Associations between Head Impact Exposure in Youth Football Players, Magnetoencephalography (MEG) and Diffusion Kurtosis Imaging (DKI) Metrics

Introduction:: Multiple concussions and cumulative exposure to repetitive subconcussive head impacts (RSHI) have been associated with long-term neurological impairment and neurodegenerative diseases in football players. Our group has prospectively investigated whether RSHI, in the absence of a clinically diagnosed concussion, leads to changes in advanced neuroimaging metrics in high school (14-18y/o) football players. These studies showed that changes in advanced imaging were associated with RSHI over a single season of high school football. To further investigate these findings, we expanded the study to include youth (8-13y/o) football players. The objective of the present analysis was to compare imaging variables between a control group and groups of players with the highest and lowest head acceleration scores throughout the season.

Materials and Methods:: 68 youth football players (all male; average age = 12.2 y) and 17 non-contact sport controls (all male; average age=11.5y) were included in this study. Football players wore sensor-embedded helmets that measured head impacts throughout the season. From this data, rotational and linear risk-weighted exposure (RWE) measures were calculated per subject. Control participants were assumed to have a 0 RWE value. In addition to the control group, four groups of 17 football players each were created: lowest rotational RWE (L-RWE_Rotational), lowest linear RWE (L-RWE_Linear), highest rotational RWE (H-RWE_Rotational), and highest linear RWE (H-RWE_Linear). DKI MRI and 8-minutes of eyes-open resting-state MEG data were acquired pre-season and post-season. MEG data underwent standard pre-processing, and the images were source localized to the structural T1w MRI in Brainstorm. The relative power per frequency band and mean kurtosis (MK) for each voxel were calculated and normalized to MNI space. Voxel-wise difference maps (post-season minus pre-season) were computed for each frequency band per subject. Voxel-wise z-scores were computed utilizing the control group’s mean and standard deviation. Whole-brain z-score maps were thresholded at 2 standard deviations above the mean of the controls resulting in the number of abnormal voxels (AV) per imaging measure. To identify significant differences between groups and to determine the imaging variable that could best differentiate the highest head impacts from the lowest and control groups, we conducted a two-step analysis consisting of a multivariate analysis of covariance (MANCOVA) followed by a linear discriminant analysis (LDA).

Results, Conclusions, and Discussions:: When evaluating between football player (H-RWE_Rotational and L- RWE_Rotational) and control groups, the MANCOVA model was significant (p< 0.05) and the model’s effect size was 0.27. A LDA analysis looks at Fisher’s linear discriminants LD1 and LD2. Firstly, LD1 is representative of a linear function that achieves the maximum separation between groups and identifies the variable(s) responsible for this differentiation, while LD2 is a linear function orthogonal to LD1 that also identifies the variable(s) responsible for maximal differentiation between all three groups. The LD1 results demonstrated higher AV in theta and MK differentiated the H- RWE_Rotational group from the other groups, while LD2 indicated lower theta values differentiated L- RWE_Rotational football players and control groups. A lower AV in low gamma differentiated the controls from the other groups. When comparing the results between the football players (H-RWE_Linear and L-RWE_Linear) and control groups, the MANCOVA model was significant (p< 0.0001) and the model’s effect size was 0.42. The LD1 results demonstrated higher AV in MK differentiated all football players from the control group, while LD2 indicated that lower AV in theta differentiated H-RWE_Linear from L-RWE_Linear football players and controls.
MEG and DKI can be valuable tools for monitoring the progression of RSHI in pediatric populations. Specifically, MK was able to distinguish controls from football players, while theta differentiated the high RSHI exposure group from the lowest RSHI exposure group and controls.



Acknowledgements (Optional): :

References (Optional): :