INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE SYMPATHETIC NERVOUS system contributes importantly to regulation of vascular tone in the systemic circulation (11, 23, 38) and in isolated vascular beds (6, 8, 24). In situations such as exercise or environmental stress, sympathetic regulation of vascular tone is highly activated, and a higher degree of variability can be expected in total peripheral resistance than during basal conditions. This higher variability in total peripheral resistance is reflected in a higher variability of arterial blood pressure (4, 5, 7, 12, 22, 25). In addition to the sympathetic nervous system, hormones like angiotensin II (14) or vasopressin (16, 37), as well as other factors such as thermoregulation (34), can also modulate vascular tone. In the last decade, power spectrum analysis of arterial blood pressure has been used to separate modulating influences of vascular tone caused by the sympathetic nervous system from modulating influences caused by other factors (26, 27, 29). The concept has evolved that each regulatory system that impinges on vascular tone is reflected in a specific frequency band of the arterial blood pressure power spectrum, depending on its frequency response characteristics. As an example, the effects of thermoregulation are slow in onset and offset and, therefore, can only impinge on very low frequency components of the blood pressure power spectrum. On the other hand, sympathetic outflow to the vasculature is generally believed to alter vascular tone more rapidly than thermoregulatory influences and, as a result, is reflected in a higher frequency band.

To assign a specific frequency band of the arterial blood pressure power spectrum to sympathetic modulation of vasomotor tone, the frequency response characteristics of sympathetic transmission to the vasculature must be delineated. In previous studies, we investigated mesenteric vascular resistance responses to sympathetic stimulations at increasing stimulation frequencies in conscious rats (32, 35). The greatest responses were found at stimulation frequencies between 0.2 and 0.5 Hz. In contrast, in humans, the frequency range that is generally believed to reflect sympathetic modulation of vasomotor tone is centered around 0.1 Hz (26, 27, 29, 30). However, evidence for this frequency range in humans comes from indirect experiments in which sympathetic drive was enhanced by applying various stressors (26, 28) or reduced pharmacologically (10). There is currently no evidence from direct sympathetic nerve stimulation experiments that supports this frequency range in humans.

Therefore, the purpose of this study was twofold: 1) we addressed the question of whether species differences between humans and rats exist in the frequency range of sympathetic modulation of vasomotor tone, and 2) we designed a unique set of experiments to characterize the frequency range in which sympathetic modulation of vasomotor tone operates in humans using direct sympathetic nerve stimulations. As an experimental approach, we used microneurographic techniques to electrically stimulate efferent sympathetic skin fibers of the median nerve with increasing stimulation frequencies and simultaneously recorded skin blood flow in the palm region of the hand innervated by these nerve fibers.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects. Twelve healthy, young (<40 yr old) male adults were included in this study. None was receiving any medication. All subjects gave written informed consent to participate in the study, and all experiments were approved by an Institutional Review Board on Human Investigation.

Experimental setup and hemodynamic recordings. During the experimental protocol, subjects were in the supine position. Electrocardiogram (ECG) electrodes were attached according to the standard bipolar limb leads, a strain gauge pneumotachometer probe was placed around the chest, and a Finapres blood pressure sensor (Ohmeda 2350 Finapres Monitor, Ohmeda, Englewood, CO) was placed on the third finger of the right hand. The median nerve of the left arm was then located by percutaneous nerve stimulation. Once the proper location of the nerve in the cubital area was identified, a microneurographic tungsten microelectrode (200 µm diameter shaft, 1-5 µm uninsulated tip) was inserted into the nerve. A subcutaneous reference electrode was placed 1-3 cm apart from the stimulation electrode. During continuous electrical stimulation (Grass S48 stimulator, Grass Instruments, Quincy, MA), the stimulation electrode was slowly moved upward or downward until skin fibers of the median nerve were found. The correct location of the stimulation electrode in skin fibers of the left median nerve was verified by predefined criteria: 1) absence of muscle contractions during electrical stimulation; 2) sensations of paresthesias ("pins and needles") in a skin area on the palm of the left hand reported by the subjects during electrical stimulation; 3) increases in neural activity recorded from the impaled nerve (Nerve Traffic Analyzer, Dept. of Bioengineering, University of Iowa, Iowa City, IA) during tactile stimulation of the skin area at the palm of the left hand; and 4) increases in neural activity during application of an arousal stimulus. All nerve stimulations were performed by an experienced investigator (E. A. Anderson) who has done >1,500 microneurographic recordings in former studies (19, 20). Recording of skin blood flow (laser-Doppler flux) was performed as described by Haynes et al. (15). Therefore, a laser-Doppler flow probe, connected to a two-channel laser-Doppler device (Laser FLO Blood Perfusion Monitor BPM 403A, Vasamedics, St. Paul, MN), was attached to the surface of the palm of the left hand at the location where the subject felt the paresthesias during nerve stimulation. The second laser-Doppler flow probe was positioned at the same location on the contralateral hand. The five analog signals obtained from the various measurement equipment (i.e., the ECG signal, the respiration signal, the pulsatile blood pressure signal, and the two laser-Doppler skin blood flow signals) were recorded by a computerized recording and analyzing system (MacLab/8s, ADInstruments, Castle Hill, Australia) with a sampling rate of 100 Hz for each channel. This sampling rate allowed detection of changes in heart rate (which was calculated off-line from the ECG signal) of <1 beats/min. To avoid aliasing effects in the laser-Doppler signals, an analog low-pass filter (that was provided by the laser-Doppler device) with a cut-off frequency of 10 Hz was applied before the signals were analog to digital converted.

Experimental protocol. The experiments were performed in a climatized room with constant room temperature (20°C) and humidity (70%). Care was taken to prevent auditory and other environmental disturbances during the experimental protocol. Subjects were placed in a supine position and instrumented with various sensors, electrodes, and laser-Doppler flow probes. Once the stimulation electrode was positioned and the proper location in skin fibers of the left median nerve was verified, the voltage used for the stimulation was determined by gradually increasing the stimulation voltage from 0.0 V to a level where the stimulation was just perceived by the subjects (usually in the range between 0.8 and 1.2 V). Electrical nerve stimulations at these low voltages were barely perceived by the subjects and not associated with pain or discomfort. After a baseline recording of the ECG, blood pressure, respiration, and laser-Doppler signal, skin blood flow in the ipsilateral and contralateral hand was obtained for 22 min without nerve stimulation; subsequent electrical stimulations were applied to the skin fibers of the left median nerve with seven increasing stimulation frequencies. Each stimulation period was preceded by a baseline recording without stimulation that served as the direct control for the following stimulation period. After each stimulation period, the cuff of the Finapres device was deflated to avoid an upward drift of the Finapres blood pressure signal that can occur if the cuff is placed on the finger for a prolonged time period (31). Stimulations were applied as bursts of six rectangular impulses of 0.8-1.2 V, 10-ms duration for each impulse, and an impulse frequency of 20 Hz (time constant = 50 ms). Therefore, the duration of each burst was 260 ms. Such bursts were applied to the skin fibers of the left median nerve with stimulation frequencies of 0.01, 0.025, 0.05, 0.075, 0.1, 0.25, and 0.5 Hz. To obtain a reasonable number of stimulations at each stimulation frequency, the recording time for each recording period was adjusted for different stimulation frequencies and ranged from 22 min for a stimulation frequency of 0.01 Hz to 3 min for a stimulation frequency of 0.5 Hz. The duration of the experimental protocol for each subject was ~2 h.

Data analysis. For each of the 12 subjects, 15 time series (7 stimulation frequencies plus 7 baseline recordings preceding each stimulation period plus an additional baseline recording at the end of the experimental protocol) were obtained for each of the 5 different signals (ECG, blood pressure, respiration, and 2 laser-Doppler signals). The ECG time series were used to generate equally spaced time series (100-Hz sampling rate) for the heart rate signals by an algorithm based on that described by Berger et al. (1). The database of this study consisted of 900 time series (12 subjects, 15 recording periods, 5 signals) each sampled at 100 Hz. All of these 900 time series were inspected manually on the computer screen, and artifacts were replaced by interpolations based on the preceding and trailing values. Typically, <1% of the data points in the time series were replaced by interpolated values. After artifact removal, mean values were generated from the 100-Hz time series of blood pressure, heart rate, and the two laser-Doppler flow signals. In addition, delta values were calculated by subtracting the mean values during stimulation from the mean values from the preceding baseline recordings. Power spectra were calculated from the 100-Hz time series using the fast Fourier transform (FFT) algorithm. The number of data points (sampled at 100 Hz) used for the FFT were (for ascending stimulation frequencies) 131,072, 65,536, 32,768, 32,768, 16,384, 16,384, and 16,384. Thus the longest time series had a duration of 22 min, whereas the shortest was 3 min long. This protocol ensured that at least 13 stimulations occurred during each stimulation period. From all spectra, the area under the curve at a frequency band centered around the respective stimulation frequency (0.01 ± 0.005, 0.025 ± 0.01, 0.05 ± 0.02, 0.075 ± 0.02, 0.1 ± 0.02, 0.25 ± 0.1, and 0.5 ± 0.1 Hz) was computed, and the differences between the area under the curve during the stimulation and the preceding baseline periods were calculated.

Statistical analysis. Mean values for blood pressure, heart rate, and laser-Doppler skin blood flows in the stimulated and nonstimulated hand obtained during stimulations and during the preceding baseline recordings were compared by paired Student's t-test. Mean values during the baseline recordings preceding the seven stimulation periods and after the last stimulation period at the end of the experiment were compared using the one-way analysis of variance for repeated measures. In case of statistical significance, post hoc Student's t-tests were performed to detect differences between individual baseline recordings. The area under the curve values centered around the respective stimulation frequencies in the spectra (i.e., spectral powers) were log transformed before testing because spectral powers are not normally distributed. The log-transformed powers during stimulation and during the preceding baseline recordings were then compared using the Student's t-test for paired observations.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Hemodynamic baseline characteristics. During the experimental protocol, the subjects lay quietly on their backs for ~2 h. Because this protocol could be associated with some degree of immobilization stress that is likely to increase in intensity during the time frame of the experiments, we investigated the time course of the hemodynamic parameters during the experimental protocol. The changes in blood pressure, heart rate, and skin blood flow in the ipsilateral and contralateral hand during the baseline recordings that preceded each stimulation period are presented in Fig. 1. Mean blood pressure increased steadily from 89 ± 3 mmHg at the beginning of the experiment to 97 ± 3 mmHg during the control recording at the end of the experimental protocol. Heart rate was 61 ± 3 beats/min before the first stimulation period and did not change significantly during the time frame of the protocol. Similarly, skin blood flow in the ipsilateral (stimulated) hand did not change during the successive baseline recordings. However, skin blood flow in the nonstimulated control hand decreased steadily during the protocol.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Time course of mean blood pressure (BP), heart rate (HR), and skin blood flow in ipslateral, stimulated (SBFstim, arbitrary units) and contralateral, nonstimulated hand (SBFcont, arbitrary units) during baseline recordings that preceded each stimulation period. bpm, Beats/min. * P < 0.05 vs. baseline recording at beginning of experimental protocol.

Hemodynamic effects of median nerve stimulation. Figure 2 demonstrates the changes in hemodynamic parameters because of median nerve stimulation for each of the seven stimulation frequencies. To correct for changes in baseline hemodynamic characteristics during the time frame of the experimental protocol, mean values for each stimulation period were compared with mean values obtained from a control recording that immediately preceded the respective stimulation period. Median nerve stimulation elicited only minor changes in blood pressure in the range of <3 mmHg. Similarly, heart rate was not changed by median nerve stimulation. Skin blood flow in the stimulated skin area was significantly reduced by median nerve stimulation at all stimulation frequencies exceeding 0.05 Hz, indicating that a sympathetic-mediated vasoconstriction occurred. In the contralateral, nonstimulated skin area of the right hand, median nerve stimulation reduced skin blood flow when the stimulation frequency was 0.25 Hz or higher. Thus it is likely that some degree of an arousal reaction occurred at the higher stimulation frequencies.

View larger version (28K): [in this window] [in a new window]   Fig. 2.   Changes () in mean blood pressure, heart rate, and skin blood flow in ipsilateral, stimulated and contralateral, nonstimulated hand by median nerve stimulation with increasing stimulation frequencies. * P < 0.05 mean values during stimulation vs. mean values during preceding baseline recording.

Frequency response of skin blood flow to sympathetic stimulation. An original recording of pulse pressure, heart rate, respiration, and skin blood flow in the ipsilateral, stimulated hand and in the contralateral, nonstimulated hand is depicted in Fig. 3. During this recording, the left median nerve was electrically stimulated with a stimulation frequency of 0.075 Hz, equivalent to a wavelength of 13.3 s. A strong periodic oscillation that corresponds to the stimulation frequency of 0.075 Hz is present in the skin blood flow signal of the ipsilateral skin area. The spontaneous fluctuations in the control recording of the contralateral skin area were not related to the nerve stimulation and may be caused by influences of thermoregulation, endocrine systems, or other factors.

View larger version (38K): [in this window] [in a new window]   Fig. 3.   Original recording of blood pressure, heart rate, respiration (Resp, arbitrary units), and skin blood flow in ipsilateral, stimulated and contralateral, nonstimulated hand during stimulation of left median nerve at a frequency of 0.075 Hz (wavelength = 13.3 s). Note strong oscillatory component in skin blood flow signal of ipsilateral, stimulated skin area.

View larger version (0K): [in this window] [in a new window]   Fig. 4.   Contour plots showing power spectra calculated from skin blood flow signals in contralateral (nonstimulated, A) and ipsilateral (stimulated, B) skin area of a representative subject. Median nerve was electrically stimulated with increasing stimulation frequencies (y-axes). Frequency axes (x-axes) are plotted on logarithmic scales, and power spectra are normalized such that largest power in each spectrum is 1 (shaded in black).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Changes in spectral power of skin blood flow in ipsilateral, stimulated and contralateral, nonstimulated skin area. Power was calculated as area under curve of power spectrum at a frequency band centered around respective stimulation frequency. Differences in power during stimulation and preceding baseline recording are presented. * P < 0.05 values during stimulation vs. values during preceding baseline recording.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

This study was designed to characterize the frequency range at which skin microcirculatory vasomotor tone can be modulated by the sympathetic nervous system in humans. Using laser-Doppler flowmetry, we were able to demonstrate that skin vascular tone can be modulated by periodic sympathetic nerve stimulations in the frequency range between 0.05 and 0.1 Hz, indicating that peripheral sympathetic transmission in humans behaves as a low-pass filter with a cut-off frequency above 0.1 Hz. Indirect evidence from human studies in which arterial blood pressure power spectra were calculated while sympathetic tone was increased by various stressors (26, 28) or reduced by clonidine applications (10) suggested that sympathetic modulation of vasomotor tone is strongest in a frequency band centered around 0.1 Hz. In contrast to these indirect studies, our current study, showing that the largest responses in skin blood flow to sympathetic nerve stimulation were detected between 0.05 and 0.1 Hz, provides the first direct evidence that a low-frequency peak (centered around 0.1 Hz) in the arterial blood pressure power spectrum can reflect sympathetic modulation of vasomotor activity in humans.

In previous studies in conscious rats, we investigated mesenteric vascular resistance responses to direct sympathetic nerve stimulation (32) and hypothalamic paraventricular nucleus stimulation (35) at different stimulation frequencies. The greatest response in mesenteric vascular resistance was found between 0.2 and 0.5 Hz. The faster vascular responses to sympathetic stimulation in rats than in humans may be because of several factors. First, different vascular beds were investigated. It is possible that skin vessels cannot respond as fast to sympathetic stimulation as mesenteric vessels. In addition, muscle sympathetic nerve activity, which is modulated primarily through cardiovascular reflexes, could behave differently from skin sympathetic nerve activity. Second, the synaptic distance from the sympathetic varicosities to the ₁-adrenergic receptors in rats may be smaller than in humans. In addition, the density of the varicosities that secrete the neurotransmitter may be higher in the intestine of rats than in the skin of humans (2). In support of this possibility, it is known that mesenteric arteries of rats have a very dense sympathetic innervation (13, 17, 18). However, conclusive studies comparing densities of varicosities or synaptic cleft widths in different vascular regions and different species have not been performed.

Because baseline blood pressure increased and skin blood flow in the contralateral, nonstimulated hand decreased during the time frame of the experimental protocol (Fig. 1), it is reasonable to assume that a general sympathetic activation occurred that was strongest at the end of the experiment, when the higher stimulation frequencies were applied. Because all stimulation frequencies, including the highest stimulation frequency at 0.5 Hz, were not associated with pain, such a general sympathetic activation could have been most likely because of the prolonged experimental protocol that lasted for ~2 h. Such a sympathetic activation could lead to a vasoconstriction that may have limited the responsiveness to sympathetic stimulation at higher frequencies that were applied toward the end of the experimental protocol. However, no reduction in baseline skin blood flow was observed in the ipsilateral, stimulated hand, and the baseline values for heart rate did not increase during the experimental protocol. Therefore, one may argue that sympathetic tone did not increase during the time frame of the protocol. The increase in baseline blood pressure may be because of an upward drift in the Finapres signal as suggested by Ristuccia et al. (31). To avoid the upward drift, we deflated the Finapres cuff after each stimulation period; however, we did not ask the subjects to exercise the finger for a few minutes as suggested by Ristuccia et al. (31) because of the risk of displacement of the laser-Doppler flow probe that was also attached to the same hand. Whether there was a sympathetic activation or not, there is still the discrepancy in the time course of baseline skin blood flow in the ipsilateral and contralateral hands (Fig. 1). It is possible that a vasoconstrictor system was activated during the experiment that affected both hands. In the stimulated hand, we most likely stimulated both efferent and afferent skin fibers of the median nerve. Afferent nerve stimulation may have triggered a reflex-mediated vasodilation in the stimulated skin area that antagonized the vasoconstriction. As a result, no changes in baseline skin blood flow were seen in the stimulated skin area, whereas skin blood flow in the nonstimulated skin area decreased. Thus we cannot completely rule out that the frequency response characteristic, as described in this study, was influenced by changing activities of local vasoconstrictor and vasodilator systems. However, the balance of those systems was unchanged in the stimulated hand, since baseline skin blood flow did not change during the time frame of the protocol.

Our results demonstrate that the linear relationship between the median nerve stimulation and the change in skin blood flow is strongest in the range of 0.075-0.10 Hz. However, nonlinear skin blood flow responses to median nerve stimulation may also play a role. For example, there is nonuniform high power in the blood flow spectra of the stimulated skin area at frequencies below the frequency of stimulation (Fig. 4B). Therefore, the question arises whether this low-frequency power reflects nonlinear resonses to the nerve stimulation. However, comparison of the skin blood flow power spectra of the stimulated skin area (Fig. 4B) with the power spectra from the nonstimulated skin area (Fig. 4A) reveals that both power spectra contain the same nonuniform high power at the same frequencies. Therefore, it is reasonable to assume that this nonuniform power at frequencies below the frequency of stimulation is not the reflection of a nonlinear response to median nerve stimulation. It may be caused by the modulatory influences of thermoregulation, endocrine systems, or other factors on skin blood flow.

In contrast to other studies employing microneurography techniques, we investigated the vascular responses to electrical nerve stimulations rather than recording sympathetic neural activity. Although microneurographic recordings provide information on spontaneous sympathetic discharge of a peripheral nerve, stimulations with simultaneous recordings of the effector responses provide information on sympathetic-mediated end-organ function. Sympathetic nerve stimulation in humans as performed in the present study is a novel technique that may provide important insights into the pathophysiology of a variety of diseases. Examples are diabetic autonomic neuropathy (which we have previously demonstrated can be detected using microneurographic recordings; Refs. 19, 20), a disease in which norepinephrine content in the arteries is reduced while the vascular responsiveness to adrenergic stimuli is enhanced (9, 36), or hepatic cirrhosis, which is often characterized by a decreased total peripheral vascular resistance despite an increased sympathetic activity due to a reduction in -adrenergic vascular responsiveness (3). In addition, it would also be interesting to study the vascular responses to sympathetic nerve stimulation during the physiological process of aging, since arterial -adrenergic responsiveness is assumed to be decreased while sympathetic nervous system activity may be increased in some older individuals (21, 33, 34). During such conditions, the magnitude of the response or the frequency characteristics of the response can be altered. The enhanced vascular responsiveness to sympathetic stimuli in patients with diabetic autonomic neuropathy is assumed to be primarily determined by a decreased neuronal catecholamine uptake (9, 36). Thus the norepinephrine molecules remain inside the synaptic cleft for a longer time period. This in turn could slow down the offset of the vasoconstrictor response after sympathetic nerve stimulation and, therefore, could cause a frequency-response shift toward lower frequencies. However, these hypothetical mechanisms need to be supported by further investigations.

In conclusion, we have provided direct evidence that sympathetic modulation of vasomotor tone in humans is most efficient in the frequency band between 0.075 and 0.1 Hz. This finding correlates well with studies providing indirect evidence that low-frequency oscillations of arterial blood pressure with a wavelength of 10 s (0.1 Hz) can reflect sympathetic modulation of vasomotor activity. Finally, we have introduced a novel method to study sympathetic responsiveness in humans that may be useful to investigate a variety of physiological and pathophysiological conditions.