Simultaneous and Noninvasive Imaging of Cerebral Oxygen Metabolic Rate, Blood Flow and Oxygen Extraction Fraction in Stroke Mice

Abstract

Many brain diseases have been linked to abnormal oxygen metabolism and blood perfusion; nevertheless, there is still a lack of robust diagnostic tools for directly imaging cerebral metabolic rate of oxygen (CMRO₂) and cerebral blood flow (CBF), as well as the oxygen extraction fraction (OEF) that reflects the balance between CMRO₂ and CBF. This study employed the recently developed in vivo ¹⁷O MR spectroscopic imaging to simultaneously assess CMRO₂, CBF and OEF in the brain using a preclinical middle cerebral arterial occlusion mouse model with a brief inhalation of ¹⁷O-labeled oxygen gas. The results demonstrated high sensitivity and reliability of the noninvasive ¹⁷O-MR approach for rapidly imaging CMRO₂, CBF and OEF abnormalities in the ischemic cortex of the MCAO mouse brain. It was found that in the ischemic brain regions both CMRO₂ and CBF were substantially lower than that of intact brain regions, even for the mildly damaged brain regions that were unable to be clearly identified by the conventional MRI. In contrast, OEF was higher in the MCAO affected brain regions. This study demonstrates a promising ¹⁷O MRI technique for imaging abnormal oxygen metabolism and perfusion in the diseased brain regions. This ¹⁷O MRI technique is advantageous because of its robustness, simplicity, noninvasiveness and reliability: features that are essential to potentially translate it to human patients for early diagnosis and monitoring of treatment efficacy.

Introduction

The brain is a highly aerobic organ that consumes oxygen extensively in order to produce chemical energy in the form of the adenosine triphosphate (ATP) molecule. The brain’s oxygen metabolism relies on an adequate oxygen supply facilitated by efficient blood circulation in the capillary bed. Deficits or abnormalities in cerebral oxygen metabolism and perfusion have been linked to many brain diseases such as stroke and tumor. Direct assessment and imaging of the cerebral metabolic rate of oxygen (CMRO₂) and blood flow (CBF) may provide potential biomarkers for clinical diagnosis and monitoring treatment efficacy of brain diseases. Another highly relevant physiological parameter is the oxygen extraction fraction (OEF) that reflects the interplay between the oxygen supplied through blood flow and the oxygen demand of brain tissues.

Positron emission tomography (PET) is the sole, established neuroimaging modality capable of imaging all parameters of CMRO₂, CBF and OEF using three radioactive tracers of ¹⁵O-isotope-labled oxygen gas (¹⁵O₂), water (H₂¹⁵O) and carbon monoxide (C¹⁵O) (Mintun et al., 1984; Ter-Pogossian et al., 1970). Due to the limitations of methodology complexity, invasiveness and radiation exposure, PET has not become a standard clinical diagnosis tool for imaging abnormal CMRO₂, CBF or OEF in patients (see more details in Discussion).

An alternative imaging approach for potentially mapping CMRO₂ and CBF is to apply in vivo ¹⁷O magnetic resonance (MR) based imaging approaches in combination with either the inhalation of non-radioactive ¹⁷O-isotope-labeled oxygen gas (¹⁷O₂) for assessing CMRO₂ or with a bolus injection of H₂¹⁷O tracer for imaging CBF (Arai et al., 1991; Atkinson and Thulborn, 2010; Fiat et al., 1992; Kwong et al., 1991; Mateescu, 2003; Mateescu et al., 1991; Mateescu et al., 1990; Mellon et al., 2010; Pekar et al., 1991; Ronen et al., 1998; Zhu et al., 2009; Zhu et al., 2002; Zhu et al., 2007). Although there is a similarity between PET and ¹⁷O MR imaging methods, only the latter specifically detects the metabolically generated H₂¹⁷O without confounding signals from ¹⁷O₂ and this merit significantly simplifies both imaging procedure and post-imaging quantification of CMRO₂ (see review articles (Zhu and Chen, 2011; Zhu et al., 2005) and the cited references therein). Several studies have also shown the ability of high-field MR scanners to greatly improve the ¹⁷O detection sensitivity (Lu et al., 2012; Thelwall et al., 2003; Zhu et al., 2001). Furthermore, a recent study has suggested that the H₂¹⁷O metabolically generated in the brain tissue could serve as an endogenous perfusion tracer during the post-¹⁷O₂-inhalation period for indirect assessment of CBF (Zhu et al., 2010).

Stroke is a leading cause of death and disability (Roger et al., 2011). The pathology, progression and treatment efficacy of stroke are extremely susceptible to the status of the cerebral oxygen metabolism (i.e., CMRO₂) and blood perfusion (i.e., CBF), as well as their balance that can be quantified by OEF. New evidence has indicated that the measures of CMRO₂ and OEF may provide a better assessment of viable brain tissue after a stroke attack for potential treatment compared to perfusion- or diffusion-weighted MRI, thus, the high-field ¹⁷O imaging approach might be potentially useful for the stroke patient management (Delapaz and Gupte, 2011; Heiss, 2011; Heiss and Sobesky, 2008; Zhu and Chen, 2011).

The objectives of this study were to exploit the feasibility of the ¹⁷O MR-based neuroimaging approach at high field for noninvasively and simultaneously imaging CMRO₂, CBF and OEF using a preclinical mouse model of middle cerebral artery (MCA) occlusion (MCAO); and to test the applicability of the ¹⁷O imaging approach for detecting the abnormal oxygen metabolism and perfusion associated with the brain stroke.

Materials and Methods

¹⁷O Isotope Mass Balance and ¹⁷O MRI

The majority (>99.9%) of the oxygen atoms in oxygen gas or water are ¹⁶O isotopes that do not have any MR signal. The sole MR-detectable oxygen isotope is ¹⁷O, which is non-radioactive and stable, but has a very low abundance (0.037%) in nature. The principle behind the ¹⁷O-MR based CMRO₂ imaging technique is to introduce the ¹⁷O-labeled oxygen gas (¹⁷O₂) into the animal or human body via inhalation (for several minutes) while imaging the dynamic change of the ¹⁷O-labeled water (H₂¹⁷O) metabolized from the ¹⁷O₂ in the brain (see review articles (Zhu and Chen, 2011; Zhu et al., 2005) and the cited references therein).

The inhaled oxygen gas will bind to hemoglobin in the blood through lung exchange, and then enter the brain via the feeding arteries and blood circulation. The labeled oxygen molecules will be metabolized in the mitochondria of brain cells to produce ¹⁷O labeled water molecules, which can be detected using a MR scanner. The net chemical reaction of this process is described as following:

$$4{\text{H}}^{+}+4{\text{e}}^{-}+{{}^{17}\text{O}}_2\iff 2{{\text{H}}_2}^{17}\text{O}$$

Note: the ¹⁷O₂ gas (reactant) is not visible to the ¹⁷O MR detection and only the final metabolic product, H₂¹⁷O, provides a MR signal. This makes the direct detection of H₂¹⁷O using ¹⁷O MRI simple, robust, and noninvasive.

After the brief ¹⁷O₂ inhalation, the inhaled gas is switched back to the regular, non-¹⁷O-labeled oxygen gas. Thus, no additional H₂¹⁷O molecules will be generated through oxygen metabolism. The produced H₂¹⁷O molecules during the ¹⁷O₂ inhalation period may serve as an endogenous tracer for CBF measurements.

The ¹⁷O MR images have been applied to directly determine the two important physiological quantities: CMRO₂ and CBF. The quantification is based on the mass balance equation of ¹⁷O labeled water content during and after the ¹⁷O₂ gas inhalation given by Eq. [1] (Zhang et al., 2004a; Zhu et al., 2002),

$$\frac{{\text{dC}}_{\text{b}}(\text{t})}{\text{dt}}=2\alpha \times {\text{CMRO}}_2+\text{CBF}({\text{C}}_{\text{a}}(\text{t})-\frac{{\text{C}}_{\text{b}}(\text{t})}{{\text{k}}_1})$$ [1]

where C_b(t) is the ¹⁷O-water content of the brain tissue metabolized from the inhaled ¹⁷O₂ gas (calibrated by the natural abundance ¹⁷O-water signal measured before the ¹⁷O₂ gas inhalation); C_a(t) is the metabolic ¹⁷O-water content of the artery blood; α is the ¹⁷O-isotope enrichment of inhaled ¹⁷O₂ gas and is a known constant; k₁ is another known constant (k₁=1.86) consistently across a wide range of brain physiology and non-physiology conditions (Zhu et al., 2010)).

Calculation of CMRO₂

It has been shown that during a short ¹⁷O₂ inhalation period (as used in this study) the temporal changes of the C_a(t) and C_b(t)/k₁ terms in Eq. [1] are similar for small animals (Zhang et al., 2004a; Zhu et al., 2002), thus, Eq. [1] becomes

$$\frac{{\text{dC}}_{\text{b}}(\text{t})}{\text{dt}}\cong 2\alpha \times {\text{CMRO}}_2.$$ [2]

The anti-derivative of Eq. [2] gives a linear equation with a slope determined by linear regression of the following equation:

$${\text{C}}_{\text{b}}(\text{t})\cong 2\alpha \times {\text{CMRO}}_2\times \text{t}.$$ [3]

Thus, CMRO₂ can be calculated as:

$${\text{CMRO}}_2=\frac{\text{Slope}\text{of}\text{Eq}.[3]}{2\alpha }$$ [4]

Calculation of CBF

After the ¹⁷O₂ inhalation, the newly formed H₂O molecules will no longer be labeled with ¹⁷O thus no more replenishing H₂¹⁷O molecules will be generated in the brain. The CMRO₂ term in Eq. [1] diminishes and C_a(t) reaches a plateau, thus:

$$\frac{{\text{dC}}_{\text{b}}(\text{t})}{\text{dt}}\cong -\frac{\text{CBF}}{{\text{k}}_1}{\text{C}}_{\text{b}}(\text{t})+{\text{k}}_2$$ [5]

where k₂ is a constant. Thus the anti-derivative of Eq. [5] gives an exponential equation with an exponential decay rate (D) that is proportional to CBF/k₁

$${\text{C}}_{\text{b}}(\text{t})={\text{k}}_3\times {\text{e}}^{-\frac{\text{CBF}\times \text{t}}{{\text{k}}_1}}+{\text{k}}_4$$ [6]

where k₃ and k₄ are constants, thus,

$$\text{CBF}={\text{k}}_1\times \text{D}.$$ [7]

Therefore, one can determine CMRO₂ from the C_b(t) time course during the period of ¹⁷O₂ inhalation, and CBF from the C_b(t) time course in the period after the ¹⁷O₂ inhalation is halted.

Calculation of OEF

$$\text{OEF}=\frac{{\text{CMRO}}_2}{{\text{C}}_{\text{a},{\text{O}}_2}\times \text{CBF}}$$ [8]

where C_a,O2 is the artery oxygen concentration (or input function) and is a constant for the brain. The common unit of the C_a,O2 is ml-O₂/dl-blood and it can be converted to μmol/ml-blood for matching the unit of CMRO₂ to CBF ratio according to Avogadro’s law, i.e., a mol of any gas occupies 22.4 liters at standard temperature and pressure. The conversion factor between them is 2.24, therefore

$$\text{OEF}=\frac{{\text{CMRO}}_2}{{\text{C}}_{\text{a},{\text{O}}_2}\times \text{CBF}}\times 2.24$$ [9]

The C_a,O2 value approximately ranges from 17 ml-O₂/dl-blood measured in the mouse brain (Frietsch et al., 2007) to 18.8 ml-O₂/dl-blood in the rat brain (unpublished data from our lab), and C_a,O2 ≈18 ml-O₂/dl-blood was applied in the present study for OEF calculation.

OEF ratio between ischemic and contralateral hemispheres

In this study, we determined the OEF values from 3D chemical shift imaging (CSI) voxels covering the left and right mouse hemispheres, then calculated the OEF ratio (R_OEF) between the paired voxels, located in the ischemic hemisphere (OEF_I) and the “mirror” voxel in the contralateral, control hemisphere (OEF_C).

$${\text{R}}_{\text{OEF}}=\frac{{\text{OEF}}_{\text{I}}}{{\text{OEF}}_{\text{C}}}$$ [10]

Animal Stroke Model

Four male C57BL/6 mice (18–30g body weight) underwent a 60-minute right middle cerebral artery (MCA) occlusion 5–7 days prior to MR scanning were used in this study. The mice were anesthetized with ketamine (87 mg/kg)/xylazine (13 mg/kg) cocktail solution via intraperitoneally injection followed by constant subcutaneous infusion at 2.6 ml/kg/hr of the same cocktail solution during the MR study. The mouse head was immobilized with tooth and ear bars. Spontaneous respiration and body temperature of the animal were monitored throughout the MR experiment; warm water and warm air were used to maintain the body temperatures of mice at ~37°C. Two to three minutes of ¹⁷O₂ (60% ¹⁷O enrichment, Sigma) gas was supplied to each mouse for ¹⁷O MR imaging experiments.

All experimental procedures and protocol were conducted under the guidelines of the National Institutes of Health and were approved by the Institutional Animal Care and Use Committee of the Washington University.

MRI/MRS Experiments

The ¹H anatomic brain images and ¹⁷O MRS imaging data were acquired using a 11.7T/26-cm clear bore horizontal animal magnet (Magnex Scientific, UK) interfaced with a Varian INOVA console (Varian Inc., Palo Alto, CA). A radiofrequency (RF) probe consisting of a single loop ¹⁷O surface coil (~1.2 cm diameter) and a larger quadrature ¹H coil was built for the ¹⁷O and ¹H imaging of the mouse brain. T₂-weighted fast spin echo image (repetition and echo times of TR/TE=4 s/15 ms; image matrix size of 256×256; field of view (FOV) of 2.5×2.5 cm²; 1 mm slice thickness, echo chain length of 8, 4 signal averages) and/or proton density gradient echo image sequences were used for acquiring the ¹H anatomic brain images.

Three-dimensional (3D) ¹⁷O MRS imaging data was acquired using the Fourier Series Window (FSW) CSI technique, in which the k-space sampling is weighted according to the Fourier coefficients of a predetermined voxel shape (Hendrich et al., 1994). The following acquisition parameters were used: TR/TE=10/0.4 ms, 50-μs hard RF pulse for a nominal 90° excitation, spectral width=30 kHz; 8.5 ms acquisition time; 2×2×2 cm³ FOV; 9×9×5 phase encodes; 40 μl (or 15 μl nominal) voxel size with a cylindrical voxel shape; and 11 s temporal resolution per 3D CSI dataset. A 17×17×9 matrix of free induced decay (FID) signals was generated from the original phase encode data for each 3D ¹⁷O image.

For ¹⁷O imaging data acquisition, a total of 100–120 3D CSI datasets were collected during pre-inhalation (Phase I), inhalation (Phase II) and post-inhalation (Phase III) periods. Two to three minutes of ¹⁷O₂ gas inhalation was applied for the noninvasive and simultaneous imaging of CMRO₂, CBF and OEF.

The imaged brain H₂¹⁷O signal (resonance peak height) was converted to the absolute brain concentration in mM (i.e., C_b(t) in Eq. [1]) based on its signal ratio to the mean natural abundance brain H₂¹⁷O signal acquired during pre-inhalation (Phase I) period and the known values of the ¹⁷O natural abundance and the brain water concentration (Zhang et al., 2004a; Zhu et al., 2002). The C_b(t) time course of the 3D CSI voxel measured during Phase II was used to perform linear regression and calculate the CMRO₂ value according to Eqs. [3] and [4]. The C_b(t) time course measured during Phase III was used to perform exponential regression and calculate the CBF value according to Eqs. [6] and [7]. For the voxels located in the ischemic brain regions, the initial few points of C_b(t) in Phase III, which not yet reached the peak H₂¹⁷O signal, were excluded from the exponential regression.

The CMRO₂ and CBF values of the same voxel were used to calculate OEF according to Eq. [9]. The OEF ratio, R_OEF, was determined using Eq. [10].

Statistical Analysis

All measurement results are presented as mean±standard deviation (SD). A paired student’s t test was applied for statistical analysis of the experimental data and for comparison of the results between the MCAO affected and intact brain regions. A p value of < 0.05 is considered to be statistically significant.

Results

Figure 1 illustrates multi-slice T₂-weighted ¹H MRI (Fig. 1A) and the corresponding 3D ¹⁷O CSI of the natural abundance brain H₂¹⁷O signal (Fig. 1B) from a representative MCAO mouse brain. The total acquisition time was 8 min for the ¹H MRI and 11 s for the 3D ¹⁷O CSI. The infarction in the right hemisphere caused by the MCA occlusion was detected as the hyper-intense lesions in the anatomic ¹H images (Fig. 1A). The spatial distribution of the natural abundance H₂¹⁷O signal in the mouse brain was consistent with the ¹H anatomic images and the ¹⁷O coil size; and the signal intensity reflected the RF field distribution of the ¹⁷O surface coil. The improved ¹⁷O detection sensitivity at 11.7 T was evident by the excellent spatial and temporal resolution of the 3D CMRO₂ and/or CBF imaging in this study.

Figure 1.

(A) Multi-slice T₂-weighted ¹H MRI (8 minutes of total acquisition time and 0.01 μl pixel size), and (B) the corresponding 3D-CSI images of the natural abundance H₂¹⁷O (11 seconds of total acquisition time and 15 μl nominal voxel size) from a representative MCAO mouse brain. The lesions with hyper-intensity in the right brain hemisphere caused by the right MCA occlusion are evident from the anatomic ¹H images. L: left side; R: right side; A: anterior image slice; and P: posterior image slice.

Figure 2A displays the stacked plots of the global brain H₂¹⁷O spectra acquired before, during and after a 2.5-min ¹⁷O₂ inhalation in a representative MCAO mouse at 11.7T. The time course of the H₂¹⁷O signal can be divided into three distinct phases: Phase I is the pre-inhalation period with a constant natural abundance brain H₂¹⁷O signal; Phase II is the inhalation period with an approximately linear increase of brain H₂¹⁷O signal; Phase III is the post-inhalation period with an exponential decay of brain H₂¹⁷O signal. Similar temporal pattern was also observed in voxel-based time courses, as shown in the example of a representative voxel (see Fig. 2B) taken from the 3D ¹⁷O CSI dataset (marked in Fig. 1B) with a temporal resolution of 11 seconds per CSI volume. The quality of the ¹⁷O MR spectra is displayed at bottom of the Fig. 2B for each of the three phases. The H₂¹⁷O signal measured in Phase I was used to quantify the absolute brain H₂¹⁷O concentrations measured during Phase II and Phase III. The time courses of the H₂¹⁷O concentrations in Phase II and III were used to calculate the CMRO₂ and CBF values, respectively.

Figure 2.

Stacked plots of ¹⁷O spectra of the H₂¹⁷O signals acquired before, during and after a 2.5-minute ¹⁷O₂ inhalations from a representative MCAO mouse brain at 11.7T. The ¹⁷O spectra of (A) the global brain signal observed with the ¹⁷O surface coil, and (B) a representative voxel (as shadowed in Fig. 1B) taken from the 3D ¹⁷O CSI data acquired with 11 seconds per 3D CSI volume are displayed. The H₂¹⁷O signal time course can be divided into three phases: pre-inhalation period (Phase I: constant phase), inhalation period (Phase II: linearly increasing phase) and post-inhalation period (Phase III: exponentially decay phase). The typical single-voxel spectra for each of three phases are shown at bottom of this figure.

Figure 3A shows the voxel-based comparison of two repeated CMRO₂ imaging measurements in the same mouse. All the data points across multiple CSI voxels (n=34) covering both ischemic and intact brain regions are located along the line of equality. The reproducibility of the CMRO₂ measurements was quantitatively assessed and the Bland-Altman plot between the differences and the mean CMRO₂ values of the two repeated measurements (Bland and Altman, 1986, 1999) is shown in Figure 3B. The mean (black line) and standard deviation (SD) of the difference between the 1^st and 2^nd CMRO₂ measurements were 0.11 μmol/g/min and 0.23 μmol/g/min, respectively. The gray lines in Fig. 3B define the limits of agreement where 95% of the difference values are expected to be less than two SD. Clearly, the majority of the data points were within the range except two outliers, which, interestingly, were from two voxels located in the periphery regions of the ¹⁷O surface coil with relatively poor detection sensitivity.

Figure 3.

Comparison of two repeated CMRO₂ imaging measurements in the same MCAO mouse. (A) The relationship between the 1^st and 2^nd CMRO₂ measurements across multiple CSI voxels (n=34) with a line of equality. (B) The Bland-Altman plot of the differences verse the mean values between the two repeated CMRO₂ measurements. The black line depicts the mean (0.11 μmol/g/min) of the difference between the two CMRO₂ measurements and the gray lines defines the limits of agreement where 95% of the difference values are expected to be less than two standard deviation (SD=0.23 μmol/g/min).

Figure 4 shows an example for comparing the time courses of the H₂¹⁷O water during the CMRO₂ and CBF measurements between the voxels located in the MCA occluded hemisphere (red circles) and the voxels located in the contralateral, intact hemisphere (blue circles). Two anatomic image slices from a representative MCAO mouse are shown in Figs. 4A and 4B. For each image slice, two voxels symmetrically located in the lesion area and in the contralateral hemisphere were identified. The time courses of the brain H₂¹⁷O concentration in such paired-voxels measured before, during and after a 2.5-min inhalation of ¹⁷O₂ were plotted together and analyzed carefully. It is clear that the H₂¹⁷O concentration increased during ¹⁷O inhalation (Phase II) followed by the H₂¹⁷O concentration decay during the post-inhalation period (Phase III). Both rates of H₂¹⁷O accumulation and decay were substantially reduced in the MCAO affected voxels compared to those of intact voxels, which revealed large reductions of CBF and CMRO₂ in ischemic brain (see Fig. 4A and Fig. 4B). Figure 4C shows the results from another MCAO mouse, in which an obvious different H₂¹⁷O dynamic was also observed in the MCAO affected brain region but without a discernible lesion in the anatomic image.

Figure 4.

Comparison of CMRO₂ and CBF measurements between the ¹⁷O CSI voxels located in the MAC occluded right hemisphere (red circles) and the voxels located in the contralateral left hemisphere of the same mouse brain (blue circles). (A) and (B) show the anatomic images, selected voxels and their corresponding dynamic brain H₂¹⁷O signal changes before, during and after a 2.5-minute ¹⁷O₂ inhalation from two representative image slices in the same MCAO mouse. Both the slope of ¹⁷O signal increase during the inhalation and the exponential decay rate during the post-inhalation phase were substantially smaller in the MCAO affected voxels compared to that corresponding voxels in intact hemisphere, indicating large reductions in both CMRO₂ and CBF. (C) Shows the similar results from a different MCAO mouse brain.

Figure 5 displays the measurement results obtained from two representative MCAO mice in imaging format. The ¹H images shown in Figs. 5A and 5B define the brain region of interest with high ¹⁷O detection sensitivity including both normal and MCAO affected brain regions. Figures 5C, 5D and 5E represent the CMRO₂, CBF and OEF maps of the same brain regions, respectively. Substantial reductions of CMRO₂ and CBF were observed in the MCA occluded right hemisphere. In contrast, OEF increased in the MCAO affected brain regions comparing with that of the contralateral control.

Figure 5.

Imaging results obtained from two representative MCAO mice (left versus right column). Anatomic brain images selected from a representative image slice (A), expanded anatomic image region of interest with high ¹⁷O detection sensitivity covering both MCA occluded and normal brain regions (B), and the corresponding CMRO₂ images (C), CBF images (D) and OEF images (E) are displayed. These images clearly demonstrate significant reductions of CMRO₂ and CBF in the right hemisphere affected by MCAO compared to the intact left hemisphere. In contrast, the values of OEF were increased in the MCAO affected brain regions.

Figure 6 summarizes the quantitative comparison results of the CMRO₂, CBF, OEF and OEF ratios in paired-voxels (i.e., the voxels in the MCA occluded hemisphere (hatched bars) vs. those in the contralateral, intact hemisphere (gray bars)) from a representative MCAO mouse brain. A total of 17 pairs of image voxels in the ¹⁷O coil sensitive region, covering multiple imaging slices, were selected. The CMRO₂ values of the voxels in the contralateral control hemisphere, CMRO_2,C, were relatively uniform with mean value of 2.63±0.16 μmol/g/min (see the dashed line in Fig. 6A), which was significantly higher than that of the voxels in the ischemic hemisphere, where mean CMRO_2,I value was 2.04±0.20 μmol/g/min (p=4×10⁻⁷). The CBF results (see Fig. 6B) show the similar pattern and the mean values of CBF_I (0.46±0.15 ml/g/min) measured in the MCAO affected hemisphere was much lower than the CBF_C (0.73±0.07 ml/g/min) of the normal hemisphere (p=2×10⁻⁵). Figures 6C reports the OEF values of the voxels in the ischemic (OEF_I=0.58±0.11) and the control (OEF_C=0.45±0.04) hemispheres (p=3×10⁻⁴), while Fig. 6D shows the OEF ratios (R_OEF=1.3±0.3) between the OEF_I and OEF_C. The overall results clearly indicate substantial CMRO₂ and CBF reductions and an OEF increase in the MCA occluded hemisphere.

Figure 6.

Comparison results of CMRO₂, CBF, OEF and OEF ratios among the paired image voxels in the MCAO affected hemisphere (hatched bars) and contralateral, intact hemisphere (gray bars) obtained from a representative MCAO mouse brain. Voxel wise comparison of (A) CMRO₂ values, (B) CBF values and (C) OEF values between MCAO affected and normal hemispheres. (D) OEF ratios between OEF_I and OEF_C. The dashed lines represent mean values of the physiological parameters in the intact hemisphere.

Figure 7 shows T₂-weighted ¹H MRI of all animals scanned in this study where, in most cases, the stroke lesions were clearly visible. Single or multiple image slices selected from each mouse were displayed and a pair of ¹⁷O CSI voxels were identified on each slice with the red circle representing the voxel in the MCA occluded brain regions and the blue circle as the corresponding voxel in the intact brain regions on the contralateral hemisphere. The CMRO₂, CBF, OEF values of these paired-voxels as well as their OEF ratios are summarized in Table 1. The paired t-test result shows statistically significant reduction of CMRO₂ (CMRO_{2, I} = 1.66±0.29 vs. CMRO_{2, C} = 2.44±0.29 μmol/g/min) and CBF (CBF_I = 0.46±0.11 vs. CBF_C = 0.88±0.11 ml/g/min), as well as elevation of OEF (OEF_I = 0.49±0.19 vs. OEF_C = 0.35±0.07) in stroked brain regions as compare to the corresponding normal brain regions.

Figure 7.

T₂-weighted ¹H MR images of all animals scanned in this study. Single or multiple image slices selected from each mouse were displayed with one pair of ¹⁷O CSI voxels identified on each image as red and blue circles. The red circles indicate the voxel located in the typical MCA occluded brain regions; and the corresponding blue circles were in the symmetric location of the contralateral hemisphere with intact brain tissue.

Table 1.

Comparison of CMRO₂, CBF and OEF values in paired voxels as shown in Fig. 7

Voxel Pair	CMRO2,I (μmol/g/min)	CMRO2,C (μmol/g/min)	CBF I (ml/g/min)	CBF C (ml/g/min)	OEFI	OEFC	R OEF
1	1.31	2.41	0.43	0.73	0.38	0.41	0.9
2	1.30	2.45	0.53	0.96	0.31	0.32	1.0
3	1.35	2.44	0.37	0.71	0.45	0.43	1.1
4	1.68	2.48	0.58	0.76	0.36	0.41	0.9
5	1.25	2.44	0.44	0.90	0.35	0.34	1.0
6	1.78	2.45	0.65	0.84	0.34	0.36	0.9
7	1.84	2.02	0.54	1.03	0.42	0.24	1.7
8	2.11	2.75	0.35	0.80	0.75	0.43	1.8
9	1.88	2.96	0.34	0.94	0.70	0.39	1.8
10	1.91	2.67	0.26	0.94	0.91	0.36	2.6
11	1.90	2.32	0.51	1.02	0.46	0.28	1.6
12	1.60	1.86	0.49	0.90	0.41	0.26	1.6
Mean	1.66	2.44	0.46	0.88	0.49	0.35	1.4
SD	0.29	0.29	0.11	0.11	0.19	0.07	0.5

Paired t-Test: p < 1.1 × 10⁻⁵ (CMRO_{2, I} vs. CMRO_{2, C}); p < 1.0 × 10⁻⁶ (CBF _I vs . CBF _C); p < 0.03 (OEF _I vs . OEF _C).

Discussion

Significance and current state of neuroimaging in acute ischemic stroke

Since the brain consumes oxygen and glucose extensively to support the neuronal activities, the deficit of oxygen supply in the brain tissue due to disturbance in the blood circulation occurring during a stroke attack would result in detrimental consequences. Developing noninvasive neuroimaging modalities capable of imaging the abnormal perfusion and oxygen metabolism in the stroke patients after an acute ischemic attack, in particular, for identifying and differentiating the irreversibly damaged brain tissue and the poorly perfused but potentially salvageable tissue (i.e., the penumbra) would significantly improve the therapeutic intervention of stroke patients (Heiss, 2011).

The combined diffusion-weighted MRI (DWI) for defining the ischemic core and perfusion-weighted MRI (PWI) with contrast agents for identifying adjacent critically hypo-perfused tissue approach is appealing to ultimately identify the ischemic penumbra based on the diffusion-perfusion mismatch (Barber et al., 1998; Heiss, 2011). However, the diffusion-perfusion mismatch approach remains challenging; and it has been reported that the detected diffusion lesion consists of irreversible infract tissues could be reversed if perfusion is restored early in the stroke patients and the measured perfusion lesion may overestimate the penumbra (Davis and Donnan, 2009; Heiss, 2011; Kidwell et al., 2003; Wardlaw, 2010). The indirect approach for detecting stroke tissue pathophysiology resulting from impaired cerebral oxygen metabolism and brain energy deficits may have contributed to this immense challenge (Heiss, 2011). An ideal neuroimaging tool for stroke diagnosis should be able to directly image both CBF and CMRO₂, subsequently OEF that reflects the balance between the blood supply and oxygen consumption, and to potentially provide a better assessment of ischemic penumbra in stroke patients (Heiss, 2011).

All three variables (CMRO₂, CBF and OEF) have been imaged using PET (Mintun et al., 1984; Ter-Pogossian et al., 1970). However, PET cannot distinguish radioactive signals of the inhaled ¹⁵O₂ molecules from those emitted by the metabolically generated H₂¹⁵O molecules. Thus, it requires multiple experiments for independently assessing the production rate of the metabolic H₂¹⁵O inside the brain during an ¹⁵O₂ inhalation; the CBF value with a bolus injection of H₂¹⁵O; the cerebral blood volume (CBV) using an C¹⁵O gas inhalation; and the artery ¹⁵O-label content to determine the artery input function through the artery blood sampling (Mintun et al., 1984; Zhu et al., 2005). These multiple PET measurements are used to derive the CMRO₂ image based on a number of assumptions and a complex quantification model (Mintun et al., 1984). In addition, due to a very short ¹⁵O half-life time (~2 minutes), an onsite cyclotron is required for the PET-based CMRO₂ measurements. These limitations in methodology complexity, invasiveness and radiation exposure make PET impractical for routine clinical diagnosis despite its value in the assessment of ischemic penumbra (Heiss, 2011).

Advantages of ¹⁷O MR imaging technique

The concept behind the in vivo ¹⁷O MR-based approach for imaging CMRO₂ using a ¹⁷O₂ tracer is similar to the PET technique that uses ¹⁵O₂. However, the in vivo ¹⁷O imaging approach is characterized by a number of unique properties and merits.

First, the ¹⁷O is a non-radioactive, stable isotope and has a very low natural abundance, thus, serves as an ideal metabolic tracer with a small background signal. The inhaled ¹⁷O₂ is NMR invisible and only the final metabolic product of ¹⁷O-labeled water can be detected and imaged; this significantly simplifies the measurement procedure and quantification for imaging CMRO₂ in situ (see review articles (Zhu and Chen, 2011; Zhu et al., 2005) and the cited references therein).

Second, several animal studies have suggested that the second term on the right side of Eq. [1] approximately approaches zero during a brief inhalation of ¹⁷O₂ in the brain of small animals due to rapid blood circulation and air exchange in the lung (Zhang et al., 2004a; Zhu et al., 2005; Zhu et al., 2002), thus, the slope of the brain H₂¹⁷O concentration accumulation during ¹⁷O₂ inhalation (Phase II) can be used to determine the absolute CMRO₂ value according to Eq. [4].

Third, the metabolic H₂¹⁷O produced in the brain mitochondria can serve as an endogenous perfusion tracer during the post-inhalation period (Phase III) and its decay rate can provide an indirect assessment of CBF (Zhu et al., 2010). This could avoid an additional CBF measurement and the use of exogenous perfusion tracers commonly applied in PET (i.e., a bolus injection of H₂¹⁵O) or in the clinical PWI (i.e., a bolus injection of paramagnetic contrast agent).

Fourth, the ¹⁷O detection sensitivity increases substantially at higher magnetic fields according to a square power relation as a function of the field strength (Lu et al., 2012; Thelwall et al., 2003; Zhu et al., 2001). The success of high-field ¹⁷O imaging approaches for the assessment of CMRO₂ has been evident in animal studies (Zhu et al., 2009; Zhu et al., 2002; Zhu et al., 2007) as well as in some preliminary human brain studies (Atkinson and Thulborn, 2010; Hoffmann et al., 2011; Zhu et al., 2006).

The unique properties of the in vivo ¹⁷O imaging approach suggest the possibility for the noninvasive and simultaneous imaging of three important physiology variables: CBF, CMRO₂ and OEF from a single, brief inhalation of ¹⁷O₂.

Feasibility for imaging CMRO₂, CBF and OEF in the MCAO mouse brain

Mice are the most common vertebrate animal models in biomedical research and have been used in many preclinical studies of various brain diseases. However, the mouse brain size is very small compared to a human brain posing a major challenge for all in vivo neuroimaging methods owing to the conflict between the needed high imaging spatial resolution and the resulting decreased detection sensitivity or reliability. Despite the progresses of the in vivo ¹⁷O-MR methods for imaging CMRO₂ in rats (Zhu et al., 2002; Zhu et al., 2007), cats (Pekar et al., 1991; Zhu et al., 2009), and swine (Mellon et al., 2010), a successful application in the mouse brain has yet to be reported prior to the present study.

In this study, we examined MCAO mice at the high field strength of 11.7T in order to achieve higher ¹⁷O detection sensitivity. The results indicated that high-field strength affords excellent sensitivity for imaging the natural abundance H₂¹⁷O signal in the mouse brain with superior temporal resolution (~11 s per 3D CSI dataset) and reasonable spatial resolution (~40 μl voxel size) as demonstrated in Fig. 1B. In the central RF field of the ¹⁷O surface coil where the ¹⁷O detection sensitivity is optimal, the signal-to-noise ratio (SNR) of the 3D CSI voxels could reach 25:1 or better. Thus, the sensitivity gained at 11.7T is essential for reliably detecting the dynamic change of brain H₂¹⁷O signal during the ¹⁷O₂ inhalation (Phase II in Fig. 2) and the post-inhalation period (Phase III); which is crucial for differentiating the metabolic and perfusion differences in normal and diseased tissues. In addition, the excellent reproducibility of CMRO₂ imaging from two repeated measurements in the same animal (Fig. 3), and the ability to detect the slower H₂¹⁷O production and clearance in the MCA occluded hemisphere (Fig. 4) further support this notion. It is interesting to note that we were able to detect the abnormal oxygen metabolism in the MCAO affected brain region even before obvious stroke lesions show up in the T₂-weighted ¹H MRI (see Fig. 4C, Fig. 7 and Table 1). This result suggests that the ¹⁷O-based CMRO₂ imaging method may be more sensitive for assessing pathophysiological changes in an early stroke stage compared to conventional structural MRI.

This study is the first attempt to image CBF using the metabolic H₂¹⁷O generated in the mitochondria as an endogenous tracer. These H₂¹⁷O molecules are washed out of the brain cells and enter the blood stream before returning to the heart through blood circulation, thus, the washout rate is expected to reflect the blood perfusion. Although it has been found that the decay rate of the metabolic ¹⁷O-water signal is lower than that of conventional tracer measurement following a bolus injection of H₂¹⁷O directly into the brain, presumably due to the limited permeability of water to across the mitochondria membranes (Zhang et al., 2004b; Zhu et al., 2002), the decay constant (D) correlates closely with the CBF value across a wide range of physiology conditions (Zhu et al., 2010). The present study clearly demonstrates the feasibility of imaging CBF during the post-¹⁷O₂-inhalation period in the same CMRO₂ imaging measurement. The results, as illustrated in Fig. 5, show excellent reliability for detecting abnormal CBF and CMRO₂ in the MCAO affected brain regions, thus, make it possible to obtain three high-quality images of CBF, CMRO₂ and OEF with only one imaging measurement and a short inhalation of ¹⁷O₂.

In comparison with the rat brain CMRO₂ (2.2 μmol/g/min) and CBF (0.53 ml/g/min) values previously measured using the high-field ¹⁷O imaging approach (Zhu et al., 2002), the CMRO₂ and CBF values measured in the intact mouse hemisphere were found to be significantly higher (see Fig. 6, Fig. 7 and Table 1). This difference is expected considering the inverse relationship between the log of body size and the CMRO₂ for different species (Siesjo, 1978).

More importantly, both CMRO₂ and CBF in the MCAO affected brain region in one hemisphere were found to be significantly smaller than that of the corresponding intact brain region on the contralateral hemisphere (see Figs. 4–7 and Table 1), indicating the deficits in blood supply and oxygen utilization in the ischemic brain regions. Interestingly, the magnitude of CBF reduction was larger than that of CMRO₂ reduction, leading to an OEF increase in the MCAO affected hemisphere as illustrated in Fig. 5, Figs. 6C–6D, Fig. 7 and Table 1. The OEF increase associated with the acute stroke ischemia has been considered as a positive sign of the penumbra that might be treated with therapeutic intervention (Heiss, 2011).

It worth to mention that the same calculation methods (i.e., Eq. [4], Eq. [7] and Eq. [9]) were used to obtain CMRO₂, CBF and OEF values for both normal and stroked brain tissues. In the ischemic brain region, the decreases in the oxygen availability are presumably due to the reduction in the blood flow and tissue perfusion rather than the lower oxygen content in the arterial blood since the blood supply of different brain regions was from the same source, the heart. Therefore, the same C_a,O2 value was utilized in the OEF calculation for different brain tissues. The C_a,O2 value in different animals may vary depending on the animal condition, i.e., the hemoglobin concentration and the oxygen saturation of the arterial blood in each animal could be different slightly, which leads to different C_a,O2 values. However, such differences should not change the conclusion of this study because the ischemic brain tissues were compared with the corresponding normal tissue of the same animal brain. Therefore, the simultaneously measured CMRO₂, CBF and OEF images as shown in this study could provide comprehensive and reliable assessments for the outcomes of acute ischemic stroke.

Potential impact on stroke patients

Several recent studies have demonstrated the feasibility of imaging the dynamic change of the metabolic H₂¹⁷O either in the human occipital lobe during two-minute ¹⁷O₂ inhalation at 7T (Zhu et al., 2006), or in the entire human brain with a much longer ¹⁷O₂ inhalation time at field strength of 1.5T–9.4T (Atkinson and Thulborn, 2010; Fiat et al., 2004; Mellon et al., 2010; Zhu and Chen, 2011). Despite the complication in quantifying CMRO₂ in human due to a much larger body size and slower lung exchange between the ¹⁷O-labeled and non-labeled oxygen gases, it should still be possible to noninvasively imaging both CMRO₂ and CBF in the human brain after establishing a more sophisticated quantification modeling (Atkinson and Thulborn, 2010; Zhang et al., 2004a; Zhu et al., 2002).

An excellent SNR of the natural abundance H₂¹⁷O signal of up to 25:1 was achievable in a ¹⁷O-CSI voxel of the mouse brain with a 40-μl voxel size and 11 seconds temporal resolution at 11.7T. This sensitivity led to reliable 3D CMRO₂ and CBF imaging as shown in this study. Herein, we estimate the ¹⁷O CSI voxel size potentially for the 7T human brain application assuming that a 7.5 cm diameter RF coil will be used and an equivalent SNR can be achieved (Zhu et al., 2006). In this case, the ¹⁷O detection sensitivity along the line perpendicular to the coil center is inversely proportional to the coil radius, resulting in 6.3 times of SNR reduction due to the larger human RF coil. In addition, the translation of the ¹⁷O imaging from 11.7T to 7T also resulted in additional 2.8 folds of SNR loss based on the field dependence of the ¹⁷O detection sensitivity (Lu et al., 2011; Thelwall et al., 2003; Zhu et al., 2001). These led to a total of 18 times in the SNR reduction for the 7T human brain application if the same ¹⁷O CSI voxel size of 40 μl were retained. On the other hand, a voxel size of 0.7 ml (=40 μl ×18÷1000) should achieve an identical SNR as observed in the present mouse study. This spatial resolution should be adequate for imaging human brain CMRO₂, CBF and OEF in order to identify the abnormal perfusion and oxygen metabolism caused by the ischemic stroke since the human brain size is more than 3000 times larger than that of a mouse brain.

Conclusion

In summary, this study demonstrates that, with a brief ¹⁷O₂ inhalation, the high-field ¹⁷O MR imaging approach is capable of simultaneous and completely noninvasive imaging of three important physiology variables: CMRO₂, CBF and OEF in a small mouse brain. In contrast to the established PET technique, the ¹⁷O imaging approach is advantageous for its noninvasiveness, robustness, simplicity, and rapid measurement; thus, it should be more suitable for survival and longitudinal animal studies. The overall results clearly indicate the high sensitivity of this method for mapping abnormal CMRO₂, CBF and OEF changes in the ischemic brain regions and for discerning that from the intact brain tissue located in the contralateral hemisphere.

The findings from this study also suggest the possibility of extending the ¹⁷O-MR based CMRO₂, CBF and OEF imaging approach to the healthy human subjects and stroke patients, potentially as a clinical diagnosis tool. Finally, this new neuroimaging modality should be valuable for investigating normal brain functions and brain diseases beyond stroke (e.g., brain tumor and many degenerative diseases), or for studying the oxygen metabolism and perfusion in other organs such as heart.

Highlights.

• The stroke brain of MCAO mouse was studied using a new in vivo ¹⁷O MR imaging method.

• CMRO₂, CBF and OEF were simultaneously and noninvasively imaged in mouse.

• CMRO₂ and CBF were significantly lower in the MCAO affected brain region

• In contrast, OEF was significantly higher in the MCAO affected brain region.

• ¹⁷O-MRI method is promising for studying abnormal oxygen metabolism and perfusion.