Joseph A. Helpern
Establishing neuroimaging biomarkers for the early detection of Alzheimer disease (AD) is crucial for the development of disease-modifying therapies. Indeed, investigation of potential neuroimaging biomarkers is one of the key strategies under the first goal of the 2012 National Alzheimer’s Project Act, as well as for President Obama’s BRAIN initiative. Our research seeks to address this need through innovations in quantifying and modeling tissue microstructure in the AD brain.
Our basic approach in this area is based on the hypothesis that changes in brain tissue microstructure are early indicators of AD. A particularly powerful, non-invasive method for probing the integrity of tissue microstructure is diffusion MRI (dMRI). Most dMRI studies to date have employed a particular technique known as diffusion tensor imaging (DTI), mainly because it is easily implemented on standard clinical MRI scanners and because the data analysis is relatively straightforward. However, DTI is limited in its information content because it assumes that brain water diffusion is Gaussian, even though we know it is markedly non-Gaussian. This non-Gaussianity is a direct consequence of complex tissue microstructure, and we conjecture that its quantification can improve our understanding of the microstructural changes associated with the earliest stages of AD.
In order to have a dMRI method that maintains the practical advantages of DTI, but also allows for the quantification of diffusional non-Gaussianity, our lab introduced a new approach called diffusional kurtosis imaging (DKI).1 Indeed, recent studies using DKI have provided new insights into microstructural changes for several neuropathologies, including multiple sclerosis,2 brain cancer,3,4 stroke,5,6 attention-deficit/hyperactivity disorder,7 traumatic brain injury,8,9 and epilepsy.10
Although the diffusion metrics obtainable with DKI are sensitive to microstructural changes, the precise relationship between diffusion and tissue microstructure can be subtle. Nonetheless, by combining DKI with tissue modeling methods, we are able to make predictions for a variety of microstructural properties, including volume fractions, orientations, and compartmental diffusivities of cellular structures (eg, axons) that may be altered with disease. In our paper, we applied DKI-based tissue modeling of the white matter to investigate the sensitivity, diagnostic accuracy, and associations of specific microstructural changes through the course of AD. Our findings suggest that widespread breakdown in myelin integrity occurs first in the transition from normal aging to the amnestic mild cognitive impairment (aMCI) stage (AUC=0.95, P<.001), whereas a loss in axonal density occurs later in the disease from aMCI to AD (AUC=0.84, P=.01). Regional analyses of these metrics reveal their marked functional