The Relationship Between Extracellular Free-Water and Gray Matter Volume in
Retired Professional Football Players With History of Mild Repetitive Head Injuries
The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters
Citation Pezzuto, Justin J. 2017. The Relationship Between Extracellular Free-Water and Gray Matter Volume in Retired Professional Football Players With History of Mild Repetitive Head Injuries. Master's thesis, Harvard Extension School.
Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:33825869
Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA
Justin J. Pezzuto
A Thesis in the Field of Biology
for the Degree of Masters of Liberal Arts in Extension Studies
Harvard University
May 2017
Abstract
The goal of this work was to investigate whether there was a relationship present between
extracellular free water and gray matter volumes in the brains of retired professional football
players. A study done by Maier-Hein et al. was able to establish the relationship in a population
of participants that presented clinical symptoms of Alzheimer’s Disease. Due to the multiple
similarities between Chronic Traumatic Encephalopathy (CTE) and Alzheimer’s Disease, the
relationship mentioned above was examined in this study using a population of retired
professional football players at high risk for CTE. Changes in the white matter microstructure as
well as general volumetric changes were observed by using MRI techniques. We did not find a
correlation between gray matter volume and extracellular water in the players, suggesting that
despite the similarities between CTE and Alzheimer’s Disease there are also distinct differences
between the diseases. Multiple brain regions were found to be different between control and
player groups providing evidence for brain alterations in retired football players. The regions that
were statistically different in CTE compared with controls were then correlated against one
another, and it was determined that multiple regions within the brain functioned as a network in
its alterations. Lastly, it was shown that diffusivities of the white matter tracts correlated with
volumes of the gray matter suggesting an interaction between white a gray matter in the brains of
retired football players.
iv
Acknowledgments
Many thanks to all who assisted me at the Psychiatry Neuroimaging Lab (PNL) in association
with and Women’s Hospital, Boston. More specifically to Dr. Ofer Pasternak and Pawel Wrobel
for supporting me to understand the ways in which the PNL functioned, and were patient enough
to deal with my ongoing trials and tribulations. Without them this process would have been
much more difficult.
To my Meme, Wanda Pezzuto. She struggled with Alzheimer’s Disease throughout her life. As a
young boy, I aspired to conduct research and explore the inner workings of Alzheimer’s Disease
as well as other neurodegenerative diseases. I am beyond excited that this thesis process allowed
me to fulfill the promise I made to her all those years ago.
To my family, who have motivated and encouraged me every step of this thesis development
process, as well as my pursuit for my degree.
v
Chapter I: Introduction ........................................................................................................................................... 1
Intention of Project .................................................................................................................................... 1
Chronic Traumatic Encephalopathy: Overview and History ........................................................ 3
Chronic Traumatic Encephalopathy Pathology ................................................................................ 5
Gross Pathology ........................................................................................................................... 5
Microscopic Pathology .............................................................................................................. 6
Comparison of Chronic Traumatic Encephalopathy and Alzheimer’s Disease ...................... 7
Taupathy ......................................................................................................................................... 8
Beta-Amyloid ............................................................................................................................... 8
TDP-43 ........................................................................................................................................... 9
Magnetic Resonance Imaging (MRI) ................................................................................. 11
Structural MRI ........................................................................................................................... 12
Diffusion MRI ........................................................................................................................... 13
Participants ................................................................................................................................................ 17
Diffusion Image Post-Processing ........................................................................................ 19
Structural Image Post-Processing ........................................................................................ 19
Diffusion Measures ................................................................................................................................ 21
Statistical Analyses ................................................................................................................................. 21
Group Comparisons of Volumes ......................................................................................... 23
Group Comparisons of Free Water ..................................................................................... 23
Correlations between Regions of Interest ......................................................................... 24
Correlations between Volumes and Diffusivity .............................................................. 24
Chapter III: Results .............................................................................................................................................. 25
Descriptive Statistics .............................................................................................................................. 25
Relationship between amount of Free Water and Total Gray Matter ..................................... 25
Region of Interest Analyses ................................................................................................................. 26
Associations between Regions of Interest ....................................................................................... 29
Relationships between Volumes and White Matter Microstructure ........................................ 33
vii
List of Tables
Table 1. Normalized Volumes. This table presents the group comparisons of the normalized
volumetric values produced by FreeSurfer segmentation. ........................................ 27
Table 2. Free-Water. This table presents the group comparisons of the Free-Water diffusivity
values located on the ENIGMA white matter skeleton ............................................. 28
Table 3. Free-Water. This table presents the group comparisons of the ADt diffusivity values
located on the ENIGMA white matter skeleton ........................................................ 28
Table 4. ENIGMA Skeleton Label Abbreviations. The following table presents the abbreviations
of regions of interest that comprise the ENIGMA white matter skeleton. ................ 31
Table 5. FAt (FreeSurfer Labels). This table presents the group comparisons of the FAt
diffusivity values located on the ENIGMA white matter skeleton ........................... 41
Table 6. Free Water (FreeSurfer Labels) This table presents the group comparisons of the FAt
diffusivity values located on the ENIGMA white matter skeleton. .......................... 42
ix
Figure 1. Gross Pathology of Chronic Traumatic Encephalopathy (CTE). .................................... 6
Figure 2. Standardized Core Processing Pipeline. ........................................................................ 18
Figure 3. Free Water values correlated with the Total Gray Matter Volume). ............................. 26
Figure 4. Normalized Volume (FreeSurfer Labels) Correlations.. .............................................. 30
Figure 5. ADt Correlations [ENIGMA Labels]. ........................................................................... 32
Figure 6. Free Water Correlations [ENIGMA Labels].. ............................................................... 32
Figure 7 Normalized Volumes correlated with ADt values......................................................... 35
Figure 8. Normalized Volumes correlated with Free Water. ........................................................ 36
Figure 9 FAt Correlations [FreeSurfer Labels] ........................................................................... 42
Figure 10 Free Water Correlations [FreeSurfer Labels]. ............................................................. 43
Chapter I
Intention of Project
Two subjects that have always been at the forefront of our society have been health and
sports. Their sections can be found neighbored in every newspaper, and stories can be viewed
repeatedly in every newscast. Recently, the prevalence of concussions and the development of a
condition known as Chronic Traumatic Encephalopathy (CTE) in the world of sports has found a
way in marrying the two topics. CTE and other neurodegenerative diseases such as Alzheimer’s
Disease share a lot of characteristics in common, but are distinctly different in nature. In a study
by Maier et al., it was determined that there is a relationship between the total amount of gray
matter and the amount of extracellular water in the brains of subjects diagnosed with
Alzheimer’s Disease. The aim of my study is to see if these results are reproducible using a
population of retired NFL football players.
Mild Traumatic Brain Injuries
Increased focus has been given to the prevalence of mild traumatic brain injuries (mTBI),
better known as concussions in sports. Concussions, along with other mTBIs occur frequently in
sports. As of 2009, approximately 1.6 to 3.8 million sports related concussions occur in the
United States every year (McKee et al., 2009). However, that statistic cannot be considered a
true figure due to the amount of concussions that are not reported by the athlete, nor detected by
coaching and training staff. A concussion occurs when a sudden force inflicted on an individual
causes the brain to quickly shift its position in the skull and violently collide with the cranium.
2
This causes damage to the tissue and its vascularity. On a cellular level this high-intensity impact
cause abnormalities with the subject’s neural cytoskeletons as well as their metabolism,
including ionic shifts (Stern et al., 2011). Nausea, vomiting, irritability, and headaches are all
symptoms of someone who has suffered a mTBI (Koerte et al., 2015). Additional symptoms that
can be present during a concussion include loss of consciousness that does not exceed 30
minutes, post-traumatic amnesia that also lasts less than a half hour, and extreme light or sound
sensitivity. (Young, Hobbs, & Bailes, 2016). Diagnosing and grading the severity of a
concussion seems to be more broad and unclear. One method used by physicians to diagnose
concussions and rank their severity is known as the Glasgow Coma Scale (GCS). This scale
utilizes neurological test results that focus on the individual’s eye reflexes, verbal and motor
abilities. The higher scores associated with those tests indicate the better functioning of the
individual (Young et al., 2016).
The dangers associated with mTBI have evolved both the rules and procedures that
govern various sports, as well as technologies used in equipment to better protect the participant.
Athletes, particularly those that participate in high contact sports such as football, boxing, ice
hockey, etc. have a noticeably increased probability of experiencing multiple mTBI’s over the
course of their careers (B. Omalu, 2014). While most athletes who experience a head injury
while playing a sport is minor, recovery time ranges from a few days to weeks. However, a
smaller group of individuals develop prolonged symptoms. Evidence also suggests that 17% of
athletes who experience multiple concussions or events of head trauma are likely to develop a
progressive neurodegenerative disease known as Chronic Traumatic Encephalopathy (CTE)
(McKee et al., 2009).
Chronic Traumatic Encephalopathy (CTE), unlike a concussion, is a permanent condition
that begins around midlife and progresses throughout the life of the affected individual. CTE ,
unlike a concussion or mTBI, is considered to be a progressive neurodegenerative disease
meaning it worsens over the course of the victims life and dismantles their nervous system (B.
Omalu, 2014).
The symptoms of Chronic Traumatic Encephalopathy were given various titles over the
years before Bowman and Blau were the first to use the term “chronic traumatic encephalopathy”
in their study of boxer whom developed behavioral and cognitive changes (Bowman & Blau,
1940). A physician named Harrison Maitland first described the symptoms viewed in boxers
with repetitive brain trauma as “punch drunk” (Martland, 1928). It had also historically been
termed as “dementia pugilistica” (Millspaugh, 1937). Both of these terms reflected back on the
association the condition shared with sports that required physical contract such as boxing,
football, hockey, wrestling etc. According to McKee et. al. of the 51 neuropathologically
confirmed cases of CTE, 46 (90%) occurred in athletes (McKee et al., 2009). Although heavily
associated with athletics other subpopulations that have been linked to develop CTE include
epileptics, as well as individuals with developmental disabilities who engage in head banging
behavior (Koerte et al., 2015). Military personnel have shown reports of CTE symptoms due to
their experiences with repeated head trauma due to physical combat or extreme proximity to
explosives (Goldstein et al., 2012).
The onset of CTE symptoms typically start to manifest in the subject between 8-10 years
after the repetitive head trauma. Deteriorations in cognitive skills such as, attention,
concentration, and memory present first in the preliminary stage of the disease (McKee et al.,
4
2009). Following this introductory period of the disease there are three additional stages of
symptoms that continue. The first stage contains psychoses, as well affective disturbances. This
group of symptoms is followed by social instability, erratic behavior, and the initial symptoms
observed in Parkinson’s Disease occupy the second stage. Dementia, complete parkinsonism,
abnormal speech and gait characteristics are symptoms included in the third and final stage
(McKee et al., 2009). Additional symptoms that have been reported throughout the progression
of CTE include dysarthria, dysphagia, ocular abnormalities, However, the point when CTE
initially develops and its severity varies from case to case. Variables that can influence CTE’s
onset and severity include but are not limited to the subject’s sport, position, the age of first
mTBI, and the duration of exposure (Gavett, Stern, & McKee, 2011). Other variables that may
have an effect on the development of CTE include age, gender, and genetic predisposition (Stern
et al., 2011). Like many other forms of dementia, CTE is diagnosed based on prevalence of
symptoms and is unable to be detected through a neuroimaging method such as an MRI or a
CAT scan. The symptoms associated with CTE may start to manifest after multiple years or even
decades after brain trauma was experienced (Koerte et al., 2015).
Besides from repetitive head trauma, there are no supplementary known indicators to
spawn the formation of CTE. It has been shown that subjects who experience head trauma at a
younger age are also more susceptible in developing neurodegenerative effects (Koerte et al.,
2015). Repetitive head trauma has been believed to have developed a neurodegenerative cascade
by the recurring distortions of axons whether by current concussions or ones that have occurred
in the past (Stern et al., 2011).
5
Chronic Traumatic Encephalopathy Pathology
CTE is currently diagnosed post-mortem by the identification of tau pathology without
amyloid beta pathology (McKee et al., 2009). The first symptoms of CTE that are exhibited
include a decline in concentration and attention followed by complaints of memory loss and
headache (Saulle & Greenwald, 2012). Also during this initial stage of CTE witnesses close to
the individual report a prevalence of mood swings and irritability (B. Omalu, 2014). Emotional
distress can be so extreme to eventually lead to long-term depression and higher risk of suicide
(Saulle & Greenwald, 2012). The severity of the disorder is related to the amount of years
playing the sport and the number of traumatic head injuries that were sustained during that
experience.
Gross Pathology
The brain of someone who was diagnosed due to CTE symptoms may appear very similar
to a normal looking brain. This means that there is no visual evidence of necrosis, hemorrhaging,
or significant atrophy of the brain’s cortex (B. Omalu, 2014). This would suggest that the
volumes of brain may appear similar to a normal individual of that age, even when analyzed with
MRI. However, the observance of a decrease in brain volume has been associated with advanced
cases of CTE. Locations as to where this volumetric reduction occurs have been observed in the
frontal and medial temporal lobes, enlargement and dilation of the ventricles, the cavum septum
pellicidum, and septal fenestrations (Stern et al., 2011). Additionally, there have been reports of
a thinning of the bottom of the hypothalamus, reduction of the mammillary bodies, and
significant volumetric decreases of the hippocampus, entorhinal cortex, and amygdala (McKee et
al., 2009). Examination of elderly subjects may show a significant decrease in volume, however
that could be attributed to the aging process.
6
Figure 1. Gross Pathology of Chronic Traumatic Encephalopathy (CTE). Above is a coronal
slice of a normal brain that depicts the expectant size and sections. Underneath is a brain from a
retired professional football player that shows the anatomical characteristics of CTE. Shown on
the lower brain is (1) Dilation of Second Ventricle, (2) Dilation of Third Ventricle, (3) cavum
septum pellicidum, (4) atrophied structures of the temporal lobe, and (5) general volumetric
reduction of the mammillary bodies. (Stern et al., 2011)
Microscopic Pathology
Because of the variability of volumetric reductions, the most accurate diagnosis of CTE is
by examining the tissue on a microscopic level. Existing even during the early stages of CTE, the
presence of neurofibrillary tangles (NFTs) and neuropil threads (NTs) reside in the brain tissue
of the individual. NFTs and NTs can be found sporadically or uniform throughout the brain
tissue. They can be found uniformly throughout the brain or could be exclusive to only certain
areas. Similar to Alzheimer’s disease, CTE is categorized as a taupathy and its effects are
7
derived from injuries caused by these microscopic hallmarks. The protein tau is known to
associate with cellular microtubules. Throughout the development of an organism, the tau protein
is phosphorylated. The phosphorylation of the microtubule binding domain of the tau protein is
necessary to maintain the stability of cellular microtubules (Hanger, Anderton, & Noble, 2009).
By stabilizing the microtubules nutrients are allowed to be transported both intercellularly and
intracellularly. An increase in tau production can also be attributed to function of neural
plasticity. Within neurons, tau is a highly soluble protein that is found mostly in axons (Hanger
et al., 2009). However, in a taupathy such as CTE, the protein becomes mutated and exhibits a
loss in function of stabilizing microtubules. Even though in some diseases, such as Alzheimer’s
Disease, tau is not the major hallmark, it is observed that taupathy induces the formation of beta-
amyloid protein (Binder, Guillozet-Bongaarts, Garcia-Sierra, & Berry, 2005).
Comparison of Chronic Traumatic Encephalopathy and Alzheimer’s Disease
Many neurodegenerative diseases function in similar ways in the brain and in turn have
comparable causes. It has been reported that head injuries may increase the risk of developing
sporadic Alzheimer’s Disease (B. I. Omalu et al., 2005). Also, as CTE develops throughout the
lifespan, the individual begins to show symptoms similar to Alzheimer’s Disease. These
symptoms include emotional outbursts, lapses in executive functioning skills, as well as
depression and reports of suicidal thought (Koerte et al., 2015). Since the symptoms between
these two diseases are so similar, the term CTE was actually used as a descriptor when referring
to the physical and mental indicators displayed in Alzheimer’s Disease (Miller, 1966). At some
point the two diseases were considered to be the same thing due to their many similarities. Many
8
of the expected pathologies of CTE are also extremely similar to those seen in Alzheimer’s
Disease, although there are also some interesting differences.
Taupathy
Both diseases are considered to be taupathies, due to their manipulation of the protein tau.
The result of this manipulation is neurofibrillary tangles (NFTs), neuropil threads (NTs) and glial
tangles (GTs). However, in CTE the tau pathology is found in the cortical laminae (layers II and
III), where in Alzheimer’s Disease it is located in layers III and V. In CTE, it is also relatively
irregular in its location prevalence throughout the brain. In Alzheimer’s Disease the tau
hallmarks are observed to be more uniform (Gavett et al., 2011). The tau abnormalities in
Alzheimer’s Disease are seen to be gathering near the lower of the brain’s sulci, around small
blood vessels and under the superficial cortex. In the later stages of Alzheimer’s Disease the
NFTs and NTs are also found in the brainstem, diencephalon, subcortical white matter, and
limbic system (Stern et al., 2011).
Beta-Amyloid
Another major micropathological hallmark of Alzheimer’s Disease is the alternative
processing of beta-amyloid protein. This mutated protein is derived from the processing or
proteolysis of a precursor protein called amyloid precursor protein (APP). The up-regulation of
this protein’s production leads to an increase of alternative cleaving and the increased
concentration of plaques (Gentleman, Nash, Sweeting, Graham, & Roberts, 1993). Beta-amyloid
is created by the sequential cleavage of APP by two different enzymes, called beta-secretase and
gamma-secretase (Selkoe, 2011). APP is essential for the normal development of an organism. In
mice studies by Zheng et. al., by deleting the gene responsible for creating APP it also resulted in
9
changes in cognitive and locomotive behavior in the organism (Zheng et al., 1995). The irregular
beta-amyloid protein formation is regularly seen in the development of Alzheimer’s Disease but
is only present in approximately 40% of CTE cases (McKee et al., 2009). The protein
congregates together and forms large plaques that interfere with neural signaling. This distinction
demonstrates that while there are many similarities between the symptoms and pathogenesis of
the two disorders, they are in turn different.
TDP-43
An additional microscopic event contributing to the pathology of CTE and Alzheimer’s
disease is the presence of the trans-activator regulatory DNA-binding protein 43). TDP-43 plays
a significant role in response to axonal injuries located in the central nervous system (Saulle &
Greenwald, 2012). Traumatic injuries to the axons also is observed to accelerate the production
and accumulation of TDP-43, leading to its dispersion throughout the cytoplasm (Gavett et al.,
2011). A study conducted by McKee et al. has detected the widespread presence of TDP-43 in
about 80% of their CTE cases. TDP-43 is also present in only about 34% of the cases of
Alzheimer’s Disease (King et al., 2010). The spreading of TDP-43 seems to be generally
uniform, and it has been detected on occasion to spread into the upper regions of the spinal cord.
However, similar to the NFTs and NTs formed by the hyperphosphorylation of tau, the presence
of TDP-43 cannot be seen until post-mortem during a histological examination of brain tissue.
Apolipoprotein E (ApoE)
Changes in genetics can also be shown to be an indicator when discussing
neurodegenerative diseases. It could be also hypothesized that genetic factors of one disorder that
10
deteriorates the brain may exist in more than one disorder. Genotyping of the gene
Apolipoprotein E (ApoE) has rendered significant results in the development of
neurodegenerative diseases including Alzheimer’s disease as well as CTE. Historically, ApoE
has been linked with the prognosis of Alzheimer’s Disease, however in studies it has had an even
greater influence on subjects with head trauma (McKee et al., 2009). However, recently the
ApoE4 gene has been noted to have an effect in CTE. Based on genetic testing of individuals
with a history of head trauma, approximately 57% of individuals that presented CTE symptoms
also contained at least one ApoE4 allele (Gavett et al., 2011). Both the ApoE3 and the ApoE4
genotype were examined in multiple studies and have proven to effect cognition in populations
with a history of brain trauma. In mouse models, the ApoE4 allele has had an increased
mortality rate than those ApoE3 genotype (Saulle & Greenwald, 2012). It should also be noted
that individuals younger than 15 that carry at least one of the ApoE4 allele will experience a
poorer outcome after TBI (Gavett et al., 2011).
The Use of Neuroimaging with Chronic Traumatic Encephalopathy
Recently the work of medical imaging, such as magnetic resonance imaging (MRI), has
been proven useful for early detection of brain alterations stemming from repetitive head trauma
as well as the onset of a neurodegenerative disease (Koerte et al., 2015). Imaging techniques
have also been useful in determining the underlying pathological mechanisms in brain injuries as
well as the development of neurodegeneration. Additional roles provided by neuroimaging
techniques include “the identification of treatable injuries, and prevention of secondary damage”
(Bruce et al., 2015). Unlike many other systems in the human body, the nervous system is
extremely feeble and vulnerable to injury. Therefore, most surgical options are commonly a last
11
resort. Owing to this MRIs and other neuroimaging techniques are a non-invasive method that
keeps the patient safe while still providing a plethora of information.
Magnetic Resonance Imaging (MRI)
There are many types of neuroimaging techniques used to explore the damage of brain
trauma. However, one of the most common and widely used techniques is Magnetic Resonance
Imaging (MRI). Originally termed nuclear magnetic resonance (NMR) imaging, to be less
feared by the public during the Cold War the term “nuclear” was removed and its name was
changed to MRI. This technique manipulates the way in which the hydrogen atoms in water
rotate, by using a combination of magnetic fields and radio waves. MRI is used frequently
through many disciplines of science and medicine because it has the ability to provide top notch
resolution of soft tissue. Apart from its use examining neuroanatomy, MRI is also used to
visualize ligament damage, intercalated lumbar discs, nerve displacement, etc. The ability to
discern between white and gray matter in the brain also makes it a wonderfully useful tool to
visualize different structures within the brain. In the case of trauma to the brain, MRI also has the
ability to detect hemorrhage and minute macroscopic damage (Shenton et al., 2012).
Magnetic Resonance Imaging combines the strength of a magnetic field, B0, and radio
frequency pulses. Our bodies remarkably contain many small magnetics in the form of hydrogen
atoms. Because the majority of the human body is water, there are plenty of hydrogen atoms
contained within it. Normally these protons contain random movement patterns and more
importantly move independently of one another. They also spin along one axis like a top in a
movement called procession. When inserted into the large magnetic field, B0, the protons align
in the same direction. Radio frequency pulses are sent throughout the subject in order to change
12
the direction of procession. During the return of the molecules to the main magnetic field
direction, energy is emitted, which is used to generate an image (Susumu Mori, 2007). The
values of the image pixels correspond to the concentration of protons (usually from hydrogen
atoms of water) in the area (S. Mori & Zhang, 2006).
Because MRI images represent a three-dimensional object it creates two dimensional
slices of the subject. When obtaining an MRI of the brain the image displays an axial, coronal
and sagittal series of slices. However, because MRIs represent a three dimensional image their
pixels have thickness in addition to length and width. These 3D pixels are called voxels and they
store all of the information collected during the MRI acquisition process.
Structural MRI
The first of the of the two types of MRI techniques I used in my project is called a
structural MRI. These are the commonly clinically used MRI that shows incredible contrast of
soft tissues and bone structures. These types of images known as T1 or T2 weighted images are
used primarily for structural and volumetric analysis of the brain. One benefit T1 MRI images
holds over CT scans is that they are more spatially sensitive which makes them useful markers of
a disease’s injury progression (Bruce et al., 2015).
FreeSurfer is a program that was developed at the Martino’s Center for Biomedical
Imaging at Massachusetts General Hospital in Boston, Massachusetts that estimates volumes and
shapes of different brain parts from structural brain images. FreeSurfer software has a variety of
uses including volumetric segmentation, segmentation of white matter, parcellation of cortical
folding patterns, mapping the thickness of cortical grey matter, and the construction of surface
models of the cerebral context (Fischl, 2012). For our purposes, we will use FreeSurfer as a
13
measurement tool in order to determine the volume of the white and gray matter of the brain.
Through a process called segmentation, “information including the statistical properties of the
anatomical structures are stored in a space where their coordinates possess anatomical meaning,
as opposed to arbitrary coordinates found in the raw structural image” (Fischl et al., 2002). The
voxels of the image are classified as to whether they belong to white matter, gray matter or
cerebrospinal fluid (CSF). Those voxels are then clustered based on brain tissue type. For
example, gray matter is clustered together and would be segregated from white matter voxel
clusters. This whole brain segmentation and clustering procedure allows programs such as
FreeSurfer to quantify the amount of white matter, gray matter, CSF, as well as the volume of
specific brain structures.
Diffusion MRI
An important imaging technique used when examining neural anatomy both on a gross
and on a microstructural level is diffusion MRI. Diffusion MRI quantifies the random
movement of water molecules (or Brownian Motion) located within the brain tissue. The
diffusion of water is not unique to white matter tracts but can be observed in any other tissue,
like kidney, skeletal muscle, and cardiac muscle. However, the degree of diffusion anisotropy in
tissues besides white matter tracts seem to be much less significant (Beaulieu, 2002). Images
produced from a diffusion MRI scan are called diffusion weighted images (DWI), with a contrast
that reflects the amount of water diffusion at that site, and along a specified direction. Combining
the information from several directions can be used to derive indicators of tissue microstructure.
Diffusion Tensor Imaging (DTI) is a model that is utilized in order to analyze and to
quantify the diffusion properties of the white matter bundles (Assaf & Pasternak, 2008). The DTI
14
model is able to determine the three-dimensional diffusion profile in which water molecules
move throughout brain tissue. Assigning a physical quantity (diffusion tensor) to the DWIs
allows us to compare data between groups of subjects and also to draw conclusions based on the
biological processes occurring in the brain tissue. DTI uses the movement of water to make
inferences about the neural anatomy (S. Mori & Zhang, 2006). For example, in the axons of
neurons water perpendicular to the axon is mostly restricted from movement, but is relatively
free to diffuse along the axon. This is known as anisotropic diffusion, meaning that diffusion is
directionally dependent (Beaulieu, 2002). This suggests a bundle of fibers that are elongated in
the direction the water is diffusing faster. When the diffusion is equal in all directions similar to a
sphere, it is said to be isotropic.
Fractional Anisotropy (FA) is one of the most widely used measures of diffusion anisotropy,
which is useful to determine whether the diffusion in that area is isotropic or anisotropic. For
example, a white matter tract, which has a distinct directionality, will have higher FA value,
closer to 1. In contrast, in gray matter, where there is less directionality, FA will have a lower
value closer to 0.
Free-Water Analysis
DTI studies have revolutionized the way in which the brain is studied. Applications in
research include using anisotropy as a marker for pathological studies, monitoring anisotropy
during brain development, gathering information about the myelination of axons, examining the
parcellation of white matter, etc. (S. Mori & Zhang, 2006).
One limitation to diffusion MRI and the quantification of anisotropy is when water is
accumulated naturally such as the presence of cerebrospinal fluid (CSF) or inflammatory
15
responses and edema. This technique evolved into another correctional technique known as free
water analysis (FW) that was able to eliminate the effect of CSF as well as other components
such as inflammation, edema and other forms of extracellular water. Free water analysis focuses
on water molecules that do not experience flow and that movement is not prohibited due to its
surrounding structures (Pasternak, Sochen, Gur, Intrator, & Assaf, 2009). According to Corsellis
et al, one of the most apparent gross pathological findings in CTE is the enlargement of the
lateral and third ventricles (Saulle & Greenwald, 2012). The ventricles serve as pumps in the
brain in order to carry CSF from the inferior to superior regions of the brain. CSF continuously
bathes the brain and it serves to provide a cushion to minimize impact from any physical trauma.
Free water affects diffusion MRI because it shows isotropic diffusion and can interfere with the
diffusion values (Pasternak et al., 2009). Therefore, it is of interest to eliminate the effect of free
water when quantifying diffusion in tissue. Edema and tumors provide similar contamination as
CSF contamination. Specific areas that show an increase in CSF contamination include the
fornix, the cingulum and the corpus callosum (Papadakis et al., 2002). All of these structures are
in close proximity to the CSF filled ventricles.
Subjects that have experienced repeated instances of head trauma have been observed to
contain excess edema. In order to extract as much information as possible from the damaged
brain a technique can be used to account for the free-water detected in the DWIs. This
correctional technique is commonly referred to as free-water imaging (Pasternak et al., 2009).
Free water imaging accurately detects and subtracts out the component of free-water that appears
in the extracellular space, and could be affected by the presence of hemorrhaging, brain tumors,
pooling of cerebrospinal fluid, and edema caused by trauma to the brain.
16
Many neurodegenerative diseases progressively work towards similar outcomes; to break
down the central nervous system. This is no different for CTE and Alzheimer’s Disease.
Bioimaging techniques, including but not limited to the use of MRIs, are non-invasive ways that
visualize brain tissue and more importantly tract the progression of disease. In this study, I aim
to use both diffusion and structural MRI methods in order to measure both the volume and white
matter microstructure of the brains belonging to retired professional football players. A similar
study, conducted by Maier-Hein et. al., found evidence that supported the case that as free-water
in the white matter increased the total volume of the brain decreased. I seek to find similar
effects in the brains of the retired football players.
17
Materials and Methods
This study is a part of the Diagnosing and Evaluating Traumatic Encephalopathy using
Clinical Tests (DETECT). This is a program funded by the National Institute of Health that is
intended to develop biomarkers for the diagnosis of CTE in vivo. Participants in the study
underwent both neuroimaging and neuropsychological testing as well as had an interview
regarding medical history, as well as genetic testing and cerebrospinal fluid (CSF) protein
analysis.
Participants
The subjects ranged in ages from 40-68 years. Any subjects who showed the presence of
any diagnosed diseases of the central nervous system were excluded. Subjects were also
excluded if their MRI scans were unable to pass quality control standards of the Psychiatry
Neuroimaging Laboratory (PNL), Department of Psychiatry, Brigham and Women’s Hospital
(BWH), Harvard Medical School, Boston.
Data acquisition
All subjects underwent MRI at BWH. The instrument used was a 3T scanner (Magneton
Verio, Siemens Healthcare, Erlangen, Germany) that contains a 32-channel head coil. A T1-
weighted Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequence was acquired
18
using the following settings: TE= 3.36 msec, TR= 1800 msec, inversion time 1100 msec, flip
angle 7 degrees, acquisition matrix 256 x 256 x 176 voxels with a voxel size of 1 x 1 x 1 mm.
Figure 2. Standardized Core Processing Pipeline used in the Psychiatry Neuroimaging
Laboratory. The pipeline is a step-by-step processing procedure all images undergo between the
collection and analysis of both structural and diffusion MRI images. All images are pre-
processed (1), and then will undergo steps associated with the type of image being analyzed [2a
for DWI and 2b for T1/T2 images]. Following those steps the image will then enter the post-
processing steps (3).
Standardized Core Processing Pipeline
The MRIs were acquired at Brigham and Women’s Hospital it was then analyzed at the
PNL using the following processing pipeline (Figure 2): The first step in the pipeline is to
convert the DICOM/NIFTI file that was outputted by the MRI scanner into a Nrrd file that is
(1) Pre- Processing
19
able to be read by software programs used in house at the PNL. The image is then manually
aligned on its axis and centered to ensure that all data generated is in the center space of the scan.
At this point the pipeline is path dependent on the image type being processed.
Diffusion Image Post-Processing
The first step in the diffusion branch of the processing pipeline includes corrections for
expected image artifacts caused by motion or eddy current. Motion is a known artifact in MRI
processing because movement during collection can cause blurring and other artifacts making it
difficult to analyze properly. The next step in the pipeline is the generation of a tensor mask. A
mask is generated in a program called 3D Slicer software package version 4.3.1
(www.slicer.org). A tensor mask is generated semi-automatically, but is edited by the user to
ensure that all regions of the brain are clearly marked and identified in the image in order for
further processing.
Structural Image Post-Processing
As noted earlier structural images have higher resolution and are used to look further into
the shape and quantity of brain matter tissue. The PNL uses two different types of structural
images. T1- and T2-weighted images. The first step in the structural pipeline is to realign the T1
image. Similar to the diffusion weighted image (DWI), the structural image needs to be masked
before further processing occurs. Unlike the DWI’s tensor mask, the structural image’s masking
process is known as an atlas mask. In the process of atlas masking, many older masks that were
generated manually are used as an atlas to guide the semi-automated process to separate the
regions of the image that contain brain and to separate it from bone tissue, fluids and other
objects presented in the image (Del Re et al., 2016).
FreeSurfer Analysis
The image is then brought into the FreeSurfer program for a process known as
segmentation. In segmentation, the program will work to separate regions of the MRI image into
different structures located within the brain. This is especially helpful when you are only
concerned on a specific part or parts of the brain to analyze, or are unsure which areas may be
effected by symptoms presented in the subject. Like most steps in the processing pipeline the
segmentation step through FreeSurfer is run through a computer script, and the results of the
segmentation process are then analyzed in order for quality control purposes. The FreeSurfer
output contains a total of 177 regions of interest. This total is comprised of 103 gray matter
regions, and 74 white matter regions. The total gray and white matter volumes were normalized
by dividing by the intracranial volume. This step was put in place to negate for body size
variability between individuals.
Tract Based Spatial Statistics (TBSS)
TBSS is a software script that uses fractional anisotropy data that is gathered from
diffusion MRI and aligns multiple subjects into a common white matter skeleton (Smith et al.,
2006). Images from different subject (e.g. patients and controls) are then projected onto the
skeleton and by using statistical analysis, notable changes between groups in the skeleton can be
determined.
21
Diffusion Measures
The MRI scanner runs a diffusion sequence to produce diffusion weighted images (DWI).
After the images are preprocessed, the data is analyzed using a free-water algorithm. The
algorithm outputs the free water measures that were used in this project. Free Water is useful
because it describes the amount of extracellular water present at that region of the skeleton that
does not experience flow. A significant accumulation of free water could suggest edema is
present in the brain due to injuries, tumors, trauma or hemorrhage. TBSS was then used to create
a white matter skeleton over the DWI. Diffusivity values that were collected and used during this
investigation that were projected onto the skeleton include the axial diffusivity (ADt), radial
diffusivity (RDt), and fractional anisotropy (FAt). The letter “t” in the modalities name signifies
that it is corrected to account for the presence of free-water. ADt and RDt examine the
diffusivities parallel and perpendicular to the fibers. FA is the normalized standard deviation of
the diffusivities (Assaf & Pasternak, 2008). FA is a value that ranges between 0 and 1. If the
value is 0, then it suggests that the diffusion is isotropic. If the value is 1, it shows the diffusion
happens on a specific axis and diffusion in other directions is restricted. Statistical analysis is
then performed by TBSS on the 63 regions of interest defined on the skeleton for each measure.
Statistical Analyses
Subjects that were included in this study were those subjects that had both usable
diffusion weighted images as well as usable structural image. As noted earlier, the structural
image provides information on the amount of volume of the different brain parts. The diffusion
image is used to collect data that provides information about the microstructure of the white
matter. Diffusion data included 62 regions of interest labels as well as an average from all
22
regions derived from TBSS, called ENIGMA. The ENIGMA labels are associated with sections
of the white matter skeleton that was generated through the tractography process. ADt, RDt and
FAt were all modalities run on the diffusion image that showed the directional diffusion present
in the white matter fibers. Additionally, a free water modality was run on diffusion weighted
image in order to show the diffusion of extracellular water within the white matter
microstructure.
As for the structural data, it was collected using region of interest labels associated with
FreeSurfer that was applied during the segmentation process. Two data sets were used that were
generated through FreeSurfer. Volumetric data was collected in order to see the amount of brain
tissue located in each voxel of the MRI. Also collected was a free-water measure that
determined the volume occupied with extracellular water in the brain. Before the data was
analyzed, z-score tests were performed in order to remove any outliers. Any points that were to
have a z-score of between 3 and -3 were removed from the sample set. All statistical analyses
conducted in this research were performed with IBM SPSS Statistics (Version 2.3)
Correlation between Free Water and Gray Matter
To determine whether there was any present relationship between the percent of free
water and volumes of the brain, a partial correlation was conducted. The diffusivity values that
were generated through the ENIGMA procedure for the images were used to represent percent of
free water values. The total gray matter volume was taken from the values created by the
FreeSurfer segmentation process. All gray matter values were normalized, by dividing the
volume of brain tissue by the intracranial volume (ICV). This technique was used in order to
account for different sizes in craniums between athletes and non-athletes. An additional reason
23
that a partial correlation technique was used as opposed to bivariate correlation was because it
allowed the option to control for a variable. The controlling variable used in this correlation was
age. The reason this value was chosen is to account for the natural degeneration of brain tissue
caused by the natural aging process.
Group Comparisons of Volumes
Normalized volumes that were produced by FreeSurfer segmentation were used in order
to determine if there were any statistically significant volumetric differences between the two
groups. The control group was then compared with the player group using a general linear model
(GLM). The univariate GLM used was selected because of its similarities with an independent t-
test. Furthermore, the univariate GLM allowed for the controlling of age. There are 177 regions
of interest were compared between the control and player groups for differences. Any region that
displayed a p-value less than or equal to 0.05 was noted as significant.
Group Comparisons of Free Water
The free-water values that were produced by FreeSurfer were used to determine any
significant differences between control and player groups. The free water values were
coregistered to the anatomical images and then their values were averaged for each FreeSurfer
region of interest. A univariate GLM was also used to test for differences in this case because of
its ability to control the age variable. Approximately 177 regions of interest were compared
between the control and player groups for differences. Any region that displayed a p-value less
than or equal to 0.05 was noted.
24
Group Comparisons for ENIGMA (ADt, FA, FAt, FW and RDt)
Values produced by the TBSS program used the ENIGMA skeleton and labels as their
regions of interest. Due to the size of the skeleton, there are sufficiently less regions of interest
(63) than compared to FreeSurfer. However, data from multiple modalities (ADt, FA, FAt, FW,
and RDt) were used in order to test for differences in the white matter diffusivity. Each modality
was tested independently to determine group differences. A univariate GLM was also used to test
for differences in this case because of its ability to control the age variable. Any region that
displayed a p-value less than or equal to 0.05 was noted.
Correlations between Regions of Interest
The regions of interest that were deemed to be statistically significant were then
correlated against one another. A partial correlation was conducted in these instances, in order to
control for the age variable. This technique was conducted separately for the normalized
volumes, free water volumes, ADt, FA, FAt, FW, and RDt. Relationships between regions of
interest were deemed statistically significant if they had a p-value of less than or equal to 0.05.
Correlations between Volumes and Diffusivity
Lastly, any volumetric region of interest that was proven to contain a statistically
significant relationship was then correlated with any diffusion region of interest that shown
significant correlations. A partial correlation was used again, because of its ability to control for
age. A correlation was deemed statistically significant if it produced a p-value of less than or
equal to 0.05.
Descriptive Statistics
The total number of participants in this study was 99. The player group consisted of 79
male football players. The mean age of the players group was 55.1 years and the standard
deviation is 7.88. They each played football for a minimum of 7 years including a minimum of
2.5 NFL seasons. Participant’s football position varied with approximately 18.6% offensive
lineman, 6.9% running backs, 3.9% tight ends, offensive skilled positions 1%, defensive lineman
10.8%, linebackers 19.6%, and defensive backs 16.7%.
The control group consists of 20 participants (mean= 55.95 years, standard deviation
7.266). The control group contains subjects that were likely not exposed to head trauma over the
course of their lives. Participants that had served in the military or had ever participated in
football, hockey, soccer, rugby, wrestling, boxing, lacrosse, basketball, martial arts, kickboxing,
cycling, or distance running were also excluded from the control group. There was no significant
statistical difference in age between control and test groups (t-test= 0.437, p= 0.663).
Relationship between amount of Free Water and Total Gray Matter
As depicted in Figure 3, there was no correlation between the percent of free water on the
skeleton of the white matter and the amount of gray matter volume within the player’s group (p=
0.933, r= -0.010).
26
Figure 3. Free Water values correlated with the Total Gray Matter Volume. This figure presents
the relationship between the percentage of free-water with the total amount of gray matter in the
brains of retired football players. No statistically significant could be established. (p= 0.933, r= -
0.010).
Region of Interest Analyses
Comparing the volumes of the different normalized volumes produced by FreeSurfer
segmentation between controls and players (Table 1) showed 5 regions where players had
significantly higher volume, and 13 regions where controls had significantly higher volume.
0.1
0.11
0.12
0.13
0.14
0.15
0.16
0.17
0.18
% F
27
Left Amygdala 0.010 6.935 0.000918902 0.000850868
Right Inferior
Lateral Ventricle
Ctx-lh-entorhinal 0.030 4.866 0.001141643 0.001029801
Ctx-lh-
inferiortemporal
Ctx-lh-
parahippocampal
Wm-lh-
middletemporal
Ctx-lh-
mediaorbitofrontal
Ctx-lh-paracentral 0.049 3.966 0.002072022 0.003320524
Wm-lh-paracentral 0.017 5.892 0.002358289 0.002570945
Wm-rh-paracentral 0.012 6.523 0.002919959 0.003174186
Table 1. Normalized Volumes. This table presents the group comparisons of the normalized
volumetric values produced by FreeSurfer segmentation. Note: The blue rows indicate group
comparisons in which the control group showed an increase of volume. The orange rows indicate
group comparisons in which the player’s group showed an increase in volume.
28
ALIC-R 0.047 4.057 0.122311519 0.113155949
CST-L 0.047 4.060 0.131679457 0.118694431
FXST-L 0.007 7.465 0.13694525 0.12388754
IC 0.048 3.989 0.120961300 0.114335750
PLIC-L 0.046 4.069 0.96098986 0.88874738
SCC 0.029 4.898 0.115306419 0.105470801
SFO-R 0.003 9.091 0.105764814 0.091398104
IFO 0.028 4.977 0.118618481 0.131641194
Table 2. Free-Water. This table presents the group comparisons of the Free-Water diffusivity
values located on the ENIGMA white matter skeleton. Note: The blue rows indicate group
comparisons in which the control group showed an increase of free water. The orange rows
indicate group comparisons in which the player’s group showed an increase in free water.
Table 3. ADt. This table presents the group comparisons of the ADt diffusivity values located on
the ENIGMA white matter skeleton. Note: The orange rows indicate group comparisons in
which the player’s group showed an increase in ADt.
When comparing the Free Water diffusivity values (Table 2) within the white matter
microstructure, the results showed 7 regions where the control group had a significant larger
amount of free water. It should be noted that IFO was the only region of interest that showed an
increased amount of Free Water present in the player’s group
(p= 0.028, t= -2.22). When comparing the ADt diffusivity values (Table 3) between groups, it
was shown that 5 regions showed a significant increase in the player’s group. However, no
regions were proven to show an increased amount in the control group.
When examining the group differences for FAt in the FreeSurfer regions of interest
(Table 5), there were only 4 instances in which the control group possessed an increased amount
Marker P-value F-Value Control Mean Player Mean
BCC 0.034 4.626 0.001458370 0.001480034
EC-L 0.003 9.024 0.001029558 0.001049909
EC-R 0.033 4.666 0.001031173 0.001043635
IFO-R 0.013 6.368 0.001113979 0.001151081
PLIC 0.047 4.044 0.001276385 0.001291473
29
of anisotropy than its player group counterpart. The player’s groups were shown to have an
increased amount of diffusivity at 7 regions of interest. The group differences for the volume of
free water (Table 6) within the brain show a noticeable increase in the parahippocampal cortex as
well as the white matter of caudal middle frontal lobe in the control group. The players group
have seen a statistically significant increase of free water in 3 regions; the left choroid plexus, the
optic chiasm, and the white matter of the parstriangularis of the right hemisphere.
Additionally, there were no statistically significant differences between groups in either
the FA, FAt, or RDt diffusion values.
Associations between Regions of Interest
Partial correlations where then run amongst the regions of interest that displayed
significant differences between the players and controls. Once again a relationship was
determined to have statistical significance if it had a p-value of less than 0.05.
There are both relationships that show positive and negative correlations present among
the regions of interest for the normalized volume (Figure 4). Positive and negative correlations
can be determined by the sign of the r coefficient that was calculated during the partial
correlation. The r coefficient is a number between 1 and -1, that shows the strength of the
correlation. One important note is that there was only one region, the cortex of the
mediaorbitofrontal lobe of the left hemisphere, that did not correlate with any other region.
30
showed a statistical significance were outlined in red. Negative correlations that showed a
statistical significance were outlined in purple. The level of significance is noted through the key
in the bottom left of the figure. p= 0.05 is followed by one asterisk (*), p= 0.01 is followed by
two asterisks (**), p= 0.001 or lower is followed by three asterisks (***).
31
BCC Body of the corpus callosum
CST Corticospinal tract
EC External capsule
SCC Splenium of corpus callosum
SFO Superior frontal-occipital fasiculus
Table 4. ENIGMA Skeleton Label Abbreviations. The following table presents the abbreviations
of regions of interest that comprise the ENIGMA white matter skeleton.
There was a total of three relationships within the regions of the skeleton when
examining the ADt (Figure 5). The external capsule of the left hemisphere positively correlated
with its right hemisphere counterpart (p= 0.001, r= 0.735), as well as the inferior fronto-occipital
fasiculus (IFO) of the right hemisphere (p= 0.001, r= 0.362). In addition, there was a positive
association in the skeleton of the right hemisphere between the external capsule and the IFO (p=
0.002, r= 0.335). It should be noted that all relationships within ADt in regions of the ENIGMA
skeleton were positively correlated with one another.
There were multiple associations between regions when looking into the amount of free
water present in locations of their white matter microstructure (Figure 6). It should be noted that
all relationships shown positive correlations with one another.
32
Figure 5. ADt Correlations [ENIGMA Labels] Positive correlations that showed a statistical
significance were outlined in red. Negative correlations that showed a statistical significance
were outlined in purple. The level of significance is noted through the key in the bottom left of
the figure. P= 0.05 is followed by one asterisk (*), p= 0.01 is followed by two asterisks (**), p=
0.001 or lower is followed by three asterisks (***).
Figure 6. Free Water Correlations [ENIGMA Labels]. Positive correlations that showed a
statistical significance were outlined in red. Negative correlations that showed a statistical
significance were outlined in purple. The level of significance is noted through the key in the
bottom left of the figure. P= 0.05 is followed by one asterisk (*), p= 0.01 is followed by two
asterisks (**), p= 0.001 or lower is followed by three asterisks (***).
33
When examining the relationships within the FAt regions located within the FreeSurfer
regions of interest (Figure 9), it can be observed that there were many regions that have a
statistically significant correlation. All of the relationships determined also were positively
correlated with one another. It should be noted that the left accumbens area has shown the most
statistically significant correlations.
When looking at the associations between regions of interest for their amount of free
water (Figure 10), of the five regions that have been proven to show differences between the
control and player group, only four relationships are shown to be significant. All four
relationships were positively correlated with one another. The third ventricle correlated with the
left choroid plexus (p=0.012, r= 0.284) and the optic chiasm (p= 0.007, r= 0.308). The optic
chiasm has shown an association with the white matter of the parstriangularis of the right
hemisphere (p= 0.046, r= 0.228). The white matter of the caudal middle frontal lobe of the right
hemisphere has shown the strongest correlation with the white matter of the parstriangularis of
the right hemisphere (p= 0.000, r= 0.614).
Relationships between Volumes and White Matter Microstructure
To reflect back on the main purpose of the study, the white matter microstructure
(ENIGMA regions) was then correlated with the normalized volumes (FreeSurfer regions). The
intent of this statistical analysis is to see if the white matter and gray matter behave in similar or
different ways in the brains of football players.
Firstly, the normalized volumes produced by the FreeSurfer segmentation process were
correlated with the ADt values (Figure 7). There were both positive and negative relationships
associated with the PLIC region. It has shown a relationship with the brain stem (p= 0.03, r= -
34
0.332), the left amygdala (p= 0.004, r= -0.319), the right inferior lateral ventricle (p= 0.000, r=
0.440), and the cortical area of the right hemisphere’s parahippocampus (p= 0.004, r= -0.323).
There were two other relationships determined to be significant. The white matter volume of the
paracentral region of the right hemisphere positively correlated (p= 0.015, r= 0.276) with the
ADt values of the white matter microstructure. The volumes of the entorhinal cortex located in
the left hemisphere also showed a positive correlation (p= 0.018, r= 0.267) with the ADt values
of the right hemisphere’s IFO region on the white matter skeleton.
The normalized volumes were also correlated with the amount of free water occupying
the white matter skeleton produced by ENIGMA. (Figure 8). There were 10 total statistically
significant relationships formed between the regions of interest. There were 3 positive
correlations and 7 negative correlations formed. It should be noted that 3 relationships were
formed in the brain stem; PLIC-L (p= 0.013, r= -0.280), SCC (p= 0.032, r= -0.247) and SFO-R
(p= 0.017, r= -0.271).
Figure 7 Normalized Volumes correlated with ADt values. Positive correlations that showed a
statistical significance were outlined in red. Negative correlations that showed a statistical
significance were outlined in purple. The level of significance is noted through the key in the
bottom left of the figure. P= 0.05 is followed by one asterisk (*), p= 0.01 is followed by two
asterisks (**), p= 0.001 or lower is followed by three asterisks (***).
36
Figure 8. Normalized Volumes correlated with Free Water. Positive correlations that showed a
statistical significance were outlined in red. Negative correlations that showed a statistical
significance were outlined in purple. The level of significance is noted through the key in the
bottom left of the figure. P= 0.05 is followed by one asterisk (*), p= 0.01 is followed by two
asterisks (**), p= 0.001 or lower is followed by three asterisks (***).
37
Discussion
The finding related to the main goal of this study was that gray matter volume and white
matter did not correlate in CTE. This was in contrast to the finding of Maier-Hein et al., where
they found a strong negative correlation between the percentage of free water in the white matter
with the volume of gray matter in subjects who presented symptoms of Alzheimer’s Disease. As
a result of our study we can suggest that the interaction between gray and white matter is
different in Alzheimer’s Disease and CTE.
There were group differences that were identified between the control and player groups.
These differences included an increase in volume of the control group in 13 regions, which
suggests that statistically the control group had more brain tissue than the player’s group. This
may or may not be a direct result of their history of head trauma. When examining the group
differences of the free water volumes in the white matter microstructure, only IFO showed a
volumetric increase in the player’s group. This suggests that the amount of extracellular water in
the white matter was not widely increased in the player’s group as originally hypothesized. In
actuality, the control group showed 7 instances where there was significant increase in free
water. These results may suggest that age could have played a role in the natural decreasing of
brain volume over the subject’s lifespan.
It should also be noted that there was a significant increase of free water in most white
matter skeleton (Table 2) regions of the control group. The free water was significantly larger in
the player’s group in only one region of the white matter microstructure, IFO. Before testing, it
38
was assumed that there would be more extracellular water in the player’s region due to excess
swelling and brain edema due to the mTBIs occurred over the course of their lifetime.
It has also been shown through statistical correlations that areas of the brain seemed to
lose volume uniformly, meaning that not only one or a few areas degraded before others were
affected (Figure 4). This showed that regions of the brain degraded in volume more as a network
than one region directly having an effect over another. This type of uniformity also was
observed in the presence of free water located in the white matter microstructure (Figure 6). All
regions of the microstructure seem to be related to each other in their development of
extracellular water. This suggests that multiple regions of the brain act as a network instead of
independently to degrade the brain tissue.
To return to the original scope of the project, the white matter measurements were then
correlated with the volumetric measurements. In Figure 7, the volumes were correlated with ADt
values from the white matter microstructure. There were 10 total statistical significant
relationships; 7 that were positive and 3 negative. This suggests that different regions of the brain
are acting independently of one another. In some regions, the axial diffusion is increasing as the
volume of the brain is decreasing, and vice versa. This provides more evidence for the lack of
positive or negative correlations between white matter tract diffusivity with the amount of gray
matter.
Free water measures were then correlated with the normalized volumes. In this study,
only the regions that shown significant differences were plotted against one another. As stated
earlier, the results were inconclusive. A total of 10 relationships were found to be significant; 7
negative correlations and 3 positive correlations. This suggests that, similar to the findings of the
39
ADt and volume correlations, the gray matter and free water seem to be increasing and
decreasing independently in the player population.
Some limitations of the study include the fact that the stage of progression of CTE is
unknown. It even possible that none of the subjects in the players group would develop the
disorder because the only way to truly diagnose it is post-mortem. Biomedical imaging is a
useful tool for prediagnosis but it does contain artifacts and measurements are not as accurate as
physically measuring the amount of tissue volume in the brain. Another limitation includes the
distinction between brain degradation due to neurodegenerative disease or simply by the natural
aging process. This means that there is no real way to determine that members of the control
group did not have lower than average gray matter volumes or present microstructural
deficiencies in their white matter. However, some benefits of examining the brain this way
include a non-invasive method of tracking stages of brain deterioration in a subject. Because the
brain is such a fragile organ, the only other way to truly examine it would be an invasive method
such as tissue or cell culture. By using MRI techniques not only viable information is being
collected, but it can be gathered more often showcasing the manner of volumetric change.
Future directions in this study would include comparing the regions that present
volumetric and microstructural abnormalities with different psychological behaviors such as
mood, executive functioning skills and memory. It would be interesting to see whether or not the
decreased volumes or diffusivity values had a positive or negative relationship to these factors.
During the study, the only variable that was controlled for was the age of the participant. Perhaps
if other variables were controlled for, such as BMI or position, there would be different results.
Although containing similar mechanisms of action and symptoms, all neurodegenerative
diseases include subtleties that deem them unique. Although the initial hypothesis of this study
40
was disproven, there are many takeaways from testing this question. Those differences that
make CTE and Alzheimer’s Disease unique, may one day lead to the breakthrough needed to
find the answer for either disease. However, we do know that the causes for the progressive
symptoms in traumatic injuries and Alzheimer’s disease are rather different. In CTE, repeated
trauma over the span of one’s lifetime triggers the molecular cascade of events leading to
degeneration of brain tissue. In Alzheimer’s disease, similar events sporadically (or genetically
through ApoE protein) occur. With such different causes, it is not surprising that they would
show different evidence within the brain. By combining multimodal imaging techniques such as
diffusion, structural and functional MRI could show underlying physiological changes that occur
during mTBI. Furthermore, more specific biomarkers could be formed by adding information
obtained through genetic and protein-based biomarkers in order to detect axonal injury,
inflammation, demyelination, apoptosis, and other symptoms used to best predict treatment and
outcome (Shenton et al., 2012). The future development of these biomarkers would become a
useful aid in the early detection and potential treatment options for future generations of athletes
of collision based sports. Alternatively, they could also be used to reactively treat a retired
player just beginning to show progression or onset of CTE.
41
Appendix
Left Cerebellum White
Right Cerebellum
White Matter
Ctx-lh-
Ctx-rh-
parstriangularis
Wm-rh-
parstriangularis
0.020 5.615 0.4178915 0.4286104
Table 5. FAt (FreeSurfer Labels). This table presents the group comparisons of the FAt
diffusivity values located on the ENIGMA white matter skeleton. Note: The blue rows indicate
group comparisons in which the control group showed an increase of FAt. The orange rows
indicate group comparisons in which the player’s group showed an increase in FAt.
42
Ctx-rh-
parahippocampal
Optic Chiasm 0.005 8.443 0.000137594 0.000147783
Wm-rh-
parstriangularis
0.042 4.263 0.002242531 0.002309160
Table 6. Free Water (FreeSurfer Labels) This table presents the group comparisons of the free
water detected within the FreeSurfer segmentation. Note: The blue rows indicate group
comparisons in which the control group showed an increase of free water. The orange rows
indicate group comparisons in which the player’s group showed an increase in free water.
Figure 9 FAt Correlations [FreeSurfer Labels]. Correlations that showed a statistical
significance were outlined in red. The level of significance is noted through the key in the
bottom left of the figure. P= 0.05 is followed by one asterisk (*), p= 0.01 is followed by two
asterisks (**), p= 0.001 or lower is followed by three asterisks (***).
43
Figure 10 Free Water Correlations [FreeSurfer Labels]. Correlations that showed a statistical
significance were outlined in red. The level of significance is noted through the key in the
bottom left of the figure. P= 0.05 is followed by one asterisk (*), p= 0.01 is followed by two
asterisks (**), p= 0.001 or lower is followed by three asterisks (***).
44
References
Assaf, Y., & Pasternak, O. (2008). Diffusion tensor imaging (DTI)-based white matter mapping
in brain research: a review. J Mol Neurosci, 34(1), 51-61. doi:10.1007/s12031-007-0029-
0
Beaulieu, C. (2002). The basis of anisotropic water diffusion in the nervous system - a technical
review. NMR Biomed, 15(7-8), 435-455. doi:10.1002/nbm.782
Binder, L. I., Guillozet-Bongaarts, A. L., Garcia-Sierra, F., & Berry, R. W. (2005). Tau, tangles,
and Alzheimer's disease. Biochim Biophys Acta, 1739(2-3), 216-223.
doi:10.1016/j.bbadis.2004.08.014
Bowman, K. M., & Blau, A. (1940). Psychotic States Following Head and Brain Injury in Adults
and Children.
Bruce, E. D., Konda, S., Dean, D. D., Wang, E. W., Huang, J. H., & Little, D. M. (2015).
Neuroimaging and traumatic brain injury: State of the field and voids in translational
knowledge. Mol Cell Neurosci, 66(Pt B), 103-113. doi:10.1016/j.mcn.2015.03.017
Del Re, E. C., Gao, Y., Eckbo, R., Petryshen, T. L., Blokland, G. A., Seidman, L. J., . . . Bouix,
S. (2016). A New MRI Masking Technique Based on Multi-Atlas Brain Segmentation in
Controls and Schizophrenia: A Rapid and Viable Alternative to Manual Masking. J
Neuroimaging, 26(1), 28-36. doi:10.1111/jon.12313
doi:10.1016/j.neuroimage.2012.01.021
Fischl, B., Salat, D. H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., . . . Dale, A. M.
(2002). Whole brain segmentation: automated labeling of neuroanatomical structures in
the human brain. Neuron, 33(3), 341-355.
Gavett, B. E., Stern, R. A., & McKee, A. C. (2011). Chronic traumatic encephalopathy: a
potential late effect of sport-related concussive and subconcussive head trauma. Clin
Sports Med, 30(1), 179-188, xi. doi:10.1016/j.csm.2010.09.007
Gentleman, S. M., Nash, M. J., Sweeting, C. J., Graham, D. I., & Roberts, G. W. (1993). Beta-
amyloid precursor protein (beta APP) as a marker for axonal injury after head injury.
Neurosci Lett, 160(2), 139-144.
Goldstein, L. E., Fisher, A. M., Tagge, C. A., Zhang, X.-L., Velisek, L., Sullivan, J. A., . . .
Wojnarowicz, M. W. (2012). Chronic traumatic encephalopathy in blast-exposed military
veterans and a blast neurotrauma mouse model. Science translational medicine, 4(134),
134ra160-134ra160.
45
Hanger, D. P., Anderton, B. H., & Noble, W. (2009). Tau phosphorylation: the therapeutic
challenge for neurodegenerative disease. Trends Mol Med, 15(3), 112-119.
doi:10.1016/j.molmed.2009.01.003
King, A., Sweeney, F., Bodi, I., Troakes, C., Maekawa, S., & AlSarraj, S. (2010). Abnormal
TDP43 expression is identified in the neocortex in cases of dementia pugilistica, but is
mainly confined to the limbic system when identified in high and moderate stages of
Alzheimer's disease. Neuropathology, 30(4), 408-419.
Koerte, I. K., Lin, A. P., Willems, A., Muehlmann, M., Hufschmidt, J., Coleman, M. J., . . .
Shenton, M. E. (2015). A review of neuroimaging findings in repetitive brain trauma.
Brain Pathol, 25(3), 318-349. doi:10.1111/bpa.12249
Martland, H. S. (1928). Punch drunk. Journal of the American Medical Association, 91(15),
1103-1107.
McKee, A. C., Cantu, R. C., Nowinski, C. J., Hedley-Whyte, E. T., Gavett, B. E., Budson, A. E.,
. . . Stern, R. A. (2009). Chronic traumatic encephalopathy in athletes: progressive
tauopathy after repetitive head injury. J Neuropathol Exp Neurol, 68(7), 709-735.
doi:10.1097/NEN.0b013e3181a9d503
Miller, H. (1966). Mental after-effects of head injury. Proceedings of the Royal Society of
Medicine, 59(3), 257.
Millspaugh, J. (1937). Dementia pugilistica. US Naval Med Bull, 35, 297-303.
Mori, S. (2007). Introduction to diffusion tensor imaging: Elsevier.
Mori, S., & Zhang, J. (2006). Principles of diffusion tensor imaging and its applications to basic
neuroscience research. Neuron, 51(5), 527-539. doi:10.1016/j.neuron.2006.08.012
Omalu, B. (2014). Chronic traumatic encephalopathy. Prog Neurol Surg, 28, 38-49.
doi:10.1159/000358761
Omalu, B. I., DeKosky, S. T., Minster, R. L., Kamboh, M. I., Hamilton, R. L., & Wecht, C. H.
(2005). Chronic traumatic encephalopathy in a National Football League player.
Neurosurgery, 57(1), 128-134; discussion 128-134.
Papadakis, N. G., Martin, K. M., Mustafa, M. H., Wilkinson, I. D., Griffiths, P. D., Huang, C. L.,
& Woodruff, P. W. (2002). Study of the effect of CSF suppression on white matter
diffusion anisotropy mapping of healthy human brain. Magn Reson Med, 48(2), 394-398.
doi:10.1002/mrm.10204
Pasternak, O., Sochen, N., Gur, Y., Intrator, N., & Assaf, Y. (2009). Free water elimination and
mapping from diffusion MRI. Magn Reson Med, 62(3), 717-730. doi:10.1002/mrm.22055
Saulle, M., & Greenwald, B. D. (2012). Chronic traumatic encephalopathy: a review. Rehabil
Res Pract, 2012, 816069. doi:10.1155/2012/816069
46
Selkoe, D. J. (2011). Alzheimer's disease. Cold Spring Harb Perspect Biol, 3(7).
doi:10.1101/cshperspect.a004457
Shenton, M. E., Hamoda, H. M., Schneiderman, J. S., Bouix, S., Pasternak, O., Rathi, Y., . . .
Zafonte, R. (2012). A review of magnetic resonance imaging and diffusion tensor
imaging findings in mild traumatic brain injury. Brain Imaging Behav, 6(2), 137-192.
doi:10.1007/s11682-012-9156-5
Smith, S. M., Jenkinson, M., Johansen-Berg, H., Rueckert, D., Nichols, T. E., Mackay, C. E., . . .
Behrens, T. E. (2006). Tract-based spatial statistics: voxelwise analysis of multi-subject
diffusion data. Neuroimage, 31(4), 1487-1505. doi:10.1016/j.neuroimage.2006.02.024
Stern, R. A., Riley, D. O., Daneshvar, D. H., Nowinski, C. J., Cantu, R. C., & McKee, A. C.
(2011). Long-term consequences of repetitive brain trauma: chronic traumatic
encephalopathy. Pm&r, 3(10), S460-S467.
Young, J. S., Hobbs, J. G., & Bailes, J. E. (2016). The Impact of Traumatic Brain Injury on the
Aging Brain. Curr Psychiatry Rep, 18(9), 81. doi:10.1007/s11920-016-0719-9
Zheng, H., Jiang, M., Trumbauer, M. E., Sirinathsinghji, D. J., Hopkins, R., Smith, D. W., . . .
Van der Ploeg, L. H. (1995). beta-Amyloid precursor protein-deficient mice show
reactive gliosis and decreased locomotor activity. Cell, 81(4), 525-531.
Abstract
Acknowledgments
Introduction
Chronic Traumatic Encephalopathy Pathology
Taupathy
Structural MRI
Diffusion Measures
Statistical Analyses
Group Comparisons of Volumes
Results
Relationship between amount of Free Water and Total Gray Matter
Associations between Regions of Interest
Relationships between Volumes and White Matter Microstructure
Discussion
Appendix
References