Alterations of brain glucose metabolism in aged APP/PS1 mice: An original voxel-based statistical analysis using BrainRAT and SPM

Background: Glucose brain metabolism is a widely used clinical biomarker in Alzheimer's disease (AD), usually examined on 2D post mortem autoradiographic data in AD animal models. We propose an innovative method to analyze these data in 3D without a priori anatomical information, to derive brain metabolic activity maps using voxel-based approach in APP/PS1 mice. Results will be discussed relative to previous findings in this field. Methods: Uptake of [14C]-2-deoxyglucose was measured in adult awake APP/PS1 (64 ± 1 weeks, n = 4) and PS1 (65 ± 2 weeks, n = 3) transgenic mice. Glucose uptake was evaluated by autoradiography on the right hemisphere while the left hemisphere was processed for Congo red staining. Data acquisition was performed with a digital camera (blockface photographs) and a high resolution flatbed scanner (autoradiography, histology). Blockface, autoradiographic and histological post mortem volumes were 3D-reconstructed using BrainRAT in-house software (freely available, http://brainvisa.info) while statistical analysis was achieved using SPM5. Results: We successfully combined BrainRAT (computerized procedures for acquisition and 3D reconstruction of anatomic and functional volumes) and SPM5 (voxel-wise statistical analysis) methodologies to data sets mapping brain metabolic activity in a transgenic mouse model of AD. We were able to extract accurate parametric mapping of both hypo- and hypermetabolic regions in the APP/PS1 animals relative to PS1 controls. Decreased glucose uptake was observed within cortex (cingulate -36%, retrosplenial -26%, somatosensory -23%), striatum (-22%), thalamus (-29%) and hippocampus (-25%). Increased glucose uptake was also detected within other cortical areas (piriform +22%, perirhinal +19%), amygdala (+23%), dorsal endopiriform (+34%) and accumbens (+20%) nuclei, dentate gyrus (+25%) and dorsal hippocampus (+26%). Conclusions: We have demonstrated the ability to robustly and automatically reconstruct post mortem functional volumes. We also provided accurate cerebral glucose uptake mapping in a murine AD model, in 3D, without a priori hypotheses. We offer an efficient and standardized method to compare local changes in mice cerebral glucose utilization across ages, lines or drugs. Furthermore, our work demonstrates the possibility of using functional brain imaging and dedicated software to help bridge the gap, in translational studies, between features of neurodegenerative diseases in human beings and animal models.Acknowledgements: Sanofi-Aventis for sharing the mouse strains.