Although MRI has been used to evaluate brain atrophy and effects of treatment in HD mouse models, most studies were cross-sectional or only reported brain atrophy at one time point (Ferrante et al., 2000
; Sawiak et al., 2009a
). Compared to histology-based stereology, MRI is noninvasive, and it provides digitized data with full brain coverage, free from distortions due to embedding and sectioning. Longitudinal in vivo
imaging of HD mice would allow a complete natural history of brain pathological changes to be developed during preclinical trials, and considerably increase the power to detect therapeutic efficacy compared to a single measurement of neuropathology. The current study of brain volumetric changes in a widely-used HD mouse model provides evidence that longitudinal structural MRI has the power to detect the response to neuroprotective treatment, suggesting that the N171-82Q mouse model is suitable for presymptomatic as well as symptomatic preclinical trials.
We also investigated correlations between the regional brain atrophy and motor dysfunction in N171-82Q HD mice. Progressive brain atrophy in striatum and cortex was positively correlated with deficits in motor function, suggesting that imaging measures could predict neuron functional changes. These results indicate that the functional change in N171-82Q mice may have originated from loss of brain volume. Interestingly, these HD mice exhibit dramatic motor deficits and short life-span, but subtle neurodegeneration. Our results suggest that motor phenotype in this mouse model may result from neuronal atrophy and dysfunction rather than neuronal loss; thus, neuroimaging measures might be ideal biomarkers for evaluation of neuroprotective treatment in preclinical trials.
Besides gray matter volume loss, we also detected loss of white matter volume at the early stage in N171-82Q mice. Clinical studies from HD patients' brains suggest two possible theories for decrease in white matter volume. For example, mutant huntingtin might have a direct effect on myelination such that, once myelinated, neuronal circuits become functional, myelin breakdown begins, which causes an excitotoxic process with failure of afferent transmissions causing the underlying neuron to be overstimulated by its efferent feedback (Gomez-Tortosa et al., 2001
; Bartzokis et al., 2007
). Alternatively, the deficit in white matter volume might be a manifestation of abnormal development instead of decreased volume from a degenerative process. Our data indicated earlier decreased white matter volume in the N171-82Q HD mice, but in contrast the volume loss is less progressive than loss of gray matter in these mice, suggesting that developmental defects might be involved in the white matter pathology in HD. Further detailed study of white matter at the earlier stage is needed.
In addition to conventional segmentation-based volume analysis, we have performed deformation-based morphometry (DBM) analysis based on the transformations generated by LDDMM to investigate potential brain atrophy in regions not easily definable, for example, the hypothalamus, and thalamus among others. DBM analysis has several advantages. It is free from intra-rater and inter-rater variations. It can be implemented in a fully automated fashion to increase throughput, which is often the bottleneck for therapeutic screening. It can locate and visualize localized atrophy, which is difficult to achieve with a segmentation-based approach (Maheswaran et al., 2009
). The accuracy of the DBM approach, however, depends on the quality of the transformation and how concentrated the potential atrophy or hypertrophy is among animals (Davatzikos, 2004
). For example, the complex geometry and relatively large morphological variations of the lateral ventricles in the mouse brain makes it challenging for LDDMM to generate accurate image transformations for the ventricles and neighboring regions (Zhang et al., 2010
). The particular portion of the lateral ventricles that is enlarged might also be different among animals. As a result, DBM may not be able to detect volume changes near the lateral ventricles. The sensitivity of the DBM approach to detect atrophy or hypertrophy could also be lower than that of the segmentation-based volume analysis. The segmentation-based approach effectively averages the volume changes of a relatively large group of voxels and therefore has a much higher signal-to-noise ratio, whereas the DBM approach examines volume change at each voxel independently, with corrections for multiple comparisons.
In summary, longitudinal in vivo MRI allows us to investigate disease onset and progression over the life-span of each mouse, which will greatly facilitate preclinical therapeutic trials in HD mice. This approach will provide significant benefits for trials that use regional brain atrophy as an endpoint in treatment trials. These findings provide the first evidence of progressive brain volume loss in different brain regions of theN171-82QHDmousemodel in response to neuroprotective therapy. The findings also guarantee further characterization of other available HD mouse models and evaluation of therapeutics in preclinical trials by MRI.