Information

Oxygenated hemoglobin in MRI

Oxygenated hemoglobin in MRI


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

I have read the following sentence:

Because this oxygenated hemoglobin is unaffected by magnetic fields, the response RF signal returned to the fMRI scanner is stronger when there is more brain activity and therefore more oxygenated hemoglobin in that brain tissue.

Oxygenated hemoglobin does not react to magnetic fields, while deoxygenated hemoglobin does. So in my view, if there is more brain activity and therefore more oxygenated hemoglobin, the signal should be weaker because the oxygenated hemoglobin does not react to the magnetic field? Why is it the other way around?


I think that the type of fMRI you are referring to is blood-oxygenation-level-dependent fMRI or BOLD fMRI.

The principle behind MRI in general is the detection of proton signals from water molecules. The proton signal is generated by magnetizing the protons in tissue, which causes their spin to change. A subsequent powerful radio wave disrupts this spin and the following relaxation phase of the protons to the original state can be detected by MRI. Water, and hence protons are everywhere in the body, including the brain and the blood.

Deoxygenated hemoglobin (hemoglobin without oxygen) in blood changes the proton signal in its immediate surroundings due to the magnetic properties of deoxyhemoglobin. This is caused by the fact that deoxygenated hemoglobin is paramagnetic and decreases the proton signal. In fact, the paramagnetic influence of deoxyhemoglobin has been regarded as noise in structural MRI scans, before its use in BOLD-fMRI became clear. Oxygenated hemoglobin does not have this property.

Radiopaedia has a nice explanation as to exactly how the BOLD signal is used in BOLD fMRI, and I quote:

When a specific region of the cortex increases its activity in response to a task, the extraction fraction of oxygen from the local capillaries leads to an initial drop in oxygenated haemoglobin [… ]. Following a lag of 2-6 seconds, cerebral blood flow (CBF) increases, delivering a surplus of oxygenated haemoglobin, washing away deoxyhemoglobin and with it the attenuating effect on the MRI signal. It is this large rebound in local tissue oxygenation which is imaged, as it is accompanied by an increase in the MRI signal. The difference is used to generate the BOLD fMRI response.

So brain activity increases the BOLD signal by picking up oxygen-changes after an increased blood flow to that specific part of the brain.

So your statement[… ] If there is more brain activity and therefore more oxygenated hemoglobin the signal should be less strong because the oxygenated hemoglobin does not react to the magnetic fieldis incorrect, because oxyhemoglobin attenuates the proton signal.


How Is Magnetic Resonance Imaging Used to Learn About the Brain?

I am a Research Scientist at Princeton University. I study how the brain develops in children. As children get older, they are exposed to new challenges, such as reading and learning how to swim. I am interested in studying how the brain changes as children meet new cognitive challenges as they grow up. Outside of the lab, I enjoy dancing, drawing, and playing with my dog. *[email protected]

Na Yeon Kim

I am a Cognitive Neuroscientist at Princeton University. I study how we pay attention to specific things—how do our brains help us find a friend in a crowd, play jigsaw puzzles, and notice interesting things from the world? I am particularly interested in how such capabilities change as children grow. Outside the lab, I enjoy playing tennis, running, and traveling to new places.

Sabine Kastner

I am a Scientist and Professor at Princeton University who studies how people use their brains to pay attention to specific activities (e.g., how can it be that you do not hear your parents calling for dinner when you are playing a video game or reading a book?). I also enjoy spending time with my two kids and love the Beatles.

Young Reviewers

Dilworth Stem Academy

This article was reviewed by Megan Tillman's eighth grade students alongside their mentors, Hector Arciniega and Carissa Romero. The class found real value in understanding how scientist use magnetic resonance imaging (MRI) to study the brain. The students enjoyed being scientists during the peer review process and cannot wait to see the article published.

Abstract

To study the brain, scientists can use a machine called an MRI (magnetic resonance imaging) scanner. An MRI scanner takes pictures of the brain in a safe way, allowing scientists to learn about the structure of the brain and its functions. MRI helps scientists learn which areas of the brain are active when you engage in different activities, such as reading a sentence like this one! First, this article explains how the MRI scanner safely takes high-quality pictures of the brain. Next, we will explain how the MRI scanner can help scientists learn how the brain functions by measuring activity in different parts of the brain. Finally, we describe what it is like to participate in a study involving the MRI scanner and the kinds of questions scientists can answer using the MRI scanner as a tool.


Oxygenated hemoglobin in MRI - Biology

Avni R 1 , Garbow JR 2 , Neeman M 1

1 Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel, 2 Mallinckrodt Institute of Radiology, Washington University, St. Louis, Missouri, USA

Oxygen transport, one of several key functions performed by the placenta, depends primarily on placental oxygen pressure gradient and the oxygen affinity of fetal and maternal blood. To compensate for its low oxygen tension environment, fetal hemoglobin has a greater oxygen affinity than adult hemoglobin. Obtaining oxygen-hemoglobin dissociation curves and extracting P50 values, characteristic of oxygen affinity, may provide useful information, both in animal models and in clinical settings. We describe a novel, non-invasive MRI method for deriving MRI-based oxygen-hemoglobin dissociation curves.

Pregnant ICR mice were analyzed using a gradual respiration challenge from hyperoxia to hypoxia on E14.5 (n=8 mice 58 fetuses), and E17.5 (n=10 mice 89 fetuses). R1 and [1-∆R2*/R2*] values derived in the placenta, fetal liver and maternal liver at each oxygen phase demonstrated the expected sigmoid-shaped curve, with a clear difference between adult and fetal tissues, manifested by a shift to the left in the curves (Figure A). Apparent P50 (AP50) values, derived from the curves, demonstrate significantly lower AP50 values in fetal liver than in maternal liver (Figure B). AP50 maps inside the placenta and fetal liver show heterogeneity within these tissues (Figure C).

In conclusion, we present here a non-invasive approach to probe and quantify oxygen transfer across the placenta. This method may be useful for evaluation of fetal health.

Figure (A) Representative examples of MRI-based oxygen hemoglobin dissociation curves in placenta, fetal liver and maternal liver (B) Mean apparent P50 values on E14.5 (C) Representative AP50 maps inside the placenta and fetal liver on E14.5.


Oxygenated hemoglobin in MRI - Biology

RSR13, an allosteric modifier of hemoglobin, reduces hemoglobin-oxygen binding affinity facilitating oxygen release from hemoglobin, resulting in increases in tissue pO2. The purpose of this study was noninvasively to monitor the time course and effect of RSR13 on tumor oxygenation, directly using in vivo electron paramagnetic resonance (EPR oximetry), and indirectly using blood oxygen level dependent magnetic resonance imaging (BOLD MRI).

Methods and materials

The study was performed in transplanted radiation-induced fibrosarcoma tumors (RIF-1) in 18 female C3H/HEJ mice, which had two lithium phthalocyanine (LiPc) deposits implanted in the tumor when the tumors reached about 200–600 mm 3 . Baseline EPR measurements were made daily for 3 days. Then, for 6 consecutive days and after an initial baseline EPR measurement, RSR13 (150 mg/kg) or vehicle (same volume) was injected intraperitoneally, and measurements of intratumoral oxygen were made at 10-min intervals for the next 60 min. In each mouse, every third day, instead of EPR oximetry, BOLD MRI measurements were made for 60 min after administration of the RSR13.

Results

Based on EPR measurements, RSR13 produced statistically significant temporal increases in tumor pO2 over the 60-min time course, which reached a maximum at 35–43 min postdose. The average time required to return to the baseline pO2 was 70–85 min. The maximum increase in tumor tissue pO2 values after RSR13 treatment from Day 1 to Day 5 (8.3–12.4 mm Hg) was greater than the maximum tumor tissue pO2 value for Day 6 (4.7 mm Hg, p < 0.01). The maximum increase in pO2 occurred on Day 2 (12.4 mm Hg) after RSR13 treatment. There was little change in R2*, indicating that the RSR13 had minimal detectable effects on total deoxyhemoglobin and hemoglobin-oxygen saturation.

Conclusion

The extent of the increase in tumor pO2 achieved by RSR13 would be expected to lead to a significant increase in the effectiveness of tumor radiotherapy. The lack of a change in the BOLD MRI signal suggests that the tumor physiology was largely unchanged by RSR13. These results illustrate a unique and useful capability of in vivo EPR oximetry and BOLD MRI to obtain repeated measurements of tumor oxygenation and physiology. The dynamics of tumor pO2 after RSR13 administration may be useful for the design of clinical protocols using allosteric hemoglobin effectors.


Results

R1 Biomarkers But Not R2* Biomarkers Relate to Hypoxia in 786𠄰-R Xenographs

The relationships of MRI biomarkers of hypoxia and tissue pathologic assessment were determined in the 786𠄰-R xenografts. Median values of native R2* and oxygen-induced ∆R2* taken across the entire image did not correlate with the hypoxic fraction measured at pimonidazole adduct formation ( Figs 3a , ​ ,3b). 3b ). Hypoxic fraction was related to median values of oxygen-induced ∆R1 (ρ, 𢄠.783 P = .013 Fig 3c ) and the perfused Oxy-R fraction (ρ, 0.902 P = .002 Fig 3d ).

Graphs show the correlations between hypoxic fraction (expressed as a percentage and calculated from pimonidazole adduct formation immunohistochemistry images) and MRI biomarkers in 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice. Hypoxia did not correlate with (a) native R2* (R2*) or (b) oxygen-induced change in R2* (∆R2*), but it did correlate with (c) oxygen-induced change in R1 (∆R1) and (d) percentage of tumor perfused Oxy-R (nine mice for ac and eight mice for d).

Graphs show the correlations between hypoxic fraction (expressed as a percentage and calculated from pimonidazole adduct formation immunohistochemistry images) and MRI biomarkers in 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice. Hypoxia did not correlate with (a) native R2* (R2*) or (b) oxygen-induced change in R2* (∆R2*), but it did correlate with (c) oxygen-induced change in R1 (∆R1) and (d) percentage of tumor perfused Oxy-R (nine mice for ac and eight mice for d).

Graphs show the correlations between hypoxic fraction (expressed as a percentage and calculated from pimonidazole adduct formation immunohistochemistry images) and MRI biomarkers in 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice. Hypoxia did not correlate with (a) native R2* (R2*) or (b) oxygen-induced change in R2* (∆R2*), but it did correlate with (c) oxygen-induced change in R1 (∆R1) and (d) percentage of tumor perfused Oxy-R (nine mice for ac and eight mice for d).

Graphs show the correlations between hypoxic fraction (expressed as a percentage and calculated from pimonidazole adduct formation immunohistochemistry images) and MRI biomarkers in 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice. Hypoxia did not correlate with (a) native R2* (R2*) or (b) oxygen-induced change in R2* (∆R2*), but it did correlate with (c) oxygen-induced change in R1 (∆R1) and (d) percentage of tumor perfused Oxy-R (nine mice for ac and eight mice for d).

Subregional Analysis Reveals the R2* and R1 Biomarker Relationship in 786𠄰-R Xenographs

The relationship between R2* and R1 biomarkers was compared. Initially, tumor-wise and voxel-wise analyses were investigated by following existing literature. Next, we used the combined OE and DCE MRI analysis to define the three subregions perfused Oxy-E tumor, perfused Oxy-R tumor, and nonperfused tumor.

Tumor-wise analysis.—Median values of native R2* and oxygen-induced ∆R2* were compared with median values of ∆R1 for each tumor. No significant correlations were observed.

Voxel-wise analysis.—Native R2* and ∆R1 did not have a significant relationship. In distinction, there was a highly significant but weak correlation between ∆R2* and ∆R1 (ρ, 0.230 P < .001 Fig E1 [online]). Voxels with greater negative gas-induced ∆R2* showed a smaller positive ∆R1, consistent with both being biomarkers of hypoxia. However, the relationship between ∆R2* and ∆R1 appeared complex and was not explained simply by the bimodal relationship predicted by the open L-shaped curve (24,27) ( Fig 1 ).

Parcellation analysis.—We defined subregional analysis on the basis of the hypoxia biomarker perfused Oxy-R. This approach was chosen because we had previously validated Oxy-R as a hypoxia biomarker in this xenograft model (21). Three subregions were defined on the basis of combined OE MRI and DCE MRI signals ( Fig 1 ). Native R2* and oxygen-induced ∆R2* were compared for each of these subregions. In all, 5815 voxels were included and analyzed, of which 488 (8.4%) were nonperfused 4547 (78.2%) were defined as perfused Oxy-E, suggestive of a normoxic profile and 780 (13.4%) were defined as perfused Oxy-R, suggestive of a hypoxic profile. Perfused Oxy-R voxels had faster native R2* (P < .001 Fig 4a ) and greater negative hyperoxia-induced ∆R2* (P < .001 Fig 4b ) than the perfused Oxy-E and nonperfused voxels. Example tumor parametric maps are shown with corresponding pathologic validation across the range of hypoxia measured ( Fig 5 ).

Box-and-whisker plots show relationship of voxel values of (a) native R2* (R2*) and (b) oxygen-induced change in R2* (∆R2*) to tumor subregions categorized by perfused Oxy-E, nonperfused (NP), and perfused Oxy-R in 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice (n = 8). Data are medians and interquartile range.

Box-and-whisker plots show relationship of voxel values of (a) native R2* (R2*) and (b) oxygen-induced change in R2* (∆R2*) to tumor subregions categorized by perfused Oxy-E, nonperfused (NP), and perfused Oxy-R in 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice (n = 8). Data are medians and interquartile range.

Representative parametric maps of change in R2* (∆R2*), change in R1 (R1), and combined oxygen-enhanced MRI and dynamic contrast-enhanced MRI (quantifying perfused Oxy-E, perfused Oxy-R, and nonperfused tumor) are shown for three 786𠄰-R tumors propagated in 8-week-old female C.B17-scid mice, showing least, middle and greatest hypoxic fractions (HF) measured by pimonidazole adduct formation.

Technique Translation to Clinical Data

To test clinical translation, we recruited seven patients with clear cell RCC at radiologic assessment that was confirmed at subsequent histopathologic analysis ( Table 2 ). The combined OE MRI and DCE MRI analysis requires reliable definition of voxels that are refractory to oxygen challenge. Data from the ML206 gas analyzer in all seven patients showed statistically significant increase in oxygen concentration to greater than 90% during gas challenge (sample trace in Fig E2 [online]).

Table 2:

Patient Demographics, Stage, and Biomarker Values

Note.—∆R1 = change in R1, ∆R2* = change in R2*, F = female, M = male, NA = not available.

*Failed quality control for oxygen-enhanced MRI ∆R1 and ∆R2*.

As an additional quality control step, we evaluated the ∆R1 in the renal cortex to act as a positive control for oxygen delivery because positive ∆R1 has been consistently reported in multiple OE MRI studies (15�) of the kidney. We evaluated renal cortex regions of interest for evidence of oxygen enhancement (Fig E3a [online]) and generated combined OE MRI and DCE MRI maps for these regions (Fig E3b [online]). These analyses showed that whereas all patients received high concentration oxygen, one patient failed to inhale the gas sufficiently to generate signal change in the renal cortex (only 3.0% of voxels were oxygen enhancing). All other patients with renal cortex in the field of view had significant positive ∆R1 in the renal cortex with between 83.7% and 100% (mean, 95.4%) of OE voxels ( Table 2 ). Patient 7 had no normal kidney included in the field of view, but equivalent analysis of the spleen confirmed successful oxygen enhancement.

Consistent Relationship between R2* and R1 Biomarkers Found in Human RCC Tumors

The analyses developed in the 786𠄰-R xenografts were applied to the patient data. Patient 6 tumor data were excluded because this patient failed quality control checks on the basis of renal cortex analysis. This tumor did not show significant oxygen enhancement in 84.3% of its voxels (Fig E4 [online]), which is consistent with a failure in gas delivery.

Tumor-wise analysis.—Median values of native R2* and gas-induced ∆R2* were compared with median values of ∆R1 for each tumor (n = 6). No significant correlations were observed.

Voxel-wise analysis.—Native R2* and ∆R1 did not have a significant relationship. However, there was a highly significant but weak correlation between ∆R2* and ∆R1 (ρ, 0.035 P < .001). Voxels with greater negative gas-induced ∆R2* showed a smaller change in R1 (Fig E5 [online]).

Parcellation analysis.—Native R2* and gas-induced ∆R2* were compared for each of three subregions, defined by their combined signals at OE MRI and DCE MRI. In total, 4112 voxels were measured, of which 436 (10.6%) were nonperfused, 2887 (70.2%) were defined as perfused Oxy-E suggestive of a normoxic profile, and 789 (19.2%) were defined as perfused Oxy-R suggestive of a hypoxic profile. Statistically significant differences were observed between the perfused Oxy-R voxels (predicted to be hypoxic) and both the perfused Oxy-E and the nonperfused voxels, with faster native R2* and a greater negative gas-induced ∆R2* in the perfused Oxy-R voxels (both P < .001) ( Figs 6a , ​ ,6b 6b ).

Box-and-whisker plots show relationship of voxel values of (a) native R2* (R2*) and (b) oxygen-induced change in R2* (∆R2*) to tumor subregions categorized by perfused Oxy-E, nonperfused Oxy-R, and perfused Oxy-R in patients with renal cell carcinoma (n = 6). Data are medians and interquartile range.

Box-and-whisker plots show relationship of voxel values of (a) native R2* (R2*) and (b) oxygen-induced change in R2* (∆R2*) to tumor subregions categorized by perfused Oxy-E, nonperfused Oxy-R, and perfused Oxy-R in patients with renal cell carcinoma (n = 6). Data are medians and interquartile range.

In an exploratory analysis, we scored tumor hypoxia by GLUT1 staining. Although the study was not powered formally, the four tumors with MRI low hypoxic fraction (9.1%, 6.6%, 1.8%, and 0.6%) had GLUT1 hypoxia scores of 4.2, 2, 10.3, and 1.7, respectively, whereas the two tumors with high MRI hypoxic fraction (31.7% and 28.8%) had GLUT1 hypoxia scores of 19.5 and 41.7, respectively ( Fig 7 ). Therefore, OE MRI helped to categorize the six patient tumors into two groups and helped to detect significant separation in GLUT1 hypoxia score (P = .003).

Relationship of perfused Oxy-R to hypoxia in patients with renal cell carcinoma. Parametric maps of perfused Oxy-E, perfused Oxy-R, and nonperfused subregions are shown for four patients with relatively low perfused Oxy-R fraction, with immunohistochemistry images for the hypoxia-regulated gene glucose transporter 1 used to obtain an indirect assessment of tumor hypoxia. For comparison, equivalent parameter maps and immunohistochemistry images (magnification, 40×) are shown for two patients with relatively high perfused Oxy-R fraction.


Background The countercurrent arrangement of capillary blood flow in the medulla of mammalian kidneys generates a gradient of oxygen tension between the renal cortex and the papillary tip that results in a state of relative hypoxia within the renal medulla. Exploration of the pathophysiological implications of medullary hypoxia has been hampered by the absence of a noninvasive technique to estimate intrarenal oxygenation in different zones of the kidney. In the present study, we demonstrate the feasibility of such a method on the basis of blood oxygenation level–dependent (BOLD) MRI, which allows sequential measurements in humans in response to a variety of physiological/pharmacological stimuli in health and disease.

Methods and Results BOLD MRI measurements were obtained in healthy young human subjects (n=7), and the effects of three different pharmacological/physiological maneuvers that induce diuresis were studied. Spin-spin relaxation rate, R2*, was measured, which is directly related to the amount of deoxyhemoglobin in blood and in turn to tissue P o 2. Furosemide but not acetazolamide (n=6 each) increased medullary oxygenation (ΔR2*=7.62 Hz P<.01), consistent with the separate sites of action of these diuretics in the nephron and with previous direct measurements of their effects in anesthetized rats with oxygen microelectrodes. A new finding is that water diuresis improves medullary oxygenation (ΔR2*=6.43 Hz P<.01) in young human subjects (n=5).

Conclusions BOLD MRI can be used to monitor changes in intrarenal oxygenation in humans in a noninvasive fashion.

Because of the countercurrent arrangement of vessels and tubules necessary to conserve water and produce concentrated urine, the tubules of the renal medulla of land mammals must actively reabsorb sodium in a milieu poor in oxygen. 1 As a result, there is normally a marked gradient in P o 2 between circulating blood in vessels of the renal cortex and the medulla, so that the medulla has been described as operating habitually on the brink of anoxia. 2 Medullary hypoxia has important implications for the mechanisms of disease in the kidneys, particularly in acute renal failure resulting from circulatory or toxic insults in which it is hypothesized to be causative. 1

By use of sensitive but fragile and expensive glass microelectrodes, direct measurements of tissue P o 2 in the cortex and medulla of anesthetized rats have shown that loop diuretics (eg, furosemide) increase medullary P o 2 by diminishing the work of transport in medullary thick limbs. 3 By contrast, acetazolamide, a proximal tubular diuretic, increases cortical P o 2 slightly but does not raise medullary P o 2. 3 These changes in P o 2 are caused largely by changes in oxygen consumption, because little alteration in regional blood flow was noted with laser Doppler probes. 3 A major roadblock to extending these observations to human subjects has been the absence of a noninvasive method to assess regional oxygenation within the kidney.

It is known that oxyhemoglobin is diamagnetic and deoxyhemoglobin is paramagnetic. 4 Microscopic field gradients in the vicinity of red blood cells and vessels are modulated by changes in deoxyhemoglobin concentration. Such magnetic field perturbations within a voxel (volume element) cause a loss of phase coherence and therefore lead to signal attenuation in gradient echo or T2* (apparent spin-spin relaxation time)–weighted sequences. This phenomenon has been called blood oxygenation level–dependent (BOLD) contrast. 5

In the present study, we have applied noninvasive BOLD MRI to evaluate the level of oxygenation in the kidney. Specific objectives for the study were (1) to determine whether BOLD MRI in kidneys of humans is feasible (2) to study the effects of two diuretics previously shown to have different specific effects on the oxygenation of the renal medulla and cortex in rats and the effect of water load on renal oxygenation, which has not been previously investigated and (3) to distinguish observed changes caused by BOLD effects from possible changes in water content.

Methods

Effect of Oxygenation Changes on BOLD MRI Signal Intensity

MRI signal intensity measurements using a gradient-echo sequence were made at several different echo times, and the slope of loge (intensity) versus echo time was calculated. This slope determines the apparent spin-spin relaxation rate, R2* (=1/T2*), which is directly proportional to the tissue content of deoxyhemoglobin (Fig 1 ). A decrease in the slope (R2*) implies an increase in the oxygenation of hemoglobin. Because the oxygenation of hemoglobin is proportional to the P o 2 of blood and therefore in equilibrium with tissue P o 2, R2* is a sensitive indicator of tissue oxygenation. Alternatively, changes in gradient-echo signal intensity made at a single but sufficiently long echo time could be used directly to indicate qualitative changes in tissue oxygenation.

Because changes in the water content of tissue in addition to changes in deoxyhemoglobin content might conceivably change R2*, it is important to control for this variable. Spin-spin relaxation rate (R2) is known to show significantly less effect from changes in deoxyhemoglobin content 6 but has been shown to be very sensitive to changes in tissue water content. 7 To distinguish changes caused by BOLD effect from changes in water content, we obtained additional spin-echo data to estimate ΔR2.

Human Studies

We studied seven healthy human volunteers (6 men and 1 women age, 20 to 40 years) who gave informed consent in a protocol approved by the Beth Israel Hospital Committee on Clinical Investigation that conformed to the Declaration of Helsinki. Two or more studies were carried out in each subject. All studies were performed on a 1.5-T whole-body scanner (Vision Magnetom, Siemens Medical Systems) by use of echo planar imaging (EPI) acquisitions. Gradient-echo EPI images were acquired with three or more echo times in the range of 29 to 140 ms. All images were acquired during breathhold in expiration. Conventional MRI techniques, especially those using long echo times, are very sensitive to motion (cardiac, respiratory, peristaltic, etc). EPI is an ultrafast technique, with a typical image acquisition time of <100 ms, 8 and is therefore ideal for abdominal imaging, especially with long echo times. Other relevant sequence parameters are as follows: field of view=300 mm×300 mm, matrix size=128×128, and slice thickness=4 mm. Each acquisition was repeated three times for averaging purposes.

After scout images were obtained and the optimal positions were chosen, gradient-echo EPI data were acquired at different echo times. All the EPI data were obtained within a 15-minute interval. One of the following three stimuli was then used in each experiment: furosemide, 20-mg IV injection administered over 2 minutes acetazolamide, 500-mg IV injection administered over 2 minutes or water diuresis, ingestion of water, 20 mL/kg body weight in about 15 minutes. In these studies, the subject came to the laboratory in the morning after having abstained from food and water overnight. After the baseline BOLD data were obtained, the subject was taken out of the magnet to drink water.

BOLD MRI measurements were then repeated. When furosemide or acetazolamide was administered, MRI data acquisition was started 5 minutes after the injection. With water diuresis, the MRI data acquisition was resumed when urine flow exceeded 5 mL/min as estimated by measurement of the quantity of urine voided at 15-minute intervals after the water load. In four of the volunteers studied with furosemide and three with water load, we also obtained spin-echo EPI images, acquired with three or more echo times (59 to 160 ms), to calculate spin-spin relaxation rate, R2.

Region of interest (ROI) analysis was used to calculate regional relaxation rates. A T1-weighted anatomic image was acquired at the same slice position to facilitate placement of the ROIs. R2* and R2 were calculated by measuring the slope of straight line fit to the loge (intensity) versus echo time data. The mean of the three acquisitions was used for each data point in the slope analysis. Fig 2b illustrates an example of propagation of the ROIs. For statistical analysis of change in R2* and R2 before versus after stimuli, a two-tailed paired t test was used.

Results

BOLD MRI Applied to Intact Human Kidneys: Effects of Diuretics

Furosemide produced a decrease in R2* in the medulla of all six subjects studied. Typical changes are illustrated in the images of kidneys before and after furosemide administration in a single subject in Fig 2a and the graph of the slopes of loge (intensity) versus echo time in Fig 2c . In the cortex, R2* was unchanged by furosemide. Results of all subjects are summarized in the Table and shown in graphical form in Fig 3 .

Acetazolamide, on the other hand, produced no significant change in R2* in medulla or cortex, as summarized for all experiments in Fig 3 .

Effects of Water Diuresis

In five subjects (four men and one women ages, 20 to 40 years), water diuresis caused a consistent and substantial decrease in R2* in renal medulla, signifying an increase in medullary P o 2. Data from all water diuresis experiments are summarized in the Table and given graphically in Fig 4 . Water diuresis did not affect BOLD MRI signals in the renal cortex. As Fig 4 shows, water diuresis did not change R2 in either medulla or cortex, implying little change in tissue water content.

Discussion

A number of MRI studies have exploited the effect of oxygen on the magnetic state of hemoglobin. 5 9 10 11 12 Recently, this principle has been used to detect changes in cerebral oxygen tension and/or blood flow. Hoppel et al 13 detected a decrease in the T2* of tissue water in rabbit brains during hypoxia. Kwong et al 14 and Bandettini et al 15 demonstrated changes in T2*-weighted echo planar MRI images of human brain after visual stimulation and breathholding that were attributed to changes in oxygen tension. These and similar studies have stimulated active research in the area of functional MRI of the brain.

Thulborn et al 10 showed that the T2 relaxation rate of whole blood is a linear function of the square of the fraction of hemoglobin that was deoxygenated. The effect was relatively insensitive to temperature and hematocrit over the physiological range but disappeared when the blood was hemolyzed. Because of the sigmoidal relationship of P o 2 to the oxygenation of hemoglobin, 16 changes in BOLD MR signal produced by changes in blood P o 2 can be expected to be most marked at low levels of P o 2 and relatively less sensitive at a P o 2 >40 mm Hg, at which point most hemoglobin is in the oxygenated form. This makes BOLD MRI ideally suited for oxygenation measurements in the renal medulla, where P o 2 is normally in the range of 15 to 20 mm Hg, 1 2 rather than in the cortex, where small changes in O2 tension might go undetected. Because the oxygen tension of blood should mirror that of the tissue being perfused, changes in BOLD signal intensity measured at a sufficiently long echo time, or in R2*, should reflect changes in the P o 2 of tissue (Fig 1 ). Calculation of R2* is more robust and precise, minimizes artifactual errors, and avoids confounding effects such as those caused by changes in the water content of tissues.

The validity of these measurements, at least in a qualitative sense, is strengthened by the correspondence of the changes we observed in humans after furosemide and acetazolamide were administered with those previously measured directly with oxygen microelectrodes in anesthetized rats. Furosemide, which inhibits active transport in the medullary thick ascending limbs, greatly increased medullary P o 2, whereas after administration of acetazolamide, which primarily inhibits proximal tubular reabsorption in the renal cortex, medullary P o 2 did not change. Significant changes in medullary R2* compared with R2 validate the presumption that the observed changes are dominated by the BOLD effect rather than changes in regional water content.

New information also is provided by the present study regarding the effects of water diuresis in humans. Water diuresis consistently increased medullary P o 2 in five healthy young subjects to a degree close to that observed after furosemide administration without altering the BOLD MRI signal from the cortex. Although water diuresis does not change total renal blood flow substantially, 17 it is possible that capillary flow to the renal medulla may be selectively increased. It is also likely that water diuresis is associated with a decrease in oxygen consumption in the medulla because of a decrease in active transport by cells lining the medullary thick ascending limbs. At least in young individuals, water diuresis is associated with a marked increase in urinary excretion of prostaglandin E2 (PGE2) and dopamine. 18 Both agents have local vasodilating effects and inhibit active reabsorptive transport in medullary tubules, actions that would increase medullary P o 2. 19 20 21 Vasopressin might also modulate active transport 22 and local blood flow to the renal medulla 23 24 in a way that would increase medullary P o 2 when its influence was removed, as in water diuresis. Because normal aging is associated with a loss of the ability to increase urinary prostaglandin E2 and dopamine during water diuresis, 18 it will be of interest to see whether aging also diminishes the effect of water diuresis to increase medullary P o 2, as estimated by BOLD MRI.

The precise quantification of tissue P o 2 in absolute terms will require suitable calibration. Furthermore, BOLD MRI cannot, of course, distinguish between changes in oxygenation produced by alterations in the supply of oxygen (blood flow) and in its consumption (active transport). Changes in the hemoglobin-O2 dissociation curve such as those produced by large changes in pH might also affect BOLD MRI, although the pH of the outer medulla of the kidney (as opposed to the distal portion of the inner medulla) probably does not differ significantly from that of peripheral blood. 25 26

The present experiments demonstrate that BOLD MRI can be used to study the effects of physiological and pharmacological perturbations and of disease processes on regional oxygenation within the kidney, sequentially and noninvasively, in human subjects.

Figure 1. Blood oxygenation level–dependent (BOLD) MRI changes with P o O2. The deoxygenation of hemoglobin changes its magnetic characteristics, leading to changes in a parameter of magnetic resonance called R2* (apparent spin-spin relaxation rate). R2* can be estimated from signal intensity measurements made at several different echo times (a through e). The slope of loge (intensity) vs echo time determines R2* and is directly related to the amount of deoxygenated blood. A decrease in the slope implies an increase in the P o O2 of blood. We can either measure the slope or obtain intensity measurements at a single echo time (eg, d) to detect a difference in P o O2. Because blood P o O2 is thought to be in rapid equilibrium with tissue P o O2, changes in BOLD signal intensity or R2* should reflect changes in the P o O2 of the tissue.

Figure 2. A, T2*-weighted echo planar images (echo time [TE]=29, 50, 80, and 100 ms) of the kidney of a human volunteer before (left) and after (right) administration of furosemide (20 mg IV). On the images made before furosemide administration, one can appreciate corticomedullary contrast that increases with echo time. Medulla appears dark (arrows) owing to the presence of more deoxygenated blood. After furosemide administration, the medulla appears lighter, implying improved oxygenation of blood and presumably of extravascular tissue in the medulla as corticomedullary contrast disappears. Note that the collecting system appears dilated after furosemide administration, which is consistent with diuresis. B, Example of a typical data set analyzed with regions of interest. Shown are prefurosemide echo planar images of the same kidney. Also included is the T1-weighted anatomic image. The regions of interest are defined over the anatomic image and then propagated through the echo planar images, and mean signal intensity (SI) within each region of interest is measured and used to generate loge (intensity) vs TE curve, which is then fit to a straight line to determine the slope (R2*). C, R2* calculated in renal medulla (top) and cortex (bottom) before and after administration of furosemide (20 mg IV). R2* decreases with decreased deoxyhemoglobin (or improved blood oxygenation). The difference in R2* seen after furosemide administration implies an increase in the oxygenation of blood in the medulla.

Figure 3. Change (before vs after diuretic) in R2* in medulla and cortex after administration of either furosemide or acetazolamide in human volunteers (n=6 for each). Data are mean±SD. NS=P≥.05. Also included is R2 data (n=4) with furosemide. Significantly greater change in R2* vs R2 indicates that this MRI technique is primarily responsible for the observed changes after administration of furosemide.

Figure 4. Changes (before vs after diuretic) in R2* (n=5) and R2 (n=3) in medulla and cortex after induction of water diuresis in human volunteers. Columns show mean±SD. NS=P≥.05. Significantly greater change in R2* vs R2 indicates that BOLD effect is primarily responsible for the observed changes after water load.

Table 1. Effects of Furosemide, Acetazolamide, and Water Diuresis on R2* and R2 in Renal Medulla and Cortex of Human Subjects

*Significantly different from before-stimulation value (P<.01).

†Not significantly different from before-stimulation value (P≥.05).

This work was supported in part by a biomedical engineering research grant from the Whitaker Foundation (Dr Prasad) and NIH (NIDDK) grant RO-18078 (Dr Epstein). We thank Katherine Spokes, John Cannillo, and Dr Wei Li for their technical assistance.


Abstract

Objectives— The contribution of endothelial function to tissue oxygenation is not well understood. Muscle blood oxygen level–dependent MRI (BOLD MRI) provides data largely dependent on hemoglobin (Hb) oxygenation. We used BOLD MRI to assess endothelium-dependent signal intensity (SI) changes.

Methods and Results— We investigated mean BOLD SI changes in the forearm musculature using a gradient-echo technique at 1.5 T in 9 healthy subjects who underwent a protocol of repeated acetylcholine infusions at 2 different doses (16 and 64 μg/min) and N G -monomethyl- l -arginine ( l -NMMA 5 mg/min) into the brachial artery. Sodium nitroprusside was used as a control substance. For additional correlation with standard methods, the same protocol was repeated, and forearm blood flow was measured by strain gauge plethysmography. We obtained a significant increase in BOLD SI during acetylcholine infusion (64 μg/min) and a significant decrease for l -NMMA infusion (P<0.005 for both). BOLD SI showed a different kinetic signal than did blood flow, particularly after intermittent ischemia and at high flow rates.

Conclusions— In standard endothelial function tests, BOLD MRI detects a dissociation of tissue Hb oxygenation from blood flow. BOLD MRI may be a useful adjunct in assessing endothelial function.

We used muscle blood oxygen level–dependent MRI (BOLD MRI) to study tissue Hb oxygenation in relation to postischemic hyperemia and endothelial stimulation. We found uncoupling of tissue Hb oxygenation from blood flow changes and conclude that BOLD MRI may provide additional information in assessing endothelial function.

The vascular endothelium regulates tone and tissue perfusion through release of vasoactive substances such as NO. Numerous studies assessed the effect of endothelial function, and particularly endothelial dysfunction, in humans. However, the contribution of endothelial function to the regulation of tissue oxygenation in humans is poorly understood. 1 Perfusion is not the sole mechanism that determines tissue oxygenation and metabolism. Therefore, blood flow measurements may have limited utility in this regard. Blood oxygen level–dependent MRI (BOLD MRI) reflects changes in the ratio of oxyhemoglobin to deoxyhemoglobin attributable to their different properties in a magnetic field. 2,3 The BOLD technique is well established for functional brain MRI 4 and has also been used to assess myocardial 5–9 and skeletal muscle ischemia. 10,11 More recently, the technique was applied to assess exercise-induced hyperemia in skeletal muscle. 12,13 Available MRI systems are able to generate BOLD-sensitive small volume data sets in <1 s. Thus, BOLD MRI can noninvasively detect changes in intravascular tissue oxygenation with high spatial and temporal resolution. Furthermore, the tissue access is potentially unrestricted. We investigated changes of tissue oxygenation as defined by BOLD MRI signal intensity (SI) compared with blood flow changes after validated standard effectors on vascular tone and applied direct vasodilation and vasodilation attributable to stimulation of the endothelium.

Methods

Protocol

We studied 9 normal healthy men (25±4 years of age 75±9 kg weight 180±9 cm height). They received no medications. Our internal review board approved all studies, and written informed consent was obtained. All studies were conducted between 9 am and noon after an overnight fast. A catheter (18G Braun AG) was introduced into the brachial artery of the nondominant arm, followed by a resting period of ≥30 minutes. We compared BOLD SI changes with changes in blood flow determined by strain gauge plethysmography (SGP). Therefore, subjects underwent the following protocols twice within 2 hours: once in the MRI scanner and once while forearm blood flow (FBF) changes were assessed with SGP. Protocols were performed in a random order and with a 60-minute intermission between techniques.

We used the following known stimulators of vasodilation.

Reactive hyperemia after intermittent ischemia, which leads to vasodilation through regional metabolic changes as well as secondary endothelium-mediated smooth muscle cell relaxation.

Administration of sodium nitroprusside (SNP), which directly exposes smooth muscle cells to NO and leads to their relaxation without primary endothelium stimulation (endothelium-independent vasodilation).

Administration of acetylcholine (ACh), which is a physiological stimulus for endothelial cells to produce NO, which again relaxes vascular smooth muscles.

In an initial series of experiments, we performed reactive hyperemia with 3 minutes of ischemia followed by a recovery period of 22 minutes and an incremental intra-arterial infusion of SNP (SNP2=2 μg/min SNP4=4 μg/min SNP6=6 μg/min for 5 minutes each) after a short break. Ischemia was induced by a rapid inflation of a cuff of a standard automatic blood pressure monitor (Omega Saegeling Medizintechnik), with a complete cessation of blood flow within 2.5 s. On another day, endothelium-dependent perfusion changes were elicited by incremental infusions of ACh (ACh16=16 μg/min ACh64=64 μg/min). Each infusion step lasted 4 minutes. After recovery, we infused the nonspecific NO synthase inhibitor N G -monomethyl- l -arginine ( l -NMMA) at a rate of 5 mg/min over 5 minutes. We used drug doses that are known to induce no systemic effects. 14 The study drug concentrations were adjusted to achieve a constant intra-arterial infusion rate of 1.2 mL/min. In the course of our studies, all 8 data sets obtained during the reactive hyperemia/SNP experiments were suitable for evaluation. During the ACh/ l -NMMA studies, significant motion artifacts occurred in 1 subject, making a reliable evaluation impossible.

Plethysmography

We used a Hokanson EC5R plethysmograph with mercury-in-silastic strain gauges that were wrapped around the forearm at its largest diameter. A wrist cuff was inflated to 50 mm Hg above systolic pressure 1 minute before the protocol started to exclude hand circulation. By intermittent application of a venous occlusion pressure of 50 mm Hg, the blood inflow into the forearm was measured every 15 s. Except for reactive hyperemia, flow values were calculated from 8 single measurements after flow reached a steady state in each section of the protocol.

Magnetic Resonance Imaging

We used a 1.5-T MRI system (Sigma CV/I GE Medical Systems), equipped with a cardiovascular-optimized gradient system (maximum gradient strength 40 mT/m slew rate 150 T/m per second). A quadrature head coil was used for excitation and reception to achieve a homogeneous radio frequency intensity profile. Subjects were asked to assume a prone “sphinx-like” position with their forearms in the head coil. A single cross-sectional slice was placed at the largest forearm diameter. We performed T2*-weighted single-shot, gradient-echo sequence, with an echo planar imaging readout using the following parameters: field of view 24×24 cm 2 pixel size 1.87×1.87 mm 2 slice thickness 10 mm echo time (TE) 18.5 ms and flip angle 20°. In the first series of experiments, we used a constant repetition time of 250 ms, and in the second set, we applied ECG triggering with a delay time of 200 ms to minimize the influence of flow-dependent saturation and T1 effects on the SI. We measured SI in a defined region of interest on the whole forearm musculature excluding large vessels and bones (Figure 1) using software designed for tracking the SI time course in functional MRI (Functool GE Medical Systems). In addition, the images of the contralateral (noninfusion) forearm were processed to identify variations in the systemic circulation.

Figure 1. T2*-weighted MRI of human forearms during ischemia (left) and during hyperemia (right). The arm undergoing intermittent occlusion of circulation is depicted on the left side. In this arm, the flow-induced intensity increase after cuff release can be observed clearly in large vessels, whereas BOLD SI variation in the muscles is too small to be identified visually. Note the irregularly shaped region of interest as drawn manually.

Comparison With Theoretical Models

To assess the contributions of different compartments to BOLD SI in human skeletal muscle, we compared our data with predictions of theoretical models of the BOLD effect. For estimating extravascular BOLD effects, we adapted the model of Bauer et al 3 to the skeletal muscle by using a typical set of tissue parameters (Table 1).For intravascular BOLD effects, we calculated BOLD SI changes on the basis of oxygenation-dependent changes of the blood relaxation rate as proposed by Silvennoinen et al. 15 Similar to earlier simulations by Meyer et al, 12 we used a constant relative blood volume of 3% and a muscle relaxation rate of 34/s.

TABLE 1. Tissue Parameters of Human Skeletal Muscle Used for Calculation of Extravascular BOLD Effects

Statistical Analysis

Descriptive statistics are expressed as mean±SEM. Statistical significance of BOLD SI changes between the sections of the protocol was assessed by ANOVA for repeated measurements. We used Bonferroni’s test to compare BOLD SI changes during infusion with mean control values. A P value <0.05 was considered significant.

Results

Postischemic Vasodilation

The SI in the T2*-weighted echo planar images yielded a signal-to-noise-ratio of 200 to 400. A low-frequency noise appeared synchronously on both forearms, resulting mainly from cardiac pulsation and breathing motion. Figure 1 shows representative axial BOLD images obtained during forearm ischemia and immediately after cuff release. The control forearm is depicted on the right side of each image. Reperfusion induced a strong SI increase in the luminal area of the visible vessels and, to a lesser extent, in the muscular tissue.

The time course of averaged BOLD SI and blood flow changes during the reactive hyperemia protocol are illustrated in Figure 2. Immediately after cuff inflation, the BOLD SI showed an exponential decay with a maximum decrease of 2.2±0.4% compared with baseline. After cuff release, the SI reached a maximum of 3.6±0.4% above baseline within ≈30 s and slowly returned to baseline values thereafter. The time course of the blood flow response was markedly different from that of the BOLD SI. The blood flow reached a maximum of 48.4±5.6 mL/min×100 mL tissue as early as 5 s after cuff release, with a rapid return to baseline (5.1±1.7 mL/min×100 mL tissue). The analysis of BOLD SI curves showed that the mean exponential time constant for the BOLD signal decrease during ischemia was 71±11 s, and the mean linear BOLD signal increase time (from 10% to 90% of the total increase after cuff release) was 9.5±1 s. The individually determined time constant for the decay in the BOLD SI after the maximum was 136±22 s, whereas the time constant for the blood flow decay subsequent to the blood flow overshoot was only 19.3±3.7 s.

Figure 2. BOLD SI changes (line) and FBF in a reactive hyperemia experiment. FBF and tissue oxygenation return to resting values on a different time scale. Squares indicate rest.

Intra-Arterial Infusions

SNP and ACh led to a dose-dependent BOLD SI increase, reflecting increased oxygenation attributable to increased blood flow. With both substances, blood flow as measured by SGP remained high throughout the infusion period, whereas the BOLD SI decreased before the end of infusion. Figure 3a shows a representative recording of BOLD SI and flow from a subject during incremental intra-arterial SNP infusions. Compared with baseline, the BOLD SI increases were 2.0±0.3%, 2.4±0.3%, and 2.6±0.3% during SNP infusions at 2, 4, and 6 μg/min, respectively (P<0.001). The corresponding FBF values were 17±2, 24±2, and 29±3 mL/min×100 mL tissue. Whereas FBF stabilized within 2 minutes at each infusion step, the BOLD SI did not reach a steady state. We induced endothelium-dependent vasodilatation with incremental intra-arterial ACh infusion. The BOLD SI and FBF changes in a representative subject are illustrated in Figure 3b. Whereas FBF rapidly approached steady state at each infusion step, BOLD SI decreased at each ACh infusion step early after reaching a maximum, despite persisting high blood flow. During ACh infusion at the high dose (ACh64), the maximal BOLD SI increase was 3.6±0.5% (P<0.005), corresponding to 8.4-fold FBF increase, whereas it was not significant during low-dose ACh16 (1.9±0.6, P>0.05 FBF increase 4.2-fold, P=0.38). Table 2 summarizes the results.

Figure 3. BOLD SI (oscillating line) and flow (solid line) changes during intra-arterial infusion of SNP (a) and incremental infusion of ACh, recovery, and l -NMMA (b).

TABLE 2. Observed Changes of Forearm BOLD SI and Blood Flow in the Settings of Reactive Hyperemia, Nitroprusside Administration, and Endothelial Stimulation

FBF and maximum BOLD SI values encountered in each section of the protocol are depicted in a scatter plot in Figure 4a. Although there was an almost-linear increase in FBF with incremental SNP doses, the BOLD SI increases reached saturation with increasing SNP dose. During intra-arterial l -NMMA, BOLD SI and FBF decreased markedly (change −1.6±0.2% [P<0.005] and −35%, respectively). Figure 4b shows the blood flow values obtained during l -NMMA and ACh infusions plotted against the BOLD SI values. The BOLD SI curve shows a steep increase at lower blood flow values and saturation at higher blood flow values. A 35% blood flow decrease during l -NMMA infusion roughly induced the same amount of BOLD SI change as the 4-fold blood flow increase during ACh16 infusion. BOLD SI at similar perfusion changes did not differ between endothelial-dependent and endothelial-independent studies (P=0.3).

Figure 4. Left, Scatter plot of FBF and BOLD SI changes during endothelium-independent blood flow changes after SNP (χ 2 of fit=0.00007). Right, Similar data comparing endothelium-dependent blood flow changes induced by incremental ACh and by l -NMMA (χ 2 of fit=0.00016). The BOLD SI saturates for increasing FBF because of saturating tissue oxygenation.

Comparison With Theoretical Models

For an estimation of the extracapillary BOLD effects, we inserted the measured FBF data at rest and after l -NMMA, ACh64, and SNP6 infusion, respectively, into Equations 1, 3, and 48, as described in Bauer et al. 3 As an approximation for a small intracapillary blood volume, we applied Equation 40 from the same publication. Compared with resting values, the calculations resulted in relative muscle relaxation rate changes (ΔR2*) of −0.013, 0.009, and 0.009 per second for l -NMMA, ACh64, and SNP6, respectively. The relative BOLD SI change was calculated by the relationship TE×ΔR2*, resulting in an absolute extravascular contribution of −0.024%, 0.016%, and 0.016% for l -NMMA, ACh64, and SNP6, respectively. These values are in the range of only 1% of the measured BOLD SI changes.

Using Equation 48 from Bauer et al to determine the flow-dependent hemoglobin (Hb) saturation and Equation 1 from Meyer et al to determine the relaxation rate of blood in the microcirculation, we could estimate intravascular BOLD effects. 3,15 Relative BOLD SI changes yielded −0.22, 0.21, and 0.20% for l -NMMA, ACh64, and SNP6, respectively, ≈10% of the observed total BOLD SI change. However, when we accounted for the relative blood volume expansion during hyperperfusion an increase in the relative blood volume from 3% to 4% resulted in a relative BOLD SI change of 1.8% for SNP6. 16,17

Discussion

To our knowledge, our study is the first to systematically correlate BOLD SI changes to endothelial-mediated blood flow changes in the human forearm. With increasing blood flow, we observed an increase in BOLD SI, with saturation at high blood flow levels. However, after intermittent ischemia, BOLD SI remained elevated even after normalization of blood flow. Because O2 consumption is constant for P o 2 levels of ≥1 mm Hg and thus does not account for the observed dissociation, the decay time constant for the BOLD MRI signal may indicate the presence of microvascular oxygen reserve.

Our reactive hyperemia results are consistent with the magnitude and dynamics of BOLD SI–related changes reported previously. 10,11 Very recent data by Towse et al also provided evidence for a dissociation of flow and BOLD SI. 18 Persisting high values of BOLD SI after blood flow normalization are met by postischemic and even posthyperemic elevations of oxygen saturation of venous blood and tissue oxygen tension. 19–21 Intermittent recruitment of capillaries after ischemia increases oxygenated blood content and leads to a more uniform and efficient O2 exchange between capillary blood and surrounding tissues. 16,22 Shortly after hyperemia, elevated tissue O2 tension may cause an intermittent decrease in O2 extraction until the preischemic equilibrium is reinstalled. The mechanism is amplified by flow-mediated NO release.

Endothelial NO mediates exercise-related responses 23,24 and reactive hyperemia. 14,25 Jordan et al recently showed that the prolonged postexercise tissue oxygenation and increased BOLD SI observed in wild-type mice was absent in endothelial NO synthase gene-disrupted mice. 13 Postexercise blood flow rapidly returned to baseline in both groups. This state of affairs may be explained by the fact that endothelial NO not only regulates vascular tone and, thus, O2 supply, but also O2 consumption. Exogenous as well as endogenous NO decreases mitochondrial O2 consumption, whereas endothelial NO synthase inhibition increases O2 uptake. 26–29 Thus, limitations in endothelial NO production may contribute to ischemia through a combination of limited O2 supply and failure to shut off O2 consumption.

Infusion of endothelium-dependent or endothelium-independent vasodilators did not lead to a stable BOLD SI plateau, whereas blood flow rapidly approached a steady state. Besides the mentioned impact of NO, this phenomenon may be explained in part by a tissue oxygen tension–dependent redistribution of blood volume into regions of lower O2 tension. In a previous study, regional blood flow and functional capillary density were inversely related to tissue O2 tension. 17

BOLD SI in our study approached saturation with increasing FBF. The response to endothelial-dependent and endothelial-independent vasodilatation was similar. The saturation of the BOLD SI suggests that microvascular Hb oxygenation gradually approached arterial Hb oxygenation as blood flow increased. Theoretically, changes in myoglobin oxygenation may also induce a BOLD effect. However, even with the l -NMMA–induced reduced blood flow, the myoglobin oxygen saturation remained ≈100% in our study. 11 Furthermore, myoglobin is fully reoxygenated within 15 s after reperfusion. Thus, we do not believe that myoglobin oxygenation saturation affected our results.

BOLD SI is affected by blood volume, blood flow, and tissue oxygenation thus, the observed BOLD SI changes could have been the result of factors other than oxygenation. However, studies of Li et al have shown that changes of BOLD SI in vivo are mainly attributable to changes of tissue oxygenation. 5 Contributions to the BOLD signal originate from intravascular and extravascular compartments. However, Meyer et al 12 showed that intravascular rather than extravascular effects mainly explain BOLD SI changes in the skeletal muscle. Variations in microvascular blood volume or in the mean angle between muscle fibers and the main field axis may alter the magnitude of calculated BOLD SI. A wide range of values has been reported particularly for capillary density, which is known to vary with the physical training level. 30 Furthermore, functional capillary density is dependent of tissue O2 tension and precapillary pressure. These variables may have altered significantly during the course of our experiments.

We could not measure FBF and BOLD SI simultaneously because SGP could not be performed within the MRI scanner. However, we standardized test conditions during both measurements. We maintained the conditions of SGP and BOLD MRI as similarly as possible. We did not perform simultaneous metabolic studies, for instance with magnetic resonance spectroscopy of myoglobin. Thus, we cannot comment on metabolic aspects of the observed changes however, coupling magnetic resonance spectroscopy with microdialysis techniques might provide interesting future perspectives.

The potential impact of our results is manifold.

The new technique allows for directly assessing oxygen availability as the regulated vascular metabolic variable itself instead of using surrogate markers such as blood flow or vessel diameters.

Mechanisms of regional oxygenation regulation in the tissue other than endothelium-dependent blood flow regulation can be investigated. Candidates could be precapillary shunts, new tissue regulators of oxygen consumption, or modifiers of Hb deoxygenation.

These mechanisms may be future targets for therapeutic approaches addressing tissue oxygenation.

BOLD MRI may provide an accurate, stable, and reproducible tool to assess endothelial function as an early hallmark of hypertension and atherosclerosis and their precursors diabetes, obesity, physical inactivity, lipid disorders, or systemic inflammation. Applied in research trials, it may overcome known limitations of established modalities.

Because BOLD MRI does not need local manipulations such as strain gauges and is largely independent from the location of the tissue to be studied, other areas such as intra-abdominal organs and the heart are within reach.

The method may be useful in monitoring changes during therapeutic interventions, particularly in areas inaccessible to other techniques.

We are thankful to Todd Anderson for his excellent revision of the manuscript and helpful discussions. We are furthermore indebted to the technical assistance of Kerstin Kretschel, Evelyn Polzin, and Ursula Wagner.


What is Deoxyhemoglobin

Deoxyhemoglobin is the hemoglobin that has released oxygen. The release of oxygen occurs at the metabolizing tissue due to the low pH, high carbon dioxide concentration, and low temperature. Deoxyhemoglobin is the tensed (T) state of hemoglobin due to the release of oxygen molecules.

Figure 2: Oxyhaemoglobin Dissociation Curve

Deoxyhemoglobin, which is purplish in color, is transported towards the heart through veins. Blood with deoxyhemoglobin is known as deoxygenated blood. It can bind with oxygen inside the lungs, forming oxyhemoglobin, which in turn, increases the pH of the blood.


Abstract

Objectives— The contribution of endothelial function to tissue oxygenation is not well understood. Muscle blood oxygen level–dependent MRI (BOLD MRI) provides data largely dependent on hemoglobin (Hb) oxygenation. We used BOLD MRI to assess endothelium-dependent signal intensity (SI) changes.

Methods and Results— We investigated mean BOLD SI changes in the forearm musculature using a gradient-echo technique at 1.5 T in 9 healthy subjects who underwent a protocol of repeated acetylcholine infusions at 2 different doses (16 and 64 μg/min) and N G -monomethyl- l -arginine ( l -NMMA 5 mg/min) into the brachial artery. Sodium nitroprusside was used as a control substance. For additional correlation with standard methods, the same protocol was repeated, and forearm blood flow was measured by strain gauge plethysmography. We obtained a significant increase in BOLD SI during acetylcholine infusion (64 μg/min) and a significant decrease for l -NMMA infusion (P<0.005 for both). BOLD SI showed a different kinetic signal than did blood flow, particularly after intermittent ischemia and at high flow rates.

Conclusions— In standard endothelial function tests, BOLD MRI detects a dissociation of tissue Hb oxygenation from blood flow. BOLD MRI may be a useful adjunct in assessing endothelial function.

We used muscle blood oxygen level–dependent MRI (BOLD MRI) to study tissue Hb oxygenation in relation to postischemic hyperemia and endothelial stimulation. We found uncoupling of tissue Hb oxygenation from blood flow changes and conclude that BOLD MRI may provide additional information in assessing endothelial function.

The vascular endothelium regulates tone and tissue perfusion through release of vasoactive substances such as NO. Numerous studies assessed the effect of endothelial function, and particularly endothelial dysfunction, in humans. However, the contribution of endothelial function to the regulation of tissue oxygenation in humans is poorly understood. 1 Perfusion is not the sole mechanism that determines tissue oxygenation and metabolism. Therefore, blood flow measurements may have limited utility in this regard. Blood oxygen level–dependent MRI (BOLD MRI) reflects changes in the ratio of oxyhemoglobin to deoxyhemoglobin attributable to their different properties in a magnetic field. 2,3 The BOLD technique is well established for functional brain MRI 4 and has also been used to assess myocardial 5–9 and skeletal muscle ischemia. 10,11 More recently, the technique was applied to assess exercise-induced hyperemia in skeletal muscle. 12,13 Available MRI systems are able to generate BOLD-sensitive small volume data sets in <1 s. Thus, BOLD MRI can noninvasively detect changes in intravascular tissue oxygenation with high spatial and temporal resolution. Furthermore, the tissue access is potentially unrestricted. We investigated changes of tissue oxygenation as defined by BOLD MRI signal intensity (SI) compared with blood flow changes after validated standard effectors on vascular tone and applied direct vasodilation and vasodilation attributable to stimulation of the endothelium.

Methods

Protocol

We studied 9 normal healthy men (25±4 years of age 75±9 kg weight 180±9 cm height). They received no medications. Our internal review board approved all studies, and written informed consent was obtained. All studies were conducted between 9 am and noon after an overnight fast. A catheter (18G Braun AG) was introduced into the brachial artery of the nondominant arm, followed by a resting period of ≥30 minutes. We compared BOLD SI changes with changes in blood flow determined by strain gauge plethysmography (SGP). Therefore, subjects underwent the following protocols twice within 2 hours: once in the MRI scanner and once while forearm blood flow (FBF) changes were assessed with SGP. Protocols were performed in a random order and with a 60-minute intermission between techniques.

We used the following known stimulators of vasodilation.

Reactive hyperemia after intermittent ischemia, which leads to vasodilation through regional metabolic changes as well as secondary endothelium-mediated smooth muscle cell relaxation.

Administration of sodium nitroprusside (SNP), which directly exposes smooth muscle cells to NO and leads to their relaxation without primary endothelium stimulation (endothelium-independent vasodilation).

Administration of acetylcholine (ACh), which is a physiological stimulus for endothelial cells to produce NO, which again relaxes vascular smooth muscles.

In an initial series of experiments, we performed reactive hyperemia with 3 minutes of ischemia followed by a recovery period of 22 minutes and an incremental intra-arterial infusion of SNP (SNP2=2 μg/min SNP4=4 μg/min SNP6=6 μg/min for 5 minutes each) after a short break. Ischemia was induced by a rapid inflation of a cuff of a standard automatic blood pressure monitor (Omega Saegeling Medizintechnik), with a complete cessation of blood flow within 2.5 s. On another day, endothelium-dependent perfusion changes were elicited by incremental infusions of ACh (ACh16=16 μg/min ACh64=64 μg/min). Each infusion step lasted 4 minutes. After recovery, we infused the nonspecific NO synthase inhibitor N G -monomethyl- l -arginine ( l -NMMA) at a rate of 5 mg/min over 5 minutes. We used drug doses that are known to induce no systemic effects. 14 The study drug concentrations were adjusted to achieve a constant intra-arterial infusion rate of 1.2 mL/min. In the course of our studies, all 8 data sets obtained during the reactive hyperemia/SNP experiments were suitable for evaluation. During the ACh/ l -NMMA studies, significant motion artifacts occurred in 1 subject, making a reliable evaluation impossible.

Plethysmography

We used a Hokanson EC5R plethysmograph with mercury-in-silastic strain gauges that were wrapped around the forearm at its largest diameter. A wrist cuff was inflated to 50 mm Hg above systolic pressure 1 minute before the protocol started to exclude hand circulation. By intermittent application of a venous occlusion pressure of 50 mm Hg, the blood inflow into the forearm was measured every 15 s. Except for reactive hyperemia, flow values were calculated from 8 single measurements after flow reached a steady state in each section of the protocol.

Magnetic Resonance Imaging

We used a 1.5-T MRI system (Sigma CV/I GE Medical Systems), equipped with a cardiovascular-optimized gradient system (maximum gradient strength 40 mT/m slew rate 150 T/m per second). A quadrature head coil was used for excitation and reception to achieve a homogeneous radio frequency intensity profile. Subjects were asked to assume a prone “sphinx-like” position with their forearms in the head coil. A single cross-sectional slice was placed at the largest forearm diameter. We performed T2*-weighted single-shot, gradient-echo sequence, with an echo planar imaging readout using the following parameters: field of view 24×24 cm 2 pixel size 1.87×1.87 mm 2 slice thickness 10 mm echo time (TE) 18.5 ms and flip angle 20°. In the first series of experiments, we used a constant repetition time of 250 ms, and in the second set, we applied ECG triggering with a delay time of 200 ms to minimize the influence of flow-dependent saturation and T1 effects on the SI. We measured SI in a defined region of interest on the whole forearm musculature excluding large vessels and bones (Figure 1) using software designed for tracking the SI time course in functional MRI (Functool GE Medical Systems). In addition, the images of the contralateral (noninfusion) forearm were processed to identify variations in the systemic circulation.

Figure 1. T2*-weighted MRI of human forearms during ischemia (left) and during hyperemia (right). The arm undergoing intermittent occlusion of circulation is depicted on the left side. In this arm, the flow-induced intensity increase after cuff release can be observed clearly in large vessels, whereas BOLD SI variation in the muscles is too small to be identified visually. Note the irregularly shaped region of interest as drawn manually.

Comparison With Theoretical Models

To assess the contributions of different compartments to BOLD SI in human skeletal muscle, we compared our data with predictions of theoretical models of the BOLD effect. For estimating extravascular BOLD effects, we adapted the model of Bauer et al 3 to the skeletal muscle by using a typical set of tissue parameters (Table 1).For intravascular BOLD effects, we calculated BOLD SI changes on the basis of oxygenation-dependent changes of the blood relaxation rate as proposed by Silvennoinen et al. 15 Similar to earlier simulations by Meyer et al, 12 we used a constant relative blood volume of 3% and a muscle relaxation rate of 34/s.

TABLE 1. Tissue Parameters of Human Skeletal Muscle Used for Calculation of Extravascular BOLD Effects

Statistical Analysis

Descriptive statistics are expressed as mean±SEM. Statistical significance of BOLD SI changes between the sections of the protocol was assessed by ANOVA for repeated measurements. We used Bonferroni’s test to compare BOLD SI changes during infusion with mean control values. A P value <0.05 was considered significant.

Results

Postischemic Vasodilation

The SI in the T2*-weighted echo planar images yielded a signal-to-noise-ratio of 200 to 400. A low-frequency noise appeared synchronously on both forearms, resulting mainly from cardiac pulsation and breathing motion. Figure 1 shows representative axial BOLD images obtained during forearm ischemia and immediately after cuff release. The control forearm is depicted on the right side of each image. Reperfusion induced a strong SI increase in the luminal area of the visible vessels and, to a lesser extent, in the muscular tissue.

The time course of averaged BOLD SI and blood flow changes during the reactive hyperemia protocol are illustrated in Figure 2. Immediately after cuff inflation, the BOLD SI showed an exponential decay with a maximum decrease of 2.2±0.4% compared with baseline. After cuff release, the SI reached a maximum of 3.6±0.4% above baseline within ≈30 s and slowly returned to baseline values thereafter. The time course of the blood flow response was markedly different from that of the BOLD SI. The blood flow reached a maximum of 48.4±5.6 mL/min×100 mL tissue as early as 5 s after cuff release, with a rapid return to baseline (5.1±1.7 mL/min×100 mL tissue). The analysis of BOLD SI curves showed that the mean exponential time constant for the BOLD signal decrease during ischemia was 71±11 s, and the mean linear BOLD signal increase time (from 10% to 90% of the total increase after cuff release) was 9.5±1 s. The individually determined time constant for the decay in the BOLD SI after the maximum was 136±22 s, whereas the time constant for the blood flow decay subsequent to the blood flow overshoot was only 19.3±3.7 s.

Figure 2. BOLD SI changes (line) and FBF in a reactive hyperemia experiment. FBF and tissue oxygenation return to resting values on a different time scale. Squares indicate rest.

Intra-Arterial Infusions

SNP and ACh led to a dose-dependent BOLD SI increase, reflecting increased oxygenation attributable to increased blood flow. With both substances, blood flow as measured by SGP remained high throughout the infusion period, whereas the BOLD SI decreased before the end of infusion. Figure 3a shows a representative recording of BOLD SI and flow from a subject during incremental intra-arterial SNP infusions. Compared with baseline, the BOLD SI increases were 2.0±0.3%, 2.4±0.3%, and 2.6±0.3% during SNP infusions at 2, 4, and 6 μg/min, respectively (P<0.001). The corresponding FBF values were 17±2, 24±2, and 29±3 mL/min×100 mL tissue. Whereas FBF stabilized within 2 minutes at each infusion step, the BOLD SI did not reach a steady state. We induced endothelium-dependent vasodilatation with incremental intra-arterial ACh infusion. The BOLD SI and FBF changes in a representative subject are illustrated in Figure 3b. Whereas FBF rapidly approached steady state at each infusion step, BOLD SI decreased at each ACh infusion step early after reaching a maximum, despite persisting high blood flow. During ACh infusion at the high dose (ACh64), the maximal BOLD SI increase was 3.6±0.5% (P<0.005), corresponding to 8.4-fold FBF increase, whereas it was not significant during low-dose ACh16 (1.9±0.6, P>0.05 FBF increase 4.2-fold, P=0.38). Table 2 summarizes the results.

Figure 3. BOLD SI (oscillating line) and flow (solid line) changes during intra-arterial infusion of SNP (a) and incremental infusion of ACh, recovery, and l -NMMA (b).

TABLE 2. Observed Changes of Forearm BOLD SI and Blood Flow in the Settings of Reactive Hyperemia, Nitroprusside Administration, and Endothelial Stimulation

FBF and maximum BOLD SI values encountered in each section of the protocol are depicted in a scatter plot in Figure 4a. Although there was an almost-linear increase in FBF with incremental SNP doses, the BOLD SI increases reached saturation with increasing SNP dose. During intra-arterial l -NMMA, BOLD SI and FBF decreased markedly (change −1.6±0.2% [P<0.005] and −35%, respectively). Figure 4b shows the blood flow values obtained during l -NMMA and ACh infusions plotted against the BOLD SI values. The BOLD SI curve shows a steep increase at lower blood flow values and saturation at higher blood flow values. A 35% blood flow decrease during l -NMMA infusion roughly induced the same amount of BOLD SI change as the 4-fold blood flow increase during ACh16 infusion. BOLD SI at similar perfusion changes did not differ between endothelial-dependent and endothelial-independent studies (P=0.3).

Figure 4. Left, Scatter plot of FBF and BOLD SI changes during endothelium-independent blood flow changes after SNP (χ 2 of fit=0.00007). Right, Similar data comparing endothelium-dependent blood flow changes induced by incremental ACh and by l -NMMA (χ 2 of fit=0.00016). The BOLD SI saturates for increasing FBF because of saturating tissue oxygenation.

Comparison With Theoretical Models

For an estimation of the extracapillary BOLD effects, we inserted the measured FBF data at rest and after l -NMMA, ACh64, and SNP6 infusion, respectively, into Equations 1, 3, and 48, as described in Bauer et al. 3 As an approximation for a small intracapillary blood volume, we applied Equation 40 from the same publication. Compared with resting values, the calculations resulted in relative muscle relaxation rate changes (ΔR2*) of −0.013, 0.009, and 0.009 per second for l -NMMA, ACh64, and SNP6, respectively. The relative BOLD SI change was calculated by the relationship TE×ΔR2*, resulting in an absolute extravascular contribution of −0.024%, 0.016%, and 0.016% for l -NMMA, ACh64, and SNP6, respectively. These values are in the range of only 1% of the measured BOLD SI changes.

Using Equation 48 from Bauer et al to determine the flow-dependent hemoglobin (Hb) saturation and Equation 1 from Meyer et al to determine the relaxation rate of blood in the microcirculation, we could estimate intravascular BOLD effects. 3,15 Relative BOLD SI changes yielded −0.22, 0.21, and 0.20% for l -NMMA, ACh64, and SNP6, respectively, ≈10% of the observed total BOLD SI change. However, when we accounted for the relative blood volume expansion during hyperperfusion an increase in the relative blood volume from 3% to 4% resulted in a relative BOLD SI change of 1.8% for SNP6. 16,17

Discussion

To our knowledge, our study is the first to systematically correlate BOLD SI changes to endothelial-mediated blood flow changes in the human forearm. With increasing blood flow, we observed an increase in BOLD SI, with saturation at high blood flow levels. However, after intermittent ischemia, BOLD SI remained elevated even after normalization of blood flow. Because O2 consumption is constant for P o 2 levels of ≥1 mm Hg and thus does not account for the observed dissociation, the decay time constant for the BOLD MRI signal may indicate the presence of microvascular oxygen reserve.

Our reactive hyperemia results are consistent with the magnitude and dynamics of BOLD SI–related changes reported previously. 10,11 Very recent data by Towse et al also provided evidence for a dissociation of flow and BOLD SI. 18 Persisting high values of BOLD SI after blood flow normalization are met by postischemic and even posthyperemic elevations of oxygen saturation of venous blood and tissue oxygen tension. 19–21 Intermittent recruitment of capillaries after ischemia increases oxygenated blood content and leads to a more uniform and efficient O2 exchange between capillary blood and surrounding tissues. 16,22 Shortly after hyperemia, elevated tissue O2 tension may cause an intermittent decrease in O2 extraction until the preischemic equilibrium is reinstalled. The mechanism is amplified by flow-mediated NO release.

Endothelial NO mediates exercise-related responses 23,24 and reactive hyperemia. 14,25 Jordan et al recently showed that the prolonged postexercise tissue oxygenation and increased BOLD SI observed in wild-type mice was absent in endothelial NO synthase gene-disrupted mice. 13 Postexercise blood flow rapidly returned to baseline in both groups. This state of affairs may be explained by the fact that endothelial NO not only regulates vascular tone and, thus, O2 supply, but also O2 consumption. Exogenous as well as endogenous NO decreases mitochondrial O2 consumption, whereas endothelial NO synthase inhibition increases O2 uptake. 26–29 Thus, limitations in endothelial NO production may contribute to ischemia through a combination of limited O2 supply and failure to shut off O2 consumption.

Infusion of endothelium-dependent or endothelium-independent vasodilators did not lead to a stable BOLD SI plateau, whereas blood flow rapidly approached a steady state. Besides the mentioned impact of NO, this phenomenon may be explained in part by a tissue oxygen tension–dependent redistribution of blood volume into regions of lower O2 tension. In a previous study, regional blood flow and functional capillary density were inversely related to tissue O2 tension. 17

BOLD SI in our study approached saturation with increasing FBF. The response to endothelial-dependent and endothelial-independent vasodilatation was similar. The saturation of the BOLD SI suggests that microvascular Hb oxygenation gradually approached arterial Hb oxygenation as blood flow increased. Theoretically, changes in myoglobin oxygenation may also induce a BOLD effect. However, even with the l -NMMA–induced reduced blood flow, the myoglobin oxygen saturation remained ≈100% in our study. 11 Furthermore, myoglobin is fully reoxygenated within 15 s after reperfusion. Thus, we do not believe that myoglobin oxygenation saturation affected our results.

BOLD SI is affected by blood volume, blood flow, and tissue oxygenation thus, the observed BOLD SI changes could have been the result of factors other than oxygenation. However, studies of Li et al have shown that changes of BOLD SI in vivo are mainly attributable to changes of tissue oxygenation. 5 Contributions to the BOLD signal originate from intravascular and extravascular compartments. However, Meyer et al 12 showed that intravascular rather than extravascular effects mainly explain BOLD SI changes in the skeletal muscle. Variations in microvascular blood volume or in the mean angle between muscle fibers and the main field axis may alter the magnitude of calculated BOLD SI. A wide range of values has been reported particularly for capillary density, which is known to vary with the physical training level. 30 Furthermore, functional capillary density is dependent of tissue O2 tension and precapillary pressure. These variables may have altered significantly during the course of our experiments.

We could not measure FBF and BOLD SI simultaneously because SGP could not be performed within the MRI scanner. However, we standardized test conditions during both measurements. We maintained the conditions of SGP and BOLD MRI as similarly as possible. We did not perform simultaneous metabolic studies, for instance with magnetic resonance spectroscopy of myoglobin. Thus, we cannot comment on metabolic aspects of the observed changes however, coupling magnetic resonance spectroscopy with microdialysis techniques might provide interesting future perspectives.

The potential impact of our results is manifold.

The new technique allows for directly assessing oxygen availability as the regulated vascular metabolic variable itself instead of using surrogate markers such as blood flow or vessel diameters.

Mechanisms of regional oxygenation regulation in the tissue other than endothelium-dependent blood flow regulation can be investigated. Candidates could be precapillary shunts, new tissue regulators of oxygen consumption, or modifiers of Hb deoxygenation.

These mechanisms may be future targets for therapeutic approaches addressing tissue oxygenation.

BOLD MRI may provide an accurate, stable, and reproducible tool to assess endothelial function as an early hallmark of hypertension and atherosclerosis and their precursors diabetes, obesity, physical inactivity, lipid disorders, or systemic inflammation. Applied in research trials, it may overcome known limitations of established modalities.

Because BOLD MRI does not need local manipulations such as strain gauges and is largely independent from the location of the tissue to be studied, other areas such as intra-abdominal organs and the heart are within reach.

The method may be useful in monitoring changes during therapeutic interventions, particularly in areas inaccessible to other techniques.

We are thankful to Todd Anderson for his excellent revision of the manuscript and helpful discussions. We are furthermore indebted to the technical assistance of Kerstin Kretschel, Evelyn Polzin, and Ursula Wagner.


Watch the video: 10 Signs Your Body Is Crying Out For Help (January 2023).