transmission effects do not dominate, an elastic taper of the aorta would appear to be unnecessary, and does not occur (Gibbons and Shadwick, 1991).

In addition to being the major blood conduit from the ventricle, the aorta in poikilotherms, like the toad, and in mammals is designed to provide a major pulsesmoothing effect. Specific differences in the structure and connective tissue composition of the artery wall have evolved to suit each to the particular pressure range encountered. Major differences in the haemodynamic properties in these two groups arise not because of functionally different aortic elastic properties but primarily because of very different heart rates.

This work was supported by a Graduate Research Grant from the University of Calgary to C.A.G. and by an operating grant and Research Fellowship from the Natural Science and Engineering Research Council of Canada to R.E.S.