Acute hypoxia increases heart rate (HR) and cardiac output (Qt) at a given oxygen consumption (V̇O2) during submaximal exercise. It is widely believed that the underlying mechanism involves increased sympathetic activation and circulating catecholamines acting on cardiac β receptors. Recent evidence indicating a continued role for parasympathetic modulation of HR during moderate exercise suggests that increased parasympathetic withdrawal plays a part in the increase in HR and Qt during hypoxic exercise. To test this, we separately blocked the β-sympathetic and parasympathetic arms of the autonomic nervous system (ANS) in six healthy subjects (five male, one female; mean ±s.e.m. age = 31.7 ± 1.6 years, normoxic maximal V̇O2 (V̇O2,max) = 3.1 ± 0.3 l min−1) during exercise in conditions of normoxia and acute hypoxia (inspired oxygen fraction = 0.125) to V̇O2,max. Data were collected on different days under the following conditions: (1)control, (2) after 8.0 mg propranolol I.V. and (3) after 0.8 mg glycopyrrolate I.V. Qt was measured using open-circuit acetylene uptake. Hypoxia increased venous [adrenaline] and [noradrenaline] but not [dopamine] at a given V̇O2 (P < 0.05, P < 0.01 and P= 0.2, respectively). HR/V̇O2 and Qt/V̇O2 increased during hypoxia in all three conditions (P < 0.05). Unexpectedly, the effects of hypoxia on HR and Qt were not significantly different from control with either β-sympathetic or parasympathetic inhibition. These data suggest that although acute exposure to hypoxia increases circulating [catecholamines], the effects of hypoxia on HR and Qt do not necessarily require intact cardiac muscarinic and β receptors. It may be that cardiac α receptors play a primary role in elevating HR and Qt during hypoxic exercise, or perhaps offer an alternative mechanism when other ANS pathways are blocked.
During progressive dynamic exercise in normoxia, heart rate and cardiac output increase with increasing workload until maximal oxygen consumption (V̇O2,max) is reached. During light exercise, an increase in heart rate secondary to vagal (parasympathetic) withdrawal and an increase in venous return mediated by the leg muscle pump are considered to increase cardiac output (Janicki et al. 1996). Parasympathetic withdrawal is held to be largely complete at a heart rate above 120-130 beats min−1 (Janicki et al. 1996). At progressively heavier levels of exercise, further elevations in heart rate and cardiac output are thought to be mediated by increased sympathetic nervous system tone, which increases both heart rate and venous return (Janicki et al. 1996). During acute hypoxia, submaximal heart rate and cardiac output are greater at the same absolute workload than in normoxic conditions, and the slopes of the cardiac output and heart rate versus oxygen consumption (V̇O2) relationship are increased compared to normoxia (Stenberg et al. 1966; Grover et al. 1967). Maximum heart rate and cardiac output are unchanged, but occur at a lower V̇O2,max and workload (Stenberg et al. 1966). It is widely believed that the mechanism underlying this alteration in cardiac function is increased sympathetic activation brought about by hypoxia. In particular, stimulation of cardiac β receptors by increased circulating catecholamines is thought to be important (Moore et al. 1986).
The main evidence for the role of increased sympathetic nervous system activation in the cardiovascular changes described above is that at a given workload, plasma catecholamine levels are increased in hypoxic exercise compared to normoxic exercise. In keeping with this, subjects who are exposed to acute hypoxia and treated with propranolol, a non-selective β-blocking agent, demonstrate a reduced cardiac output and heart rate during submaximal exercise, compared to subjects treated with a placebo (Wolfel et al. 1998). Control of heart rate is a result of the balance between sympathetic and parasympathetic activity, and increased parasympathetic withdrawal could, in part, explain some of the elevation in heart rate observed during acute hypoxia. Recently, it has been appreciated that during normoxic exercise there is evidence both from the spectral analysis of heart rate variability (Saito & Nakamura, 1995; Tulppo et al. 1998) and from direct electrophysiological measurements (Kannankeril & Goldberger, 2002) for continued parasympathetic modulation of heart rate during moderate exercise. The role of the parasympathetic nervous system during acute hypoxic exposure is controversial, and it has been reported that parasympathetic activity is increased (Hartley et al. 1974), decreased (Richalet et al. 1985) and unchanged (Hammill et al. 1979). The combined effects of parasympathetic activity and acute hypoxic exercise on heart rate and cardiac output have not been established. However, there is evidence for the roles of both increased parasympathetic withdrawal and β-sympathetic stimulation in the elevation of resting heart rate in subjects who are acutely exposed to a simulated altitude of 6000 m (Koller et al. 1988).
In many of these studies, there has often been a limited ability to obtain repeated measurements of heart rate, cardiac output and blood catecholamine concentrations in the same subjects under a variety of pharmacological manipulations of the autonomic nervous system (ANS). We therefore chose to investigate the role of both the β-sympathetic and the parasympathetic nervous system in the elevation of heart rate and cardiac output during acute hypoxic exercise. We hypothesised that β-sympathetic inhibition with propranolol would not completely abolish the increase in heart rate and cardiac output observed during acute hypoxic exercise. To test this, we compared the effects of propranolol (a non-specific blocker of β receptors) to those of glycopyrrolate (a muscarinic parasympathetic blocking agent), and control conditions in a repeated-measures study design, during exercise to maximal capacity in normoxia and acute hypoxia (inspired oxygen fraction (FI,O2) = 0.125).
This study was approved by the Human Subjects Committee of the University of California, San Diego, and all procedures used conformed with the Declaration of Helsinki. Six healthy, non-smoking subjects (five males, one female) were included in the study. They all exercised regularly, but were not competitive athletes. The subjects had no prior history of respiratory or cardiac disease. After obtaining their written informed consent to participate, a history was taken and a physical examination was performed to exclude cardiopulmonary abnormalities. All subjects were screened for obstructive or restrictive pulmonary disease using spirometry (PDS Instrumentation, PDS VRS-200, Louisville, CO, USA).
After familiarisation with the testing apparatus, the subjects underwent two progressive cycle exercise tests to determine V̇O2,max in normoxia and hypoxia (FI,O2= 0.125 %). The order of the two conditions (room air and hypoxia) was balanced between subjects, and the subjects were given approximately 1 h to recover between testing sessions. These preliminary tests allowed the selection of workloads that elicited approximately 30, 60, 90 and 100 % of V̇O2,max in both inspired oxygen conditions. On each experimental day, subjects either received no drug (control), non-selective β-sympathetic inhibition with propranolol, or muscarinic (parasympathetic) inhibition with glycopyrrolate. Two exercise tests were then performed, one while breathing room air (normoxia, N) and the other while breathing hypoxic gas (hypoxia, H; FI,O2= 0.125). The order of the drug conditions (control, β-sympathetic inhibition or parasympathetic inhibition) was chosen at random, but the inspired air conditions (N or H) were performed in the same order as in the preliminary test. The exercise test consisted of 5 min of rest while seated on the bicycle ergometer, followed by 5 min at each of the previously determined normoxic or hypoxic workloads. The workloads, while adjusted for FI,O2, were the same within each FI,O2 for each drug condition. Data were therefore collected under six different conditions: control-N, control-H, β-sympathetic inhibition-N, β-sympathetic inhibition-H, parasympathetic inhibition-N and parasympathetic inhibition-H. Each experimental day was separated by at least one intervening day to allow for drug washout.
Preliminary exercise tests
V̇O2,max in normoxia and acute hypoxia was determined on an electronically braked cycle ergometer (Excaliber, Quinton Instruments, Gronigen, The Netherlands). After a 5 min warm-up at 50 W, the subjects rode a progressive exercise test (25 W min−1) until they were unable to continue. Heart rate was monitored by cardiac monitor (Lifepak 6, Physio-control, Redmond, WA, USA) and arterial oxygen saturation (Sa,O2) was monitored for safety using a Nellcor N-395 (Oxismart XL) pulse oximeter equipped with an RS-10 forehead sensor. In our laboratory, this pulse oximeter has been shown to have an average bias of 0.3 ± 2.5 % in comparison to co-oximetry (Yamaya et al. 2002).
The subjects breathed through a non-rebreathing valve (2700, Hans-Rudolph, Kansas City, MO, USA). Inspired gas for the determination of FI,O2 was sampled immediately prior to and after the completion of the exercise test, and the results were averaged. The difference between the two measures was ± 0.005 % oxygen in normoxia and hypoxia. For hypoxic exercise tests, the subjects breathed from a 70 l reservoir that was continuously replenished from a premixed tank of 12.5 % oxygen with balanced nitrogen. Expired gas was sampled continuously from a heated mixing chamber, and oxygen and carbon dioxide concentrations were measured (mass spectrometer 1100, Perkin-Elmer, Pomona, CA, USA). Expired gas flow was measured using a pneumotach (no. 3 Fleisch) and a differential pressure transducer (Validyne, DP45-14, Northridge, CA, USA). The electrical signals from the mass spectrometer and the pneumotach were logged at 100 Hz using a 12 bit analog-to-digital converter. Ventilation (V̇E), V̇O2 and carbon dioxide production (V̇CO2,) were calculated using a commercially available software package (Consentius Technologies, Salt Lake City, UT, USA). V̇O2,max was calculated as the average of the four highest consecutive 15 s measures of V̇O2. All subjects fulfilled at least two of the following four criteria for V̇O2,max: (1) no further increase, or a decrease, in V̇O2 with increasing workload, (2) no further increase in heart rate despite an increase in workload, (3) respiratory exchange ratio > 1.10, and (4) heart rate greater than the age-predicted maximum. These data were used to select workloads representing 30, 60, 90 and 100 % of V̇O2,max for both normoxia and hypoxia.
A total of 8 mg of propranolol was administered intravenously while monitoring ECG and blood pressure. The first 1 mg was administered in 0.1 mg boluses over 10 min. An additional 5 mg was given over 5 min prior to the first exercise test. After this test and a 1 h resting period, another 2 mg of the drug was administered intravenously prior to the second exercise test to ensure maintenance of β-sympathetic inhibition.
Parasympathetic inhibition with 0.8 mg of glycopyrrolate was performed by administration of glycopyrrolate intravenously over 10 min prior to the first exercise test. Another 0.2 mg of glycopyrrolate was given prior to the second exercise test to ensure maintenance of parasympathetic inhibition.
After administration of the receptor inhibiting agents, the subjects performed two tests consisting of 5 min of upright rest seated on the bicycle ergometer followed by 5 min of exercise at each workload corresponding to 30, 60, 90 and 100 % of the previously determined normoxic and hypoxic V̇O2,max. During the hypoxic testing, the subject breathed the hypoxic gas mixture for 10 min prior to the start of data collection. The subjects were allowed to rest for approximately 1 h between the two exercise tests. Data were collected in the last 2 min at each exercise intensity, to allow for the achievement of steady-state conditions.
Metabolic and cardiovascular data
Metabolic data including V̇E, V̇O2 and power output were measured using the same set-up as described for the preliminary V̇O2,max testing. Arterial oxygen saturation was monitored as described above.
Heart rate was measured by direct ECG monitoring of lead II. Cardiac output was measured at rest and during exercise at each workload under all six conditions using an open-circuit acetylene (C2H2)-uptake technique (Becklake et al. 1962; Barker et al. 1999). The details of this method have been reported previously and show excellent agreement with the direct Fick methods for measurement of cardiac output (Barker et al. 1999). Briefly, subjects were given a gas mixture containing C2H2 (0.5 %), helium (1 %) and oxygen (20.9 % or 12.5 % for normoxia and hypoxia, respectively), in nitrogen to breathe in an open circuit until the helium showed complete mixing plus 12 breaths. V̇E, end-tidal PCO2, the partial pressure of C2H2 and the partial pressure of helium were measured using a second Perkin Elmer mass spectrometer. The difference between inspired C2H2 and end-tidal C2H2 (corrected for mixing with the ratio of inspired helium/end-tidal helium) was calculated for each breath. This difference was then extrapolated back to the first breath to account for C2H2 recirculation (Barker et al. 1999). The blood-gas partition coefficient for C2H2 in blood for each subject was measured on each experimental day at rest and during maximal exercise using a 7 ml venous blood sample, as has been reported previously (Wagner et al. 1974).
A 5 ml sample of blood was obtained via a peripheral vein at rest and during each exercise level. Samples were maintained in ice until the end of the exercise session and then were spun in a refrigerated centrifuge and frozen (-70 °C) for later batch analysis. Thawed plasma samples were assayed for adrenaline, noradrenaline and dopamine using a catechol-O-methyltransferase-based radioenzymatic procedure with a concentration step to increase assay sensitivity (Kennedy & Ziegler, 1990). The sensitivities of this assay for noradrenaline and adrenaline are 10 and 6 pg ml−1, respectively.
ANOVA for repeated measures with two levels of FI,O2 (conditions N and H), three levels of drug intervention (control, β-sympathetic inhibition and parasympathetic inhibition) and four levels of exercise intensity (rest, 30, 60 and 90-100 % V̇O2,max) was used to test statistically for overall differences in the major dependent variables. All subjects completed all workloads up to 90 % of V̇O2,max. However, not all subjects were able to complete an additional 5 min at 100 % of V̇O2,max following the 5 min at 90 % of V̇O2,max, so the data from the highest exercise intensity reached, either 90 or 100 % of V̇O2,max, was used for statistical purposes. A second ANOVA was performed for heart rate, cardiac output, stroke volume and [catecholamines] after normalising data for V̇O2, to compare responses at the same absolute V̇O2 in normoxia and hypoxia. The slope and intercept of the relationship (linear regression) between individual subject data points for the major dependent variables and V̇O2 over all exercise levels were compared between normoxia and hypoxia across drug interventions (Glantz & Slinker, 1990).
In the results section, the data are reported in the following order: for metabolic and catecholamine data, the overall main effects (FI,O2, drug intervention, exercise intensity) and any significant FI,O2× drug interaction averaged across all conditions studied are reported. In addition, any significant overall main effects (FI,O2, drug intervention) and any significant drug ×FI,O2 interaction are broken down by exercise intensity (e.g. rest, submaximal exercise and maximal exercise). For the cardiovascular data, the overall main effects and any significant FI,O2× drug interaction averaged across all conditions studied are reported, followed by the results of the comparisons of slope and intercept of the relationships between the these variables and V̇O2. Next, any significant overall main effects (FI,O2, drug intervention) and any significant drug ×FI,O2 interaction are broken down by exercise intensity (e.g. rest, submaximal exercise and maximal exercise). α was accepted at P < 0.05, two tailed.
All subjects tolerated the study without difficulty. Subject descriptive data are given in Table 1. This study was part of a larger investigation examining the effects of 2 weeks of acclimatisation to altitude on cardiac output. The results of the altitude portion of the study have been reported previously (Bogaard et al. 2002).
Table 1. Subject descriptive characteristics
Values are means ±s.e.m. BSA, Body surface area; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity.
31.7 ± 1.6
1.72 ± 0.01
69.5 ± 2.7
1.8 ± 0.04
4.4 ± 0.2
FEV1 (% predicted)
99.3 ± 2.6
5.1 ± 0.3
FVC (% predicted)
103.3 ± 2.8
Normoxic V̇O2,max (ml kg−1 min−1)
45.2 ± 3.6
Hypoxic V̇O2,max (ml kg−1 min−1)
36.6 ± 2.7
Metabolic and ventilatory data
Metabolic and ventilatory data for each of the six experimental conditions are given in Table 2. On average, subjects attained 94 ± 1 % of their previously determined V̇O2,max during the highest workload reached during the normoxic exercise tests and 94 ± 1 % of their previously determined hypoxic V̇O2,max. Resting V̇O2 was not altered by hypoxia (P= 0.13). However, β-sympathetic inhibition significantly reduced resting V̇O2 (P < 0.01), which was not different between normoxia and hypoxia (P= 0.18). There was no effect of parasympathetic inhibition on resting V̇O2. Overall, peak V̇O2 was reduced by some 25 % in hypoxia (control-N, 3.08 ± 0.25 l min−1; control-H, 2.47 ± 0.17 l min−1; P < 0.001) and was not affected by drug intervention in either normoxia or hypoxia (β-sympathetic inhibition-N, 3.02 ± 0.23 l min−1; parasympathetic inhibition-N, 3.15 ± 0.27 l min−1; β-sympathetic inhibition-H, 2.52 ± 0.19 l min−1; parasympathetic inhibition-H, 2.47 ± 0.19 l min−1; P= 0.74).
Table 2. Ventilatory and metabolic data
Exercise intensity (% of V̇)
All dependent variables showed a significant main effect of exercise intensity (P < 0.0001, symbols omitted for clarity). Values are means ±s.e.m.* Significant main effect for F, † significant F× exercise intensity interaction, all P < 0.05. Ventilation/V̇ showed significant main effects for F (P < 0.001) and drug intervention (P < 0.005). See text for details.
V̇E (1 min−1 BTPS)†
12.1 ± 1.6
34.5 ± 2.8
63.0 ± 3.2
131.1 ± 14.8
14.1 ± 1.2
43.2 ± 2.3
70.4 ± 3.8
123.1 ± 8.9
β-Sympathetic inhibition, normoxia
9.3 ± 0.4
34.7 ± 1.9
60.7 ± 3.2
119.0 ± 8.3
β-Sympathetic inhibition, hypoxia
10.6 ± 0.7
39.0 ± 2.7
66.1 ± 4.8
118.3 ± 10.9
Parasympathetic inhibition, normoxia
13.7 ± 2.4
39.9 ± 1.8
64.9 ± 3.3
128.6 ± 12.6
Parasympathetic inhibition, hypoxia
13.9 ± 1.7
45.0 ± 3.2
71.4 ± 4.4
120.4 ± 10.1
V̇O2 (1 min−1 STPD)*†
0.36 ± 0.04
1.37 ± 0.09
2.24 ± 0.19
3.08 ± 0.25
0.36 ± 0.04
1.38 ± 0.12
1.94 ± 0.16
2.47 ± 0.17
β-Sympathetic inhibition, normoxia
0.30 ± 0.01
1.54 ± 0.11
2.32 ± 0.18
3.02 ± 0.23
β-Sympathetic inhibition, hypoxia
0.28 ± 0.03
1.31 ± 0.11
1.94 ± 0.18
2.52 ± 0.19
Parasympathetic inhibition, hormoxia
0.39 ± 0.04
1.62 ± 0.12
2.36 ± 0.24
3.15 ± 0.27
Parasympathetic inhibition, hypoxia
0.39 ± 0.04
1.36 ± 0.08
1.92 ± 0.14
2.47 ± 0.19
CO2 (1 min−1 STPD)*†
0.29 ± 0.04
1.12 ± 0.06
2.12 ± 0.18
3.39 ± 0.26
0.33 ± 0.03
1.18 ± 0.09
1.84 ± 0.13
2.76 ± 0.21
β-Sympathetic inhibition, normoxia
0.22 ± 0.01
1.17 ± 0.11
2.12 ± 0.20
3.40 ± 0.29
β-Sympathetic inhibition, hypoxia
0.25 ± 0.02
1.11 ± 0.10
1.89 ± 0.19
2.88 ± 0.27
Parasympathetic inhibition, normoxia
0.29 ± 0.05
1.24 ± 0.10
2. 14 ± 0.23
3.44 ± 0.34
Parasympathetic inhibition, hypoxia
0.32 ± 0.04
1.19 ± 0.09
1.91 ± 0.15
2.80 ± 0.23
Respiratory exchange ratio*†
0.80 ± 0.04
0.82 ± 0.02
0.95 ± 0.02
1.10 ± 0.02
0.92 ± 0.04
0.86 ± 0.03
0.96 ± 0.03
1.12 ± 0.03
β-Sympathetic inhibition, normoxia
0.73 ± 0.03
0.76 ± 0.03
0.91 ± 0.03
1.12 ± 0.03
β-Sympathetic inhibition, hypoxia
0.88 ± 0.06
0.85 ± 0.03
0.97 ± 0.04
1.13 ± 0.04
Parasympathetic inhibition, normoxia
0.74 ± 0.08
0.76 ± 0.02
0.91 ± 0.02
1.08 ± 0.02
Parasympathetic inhibition, hypoxia
0.82 ± 0.04
0.88 ± 0.03
1.00 ± 0.02
1.13 ± 0.02
Averaged over all conditions, V̇E was significantly increased relative to V̇O2 by hypoxia (P < 0.001), and was significantly affected by drug intervention (P < 0.005, see below). Resting V̇E was not significantly different between normoxia and hypoxia, and was not affected by the different drug interventions. During submaximal exercise, V̇E was significantly increased relative to V̇O2 in hypoxia compared to normoxia (P < 0.05) and was significantly affected by drug intervention (P < 0.005): β-sympathetic inhibition lowered submaximal V̇E relative to V̇O2 (P < 0.05), but parasympathetic inhibition had no effect. Peak exercise V̇E was unchanged between normoxia and hypoxia (P= 0.26) and there was no effect of β-sympathetic/parasympathetic inhibition on peak V̇E (control-N, 131 ± 15 l min−1 BTPS; control-H, 123 ± 9 l min−1; β-sympathetic inhibition-N, 119 ± 8 l min−1; β-sympathetic inhibition-H, 118 ± 11 l min−1; parasympathetic inhibition-N, 129 ± 13 l min−1; parasympathetic inhibition-H, 120 ± 10 l min−1; P= 0.33). Thus, the observed overall significant effect of the drug intervention was due to the effect of β-sympathetic inhibition in reducing submaximal exercise V̇E.
Haemoglobin saturation at maximal exercise averaged 98.6 ± 0.3 % in normoxia and was reduced to 78.7 ± 1.2 % in hypoxia (P < 0.0001). As expected, Sa,O2 was decreased with increasing exercise intensity (P < 0.0001). However, there was no effect of drug intervention on Sa,O2 (P= 0.88).
The effect of the different conditions on venous adrenaline concentration can be seen in Fig. 1A. Averaged over all conditions, adrenaline concentration was increased relative to V̇O2 in hypoxia (P < 0.05), affected by drug intervention (P < 0.05, see below) and was increased by exercise (P < 0.01). Resting adrenaline concentration was not significantly affected by hypoxia (P= 0.24) or drug intervention (P= 0.43). During submaximal exercise at any given V̇O2, adrenaline was increased by hypoxia (P < 0.01). However, the effect of hypoxia on adrenaline concentration during submaximal exercise was not different between the three different drug conditions (P= 0.09) and there was no significant FI,O2× drug interaction (P= 0.23). Peak venous adrenaline concentration was increased during hypoxic exercise during β-sympathetic inhibition, an effect that was significant when compared to control conditions (P < 0.05), accounting for the observed overall significant effect of drug intervention.
The effect of the different conditions on venous noradrenaline concentration can be seen in Fig. 1B. Averaged over all conditions, noradrenaline concentration was increased relative to V̇O2 in hypoxia (P < 0.01), affected by drug intervention (P < 0.05, see below) and increased by exercise (P < 0.0001). There was no significant FI,O2× drug interaction. Resting noradrenaline concentration was not altered by hypoxia (P= 0.27) or by the different drug interventions (P= 0.09). During submaximal exercise at any given V̇O2, noradrenaline concentration was increased by exercise (P < 0.01) and by hypoxia (P < 0.005). In addition, noradrenaline concentration showed a small but significant reduction with parasympathetic inhibition (P < 0.05), but this was not different between normoxia and hypoxia (FI,O2× drug interaction P= 0.75). β-Sympathetic inhibition had no significant effect on venous noradrenaline concentration (P= 0.25). During maximal exercise there was a tendency for peak noradrenaline concentrations to be increased by hypoxia, but this did not reach statistical significance (P= 0.06), nor was there any significant effect of drug intervention on peak noradrenaline concentrations (P= 0.94). Thus, the overall significant effect of drug intervention on noradrenaline concentration was due to a reduction in submaximal exercise noradrenaline concentration during parasympathetic inhibition.
The effect of the different interventions on venous dopamine concentration can be seen in Fig. 1C. Averaged over all conditions, dopamine concentration was not changed relative to V̇O2 by hypoxia (P= 0.2), or affected by drug intervention (P= 0.1), but it was significantly increased by exercise (P < 0.0001). During submaximal exercise at any given V̇O2, dopamine concentration was increased by exercise (P < 0.001) but not by hypoxia (P= 0.18). Dopamine concentration during submaximal exercise was not significantly affected by drug intervention (P= 0.09), and there was no significant FI,O2× drug interaction (P= 0.50). Peak dopamine concentration was not altered by hypoxia (P= 0.74) or by drug intervention (P= 0.56).
Heart rate and cardiac output data
The effect of the different interventions on heart rate can be seen in Fig. 2A. Averaged over all conditions, heart rate was increased relative to V̇O2 by hypoxia (P < 0.05), increased by exercise (as expected; P < 0.0001) and affected by drug intervention (P < 0.0005). However there were no significant FI,O2× drug interactions (P= 0.7).
The relationship between heart rate and V̇O2 in normoxia and hypoxia during all levels of exercise under each drug intervention can been seen in Fig. 3Aa-c. Under all three drug conditions, hypoxia increased the intercept of the heart rate-V̇O2 relationship (all P < 0.01) and increased the slope under control conditions and with propranolol (both P < 0.005). During parasympathetic inhibition (Fig. 3Ac), the slope of the heart rate-V̇O2 relationship was significantly reduced, probably because of the effect of both hypoxia and parasympathetic inhibition on heart rate during submaximal exercise (once maximal heart rate is reached there is no further response). The changes in heart rate between normoxia and hypoxia were not significantly different under the different conditions of drug intervention (P= 0.6).
Resting heart rate was not significantly altered by hypoxia (P= 0.25). Parasympathetic inhibition increased resting heart rate from 79 ± 14 beats min−1 in normoxic rest to 124 ± 18 beats min−1 (P < 0.001). β-Sympathetic inhibition lowered normoxic resting heart rate to 60 ± 7 beats min−1 (P < 0.001). During submaximal exercise at any given V̇O2, heart rate was increased by hypoxia (P < 0.001). β-Sympathetic inhibition lowered the heart rate compared to control conditions (P < 0.05) and parasympathetic inhibition elevated heart rate compared to control conditions (P < 0.005). However, the effect of hypoxia on submaximal heart rate was not different between the three different drug conditions: there was no significant FI,O2× drug interaction (P= 0.15) and individual post hoc tests comparing submaximal heart rate under conditions of normoxia and hypoxia for β-sympathetic and parasympathetic inhibition to control conditions showed no significant effect of either form of inhibition on the response to hypoxia. Averaged over all drug conditions, peak heart rate showed a small but significant reduction in hypoxia (N, 172 ± 4 beats min−1; H, 169 ± 4 beats min−1; P < 0.05) and, as expected, was altered by β-sympathetic/ parasympathetic inhibition (P < 0.0001). β-Sympathetic inhibition reduced peak heart rate (control, 181 ± 3 beats min−1; propranolol, 149 ± 3 beats min−1; P < 0.0001), whereas parasympathetic inhibition had no significant effect. Again, there was no significant FI,O2× drug interaction (P= 0.07).
Averaged over all conditions, cardiac output was increased relative to V̇O2 by hypoxia (P < 0.05), and was affected by drug intervention (P < 0.05). However, as was found for the heart rate data, there was no significant FI,O2× drug interaction (P= 0.5). The relationship between cardiac output and V̇O2 in normoxia and hypoxia during all levels of exercise for each drug condition can been seen in Fig. 3Ba-c. In all conditions, hypoxia increased the slope of the cardiac output-V̇O2 relationship (all P < 0.05). The intercept was unchanged by hypoxia under control conditions and during β-sympathetic inhibition, but showed a very small but statistically significant reduction with glycopyrrolate (P < 0.05). As for heart rate, across all exercise conditions, the changes in cardiac output between normoxic and hypoxia conditions were not significantly different between different drug intervention conditions (P= 0.35).
Resting cardiac output was not affected by hypoxia (P= 0.62), but was reduced by β-sympathetic inhibition (P < 0.05). This reduction was not different in normoxia compared to hypoxia (P= 0.88). There was no effect of parasympathetic inhibition on resting cardiac output. Submaximal cardiac output was increased relative to V̇O2 by hypoxia (P < 0.005) largely because of an increase at 30 % V̇O2,max, and was reduced by β-sympathetic inhibition (P < 0.05). There was a significant reduction in maximal cardiac output with hypoxia (P < 0.05). There was also a significant reduction in cardiac output during β-sympathetic inhibition (P < 0.05), but no significant change from control conditions was observed with parasympathetic inhibition.
Averaged over all conditions, stroke volume was unchanged relative to V̇O2 by hypoxia (P= 0.47), although it was significantly affected by drug intervention (P < 0.0001). The relationship between stroke volume and V̇O2 in normoxia and hypoxia during all levels of exercise for each drug intervention paralleled the heart rate and cardiac output data (data not shown). As we found for heart rate and cardiac output, the changes in stroke volume between normoxic and hypoxic conditions were not significantly different between different conditions of drug intervention (P= 0.7).
Resting stroke volume was not altered by hypoxia (P= 0.85) or β-sympathetic inhibition. However, parasympathetic inhibition significantly reduced resting stroke volume (P < 0.005), in keeping with the reciprocal changes in resting heart rate. Stroke volume during submaximal exercise was increased by hypoxia (P < 0.05), increased by exercise (P < 0.05) and, in keeping with the changes in heart rate, was significantly affected by drug intervention (P < 0.0001). The effect of β-sympathetic/ parasympathetic inhibition was not different during submaximal exercise between normoxia and hypoxia (FI,O2× drug interaction P= 0.63). Maximal stroke volume was reduced by hypoxia (P < 0.05) and was increased by β-sympathetic inhibition (P < 0.05). In keeping with the heart rate data, maximal stroke volume was not altered by parasympathetic inhibition. The effect of hypoxia on maximal stroke volume was not affected by drug intervention (FI,O2× drug interaction P= 0.39).
The major new findings of this study are that acute exposure to hypoxia increases submaximal exercising heart rate and cardiac output even under conditions where muscarinic and β-sympathetic receptors in the heart are separately blocked. The evidence for this statement is that although exercising heart rate and cardiac output were increased in hypoxia and were affected by β-sympathetic/ parasympathetic inhibition, there was no significant interaction between receptor inhibition and FI,O2. The changes in exercising heart rate and cardiac output between normoxia and hypoxia are similar irrespective of the nature of the ANS inhibition imposed during this study. We believe that the nature of this repeated-measures study design, where all six subjects were studied under all 24 conditions (two levels of FI,O2, four levels of exercise and three levels of drug intervention) reveals effects of hypoxia on the heart that have not been previously been appreciated. We did not see any significant increase in resting heart rate, cardiac output or catecholamine concentration during hypoxia, but this is probably because of the very short duration of the hypoxic exposure (10 min) prior to resting data collection.
Heart rate responses to hypoxia and β-sympathetic/parasympathetic inhibition
The elevation of resting and submaximal exercising heart rate during acute exposure to hypobaric or normobaric hypoxia is well described and has been widely attributed to β-adrenergic sympathetic nerve stimulation (Grover et al. 1967; Maher et al. 1975; Boutellier & Koller, 1981). The evidence in support of this is considerable. For example, plasma noradrenaline concentrations are increased in acute hypoxic exercise (e.g. Escourrou et al. 1984), and 24 h urinary noradrenaline excretion is increased on arrival at 4300 m (Mazzeo et al. 1991) and remains elevated with more prolonged exposure (Fischetti et al. 2000). In addition, non-specific β-blockade with propranolol reduces both submaximal and maximal heart rate during hypoxic exercise (Moore et al. 1986; Wolfel et al. 1998). In our study, β-sympathetic inhibition with propranolol lowered the heart rate at rest and during all levels of exercise. However, the increase in heart rate during exercise in hypoxia was not prevented, as a similar increase in heart rate from normoxic conditions was observed, albeit from a lower baseline. In support of this, the changes in heart rate between normoxia and hypoxia over all observations (rest and all levels of exercise) were unaffected by β-sympathetic inhibition.
Our findings under control conditions and β-sympathetic inhibition might implicate an increase in parasympathetic withdrawal as a factor that is at least partially responsible for the increases in heart rate during acute hypoxic exercise. The effects of acute hypoxia on the parasympathetic nervous system are unclear, and parasympathetic activity has been reported to be increased (Hartley et al. 1974), decreased (Richalet et al. 1985) and unchanged (Hammill et al. 1979). However, in the present study we found a similar increase in heart rate during exercise in hypoxia with parasympathetic inhibition as with β-sympathetic inhibition or control conditions. In addition, as for β inhibition, the changes in heart rate between normoxia and hypoxia over all observations (rest and exercise) were unaffected by parasympathetic inhibition. Thus, this mechanism alone cannot be responsible for the changes observed with acute hypoxic exercise.
Cardiac output responses to hypoxia and β-sympathetic/parasympathetic inhibition
The changes in cardiac output under the various interventions were similar in direction to those described for heart rate. Overall, hypoxia increased submaximal cardiac output at any given workload, but this was mostly due to a marked increased in cardiac output/work rate at 30 % V̇O2,max, with a lesser effect at 60 % V̇O2,max. Maximal cardiac output was reduced by acute hypoxia (N, 19.7 l min−1vs. H, 17.4 l min−1). This is probably because of the effect of hypoxia on stroke volume, where local metabolic factors acting to vasodilate muscle vascular beds compete with vasoconstricting α-adrenergic stimulation and have a variable effect on central blood volume and venous return to the heart. Although the results were not significant, there was a tendency for stroke volume to decrease with hypoxia at 60 % V̇O2,max, and at 100 % V̇O2,max, the reduction was statistically significant. Where the increase in heart rate was not sufficient to compensate for the reduction in stroke volume, the net result was a reduction in maximal cardiac output under hypoxic conditions. However, the take-home message with respect to cardiac output and stroke volume remains the same as for heart rate: the changes in these variables observed with hypoxia are similar irrespective of the nature of β-sympathetic or parasympathetic inhibition imposed.
This can perhaps be best appreciated visually in Fig. 3. Here, cardiac output is plotted relative to V̇O2 in normoxia and hypoxia for all three conditions. The effect of hypoxia in control conditions was to increase the intercept of the cardiac output-V̇O2 relationship by 0.74 l min−1 and to increase the slope by 0.48 l min−1. β-Sympathetic inhibition had virtually the same effect, resulting in a 0.74 l min−1 increase in intercept and a 0.23 l min−1 increase in slope. Glycopyrrolate resulted in a small reduction in intercept, but a 0.99 l min−1 increase in slope. These changes are not significantly different from one another. Thus, in a similar manner to the discussion of heart rate findings, we would argue that although cardiac output is affected by hypoxia, and is affected by β-sympathetic or parasympathetic inhibition, the manner in which the cardiac output is affect by these drugs is not substantially different between normoxic and hypoxic exercise. Thus, the increases in the cardiac output-V̇O2 relationship that occur with hypoxic exercise can occur independently of the effects of inhibition of muscarinic and β-adrenergic receptors on the heart.
Possible mechanisms for the observed increases in heart rate and cardiac output
If the effects of acute hypoxia on exercising cardiac output and heart rate still occur despite inhibition of muscarinic and β-adrenergic receptors on the heart, we can speculate as to the mechanism underlying the well-described increases in heart rate and cardiac output during hypoxic exercise. It is possible that when the cardiac β receptors are blocked by propranolol, hypoxia elevates the heart rate as a result of increased parasympathetic withdrawal, whereas when the muscarinic receptors are blocked by glycopyrrolate, the β receptors are stimulated, thus increasing heart rate and cardiac output. The concept of redundant systems of autonomic control of heart rate during hypoxia has been described in anaesthetised dogs (Krasney, 1967), although there are marked species difference in ANS function.
Alternatively, a mechanism that is not affected by either β-sympathetic or muscarinic inhibition could be responsible for the changes observed. One possible candidate is α1-adrenergic receptors in the heart, which could be stimulated by noradrenaline and would not be blocked by either propranolol or glycopyrrolate. α1-Adrenergic receptors play a role in increasing myocardial contractility, particularly after exercise training (Korzick & Moore, 1996), and have been demonstrated to have a chronotropic action in children (Tanaka et al. 2001). Although the persistent ability of α1-adrenergic receptors to modulate heart rate in adult humans is not fully established, there is evidence to suggest, particularly in young adults (which comprised our study population), that the response may be similar to that described in children (Saitoh et al. 1995). Normally, any effects of α1-adrenergic-receptor stimulation would be masked by vagal innervations and negative chronotropic effects of the baroreflex pathway; however, the unique nature of our study design raises the possibility that α-cardiac receptors could be involved. A combination of increased parasympathetic withdrawal and increased β-receptor stimulation may be responsible for the response when these arms of the ANS are intact, or the integrated response to hypoxia may involve all three components, α1 receptors, β receptors and increased parasympathetic withdrawal.
We cannot choose among these alternatives based on the available data. One approach that would allow the differentiation of these possible mechanisms would be to compare combined β-sympathetic and parasympathetic inhibition to control conditions in an experimental design similar to that used in the present study. The lack of an increase in heart rate and cardiac output during hypoxic exercise when both β-sympathetic and parasympathetic receptors are inhibited would support the idea that both the β-sympathetic and parasympathetic systems are operating in a redundant fashion. The persistence of the increase in heart rate and cardiac output during hypoxic exercise under these conditions would suggest that α-adrenergic receptors, either cardiac or in the vascular beds, are important. This could be confirmed by α-adrenergic inhibition. In addition to the effects on the adrenergic receptors of the heart, circulating catecholamines act on vascular smooth muscle α-adrenergic receptors to vasoconstrict the vascular beds. Hypoxia acts directly to override the sympathetically mediated vasoconstriction associated with exercise (Rowell & Blackmon, 1987). Thus the net effect on the peripheral circulation is a result of the balance of these competing influences in skeletal muscle modified by the effect that these factors have on the splanchnic circulation and other vascular beds (e.g. cutaneous), which compete for blood flow. Therefore, another possible explanation for our finding relates to the effect that increased α-adrenergic stimulation has on central blood volume. Increased cardiac output resulting from either increased venous return, occurring with sympathetically medicated vasoconstriction, or from a reduction in afterload with hypoxia-mediated vasodilatation, cannot be ruled out.
Ideally, systemic arterial pressure would have been monitored invasively during our study and provided additional information about the effects of hypoxia on haemodynamics. However, this would have been logistically very difficult given the large number of exercise tests performed. Without an indwelling arterial cannula, we were unable to obtain a complete set of blood pressure data on our subjects and thus the results must be interpreted with caution. However, there were no significant drug ×FI,O2 interactions for mean arterial pressure (P= 0.4) or systemic vascular resistance (P= 0.2), suggesting that the increases in heart rate and cardiac output with hypoxia are not related to changes in afterload.
Completeness of β-sympathetic and parasympathetic inhibition
The validity of these findings depends on the completeness of the receptor inhibition imposed. Although our subjects were not challenged with isoproterenol to confirm β-blockade, we would argue that the receptor inhibition was complete for the following reasons. Firstly, the effects of β-sympathetic and parasympathetic inhibition on heart rate and exercise capacity were similar to those described in previous studies (Hartley et al. 1974; Moore et al. 1986) and the dose of drug administered was consistent with dosages reported to cause complete receptor block (Jose & Taylor, 1969; Boushel et al. 2001). The half-life of propranolol is approximately 4 h. Data collection was commenced immediately after the administration of the drug and after 10 min of equilibration time. A supplementary dose (2 mg) of propranolol was given before starting the second test (approximately 1.5 h after the first). The total duration of the study was less than 2 h, and thus we feel confident that there was no significant time-dependent loss of β-sympathetic inhibition. Similarly, the half-life of glycopyrrolate is ≈2 h, and again a supplementary dose was administered prior to the start of the second test; thus, time-dependent loss of muscarinic inhibition is unlikely to have occurred. The order of normoxia or hypoxia was balanced between subjects, and the observed changes in heart rate and cardiac output are therefore not likely to have been an ordering effect. In addition, during β inhibition, maximal heart rate was reduced by some 24 % from control conditions, with no significant increase in hypoxia, despite a doubling of venous adrenaline and an ≈10 % increase in noradrenaline concentrations. During parasympathetic block with glycopyrrolate, resting normoxic heart rate was increased from 79 ± 14 beats min−1 under control conditions to 124 ± 18 beats min−1, a level that is consistent with complete receptor inhibition.
Another issue that needs to be addressed is whether or not this study had adequate statistical power to support the conclusions reached, particularly in light of the small number of subjects studied. The repeated-measures design of this study, combined with a very high within-subjects correlation under the different FI,O2 conditions for the dependent variable of interest (for example, R > 0.9 for both heart rate and cardiac output) largely offsets the potential lack of power risked by using a small number of subjects. This is evidenced by highly significant main effects for many of the dependent variables, such as the effect of drug intervention on heart rate (P < 0.0001). In addition, post hoc power calculations revealed a power of 0.85 to detect a significant difference in heart rate of 2 beats min−1 between normoxia and hypoxia across drug interventions. Similar power was obtained for calculations involving the other major dependent variables.
We found that the increases in heart rate and cardiac output in response to acute hypoxia (FI,O2) were not prevented by either parasympathetic or β-sympathetic inhibition. This points to a complex mechanism of heart rate and cardiac output regulation in response to hypoxia, possibly involving multiply redundant systems. Cardiac α1 receptors may play a primary role, or may be a secondary mechanism that is activated only when other ANS pathways are blocked. Alternatively, the effects of hypoxia and/or catecholamines acting on vascular beds may be important in determining the heart rate and cardiac output response to hypoxic exercise.
This study was supported by NIH grants HL17731 and M01 RR00827. A Fulbright scholarship, the Netherlands Organization for Scientific Research (NWO) and the Haak-Bastiaanse Kuneman Foundation sponsored H. Bogaard. The technical assistance of Brian Kennedy, Harrieth Wagner, Jeff Struthers and Nick Busan are gratefully acknowledged. We would like to thank our subjects for their enthusiastic participation.