Short-term haemodynamic variability in the conscious areflexic rat

Abstract

1. Simultaneous measurements of arterial pressure and cardiac output (n = 8), mesenteric blood flow (n = 7) or hindquarters (n = 8) blood flow were performed during 1 h periods in conscious rats, before and after acute pharmacological blockade of the autonomic, renin-angiotensin and vasopressin systems. In the latter condition (areflexic state), arterial pressure was maintained with a continuous infusion of noradrenaline.

2. In the areflexic state, spontaneous fluctuations in arterial pressure were markedly exaggerated, especially depressor episodes. At the onset of these falls in arterial pressure, there was an abrupt and transient decrease in stroke volume and cardiac output. Systemic vasodilatation then developed while cardiac output returned to normal. Regional vasodilatations were also delayed from the onset of the falls in arterial pressure and were usually large enough to maintain blood flow.

3. Both time and frequency domain analyses confirmed that changes in systemic and regional vascular conductances lagged by about 1 s behind arterial pressure changes.

4. These results indicate that, in the absence of neurohumoral influences, autoregulatory-like mechanisms become dominant in the control of systemic and regional circulations and contribute to exaggeration of the spontaneous short-term variability of arterial pressure.

Exaggerated short-term variability of arterial pressure, commonly referred to as arterial pressure lability, is the main consequence of any intervention interrupting the arterial baroreceptor reflex loop, whether it concerns its afferent limb (sino-aortic baroreceptor denervation; Cowley, Liard & Guyton, 1973; Alper, Jacob & Brody, 1987) or its sympathetic efferent limb (chemical sympathectomy; Julien, Kandza, Barrès, Lo, Cerutti & Sassard, 1990). In both sino-aortic baroreceptor denervated and sympathectomized rats, depressor episodes, usually accompanying body movements, have been shown to make an important contribution to arterial pressure lability (Trindade & Krieger, 1984; Julien, Zhang & Barrès, 1993). These hypotensive episodes are associated with strong vasodilatations in the mesenteric and hindquarters circulations that could involve an autoregulatory component (Zhang, Julien, Gustin, Cerutti & Barrès, 1994; Zhang, Barrès & Julien, 1995). However, the interpretation of haemodynamic fluctuations in these models is complicated by the persistence of arterial pressure control systems. Specifically, in sino-aortic baroreceptor denervated rats, the central command of sympathetic nervous activity, which is left unopposed by baroreceptor inhibitory influences, probably interferes with the mechanisms producing arterial pressure lability (Barrès, Lewis, Jacob & Brody, 1992; Zhang, Julien & Barrès, 1996). In sympathectomized rats, the reflex vagal control of heart rate is preserved (Julien et al. 1990) and contributes to buffer arterial pressure fluctuations (Ferrari, Franzelli, Daffonchio, Perlini & DiRienzo, 1996). Moreover, in these animals arterial pressure is maintained primarily by the renin-angiotensin system (Julien et al. 1990; Lo et al. 1991), and it is likely that intermittent release of renin occurs during the spontaneous decreases in arterial pressure (Bertolino, Julien, Medeiros, Vincent & Barrès, 1994).

In this study, we examined whether acute pharmacological blockade of the major pressor systems controlling arterial pressure would unmask spontaneous autoregulatory responses and thereby exaggerate arterial pressure variability in conscious rats. In addition, the continuous monitoring of systemic and regional haemodynamics allowed us to delineate the relative contribution of changes in cardiac output and regional vascular conductances to the initiation and maintenance of spontaneous fluctuations in arterial pressure. As the extent of arterial pressure lability critically depends on the level of vascular tone (Jacob, Alper, Grosskreutz, Lewis & Brody, 1991; Julien et al. 1993), the arterial pressure level was maintained with a constant infusion of noradrenaline during neurohumoral blockade.

METHODS

Male Sprague-Dawley rats (Iffa-Credo, L'Arbresle, France) weighing 300-400 g were anaesthetized with sodium pentobarbitone (60 mg kg⁻¹i.p.) and were chronically instrumented for the measurement of either ascending aortic blood flow (referred to as cardiac output, n = 8), mesenteric blood flow (n = 7) or distal aortic blood flow (referred to as hindquarters blood flow, n = 8).

Chronic instrumentation

For the measurement of cardiac output, an ultrasonic transit-time flow probe (Model 2.5 SB; Transonic Systems, Ithaca, NY, USA) was placed around the ascending thoracic aorta through the right third intercostal space (Smith & Hutchins, 1979). The probe cable was passed through the thoracotomy and routed subcutaneously behind the right forelimb up to the back of the neck. The connector end of the flow probe was exteriorized at the time of catheter implantation (see below). Rats were given penicillin G (100 000 i.u. i.p.) 3 days post-operatively and were allowed 10-15 days to recover.

For the measurement of regional blood flows, a miniaturized 20 MHz pulsed Doppler flow probe (DBF-120A-XS; Crystal Biotech, Holliston, MA, USA) was implanted around the superior mesenteric artery or the lower abdominal aorta (Zhang et al. 1994, 1995). The wires of the probe were led subcutaneously to the nape of the neck and their free ends were embedded in dental cement and left under the skin until catheterization. Rats were treated with a single injection of penicillin G (60 000 i.u. i.p.) and were given 8-15 days to recover.

Subsequently, rats were re-anaesthetized with halothane (1.5-2 % in oxygen), and polyethylene catheters were inserted into the lower abdominal aorta and inferior vena cava via the left femoral artery and vein, respectively (Zhang et al. 1995). Catheters were led under the skin to exit between the scapulae. The electrical wires of the Doppler probe were then exteriorized and soldered to a connector plug. The connector end of either the cardiac or the Doppler probe was protected in a small cap sewn to the skin. Antibiotic (neomycin sulphate) was applied topically.

Experimental protocol

After catheterization, rats were placed in large individual recording cages. Two days later, experiments were performed in conscious freely moving rats accustomed to their environment, and provided with food and water ad libitum. The arterial catheter was connected to a precalibrated pressure transducer (Statham P23 ID; Gould, Cleveland, OH, USA). Instantaneous cardiac output (ml min⁻¹) was measured with an ultrasonic transit-time flowmeter (Model T106; Transonic Systems). Indices of regional blood flows (kHz of Doppler shift) were monitored with a directional pulsed Doppler flowmeter (Department of Bioengineering, University of Iowa, Iowa City, IA, USA). Phasic pressure and flow outputs were fed simultaneously to an amplifier-recorder (Model 8802; Gould) and to a personal computer (486 DX 2/66) equipped with an analog-to-digital converter board (AT-MIO-16; National Instruments, Austin, TX, USA). Using LabVIEW 3.1.1 software (National Instruments), data were sampled every 2 ms and stored on CD-ROM.

The recording sessions consisted of: (1) a 60 min baseline period initiated when the animals were quiet and cardiovascular parameters had stabilized, (2) a 30 min period initiated 5 min after neurohumoral blockade, and (3) a 60 min period initiated after arterial pressure had been restored to its basal level with a continuous infusion of noradrenaline. The latter condition was referred to as the areflexic state. Neurohumoral blockade was achieved by the combined administration of the ganglionic blocker chlorisondamine chloride (2.5 mg kg⁻¹; Ciba Geigy), the peripheral cholinoceptor antagonist atropine methyl nitrate (2 mg kg⁻¹; Sigma), the angiotensin-converting enzyme inhibitor perindopril (2 mg kg⁻¹; Laboratoires Servier, Neuilly-sur-Seine, France) and the selective V₁-vasopressin receptor antagonist [β-mercapto-β,β-cyclopentamethylenepropionyl¹, O -Me-Tyr²,Arg⁸]-vasopressin (10 μg kg⁻¹; Sigma). These drugs were administered as an i.v. bolus injection (250 μl kg⁻¹). The arterial pressure level was restored with a constant i.v. infusion of noradrenaline (0.4-1 μg kg⁻¹ min⁻¹). Infusion flow rates ranged from 0.4 to 1.1 ml h⁻¹. Administration of blockers was repeated after 2 h, if necessary, so as to ensure complete blockade across the whole study (Lo et al. 1991). At the end of the experiments animals were killed with an i.v. overdose of sodium pentobarbitone.

Data analysis

Beat-to-beat time series were generated and processed off-line on a work station (SPARC 1; Sun Microsystems, Mountain View, CA, USA). For each cardiac cycle, the computer calculated mean arterial pressure, heart rate, cardiac output and stroke volume. All time series were synchronized. Total and regional vascular conductances were estimated as the ratio of cardiac output or mean Doppler flow to mean arterial pressure. For all parameters, the overall variability was defined as the variation coefficient.

To characterize combinations of mean arterial pressure and blood flows, three-dimensional frequency distributions were constructed by plotting the frequency of occurrence against the data pairs. Results are presented as contour plots.

Temporal relationships between mean arterial pressure and either total or regional vascular conductances were statistically evaluated in both the time and frequency domains. In the time domain, a Spearman rank correlation procedure was applied to each of the consecutive 60 s time segments constituting the 1 h recording periods. Mean values for both positive and negative correlation coefficients were calculated separately and multiplied by their respective frequency of occurrence. These weighted coefficients were computed for variable delays introduced between parameters (from -20 to 20 cardiac beats). Group means were then calculated and plotted against the time delay so that cross-correlation functions were obtained (Zhang et al. 1994, 1995). Regarding correlations obtained in the areflexic state, a polynomial regression was applied to the data points. The delay yielding the best correlation coefficient was calculated using the best fitting polynomial.

In the frequency domain, phase angles between arterial pressure and vascular conductances were calculated using cross-spectral techniques, as previously described in detail (Cerutti, Barrès & Paultre, 1994; Julien, Zhang, Cerutti & Barrès, 1995). In brief, discrete time series were generated from beat-to-beat data by computing interpolated values every 100 ms. For a 1 h recording period, 34 data sets of 2048 points (204.8 s) overlapping by half were processed. The frequency resolution was therefore 0.005 Hz. The coherence function, ranging from 0 to 1, was estimated using an overlapped Fourier transform processing. The coherence quantifies the reliability of phase estimates and its significance threshold was 0.2 in this study (Benignus, 1969). The phase function was bounded between -π and π radians. A positive phase at a given frequency indicates that fluctuations of mean arterial pressure precede those of vascular conductance, with respect to the out-of-phase pattern. The calculation of pure time delays between pressure and conductances was based on the identification of linear segments in the phase functions and calculation of their slope by means of linear regression analysis (Saul, Berger, Albrecht, Stein, Chen & Cohen, 1991). The analysis was restricted to the frequency range 0.005-0.2 Hz. This window included all fluctuations of a period between 5 and 200 s, which is consistent with the duration of the haemodynamic events responsible for most of the arterial pressure lability in areflexic rats (see Results).

Results are presented as means ±s.e.m. Statistical comparisons between the intact and areflexic states were performed using the Wilcoxon signed-rank test.

RESULTS

Table 1 summarizes the effects of neurohumoral blockade combined with noradrenaline infusion on mean values and variability of cardiovascular parameters recorded in conscious unrestrained rats. In the areflexic state, mean arterial pressure, stroke volume and total peripheral conductance values were similar to those observed in the intact state. However, cardiac output was decreased by 14 ml min⁻¹ and this could be attributed particularly to a significant decrease in heart rate. Blood flows and vascular conductances in both regional circulations did not significantly differ from basal values. The spontaneous variability of mean arterial pressure was not affected by neurohumoral blockade alone (data not shown) but was more than doubled during noradrenaline infusion. This lability of mean arterial pressure was associated with a marked increase in the variability of total peripheral conductance and, to a lesser extent, with an increase in the variability of cardiac output and of stroke volume. As expected, the variability of heart rate tended to decrease. However, slow changes in heart rate were sometimes observed during neurohumoral blockade alone or when noradrenaline infusion was combined with neurohumoral blockade. These trends probably partly offset the decrease in the short-term variability of heart rate. Variabilities of regional blood flows and vascular conductances were not altered in the areflexic state.

Table 1.

Effect of neurohumoral blockade combined with noradrenaline infusion (areflexic state), on mean values and short-term variability of haemodynamic parameters in conscious rats

	Mean		Variation coefficient (%)
	Intact	Areflexic	Intact	Areflexic
Group I (n = 8)
MAP (mmHg)	109 ± 3	108 ± 2	4·5 ± 0·4	12·3 ± 1·2 **
HR (beats min−1)	363 ± 13	330 ± 10 **	5·1 ± 0·9	2·4 ± 0·3 **
CO (ml min−1)	94 ± 5	80 ± 5 *	7·0 ± 0·5	8·4 ± 0·3 **
SV (μl)	260 ± 16	244 ± 15	6·6 ± 0·6	8·0 ± 0·4 *
TPC (ml min−1 mmHg−1)	0·86 ± 0·05	0·76 ± 0·05	8·0 ± 0·7	13·7 ± 1·3 **
Group II (n = 7)
MAP (mmHg)	102 ± 4	99 ± 5	4·8 ± 0·3	10·1 ± 0·6 **
HR (beats min−1)	385 ± 8	310 ± 11 **	4·6 ± 0·5	3·6 ± 0·6
MeBF (kHz)	3·9 ± 1·0	3·9 ± 1·0	13·7 ± 1·7	12·1 ± 1·4
MeC × 1000 (kHz mmHg−1)	39·8 ± 10·8	40·9 ± 11·0	15·2 ± 1·9	14·3 ± 1·2
Group III (n = 8)
MAP (mmHg)	107 ± 2	107 ± 2	4·5 ± 0·2	11·6 ± 0·5 **
HR (beats min−1)	376 ± 7	322 ± 16 **	5·4 ± 0·4	3·8 ± 0·6
HqBF (kHz)	2·8 ± 0·2	3·9 ± 0·8	15·5 ± 1·4	14·1 ± 2·1
HqC × 1000 (kHz mmHg−1)	26·4 ± 1·8	37·3 ± 7·6	15·9 ± 1·3	16·9 ± 1·9

Rats were chronically instrumented for the measurement of either cardiac output (group I), mesenteric blood flow (group II), or hindquarters blood flow (group III). Values are means ± s.e.m.

^*P < 0·05

^**P < 0·02 compared with Intact. MAP, mean arterial pressure; HR, heart rate; CO, cardiac output; SV, stroke volume; TPC, total peripheral conductance; MeBF, mesenteric blood flow; MeC, mesenteric vascular conductance; HqBF, hindquarters blood flow; HqC, hindquarters vascular conductance.

Examples of continuous recordings of systemic haemodynamic variables are presented in Fig. 1. In the intact state, the low variability of arterial pressure was associated with a low variability of cardiac output and total peripheral conductance. Sharp decreases in stroke volume occasionally occurred, but were opposed by concomitant increases in heart rate so that cardiac output did not change. In the areflexic state, arterial pressure became unstable, mainly due to the occurrence of depressor episodes that seemed to be associated with large systemic vasodilatations, as indicated by the sharp increases in total peripheral conductance. However, cardiac output also exhibited marked fluctuations which were secondary to fluctuations of stroke volume because heart rate did not change concomitantly.

Figure 1. Representative recordings of cardiovascular variables.

Digital reconstruction of original tracings obtained during 10 min of spontaneous activity in the same conscious rat before (Intact) and after acute pharmacological blockade of the major pressor systems combined with noradrenaline infusion (Areflexic). Beat-to-beat values were averaged over 1.2 s consecutive periods. PAP, pulsatile arterial pressure; MAP, mean arterial pressure; HR, heart rate; CO, cardiac output; SV, stroke volume; TPC, total peripheral conductance.

The frequency distributions of mean arterial pressure and blood flow value pairs (Fig. 2) were elongated vertically and horizontally relative to the flow axis, in the intact and areflexic states, respectively. The horizontal stretching of the distribution in the areflexic state indicates that cardiac output and regional blood flows were relatively insensitive to changes in mean arterial pressure, and so is suggestive of flow autoregulation. However, it should be noted that some low blood flow values in these animals were associated with low mean arterial pressure values.

Figure 2. Contour plots of three-dimensional frequency distributions.

Plots were constructed from 60 min beat-to-beat recordings performed in 3 conscious rats (top, middle and bottom pairs of graphs), each studied in both intact and areflexic states. First contour lines were set at 0.1, 0.3, 0.6 and 1 % followed by 1-2 % increments.

Figure 3 shows representative beat-to-beat recordings of haemodynamic variables during depressor episodes that accompanied a change in body position in areflexic rats. At the onset of the decrease in mean arterial pressure, there was an abrupt and transient drop in cardiac output (Fig. 3A). Because heart rate did not change concomitantly, this decrease in cardiac output could be attributed entirely to a decrease in stroke volume. Total peripheral conductance showed an immediate transient decrease and then increased while cardiac output returned to normal. In both regional circulations, there was an initial, transient decrease in blood flow when mean arterial pressure started to drop (Fig. 3B and C). Vasodilatation then developed and blood flow returned to normal in each vascular bed or even increased while mean arterial pressure remained lowered.

Figure 3. Changes in haemodynamic variables during depressor episodes in conscious areflexic rats.

Tracings are beat-to-beat data collected over 30 s in 3 different rats instrumented for the measurement of cardiac output (A), mesenteric (B) or hindquarters (C) blood flow. MAP, mean arterial pressure; HR, heart rate; CO, cardiac output; TPC, total peripheral conductance; MeBF, mesenteric blood flow; MeC, mesenteric vascular conductance; HqBF, hindquarters blood flow; HqC, hindquarters vascular conductance. In each case, the MAP trace is the heavy line.

In both intact and areflexic states, strong negative correlations between mean arterial pressure and either total (Fig. 4A) or regional (Fig. 4B and C) vascular conductances were observed, thus confirming that pressure and conductance changes were generally in opposite directions. Whatever the circulation and the experimental condition, the percentage occurrence of significant positive correlations was usually less than 1 %. These correlations are therefore not shown on the graphs. In the intact state, the best negative correlation coefficients were obtained at about zero delay, pointing to the synchronicity of these opposite changes. In the areflexic state, the best correlation coefficients were obtained for positive delays, indicating that changes in mean arterial pressure preceded changes in vascular conductances. As a fast oscillation (probably secondary to respiration-related fluctuations) was superimposed on the cross-correlation function, a polynomial fitting was applied to the correlation coefficients. In the systemic circulation, the optimum delay calculated from individual fitted curves was 5.1 ± 1.1 cardiac beats, i.e. 951 ± 212 ms, as calculated from the average heart rate value of each rat. In the mesenteric circulation, the best correlation coefficients were obtained at about zero delay in two of seven rats. In the other five rats, the optimum delay was 5.6 ± 1.5 cardiac beats, i.e. 1058 ± 260 ms. In the hindquarters circulation, the optimum delay was observed at 4.6 ± 0.6 cardiac beats, i.e. 889 ± 124 ms.

Figure 4. Average cross-correlation functions between mean arterial pressure and vascular conductances.

Spearman rank correlation coefficients between mean arterial pressure and either total peripheral conductance (A; n = 8), mesenteric (B; n = 7) or hindquarters (C; n = 8) vascular conductance were computed as a function of the time delay (cardiac beats) in the intact (^) and areflexic (•) states. Data points are means ±s.e.m. Correlation coefficients reach the P < 0.001 significance level at values < -0.19. Vascular conductances are the dependent variables and are placed in advance of mean arterial pressure for negative delays.

Average results of cross-spectral analysis between mean arterial pressure and vascular conductances are presented in Fig. 5. In the systemic circulation of intact rats (Fig. 5A), coherence was highest in the frequency range 0.1-0.2 Hz, corresponding phase angles tending to π radians. In the areflexic state, coherence was maximum in the very low frequency range and tended to decline from 0.05 Hz, whereas the phase function decreased over the whole frequency band. In each rat, linear regression analysis was applied to phase values associated with a significant coherence (> 0.2) and confirmed that phase angles decreased linearly (R = 0.81 ± 0.06; 28 ≤n ≤41; P < 0.0008). The average slope of these linear segments was 7.0 ± 1.5 rad Hz⁻¹, which indicates the presence of a pure time delay of 1117 ± 236 ms, total peripheral conductance lagging behind mean arterial pressure. Regarding the relations of mean arterial pressure to mesenteric vascular conductance (Fig. 5B), coherence was relatively stable in the very low frequency range and then started to decrease between 0.1 and 0.15 Hz in both intact and areflexic states. The phase function first showed a tendency to decrease linearly that was significant in four of seven rats in the intact condition (R = 0.68 ± 0.03; 18 ≤n ≤27; P < 0.0028) and in five of seven rats in the areflexic condition (R = 0.76 ± 0.07; 21 ≤n ≤41; P < 0.0043). The average slopes were 4.4 ± 0.2 and 5.6 ± 2.0 rad Hz⁻¹, which corresponded to pure time delays of 693 ± 32 and 885 ± 312 ms in the intact and areflexic states, respectively. Then, phase angles declined steeply towards null or even negative values. In the hindquarters circulation (Fig. 5C), coherence between vascular conductance and mean arterial pressure was rather low in both intact and areflexic conditions. Phase function did not show any significant linear trend in the intact state and phase values were close to π radians between 0.05 and 0.2 Hz. In the areflexic state, linear sections were identified in seven of eight rats (R = 0.76 ± 0.05; 17 ≤n ≤41; P < 0.0017). The average slope was 6.7 ± 1.0 rad Hz⁻¹, corresponding to a fixed time delay of 1068 ± 155 ms.

Figure 5. Frequency domain analysis of relationships between mean arterial pressure and vascular conductances.

Graphs depict average coherence and phase spectra computed between mean arterial pressure and either total peripheral conductance (A; n = 8), mesenteric (B; n = 7) or hindquarters (C; n = 8) vascular conductance. Means (heavy lines) ±s.e.m. (fine lines) are shown.

DISCUSSION

The present experiments demonstrate that arterial pressure lability can easily be induced in conscious rats by acutely interrupting the reflex control of arterial pressure with neurohumoral blockade, while maintaining the arterial pressure level with a continuous infusion of noradrenaline. This first observation confirms a previous study by Jacob et al. (1991) who reported that ganglionic blockade combined with either captopril or a V₁-vasopressin receptor antagonist induced arterial pressure lability in conscious rats when vasoconstrictor agents were infused to restore the arterial pressure level. The latter authors also noted that under these conditions, the spontaneous variability of vascular resistances in the hindquarters, renal or mesenteric circulations was not consistently increased. However, when the sum of these individual resistances was calculated, an increased variability was demonstrated. In the areflexic rats of the present study, arterial pressure lability was associated with an exaggerated variability of total peripheral conductance whereas the variability of the hindquarters or mesenteric vascular conductances was unchanged, compared with the intact state. This observation strongly suggests that the spontaneous variations in regional vascular conductances were synchronous, thereby contributing to the increase in the variability of the total peripheral conductance. This interpretation would accord with our previous findings in sympathectomized rats (Zhang et al. 1994). In these animals, the variabilities of hindquarters and mesenteric conductances were not increased, but were strongly positively correlated. By contrast, the variations seen in the regional vascular conductances of intact rats were only weakly related or even inversely related, so giving a low variability of subdiaphragmatic aortic conductance, an approximate index of total peripheral conductance.

As in the sympathectomized rat, and to a lesser degree, in the sino-aortic baroreceptor denervated rat (Julien et al. 1993), the arterial pressure lability seen in the areflexic rat was mainly due to the frequent occurrence of depressor episodes. Although the influence of behaviour was not studied quantitatively, visual surveillance of the animals during the recording sessions indicated that movements and postural changes invariably induced large falls in arterial pressure. The continuous recording of cardiac output allowed us to observe that there was an abrupt and transient decrease in stroke volume at the very beginning of these depressor events. As the autonomic control of the heart was interrupted, it is likely that decreases in cardiac filling rather than decreases in myocardial contractility were responsible for these drops in stroke volume. The mechanism of this haemodynamic transient is unclear. Abrupt decreases in stroke volume have been reported to occur in humans during the few heartbeats after the start of a dynamic exercise or immediately after standing up (Adams, Guz, Innes & Murphy, 1987; Sprangers, Wesseling, Imholz, Imholz & Wieling, 1991), whereas the reduction of stroke volume during a passive head-up tilt is much more progressive (Sprangers et al. 1991). Usually, the increase in heart rate that accompanies, or even precedes, the decrease in stroke volume is sufficient to maintain cardiac output and arterial pressure (Sprangers et al. 1991). Interestingly, a similar haemodynamic pattern, i.e. a decrease in stroke volume preceded by an increase in heart rate, has been demonstrated in sino-aortic baroreceptor denervated dogs during the few seconds required to stand fully from a supine position (Cowley et al. 1973), which indicates that the chronotropic changes are not reflexly mediated but rather depend on central command. Similarly, in conscious, intact rats, heart rate and stroke volume changes are usually inversely related, as indicated by the cross-correlation function computed between these two variables (Janssen, Oosting, Slaaf, Persson & Struijker-Boudier, 1995). It would therefore not be surprising if decreases in stroke volume, which are usually inconsequential in normal rats, were to have a profound impact on cardiac output in areflexic rats, because of the lack of a concomitant rise in heart rate. Indeed, this would partly explain the 20 % increase in the spontaneous variability of cardiac output in areflexic rats. In addition, a significant increase in the variability of stroke volume was observed in the areflexic state. This could be due, at least in part, to an amplification of the respiration-linked fluctuations (data not shown).

At the onset of the arterial pressure decreases, there was generally a brief sharp decrease in total peripheral conductance which coincided with the drop in cardiac output (Fig. 3). This decrease in conductance seems too abrupt to reflect an arteriolar vasoconstriction, and is more probably the result of the Windkessel buffering properties of the large compliant arteries, especially the aorta (Toorop, Westerhof & Elzinga, 1987). Thereafter, systemic vascular conductance increased progressively, thus amplifying and prolonging the fall in arterial pressure. Examination of the haemodynamic behaviour of mesenteric and hindquarters circulations during depressor episodes indicated that regional vasodilatations were also delayed from the onset of the arterial pressure decreases. These selective observations therefore suggest the following sequence of haemodynamic events: an initial drop in cardiac output initiates a fall in arterial pressure, which is then maintained by a large systemic vasodilatation, as a result of synchronous vasodilatation in the regional circulations. Because these vasodilatations were usually large enough to maintain cardiac output and regional blood flows (Fig. 2), they may be attributed to autoregulatory responses of the resistance vessels.

To assess the reproducibility of the temporal relationships between arterial pressure and vascular conductances, cross-correlation analysis was applied to 60 s time segments, so allowing us to explore haemodynamic changes taking place within this time frame (Zhang et al. 1994, 1995). In the intact condition, strong negative correlations were maximum for a null delay, which demonstrated the direct effect of vasomotor changes on arterial pressure. As these correlation coefficients rapidly attenuated when delays were introduced between variables, it seems reasonable to propose that the haemodynamic changes responsible for these correlations were of a short duration. It is likely that synchronous rhythmic fluctuations in vascular conductances and arterial pressure, such as those centred at 0.4 Hz (Janssen et al. 1995; Julien et al. 1995), contributed to the observed correlations. This contribution was especially evident in the mesenteric circulation. In this case, the cross-correlation function exhibited a clear oscillation of a period of 15-17 cardiac beats which, considering the average heart rate of the rats, corresponded to approximately 2.5 s. By contrast, in the areflexic state, the best correlation coefficients were obtained for positive delays, indicating that changes in arterial pressure preceded changes in vascular conductances.

The precise determination of optimum delays was sometimes impeded by the presence of a fast oscillation superimposed on the cross-correlation function. For this reason, we used cross-spectral analysis to confirm the estimated time lags between changes in vascular conductances and arterial pressure. To achieve an acceptable resolution, the 1 h beat-to-beat time series were split into 205 s time segments after equidistant sampling at 10 Hz (Cerutti et al. 1994). Linear sections of the phase function were identified, which permitted the calculation of fixed time delays (Saul et al. 1991). This calculation assumed a strict mechanical coupling between arterial pressure and vascular conductances for fluctuations of an infinite duration, i.e. a phase of π radians at 0 Hz: in most cases, this was indeed verified for the first frequency component. Phase estimates were considered only when corresponding coherence values were significant. This could appear to be a limitation of the method because coherence analysis assumes linearity between variables, which is not necessarily the case between arterial pressure and vascular conductances (O'Leary, 1991). This possibly led us to underestimate the strength of the association between pressure and conductance changes but, even under these conditions, linear portions of the phase function could be reliably identified. Indeed, we confirmed that, in the intact state, fluctuations in total peripheral conductance were not delayed relative to fluctuations in arterial pressure, whereas, in the areflexic state, changes in conductance lagged behind changes in arterial pressure by about 1 s. Interestingly, the delays calculated from cross-spectral analysis were close to those obtained from cross-correlation analysis, indicating that in this particular experimental setting, the time and frequency domain approaches provide the same qualitative and quantitative information. Therefore, we can conclude that delayed changes in total peripheral conductance form the haemodynamic basis of arterial pressure lability in the areflexic rat. A similar haemodynamic pattern was identified in the subdiaphragmatic aortic circulation of chronically sympathectomized rats (Zhang et al. 1994).

Regarding the hindquarters circulation, coherence and phase functions were roughly similar to those calculated in the systemic circulation. Vascular conductance and arterial pressure fluctuations were out of phase in the intact state, whereas in the areflexic state, changes in arterial pressure were reflected in vascular conductance after a delay of approximately 1 s. This latter observation points to a strict autoregulatory-like behaviour, and is at variance with what we observed in the hindquarters circulation of sympathectomized rats where non-autoregulatory vasodilatations were often present (Zhang et al. 1994). The possibility that our pharmacological protocol interrupted a vasodilator pathway remains to be investigated.

In the mesenteric circulation, the picture was somewhat different. In both intact and areflexic states, there was a linear trend in the phase function up to a break-off point at 0.12-0.13 Hz. Afterwards, phase angles declined precipitously whereas coherence values decreased gradually. This behaviour would be consistent with the characteristics of a second-order physical system with a resonance frequency (Gille, Decaulne & Pélegrin, 1992), and would lead to the appearance of a regular oscillation centred at the resonance frequency. Such an oscillation has indeed been described in the mesenteric circulation of conscious intact rats at the frequency of 0.13 Hz, and in the resting state, this oscillation has been found to account for more than one-third of the total variance of the mesenteric blood flow (Janssen et al. 1995). In the mesenteric vascular conductance of areflexic rats, a slow oscillation (period of 8-9 s) was also observed. It was frequently triggered or amplified by the drops in arterial pressure (Fig. 3). Overall, these observations suggest that autoregulatory mechanisms prevail in the mesenteric circulation of the rat at low frequencies, irrespective of neurohumoral influences. At higher frequencies, conductance fluctuations display a passive character in the areflexic state, phase angles tending to 0 (Fig. 5). In the intact state, phase angles eventually decrease to -π radians (data not shown). Oscillations of mesenteric vascular conductance are then driven by the sympathetic nervous system in the frequency band around 0.4 Hz (Julien et al. 1995).

In summary, the results of the present study support the conclusion that in the absence of neurohumoral influences, autoregulatory-like mechanisms become dominant in the short-term control of regional and systemic circulations. Pressure autoregulation of blood flow has long been recognized as an important factor in volume-dependent forms of hypertension (Guyton, Granger & Coleman, 1971), and the conscious areflexic rat has successfully been used to quantify autoregulatory responses to changes in blood volume (Hinojosa-Laborde, Greene & Cowley, 1988; Hinojosa-Laborde, Frohlich & Cowley, 1991). However, as far as we know, there have been no previous studies describing the spontaneous occurrence of short-term autoregulatory changes in the systemic or regional circulations of conscious animals, although in some (Persson et al. 1993), but not all (Skarlatos et al. 1993), studies an autoregulatory-like pressure-flow pattern was suggested in the renal circulation of conscious dogs. The latency of autoregulatory responses suggested by the present experiments is much shorter than that measured in vitro (VanBavel, Giezeman, Mooij & Spaan, 1991) or in vivo in anaesthetized animals (Borgdorff, 1983). We propose that blood flow autoregulation becomes an important source of short-term arterial pressure variability when the reflex control of arterial pressure is interrupted. This observation has potentially important clinical implications, especially in patients suffering from autonomic disorders.