INTRODUCTION

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

ALTHOUGH OSCILLATIONS in blood pressure and heart rate were identified over 100 years ago, it was the notion that certain frequencies may be indicative of either sympathetic or parasympathetic tone (3) that stimulated great clinical interest in using measures of cardiovascular variability as diagnostic tools. Extensive effort has subsequently been invested in describing changes in cardiovascular variability in a range of physiological and pathological conditions. Unfortunately, the momentum of this clinical research has continued without reflection and understanding of the fundamental causes of such variability and without reference to the possibility that not all changes in cardiovascular variability are indicative of changes in autonomic function. The aim of this review, therefore, is to "take stock," to describe recent work on the properties of the component parts that govern variability at different frequencies, and to point out pitfalls and possibilities for future research.

There is no doubt that analysis of variability in heart rate and blood pressure has proven useful in understanding cardiovascular regulation in a range of conditions, including heart failure, diabetes, and hypertension. Initial studies simply used global indexes of variability (such as the standard deviation of heart rate) rather than defining variability at any particular frequency (77, 93). Nevertheless, these studies have shown relationships between reduced heart rate variability and coronary artery disease (17), atrial fibrillation, and heart failure (25). Importantly, reduced heart rate variability is associated with increased mortality after myocardial infarction (60, 61) and shown to be a better predictor of death due to progressive heart failure than other conventional clinical measurements (92). Rather than structural changes in the sinoatrial node limiting heart rate, alterations in autonomic outflow were proposed as the origin for reduced variability. Because sympathetic activity is well established to be elevated in heart failure (31, 32) and coronary artery disease (78) and may be associated with the initiation of hypertension (54), a diagnostic index of autonomic tone could be of immense value. However, after nearly 20 years of reports describing changes in cardiovascular variability and its possible origins, it is apparent that no test has been accepted as providing prognostic or diagnostic information over that which can be obtained by more conventional tests. Why is this so? Is it that measurements of variability have been inappropriately applied or that the data obtained misinterpreted? If so, are there specific conditions under which a measure of variability is a definitive index of autonomic function?

	WHAT FREQUENCIES ARE PRESENT?

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

The first recorded measurements of oscillations in the cardiovascular system were described by Mayer in 1876. It is interesting to examine his original paper in light of current research because the oscillation at 0.1 Hz in humans is often referred to as the "Mayer wave." Mayer's experiments were conducted in anesthetized spontaneously breathing rabbits in which he observed pronounced oscillations in blood pressure (amplitude between 15 and 40 mmHg). There are several points that indicate it is unlikely these oscillations are of the same etiology as those to which we often give his name. First, the frequency he observed was ~0.05 Hz, whereas the sympathetically mediated oscillation in blood pressure seen in rabbits is ~0.3 Hz (52). Such slow oscillations have generally been observed under conditions of low blood pressure or deteriorating preparations and are thought to be myogenic in origin (64, 71). Finally, sympathetically mediated oscillations in blood pressure have never been reported in anesthetized rabbits, and indeed in conscious rabbits the amplitude of the oscillations is generally below 5 mmHg (unpublished observations).

It was another 75 years after Mayer before interest in cardiovascular variability was rekindled with the suggestion that baroreceptors and sympathetic outflow were important in regulating the oscillations. In 1951, Guyton and Harris (37) proposed the term "vasomotor waves" to refer to slow oscillations in blood pressure that were unrelated to respiration. This definition was based on a search of recordings from their experiments previously undertaken in anesthetized dogs. They found that oscillations were most evident in animals that had lost 20-25% of their blood volume. Importantly, baroreceptor denervation or spinal anesthesia abolished these oscillations. There are some similarities with Mayer's work in that the period of the oscillations was between 15 and 30 s (0.06-0.03 Hz) and thus was considerably slower than the oscillation that we now know to involve the sympathetic nervous system. Furthermore, the amplitude of the oscillations was between 20 and 40 mmHg. Although Guyton and colleagues were adamant that such oscillations were autonomic in origin, as is discussed later, a reduction in the strength of an oscillation with baroreceptor denervation or sympathectomy is by itself not evidence that such oscillations are generated by the autonomic nervous system. Rather, all that can be concluded is that the autonomic nervous system is involved.

A variety of animal and human research has now established two clear frequency bands in heart rate and blood pressure with autonomic involvement. These bands are an oscillation associated with respiration and one of slower frequency at 0.1 Hz in the human (69, 110), 0.14 Hz in the dog (2, 21), 0.3 Hz in the rabbit (Fig. 1) (52), and 0.4 Hz in the rat (15). One approach with regard to heart rate variability has been to calculate the ratio of the two spectral powers in each frequency band as an index of sympathovagal balance (68, 83, 95). Subsequently, there have been numerous papers reporting changes in the sympathovagal balance in such varied conditions as anesthesia (45), sleep (7), and the menstrual cycle (104). However, as discussed below, a number of major criticisms have been raised that question the validity of such an index.

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Example of a 0.3-Hz oscillation in one rabbit. Top trace, mean arterial pressure (MAP); bottom trace, renal sympathetic neuron activity (SNA) measured in arbitrary units. Redrawn from Malpas and Burgess (71).

	RESPIRATORY OSCILLATION IN HEART RATE AND BLOOD PRESSURE

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

The respiratory oscillation in blood pressure can be ascribed to the cyclic variation in intrathoracic pressure, with breathing mechanically perturbing venous return, cardiac output, and thus blood pressure. The important point is that such changes are detected by baroreceptors that cause changes in autonomic activity to the heart and therefore change heart rate. Because of the limited ability of the heart to respond to sympathetic influences at the respiratory rate (see FREQUENCY RESPONSE OF HEART RATE), it is thought that the respiratory-associated oscillation in heart rate is mediated via the vagus. Vagal blockade with atropine abolishes this oscillation in heart rate (106, 124). One proposal is that heart rate does not inherently contain a respiratory-related oscillation without it also being present in blood pressure, suggesting it is a function of the baroreflex. However, there is good evidence from both human and animal experiments that a major cause of sinus arrhythmia is central coupling of respiratory drive to cardiac vagal motor neurons (38, 96). Evidence for the central mechanism in humans may be seen at the onset of an inspiratory breathhold (26) or just before the termination of a prolonged breathhold (127), where changes in heart rate occur in the absence of respiratory movements. Under some conditions (e.g., atrial pacing), it appears that respiratory-related oscillations in heart rate can contribute to respiratory oscillations in blood pressure in a feed-forward relationship (5, 122). Removal of the respiratory oscillation in heart rate with vagal blockade also reduces the oscillation in blood pressure by ~50% (121). In humans, the relative contribution of central versus baroreflex-imposed changes in heart rate are difficult to determine precisely under the closed-loop (baroreceptor intact) condition of the studies.

Because there is good agreement that the respiratory oscillation in heart rate is mediated via the vagus, a logical proposal may be that changes in heart rate variability at this frequency are indicative of "vagal tone" (67). This appears incorrect for several reasons. First, if the changes in vagal activity are even partly induced by baroreceptor sensing of a respiratory oscillation in blood pressure, any measurement of the strength of the oscillation in heart rate at this frequency must be indicative of all components in this reflex. Second, a number of studies have shown that other nonneural factors can influence the strength of this oscillation. Factors such as reduced respiratory capacity (106) and body position (122) that alter the amplitude of the oscillation in blood pressure in turn alter the amplitude of the heart rate oscillation at this frequency. In this case, the oscillation is still being produced via the vagus and the baroreflex pathway, but because the amplitude of the oscillation in baroreceptor input is reduced, those in the heart rate are also smaller. Thus, in terms of developing a diagnostic tool that measures the heart rate variability associated with respiration, the main problem is that it will be difficult to compare between different patient groups or individuals who may have different levels of respiratory capacity. Controlling ventilation for both rate and depth is likely to improve the reproducibility of measuring the heart rate variability associated with respiration; however, it remains to be established whether such controls are viable when comparing across different patient groups. Whereas a longitudinal comparison within a patient is potentially possible, again one would have to be sure that respiratory variables were unchanged to ensure that changes in heart rate variability were solely reflective of baroreflex/vagally mediated changes in heart rate.

	SLOW OSCILLATIONS

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

As has been previously noted, the 0.1-Hz oscillation in blood pressure in humans is analogous to 0.3 Hz in rabbits (Fig. 2) and 0.4 Hz in rats. This species difference is accounted for by the larger physical distance between the components producing the oscillations (e.g., arterial baroreceptors, the brain, and various target organs) in humans compared with rabbits and rats, which means that conduction times for the afferent and efferent signals will be longer (102). The slow oscillation in blood pressure and heart rate (often referred to as the low-frequency component) is probably the most contentious aspect with respect to cardiovascular variability. Two opposing viewpoints have been proposed: 1) that spectral analysis of heart rate and blood pressure offers a useful marker of sympathetic and parasympathetic tone (3, 69), or 2) that they are largely an index of baroreflex gain (10, 27). With regard to the slow oscillation in blood pressure, it is accepted by both sides that it does involve the action of the sympathetic nervous system on the vasculature as high spinal transection or ganglionic blockade abolishes the oscillation (19). The debate is really whether arterial baroreceptors are also involved in its generation, and both sides have been able to mount convincing arguments for each proposal based on a range of experimental evidence. The fundamental basis for each view lies in the hypothesized origin of the slow oscillation; those researchers believing that the oscillation reflects sympathetic tone have proposed a central oscillator model, whereas those advancing the index of baroreflex gain have proposed a baroreflex feedback model.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   A: power spectrum of renal SNA from a single rabbit illustrating the dominant cardiac (4 Hz) and respiratory-related (1.7 Hz) oscillations. B-D: power spectrums of SNA from individual conscious rabbits before and during the initial phase of hemorrhage [B: 1.35 ml · min1 · kg1 for 10 min, mean increase in SNA 21% (73)], after blood volume expansion [C: 1.5 ml · min1 · kg1 for 15 min, mean decrease in SNA 25% (62)], and during noise stress [D: mean increase in SNA 21% (72)]. Note difference in the strength of the 0.3-Hz rhythm in SNA between rabbits under resting conditions.

Central Oscillator Theory

Baroreflex Feedback Loop Theory

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Schematic figure representing component sections responsible for slow oscillation in blood pressure. In this proposal, arterial baroreflexes play a critical role in generating slow oscillation through a reflex resonance loop involving SNA and the vasculature, but importantly other reflex pathways or the central nervous system can modulate the strength of this oscillation. Insets: transfer function gain between arterial pressure (AP) and aortic nerve activity (ADN) (105), between arterial pressure and SNA (44, 56), and between SNA and renal blood flow (RBF) (36). All data were obtained in anesthetized rabbits, and in all cases, the gain was normalized to the values obtained at the lowest frequencies. Time delays shown indicate the lag between in input and the output.

Strong support for the baroreflex model comes from evidence that removal of different sections of the feedback loop reduced the strength of the slow frequency oscillation. Sympathectomy or combined - and -adrenoceptor blockade decreased spectral power at 0.4 Hz in rats (analogous to 0.1 Hz in the human) (20) and critically sinoaortic denervation decreased the 0.4-Hz oscillation in blood pressure (18, 19, 29, 50). The oscillation in heart rate appears to be driven via the baroreflex mechanism because it is not produced in the absence of a 0.1-Hz pressure oscillation (22). One concern with such "removal" experiments is that the interventions do not produce pure responses and are likely to cause compensation in other cardiovascular control mechanisms. Therefore, it is pertinent that a range of studies have found that fluctuations could also be produced. Bertram et al. (14) used electrical stimulation of the aortic depressor nerve to produce fluctuations in blood pressure. Importantly, by stimulating at different frequencies, Bertram et al. (14) found evidence for a resonant frequency near 0.4 Hz in the rat. The resonant frequency is derived from the time delay between the stimulus and the blood pressure response and refers to the frequency at which the stimulus and response are in phase. In humans, fluctuations at 0.1 Hz in blood pressure could be produced with perturbations in baroreceptor input via neck suction (10).

Strength of Slow Oscillation in Blood Pressure Does Not Indicate Mean Level of SNA

What Can Mathematical Models Tell Us?

A Unifying Theory to Account for Slow Oscillation in Blood Pressure?

A new theory is proposed to account for the slow oscillation in blood pressure to unify the range of seemingly conflicting data. It is suggested that arterial baroreflexes play a critical role in generating the slow oscillation through a reflex resonance loop involving SNA and the vasculature, but importantly other reflex pathways or central nervous system components involved in regulating SNA can modulate the strength of these oscillations (Fig. 3).

Unfortunately, rather than simplifying the explanation for the slow oscillation, this hypothesis makes it extremely complex. The important implication for measurement of the slow oscillation in heart rate and blood pressure as an index of baroreflex sensitivity is that one cannot guarantee that factors other than baroreflex sensitivity are not responsible for the changes observed.

Does Vasculature Influence Slow Oscillation in Blood Pressure?

Do Changes in Slow Oscillation in Blood Pressure Reflect Oscillations in Global Sympathetic Activity?

With reference to the slow oscillation in blood pressure, if we accept that the oscillation involves the arterial baroreflex, then it follows that, because baroreflexes differentially modulate sympathetic drive to different organs (89-91), it is SNA to a few key organs that may dominate in the production of the slow oscillation in blood pressure. In particular, SNA to the lungs, kidney, and spleen is highly baroreceptor sensitive (90, 109). However, SNA to the skin is only weakly regulated by baroreceptor activity (89). The importance of neural regulation of the renal vasculature in determining the strength of the slow oscillation in blood pressure was illustrated in a recent study where hemorrhage was used to activate the sympathetic nervous system in rabbits and resulted in increases in the strength of the slow oscillation, and the study found that renal denervation greatly reduced the strength of the induced oscillation (71). It is unknown whether other organs such as the gut with its quantitatively large portion of cardiac output and evidence of barosensitive neurons (51) also play an important role in governing the slow oscillation. With regard to humans, there is some debate because studies have found little evidence of baroreflex modulation of skin SNA (28, 129), but stimuli from baroreceptors appear to produce corresponding oscillations in the skin vasculature up to 0.1 Hz that lead oscillations at the same frequency in blood pressure (9). One caveat with this work, however, is that even if skin blood flow does contain a neurally mediated oscillation, it cannot be assumed that this contributes to the associated oscillation in blood pressure because other inputs from sources such as the kidney may still dominate.

Several studies have explored the possibility of measuring the 0.1-Hz oscillation in skin blood flow as an index of sympathetic activity to the vasculature or baroreflex gain (11, 101). Not withstanding the limitations of measuring the strength of such oscillations in the first place, changes in the strength of this oscillation in skin blood flow are unlikely to be due to SNA directed solely toward the skin vasculature but are more likely to reflect the effect of SNA to organs that are strongly baroreceptor related, e.g., the kidney. As a result, any change in the strength of the oscillation in skin blood flow would be more likely to be driven by the oscillation in blood pressure in a direct pressure-flow relationship rather than through changes in skin vascular resistance.

	FREQUENCY RESPONSIVENESS OF COMPONENTS INVOLVED IN PRODUCING OSCILLATIONS

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

In discussing the generation and regulation of various oscillations in blood pressure and heart rate, it must be stressed that if oscillations are the result of feedback (e.g., baroreflex feedback), then it is the ability of the individual components of the feedback pathway to respond to inputs that play the critical role in determining the strength and frequency of such oscillations. Conceptually, all of the steps involved in the pathway can affect the gain between the input and the output, because the total gain is a combination of all the gains relating to the respective steps, e.g., blood pressure to sympathetic response and sympathetic activity to vasculature response (Fig. 3). In contrast, a rate-limiting step, which determines how quickly the system can respond to dynamic changes in the input, mainly affects the parameters of the low-pass filter (natural frequency and damping coefficient).

Using a model of isolating the aortic baroreceptors from arterial pressure, Sato et al. (105) were able to show that alterations in pressure to the baroreceptors were transduced to changes in afferent aortic nerve activity with increasing gain up to 2 Hz and a cutoff frequency of 3-4 Hz. These results suggested that the baroreceptors responded primarily to dynamic rather than static changes in pressure. With regard to the central component of the baroreflex loop, it has been established that the response between a change in blood pressure to a change in SNA displays a high-pass characteristic above 0.1 Hz where the size of the SNA response to baroreceptor input becomes greater as the input frequency increases (Fig. 3) (44). It has been suggested that the high-pass characteristic of the sympathetic response is attributable solely to the dynamic transduction properties of the baroreceptors and that the central nervous system processing is a simple all-pass filter rather than a high-pass filter and does nothing in particular except for inverting the sign of the signal and adding some delay time (~440 ms in the rabbit) (59). However, more recent experiments in which baroreceptors were stimulated and SNA to the heart and kidney simultaneously was measured revealed that the frequency response is different to each organ indicating differential central processing (57).

Whereas the various components that give rise to the total frequency response can be described, it should be noted that there is evidence indicating that the components do not necessarily simply summate to give the complete response. In particular, Ikeda et al. (44) proposed that the rapid neural component of the baroreflex arc compensates for the slow peripheral vasculature response resulting in enhanced blood pressure stability.

With regard to the vasculature, the presence of an oscillation between 0.1 and 0.4 Hz (depending on the species) requires that sympathetic transmission at the ₁-receptor or other receptor (e.g., purinergic) sites be fast enough to transmit oscillations in this frequency range. Initial experiments performed in 1968 by Rosenbaum and Race (103) using patterned stimulation of the lumbar sympathetic trunk in an isolated hindlimb of the dog suggested that the vasculature was particularly sluggish to frequencies present in SNA above 0.017 Hz (~60 s wavelength). However, more recent experiments by Stauss and colleagues (116) using electrical stimulation of sympathetic nerves at different frequencies showed that the mesenteric vasculature was easily able to follow SNA frequencies up to 0.5 Hz but had negligible responses beyond 1.0 Hz. They suggested that the discrepancy with the earlier study may have arisen because Rosenbaum and Race did not directly measure blood flow but rather perfusion pressure in the vascularly isolated hindlimb. To overcome the artificial nature of electrical stimulation of nerves, Stauss and colleagues (119) subsequently performed electrical stimulation of the paraventricular nucleus of the hypothalamus at multiple frequencies while recording splanchnic nerve activity and mesenteric blood flow. They found that the central nervous system component does not limit the frequency range of the reflex but that the vasculature was unable to respond to stimulation frequencies above 1.0 Hz with an oscillation. Thus the rate-limiting step lies at the ₁-receptor or other receptor transmission and subsequent smooth muscle contraction.

The nature of the frequency response characteristic of the vasculature has been most precisely determined for the renal vasculature. Initial experiments suggested the renal vasculature was able to follow frequencies in SNA <0.7 Hz with frequencies above this level producing steady vascular tone (74). This indicated that a simple first-order model with a time delay may be sufficient for describing the neural control of renal blood flow. However, using a new form of patterned stimuli that allowed for stimulation at all frequencies between 0.001 and 1 Hz, a much more complex vasculature response was revealed (36). A low-pass filter characteristic was still evident but also with regions of constant or increasing gain, one extending from 0.001 to 0.006 Hz and the other from 0.01 to 0.2 Hz (Fig. 3). The mathematical model developed for this response was consistent between animals, and the time delay between the stimulus and the renal blood flow response was very similar among animals, with an average of 672 ± 22 (SE) ms. Such little biological variation suggests that the contributors to this time delay comprised a series of delays, e.g., neurotransmission and signal transduction, the properties of which are normally fixed. In general, models that have been constructed to account for oscillations in blood pressure have used two time delays: one between efferent SNA and blood pressure and a second time delay on the afferent side of the baroreflex (16). Measuring such time delays is an important step in understanding the origin of oscillations in the cardiovascular system and in producing accurate models.

One question arising from the work of Stauss and colleagues (116, 119) is whether all target organs and species have similar frequency response characteristics. Certainly, differences in such responsiveness could greatly affect their ability to contribute to the slow oscillation in blood pressure. Recently, with the use of microneurographic techniques in humans to electrically stimulate efferent sympathetic skin fibers and simultaneously record skin blood flow, it was found that the skin blood flow could oscillate in response to stimulation at 0.1 Hz but not with stimulation at 0.5 Hz (113). The sluggishness of the human skin vasculature, however, does not seem to be species dependent because the skin vasculature of the rat has also been shown to have a slower frequency response than that of the mesenteric vasculature (120). The observed differences in frequency responsiveness between organs within the same species suggests that at least the skin vasculature is unlikely to play a role in supporting the 0.4-Hz oscillation in blood pressure in the rat.

Having ascertained the frequency response of various target organs to SNA, the logical question is what are the mechanisms giving rise to this phenomenon. The ability of the vasculature to respond to the faster frequencies of SNA with steady tone and to lower frequencies with an oscillation is indicative of an integrating-like phenomenon. Clearly, it results from a complex series of interactions between the characteristics of the neuromuscular coupling, second messenger pathways in the smooth muscle [i.e., the excitation-contraction coupling (39, 40, 112)] and the interaction with the intrinsic regulatory systems of the vasculature such as nitric oxide. To examine whether norepinehrine reuptake is the frequency-limiting step of the vascular response, Bertram and colleagues (13) applied rhythmic stimulation of the lumbar sympathetic chain and measured iliac blood flow before and after administration of desipramine, a specific norepinephrine uptake inhibitor. The transfer properties of the rat iliac vasculature showed characteristics of a second-order low-pass filter (natural frequency ~0.13 Hz), but this was not altered after norepinephrine uptake inhibitor, suggesting that norepinephrine uptake is not the rate-limiting step.

The cellular mechanisms underlying the dynamic responses of the vasculature have been examined in vascular smooth muscle cells isolated from the aorta of 4-wk-old rats (115). In an elegant series of experiments, rhythmic contraction and relaxation of the cells was elicited by periodically inducing depolarization by increasing the concentration of K⁺ or by ₁-adrenoceptor stimulation with phenylephrine. At low stimulation frequencies, the cells responded to the periodic applications of K⁺ and phenylephrine with likewise contraction and relaxation. At higher stimulation frequencies, the cells contracted and were unable to relax before the next stimulus occurred (tonic contraction). Tonic contraction occurred at application frequencies >0.1 Hz when the cells were stimulated with phenylephrine but at 0.5 Hz in the case of K⁺-induced depolarization. These experiments demonstrate a key feature: that vascular smooth muscle cells have an intrinsic ability to contract and relax at quite high frequencies, as indicated by the response to K⁺, but that the modulation of vascular tone by sympathetic activity is limited by signal transduction mechanisms in the smooth muscle.

	FREQUENCY RESPONSE OF HEART RATE

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

With regard to heart rate, the frequency response characteristics to vagal and sympathetic activity have been well described. A range of researchers have identified two frequency response characteristics of the sinoatrial node to parasympathetic activity: one associated with low levels of vagal tone and having a cutoff frequency of 0.065 Hz and another capable of responding to stimulation frequencies up to 0.8 Hz (Fig. 4) (8, 79).

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Schematic figure representing component sections responsible for variability in heart rate. Note that the central respiratory drive is likely to be a major component in producing sinus arrhythmia. Insets: transfer function gain between vagal or sympathetic stimulation at different stimulation rate obtained in anesthetized dogs (8). Note that there is evidence that the transfer function is altered with changes in posture (107, 122).

With regard to sympathetic activity, it must be stressed that the frequency responsiveness of the heart and vasculature to changes in sympathetic activity are quite different despite evidence that the pattern of sympathetic outflow to the heart and organs such as the kidney is similar (90). Certainly, the heart rate response to sympathetic nerve stimulation is slower than that of the vasculature. With regard to steady-state changes in SNA, the heart rate response is characterized by a time delay of 1-3 s followed by a slow increase with a time constant of 10-20 s. In the frequency domain the heart rate response is characterized by a low-pass filter system with a cutoff frequency around 0.015 Hz coupled to a 1.7-s time delay (8). The slow development of the heart rate response has been attributed to the slow norepinephrine dissipation rate and/or to the sluggishness of the adrenergic signal transduction system (63). Administration of the neuronal uptake blocker desipramine significantly slowed the heart rate response to sympathetic nerve stimulation, suggesting that the removal rate of norepinephrine at the neuroeffector junction is a rate-limiting step (87). This is in contrast to the vasculature, where the reuptake blocker did not affect the frequency response (13). Mokrane and Nadeau (80) recently identified two components in the heart rate response to SNA. With low intensities of sympathetic activation, the -adrenergic response was faster than at higher intensities of nerve stimulation.

It should be noted that there is good evidence that vagal and sympathetic influences, whereas antagonistic with regard to heart rate, do appear to interact in a dynamic fashion. In particular, sympathetic stimulation combined with vagal stimulation always increased the gain of the transfer function and by virtue of this interaction appears to extend its dynamic range of operation (55, 58). This interaction may account for the observation that -adrenergic blockade can enhance respiratory sinus arrhythmia (123). Thus, although the response of the heart to SNA is sluggish, it restrains the variability at all frequencies. This therefore challenges that notion that respiratory sinus arrhythmia is mediated solely by fluctuations in vagal nerve traffic.

	OTHER FORMS OF ANALYSIS

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

Whereas this review has focused on research that has measured cardiovascular variability, in particular, frequency ranges using predominantly spectral techniques, it should be noted that an emerging area of interest is the use of nonlinear system theory (chaos theory and fractal analysis) to measure variability (48, 49, 97, 130). It is proposed that heart rate in healthy subjects is fractal-like because it displays similar fluctuations (scale invariant) over a wide range of time scales (130). A reduction in this fractal-like variability (scaling exponent) reflects a loss of the short-term correlation properties of the R-R intervals and is proposed to be deleterious (34). This form of analysis has been found to be more sensitive than conventional spectral analysis in forming correlations between reduced heart rate variability and mortality (41, 65). In particular, in a recent study of 446 survivors of myocardial infarction, a fractal measure of heart rate variability was a significant predictor of mortality and remained an independent predictor of death after adjustment for other postinfarction risk factors, such as age, ejection fraction, and medication (43). Alterations in the fractal nature of heart rate have been observed after blockade of the parasympathetic nervous system (126) and during maneuvers designed to alter sympathetic tone, e.g., exercise or tilting (125). Reductions in the short-term scaling exponent and altered heart rate dynamics have also been observed with norepinephrine infusions (126). Whereas further research is required to show whether the observed changes in fractal indexes are indexes of autonomic control or of generalized heart rate variability of multiple origins, it is likely that such measurement may provide useful prognostic information.

	DO THESE OSCILLATIONS MATTER FUNCTIONALLY?

TOP ABSTRACT INTRODUCTION WHAT FREQUENCIES ARE PRESENT? RESPIRATORY OSCILLATION IN... SLOW OSCILLATIONS FREQUENCY RESPONSIVENESS OF... FREQUENCY RESPONSE OF HEART... OTHER FORMS OF ANALYSIS DO THESE OSCILLATIONS MATTER... REFERENCES

Whereas it has been the hope of many research groups that measurement of cardiovascular variability may be used as an index of autonomic function, one can also ask a somewhat teleological question: Do they have a functional purpose? Indeed, given the complexity of the central nervous system in producing differential neural control over various end-organ functions, why is it that we appear to accept that oscillations are simply the result of "the clanking of cogs" in the circuit? An alternative viewpoint is that such variability does serve a purpose as yet unknown. It is interesting to speculate that the presence of neurally mediated oscillations in blood pressure are antihypertensive through facilitation of the excretion of fluid and electrolytes and reduced renin release.

In conclusion, variability in cardiovascular signals reflects the dynamic interplay between perturbations to cardiovascular function and the dynamic response of cardiovascular regulatory systems. It is apparent that altered variability in the cardiovascular system is associated with a range of cardiovascular diseases and increased mortality and thus may prove to be a useful prognostic index. The central question, however, is whether the autonomic nervous system can be linked to such changes? It is suggested that, whereas SNA, the vagus, and baroreflexes clearly are important in producing variability at particular frequency bands, other factors including the vasculature's response, the central nervous system, and other reflex pathways also appear to play a role. A unifying theory is proposed that links the discrepancies between research indicating a central origin for the slow oscillation in blood pressure and heart rate at 0.1 Hz and research suggesting a resonant loop in the baroreflex control of blood pressure. Given the potential impact of other nonbaroreflex or nonautonomic pathways in affecting cardiovascular variability, it difficult to definitively relate changes in the strength of an oscillation as due to a change in autonomic control. Thus committee reports on the interpretation of variability in the cardiovascular system, which suggest that a measure of autonomic balance is obtainable, appear incorrect (12, 120a).