A method for determining baroreflex-mediated sympathetic and parasympathetic control of the heart in pregnant and non-pregnant sheep

Abstract

1. The cardiac baroreflex was measured in four non-pregnant and six pregnant ewes before and during β-adrenoreceptor blockade with propranolol and before and during vagal blockade with atropine. Arterial pressure was raised by phenylephrine and lowered by sodium nitroprusside. The relationships between mean arterial pressure (MAP) and heart rate (HR), between MAP and heart rate variability (HRV) measured as the coefficient of variation (c.v.) of the mean pulse interval (PI), and between MAP and HRV measured by power spectral analysis were determined.

2. The MAP-HR relationship showed that in pregnant ewes the gain of the cardiac baroreflex was reduced when compared with non-pregnant ewes. Threshold and saturation pressures were higher, maximum achievable HR was lower and there was a decrease in the operating range.

3. V-shaped relationships were obtained between MAP and HRV (measured as the c.v. of PI) and between MAP and power spectral density in the frequency range 0.04–0.08 Hz. Using selective autonomic blockade the negative, or downward, slope of the V shape was shown to be a measure of baroreceptor-induced, sympathetically mediated effects on HRV. The upward, or positive, slope of the V shape was a measure of baroreceptor-induced, vagally mediated effects. Similar results were also obtained from the cardiac power spectrum, but it was less sensitive. The MAP at which the two slopes intersected was the same as the resting MAP.

4. In pregnant ewes, the slope of the downward limb of the V-shaped relationship between HRV (when measured as the c.v. of PI) and MAP was less than in non-pregnant ewes.

5. The relationship between MAP and the coefficient of variation of the mean pulse interval can therefore be used to measure the degree to which baroreceptor-induced sympathetic and parasympathetic activity affects the heart.

6. The resting MAP is the pressure at which the net effect of these sympathetic and parasympathetic influences on the heart is at a minimum. Studies of both the MAP-HR and MAP-HRV relationships in pregnant and non-pregnant sheep show that in pregnant sheep, there is attenuation of baroreceptor-mediated sympathetic effects on the heart.

A number of clinical and experimental studies have examined the effects of pregnancy on the cardiac baroreflex. The results obtained vary. For example, it has been reported that baroreflex sensitivity was decreased during pregnancy in women (Souma et al. 1983), in pregnant ewes (Ismay et al. 1979; Keller-Wood, 1995) and in pregnant rabbits (Cumbee et al. 1986). In other studies, however, it was found to be increased in women (Leduc et al. 1991), or unchanged in pregnant sheep (Magness & Rosenfeld, 1988). We therefore decided to compare baroreflex-mediated effects on the control of the heart rate of pregnant and non-pregnant ewes, using three different methods of analysis. By selective autonomic blockade we also measured the extent to which the sympathetic and parasympathetic nerves were responsible for baroreceptor effects on the heart.

First we examined the arterial pressure-heart rate reflex by fitting an equation described by Kent et al. (1972) to the values for mean arterial pressure (MAP) and heart rate (HR), both obtained as mean values from 100 s records. Second, using the same 100 s records we determined the relationship between MAP and heart rate variability (HRV) measured as the coefficient of variation (c.v.) of the mean pulse interval (PI). Finally, using the same records we measured the relationship between MAP and HRV from changes in power in the frequency range 0.04-0.08 Hz of the cardiac power spectrum. Arterial pressure was raised by phenylephrine and lowered by sodium nitroprusside.

Our results show that by using the simple relationship between the c.v. of the mean PI and MAP, it is possible to derive the relative sensitivities of baroreceptor-mediated sympathetic and parasympathetic cardiac effects. While a similar relationship can be derived from the relationship between spectral power in the frequency range 0.04-0.08 Hz and MAP, this method is less sensitive. By two of these three methods (the HR-MAP relationship and HRV measured as the c.v. of PI) we have shown that in pregnancy there is attenuation of baroreceptor-mediated sympathetic effects on the heart.

METHODS

These experiments were approved by the Animal Care and Ethics Committee, University of New South Wales. Six pregnant and four non-pregnant chronically catheterized ewes were used. Gestation age was 128-138 days.

Surgical preparation

Surgery was performed under aseptic conditions, after the ewe had fasted for 16 h. Anaesthesia was induced with i.v. sodium thiopentone (1.5 g; Pentothal, Abbott) and maintained with 2-3% halothane (Fluothane, ICI) in oxygen. Catheters were placed in the trachea for monitoring respiration, in a femoral artery for measurement of blood pressure and in both femoral veins for sampling of blood and infusion of drugs. In pregnant ewes fetuses were also cannulated using methods described previously (Lumbers & Stevens, 1983). Procaine penicillin (600 mg) and dihydrostreptomycin sulphate (750 mg) were given i.m. Ewes were housed in metabolic cages. The ambient temperature was 18-23°C. They were given free access to 6 l of water and lucerne chaff and oats. Daily fluid and food intake and urine output were measured. Arterial and venous catheters were flushed with heparinized saline (100 U ml⁻¹) daily. No experiments were performed until at least 5 days after surgery.

Experimental protocol

Maternal arterial and tracheal pressures were continuously monitored using Utah Medical Disposable transducers connected to Neomedix preamplifiers (Neomedix Systems, Sydney, Australia) and a Yokogama ORP 1200 8-channel recorder (Tech-Fast, Sydney) with a paper speed of 2 mm s⁻¹. At appropriate times, 100 s records of arterial and tracheal pressure were logged using an IBM compatible PC (486) and a Metrabyte DAS16 card (Keithley, MA, USA).

Control observations were collected for 30 min. Then arterial pressure was raised or lowered by i.v. phenylephrine (0.1-0.4 mg min⁻¹; Neosynephrine, Winthrop) or i.v. sodium nitroprusside (0.5-1.4 mg min⁻¹; Mulgrave, Victoria, Australia). In each ewe, data were collected at between six and eight different levels of mean arterial pressure. Arterial pressure was maintained for 3-5 min at each level and records were taken during the last 100 s when pressure was steady.

After a 2 h recovery period, this experiment was repeated during cardiac vagal blockade with atropine (Astra Pharmaceuticals, Sydney) at 4 mg i.v., 10 min before an experiment, followed by a continuous infusion at 24 mg h⁻¹ and repeated injections of 1.2 mg prior to each 100 s record. The experiment was also repeated again during β-adrenoreceptor blockade with propranolol (Inderal, ICI). The ewe was given an injection of 20 mg propranolol 10 min before an experiment and a continuous infusion of 30 mg h⁻¹. In each ewe, eight to ten 100 s records of the different levels of arterial pressure were obtained during atropine or propranolol treatment. These doses of atropine and propranolol block cardiac vagal and cardiac sympathetic activity, respectively (Ismay et al. 1979). Parasympathetic and sympathetic blockades were performed on different days, allowing 24 h between each experiment. At the end of an experiment, ewes were killed by i.v. injection of 20 ml pentobarbitone sodium (350 mg ml⁻¹, Lethobarb, Vibac, Australia).

Analysis of data

Pulse interval (PI) was calculated from the time between each maximum of the arterial pressure wave. A peak picking method was used to derive the PI from the arterial pressure record. From each 100 s record the mean PI and its standard deviation (s.d.) were calculated. Because the s.d. is dependent on the magnitude of the mean (Zar, 1984), the coefficient of variation (c.v.;%) of the mean PI was calculated and used as an index of the variability in heart rate that occurred over the 100 s period (HRV). Heart rate (HR) was determined from the mean PI.

The relationship between MAP and HR was determined using the following equation (Kent et al. 1972):

(1)

where A₁ is the range of HR (beats min⁻¹), A₂ is the slope or sensitivity of the relationship (1/mmHg), A₃ is the mid-point (in mmHg), i.e. the point from which equal pressor and depressor responses can occur, and A₄ is the minimum HR. Derivations of this equation (Korner, 1989) use MAP rather than systolic pressure to determine arterial pressure and HR relationships.

The gain of the baroreflex was calculated from the first derivative of eqn (1), and the threshold pressure (the lowest pressure that produces a significant decline in heart rate) and saturation pressure (the pressure necessary to achieve maximal inhibition of heart rate) were calculated from the third derivative of eqn (1) (Kent et al. 1972).

Each 100 s record of beat-to-beat intervals was interpolated into an 8192 point wave, suitable for fast Fourier transform and determination of HRV in the frequency domain. In a previous study we have shown in adult and fetal sheep that the same spectra are obtained when beat-to-beat intervals are used as when R-R intervals are used (Yu et al. 1998). A rectangular window was used for calculating the power spectrum. Results were exported to an Excel spreadsheet for further analysis. Power spectral density (PSD) was expressed in arbitrary units.

Respiration was measured from the fluctuations in the tracheal pressure record. To find out if this provided a record of the rate at which the animal breathed, the breathing rate was checked by watching the ewe and counting her respirations. The animals breathed at very fast rates, 78 breaths min⁻¹, because they were in a warm environment of 18-23°C, which is above their thermoneutral zone. This respiratory frequency was then used to look for a corresponding peak in the cardiac power spectrum (1.2-1.4 Hz). There was no prominent peak in the cardiac power spectrum at this frequency in either pregnant or non-pregnant ewes.

From preliminary experiments and to relate our data to values obtained by others, the following three frequency ranges were studied in each cardiac power spectrum: 0.04-0.08 Hz, called low low frequency (LLF); 0.08-0.15 Hz, called low frequency (LF); and that frequency range in the heart rate spectrum identical to the respiratory frequency, called high frequency (HF). For each frequency range, in each experiment, there were a similar number of points in the power spectrum, i.e. in LLF, 4 ± 1 samples, in LF, 7 ± 2 samples and in HF, 9 ± 2 samples were collected. The power in each range was the sum of these values. Total power was the sum of the power in the three ranges.

Statistical analysis

Data are presented as means ±s.e.m. Student's paired t test was used to compare values obtained before and after autonomic blockade. Student's non-paired t test was used to compare data obtained in non-pregnant ewes with data from pregnant ewes. The relationships between MAP and HRV and between MAP and PSD were determined by linear regression analysis (Zar, 1984).

RESULTS

MAP, HR and HRV at rest

Arterial P_O2 was higher (P < 0.05) and P_CO2 lower in pregnant ewes than in non-pregnant ewes (P < 0.05; Table 1). There were no changes in either pregnant or non-pregnant ewes in arterial P_O2, P_CO2 and pH during treatment with atropine or propranolol.

Table 1. Arterial pH, P_O2, P_CO2, MAP, HR, HRV and PSD in LLF, LF and HF in control, after cardiac vagal blockade and after β-adrenoreceptor blockade in 4 non-pregnant and 6 pregnant ewes.

	Non-pregnant ewes			Pregnant ewes
	Control	Atropine	Propranolol	Control	Atropine	Propranolol
pH	7.4 ± 0.01	7.4 ± 0.05	7.4 ± 0.05	7.4 ± 0.06	7.38 ± 0.05	7.4 ± 0.05
PO2(mmHg)	98.5 ± 1.2	96 ± 0.8	99 ± 0.8	110.8 ± 1.7†	110 ± 1.7	109 ± 1.7
PCO2(mmHg)	46.7 ± 1.8	46 ± 1.3	47 ± 1.8	37.5 ± 0.8†	36.8 ± 0.3	34.5 ± 5.7
MAP(mmHg)	102 ± 0.6	128 ± 2.7**	100 ± 0.8	94 ± 0.4†	107 ± 4.6*	95 ± 0.7
HR(beats min−1)	83.5 ± 1.7	113 ± 2.5***	65 ± 2.4*	104.5 ± 3.0††	141 ± 5**	62.6 ± 1.3***
HRV(%)	6.8 ± 0.2	2.8 ± 0.6***	5.1 ± 0.5*	4.5 ± 0.1††	2.5 ± 0.1**	3.0 ± 0.05*
PSD LLF	215 ± 6.1	2.4 ± 0.4***	195 ± 7.8	185 ± 14	3.6 ± 0.5***	168 ± 30*
PSD LF	136 ± 5.4	1.5 ± 0.2***	175 ± 36.8	86.5 ± 18††	2.5 ± 0.8***	91.6 ± 5
PSD HF	87.8 ± 2.7	3.5 ± 0.8***	72 ± 10.5	24.4 ± 2.8††	4.5 ± 0.5**	28.5 ± 5.3
PSD total	434 ± 10.4	6.5 ± 1.2***	445 ± 53.6	276.5 ± 15.2††	10.5 ± 2.4***	283 ± 37.4

Arterial pH, P_O2, P_CO2, mean arterial pressure (MAP), heart rate (HR), heart rate variability (HRV, measured as the c.v. of PI) and power spectral density (PSD, expressed as arbitrary units) in LLF (0.04–0.08 Hz), LF (0.08–0.15 Hz) and HF (about 1.2 Hz) were determined in 4 non-pregnant and 6 pregnant ewes in control, after cardiac vagal blockade (with atropine) or after β-adrenoreceptor blockade (with propranolol). In these animals, HF corresponded to the respiratory frequency measured from the tracheal pressure record. Values are means ±s.e.m.

^*P < 0.05

^**P < 0.01

^***P < 0.001 compared with control values.

^†P < 0.05

^††P < 0.01 compared with values obtained in non-pregnant ewes.

HR was higher (P < 0.01) and MAP and HRV (measured as c.v. of PI) were lower in pregnant than in non-pregnant ewes (P < 0.05 and P < 0.01, respectively; Table 1). In both non-pregnant and pregnant ewes, MAP increased during treatment with atropine (P < 0.01and P < 0.05, respectively) but was not affected by propranolol (Table 1). HR increased during cardiac vagal blockade (P < 0.001and P < 0.01) and decreased during β-adrenoreceptor blockade (P < 0.05and P < 0.001; Table 1).

HRV (measured as c.v. of PI) was significantly reduced by both atropine (P < 0.001 and P < 0.01, respectively) and propranolol (P < 0.05 and P < 0.05, respectively; Table 1).

In both non-pregnant and pregnant ewes the main peak in the cardiac power spectrum was in the frequency range of 0.04-0.08 Hz (PSD LLF; Table 1). However, total power was significantly less in pregnant ewes (P < 0.01) because PSDs in LF and HF were less than in non-pregnant ewes (P < 0.01 and P < 0.01, respectively; Table 1).

Power in all three frequency ranges was decreased after atropine in both non-pregnant (P < 0.001, P < 0.001 and P < 0.001 for LLF, LF and HF, respectively) and pregnant ewes (P < 0.001, P < 0.001 and P < 0.01; Table 1). During treatment with propranolol, power in LLF was reduced in pregnant ewes (P < 0.05; Table 1) but not in non-pregnant ewes.

The MAP-HR relationship

This was measured in four non-pregnant and six pregnant ewes. Individual and mean data for gain, threshold, mid-point and saturation pressures, maximum and minimum HRs, and operating ranges are shown in Table 2. The effects of β-adrenoceptor blockade with propranolol on these parameters are also listed in Table 2. Compared with non-pregnant ewes, pregnant ewes had a lower gain (P < 0.001, Table 2A)and higher threshold, saturation and mid-point pressures (P < 0.01, P < 0.05 and P < 0.01, respectively; Table 2A). Overall, the operating range was decreased by 9 mmHg (P < 0.05; Table 2B).Pregnant ewes could not achieve such high HRs when MAP was lowered to threshold pressure (P < 0.001; Table 2B).Also HRs did not fall to as low a level as those measured in non-pregnant ewes when MAP reached saturation pressure (P < 0.001; Table 2B and Fig. 1).

Table 2. Characteristics of cardiac baroreflexes in 4 non-pregnant and 6 pregnant ewes.

A.	Control				Propranolol
	Gain	Threshold(mmHg)	Saturation(mmHg)	Mid-point(mmHg)	Gain	Threshold(mmHg)	Saturation(mmHg)	Mid-point(mmHg)
Non-pregnant
Ewe 1	−1.8	56	128	92	−2.4	65.3	110	88
Ewe 2	−1.8	55	130	92	−2.1	74.5	100	87
Ewe 3	−1.7	59	135	97	−2.1	75.0	105	90
Ewe 4	−2.1	58	120	89	−2.4	57.0	84	71
Means ± S.E.M	−1.85 ± 0.1	56.8 ± 1.1	128 ± 3	92 ± 1.7	−2.3 ± 0.2*	69 ± 5.3*	99 ± 5.6**	84 ± 4.4
Pregnant
Ewe 1	−1.3	73	138	105	−1.45	65	102	83
Ewe 2	−1.2	72	136	104	−1.57	72	120	96
Ewe 3	−1.25	70	134	101	−1.50	70	125	98
Ewe 4	−1.44	68	136	102	−1.73	87	125	105
Ewe 5	−1.2	80	138	108	−1.43	86	122	104
Ewe 6	−1.33	87	138	112	−1.34	80	111	96
Means ± S.E.M.	−1.3 ± 0.04†††	75 ± 3††	137 ± 0.7†	104 ± 1.7††	−1.5 ± 0.06*	77 ± 3.7	117 ± 3.7**	97 ± 3.2

B.	Control			Propranolol
	Maximum HR(beats min−1)	Minimum HR(beats min−1)	Operating range(mmHg)	Maximum HR(beats min−1)	Minimum HR(beats min−1)	Operating range(mmHg)
Non-pregnant
Ewe 1	162	45	72	133	49	45
Ewe 2	156	42	75	98	53	26
Ewe 3	160	36	76	94	52	30
Ewe 4	170	51	62	94	49	27
Means ± S.E.M	162 ± 3	43 ± 3	71 ± 3.2	104 ± 9.5**	50 ± 1.2	32 ± 4.4**
Pregnant
Ewe 1	136	60	65	100	52	37
Ewe 2	121	62	63	118	51	48
Ewe 3	142	57	64	112	60	55
Ewe 4	137	53	68	103	49	38
Ewe 5	116	59	58	104	53	36
Ewe 6	120	63	51	97	60	31
Means ± S.E.M.	129 ± 4.5†††	59 ± 1.5†††	62 ± 2.5†	106 ± 3.2**	54 ± 2	40 ± 3.6***

Data were obtained before and during β-adrenoreceptor blockade (with propranolol).

^*P < 0.05

^**P < 0.01 compared with control values.

^†P < 0.05

^††P < 0.01

^†††P < 0.001 compared with values obtained in non-pregnant ewes. Gain was expressed as beats min⁻¹ mmHg⁻¹. Threshold, threshold pressure; Saturation, saturation pressure.

Figure 1. The HR-MAP relationship in 4 non-pregnant and 6 pregnant ewes.

•, non-pregnant ewes; ○, pregnant ewes. Values are means ±s.e.m.

In both non-pregnant and pregnant ewes, β-adrenoceptor blockade caused an increase in gain (P < 0.05 and P < 0.05, respectively; Table 2A).The saturation pressures were lower (P < 0.01 and P < 0.01, respectively; Table 2A) and the threshold pressure was greater but only in non-pregnant ewes (P < 0.05; Table 2A).

During β-adrenoceptor blockade the maximum achievable HRs of both non-pregnant and pregnant ewes (P < 0.01 and P < 0.01, respectively; Table 2B)and operating ranges were reduced (P < 0.01 and P < 0.001, respectively; Table 2B) but the minimum achievable HRs did not change.

The MAP and HR relationships described above were abolished in both groups by atropine.

Effects of the MAP on HRV (measured as c.v. of PI)

The relationship between MAP and HRV (c.v. of PI) in both non-pregnant and pregnant ewes was V-shaped (Figs 2A, 3A and 4).

Figure 2. Slopes of the relationships between MAP and HRV in 4 non-pregnant ewes in control, during β-adrenoreceptor blockade and during vagal blockade.

MAP-HRV relationships (means ±s.e.m.) are shown in 4 non-pregnant ewes under control conditions (A) and during selective β-adrenoreceptor blockade (B; slope = 0.07 ± 0.01% mmHg⁻¹, r²= 0.97, P < 0.001) or vagal blockade (C; slope 1 = -0.14 ± 0.02% mmHg⁻¹, r²= 0.89, P < 0.001; slope 2 = -0.03 ± 0.003% mmHg⁻¹, r²= 0.82, P < 0.01).

Figure 3. Slopes of the relationships between MAP and HRV in 6 pregnant ewes in control, during β-adrenoreceptor blockade and during vagal blockade.

MAP-HRV relationships (means ±s.e.m.) are shown in 6 pregnant ewes under control conditions (A) and during selective β-adrenoreceptor blockade (B; slope = 0.07 ± 0.01% mmHg⁻¹, r²= 0.98, P < 0.001) or vagal blockade (C; slope 1 = -0.067 ± 0.004% mmHg⁻¹, r²= 0.96, P < 0.001; slope 2 = -0.027 ± 0.001% mmHg⁻¹, r²= 0.88, P < 0.01).

Figure 4. V-shaped relationship between MAP and HRV in 4 untreated non-pregnant ewes and 6 untreated pregnant ewes.

The slopes (negative and positive) of the relationships between MAP and HRV superimposed on the V-shaped relationships between MAP and HRV (measured as c.v. of PI) obtained in each non-pregnant ewe prior to any autonomic blockade (A) and in each untreated pregnant ewe (B).

During β-adrenoceptor blockade, only a positive relationship between MAP and HRV (c.v. of PI) was found (Figs 2B and 3B), which was the same in non-pregnant (0.07 ± 0.01% mmHg⁻¹, Fig. 2B)and pregnant ewes (0.07 ± 0.01% mmHg⁻¹; Fig. 3B and Table 3A). During β-blockade, only cardiac vagal tone can affect HRV, and therefore this relationship measures the degree to which the baroreceptors influence HRV through the vagus.

Table 3. Slopes of the relationships between MAP and HRV (A) and between MAP and power in LLF (B) in 4 non-pregnant and 6 pregnant ewes during cardiac vagal blockade (with atropine) or β-adrenoreceptor blockade (with propranolol).

	Atropine				Propranolol
	Slope 1(% mmHg−1)	r2	Slope 2(% mmHg−1)	r2	Slope(% mmHg−1)	r2
A. Slopes of MAP–HRV relationships
Non-pregnant
Ewe 1	−0.10	0.99	−0.04	0.78	0.05	0.88
Ewe 2	−0.18	0.90	−0.03	0.80	0.05	0.82
Ewe 3	−0.14	0.88	−0.03	0.82	0.10	0.90
Ewe 4	−0.13	0.80	−0.03	0.85	0.09	0.92
Means ± S.E.M.	−0.14 ± 0.02	0.89 ± 0.04	−0.03 ± 0.003	0.82 ± 0.015	0.07 ± 0.01	0.88 ± 0.02
Pregnant
Ewe 1	−0.070	0.95	−0.028	0.85	0.05	0.80
Ewe 2	−0.068	0.92	−0.030	0.90	0.05	0.85
Ewe 3	−0.070	0.97	−0.027	0.92	0.06	0.82
Ewe 4	−0.070	0.88	−0.025	0.95	0.07	0.90
Ewe 5	−0.067	0.90	−0.026	0.80	0.10	0.78
Ewe 6	−0.065	0.95	−0.025	0.84	0.09	0.85
Means ± S.E.M.	−0.067 ± 0.004††	0.93 ± 0.01	−0.027 ± 0.002	0.87 ± 0.023	0.07 ± 0.01	0.83 ± 0.03
B. Slopes of MAP–PSD relationships
Non-pregnant
Ewe 1	−3.2	0.93	−1.15	0.85	13.0	0.87
Ewe 2	−3.2	0.90	−1.1	0.91	12.0	0.90
Ewe 3	−3.7	0.96	−1.1	0.82	15.0	0.90
Ewe 4	−4.1	0.95	−1.05	0.81	11.3	0.85
Means ± S.E.M.	−3.6 ± 0.2	0.93 ± 0.01	−1.1 ± 0.02	0.85 ± 0.02	12.8 ± 0.8	0.88 ± 0.02
Pregnant
Ewe 1	−3.20	0.95	−0.80	0.85	6.3	0.84
Ewe 2	−3.10	0.91	−0.70	0.83	7.2	0.86
Ewe 3	−1.85	0.93	−0.63	0.88	4.5	0.82
Ewe 4	−2.50	0.97	−0.67	0.97	8.0	0.88
Ewe 5	−2.80	0.92	−0.60	0.93	7.4	0.90
Ewe 6	−1.35	0.99	−0.62	0.93	7.3	0.87
Means ± S.E.M.	−2.6 ± 0.25†	0.94 ± 0.01	−0.67 ± 0.03†††	0.89 ± 0.22	6.8 ± 0.5†††	0.86 ± 0.02

Power in LLF was at 0.04–0.08 Hz.

^†P < 0.05

^††P < 0.01

^†††P < 0.001 compared with values obtained from non-pregnant ewes. Values were derived using linear regression analysis.

During vagal blockade with atropine, only negative relationships between MAP and HRV were found (Figs 2C and 3C). Before any form of blockade, in both pregnant and non-pregnant ewes, the negative or downward slope of the V shape (Figs 2A and 3A) was best described by a single relationship, the slope of which was -0.04 ± 0.006 in non-pregnant and -0.06 ± 0.01 in pregnant ewes. Inspection of the values in Figs 2C and 3C suggests, however, that there are two components to the relationship between MAP and HRV after atropine. One slope describes the MAP-HRV relationship at pressures of < 100 mmHg; mean slopes were -0.067 ± 0.004% mmHg⁻¹ for pregnant and -0.14 ± 0.02% mmHg⁻¹ for non-pregnant ewes (Table 3A). The other slope describes the relationships between MAP and HRV at pressures ≥ 100 mmHg. In pregnant ewes it was -0.027 ± 0.002% mmHg⁻¹; in non-pregnant ewes, this slope was -0.03 ± 0.003% mmHg⁻¹ (Table 3A).

After atropine the slope relating lower arterial pressures to HRV was significantly less in pregnant compared with non- pregnant ewes (P < 0.01; cf. Fig. 3C with Fig. 2C). There was no difference in the two groups of animals between the slopes relating higher pressures to HRV. In the presence of atropine only cardiac sympathoadrenal control of the heart is present. Therefore these two negative slopes measure the degree to which baroreceptors influence HRV-mediated changes in sympathetic activity. The two negative slopes were obtained under different conditions: the slope of low MAP-HRV was obtained when MAP was lowered by i.v. sodium nitroprusside and the slope of high MAP-HRV was obtained when MAP was raised by i.v. phenylephrine.

In order to compare the intersection of the negative and positive slopes obtained after atropine and propranolol, respectively, with the nadir of the V-shaped relationship obtained under resting conditions, and also to compare the sensitivities of HRV measured by the c.v. of PI and by power spectrum analysis (see below), all data obtained after atropine treatment were combined and a single linear relationship derived for each animal (Fig. 4). Table 4 shows that the point at which this single negative slope intersects with the positive slope obtained after propranolol is the same as the resting MAP. The mean combined negative slope obtained after atropine was much greater than the downward slope of the V-shaped relationship found in the same animals before blockade for non-pregnant ewes (-0.12 ± 0.01% mmHg⁻¹ after blockade compared with -0.04 ± 0.006% mmHg⁻¹ before blockade; P < 0.001) but not different in pregnant ewes (-0.06 ± 0.01% mmHg⁻¹ after compared with -0.06 ± 0.005% mmHg⁻¹ before). Propranolol had no effect on the positive relationships between HRV and MAP.

Table 4. HRV and MAP at the point of intersection of the two limbs of the V-shaped relationship between MAP and HRV (measured after selective autonomic blockade) with control in non-pregnant and pregnant ewes.

	HRV(%)	MAP(mmHg)	Control MAP(mmHg)
Non-pregnant
Ewe 1	6.5	98	102
Ewe 2	6.5	100	104
Ewe 3	6.7	98	102
Ewe 4	6.4	100	98
Means ± S.E.M.	6.5 ± 0.06	99 ± 0.6	102.4 ± 1.2
Pregnant
Ewe 1	4.4	97	94
Ewe 2	4.5	95	92
Ewe 3	5.3	98	96
Ewe 4	4.8	94	92
Ewe 5	4.0	92	95
Ewe 6	4.1	93	92
Means ± S.E.M.	4.5 ± 0.2***	95 ± 1**	94 ± 0.6***

^**P < 0.01

^***P < 0.001 compared with non-pregnant ewes.

Effect of the MAP on power spectral density

The greatest power was measured in the frequency range 0.04-0.08 Hz (Table 1). Not surprisingly, there was also a V-shaped relationship between MAP and PSD in this range in both non-pregnant (Fig. 5A) and pregnant ewes (Fig. 6A). As with HRV (measured as c.v. of PI), this relationship became positive during propranolol treatment and negative during atropine treatment (Figs 5B and C and 6B and C). As with HRV, the negative relationship between PSD in LLF and MAP found after atropine could be resolved into two slopes.

Figure 5. Slopes of the relationships between MAP and PSD at 0.04-0.08 Hz in control and during β-adrenoceptor and vagal blockade in non-pregnant ewes.

The slopes (means ±s.e.m.) of the relationships between MAP and PSD (expressed as arbitrary units) at 0.04-0.08 Hz in control (A), during β-adrenoceptor blockade (B; slope = 12.8 ± 0.8, r²= 0.98, P < 0.001) and during vagal blockade (C; slope 1 = -3.6 ± 0.2, r²= 0.94, P < 0.001; slope 2 = -1.1 ± 0.02, r²= 0.80, P < 0.01) in 4 non-pregnant ewes.

Figure 6. Slopes of the relationships between MAP and PSD at 0.04-0.08 Hz in control and during β-adrenoceptor and vagal blockade in pregnant ewes.

The slopes (means ±s.e.m.) of the relationships between MAP and PSD (expressed as arbitrary units) at 0.04-0.08 Hz in control (A), during β-adrenoceptor blockade (B; slope = 6.8 ± 0.5, r²= 0.96, P < 0.001) and during vagal blockade (C; slope 1 = -2.6 ± 0.25, r²= 0.97, P < 0.001; slope 2 = -0.67 ± 0.03, r²= 0.97, P < 0.001) in 6 pregnant ewes.

After atropine, at pressures < 100 mmHg, the mean slope was -3.6 ± 0.2 (Table 3B)for non-pregnant ewes and -2.6 ± 0.25 for pregnant ewes (P < 0.05; Table 3B).At pressures ≥ 100 mmHg the slope was -1.1 ± 0.02 for non-pregnant ewes and -0.67 ± 0.03 for pregnant ewes (Table 3B; P < 0.001). After propranolol, positive slopes were less in pregnant than in non-pregnant ewes (P < 0.001; Table 3B and Figs 5B and 6B). In the two higher frequency ranges studied, no consistent results were obtained.

In one ewe the effects of combined treatment with atropine and propranolol on the PSD-MAP relationship was studied. PSD in LLF was 345 at resting MAP; it was 8 at 45 mmHg and 10 at 150 mmHg. Therefore combined blockade abolished both the positive and negative slopes of the PSD-MAP relationships.

Comparison of HRV (c.v. of PI) and PSD-derived estimates of baroreflex-mediated sympathetic and parasympathetic effects on the heart

Comparison of Figs 2C and 3C with Figs 5C and 6C suggests that in both groups of ewes, the negative slopes of the relationships between HRV (measured as the c.v. of PI) during treatment with atropine were steeper than the same slopes measured from power spectrum analysis. These negative slopes measure sympathetically mediated baroreflex effects on HRV. To confirm that these slopes were less when HRV was derived from the cardiac power spectrum, the ratios of negative to positive slopes, i.e. sympathetic nervous system: parasympathetic nervous system (SNS/PNS × 100) derived from each HRV-MAP relationship were compared with the ratios of the two slopes derived from each PSD-MAP relationship. As Table 5 shows, this ratio was greater in both groups of animals when HRV measured as the c.v. of PI, rather than HRV measured from the power spectrum was used (P < 0.01). In pregnant ewes the SNS: PNS ratios derived from the HRV (c.v. of PI)-MAP relationships were less than in non-pregnant ewes (P < 0.01). This difference was not detected when the ratios were derived from PSD-MAP relationships.

Table 5. Comparison of the ratio of sympathetic to parasympathetic baroreceptor-mediated changes in heart rate variability in non-pregnant and pregnant ewes.

	SNS: PNS using HRV (c.v. of PI)(%)	SNS: PNS using HRV(PSD)(%)
Non-pregnant
Ewe 1	200	18.5
Ewe 2	167	19.2
Ewe 3	250	18.7
Ewe 4	144	21.0
Means ± S.E.M.	190 ± 23**	19.2 ± 0.6
Pregnant
Ewe 1	120	15.8
Ewe 2	100	13.2
Ewe 3	116	20.0
Ewe 4	89	7.5
Ewe 5	50	21.6
Ewe 6	83	31.5
Means ± S.E.M.	93 ± 10.4***††	18.7 ± 3.3

The ratios of sympathetic (SNS) to parasympathetic (PNS) baroreceptor-mediated changes in heart rate variability (HRV) (SNS/PNS × 100) were measured using either the coefficient of variation (c.v.) of the pulse interval (PI) or from the cardiac power spectrum (PSD).

^**P < 0.01

^***P < 0.001 compared with values obtained using PSD.

^††P < 0.01 compared with values obtained in non-pregnant ewes.

DISCUSSION

There were two reasons why HR was determined from the arterial pressure record. In other studies, we wanted to make comparisons between baroreflex responses of fetal sheep with those of adult sheep, and fetal ECG is likely to be contaminated with maternal ECG. Secondly, we have found it extremely difficult to obtain ECG recordings in which the baseline was sufficiently stable and where there were consistently large enough R waves for detection.

In a previous study (Yu et al. 1998), the slope of the relationship between PI and the R-R interval was 1.0 in both adult and fetal sheep (r²= 0.98 and r²= 0.99, respectively) and the differences in power spectral densities determined from PI and the R-R interval in adult sheep were 0.12% in the range 0.04-0.08 Hz, 0.5% in the range 0.08-0.15 Hz, 0.054% in the range 0.6-0.08 Hz and 0.2% in total power.

The relationships between MAP, HR and HRV were studied by infusion of drugs so that arterial pressure changed to a new level. We then waited 2-3 min after the pressure had changed, so that it was steady. The reason for this was to make sure that any contribution to the changes in HR mediated by changes in cardiac sympathetic activity was measured. It is well known that when arterial pressure is raised the initial rapid response in terms of HR is mediated by the vagus. It is accepted that there could be some adaptation of baroreceptors over this time, but all measurements were carried out in the same way in both groups of animals.

The apparent lack of sensitivity of PSD in detecting sympathetically mediated changes in HRV, as evidenced by the lesser slope of the relationship between PSD and MAP (measured after atropine) compared with the slope of the relationship between HRV and MAP (Figs 2C, 3C, 5C and 6C and Table 3)could possibly be due to the fact that we did not analyse frequencies < 0.03 Hz. Thus there may have been very slow oscillations < 0.03 Hz in sympathetic tone which were not measured by PSD, but would have contributed to the overall variability as measured by the c.v. in PI. As explained in Yu et al. (1998), we excluded frequencies < 0.03 Hz because of contamination by a DC component and because only 100 s records were taken.

The effects of pregnancy on the cardiac baroreflex of women (Souma et al. 1983) and sheep (Magness & Rosenfeld, 1998) and on renal sympathetic activity (RSNA) in the rat (Heesch & Rogers, 1995) have been well described. In general most authors have shown that the gain of the cardiac baroreflex is reduced in pregnancy (Ismay et al. 1979; Heesch & Rogers, 1995; Brooks et al. 1995). The present findings confirm this observation and describe in detail other changes. First, despite the fact that the resting arterial pressure was lower in pregnant compared with non-pregnant ewes, threshold and saturation pressures were higher and the operating range of the cardiac baroreflex was reduced. These changes could be related to effects on the baroreflex of the high levels of angiotensin (Lee et al. 1980).

The fact that the maximum achievable heart rate was reduced in pregnant ewes yet β-adrenoceptor blockade reduced it to the same level in both pregnant and non-pregnant ewes might suggest that baroreceptor-mediated sympathetic effects on the heart are attenuated, or else that the heart cannot respond as effectively. The latter is unlikely as the resting heart rate of pregnant ewes is higher, not lower.

Although these findings are not unique (Heesch & Rogers, 1995), inconsistent and variable descriptions of the effects of pregnancy on the cardiac baroreflex exist. In part this may be due to the way in which the cardiac baroreflex has been analysed. Korner et al. (1974) point out that PI-MAP relationships overemphasize the degree to which the cardiac baroreflex is mediated by the vagus, whereas HR-MAP relationships give a more balanced analysis of the role of the sympathetic and vagally mediated components of the cardiac baroreflex. Also, we waited 2-3 min at each level of MAP in order to ensure that the baroreflex-mediated sympathetic effects on HR were fully expressed (Korner et al. 1974). This would partly explain why in earlier studies from this laboratory using PI-MAP relationships the only significant effector of the cardiac baroreflex in the sheep was the vagus (Ismay et al. 1979), whereas in the present study we, like others, have been able to show a significant role of cardiac sympathetics in baroreflex-mediated changes in HR (Table 2 and Table 3). As well as this, we can see by using two different methods of analysis that the ability of the sympathoadrenal system to respond to unloading of the baroreceptors is attenuated in pregnant sheep. Using the HRV (measured as c.v. of PI)-MAP relationship we could also determine the extent to which baroreceptor-induced sympathetic activity influenced HRV and could see again that this effector limb of the baroreflex was weaker in pregnant ewes. The MAP-RSNA reflex is also weaker in the pregnant rabbit and rat (Crandall & Heesch, 1990; Brooks et al. 1995; Heesch & Rogers, 1995) and it is suggested that the renal sympathetic nerve response to unloading the baroreceptors is impaired. These investigators suggest that this change is mediated centrally by inhibitory neuroactive steroids (e.g. 3α-hydroxydihydroprogesterone (3αOH-5α-DHP)). This steroid is present in high levels in the circulation in pregnancy and has been shown to potentiate GABA-mediated increases in chloride conductance in vitro (Paul & Purdy, 1992). Since GABAergic influences in the rostroventral lateral medulla are responsible for sympathetic inhibition when the arterial baroreflex is activated (Guyenet, 1990), modulation of the GABA_A receptors by progesterone metabolites at this site would be likely to alter the properties of the baroreflex. Infusions of 3αOH-5α-DHP into rats reduce the maximum achievable level of RSNA and the slope of the arterial baroreflex (Heesch & Rogers, 1995).

We characterized the cardiac baroreflex using HR-MAP relationships obtained in pregnant and non-pregnant ewes in order to establish that it was functional in every animal studied. We then used the same data to see what effect if any the baroreceptors had on heart rate variability measured as the c.v. of PI and from the cardiac power spectrum. Numerous studies have attempted to examine the interactions between the baroreceptors and the cardiac power spectrum in exercise (Sleight, 1989), cardiac disease, (Casadei et al. 1996) and pregnancy (Ekholm & Ekkola, 1996). However this is the first study to take simple parameters such as the c.v. of PI and the level of MAP and determine the relationship between the two. Both time-domain and frequency-domain analysis of the variability in HR revealed V-shaped relationships with MAP. What was particularly exciting was that we could, using selective autonomic blockade, resolve this V shape into two linear components (Figs 2 and 3 and Table 3): a sympathetic component measured during vagal blockade with atropine and a vagal component measured during β-adrenoceptor blockade with propranolol Since combined vagal and sympathetic blockade abolished both limbs at the apex of the V, it is unlikely that the changes in HRV and PSD occurring in response to changes in MAP depended on other baroreceptor-mediated endocrine responses. There are a number of important consequences of this finding.

The negative relationship between MAP and HRV or between MAP and PSD (LLF) measured after atropine could be resolved into two linear components. It was much steeper at lower pressures, when sodium nitroprusside was used to lower MAP than at higher pressures, when phenylephrine was used to raise MAP. The reason why there were these two components is not clear. It could be related to the vasoactive drugs used to alter pressure. However, sodium nitroprusside, which was used to lower arterial pressure, is an NO donor. Therefore it would be expected to be associated with lesser, not greater sensitivity of the baroreflex, as NO appears to reduce the sensitivity of the arterial pressure-heart rate relationship (Liu et al. 1996; Murakami et al. 1998).

In control (i.e. untreated ewes) there was no evidence of two distinct components to the negative limb of the HRV-MAP relationship.This may be due to the fact that in control experiments, alterations in both vagal and sympathetic activity modulate HRV. Thus changes in vagal tone could affect the negative slope of the V shape. In fact this is likely because after atropine the negative slope obtained was greater than in control experiments in non-pregnant sheep.

The MAP at which the two slopes intersected (Table 4) was the same as the resting arterial pressure and at this pressure the net effect of the two efferent arms of the baroreflex on HRV (measured as c.v. of PI) was at a minimum (Fig. 4 and Table 4). This suggests that the resting MAP in an individual is that pressure at which net baroreceptor-induced PNS and SNS effects on the heart are lowest. Thus when arterial pressure is altered and there is baroreceptor resetting, the apex of the V-shaped relationship between HRV (measured as c.v. of PI) and MAP will be shifted to a higher or lower pressure. A change in the slope of one of the limbs making up the V shape would cause this shift.

Another advantage that we gained using the HRV (measured as c.v. of PI)-MAP relationships was that we could measure the dynamics of baroreflex-mediated sympathetic and parasympathetic effects on HRV, because we could determine the slope of these relationships, which is a measure of the rate of change or gain. It was therefore easy to see that pregnancy affected the gain of the sympathetic limb of the V shape, i.e. it was weaker in pregnant ewes when MAP was < 100 mmHg (Fig. 3C and Table 3) than in non-pregnant animals. Thus the SNS: PNS ratio was also weaker (Table 5). Like the HR-MAP relationship, therefore, the HRV (measured as c.v. of PI)-MAP relationship showed that pregnancy impaired baroreflex-induced sympathetically mediated control of the heart.

We obtained similar results using the cardiac power spectrum and studying the effects of MAP on power in the frequency range 0.04-0.08 Hz. However, one weakness of the PSD-MAP relationships that we measured (Figs 5 and 6) was that the difference in the negative (sympathetic) slope between pregnant and non-pregnant ewes was no longer apparent. This prompted us to compare the relative sensitivities of the time-domain and frequency-domain measurements of heart rate variability. As Table 5 shows, the higher values of the SNS:PNS ratios obtained from measurement of HRV-MAP relationships during selective blockade indicates that time-domain analysis is a more sensitive index of sympathetically mediated effects on HRV than is frequency-domain analysis. This increased sensitivity provided by time-domain analysis is also evident in Table 1, which shows that β-adrenoceptor blockade affects HRV measured as c.v. of PI but not when measured as PSD. We have made similar observations concerning the insensitivity of PSD as a measure of altered sympathetically mediated effects on HRV in fetal sheep (Yu et al. 1998).

In summary, we have compared three methods for studying baroreceptor-mediated influences on the heart rates of conscious pregnant and non-pregnant ewes. The classic HR-MAP relationship showed that the sympathetically mediated component of the cardiac baroreflex was attenuated in pregnant ewes. The inability of the pregnant ewe to increase her HR to the same extent as non-pregnant ewes suggests that in pregnancy, the range over which cardiac output can be increased by increasing HR is limited.

We have described an effect of the cardiac baroreflex on HRV using two methods of HRV measurement: the c.v. of the mean PI measured over 100 s, and the cardiac power spectrum. A striking V-shaped relationship was obtained with both methods. The two limbs of the V shape provided a measure of baroreflex-mediated sympathetic and parasympathetic effects on the heart. The baroreflex-mediated sympathetic effects on HRV and PSD were non-linear but parasympathetically mediated effects were linear. However, only the HRV (measured as c.v. of PI)-MAP relationship was sensitive enough to detect the effects of pregnancy on the heart that were also evident using the HR-MAP relationship. This method of analysing baroreflex-mediated effects on the heart also provided a measure of the extent to which they were mediated by the sympathetic and parasympathetic limbs of the reflex. Finally, it seems as though the resting arterial pressure was the pressure at which the net effect of the two autonomic cardiac effectors was at a minimum.