GlucoCEST MRI allows visualisation of regional glucose uptake in tumours based on the chemical interaction between hydroxyl protons in glucose and water¹. However, the gold standard method for the detection of tumours, fludeoxyglucose positron emission tomography[18F-FDG-PET], does not provide sufficient contrast to differentiate the majority of brain gliomas. The high glucose background in brain makes the technique ineffective at identifying highly metabolic cancer nodes. Nonetheless, due to the fact that FDG and natural glucose do not share the same metabolic path and that glucoCEST is sensitive to sugars formed along the glycolytic pathway², it is possible that glucoCEST could provide valuable information on brain gliomas.

On these premises and following our previous study^3, in this work we explore the feasibility of using glucoCEST as a tool for early detection of primary brain tumours.

Cancer cells from four different patient biopsies and cultured in the laboratory were injected intra-cranially into immune suppressed mice. The xenograft tumours from these cells showed a diffused phenotype. Additionally, U87 cancer cells were included in the study, which are known to form solid tumours. Mice were fasted for 12 hours prior to the experiments in order to reduce and stabilize blood glucose levels. Animals were anaesthetised with 1.3% isoflurane for the duration of the experiment. A pressure pad placed under the animal’s chest was used to monitor the respiration rate. Body temperature was monitored with a probe placed on top of the animal and was maintained at 37°C with a flow of warm air controlled by an SA Instruments monitoring system. Mice were cannulated via the intra-peritoneal (IP) route for the administration of glucose while in the scanner. A dose of 1g/kg D-glucose was given from a solution of 10% glucose in saline. All experiments were done under a Home Office approved license.

High resolution structural images (74μm²×0.5mm) were acquired using a T2 weighted Spin Echo sequence (T2wSE, TR=3s and TE=20ms). Z-spectra were acquired using a saturation train of 80 Gaussian pulses prior to a turbo-flash readout. Each Gaussian pulse was 50ms long with flip angle of 540° and 91% duty cycle, providing an equivalent of 0.9 𝜇𝑇 B1 power and 4 seconds saturation length. Saturation was applied at 57 frequency offsets ranging from -4.5 to 4.5 ppm in a linearly spaced pattern. The total temporal resolution was 4 minutes per Z-spectra. Readout parameters were 2.73ms TR, 1.52ms TE and 20°excitation angle. The K-space was sampled from low to high frequencies for a matrix size of 64 by 64 voxels per slice and a field of view of 20mm by 20mm, with 1.3mm slice thickness. Three parallel slices of the brain were scanned for each saturation train length. A minimum of three CEST baselines were scanned before the infusion of glucose IP, after which another nine CEST measurements were taken. The glucoCEST signal was calculated as the subtraction of the mean MTR asymmetry between the first and last three CEST images (post minus pre glucose administration). Z-spectra were fitted with a smoothing spline and corrected for B0 drifts on a pixel by pixel basis. GlucoCEST enhancement maps (GCE) were obtained by integration of MTR asymmetry between 0.75 and 1.5 ppm (Figure 1).

Figure 2 shows the case of an animal injected with U87 cancer cells that displays an hyper-intense glucoCEST signal in the right side of the cortical region from where the tumour grows at a later stage.

This same effect is manifested particularly in animals with diffused phenotype tumours (Figure 3). Eight weeks post inoculation, glucoCEST displays an intense signal in the top right hand side of the brain where the tumour will grow (see week 12). At this stage T2wSE does not provide signs of any anomaly. At a later stage a normo-intense signal is observed in areas where cancer has developed but a hyper-intense signal around the tumour area, conceivably indicating further expanse of tumour towards those regions.

It is worth emphasising that in the majority of the cases GCE images displayed no significant contrast in regions where the cancer had been consolidated. This was especially true at late stage of tumour development when GCE showed normo-intense signal in the cancer but hyper-intense in areas where the spread of the cancer had not yet fully developed (Figure 4).