Diffusion MRI was used to examine whether any change in the white matter tracts of patients with mild cognitive impairment (MCI) can predict conversion to Alzheimer’s disease (AD) in a longitudinal study. Our data show increases in mode of anisotropy (MO) in a region of crossing fibres in the centrum semiovale for MCI patients who later converted to AD.
Background
Mild cognitive impairment (MCI) is associated with an increased risk for developing dementia in a short time period1. Nevertheless, MCI is a heterogeneous condition, and valid biomarkers able to predict conversion are desperately needed. MRI is sensitive to different aspects of tissue modifications across Alzheimer’s disease (AD) in both white and grey matter2. A previous study extensively compared indices derived from diffusion tensor image (DTI) across AD, MCI and age-matched controls3. While significant changes in all indices were found in AD compared to the other groups, the only parameter that was significantly different between MCI and healthy controls was the mode of anisotropy (MO), which quantifies the shape of the diffusion tensor. The aim of this longitudinal study is to assess whether any index derived from DTI is able to predict conversion from MCI to AD.Methods
Data from 54 MCI patients (M/F=29/25, mean age = 69.8 years (SD 7.77), range: 47-82 years) were acquired at 3T. The protocol included a T1-weighted volume, DTI and resting state functional MRI. Patients were reassessed after a period of 24 months (± 6 months), and AD conversion recorded. Patients were thus classified as MCI converters and non-converters.
All diffusion-weighted images were corrected for involuntary motion and eddy current distortions using affine registration and the FMRIB's Linear Registration Tool (FLIRT). The images were skull-stripped using FMRIB's Brain Extraction Tool (BET), and FMRIB’s dtifit was used to fit tensor indices: Fractional Anisotropy (FA), Axial Diffusivity, Radial Diffusivity, Mean Diffusivity and Mode of Anisotropy (MO). The FA data were then warped into MNI space using the Advanced Normalization Tools (ANTs) 2.1.0., and the resultant warpings were applied to the remaining tensor indices.
Voxel-wise analysis was performed using Tract-Based Spatial Statistics (TBSS). Permutation inference4 was performed on the skeletonised images with randomise, available with FSL, using the Threshold-Free Cluster Enhancement (TFCE) method. Results were accepted as significant for p < 0.05 after correction for multiple comparisons.
At follow-up, 12 patients (M/F=4/8, mean age = 73.1 years (SD 5.60), range: 63-82 years) had converted to AD, while 42 patients (M/F=25/17, mean age = 68.8 years (SD 8.10), range: 47-81 years) remained as MCI.
Increased MO was found in a region of crossing fibres in the right centrum semiovale and internal capsule (Fig 1). No other tensor indices showed significant differences between MCI converters and non-converters.