Functional MR imaging is a technique that is used to map or localize brain areas involved in a particular task. The approach is based on a technique known as BOLD contrast.1 The BOLD effect is attributable to the magnetic properties of hemoglobin, which vary depending upon its state of oxygenation. When diamagnetic oxyhemoglobin releases the O2, resulting in paramagnetic deoxyhemoglobin, a local magnetic field distortion occurs that changes the proton relaxation behavior inside and around blood vessels in gradient-echo MR images.2 Thus, in T2*-weighted images, signal intensity decreases when there is an increase in deoxyhemoglobin content of blood and increases when there is an increase in oxyhemoglobin content of blood.2 Cerebral venous blood oxygenation, and consequently the intensity of the BOLD signal, depends on the balance between the demand for O2 or the CMRO2 and the supply of O2 or CBF—processes of supply and demand that are coupled in healthy brains. When the level of neuronal activity in a region of brain increases, so does the CMRO2; therefore, the CBF to that region of the brain increases. However, a large CBF increase is required to support a modest increase in CMRO2, which creates a disparity between O2 delivery and O2 utilization and increases the amount of O2 in cerebral venous blood.3 This is the basis of studies3,4 involving stimulus-evoked fMR imaging of the brain, in which relative local increases in BOLD signal are measured to estimate relative local increases in CBF in response to increased neuronal activity. The greater the increase in CBF for any increase in CMRO2, the greater the BOLD signal intensity and vice versa.5
Functional MR imaging must be performed in anesthetized animals. Presently, maintenance of anesthesia in animals is usually achieved by use of an inhalant anesthetic agent delivered in 100% O2. During anesthesia, animals breathe spontaneously or are mechanically ventilated. Therefore, an anesthetized animal rapidly becomes hyperoxemic; PaO2 values may become as high as 660 mm Hg at sea level, and depending on the type of ventilation, the PaCO2 value can be highly variable.6 During spontaneous breathing, an anesthetized animal typically develops hypercapnia because of respiratory depression induced by most general anesthetics,6 whereas during mechanical ventilation, PaCO2 can be controlled.
Global CBF changes that result from PaCO2 changes (even PaCO2 changes that are within the physiologic range) strongly influence BOLD MR image contrast during fMR imaging in conscious humans.7 When the PaCO2 increases, there is an increase in CBF, which provides a greater supply of O2 to the brain; in the absence of a change in CMRO2, the level of venous blood oxygenation and BOLD signal increase.2 In contrast, in conditions of hypocapnia, the CBF decreases and the concentration of deoxyhemoglobin in venous blood increases, thereby decreasing the BOLD signal.8 When conscious humans breathe 100% O2, the BOLD signal increases slightly because of the increased concentration of oxyhemoglobin in the venous blood, despite the known decrease in CBF induced by hyperoxemia.9,10
It is evident that varying PaCO2 and PaO2 values could potentially interfere with fMR imaging. The potential interference with fMR imaging holds especially true in anesthetized animals, in which cerebrovascular responses may be altered by anesthetic drugs and arterial concentrations of CO2 and O2 may be different from values obtained in animals that are awake and breathing room air. A few studies11,12 have investigated the degree of BOLD signal changes in the brain with varying concentrations of arterial CO2 during anesthesia, but those experiments were performed in rats. To our knowledge, there are no reports of BOLD signal changes in various regions of the brains of anesthetized dogs in which arterial CO2 and O2 concentrations were independently controlled, yet such information would be useful for fMR imaging procedures in anesthetized dogs.
The purpose of the study reported here was to assess the effects of alterations in PaCO2 and PaO2 on BOLD signal intensity in different regions of the brain determined by use of SW imaging (a BOLD-sensitive method) in isoflurane-anesthetized dogs. We undertook to quantify changes in BOLD signal intensity during eucapnia and moderate hypocapnia and hypercapnia in combination with normoxemia and hyperoxemia in healthy dogs anesthetized with isoflurane. The hypothesis was that differences in BOLD signal intensity would be detected at different target values of PaO2 and PaCO2.
Materials and Methods
Animals and study design—Six healthy adult castrated male mixed-breed dogs were included in the study (mean ± SD age, 2.3 ± 0.5 years; weight, 28.5 ± 4.7 kg). Health status was determined on the basis of results of a physical examination, a CBC, and serum biochemical analyses. Food but not water was withheld for 12 hours prior to anesthesia. Dogs were housed in individual runs in the Central Animal Facility building of the University of Guelph and were taken to the MR imaging unit of the Ontario Veterinary College on the morning of the experiment day. Each dog was anesthetized only once. After complete recovery from anesthesia, each dog was returned to its run on the evening of the experiment day. The Animal Care Committee of the University of Guelph approved the procedures and the experimental design of the study.
A split plot design (based on a modified Latin square design for randomization) was used. During 1 MR imaging session in each dog, targeted values of PaCO2 (20, 40, or 80 mm Hg) and PaO2 (100 or 500 mm Hg) were combined to establish 6 experimental conditions as follows: hc-no (PaCO2, 20 mm Hg; PaO2, 100 mm Hg), hc-Ho (PaCO2, 20 mm Hg; PaO2, 500 mm Hg), ec-no (PaCO2, 40 mm Hg; PaO2, 100 mm Hg [control treatment]), ec-Ho (PaCO2, 40 mm Hg; PaO2, 500 mm Hg), Hc-no (PaCO2, 80 mm Hg; PaO2, 100 mm Hg), and Hc-Ho (PaCO2, 80 mm Hg; PaO2, 500 mm Hg). Each dog was randomly assigned to a different sequence of experimental conditions.
To achieve the 2 PaO2 target values, the carrier gas administered was switched from 23% O2 for normoxemia (PaO2, 100 mm Hg) to 100% O2 for hyperoxemia (PaO2, 500 mm Hg). To achieve the 3 PaCO2 target values, a Bain anesthetic circuit was used for hypercapnia (PaCO2, 80 mm Hg) with fresh gas flows ranging from 50 to 130 mL/kg/min to permit rebreathing of CO2 and a circle circuit was used for eucapnia (PaCO2, 40 mm Hg) and hypocapnia (PaCO2, 20 mm Hg) with fresh gas flows ranging from 100 to 400 mL/kg/min. The tidal volume and respiratory rate were adjusted to induce hypoventilation for hypercapnia, normoventilation for eucapnia, and hyperventilation for hypocapnia. When the target values were reached (determined on the basis of arterial blood gas analysis results), a period of ≥ 5 minutes was allowed to elapse for stabilization of the experimental condition before MR imaging was commenced.
Anesthesia and monitoring—A 20-gauge catheter was placed in a cephalic vein in each dog on the morning of the experiment day. Anesthesia was induced via IV administration of propofola (6 to 8 mg/kg [titrated to effect]). Once palpebral reflexes, swallowing reflex, and jaw tone were no longer detectable, each dog was intubated and the endotracheal tube was connected to a Bain circuit. Isofluraneb was administered at a constant end-tidal concentration of 1.7% (approx 1.3 × MAC of isoflurane in dogs)13 throughout the experimental period; initially, isoflurane was delivered in 100% O2 at a fresh gas flow of 200 mL/kg/min. Mechanical ventilation was initiated immediately after intubation with a volume ventilatorc that was set to initially provide a tidal volume of 15 mL/kg and a respiratory rate of 10 breaths/min. After a dog was anesthetized and before transfer to the MR imaging scanner, a 20-gauge catheter was placed in a dorsal pedal artery for continuous blood pressure monitoring (ie, systolic, diastolic, and mean arterial blood pressures).
Each dog was positioned in sternal recumbency for the brain MR imaging. A fiber optic pulse-oximeterd was placed on the tongue for continuous monitoring of heart rate and SpO2. A multiparameter monitore was used to continuously display systolic, diastolic, and mean arterial blood pressures; heart rate; and inspired and expired concentration of gases. Arterial blood pressures were monitored with a pressure transducer positioned at the level of the right atrium and zeroed at this level before every pressure recording. Side-stream capnography was used to measure inspired O2 concentration, end-tidal CO2 concentration, and end-tidal isoflurane concentration. The sampling line was positioned between the proximal end of the endotracheal tube and the anesthetic circuit and was advanced 10 cm distally through the endotracheal tube (sampling rate, 200 mL/min). The capnograph was calibrated with commercial calibration gases prior to use on the morning of each experimental day. Rectal temperature was intermittently monitored by use of a digital thermometer.f If the rectal temperature reading was > 38.5°C, a fan within the MR imaging scanner was turned on; if the rectal temperature reading was < 37°C, active warming with warm oat bags and blankets was initiated.
For replacement of fluid losses during anesthesia, a crystalloid solutiong was administered IV at a rate of 5 mL/kg/h. Dopamineh (5 to 7 μg/kg/min) was administered as needed to maintain MAP at 90 to 100 mm Hg. Atracuriumi was administered IV at a dose of 0.2 mg/kg before the MR imaging scan was started, and the same dose was repeated every 30 minutes to avoid respiratory efforts or any other movement of the dog during the experimental period. At the end of the experimental period, glycopyrrolatej (5 μg/kg, IV, once) was administered followed 5 minutes later by administration of edrophoniumk (0.2 mg/kg, IV, once) to reverse any residual effects of atracurium. Each dog was extubated when spontaneous breathing recommenced and a swallowing reflex was detectable. The duration of anesthesia (the interval from induction until extubation) was recorded for each dog.
Immediately before and every 10 minutes during the MR imaging procedure for each experimental condition, physiologic variables including heart rate, SpO2, arterial blood pressures, and end-tidal CO2 and isoflurane concentrations were recorded. Arterial blood samples (2 mL each) were collected and rectal temperature was measured immediately before and after each MR imaging procedure for each experimental condition. The arterial blood samples were immediately analyzed for pH, PaCO2, PaO2, bicarbonate concentration, and base excess by use of a blood gas analyzerl with a correction for rectal temperature. The arterial O2 content of each blood sample was also measured by use of an oximeter.m Hematocrit and total protein concentration of each arterial blood sample were also assessed.
Image acquisition and processing—A 1.5-T MR imaging scannern with a quadrature transmit-receive knee coil was used for imaging of the brain. For each dog during each of the 6 experimental conditions, T1-weighted and SW images were obtained. A 3-D T1-weighted sequence in the transverse plane was acquired first for each condition and was used for localization of anatomic structures (repetition time, 10.6 milliseconds; echo time, 4.7 milliseconds; inversion time, 175 milliseconds; flip angle, 30°; field of view, 20 cm; slice thickness, 1 mm; number of slices, 94; and matrix size, 256 × 160). High-resolution SW images (a T2*-weighted sequence) were also obtained in the transverse plane by use of a modified 3-D time-of-flight pulse sequence (repetition time, 54 milliseconds; echo time, 38 milliseconds; flip angle, 35°; field of view, 20 cm; slice acquisition thickness, 1.4 mm; number of slices, 68; matrix size, 256 × 192; receiver bandwidth, 31 kHz; and final voxel size, 0.78 × 1.04 × 0.7 mm).
The SI was measured in 3-D ROIs in the SW images by use of a dedicated software program.o The selected ROIs were defined manually by referring to a veterinary anatomy atlas14 and comparing the SW images with their corresponding T1-weighted images. The ROIs were rectangular, and the size of each ROI was maintained constant in all images obtained from a given dog. The ROIs included the thalamus (left and right sides), diencephalic gray matter (left and right sides), diencephalic white matter (left and right sides), medulla oblongata, cerebellar gray matter, and cerebellar white matter (left and right sides; Figure 1). The 3-D ROIs extended over 3 consecutive images, and their location in each image was assessed to ensure that they contained only the desired tissue. The histogram, kurtosis, and skewness of each individual 3-D ROI were obtained to determine the distribution of the SI within the sample. Because the distribution of SI in some ROIs was not normal, the median SI was obtained for each 3-D ROI over each of the 3 consecutive images and then those median values were averaged to calculate the mean of the median values, which was then used for statistical analysis. For each experimental condition within each dog, the PSIC value was obtained by use of a formula as follows:
where SIT is the mean of the median SIs for the condition of interest and SIC is the mean of the median SIs for the control condition (ec-no).
Representative SW (A, C, and E) and correspondingT1-weighted (B, D, and F) images of the brain of a dog obtained at the level of the thalamus (A, B), medulla oblongata (C, D), and cerebellum (E, F). In the SW images, ROIs are indicated by numbers as follows: 1, diencephalic white matter (right and left sides); 2, thalamus (right and left sides); 3, diencephalic gray matter (right and left sides); 4, medulla oblongata; 5, cerebellar gray matter; and 6, cerebellar white matter (right and left sides). In all images, dorsal is toward the top of the image and left side of the brain is toward the right as indicated in panel A.
Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.24
Representative SW (A, C, and E) and correspondingT1-weighted (B, D, and F) images of the brain of a dog obtained at the level of the thalamus (A, B), medulla oblongata (C, D), and cerebellum (E, F). In the SW images, ROIs are indicated by numbers as follows: 1, diencephalic white matter (right and left sides); 2, thalamus (right and left sides); 3, diencephalic gray matter (right and left sides); 4, medulla oblongata; 5, cerebellar gray matter; and 6, cerebellar white matter (right and left sides). In all images, dorsal is toward the top of the image and left side of the brain is toward the right as indicated in panel A.
Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.24
Representative SW (A, C, and E) and correspondingT1-weighted (B, D, and F) images of the brain of a dog obtained at the level of the thalamus (A, B), medulla oblongata (C, D), and cerebellum (E, F). In the SW images, ROIs are indicated by numbers as follows: 1, diencephalic white matter (right and left sides); 2, thalamus (right and left sides); 3, diencephalic gray matter (right and left sides); 4, medulla oblongata; 5, cerebellar gray matter; and 6, cerebellar white matter (right and left sides). In all images, dorsal is toward the top of the image and left side of the brain is toward the right as indicated in panel A.
Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.24
The relative percentage of SI variation between all condition pairs was obtained by subtracting the PSIC of one treatment from the PSIC of the other. The quality of the SW images was assessed by calculating the SNR. The SNR was calculated for each individual ROI by use of the mean of the median SIs of the ROI over 3 consecutive images divided by the SD of the background noise measured in a large image area external to the brain (also over 3 consecutive images), taking care to avoid artifacts. The contrast of the SW images was assessed by measuring the contrast-to-noise ratio, which was obtained by subtracting the SNR of the diencephalic white matter from the SNR of the thalamus.
Statistical analysis—Statistical analysis was performed by use of computer software.p A Shapiro-Wilk test was performed on all analyzed variables to test for data normality. The values of the physiologic variables (heart rate; SpO2; end-tidal CO2 and isoflurane concentrations; MAP; PaO2; PaCO2; arterial blood pH, O2 content, bicarbonate concentration, base excess, Hct, and total protein concentration; and rectal temperature) were analyzed by use of an ANOVA for repeated measures, including dog as random effect and period and carryover as fixed effects in the model. Multiple paired t tests with Tukey-Kramer adjustment were performed for comparisons between treatments. For each treatment, the mean dose of dopamine used during the MR imaging was calculated and analyzed by use of an ANOVA. Tukey-Kramer adjustment was performed post hoc for dopamine dose differences between treatments. Analysis of covariance was performed on the mean of the median SIs of the ROIs. Random effects of dog and period and their interaction, as well as fixed effects of treatment, body side, and carryover and their interaction, were included in the model. The mean background noise was included as a covariable in the model. The SD of the background noise was also included in the model as a weighing variable. If the effect of body side was not significant, the SIs of both sides were averaged and used for further analysis. By use of the mean values of the median SIs, multiple pairwise comparisons were made with paired t tests between pairs of the 6 experimental conditions for each ROI within each dog, and if significance was found for a pair, post hoc Tukey-Kramer adjustment was performed. A value of P ≤ 0.05 was considered significant. Simple descriptive statistics were performed on the SNRs of each ROI and the contrast-to-noise ratio of the SW images. For each ROI, the mean SNR was calculated by averaging the values for all conditions.
Results
All results are reported as mean ± SD. The mean duration of anesthesia (from induction until extubation) was 385 ± 21 minutes. None of the dogs developed complications during the experimental period, and all recovered uneventfully from anesthesia. The mean total MR imaging scan time for all experimental conditions was 25 ± 1 minutes. Minor differences in physiologic variables were detected among conditions (Table 1). The mean ± SD dopamine dose used for all dogs during all conditions was 2.04 ± 1.3 μg/kg/min, and there were no significant differences in the dose used among conditions. There was no significant difference in SI between brain sides for any of the ROIs that were measured bilaterally.
Physiologic variables (mean ± SD) measured during MR imaging in 6 isoflurane-anesthetized dogs in which 6 experimental conditions (ec-no [control condition], ec-Ho, hc-no, hc-Ho, Hc-no, and Hc-Ho) were established in random sequence.
| Variable | Experimental condition | |||||
|---|---|---|---|---|---|---|
| hc-no | hc-Ho | ec-no | ec-Ho | Hc-no | Hc-Ho | |
| PaCO2 (mm Hg) | 21.6 ± 1.5a | 21.5 ± 1.3a | 41.7 ± 1.8b | 42.4 ± 0.9b | 81.5 ± 4.7c | 80.0 ± 3.7c |
| PETCO2 (mm Hg) | 20.5 ± 0.7a | 21 ± 1.2a | 43.4 ± 2.3b | 43.9 ± 2.7b | 86.2 ± 3.1c | 88 ± 3.1c |
| PaO2 (mm Hg) | 143.7 ± 14.3a | 526.3 ± 36.4b | 133.7 ± 15.8a | 513.8 ± 24.8b,d | 104.2 ± 16.7c | 488.8 ± 15.9d |
| CaO2 (mL of O2/dL of blood) | 18.7 ± 1.4a,b | 19.1 ± 1a,b | 18 ± 2a | 18.5 ± 1.5a,b | 19.4 ± 1.8b | 19.7 ± 1.1b |
| SpO2 (%) | 98.4 ± 0.9a | 98.8 ± 0.9a | 96.6 ± 0.9b | 99 ± 0.6a | 94.2 ± 0.9c | 98.9 ± 0.8a |
| Heart rate (beats/min) | 133.8 ± 15.1a | 130.4 ± 18.1a | 137.4 ± 12.4a | 133.8 ± 12.6a | 126.5 ± 9.7a | 125.3 ± 10.3a |
| MAP (mm Hg) | 98.7 ± 12.9a | 105.9 ± 17.5a | 103.9 ± 15.4a | 100.3 ± 13.1a | 91 ± 10.7a | 92 ± 8.7a |
| pH | 7.539 ± 0.04a | 7.536 ± 0.05a | 7.341 ± 0.026b | 7.331 ± 0.037b | 7.126 ± 0.029c | 7.128 ± 0.028c |
| Bicarbonate (mmol/L) | 18.2 ± 1.9a | 18.0 ± 1.6a | 21.6 ± 1.5b | 21.9 ± 1.7b | 24.9 ± 1.6c | 24.8 ± 1.6c |
| Base excess (mmol/L) | −1.9 ± 2.3a | −1.7 ± 2.8a,d | −3.1 ± 1.6b | −2.8 ± 1.9b,d | −6.1 ± 1.7c | −6.0 ± 1.6c |
| Hct (%) | 38.5 ± 3.4a | 39.3 ± 2.3a,c | 37.6 ± 3.8a | 38.7 ± 3.5a | 42.7 ± 3.6b | 41.4 ± 2.8b,c |
| Total protein (g/dL) | 5.6 ± 0.3a,b | 5.6 ± 0.3a | 5.6 ± 0.3a,b | 5.6 ± 0.2a | 5.8 ± 0.3b | 5.6 ± 0.3a |
| Temperature (°C) | 37.9 ± 0.4a | 38.1 ± 0.6a,c | 38.1 ± 0.6a | 38.1 ± 0.6a,c | 38.4 ± 0.9b | 38.3 ± 0.8b,c |
The experimental conditions were based on 6 combinations of targeted values of PaCO2 (20 [hypocapnia], 40 [eucapnia], or 80 [hypercapnia] mm Hg) and PaO2 (100 [normoxemia] or 500 [hyperoxemia] mm Hg). For each condition, each value represents the mean of measurements obtained immediately before and after the 25-minute MR imaging scan (PaCO2; PaO2; arterial blood O2 content [CaO2], pH, bicarbonate concentration, base excess, Hct, and total protein concentration; and rectal temperature) or measurements obtained at 10-minute intervals (PETCO2, SpO2, heart rate, and MAP).
Within a variable, values with different superscript letters differ significantly (P > 0.05).
No significant differences in mean SI were detected between the control condition and any of the other conditions. In the thalamus, the SI was significantly greater in dogs undergoing ec-Ho (14 ± 9%), Hc-no (8.7 ± 7%), and Hc-Ho (14.8 ± 10%), compared with findings in dogs undergoing hc-no (P = 0.002, 0.047, and 0.003, respectively), and in dogs undergoing ec-Ho (9 ± 4%) and Hc-Ho (9.8 ± 4%), compared with findings in dogs undergoing hc-Ho (both P < 0.001). In the diencephalic gray matter, the SI was significantly greater in dogs undergoing ec-Ho (13.6 ± 9%), Hc-no (8 ± 7%), and Hc-Ho (13.8 ± 10%), compared with findings in dogs undergoing hc-no (P = 0.011, 0.047, and 0.022, respectively), and in dogs undergoing ec-Ho (5.5 ± 4%), compared with findings in dogs undergoing hc-Ho (P = 0.046). In the diencephalic white matter, the SI was significantly greater in dogs undergoing ec-Ho, compared with findings in dogs undergoing hc-no (8.2 ± 7%) and hc-Ho (4.4 ± 3%; P = 0.011 and 0.024, respectively). In the medulla oblongata, cerebellar gray matter, and cerebellar white matter, the SI did not differ significantly among experimental conditions. The PSIC was calculated for all conditions (Figure 2).
Mean ± SD PSIC of each of 6 ROIs in the brains of 6 isoflurane-anesthetized dogs assessed during each of 6 experimental conditions (ec-no [control condition], ec-Ho, hc-no, hc-Ho, Hc-no, and Hc-Ho). The experimental conditions were based on 6 combinations of targeted values of PaCO2 (20 [hypocapnia], 40 [eucapnia], or 80 [hypercapnia] mm Hg) and PaO2 (100 [normoxemia] or 500 [hyperoxemia] mm Hg). Each dog was randomly assigned to a different sequence of experimental conditions. *Value is significantly (P ≤ 0.05) different from the value during hc-no. †Value is significantly (P ≤ 0.05) different from the value during hc-Ho.
Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.24
Mean ± SD PSIC of each of 6 ROIs in the brains of 6 isoflurane-anesthetized dogs assessed during each of 6 experimental conditions (ec-no [control condition], ec-Ho, hc-no, hc-Ho, Hc-no, and Hc-Ho). The experimental conditions were based on 6 combinations of targeted values of PaCO2 (20 [hypocapnia], 40 [eucapnia], or 80 [hypercapnia] mm Hg) and PaO2 (100 [normoxemia] or 500 [hyperoxemia] mm Hg). Each dog was randomly assigned to a different sequence of experimental conditions. *Value is significantly (P ≤ 0.05) different from the value during hc-no. †Value is significantly (P ≤ 0.05) different from the value during hc-Ho.
Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.24
Mean ± SD PSIC of each of 6 ROIs in the brains of 6 isoflurane-anesthetized dogs assessed during each of 6 experimental conditions (ec-no [control condition], ec-Ho, hc-no, hc-Ho, Hc-no, and Hc-Ho). The experimental conditions were based on 6 combinations of targeted values of PaCO2 (20 [hypocapnia], 40 [eucapnia], or 80 [hypercapnia] mm Hg) and PaO2 (100 [normoxemia] or 500 [hyperoxemia] mm Hg). Each dog was randomly assigned to a different sequence of experimental conditions. *Value is significantly (P ≤ 0.05) different from the value during hc-no. †Value is significantly (P ≤ 0.05) different from the value during hc-Ho.
Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.24
The mean ± SD of the SNRs for individual ROIs were as follows: thalamus, 27.9 ± 2.8; diencephalic gray matter, 27.5 ± 3.1; diencephalic white matter, 23.7 ± 2.1; medulla oblongata, 28.5 ± 3.1; cerebellar gray matter, 30.4 ± 2.9; and cerebellar white matter, 27.7 ± 2.3. The mean ± SD contrast-to-noise ratio was 4.2 ± 1.4.
Discussion
In healthy dogs anesthetized with isoflurane, the SI in SW images of the regions of the thalamus and diencephalic gray matter significantly decreased during all conditions that involved hypocapnia, compared with findings during conditions that involved hypercapnia. These changes may indirectly reflect the fact that greater changes in CBF and oxygenation occur in these regions of the brain, at least in response to the values of PaCO2 and PaO2 that were obtained in the present study in isoflurane-anesthetized dogs.
Regulation of cerebral circulation involves several factors including chemical factors such as CO2, hydrogen ion concentration, and O2, all of which affect cerebral vascular tone; autoregulation that maintains a constant CBF in the face of changes in MAP between 60 and 140 mm Hg; changes in CMRO2, which are influenced by neuronal activity, body temperature, or anesthetic agents; and neurogenic control that occurs via sympathetic innervation.15 To study the possible effects that different concentrations of CO2 and O2 could have on the BOLD SI (which indirectly indicates changes in CBF and cerebral venous blood oxygenation), the present study was designed to control other factors that could influence the CBF.
Isoflurane-induced depression of the CNS decreases the CMRO2, although the CBF increases because of direct vasodilatory effects (uncoupling of CMRO2 and CBF),16 and impairs cerebral autoregulation in a dose-dependent manner.17,18 In dogs, cerebral autoregulation is preserved during administration of 1.4% isoflurane (approx 1 × MAC) but not during administration of 2.8% isoflurane (approx 2 × MAC).18 Similarly in rats, cerebral autoregulation is maintained during isoflurane anesthesia at 1 × MAC and attenuated during isoflurane anesthesia at 2 × MAC.17 In humans, static and dynamic cerebral autoregulation are impaired during isoflurane anesthesia at 1.5 × MAC.19 In the dogs of the present study, it was possible that cerebral autoregulation was impaired to some extent because end-tidal isoflurane concentration was maintained at 1.7% (1.3 × MAC).13 Therefore, both the end-tidal isoflurane concentration and MAP were maintained constant throughout the MR imaging scans in all dogs to avoid any possible change in CBF in relation to these factors. Dopamine was administered as needed to maintain MAP. There are conflicting reports on the effects of dopamine infusions on CBF during isoflurane anesthesia. In sheep anesthetized with 2% isoflurane, a significant increase in CBF is evident during a high-dose infusion of dopamine (40 μg/kg/min) but not at lower doses.20,21 A study22 in dogs anesthetized with ketamine and pentobarbital, paralyzed with gallamine triethiodide, and mechanically ventilated with 65% to 70% nitrous oxide mixed with 30% to 35% O2 revealed that the changes in CBF induced by dopamine infusion were dose dependent, with an increase in CBF induced by moderate doses (2 to 6 μg/kg/min). In the present study, the dose of dopamine administered did not differ significantly among experimental conditions; thus, the differences in BOLD signal observed among conditions were unlikely a result of dopamine-induced changes in CBF.
In the present study, the depth of anesthesia was kept constant and the dogs were not exposed to any type of stimulation to avoid changes in CMRO2. Body temperature also influences CMRO2, with hyperthermia increasing and hypothermia decreasing cerebral O2 demands and CBF23–25; as a result, body temperature can also alter the BOLD contrast. Rectal temperature was slightly higher in dogs during conditions involving hypercapnia in the present study, which was attributed to a greater conservation of heat as a result of the forced reinhalation of expired gases to increase the PaCO2. Nevertheless, the mean temperature was within the physiologic range for dogs during all conditions (38.4 ± 0.6°C)26; some differences in rectal temperature were significant, but the maximum difference detected among conditions was of the order of 0.5°C, which might have induced a difference in CMRO2 of approximately 3%.24 This difference in CMRO2 was likely too small to have notably influenced the BOLD signal. Consequently, the changes in SI were unlikely affected by any factor apart from differences in PaCO2 and PaO2.
The interval that was allowed to elapse after the target PaCO2 and PaO2 values were achieved before commencement of the MR imaging was 5 minutes. The duration of this interval should have been sufficient for the cerebrovascular responses to occur and for stabilization of the BOLD signal; in a previous study11 in rats anesthetized with isoflurane, the response to hypercapnia was more immediate in the conscious condition than in the anesthetized state, with BOLD signal risetimes twice as fast in the conscious state (< 30 and 60 seconds, respectively). Nonetheless, T1-weighted images were obtained first for each condition in the present study, a process that took approximately 10 minutes to complete; hence, the time allowed for stabilization of the cerebrovascular responses before SW images were obtained for each condition was approximately 15 minutes.
In the present study, the magnitude of increase in BOLD SI (compared with control condition findings) was similar during ec-Ho and during Hc-Ho, which suggested that the increase in SI was not attributable to an increase in CBF that was induced by CO2, but probably a result of an increase in venous cerebral blood oxygenation induced by higher PaO2. This hypothesis was further supported by the fact that no change in SI was observed during Hc-no in any ROI. However, the decreases in SI observed in the thalamus and diencephalic gray matter during both conditions involving hypocapnia cannot be explained solely by changes in PaO2 because the magnitude of the decrease associated with normoxemia and with hyperoxemia was similar. Therefore, the decrease in SI during conditions involving hypocapnia in these regions of the brain was probably a result of decreased CBF that was induced by low PaCO2. On the basis of these data, it appears that the cerebrovascular response to high PaCO2 (cerebral vasodilation) was blunted or absent but that the cerebrovascular response to low PaCO2 (cerebral vasoconstriction) was still present—at least in the thalamus and diencephalic gray matter—in dogs anesthetized with isoflurane. In a study27 in sheep, the cerebrovascular reactivity to CO2 was maintained during anesthesia with 2% isoflurane. In dogs anesthetized with 1.4% isoflurane, the cerebral vasculature constricts during hypocapnia and dilates during hypercapnia, whereas with 2.8% isoflurane, vasoconstriction in response to hypocapnia was retained but vasodilation in response to hypercapnia was absent.28 The decrease in cerebrovascular responsiveness to hypercapnia during isoflurane anesthesia could be attributed to the direct arteriolar vasodilation induced by isoflurane,29 which is dose dependent30,31 and might impair further vasodilation induced by CO2. Hence, it is possible that at the dose of isoflurane used in the present study, the cerebrovascular response to hypercapnia was present but it yielded a small BOLD signal increase that could not be detected or that it was absent.
The effects of different anesthetic protocols on fMR imaging in dogs has been investigated by use of a visual stimulus.32 In that study, dogs were administered isoflurane at an end-tidal concentration of 1.5 ± 0.2%, propofol via continuous IV infusion, or a combination of fentanyl and midazolam via continuous IV infusion; for the 3 protocols, the mean percentage change in BOLD signal was 0.3% to 1.1%. The variation in BOLD signal in images obtained during isoflurane anesthesia was reduced, compared with the variation in images obtained during anesthesia with the other 2 protocols. In the present study, PSIC ranged from 0.03% to 8% (greater or less with respect to the control condition findings) with different combinations of PaCO2 and PaO2 in different regions of the brain of dogs that did not receive any stimulation. These results indicated that BOLD signal changes of similar magnitude to those obtained in fMR imaging studies in anesthetized animals could be obtained by varying PaCO2 and PaO2 values. Therefore, PaCO2 and PaO2 should be controlled and taken into account when conducting MR imaging that involves BOLD signal analysis.
The overall metabolic rate of gray matter is approximately 4 times as great as that of white matter; correspondingly, the number of capillaries and rate of blood flow in gray matter are also approximately 4 times as great.15 This is in accordance with the results of the present study, in which there were consistently greater BOLD SIs in the ROIs composed of gray matter (ie, thalamus, diencephalic gray matter, and cerebellar gray matter) than the SIs in the ROIs composed of white matter (ie, medulla oblongata, diencephalic white matter, and cerebellar white matter). This indicated that there are greater amounts of oxyhemoglobin and therefore greater blood perfusion to those areas of gray matter in the brains of isoflurane-anesthetized dogs. It is also known that there are differences in vascular responses to O2 and CO2 between gray and white matter. Gray matter has a greater vasoconstrictive response to hyperoxemia and a greater vasodilatory response to hypercapnia than does white matter, as determined via continuous arterial spin-labeled-perfusion MR imaging.33 In the present study, we did not compare SIs in areas of gray matter with findings in areas of white matter, but the relative changes in SI during hypercapnia appeared similar in all regions of the brain. During hypocapnia, however, it appears that the areas of gray matter had relative greater decreases in SI than did the areas of white matter. In a study34 in which fMR imaging was performed in conscious human volunteers, the increases in BOLD signal in response to 2 levels of hypercapnia (46 mm Hg and 53 mm Hg) were greater in cortical versus central gray matter and nonsignificant in white matter. Relative changes in BOLD SI among experimental conditions were more pronounced in the thalamus and diencephalic gray matter than in other regions of the brain in the dogs of the present study, which indirectly indicated that relatively greater changes in CBF and venous blood oxygenation may also occur in areas of gray matter in isoflurane-anesthetized dogs in response to changes in PaO2 and PaCO2. The relative SI changes in cortical and central areas of gray matter were highly similar in the present study in contrast to findings of the study34 performed in conscious humans. This discrepancy could be a consequence of global isoflurane-induced depression of the CNS, especially of cortical regions responsible for consciousness, or of a different regional vasodilatory effect of isoflurane in the dogs of our study. It could also be attributable to a lack of sensitivity of the methods used in the present study for detection of small SI changes.
In rats, cerebrovascular reactivity to hypercapnia is reduced during isoflurane anesthesia, compared with the response to hypercapnia in the conscious state.11,12 In rats that are anesthetized with 2% isoflurane and breathing spontaneously, the BOLD SI increases by 5.4% and 7.2% in the entire brain,12 by 3.2% and 4.9% in cortical regions,11 and by 2.4% and 2.7% in subcortical regions11 during inhalation of 5% and 10% of CO2 mixed with air, respectively. These increases in BOLD signal were of lesser magnitude than those observed in awake animals under similar breathing conditions.11,12 In the present study, variations in the BOLD signal during conditions involving hypercapnia, compared with control condition findings, were not detected in any examined region of the brain. This discrepancy between results of the present and previous studies could reflect the greater magnetic field strength used in the studies in rats (4.7 T), compared with that used in our study (1.5 T), or the use of a different type of MR imaging sequence, voxel size, number of acquisitions, coil, or bandwidth. All these factors may affect SI and SNR. The SNR has a direct impact on the degree of BOLD signal change that can be detected in fMR imaging studies35 and most likely also in investigations that use SW imaging. Signal-to-noise ratios in the present study ranged from 23.7 to 30.4 in all regions of the brain, which probably allowed a high detection rate of BOLD signal changes ≥ 5%.35 Therefore, it is possible that increases in the BOLD SI < 5% in dogs during hypercapnia were not detected in our study.
In conscious human volunteers, breathing 100% O2 decreases CBF from 9% to 31%, compared with CBF in individuals breathing 21% O2.36 This decrease in CBF induced by hyperoxemia is attributed to 2 distinct mechanisms: indirectly, via induction of hyperventilation in conscious individuals, which decreases PaCO2, and directly, via induction of cerebral vasoconstriction.33,36,37 Most MR imaging studies to investigate the effects of hyperoxemia on the BOLD signal have been performed in conscious human volunteers9,37–39 or in anesthetized rats that are breathing spontaneously,39 making it impossible to separate the relative effects induced by hypocapnia from those induced by hyperoxemia. In a recent study10 in human volunteers, the effects on the brain BOLD MR imaging signal of end-tidal CO2 concentration were separated from the effects of endtidal O2 concentration by use of a method of sequential gas delivery. However, in that study, conscious individuals were investigated and only mild hypercapnic conditions (end-tidal CO2 concentration, 40 mm Hg) were achieved. In the present study, dogs were anesthetized, mechanically ventilated, and paralyzed with a neuromuscular blocking agent and PaO2 and PaCO2 were independently controlled at target values; thus, it was possible to assess the effects of each of these 2 gases on the BOLD SI. At the same target values of PaCO2, there was an increase in SW imaging signal during hyperoxemia, compared with findings during normoxemia, which suggested that an increase in cerebral venous blood oxygenation occurred as determined in previous studies10,38 in humans. However, the increase in the present study was not significant, perhaps because of a low sensitivity of the study method to detect SI changes < 5%, as discussed previously. Another possibility is that hyperoxemia induced cerebral vasoconstriction with resulting decreases in CBF and brain tissue PO2, which could have partially counteracted the increase in cerebral venous blood oxygenation. Nevertheless, this is unlikely because breathing 100% O2 at pressures ≥ 1 atmosphere absolute increases the PO2 in the striatum of anesthetized rats.40
Cerebral blood flow is inversely correlated with the arterial O2 content when the latter is low as a result of reduced hemoglobin concentration rather than decreased PaO2.41 The increase or decrease in CBF when there is a decrease or increase in Hct, respectively, is not attributable to changes in blood viscosity but rather to changes in oxygen delivery to the brain.42 In the present study, Hct was higher in dogs during both conditions involving hypercapnia, compared with values in all other conditions, which could be a result of CO2-induced catecholamine release with subsequent splenic contraction and release of RBCs into the systemic circulation. However, this increase in Hct was not accompanied by a significant increase in the arterial O2 content; thus, its effect on CBF was probably unimportant.
Results of the present study indicated that there are cerebral regional differences in cerebrovascular responses to CO2 and O2 as assessed by changes in BOLD SI in dogs anesthetized with isoflurane, with comparatively greater responses in the thalamus and diencephalic gray matter. The magnitude of BOLD signal changes achieved in dogs that are breathing 100% O2 are of similar magnitude to those achieved in anesthetized dogs during fMR imaging studies. Therefore, on the basis of these findings, it is suggested that a lower inspired fraction of O2 should be used in fMR imaging studies to obtain greater BOLD signal changes in response to the functional stimulus, especially when a 1.5-T magnet is used. Another possible approach for fMR imaging studies in isoflurane-anesthetized animals would be to use basal conditions of hypocapnia, so that cerebral vessels would have a greater capacity to dilate in response to the functional stimulus.
ABBREVIATIONS
| BOLD | Blood oxygenation level dependent |
| CBF | Cerebral blood flow |
| CMRO2 | Cerebral metabolic rate for O2 consumption |
| ec-Ho | Conditions of eucapnia and hyperoxemia |
| ec-no | Conditions of eucapnia and normoxemia |
| fMR | Functional magnetic resonance |
| hc-Ho | Conditions of hypocapnia and hyperoxemia |
| Hc-Ho | Conditions of hypercapnia and hyperoxemia |
| hc-no | Conditions of hypocapnia and normoxemia |
| Hc-no | Conditions of hypercapnia and normoxemia |
| MAC | Minimum alveolar concentration |
| MAP | Mean arterial blood pressure |
| MR | Magnetic resonance |
| PSIC | Percentage of signal intensity variation with respect to control condition value |
| ROI | Region of interest |
| SI | Signal intensity |
| SNR | Signal-to-noise ratio |
| SpO2 | Oxygen saturation as measured by pulse oximetry |
| SW | Susceptibility weighted |
Propofol injection, Novopharm Ltd, Toronto, ON, Canada.
IsoFlo, Abbott Animal Health, Abbott Laboratories Ltd, Saint-Laurent, QC, Canada.
Narkomed 2B anesthesia system, North American Draeger, Draeger Medical Inc, Telford, Pa.
Nonin 8600 V, Nonin Medical Inc, Plymouth, Minn.
Criticare model 1100, Criticare System Inc, Waukesha, Wis.
Omron Healthcare Inc, Vernon Hills, Ill.
Plasma-Lyte A injection, Baxter Corp, Mississauga, ON, Canada.
Dopamine hydrochloride and 5% dextrose injection USP, Baxter Corp, Toronto, ON, Canada.
Atracurium besylate injection, Sandoz Canada Inc, Boucherville, QC, Canada.
Glycopyrrolate injection USP, Sandoz Canada Inc, Boucherville, QC, Canada.
Enlon, Baxter Corp, Toronto, ON, Canada.
ABL 700 series analyzer, Radiometer, Copenhagen, Denmark.
OSM 3 hemoximeter radiometer, Copenhagen, Denmark.
Signa 1.5T Excite II, 11.0 software platform, General Electric Medical Systems, Waukesha, Wis.
IMAGEJ, version 1.39d, National Institutes of Health, Bethesda, Md.
SAS, version 9.1.3, SAS Institute Inc, Cary, NC.
References
- 1.↑
Chavhan GB. Miscellaneous MR techniques. In: Chavhan GB, ed. MRI made easy. Tunbridge Wells, Kent, England: Anshan Ltd, 2007;246–247.
- 2.↑
Ogawa S, Lee TM, Kay AR, et al.Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci U S A 1990;87:9868–9872.
- 3.↑
Buxton RB, Frank LR. A model for the coupling between cerebral blood flow and oxygen metabolism during neural stimulation. J Cereb Blood Flow Metab 1997;17:64–72.
- 4.
Jezzard P, Ramsey NF. Functional MRI. In: Tofts P, ed. Quantitative MRI of the brain. John Wiley & Sons Ltd, 2003;413–453.
- 5.↑
Davis TL, Kwong KK, Weisskoff RM, et al.Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci U S A 1998;95:1834–1839.
- 6.↑
McDonell WN, Kerr CL. Respiratory system. In: Tranquilli WJ, Thurmon JC, Grimm KA, eds. Lumb and Jones' veterinary anesthesia and analgesia. 4th ed. Blackwell Publishing Ltd, 2007;117–151.
- 7.↑
Posse S, Kemna LJ, Elghahwagi B, et al.Effect of graded hypo- and hypercapnia on fMRI contrast in visual cortex: quantification of T(*)(2) changes by multiecho EPI. Magn Reson Med 2001;46:264–271.
- 8.↑
Posse S, Olthoff U, Weckesser M, et al.Regional dynamic signal changes during controlled hyperventilation assessed with blood oxygen level-dependent functional MR imaging. AJNR Am J Neuroradiol 1997;18:1763–1770.
- 9.
Losert C, Peller M, Schneider P, et al.Oxygen-enhanced MRI of the brain. Magn Reson Med 2002;48:271–277.
- 10.↑
Prisman E, Slessarev M, Han J, et al.Comparison of the effects of independently-controlled end-tidal PCO(2) and PO(2) on blood oxygen level-dependent (BOLD) MRI. J Magn Reson Imaging 2008;27:185–191.
- 11.↑
Brevard ME, Duong TQ, King JA, et al.Changes in MRI signal intensity during hypercapnic challenge under conscious and anesthetized conditions. Magn Reson Imaging 2003;21:995–1001.
- 12.↑
Sicard K, Shen Q, Brevard ME, et al.Regional cerebral blood flow and BOLD responses in conscious and anesthetized rats under basal and hypercapnic conditions: implications for functional MRI studies. J Cereb Blood Flow Metab 2003;23:472–481.
- 13.↑
Steffey EP, Howland D Jr. Isoflurane potency in the dog and cat. Am J Vet Res 1977;38:1833–1836.
- 14.↑
Beitz AJ, Fletcher TF. The brain. In: Evans HE, ed. Miller's anatomy of the dog. 3rd ed. Philadelphia: Saunders, 1993;894–952.
- 15.↑
Guyton AC, Hall JE. Cerebral blood flow, cerebrospinal fluid and brain metabolism. In: Guyton AC, Hall JE, eds. Textbook of medical physiology. 11th ed. Philadelphia: Elsevier Saunders, 2006;761–768.
- 16.↑
Drummond JC, Todd MM, Scheller MS, et al.A comparison of the direct cerebral vasodilating potencies of halothane and isoflurane in the New Zealand white rabbit. Anesthesiology 1986;65:462–467.
- 17.↑
Hoffman WE, Edelman G, Kochs E, et al.Cerebral autoregulation in awake versus isoflurane-anesthetized rats. Anesth Analg 1991;73:753–757.
- 18.↑
McPherson RW, Traystman RJ. Effects of isoflurane on cerebral autoregulation in dogs. Anesthesiology 1988;69:493–499.
- 19.↑
Strebel S, Lam AM, Matta B, et al.Dynamic and static cerebral autoregulation during isoflurane, desflurane, and propofol anesthesia. Anesthesiology 1995;83:66–76.
- 20.
Myburgh JA, Upton RN, Grant C, et al.The cerebrovascular effects of adrenaline, noradrenaline and dopamine infusions under propofol and isoflurane anaesthesia in sheep. Anaesth Intensive Care 2002;30:725–733.
- 21.
Myburgh JA, Upton RN, Grant C, et al.The effect of infusions of adrenaline, noradrenaline and dopamine on cerebral autoregulation under isoflurane anaesthesia in an ovine model. Anaesth Intensive Care 2003;31:259–266.
- 22.↑
von Essen C, Zervas NT, Brown DR, et al.Local cerebral blood flow in the dog during intravenous infusion of dopamine. Surg Neurol 1980;13:181–188.
- 23.
Busija DW, Leffler CW, Pourcyrous M. Hyperthermia increases cerebral metabolic rate and blood flow in neonatal pigs. Am J Physiol 1988;255:H343–H346.
- 24.↑
Michenfelder JD, Theye RA. Hypothermia: effect on canine brain and whole body metabolism. Anesthesiology 1968;29:1107–1112.
- 25.
Krafft P, Frietsch T, Lenz C, et al.Mild and moderate hypothermia (alpha-stat) do not impair the coupling between local cerebral blood flow and metabolism in rats. Stroke 2000;31:1393–1400.
- 26.↑
Haskins S, Pascoe PJ, Ilkiw JE, et al.Reference cardiopulmonary values in normal dogs. Comp Med 2005;55:156–161.
- 27.↑
Myburgh JA, Upton RN, Ludbrook GL, et al.Cerebrovascular carbon dioxide reactivity in sheep: effect of propofol or isoflurane anaesthesia. Anaesth Intensive Care 2002;30:413–421.
- 28.↑
McPherson RW, Briar JE, Traystman RJ. Cerebrovascular responsiveness to carbon dioxide in dogs with 1.4% and 2.8% isoflurane. Anesthesiology 1989;70:843–850.
- 29.↑
Iida H, Ohata H, Iida M, et al.Isoflurane and sevoflurane induce vasodilation of cerebral vessels via ATP-sensitive K+ channel activation. Anesthesiology 1998;89:954–960.
- 30.
Matta BF, Heath KJ, Tipping K, et al.Direct cerebral vasodilatory effects of sevoflurane and isoflurane. Anesthesiology 1999;91:677–680.
- 31.
Matta BF, Mayberg TS, Lam AM. Direct cerebrovasodilatory effects of halothane, isoflurane, and desflurane during propofolinduced isoelectric electroencephalogram in humans. Anesthesiology 1995;83:980–985.
- 32.↑
Willis CK, Quinn RP, McDonell WM, et al.Functional MRI as a tool to assess vision in dogs: the optimal anesthetic. Vet Ophthalmol 2001;4:243–253.
- 33.↑
Floyd TF, Clark JM, Gelfand R, et al.Independent cerebral vasoconstrictive effects of hyperoxia and accompanying arterial hypocapnia at 1 ATA. J Appl Physiol 2003;95:2453–2461.
- 34.↑
Rostrup E, Larsson HB, Toft PB, et al.Functional MRI of CO2 induced increase in cerebral perfusion. NMR Biomed 1994;7:29–34.
- 35.↑
Parrish TB, Gitelman DR, LaBar KS, et al.Impact of signal-tonoise on functional MRI. Magn Reson Med 2000;44:925–932.
- 36.↑
Watson NA, Beards SC, Altaf N, et al.The effect of hyperoxia on cerebral blood flow: a study in healthy volunteers using magnetic resonance phase-contrast angiography. Eur J Anaesthesiol 2000;17:152–159.
- 37.
Rostrup E, Larsson HB, Toft PB, et al.Signal changes in gradient echo images of human brain induced by hypoand hyperoxia. NMR Biomed 1995;8:41–47.
- 38.
Bulte DP, Chiarelli PA, Wise RG, et al.Cerebral perfusion response to hyperoxia. J Cereb Blood Flow Metab 2007;27:69–75.
- 39.↑
Sicard KM, Duong TQ. Effects of hypoxia, hyperoxia, and hypercapnia on baseline and stimulus-evoked BOLD, CBF, and CMRO2 in spontaneously breathing animals. Neuroimage 2005;25:850–858.
- 40.↑
Demchenko IT, Luchakov YI, Moskvin AN, et al.Cerebral blood flow and brain oxygenation in rats breathing oxygen under pressure. J Cereb Blood Flow Metab 2005;25:1288–1300.
- 41.↑
Brown MM, Wade JP, Marshall J. Fundamental importance of arterial oxygen content in the regulation of cerebral blood flow in man. Brain 1985;108:81–93.
- 42.↑
Waschke KF, Krieter H, Hagen G, et al.Lack of dependence of cerebral blood flow on blood viscosity after blood exchange with a Newtonian O2 carrier. J Cereb Blood Flow Metab 1994;14:871–876.
