Dynamic changes in cardiac vagal tone as measured by time-series analysis GEORGE E. BILLMAN AND JEAN-PIERRE Department of Physiology, Ohio State University, BILLMAN, GEORGE E., AND JEAN-PIERRE DUJARDIN. Dynamic changes in cardiac vagal tone as measured by time-series analysis. Am. J. Physiol. 258 (Heart Circ. Physiol. 27): H896-
H902, 1990.-A time-series analysis of heart rate variability was evaluated as a marker of cardiac vagal tone using wellcharacterized autonomic interventions. Heart period (R-R interval) was recorded in 14 mongrel dogs from which the amplitude of the respiratory sinus arrhythmia (0.24-1.04 Hz) was determined. Exercise elicited significant (P < 0.01) reductions in the index of vagal tone (control 6.3 & 0.3 In ms2 vs. exercise 2.4 & 0.4 In ms2) that were accompanied by significant (P < 0.01) increases in heart rate (control 123.1 t 5 vs. exercise 201.0 t 7.7 beats/min). The vagal tone index remained >O throughout exercise. After propranolol HCl pretreatment, the vagal tone index rapidly decreased toward zero (control 6.2 t 0.5; exercise 0.7 * 0.3 In ms2), despite significantly lower increases in heart rate (control 109.3 $- 4.2; exercise 178.0 * 7.6 beats/min). Atropine given during exercise evoked significantly greater increases in heart rate in the control (+48.7 t 7.9 beats/min) vs. propranolol (+14.2 t 6.7 beats/min) conditions. These data suggest that 1) high levels of cardiac vagal tone remain during exercise; 2) vagal withdrawal is largely responsible for the heart rate increase after ,&adrenergic receptor blockade; and 3) time-series analysis of the R-R interval can provide a dynamic and noninvasive index of cardiac vagal tone. cardiac parasympathetic tone; heart rate variability; ,8-adrenergic receptor blockade; respiratory sinus arrhythmia; exercise
PERIODIC FLUCTUATIONS in the heart beat were first described well over 200 years ago (14) and have since been used as noninvasive measures of cardiac autonomic activity (1, 4, 7, 11, 17). In particular, the oscillations correlating with respiration, the so-called respiratory sinus arrhythmia, are generally accepted as markers of parasympathetic activity (1, 7, 11, 17, 20, 21). Recently, a number of techniques have been applied to quantify this beat-to-beat heart variability to serve as an index of autonomic tone. The easiest approach involves the calculation of the variance or standard deviation of R-R intervals over some time period, usually hours. This technique has proven to be useful in the identification of gross changes in autonomic activity and has also been shown to be a useful predictor of cardiac mortality in certain disease states (4, 15, 18). This technique, however, cannot distinguish between sympathetic and parasympathetic activity. Therefore, a number of investigators (1, 7, 20, 21) have used power spectral analysis techniques (fast Fourier transform) to H896
0363-6135/90
DUJARDIN Columbus, Ohio 43210 quantify the nonrandom components of the R-R interval variability. In humans and conscious dogs a high frequency peak (>0.15 Hz) and one or two low frequency (0.1 Hz and 0.03 Hz) peaks have been consistently observed in the power spectrum. Pharmacological and surgical (denervation) interventions suggest that the high frequency peak corresponds to parasympathetic activity (respiratory sinus arrhythmia) whereas the low frequency peaks may represent a combination of sympathetic and parasympathetic activity (1, 20, 21). This technique, however, was limited by complicated and time-consuming data analysis that which makes rapid “real time” analysis problematic. Thus dynamic changes in autonomic activity in response to physiological perturbations, such as exercise, would be difficult to evaluate using this technique. Porges and co-workers (19, 22, 23) recently developed a time-series analysis technique that accurately evaluates the amplitude of the respiratory sinus arrhythmia yet allows for real time data analysis over short time intervals. Thus dynamic changes in cardiac vagal tone in response to physiological perturbations could be evaluated using this technique. The autonomic response to exercise is well characterized. It is generally agreed that, in humans and probably other species (24, 29, parasympathetic tone is gradually withdrawn as exercise intensity increases, whereas sympathetic activity increases once heart rate reaches 100 beats/min (-56% of maximum). Thus the heart rate increase associated with low-intensity exercise results almost exclusively from decreased parasympathetic activity, whereas the heart rate increase at higher levels of exercise results from both a decreased parasympathetic activity and augmented sympathetic activity (24, 25). Therefore, it was the purpose of this series of experiments 1) to evaluate time-series signal processing of heart period variability (the amplitude of respiratory sinus arrhythmia) as an index of cardiac vagal tone during pharmacological interventions (atropine and phenylephrine), which have been shown to increase or decrease cardiac parasympathetic activity, and 2) to evaluate dynamic changes in the vagal tone index in response to exercise, a well-characterized autonomic intervention. In particular, the hypothesis that time-series analysis techniques could provide an accurate and dynamic noninvasive measurement of cardiac vagal tone in response to physiological perturbations was tested. METHODS
Fourteen mongrel dogs weighing 11.6-21.8 kg were used in this study. The dogs were given Innovar-Vet
$1.50 Copyright 0 1990 the American Physiological
Society
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
SUBMAXIMAL
EXERCISE
REDUCES
(0.02 mg/kg fentanyl citrate and 1 mg/kg droperidol iv, Pittman-Moore) as a preanesthetic followed by pentobarbital sodium (10 mg/kg iv, Harvey Laboratories) to induce anesthesia. Using strict aseptic techniques, a left thoracotomy was made in the fourth intercostal space. The heart was exposed and supported by a pericardial cradle. Insulated silver-coated copper wires were sutured to the epicardial surface of the left and right ventricle and were later used to record a ventricular electrogram. The heart was also instrumented to measure left ventricular pressure and coronary blood flow. These variables were not used in the present study. All the leads to the cardiovascular instrumentation were tunneled under the skin to exit on the back of the animal’s neck. Pentazocine lactate (Talwin, Winthrop-Breon Lab, 30 mg im) was given as needed to control postoperative pain. In addition, the long-acting local anesthetic bupivacaine HCl (Marcaine, Winthrop-Breon Lab) was used to block the intercostal nerves (pain fibers) in the area of the incision to minimize discomfort to the animal. Each animal was placed on antibiotic therapy (penicillin G x lo6 U im, Burns Veterinary Supply) twice daily for 7 days. The principles governing the care and treatment of animals, as expressed by the American Physiological Society, were followed at all times during this study. In addition, the procedures used in this study were approved by the Ohio State University Institutional Animal Care and Use Committee. Exercise protocol. Three to 4 wk after surgery the studies began. The animals were walked on a motordriven treadmill for several days to familiarize them with the laboratory and to extinguish any orienting response associated with the novel environment. The response to exercise was then assessed using a submaximal exercise protocol as previously described by Stone (27). Briefly, the treadmill exercise lasted a total of 18 min and was divided into 3-min blocks. The protocol began with a 3min warm-up period during which the animal ran at 4.8 km/h at 0% grade. The speed was increased to 6.4 km/ h, and the grade of the treadmill was increased every 3 min as follows: 0,4,8,12, and 16%.Each animal received three exercise tests (24 h apart) that were averaged such that only one set of data was obtained for each animal. Once the control response was determined, the exercise test was repeated after P-adrenergic receptor blockade with propranolol HCl(1 mg/kg iv, Sigma Chemical). The ,&adrenergic receptor agonist isoproterenol HCl (Isuprel, Winthrop, 1 pg/kg iv) was injected before and 5 min after propranolol to confirm the completeness of the blockade. This dose of propranolol completely suppressed the chronotropic and inotropic effects of isoproterenol. On subsequent days a catheter was percutaneously placed in the external jugular vein so that atropine sulfate (50 pg/kg, Lypho-Med) could be administered while the animal was running. Atropine sulfate was given during the last exercise level both during control and during ,&adrenergic receptor blockade exercise tests. Table study protocol. The effects of interventions established to increase or decrease cardiac vagal tone were evaluated as described below. The dogs were trained to
CARDIAC
VAGAL
TONE
H897
lie quietly and unrestrained on a laboratory table. The animals were given bolus injections of phenylephrine (Neo-Synephrine, Winthrop, 10 pg/kg) to raise arterial pressure 30-50 mmHg and thereby reflexively increase cardiac vagal tone. Conversely, the animals were given atropine sulfate (Lypho-Med, 50 pg/kg) both before and after P-adrenergic receptor blockade. Heart rate variability was monitored for 5 min after the drug injections to ensure that peak changes had occurred. Finally, as a terminal experiment, nine dogs were premeditated with morphine sulfate (2 mg/kg im, Eli Lilly). A catheter was percutaneously placed in a cephalic vein and used to administer the anesthesia: a-chloralose (50 mg/kg, Fisher Chemical) and urethan (500 mg/kg, Sigma Chemical). One hour later both cervical vagus nerves were cut. Data analysis. All data were recorded on a Gould m .odel 2800s 8-channel chart recorder and a Teat model MR30 FM cassette tape recorder. An estimate of cardiac vagal tone was obtained using a Delta-Biometrics vagal tone monitor triggering off the electrocardiogram (ECG; the R-R interval). This device employs the time-series signal processing techniques as developed by Porges (22) to estimate the amplitude of respiratory sinus arrhythmia. This method deals with many of the statistical problems associated with extracting the amplitude of the respiratory sinus arrhythmia superimposed on a complex and changing base line of heart rate. The ECG signal was digitized at 1 kHz and sequential R-R intervals were timed to the nearest ms. The no nperiodi .c base-line fluctuations were removed using a moving third-order 2lpoint polynomial function (22). This procedure prevented the leakage of trends and harmonics of nonsinusoidal periodic activity (i.e., transient changes) into the respiratory frequency component. A more detailed presentation may be found in the APPENDIX. Once the “filtering” procedures had been performed, the output of the moving polynomial was processed with a digital bandpass filter to extract the variance in the 0.24- to 1.04-Hz frequency band. A 30-s time epoch was selected based on an assumption of stability. The variance measure was then transformed to its natural logarithm to “normalize” the distribution of the variance estimates to reduce the impact of large differences, i.e., outlying values. An output was obtained every 30 s of the following variables: heart rate, R-R interval, R-R interval variance, and vagal tone index. Heart rate and cardiac vagal tone index were the only variables of interest in the present study. The vagal tone monitor was set to evaluate the 0.24. to 1.04Hz frequency component of the heart period function. This frequency band was selected because it was inclusive of all the breathing frequencies noted in the animals used in this study. Control data were obtained before exercise began with the animal standing on the treadmill (last 3 min before exercise began). In a similar manner, recovery data were obtained 3 min after the cessation of exercise. The analysis was performed on-line. The six 30-s intervals for a given level of exercise were averaged and reported as one value for that level. Thus each data point reported represents 3 min of data collection. As noted above, these data were then averaged across exercise presentations so that onlv one set of values was used for
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H898
SUBMAXIMAL
EXERCISE
REDUCES
a given animal (i.e., the three control exercise tests were averaged for each dog). Phenylephrine, atropine, and vagotomy data were obtained 1 min before and 1 min after the drugs had been given. The postintervention time period corresponded to the time at which the maximal effect was observed. All data were analyzed using analysis of variance (ANOVA) procedures (28). The exercise data were analyzed using a two-factor ANOVA for repeated measures (drug X exercise level, 2 X 8), whereas a one-way ANOVA for repeated-measures test was used on the atropine, phenylephrine, and vagotomy data. When the F ratio was found to exceed the critical value (P < 0.05), Scheffe’s test was used to compare the means. All data are presented as means t SE. RESULTS
Effects of pharmacological interventions at rest on car(PE) elicited a significant diac vagal tone. Phenylephrine (P < 0.01) increase in the vagal tone index (control 5.4 k 0.3, PE 8.2 t 0.2 ln ms2) that was accompanied by a significant reduction in heart rate (control 130.8t 6.9, PE 74.9 * 3.7 beats/min). In a similar manner, atropine sulfate significantly decreased the vagal tone index (control 5.8 kO.5 In ms2, atropine 0.0 t 0.0 In ms2) reaching 0.0 In ms2 in each animal tested. This decrease in vagal tone was accompanied by a large increase in heart rate (control 104 t 6.6, atropine 211.6 t 10.8 beats/min). Similar results were obtained if atropine was given after pretreatment with propranolol. Heart rate significantly increased (P-adrenergic blockade 107.7 t 3.6 vs. atropine + ,&adrenergic blockade 178.7 t 7.9 beats/min), whereas vagal tone index significantly decreased (p-adrenergic blockade 5.8 k 0.7 vs. atropine + P-adrenergic blockade 0.2 t 0.1 In ms2). Bilateral vagotomy yielded results very similar to those
M
CARDIAC
VAGAL
TONE
noted for atropine sulfate. The cardiac vagal tone index significantly decreased (control 6.1 t 0.5 vs. vagotomy 0.1 t 0.1 In ms2), whereas heart rate significantly increased (control 110.8 t 7.4 vs. vagotomy 186.0 t 9.8 beats/min). Vagal tone fell to zero in eight of nine animals. The small vagal tone index values obtained after vagotomy and atropine plus @-adrenergic approach the measurement limits of the vagal tone monitor (i.e., &l ms) and may reflect non-neural influences such as the impact of intrathoracic pressure on the sinoatrial node. For example, after vagotomy 0.1 U on the natural log scale corresponds to 1.5 ms. This contrasts with the vagal tone index of 6.1, which would represent a periodic oscillation of -30 ms. Effect of exercise on cardiac vagal tone. The cardiac vagal tone index response is shown in Fig. 1. As exercise progressed, cardiac vagal tone significantly decreased. As would be expected, heart rate increased in response to exercise (Fig. 2). In an analogous fashion, cardiac vagal tone and heart rate response to exercise as measured by
time-series analysis alsO decreasedafter p-a&ener@c
receptor blockade. Heart rate increased in response to exercise after ,&adrenergic receptor blockade (Fig. 2), but, as one would predict, this heart rate increase was significantly reduced by propranolol as compared with the control situation. The vagal tone index was significantly reduced after P-adrenergic blockade (compared with the control no-drug condition), suggesting that parasympathetic activity was withdrawn more rapidly and to a greater extent when the effects of cardiac sympathetic nerves were blocked. Thus P-adrenergic receptor blockade may place a “compensatory demand” on the parasympathetic system such that vagal tone must be withdrawn to facilitate the exercise-induced increase in cardisc output. Atropine sulfate was administered during the last level of exercise for the control and propranolol conditions to
CONTROL
FIG. 1. Effect of exercise on cardiac vagal tone index before and after p-adrenergic receptor blockade (propranolol HCl, 1.0 mg/kg). Note very low values achieved after /3-adrenergic receptor blockade. ** P < 0.01 control vs. &adrenergic receptor blockade. No other significant differences were noted.
1 o/o
I
I
I
I
4.8/O
6.4/O
6.4/4
6.418
EXERCISE Speed
(kph)/Grade
I 6.4/12
I
1
6.4/16
recov.
LEVEL (%I
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
SUBMAXIMAL
EXERCISE
REDUCES
CARDIAC
VAGAL
TONE
H899
** H 200
T
**
CONTROL
0
180
160
FIG. 2. Effect of exercise on heart rate before and after ,&adrenergic receptor blockade (propranolol HCl, 1 mg/kg). ** P < 0.01 control vs. P-adrenergic receptor blockade. No other significant differences were noted.
140
120
100 I
o/o
I
I
I
4.810
6.4/O
6.414
EXERCISE Speed
(kph)/Grade
I 6.418
1 6.4/
I 12
6.4116
I recov.
LEVEL (%I
test this hypothesis. The results are displayed in Fig. 3. Heart rate significantly (P c 0.01) increased when atropine was given during a control exercise test (change 48.7 k7.9 beats/min), whereas atropine did not significantly alter heart rate after propranolol pretreatment (14.2 t 6.7 beats/min). In an analogous fashion, the vagal tone index significantly decreased (change -3.2 t 0.5 In ms2) when atropine was given during a control test but did not change (change -0.2 t 0.2 In ms2) when atropine was given after ,&adrenergic receptor blockade. DISCUSSION
The present study demonstrates that a time-series analysis technique can be used to evaluate the dynamic response of cardiac vagal tone to physiological perturbations. Submaximal exercise elicited significant reductions in the amplitude of the respiratory sinus arrhythmia (0.24-1.04 Hz of R-R interval variations), which was accompanied by the appropriate elevations in heart rate. Although this vagal component of the R-R variability was reduced by exercise, significant levels of cardiac vagal tone remained throughout the exercise protocol. This observation was confirmed by the large heart rate increase elicited by atropine injection administered while the animal was running. In contrast, after ,&adrenergic receptor blockade with propranolol, exercise elicited significantly greater reductions in cardiac vagal tone despite lower absolute heart rate values. The cardiac vagal index, in fact, was not significantly different from zero, which may reflect the almost total withdrawal of parasympathetic influences from the heart. Indeed, the observation that atropine administered while the animal was running after propranolol pretreatment did not significantly alter heart rate supports this conclusion. Taken together, these data suggest that time-series analysis techniques appropriately measured the dynamic changes in cardiac
vagal tone, (i.e., a gradual reduction) during exercise. Interestingly, when sympathetic effects were removed (by ,&adrenergic receptor blockade), time-series analysis indicated that the heart rate increase associated with exercise was probably limited to the extent that cardiac parasympathetic tone can be withdrawn. These data are consistent with results of previous studies that investigated the parasympathetic and sympathetic contributions to the exercise response as evaluated by pharmacological agents. For example, Robinson et al. (24) found that in humans, most of the initial heart rate response to exercise could be attributed to release of vagal tone. Once heart rate reached ~100 beats/min, sympathetic activity became increasingly important. These results have since been confirmed by severa other investigators in humans and other species (25). Because the results of the present study were consistent with previous findings, one may conclude that dynamic changes in cardiac vagal activity can be noninvasively evaluated using time-series analysis techniques. This illustrates the major advantage of this technique over other measures of heart period variability: namely, the measurement of rapid changes in cardiac vagal activity in response to some stimulus. This technique therefore allows one to investigate changes in vagal tone rather than steady-state events. A number of investigators have demonstrated that power spectral analysis (fast Fourier analysis) of the periodic fluctuations in R-R interval could provide insight as to autonomic regulation of the heart (1, 7, 20, 21). In humans and conscious dogs, for example, a high frequency peak (>0.15 Hz) in the power spectrum has been consistently observed and correlates with cardiac vagal activity (1,20,21). Recently, Porges (22) developed a time-domain analysis technique to assess accurately the amplitude of respiratory sinus arrhythmia. This
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H900
SUBMAXIMAL
EXERCISE
cl
REDUCES
CONTROL
t
240
CONTROL
BETA
BLOCKADE
3. Effect of atropine sulfate on cardiac vagal tone index (top) and heart rate (bottom) before and after /3-adrenergic receptor blockade (propranolol HCl, 1.0 mg/kg). Atropine was injected during last exercise level (6.4 km. h per 16% grade). Note large heart rate increase in control but not during /3-adrenergic receptor blockade exercise tests. ** P < 0.01 before vs. after atropine conditions. No other significant differences were noted. FIG.
method successfully dealt with two historical problems (see Ref. 23) that have limited the application of spectral analysis to the study of physiological rhythms such as respiratory sinus arrhythmia. Briefly, these two problems have been 1) that physiological activities such as heart rate patterns, even in assumed stable-state conditions, are time series that are not stationary, i.e., they have changing means and variance as a function of time; and 2) that the experimental manipulations are, in themselves, creating “nonstationarities” in the physiological systems. Commonly used methods of quantifying the amplitude of respiratory sinus arrhythmia have relied on either simplistic descriptive methods of peak-to-trough or inappropriate statistical assumptions regarding stationarity to employ spectral techniques. The peak-totrough method is vulnerable to the impact of trends that can either obfuscate or amplify an oscillation. Similarly, the spectral method is only accurate if the time series is actually composed of sinusoidal components. Because physiological activity is neither sinusoidal nor without complex trends, both methods have serious limitations. These limitations become exaggerated when the experi-
CARDIAC
VAGAL
TONE
mental manipulation, in itself, is creating a complex trend. Thus, in an exercising paradigm, the metabolic demands on the heart would result in an increase in heart rate that would produce a trend line that would violate the assumption of stationarity for the spectral method and potentially confound the scoring of respiratory sinus arrhythmia with the peak-to-trough method. Porges (22) developed a technique that removes nonstationarities associated with complex aperiodic base lines and also removes slow periodic activity that may produce harmonics in the frequency band of interest (i.e., respiration). For example, periodic activity that is not sinusoidal produces harmonics at integer frequencies in the spectrum. In adult humans, there is often a heart rate rhythm observed in a frequency band between 0.08 and 0.10 Hz. This frequency band is assumed to be related to the homeostatic control of blood pressure via the baroreceptors. This periodicity is far from sinusoidal and, therefore, would produce a harmonic at integer frequencies such as 0.16-0.20 Hz. Note that the frequency band between 0.16 and 0.20 Hz is where respiration normally occurs. Thus spectral methods of respiratory sinus arrhythmia, even if the data appear to be stationary, would still generate spectral densities in the breathing frequency band that would represent the sum of the harmonics and the respiratory sinus arrhythmia. Porges (22) solved this problem by a complex detrending method that dynamically fits into the trends and slower periodicities. When there are no trends or slower periodicities, the method is perfectly correlated with the spectral techniques. However, when trends and periodicities are added, only detrending (filtering) techniques such as the one developed by Porges can accurately extract a criterion signal. It should be noted that if the Porges detrending procedure is first applied to the data set, spectral analysis will yield results equivalent to timeseries analysis. It must also be acknowledged that the time-series analysis technique employed in the present study is subject to a limitation also noted for spectral analysis. Both techniques break down when a complex trend has a rate of change in the frequency band of interest (i.e., high frequency trends, >0.24 Hz). There is as yet no satisfactory solution to this problem. In the present study, relatively short data epochs (i.e., 30 s averaged over 3-min periods), transformation of the variance to natural logarithm, and averaging the resulting natural log units were used to reduce the impact of any spurious rapid shifts of the trend on the frequency band of interest. Finally, unlike spectral analysis, the Porges technique calculates variance in the time domain rather than the frequency domain, over short data epochs (brief time intervals) (23). In other words, because time-domain analysis requires relatively few data points to estimate respiratory sinus arrhythmia, a rapid “real time” output can be achieved. Using this technique, Porges and co-workers demonstrated the amplitude of respiratory sinus arrhythmia correlated to cardiac vagal activity (19, 22, 23). For example, atropine sulfate eliminated this respiratory frequency component of heart rate variability, whereas phenylephrine enhanced it (19). In addition, antiacetyl-
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
SUBMAXIMAL
EXERCISE
REDUCES
cholinesterases have been shown to increase the amplitude of respiratory sinus arrhythmia as measured by this technique (10). In the present study, we demonstrated that atropine sulfate or bilateral vagotomy completely eliminated the 0.24-1.04 Hz frequency peak obtained by time-series analysis. Conversely, activation of the baroreceptor reflex with an increase in arterial pressure elicited reflex reductions in heart rate accompanied by appropriate increases in the cardiac vagal tone index. When taken together, these data strongly suggest that cardiac vagal tone was indeed appropriately measured by this time-series analysis technique. As one would predict, cardiac vagal tone decreases in response to exercise. However, because increased respiration and panting occur during exercise (3, 8, 26), it is possible that these respiratory effects could indirectly alter (reduce) the cardiac vagal index. For example, increases in respiratory rate have been shown to reduce respiratory sinus arrhythmia amplitude (as determined by peak-to-trough measurements, 16). Respiratory frequency increased during exercise both before (control 19.1 t 1.1 breaths/min, peak exercise 24.7 t 1.3 breaths/ min) and after ,&adrenergic receptor blockade (control 19.8 & 2.4 breaths/min, peak exercise 26.5 t 3 breaths/ min), which would tend to reduce respiratory sinus arrhythmia amplitude. However, if it is possible to extrapolate the data obtained in adult human subjects by Hirsch and Bishop (16) to exercising mongrel dogs, then one would calculate that the change in the respiratory frequency noted in the present study could account for at most only 0.6 In ms2 units of reduction, far less than was noted in the present study. Therefore, respiratory rate effects alone could not explain the large reductions in respiratory sinus arrhythmia noted during exercise. In a similar manner, respiratory sinus amplitude decreased to a greater extent after ,&adrenergic receptor blockade, yet respiratory frequency increased to a similar extent after this intervention. It has also been demonstrated that tidal volume will affect the amplitude of the respiratory sinus arrhythmia (i.e., amplitude increases as tidal volume increases; Ref. 16). Because tidal volume has been shown to either not change (8) or to increase with exercise (3, 26), the amplitude of respiratory sinus arrhythmia would be expected to increase rather than decrease during exercise. Furthermore, Eckholdt and Schubert (13) found that during sleep, respiratory sinus arrhythmia varied independent of changes in tidal volume suggesting that these two factors were not causally related. In a-similar manner Arai et al. (2) found that in humans, exercise elicited a M-fold increase in respiratory frequency and a more than fourfold increase in tidal volume. They estimated that the high-frequency peaks (vagal component) of the heart rate power spectrum would increase -25% if only respiratory effects were considered. Yet, they found a 90% reduction in these peaks, which suggested that cardiac vagal tone was markedly reduced during exercise. Thus, when considered together, the effects of exercise on respiration cannot account for the large reduction in the amplitude of respiratory sinus arrhythmia noted during exercise. Snontaneous occurrences of nanting at
CARDIAC
VAGAL
TONE
H901
rest did not alter either the cardiac vagal tone index or shift the 0.24-1.04 Hz frequency peak as measured by fast Fourier analysis. Therefore, panting probably did not contribute significantly to vagal tone index reduction noted in the present study. In summary, the present study demonstrates that time-series analysis of the R-R interval fluctuations may be a powerful and noninvasive means of evaluating cardiac vagal tone not only at rest but also in response to physiologically relevant environmental stressors. This may be a particularly useful technique for the evaluation of the autonomic nervous system of patients with heart disease. In particular, reductions in parasympathetic activity increase the likelihood of dying suddenly (9) and individuals with diseased hearts often have impaired vagal control of the heart (5, 6, 12). This technique may therefore be useful in the identification of patients particularly at risk of dying suddenly (5). This intriguing possibility clearly merits further investigation. APPENDIX
A moving polynomial “filter” was used to remove nonperiodic trends and any periodic component (including harmonics of periodic nonsinusoids) with a dominant frequency below the frequency band of interest. To get to the level of accurately quantifying respiratory sinus arrhythmia with the Porges (22) method, the data needs to be processed through the following operations. I) The ECG needs to be sampled at a rapid rate to detect the inflection point of the R wave (1 kHz in the vagal tone monitor). 2) The R-R intervals need to be timed to the nearest millisecond. 3) Effective artifact correction needs to be employed to deal with beats that are not effectively timed (sequential R-R interval changes ~75137% of preceding beats are flagged). 4) The time series of R-R intervals needs to be sampled at equal time intervals to generate a time series of equally spaced estimates. At this point a data set is created that will be detrended by the moving polynomial filter. The order and number of coefficients of the polynomial is critical to the transfer function of the filter. A cubic order polynomial has been selected because it can “bend” to fit shifting trends without matching the oscillations being studied. The number of coefficients determines the transfer function of the filter. The reciprocal of the duration of the polynomial reflects the slowest frequency influencing the residual time series. The 21-point polynomial based on a time series of data sampled every 250 ms would have a duration of 5.25 s or NO.19 Hz. As a result, any frequency slower than 0.19 Hz will not influence the time series. Thus the sampling rate, number of coefficients, and order of the polynomial determine the transfer function of the filter and its ability to accurately remove lower frequency components (i.e., a high-pass transfer function) as well as to remove the harmonic components of these lower frequency components. The formula for the cubic polynomial is shown below, where the constants “a” are determined by the principal of least squares.
i
( Yt - a~ - alt - azt2 - a3t3)2
t-m
where Yt t, and m midpoint important to unity. values.
represents the raw untransformed data point at time is the number of time windows on each side of the of the polynomial. The moving polynomial has two properties. First, the weights of the coefficients sum Second, the weights are symmetric about the middle
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H902
SUBMAXIMAL
EXERCISE
REDUCES
The authors thank Richard S. Hoskins for technical assistance and Terry Carsner for typing this manuscript. The research reported was supported by the National Heart, Lung, and Blood Institute Grant HL36336.
Address for reprint requests: G. E. Billman, Dept. of Physiology, Ohio State University, 4196 Graves Hall, 333 West 10th Ave., Columbus, OH 43210. Received 27 December 1988; accepted in final form 30 October 1989. REFERENCES 1. AKSELROD, S., D. GORDON, BARGER, AND R. J. COHEN.
F. A. VBEL, D. C. SHANNON, A. C. Power spectrum analysis of heart rate probe of beat-to-beat cardiovascular
fluctuation: a quantitative control. Science Wash. DC 213: 220-222, 1981. 2. AM, Y., P. SAUL, P. ALBRECHT, H. L. HARTLEY, L. S. LILLY, R. J. COHEN, AND W, S. CALUCCI. Modulation of cardiac autonomic activity during and immediately after exercise. Am. J. Physiol. 256 (Heart Circ. Physiol. 25): H132-H141, 3. BAINTON, C. R. Effect of speed vs.
Auf die Sinusarrhythmie
5.
6.
7.
8.
9.
10.
11. 12.
13.
TONE des Herzens. Acta Biol. Med. Ger. 34:
Statistical Essays Vol. II. Haemastaticks. London: Innings and Manby, 1733. L. E., JR., S. T. CARVER, AND A. PLAKUN. Slow heart 15. HINKLE, rates and increased risk of cardiac death in middle-aged man. Arch. Intern. Med. 219: 732-748,1972. 16. HIRSCH, J. A., AND B. BISHOP. Respiratory sinus arrhythmia in humans: how breathing pattern modulates heart rate. Am. J. Physiol. 241 (Heart Circ. Physiol. 10): H620-H629, 1981. 17. KATONA, P., AND F. JIH. Respiratory sinus arrhythmia: noninvasive measure of parasympathetic cardiac control. J. Appl. Physiol. 39: 801-805,1975. 18. KLEIGER, R. E.,
J. P. MILLER, J. T. BIGGER, JR., AND A. J. Moss. Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am. J. Curdiol. 59:
256-262,1987. 19. MCCABE, P.
1989.
J. T., JR., R. E. KLEIGER, J. L. FLEISS, L. M. ROLNITZKY, R. C. STEINMAN, AND J. P. MILLER. Components of heart rate variability measured during healing of acute myocardial infarction. Am. J. Curdiol. 61: 208-215, 1988. BILLMAN, G. E., AND R. S. HOSKINS. Time-series analysis of heart rate variability during submaximal exercise: evidence for reduced cardiac vagal tone in animals susceptible to ventricular fibrillation. Circulation 80: 146-157, 1989. BILLMAN, G. E., P. J. SCHWARTZ, AND H. L. STONE. Baroreceptor reflex control of heart rate: a predictor of sudden cardiac death. Circulation 66: 874-880, 1982. CHESS, G. F., R. M, K. TAM, AND F. R. CALARESU. Influence of cardiac neural inputs on rhythmic variations of heart period in the cat. Am. J. Physiol. 228: 775-780, 1975. CLIFFORD, P. S., J. T. LITZOW, J. H. VON COLDITZ, AND R. L. COON. Effect of chronic pulmonary denervation on ventilatory responses to exercise. J. Appl. Physiol. 61: 603-610, 1986. CORR, P. B., K. A, YAMADA, AND F. X. WITKOWSKI. Mechanisms controlling cardiac autonomic function and their relationships to arrhythomogenesis. In: The Heart and Cardiovascular System, edited by H. A. Fozzard, E. Haber, R. B. Jennings, A. M. Katz, and H. E. Morgan. New York: Raven, 1986, p. 1343-1404. DELLINGER, J. A., H. T. JANSEN, D. J. ZABER, AND S. G. BERNBAUM. Vagal tone monitoring: a potential indicator of anticholinesterase exposure in Macaca mulatta. Toxicology 49: 227-235,1988. ECKBERG, D. Human sinus arrhythmia as an index of vagal cardiac outflow. J. Appl. Physiol. 54: 961-966,1975. ECKBERG, D. L., M, DRABINSKY, AND E. BRAUNWALD. Defective cardiac parasympathetic control in patients with coronary artery disease. N. Engl. J. Med. 285: 877-883,197l. ECKHOLDT, K., AND E. SCHUBERT. Zum Enflu B der Atemtiefe
VAGAL
767-771,1973. 14. HALES, S.
grade and shivering on ventilation in dogs during active exercise. J. Appl. Physiol. 33: 778-787,
1973. 4. BIGGER,
CARDIAC
20.
21.
22.
23.
M., B. G. YONGUE, P. K. ACKLES, AND S. W. PORGES. Changes in heart period, heart-period variability and spectral analysis estimate of respiratory sinus arrhythmia in response to pharmacological manipulations of the baroreceptor reflex in cats. Psychophysiology 22: 195-203,1985. PAGANI, M., F. LOMBARD, S. GUZZETTI, 0. RUINOLDI, R. FURLAN, P. PIZZINELLI, G. SANDRONE, G. ANALFALTO, S. DELL’ORTO, R. PICCIALUGA, M. TURIEL, G. BARELLI, S. CERULTI, AND A. MALLIANI. Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circ. Res. 59: 178-193, 1986. POMERANZ, B., R. J. B. MACAULAY, M. A. CAUDILL, I. KUTZ, D. ADAM, D. GORDON, K. M. KILBORN, A. C. BARGER, D. C. SHANNON, R. J. COHEN, AND H. BINSON. Assessment of autonomic function in human heart rate spectral analysis. Am. J. Physiol. 249 (Heart Circ. Physiol. 18): H867-H875,1985. PORGES, S. W. (Inventor). Method and Apparatus for Evaluating Rhythmic Oscillations in Periodic Response Systems. US Patent 4,510,944. 16 Apr. 1985. PORGES, S. W. Respiratory sinus arrhythmia: physiological basis, quantitative methods and clinical implications. In: Cardiac, Respiratory and Cardiosomatic Psychophysiology, edited by P. Grossman, K. Janssen, and D. Vaith. New York: Plenum, 1986, p. lOl115. ROBINSON, B. WALD. Control
F., S. E. EPSTEIN, G. D. BEISER, AND E. BRAUNof heart rate by the autonomous nervous system: studies in man on the interrelationships between baroreceptor mechanisms and exercise. Circ. Res. 19: 400-411, 1966. 25. ROWELL, L. B. Human Circulation Regulation During Physical Stress. New York: Oxford University Press, 1986, p. 213-256. 26. STAVERT, D. M., P. REISCHI, AND B. J. O’LOUGHLIN. A respiratory mask for resting and exercising dogs. J. Appt. Physiol. 52: 500-504, 24.
1982.
27. STONE, H. L. Cardiac function and exercise training in conscious dogs. J. Appl. Physiol. 42: 824-832,1977. 28. WINER, B. J. Statistical Principles in Experimental Design. New York: McGraw-Hill, 1971, p. 514-603.
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H902, 1990.-A time-series analysis of heart rate variability was evaluated as a marker of cardiac vagal tone using wellcharacterized autonomic interventions. Heart period (R-R interval) was recorded in 14 mongrel dogs from which the amplitude of the respiratory sinus arrhythmia (0.24-1.04 Hz) was determined. Exercise elicited significant (P < 0.01) reductions in the index of vagal tone (control 6.3 & 0.3 In ms2 vs. exercise 2.4 & 0.4 In ms2) that were accompanied by significant (P < 0.01) increases in heart rate (control 123.1 t 5 vs. exercise 201.0 t 7.7 beats/min). The vagal tone index remained >O throughout exercise. After propranolol HCl pretreatment, the vagal tone index rapidly decreased toward zero (control 6.2 t 0.5; exercise 0.7 * 0.3 In ms2), despite significantly lower increases in heart rate (control 109.3 $- 4.2; exercise 178.0 * 7.6 beats/min). Atropine given during exercise evoked significantly greater increases in heart rate in the control (+48.7 t 7.9 beats/min) vs. propranolol (+14.2 t 6.7 beats/min) conditions. These data suggest that 1) high levels of cardiac vagal tone remain during exercise; 2) vagal withdrawal is largely responsible for the heart rate increase after ,&adrenergic receptor blockade; and 3) time-series analysis of the R-R interval can provide a dynamic and noninvasive index of cardiac vagal tone. cardiac parasympathetic tone; heart rate variability; ,8-adrenergic receptor blockade; respiratory sinus arrhythmia; exercise
PERIODIC FLUCTUATIONS in the heart beat were first described well over 200 years ago (14) and have since been used as noninvasive measures of cardiac autonomic activity (1, 4, 7, 11, 17). In particular, the oscillations correlating with respiration, the so-called respiratory sinus arrhythmia, are generally accepted as markers of parasympathetic activity (1, 7, 11, 17, 20, 21). Recently, a number of techniques have been applied to quantify this beat-to-beat heart variability to serve as an index of autonomic tone. The easiest approach involves the calculation of the variance or standard deviation of R-R intervals over some time period, usually hours. This technique has proven to be useful in the identification of gross changes in autonomic activity and has also been shown to be a useful predictor of cardiac mortality in certain disease states (4, 15, 18). This technique, however, cannot distinguish between sympathetic and parasympathetic activity. Therefore, a number of investigators (1, 7, 20, 21) have used power spectral analysis techniques (fast Fourier transform) to H896
0363-6135/90
DUJARDIN Columbus, Ohio 43210 quantify the nonrandom components of the R-R interval variability. In humans and conscious dogs a high frequency peak (>0.15 Hz) and one or two low frequency (0.1 Hz and 0.03 Hz) peaks have been consistently observed in the power spectrum. Pharmacological and surgical (denervation) interventions suggest that the high frequency peak corresponds to parasympathetic activity (respiratory sinus arrhythmia) whereas the low frequency peaks may represent a combination of sympathetic and parasympathetic activity (1, 20, 21). This technique, however, was limited by complicated and time-consuming data analysis that which makes rapid “real time” analysis problematic. Thus dynamic changes in autonomic activity in response to physiological perturbations, such as exercise, would be difficult to evaluate using this technique. Porges and co-workers (19, 22, 23) recently developed a time-series analysis technique that accurately evaluates the amplitude of the respiratory sinus arrhythmia yet allows for real time data analysis over short time intervals. Thus dynamic changes in cardiac vagal tone in response to physiological perturbations could be evaluated using this technique. The autonomic response to exercise is well characterized. It is generally agreed that, in humans and probably other species (24, 29, parasympathetic tone is gradually withdrawn as exercise intensity increases, whereas sympathetic activity increases once heart rate reaches 100 beats/min (-56% of maximum). Thus the heart rate increase associated with low-intensity exercise results almost exclusively from decreased parasympathetic activity, whereas the heart rate increase at higher levels of exercise results from both a decreased parasympathetic activity and augmented sympathetic activity (24, 25). Therefore, it was the purpose of this series of experiments 1) to evaluate time-series signal processing of heart period variability (the amplitude of respiratory sinus arrhythmia) as an index of cardiac vagal tone during pharmacological interventions (atropine and phenylephrine), which have been shown to increase or decrease cardiac parasympathetic activity, and 2) to evaluate dynamic changes in the vagal tone index in response to exercise, a well-characterized autonomic intervention. In particular, the hypothesis that time-series analysis techniques could provide an accurate and dynamic noninvasive measurement of cardiac vagal tone in response to physiological perturbations was tested. METHODS
Fourteen mongrel dogs weighing 11.6-21.8 kg were used in this study. The dogs were given Innovar-Vet
$1.50 Copyright 0 1990 the American Physiological
Society
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
SUBMAXIMAL
EXERCISE
REDUCES
(0.02 mg/kg fentanyl citrate and 1 mg/kg droperidol iv, Pittman-Moore) as a preanesthetic followed by pentobarbital sodium (10 mg/kg iv, Harvey Laboratories) to induce anesthesia. Using strict aseptic techniques, a left thoracotomy was made in the fourth intercostal space. The heart was exposed and supported by a pericardial cradle. Insulated silver-coated copper wires were sutured to the epicardial surface of the left and right ventricle and were later used to record a ventricular electrogram. The heart was also instrumented to measure left ventricular pressure and coronary blood flow. These variables were not used in the present study. All the leads to the cardiovascular instrumentation were tunneled under the skin to exit on the back of the animal’s neck. Pentazocine lactate (Talwin, Winthrop-Breon Lab, 30 mg im) was given as needed to control postoperative pain. In addition, the long-acting local anesthetic bupivacaine HCl (Marcaine, Winthrop-Breon Lab) was used to block the intercostal nerves (pain fibers) in the area of the incision to minimize discomfort to the animal. Each animal was placed on antibiotic therapy (penicillin G x lo6 U im, Burns Veterinary Supply) twice daily for 7 days. The principles governing the care and treatment of animals, as expressed by the American Physiological Society, were followed at all times during this study. In addition, the procedures used in this study were approved by the Ohio State University Institutional Animal Care and Use Committee. Exercise protocol. Three to 4 wk after surgery the studies began. The animals were walked on a motordriven treadmill for several days to familiarize them with the laboratory and to extinguish any orienting response associated with the novel environment. The response to exercise was then assessed using a submaximal exercise protocol as previously described by Stone (27). Briefly, the treadmill exercise lasted a total of 18 min and was divided into 3-min blocks. The protocol began with a 3min warm-up period during which the animal ran at 4.8 km/h at 0% grade. The speed was increased to 6.4 km/ h, and the grade of the treadmill was increased every 3 min as follows: 0,4,8,12, and 16%.Each animal received three exercise tests (24 h apart) that were averaged such that only one set of data was obtained for each animal. Once the control response was determined, the exercise test was repeated after P-adrenergic receptor blockade with propranolol HCl(1 mg/kg iv, Sigma Chemical). The ,&adrenergic receptor agonist isoproterenol HCl (Isuprel, Winthrop, 1 pg/kg iv) was injected before and 5 min after propranolol to confirm the completeness of the blockade. This dose of propranolol completely suppressed the chronotropic and inotropic effects of isoproterenol. On subsequent days a catheter was percutaneously placed in the external jugular vein so that atropine sulfate (50 pg/kg, Lypho-Med) could be administered while the animal was running. Atropine sulfate was given during the last exercise level both during control and during ,&adrenergic receptor blockade exercise tests. Table study protocol. The effects of interventions established to increase or decrease cardiac vagal tone were evaluated as described below. The dogs were trained to
CARDIAC
VAGAL
TONE
H897
lie quietly and unrestrained on a laboratory table. The animals were given bolus injections of phenylephrine (Neo-Synephrine, Winthrop, 10 pg/kg) to raise arterial pressure 30-50 mmHg and thereby reflexively increase cardiac vagal tone. Conversely, the animals were given atropine sulfate (Lypho-Med, 50 pg/kg) both before and after P-adrenergic receptor blockade. Heart rate variability was monitored for 5 min after the drug injections to ensure that peak changes had occurred. Finally, as a terminal experiment, nine dogs were premeditated with morphine sulfate (2 mg/kg im, Eli Lilly). A catheter was percutaneously placed in a cephalic vein and used to administer the anesthesia: a-chloralose (50 mg/kg, Fisher Chemical) and urethan (500 mg/kg, Sigma Chemical). One hour later both cervical vagus nerves were cut. Data analysis. All data were recorded on a Gould m .odel 2800s 8-channel chart recorder and a Teat model MR30 FM cassette tape recorder. An estimate of cardiac vagal tone was obtained using a Delta-Biometrics vagal tone monitor triggering off the electrocardiogram (ECG; the R-R interval). This device employs the time-series signal processing techniques as developed by Porges (22) to estimate the amplitude of respiratory sinus arrhythmia. This method deals with many of the statistical problems associated with extracting the amplitude of the respiratory sinus arrhythmia superimposed on a complex and changing base line of heart rate. The ECG signal was digitized at 1 kHz and sequential R-R intervals were timed to the nearest ms. The no nperiodi .c base-line fluctuations were removed using a moving third-order 2lpoint polynomial function (22). This procedure prevented the leakage of trends and harmonics of nonsinusoidal periodic activity (i.e., transient changes) into the respiratory frequency component. A more detailed presentation may be found in the APPENDIX. Once the “filtering” procedures had been performed, the output of the moving polynomial was processed with a digital bandpass filter to extract the variance in the 0.24- to 1.04-Hz frequency band. A 30-s time epoch was selected based on an assumption of stability. The variance measure was then transformed to its natural logarithm to “normalize” the distribution of the variance estimates to reduce the impact of large differences, i.e., outlying values. An output was obtained every 30 s of the following variables: heart rate, R-R interval, R-R interval variance, and vagal tone index. Heart rate and cardiac vagal tone index were the only variables of interest in the present study. The vagal tone monitor was set to evaluate the 0.24. to 1.04Hz frequency component of the heart period function. This frequency band was selected because it was inclusive of all the breathing frequencies noted in the animals used in this study. Control data were obtained before exercise began with the animal standing on the treadmill (last 3 min before exercise began). In a similar manner, recovery data were obtained 3 min after the cessation of exercise. The analysis was performed on-line. The six 30-s intervals for a given level of exercise were averaged and reported as one value for that level. Thus each data point reported represents 3 min of data collection. As noted above, these data were then averaged across exercise presentations so that onlv one set of values was used for
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H898
SUBMAXIMAL
EXERCISE
REDUCES
a given animal (i.e., the three control exercise tests were averaged for each dog). Phenylephrine, atropine, and vagotomy data were obtained 1 min before and 1 min after the drugs had been given. The postintervention time period corresponded to the time at which the maximal effect was observed. All data were analyzed using analysis of variance (ANOVA) procedures (28). The exercise data were analyzed using a two-factor ANOVA for repeated measures (drug X exercise level, 2 X 8), whereas a one-way ANOVA for repeated-measures test was used on the atropine, phenylephrine, and vagotomy data. When the F ratio was found to exceed the critical value (P < 0.05), Scheffe’s test was used to compare the means. All data are presented as means t SE. RESULTS
Effects of pharmacological interventions at rest on car(PE) elicited a significant diac vagal tone. Phenylephrine (P < 0.01) increase in the vagal tone index (control 5.4 k 0.3, PE 8.2 t 0.2 ln ms2) that was accompanied by a significant reduction in heart rate (control 130.8t 6.9, PE 74.9 * 3.7 beats/min). In a similar manner, atropine sulfate significantly decreased the vagal tone index (control 5.8 kO.5 In ms2, atropine 0.0 t 0.0 In ms2) reaching 0.0 In ms2 in each animal tested. This decrease in vagal tone was accompanied by a large increase in heart rate (control 104 t 6.6, atropine 211.6 t 10.8 beats/min). Similar results were obtained if atropine was given after pretreatment with propranolol. Heart rate significantly increased (P-adrenergic blockade 107.7 t 3.6 vs. atropine + ,&adrenergic blockade 178.7 t 7.9 beats/min), whereas vagal tone index significantly decreased (p-adrenergic blockade 5.8 k 0.7 vs. atropine + P-adrenergic blockade 0.2 t 0.1 In ms2). Bilateral vagotomy yielded results very similar to those
M
CARDIAC
VAGAL
TONE
noted for atropine sulfate. The cardiac vagal tone index significantly decreased (control 6.1 t 0.5 vs. vagotomy 0.1 t 0.1 In ms2), whereas heart rate significantly increased (control 110.8 t 7.4 vs. vagotomy 186.0 t 9.8 beats/min). Vagal tone fell to zero in eight of nine animals. The small vagal tone index values obtained after vagotomy and atropine plus @-adrenergic approach the measurement limits of the vagal tone monitor (i.e., &l ms) and may reflect non-neural influences such as the impact of intrathoracic pressure on the sinoatrial node. For example, after vagotomy 0.1 U on the natural log scale corresponds to 1.5 ms. This contrasts with the vagal tone index of 6.1, which would represent a periodic oscillation of -30 ms. Effect of exercise on cardiac vagal tone. The cardiac vagal tone index response is shown in Fig. 1. As exercise progressed, cardiac vagal tone significantly decreased. As would be expected, heart rate increased in response to exercise (Fig. 2). In an analogous fashion, cardiac vagal tone and heart rate response to exercise as measured by
time-series analysis alsO decreasedafter p-a&ener@c
receptor blockade. Heart rate increased in response to exercise after ,&adrenergic receptor blockade (Fig. 2), but, as one would predict, this heart rate increase was significantly reduced by propranolol as compared with the control situation. The vagal tone index was significantly reduced after P-adrenergic blockade (compared with the control no-drug condition), suggesting that parasympathetic activity was withdrawn more rapidly and to a greater extent when the effects of cardiac sympathetic nerves were blocked. Thus P-adrenergic receptor blockade may place a “compensatory demand” on the parasympathetic system such that vagal tone must be withdrawn to facilitate the exercise-induced increase in cardisc output. Atropine sulfate was administered during the last level of exercise for the control and propranolol conditions to
CONTROL
FIG. 1. Effect of exercise on cardiac vagal tone index before and after p-adrenergic receptor blockade (propranolol HCl, 1.0 mg/kg). Note very low values achieved after /3-adrenergic receptor blockade. ** P < 0.01 control vs. &adrenergic receptor blockade. No other significant differences were noted.
1 o/o
I
I
I
I
4.8/O
6.4/O
6.4/4
6.418
EXERCISE Speed
(kph)/Grade
I 6.4/12
I
1
6.4/16
recov.
LEVEL (%I
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
SUBMAXIMAL
EXERCISE
REDUCES
CARDIAC
VAGAL
TONE
H899
** H 200
T
**
CONTROL
0
180
160
FIG. 2. Effect of exercise on heart rate before and after ,&adrenergic receptor blockade (propranolol HCl, 1 mg/kg). ** P < 0.01 control vs. P-adrenergic receptor blockade. No other significant differences were noted.
140
120
100 I
o/o
I
I
I
4.810
6.4/O
6.414
EXERCISE Speed
(kph)/Grade
I 6.418
1 6.4/
I 12
6.4116
I recov.
LEVEL (%I
test this hypothesis. The results are displayed in Fig. 3. Heart rate significantly (P c 0.01) increased when atropine was given during a control exercise test (change 48.7 k7.9 beats/min), whereas atropine did not significantly alter heart rate after propranolol pretreatment (14.2 t 6.7 beats/min). In an analogous fashion, the vagal tone index significantly decreased (change -3.2 t 0.5 In ms2) when atropine was given during a control test but did not change (change -0.2 t 0.2 In ms2) when atropine was given after ,&adrenergic receptor blockade. DISCUSSION
The present study demonstrates that a time-series analysis technique can be used to evaluate the dynamic response of cardiac vagal tone to physiological perturbations. Submaximal exercise elicited significant reductions in the amplitude of the respiratory sinus arrhythmia (0.24-1.04 Hz of R-R interval variations), which was accompanied by the appropriate elevations in heart rate. Although this vagal component of the R-R variability was reduced by exercise, significant levels of cardiac vagal tone remained throughout the exercise protocol. This observation was confirmed by the large heart rate increase elicited by atropine injection administered while the animal was running. In contrast, after ,&adrenergic receptor blockade with propranolol, exercise elicited significantly greater reductions in cardiac vagal tone despite lower absolute heart rate values. The cardiac vagal index, in fact, was not significantly different from zero, which may reflect the almost total withdrawal of parasympathetic influences from the heart. Indeed, the observation that atropine administered while the animal was running after propranolol pretreatment did not significantly alter heart rate supports this conclusion. Taken together, these data suggest that time-series analysis techniques appropriately measured the dynamic changes in cardiac
vagal tone, (i.e., a gradual reduction) during exercise. Interestingly, when sympathetic effects were removed (by ,&adrenergic receptor blockade), time-series analysis indicated that the heart rate increase associated with exercise was probably limited to the extent that cardiac parasympathetic tone can be withdrawn. These data are consistent with results of previous studies that investigated the parasympathetic and sympathetic contributions to the exercise response as evaluated by pharmacological agents. For example, Robinson et al. (24) found that in humans, most of the initial heart rate response to exercise could be attributed to release of vagal tone. Once heart rate reached ~100 beats/min, sympathetic activity became increasingly important. These results have since been confirmed by severa other investigators in humans and other species (25). Because the results of the present study were consistent with previous findings, one may conclude that dynamic changes in cardiac vagal activity can be noninvasively evaluated using time-series analysis techniques. This illustrates the major advantage of this technique over other measures of heart period variability: namely, the measurement of rapid changes in cardiac vagal activity in response to some stimulus. This technique therefore allows one to investigate changes in vagal tone rather than steady-state events. A number of investigators have demonstrated that power spectral analysis (fast Fourier analysis) of the periodic fluctuations in R-R interval could provide insight as to autonomic regulation of the heart (1, 7, 20, 21). In humans and conscious dogs, for example, a high frequency peak (>0.15 Hz) in the power spectrum has been consistently observed and correlates with cardiac vagal activity (1,20,21). Recently, Porges (22) developed a time-domain analysis technique to assess accurately the amplitude of respiratory sinus arrhythmia. This
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H900
SUBMAXIMAL
EXERCISE
cl
REDUCES
CONTROL
t
240
CONTROL
BETA
BLOCKADE
3. Effect of atropine sulfate on cardiac vagal tone index (top) and heart rate (bottom) before and after /3-adrenergic receptor blockade (propranolol HCl, 1.0 mg/kg). Atropine was injected during last exercise level (6.4 km. h per 16% grade). Note large heart rate increase in control but not during /3-adrenergic receptor blockade exercise tests. ** P < 0.01 before vs. after atropine conditions. No other significant differences were noted. FIG.
method successfully dealt with two historical problems (see Ref. 23) that have limited the application of spectral analysis to the study of physiological rhythms such as respiratory sinus arrhythmia. Briefly, these two problems have been 1) that physiological activities such as heart rate patterns, even in assumed stable-state conditions, are time series that are not stationary, i.e., they have changing means and variance as a function of time; and 2) that the experimental manipulations are, in themselves, creating “nonstationarities” in the physiological systems. Commonly used methods of quantifying the amplitude of respiratory sinus arrhythmia have relied on either simplistic descriptive methods of peak-to-trough or inappropriate statistical assumptions regarding stationarity to employ spectral techniques. The peak-totrough method is vulnerable to the impact of trends that can either obfuscate or amplify an oscillation. Similarly, the spectral method is only accurate if the time series is actually composed of sinusoidal components. Because physiological activity is neither sinusoidal nor without complex trends, both methods have serious limitations. These limitations become exaggerated when the experi-
CARDIAC
VAGAL
TONE
mental manipulation, in itself, is creating a complex trend. Thus, in an exercising paradigm, the metabolic demands on the heart would result in an increase in heart rate that would produce a trend line that would violate the assumption of stationarity for the spectral method and potentially confound the scoring of respiratory sinus arrhythmia with the peak-to-trough method. Porges (22) developed a technique that removes nonstationarities associated with complex aperiodic base lines and also removes slow periodic activity that may produce harmonics in the frequency band of interest (i.e., respiration). For example, periodic activity that is not sinusoidal produces harmonics at integer frequencies in the spectrum. In adult humans, there is often a heart rate rhythm observed in a frequency band between 0.08 and 0.10 Hz. This frequency band is assumed to be related to the homeostatic control of blood pressure via the baroreceptors. This periodicity is far from sinusoidal and, therefore, would produce a harmonic at integer frequencies such as 0.16-0.20 Hz. Note that the frequency band between 0.16 and 0.20 Hz is where respiration normally occurs. Thus spectral methods of respiratory sinus arrhythmia, even if the data appear to be stationary, would still generate spectral densities in the breathing frequency band that would represent the sum of the harmonics and the respiratory sinus arrhythmia. Porges (22) solved this problem by a complex detrending method that dynamically fits into the trends and slower periodicities. When there are no trends or slower periodicities, the method is perfectly correlated with the spectral techniques. However, when trends and periodicities are added, only detrending (filtering) techniques such as the one developed by Porges can accurately extract a criterion signal. It should be noted that if the Porges detrending procedure is first applied to the data set, spectral analysis will yield results equivalent to timeseries analysis. It must also be acknowledged that the time-series analysis technique employed in the present study is subject to a limitation also noted for spectral analysis. Both techniques break down when a complex trend has a rate of change in the frequency band of interest (i.e., high frequency trends, >0.24 Hz). There is as yet no satisfactory solution to this problem. In the present study, relatively short data epochs (i.e., 30 s averaged over 3-min periods), transformation of the variance to natural logarithm, and averaging the resulting natural log units were used to reduce the impact of any spurious rapid shifts of the trend on the frequency band of interest. Finally, unlike spectral analysis, the Porges technique calculates variance in the time domain rather than the frequency domain, over short data epochs (brief time intervals) (23). In other words, because time-domain analysis requires relatively few data points to estimate respiratory sinus arrhythmia, a rapid “real time” output can be achieved. Using this technique, Porges and co-workers demonstrated the amplitude of respiratory sinus arrhythmia correlated to cardiac vagal activity (19, 22, 23). For example, atropine sulfate eliminated this respiratory frequency component of heart rate variability, whereas phenylephrine enhanced it (19). In addition, antiacetyl-
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
SUBMAXIMAL
EXERCISE
REDUCES
cholinesterases have been shown to increase the amplitude of respiratory sinus arrhythmia as measured by this technique (10). In the present study, we demonstrated that atropine sulfate or bilateral vagotomy completely eliminated the 0.24-1.04 Hz frequency peak obtained by time-series analysis. Conversely, activation of the baroreceptor reflex with an increase in arterial pressure elicited reflex reductions in heart rate accompanied by appropriate increases in the cardiac vagal tone index. When taken together, these data strongly suggest that cardiac vagal tone was indeed appropriately measured by this time-series analysis technique. As one would predict, cardiac vagal tone decreases in response to exercise. However, because increased respiration and panting occur during exercise (3, 8, 26), it is possible that these respiratory effects could indirectly alter (reduce) the cardiac vagal index. For example, increases in respiratory rate have been shown to reduce respiratory sinus arrhythmia amplitude (as determined by peak-to-trough measurements, 16). Respiratory frequency increased during exercise both before (control 19.1 t 1.1 breaths/min, peak exercise 24.7 t 1.3 breaths/ min) and after ,&adrenergic receptor blockade (control 19.8 & 2.4 breaths/min, peak exercise 26.5 t 3 breaths/ min), which would tend to reduce respiratory sinus arrhythmia amplitude. However, if it is possible to extrapolate the data obtained in adult human subjects by Hirsch and Bishop (16) to exercising mongrel dogs, then one would calculate that the change in the respiratory frequency noted in the present study could account for at most only 0.6 In ms2 units of reduction, far less than was noted in the present study. Therefore, respiratory rate effects alone could not explain the large reductions in respiratory sinus arrhythmia noted during exercise. In a similar manner, respiratory sinus amplitude decreased to a greater extent after ,&adrenergic receptor blockade, yet respiratory frequency increased to a similar extent after this intervention. It has also been demonstrated that tidal volume will affect the amplitude of the respiratory sinus arrhythmia (i.e., amplitude increases as tidal volume increases; Ref. 16). Because tidal volume has been shown to either not change (8) or to increase with exercise (3, 26), the amplitude of respiratory sinus arrhythmia would be expected to increase rather than decrease during exercise. Furthermore, Eckholdt and Schubert (13) found that during sleep, respiratory sinus arrhythmia varied independent of changes in tidal volume suggesting that these two factors were not causally related. In a-similar manner Arai et al. (2) found that in humans, exercise elicited a M-fold increase in respiratory frequency and a more than fourfold increase in tidal volume. They estimated that the high-frequency peaks (vagal component) of the heart rate power spectrum would increase -25% if only respiratory effects were considered. Yet, they found a 90% reduction in these peaks, which suggested that cardiac vagal tone was markedly reduced during exercise. Thus, when considered together, the effects of exercise on respiration cannot account for the large reduction in the amplitude of respiratory sinus arrhythmia noted during exercise. Snontaneous occurrences of nanting at
CARDIAC
VAGAL
TONE
H901
rest did not alter either the cardiac vagal tone index or shift the 0.24-1.04 Hz frequency peak as measured by fast Fourier analysis. Therefore, panting probably did not contribute significantly to vagal tone index reduction noted in the present study. In summary, the present study demonstrates that time-series analysis of the R-R interval fluctuations may be a powerful and noninvasive means of evaluating cardiac vagal tone not only at rest but also in response to physiologically relevant environmental stressors. This may be a particularly useful technique for the evaluation of the autonomic nervous system of patients with heart disease. In particular, reductions in parasympathetic activity increase the likelihood of dying suddenly (9) and individuals with diseased hearts often have impaired vagal control of the heart (5, 6, 12). This technique may therefore be useful in the identification of patients particularly at risk of dying suddenly (5). This intriguing possibility clearly merits further investigation. APPENDIX
A moving polynomial “filter” was used to remove nonperiodic trends and any periodic component (including harmonics of periodic nonsinusoids) with a dominant frequency below the frequency band of interest. To get to the level of accurately quantifying respiratory sinus arrhythmia with the Porges (22) method, the data needs to be processed through the following operations. I) The ECG needs to be sampled at a rapid rate to detect the inflection point of the R wave (1 kHz in the vagal tone monitor). 2) The R-R intervals need to be timed to the nearest millisecond. 3) Effective artifact correction needs to be employed to deal with beats that are not effectively timed (sequential R-R interval changes ~75137% of preceding beats are flagged). 4) The time series of R-R intervals needs to be sampled at equal time intervals to generate a time series of equally spaced estimates. At this point a data set is created that will be detrended by the moving polynomial filter. The order and number of coefficients of the polynomial is critical to the transfer function of the filter. A cubic order polynomial has been selected because it can “bend” to fit shifting trends without matching the oscillations being studied. The number of coefficients determines the transfer function of the filter. The reciprocal of the duration of the polynomial reflects the slowest frequency influencing the residual time series. The 21-point polynomial based on a time series of data sampled every 250 ms would have a duration of 5.25 s or NO.19 Hz. As a result, any frequency slower than 0.19 Hz will not influence the time series. Thus the sampling rate, number of coefficients, and order of the polynomial determine the transfer function of the filter and its ability to accurately remove lower frequency components (i.e., a high-pass transfer function) as well as to remove the harmonic components of these lower frequency components. The formula for the cubic polynomial is shown below, where the constants “a” are determined by the principal of least squares.
i
( Yt - a~ - alt - azt2 - a3t3)2
t-m
where Yt t, and m midpoint important to unity. values.
represents the raw untransformed data point at time is the number of time windows on each side of the of the polynomial. The moving polynomial has two properties. First, the weights of the coefficients sum Second, the weights are symmetric about the middle
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
H902
SUBMAXIMAL
EXERCISE
REDUCES
The authors thank Richard S. Hoskins for technical assistance and Terry Carsner for typing this manuscript. The research reported was supported by the National Heart, Lung, and Blood Institute Grant HL36336.
Address for reprint requests: G. E. Billman, Dept. of Physiology, Ohio State University, 4196 Graves Hall, 333 West 10th Ave., Columbus, OH 43210. Received 27 December 1988; accepted in final form 30 October 1989. REFERENCES 1. AKSELROD, S., D. GORDON, BARGER, AND R. J. COHEN.
F. A. VBEL, D. C. SHANNON, A. C. Power spectrum analysis of heart rate probe of beat-to-beat cardiovascular
fluctuation: a quantitative control. Science Wash. DC 213: 220-222, 1981. 2. AM, Y., P. SAUL, P. ALBRECHT, H. L. HARTLEY, L. S. LILLY, R. J. COHEN, AND W, S. CALUCCI. Modulation of cardiac autonomic activity during and immediately after exercise. Am. J. Physiol. 256 (Heart Circ. Physiol. 25): H132-H141, 3. BAINTON, C. R. Effect of speed vs.
Auf die Sinusarrhythmie
5.
6.
7.
8.
9.
10.
11. 12.
13.
TONE des Herzens. Acta Biol. Med. Ger. 34:
Statistical Essays Vol. II. Haemastaticks. London: Innings and Manby, 1733. L. E., JR., S. T. CARVER, AND A. PLAKUN. Slow heart 15. HINKLE, rates and increased risk of cardiac death in middle-aged man. Arch. Intern. Med. 219: 732-748,1972. 16. HIRSCH, J. A., AND B. BISHOP. Respiratory sinus arrhythmia in humans: how breathing pattern modulates heart rate. Am. J. Physiol. 241 (Heart Circ. Physiol. 10): H620-H629, 1981. 17. KATONA, P., AND F. JIH. Respiratory sinus arrhythmia: noninvasive measure of parasympathetic cardiac control. J. Appl. Physiol. 39: 801-805,1975. 18. KLEIGER, R. E.,
J. P. MILLER, J. T. BIGGER, JR., AND A. J. Moss. Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am. J. Curdiol. 59:
256-262,1987. 19. MCCABE, P.
1989.
J. T., JR., R. E. KLEIGER, J. L. FLEISS, L. M. ROLNITZKY, R. C. STEINMAN, AND J. P. MILLER. Components of heart rate variability measured during healing of acute myocardial infarction. Am. J. Curdiol. 61: 208-215, 1988. BILLMAN, G. E., AND R. S. HOSKINS. Time-series analysis of heart rate variability during submaximal exercise: evidence for reduced cardiac vagal tone in animals susceptible to ventricular fibrillation. Circulation 80: 146-157, 1989. BILLMAN, G. E., P. J. SCHWARTZ, AND H. L. STONE. Baroreceptor reflex control of heart rate: a predictor of sudden cardiac death. Circulation 66: 874-880, 1982. CHESS, G. F., R. M, K. TAM, AND F. R. CALARESU. Influence of cardiac neural inputs on rhythmic variations of heart period in the cat. Am. J. Physiol. 228: 775-780, 1975. CLIFFORD, P. S., J. T. LITZOW, J. H. VON COLDITZ, AND R. L. COON. Effect of chronic pulmonary denervation on ventilatory responses to exercise. J. Appl. Physiol. 61: 603-610, 1986. CORR, P. B., K. A, YAMADA, AND F. X. WITKOWSKI. Mechanisms controlling cardiac autonomic function and their relationships to arrhythomogenesis. In: The Heart and Cardiovascular System, edited by H. A. Fozzard, E. Haber, R. B. Jennings, A. M. Katz, and H. E. Morgan. New York: Raven, 1986, p. 1343-1404. DELLINGER, J. A., H. T. JANSEN, D. J. ZABER, AND S. G. BERNBAUM. Vagal tone monitoring: a potential indicator of anticholinesterase exposure in Macaca mulatta. Toxicology 49: 227-235,1988. ECKBERG, D. Human sinus arrhythmia as an index of vagal cardiac outflow. J. Appl. Physiol. 54: 961-966,1975. ECKBERG, D. L., M, DRABINSKY, AND E. BRAUNWALD. Defective cardiac parasympathetic control in patients with coronary artery disease. N. Engl. J. Med. 285: 877-883,197l. ECKHOLDT, K., AND E. SCHUBERT. Zum Enflu B der Atemtiefe
VAGAL
767-771,1973. 14. HALES, S.
grade and shivering on ventilation in dogs during active exercise. J. Appl. Physiol. 33: 778-787,
1973. 4. BIGGER,
CARDIAC
20.
21.
22.
23.
M., B. G. YONGUE, P. K. ACKLES, AND S. W. PORGES. Changes in heart period, heart-period variability and spectral analysis estimate of respiratory sinus arrhythmia in response to pharmacological manipulations of the baroreceptor reflex in cats. Psychophysiology 22: 195-203,1985. PAGANI, M., F. LOMBARD, S. GUZZETTI, 0. RUINOLDI, R. FURLAN, P. PIZZINELLI, G. SANDRONE, G. ANALFALTO, S. DELL’ORTO, R. PICCIALUGA, M. TURIEL, G. BARELLI, S. CERULTI, AND A. MALLIANI. Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circ. Res. 59: 178-193, 1986. POMERANZ, B., R. J. B. MACAULAY, M. A. CAUDILL, I. KUTZ, D. ADAM, D. GORDON, K. M. KILBORN, A. C. BARGER, D. C. SHANNON, R. J. COHEN, AND H. BINSON. Assessment of autonomic function in human heart rate spectral analysis. Am. J. Physiol. 249 (Heart Circ. Physiol. 18): H867-H875,1985. PORGES, S. W. (Inventor). Method and Apparatus for Evaluating Rhythmic Oscillations in Periodic Response Systems. US Patent 4,510,944. 16 Apr. 1985. PORGES, S. W. Respiratory sinus arrhythmia: physiological basis, quantitative methods and clinical implications. In: Cardiac, Respiratory and Cardiosomatic Psychophysiology, edited by P. Grossman, K. Janssen, and D. Vaith. New York: Plenum, 1986, p. lOl115. ROBINSON, B. WALD. Control
F., S. E. EPSTEIN, G. D. BEISER, AND E. BRAUNof heart rate by the autonomous nervous system: studies in man on the interrelationships between baroreceptor mechanisms and exercise. Circ. Res. 19: 400-411, 1966. 25. ROWELL, L. B. Human Circulation Regulation During Physical Stress. New York: Oxford University Press, 1986, p. 213-256. 26. STAVERT, D. M., P. REISCHI, AND B. J. O’LOUGHLIN. A respiratory mask for resting and exercising dogs. J. Appt. Physiol. 52: 500-504, 24.
1982.
27. STONE, H. L. Cardiac function and exercise training in conscious dogs. J. Appl. Physiol. 42: 824-832,1977. 28. WINER, B. J. Statistical Principles in Experimental Design. New York: McGraw-Hill, 1971, p. 514-603.
Downloaded from www.physiology.org/journal/ajpheart by ${individualUser.givenNames} ${individualUser.surname} (129.186.138.035) on January 17, 2019.
