Multi-Modal Signatures of Tau Pathology, Neuronal Fiber Integrity, and Functional Connectivity in Traumatic Brain Injury

Abstract

[¹⁸F]AV-1451 (aka ¹⁸F-Flortaucipir, [¹⁸F]T807) was developed for positron-emission tomography (PET) imaging of paired helical filaments of hyperphosphorylated tau, which are of interest in a range of neuropathologies, including traumatic brain injury (TBI). Magnetic resonance imaging (MRI) techniques like diffusion tensor imaging (DTI) and resting state functional connectivity assess structural and functional characteristics of the brain, complementing the molecular information that can be obtained by PET. The goal herein was to explore the utility of such multi-modal imaging in a case series based on a population of TBI subjects. This study probes the interrelationship between tau deposition, white matter integrity, and gray matter functional connectivity across the spectrum of TBI. Nineteen subjects (11 controls, five former contact sports athletes, one automotive accident, and two with military-related injury) underwent [¹⁸F]AV-1451 PET and magnetic resonance scanning procedures. [¹⁸F]AV-1451 distribution volume ratio (DVR) was estimated using the Logan method and the cerebellum as a reference region. Diffusion tractography images and fractional anisotropy (FA) images were generated using diffusion toolkit and FSL. Resting-state functional MRI (fMRI) analysis was based on a graph theory metric, namely weighted degree centrality. TBI subjects showed greater heterogeneity in [¹⁸F]AV-1451 DVR when compared with control subjects. In a subset of TBI subjects, areas with high [¹⁸F]AV-1451 binding corresponded with increased FA and diminished white matter tract density in DTI. Functional MRI results exhibited an increase in functional connectivity, particularly among local connections, in the areas where tau aggregates were more prevalent. In a case series of a diverse group of TBI subjects, brain regions with elevated tau burden exhibited increased functional connectivity as well as decreased white matter integrity. These findings portray molecular, microstructural, and functional corollaries of TBI that spatially coincide and can be measured in the living human brain using noninvasive neuroimaging techniques.

Introduction

Traumatic brain injury (TBI) is a highly prevalent, heterogeneous condition that results in significant morbidity and acute and chronic neuropsychiatric symptoms. Head trauma can produce focal brain damage and diffuse axonal injury,^{1, 2} yet the pathophysiology of TBI is complex and remains poorly understood. Multi-modal imaging of TBI specific pathology may help address some of the questions around how the brain changes after injury. The goal of this work was to examine how the brain changes following injury by utilizing magnetic resonance imaging (MRI) to examine structural and functional related changes and positron-emission tomography (PET) imaging of tau protein aggregates to examine resulting damage to axons.

Severe traumatic brain injury has been shown to elevate amyloid deposits in the central nervous system, increasing the risk for neurodegenerative conditions including Alzheimer's disease, while repetitive concussive events, like the ones incurred by athletes engaged in high impact sports or military personnel, increase the risk of chronic traumatic encephalopathy (CTE). Neurodegeneration after TBI leading to CTE is characterized by accumulation of varying amounts of amyloid-beta plaques and neurofibrillary tangles composed of hyperphosphorylated tau.^2,3 A wide array of PET radiotracer have been developed to image pathological tau in vivo. In particular, [¹⁸F]AV-1451 has demonstrated high affinity and selectivity for tau over amyloid.^4,5 It should be noted that upon initiation of this study, [¹⁸F]AV-1451 was the best tau radiotracer for PET imaging available, but it has been shown since then to exhibit off target binding which may confound the interpretation of uptake.^6–10 Despite potential challenges surrounding off-target binding of [¹⁸F]AV-1451, for individuals with a history of TBI, tau PET imaging has the potential to advance our understanding of disease mechanisms and provide early stage diagnostic and prognostic biomarkers.

In addition to PET imaging, other multi-modal structural MRI techniques—particularly diffusion imaging—have also contributed to our understanding of the pathophysiology of TBI by allowing in vivo assessment of white matter tracts in humans. This technique has predominantly been used in two ways: First, tractography allows the trajectory of white matter tracts to be defined and second, standardized measures of diffusion properties (such as fractional anisotropy or mean diffusivity) can be used to assess white matter microstructural integrity at specific locations and thus compared with [¹⁸F]AV-1451 uptake. Further, in this study, we postulate that tau deposits in gray and white matter tissue after TBI generates functional changes in neural circuits of the human brain. Thus, we used a multi-modal imaging strategy to assess the relationship between tau deposits and functional network changes in TBI individuals, by integrating the information of PET and functional connectivity magnetic resonance images. As dysfunctional connectivity is expected to impact both short and long-range distributed functional networks, we used a graph theory approach able to specifically study local and distant voxel-wise connectivity changes across the entire brain functional connectome.

In this case series, we illustrate promising utility of multi-modal neuroimaging to delineate structural, functional, and molecular brain changes across the spectrum of TBI, seeking to reveal a link between clinical phenotype and underlying pathological substrates in the brain.

Methods

Subjects

A total of 19 subjects underwent [¹⁸F]AV-1451 PET and MRI scanning procedures approved by the Massachusetts General Hospital Institutional Review Board. Subject classification, sex, age, weight, and body mass index (BMI) are shown in Table 1. In summary, 11 control subjects with no self-reported history of TBI and eight subjects with a history of TBI were included in this study. TBI subjects were taken from a consecutive recruitment of a convenience sampling from a brain injury program clinic that met criteria. In general, attempts were made to age- and sex-match control subjects to TBI, but, due to quicker recruitment of control subjects, this was difficult. Background information including neurocognitive data from clinical assessment specific to the TBI subjects is shown in Table 2. In short, the TBI group consisted of four former American-style football players, one former rugby player with diagnosed amyotrophic lateral sclerosis (ALS), one subject who sustained a clinically severe injury in an automotive accident approximately 2.5 years prior to participation in this study, and two subjects with military-related injury.

Table 1.

Subject Information

Subject	Sex	Age	Mass (kg)	BMI	Race	MMSE
CNTL1	M	33	73.5	21.9	Asian	NA
CNTL2	M	42	85.3	30.3	NA	NA
CNTL3	F	26	64.4	24.3	Caucasian	29
CNTL4	F	46	76.2	26.3	African American	29
CNTL5	M	67	93.9	33.4	Caucasian	28
CNTL6	M	61	68.0	22.8	NA	30
CNTL7	M	64	93.0	29.4	Caucasian	29
CNTL8	M	29	74.4	22.2	African American	28
CNTL9	M	47	136.1	38.5	Caucasian	30
CNTL10	M	52	78.9	28.0	Caucasian	30
CNTL11	M	53	69.8	22.0	Caucasian	29
TBI1	M	4th decade	102.0	30.5	NA	30
TBI2	M	7th decade	90.7	27.8	NA	NA
TBI3	M	6th decade	127.0	34.0	Caucasian	NA
TBI4	M	6th decade	105.2	28.2	African American	29
TBI5	M	5th decade	81.6	25.8	NA	NA
TBI6	M	34	68.0	24.9	Caucasian	27
TBI7	M	41	76.2	27.1	Caucasian	28
TBI8	M	31	123.4	39.0	Caucasian	29

BMI, body mass index; MMSE, Mini-Mental State Examination; M, male; NA, not available; F, female; TBI, traumatic brain injury.

Table 2.

Background Information on TBI Subjects

Subject	Injury cause	Summary of symptoms
TBI1	American style football	Repetitive concussion with no reported amnesia
TBI2	American style football	Dozens of concussions with >5 resulting in loss of consciousness; experienced retrograde and posttraumatic amnesia; neurocognitive testing noted elemental memory, verbal fluency, attention deficits and anxiety
TBI3	American style football	Repetitive concussions over a decade of play
TBI4	American style football	Repetitive concussions over a long career of play; experienced numerous episodes of confusion, one blackout in vision, but no true loss of consciousness; Neurocognitive testing noted hypomania, mild depression, relative weakness in attention, and impairment in strategic learning and memory in the context of inefficient processing
TBI5	Rugby/ALS	Repetitive concussion through a decade of playing rugby; later developed amyotrophic lateral sclerosis (ALS)
TBI6	Automotive accident	Loss of consciousness and prolonged posttraumatic amnesia; quadriparesis, confusion regarding past events, poor memory and impaired executive functioning
TBI7	Military-related	Several concussions that occurred over approximately 20 years prior to the imaging session caused by years of service in the armed forces; no amnesia or loss of consciousness reported; neurocognitive testing noted reduced concentration and irritability
TBI8	Military blast	Multiple blast concussions; within 10 feet of a mortar explosion; sustained other blast injuries while in close proximity to improvised explosive devices; no amnesia or loss of consciousness; neurocognitive symptoms included sleep dysfunction, headaches, anxiety, and concentration-related issues

TBI, traumatic brain injury.

Image acquisition procedures

PET imaging

[¹⁸F]AV-1451 was synthesized according to our previously published simplified methods.¹¹ Briefly, tBOC protected precursor dissolved in 1.5 mL Dimethyl sulfoxide was reacted with [¹⁸F]fluoride at 130°C for 10 min. Following initial purification using hydrophilic–lipophilic balance (HLB) solid phase extraction, product was injected onto reverse phase high performance liquid chromatography for further purification. The product peak was captured and loaded onto a second HLB cartridge, rinsed with 10 mL H₂O, eluted with 1 mL United States Pharmacopeia (USP) ethyl alcohol (EtOH), and formulated with 10 mL 0.9% USP sodium chloride. Final product had high specific activities (95–280 GBq/μmol).

PET image data was acquired using an ECAT EXACT HR+ scanner. Prior to collection of emission data, a 6-min transmission scan was acquired using Ge-68 rod sources for purposes of attenuation correction. Three-dimensional (3D) PET emission data collection was initiated with the bolus injection of 187 ± 7 MBq (SA: 90 ± 37 MBq/nmol) of [¹⁸F]AV-1451 and was acquired for 120 min with a 15-min break from 60–75 min. Pre-break data (0–60 min post-injection) was binned into time frames of 6 × 10 sec, 8 × 15 sec, 6 × 30 sec, 8 × 1 min, 8 × 2 min, and 8 × 5 min; post-break data (75–120 min) was binned into 7 × 5 min frames. Dynamic PET sinograms were reconstructed using filtered back projection while applying corrections for scatter, attenuation, deadtime, random coincident events, and detector normalization. Final reconstructed images had a grid size of 128 × 128 × 63 and voxel dimensions of 2.06 × 2.06 × 2.42 mm.

MRI

Magnetic resonance images were acquired using a Siemens 3 Tesla Trio Tim scanner with a 32-channel RF head coil.

Anatomical

A high-resolution anatomical MRI was acquired using a 3D Magnetization Prepared Rapid Acquisition Gradient Echo (MPRAGE) sequence with the following parameters: repetition time (TR) = 2.10 msec, echo time (TE) = 2.89 msec, flip angle = 12°, inversion time = 1100 msec; acquisition matrix size = 256 × 256 in-plane, 206 slices; voxel size 1 mm isotropic.

DTI

Diffusion tensor magnetic resonance data was acquired using a multi-slice diffusion-weighted balanced spin echo planar pulse sequence with 60 diffusion directions. The imaging parameters were TR/TE 8800/84 msec, b = 700 sec mm⁻², pixel bandwidth = 1457 Hz, flip angle = 90°, 72 slices, acquisition matrix = 104 × 104 with isotropic spatial resolution of 2.5 mm.

Resting-state fMRI

Changes in blood oxygenation level dependent (BOLD) T2* signals were measured using an interleaved gradient-echo echo planar imaging sequence. The subjects lay quietly for 5 min, during which 90 whole–brain volumes were obtained with the following parameters: TR = 3 sec, TE = 30 msec; flip angle = 90°; 64 × 64 pixel matrix; and 3 × 3 × 3.75 mm voxel dimensions.

Image processing and analysis procedures

PET/[¹⁸F]AV-1451

Reconstructed dynamic PET images were rigidly realigned to a common orientation to reduce head movement effects followed by co-registration of the post break PET session to the pre break PET session. Next, the realigned images were summed from 0–10 min, representing a diffuse distribution of radiotracer, and were then rigidly aligned to the MPRAGE image. PET images were co-registered and warped to the Montreal Neurological Institute (MNI) reference space using the transformation applied to the MPRAGE as described below. Final co-registrations and transformations to MNI space were applied back to the original dynamic PET images to reduce interpolation errors. Regions of interest (ROIs) were delineated using standard atlases available through FSL including (MNI structural atlas, Harvard-Oxford cortical and subcortical structural atlas, and the probabilistic cerebellar atlas).¹² ROI masks were transformed to PET native space for each individual subject to allow extraction of time activity curves (TACs) for analysis of radiotracer uptake throughout the course of scanning.

[¹⁸F]AV-1451 uptake was quantified using the distribution volume ratio (DVR) calculated using the Logan method¹³ with a 55-min t* and non-vermis cerebellar gray matter available through FSL as suggested in our recently published pharmacokinetic analysis of [¹⁸F]AV-1451.¹⁴ Logan DVR estimates were performed on ROI-based TACs and dynamic PET images for the creation of voxelwise [¹⁸F]AV-1451 DVR maps. ROI based estimates of DVR and max DVR were compared in regions potentially affected by head injury including of frontal lobe, occipital lobe, parietal lobe, cerebral white matter, cerebral cortex, thalamus, caudate, putamen, medial temporal cortex (MTC), and nucleus accumbens.

MRI/DTI

MPRAGE images were transformed into MNI space using sequential linear affine and nonlinear warping methods available through the FSL (Functional MRI of the Brain Software Library)¹² software package. Diffusion-Toolkit/TrackVis was used for fiber tracking reconstruction, display, and analysis of diffusion data.¹⁵ Final tractography images were aligned to a common space using affinemethods based upon the subjects MEMPRAGE image allowing inter-subject comparisons of tractography images. FSL was used to generate FA images which were then transformed to MNI space.¹²

White matter tracts were quantified using two methods. First, tractography images were visually compared using TrackVis to make inter-subject comparisons and bilateral comparisons in the same subject when appropriate. Second, TrackVis statistics tools were used to count the number of tracts penetrating ROIs guided by [¹⁸F]AV-1451 uptake. When inter-subject comparisons were made using TrackVis, all visualization and statistics parameters were identical between the subjects.

MRI/fMRI

Pre-processing

The resting-state fMRI data were preprocessed using the FSL (FMRIB Software Library v5.0). The first volume for each subject was discarded due to magnetic saturation effect. Next slice-time correction was applied for temporal alignment. Motion correction was applied and Friston 24 parameters were computed.¹⁶ The frame-wise displacement (FD), which represents the scalar quantity of instantaneous head motion of each volume relative to its earlier neighboring volume, was also calculated based on the head motion parameters.¹² Because several recent studies have shown that high levels of head motion can significantly influence the estimation of resting-state FD,^17–19 the following steps were adopted to reduce these motion effects. First, the fMRI data were excluded from further analysis if the maximum displacement in one or more of the orthogonal directions was >1 mm or a maximum rotation >1.0° and second, the data were also excluded if the average FD of the subject exceeded 0.5 mm. After realignment, the individual structural images were linearly registered to the mean functional images.

Next, the structural images were segmented into gray matter, white matter, and cerebrospinal fluid (CSF) and projected to the mean BOLD volume. The nuisance variables, including 24 head motion parameters and the average BOLD signals of the CSF and white matter, were linearly regressed out from the data. A band-pass filter was applied between 0.01–0.08 Hz.²⁰ The functional volumes were then spatially normalized to the MNI space and resampled to 3 × 3 × 3 mm³ voxels using the normalization parameters for their respective structural images. To make group comparisons all voxels were spatially smoothed with a 6 mm full width at half maximum isotropic Gaussian kernel. Finally spline interpolation for the scrubbing was applied to all the time-points that presented a FD greater than 0.5 mm. Subjects with less than 80% of the time-points with FD values lower than 0.05 mm were not included for further analysis.

Post-processing

Local and distant connectivity

Weighted degree graph theory analysis was used to compute the local and distant connectivity. Functional MRI images were down-sampled to 8 mm, then correlation matrices were computed at the voxel level, resulting in connectivity matrices of 5138 × 5138. Only positive correlations were retained due to ambiguous interpretations of negative correlations^21,22 and to prevent further misinterpretations in the weighted degree connectivity values. R to z Fisher transformation was applied. We calculated the local and distant weighted degree centrality value for each voxel of the brain based on an Euclidean distance optimal segregation previously described.²³ The weighted degree measures the sum of the z values for all the links in a node, with local defined as within a 16 mm radius of each node and distant everything outside that radius.²³

Relationship between [¹⁸F]AV-1451 binding and fMRI connectivity

[¹⁸F]AV-1451 images were transformed to the individual T1 space by rigid alignment with a mutual information algorithm and the T1 was projected to the MNI standard space. To be able to work in the same resolution of 8 mm, a down-sampling of the images was performed. Due to the heterogeneity of the TBI cohort we implemented individual analysis for this study. In order to study the individual distribution of [¹⁸F]AV-1451 in each participant, we transformed their DVR intensities into z-score values and investigate their relationship with local and distant connectivity as follows:

a) At the subject level comparing to controls

Functional connectivity scores (local and distant) associated with [¹⁸F]AV-1451 uptake were calculated by computing the average degree connectivity within regions in which [¹⁸F]AV-1451 z-score were greater than 1.96SD. Therefore, we focus our investigation on specific regions of presumably abnormal tau depositions, both in TBI individuals (TBI-related tau binding) and adult controls (non-TBI-related tau binding).

b) At the subject level compared with [¹⁸F]AV-1451 binding

To investigate the spatial relationship between functional connectivity scores and [¹⁸F]AV-1451 uptake, we extract the mean weighted degree of local and distant connectivity in regions of interest, which were preciously selected based on the z-score values of [¹⁸F]AV-1451 images. We used specific intervals of the z-score values to capture all possible types of tau-related lesions across the cortex. Thus, mean local and distant weighted degree was displayed for each of the tau-related z-score intervals: <1, 1.0–1.2, 1.2–1.4, 1.4–1.6, 1.6–1.8, 1.8–2.0, > = -2. Of note, as data points in this approach are not independent between each other and our interest was to investigate their spatial relationships, we reported the curvilinear fits as descriptive findings.

Results

Reference region/nondisplaceable analysis

Figure 1 shows the mean ± standard deviation standardized uptake value time course of [¹⁸F]AV-1451 in the non-vermis cerebellum for control and TBI subjects which were grouped together only for this comparison. No statistically significant difference was observed between the two groups.

FIG. 1.

Comparison of mean ± standard deviation time activity curves for control (red) and traumatic brain injury (blue) subjects in the cerebellum. No significant differences in cerebellar time courses were observed. The cerebellum was used as a reference region in this work for estimation of distribution volume ratio. Although subjects were considered individually for the purposes of this work due to their heterogeneity, they were grouped for this comparison. Color image is available online.

[¹⁸F]AV-1451 PET distribution volume ratio

Regional maximum Logan DVR estimates are shown in Figure 2. High heterogeneity of maximum DVR was found within the TBI subjects as coefficient of variation (CV) for max DVR was greater than 10% in the parietal lobe, white matter, cortex, thalamus, caudate, putamen, MTC, and accumbens for the TBI subjects and was greater than 10% only in the white matter and MTC for the control subjects. Highest DVR was found in the acute injury subject of 2.0.

FIG. 2.

Regional maximum distribution volume ratio (DVR) comparison for all subjects. Shown are maximum DVR in the regions of the frontal, occipital, and parietal lobes, white matter, cortex, thalamus, caudate, putamen, medial temporal cortex (MTC), and accumbens. Control subjects are shown in red and TBI subjects are shown in blue. Color image is available online.

Figure 3 shows five axial slices of DVR for all 19 subjects and equivalent SUVR images are shown in Supplementary Figure S1. Heterogeneity in distribution of [¹⁸F]AV-1451 was generally high in both control (CNTL) subjects and TBI subjects. CNTL1 demonstrated high uptake in the subcortical regions. CNTL5, a 67-year-old male, exhibited high DVR in the subcortical areas including the striatum as well as the thalamus and hippocampus. Some cortical areas also exhibited high uptake particularly in the anterior cortices in this subject. CNTL6, a 61-year-old male, demonstrated high DVR in the hippocampus and thalamus as well as the posterior cortical regions. CNTL 7 showed high uptake in the striatum and thalamus. CNTL11 showed high uptake in subcortical regions including the striatum, thalamus, and hippocampus. The TBI subjects showed very high heterogeneity in [¹⁸F]AV-1451 DVR.TBI1, a former American style football player in the 4th decade of life, demonstrated elevated [¹⁸F]AV-1451 distribution in the striatum, thalamus, hippocampus, and a focal region in the left anterior cortex near the gray-white matter interface of the precentral gyrus. TBI2, a former American style football player in the 7th decade of life, exhibited high subcortical [¹⁸F]AV-1451 uptake and diffuse cortical binding.

FIG. 3.

Axial slices of distribution volume ratio (DVR) thresholded from 0–2 for all subjects. Also listed is the subject information including sex, age, and injury or sport, if applicable. Color image is available online.

TBI3, a former American football player in the 6th decade of life, showed high subcortical binding in the striatum, thalamus, and hippocampus. This subject also displayed high [¹⁸F]AV-1451 binding in the left occipital lobe. TBI4, a former American style football player in the 6th decade of life, demonstrated low overall uptake of [¹⁸F]AV-1451 with exception to the choroid plexus. TBI5, a former rugby player in the 5th decade of life with diagnosed amyotrophic lateral sclerosis, demonstrated overall low [¹⁸F]AV-1451, however, this subject showed high levels of uptake in the posterior cingulate (better illustrated in Fig. 4). TBI6, a 34-year-old male who had been in a car accident approximately 2.5 years prior to the imaging study, demonstrated the highest binding of all subjects with focal uptake in the corpus callosum protruding into the posterior cingulate. Additionally, high uptake was found in the thalamus and brainstem. The final subject, TBI8, a 31 year old male who suffered military blast exposure – demonstrated globally low [¹⁸F]AV-1451 uptake with no evident focal enhancements in binding.

FIG. 4.

Comparison of fractional anisotropy (FA) images between the acute head injury subject (A-TBI6, left column) and a matched control subject (B-CNTL1, right column). Highest [¹⁸F]AV-1451 uptake was found in the corpus callosum as shown in the thresholded DVR images (A-iii, A-iv). (i) Axial view of the FA. (ii) Coronal view of the FA highlighting the subject's corpus callosum revealing lower FA in the acute head injury subject. Color image is available online.

Diffusion imaging and relation to [¹⁸F]AV-1451 uptake

Figure 4 shows an examination of fractional anisotropy (FA) and [¹⁸F]AV-1451 uptake in TBI6 who was in an automotive accident approximately 2.5 years prior to PET and MRI and a matched control subject (CNTL1) for a comparison. TBI6 exhibited very high uptake of [¹⁸F]AV-1451 in the corpus callosum as shown in the thresholded DVR images at the bottom of Figure 4. FA images revealed low FA the corpus callosum that corresponded to the high [¹⁸F]AV-1451 DVR. Examination of tract density revealed the number of tracts extending from the corpus callosum was 813 for the acute TBI subject and 2419 for the matched control.

Figure 5 shows tractography and [¹⁸F]AV-1451 uptake in former athletes. For comparison, diffusion tensor images for all subjects are shown in Supplementary Figure S2. Figure 5A shows the neuroimaging results for TBI3 who played American football highlighting the lack of bilateral symmetry in the tractography images in the region of high [¹⁸F]AV-1451 uptake. Figure 5A.ii. exhibits the tracts extending from the high focal uptake region measured by PET (green ROI) and the contralateral side (blue ROI). The number of tracts extending from the ROI on the side with high [¹⁸F]AV-1451 uptake was 23 and 169 on the contralateral side. Figure 5B shows a bilateral symmetry comparison for TBI1, a former American style football player. Highest cortical [¹⁸F]AV-1451 uptake was found in the white-gray matter interface of the precentral gyrus on the left side. Tractography imaging demonstrated lower tract density in the left hemisphere corresponding to the elevated [¹⁸F]AV-1451 uptake. The number of tracts that penetrated the left gray matter of this subject was 1998 and 2645 penetrated the same region on the contralateral side.

FIG. 5.

Diffusion tensor imaging (DTI) and [¹⁸F]AV-1451 uptake in professional athletes. (A) Subject TBI3 demonstrated high [¹⁸F]AV-1451 uptake in the lateral occipital cortex extending into the inferior temporal cortex and demonstrated decreased tract density in this region when compared with the subject's contralateral side (i). Region of interest (ROI) analysis comparing symmetry of tracts penetrating the region demonstrating high [¹⁸F]AV-1451 uptake revealed decreased tract density (23) when compared with the contralateral ROI (169). (B) Subject TBI1 exhibited high [¹⁸F]AV-1451 uptake in the white–gray matter interface of the left precentral gyrus (iii, iv) and demonstrated decreased tract density in that region and throughout the left hemisphere (i, ii). (C) Subject TBI5 demonstrated high [¹⁸F]AV-1451 uptake in the posterior cingulate (i) and showed tract thinning in this same region (ii). The right panels display the local and distant functional connectivity weighted degree for the tramatic brain injury (TBI) subjects (red markers) compared with the control subjects (blue markers). Color image is available online.

Tractography and [¹⁸F]AV-1451 uptake for TBI5 is shown in Figure 5C. TBI5 was a former rugby player who later developed ALS and showed high uptake of [¹⁸F]AV-1451 in the posterior cingulate wrapping around to the precuneus as demonstrated in the PET image (Fig 5C-ii). Tractography imaging showed thinning of tracts in the region corresponding to elevated [¹⁸F]AV-1451 uptake. Of note, none of the most important structural or functional areas, previously described by different neuroimaging studies in ALS, such as the primary motor cortex, corticospinal tract, corpus callosum, frontal lobes, basal ganglia, thalamus, brainstem and spinal cord,²⁴ involve the posterior cingulate cortex.

Relationship between [¹⁸F]AV-1451 binding and fMRI connectivity

Figure 6 shows the relationship between weighted degree centrality of functional connectivity and ¹⁸F-AV1451 uptake in TBI subjects. It should be noted that subjects with less than 80% of the time-points with FD values lower than 0.05 mm were not included for further analysis. As a result, six controls and six TBI participants with optimal FD quality assessment were included in the functional connectivity analysis. As seen in the figures, we observed increasing number in local functional connections as [¹⁸F]AV-1451 uptake increases.

FIG. 6.

Relationship between [¹⁸F]AV-1451 binding and functional magnetic resonance imaging connectivity: Six different traumatic brain injury (TBI) subjects were evaluated. Mean local and distant weighted degree is shown for each of the intervals: <1, 1–1.2, 1.2–1.4, 1.4–1.6, 1.6–1.8, 1.8–2, >2. As data points in this approach are not independent between each other and our interest was to investigate their spatial relationships, we reported the curvilinear fits as descriptive findings. Color image is available online.

Discussion

TBI is a heterogeneous condition and the goal of this work was to explore the use of three potentially complimentary neuroimaging methods including [¹⁸F]AV-1451 PET for measurement of tau deposition, magnetic resonance diffusion imaging for determination of white matter damage, and fMRI for characterization of resting-state functional networks. This case series was performed in a group of subjects with a range of brain injury including those that experienced head trauma from repetitive sport activities, automobile accidents and military related blast exposures. Due to the heterogeneity of injury type, severity, and location, a groupwise analysis of this data set did not seem appropriate with the exception of reference region validation. We therefore presented these data as a case series.

[¹⁸F]AV-1451 was used in this work because at the time of this studies initiation, it was believed to be the best marker of 3R and 4R tau aggregates, which is believed to be a biomarker of interest in TBI, over other misfolded proteins such as amyloid. Due to the heterogeneity of injury, which presumably translates to variability in the location of tau pathology, we did not directly compare [¹⁸F]AV-145 uptake between groups in a voxelwise fashion. Figure 2 shows the maximum measured DVR in various cerebral regions, a presentation of the findings which highlights that focal [¹⁸F]AV-1451 uptake, which can be lost when DVR is averaged over an entire region. Subcortical regions exhibited high variability in maximum DVR in TBI subjects as well as control subjects. With a larger study population, perhaps TBI type or severity correlates may reveal the reason for this variability; however, it is most likely due to off target binding which will be discussed further.

Proper use of the distribution volume ratio requires a valid reference region or a region that lacks specific binding. We examined several areas including the pons and medulla, however, we found the cerebellum to have the lowest variation among controls and subjects with TBI history. Although the cerebellum was empirically deemed appropriate in this work, studies in humans and mice with blast related injury indicate the cerebellar changes may occur after injury including tau accumulation.²⁵ Larger in vivo and/or autoradiography studies are therefore needed that validate the use of the cerebellum in the different TBI types.

The use of in vivo measures of tau along with multi-modal structural and functional MRI techniques offers great promise in elucidating the pathophysiology of TBI related disorders and may aid diagnostic clarification with other neurologic and psychiatric disorders that share partially overlapping clinical features.^26–28 Many patients with traumatic brain injury, for example, are faced with persistent impairments in cognitive control and executive functions.²⁹ Neurodegenerative disorders including Alzheimer's disease (AD) and CTE can be long-term consequences of TBI, given that single and repetitive injuries predispose individuals to cognitive decline in later life. In TBI-related cognitive impairment, the integrity of tubulin associated unit (or tau) protein is vital to white matter structure, and when damaged, tau protein can abnormally hyperphosphorylate and oligomerize into insoluble filaments known as tau pathology.^30,31 The PET tracer [¹⁸F]AV-1451 strongly binds to tau lesions primarily made out of paired helical filaments in pathological brains, allowing in vivo clarification of a differential diagnosis with partially overlapping clinical features including AD, CTE or normal aging. Another potential application may relate to clarifying the difference between a prolonged post-concussive symptoms and post-traumatic stress disorder in military personal,³² given that both entities have overlapping symptomatology including concentration, memory and mood changes but have distinct underlying pathophysiologies. Our current sample of patients with TBI demonstrated an interesting spatial correlation between [¹⁸F]AV-1451 PET uptake, loss of DTI measured white matter integrity and an increase in fMRI local and distant functional connectivity, indicating a possible association between these three biomarkers.

As in our patient sample, the relationship between axonal injury and deposition of a PET neurotracer in TBI has been described previously, using Pittsburgh compound B ([11C]PiB), where Scott and colleagues demonstrated a relationship between axonal injury and amyloid accumulation, suggesting a direct relationship between [¹¹C]PiB binding and white matter damage in connected tracts.³³ [¹⁸F]AV-1451 imaging has been previously reported in case studies of two retired American football players.³⁴ Both subjects experienced declines in cognitive function and one exhibited imaging features suggestive of either CTE or progressive supranuclear palsy to support CTE based on binding patterns, a finding that demonstrates the utility of tau imaging to differentiate between different types of dementia. This distinction will be particularly necessary with the promise of future disease modifying therapeutics.²⁶

Several studies have combined fMRI with DTI in order to evaluate the association between axonal injury and aberrant network functional connectivity in TBI subjects,^35–37 with very different results.¹ (for a review read Sharp, 2014). Here, we report an increase in local and distant functional connectivity in the areas with the highest amount of tau deposition. Increases, decreases and no significant changes in functional connectivity have been observed in a number of different TBI studies, the vast majority in relation to particular network analyses.^35–39 The papers describing decreases in functional connectivity in TBI subjects compared with healthy controls, studied long-range connections at the network level, which is a different approach from the one we used here which focused on characterizing connectivity changes in brain areas with concurrent tau. Our results could be explained by the fact that large-scale white matter tracts do not tend to form new connections, but brain plasticity can produce new synapsis at the microscopic level.^40,41 Shumskaya and colleagues³⁹ suggested that abnormal increased connectivity might reflect compensatory processes attempting to enhance the function of damaged brain areas or may relate to aspects of the physical post-concussive symptoms, such as headache and increased sensitivity to noise/light.³⁹

Several limitations of this study should be noted. Quantification of [¹⁸F]AV-1451 PET data relied on the use of the cerebellum as a reference tissue devoid of radiotracer binding. This has generally been accepted as valid for tau imaging studies in Alzheimer's disease and many other tauopathies because it is fairly well established that tau burden in the cerebellum is negligible; however, the nature of head trauma is such that the cerebellum cannot be assumed spared of injury in all cases so postmortem validation of this assumption in a generous population of TBI subjects is warranted. Although BMI is noted in the manuscript, it should be recognized that a similar BMI in a former athlete and a control subject might be meaningless due to different body types. Recent studies also have called into question the binding specificity of existing tau tracers including [¹⁸F]AV-1451, so it is imperative to also empirically exclude the possibility that radiotracer retention observed in vivo reflects a binding substrate other than tau.

Studies have shown that [¹⁸F]AV-1451 binds to MAO-A and MAO-B; however, another study has shown MAO-B inhibitors are incapable of blocking [¹⁸F]AV-1451 uptake.^6,7,42 Additionally, others have shown off-target uptake in iron-associated regions as well as uptake correlation with age-related increases in iron deposits.^8,9 Taken together, this results in an inability to interpret the differences in binding in the subcortical regions of the brain. For completeness, however, the regional maximum DVR for thalamus, caudate, and putamen are included in this work, but we do not believe that increased [¹⁸F]AV-1451 uptake in these regions reflects tau deposition. Further, [¹⁸F]AV-1451 binding in postmortem brain tissue has been found in regions absent of tau deposits such as the choroid plexus potentially explaining the binding we observed in TBI4. Overall, it is clear [¹⁸F]AV-1451binds with high affinity to tau, but it is also clear that off target binding to MAO-A, neuromelanin, iron, and other potential targets are problematic and needs to be considered when interpreting [¹⁸F]AV-1451binding.^10,43 All of these considerations converge of the need for a thorough postmortem characterization of tau radiotracer binding in tissue samples from subjects exhibiting a wide range of brain injury types and severity. Lastly, these subjects were not tested for APOE status and did not undergo an amyloid PET scan which may have aided in interpretation of the [¹⁸F]FAV-1451 scan.

In summary, in this manuscript we report a novel combination of advanced multi-modal imaging techniques including [¹⁸F]AV-1451 PET, DTI, and fMRI in a heterogeneous sample of TBI pathologies. Results suggest that disruptions in white matter integrity and functional connectivity are spatially congruent with tau pathology. These findings demonstrate that synergistic neurobiological information may be gained by interpreting the neuroimaging modalities in unison rather than in isolation. Given that TBI affects 1.5 million people in the United States each year, future research efforts should be aimed at leveraging the multi-modal approach illustrated here to advance our understanding of the full spectrum of TBI to ultimately personalize diagnostic and therapeutic efforts.

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S2