Respiratory sinus arrhythmia (RSA) is an acceleration of heart rate during inspiration and deceleration with expiration. We asked whether or not in humans some of the volume-related information necessary for RSA originated from the chest wall. Men and women, 19–20 years old, were breathing supine. Rib cage and abdomen displacement provided an index of tidal volume (VT) and RSA was computed breath-by-breath from the peak and trough of instantaneous heart rate. First, measurements were taken during breathing at rest (protocol a, 129 male and 164 female). Then, in subgroups of the original subject population, measurements were collected for the first five breaths immediately following a brief breath-hold period (protocol b), predominantly with the rib cage or predominantly with the abdomen (protocol c), above functional residual capacity or below it (protocol d). As long as VT was constant, severe chest wall distortion (protocol c) did not modify RSA. A drop in absolute lung volume (protocol d) or an increase in VT (protocol b) respectively decreased and increased RSA. The results, globally taken, are compatible with the notion that in humans changes in lung volume are detected by lung mechanoreceptors, whereas chest wall reflexes play no role in RSA. No difference in RSA emerged between genders during resting breathing or modest breath-hold hyperventilation.
In humans, the mechanism responsible for the lung-volume information necessary for RSA is still unclear. We hypothesized that the configuration of the chest wall could be a contributing factor. However, the results of various respiratory manoeuvres, globally taken, have indicated that the chest wall gives no contribution to RSA.
This implies that in humans during eupnea the vagal information of pulmonary origin, although irrelevant for the control of the breathing pattern, plays a major role in the respiratory-cardiac co-ordination.
The stability of RSA in face of major changes in chest wall configuration may suggest that RSA is a mechanism of greater importance than commonly recognized, possibly designed to fine-tune the pulmonary ventilation-perfusion matching.
Probably the best known variation in heart rhythm is the respiratory sinus arrhythmia (RSA). It consists of the acceleration of heart rate (HR) during inspiration and its reduction during expiration. This pattern has been documented in many species; in humans, it must have been recognized centuries ago and recorded at least since the nineteenth century . The evolutionary development of RSA is not known, but animal studies have shown that RSA improves the efficiency of gas exchange by ameliorating the coupling between pulmonary blood flow and alveolar ventilation, or ventilation–perfusion matching [2,3]. The mechanistic basis of the phenomenon has been worked out mostly on the basis of studies on anaesthetized animals, of which the classic work on dogs  with pulmonary vagal denervation is emblematic. From that and other subsequent work, the simplest mechanistic interpretation of RSA is that, as lung volume increases, the central projection of the pulmonary vagal afferents to the nucleus of the tractus solitarius temporarily inhibits the closely located nucleus ambiguus (or cardio-inhibitory centre); this reduces the vagal output to the sino-atrial node and increases HR. The opposite occurs in expiration. Hence, RSA would be a reflex inhibition of vagal tone mediated by the pulmonary stretch receptors.
Additional mechanistic possibilities do exist; among them, since the earliest experiments with cross circulation between anaesthetized dogs , the possibility emerged of a central interaction between the respiratory and cardio-inhibitory centre . Later studies have indicated that many factors can influence RSA, broadly categorized as peripheral and central mechanisms ; these factors are not mutually exclusive and could gain a different degree of importance in different circumstances. Because it involves the inhibition of the parasympathetic arm of the autonomic control of the heart, RSA often is taken as an indicator of vagal tone, not only in physiology  but also in biomedical and psychological assessments (e.g. ).
In humans, mechanistic studies on the origin of RSA are very few. It is clear that RSA is lung volume-dependent and that its magnitude decreases with the increase in breathing rate [7,10,11]. What remains unclear is the importance of pulmonary stretch receptors in determining the afferent limb of the reflex response, given that in humans these receptors have minimal, if any, role in the regulation of breathing during eupnea [12–15]. However, in patients who underwent lung transplant (and therefore were deprived of the pulmonary afferent information but maintained functional cardiac control) RSA was greatly reduced; what was left was just slightly above the RSA detectable in heart-transplant patients . These results indicated an essential contribution of vagal pulmonary afferents to RSA, although did not exclude the possibility that, in intact subjects, other peripheral and central factors may also be contributing. For example, in normal subjects, passive positive-pressure ventilation large enough to inhibit spontaneous respiratory activity greatly attenuated RSA, suggesting a major role of a central component . Hence, it is possible that the pulmonary afferents play a permissive role to extra-pulmonary contributors to RSA. In the present study, we considered the possibility that the chest wall may send information on breathing pertinent to RSA.
During breathing, the chest wall changes its volume in unison with the lungs, but whether or not it may participate to RSA has been investigated neither in animals nor in humans. Ventilatory reflexes originated by the chest wall muscle spindles and tendon organs are present in humans [17,18]. In addition, the chest wall is responsible for breathing-related changes in pleural and abdominal (abd) pressures, which determine the thoraco-abd pump and venous return to the heart. Hence, the chest wall could have a role in RSA with neural afferent information and mechanically through changes in thoraco-abd pressures. In the present study, we asked whether or not in humans the configuration of the chest wall may influence RSA. In one experiment, the subjects were asked to breathe either predominantly with the thorax or predominantly with the abdomen, in order to grossly distort the chest wall, while maintaining similar absolute and tidal volumes. In a second experiment, subjects were invited to breathe with similar tidal volumes either above or below functional residual capacity (FRC), which involved recruitment of different sets of muscles. In this case, both chest wall configuration and absolute lung volume were altered, with no changes in tidal volume. Finally, RSA was examined after a brief period of breath-hold, which increased the chemical drive and caused a modest rise in tidal volume with no major changes in chest wall configuration.
The parasympathetic arm of the autonomic control of the heart, commonly evaluated from the high frequency–low frequency (H–L) ratio of the power spectrum of instantaneous HR , is thought to be more pronounced in women than in men [20–23]. Therefore, RSA, if it were strictly a manifestation of the inhibition of parasympathetic control, could be more marked in women, although cardiac vagal tone and RSA may be parallel, not consequential, phenomena . The published data of RSA in men and women do not permit a firm conclusion because of the small number of subjects investigated (e.g. [6,11,22]) and the large variance in age, a factor known to influence cardiac parasympathetic control . Therefore, the additional purpose of the present study on almost 300 subjects of similar age was to examine the possibility of a male–female difference in RSA during eupneic breathing.
MATERIALS AND METHODS
The subject population consisted of university undergraduate students of either gender, between 19 and 21 years of age, non-smokers, free from cardio-respiratory disorders or any medical treatment. The study was approved by the Ethics Committee of the Institution. Subjects were explained the type of measurements and the respiratory manoeuvers to be performed, and signed an informed written consent to this effect; however, they remained unaware of the purpose of the study.
All experiments were conducted in the early hours of the afternoon. The subject was instrumented for measurements of ECG and breathing movements. ECG was obtained by standard peripheral derivations. The inter-beat interval (IBI, ms) was electronically computed and converted into instantaneous heart rate (HR′, beats/min) according to HR′=60/(IBI/1000). The accuracy of the electronic conversion of IBI into HR′ was tested by delivering electrical signals with a stimulator at known rates; no discrepancy was detected over the range of HR measured.
Breathing movements were recorded by impedance pneumographs. The pneumographs consisted of two Piezo Respiratory Belt Transducers (ADInstrument), one positioned around the rib cage (rc) (approximately at the level of the fifth intercostal space, the second one positioned around the abdomen at the level of the umbilicus. The piezo-electric transducer responds linearly to changes in length and permits to measure changes in thoracic or abd circumference during respiration. In addition to the detection of breathing rate, the electronic sum of the rc and abd signals provides a good representation of the changes in tidal volume (VT). In fact, in supine men during breathing the chest wall moves with one degree of freedom, where at any lung volume the expansion of rc is accompanied by a volume-equivalent decrease in abd, and vice versa [24,25] with minimal additional distortion between the two compartments . In the present study, we considered the rc–abd sum as an index of VT, in arbitrary units.
The rc and abd signals were digitally acquired at 100 Hz and the ECG at 400 Hz (Powerlab®; ADInstrument). These signals and the electronically computed beat-to-beat HR′ and rc–abd sum were displayed on a computer monitor and saved for subsequent analysis.
The subject was laying supine on a comfortable mattress with no visual access to the computer monitor. For 30–60 min, the subject practised the specific respiratory manoeuver(s) assigned. Data during breathing at rest were collected in all subjects, whereas the remaining protocols were performed by subgroups. Number and characteristics of the subjects for each of the respiratory manoeuvers are given in Table 1.
|.||Males .||Females .|
|Parameter .||n .||Age .||Height (cm) .||W (kg) .||n .||Age (years) .||Height (cm) .||W (kg) .|
|.||Males .||Females .|
|Parameter .||n .||Age .||Height (cm) .||W (kg) .||n .||Age (years) .||Height (cm) .||W (kg) .|
The subject was given the option to wear an ear set and close the eyes to minimize external disturbances, but attention was paid to maintain wakefulness. Data were collected continuously until the breathing pattern appeared stable and reproducible. Analysis was performed on at least two epochs of 10 consecutive breaths and the results averaged (Figure 1).
Experimental recordings in two subjects during resting breathing
This test meant to extend the comparison between genders to a ventilatory pattern under a modest chemical stimulus, with increased VT and no important changes in chest distortion. After a period of resting breathing, the subject was instructed to hold his breath at FRC for 10 s, monitored by a stopwatch (Figure 2). Separate measurements had indicated that the end-tidal CO2 of the first breath upon termination of the breath-hold period was approximately 5 mm Hg above the resting value, and the alveolar O2 was correspondingly lower. A chemical stimulus was expected to cause some increase in VT without appreciable changes in chest wall distortion. The subject performed the manoeuver several times. Then, based on the stability of the breathing pattern preceding the breath-hold, the two best runs were picked for analysis. Each of the first five breaths following the breath-hold were analysed singly and the results compared with the average values of the ten breaths immediately preceding the breath-hold.
RSA (in bpm) during the first five breaths following a brief period (10 s) of breath-holding (indicated by the thick horizontal line)
This exercise consisted of breathing predominantly with the thorax or predominantly with the rc while maintaining a similar VT, as assessed by the rc+abd signal. The goal was to compare RSA between two very different conditions of chest wall configuration with minimal variations in absolute lung volume and VT (Figure 3). Each manoeuver was repeated several times alternating rc and abd breathing. For each of the two conditions the two runs with the greatest stability of the breathing pattern were chosen for analysis; then, ten consecutive breaths were analysed and data averaged.
RSA (in bpm) while breathing a similar
VT (rc+abd) either predominantly with the rib cage (thoracic breathing at left) or predominantly with the diaphragm (abdominal breathing at right)
In humans at rest breathing consists of an inspiration above FRC followed by relaxation of the respiratory muscle and consequent passive expiration to FRC. Differently, breathing below FRC requires, first, active expiration by contraction of the abd muscles, followed by relaxation to FRC. Hence, while breathing below FRC the mean absolute lung volume is lower than normal, by a magnitude that approximately corresponds to VT. This manoeuver was found difficult by many subjects; in the end, useful data (that is, with similar tidal volumes during breathing above and below FRC) were collected in 34 males and 35 females (Table 1), or less than half of the subjects that attempted it. Each manoeuver was repeated several times, alternating breathing above and below FRC. For each manoeuver, analysis was performed on ten consecutive breaths of the two runs with the greatest stability of the breathing pattern and data averaged.
For each of the breaths analysed, the peak and trough HR′ were identified and measured; their difference represented RSA (beats/min) [11,16]. Because rc, abd and VT (VT=abd + rc) were in arbitrary units, first, the data of each breath were expressed as percent of the corresponding average value during resting breathing. This step permitted to normalize the rc and abd signals irrespective of their absolute values, which varied among subjects according to gender and body size.
For each subject, the results of the two runs in protocols b, c and d were averaged. Then, results of all subjects of the same gender were pooled together to obtain the group grand mean. Statistical comparison of the data between the two genders (protocol a) was done by two-tailed t-test. Differently, statistical difference in RSA between manoeuvers (e.g., abd compared with thoracic breathing in protocol c, breathing above or below FRC in protocol d) and between genders were evaluated by two-way ANOVA, of which the manoeuver and gender represented the two grouping factors. Group data are reported as means±1 S.E.M., unless otherwise indicated. Differences between groups were considered statistically significant at P< 0.05.
During breathing at rest all 293 subjects presented a clearly identifiable RSA, although variability among subjects was large. Of the total population the average RSA was 8.6±1.3 bpm (beats/min), a few cases exceeding 20 bpm (Figure 4). One extreme example of very large RSA is the subject presented in Figure 1 (bottom panel). In males (n=129) and females (n=164), RSA averaged, respectively, 8.1±0.4 bpm (min 2, max 31.7) and 9.9±0.5 bpm (min 2.5, max 35.2), with no significant difference between genders.
RSA (in bpm) as function of body weight in males (filled triangles) and females (open circles)
Resting HR averaged 68.7±1 bpm (male) and 69.6±0.8 bpm (female) (P > 0.05); the corresponding values for breathing frequency (F) were 14±0.4 breaths/min and 14.3±0.3 breaths/min (P > 0.05). No significant correlation emerged between RSA and body height or weight (Figure 4) nor between RSA and mean HR (Figure 5, left panel). Differently, in both genders RSA decreased significantly with F (Figure 5, right panel). The F–RSA data points did not differ significantly between genders. The linear regression through the data points of both genders combined (dashed line) was RSA (bmp)=−0.56 F (breaths/min) + 16.92 (r2=0.95).
RSA (in bpm) as function of mean heart rate (left panel) or breathing rate (right panel) in males (filled triangles) and females (open circles)
RSA after breath-holding
Following 10 s breath-holding all subjects increased VT; then, after two to three breaths, tidal volume returned to normal. During this time, also RSA increased, exclusively because of the increase in the peak HR′, whereas the trough HR′ remained unaltered (Figure 6, top panels). In the following breaths peak HR′ decreased and so did RSA. The time course of RSA during the first five breaths following breath-hold closely resembled that of VT (Figure 6, bottom panel), with good correlation between the percent changes in VT and those of RSA in both males (r2=0.99) and females (r2=0.96) (Figure 7). The slope of the VT (%)–RSA(%) relationship (Figure 7) was 1.47 (±0.07 S.D.) and 1.03 (±0.12 S.D.) in men and women, respectively, slightly, yet significantly, different from each other (P< 0.02).
First five breaths following a brief breath-hold period
RSA as function of tidal volume during the first five breaths immediately following a 10-s period of breath-hold
Thoracic or abd breathing
In both males and females during thoracic breathing, the rc compartment contributed 60–70% of VT, and during abd breathing it contributed approximately 10% (Figure 8, left). Neither VT nor HR differed significantly between the two breathing patterns. Peak and trough HR′ were very similar during thoracic and abd breathing in both men and women; hence, RSA did not vary between manoeuvres (Figure 8, right).
Thoracic breathing versus abdominal breathing
Breathing above or below FRC
In both males and females, breathing below FRC resulted in ~10–15% greater contribution of the thorax to VT, because of the recruitment of the abd expiratory muscles (Figure 9, left). At the same time, breathing below FRC decreased the magnitude of RSA by approximately 45%; in males, the drop was from 8 ±0.8 bpm to 4.2±0.4 bpm (P< 0.0001), in females from 8.9 ±1 bpm to 5.3±0.6 bpm (P< 0.0001) (Figure 9, right). In both genders, the decrease in RSA resulted from some drop in peak HR′ and rise in trough HR′, with unaltered mean HR, which averaged 69±2 bpm both in men and in women for both breathing patterns.
Breathing above FRC versus breathing below FRC
The findings that RSA had large inter-subject variability (Figure 4) and was sensitive to breathing rate (Figure 5) were confirmatory of previous reports [10,11,16]. Equally, the fact that the increase in RSA with the increase in VT occurred almost exclusively because of the rise in peak HR′ (Figure 6) agrees with the most common mechanistic interpretation of RSA as a lung volume-dependent inhibition of the cardio-inhibitory centre [3,8]. The main novel question addressed by the current experiments was the possible role of the chest wall on RSA. The results indicated that even major changes in chest wall configuration had no impact on RSA. In addition, the analysis of a large group of young subjects of very similar age did not reveal obvious gender-related differences in RSA, whether at rest or during the modest hyperventilation after a brief breath-hold.
We elected to quantify RSA from the within-breath difference between HR peak and trough, rather than from the high-frequency (HF) band (0.15–0.45 Hz) of the power spectrum of a sequence of IBIs [19,20]. The measurements of peak and trough HR′ within each breath [11,16] permits an accurate breath-by-breath quantification of RSA not possible from analysis of the frequency spectrum.
Breathing was recorded by non-invasive means, rather than by pneumotachography or body plethysmography, to minimize the possibility that the instrumentation may cause some discomfort and changes in breathing pattern and blood gases, with repercussion on RSA [27,28]. The estimate of VT from rc and abdomen movements is valid in humans  because in the supine posture the chest wall moves as a two-compartment model with only one degree of freedom even during forced thoraco-abd manoeuvres . The major drawback of this approach is that we could not determine with certainty the change in absolute lung volume (protocol d). This would have required separate approaches and instrumentations (body plethysmography, dilution techniques) not feasible for supine subjects or difficult to couple to natural resting breathing.
Chest wall configuration
Breathing predominantly with the thorax or predominantly with the abdomen-diaphragm obviously introduced large differences in the configuration of the chest wall (protocol c). Chest wall afferents are known to influence the phrenic neural output and breathing [17,18], but in this protocol VT was voluntarily maintained nearly constant. Therefore, the fact that RSA was unaltered (Figure 8) should lead to the inescapable conclusion that the configuration of the chest wall, and the mechanical and neural implications of it, bear no relevance to RSA.
At first sight, this conclusion may seem in contradiction with what resulted when breathing forcibly below FRC (protocol d) because this pattern of breathing caused chest distortion and decreased RSA by half (Figure 9). Any interpretation of this result needs to take into account that mean HR did not change between the two breathing patterns and that the reduction in RSA when breathing below FRC was contributed by a decrease in peak HR′. The intermittent compression of the abdomen when breathing below FRC may have altered venous return and influenced HR through the Bainbridge reflex. However, this reflex, which consists in inhibition of the parasympathetic control of the heart following stimulation of the atrial mechanoreceptors, is of minor importance in humans [29,30]. Furthermore, the reflex causes an increase in HR, which is not what was observed.
A more likely explanation is that breathing below FRC substantially lowered the activity of the pulmonary stretch receptors. We do not have measurements of absolute lung volume, but it is most probable that breathing below FRC caused a drop in lung volume equivalent to VT, or approximately 500 ml out of 2.6 l . This means that the afferent inputs of the pulmonary stretch receptors to the nucleus of the tractus solitarius decreased by approximately 20%. This drop, in turn, according to the mechanistic view of RSA [3,8] decreased the inspiratory inhibition on the nucleus ambiguus, which raised the parasympathetic outflow and lowered peak HR′. This interpretation of the results of protocol d explains the fact that RSA decreased with no changes in mean HR. In summary, the drop in RSA during breathing below FRC is compatible with the conclusion from protocol c that chest wall afferents play no role in RSA.
During eupneic breathing, we could not detect any significant difference in RSA between genders, despite the very large number of subjects (Figure 4). With the exception to be discussed below, differences in RSA did not emerge even during the respiratory manoeuvres oriented to modify chest wall configuration (protocols c and d) or increase ventilation (protocol b). This negative result may seem surprising in light of the gender differences in vagal tone, as evaluated from the HF band of the spectral analysis of IBIs. In fact, various studies have indicated a higher HF–LF ratio in women than in men [20–23], although the higher sympathetic control in men, which raises the LF band of the spectrum, may be a contributing factor . One possibility is that the stronger cardiac vagal control in women may be accompanied by a weaker central integration of the pulmonary afferent information; in such a case, the similar RSA between men and women may be the net effect of opposing factors, the vagal control on one side (stronger in women) and the pulmonary afferent information and its central integration on the other side (stronger in men). It is worth noticing that when RSA was compared between genders not in absolute terms but in relation to changes in VT, a gender-difference in RSA emerged clearly (Figure 7). The possibility that RSA and vagal tone may not be uniquely related has been suggested before . Finally, we cannot exclude that gender differences in RSA may depend on the age of the subjects. In this respect, it is of interest that the higher HR of women is found consistently only after 20 years of age, presumably because before that age neural differences in cardiac control are still blurry (reviewed in ).
Measurements in a large population of young men and women have indicated no difference in RSA between genders. Several factors could have contributed to this result and the possibility that gender differences in RSA may become clear at a later age remains open. Contrary to the hypothesis put forward at the onset of the study, the results gave no indication that in humans the configuration of the chest wall plays a role in RSA. Hence in humans, the lung volume-related information essential for RSA must originate from the mechanoreceptors in the lungs. This should imply that in humans during eupnea the pulmonary vagal information, although irrelevant for the control of the breathing pattern, plays a major role in the respiratory cardiac co-ordination and ventilation–perfusion matching.
Jacopo Mortola conceived the study, analysed the data and wrote the manuscript. Domnica Marghescu and Rosmarie Siegrist-Johnstone set up and checked the instrumentation, contributed to data collection and reviewed the manuscript.
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.