Ebook: Blood Pressure and Heart Rate Variability

In the past twenty years technological progress has allowed us to develop devices which make it possible to record blood pressure and heart rate outside the laboratory environment and thus to collect observations in the subject’s real living conditions. Furthermore, technological progress has provided computer tools to analyse data on a beat-to-beat basis, offering a large body of information on the complex patterns of blood pressure and heart rate variability in health and in disease which has clarified a number of mechanisms responsible for cardiovascular regulation during the day and night.

This book aims at providing an updated state of the art on the analysis of blood pressure and heart rate variability from the different perspectives of physiologists, clinicians and engineers by gathering the contributions of experts taken from both the biomedical and the engineering environment.

In this framework, a special symposium was held in Verona (Italy) in 1991* where the experts contributing to this book were invited to make a clearcut point on some of the matters pertinent to the dynamic analysis of cardiovascular signals and to debate the many open questions that still exist in this field.

In particular, attention was focused on: 1) the methodology of ambulatory blood pressure and heart rate analyses, 2) the mathematical models that can be derived from the data and the implications they have for the understanding of normal and deranged cardiovascular control and 3) the present and future use of computer analysis of dynamically collected observations for the diagnosis of cardiovascular diseases.

We thank the authors for their effort. We also thank Centro Auxologico Italiano, Fnd. Pro Juventute Don Carlo Gnocchi, Ospedale Maggiore ed Universita’ di Milano, Milano, Italy for organising the symposium and Glaxo Italy for its generous contribution. We hope that the attempt made in this volume to clarify some of the controversial matters that characterise dynamical analysis of blood pressure and heart rate may stimulate new ideas for the future among people working in this field.

Marco Di Rienzo

Giuseppe Mancia

Gianfranco Parati

Antonio Pedotti

Alberto Zanchetti

* “Computer Analysis of the Blood Pressure Signal. Research and Clinical Applications” a satellite symposium to the V European Meeting on Hypertension. Verona, 1991.

Within the broad range of physiological oscillations is a sub-group which occurs in physiological systems involved in homeostasis. Such systems have the potential for oscillatory behaviour because their distributed nature allows the existence of significant time delays in reflex arcs. Physiological systems of this type can be considered as a number of different elements linked together, typically by neural pathways, to give the negative feedback necessary for homeostasis. Often such systems behave as spontaneous oscillators (Hyndman, Kitney and Sayers, 1971), the frequency of oscillation being determined by the characteristics of the system. A change in these characteristics can be considered to be equivalent to changing the setting of an oscillator so that it takes up a new and normally stable oscillatory mode. It is possible to envisage the human body as containing a number of systems which frequently oscillate. There is also evidence that biological systems which oscillate at frequencies higher than that consistent with the 24-hour day interact with each other, e.g. the interaction between heart rate control and respiration.

Historically, there have been a number of different theories relating the nature of biological rhythms. The current consensus appears to be that biological systems which exhibit oscillations can be divided into autonomous or endogenous systems (i.e. systems which are self-sustaining), which usually oscillate at periods of less than 1 hour, and non-autonomous or exogenous systems (systems which require external stimulation).

The analysis of Blood Pressure (BP) and Heart Rate (HR) variability in the frequency domain is currently used to investigate the mechanisms responsible for cardiovascular control. The potentialities of this methodology are greatly enhanced by techniques able to track the changes of BP and HR spectral characteristics for prolonged time periods. Up to now, specific techniques for dynamic spectral analysis are complex and provide results not always easy to interpret by visual inspection. Simplified tracking procedures may be obtained by the application of standard spectral estimators in a sequential fashion. Aim of this paper is to address the methodological and practical aspects of sequential spectral analysis in the study of long term BP and HR recordings. A short review of the results we obtained by this technique in the analysis of data from humans and animals is provided as well.

The basic methodological implications of spectral analysis of beat-to-beat discrete values of systolic or diastolic arterial blood pressure and RR intervals on ECG tracings are briefly summarized. Different applications are involved in the short, medium and long-term spectral estimation, in both the parameter extraction phase and in the modelling of complex pathophysiological relationships among the cardiovascular variability signals. Autospectral, crossspectral and coherence parameters find many applicative examples for a quantitative determination of stationary conditions in the functioning of neural, mechanical, vascular, humoral mechanisms of control and others. A few examples are also reported for the detection of transients in the sympatho-vagal balance which have a considerable importance even from the clinical standpoint. That is done by means of the employment of time-varient techniques of spectral estimation.

A particular emphasis is dedicated to the comprehension of various functions involved in the physiology of the above mentioned mechanisms, through proper linear modelling which makes use of a few of the obtained spectral parameters: mechanical and neural effects may be properly quantified via a closed-loop model, whose transfer functions are estimated through least squares algorithms. An interesting and novel approach is to apply non-linear models which may take into account the well known chaotic behaviour of the beat-to-beat cardiovascular discrete series. Simple models and ways of representation are introduced and commented at this regard.

We modelled the nonpulsatile circulation with baroreflex and cardiopulmonary reflexes. Special care was taken to model as accurate as possible the various delays and time constants to obtain reliable dynamic responses in the model. In experimenting with the model circulation we found that blood pressure was stable under a wide range of adverse conditions and could only be made to vary as observed in human beings by modulating the sensitivity of the baroreflex. Pseudo-static modulation caused blood pressure in the model to vary over a three-to-one range, more than is necessary to mimic circadian blood pressure variability. Dynamic baroreflex modulation with 1/ f random noise allowed the computation of blood pressure variability spectra in all practical aspects identical to those observable in human beings.

Cardiovascular regulation involves the interplay of many physiologic variables which interact through multiple feedback and control loops. This work examines the coupling between heart rate (HR), instantaneous lung volume (ILV) and arterial blood pressure (BP) through the use of closed loop system identification techniques in order to assess the underlying physiology without disrupting normal regulatory function.

We have developed an identifiable, closed loop model of cardiovascular regulation that contains four transfer relations representing both autonomic (ILV to HR, BP to HR) and mechanical (ILV to BP, HR to BP) coupling mechanisms. The model also contains two additive perturbations: one to HR and one to BP. When the experimental subject is physiologically stable and when the data fluctuate within a small neighborhood of their mean values, it is hypothesized that the interaction between HR, ILV and BP in the model should obey linear, time-invariant dynamics. These dynamics and the spectra of the noise sources can be uniquely estimated even in the presence of feedback if one assumes that each transfer relation is causal and if the data are sufficiently broad-band. One achieves the first condition by parameterizing the transfer functions with the appropriate constant- coefficient difference equations. The data set is rendered broad-band through the application of a wide-spectrum perturbation such as random-interval breathing.

This approach is applied here to assess the importance of respiration and HR in the genesis of BP fluctuations below 0.5 Hz. The results suggest that BP variability at low frequencies does not depend strongly on HR fluctuations, consistent with independent experimental evidence. The present application shows that the use of closed loop system identification offers a new, minimallyinvasive approach to estimating the dynamics of cardiovascular regulation. The techniques developed here are very general and can be applied to other biological and physiological systems.

This work presents an algorithm based on blood pressure signal analysis for the dynamic estimation of the respiratory rate and for tracking the effects of respiration on blood pressure and heart rate. This method was used to assess the respiratory influences on long term blood pressure tracings recorded in free-moving animals, where a direct recording of the respiratory activity was not available.

It is common knowledge that the blood pressure data detected in the same patient at constant intervals, after the artefacts rejection, are dependent and strictly correlated. Such a dependence is as much stronger as closer BP values are. For this reason we used the Fourier analysis, curtailed to the third term, based on time series. We obtained, in previous studies, a good fit between experimental data and the statistical model. We analysed a large normal population as control selected from a multicenter Italian epidemiologic study. In addition we take into account a wide group of hypertensives subdivided in accordance with the target organ damage. The Fourier model evidenced that it is likely to have a connection between the different factors involved in the regulatory mechanisms of blood pressure control and the first term syncronized on a 24 hour basis, i.e. able to fluctuate once a day, the second one twice, the last one three times, every eight hours, daily. The analysis allows a detailed evaluation of the blood pressure nocturnal fall and the incidental presence of a relative minimum in the early afternoon. We can also have the opportunity to evaluate the blood pressure peaks during late morning and late afternoon. The further step was to determine for each model, the tangents to the inflection points.

As already reported, the FINAPRES device is able to monitor blood pressure (BP) variations in a beat-to-beat fashion both in resting conditions and during different laboratory manoevres, providing BP values close to those simultaneously recorded by an intra-arterial line. Whether this applies not only to the assessment of BP mean values but also to the evaluation of specific components of BP variability, as determined by spectral analysis, is not known however. We have addressed this issue by assessing the spectral characteristics of BP and heart rate (pulse interval, PI) of continuous finger and intra-arterial BP recordings simultaneously obtained in 14 untreated essential hypertensive patients. Spectral powers were computed by Fast Fourier Transform (FFT) over three frequency bands defined as Low Frequency (LF, 0.025-0.07 Hz), Mid Frequency (MF, 0.07-0.14 Hz) and High Frequency (HF, 0.14-0.35 Hz) band. All powers of mean arterial pressure (MAP) and diastolic (D) BP estimated from analysis of finger BP tracings were superimposable to those obtained by analysing intra-arterial recordings. This was the case also for HF powers of systolic (S) BP, while LF and MF powers of intra-arterial SBP were over-estimated by the analysis of finger BP tracings (+ 13.7± 4.4 mmHg²; p<0.01 and +2.3±0.9 mmHg²; p<0.05). Thus beat-to-beat finger BP recording seems to represent an acceptable substitute of invasive recording also in the analysis of BP and PI spectral components. Some caution is needed, however, in interpreting LF and MF powers of SBP which are overestimated by the analysis of finger BP tracings.

In order to verify whether finger photoplethysmographic beat-to-beat recording of blood pressure (BP) can be used to study short-term BP variability, we analyzed by spectral analysis 512-beats SBP stationary sequencies recorded on 36 patients during supine bed rest, active standing, and controlled respiration (CR) at 20 breaths/min. A space state model of the mean was adopted to account for very low frequencies. Spectral analysis of low-(LF, ≈ 0.10 Hz) and high-frequency (HF, ≈ 0.25 Hz) peaks was performed by an autoregressive (AR) method. LF peak spectral power increased during orthostatism (51.35± 10.10 vs 28.70± 25.10, percent power in normalized units), and HF peak increased during CR (77.10± 28.99 vs 58.80± 27.70). A mixed model analysis of variance, performed on sequencies recorded at rest and during active standing, showed that within- and between- subject variabilities were significantly lower than the variability due to the experimental protocol. These results show that noninvasive BP recording from the finger can be reliably used to study short-term BP variability by spectral analysis. This approach will presumably prove most useful in the comparison of groups of patients and in the study of BP response to selected stimuli.

Time course of beat-by-beat error measurement associated with continuous non invasive monitoring of arterial blood pressure by the Finapres device, has been analyzed during short-term (up to 7 min) recordings either at rest (stationary conditions) or after phenylephrine injection (dynamic conditions) in 22 post-myocardial infarction patients. The equipment set point Self-Adjustment was deactivated during recordings. The systematic and random components of the error have been estimated over one min consecutive intervals for systolic and diastolic pressure. Long-term (> 20 min) reproducibility has been assessed as well. Quite a stable performance has been found during short-term stationary conditions associated to a high interpatients variability of systematic error and to a not negligible random error. A slight drift of systematic error, either for systolic or diastolic pressure, has been detected in long-term analysis. During dynamic conditions a significant difference in accuracy, between the first and second min after phenylephrine injection associated with a global loss of stability, has been observed during short term recordings consistent over the long-term period.

Thus, while effective in stationary recordings despite Self-Adjustment disconnection, Finapres performance should be considered cautiously during dynamic evaluation.

Pressor tests are considered a useful method to better understand the pathophysiology of hypertension. However, the blood pressure response to the pressor tests is frequently evaluated with indirect blood pressure measurements, which have been proven to be insufficiently accurate. The intraarterial Oxford Instruments system represents a unique method to reliably assess blood pressure changes during these manoeuvres. A computer programme is presented for the automatic analysis of blood pressure and heart rate response to pressor tests performed during intraarterial monitoring.

Cold pressor, head-up tilting, hand grip and bicycle ergometry tests were studied. The analysis of the blood pressure signals was performed on a Data General MV 10000 mainframe and systolic, diastolic, mean blood pressure and heart rate were evaluated beat by beat. The following calculations were made: (1) Mean baseline blood pressure and heart rate values over 30 seconds before each test. (2) The same parameters every 30 seconds for the cold pressor test, every 60 seconds for the hand grip test and for the head-up tilting, every 120 seconds for the bicycle ergometry during the whole test. (3) The lowest and highest 5 consecutive beats average of systolic, diastolic, mean blood pressure and heart rate. (4) The difference between the preceding averages and mean baseline blood pressure and heart rate. The time during the test when the maximal BP and HR variation occurs was recorded. A beat by beat plotting of every test was also obtained.

These data allow a very accurate and synthetic evaluation of blood pressure and heart rate response to the various pressor tests.

Computer analysis of blood pressure and heart rate variability has been recently proposed as a new method to obtain quantitative information on neural control of circulation and, specifically, on the sensitivity of the baroreceptor-heart rate reflex. This approach is particularly appealing because it allows a dynamic assessment of baroreflex gain in daily life conditions, thus overcoming several drawbacks of traditional laboratory procedures. In this chapter techniques developed to this aim both in the time and in the frequency domain will be briefly described, discussing their advantages and limitations. Examples of the results obtained by using them to analyse 24 hour ambulatory blood pressure and heart rate recordings in normal subjects and in hypertensive patients of different ages will also be shown. Finally, perspectives for further technical improvement of these methods will be mentioned.

The analysis of oscillations in heart rate (HR) and blood pressure (BP) has been used to assess indirectly the control of the circulation by the autonomic nervous system. Little is known of the effect of physical fitness on these measures. Fitness is associated with reductions in rest and exercise heart rate, probably to be due to increased vagal drive and reduced sympathetic tone. Training has been assessed in two conditions associated with abnormal autonomic control.

In hypertensive subjects BP is lowered by strenuous physical training. This is associated with an increase in HR variability and an increased baroreflex sensitivity assessed either by the bolus phenylephrine method or by two-way autoregressive bivariate cross-spectral analysis of HR and BP fluctuations.

In chronic heart failure a controlled trial of physical training has shown a significant reduction in low frequency HR oscillations and increases in high frequency oscillations, associated with a reduction in resting noradrenaline spillover assessed by radiolabelled tracer techniques. The changes in HR variability correlated well with both compliance with training and with improvement in exercise capacity.

Training can alter the patterns of HR and BP variations in a variety of conditions and the changes suggest a shift away from sympathetic predominance.

Heart rate variability is thought to be mediated by both vagal and sympathetic activity, through both phasic and tonic stimulation of the sino-atrial node. Decomposition of heart rate variability by power spectral analysis techniques has shown fluctuations related to the respiratory pattern, expression of respiratory sinus arrhythmia (RSA) and oscillations related to blood pressure (between 0.03 and 0.15 Hz). Under physiologic conditions, low- and high frequency peaks can be considered as relative markers of sympathetic and parasympathetic activity, respectively.

The purpose of our study was to investigate whether high frequency oscillations are dependent on vagal activity also in critical conditions (i.e. low heart rate variability). In 6/6 recently heart-transplantated subjects we found a residual RSA which increased during moderate bicycle exercise. To assess its mechanical determinants, in 15 urethane-anaesthetized, vagotomized and mechanically-ventilated rabbits we measured the influence on RSA of changes in right atrial pressure, induced by changes in tidal volumes (Vt: 20-60m1) and respiratory frequency (RF: 10-30/min). RSA was present in all recordings, it was dependent on both RF and Vt (p<0.01 and p<0.001, respectively) and correlated with right atrial pressure variability (r=0.62, p<0.001).

Similarly, patients with severe autonomic dysfunction and low heart rate variability show a well evident RSA which can not be abolished by injection of atropine. In absence of parasympathetic drive to the heart RSA does persist and is proportional to changes in ventilation through phasic changes in right atrial pressure, supporting the hypothesis that atrial stretch can be a secondary source of RSA.

In order to investigate if slow oscillations in blood pressure are related to oscillations induced by the parasympathetic or sympathetic nervous system, we made a fourier analysis of blood pressure, sympathetic efferent nerve activity (splanchnic nerve activity) and vagal afferent and efferent activity (thoracic vagal nerve activity) in 6 conscious WKY rats. The control measurement was made after a 0.2 ml saline bolus i.v. injection and during a constant saline perfusion of 2 ml/h. In a second protocol we gave a 20 μg bolus injection (0.2 ml) of Metyl-Scopolamine. Thereafter, the measurement began during a constant perfusion of 200 μg/h Metyl-Scopolamine (2 ml/h). One day after these protocols, a bolus infusion of 100 μg Prazosin (0.2 ml) was given. A further measurement was then made during a constant i.v. Prazosin infusion of 3 mg/h (6 ml/h).

To test if the slow frequency oscillations prevail over the entire measuring period, three dimensional plots of frequency vs. power vs. time were performed (10 min. at 40 Hz). In the blood pressure plots two peaks were regularly found: one coinciding with the heart rate (between 6-7 Hz) and another corresponding to oscillations in sympathetic and vagal nerve activity at 0.5-1.5 Hz. The latter peak was blunted by Metyl-Scopolamine and the frequency of these oscillations were shifted to the right (1.5-2.5 Hz). During application of Prazosin the oscillations remained in vagal and sympathetic nerve activity, however the corresponding blood pressure peak was accentuated in its power.

In conclusion, blood pressure, sympathetic nerve activity and vagal nerve activity have corresponding slow frequency oscillations. The oscillations in blood pressure can be blunted by blockade of the parasympathetic system and augmented by sympathetic blockade.

Sympathetic activation is known to promote wide blood pressure variations but the relationship between sympathetic influences and blood pressure variability under spontaneous behavioral conditions is less clear. To address this point we studied in conscious unrestrained rats the effects of chemical sympathectomy on the spontaneous variability of systolic blood pressure. Both overall blood pressure variability and its spectral components were examined in sympathectomized as well as in intact control rats. The spectral profiles were extracted by the FFT tecnique and the powers were computed in the high (HF, 3.0-0.8 Hz), mid (MF, 0.6-0.1 Hz) and low (LF, 0.1-0.025 Hz) frequency band.

The results showed that 1) overall blood pressure variability is greater in sympathectomized compared to control rats, 2) the spectral components of blood pressure variability present diversified sympathectomy-related alterations. In fact, only the LF component shows the same trend of overall variability (i.e. it is greater in sympathectomized rats) while the MF component is decreased and the HF component does not change. Thus under spontaneous behavioral conditions, sympathetic activity limits overall blood pressure variability, fails to affect HF, contributes to generate MF and opposes LF component of variability. These findings document the complex contribution of sympathetic influences on the blood pressure spectral profile.

The mechanisms underlying systolic (SBP) and diastolic (DBP) blood pressure and heart rate (HR) beat-to-beat variability were investigated using spectral analysis in conscious genetically normotensive (LN) adult rats from the Lyon strain. Basal SBP, DBP and HR spectra exhibited peaks in low- (LF : 0.38 to 0.45 Hz) and high- (HF : 1.04 to 1.13 Hz) frequencies. The LF peak of SBP, and even more of DBP, could be attributed to the sympathetic nervous system influence as it disappeared after destruction of the sympathetic nerves or a combined alpha- and beta-adrenoceptor blockade whereas it was higher after blockade of the parasympathetic system. The HF peak of HR, linked to the respiratory rate, was abolished by the parasympathetic system blockade whereas those of SBP and DBP were enhanced. We conclude that the LF peak of DBP and the HF peak of HR are likely to represent useful estimates respectively of the sympathetic vascular control and of the parasympathetic cardiac control.

Fluctuations in arterial blood pressure (ABP), the pattern of their power spectrum and the frequency specificity of the various pressor mechanisms, may provide important information regarding the maintenance of normal ABP levels. The study of these parameters may contribute to the characterization of a cardiovascular system prone to hypertension, and to the understanding of its pathogenesis.

We investigated ABP fluctuations under steady state conditions and following acute perturbations, with and without blockade of various ABP control branches, in spontaneously hypertensive rats (SHR) and age-matched (1 to 6 mo) normotensive WKY. A continuous ABP signal was recorded from the caudal artery in conscious rats. Though heart-rate (HR) power spectra were similar in both strains, the low frequency ABP fluctuations (below 0.18Hz), reflecting vasomotor control, were markedly reduced in SHR at all ages, including the prehypertensive stage of 1 mo. This steady state abnormality was shown to be of neural origin, most probably α-sympathetic. Furthermore, SHR and WKY responded differently to a blood pressure perturbation, even at the age of 1 mo. In particular, a sudden drop in blood volume caused an exaggerated increase in low frequency ABP fluctuations in SHR: 8.7x before hemorrhage (bh) in SHR vs 1.5x in WKY, in the 0.004-0.04Hz range, and, 5.7x bh in SHR vs 1.4x in WKY, in the 0.04-0.07Hz range. Pharmacological blockade of either of the main pressor mechanisms, α or renin-angiotensin (R-A), clearly affected this abnormal response, displaying an interesting frequency specificity. Under α_l -blockade with prazosin, the difference in response to hemorrhage was eliminated: the response was reduced in SHR in the 0.04-0.07Hz range (1.5x bh in SHR and 2.1x bh in WKY) and enhanced in WKY in the 0.004-0.04Hz range (4.6x bh in WKY vs 4.3x bh in SHR). R-A-blockade with CEI reduced the response to hemorrhage in SHR to the level in WKY, in both frequency ranges. R-A-control is thus prominent mainly below 0.04Hz, enhancing its activity when α-control is impaired (under α-blockade or in SHR), its contribution between 0.04-0.07Hz being mainly via its effect on α-release. The α-control manifests itself between 0.04-0.1Hz and seems to contribute to the abnormality in blood pressure control in the spontaneously hypertensive system.

Miniaturized Doppler flow probes were chronically implanted around the abdominal aorta, left renal and superior mesenteric arteries of 20 normotensive WKY rats to allow regional blood flow velocities to be monitored for prolonged periods in the unanesthetized unrestrained condition. Arterial blood pressure (BP) was simultaneously recorded via a chronic femoral arterial catheter. The four signals were stored on a magnetic tape for subsequent beat-to-beat computer analysis. Computer outputs included: 1) mean values of arterial blood pressure and of hindquarter, renal and splanchnic blood flow velocity (BFV); 2) mean values of the index of vascular resistance in each regional bed calculated over all consecutive periods of 0.8 sec as the ratio of mean BP to BFV, and 3) variability, as the percent variation coefficient, calculated for each of the above parameters by averaging the signal every 0.8 sec, this coefficient thus largely reflecting short-term variability. The results indicate that regional hemodynamics (either flow or resistance) varied to a largely similar extent in all three beds and that each of the regional hemodynamic parameters was much more variable than systemic blood pressure, suggesting the frequent occurrence of simultaneous and opposite vasomotor changes in different regional beds. This may minimize the variability of total peripheral resistance and might represent a previously unrecognized BP-stabilizing mechanism. Preliminary evidence suggests that the hemodynamic changes simultaneously occurring in the hindquarter and splanchnic territories may indeed be linked by a divergent trend.

Power spectral analysis (PSA) of heart rate (HR) and blood pressure (BP) may provide useful informations about neural control of the cardiovascular system. Aim of the study was to evaluate PSA of HR and BP in a group of normal subjects (N) and hypertensive patients (H), with and without left ventricular hypertrophy (LVH), in the following situations: 1) at rest, in the morning and also at different hours during the day and night, 2) during application of a negative pressure (LBNP) to the lower body, in order to induce sympathetic activation. PSA of HR was performed on sequences of 512 beats, in a total number of 12 N, 19 H without LVH, 20 H with LVH. In 10 of the subjects PSA of BP variability was also performed. Absolute and normalized power spectral density (PSD) of the peaks at 0.10 (low frequency peak = LF) and at 0.25 (high frequency peak = HF) Hz, as well as their ratio (index of sympathovagal interaction: SVI) were calculated. H with LVH had a SVI index significantly higher in respect to that found in N and in H without LVH (p<0.05); in addition, SVI index was significantly less during night in N and in H without LVH, compared to H with LVH (p<0.05). In the group of H with LVH, changes in SVI during LBNP were markedly reduced. In the group of patients studied so far, results of PSA of BP were similar to those obtained performing PSA of HR, thus suggesting an increased sympathetic drive to the heart, and, probably, also to the vasculature, in H with LVH.

Because of the cardiac denervation, transplanted heart represents a reference model for the study of the autonomic control of the heart. Heart rate (HR) variability analysis has been proposed as a method for assessing the vagus — sympathetic influence on the heart. In this study, both circadian and beat-tobeat HR variability was investigated in 24-hour Holter monitoring from 8 orthotopic cardiac transplant recipients, and 8 normal subjects. Circadian rhythm of HR was assessed by computerised periodic analysis, while beat-to-beat HR variability was assessed by autoregressive power spectrum analysis. Average 24-hour HR of transplanted heart was higher but not statistically different from controls. A significant circadian rhythm of HR was found in all controls and in all but one transplanted patients (pts). The semiamplitude of the periodic function was significantly lower in transplanted pts than in controls. A beat-to-beat variability was also present in all transplanted pts; as compared to normal subjects 1) the total R-R spectral power was markedly reduce (p < .01) throughout the 24-hour period, 2) the HR variability at the ‘very low’ frequencies (VLF: ≤ .05 Hz) was markedly reduced (p < .01) while it was preserved in the ‘low’ (LF: up to .14 Hz), and in the ‘high’ respiratory frequency range (HF: .15 to .35 Hz) in all pts, particularly during the night (the power spectrum of R-wave amplitude, an indirect marker of respiration, allowed the identification of peaks related to respiration in the spectra of transplanted heart). In conclusion, in cardiac transplantation, a persistence of HR variability can be shown in spite of heart denervation. While the persistence of circadian rhythm of HR may be attributed to a parallel oscillation in blood cathecolamine levels and its attenuation to the absence of nervous reinforcement, beat-to-beat fluctuations (namely at HF, and possibly at LF) might be interpreted as the result of intrinsic cardio-cardiac reflexes, elicited by changes in atrio- ventricular volume and/or pressure load.

In normally breathing subjects, periodicities in blood pressure (BP) and heart rate (HR) often occur in three main frequency bands: high-frequency (0.20-0.45 Hz), mid-frequency (0.08-0.12 Hz), and low-frequency regions (0.02-0.06 Hz). While the high-frequency oscillations occur in synchrony with respiratory movements and mid-frequency oscillations are probably due to sympathetic baroreflex-feedback, the sources of low-frequency oscillations are less well known. A possible explanation for low-frequency oscillations in the circulation ensues from periodicities in the breathing pattern, of which sleep apnoea syndrome (SAS) is an extreme example. We applied spectral analysis to the pronounced low-frequency BP- and HR-variations that occur during sleep in this syndrome. The obtained power and cross spectra could be explained by a simple model that assumes peripheral chemoreceptor reflexes to be the main link between low-frequency variations in the respiratory and circulatory systems in SAS.