Data Acquisition
Participants’ information was acquired from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). The ADNI led by Principal Investigator Michael W. Weiner, MD was launched in 2003 as a public-private partnership. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer's disease (AD). We extracted baseline visit data for which demographic data, post-processed DTI, and plasma p tau 181 tau levels from baseline visit were available. We also downloaded results of white matter hyperintensity analysis and brief volumetric data for better understanding. Our cross-sectional study consisted of 21 patients with AD, with their baseline plasma p tau 181, CSF Amyloid β, CSF total tau, and CSF p tau levels and their post-processed Diffusion Tensor Imaging (DTI).
Participants were classified as AD patients if their mini-mental state examination (MMSE) was between 20 and 24, clinical dementia rating (CDR) score was between 0.5 and 1, and met the National Institute of Neurologic and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association criteria (12).
Plasma p tau 181 Measurements
Plasma samples were analyzed at the University of Gothenburg, Sweden by using the Single-Molecule array (Simoa) technique. The detailed procedure is described in (adni.loni.usc.edu). (11).
DTI Imaging Processing
We extracted the results of DTI ROI analysis from ADNI. For each participant, all images were corrected, normalized and extracerebral tissue was removed by the Extraction Tool (BET) in FSL (13). To align data from different subjects into the same 3D coordinate space, each T1- weighted anatomical image was linearly aligned to a version of the Colins27 brain template (14) using FSL’s flirt (15) with 6 degrees of freedom to allow translations and rotations in 3D. The Colin27 brain was zero-padded to have a cubic isotropic image size 220x220x220 1mm^3) and then downsampled (110x110x110 2mm^3) to be more similar to the DWI resolution.
To adjust echo-planar imaging (EPI) induced susceptibility artifacts, which can cause distortions at tissue-fluid interfaces, skull-stripped b0 images were linearly aligned to their respective T1-weighted structural scans using FSL’s flirt with 9 degrees of freedom and then elastically registered to their aligned T1 scans using an inverse consistent registration algorithm with a mutual information cost function (16) as described in (17). The resulting 3D deformation fields were then applied to the remaining 41 DWI volumes before mapping diffusion parameters. To account for linearly registering the average b0 from the DWI images to the structural T1-weighted scan, a corrected gradient table was calculated.
A single diffusion tensor was modeled at each voxel in the brain from the eddy- and EPI-corrected DWI scans using FSL’s dtifit command, and scalar anisotropy and diffusivity maps were obtained from the resulting diffusion tensor eigenvalues (λ1, λ2, λ3). Fractional anisotropy (FA) was calculated from the standard formula.
We registered the FA image from the JHU DTI atlas (18) to each subject using a previously described mutual information-based elastic registration algorithm(16). We then applied the deformation to the stereotaxic JHU “Eve” WM atlas labels (http://cmrm.med.jhmi.edu/cmrm/atlas/human_data/file/Atlas Explanation2.htm) using nearest neighbour interpolation to avoid intermixing of the labels.
This placed the atlas ROIs in the same coordinate space as our DTI maps. We were then able to calculate the average FA and MD within the boundaries of each of the ROI masks for each subject. Of the 56 WM ROIs, we excluded 4 ROIs, the left and right middle cerebellar peduncle, and the pontine crossing tract, as they often fall wholly or partially out of the field of view (FOV). We note that this is also occasionally true of the left and right medial lemniscus, inferior and superior peduncles. We only included non-zero voxels within the FOV in our calculations of mean FA and MD. In addition to the 52 JHU labels, five more ROIs were evaluated: the bilateral fornix, bilateral genu, bilateral body, and bilateral splenium of the corpus callosum and the full corpus callosum, to get full summary measures of the regions.
Tensor based spatial statistics (19) was also performed, and the mean FA in regions of interest along the skeleton was extracted. TBSS was performed according to protocols outlined by the ENIGMA-DTI group: http://enigma.loni.ucla.edu/wpcontent/uploads/2012/06/ENIGMA_TBSS_protocol.pdf
In short, all subjects were registered to the ENIGMA-DTI template in ICBM space, and standard tbss steps were performed to project individual FA maps onto the skeletonized ENIGMA-DTI template. ROI extraction was also performed according to the following protocol to extract the mean FA in ROIs along with the skeleton: http://enigma.loni.ucla.edu/wpcontent/uploads/2012/06/ENIGMA_ROI_protocol.pdf.
Cognitive assessments
The patients' cognitive condition was assessed by the Mini-Mental State Exam (MMSE), which is a common test of cognitive function among the elderly, including tests of orientation, attention, memory, language, and visual-spatial skills. MMSE scores were extracted for each patient from the ADNI Mini-Mental Examination.
Statistical Analyses
We used SPSS16 software for statistical analyses. First we tested the normality of variables by Kolmogorov Smirnov and Shapiro-Wilk tests. Then we log-transformed the non-normal variables to normal distribution. We analyzed the difference between groups by one-way ANOVA using Bonferroni correction for multiple comparisons. Then we used a partial correlation adjusted for age, sex, and APOE for checking the relation between plasma p tau 181 and other demographic variables once among all participants and then within groups. We used the same model for investigating the association between plasma p tau 181 and DTI values in each ROI. The bootstrap method was used for addressing type I error due to multiple comparisons in the correlation models.