This study was approved by the Mayo Clinic and Phoenix VA Institutional Review Boards. Study participants were between 18 and 65 years of age. PTH participants were enrolled from Mayo Clinic Arizona and the Phoenix VA Health Care System and HCs were recruited through advertisements and word-of-mouth. All participants completed an informed consent process and signed informed consent forms. Data were collected over a three-year period between 2020 and 2023. Participants were diagnosed with acute PTH due to mTBI using the ICHD-3 criteria [23]. PTH participants were enrolled between the day of their injury until 59 days post mTBI. Participants completed questionnaires and structured interviews that collected data on patient demographics, current and prior mTBI characteristics, and headache characteristics [24]. Exclusion criteria for HC and PTH included gross anatomical abnormalities on brain MRI, history of severe psychiatric disorder as assessed by the principal investigator, such as but not limited to schizophrenia and bipolar disorder, history of speech disorders, pregnancy, and history of moderate or severe TBI, contraindications to MRI or history of medical condition that contraindicates research participation, chronic headache within 12 months prior to the mTBI that led to the current PTH (including PPTH, chronic migraine, medication overuse headache, new daily persistent headache, hemicrania continua, or chronic tension type headache), use of headache preventive medication within 3 months prior to screening, use of onabotulinumtoxinA in the head, neck or face in the prior 12 months, and use of opioids or barbiturates on at least 4 days per month during the 6 months before screening. HC participants were excluded if they had history of mTBI or migraine. HC participants with tension-type headaches on three or fewer days per month were not excluded.

Neuroimaging was performed on a 3T Siemens scanner (Siemens Magnetom Skyra, Erlangen, Germany) at Mayo Clinic Arizona. A 20-channel head and neck coil was used for all imaging. T1-weighted images were acquired for anatomical reference using a magnetization prepared rapid image acquisition gradient echo (MPRAGE) sequence with 1 mm isotropic resolution with echo time (TE = 3.03 ms), repetition time (TR = 2400 ms), and flip angle (FA = 8 degrees) covering 160 × 256 × 256 mm3. To obtain T2* maps, 12 T2-weighted gradient echo (GRE) magnitude images were used with varying echo times (TE = 2.81, 5.71, 8.06, 10.46, 12.93, 15.4, 17.86, 29.78, 42.34, 54.9, 67.46 and 80 ms). Each GRE was a stacked 2D axial image with FA of 60 degrees, TR of 3200ms, in-plane resolution of 0.94 × 0.94 mm, slice thickness of 4 mm and FOV of 240 × 240 × 132 mm3.

Two five-minute runs of resting state blood oxygen level dependent (BOLD) imaging with TE of 27 ms, TR of 2500 ms, FOV 64 × 64 × 64 mm3, FA of 90 degrees, and voxel size 4 × 4 × 4 mm were collected while participants relaxed with their eyes closed. For each participant, the T1 and T2 imaging sequences were reviewed by a neuroradiologist, and imaging was excluded from further analyses if abnormal brain imaging findings were present.

Data pre-processing was done using SPM12 software (Wellcome Department of Cognitive Neurology, Institute of Neurology, London, UK) in conjunction with MATLAB version R2019b (Mathworks, Natick, MA, USA). T2* maps were smoothed with a 6 mm kernel and resampled to match the T1-weighted images. The T1-weighted images were used to normalize to Montreal Neurological Institute (MNI) space along with the co-registered T2* to yield 1 mm isotropic resolution maps. A 4 mm dilated mask was used to remove cerebrospinal fluid (CSF) contamination.

Functional images were motion corrected, realigned to the first image of the set, coregistered to the anatomical T1-weighted image, realigned to the average MNI template, and smoothed using an 8 mm FWHM Gaussian kernel using SPM12. Further post processing in AFNI 3dTproject included band pass filtering (0.01–0.1 Hz) after removal of nuisance signals from framewise displacement, cerebrospinal fluid signal, and linear drift. The first five frames were excluded to allow the signal to reach steady state. Frames with excess of 2 mm motion were also removed from the analysis. Region of interest to brain correlation maps were calculated using MATLAB. Regions of significant T2* difference between PTH and HC were used to select the regions of interest for the functional connectivity analysis. Left and right hemisphere clusters were interrogated separately. The Pearson correlation coefficients were calculated between the region of interest and the rest of the brain using in-house software written in MATLAB. The correlation maps were Fisher transformed to Z-scores for group comparisons.

The number of participants (N = 60 per group) was not chosen a-priori based on a power analysis. All available data was used for these analyses. T2* differences between 60 PTH participants and 60 age-matched HC were evaluated in MNI space using a t-test within SPM12 (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/) compensating for age and sex. Cluster analysis was performed in SPM 12 using uncorrected p < 0.005 and volume threshold of 90 mm3 (90 voxels) to interrogate T2* differences in PTH compared to HC. Cluster labelling was assisted with automated anatomical atlas ROI_MNI_V7.

The average T2* values from regions with significant T2* differences were included in a linear model to assess association with clinical variables within the PTH group. The clinical variables examined were number of lifetime mTBIs, time since most recent mTBI, the Sport Concussion Assessment Tool (SCAT) symptom severity checklist score, and headache frequency. Headache frequency was calculated as the percentage of days with headache since mTBI.

The regions with significant T2* differences between PTH and HC were used as seeds for a for full brain resting state functional connectivity correlation analysis. A full factorial analysis was conducted in SPM12 with two factors: group factor with two levels for PTH and HC and region of interest factor with the number of levels equal to the number of regions of interest used, as well as covariates to compensate for age and sex. Group differences for each region of interest were examined with a F-statistic with post hoc t-tests used to assess direction of the effect. Significance for group differences was set to p < 0.05 with family wise error correction for multiple comparisons.