Cutaneous Vasoconstriction Affects Near-infrared Spectroscopy Determined Cerebral Oxygen Saturation during Administration of Norepinephrine

• Cerebral oximetry by spatially resolved near-infrared spectroscopy is used to monitor and optimize cerebral oxygenation during procedures that might impair cerebral perfusion

• Changes in cutaneous perfusion could interfere with the accuracy of cerebral oximetry

• The effects of hypocapnia, hyperthermia, hypoxia, and norepinephrine on cerebral and forehead skin blood flow and oxygenation were studied in human volunteers

• Cerebral oximetry accurately detected cerebral deoxygenation produced by hypoxia and hyperventilation, but hyperthermia and norepinephrine affected skin oxygenation and directionally influenced cerebral oximetry readings

IT is crucial to maintain cerebral blood flow (CBF) during anesthesia, but it is difficult to determine CBF in clinical settings. On the other hand, it is straightforward to monitor frontal lobe oxygenation (scO²) by noninvasive near-infrared spectroscopy (NIRS) that provides real-time assessment and is reported to monitor scO²correctly.1Thus, perioperative optimization may be directed to preserve scO²2and maintained scO²secures rapid postoperative recovery in both cardiac3and elderly patients.4For example, vasopressors are administered to prevent reductions in mean arterial pressure (MAP) that could affect CBF and in turn scO². However, vasopressors appear to affect scO²differently. Phenylephrine reduces scO²in anesthetized patients,5but that is not the case for ephedrine.6In healthy subjects the administration of norepinephrine reduces scO²7with a concomitant reduction in blood flow velocity in the middle cerebral artery (MCA v^mean) and internal jugular venous saturation (sjO²), suggesting that sympathetically mediated cerebral vasoconstriction reduces CBF to an extent that affects scO². But sympathetic innervation of the cerebral vasculature remains controversial, and whether administration of norepinephrine reduces MCA v^meanremains debated.8Alternatively, the reduction in s^cO²following administration of norepinephrine is influenced by cutaneous vasoconstriction.9Also both CBF and scO²are sensitive to changes in PaCO²10,11and an increase in ventilation provoked by administration of norepinephrine could contribute to the reduction in scO².12 

An influence from skin oxygenation to the NIRS signal has been identified.13,14In a hot environment, skin blood flow (SBF) increases, especially over the forehead,15whereas there is no change in CBF or muscle blood flow. However, the NIRS-determined muscle oxygenation increases during whole-body heating.13For evaluation of cerebral oxygen saturation in a hot environment, it appears important that hyperthermia elicits a hyperventilation-induced lowering of PaCO²16,17and, in turn, CBF.18Thus, when scO²is unaffected by hyperthermia19while the internal jugular venous-derived cerebral oxygenation is reduced,20the maintained scO²indicates that scO²is influenced by SBF, and this hypothesis was tested in the present study. For this purpose we assessed cerebral perfusion and oxygenation while PaCO²and PaO²were manipulated with separate assessment of SBF and skin hemoglobin and oxygenation.

Fifteen healthy males (age 25 ± 5 yr [mean ± SD], height 182 ± 7 cm, and mass 76 ± 8 kg) participated in the study. The study was approved by the local ethics committee (H-4–2010–132, Copenhagen, Denmark) in accordance with Declaration of Helsinki including oral and written informed consent.

After arrival to the laboratory at 8:00 AM, the subjects were resting supine for 20 min before catheterization. Under local anesthesia (2% lidocaine), a catheter (Edwards Life Sciences, Irvine, CA) was inserted retrograde in the right internal jugular vein by Seldinger technique guided by ultrasound, and the tip of the catheter was placed at its bulb. The position of the catheter at the jugular bulb was verified during placement if the subjects reported a slight pain behind the ear. The placement was also confirmed by quick infusion of saline if the subject experienced an auditory response.21,22A 20 G catheter was placed in the brachial (n = 12) or radial artery (n = 3) of the nondominant arm and a central venous catheter (Cavafix MT134, Braun, Melsungen, Germany) was advanced through an arm vein. Catheters were connected to transducers (Edwards Life Sciences) placed at the level of the heart (fourth intercostal space; Dialogue-2000; IBC-Danica Electronic, Copenhagen, Denmark) to monitor MAP and heart rate. Data were analog-digital converted and sampled at 200 Hz (DI-720; Dataq Instruments Inc., Akron, OH) by computer software (Windaq; Dataq Instruments Inc.).

To assess scO²we used a spatially resolved NIRS (INVOS; Somanetics, Troy, MI), meaning that it records differences in absorption of photons returning from deep and superficial tissues23using light at 703 and 808 nm and an emitter-detector separation of 3 and 4 cm with the depth sensitivity corresponding to approximately one-third of the emitter-detector separation.24Although the algorithm of the instrument is not disclosed, the apparatus is widely used in clinical cerebral monitoring2,3,10,25and evaluation of its performance is therefore important. The general approach of this instrumentation is that by analyzing the signals from the two emitter-detector separations, the signal of superficial tissue layers is suppressed, providing an estimate of deep tissue, e.g. , the frontal lobe oxygenation. We assumed that frontal lobe activation was unchanged throughout the protocol and therefore did not influence cerebral metabolic rate of oxygen and, in turn, the NIRS signal. To avoid any influences from the frontal or sagittal sinuses to the NIRS signal, the sensor was placed high on the forehead in randomized and balanced order. The SBF was monitored by LDF (moorVMS-LDF; Moor Instruments, Axminster, United Kingdom) using light at 785 nm. The LDF sensor was integrated with a tissue oxygenation sensor (moorVMS-OXY; Moor Instruments) that uses white light spectroscopy (wavelength range, 400–700 nm) to assess skin oxygen saturation^,skin hemoglobin concentrations, and temperature. The instrument self-calibrates when turned on prior and LDF and white light spectroscopy are established as valid measurements of the microcirculation of the skin.26,27We assumed that changes in SBF beneath the LDF sensor reflected that beneath the NIRS sensor, with the two sensors placed ipsilateral and distanced 3 to 4 cm apart.

To evaluate whether the interventions affected CBF, MCA v^meanwas determined by transcranial Doppler sonography (2 MHz probe, Multi-Dop, DWL, Singen, Germany). Using adhesive ultrasound gel, the best signal-to-noise ratio was obtained at the temporal insonation window. Placement of the probe was randomized and balanced across subjects with respect to hemisphere, and eight subjects had the probe placed on the right side. Previous reports exclude influence of PaCO²on the diameter of the MCA and we, therefore, assumed that changes in MCA v^meanreflect those in CBF.28 

Transcranial Doppler, SBF, NIRS, and skin oxygenation in addition to the blood pressure signals were all analog-digital converted and sampled on the DI-720. Following recording, values were averaged over at least 30 s before entering the statistical analysis.

Arterial and jugular venous blood samples were obtained simultaneously in preheparinised syringes. Atmospheric air was immediately removed from the syringe. The samples were analyzed for PaO²and PaCO², oxygen saturation (saO²; sjO²), and total hemoglobin (ABL700; Radiometer, Copenhagen, Denmark). To express the oxygen content in the arterial or venous blood, CO²= 0.0031 · PO²+ Hb · SO²  was used. The arterial to venous difference (av^dO²) for the brain was calculated. With the assumption that CMRO²was constant, changes in CBF were calculated using the Fick equation,

The cerebrovascular resistance index was calculated as MAP divided by MCAv^meanand the skin vascular resistance index was MAP divided by SBF. Using a simplified equation, s^capO²was calculated:1,29 

The calculation of s^capO²is based on the cerebral oxygen extraction and motivated by the assumption that oxygen extraction remains constant along the capillary network.30,31Gas is exchanged linearly along the entire vascular pathway from the arteries to the veins.32,33Thus, it can be assumed that sO²decreases linearly from the arterial to the venous end of the capillary and average cerebral oxygen saturation is then the midway point between arterial and venous saturations.29 

Following catheterization, the subjects rested supine for 30 min with elevated head rest with room temperature kept constant (22–24°C). The experiment included three trials performed in randomized and balanced order, followed by a hypoxic trial. Before each trial resting values were obtained with 43 ± 14 min rest between the trials. The trials included the following.

(1) The subjects were asked to hyperventilate to reduce the end-tidal pressure of carbon dioxide (PETCO²; INNOCOR, Odense, Denmark) by 1.5 kPa. Blood samples were collected after 7 ± 2 min reflecting when the targeted PETCO²was established.

(2) To increase skin temperature, the subjects were exposed to whole-body heating. The subjects were covered with a heat-isolating aluminum blanket together with a dedicated heating blanket (Bair Hugger 505; Arizant Healthcare, Eden Prairie, MN) that was inflated by air at 42°C. Linens covered and isolated the head. Blood samples were obtained after 26 ± 7 min, reflecting when there was no further increase in SBF (skin temperature increased by 1.7 ± 0.7°C) and it was confirmed that PaCO²remained stable.

(3) Norepinephrine (0.15 μg · kg⁻¹· min⁻¹) was infused through the central venous catheter for 15 min; blood samples were collected afterward. To avoid potential effects of norepinephrine on PaCO²,7the subjects were instructed to restrain the ventilatory depth for 11 ± 5 min in order to maintain the arterial carbon dioxide. If the subject was not able to control PETCO²in this way^,a tube was added to the inspired gas mixture and thus increased the ventilatory death space. Administration of norepinephrine with and without PETCO²correction was performed in randomized order.

(4) After these trials, the subjects were exposed to hypoxia (FⁱO²: 0.12%) for which the inspired air was balanced by N²(Altitrainer, SMTEC, Nyon, Switzerland) for 10 min, while blood samples were collected.

Based on similar studies,7,8,34a sample size of 15 subjects was assumed to provide β < 0.20. One-way ANOVA followed by a Tukey post hoc  test on repeated measures evaluated variables. To express relationships between variables we used Pearson's correlation. The results are presented as changes from baseline and variables are provided with their 95% CI. We performed a two-tailed hypothesis testing and the statistically significant level was set to P < 0.05 with data analyzed in SAS 9.2 (SAS Institute Inc., Cary, NC). Because of insufficient signal quality, MCV v^meanwas excluded in one subject during hyperventilation and in another subject during the control before whole-body heating. Similarly, scO²was excluded in one subject during whole-body heating.

As intended, hyperventilation induced a 1.1 kPa decrease in PaCO²whereas PaO²increased (table 1). Skin oxygen saturation increased by 1.8% (0.5% to 3.1%), although there were no changes in skin hemoglobin concentrations, skin vascular resistance index, or SBF.

There were no changes in MAP, but MCA v^mean, sjO², and CBF^Fickall decreased. Also, the NIRS-determined scO²was reduced by 2.1% (0.4–3.7%; fig. 1) and s^capO²was reduced by 5.3% (3.8–6.9%).

With exposure to whole-body heating there was an increase in skin temperature by 1.6°C (0.7–2.6°C) (table 2) and a concomitant increase in SBF by 23.5 AU (10.4–36.6) and skin oxygen saturation by 3.1% (1.3–4.9%).

We did not observe any changes in MAP, MCA v^mean,sjO², or CBF^Fickin response to whole-body heating (table 1). Yet, there was an increase by 3.6% (2.1–5.1%; fig. 1) in scO², whereas s^capO²remained unchanged.

With administration of norepinephrine, skin vascular resistance index increased while total hemoglobin of the skin decreased (table 2). Thus, skin oxygen saturation was reduced by 2.2% (1.0–4.3%).

Administration of norepinephrine induced in average 22 mmHg increase in MAP, whereas MCA v^mean,sjO², and CBF^Fickwere unaffected. Yet, the scO²was reduced by 6.2% (4.2–8.0%; fig. 2), although s^capO²remained unchanged.

The subjects managed to keep PaCO²at the level established before administration of norepinephrine, eventually with enlargement of the ventilatory dead space (n = 6). Under these circumstances there was a tendency for a reduction in skin oxygen saturation (−3.8–0.5%; P = 0.066), but no changes in SBF or skin temperature, although skin hemoglobin concentrations and skin vascular resistance index were affected (table 2).

The MCA v^meanincreased by 6.9 cm/s. With PETCO²correction, scO²remained reduced by 4.2% (2.2–6.0%; fig. 2) with no change in s^capO²(−3.7–4.9%).

Hypoxia was associated with a reduction in skin oxygen saturation by 8.9% (6.3–11.6%; fig. 2). There was an increase in skin vascular resistance index because SBF decreased (table 2), but, on the other hand, skin hemoglobin concentrations remained unaffected.

With hypoxic exposure, PaO²and s^aO²were reduced. Neither MCA v^meannor CBF^Fickchanged. Also scO²(by 10.2%; 6.6–13.7%) and s^capO²(by 7.9%; 6.4–9.4%; fig. 1) were reduced.

The results provide, in our opinion, compelling evidence that skin oxygen saturation affects the applied spatially resolved INVOS NIRS-determined evaluation of scO². Whole-body heating and administration of norepinephrine provide a discrepancy between scO²and s^capO². On the other hand, during hyperventilation and hypoxia, there were corresponding reductions in scO²and s^capO². Thus, the reduction in scO²during administration of norepinephrine can be explained by cutaneous vasoconstriction, whereas the increase in scO²during whole-body heating seems because of an increase in skin oxygen saturation. These observations have implications for use of NIRS clinically and for preservation of scO²and in turn CBF with administration norepinephrine during anesthesia.

The calculation of cerebral oxygenation from arterial and jugular bulb saturation is of importance to this study. Here, we used an approach based on cerebral oxygen extraction and defined the average cerebral blood hemoglobin oxygenation as the midway point between the arterial and jugular venous saturations.29Others have proposed a 3:1 ratio primarily based on anatomical evidence between arterial and venous blood for validation of NIRS.35The contribution ratio may be different between NIRS devices and it is unlikely that the composition of cerebral blood volume remains stable at different levels of PcO²and, consequently, CBF.1,36The NIRS-derived arterial-to-venous ratios are somewhat contradicted by positron emission tomography data,37but contribution from skin may have influenced NIRS validation of the cerebral blood volume distribution. Importantly, using a 3:1 ratio between arterial and venous blood, or s^jugO²alone, does not change the conclusion that skin has a significant contribution to the NIRS signal.

The penetration depth sensitivity of NIRS light is proportional to the emitter-detector separation.24The presented results raise the question whether a separation distance of 3 or 4 cm is enough to secure spatial resolution and thereby exclude crosstalk between scO²and skin oxygen saturation. With whole-body heating, we obtained an increase in scO²and skin oxygen saturation, whereas calculated s^capO²remained stable (fig. 1). Thus, the increase in scO²associated with whole-body heating can be explained by the increase in skin oxygen saturation. Nonspatially resolved NIRS is affected by skin oxygen saturation with local and whole-body heating13and with intradermal injections of epinephrine.14Here, with spatially resolved NIRS, we observed a correlation between scO²and skin oxygen saturation during administration of norepinephrine with and without PETCO²correction(fig. 2) and a similar correlation was observed during whole-body heating (fig. 3). In support for an influence of skin oxygen saturation on scO², a marked increase in SBF and reduced skin vascular resistance index were observed. Marked hyperthermia can elicit hyperventilation, but PaCO²was maintained during whole-body heating and therefore did not affect cerebrovascular resistance index and sjO²as previously described.1Core temperature was not recorded, but the results show no changes in PaCO², MCA v^mean, or CBF^Fick, typically seen during whole-body heating.9 

Previous studies suggest that administration of vasopressors exert a negative impact on scO².5,–,7It is debated whether the reduction in scO²is because of a decrease in cardiac output,38sympathetically mediated cerebral vasoconstriction, or changes in SBF.34It is less ambiguous how α-adrenergic drugs affect the microcirculation of the skin.39The present results demonstrate a decrease in skin oxygen saturation with administration of norepinephrine (fig. 2) and lower skin hemoglobin concentrations and increased skin vascular resistance index. At the same time, scO²decreased whereas calculated s^capO²was unaffected, suggesting that cutaneous vasoconstriction affects the NIRS-determined scO². In support, when PETCO²was maintained, the reductions in scO²and skin oxygen saturation remained and the reduction in scO²during administration of norepinephrine could illustrate the influence of skin oxygen saturation considering that s^capO²was stable.

Administration of phenylephrine provokes a marked reduction in scO²despite unaffected internal carotid artery flow, PaO²and PaCO².34We acknowledge that a decrease in scO²with administration of norepinephrine could be because of an alteration in the arterial and venous contributions to the NIRS signal, but also, cutaneous vasoconstriction could be responsible for the reduction in scO². Administration of ephedrine, on the other hand, does not affect scO².5,–,7One explanation for this discrepancy could be that ephedrine does not have the same vasoconstrictive effects on smooth muscles in skin vasculature as norepinephrine. Reduction in cardiac output has a negative impact on scO^2,40but the present results suggest that the reduction in scO²is explained by a reduction in skin oxygen saturation. The stability of MCA v^meanand CBF^Fickalso emphasize that cerebral blood flow is not challenged, which would have been the case if cardiac output was lowered.40It remains unknown to what extent the cerebral vasculature is influenced by sympathetic innervation, but the results do not support sympathetic vasoconstriction in brain (table 1).

With systemic hypoxic exposure, s^capO²was reduced by 7.9% and scO²by 10.2% (P < 0.0001). Therefore, it seems that NIRS estimates brain oxygen saturation correctly during systemic hypoxic exposure. The influence of the skin oxygen saturation is minimal with hypoxia, but we cannot exclude that the 2.3% discrepancy between scO²and s^capO²can be explained by the 8.9% reduction of skin oxygen saturation (P = 0.03) because of cutaneous vasoconstriction (table 2). The threshold for hypoxemia-induced arterial dilatation was not reached,41but we cannot exclude cerebral vasodilatation because of hypoxic exposure. On the other hand, such vasodilation was likely counterbalanced by hypocapnia, because MCV v^meandid not change significantly (table 1).

Because of the higher oxygen pressure during hyperventilation, skin oxygen saturation increased, whereas s^capO²and scO²decreased (fig. 1 ).  The inverse relationship between scO²and skin oxygen saturation with hyperventilation could explain why scO²underestimates s^capO²by 3.2%, however, systematic errors because of a low signal-noise ratio cannot be excluded. These findings support that PaCO²has an influence on cerebral oxygen saturation as assessed by NIRS11and more so than the influence of the skin oxygen saturation. Regardless of the crosstalk between skin oxygen saturation and scO², INVOS is sensitive to changes in cerebral oxygen saturation during hyperventilation and systemic hypoxic exposure.

Despite the influence of SBF for the NIRS signal, it remains useful for perioperative optimizing to improve patient outcome after anesthesia,2,25although administration of α-agonists as vasopressors may confound interpretation of NIRS. Further studies are needed to evaluate whether difference in emitter-detector separation of more than 3 and 4 cm can avoid crosstalk between skin oxygen saturation and spatially resolved scO². Alternatively, different algorithms for the analysis of NIRS data are required to suppress skin artifacts. In a multiple regression analysis, skin oxygenation contributes approximately 30% to the NIRS signal; however, further investigations are needed to quantify in details the contribution from the different components beneath the NIRS sensor. Furthermore, it is needed to evaluate whether administration of norepinephrine is associated with reduction in scO²in situations with low cardiac out or hypotension.

The results suggest that the reduced INVOS NIRS-determined evaluation of scO²observed with administration of norepinephrine is because of skin vasoconstriction rather than cerebral deoxygenation. Thus, in situations with administration of norepinephrine and whole-body heating, an emitter-detector separation of 3 or 4 cm seems not to be large enough to avoid crosstalk between skin oxygen saturation and the INVOS NIRS-determined scO², and further technical measures to eliminate the skin contribution should be considered. Nevertheless, the results confirm that spatial resolved NIRS is able to detect cerebral deoxygenation associated with hyperventilation and systemic hypoxic exposure.