Altered glucose uptake is associated with altered pathologies - malignant tumors, for example, gain energy by anaerobic glycolysis (Warburg effect [1]). The clinical gold standard to monitor metabolic rates of glucose is FDG-PET, which uses the radioactively labeled [¹⁸F]-fluordeoxyglucose (FDG). As a non-radioactive alternative to FDG-PET we propose to use dynamic ¹⁷O-MRS of ¹⁷O-labeled D-glucose [2-4] to assess the metabolic glucose pathway.

To optimize the acquisition parameters for dynamic ¹⁷O-MRS in mouse brain, phantom experiments (Fig. 1) were conducted with ¹⁷O-labeled D-glucose (Nukem Isotopes Alzenau, Germany) on a 9.4 Tesla animal MR system (Bruker BioSpin). For RF excitation and signal reception a two-turn 2 cm diameter surface coil tuned to the ¹⁷O frequency (54.271 MHz) was built, and spectra were acquired with an FID sequence.
Animal preparation:
Each mouse was anesthetized (4% induction/1.25% continous rate of isoflurane) prior to introducing a catheter into the tail vein. During the MRS experiment the mouse was continously anesthetized with 1.25% isoflurane, breathing rate and body temperature were monitored and kept as stable as possible.
After placing the surface coil on the mouse brain, baseline spectra were acquired over 27 min. The following MRS parameters were used: spectral bandwidth 62.5 kHz, pulse length 62.5µs, acquisition delay 25µs, TE = 56µs, TR = 5.4ms, flip angle α ≈ 90, 5791 averages, T_acq = 5ms, 313 points were sampled with a dwell time of 16µs. Then, a solution was injected through the catheter which consisted of 80mg of either ¹⁷O-labeled 1-glucose, 1,6-glucose, 6-glucose, or unlabeled glucose dissolved in 200µl 0.9% NaCl, and ¹⁷O spectra were acquired over the course of up to 163 min (temporal resolution: 42 s). Glucose uptake was measured from the signal change of the H₂¹⁷O water peak. In the phantom, the ¹⁷O spectra (Fig.1) were quantified in the time domain using the HLSVD algorithm of JMRUI [5]. In the dynamic in vivo spectra, at first the H₂¹⁷O peak was separated from the short-T₂ glucose peak at -11 ppm by discarding initial points from the FID emulating an echo time of TE = 216ms. H₂¹⁷O signal intensities were quantified from the peak height of the magnitude spectra, and converted into absolute H₂¹⁷O concentrations by normalizing to the natural abundance of ¹⁷O (assuming a mouse brain weight of 0.4g and water content of 73%). The dynamics of 1-glucose and 6-glucose peaks were plotted after suppressing the water peak with HLSVD (Fig. 2). Since the anomeric 1-OH position has a temperature and pH-dependent chemical exchange with unlabeled water in solution [6,7], only the time course of the 6-glucose experiment was chosen to further investigate glucose metabolism. Using a pharmacokinetic model [8], the time course of the H₂¹⁷O data was fitted yielding cerebral rates of glucose metabolism CMR_Glc (Fig. 3). Glucose enrichment factor α in blood was estimated from separated measurement of blood sugar concentration and included as constant fit-parameter. Note that, the model was modified by considering that 1mol 6-glucose is converted into 1mol H₂¹⁷O.

In the phantom spectra the following line widths and chemical shifts relative to the water peak (FWHM =2 ppm) were obtained: Δf_1-OH = 40±2 ppm / FWHM_1-OH = 27ppm, Δf_6-OH = -11 ± 2 ppm / FWHM_6-OH =13ppm. The chemical shifts of the 1-OH and 6-OH position are in good agreement with literature [2], if a mixture of 36% α and 64% β glucose is assumed. In vivo line widths and chemical shifts were similar to those of the phantom measurements. In all in vivo experiments with labeled glucose a H₂¹⁷O signal increase of about 12-47% was observed, whereas no increase of H₂¹⁷O signal was seen with unlabeled glucose. From the model fit, a CMR_Glc of 0.43 ± 0.21 µmol/g/min was found which is about 65% higher than literature value of 0.26 ± 0.10 µmol/g/min measured with ¹⁸F-FDG PET in mouse under 1.0% isoflurane [9]. The delay of about 13 min between bolus injection and signal increase in glucose-6 experiment is 56% longer compared to a ¹³C glucose bolus experiment performed in humans without anesthetic [10]. In this work, dynamic ¹⁷O-MRS spectra are acquired in vivo in mouse brain for the first time. From the concentration time courses metabolic rates were obtained which indicate that the labelled oxygen in glucose is transformed into H₂¹⁷O water by glycolysis. The deviations from literature values can be a result of uncertainties in the model fit, or chemical exchange processes that occur before glycolysis.These measurements are a first step towards a MR-based method for the quantification of glucose metabolism.