Spatial regression analysis of MR diffusion reveals subject-specific white matter changes associated with repetitive head impacts in contact sports

• Published in: Scientific reports Vol. 10; no. 1; p. 13606
• Main Authors: Asselin, Patrick D., Gu, Yu, Merchant-Borna, Kian, Abar, Beau, Wright, David W., Qiu, Xing, Bazarian, Jeff J.
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 12.08.2020; Nature Publishing Group; Nature Portfolio
• Subjects: 631/114/1314; 631/114/2397; 692/308/409; 692/617/375/1345/3190; Adolescent; Athletes; Brain Concussion - diagnostic imaging; Brain Concussion - etiology; Cognition; Diffusion Tensor Imaging; Female; Football; Head Protective Devices; Humanities and Social Sciences; Humans; Magnetic resonance imaging; Male; multidisciplinary; Neuroimaging; Prospective Studies; Protective equipment; Radiographic Image Interpretation, Computer-Assisted; Regression analysis; Science; Science (multidisciplinary); Seasons; Secondary analysis; Spatial analysis; Spatial Regression; Substantia alba; White Matter - diagnostic imaging; White Matter - pathology; Young Adult
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-020-70604-y

Repetitive head impacts (RHI) are a growing concern due to their possible neurocognitive effects, with research showing a season of RHI produce white matter (WM) changes seen on neuroimaging. We conducted a secondary analysis of diffusion tensor imaging (DTI) data for 28 contact athletes to compare WM changes. We collected pre-season and post-season DTI scans for each subject, approximately 3 months apart. We collected helmet data for the athletes, which we correlated with DTI data. We adapted the SPatial REgression Analysis of DTI (SPREAD) algorithm to conduct subject-specific longitudinal DTI analysis, and developed global inferential tools using functional norms and a novel robust p value combination test. At the individual level, most detected injured regions (93.3%) were associated with decreased FA values. Using meta-analysis techniques to combine injured regions across subjects, we found the combined injured region at the group level occupied the entire WM skeleton, suggesting the WM damage location is subject-specific. Several subject-specific functional summaries of SPREAD-detected WM change, e.g., the L ∞ norm, significantly correlated with helmet impact measures, e.g. cumulative unweighted rotational acceleration (adjusted p  = 0.0049), time between hits rotational acceleration (adjusted p value 0.0101), and time until DTI rotational acceleration (adjusted p  = 0.0084), suggesting RHIs lead to WM changes.