An association between vasomotion and oxygen extraction

Clare E. Thorn, Hayley Kyte, Dick W. Slaff, and Angela C. Shore Diabetes and Vascular Medicine, Institute of Biomedical and Clinical Sciences, Peninsula College of Medicine and Dentistry, University of Exeter, National Institute for Health Research Peninsula Clinical Research Facility, Exeter, United Kingdom; and Departments of Biomedical Engineering, Universities of Maastricht and Eindhoven, Maastricht and Eindhoven, the Netherlands

optical reflectance spectroscopy; vasomotion; oxygen extraction; hypoxic vasodilation THE REGULATION OF BLOOD FLOW to organs is primarily driven by a need to maintain sufficient transport of nutrients to the tissues and washout of metabolic waste products.However, this regulation occurs at two levels, to meet both the local needs of an individual organ and also the global needs of the body.This is predominantly achieved by the dilation and constriction of blood vessels, thus altering their conductivity.From the structure of the vascular walls, it can be deduced that this can be mediated through a neurogenic, myogenic, or endothelial response.This dynamic interplay of regulatory mechanisms results in a system that induces oscillations in blood pressure, volume, flow, and oxygenation (56).By studying the cyclical nature of these variables, researchers have attempted to identify the complex mechanisms that respond to changes in perfusion pressure and metabolic needs.Rhythmic variations in blood flow independent of heart rate can be observed at a range of frequencies.The effect of fluctuation in blood pressure approximately every 10 s, known as Mayer waves, can be observed in tissue as a 0.1-Hz oscillation in blood volume and flow (3,23,57).Rhythmic variations in blood flow between 0.02 and 0.05 Hz are thought to be induced by neurogenic activity (23,28,54).The endothelium is considered to modulate vascular tone at low frequencies Ͻ0.02 Hz via nitric oxide (NO) and endothelium-derived hyperpolarizing factor (25,27,33,48).
Vasomotion is a specific form of spontaneous local oscillation of vascular tone generated within the vascular wall (35) that may be synchronized by neuronal or humoral inputs.These fluctuations in vascular tone have been observed for more than 150 years in the intact animal and in isolated arteries (5,12,17,37,46,54).Vasomotion has also been demonstrated indirectly in vivo by the techniques of laser Doppler fluximetry (LDF), optical reflectance spectroscopy (ORS), near infrared spectroscopy, cell velocity measurements, and capillary pressure measurements (18,26,45,51,59).Although vasomotion is strictly a local phenomenon, the regulation of contractile activity of vascular smooth muscle cells is dependent on the complex interplay between vasodilator and vasoconstrictor stimuli from circulating hormones, neurotransmitters, endothelial-derived factors, and blood pressure (38).However, the complexity and the transient nature of vasomotion has constrained our ability to observe this phenomenon and has therefore limited our understanding (1).Although a recent review concludes that little is known about the consequences of such oscillations, particularly in microcirculatory function (38), there is some evidence indicating a possible role of vasomotion in tissue oxygenation (15,35,46,47,60).Theoretical modeling of vasomotion has also suggested that these oscillations might ensure adequate oxygen delivery to all tissues (24,60), with the largest effect seen for low frequencies Ͻ0.05 Hz related specifically to endothelial and sympathetic activity (15).
The primary goal of this study was therefore to establish whether vasomotion induced specifically by endothelial activity has an influence on the oxygen extraction and hemodynamics of blood in the arterioles, capillaries, and venules of the cutaneous microcirculation of dorsal forearm.Previously, the authors have described how the effects of vasomotion in the skin can be observed by ORS (58).Spontaneous cyclical changes in blood volume and mean blood saturation (S mb O 2 ) can be derived from the spectroscopic quantitative measurement of the concentration of the two chromophores oxyhemoglobin ([HbO 2 ]) and deoxyhemoglobin ([Hb]).However, all oscillations in S mb O 2 are not induced by vasomotion, and changes in S mb O 2 do not necessarily reflect changes in oxygen extraction.We have previously identified two distinctly different types of spontaneous swings in S mb O 2 .Type I swings were thought to be induced by fluctuations in arterial blood volume, resulting from the effects of respiration, endothelial, sympathetic, and myogenic activity.There was no apparent change in [Hb] and therefore no apparent change in oxygen uptake.In contrast, type II swings resulted from a fall in [HbO 2 ] accompanied by a rise in [Hb] and were only induced by endothelial and sympathetic activity.The type II swings may therefore indicate either a measure of oxygen extraction or a change in blood flow.This study focused on the type II swings induced by endothelial activity with the aim to use both ORS and LDF to determine whether this type of vasomotion effects oxygen extraction.

METHODS
Optical reflectance spectroscopy.The ORS instrumentation used in this study was the O2C (Lea Medizintechnik, Giessen, Germany) with an LF-2 surface probe.Continuous white light (500 -850 nm) is guided to the skin via a 200-m-diameter optical fiber.Backscattered light is then collected by a 400-m detector fiber encapsulated in the same probe at a distance of 2 mm from the light source.The sampled volume of tissue is considered to be to a depth of approximately half the probe spacing (32,36); therefore, a skin thickness of 1 mm is considered to be sampled.The collected light is spectrally analyzed in steps of 1 nm (500 -620 nm) to derive relative concentrations of oxyand deoxyhemoglobin.The derived concentration parameters [HbO 2] and [Hb] can also provide a measure of changes in blood volume in the skin obtained from the concentration of total hemoglobin [rHb], being the sum of [HbO 2] and [Hb].Similarly, the mean blood oxygen saturation defined by SmbO2 ϭ [HbO2] ϫ 100/([HbO2] ϩ [Hb]) can be determined.However, it is important to recognize that ORS calculates the mean values of [HbO 2] and [Hb] across all vessels in the microcirculation of the skin.Therefore, the derived SmbO2 is a measure of mean blood oxygen saturation across arterioles, capillaries, and venules.
Laser Doppler fluximetry.Incorporated within the O2C probe is an optical fiber that guides the output from 830 nm (30 mW) laser diode to the surface of the skin, adjacent to the ORS white light source.The moving red blood cells (RBCs) in the cutaneous microcirculation Doppler shift the laser light, which returns to a detector 2 mm from the laser source.This probe separation is the same as for ORS.Therefore, the area of tissue sampled by LDF overlaps that studied by ORS.Extrapolation from the model simulations by Clough et al. ( 4) suggests that the longer laser wavelength and higher power of the LDF may result in not only samples from the superficial layers as ORS does but also in samples from slightly deeper vessels.From the Doppler shift and intensity of back scattered light, the O2C derives in arbitrary units the mean RBC velocity and the blood flux, being the product of mean RBC velocity and concentration of moving RBCs.
Experimental protocols.The investigation was carried out with the approval of the Devon and Torbay Research Ethics Committee and in accordance with the Declaration of Helsinki.Twenty four healthy,  normotensive, male volunteers took part in the study; the clinical characteristics are in Table 1.The wide range of body mass indexes (BMIs) was considered to represent the variety in current population.All volunteers gave written informed consent and were studied in the morning, having fasted and refrained from caffeine, alcohol, and smoking from 2200 h the previous evening.Participants were not taking any vasoactive medication.Subjects lay supine and were acclimatized for 30 min in a thermostatically controlled room at a temperature of 22.0 Ϯ 0.5°C.Right forearm skin temperature was monitored continuously to ensure it remained constant for the duration of the study.The O2C probe was placed ϳ10 cm distally to the lateral epicondyle of the humerus over the extensor muscles.Upon stabilization of skin temperature, 30 min of baseline data were recorded from the dorsal forearm skin using the O2C. and the derived SmbO2 are given in Fig. 1.Power spectra for all parameters were derived by fast-Fourier analysis (Matlab) with a frequency range of 0.001-0.25 Hz.The highest frequency is determined by the sampling rate of 0.5 Hz and the frequency resolution of 0.001 Hz derived from the 17-min period of analysis.The mean values of the signals were subtracted (detrended) and multiplied by a Hamming window to reduce spectral leakage, before Fourier analysis.Each subject's raw data were examined manually to identify patterns in fluctuation in [HbO2], [Hb], [rHb], and hence SmbO2.Data were defined to contain type II swings in SmbO2 considered to be the result of endothelial activity if: 1) the frequency of oscillation was Ͻ0.02 Hz, 2) there was an anti-phase change in [HbO2] and [Hb] with small changes in total blood volume [rHb] synchronized predominantly with changes in [HbO2], and 3) the rise in SmbO2 was synchronized with a short surge in flux.Although vasomotion is rhythmic, it is still relatively variable over time, as shown by the examples from two subjects in Fig. 1.However, consistent, rhythmic oscillations occurred over periods of Ͻ17 min across all subjects.To fully comply with the three requirements identified above, analysis was therefore limited to five cycles that fulfill the criteria (Fig. 1).
Statistics.Data presented in the text are means Ϯ SD, unless specified as mean and range.Statistical analysis was performed by SPSS version 15.0 (SPSS) with data sets tested for normality using the Kolmogororv-Smirnov test.For normally distributed data, group comparisons were made by parametric unpaired t-test.For nonnormally distributed data and sample sizes Ͻ12, nonparametric statistics were applied.

RESULTS
Vasomotion attributed to endothelial activity has been studied by ORS and LDF in dorsal forearm skin of healthy males (Table 2).In 19 out of 24 subjects, while at rest and without intervention, spontaneous oscillations in S mb O 2 , [HbO 2 ], [Hb], [rHb], and flux were observed (Fig. 1).The S mb O 2 fluctuated with amplitude (relative to mean) of 9.0 Ϯ 4.0% averaged across all subjects.Fourier analysis and manual inspection of the data revealed low-frequency oscillations in S mb O 2 , [HbO 2 ], [Hb], and flux at Ͻ0.02 Hz, thought to be induced by endothelial activity (Fig. 2).These oscillations in saturation were identified as type II swings in S mb O 2 characterized by an anti-phase swing in [HbO 2 ] and [Hb] with some small changes in total blood volume [rHb] synchronized predominantly with changes in [HbO 2 ].Fluctuations in S mb O 2 were predominantly observed in subjects with skin temperatures Ͼ29.3°C (X 2 ϭ 6.14, P Ͻ 0.02).A schematic diagram of the observed correlation between changes in [HbO 2 ] and S mb O 2 with changes in flux is shown in Fig. 3.The data suggest a consistent minimum threshold of S mb O 2 in each individual (mean: 39.4%, range: 24.0 -50.6%) that precedes a sudden sharp rise in S mb O 2 .This minimum S mb O 2 threshold correlates very tightly (Pearson's correlation coefficient r 2 ϭ 0.95, P Ͻ 0.001) with the averaged  For each subject, a t 1/2 was calculated across five oscillations, and this was plotted against their BMI (Fig. 6).A comparison of the mean half-life in the 8 lean subjects (BMI Ͻ26.0 kg/m) of 12.2 s and the 11 obese subjects (BMI Ͼ29.5 kg/m 2 ) of 18.8 s was statistically significant (Mann Whitney, P Ͻ 0.004).If the t 1/2 is to be a measure of oxygen extraction alone, then for every molecule of oxyhemoglobin that releases oxygen there should be an increase of one molecule of deoxyhemoglobin.This correlation is clearly demonstrated in Fig. 7 where, for periods of constant flux and blood volume, the change in concentration of Hb has been plotted  against the change in concentration of HbO 2 (Pearson's correlation coefficient, r 2 ϭ 0.80, P Ͻ 0.001).

DISCUSSION
In this study, the influence of vasomotion on oxygen extraction has been studied indirectly by observing its consequences for blood flux derived by LDF and the concentration of oxyand deoxyhemoglobin in the tissue by ORS.An association has been observed between vasomotion and oxygen extraction, but this does not imply causality.The authors have previously described how fluctuations in S mb O 2 derived by ORS do not necessarily imply changes in oxygen extraction.In simultaneously recording both LDF and ORS, this study has for the first time in human skin been able to identify possible regulatory components of flow patterns.The study aims to unify the volume of tissue interrogated by LDF and ORS by using the same light source-detector separation of 2 mm.However, the authors recognize that the longer wavelength and relatively high power used for LDF may also sample from slightly deeper vessels (4).The results suggest that the consequences of vasomotion in the microcirculation are brief surges in arterial inflow followed by a passive period of oxygen extraction.This supports the findings of Oude Vrielink et al. ( 37) that vasomotion in animals appears to be a series of rhythmic dilations.Their observations from intravital video microscopy of arterioles of the tenuissimus muscle of rabbits describe the onset of dilation rather than the onset of constriction occurring synchronously and the dilation phase of vasomotion to be shorter than the constriction phase.The present authors believe that this is the first time the pattern of low-frequency vasomotion has been described in humans.Previous studies have predominantly focused on the spectral analysis of vasomotion in the skin but have not described the pattern.Oude Vrielink et al. further suggested, but provided no evidence to support, that it may be a low threshold in oxygen supply that triggers an increase in arterial inflow.Interestingly, in the present study, the increase in blood volume in the tissue under investigation during this period of increased flux is relatively small.The present authors have previously demonstrated that oscillations in blood volume, particularly arising from vasomotion in the arterioles attributed to neurogenic activity, can be observed directly by ORS in the skin (58).However, in the present study, since the increase in blood volume is relatively small, it is possible that the true vasomotion, being a dilation of the vessel and therefore an increase in blood volume, is occurring upstream and that this study is demonstrating the effects downstream.Further research is required.The key question is whether it is a local low threshold in S mb O 2 that is triggering the response upstream, conforming with the hypothesis that information should be conveyed from distal to proximal vessels in the vascular network (39).
It is well established that there exist multiple signaling mechanisms by which information about the dynamically varying tissue function status can be conveyed to individual vessel segments (37).Tissue metabolites, such as acidosis, ATP, and NO released from tissue cells, can reach smooth muscle cells of blood vessels by diffusion, causing local changes in vessel diameter (7,34).Similarly, the intravascular concentration of these vasoactive metabolites can play a regulatory role on vascular tone.Our study suggests that a low threshold in S mb O 2 induces a surge in arterial inflow at a frequency of Ͻ0.02 Hz, related to endothelial activity.Hypoxic vasodilation has been observed for more than 100 years and is a physiological response that matches blood flow and oxygen delivery to tissue metabolic demand (8).The direct role of the RBC as a sensor of oxygen demand and a regulator of oxygen supply was postulated in 1995 by Ellsworth et al. (10) to be mediated through the interaction of ATP, released from the RBC, with the endothelium.This concept, that RBCs may facilitate efficient delivery of oxygen to tissue, was further advanced by Stamler and colleagues (21) who postulated it to be mediated through an S-nitrosothiol-based signal.Although hemoglobin has now been established as an oxygen sensor that induces hypoxic vasodilation, there are several pathways through which this can be mediated.The relative importance of each individual mechanism has been debated intensely for a number of years and produced many review articles (2,13,14,20,29,42,53,61).However, the experimental evidence applicable to this human in vivo study is variable.Proposed mechanisms currently include ATP release (16,19), S-nitrosohemoglobin (SNO-Hb)-dependent vasodilation (8,40), and the nitrite reductase activity of deoxyhemoglobin (30).
For RBCs to initiate improved perfusion locally, there must also be a feedback mechanism providing conducted vasodilation.Because deoxygenation mainly occurs in the capillaries where there is no smooth muscle, resistance must be reduced  either upsteam or downstream to increase local blood flow.Conducted vasodilation has been shown in the arterioles of intact hamster cheek pouch retractor muscle in response to ATP, as far as 1,200 m upstream at a rate of 50 m/s (31).Conducted vasodilation in response to acetylcholine has also been well documented in the literature (9,49,50).However, in a recent study in mouse cremaster muscle, an upstream conducted vascular response to a local change in tissue oxygen tension suggested the endothelium-derived vasodilation to be NO independent (41).
This study demonstrates for the first time human in vivo evidence of a fast arterial inflow to a region of cutaneous tissue that appears to be triggered by an average low threshold in local S mb O 2 of 39.4% with a range of 24.0 -50.6% across the subjects.These results are consistent with findings in dogs that hypoxic vasodilation appears to occur as the hemoglobin desaturates from 60 to 40% with blood flow increasing to an average of 3.4 times normal (44).In this study, the minimum threshold oxygen saturation is highly correlated with the mean S mb O 2 (r 2 ϭ 0.95, P Ͻ 0.001; Fig. 4).The linear regression has a gradient of 1.06 and intercept of Ϫ6.7%.This suggests that, independent of their mean S mb O 2 (range 30.2-52.4%), in all subjects with vasomotion attributed to endothelial activity, a decrease in oxygen saturation of ϳ7% triggers a vasodilatory response.Again, it should be reiterated that S mb O 2 is a measure of mean blood oxygen saturation across arterioles, capillaries, and venules, and the relative size of these compartments may vary between individuals.
It needs to be emphasized that this surge in flux is brief, ϳ10 -15 s, when compared with the total vasomotion cycle length of 50 -100 s.The application of ATP has been shown to induce hypoxic vasodilation upstream from a hamster cheek pouch retractor muscle arteriole in ϳ7 s and from a capillary in ϳ20 s (11).Similarly, human RBCs rapidly elicit the relaxation of rabbit aortic rings in conditions of hypoxia in ϳ8 to 12 s, and this is attenuated in SNO-Hb-depleted RBCs (8).In steady-state hypoxia, vasodilation in rat or rabbit aortic rings by RBCs and nitrite was relatively slow, requiring 2-7 min, but, by dropping the oxygen with time, a 10% tension decrease could be observed in 10 -15 s (6).In this in vivo human study, this fast but brief surge in flux is followed by a prolonged period where both blood volume and flux are seen to remain relatively constant.During this period, there is a steady decrease in [HbO 2 ] and rise in [Hb] that might reasonably be assumed to be a consequence of oxygen extraction alone.Oxygen extraction has been approximated by the t 1/2 of the decay in S mb O 2 during this period.There is a significant relationship between t 1/2 and BMI but not with any of the other parameters in Table 2.It must, however, be emphasized that the subject group does not have a continuous range of BMIs but is subdivided into normal and obese; therefore, linear correlation was not applied.A comparison of the t 1/2 in the 8 lean subjects (BMI Ͻ26.0 kg/m) of 12.2 s and the 11 obese subjects (BMI Ͼ29.5 kg/m) of 18.8 s was statistically significant.This implies that the oxygen extraction in the obese subjects is reduced or less efficient than in the lean subjects.Although there is a possibility that the sample volume of tissue in the two groups may be different, the O2C probe with a detector spacing of 2 mm has a sampling depth of up to 1 mm, incorporating cutaneous tissue in both the lean and obese.Dorsal forearm, nonacral skin was specifically se-lected as the site of interest in this study, to focus on endothelium-dependent rather than neuronal vasoregulation.In dorsal forearm, nonacral skin, the complications of high basal sympathetic vasoconstrictor drive to arteriovenous anatastomoses are limited; however, the authors are aware of the possibility of sympathetic vasodilator fibers (22,43).For over 20 years, researchers have attributed vasomotion at frequencies between 0.02 and 0.05 Hz to neurogenic activity (23,54,55).From the Fourier power spectrum of ORS and LDF in Fig. 2, it can be seen that, in this study, vasomotion is dominant at frequencies Ͻ0.02 Hz attributed to endothelial activity in contrast to minimal power between 0.02 and 0.05 Hz resulting from neurogenic activity.
Although the range of skin temperatures across subjects varied from 25.5 to 33.5°C, the maximum variation of skin temperature within a study was only 0.7°C.Spontaneous periodic fluctuations in volar finger skin temperature have been demonstrated to correlate with vasomotion at ϳ0.014 Hz related to endothelial function, however, only with an amplitude of 0.04°C (52).The presence of vasomotion in this study was predominantly observed in skin temperatures Ͼ29.3°C.In conjunction with other unpublished data, the authors hypothesize that there may be a limited skin temperature range in which vasomotion may be observed and requires further research.
In conclusion, this study has clearly demonstrated oscillations in S mb O 2 that occur spontaneously in the dorsal skin of the forearm and are thought to be the consequence of vasomotion related to endothelial activity.The results suggest that the effects of vasomotion in the cutaneous microcirculation are brief surges in arterial inflow followed by passive periods of oxygen extraction.It appears that the trigger for increased arterial inflow is a low threshold in mean blood oxygen saturation.These observations support the hypothesis that RBCs may act as sensors of local tissue hypoxia, through the oxygenation status of the hemoglobin.After a temporary surge in arterial inflow, while blood volume and flux remain relatively constant, a steady decrease in [HbO 2 ] and equal and opposite increase in [Hb] may provide a noninvasive measure of in vivo tissue oxygen extraction.

Fig. 1 .
Fig. 1.Examples from two subjects of out of phase swings in the concentration of oxyhemoglobin ([HbO2]) and deoxyhemoglobin ([Hb]) in dorsal forearm at a frequency of 0.013 Hz (top) and 0.08 Hz (bottom).The subsequent swings in mean blood saturation (SmbO2) are identified as type II swings and synchronize with fluctuations in flux.There are small changes in the concentration of total hemoglobin ([rHb]) synchronized predominantly with changes in [HbO2].Shaded areas identify regions where flux and blood volume remain relatively constant and the fall in SmbO2 is considered to be a measure of oxygen extraction.au, Arbitrary units.
Data analysis.Periods of 17 min artifact-free baseline data were analyzed in Matlab to identify the frequencies of spontaneous oscillations observed in S mbO2, [HbO2], [Hb], [rHb], flux, and mean RBC velocity.Two examples of the signals of flux, [HbO2], [Hb], [rHb],

H444VASOMOTION
AND OXYGEN EXTRACTION AJP-Heart Circ Physiol • VOL 301 • AUGUST 2011 • www.ajpheart.orgby 10.220.33.5 on June 12, 2017 http://ajpheart.physiology.org/Downloaded from S mb O 2 across 17 min of data (Fig.4).If this intermittent sharp rise in S mb O 2 can be attributed to the short fivefold surge in flux, then, assuming arterial inflow with 100% oxygenated hemoglobin, the integral of the increase in flux (F AUC ) should correlate with the integral increase in [HbO 2 ] (H AUC ).This correlation is confirmed in Fig.5where F AUC and H AUC have been calculated for five cycles in S mb O 2 in each of the 19 subjects displaying vasomotion (Pearson's correlation coefficient r 2 ϭ 0.50, P Ͻ 0.001).Following each intermittent surge in flux, there is a constant period where both flux and blood volume [rHb] remain relatively constant, but there is still a steady decrease in [HbO 2 ] and rise in[Hb].In the sample of skin interrogated by ORS and LDF, if there is no change in either blood flux or blood volume but there is an increase in [Hb] then it might reasonably be assumed that this is a consequence of oxygen extraction alone.A measure of this oxygen extraction [average half-life (t 1/2 )] has been approximated to the time for the mean blood oxygen saturation to fall by half, i.e., 1/2⌬S mb O 2 .

Fig. 2 .Fig. 3 .
Fig. 2.An example of the Fourier power spectra of optical reflectance spectroscopy and laser Doppler fluximetry baseline data from dorsal forearm skin.The frequency range is 0.001-0.25 Hz in B and expanded 0.001-0.05Hz in A.

Fig. 4 .Fig. 5 .
Fig. 4. Minimum threshold SmbO2 at which a surge in flux is induced vs. the mean SmbO2 over 17 min.

Fig. 6 .
Fig. 6.The t1/2 of SmbO2 (a measure of oxygen extraction) vs. body mass index (BMI) for 19 subjects during periods of constant blood flux and volume but increase in [Hb] and fall in [HbO2].

Fig. 7 .
Fig. 7.The change in [Hb] plotted against the change in [HbO2] from 19 subjects for periods where blood volume and flux remain constant and a measure of oxygen extraction is derived.

Table 1 .
Comparison of clinical characteristics of 24 healthy male subjects, 12 normal and 12 obese