Exercise-induced muscle chemoreflex modulation of spontaneous baroreflex sensitivity in man



  • 1The goal of this study was to determine the effect of exercise-induced muscle chemoreflex activation on baroreflex sensitivity (BRS). This is a retrospective study using data obtained during two prior studies.
  • 2Twenty-three subjects with a mean (s.e.m.) age of 28 (1.5) years took part in the study. Sequence analysis was performed on the systolic blood pressure (SBP) responses, measured by a Finapres, and R-R intervals, measured from the ECG.
  • 3Electrically evoked isometric exercise (Stim) of the triceps surae was performed for 2 min at 30 % maximum voluntary contraction force. During exercise and for a further 2 min thereafter, circulation to the lower leg was occluded by inflation of a thigh cuff to above 200 mmHg.
  • 4Prior to exercise mean (±s.e.m.) BRS was 10.92 ± 6.3 ms mmHg−1, and BRS remained at this level during evoked exercise (10.90 ± 7.1 ms mmHg−1). BRS increased to 12.34 ± 6.0 ms mmHg−1 during post-exercise circulatory occlusion (PECO) (P < 0.05, MANOVA, post hoc Student's paired t test vs. Stim) and fell to 9.27 ± 4.4 ms mmHg−1 during recovery (P < 0.01vs. PECO value, P= 0.059vs. resting value).
  • 5These data indicate that during PECO following electrically evoked plantar flexion, where only muscle chemosensitive afferents were likely to be stimulated, BRS was increased.

The effect of exercise on the arterial baroreflex has been studied extensively (Sagawa, 1983; Rowell, 1993) and it is believed that the autonomic responses to exercise are, in part, mediated by resetting of the arterial baroreflex (Scherrer et al. 1990; Papelier et al. 1997). It is not yet clear which neural mechanisms modulate the human arterial baroreflex during exercise and in the past, central command has been proposed as the signal (Ebert, 1986; Rowell & O'Leary, 1990). More recently, however, a number of studies (Iellamo et al. 1994, 1997; Papelier et al. 1997) have suggested that muscle afferent activation may also modify the human arterial baroreflex.

Iellamo and colleagues (Iellamo et al. 1997) used the sequence analysis technique to measure human spontaneous baroreflex sensitivity (BRS). They reported that during low level, electrically evoked dynamic exercise, which was assumed to activate only muscle mechanoreceptors, there was a decrease in human spontaneous BRS. However, when the same exercise was performed with circulatory occlusion, which was expected to initiate muscle chemoreflex activation as well as the mechanoreflex, BRS returned to control levels. Thus, Iellamo et al. (1997) concluded that the role of the muscle chemoreflex was to maintain or preserve BRS in exercise.

To date, the experiments on man that have been performed to examine the influence of the muscle chemoreflex alone on BRS have been poorly controlled. Therefore, the purpose of the present study was to use a well-established method (Bull et al. 1989) to investigate the effect of muscle chemoreflex activation on BRS. Involuntary exercise, in the form of electrically evoked isometric plantar flexion, was used because voluntary exercise elicits both central command and skeletal muscle reflexes. Post-exercise circulatory occlusion (PECO) followed the electrically evoked exercise in order to separate the contribution of metabolically sensitive muscle afferents from that of mechanically sensitive afferents. BRS was determined by performing sequence analysis on the systolic blood pressure (SBP) responses and intervals between successive R waves of the ECG (R-R intervals) that were recorded during the electrically evoked isometric exercise and PECO.


This is a retrospective study that uses data obtained during two prior studies (Fisher & White, 1999; Ubolsakka et al. 2000). These studies had local ethical committee approval and all subjects gave written informed consent before participation. All work conformed with the Declaration of Helsinki.


Data were collected from 23 healthy subjects (15 men) with a mean (s.e.m.) age of 28 (1.5) years. Subjects were recruited from the students and staff of the University of Birmingham and were recreationally active.

Established methods were used to measure maximal voluntary contraction (MVC) and evoked pressor responses at 30 % MVC (Bull et al. 1989; Fisher & White, 1999). Subjects were habituated to the procedures during preliminary visits to the laboratory. Before the start of the protocol the subject rested for 10 min, positioned in the dynamometer, in order to attain a stable basal circulatory state. The 8 min protocol consisted of four phases. It began with a 2 min control rest period which was followed immediately by 2 min of electrically evoked contraction. A thigh cuff was inflated to above 200 mmHg just prior to the start of the contraction, and was maintained throughout the 2 min contraction. At the end of the contraction the thigh cuff remained inflated at above 200 mmHg for 2 min post-exercise circulatory occlusion (PECO) and was then released for a 2 min recovery period.

The R-R interval and blood pressure were recorded continuously throughout the 8 min protocol. Beat-to-beat blood pressure was recorded using a 2300 Finapres (Ohmeda) (Parati et al. 1989) from the middle finger of the left hand, which was supported on an adjustable arm rest at the level of the heart throughout the experiment. The R-R interval was recorded using a three lead electrocardiogram (ECG) and heart rate monitor (Cardiorater CR7, Cardiac Records Ltd). Analog blood pressure and ECG signals were transmitted to an analog-to-digital converter (Cambridge Electronic Design 1401 plus). For each signal, the sampling frequency of analog-to-digital conversion was 1000 Hz. Blood pressure and ECG data were displayed and analysed on a personal computer (Vale Platinum TX).

Calculation of BRS by the spontaneous sequence analysis technique

Purpose-written software was used to search beat-to-beat SBP and R-R intervals for sequences of three or more consecutive beats in which SBP progressively increased and the subsequent R-R interval lengthened (up sequence) or SBP progressively decreased and the subsequent R-R interval shortened (down sequence). The minimum change was 1 mmHg for SBP and 1 ms for R-R interval and a lag of one beat was used. Linear regressions relating SBP to R-R interval were plotted for each sequence and only those with linear r values > 0.92 were used. Slopes derived from all up and down sequences within 2 min phases were pooled, so that one measure of BRS was obtained for each subject for each phase (rest, exercise, PECO, recovery) of the protocol.

As it could be argued that BRS may be affected by alterations in breathing frequency we felt it was important to examine the effect of the protocol on this variable. In a separate study on a group of 14 healthy young subjects, similar to those used in the retrospective analysis, breathing frequency was measured during the evoked protocol. In a subgroup of nine of these subjects the protocol was repeated with the substitution of a voluntary contraction for the evoked contraction. Breathing frequency was measured from analysis of respiratory gas. A rapid response CO2 analyser sampled continuously from a mouthpiece worn by the subjects for 5 min prior to, and throughout, the protocol. A nose clip was also worn.


A repeated measures multivariate analysis of variance (MANOVA) was used to compare differences in the measured variables during the four phases of the protocol. If a significant effect was indicated a post hoc Student's paired t test was performed. Data are reported as group means (±s.e.m) unless otherwise stated. The criterion for statistical significance was P < 0.05.


Breathing frequency

There was no significant difference (repeated measures MANOVA, P= 0.186) in respiratory rate between rest, electrically evoked isometric contraction, PECO and recovery phases (15.0 ± 1.4, 14.4 ± 1.3, 13.9 ± 1.2 and 13.5 ± 0.9 breaths min−1, respectively). However, in the subgroup of nine subjects who repeated the protocol with voluntary isometric contraction, there was a significant experimental phase effect (repeated measures MANOVA, P= 0.007). Post hoc t tests showed that the respiratory rate was significantly faster in the contraction phase when compared with rest (P < 0.001), PECO (P= 0.012) and recovery (P= 0.007) phases.

Cardiovascular responses

Resting systolic (SBP) and diastolic blood pressure (DBP) and heart rate were 130 ± 3 mmHg, 76 ± 2 mmHg and 72 ± 3 beats min−1, respectively. Electrically evoked isometric plantar flexion produced a rise in blood pressure and heart rate (Table 1 and Fig. 1). During the PECO phase, on cessation of the contraction, there was an immediate fall in blood pressure, but to a level that was significantly higher than that at rest. This was maintained until the circulation was restored to the lower leg. Heart rate fell rapidly to pre-contraction levels during PECO (Table 1 and Fig. 1).

Table 1. Group mean (s.e.m.) cardiovascular changes at the end of electrically evoked contraction (Stim) and postexercise circulatory occlusion (PECO)
  1. SBP, systolic blood pressure; DBP, diastolic blood pressure; MAP, mean arterial pressure; HR, heart rate. n= 23.

SBP (mmHg)29 (6.1)19 (4.0)
DBP (mmHg)13 (2.8) 9 (1.8)
MAP (mmHg)19 (3.9)12 (2.6)
HR (beats min−1) 6 (1.2)−1 (0.2)
Figure 1.

Mean (±s.e.m.) changes in mean arterial pressure (MAP; □) and heart rate (HR; ♦) during the 8 min protocol.

Spontaneous sequence analysis of the baroreflex

Of all beats recorded in each 2 min phase of the protocol an average of 40, 31, 32 and 36 % showed baroreflex activity (i.e. R-R interval and SBP changed in the same direction) during rest, evoked exercise, PECO and recovery, respectively. Figure 2 shows BRS, estimated by the mean slope of the regression line between SBP and R-R interval changes, for each phase of the experiment. Group mean slope and intercept data for each phase of the experiment are shown in Fig. 3 and Fig. 4, respectively. There was no significant difference in BRS between the rest (10.92 ± 6.3 ms mmHg−1) and evoked exercise (10.90 ± 7.1 ms mmHg−1) phases. However, during evoked exercise, although the gain was maintained, the line was shifted to the right along the pressure axis (Fig. 2) and the intercept decreased significantly (P < 0.05; Fig. 4). When the evoked exercise and PECO phases were compared BRS was significantly increased to 12.34 ± 6.0 ms mmHg−1 (P < 0.05) during PECO (Fig. 3). BRS was also significantly increased (P < 0.01) during PECO, compared with the value of 9.27 ± 4.4 ms mmHg−1 measured during recovery (Fig. 3).

Figure 2.

Mean regression lines calculated from spontaneous baroreflex sequences for rest (A), electrically evoked exercise (B), post-exercise circulatory occlusion (C) and recovery (D).

Figure 3.

Mean (±s.e.m.) BRS during rest (Rest), electrically evoked exercise (Stim), post-exercise circulatory occlusion (PECO) and recovery (Rec). * Significant difference from Stim and Rec, P < 0.05.

Figure 4.

Mean (±s.e.m.) intercept values for baroreflex regression lines during rest (Rest), electrically evoked exercise (Stim), post-exercise circulatory occlusion (PECO) and recovery (Rec). * Significant difference from Rest and Rec, P < 0.05.


The primary finding of this investigation is that during PECO following electrically evoked isometric contraction, when only the muscle chemoreflex can be activated, the slope of the regression line relating SBP and R-R interval was significantly steeper, i.e. BRS was significantly increased.

Our findings are in contrast to those of Iellamo et al. (1997) who used PECO after electrically evoked ischaemic dynamic knee extension and reported no significant difference in BRS during rest, exercise and PECO phases. This may be because of the low force level that they chose to use (∼15 % MVC) and the intermittent nature of the exercise, resulting in lower levels of metabolite accumulation than would be expected than in the present study. In addition, the small subject sample (n= 6) would reduce the statistical power of their analysis.

Other studies (Iellamo et al. 1994; Papelier et al. 1997) on man have used PECO following voluntary exercise in an attempt to study the effects of the muscle chemoreflex on BRS. However, their respective methodologies may confound the interpretation of their findings. Iellamo et al. (1994) observed no change in BRS during post-exercise forearm vascular occlusion following voluntary handgrip. They used voluntary isometric handgrip at a force level of 30 % MVC, which is much lower than the 70 % MVC that is needed to occlude completely blood supply to the forearm (Humphreys & Lind, 1963). In addition, circulatory occlusion was applied only 8 s before the end of the handgrip. As a consequence, metabolite accumulation was unlikely to be large leading to minimal muscle chemoreflex activation during PECO. This is supported by the observation that in their study mean arterial pressure (MAP) only increased by ∼12 mmHg during exercise whilst the expected rise for this intensity and duration of handgrip is ∼25 mmHg (Perez-Gonzalez et al. 1981).

Papelier et al. (1997) used PECO after 7 min of voluntary dynamic exercise on a cycle ergometer at 150 W. Circulatory occlusion was applied just prior to the end of the dynamic exercise. Carotid sinus baroreflex stimulus- response curves were generated by step changes in neck pressure applied sequentially, going from positive to negative pressures, over a 160 s period. Papelier et al. (1997) observed that the stimulus-response curve for MAP vs. carotid sinus pressure during PECO was steeper with positive neck pressures (applied first) and much flatter with negative neck pressures (applied later in the sequence). They concluded that the relationship between the muscle chemoreflex and the carotid baroreflex was not linear. However, it is probable that muscle chemoreflex activity was increasing progressively, due to metabolite efflux from muscle fibres, whilst BRS was being assessed, because of the prolonged period of time required to make their measurements. Their conclusions regarding BRS changes during sequential hypotensive and hypertensive stimulation may be confounded, therefore, by a time-dependent change in muscle chemoreflex activity.

In the present study, during electrically evoked isometric plantar flexion, when both muscle mechanoreceptors and chemoreceptors were activated, we found no change in the slope of the line representing BRS but observed that it was shifted rightwards along the pressure axis (Fig. 2 and Fig. 4). Additionally, we observed that involuntary exercise did not cause a change in breathing frequency during any phase of the experimental protocol. This finding of a rightward shift in the regression line with unaltered sensitivity suggests a resetting of the baroreflex and agrees with the findings of many previous studies in man which have also reported baroreflex resetting during voluntary dynamic exercise (Potts et al. 1993; Papelier et al. 1994; Norton et al. 1999), voluntary isometric exercise (Ebert, 1986; Scherrer et al. 1990; Iellamo et al. 1994) and involuntary ischaemic dynamic exercise (Iellamo et al. 1997).

It is unlikely that the BRS changes we observed could be explained by altered breathing frequency since this remained unchanged throughout the involuntary protocol. The results of the exercise and PECO phases of the experiment may be explained by the following model. There are three components that may influence the level of autonomic outflow during evoked isometric exercise: (1) mechanoreflex activation which is known to inhibit cardiac vagal motoneurone activity by reducing the excitability of the vagal motoneurone pool and thus causing heart rate to rise (McWilliam & Yang, 1991; McMahon et al. 1992); (2) increasing chemoreflex activation which causes sympathoexcitation and a progressive increase in blood pressure (Victor et al. 1988; Seals & Victor, 1991); and (3) baroreflex activation, due to the increasing blood pressure, which will increase the excitation of the cardiac vagal motoneurone pool (Spyer, 1994). During exercise the resultant cardiac vagal motoneurone activity will depend upon the relative intensity of these competing excitatory and inhibitory inputs and their integration, at the nucleus tractus solitarii (NTS) (Spyer, 1994). A measurement technique that is sensitive to vagal efferent-mediated events should reflect this. The spontaneous sequence analysis technique used to measure BRS is such a technique (Parati et al. 2000).

An attractive explanation of our findings during exercise, based on the model outlined above, is that the inhibitory effect of muscle mechanoreceptor activation is offset by baroreflex excitation, caused by the pressor response (Potts & Mitchell, 1998). The result is unchanged BRS about a higher arterial pressure. Of course, our data are averaged over 2 min of exercise, during which muscle chemoreflex activation is likely to increase (blood pressure rises) due to metabolite efflux. It might be expected, therefore, that at exercise onset of sufficient intensity and before muscle chemoreflex activation is established, muscle mechanoreceptor-mediated inhibition would predominate and BRS would then decrease.

The model can be extended to explain our findings during PECO. In the absence of central command, as in our experiments, muscle mechanoreceptor activation alone must fall abruptly on cessation of exercise thus reducing the inhibitory effects of the muscle mechanoreflex on vagal tone. Muscle chemoreceptor activation remains or, indeed rises, and this sustains the rise in blood pressure during PECO (Victor et al. 1988; Seals & Victor, 1991). This maintains the elevated baroreflex-mediated excitation input of the cardiac vagal motoneurones. As a result of these changes in the two inputs, there is a net increase in the excitation of cardiac vagal motoneurones and an increase in vagal tone, over which the muscle chemoreflex has little influence (O'Leary & Seamans, 1993). The increased vagal tone in PECO will then overpower the muscle chemoreflex-induced sympathoexcitatory effects on the heart, thus restoring heart rate towards baseline (O'Leary, 1996). Finally, it is worth considering why heart rate is prevented from falling well below baseline on cessation of exercise despite the continued marked elevation in blood pressure during PECO, and indeed why heart rate may show a small recovery during the later stages of PECO (Fig. 1). This may be because vagal tone is modulated by an input from polymodal (normally mechanoreceptive) afferents which are increasingly sensitised to threshold by accumulating metabolites (Kaufman & Rybicki, 1987; Sinoway et al. 1993) thus re-establishing some inhibition of cardiac vagal motoneurones.

There may be some limitations associated with the calculation of BRS using the spontaneous sequence analysis technique because it is a method that explores baroreflex gain during small spontaneous beat-to-beat oscillations of SBP. However, during exercise and PECO, the strength of the non-invasive spontaneous sequence analysis technique is that it permits measurements under natural conditions and provides a closed-loop estimation of BRS, i.e. changes in blood pressure induce changes in R-R interval which, in turn, are able to modify blood pressure. Although other methods, such as those using bolus injections of vasoactive drugs or neck chamber devices, might allow an estimation of baroreflex gain during greater changes of SBP, they are not without their own disadvantages (Casadei & Paterson, 1999; Parati et al. 2000).

In conclusion, our findings provide support for vagal modulation of baroreflex-sensed blood pressure changes during exercise and PECO. Increased BRS during PECO, following electrically evoked isometric exercise of the human calf muscles, is compatible with a reduction in the muscle mechanoreceptor-mediated inhibition of cardiac vagal motoneurones at a time of increased baroreceptor-induced excitation input. Functionally, activation of the muscle chemoreflex at some time after the onset of exercise may act to counter muscle mechanoreceptor inhibition of baroreflex driven, vagally mediated, events and thereby protect blood pressure from falling.


We gratefully acknowledge the help of Dr C. Ring with the spontaneous sequence analysis software and we thank Professor J. H. Coote for a most helpful discussion and his constructive comments on the manuscript. This work was supported by the British Heart Foundation BHF PG/99148.