INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE MICROCIRCULATORY network of the skin, like other body tissues, continuously exhibits rhythmic changes in diameter and flow (7, 30). The fundamental mechanism of these fluctuations is not yet clarified, particularly for the skin microvessels. Some researchers hypothesized that skin blood flow is exclusively under local control (6, 22), whereas others have suggested both general (central) adjustments and baroreflex-induced changes (1, 12, 13, 21, 25). Part of this discrepancy is probably due to the lack of methods to observe and analyze the spontaneous fluctuations in vascular tone and to compare them with other signals. Laser-Doppler flowmetry (LDF) is an accepted and validated method for noninvasive measurement of changes in cutaneous blood flow (3, 14). Spectral analysis techniques can characterize different cardiovascular, neural, and respiratory fluctuations (4, 16, 19, 24). We have previously found in the LDF signal (5) the presence of the so-called low-frequency (LF; ~0.1 Hz) fluctuations, considered a marker of sympathetic activity in the heart and blood pressure (16). These LF fluctuations in the skin circulation were modified by changes in sympathetic tone (posture and local anesthetics; Refs. 4, 5, 19), suggesting that oscillations in microvascular blood flow may therefore be controlled by both central and local mechanisms.

In the present study, we therefore assessed whether spontaneous fluctuations in the skin microcirculation, obtained by LDF, are related to central control mechanisms, by evaluating 1) their synchronism in different body areas (synchronism should be evidence of a central, rather than a local mechanism); 2) their relationship with spontaneous fluctuations observed in signals known to be related to general hemodynamics and/or autonomic activity, such as the blood pressure, R-R interval, respiration, and muscle sympathetic nerve activity (MSNA; absence of synchronization with rhythms of known autonomic origin would be an argument against an autonomic origin of these skin circulation rhythms; 3) their ability to respond to arterial baroreflex stimulations at different frequencies of stimulation, since the autonomic modulation of the cardiovascular system appears to be frequency dependent (2, 18, 23) (for this last purpose, we used a "pure" autonomic stimulus, namely, neck suction, i.e., modifications in cardiovascular system without hemodynamic changes other than reflex; Ref. 2).

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects

Data Collection

Skin blood flow. Skin arteriolar blood flow was measured by two identical laser-Doppler flowmeters (Perimed model PF2b, Sweden, powered by 1-mW helium-neon laser tubes, supplying a 632.8-nm red light) with identical skin probes, on two corresponding volar surfaces of right and left index fingers. Details of the LDF methodology and validation techniques have previously been published (3).

Muscle nerve recording. MSNA was evaluated by conventional microneurography (24). A tungsten microelectrode (200-µm shaft diam, 1- to 5-µm uninsulated tip; University of Iowa) was inserted in the peroneal nerve and positioned in the sympathetic fibers. An identical reference electrode was inserted subcutaneously 1-2 cm from the recording electrode. Electrodes were connected to a preamplifier (gain 1,000) and amplifier (variable gain 10-50). Neural activity was fed through a band-pass filter (band width 700-2,000 Hz) and then a resistance-capacitance integrating network (time constant 0.1 s) to obtain a mean voltage neurogram. Assessment of the correct position in the fibers supplying the muscle was made by observing typical heart period-related spikes, which increased after simple maneuvers increasing sympathetic activity (mental arithmetic, Valsalva maneuver, and apnea).

Baroceptor stimulation. Neck suction was applied by means of a molded lead collar connected with a vacuum cleaner whose power was modulated by a second computer (Apple II) equipped with a 12-bit digital-to-analog board through a phase-control power unit (2). This computer generated a special sinusoidal function, which remained at the frequency of 0.02 Hz for 2 min, then continuously increased up to 0.20 Hz (so-called "chirp" function) over 4 min, and finally remained at 0.20 Hz for 2 more min. The software generated a correction function to compensate for frequency-dependent variations in the vacuum pressure; as a consequence, the pressure within the neck oscillated steadily from 0 to 30 mmHg peak to peak regardless of frequency. The pressure within the neck collar was continuously recorded (Viggo-Spectramed Statham P 23d pressure transducer, Bilthoven, The Netherlands).

Protocol

Data Analysis

Univariate spectral analysis. Power spectral analysis was applied to all signals recorded at baseline, using an autoregressive model, as previously described (19). Spectral components were obtained by a decomposition method (2, 16, 19) to measure the area below each spectral peak and to obtain the distribution of low-, high-, and very low frequency components (LF, HF, and VLF, respectively). The power spectra of all signals except respiration show at least two separate peaks (2, 16, 19): the higher frequency peak is similar in both shape and central frequency to that of respiration. Although, in the R-R interval sequence, this peak seems to reflect mostly the parasympathetic efferent activity, in the circulatory signals, it is interpreted as a mainly mechanical consequence of respiration induced increases in venous return (9, 16, 19). This respiratory component (HF) was identified by its coincidence with the peak of the respiratory spectrum, which served as a reference. The LF component (0.03-0.15 Hz), with usual peak at ~0.10 Hz in the R-R interval, has been suggested to reflect changes in sympathetic tone (2, 9, 16), particularly in the blood pressure, but only when slow breaths are absent. Finally, the oscillatory components between 0.001 and 0.03 Hz (designated VLF), whose origin is still unknown, were also measured.

Bivariate spectral analysis (coherence and phase analysis). The fact that two signals could oscillate at the same frequency and that oscillations from such signals undergo similar changes under different experimental settings suggests a common control mechanism. Nevertheless, oscillations may have the same frequency without being linked to each other in a stable way (as would be the case if they were due to the same mechanism). To verify the presence of a stable relationship between the oscillations of pairs of signals, we performed a coherence analysis (by autoregressive algorithms; Ref. 2) between the various signals obtained at baseline. The "squared coherence" function determines whether two signals oscillating at the same frequency maintain a stable phase relationship. This function spans from 0 (i.e., no association) to 1 (i.e., maximal association). If a significant (>0.5) coherence is found, then the two signals have a stable phase relationship for a given frequency of oscillation and hence can be considered as synchronized with each other. In this case, it is then possible to calculate the phase delay (expressed in radians) between pairs of coherent oscillations.

Time-varying spectral analysis. The recordings obtained during chirp sinusoidal neck suction could not be evaluated by the previous spectral methods, since the stimulation and response signals were continuously changing over time. On the other hand, recordings of 2 min, for example, at each neck suction frequency would have been too time consuming considering this complex experimental protocol and not well tolerated by the subject. Unlike a previous study dealing simply with the R-R interval and blood pressure fluctuations (2), whose characteristic predominance in 0.1 Hz and respiratory rhythms is known, it was necessary when dealing with the microcirculation to obtain the widest possible range of stimulation frequencies, particularly since previous reports (20) indicated that the natural resonance frequency of a microvessel is very low, i.e., ~0.03 Hz. To solve this problem, we therefore applied a function capable of performing a stimulation at every frequency; this type of signals can now be analyzed by a new family of spectral methods, allowing computation of the instantaneous spectra. For this purpose, we used the Wigner-Ville transform, which has recently been used with success by different groups (18, 27) in comparable experimental or clinical conditions. The algorithms used have been extensively described and tested (17, 27), and our implementation of the Wigner-Ville transform was written and validated according to previously described methods (27). Finally, to assess the dependence on neck suction oscillations in the various signals and at each frequency, we calculated the cross-spectrum of each signal vs. the neck suction. This cross-spectrum was obtained by multiplication of each signal spectrum by the spectrum of the neck suction, with the neck suction spectrum maximal peak normalized to a value of 1. As a consequence, the resulting spectra indicated to what extent and at what frequency each signal was dependent on arterial baroceptor stimulation.

Statistical Analysis

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Synchronism of Skin Microcirculatory Fluctuations Obtained in Two Index Fingers

Visual inspection of time series of signals. The fluctuations observed in two similar skin areas (left vs. right index finger) were consistently synchronous (Fig. 1).

View larger version (46K): [in this window] [in a new window]   Fig. 1.   Time series of respiration, R-R interval, systolic and diastolic blood pressure (SBP and DBP, respectively), left and right index finger laser-Doppler flowmetry (LDF), and muscle sympathetic nerve activity (MSNA, in arbitrary units) in 1 normal subject. X-axis is no. of heartbeats (R-R intervals). Note similarity in 2 LDF recordings. All signals, except respiration, show synchronous presence of slow (in 0.1-Hz range) fluctuations, including MSNA bursts.

Power spectral analysis, univariate. The spectral analysis (Table 1) identified three kinds of oscillations in the skin microcirculation. The range of 0.1 Hz (LF; 0.081 ± 0.006 and 0.080 ± 0.08 Hz in left and right side, respectively) was invariably present in all subjects and accounted for the greatest percentage of overall variability (61.6% for right side, 57.7% for left side). A consistent fluctuation synchronous with respiration (HF) was also present in all tracings but was only 15.2 (right side)-18.2% (left side) of total variability compared with that observed at lower frequencies. Finally, a third fluctuation at even lower frequency (VLF; ranging from 0.015 to 0.028 Hz) was observed in 5 of 10 subjects, but only in 3 was it detected on both the right and left sides. Figure 2 shows the spectra obtained from the same data as Fig. 1. The spectra obtained from the two LDF signals are very similar, confirming the visual impression of similar waves present in these two signals. Figure 1 shows clear fluctuations in the VLF range, but these did not appear in the LDF spectra, probably due to their variable period.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Autoregressive power spectra of data in Fig. 1. In each panel, continuous line is sum spectrum; dashed lines are individual peaks. All signals, except respiration, show a clear peak around 0.1 Hz (low frequency, LF). Notice also marked similarity of 2 LDF spectra.

Power spectral analysis, bivariate (coherence analysis). The coherences between the pairs of oscillations in these signals were significant in 10 of 10 subjects for the LF oscillations in the two LDF signals (right vs. left index fingers) and in 10 of 10 subjects for the respiratory frequency. The phase delay between left vs. right LDF signals was close to zero for both LF (0.026 ± 0.106 rad) and HF (0.078 ± 0.104 rad), hence indicating that 1) these two signals were synchronized at these two bands and 2) the corresponding waves occurred with no delay on one side with respect to the other, as if they were generated or propagated simultaneously. The example in Fig. 3 shows the coherence between the left and right LDF signals obtained from the data of Figs. 1 and 2.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Coherence and phase analysis of pairs of signals/spectra of Figs. 1 and 2. Thick line is coherence function (spanning from 0 to 1); thin line is phase function (spanning from to +). Significant coherence (i.e., thick line >0.5) allows evaluation of phase at this frequency. If phase is positive, 1st signal leads 2nd and vice versa. Note a significant coherence between all pairs of signals in LF region. Left and right LDF have ~0 phase delay in LF band, i.e., LF fluctuations occur simultaneously. LF fluctuations in blood pressures lag behind those in LDF. LF fluctuations in MSNA lead those in LDF.

Comparison of Microcirculatory Fluctuations with Other Signals

Visual inspection of time series of signals. Not only were the fluctuations in the two similar skin areas (left and right index fingers) similar to each other, but this similarity was also seen in other signals, such as systolic and diastolic blood pressure, MSNA, and R-R interval; in addition, fluctuations synchronous with respiration could be observed in all signals.

Power spectral analysis, univariate. Table 1 reports the results of power spectral analysis of all the signals recorded. The visual similarity of the fluctuations in the various signals is confirmed by the presence of LF and HF waves in all signals, although the relative proportion of LF and HF was different in each signal; VLF components were more evident in the LDF than in any other signal. The LF power in the LDF signals correlated with the LF power in the MSNA (r = 0.845, P < 0.01) and also with the HF power in the MSNA (r = 0.808, P < 0.01) but not with the LF power in the systolic or diastolic blood pressure and R-R interval. Figure 2 shows the spectra obtained from the signals of Fig. 1. The 0.1-Hz fluctuations are well represented in all signals, including the skin microcirculation and the MSNA. Respiratory fluctuations are also evident in most of the signal spectra. However, their proportions decrease from the R-R interval with respect to the systolic pressure, diastolic pressure, and skin blood flow. These fluctuations, though present, are not apparent in the spectra of Fig. 2 because of their small proportion.

Power spectral analysis, bivariate (LDF vs. other signals). In most subjects, the coherence function was significant in the LF and HF regions when the LDF signal was compared with other signals (Fig. 3).

View larger version (11K): [in this window] [in a new window]   Fig. 4.   Schematic diagrams of phase relationships between microvascular (by LDF) SBP, DBP, respiration, and MSNA in 0.1-Hz (LF) region (A) and in respiratory (high frequency, HF) region (B). Taking LDF as reference (0 rad), delay of same fluctuation for each signal is reported on circumference (equivalent to a complete fluctuation period, i.e., 2) at a distance correspondent to phase delay with respect to LDF. If a signal is positioned clockwise with respect to LDF, its fluctuations lag behind those of LDF. For example, since phase lag between LDF vs. DBP was +0.87 rad in LF region, DBP is positioned clockwise with respect to LDF. These diagrams also allow to establish phase relationship between any other pair of signal in LF and HF regions. In LF region LDF fluctuations precede those of blood pressure and R-R interval but lag behind those in MSNA, suggesting that LF fluctuations are primarily generated by sympathetic activity in microvasculature. In respiratory region (B), fluctuations in blood pressure lag behind those in LDF, suggesting that these might be passively transmitted from respiration to larger arteries and then to microvasculature and finally to MSNA. Respiration precedes all fluctuations and maintains a nearly phase opposition with R-R interval in respiratory frequency range of our subjects (~0.25 Hz, Table 1).

Sympathectomized Subjects

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Microvascular fluctuations recorded in a sympathectomized subject. Notice different pattern in 2 limbs, different spectra, and lack of coherence between left and right side.

The other two patients who underwent sympathectomy 10 mo and 1 year earlier showed some degree of synchronization in the microvascular fluctuations of both sides and different fingers. However, they also reported signs of partial reinnervation, such as reappearance of sweating in response to psychological stimulation. As a consequence, their results were between those of the first subject and those of the normal subjects.

Changes in Microcirculatory Fluctuations Induced by Frequency-Dependent Arterial Baroreflex Modulation

Although all signals showed their maximal response in the LF region, the frequency of maximal response was significantly (P < 0.01) lower for the LDF (0.058 and 0.055 Hz on left and right sides, respectively) and blood pressures (0.059 and 0.057 Hz for systolic and diastolic, respectively) than for MSNA and R-R intervals (0.077 and 0.088 Hz, respectively). However, the maximal response occurred at a similar frequency in the four "vascular" signals (left and right LDF, systolic and diastolic blood pressures; NS). In addition, a secondary peak of response was found (Table 2) at similar frequencies in the two LDF signals (0.026 and 0.023 Hz on left and right sides, respectively; NS) and at a high frequency in the MSNA signal (0.193 Hz).

As a consequence, with respect to their ability to respond to the arterial baroceptor modulation, all four vascular signals behaved as a low-pass system, the R-R interval was similar to an all-pass system, and the MSNA was closer to a high-pass system.

Figure 6 shows an example of the data obtained. It is clearly shown that each signal responded with its own characteristics. Although the R-R interval was able to follow the stimulation at almost every frequency, the blood pressure response decreased just above 0.1 Hz and was almost unresponsive in the HF band. Similarly, the microvessels responded well at the lowest frequencies of stimulation but poorly above the LF region. Nevertheless, it is evident from Fig. 6 that all the cardiovascular signals exhibit their maximal response in the 0.1-Hz region. We noticed that the blood pressure and, to a greater extent, the microvessels maintained a slow rhythm independent of that imposed by the baroceptor stimulation. This is particularly evident when neck suction was oscillating at frequencies >0.1 Hz, where the only fluctuations left were those at LF or VLF, suggesting that factors independent from the baroceptor stimulation (due to local factors or to a central oscillator or to a combination of both) are also operating at the same time (11).

View larger version (86K): [in this window] [in a new window]   Fig. 6.   Effect of chirp neck suction (i.e., continuously increasing frequency from 0.02 to 0.2 Hz of baroceptors stimulus) on cardiovascular, respiratory, and MSNA fluctuations in 1 subject. During these recordings, respiration was maintained at 0.25 Hz to avoid random respiratory frequencies superimposed on neck stimulus. For each signal, 2-dimensional graph shows time series (at base), whereas 3-dimensional graph shows time-varying spectral analysis (x-axis is time series, as in 2-dimensional graph; y-axis is power density; z-axis is frequency). Although all signals are modified by neck suction, with different sensitivity at different frequencies of stimulation, maximal response is invariably in LF region. Although R-R interval continues to be influenced well above 0.1-Hz region, blood pressure show maximal sensitivity at 0.1 Hz and is almost insensitive to neck suction at higher frequencies. MSNA is sensitive mostly in 0.1-Hz region and above it. LDF is sensitive to neck suction in regions of 0.03 and 0.1 Hz. Note bifurcation of these 2 components in LDF spectrum when neck suction frequency is above 0.03-Hz range. SBP and DBP are also sensitive up to LF region. Above 0.1 Hz, these signals show no further sensitivity to neck suction. Notice that peaks in spectrum of respiration are distributed in a region different from that of neck suction, confirming that observed patterns were not influenced by respiration. At lower frequencies (particularly in 0.02- to 0.03-Hz region), neck suction induces a clear amplitude and width modulation of 0.1-Hz rhythm in MSNA. This is in turn transformed into a continuous slow wave in blood pressure and LDF signal (example of nonlinear interaction).

The cross-spectra of each signal vs. the neck suction showed only the net response of each signal to the arterial baroceptor stimulation. Figure 7 is an example of the cross-spectra obtained from the same data of Fig. 6, showing the low-pass behavior of the LDF and systolic and diastolic blood pressures, the all-pass behavior of the R-R interval, and the high-pass behavior of the MSNA. The pattern of response of the previous figure resulted here more evident.

View larger version (82K): [in this window] [in a new window]   Fig. 7.   Bivariate time-variant spectral analysis (Wigner-Ville cross-spectra) shows only net effects of neck suction on various signals at each frequency (same data as in Fig. 6). For each signal, linear dependence on neck suction stimulation at each frequency appears more clearly than in Fig. 6. Sensitivity to baroceptor stimulation is all-pass type for R-R interval, high-pass type for MSNA, and low-pass type for LDF, SBP, and DBP.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Although variations in peripheral blood flow have been well described for over half a century (7), the problem of the origin and significance of these fluctuations remains debated (see introduction). It is now accepted that the LDF signal is proportional (within same subject) to the blood flow in the most superficial (in range of 1 mm) layers of the skin (3). The arterioles and/or the arteriovenous anastomoses, i.e., structures rich in autonomic innervation, are the main determinants of the microvascular fluctuations (30). To assess the influence of central control mechanisms on the fluctuations in skin microvessels, we evaluated the synchronism of LDF fluctuations in different body areas, their relationship with spontaneous fluctuations observed in signals known to be related to general hemodynamics and/or autonomic activity, such as the blood pressure, R-R interval, respiration, and MSNA, and their ability to respond to arterial baroreflex stimulations at varying frequencies, taking into account the possibility that the autonomic modulation of the cardiovascular system could be frequency dependent (2, 18, 23).

Synchronism of LDF Fluctuations in Different Body Areas

Nonrespiratory (LF) and respiratory (HF) fluctuations. These fluctuations are highly coherent with zero phase lag; they have similar absolute and relative (i.e., percentage of variance) power, which correlated in each subject, and similar frequency. This similarity in frequency can easily be understood for the respiratory components (since both left and right LDF HF originate from the activity of respiration) but was quite surprising for the LF, which are nonrespiratory by definition. If these fluctuations were of local origin, there is no reason why they should have the same frequency and amplitude and occur simultaneously in two different body areas.

VLF fluctuations. A direct correspondence between two LDF signals is not always evident on visual inspection of the signals, since other rather irregular fluctuations of lower frequencies (0.017-0.028 Hz, VLF) are superimposed on LF and HF fluctuations. We have observed such slower waves in 5 of 10 subjects, but only in 3 of 10 were they present in both right and left LDF signals. The relatively short period of analysis (only few cycles can be observed in 12 min), in addition to likely local influences, might explain the relatively poor synchronization in these fluctuations. Burton (7) suggested that the slowest fluctuations in skin blood flow were due to thermoregulatory influences. If this hypothesis were true, then the VLF should be present particularly in those signals directly related to the thermoregulatory adjustments, such as those in the skin blood flow. Because these fluctuations were in fact present in the LDF more than in any other signal that we have recorded and appeared to be locally mediated, we might speculate that they could indeed be related to thermoregulatory activity (among other factors). However, this statement should be matter of a specific investigation.

Relationship Between Spontaneous Fluctuations in LDF and Other Cardiovascular and Neural Signals

Nonrespiratory LF fluctuations. One of the main findings of the present study is that the LF fluctuations in the LDF signals are coherent with those present in most cardiovascular signals and the MSNA, indicating that all such fluctuations are related to each other (Fig. 3). The significant association between these waves allowed us to measure the phase lag between each signal. We found that LF waves first occur in the MSNA, then in the LDF signals, then in the blood pressure, and finally in the R-R interval (Fig. 4). Taken together, even with the limitations due to the lack of skin sympathetic activity signal, it seems that, under normal conditions, these LF originate in the microvasculature, in keeping with the well-known fact that most of the sympathetic modulation of blood vessel diameter occurs at arteriolar level (30). Consideration of the other signals in Fig. 4 also shows that LF blood pressure fluctuations precede those in the R-R interval by ~1 rad. This finding is in close agreement with similar results obtained by de Boer et al. (Ref. 9; they found a delay of 60-90° corresponding to 1-1.5 rad, similar to our finding) and with those of Saul et al. (23) for the same frequency. These findings cannot be due to a spurious delay from arterial blood pressure to the Finapres signal. Theoretical measures suggest a delay of ~0.3 s; preliminary data from our laboratory (unpublished data) indicate a similar delay. In support of the hypothesis of a peripheral origin of the LF, we also found a significant intersubject correlation between the power of the LF fluctuations in the LDF signal vs. those in the MSNA, whereas the correlation with the LF power in the other signals was not significant despite significant coherence. This suggests that there is a close relationship between the skin blood flow fluctuations in the LF band and the sympathetic neural activity directed to the limbs. Although we could record the muscle nerve activity in all subjects, we could never record the skin sympathetic activity for a long enough period to be analyzed. Previous reports (10, 15, 28, 29) on the relationship between skin sympathetic activity/skin blood flow vs. baroreflex or central command stimuli are conflicting. Those studies were based on inspection of short-term recordings, and no mathematical methods were applied to analyze in depth the relationship (if any) between the LF of the various signals. However, the presence of a correlation between the power of LDF LF fluctuations and the power of MSNA LF fluctuations further indicates that the LDF LF fluctuations are not simply of local origin and must have some relationship with the sympathetic efferent activity to the limbs.

Respiratory HF fluctuations. We found that the HF fluctuations (respiratory band) in the LDF signals are coherent with those present in most cardiovascular signals and in the MSNA (Fig. 3), although the presence of significant values was less frequent than for the LF fluctuations. Phase analysis (Fig. 4) revealed a different behavior with respect to the LF. After originating in the respiration, the HF fluctuations appear in the blood pressures, then in the LDF, then in the R-R interval, and finally in the MSNA. This suggests that the HF waves are transmitted "downstream" from the larger arteries to the microvessels (due to respiratory changes in venous return and stroke volume) and are also sensed by the baroceptors and transformed into R-R interval changes and respiratory modulation of MSNA. This hypothesis is also in agreement with previous findings of a transmission of respiratory sinus arrhythmia from blood pressure fluctuations (9, 23). In fact, consideration of the other signals in Fig. 4 shows that blood pressure HF fluctuations precede those in the R-R interval (by ~0.5 rad). The R-R interval is nearly in phase opposition with respiration at ~0.25 Hz. As for the LF, this finding is in agreement with similar results obtained by other authors for comparable frequencies (9, 23).

Response of LDF Fluctuations to Frequency-Dependent Arterial Baroreflex Stimulation

All cardiovascular signals and the MSNA have their maximal sensitivity to the arterial baroceptor stimulation at a frequency in the LF region. This frequency is not significantly different for the LDF and blood pressures, whereas it is slightly but significantly lower than the frequency of maximal response of R-R interval and MSNA (Table 2). This parallels the result of significant coherences in the LF region obtained in the comparison between the LDF and all other signals (Fig. 3) and indicates that perhaps the whole cardiovascular system has a peculiar attitude to resonate at this specific frequency. The tridimensional spectra relative to the time series of Figs. 6 and 7 and the data of Table 2 clearly show this concept. Despite each signal having its own pattern of response to neck suction, all signals showed their greatest response at a frequency in the LF region.

Additional Findings

Interaction between autonomic modulation and natural resonance frequency of microvessels. There is a maximal response of the LDF signal in the LF region, but, in addition, there is also a second peak in the VLF region at 0.023 and 0.026 Hz (right and left side, respectively; Table 2). This is close to the frequency at which Rosenbaum and Race (20) found that the resistance vessels have a resonant frequency (~0.03 Hz). This result suggests that the LDF LF fluctuations are most likely the result of an external modulation (as one provided by sympathetic neural activity at 0.1 Hz). However, slower stimulation of whatever origin (including neck suction) can easily produce LDF fluctuations in the VLF region at the natural resonance frequency the skin microvessels. This might explain the frequent occurrence of VLF fluctuations in the LDF more than in other signals.

Nonlinear interaction between faster MSNA fluctuations and slower vascular waves. Although the stimulation of the arterial baroceptors in the HF region was able to increase the MSNA at high frequencies, this did not result in an increase in the LDF HF oscillations, since the behavior of the LDF was clearly similar to a low-pass system. This cannot be attributed to inability of the LDF fluctuations to respond to the baroceptor stimulation, since this signal, similar to that for MSNA, was highly sensitive to the stimulation in the LF (and also in VLF) region (Table 2, Figs. 6 and 7). Similar considerations also apply to the blood pressure fluctuations, since these signals did not respond to stimulation of the arterial baroceptors in the HF region but only when the stimulations was limited to the LF region. Hence, for some reason, the faster (i.e., in HF region) components in the sympathetic nerve discharge cannot be "transferred" into oscillations at the same frequency in the microcirculation. One might hypothesize that such fluctuations either are simply ineffective in terms of vascular modulation or they might contribute to the generation of other fluctuations at lower frequencies (i.e., nonlinear interaction).

Limitations of Study

Conclusions