Does Tissue Oxygen-Tension Reliably Reflect Cerebral Oxygen Delivery and Consumption?

Abstract

We investigated the value of brain oxygen partial pressure (P_brO₂) with respect to predicting cerebral energetic failure in a rabbit model of global cerebral ischemia and hypoxia. Local cortical blood flow (l_coCBF), P_brO₂, extracellular lactate, pyruvate, and glutamate concentrations, as well as microvascular hemoglobin saturation (S_mvO₂), cytochrome oxidase redox level (Cyt a+a₃ oxidation), and brain electrical activity, were assessed during variable degrees of cerebral ischemia and hypoxia, induced by cisternal infusion of artificial cerebrospinal fluid or an admixture of nitrous oxide to inspiratory gas in 10 animals each. Arteriovenous difference in oxygen content, cerebral metabolic rate for oxygen, and oxygen extraction were derived from multimodal data. P_brO₂, S_mvO₂, and Cyt a+a₃ oxidation were closely related to cerebral blood flow and indices of oxidative metabolism. P_brO₂ ≤8 mm Hg corresponded to l_coCBF ≤15 mL · 100 g⁻¹ · min⁻¹, S_mvO₂ ≤9%, Cyt a+a₃ oxidation ≤20%, and progressive loss of brain electrical activity. Adequate tissue oxygenation was reflected by cerebral metabolic rate for oxygen ≥2.8 mL · 100 g⁻¹ · min⁻¹, arteriovenous difference in oxygen content ≤12.5 mL O₂ · 100 mL⁻¹, and oxygen extraction ≤60%. Meaningful interpretation of low P_brO₂, especially with respect to definition of energetic thresholds, requires complementary information from simultaneous assessment of l_coCBF and tissue oxygen extraction.

Ischemic and hypoxic thresholds of brain tissue have been characterized by cerebral blood flow (CBF) ⁽¹⁾, brain oxygenation ⁽²⁾, and energy metabolism ⁽³⁾. Complementary techniques for the assessment of CBF ⁽⁴⁾, brain electrical activity (BEA), arteriovenous difference in oxygen content (AVDO₂) ⁽⁵⁾, metabolic rate for oxygen (CMRO₂), brain energy metabolism ^(6,7), and tissue oxygenation (brain oxygen tension/P_brO₂, microvascular hemoglobin saturation/S_mvO₂) ^(8,9) have been used. Bedside systems for the continuous evaluation of CBF, P_brO₂, AVDO₂, and cerebral metabolism (microdialysis) have facilitated clinical monitoring ⁽⁶⁾. Recent clinical ⁽¹⁰⁾ and experimental ⁽¹¹⁾ data suggest that alternative variables such as P_brO₂ may surpass CBF in defining cerebral ischemic thresholds. Questions remain regarding the exact relationships between CBF, arterial Po₂, P_brO₂, intracellular Po₂ (P_icO₂), and CMRO₂. It has been proposed that P_brO₂ reflects AVDO₂ or oxygen extraction (OEF) rather than exclusively mirroring CBF ⁽¹¹⁾. Because coupling between CBF and AVDO₂ dynamically changes depending on the actual level of tissue perfusion ⁽⁵⁾, coupling between P_brO₂ and CBF should also vary as a function of CBF. Therefore, exclusive assessment of P_brO₂ may not adequately represent P_icO₂, if the diffusion gradients between capillaries and the intracellular compartment are altered. Interpretation of P_brO₂ under pathologic conditions warrants careful analysis and may necessitate support from supplementary variables of cerebral metabolism, especially with respect to refined definition of energetic thresholds. In this study, we intended to clarify the role of P_brO₂ by providing data derived from multiple complementary techniques, simultaneously analyzing local cortical CBF (l_coCBF), P_brO₂, AVDO₂, CMRO₂, S_mvO₂, cytochrome oxidase redox level (Cyt a+a₃ oxidation), extracellular lactate (L), pyruvate (P), and glutamate (Glu) concentrations as well as BEA during global cerebral ischemia and systemic hypoxia.

Animals and Methods

All animal procedures were approved by the institutional animal investigations committee. Twenty adult male New Zealand white rabbits (Charles River Laboratories, Kißlegg, Germany) weighing between 3.5 and 4 kg were fasted overnight and anesthetized by IM injection of ketamine (20 mg/kg; Parke-Davis, Berlin, Germany) and xylazine (7 mg/kg; Bayer, Leverkusen, Germany), followed by continuous IV administration of fentanyl (2–3 μg · kg⁻¹ · h⁻¹; Janssen, Neuss, Germany) and midazolam (0.2–0.5 mg · kg⁻¹ · h⁻¹; Hoffmann-LaRoche, Grenzach-Wyhlen, Germany). Minimal effects on cerebral circulatory physiology result from this anesthetic regimen. Animals were intubated and ventilated with O₂/N₂O, maintaining baseline Pao₂ between 100 and 120 mm Hg (CIV 101; Columbus Instruments, Columbus, OH). Paco₂, and arterial and venous pH were assessed regularly (ABL; Radiometer, Copenhagen, Denmark) and kept within their physiologic ranges (Paco₂: 35 ± 4 mm Hg, pH_a: 7.37 ± 0.06; mean ± sd). Mean arterial blood pressure was monitored in the inferior aorta, central venous pressure assessed in the inferior vena cava. Fluid administration was adjusted by regular assessment of central venous pressure and urine output. The electrocardiogram was obtained from needle electrodes placed in the lateral chest wall. BEA was recorded via eight bilateral frontoparietal miniature screw electrodes. The signal was digitally filtered, fast Fourier transformed, and was expressed as spectral edge frequency (SEF₉₅), containing 95% of the spectral power within 12-s epochs of the continuous recording. Brain temperature was measured by using a thermocouple (GMS, Kiel-Mielkendorf, Germany) and maintained at 36.5°–37.5°C, body temperature was adjusted accordingly by using a thermal blanket. Continuous registration of all vital variables was performed by using a customized monitor setup (CS/3; Datex-Ohmeda, Duisburg, Germany) allowing synchronous data trans-fer to a computer.

A cisternal infusion technique ^(12,13) was used for controlled variation of cerebral perfusion by dynamic adjustment of cerebral perfusion pressure (CPP). After rigid head fixation, a circular frontal trephination (Ø: 6 mm) exposing the translucent dura was performed to provide a closed cranial window for the spectroscopic system. Two 22-gauge silicone catheters were introduced into the cerebellomedullary cistern, one attached to an infusion pump for infusion of artificial cerebrospinal fluid (304 mOsmol/L, 37°C, 142.5 mmol/L NaCl, 2.5 mmol/L KCl, 1.5 mmol/L CaCl₂, 2.9 mmol/L glucose, 14.5 mmol/L NaHCO₂, equilibrated to pH 7.3), one connected to a pressure transducer (Medex Medical, Ratingen, Germany) for continuous recording of intracranial pressure (ICP). Rapid equilibration of ICP between infra- and supratentorial compartments was confirmed during preliminary experiments. Establishing steady-state conditions (spontaneous baseline CPP: 85–105 mm Hg) before and after each run, CPP was varied in random order between 10 and 180 mm Hg (creating plateaus at 10 mm Hg intervals across the entire cerebrovascular pressure flow autoregulatory range of approximately 50–130 mm Hg in rabbits). To reduce CPP from baseline, cerebrospinal fluid infusion rate was varied between 0 and 120 mL/h. CPP was increased by IV noradrenaline (0.2 mg/mL, 1–8 mL/h). Each CPP plateau was established for a minimum of 8 min to permit adaptation of cerebral hemodynamics. Finally, CPP was reduced to 0, inducing cerebral circulatory arrest, thus establishing extreme values for all investigated variables. Likewise, Pao₂ was randomly varied between 140 and 10 mm Hg (10 mm Hg plateaus, each maintained for at least 8 min) in 10 animals, replacing inspiratory oxygen (baseline inspiratory fraction: 30%–40%) by N₂O. Data were averaged for each CPP/Pao₂ plateau. Animals were euthanized by intracardiac injection of 50 mmol KCl after the administration of 200 mg of thiopental.

Assessment of CBF, P_brO₂, and Brain Energy Metabolites

Oxygen microelectrodes (Licox-Revoxode; GMS), microdialysis catheters (CMA 70; Axel-Semrau GmbH, Sprockhövel, Germany), and custom isocaloric thermal-diffusion sensors (2 round gold disk electrodes integrated in a 3 × 12 mm silastic leaf) for quantitative assessment of l_coCBF were inserted via separate burrhole craniotomies. l_coCBF was calculated as follows: K × (1/V − 1/V₀). K is a constant, V denotes the voltage difference during actual measurements, and V₀ signifies the voltage difference at zero flow, obtained during final calibration 10 min after cardiac arrest. Adequate function of Po₂ electrodes in situ was confirmed by their response (increase in P_brO₂ ≥25 mm Hg within 90 s) after an oxygen challenge (increasing F_iO₂ to 100%). Post hoc, each catheter was tested for zero drift (<1 mm Hg) by using an absolute-zero solution (Na₂S₂O₄) and for sensitivity drift (<5%) at a mean room-air Po₂ of 156.9 ± 1.7 mm Hg. Microdialysis samples (CMA 107 pump, flow rate: 1 μL/min, yielding recovery rates of approximately 20%–30%⁽⁶⁾; dialysate: Ringer’s solution containing 147 mval Na⁺, 4 mval K⁺, 155.5 mval Cl⁻, 2.25 mval Ca²⁺) were collected at 8-min intervals and frozen at −70°C. Extracellular L, P, and Glu concentrations were determined by using enzymatic assays and subsequent photometric quantification (CMA 600; Axel-Semrau GmbH). Systemic arterial L reference concentrations were analyzed regularly (ABL; Radiometer). Intracortical placement of all catheters (except for CMA 70, integrating a predominant cortical and a minor medullary fraction) was performed under microscopic magnification.

Reflection Spectroscopy

Raw tissue reflection spectra were collected in the VIS-NIR range between 490 and 1000 nm by using a high-resolution optical multichannel analyzer (OS30; Hamamatsu Photonics, Munich, Germany) coupled to a fiberoptic reflectance sensor using 6 concentric 200-μm illumination bundles and 1 200-μm central receiving bundle fitted within a stainless steel tip (Top-Sensor-Systems, Eerbeek, Netherlands). A 5-W halogen light source was used for tissue illumination (increasing tissue surface temperature <0.5°C). Reflected photons were directed to a linear CCD array (spectral resolution: 0.5 nm), averaged (sampling time: 124 ms/scan), transformed to a log (1/R) scale, mean-centered, and normalized. All spectra were referenced against a standard titanium-oxide reflector. Spectral signatures of hemoglobin (Hb) and Cyt a+a₃ implemented from high resolution (0.5 nm) in vitr. reference spectra were examined in fourth derivatives of the actual tissue summation spectra using a multivariate calibration model (NPLS) for nonlinear spectral multicomponent analysis ⁽⁸⁾. Relative mean spectral residual errors of <0.3% were achieved by least-squares fitting. Because of the optical geometry of the sensor and the interrogated wavelengths, only microvessels <30 μm (small venules, capillaries, and arterioles) significantly contribute to the reported S_mvO₂ levels and tissue Hb (tHb) concentrations ⁽¹⁴⁾. S_mvO₂ and Cyt a+a₃ oxidation were expressed in percent, tHb concentrations in units/milliliter tissue. SO₂ within pial arterioles (S_caO₂) was (a) measured spectroscopically, and (b) calculated from actual Pao₂ according to the rabbit oxygen-dissociation curve (ODC) of Hb ⁽¹⁵⁾, dynamically corrected by actual pH_a and Paco₂. The average P₅₀ (Pao₂ at 50% Hb saturation) under standard baseline conditions (Paco₂: 35 ± 2 mm Hg, pH: 7.38 ± 0.05) was 32.7 ± 0.4 mm Hg. The average values of Hill’s n were 2.9 ± 0.02. Polynomial fitting was applied to ODCs to illustrate the relationship between Pao₂/P_brO₂ and SO₂ (Fig. 1). Because cortical microvasculature (and thus, S_mvO₂) is predominantly (≥70%) venous ⁽¹⁶⁾, AVDO₂, CMRO₂, and local microvascular OEF may be calculated as follows:

Figure 1: Systemic, pH_a, and Paco₂-corrected arterial oxygen dissociation curve in the rabbit and corresponding cerebral capillary/venous oxygen dissociation curve obtained from simultaneous assessment of cortical Po₂ (P_brO₂) and hemoglobin saturation in cortical microvasculature (S_mvO₂, representing ≥70% venules) during progressive hypoxia (n = 10; mean ± sd). P₅₀ denotes the systemic arterial (Pao₂) and cortical Po₂ (P_brO₂) values observed at half-maximal hemoglobin saturation in systemic arterial blood (S_aO₂) and cortical microvessels (S_mvO₂). Whereas the slopes of systemic arterial and cerebral microvascular (venular) oxygen dissociation curves differ, P₅₀ remains constant (32.7 mm Hg). Polynomial curve fits were applied to illustrate the relationship between mean Pao₂/P_brO₂ and SO₂. The resulting polynomial equations and multiple correlation coefficients (R²) are plotted within the corresponding graphs.

MATHMATHMATH

Data were expressed as the mean ± sd or as median and range in those instances with substantial deviation from normal distribution of data. Nonparametric analysis of variance (Kruskal-Wallis test) was used to describe differences in distribution among groups. Statistical significance was assumed for P < 0.01. Statistical analysis was performed by using the Statistica 5.0 package (StatSoft, Hamburg, Germany).

Results

CBF, Oxygenation, and Metabolism During Global Ischemia

During variation of CPP between 186 ± 14 to 14 ± 4 mm Hg (baseline: 59 ± 4 mm Hg), l_coCBF (baseline average: 50.4 ± 3.8 mL · 100 g⁻¹ · min⁻¹) ranged from 130 ± 12 to 0 ± 2 mL · 100 g⁻¹ · min⁻¹ (Fig. 2). S_mvO₂ ranged from peak values of 70% ± 4% to 0% ± 1% (baseline average: 54% ± 5%). Similarly, P_brO₂ and tHb concentrations ranged from peak values of 120 ± 8 mm Hg and 6 ± 2 U/mL to 0 ± 1 mm Hg and <3 U/mL (baseline averages: 37 ± 5 mm Hg and 3 ± 1 U/mL). Considerable interindividual CPP variability during cerebral circulatory arrest and corresponding nadirs of P_brO₂ and S_mvO₂ was observed (range, 10–18 mm Hg). Cyt a+a₃ oxidation decreased from 79% ± 4% (hyperperfusion) and 67% ± 4% (baseline average) to <4% during cerebral circulatory arrest (Fig. 2). BEA (SEF₉₅) did not differ significantly between hyperperfusion (11.2 ± 1.2 Hz) and baseline conditions (9.8 ± 1.1 Hz). Loss of BEA (SEF₉₅ <0.4 Hz) preceded cessation of l_coCBF (range, 5–11 mL · 100 g⁻¹ · min⁻¹). The lactate-pyruvate ratio (L/P ratio) and extracellular Glu concentrations increased from 9.6 ± 17.9 (7.6 ± 11.1) μmol/L during hyperperfusion to 17.6 ± 16.4 (18.1 ± 16.7) μmol/L under baseline conditions. Peak values for L/P ratio (877 ± 359) and Glu (673 ± 268) μmol/L were observed during complete global ischemia (l_coCBF <1 mL · 100 g⁻¹ · min⁻¹) (Fig. 3).

Figure 2: Relationship between cerebral perfusion pressure (CPP), local cortical blood flow (l_coCBF), brain oxygen-tension (P_brO₂), cytochrome oxidase redox level (Cyt a+a₃ oxidation), and brain electrical activity (SEF₉₅) during random variation of CPP (n = 10; mean ± sd). Changes in l_coCBF and P_brO₂ show characteristic features of cerebrovascular pressure-flow autoregulation. Brain electrical activity (SEF₉₅) is closely coupled to the state of cellular respiration (Cyt a+a₃ oxidation). Significant differences (P < 0.01) between average baseline values (white dots) and groups of data recorded at different target CPP plateaus (black dots) are indicated by brackets and double asterisks.

Figure 3: Relationship between brain oxygen tension (P_brO₂), microvascular hemoglobin saturation (S_mvO₂), extracellular glutamate concentration (Glu), lactate-pyruvate ratio (L/P ratio), and local cortical blood flow (l_coCBF) during random variation of CPP (n = 10; mean ± sd). Whereas P_brO₂ is almost linearly coupled to l_coCBF, a sigmoid association is observed between S_mvO₂ and cortical perfusion. Extracellular glutamate concentrations and L/P ratio steeply increase beyond critical tissue perfusion levels (l_coCBF ≤20 mL · 100 g⁻¹ · min⁻¹). Significant differences (P < 0.01) between average baseline values (white dots) and groups of data recorded at various target CPP plateaus (black dots) are indicated by brackets and double asterisks.

CBF, Oxygenation, and Metabolism During Systemic Hypoxia

Random variation of Pao₂ between hyperoxic (345 ± 13 mm Hg) and normoxic levels (100 ± 8 mm Hg) yielded corresponding changes in average P_brO₂ between 120 ± 9 (hyperoxia) and 38 ± 4 mm Hg (normoxia). During deep hypoxia (Pao₂: 10 ± 3 mm Hg), P_brO₂ decreased to 3 ± 2 mm Hg. S_mvO₂ varied between 71% ± 5% (peak values during hyperoxia), 55% ± 4% (baseline average), and 0% ± 2% (trough levels during profound hypoxia), along with corresponding changes in l_coCBF from 51 ± 7 (normoxia) to 89 ± 9 mL · 100 g⁻¹ · min⁻¹ (hypoxia). Compensatory increases in l_coCBF >5% of baseline were registered beginning at P_brO₂ levels ≤18 mm Hg (Pao₂ ≤40 mm Hg), whereas tHb (as an estimate of cerebral blood volume) increased from 3 ± 1 (baseline) to 7 U/mL tissue (7 ± 2) as a result of hypoxic vasodilatation. Cyt a+a₃ oxidation decreased from 79% (79% ± 5%; hyperoxia) via 67% (67% ± 5%) at P_brO₂ values of 38 mm Hg (normoxia) to <6% (range, 3%–9%) during profoundly hypoxic conditions (P_brO₂ <3 mm Hg). Correspondingly, SEF₉₅ varied between 11.3 ± 1.5 (hyperoxia), 9.6 ± 1.2 (normoxia), and 0.3 ± 0.6 Hz (hypoxia). Glu (411 ± 159 μmol/L) and L/P ratio (443 ± 179) were markedly increased under severely hypoxic conditions compared with baseline values of 15 ± 14 μmol/L (Glu) and 17 ± 17 (L/P ratio).

Indices of Cerebral Energy State Derived from Multiparametric Monitoring

Baseline AVDO₂ (CMRO₂) ranged from 5.2 to 6.6 mL O₂ · 100 mL⁻¹ (3.9–3.1 mL · 100 g⁻¹ · min⁻¹). During profound ischemia (l_coCBF ≤5 mL · 100 g⁻¹ · min⁻¹), AVDO₂ increased to 16.4 ± 2.6 mL O₂ · 100 mL⁻¹, whereas CMRO₂ approached zero. OEF increased from 0.35–0.45 under baseline conditions (l_coCBF: 51 ± 6 mL · 100 g⁻¹ · min⁻¹) to 0.81, recorded at the ischemic flow threshold of 20 ± 3 mL · 100 g⁻¹ · min⁻¹. Beyond this critical perfusion level, compensatory OEF (>0.86) was exhausted, resulting in a corresponding reduction of CMRO₂ (Fig. 4).

Figure 4: Relationship between local cortical blood flow (l_coCBF), brain oxygen tension (P_brO₂), arteriovenous oxygen difference (AVDO₂), and cerebral metabolic rate for oxygen (CMRO₂) during variable degrees of cerebral ischemia (upper graph) and hypoxia (lower graph) (n = 10; mean values). The hatched boxes designate critical energetic thresholds, indicated by significant suppression oxygen dependent metabolism (CMRO₂) and exhausted compensatory adaptation of either microvascular oxygen extraction (AVDO₂) during ischemia or l_coCBF during hypoxia.

Baseline AVDO₂ (5.8 ± 1.5 mL O₂ · 100 mL⁻¹) and CMRO₂ (3.7 ± 0.6 mL · 100 g⁻¹ · min⁻¹) observed under normoxia (P_brO₂: 36 mm Hg; F_iO₂: 26%) decreased to trough values of <0.5 mL O₂ · 100 mL⁻¹ (AVDO₂) and 0.2 mL · 100 g⁻¹ · min⁻¹ (CMRO₂) during hypoxia, corresponding to P_brO₂ levels <1.5 mm Hg (F_iO₂ below 2%). Whereas OEF was reduced from baseline average levels (35%–40%) to 25%–30% during hyperoxia, it remained constant during hypoxia (P_brO₂ ≤15 mm Hg).

Discussion

Critical cerebral ischemia and hypoxia were assessed by five complementary monitoring modalities (including differential fluid pressure measurements [mean arterial blood pressure, ICP, CPP], vascular variables [l_coCBF, cerebral capillary Hb concentration], evaluation of tissue oxygen supply [P_brO₂, S_mvO₂, Cyt a+a₃ oxidation], and secondary metabolic [Glu, L/P ratio] as well as functional [BEA] effector systems) under defined, reproducible conditions. It is emphasized that our data represent only acute changes in cerebral microcirculation and metabolism, because there are no time-based controls. Assessment of l_coCBF by thermal-diffusion techniques yields quantitative data if true zero calibration values are obtained. This method may not reliably represent the global pattern of cerebral circulation, especially with heterogeneous tissue flow characteristics.

To reduce flow heterogeneity, we induced global ischemia through variation of CPP. l_coCBF values presented in this study seem to reflect global changes of cerebral perfusion with reasonable accuracy and are consistent with data obtained in similar experiments ^{(1,2,8,9,11,13,17,18)}. Sylvia et al. ⁽³⁾ reported a sigmoid relationship between Pao₂ and Cyt a+a₃ oxidation during hypoxia in the rat brain, reaching a critical energetic threshold at Pao₂ of 30 mm Hg, corresponding to 80% Cyt a+a₃ reduction, progressive phosphocreatine, and subsequent adenosine triphosphate depletion as well as a steep increase in the L/P ratio. Cyt a+a₃ redox levels were shown to be linearly related to cerebral phosphocreatine concentrations. These findings correspond favorably with data from recent positron emission tomography studies ⁽¹⁹⁾, indicating that oxidative metabolism is maintained down to arterial Po₂ levels of 30–35 mm Hg. During systemic hypoxia, P_brO₂ values of 10–14 mm Hg (corresponding to Pao₂ levels ranging between 29 and 35 mm Hg) were associated with 75% Cyt a+a₃ reduction. Various experimental data suggest that Cyt a+a₃ is not fully oxidized under physiologic conditions ^(3,8,17,20). We obtained 50% Cyt a+a₃ reduction at P_brO₂ values ranging from 15 to 20 mm Hg (P₅₀ for Cyt a+a₃), corresponding to an estimated mean P_icO₂ of approximately 0.5 mm Hg [averaged mean value based on data extracted from Refs. ^(20–24)]. Comparable P_icO₂ values seem to be adequate for optimal oxidative adenosine triphosphate synthesis in a variety of different species ^(21–23). These very low P_icO₂ values (compared with venous Po₂) indicate that oxy- genated blood may not achieve equilibrium with tissue during its passage through cerebral capillaries ⁽²⁴⁾.

Another important gauge of critical oxygen supply is the P_icO₂, at which a decrease in cellular oxygen utilization is observed. Chance et al. ⁽²⁰⁾ characterized this relationship by assessment of nicotinamide adenine dinucleotide (NADH) fluorescence in rat brain mitochondria as a function of extra- and intracellular Po₂. P_icO₂ values of 0.5 mm Hg, corresponding to a P_brO₂ of 15 mm Hg at an F_iO₂ of 8%, induced a distinct increment in NADH reduction. P_icO₂ of 0.2 mm Hg (corresponding to an F_iO₂ of 4%) just sustained cellular respiration, with 30% to 60% NADH reduction. During anoxia (P_icO₂: 0.06 mm Hg, abolition of cellular respiration), NADH reduction reached 87% of baseline. Similar findings were reported for global ischemia ⁽⁴⁾. These results indicate an extremely efficient use of intracellular oxygen by respiratory enzymes. It was therefore suggested that P_icO₂ predicts cellular energy homeostasis and function ^(21,22). Whereas P_icO₂ is difficult to quantify directly ⁽²⁴⁾, reliable techniques have been devised for continuous assessment of P_brO₂^(10,11), which depends on radial diffusion of oxygen from erythrocyte/plasma via the endothelium to the interstitial space. Duling et al. ⁽¹⁸⁾ have reported a physiologic Po₂ gradient of 0.9 mm Hg/μm distance across the wall of arterioles with luminal diameters of 40–70 μm (wall thickness: 8–15 μm). This gradient is minimal for capillaries and small venules ⁽²⁵⁾. In the extracellular space, Po₂ decreases nonlinearly with increasing distance from the vessel. At radial distances of 30 μm from the vessel, mean Po₂ is approximately 40 mm Hg for small arterioles, 30 mm Hg for capillaries, and 25 mm Hg for venules ⁽²⁵⁾. Under physiologic conditions, polarographic oxygen microelectrodes integrate P_brO₂ from all cortical microvessels (mean P_brO₂ during normoperfusion and normoxygenation in this study: 30 mm Hg). In agreement with recent data indicating the venous fraction within cortical microvasculature to exceed 70%⁽¹⁶⁾, it seems tenable to suggest that P_brO₂ predominantly reflects venous Po₂. However, this is contrasted by the diverse relationships between P_brO₂, OEF, and l_coCBF observed during our experiments, which may in part be explained by the fact that P_brO₂ measurements are significantly influenced by the relative predominance of venous or arterial vessels present in close vicinity to the catheter. Arteriolar predominance will result in coupling between P_brO₂ and CBF, whereas a purely venous environment will produce close correspondence to OEF (AVDO₂). During clinical monitoring, both components will overlap, thus rendering results widely unpredictable without additional information on either corresponding CBF or OEF (AVDO₂).