Gender Differences in the Dependence of Body Weight and Brain Connectivity Revealed by Diffusion Tensor Imaging

Introduction: Recent studies showed that obesity is associated with changes in brain structure (1,2). It was also suggested that cerebral changes may be similar to those observed in drug addiction. However, the underlying mechanisms are unclear. To address these questions, we investigated the dependence between body weight and brain structure using diffusion-tensor imaging (DTI). Correlations between the body mass index (BMI) and fractional anisotropy (FA) were computed in female and male subjects separately. Axial and the radial diffusivity were additionally studied, which had been investigated by Song et al. (3,4) in animal models including the shiverer mouse. Methods: High-resolution diffusion-weighted (DW) images (60 directions, b = 1000 s/mm2, GRAPPA accel. fact. 2, NEX = 3) were acquired on a whole-body 3T TIM Trio scanner (Siemens, Erlangen, Germany) with a 12-channel head array coil.. The measurement of 72 axial slices with an isotropic resolution of 1.7 mm covered the entire brain. A set of 49 subjects was analyzed (23 female; age=26.4±5.0 y, BMI 18.5-50.7, mean 29.1, std=7.0). Brain masks were extracted from T1-weighted structural scans and co-registered into the Talairach space. The DW images were corrected for subject motion and registered to the T1 anatomy with isotropic voxel size of 1mm using rigid-body transformation computed with FLIRT (5). For each voxel, a diffusion tensor was fitted to the data, and the FA was computed. Non-linear registration was used for inter-subject alignment (6). Using the deformed FA images, a mean image was created and thinned to create a mean skeleton, which represents the centers of all tracts common to the group (7). Finally, each subject's aligned FA were projected to the mean skeleton and fed into voxel-wise cross-subject statistics based on a general linear model containing BMI, gender, and age as covariates. In order to investigate gender differences, the statistical analysis was also performed for the female and male subjects separately. In addition to the FA, axial and radial diffusivity was investigated. Axial diffusivity is defined by the eigenvalue of the primary eigenvector and is also referred to as parallel diffusion. Radial diffusivity is the mean of the eigenvalues of the second and third eigenvectors, and it is also attributed to perpendicular diffusion.