Breathing 100% O2 at 1 atmosphere absolute (ATA) is known to be associated with a decrease in cerebral blood flow (CBF). It is also accompanied by a fall in arterial Pco2 leading to uncertainty as to whether the cerebral vasoconstriction is totally or only in part caused by arterial hypocapnia. We tested the hypothesis that the increase in arterial Po2 while O2 was breathed at 1.0 ATA decreases CBF independently of a concurrent fall in arterial Pco2. CBF was measured in seven healthy men aged 21–62 yr by using noninvasive continuous arterial spin-labeled-perfusion MRI. The tracer in this technique, magnetically labeled protons in blood, has a half-life of seconds, allowing repetitive measurements over short time frames without contamination. CBF and arterial blood gases were measured while breathing air, 100% O2, and 4 and 6% CO2 in air and O2 backgrounds. Arterial Po2 increased from 91.7 ± 6.8 Torr in air to 576.7 ± 18.9 Torr in O2. Arterial Pco2 fell from 43.3 ± 1.8 Torr in air to 40.2 ± 3.3 Torr in O2. CBF-arterial Pco2 response curves for the air and hyperoxic runs were nearly parallel and separated by a distance representing a 28.7–32.6% decrement in CBF. Regression analysis confirmed the independent cerebral vasoconstrictive effect of increased arterial Po2. The present results also demonstrate that the magnitude of this effect at 1.0 ATA is greater than previously measured.

  • cerebral blood flow
  • oxygen
  • carbon dioxide
  • arterial spin-labeling
  • magnetic resonance
  • atmospheres absolute

cerebral blood flow (CBF) in normal subjects is under the continuous dominant control of arterial Pco2 (PaCO2), but it can be modulated by the background level of arterial Po2 (PaO2). Acute exposure to a high inspired Po2 results in a cascade of physiological events leading to increased brain tissue Pco2, increased ventilation, decreased PaCO2, and decreased CBF (29, 30). The magnitude of the decrease in CBF in response to breathing 100% O2 at 1.0 atmosphere absolute (ATA), as measured by different methods, ranges from 13 to 27% of the air-breathing value. With the use of the Kety-Schmidt method, a 13–15% decrement has been measured (27, 34), whereas with 133Xenon scintigraphy, a 21% decrease has been reported (40). Two recently reported MRI studies using phase contrast angiography found this decrease in CBF to be in the range of 16–27% (49, 59). Because breathing O2, even at 1.0 ATA, is associated with arterial hypocapnia, it remains unclear whether the decreased CBF during hyperoxia is a consequence of the hypocapnia alone or whether there are separate effects of hypocapnia and hyperoxia. We tested the hypothesis that an increase in PaO2 while breathing O2 at 1.0 ATA decreases CBF independently of the accompanying fall in PaCO2 using a noninvasive MRI method to measure CBF. Although the CO2 and O2 effects on CBF are physiologically inseparable, measurements of CBF over a range of PaCO2 values in air and O2 backgrounds made it possible to determine their separate influences analytically.


The experimental protocol was approved by the institutional review board, and informed consent was obtained from each subject. Seven men between the ages of 21 and 62 yr (mean ± SE: 39 ± 14 yr) were studied after medical history and physical examination determined no occult cardiovascular, respiratory, neurological, or psychiatric disorder.

An open breathing system was employed to deliver mixed gases by using a low dead space, nonrebreathing valve (Hans Rudolph model 1400) and 100-liter Douglas bag as the inspiratory reservoir. Premixed gases in high-pressure cylinders containing air, 4 and 6% CO2 in air, and 4 and 6% CO2 in O2 were obtained from BOC Gases (Murray Hill, NJ). O2 (100% O2) was administered from the hospital wall supply. These gases were selected to provide comparable ranges of PaCO2 in air and O2 backgrounds.

The experimental protocol was designed to minimize subject time in the magnet yet allow for stabilization of CBF after the inspired gas mixture was changed. The study was divided into two runs, an “air” run and a “hyperoxic” run. At the initiation of each run, a 10-min period of breathing was allowed for CBF stabilization while subjects breathed either 21% O2 (air run) or 100% O2 (hyperoxic run). Subsequently, when gases were inspired containing increasing concentrations of CO2, a 5-min period for stabilization on the new gas was allowed. A continuous arterial spin-labeled (CASL)-perfusion-MRI imaging session of 6 min followed stabilization on each gas for the purpose of measuring CBF. CBF measurements occurred in the following order: air run: 1) air → 2) air-4% CO23) air-6% CO2; hyperoxic run: 4) 100% O25) 96% O2-4% CO26) 94% O2-6% CO2.

For arterial blood sampling, a 22- or 20-gauge radial arterial catheter was placed by using sterile technique and 2 ml of 1% lidocaine for local anesthesia. Arterial blood gas samples were withdrawn near the end of the 6-min MRI scanning phase. The 3-ml sample was placed in ice water and analyzed within 2–3 min of sampling by using a NOVA pHOX analyzer (NOVA Biomedical, Waltham, MA). The following parameters were measured: Po2, Pco2, pH, hemoglobin, and Hct. Arterial blood O2 content (CaO2) was calculated.

The duration of each run was ∼40 min, including time for structural imaging. Subjects were allowed a 15- to 30-min rest period in air between the two runs.

CBF measurements were made by using the CASL-perfusion MRI method (4, 12, 16). This technique has been demonstrated in humans to have a high degree of accuracy compared with PET methods (65) as well as a high degree of precision (19). All CASL-perfusion MRI studies were performed in a 1.5-T GE Signa Echospeed scanner (GE Medical Systems, Milwaukee, WI). CASL control labeling was applied at the level of the cervicomedullary junction by using a postlabeling delay of 1.5 s (2). Images were acquired by using a gradient-echoplanar sequence [field of view = 24 × 15 cm, matrix 64 × 40, bandwidth 62.5 kHz, repitition time/echo time = 4,000/22 ms, slice thickness = 8 mm with a 2-mm gap]. Eight slices were acquired from inferior to superior in an interleaved order, and each slice acquisition took 45 ms. The imaging volume was chosen to include supratentorial structures. Each perfusion measurement consisted of 90 acquisitions with a scan time of ∼6 min. Raw image data were saved onto DAT tape. A 1-min chemical shift image sequence was also performed to obtain data used for correcting geometric distortion in echoplanar images.

CASL-perfusion MRI data were reconstructed from raw data offline and corrected for geometric distortion by using custom software written in the Interactive Data Language software program (Research Systems, Boulder, CO). The reconstructed images in each scan were separated into 45 pairs of label and control images and then pairwise subtracted (1). The resulting series of 45 perfusion difference images were corrected for motion and physiological fluctuation by using an algorithm based on principal component analysis (3), followed by averaging across the image series to produce a single set of perfusion-sensitive images.

Absolute CBF was quantified from the mean difference perfusion images by using a modification of a previous approach (2) according to the following equation (12) Math where SCASL is the difference between the control and labeled image intensities, SCSF is the average intensity of the control image in the manually defined ventricular region, T1CSF (4.2 s) is the longitudinal relaxation time of CSF, T1a and T2a are the longitudinal and transverse relaxation times of arterial blood, respectively, w is the postlabeling delay (1.5 s), α is the labeling efficiency (71%), λ is the water fraction of arterial blood (0.76), and ρ is the density of brain tissue (1.05 g/ml). T1a and T2a are assumed to be constant for this study at T1a = 1,100 ms and T2a = 240 ms. Use of fixed T1a rather than measuring tissue T1 assumes that the labeled spins remain primarily in the microvasculature rather than exchanging completely with tissue water. However, because blood and brain T1 values are almost equal at 1.5 T, violations of this assumption would cause only a small error. The CASL-perfusion MRI technique, therefore, measures the rate of microvascular blood flow within brain parenchyma (12). An example of quantitative perfusion images obtained from a 6-min acquisition is shown in Fig. 1.

Fig. 1.

Colorized continuous arterial spin labeled (CASL)-perfusion MRI images with scale depicting cerebral blood flow (CBF) changes in response to air, air-4% CO2, air-6% CO2, 100% O2, 96% O2-4% CO2, and 94% O2-6% CO2.

Global CBF (CBFGlobal) was determined by averaging perfusion values across all brain voxels. Second, brains were segmented into gray and white matter pixels based on the algorithm of Ashburner and Friston (7) within statistical parametric mapping (SPM) 99 (54a). CBF was determined for each tissue type yielding gray matter CBF (CBFGray) and white matter CBF (CBFWhite).

Statistical analysis was performed by using JMP professional software version 5 (SAS, Cary, NC). The paired t-test was used to estimate differences in means for paired gas samples between runs. Multiple linear regression methods were used to model the contributions of baseline CBF during air breathing (CBFAir-4–6%), change in PaCO2 (ΔPaCO2), and change in PaO2 (ΔPaO2), to the change in CBF (ΔCBF) when moving between the paired gases (air-4% CO2 → 96% O2-4% CO2, and air-6% CO2 → 94% O2-6% CO2). Terms, P values, and power estimates are presented for this analysis.

Multiple linear and nonlinear regression approaches were also used to model the effects of baseline CBF while breathing 21% O2 (CBFAir), ΔPaCO2 from baseline on 21% O2, and ΔPaO2from baseline on 21% O2, on CBFGray, CBFWhite, and whole brain. Coefficients of determination (R2) are presented for the models along with P values for each partial regression coefficient from ANOVA. A probability of P ≤ 0.05 was chosen to test significance for the whole model and for partial regression coefficients.


All subjects completed the study. Figure 1 depicts a series of colorized CASL-perfusion MRI image sets used for CBF measurements in a single subject. Qualitatively, from visual inspection of the images, it can be seen that CBF increases during the air run with the addition of increasing inspired Pco2. Also, when the air and hyperoxic runs are compared, the matched pairs of images reveals the lower levels of CBF at all pairings in the hyperoxic run.

Table 1 summarizes the Pco2, Po2, CaO2, as well as CBFGlobal, CBFGray, and CBFWhite regions. Breathing 100% O2 vs. air at 1 Atm was associated with a relative hypocapnia or decrement in PaCO2 of ∼3 Torr (P = 0.01). The mean relative differences in CBFGlobal in the group pairings (n = 7) of air vs. 100% O2, air-4% CO2 vs. 96% O2-4% CO2, and air-6% CO2 vs. 94% O2-6%C O2 were –32.6% (P < 0.0001), –28.7% (P = 0.0008), and –29.4% (P < 0.0001), respectively. It is important to note that, although there exists a statistical difference in PaCO2 tensions between the air and 100% O2 exposures, by the addition of CO2 at 4 and 6% inspired concentrations, we were able to eliminate any statistical differences in these levels in the subsequent comparisons.

View this table:
Table 1.

Arterial Pco2, O2 content, and CBF by inspired gas

The major goal of this study was to determine whether the effects of PaO2 and PaCO2 on CBF could be discriminated and identify whether O2 has a vasoconstrictive effect on CBF independent of the mild arterial hypocapnia, which occurs in response to hyperoxia in normal subjects at rest. Figure 2 depicts graphically the relationship between CBF and PaCO2 for the seven subjects while breathing air or O2 background gases alone and then with 4 and 6% CO2. Two distinct, nearly parallel curves are defined. The initial points on each curve depict a difference in CBF while breathing air vs. 100% O2. As mentioned above, PaCO2 is decreased during the transition from air breathing to O2 breathing at 1.0 ATA resulting in the shift of the lower (hyperoxic run) curve to the left. Any difference in CBF between these initial study points (air vs. 100% O2) thus reflects the combined effects of arterial hypocapnia and hyperoxia on the cerebral vasculature (Table 1). With the addition of 4 and 6% CO2, mean differences in PaCO when breathing the air or hyperoxic mixtures were not significant (Table 1), and the differences between the curves at these points of comparison essentially reflect the effect of the difference in PaO2 2 alone.

Fig. 2.

CBF is plotted vs. arterial Pco2 for the air and hyperoxia runs. Values are means ± SE for CBF and Pco2 values for each gas. Means are connected to describe curves representing the CBF response to Pco2 in the presence of air and hyperoxia.

Preliminary univariate analysis indicated that the absolute magnitude and the percentage of the decrease in CBF in response to hyperoxic breathing are also dependent on the baseline level of CBF during air breathing (Fig. 3). Therefore, multiple regression analysis of the change in CBF when moving from air-4% CO2 → 96% O2-4% CO2 and air-6% CO2 → 94% O2-6% CO2 (ΔCBF4–6%) was conducted as a function of the initial CBF during the breathing of air-4% CO2 or air-6% CO2 (CBFAir-4–6%), ΔPaO2, and ΔPaCO2 during the same transitions. A statistical summary for each of the covariates is listed in Table 2. It can be seen then that both CBFAir-4–6% (P = < 0.0001) and ΔPaO (P = 0.0081) contribute significantly to the magnitude of ΔCBF. It can also be seen that any change in PaCO did not significantly contribute to CBF because P = 0.17 for the covariate ΔPaCO2.

Fig. 3.

Absolute (left) and relative (right) changes in CBF from baseline measurements made on 4 and 6% CO2 (ΔCBF4–6%) in air background to measurements made on 4 and 6% CO2 in O2 background are plotted vs. baseline CBF measurements made on 4–6% CO2 in air background.

View this table:
Table 2.

ΔCBF multiple regression analysis

Thus, by comparing results at matched testing points, the decrease in CBF persists when the hypocapnia associated with breathing 100% O2 has been eliminated (Table 1, Fig. 2). An independent cerebral vasoconstrictive effect of hyperoxia is also confirmed by using multiple regression analysis over the experimentally imposed ranges of PaCO2 and PaO and by accounting for the impact of baseline CBF when air was breathed on the magnitude of any change.

To examine the effects of hyperoxic breathing on different tissue types within the brain, multiple linear and nonlinear regression methods were used to analyze the contributions of the following covariates to CBF: CBFAir-21% (baseline CBF on air, before the addition of CO2), ΔPaO2, and ΔPaCO2.

Regression models were created for global, gray matter, and white matter. The following equations resulted Math(1) which had model R2 and P values of 0.84 and <0.0001, respectively, and individual P values of <0.0001 for CBFAir, ΔPaO2, and ΔPaCO2, and a P value of 0.0043 for the intercept Math(2) which had model R2 and P values of 0.89 and <0.0001, respectively, and individual P values of <0.0001 for CBFAir, ΔPaO2, and ΔPaCO2, and P values of 0.0012 and 0.0088 for the intercept and ΔPaCO22, respectively, and Math(3) which had model R2 and P values of 0.73 and <0.0001, respectively, and individual P values of <0.0001 for CBFAir and ΔPaO2, and P values of 0.0029 and 0.0013 for the intercept and ΔPaCO2, respectively.

Plots of actual vs. predicted CBF values from the three models are further depicted in Fig. 4, A–C. The effect of a change in PaO2 could not be modeled in a nonlinear fashion because the Po2 created in our experimental design essentially represents only two levels (high and low) in the presence of air breathing or 100% O2 breathing. Nonlinear modeling of CO2 tensions was attempted for global, gray, and white matter and demonstrated significance when modeled as a binomial in gray matter alone.

Fig. 4.

Actual (measured) CBF vs. predicted (model derived) CBF for global (A), gray matter (B), and white matter (C) along with 95% confidence intervals (dotted lines).


Independent cerebral vasoconstrictive effect of hyperoxia. As stated previously, the normal CO2-related interaction of respiratory and cerebral circulatory control causes the concurrent effects of hyperoxia and arterial hypocapnia to be physiologically inseparable. The linkage of these effects has been attributed to hyperventilation induced by CO2 retention in respiratory control centers (22, 31, 33) that is in turn caused by a reduced CO2-carrying capacity of oxyhemoglobin in association with high concentrations of physically dissolved O2 (Haldane effect) (38). Nevertheless, the present results (Fig. 2) clearly demonstrate an independent cerebral vasoconstrictive effect of hyperoxia across a wide range of PaCO2. In agreement, Kolbitsch et al. (28) recently demonstrated a significant reduction in regional CBF during oxygen breathing at 1.0 ATA when end-tidal (alveolar) Pco2 was maintained at the air-breathing control level. Reivich (45, 46) also found what appeared to be an independent vasoconstrictor effect of hyperoxia during oxygen breathing at 3.5 ATA with PaCO2 maintained at ∼15 Torr by voluntary hyperventilation. Similar results were observed at 2.0 ATA with Pco2 maintained at ∼19 Torr.

In contrast to the above results, an independent vasoconstrictor effect of hyperoxia is not supported by the observations that CBF remained unchanged in normal men during the transition from air breathing to 80% O2 at 1.0 ATA with alveolar Pco2 controlled at 43 Torr (29, 30) or in unanesthetized ponies breathing O2 at 1.0 ATA with PaCO maintained at air-breathing control levels (10). The 2known association of cerebral hypoxia with extreme hypocapnia during air breathing (46) suggests that the apparent independent vasoconstrictor effect of hyperoxia found by Reivich at hypocapnic levels of PaCO2 could also be explained by removal of the cerebral vasodilator effect of hypoxia (30, 45).

Analysis of the present data also demonstrates an independent vasoconstrictor effect of hyperoxia on CBFGray and CBFWhite as well as on CBFGlobal. The three linear models that were developed for the purpose of comparing the effects of hyperoxic breathing in global, gray, and white matter describe a greater vasoconstrictive response to increasing Po2 and greater vasodilatory response to increasing Pco2 in gray vs. white matter. Gray matter is metabolically more active than white matter, and thus these differences are not surprising. The higher CBF values recorded for gray matter and lower values for white matter (Table 1) are consistent with values recorded for PET (65) during air breathing. The CBFGray-to-CBFWhite ratio of 1.7 from our present work is consistent with findings in the literature from previous work using PET and single-photon-emission computed tomography of 1.6–1.8 (24, 25, 41).

O2 in the regulation of CBF. Mechanisms proposed as involved in the localized regulation of CBF in response to hyperoxia and hypoxia are complex and may include roles for the parenchyma (43, 51), cerebrovascular endothelium (15, 18, 42), specific brain O2-sensitive neurons (23), and even the red blood cells (17). When PaO2 is reduced by reducing the inspired Po2, CBF does not increase until PaO2 is <50 Torr (30, 46). In contrast, CBF has been shown to inversely correlate with CaO2 when CaO2 is low due to reduced hemoglobin concentration and PaO2 is not manipulated (9). This correlation occurs whether O2 is carried by the intact red blood cells or in a dissolved state(58). In the present report, there was also an inverse relation between CaO2 and CBF. When inspired percent O2 was increased from 21 to 100% at 1.0 ATA, there was a 13–14% increase in CaO2, which was concurrent with a 33% decrease in CBF. Although these correlations exist, they do not necessarily mean that CaO2 acts as a specific factor in brain O2 flow control. Rather, it has been suggested that Po2-dependent cells in endothelium may act as local O2 sensors modulating vascular tone (44). The vasoconstrictive effect of hyperoxia may be related to inactivation of NO by enhanced generation of O2-free radicals(44).

Magnitude of hyperoxic cerebral vasoconstriction. In the present study, the transition from breathing air to 100% O2 at 1.0 ATA caused a larger decrease in CBF (33%) than the 13–15% decrement, which has been consistently reported previously (27, 34) by using the Kety-Schmidt technique, or the 21% decrease reported by Ohta (40) by using 133Xe scintigraphy. The present results are more consistent with, yet still larger than, the 16–27% decrease in CBF during O2 breathing at 1.0 ATA recently reported by Watson (59) and Rostrup (49) by using another MRI technique, dynamic susceptibility contrast. Even during O2 breathing at 3.5 ATA, a previously reported CBF decrement of 25% (34) is smaller than the present value for O2 breathing at 1.0 ATA. Only a few studies contain CBF measurements during O2 breathing at >1 ambient pressure. Lambertsen et al. (34) found CBF decrements of 15 and 25% at 1.0 and 3.5 ATA, respectively, whereas Reivich (45, 46) reported values of 22 and 27% at 2.0 and 3.5 ATA, respectively, in the presence of arterial hypocapnia. More recently, Ohta (40) measured CBF decrements of 9, 21, 23, and 19% at O2 pressures of 0.5, 1.0, 1.5, and 2.5 ATA, respectively. The latter results suggest that the cerebral vasoconstrictive effect of hyperoxia may be near maximal at 1.0 ATA. At the present time, it is not practical to use CASL-perfusion MRI or other MRI methods at ambient pressures that are >1 atm.

Comparison with previous CBF measurements. Quantitative investigation of CBF and metabolism in man became possible when Kety and Schmidt developed an inert gas method for measuring CBF by calculating rate of N2O uptake by the brain (26). The method required repeated sampling of brain venous blood from the superior bulb of the internal jugular vein concurrently with arterial blood while the subject breathed 15% N2O for at least 10 min. Subsequent demonstrations that cerebral O2 consumption is not changed by O2 breathing at 1.0 (27, 32) or 3.5 ATA (32) made it possible to use arterial-venous O2 content differences across the brain to measure relative changes in CBF under these conditions (53). Continued investigation led to the development of additional CBF measurement methods that used other inert gas tracers or radioactive labels to determine rate of uptake of the tracer or rate of washout after a state of near saturation has been established (46). All of these methods require either the sampling of brain venous blood or the monitoring of brain venous concentrations of the selected tracer.

Blood from the internal jugular bulb is known to contain contributions from extracranial or extracerebral sources (52, 53). While air was breathed at 1.0 ATA, extracranial contributions to internal jugular venous flow have been estimated to range from 3 to 7% (52), with the occasional possibility of gross contamination from extracranial venous effluent (35). Scheinberg (50) found that contamination of cerebral venous blood by facial or neck blood caused falsely low values of CBF, arterio-venous O2 difference, and O2 consumption. He estimated that some degree of extracerebral contamination occurred in ∼20% of the subjects. In contrast, the CASL-perfusion MRI technique allows CBF to be determined from designated intracranial sources by measuring the rate of microvascular blood flow within brain parenchyma (12).

It is possible, yet unlikely, that the potential extracerebral contamination of brain venous blood in some of the previous studies can account for the reported absence of an independent cerebral vasoconstrictive effect of hyperoxia or the smaller magnitude of this effect found previously. At the present time, there are no obvious explanations for the differences between the results of this study and those of previous investigations. In some aspects, the results of the present study are remarkably consistent with previous data. Average values of 53.6 ml · 100 g–1 · min–1 for CBFGlobal at an PaCO2 of 43.3 Torr during air breathing (Table 1) are essentially identical to corresponding values of 53 ml · 100 g–1 · min–1 at 43 Torr in one of the subject groups studied by Kety and Schmidt (27). In addition, the average CBF value of 36.1 ml · 100 g–1 · min–1 at an PaCO2 of 40.2 Torr during O2 breathing (Table 1) falls directly on the curve established for CBF vs. PaCO2 in an O2 background while using cerebral clearance of 133Xe to measure CBF (14).

Although the present CBF-PaCO2 relationships for both air and O2 breathing without added CO2 are consistent with the results of previous studies using other methods, the slopes of the present CBF-PaCO2 curves (Fig. 2) are more shallow than those found previously in normal subjects. Using arterio-venous O2 differences across the brain to calculate relative changes in CBF during air breathing at 1.0 ATA, Lambertsen et al. (33) found that CBF increased by 57.6% for an PaCO2 increment of 8.9 Torr. During air breathing in the present study, CBFGlobal increased by 30.8% for an 8.8-Torr increase in PaCO2 (Table 1). By using clearance of 133Xe to measure CBF during O2 breathing with added CO2, CBF more than doubled (102.5%) for an PaCO2 increase of 11.4 Torr (14). Corresponding values for the present study during O2 breathing were 37.4% and 10.9 Torr, respectively (Table 1).

Effect of CO2 breathing on PaO2. The breathing of CO2 in an air background resulted in small increases in PaO2 from 91.7 ± 6.8 Torr on air to 127.0 ± 5.4 Torr on air-4% CO2 and 144.5 ± 6.6 Torr on air-6% CO2, as well as corresponding increments in CaO2 from 17.9 ± 0.8 ml/dl on air to 18.5 ± 0.9 ml/dl on air-4% CO2 and 18.7 ± 0.8 ml/dl on air-6% CO2, differences that were not observed during the breathing of CO2 in an O2 background (Table 1). An increase in PaO2 during air breathing with added CO2 has been attributed to increased alveolar ventilation with improved alveolar ventilation-blood flow matching (55). The present average PaO2 values of 127.0 and 144.5 Torr at inspired Pco2 levels of ∼28 and 43 Torr agree well with previously measured values of 128.5 and 133.3 Torr at controlled inspired Pco2 levels of 30 and 40 Torr, respectively (13).

CASL-perfusion MRI. CASL-perfusion MRI is ideally suited for this type of study at 1.0 ATA because it is completely noninvasive and the tracer, magnetically flipped protons in arterial blood, has a half-life on the order of several seconds. Multiple measurements can be obtained, over temporally short time frames, without contamination from retained tracer. Several excellent reviews have been published with detailed theoretical descriptions of the technique (8, 11, 61, 63).

The CASL-perfusion MRI technique has been validated against “gold standards” such as PET (37, 65) and dynamic susceptibility contrast MRI methods (36) in human brain studies. Calamante et al. (11) offer a detailed listing of CBF measurements in various species and comparisons with those made in the same species with more traditional methods, including microspheres. Walsh et al. (56) compared regional CBF measurements by using the arterial spin-labeling technique with those obtained from radioactive microspheres in the rat and found a mean difference of 1.5%. The CASL-perfusion MRI technique has been applied clinically to study CBF in stroke (12), epilepsy (60), and Alzheimer's dementia (6), as well as blood flow in other organ systems such as lung (39, 48), kidney (47), and muscle perfusion (21). The technique has been additionally utilized in human functional brain-mapping research (62), research into human cerebral physiology (20), and in the laboratory animal for the study of blood flow in various organ systems (54, 64).

The most significant concern for CASL-perfusion MRI is that prolonged arterial blood transit times, such as might exist in the presence of cerebrovascular disease, may cause an underestimation of flow (11). At the extreme, transit time could be greater than the time for the label to decay (T1). This deficiency has been exploited to assist in the diagnosis of cerebrovascular disease through the noninvasive mapping of inhomogeneities in regional CBF (57).

In conclusion, the observed decrease in CBF while breathing 100% O2 at 1.0 ATA represents the combined effects of arterial hyperoxia and hypocapnia. Furthermore, the present data support the hypothesis that breathing O2 at 1.0 ATA causes cerebral vasoconstriction independently of any vasoconstriction associated with the accompanying arterial hypocapnia. These data also document that gray matter cerebral vasculature is relatively more sensitive to the vasoconstrictive properties of hyperoxia and vasodilatory properties of hypercarbia over the ranges tested. The magnitude of hyperoxia-induced cerebral vasoconstriction is considerably greater than previously reported and is highly dependent on the magnitude of the baseline CBF. If the systemic vasculature responds to hyperoxia in a similar degree, these results have relevance to the use of hyperoxic inspired gas mixtures as an effective means for reducing decompression stress in undersea and aerospace medicine.


This grant was supported by Navy contract no. N61331-99-C-0040.


We acknowledge the assistance of Julio Gonzalez, David Alsop, and Joseph Maldjian for technical assistance on this project.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 6.
  6. 7.
  7. 8.
  8. 9.
  9. 10.
  10. 11.
  11. 12.
  12. 13.
  13. 14.
  14. 15.
  15. 16.
  16. 17.
  17. 18.
  18. 19.
  19. 20.
  20. 21.
  21. 22.
  22. 23.
  23. 24.
  24. 25.
  25. 26.
  26. 27.
  27. 28.
  28. 29.
  29. 30.
  30. 31.
  31. 32.
  32. 33.
  33. 34.
  34. 35.
  35. 36.
  36. 37.
  37. 38.
  38. 39.
  39. 40.
  40. 41.
  41. 42.
  42. 43.
  43. 44.
  44. 45.
  45. 46.
  46. 47.
  47. 48.
  48. 49.
  49. 50.
  50. 51.
  51. 52.
  52. 53.
  53. 54.
  54. 54a.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
View Abstract