In Vitro and In Vivo Studies of ¹⁷O NMR Sensitivity at 9.4 and 16.4 Tesla

Abstract

In vivo ¹⁷O magnetic resonance spectroscopy (MRS) has been successfully applied for imaging the cerebral metabolic rate of oxygen consumption (CMRO₂) through the detection of metabolically produced H₂¹⁷O from the inhaled ¹⁷O-labeled oxygen in animals at high field. In this study, we compared the ¹⁷O sensitivity for detecting natural abundance H₂¹⁷O signals from a phantom solution and rat brains at 9.4 and 16.4 Tesla. The ¹⁷O signal-to-noise ratio (SNR) measured at 16.4T was 2.9- and 2.7~2.8-fold higher than that at 9.4T for the phantom and rat brain studies, respectively. Similarly, three-dimensional ¹⁷O MRS imaging data showed a more than 2.7-fold higher SNR in the central rat brain region at 16.4T than that at 9.4T. The substantial ¹⁷O SNR gain at ultrahigh field significantly improved the reliability for imaging CMRO₂, and will provide an opportunity for in vivo assessment of altered oxidative metabolism associated with brain functions and neurological diseases.

INTRODUCTION

The rate of oxygen utilization in the mitochondrial respiratory chain serves as an important index of the cellular metabolic status. However, the lack of non-destructive and robust in vivo imaging approaches for the investigation of cellular oxidative metabolism has limited our understanding of the mechanism underlying metabolic alteration between normal and diseased tissues. Although ¹⁵O positron emission tomography (PET) has been the standard imaging modality for the measurements of oxygen consumption rate for more than 20 years, its application has been limited to research because of the complex experimental procedures, data interpretation and compartmental modeling requirement (1,2). In contrast, ¹⁷O magnetic resonance spectroscopy/imaging (MRS/I) provides the specificity by direct detection of the dynamics of metabolic H₂¹⁷O production from the reduction of ¹⁷O-labeled oxygen molecules (¹⁷O₂) but without confounding signal from ¹⁷O₂, thus, allows more straightforward measurement and quantification of the cellular oxygen consumption rate in vivo (3).

Recently, in vivo ¹⁷O MRS approaches for the measurement of the cerebral metabolic rate of oxygen consumption (CMRO₂) (4–11) and the myocardial oxygen consumption rate (MVO₂) (12–14) have been developed in animals and human subjects. In comparison with the indirect detection of H₂¹⁷O via ¹⁷O-induced changes in the proton T_1ρ or T₂, the direct ¹⁷O MRS/I has the advantages of being highly specific to the tissue H₂¹⁷O and the availability of using natural abundance H₂¹⁷O signal as an internal reference for quantification of tissue H₂¹⁷O concentration. In addition, the direct ¹⁷O MRS/I detection approach may avoid the challenges encountered by the indirect method, i.e., the complicated quantification of H₂¹⁷O concentration due to the sensitivity of the proton relaxation time to biochemical and physiological changes in tissues, such as pH (15). However, the efforts of direct monitoring and imaging the accumulation rate of H₂¹⁷O during an inhalation of ¹⁷O₂ have been suffered from the low sensitivity of ¹⁷O detectionbecause of its low gyromagnetic ratio and low H₂¹⁷O content in nature.

It is known that the optimal signal-to-noise ratio (SNR) for the detection of magnetic resonance signal depends on magnetic field strength (B₀), longitudinal relaxation time (T₁), T₂* and the RF coil quality factor (Q) according to the following relationship:

$$\text{SNR}\propto {\text{B}}_0^{\beta }\sqrt{\frac{{\text{QT}}_2^{\ast }}{{\text{T}}_1}}$$ [1]

where, the constant β was suggested to be close to the value of 7/4~2 for ¹⁷O spins (16,17). Obviously, increasing B₀ is one efficient way to improve the SNR. Two previous studies have compared the in vitro and in vivo ¹⁷O sensitivity between 4.7T and 9.4T, and between 4.7T and 11T, respectively (16,17), suggesting great advantages of high field. The unique NMR properties of ¹⁷O, i.e., the independence of ¹⁷O relaxation times (T₁ and T₂) on the field strength (16,17) and extremely short T₁ for more signal averaging, also contribute to a large gain of ¹⁷O detection sensitivity at high/ultrahigh field (3).

In this study, ¹⁷O sensitivity was measured and compared using the natural abundance H₂¹⁷O signals in a phantom solution and rat brains at 9.4T and 16.4T, without and with spatial localization, to quantify the possible SNR gain for in vivo ¹⁷O MRS/I application using the newly developed large-bore animal scanner with ultrahigh field strength of 16.4T.

MATERIALS and METHODS

Phantom and Animal Preparation

A 6-mm-diameter glass sphere filled with natural abundance ¹⁷O distilled water was prepared for all phantom studies at 9.4T and 16.4T. For in vivo studies, following the initial induction with 5% isoflurane and oral intubation, male Sprague Dawley rats (250~300 g body weight) were anesthetized with 2% (v/v) isoflurane in a mixture of O₂ and N₂O gases (~2:3) during surgery and experiment. To monitor physiological parameters and obtain blood samples, the femoral artery and vein were catheterized during the surgery. Rectal temperature was maintained at 37±1°C using heated circulating water. Blood gases were sampled for monitoring physiological conditions. The arterial blood pressure and heart rate were monitored during the experiment. The animals were maintained under physiological conditions during all the measurements.

Ten rats (five for each field strength) were used in this study to measure the global and localized SNR at both fields from the natural abundance brain H₂¹⁷O signals. One ¹⁷O₂ inhalation experiment was performed on another rat under alpha-chloralose anesthesia (20) at 16.4T and using ¹⁷O-labeled O₂ (74.8% enrichment, ISOTEC) with N₂O gases (~2:3) to imaging and quantify CMRO₂ from the metabolically generated H₂¹⁷O signal in the brain. During the sequential acquisitions of localized ¹⁷O MRS imaging, the respiration gas was switched from the normal gas mixture with ¹⁶O₂ to the ¹⁷O-labeled oxygen gas mixture. After 2.75 minutes of ¹⁷O₂ inhalation, the gas line was switched back to the normal gas mixture for ~15 minutes until the completion of data acquisition.

All animal surgical procedures and experimental protocols were approved by the Institutional Animal Care and Use Committee of the University of Minnesota.

In vitro and In vivo ¹⁷O MRS Imaging

All measurements were performed on either a 9.4T/31cm or a 16.4T/26cm horizontal bore magnet (Magnex Scientific) interfaced to VNMRJ consoles (Varian, CA). For in vitro studies, one ¹⁷O radiofrequency (RF) probe consisting of an oval shaped (18mm×13mm) surface coil was designed and constructed. Its resonance frequency can be tuned to either 54.25 MHz for 9.4T or 94.65 MHz for 16.4T application. The glass sphere phantom was fixed on the RF probe for all phantom studies at both fields. For in vivo studies, two passively decoupled dual-coil RF probes with identical geometry were constructed and placed over the rat brain to obtain optimum in vivo sensitivity for both 9.4T and 16.4T experiments. Each RF probe included an oval shaped single-loop ¹⁷O surface-coil (10mm×20mm) for acquiring in vivo ¹⁷O MRS/MRSI and a butterfly shaped ¹H surface-coil for shimming and acquiring anatomical images.

A single-pulse-acquire sequence was applied to obtain the optimal ¹⁷O SNR with a nominal 90° RF excitation pulse and following acquisition parameters: 20 kHz spectral width (SW), 1024 number of points (NP) for each FID, 50 ms repetition time (TR) with the number of signal averages (NT)=128 for phantom measurements and NT=512, 128 or 16 for in vivo studies. The spatial localization of ¹⁷O MRS was achieved by using the three-dimensional (3D) Fourier series window MRS imaging technique (11,18–20) with 15 ms TR, 0.41 ms echo time (TE), 512 NP, 20 kHz SW, 928 total NT, 9×9×5 3D phase encodes and 30×30×30 mm³ field of view (FOV).

For the ¹⁷O₂ inhalation experiment at 16.4T, an identical set of 3D ¹⁷O MRS imaging acquisition parameters as previously reported for rat brain applications at 9.4T (20) was used in this study. The FIDs were processed by exponential filtering with a line broadening of 100 Hz to enhance the SNR, followed by Fourier transformation.

The ¹⁷O sensitivity was evaluated using the SNR of H₂¹⁷O resonance peak calculated by dividing peak intensity by the standard deviation of the noise. In addition to the SNR gain at higher field, the relative fluctuation of ¹⁷O signals among repeated measurements at a given magnetic field is also an important factor for the evaluation of dynamic studies using direct ¹⁷O MRS/I approaches. Therefore, the relative fluctuation of ¹⁷O signals was also assessed by the variance from both the phantom and rat brain measurements by computing the standard deviation of the global natural abundance H₂¹⁷O signals from 10 repetitions of data acquisition at 9.4T and 16.4T. All results presented in this study were shown as mean ± standard deviation.

RESULTS

Table 1 summarizes the SNR and measurement variance results of global ¹⁷O signals measured in the rat brain and phantom at 9.4T and 16.4T. The average ¹⁷O SNR at 16.4T was 2.9- and 2.7~2.8-fold higher than that at 9.4T for the phantom and rat brain studies, respectively. In contrast, the ¹⁷O signal variance was 2–4-fold smaller at 16.4T than 9.4T. The overall results clearly demonstrate excellent sensitivity and small variance (i.e. high signal stability) at both fields for obtaining the ¹⁷O signals of natural abundance H₂¹⁷O from either the phantom or rat brains. Nevertheless, the 16.4T scanner offers striking improvements of several folds in both sensitivity and stability.

Table 1.

¹⁷O SNR and signal stability comparisons between 9.4T and 16.4T (n=5)

	Sphere Phantom (NT=128)	Rat Brain (NT=16)	Rat Brain (NT=128)	Rat Brain (NT=512)
SNR9.4T	36.8±3.5	35.6±2.1	97.5±10.4	197.1±21.8
SNR16.4T	108.4±11.3	96.4±8.8	269.2±28.9	531.3±51.9
SNR16.4T/SNR9.4T	2.9	2.7	2.8	2.7
Variance at 9.4T	2.19%	2.16%	0.90%	0.47%
Variance at 16.4T	0.53%	1.01%	0.33%	0.14%

The 1D ¹⁷O SNR profiles along an axis parallel to the RF coil plane obtained from the central slides of 3D ¹⁷O MRS images in rat brains at both 9.4T and 16.4T were demonstrated in Fig. 1a. As shown in Fig. 1b, significantly improved ¹⁷O sensitivity, i.e., an averaged more than 2.5-fold SNR gain was observed from the natural abundance H₂¹⁷O in the rat brain at 16.4T compared to that at 9.4T. Similarly, the 2D ¹⁷O MRS images of natural abundance H₂¹⁷O in the rat brain taken from the central slides of 3D ¹⁷O MRS imaging data also indicated a more than 2.7-fold higher SNR in the central brain voxels at 16.4T than that at 9.4T (Fig. 2).

Figure 1.

One-dimensional ¹⁷O SNR profiles along an axis parallel to the RF coil plane obtained from the central slides of 3D ¹⁷O MRS imaging in rat brains at 9.4T and 16.4T, respectively (a), and their SNR comparison (b). The vertical dashed lines present the anatomical edges of the rat brain.

Figure 2.

Representative anatomical image and the corresponding 2D ¹⁷O spectral image of natural abundance H₂¹⁷O of the rat brain obtained at 9.4T and 16.4T, respectively (a). The spectra of central brain voxel acquired at 9.4T and 16.4T were shown in (b) for comparison.

The large sensitivity improvement at ultrahigh field indicated the potential of ¹⁷O MRS imaging for the quantification of regional CMRO₂ from detecting the dynamic changes of the metabolic H₂¹⁷O in small animal brains during a short inhalation of ¹⁷O₂. Figure 3a shows the stacked plots of H₂¹⁷O spectra from one representative voxel with the voxel size of ~0.1 ml, located near the center of the ¹⁷O RF coil-sensitive region inside the rat brain before, during and after a 2.75-minute inhalation of ¹⁷O₂. The time course of ¹⁷O signal intensity of brain H₂¹⁷O in the representative rat shows three distinct phases: constant baseline (from natural abundance H₂¹⁷O); approximately linear increasing during ¹⁷O₂ inhalation; and decreasing exponentially after the cessation of inhalation until reaching a new steady state. Figure 3a also illustrates the excellent sensitivity and small variance of the brain ¹⁷O signal measurements, which were important for detecting the dynamic change of the metabolic H₂¹⁷O content in the rat brain during and after ¹⁷O₂ inhalation. The linear fitting of the metabolic H₂¹⁷O concentration change in the same voxel during ¹⁷O₂ inhalation was shown in Fig. 3b. Based on the simplified quantification model (20,21), the CMRO₂ value in this particular voxel was calculated with linear regression method (=2.1 μmol/g/min), which was consistent with those reported in the previous studies under the same physiological condition (11,20).

Figure 3.

Stacked plots of brain H₂¹⁷O spectra from a representative voxel (~0.1 ml voxel size and located near the center of ¹⁷O coil-sensitive region inside the rat brain) of 3D ¹⁷O MRS imaging data acquired before, during and after 2.75 minutes ¹⁷O₂ inhalation in a rat brain at 16.4T (a), and the linear fitting of the metabolic H₂¹⁷O concentration in this voxel during the ¹⁷O₂ inhalation for determining the CMRO₂ value (b).

As a further proof of the independent ¹⁷O relaxation times on the field strength, we measured T₁ value of natural abundance H₂¹⁷O in the rat brain at fully relaxed condition using the inversion recovery pulse sequence at 16.4T. The in vivo T₂* value of natural abundance H₂¹⁷O was also evaluated from the line width, i.e., full width at half maximum (Δν_1/2), of the H₂¹⁷O resonance peak without line broadening. As a result, the measured T₁ (5.3±0.4 ms), Δν_1/2 (174±5 Hz) and calculated T₂* (1.83±0.05 ms) values for rat brains at 16.4T were consistent with those previously reported at 11T, 9.4T and 4.7T (16,17).

DISCUSSION and CONCLUSION

In the present study, we examined the in vitro and in vivo ¹⁷O sensitivity at two high field strengths. The results demonstrate an approximated 2.5~2.9-fold SNR gain at 16.4T compared to 9.4T for the phantom and rat brain studies, which indicated an approximated square power (β=2) dependence of ¹⁷O SNR on B₀. This finding of quadratic SNR gain over the field strength was consistent with the previous studies (16,17,22) and close to the theoretical prediction of 7/4 (23). Also shown in Table 1, a smaller measurement variance (2~4-fold reduction) was obtained at 16.4T as compared with that of 9.4T indicated an improved signal stability at 16.4T. All of these improvements suggest the significant advantage provided by high/ultrahigh magnetic fields, and more reliable measurements of CMRO₂ are expected from the direct detection of small dynamic changes of metabolically generated H₂¹⁷O signal during a brief ¹⁷O₂ inhalation at high/ultrahigh magnetic fields.

Compare to the 9.4T, a larger error in inter-subject measurements at 16.4T were observed (see Figure 1a and Table 1). Two possible explanations for this observation: a larger SNR variation among different subjects measured at 16.4T; and the non-uniform sensitivity profile for detecting the ¹⁷O NMR signal due to inhomogeneous RF (B₁) field of the surface coil applied in this study. Therefore, the measured ¹⁷O signal and SNR were sensitive to a number of factors such as the relative position between the sample and RF coil, sample loading factor, RF coil tuning and matching performance etc. Nevertheless, both SNR and the variance between repeated measurements were much improved at 16.4T than that at 9.4T, which are most crucial for in vivo ¹⁷O MR application.

Interestingly, the results summarized in Table 1 also indicate that the variance was inversely proportional to $\sqrt{\text{NT}}$ at both 9.4T and 16.4T. This observation was consistent with a positive linear relationship between SNR and $\sqrt{\text{NT}}$ (see Table 1). They collectively suggest an inverse relationship between variance and SNR, and the white noise is the dominant noise source of the in vivo ¹⁷O MRS signal measured in the rat brain.

The measured ¹⁷O T₁ and T₂* values for the natural abundance H₂¹⁷O in the rat brain were similar between 16.4T and 9.4T, which provide further evidence of the independence of ¹⁷O relaxation times on the field strength. Moreover, the short ¹⁷O T₁ value makes it possible for much faster data acquisition and more signal averages, and consequently, contributes to the improvement of SNR per unit of sampling time, which is essential for the success of CMRO₂ measurement and quantification. Owing to the unique field independent ¹⁷O relaxivity, Eq. [1] can be further simplified to Eq. [2],

$$\text{SNR}\approx \text{k}·{\text{B}}_0^{\beta }$$ [2]

where k is a constant. Figure 4 displays the relationship between the relative ¹⁷O SNR, which was the SNR of the rat brain normalized to the SNR value measured at 9.4T, and the field strength (B₀) across a wide range up to 16.4T. The relative SNR value of 2.75 at 16.4T was obtained in the present study (averaged from the 2.7 and 2.8 values listed in Table 1). The relative SNR value of 0.238 at 4.7T was obtained from a previous rat brain study (17). The zero SNR value at zero field strength was based on the NMR principle/theory. The solid line in Fig. 4 is the fitting curve according to the function expressed by Eq. [2]. The regression results k=0.015 and β =1.9 for a wide field strength range from zero to 16.4T. The high β power value suggests that compared to in vivo ¹H, ³¹P and ¹³C MRS, the in vivo ¹⁷O MRS can benefit the most from the sensitivity gain at high/ultrahigh field.

Figure 4.

Field dependence of the relative ¹⁷O SNR measured in the rat brain and normalized to the SNR value measured at 9.4. The solid line shows the fitting curve according to Eq. [2], covering the field strength from zero to 16.4T.

In conclusion, the significant SNR improvements achievable at 16.4T could benefit the 3D CMRO₂ imaging based on the ¹⁷O MRS/I methods, which should provide an opportunity for detecting altered oxidative metabolism associated with brain function and neurological diseases with improved spatial and temporal resolutions.