Relationship Between Repeated Measures of Hemodynamics, Muscle Sympathetic Nerve Activity, and Their Spectral Oscillations
Background We determined the intraclass correlation coefficients (ICC) of repeated measures of the mean levels and variability of RR and muscle sympathetic nerve activity (MSNA) in 7 normal subjects. We examined whether spontaneous fluctuations in RR and MSNA over repeated measurements were mirrored by changes in spectral components of RR and MSNA.
Methods and Results Twenty-minute recordings of respiration, RR, blood pressure (BP), and MSNA were performed at day 1, 1 week, 1 month, and 3 months and divided into two 10-minute periods for the analysis of short-term reliability. Comparison between these recordings also determined the long-term reliability. Linear regressions examined the relationship between changes in these measurements and changes in spectral components of RR and MSNA. All analyses were carried out blinded to subject and session. Short-term ICC of RR, BP, MSNA and of the variabilities of RR and MSNA (in % of total power) ranged between .98 and .70 and indicated a good short-term reliability. The long-term reliability of RR variability was comparable to MSNA variability (range of ICC, .34 to .52). Spontaneous decreases in RR during the repeated recordings were accompanied by increases in sympathetic drive, as evidenced by increases in the ratio of low-frequency to high-frequency variability (LF/HF ratio) of RR interval (r=−.43, P<.01) and by increases in MSNA (r=−.36, P=.01). The changes in the LF/HF ratio of RR were mirrored by parallel changes in the LF/HF ratio of MSNA (r=+.30, P<.05). Spontaneous decreases in BP were accompanied by increases in the LF/HF ratio of MSNA (r=−.52; P=.0001).
Conclusions Heart rate, MSNA, and their variability are stable in the short-term, but less so over the long term. Spontaneous changes in repeated measurements of RR interval and blood pressure over the long term are accompanied by parallel changes in the normalized spectral components of RR and MSNA variability. Thus even over an extended period, there is a synchrony between changes in absolute cardiovascular measures and changes in their spectral components.
Spectral analysis of heart rate is widely used as an indirect marker of cardiac autonomic regulation in a variety of physiological and pathological conditions.1 2 3 4 5 These spectral measures have been used to compare autonomic cardiovascular drives at different times within individuals as well as to compare autonomic cardiovascular control between individuals.1 2 3 4 5 There is only limited information on the long-term reproducibility of these measures.1 3 6 There are no data comparing the reproducibility and relationship of spectral analysis of heart rate variability with that of direct intraneural measures of sympathetic activity and its variability. Most important, there are no studies examining the relationship between changes in spectral components of cardiovascular measurements and changes in the absolute values of these measurements. In other words, do changes in spectral components of cardiovascular variability represent a physiological response to spontaneous changes in blood pressure that occur over repeated measures?
Recent studies in normal subjects have shown that muscle sympathetic nerve activity (MSNA) oscillates at low-frequency (LF) and high-frequency (HF) components almost identical to those present in heart rate variability.7 8 9 A similar relationship (or coherence) was observed between spectral components of MSNA and the skin microcirculation.10 In normal humans, short-term changes in autonomic drive in response to blood pressure changes induced by intravenous vasoactive substances are accompanied by consistent and parallel changes in spectral measures of RR interval and of MSNA.8 Namely, during sympathetic activation, there is a relative dominance of the LF component and an increased LF/HF ratio of RR and MSNA variability. During vagal activation (and sympathetic inhibition), the HF spectral components of RR and MSNA are increased, and the LF/HF ratio decreases. Thus the LF/HF ratio of RR and the LF/HF ratio of MSNA appear to reflect the balance between excitatory (sympathetic) and inhibitory (vagal) cardiovascular drives.8 It is not known whether similar systematic changes in spectral measures of RR and MSNA also accompany spontaneous long-term fluctuations in cardiovascular measures.
We tested the hypothesis that spontaneous changes in cardiovascular variables, observed over repeated measures, would be tracked by parallel changes in the LF and HF spectral components of these variables. We therefore measured RR, blood pressure, and MSNA over 20 minutes on four occasions (at days 1, 7, 30, and 90) in 7 normal subjects. This allowed us to answer the following questions: (1) How reproducible are spectral measures of RR and MSNA, both over the short-term (first 10 minutes compared with second 10 minutes of the same session) and long-term (four measures over 3 months)? (2) How does the reproducibility of spectral measures of RR compare with the reproducibility of MSNA and its spectral components? What is the effect of normalization for total power? (3) Are spontaneous changes in long-term repeated measures of RR and blood pressure accompanied by consistent changes in the LF and HF spectral components of RR and MSNA variability? (4) What are the potential mechanisms regulating the long-term changes in resting measurements of spectral components of RR and MSNA?
We studied 7 normal subjects (mean age, 36±10 (SD) years; 5 men, 2 women). All subjects were healthy and none was receiving any medication. The study was approved by the Institutional Human Subjects Review Committee, and written informed consent was obtained.
Blood pressures were measured every minute with an automatic sphygmomanometer (Lifestat 200, Physio-Control). ECG and respiration (Biotach and Pneumotrace, Gould Electronics) were recorded continuously on all subjects. Intraneural measurements of muscle sympathetic nerve activity were obtained from microneurographic recordings of multiunit postganglionic sympathetic fibers (MSNA) measured in the peroneal nerve posterior to the fibular head.11 We used tungsten microelectrodes (shaft diameter 200 μm, tapering to an uninsulated tip of 1 to 5 μm). A subcutaneous reference electrode was inserted 2 to 3 cm away from the recording electrode inserted into the nerve fascicle. The neural signals were amplified, filtered, rectified, and integrated to obtain a mean voltage display of sympathetic nerve activity. All nerve recordings met standard criteria.11
The subjects were studied in the supine position on a comfortable bed. Measurements of heart rate (ECG), blood pressure, respiration, and MSNA were obtained over 20 minutes in carefully standardized conditions. Studies were conducted in the same room at the same time of day and at least 3 hours after the last meal.12 All studies were performed under quiet resting conditions by the same investigator (P.v.d.B.). Measurements were obtained during relaxed free breathing, and subjects were instructed to avoid talking during the recordings. The temperature of the room was maintained constant throughout the studies. In addition, all subjects were asked to void before the recording.12 These recordings were repeated after 1 week, 1 month, and 3 months in all subjects. Technically excellent recordings of all variables were obtained in all subjects during all four study sessions.
The 20-minute recordings of heart rate, MSNA, blood pressure, and respiration were divided into two consecutive 10-minute periods for the analysis of the short-term reproducibility. Comparison of the recordings performed at day 1, 1 week, 1 month, and 3 months determined the long-term reproducibility of the data.
A random code was attributed to the recordings so that all data analyses were performed completely blinded to the identity of the subject and the session during which the recording had been performed.
Sympathetic bursts were identified by a single experienced observer (P.v.d.B.). Muscle sympathetic nerve activity was expressed as frequency (bursts/min) and, for the spectral analysis, as mean integrated activity (ie, area under the sympathetic bursts).
Analog-to-digital conversion was performed in real time at 600 samples/s per channel. The data were then analyzed off-line with a personal computer (433DX/T, IBM). The principles of the software for data acquisition and spectral analysis have been described elsewhere.3 13 14 15 In summary, a derivative/threshold algorithm provided the continuous series of RR intervals (tachogram) derived from the electrocardiogram. Isolated artifacts and rhythm disturbances were detected and removed. All interpolated values were visually checked. Stationary segments devoid of arrhythmias (150 to 300 RR intervals) were analyzed with autoregressive algorithms. These algorithms provide the number, center frequency, and power of the oscillatory components. A potential advantage of the autoregressive method is that it allows an accurate spectral estimation on short segments of data, which are more likely to be stationary. Furthermore, statistical criteria3 13 14 15 such as Akaike’s test and Anderson’s test allowed a determination of the optimal model order fitting the data and verified that all information contained in the time series had been extracted in the computation.
Previous studies1 2 3 have shown that two major oscillatory components are usually detectable in RR interval. One of these oscillatory components is synchronous with respiration. This component is called HF (high frequency) and its frequency varies with the respiration frequency (usually about 0.25 Hz). The other component is described as the LF (low-frequency),1 2 3 the center frequency of which is about 0.10 Hz but can vary considerably (from 0.04 to 0.13 Hz).1 2 3
In this study, the LF and HF components were expressed in absolute (ms2) and normalized units (nu). The normalized units were obtained by calculating the percentage LF and HF variability with respect to the total power after subtracting the power of the very-low-frequency component (frequencies below .03 Hz).3 15 16 17
The signals of sympathetic nerve activity and respiratory activity were sampled once every cardiac cycle, thus obtaining a neurogram and a respirogram synchronized with the tachogram. Before sampling, the neurogram was preprocessed to determine the time-integrated value of the signal per cardiac cycle.8 9 The respirogram was sampled in correspondence with the R wave of the ECG.
Statistical analysis was performed by a statistician (B.Z.). Distribution was normal for all variables except for the absolute HF variability and the ratio of LF to HF variability (LF/HF) of RR interval and muscle sympathetic activity. The distribution of these latter measures was normalized using a logarithmic transformation.
A repeated-measures ANOVA first determined if significant trends were present over time. An ANOVA for a nested random effects model was then used to estimate the variance components for subject, time, and random error.16 17 18 These variance components were then used to determine the within (short-term) and between (long-term) intraclass correlation coefficients.16 17 18 Unlike the Pearson correlation coefficients, intraclass correlation coefficients should not be squared to be interpreted as a proportion of variance.16 17 18 The intraclass correlation coefficient itself is the proportion of the total variance present in repeated recordings accounted for by bona fide subject-to-subject variance.16 17 18 The closer the intraclass correlation coefficient is to unity, the more reliable (precise, interchangeable) is the measure.16 17 18 A probability value of .75 or greater indicates a good reliability.
Linear regression analyses determined the relationships between changes in absolute measurements of blood pressure, RR interval and MSNA, and changes in the normalized spectral powers of RR and MSNA. These changes were measured with respect to the first 10-minute recording.
Only systolic blood pressure disclosed a significant decline over time (P<.01), whereas all other variables remained unchanged (Fig 1⇓).
Short-term Intraclass Correlation Coefficients
In this analysis, the four 20-minute recordings of heart rate, muscle MSNA, blood pressure, and respiration performed at day 1, 1 week, 1 month, and 3 months were divided into two consecutive 10-minute periods for the analysis of the short-term reproducibility (Fig 2⇓).
RR interval, systolic and diastolic blood pressures, the number of bursts, and the integrated MSNA achieved the largest intraclass correlation coefficients (range, 1.0 to .93). Integrated muscle sympathetic nerve activity offered a more reliable estimate of MSNA (intraclass correlation coefficient, 1.0) than the burst count (intraclass correlation coefficient, .93). Breathing frequency and the center frequency of the HF components of the variability of both RR interval and muscle sympathetic activity were also highly reliable (range of intraclass correlation coefficients, .95 to .80).
The LF and HF components of RR and MSNA were expressed in absolute and normalized units (% of total power after subtraction of the power of the very low frequency component). Normalization did not consistently improve the short-term reliability of the spectral estimates of RR interval and MSNA. The ratio of LF to HF variability (LF/HF) of RR interval and of MSNA were highly reliable (intraclass correlation coefficients, .82 and .86, respectively).
Long-term Intraclass Correlation Coefficients
Comparison of the recordings performed at day 1, 1 week, 1 month, and 3 months were used to determine the long-term reproducibility of the measurements obtained (Fig 3⇓).
Breathing frequency, the center frequency of the HF components of the variability of both RR interval and muscle sympathetic activity, systolic blood pressure, diastolic blood pressure, and RR interval achieved the highest intraclass correlation coefficients (range, .73 to .61). These measures were more reliable than the number of sympathetic bursts (intraclass correlation coefficient, .34). Normalization markedly improved the long-term reliability of the LF powers of RR interval that became comparable to that of muscle sympathetic nerve activity (Fig 3⇑). The long-term ICC of RR was lower than that of blood pressure (Fig 3⇑) in part because of a marked reduction of RR (approximately 350 ms) in two cases (Fig 4⇓).
Relationships Between Long-term Changes in Absolute Cardiovascular Measures and Changes in RR and MSNA Variabilities
A shortening of RR interval was associated with a reduced absolute LF and a reduced absolute HF variability of RR interval (r=+.56, P<.0001 and r=+.87, P<.0001, respectively), an increased normalized LF variability of RR interval (r=−.40, P<.01), a reduced normalized HF variability of RR interval (r=+.36, P=.01), an increased LF/HF ratio of RR interval (r=−.43, P<.01), and an increased number of sympathetic bursts (r=−.36, P=.01) (Fig 4⇑). Thus increases in heart rate were accompanied by an increase in LF/HF ratio of RR and by increased MSNA burst frequency. After exclusion of the two cases showing a marked change in RR (Fig 4⇑), there was no reduction in either the correlation coefficients or the significance of the relationships described.
In addition, an increase in MSNA was accompanied by higher absolute LF and HF variabilities in MSNA (r=+.35, P=.01 and r=+.43, P<.01, respectively). The relationship between MSNA and normalized MSNA spectral components was not significant. The LF/HF ratio of RR correlated with the LF/HF ratio of MSNA (r=+.30, P<.05) (Fig 5⇓). Changes in SBP did not correlate either with changes in RR or changes in absolute MSNA. Remarkably, however, as SBP decreased, the normalized LF of MSNA increased (r=−.41, P<.01), the normalized HF of MSNA decreased (r=+.35, P=.01), and the LF/HF ratio of MSNA increased (r=−.52, P=.0001) (Fig 6⇓). Similar results were found for diastolic blood pressure.
In this study, we sought to determine the short-term and long-term stability of blood pressure, RR interval, MSNA, and spectral components of RR interval and of MSNA. To determine potential mechanisms governing changes in repeated resting measurements of spectral components of MSNA and RR interval, we examined whether spontaneous changes in hemodynamics and MSNA were tracked by parallel changes in the spectral components of RR variability and MSNA variability.
Over the short term (within the same session), integrated muscle sympathetic nerve activity demonstrated excellent stability. This was also true for blood pressure, RR interval, sympathetic burst frequency, and breathing frequency, all showing intraclass correlation coefficients of >.90. With regard to spectral analysis of variability, both the LF/HF ratio of sympathetic nerve activity and of RR achieved scores of >.8. Over the short term, the LF/HF ratio of MSNA was more stable than that of RR interval.
Repeated measurements on four different occasions in these normal subjects over 3 months showed that the most stable of all the measurements we obtained was that of breathing frequency. During these repeated measurements, all hemodynamic measures as well as sympathetic burst frequency showed considerable spontaneous changes. Absolute measures of the LF and HF components of RR interval variability and MSNA variability were not consistently reproducible over multiple measurements, in comparison to the measurements of the LF/HF ratio of MSNA and RR interval. The LF/HF ratio of MSNA was surprisingly more stable than either sympathetic burst frequency or the LF/HF ratio of RR interval.
The important finding in this study was the link between spontaneous changes in hemodynamic measurements and systematic changes in power spectral densities of both RR interval and MSNA variabilities. Spontaneous decreases in RR interval (tachycardia) were accompanied by increases in burst frequency of sympathetic nerve activity. Decreases in RR interval were also accompanied by a decrease in the normalized HF component of RR variability and an increase in the normalized LF component of RR variability, as well as an increase in the LF/HF ratio of RR interval. These findings, obtained during spontaneous changes in RR interval over repeated measurements, are very similar to those observed during reflex changes in RR interval in response to short term pharmacological changes in blood pressure.8
Spontaneous changes in blood pressure did not correlate with absolute changes in MSNA burst frequency due, at least in part, to the technical limitation of comparing activities from different neural recording sites. Remarkably, however, blood pressure changes correlated with changes in the normalized spectral components of MSNA. As systolic blood pressure fell, there was an accompanying decrease in the normalized HF component of MSNA and an increase in the normalized LF component of MSNA as well as an increase in the LF/HF ratio of MSNA. MSNA responses to blood pressure changes therefore appear to be more sensitively tracked by the spectral indices of MSNA rather than by absolute measures of MSNA. Again, this systematic change in spectral components of MSNA oscillations is also observed when blood pressure is altered by pharmacological methods.8
The present study demonstrates that even during spontaneous physiological changes observed during multiple measurements over an extended period of time, the same paradigm applies as is seen during acute pharmacological changes in blood pressure. Namely, during spontaneous changes in blood pressure and heart rate, there is a systematic change in the oscillatory mechanisms modulating the variabilities of RR interval and MSNA. The factors determining the absolute values of hemodynamics and MSNA are extremely complex and not completely understood. Thus it is not surprising that repeated measurements of these variables over extended periods are far from identical. Hence it is also no surprise that spectral measurements of their variabilities are also far from identical over repeated measures. However, this study shows that changes in spectral components of RR and MSNA variability are determined, consistently and at least in part, by spontaneous changes in RR interval and blood pressure, respectively.
These findings provide novel insights into the mechanisms regulating blood pressure at rest. It is generally assumed that blood pressure varies on repeated resting measurements primarily because of alterations in cardiovascular autonomic drive. Our data, however, suggest that other factors are predominantly involved in mediating blood pressure changes. The major determinants of blood pressure alterations are likely to be varied and include cortical, environmental, and circadian factors. The effect of these influences on blood pressure is modulated or buffered by reflex autonomic mechanisms.19 20 21 The present study suggests that the autonomic nervous system responds to these increases and decreases in pressure in much the same way in which it would respond to pharmacological alterations in pressure in a “baroreflex-like” fashion. Thus day-to-day changes in spectral components of RR and MSNA are a response to, rather than a cause of, blood pressure changes.
Over these repeated measurements, changes in the LF/HF ratio of RR variability were tracked by similar changes in the LF/HF ratio of MSNA. Thus the oscillatory characteristics of the RR interval (which is regulated by both sympathetic and vagal influences) are linked to the oscillatory characteristics of sympathetic neural traffic. This finding adds further support to the concept of common mechanisms governing both vagal and sympathetic cardiovascular control at a central level.8 9
With regard to absolute measures of spectral components of MSNA and RR variability, as the absolute values of either RR or MSNA decreased, the absolute values of their spectral components, both LF and HF, also decreased. This is explained by the decrease in variance of RR and MSNA.
Our findings also have implications for the use of power spectral analysis in drawing conclusions regarding autonomic cardiovascular control in experimental studies. First, since spontaneous fluctuations in spectral measurements of RR interval and MSNA are linked to changes in resting blood pressure and RR interval, it is crucial that measurements of blood pressure and RR interval be obtained simultaneously with measurements of power spectra. Second, although the spectral power measurements are very stable over the short term, long-term stability of these measures is less impressive, in part because of fluctuations in blood pressure and heart rate. Third, the relationship between absolute hemodynamic measures and measurements of spectral power are best seen when spectral components are normalized for total power or when using the LF/HF ratio.
The strengths of the present study include the fact that all measurements were obtained at the same time of day by the same investigator under highly standardized conditions. These studies were carried out in normal subjects without any intervention. Thus our data are free from potential confounding influences of medications or of fluctuations in disease severity. Most important, analysis of these data was carried out blinded to subject and session.
Our findings should not be misinterpreted as implying that power spectral variability can be equated to direct measurements of sympathetic or other autonomic function. They indicate instead that changes in tonic autonomic cardiovascular drives, which determine absolute values of RR interval and blood pressure, are linked to changes in the oscillatory (phasic) characteristics of RR interval and MSNA. The mechanisms underlying the link between tonic and phasic characteristics of cardiovascular regulation are unknown. It may be that reflex responses to apparently random changes in blood pressure induce changes in oscillatory characteristics of neural circulatory control, thus creating an association between blood pressure changes and spectral variability of RR and MSNA.8 19 Alternatively, central mechanisms governing autonomic cardiovascular regulation may function such that changes in strength of signal are accompanied by modulation of the oscillatory characteristics of that signal.
The relationship between oscillations of RR interval and of MSNA with blood pressure, evident over repeated measures, suggests that these oscillatory changes may have functional significance. This question is outside the scope of the present study. Nevertheless, studies of sensory neurons by van Steveninck et al22 suggest that variability patterns in neuronal firing may allow an increase in the information content carried by the neuron. We speculate that simultaneous coding of tonic and oscillatory messages in MSNA may increase the functional (vasoconstrictor) response to the neural signal, in comparison to that which would be observed were the neural message to consist of the tonic signal alone.
In conclusion, we have shown that absolute measures of hemodynamics and measurements of the LF/HF ratio of RR interval and MSNA variability are very stable within the same session. These measurements are less stable over repeated measurements within the same subject in the long term. Spontaneous fluctuations in RR interval and blood pressure are accompanied by systematic and coordinated changes in the LF/HF ratio of RR variability and of MSNA variability. Thus even during spontaneous alterations in cardiovascular control over an extended period, there is a synchrony between changes in absolute cardiovascular measures and changes in their spectral components.
Dr van de Borne is a recipient of a Belgian NATO Research Fellowship (17/B/94/BE), a Dr André Loicq Foundation Travel and Research Award, Belgium, a Bekales Research Award, Belgium, and a Michael J. Brody Fellowship from the University of Iowa, Iowa City. These studies were supported by an American Heart Association Grant-in-Aid, NIH HL-14388, and a Sleep Academic Award from the NIH (Virend Somers). We thank Mary Clary, RN, for technical assistance during these studies and Linda Bang for typing the manuscript.
- Received June 11, 1997.
- Revision received August 20, 1997.
- Accepted September 1, 1997.
- Copyright © 1997 by American Heart Association
Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation.. 1996;93:1043-1065.
Malliani A, Pagani M, Lombardi F, Cerutti S. Cardiovascular neural regulation explored in the frequency domain. Circulation.. 1991;84:482-492.
Pagani M, Lombardi F, Guzzetti S, Rimoldi O, Furlan R, Pizzinelli P, Sandrone G, Malfatto G, Dell’Orto S, Piccaluga E, Turiel M, Baselli G, Cerutti S, Malliani A. Power spectral analysis of heart rate and arterial pressure variabilities as a marker of symaptho-vagal interaction in man and conscious dog. Circ Res.. 1986;59:178-193.
Pitzalis MV, Mastropasqua F, Massari F, Forleo C, Di Maggio M, Passantino A, Colombo R, Di Biase M. Short- and long-term reproducibility of time and frequency domain heart rate variability measurements in normal subjects. Cardiovasc Res.. 1996;32:226-233.
Saul JP, Rea RF, Eckberg DL, Berger RD, Cohen RJ. Heart rate and muscle sympathetic nerve variability during reflex changes of autonomic activity. Am J Physiol.. 1990;258:H713–H721.
Pagani M, Montano N, Porta A, Malliani A, Abboud FM, Birkett C, Somers VK. The relationship between spectral components of cardiovascular variabilities and direct measures of muscle sympathetic nerve activity in humans. Circulation.. 1997;95:1441-1448.
van de Borne P, Montano N, Pagani M, Oren R, Somers VK. Absence of low-frequency variability of sympathetic nerve activity in severe heart failure. Circulation.. 1997;95:1449-1454.
Bernardi L, Hayoz D, Wenzel R, Passino C, Porret CA, Weber R, Noll G. Synchronous vasomotion in the skin microvessels: evidence for central autonomic control mechanism. Eur Heart J.. 1995;16:187. Abstract.
Mark AL, Victor RG, Nerhed G, Wallin BG. Microneurographic studies of the mechanisms of sympathetic nerve responses to static exercise in humans. Circ Res.. 1985;57:461-469.
Montano N, Ruscone TG, Porta A, Lombardi F, Pagani M, Malliani A. Power spectrum analysis of heart rate variability to assess the changes in sympathovagal balance during graded orthostatic tilt. Circulation.. 1994;90:1826-1831.
Brovelli M, Baselli G, Cerutti S, Guzzetti S, Liberati D, Lombardi F, Malliani A, Pagani M, Pizzinelli P. Computerized analysis for an experimental validation of neurophysiological models of heart rate control. Comput Cardiol. 1983;205-208.
Fleiss JL, Slakter MJ, Fischman SL, Park MH, Chilton NW. Inter-examiner reliability in caries trials. J Dent Res.. 1979;58:604-609.
Brown DR, Brown LV, Patwardhan A, Randall DC. Sympathetic activity and blood pressure are tightly coupled at 0.4 Hz in conscious rats. Am J Physiol.. 1994;267:R1378–R1384.
Parati G, Saul JP, Di Rienzo M, Mancia G. Spectral analysis of blood pressure and heart rate variability in evaluating cardiovascular regulation: a critical appraisal. Hypertension.. 1995;25:1276-1286.
van Steveninck RRDR, Lewen GD, Strong SP, Koberle R, Bialek W. Reproducibility and variability in neural spike trains. Science.. 1997;275:1805-1808.