A prospective functional MR imaging study of mild traumatic brain injury in college football players

Abstract

Background and purpose: Although concussion is common among athletes, evidence-based methods for clinical evaluation, treatment, and recovery are lacking. We used a prospective, functional neuroimaging approach to assess sports-related concussion in which imaging was performed before injury so that brain changes resulting from concussion could be better understood.

Methods: Neurophysiologic correlates of sports-related concussion were investigated in eight college football players by using functional MR imaging. Preseason baseline levels of blood oxygen level-dependent (BOLD) activity were acquired during the performance of a test battery that included mathematical, memory, and sensorimotor coordination tasks. Four players who had a concussion repeated these baseline procedures within 1 week of injury. The remaining control players were retested at the end of the season.

Results: Specific neural signatures of concussion were detected in individual players by comparing postconcussion results to preconcussion baseline values. The validity of these indicators was confirmed by comparing them with the same measures in noninjured control subjects. When compared with control subjects, concussed players had marked within-subject increases in the amplitude and extent of BOLD activity during a finger-sequencing task. Effects were observed primarily in the parietal and lateral frontal and cerebellar regions.

Conclusion: Differences in neural functioning were observed in the absence of observed deficits in behavioral performance, suggesting that this approach may increase sensitivity to concussion compared with neuropsychological evaluation alone. Though preliminary, the proposed prospective neuroimaging approach may have great potential for understanding mild traumatic brain injury and identifying mechanisms underlying recovery.

Fig 1.

Average performance scores (percentages) for concussed players do not change between baseline (white) and postconcussion (dark gray) sessions. Similarly, no change was observed for the control players when we compared the first baseline session (light gray) with the second baseline session (black). Performance on the digit span and addition tasks were similar and consistently above 80% correct for all sessions. The subtraction task appeared considerably more difficult with percentages of correct values well below those for the other tasks.

Fig 2.

Significant mean task-related BOLD responses for all subject and all tasks except addition, which failed to show any consistent activity. Maps were derived from the average of the baseline images of each subject. Activity is overlaid in color on a 3D rendering of a single subject’s brain viewed from the top. Activity displayed for the sequencing tasks include contralateral precentral and postcentral gyrus, bilateral middle frontal gyrus (MFG), medial frontal gyrus (medFG), and inferior parietal lobe (IPL). The subtraction and digit span tasks activated similar areas, including prefrontal cortex (PFC), IPL, amd MFG. The digit span task also recruited areas in visual cortex.

Fig 3.

Representative individual Z-score differences between baseline and either a postconcussion session (concussed, left) or postseason baseline sessions (control, right). Colored areas show regions of activity that significantly increased from the baseline value of the bimanual sequencing task. Although both concussed and control subjects demonstrate some increases, those of the concussed player are considerably larger. Activity is significantly increased in the medial frontal gyrus (medFG), middle frontal gyrus (MFG), inferior parietal lobe (IPL), and bilateral cerebellum.

Fig 4.

The mean voxel volume of significant, within-subject BOLD increases in specific ROIs. Error bars show the between-subject standard error. Each plot shows activity changes for each task in an ROI. Stylized images on the left show the regions encompassed by each ROI for two axial sections. The volume of increase was greater for the concussed players (black) than for the control players (gray), particularly for sequencing tasks. Differences between groups were most prominent in frontal and parietal regions. Biman indicates bimanual; DSpan, digit span; Subt, subtraction.

Fig 5.

Amplitude difference (mean voxel intensity) between baseline and postbaseline sessions averaged within ROI and across subjects. Each plot shows activity changes for each task in an ROI. Stylized images on the left show the regions encompassed by the ROI for two axial sections. In general, concussed players (black) showed greater increases in amplitude compared with control subjects (gray); this was particularly evident in the bilateral lateral frontal and parietal regions. Biman indicates bimanual; DSpan, digit span; Subt, subtraction.