Introduction

Deuterium Metabolic Imaging (DMI) is a novel approach providing high spatial resolution 3D metabolic data by combining ²H MRSI with administration of ²H-labeled substrates (1). DMI of brain glucose metabolism has so far been focused on human and rat brain. However, preclinical neurological research relies for a great part on mouse models of disease, which motivates implementing DMI for brain studies in mice. When DMI is used to map steady state ²H-labeling in downstream metabolites of glucose, different options for glucose administration are available. We explored both intraperitoneal (IP) infusion, and intravenous (IV) infusion via a tail vein catheter. IV infusion is the most efficient method to quickly attain high levels of labeled glucose in the blood. However, an IV infusion line is not always easy to establish in mice, and particularly challenging when repeated DMI scans are required. We explored how the SNR of ²H-labeled glutamate+glutamine (Glx), which is labeled via oxidative metabolism of ²H-glucose, is affected by IV and IP infusions. [6,6’-²H₂]-glucose is most commonly used for DMI studies, and its metabolism leads to one unlabeled and one ²H-labeled pyruvate/lactate molecule per glucose molecule. Having a deuteron on the 1st carbon position of glucose results in two labeled trioses and would increase the SNR for detection of ²H-labeled lactate and Glx. We therefore explored the use of [1,2,3,4,5,6,6’-²H₇]-glucose, and compared the difference in SNR for ²H-labeled Glx with [6,6’-²H₂]-glucose infusions. Lastly, DMI was used to map ²H-lactate production in a mouse model of glioblastoma to illustrate the capability of DMI to visualize localized differences in ²H-glucose metabolism in mouse brain, in vivo.

Materials and Methods

Healthy and glioma-bearing B6(Cg)-Tyrc-2J/J mice, (n=8) were studied. Glioma-bearing mice were generated by intracerebral injection of GL261 cells (100,000 cells), as described previously for rat glioma models (2). Glucose administration was performed via an infusion line placed in the lateral tail vein (IV), or in the lower right quadrant of the abdomen (IP). Solutions (1 M, in sterile water) of [6,6’-²H₂]-glucose (“D2”) and [1,2,3,4,5,6,6’-²H₇]-glucose (“D7”) (Cambridge Isotope Laboratories, MA, USA) were administered using a syringe pump (Harvard Apparatus, MA, USA), following a bolus-continuous infusion protocol, for both IV and IP infusions. DMI acquisition started at 90 min and 120 min following the start of the IV and IP infusions, respectively. All animal studies were performed on an 11.74 T Magnex magnet (Magnex Scientific Ltd.) interfaced to a Bruker Avance III HD spectrometer running on ParaVision 6 (Bruker Instruments), as described previously (1). A double-turn ²H radiofrequency (RF) coil was used, shaped to fit the head of mice, and sized (14x12 mm²) to cover the whole mouse brain, combined with a single turn ¹H surface coil for shimming and MRI as illustrated in Fig. 1. DMI data were acquired at nominal spatial resolutions of 5 uL and 8 uL using a 11x11x11 matrix and FOVs of 18.8x18.8x18.8 mm³, and 22x22x22 mm³, respectively, with spherical k-space encoding. Using a TR of 400 ms and 8 averages, acquisition of 3D DMI dataset required ~36 min. All ¹H MRI and ²H DMI data were processed in DMIWizard, a home-written graphical user interface in Matlab. Preliminary SNR analysis based on peak height of Glx in voxels located within the mouse brain was performed to reveal differences between ²H-labeled glucose types and infusion protocols.

Results

Figure 2 shows DMI-based metabolic maps of ²H-labeled Glx overlaid on coronal anatomical MR images, and examples of individual ²H MR spectra originating from the indicated 5 uL voxel. The data shown in Figure 2A were acquired 90 min after the start of IV infusion of D7-glucose. The data in Figure 2B were acquired 120 min after IP infusion of D2-glucose. Note the broader peak of D7-glucose compared to D2-glucose in the ²H MR spectra, and the visually lower SNR for Glx and lactate. Figure 3 provides an overview of different levels of ²H-labeled Glx SNR across different infusion methods, types of ²H-labeled glucose, and spatial resolution of DMI. DMI of glucose metabolism in a glioblastoma-bearing mouse is shown in Figure 4, which depicts the map of ²H-labeled lactate, illustrating the local increase in labeled lactate in the tumor region.

Discussion

Using a simple RF coil design we showed that DMI of glucose metabolism in mouse brain is feasible, achieving a nominal spatial resolution of 5 uL in a 35 min acquisition. A clear SNR advantage is achieved by using glucose that is labeled in both the 1st and 6th carbon position. Note that we used D7-glucose because [1,6,6’-²H₃]-glucose is not readily commercially available. The deuterons on carbons 2,3,4 and 5 do not contribute to labeling of Glx or lactate. During IV infusion of ²H-labeled glucose Glx reaches steady state faster than IP infusion (data not shown). However, when taking this delay into account and starting DMI acquisition 120 min after start of IP infusion, DMI data can be acquired with similar SNR than during IV infusion.

Conclusion

DMI of glucose metabolism in mouse brain is feasible, with the highest SNR being achieved by using D7-glucose administered via intravenous infusion. Regional differences in glucose metabolism were clearly visualized using DMI in the GL261 mouse model of glioblastoma.

Acknowledgements

This research was supported by NIH grant R01- EB025840.