QUANTITATIVE 1H MR TISSUE OXIMETRY (QMRO)
Scott C Beeman1, Joseph JH Ackerman1, and Joel R Garbow1

1Washington University in St. Louis, St. Louis, MO, United States

Synopsis

A direct and non-invasive measure of tissue O2 would be a major advance. O2 is paramagnetic and can thus, in principle, be quantified with NMR/MRI. However, such measurements are challenged/masked by two competing effects: (i) magnetization transfer between 1H spins of tissue water and the solid-like macromolecular matrix (e.g., proteins, cell membranes) and (ii) blood flow, which can bring equilibrium-polarized 1H spins into the interrogated tissue volume. We describe a strategy for mitigating these confounds and quantify the direct relationship between pO2 and the MR-measured longitudinal relaxation rate constant, R1.

INTRODUCTION

Tissue oxygen content is a critical determinant of metabolic functional status. Development of a non-invasive method to quantitatively map (image) tissue oxygen content (pO2) would be a major advance, enabling assessment of physiologic competence in response to stimulus, disease, and therapeutic intervention. It remains under appreciated that water 1H longitudinal relaxation in mammalian tissue is well modeled as bi-exponential and it is the experimentally accessible exponential rate constant R1,slow, characterizing the long-lived relaxation component, that is linearly related to tissue pO2: R1,slow = R1,0,slow + r1,slow•pO2. Following appropriate calibration to determine R1,0,slow and r1,slow (O2 relaxivity), an MRI-determined R1,slow map can, in principle, be converted to a map of tissue pO2. However, MR-based quantification of tissue pO2 is challenged by two competing apparent longitudinal relaxation mechanisms: (i) magnetization transfer (MT, characterized by R1,fast) between the 1H spins of tissue water and of the solid-like macromolecular matrix (e.g., proteins, cell membranes) and (ii) blood flow, which can bring equilibrium-polarized 1H spins into the interrogated tissue volume. Herein, we demonstrate that by mitigating the relaxation effects of (i) MT (via discrimination of R1,slow from R1,fast) and (ii) blood flow (via non-slice-selective inversion pulses and gradient-enabled IVIM-based vascular spin dephasing), the direct relationship between pO2 and tissue water longitudinal relaxation can be quantified and tissue pO2 determined/mapped.

METHODS

Phantoms. Glutaraldehyde-cross-linked 15% (wt) x-BSA/PBS (bovine serium albumin/phosphate-buffered saline) tissue mimics/phantoms were prepared at various pO2 by bubbling varying mixtures of O2 and N2. Sample pO2 was measured by an Integra® Licox® Clark-type (platinum) electrode. •In vivo. Normal female BALB/c mice breathed alternate mixtures of 100% O2, 12.5% O2/87.5% N2, and 95% O2/5% CO2 (mixed to modulate tissue pO2). An OxyLite® optical microprobe (“gold standard”) was implanted into either the right thalamus or thigh muscle to determine tissue-pO2 in vivo concurrent with MR monitoring. •MRI. Modified Fast Inversion Recovery (MFIR) Point RESolved Spectroscopy (PRESS) data were acquired in x-BSA phantoms and mice: x-BSA - 4.7 tesla; 64 TI from 0.0075 to 6 seconds; non-slice-selective inversion pulse; voxel dimensions = 4x4x4 mm3, two=averages, scan time = 9 min 4 sec; muscle/thalamus in vivo - 1x1x1 mm3 one average, scan time = 4 min 33 sec. •R1 Determination. Relaxation rate constants were estimated using a bi-exponential relaxation model yielding: (i) a “fast” relaxation rate constant (R1,fast, i.e., MT-dominated relaxation of 1H spins proximal to the macromolecular matrix, ~20 sec-1) and (ii) a “slow” relaxation rate constant (R1,slow, i.e., 1H spins remote from the macromolecular matrix, ~0.6 sec-1). Parameter estimates were obtained using Bayesian-probability-theory-based methods: http://bayesiananalysis.wustl.edu/index.html.

RESULTS

•x-BSA phantoms. In the x-BSA phantoms, only R1,slow is sensitive to dissolved O2 (Figure 1). •In vivo. Employing a blood-flow-mitigating PRESS sequence (non-slice-selective inversion pulse and gradient-enabled IVIM-suppression of vascular 1H spins) and bi-exponential data modeling, brain and muscle r1,slow (pO2 tissue relaxivity in vivo) was determined as 0.56 x 10-3 ± 0.09 x 10-3 and 0.69 x 10-3 ± 0.15 x 10-3, respectively (summarized in Table inlaid in Figure 2).

DISCUSSION

We report a milestone in the development of a translatable MRI/MRS-based method for quantitatively mapping tissue pO2. The direct relationship between pO2 and R1 (r1,slow) can be, and has been herein, resolved/quantified by: (i) isolating R1,slow from MT-dominated R1,fast and (ii) using non-slice-selective inversion pulses and gradient-enabled IVIM-suppression of vascular 1H spins) to mitigate blood flow - Quantitative MR Oximetry (QMRO).

Acknowledgements

No acknowledgement found.

References

No reference found.

Figures

Figure 1. Plots of pO2 vs. (A) R1,slow and (B) R1,fast calculated from data collected in 15% x-BSA. R1,slow is linearly correlated to pO2 (R2 = 0.89). O2 r1,slow = 1.0 x 10-3 ± 0.1 x 10-3 mmHg-1 sec-1. R1,fast was not sensitive to pO2. (R2 = 0.22).

Figure 2. Plots of R1,slow vs. pO2, for in vivo mouse brain (A, R2 = 0.9) and muscle (B, R2 = 0.8). R1,0,slow and r1,slow are shown (in the form of the mean ± the standard deviation of the Bayesian estimated parameter PDFs).



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1568