Under normal physiological conditions, the renal medulla is operating on the verge of hypoxia and therefore at risk of hypoxic injury. High metabolic activity and shunting between the arterial and venous system render medullary PO2 at about half that of the cortical region (1). Blood oxygenation level dependent (BOLD) MRI is the only non-invasive method available to evaluate relative regional renal oxygenation status (2). Currently, the transverse relaxation rate, R2*, is the primary measure of changes in regional oxygenation within the kidneys. Here we provide preliminary validation of a quantitative model (3) based on R2’ (R2*-R2) to derive renal blood PO2 in rat kidneys.

Animals Six male Sprague-Dawley rats (405 ± 20 g) were used for this study (Charles River, Chicago, IL). A catheter (PE-50 tubing) was placed in the femoral vein for administration of drugs. All procedures were conducted under anesthesia using Inactin (thiobutabarbital sodium, 100 mg/kg i.p., Sigma-Aldrich, St. Louis, MO). L-NAME and Furosemide were administered intravenously at a concentration of 10 mg/kg body weight for each (Sigma-Aldrich, St. Louis, MO) to modulate renal perfusion and/or oxygenation.

MRI Acquisition All experiments were performed on a 3T whole body scanner (Magnetom Verio, Siemens Healthcare, Erlangen, Germany). MRI measurements were made using an mGRE (12 Echoes: 3.56 - 41.29 ms increments of 3.43 ms, TR: 69.0 ms, Matrix: 256x128, PhaseFOV: 50%, resolution: 0.47mm x 0.57 mm, slice thickness: 2.0 mm) and spin echo sequence (8 echoes: 9.9 – 79.2 ms increments of 10ms, TR: 500ms), repeated three times, followed by an arterial spin labeling (ASL) acquisition (FAIR-SSFP, TI: 1200ms, slice thickness: 2.5mm, matrix: 128x128) (4,5). The timeline for drug administration and MRI sequencing is shown in Figure 1.

MRI Analysis Regions of interest were manually defined in the cortex, inner stripe of the outer medulla (ISOM) and inner medulla using a custom image-processing library written in Python. R2* was fit to the 12 mGRE echoes and regions near large susceptibility artifacts are removed automatically (6). R2 is fit from the 8 spin echoes. R2’ is calculated for each region by averaging the R2* and R2 values, then subtracting them.

Statistical Methods Comparisons were made between each successive time stage to assess if a significant change took place. Cohen's d was used to assess the magnitude of the change and Wilcoxon's signed-rank test was used to assess the statistical significance.

SHb and Blood PO2 A statistical model was used for estimating the oxygen saturation of hemoglobin (SHb) as well as the blood PO2 of each region within the kidney (3). The following parameters were assumed (7–9): cortex (vascular fraction: 0.27, hematocrit: 0.4, extravascular diffusion coefficient: 1.45, intravascular diffusion coefficient: 1, average vesicle radius: 10 (um)); medulla (vascular fraction: 0.18, hematocrit: 0.25, extravascular diffusion coefficient: 1.04, intravascular diffusion coefficient: 1, average vesicle radius: 10 (um)).

Consistent with prior reports based on R2* (10,11), the estimated SHb and blood PO2 showed regional responses to vasoconstrictor (L-NAME) and reductions in oxygen consumption (furosemide). Large responses were observed in the inner stripe of outer medulla where SHb and blood PO2 were reduced with L-NAME and subsequently increased following furosemide administration (Figures 2, 3). Additionally, renal blood flow, as estimated by ASL MRI, showed responses to the pharmacological maneuvers (5,12), with cortex showing the most changes (Figure 4).

Our data supports the feasibility of applying this quantitative model to estimate renal blood PO2 based on BOLD MRI data acquisitions. At baseline, blood PO2 in the inner stripe of the outer medulla was found to be 30.5 ± 1.2 mmHg and 51.9 ± 5.2 mmHg in the cortex, which is inline with previously measured values using microelectrodes (7). However, microelectrode studies are unable to differentiate between capillary blood PO2 and tissue PO2.

There are limitations to this approach. The method for converting from MRI measured parameters (R2, R2*) to physiological ones (SHb, PO2) requires a number of parameters that are regionally inhomogeneous and may not be readily available. This is especially true for changes in these parameters (e.g. regional hematocrit) in response to the pharmacological maneuvers. Similarly changes in these parameters in certain diseased states could make comparisons to healthy subjects challenging.

Our preliminary data have shown the estimation of SHb and blood PO2 measurements based on the recently proposed statistical model to be consistent with previous microelectrode studies and sensitive to changes in renal oxygenation via pharmacological maneuvers. Future studies are necessary to directly compare these MRI derived estimates of blood PO2 against invasive measurements such as phosphorimetric oxygen probes (13).