Acute kidney injuries (AKI) of various origins share one common feature in the initiating chain of events: imbalance between local tissue oxygen delivery and oxygen demand.[1,2] Quantitative parametric MRI (T₂* mapping) offers a non-invasive approach to probe renal oxygenation but provides a surrogate rather than a quantitative measure of oxygen saturation. Changes in tissue pO₂ and T₂* may be closely related, but their link is influenced by various effects, including changes in vascular volume fraction. Previously we reported T₂* alterations of renal arterio-venous occlusion were more pronounced than those induced by hypoxia, while arterial occlusion induced a smaller T₂* effect than hypoxia.[3,4] This observation might be explained by variations in renal blood volume. The suitability of the intravascular contrast agent (CA) ferumoxytol (ultra small paramagnetic iron oxide, USPIO) for renal blood volume estimation has been demonstrated in mice [5]. Recently quantitative blood oxygenation level-dependent (qBOLD) approaches were developed that intend to map local blood oxygen saturation (SO₂), based on MR measurements of T₂*, T₂, BVf and macroscopic magnetic field distortions (B₀ map) [6-9]. We hypothesized that ferumoxytol-based monitoring of renal BVf in combination with qBOLD model analysis is essential for the unambiguous interpretation of renal T₂* and ultimately the estimation of SO₂. To test the feasibility and establish the importance of this new approach we monitored renal T₂*/T₂ during baseline and short periods of venous occlusion (VO). In the same animal this was performed under naïve conditions and then again with USPIO. We subsequently estimated BVf and SO₂ based on the approach of Christen et al [8].

Animal model: 6 male Wistar rats were anesthetized (urethane) and kept at 37°C core body temperature during surgery and MRI.[4] For VO a remotely controllable hydraulic occluder was placed around the renal vein. A short-term reversible ischemia was induced by closing the hydraulic occluder for 3 minutes, followed by a reperfusion phase of ~20 minutes. VO was confirmed by time-of-flight MR angiography of the kidney. Subsequently ferumoxytol was injected i.v. at a dose of 4 mg Fe/kg and after a mixing time of 2 minutes the short-term reversible VO was repeated.

MR imaging: in vivo MRI was performed using a 9.4 T animal scanner (Bruker, Germany) in conjunction with birdcage RF resonator and a 4-channel receive RF coil array (Bruker, Germany) customized for rats. Local B0 shimming on a voxel tailored to the kidney was performed first. Parametric T₂* mapping used respiratory gated multi gradient echo (MGE) imaging (TR = 50 ms, echoes = 10, first TE = 1.43 ms, echo spacing = 2.14 ms, averages = 4) [4]. A coronal oblique slice across the kidney was acquired with a spatial in plane resolution of (226x445) µm² and a slice thickness of 1.4 mm.

Quantitative renal MR blood oximetry: We estimated BVf from pre- and post-USPIO MR measurements of T₂*. For this, we selected data from matching timepoints in the baseline-occlusion-recovery cycle under naïve conditions and after USPIO administration. BVf and SO₂ were then calculated based on the approach of Christen et al [8]. For this feasibility study no B₀ mapping was performed, as the effect of macroscopic magnetic field distortions was considered to be constant throughout the experiment. Calculation of absolute values for SO₂ would require this information; here we report SO₂ in a.u. with the aim to demonstrate that a temporal change can be detected.

USPIO administration decreased cortical and medullary intensity in renal T₂*-weighted images (Fig.1). The effect of venous occlusion was substantial without as well as with USPIO. Figure 2 shows maps of renal T₂*, BVf and SO₂ at baseline, during venous occlusion (VO) and the recovery phase. The reduction in renal T₂* during VO could be unraveled into a significant SO₂ decrease in cortex and outer medulla combined with a substantial blood volume increase.Our results demonstrate that combined monitoring of T₂*, T₂ and blood volume fraction is feasible and permits estimation of oxygen saturation of Hb (SO₂). This is an important step towards quantitative renal MR blood oximetry, because - as hypothesized and now proven by our data - blood volume fraction may change considerably and confound the interpretation of renal T₂* changes as a surrogate of changes in blood oxygenation. BVf may vary e.g. due to changes in renal perfusion pressure, vasoconstriction/-dilation or tubular distension. Unambiguous characterization of renal oxygenation by T₂* hence requires multi-parametric MR including BVf. Combining T₂*-mapping with USPIO, paralleled by calibration via invasive but quantitative physiological measurements using MR-PHYSIOL [10] might help to gain a better insight into renal oxygenation and hemodynamics.