Cognitive and Salience Network Connectivity Changes following a Single Season of Repetitive Head Impact Exposure in High School Football

Overview

The imaging Telemetry And Kinematic modeLing (iTAKL) study followed high school football players throughout a football season to study the effects of repetitive HIE on the brain. The players underwent cognitive testing and MRI during the same study visit, both preseason and postseason. HIE was measured with the Head Impact Telemetry System (HITS; Simbex), which uses accelerometers embedded in the helmets. Preseason assessments were several days to 1 month before the first contact practice, and postseason assessments were performed at the earliest opportunity after the conclusion of the season (median, 26 days; interquartile range, 14–38 days). Control athletes from local sports teams were recruited for preseason and postseason cognitive testing; they also underwent pre- and postseason MRI during their study visits. Additional control athletes were recruited for cognitive testing by collaborators as part of a similar high school study conducted at the Geisel School of Medicine at Dartmouth (Hanover, New Hampshire). The control athletes participated in a variety of noncontact and limited-contact sports, including tennis, track, cross-country, rowing, swimming, golf, and baseball.

Standard Protocol Approvals, Registrations, and Participant Consents

Study protocols at each site were approved by the appropriate institutional review board at Wake Forest School of Medicine and the Geisel School of Medicine at Dartmouth. Written informed consent was obtained from a parent or legal guardian, and written assent was obtained from each study participant.

Football players 13–18 years of age at the time of enrollment were recruited from the junior varsity and varsity teams of a local high school during the 2012–2019 football seasons. Potential participants were excluded if they or their parents reported that the participant had any of the following: 1) developmental disorders, 2) neurologic disorders, 3) psychological disorders, 4) medications known to alter brain rhythm, or 5) MRI contraindications. To focus on repetitive HIE, we excluded players who experienced a clinically diagnosed concussion during the season or had reported a history of previous concussion from the analysis. Furthermore, we excluded football players who had their postseason assessments >6 weeks after the end of the season to limit the influence of factors unrelated to football participation (i.e., participation in a winter sport, deconditioning, and so forth). Board-certified neuroradiologists evaluated all MRI scans for clinically relevant abnormalities; participants with overt structural abnormalities, including but not limited to tumors, previous or acute infarcts, AVMs, or aneurysms were excluded from the study. For players with multiple years of participation in the study, we used the data from their first season for the analysis. Of 95 football players, 71 were considered for analysis (Fig 1).

• Download figure
• Open in new tab
• Download powerpoint

FIG 1.

Flow chart of football group exclusion steps and sample sizes for all the analyses.

A group of male high school athletes participating in noncontact and limited-contact sports, with no prior tackle football experience, were used as controls. Thirteen controls from local high school teams were enrolled in the iTAKL study and underwent the same cognitive assessments and imaging protocol as the football players at preseason and postseason time points coinciding with the pre- and postseason study visits for the football athletes. For the cognitive analysis, an additional group of 21 athletes enrolled in New Hampshire with the same pre- and postseason cognitive assessments was included to augment the size of the control group and increase statistical power. These additional control athletes were subject to the same exclusion criteria as participants in the iTAKL study and did not differ from the iTAKL controls in age or cognitive performance (Online Supplemental Data).

Cognitive Assessment

Participants took the Immediate Post-Concussion Assessment and Cognitive Test (ImPACT; https://link.springer.com/referenceworkentry/10.1007/978-0-387-79948-3_1873), a widely used computerized neuropsychological tool at their pre- and postseason study visits.¹⁹ Test administration was completed individually under the supervision of trained personnel. Data quality was reviewed and checked for validity using the Impulse Control composite scores (invalid if >30). Sixty-three of 71 football players and all 34 controls had complete and valid data for both time points (Fig 1).

Head Impact Measurement

During all practice and game sessions, football players wore helmets fitted with HITS. The HITS is a common impact measurement system in American football and uses an array of 6 single-axis accelerometers that are spring-loaded to maintain contact with the head.^20⇓-22 To avoid recording spurious impacts, we set the lower limit of data acquisition to a peak linear head acceleration of 10 g. Additionally, videographic recording and review of each session was conducted to ensure that all impacts not directly related to helmeted athletic participation were removed from the data set. The composite metric risk-weighted exposure combined probability (RWEcp) was used to capture cumulative HIE, as detailed in prior studies.^23⇓-25 Of the 63 football players with complete cognitive data, 58 had HITS data sufficient to compute the HIE metrics; of the 57 football players with complete imaging data, 51 had HITS data sufficient to compute HIE metrics. A total of 50 football players had complete cognitive data, imaging data, and HIE metrics (Fig 1).

MRI Acquisition

All football players were scanned during their pre- and postseason assessments; 13 iTAKL controls were scanned at similar time points, which, as mentioned above, were timed to coincide with the football players’ study visits and did not always coincide with the start and end of their respective sport season. All MRI sequences throughout the study were acquired on the same 3T Magnetom Skyra MRI scanner (Siemens) using a 32-channel head/neck radiofrequency coil. Participants were supine and were instructed to remain still with their eyes open during acquisition. The resting-state BOLD data were acquired for 6.3 minutes using a gradient-echo-planar imaging sequence with the following parameters: TR = 2000 ms, repetitions = 190, TE = 25 ms, flip angle = 90°, field of view = 224 × 224 mm, 3.5-mm³ isotropic resolution. Field map imaging was performed using a double-echo gradient-echo sequence (TR = 488.0 ms, TE = 4.92/7.38 ms, voxel size = 3.5 × 3.5 × 3.5, flip angle = 60°) that generated 2 magnitude images and 1 phase image. Structural T1-weighted images were acquired for spatial normalization using a 3D MPRAGE sequence with the following parameters: TR = 2300 ms, TE = 2.98 ms, flip angle = 90°, FOV = 256 × 256 mm, 1-mm³ isotropic resolution.

Imaging Data Preprocessing

Field-map distortion correction of the BOLD data was performed using the standard tool FUGUE in FSL (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FUGUE). All further preprocessing was performed with Statistical Parametric Mapping 12 (SPM12) software (http://www.fil.ion.ucl.ac.uk/spm/) as implemented in the CONN functional connectivity toolbox (Version 20.b; https://web.conn-toolbox.org/) preprocessing pipelines,²⁶ running on Matlab R2015b (MathWorks). Imaging data were processed using a standard CONN preprocessing pipeline. T1-weighted images were realigned and normalized to the Montreal Neurological Institute template and used to generate tissue segmentations of gray matter, white matter, and CSF. Resting-state BOLD data-processing included realignment, slice-timing correction, direct segmentation and normalization into Montreal Neurological Institute space, and spatial smoothing with an 8-mm³ full width at half maximum Gaussian kernel. For motion artifact detection and removal, volumes or scans with the following cutoffs were flagged as outliers: 1) scan-to-scan global BOLD signal change of >3 SDs, or 2) >1-mm frame-wise displacement. The denoising step included scrubbing of the flagged outlier scans, regression of the 6 motion parameters and their first-order derivatives, and the removal of white matter and CSF-related noise on the basis of the first 5 principal components estimated from the BOLD signal of these segments using the anatomical component correction (aCompCor) process, which is a component-based noise-correction method.^27,28 After scrubbing and nuisance regression, images were detrended and bandpass-filtered between 0.008 and 0.1 Hz. After denoising, 2 football players with >20% of outlier volumes per scanning session (>38/190 volumes) were excluded from the imaging analysis due to their excessive motion; an additional 12 football players were excluded during a visual quality check of the BOLD data, principally for incomplete brain coverage.

Functional Connectivity Measures

Fifty-seven football players and the 13 controls were included in the functional connectivity analysis, performed using the CONN functional connectivity toolbox (Version 20.b).²⁶ Interhemispheric connectivity (IHC) at each voxel was computed by calculating the Fisher-transformed correlation coefficient between the BOLD signal at a given voxel and that of the corresponding voxel in the contralateral hemisphere. To calculate IHC for our ROI, the dorsal ACC, we averaged the IHC of all voxels in the right dACC (Brodmann area 32) as delineated by the Brodmann atlas available within CONN. Within-network connectivity of the SN was calculated by averaging the Fisher-transformed pair-wise correlations for all the ROI pairs in an ROI-to-ROI connectivity matrix of SN regions, defined in the standard networks atlas in CONN.²⁹ These SN ROIs were obtained from an independent component analysis of 497 subjects from the Human Connectome Project (https://www.humanconnectome.org/) and included 7 regions: the joint anterior cingulate cortices, right and left anterior insulae, right and left supramarginal gyri, and right and left rostral prefrontal cortices (RPFC). Connectivity measures were exported from CONN for further statistical analysis.

Statistical Analyses

Functional Connectivity

Due to the disproportionate sample sizes, with only 13 control athletes versus 57 football players, we refrained from comparing connectivity-change differences between the groups, because the limited control data may lead to unreliable inferences. Preseason-to-postseason changes in the IHC of the dACC and SN connectivity were assessed independently in the football and control groups using paired-sample t-tests in jamovi. Additionally, connectivity measures of the football group were examined for correlations with ImPACT Δ scores (post- minus preseason). Due to the small sample size of the control group, analyses are underpowered but were performed to reveal the direction of change. To examine which subregions of the SN drove connectivity changes, we performed ROI-to-ROI connectivity analysis within the SN for each group independently in CONN to visually illustrate significant changes at the connection level (P value uncorrected, < .05).

Correlations with RWEcp

For the football players, Δ scores for ImPACT composites, IHC of the dACC, and average within-network connectivity for the SN were calculated. The Pearson or Spearman correlation was used to evaluate the linear correlation between these change measures and RWEcp.