Chronic Traumatic Encephalopathy and Neurodegeneration in Contact Sports and American Football

What is a tauopathy?

Tau is a protein that fortifies microtubules in neurons and exists in all individuals. Tau is modified after translation in several ways such as glycosylation, nitration, and truncation, but one post-translation moderation important in neurodegeneration is hyperphosphorylation of tau. Hyperphosphorylated tau, known as p-tau, is associated with several disorders known as tauopathies, including AD, frontotemporal dementia, Parkinson’s disease (PD), progressive supranuclear palsy (PSP), corticobasal degeneration, multiple system atrophy, and others [49]. While the finding of p-tau can be pathologic, it is also present in normal aging. Braak et al. [50] reported that only 10 of 2,332 consecutively autopsied brains (0.004%) were absent of any abnormal tau, and all were under the age of 30 years. The NINDS CTE Neuropathology Consensus group [44] identified p-tau in a specific region and pattern to diagnose CTE, and stated that the p-tau of CTE was different from the age-related tau astrogliopathy or primary age-related tauopathy, previously noted by Braak et al. [50] Overall, due to their relative rarity, these tauopathies are often grouped together as a single endpoint to predict the occurrence of neurodegenerative diseases. The overlap between these tauopathies is an active area of research, and describing discrete differences between each tauopathy remains elusive.

Classic CTE

A recent systematic review of CTE by Gardner et al. [45] proposed two distinct variants of CTE to be described: “classic” and “modern” CTE. This proposal came after the authors concluded distinct clinicopathological differences between the earlier accounts of CTE compared to CTE mentioned over the last two decades. Compared to “classic” CTE, “modern” CTE includes a wider range of clinical symptoms and only one, but more specific neuropathologic finding, mostly seen in former athletes.

While “classic” CTE encompasses almost 80 years of research, it can be boiled down to the work by Roberts et al. [32] in 1969 and Corsellis et al. [1] in 1973. As mentioned above, Roberts evaluated 250 retired boxers in the United Kingdom and reported that central nervous system disorders were present in 17% of cases. Though 37 cases were reported, clinical and pathologic details were included in only 11, and a limited description of each was provided.

Four years later, Corsellis et al. published the first neuropathologic criteria for diagnosing CTE based on autopsies in 15 boxers. These criteria included cerebral atrophy, enlarged lateral/third ventricles, thinning of the corpus collosum, a cavum septum pellucidum with fenestrations, cerebellar scarring, and agyrophilic neurofibrillary degeneration. The specimens used by Corsellis were later reexamined by Roberts et al. [33] who found that nearly all had amyloid-β deposition, suggesting the possibility of AD. Gardner and colleagues continued to note that “classic” CTE does not represent a progressive disease with worsening stages [45]. Moreover, the early descriptions of “classic” CTE were limited by a lack of control for confounding variables, including psychiatric illness, family history variables, substance abuse, or genetic risk factors, limiting the generalizability of their findings.

Early modern definitions of CTE

In 2009, the first contemporary CTE staging system was proposed by McKee and colleagues based on autopsies of 85 individuals with repetitive mild traumatic brain injury (mTBI), 68 (80%) of whom were found to have CTE [43]. Microscopic examinations were performed by a single, blinded neuropathologist— McKee herself— and confirmed by two additional Boston University (BU) colleagues, Dr. Thor D. Stein and Dr. Victor E. Alvarez. The preliminary CTE definition used before the study began included: 1) perivascular foci of p-tau immunoreactive neurofibrillary tangles (NFTs) and astrocytic tangles (ATs); 2) irregular cortical distribution of p-tau immunoreactive NFTs and ATs at the depth of cerebral sulci; 3) clusters of subpial and periventricular NFTs in the cerebral cortex, diencephalon, basal ganglia, and brainstem; and 4) NFTs in the cerebral cortex preferentially in the superficial layers. After the study results, McKee and colleagues proposed four neuropathologic CTE Stages I-IV. An attempt was also made to match clinical symptoms with neuropathologic findings, with corresponding Stages I-IV representing worsening symptoms.

In 2011, Omalu et al. [9] described a distinctly different set of CTE guidelines from McKee’s, which included four phenotypes of CTE based on 17 autopsy findings— 8 professional football players, 4 wrestlers, 1 mixed martial arts fighter, 3 high school football players, and 1 boxer. Unlike McKee’s classification system, each phenotype was not meant to be an advanced form of the prior, and no clinical sign or symptom criteria were mentioned.

Current NINDS CTE neuropathology consensus guidelines

In 2013, the NIH and NFL funded an effort to define the pathological criteria of CTE and long-term outcomes of TBI, which in 2016 produced the first published consensus guidelines of CTE. At the outset, the primary objective was to determine whether CTE was a distinctive tauopathy that could be reliably distinguished from other tauopathies. A total of 25 cases of various tauopathies were selected by four neuropathologists not involved in grading: two from the BU group, one from the Mayo Clinic Jacksonville, and one from Mt. Sinai. A total of 7 neuropathologists experienced in neurodegenerative diseases participated in specimen examinations, including faculty from Washington University School of Medicine, Mayo Clinic Jacksonville, Brigham and Women’s Hospital, University of Washington School of Medicine, Uniformed Services University of Health Sciences, Boston University, and Columbia University. The neuropathologists rated each specimen as 1-unsure, 2-possible, 3-probably, or 4-definite after gross inspection. After initial evaluations, evaluators were then sent gross findings and clinical summaries for each case, and asked to reevaluate the diagnosis and provide a second level of conviction.

The expert panel produced an agreed upon neuropathologic definition of CTE that included: “pathognomonic lesion consist of p-tau aggregates in neurons, astrocytes, and cell processes around small vessels in an irregular pattern at the depths of the cortical sulci.” Several supportive neuropathological features of CTE were also included, along with mention of age-related p-tau astrogliopathy that may be present, which were non-diagnostic and non-supportive of a CTE diagnosis.

When assessing agreement for any tauopathy in the 25 cases presented at the consensus conference, the agreement level was 67%. When assessing agreement for a diagnosis of CTE specifically, the agreement level was 78%, meaning that 22% disagreed on a CTE diagnosis. Among non-CTE cases, more agreement was seen for AD, corticobasal degeneration, and primary age-related tauopathy, yet frequent discrepancies were seen with PSP and argyrophilic grain disease. When a clinical history and pathologic summary were provided, the degree of certainty rose from 3.1 to 3.7, meaning the clinical context made the evaluators more certain of the diagnosis. Comparing these inter-rater agreement values for CTE to AD, a 1997 study by Nagy et al. [51] evaluated a method to diagnose AD based on the sequential accumulation of NFTs in the cortex, without taking into account age, and reported a kappa of 0.90 among 41 brains. A similar study of AD by the same authors assessed the reliability of AD staging in thin paraffin sections and reported kappa values ranging from 0.6–0.8 for both the interrater and the intrarater reliability. Interestingly, a larger study concluded that absolute agreement was 91% when NFTs were substantial, but dropped to 50% in early stages of AD [52].

Soccer

The first identified case of cytoskeletal and pathological changes observed in a former amateur soccer player was presented by Geddes et al. [54], who identified pathological findings of neocortical NFTs and neuropil threads in the absence of amyloid-β protein. This finding occurred in 1999, prior to the consensus criteria established in 2016, thus the degree of accordance with the current pathological criteria (p-tau in sulcal depths) is unknown. Personal or medical histories were not provided for this case. Additionally, this athlete suffered a subdural hematoma resulting in death, which also influenced the neuropathologic findings and decreased the generalizability of the findings.

In the 2014 case series entitled “The Neuropathology of Sport,” McKee and colleagues [55] reported a single case of CTE identified in a 29-year-old semi-professional soccer player without a history of CTE. The athlete was diagnosed with amyotrophic lateral sclerosis (ALS) three months after initially presenting with fatigue and lower extremity/hand weakness at the age of 27 and died of respiratory insufficiency 21 months after diagnosis. CTE stage II with motor neuron disease was identified postmortem, though many of the pathological findings were consistent with changes seen in ALS (degeneration of the anterior horn cells).

The largest, non-football study of CTE involved 6 retired, male soccer players with dementia who were followed between 1980 and 2010 [56]. Of these six cases selected based on an a-priori criteria of cognitive decline (inclusion criteria of dementia), CTE pathology according to the latest consensus criteria was confirmed in four patients [44]. Across the six cases, clinical presentation varied widely and included mood/anxiety, gait/motor, and behavioral changes, with memory impairment recorded as the most prevalent. Similar to many of the studies highlighted above, limitations were inherent in the means of clinical data collection, as the authors relied solely on retrospectively collected collateral history obtained through relatives and medical records. Furthermore, the criteria used to determine behavioral, mood, and cognitive changes were not well-defined or provided. While repetitive head impacts were highlighted as the basis for the observed neuropathology in the series (all were described as “skilled headers of the ball”), findings regarding history of head injury were mixed. Two athletes independently sustained head-to-head collisions during play that resulted in loss of consciousness, and CTE was found in one of these athletes but not the other. Additionally, no CTE was found in a player who also participated in amateur boxing. Interestingly, all six cases exhibited AD pathology and aspects of age-related tau astrogliopathy. Further parallel pathology observed across the six cases included TDP-43 (n = 5/6), cerebral amyloid angiopathy (n = 5/6), hippocampal sclerosis (N = 2/6), corticobasal degeneration (N = 1), dementia with Lewy bodies (n = 1/6), and vascular pathology (n = 1/6). No “pure CTE” was seen. The various concomitant neuropathology combined with less than ideal clinical history collection further cloud interpretation of these findings.

The studies of CTE in soccer are sparse, but more common are studies that assess the impact of heading on long-term neurologic function. Several studies have failed to demonstrate a significant relationship between frequency of heading the ball and adverse cognitive sequelae across a range of time and developmental periods [57–63]. While most systematic or meta-analytic reviews have concluded that no evidence currently exists to suggest a relationship between heading and cognitive impairment, [64–66] the studies that reported a positive relationship were older, taken at a time when the sport used a heavier leather ball prone to water absorption, and included athletes with a history of alcohol abuse [67–70]. A recent systematic review by Tarnutzer and colleagues [71] concluded that the majority of studies reporting persistent impairment from heading were less likely to control for Type I error and select an appropriate control sample when compared with studies showing no impairment. Studies demonstrating an association between heading and cognitive deficits were also more likely to have lower quality methods of assessing and quantifying heading (e.g., retrospective query) than studies reporting no such correlation.

Rugby

Similar to soccer, the association between CTE pathology and rugby has been limited to case series [53]. In the aforementioned study by McKee and colleagues [55], a 77-year-old former rugby player presented with stage IV CTE and dementia. The athlete began playing rugby at the age of 13 years and played for 19 years; level of play was not mentioned. Macroscopic changes associated with severe atrophy and widespread p-tau immunoreactive neurofibrillary pathology and neuronal loss in diffuse cortical and subcortical areas was reported, yet histological staining for other neuropathology was not included. The athlete first exhibited cognitive difficulties in his mid-50s, including memory loss, executive dysfunction, and attention difficulties. This was reportedly followed by depression and anxiety, as well as behavioral changes of explosivity and impulsivity that reportedly progressed to physical/verbal abuse and paranoia by his mid-60s. The method in which these data were collected was not reported, nor was information regarding premorbid mood and behavioral problems, prior head injuries, family history, and substance abuse.

To date, a single pathologically confirmed comprehensive report of CTE in a former professional rugby player was provided by Stewart and colleagues [72]. The individual reportedly possessed a number of vascular risk factors, asthma, alcohol abuse, and “countless head injuries” as reported by family. No psychiatric history was provided. Onset of cognitive difficulties in attention, organization, and memory started at the age of 51 with positive neurological exam findings, including axial rigidity, asymmetrical upper limb rigidity, bradykinesia and ideomotor apraxia. Age-related white matter changes were noted on the MRI scan and a PET-scan revealed reduced accumulation of dopamine isotope in the caudate and putamen nuclei bilaterally, as well as evidence of frontal and posterior cingulate hypometabolism. Following death at the age of 57, widespread p-tau pathologies were recorded throughout the neocortex, hippocampus, and striatum. These findings were greatest in perivascular, subpial, and sulcal depth locations. An antemortem clinical diagnosis of PSP had been made, while postmortem pathological findings included classical Lewy bodies, PSP-like globose tangles, amyloid-related pathologies, and CTE pathology. Though the comorbidity of PSP does obscure the interpretation of clinicopathological interpretation, a strength of this case report is the inclusion of a comprehensive history and findings from clinical examination.

Studies of long-term outcomes in former rugby participation have been generally equivocal. In a retrospective study of 52 retired male Scottish international rugby players between the ages of 26 and 79 years (mean = 53.5) and 29 age-matched controls, McMillan et al. [73] failed to demonstrate significant differences in various health conditions, social or work functioning, and self-reported measures of mental health, which fell within the “normal range.” The retired rugby athlete group did exhibit poorer performance on a metric of verbal learning (Rey Auditory Verbal Learning Test) and dominant hand fine motor dexterity; however, there was no correlation between frequency of concussion (mean = 13.9) and cognitive functioning.

A second cross-sectional study examined long-term outcomes in 259 former elite, male rugby players (mean age = 60.1 + 16.1), as compared to population-based standardized morbidity rates (SMR) based on multiple population-based health-related surveys [74]. Former elite rugby athletes who averaged 22 years of sport exhibited similar long-term rates of dementia and depression following retirement compared to population-based SMR. A fair criticism of the study, as acknowledged by the authors, was the significant difference in age of the population-based reference group (older on average). However, these outcomes remained the same when controlling for age. In the same study, former elite rugby players were two times more likely to experience anxiety. Interestingly, 85% of former rugby athletes did not endorse experiencing anxiety and depression on a separate health-related quality of life instrument (EQ-5D-5L). The authors of the study attributed these mixed results to the timeframe of symptoms solicited; the significant findings asked questions of lifetime prevalence, whereas the non-significant findings asked about current symptomatology. Regardless, SMR were significantly higher in former elite rugby players, which raises suspicion. As part of the study, there was no discussion of CTE or neurodegenerative diagnoses other than “dementia.”

Regarding structural brain changes, Wojtowicz and colleagues [75] examined cortical and frontal thickness, as well as subcortical brain volumes in 24 relatively young (mean = 33.3 years-old), active, and retired professional rugby athletes, as compared to age- and education-matched controls. No between-group differences in psychiatric symptoms (depression and anxiety), cognitive functioning across a number of domains (with the exception of a single visual memory-related metric), or whole brain cortical and frontal lobe thickness were seen. Although rugby athletes did exhibit smaller bilateral hippocampi and left amygdala, follow-up analysis suggested that differences in subcortical nuclei volumes were likely attributable to alcohol use, especially in the rugby group.

Ice hockey

As part of a larger case series of individuals reportedly exposed to repetitive head mTBI [76], eight former hockey players (5 professional, 2 college/amateur, and 1 high school) were examined for CTE pathology. Of the eight, three were identified as exhibiting no CTE-related pathology and three were classified as possessing early disease stage pathology (Stage II). Of the two exhibiting more “progressed” pathology consistent with later stages of the disease (stage III and IV), both possessed multiple concomitant neuropathologies (e.g., diffuse amyloid-β, alpha synuclein, and neuritic amyloid-β plaques) and parallel neurodegenerative diagnoses (AD and Lewy body dementia (LBD)). Similar to the previously discussed soccer players, the concomitant pathology clouds interpretation of the relationship between hockey exposure and adverse outcomes. Furthermore, clinicopathological correlations (behavioral, cognitive, mood changes) of the former hockey players were not reported. This relationship between neuropathology and clinicopathological correlations is further confounded by the absence of CTE pathology during a recent autopsy of a 49-year-old retired professional hockey player who suffered from symptoms/conditions commonly said to be associated with CTE (i.e., self-reported memory loss, depression, and suicide) [77].

Investigations into long-term outcomes associated with participation in elite hockey have been limited. Esopenko et al. [78] examined cognitive and psychosocial functioning in 33 retired professional hockey players, with additional consideration of concussion exposure and genetic (apolipoprotein ɛ4) influence. Results revealed no significant group differences across a wide array of cognitive functions (i.e., attention, verbal memory, visuospatial functioning, inhibitory control, and reaction time). Group differences were observed on select measures of psychosocial factors comprised of multiple psychiatric and behavioral inventories, executive function and visual reasoning, with a significant association between greater concussion history and lower cognitive performance on these measures. Increased psychosocial disturbances were associated with the presence of the apolipoprotein ɛ4 allele. The degree to which differences were observed and further examination into how specific inventories differed was not performed.

Other limited-contact sports

The study of CTE in other common sports, such as baseball and basketball, is even more limited. A study of 66 individuals with sport exposure from a neurodegenerative disorder brain bank revealed the presence of Stage II CTE in a single former semi-pro baseball athlete. No information on baseball head impacts was provided [79]. AD-like (amyloid-β) pathology was also identified in this 87-year-old former semi-professional baseball. No antemortem cognitive or functional impairment was demonstrated at any point. CTE pathology was not observed in six other former baseball athletes (2 semi-professional, 1 college, and 3 high school level) with various other positive neuropathological findings in this sample. A single former high school basketball athlete was also identified in the sample as possessing stage I CTE, along with co-occurring ALS pathology. This is consistent with the established overlap of CTE and ALS pathology [75, 81]. Concussion history for the sample was not reported and both the former baseball and basketball athletes had a positive history of reported alcohol use, of which the degree of consumption was unknown.

In conclusion, of the few larger series of case studies, CTE has been identified postmortem in a select number of non-football cases. CTE in non-football athletes was significantly obscured by confounding neurodegenerative disorders, the presence of neuropathology in athletes who were asymptomatic at death, and methodological flaws, such as not controlling for long-standing medical conditions and poor sampling/data recording techniques (e.g., remote retrospective query of behaviors and cognitive functioning rather than objective measures). Findings demonstrating adverse long-term cognitive, behavioral, and emotional outcomes in non-football sports have been mixed, with better-designed studies much less likely to demonstrate negative outcomes.

Studies of CTE in football players

The earliest CTE study by the BU group was published in 2009 [43], where 5/51 (10%) of all CTE positive athletes were football players, and the remaining 46 were boxers. All 5 football players died in middle age (36–50 years) and played broadly similar positions (3 offensive linemen, 1 defensive lineman, 1 linebacker). Concussion information was obtained from next of kin and often included non-medically confirmed concussion history data. For example, one player’s wife reported 8 NFL concussions, though only one was medically confirmed. Most common was a diagnosed mood disorder (depression), along with memory loss, paranoia, poor judgment, anger, irritability, apathy, confusion, and poor concentration. Cause of death was reported for 4 of 5 football players and included suicide (2), high-speed police chase resulting in motor vehicle collision, and accidental gunshot wound. The authors concluded that there was “overwhelming evidence” that CTE was the result of repeated sub-lethal brain trauma that often occurred well before the development of clinical manifestations [43]. The authors also mentioned that pathologically, CTE shared features with AD, and that the Aβ and NFTs found in CTE are immunocytochemically identical to those found in AD, suggesting a possible common pathogenesis [43].

Four years later, a second study from the BU group reported on brain autopsies of 85 individuals with repetitive mild head impacts. In this sample, 68 (80%) tested positive for CTE based on the presence of p-tau [76]. Of the 68 CTE positive players, 50 played football— 35 professionally, 9 college, and 6 high school. The breakdown of CTE stage for the 35 football players included: 3 Stage I, 3 Stage II, 9 Stage III, and 7 Stage IV, with the remaining 11 having concomitant pathology and one player with no disease. Thirty-one of the 34 former professional football players had stage III–IV CTE or CTE plus co-morbid disease (89%). Positions played by NFL players positive for CTE included offensive linemen (26%), running backs (20%), defensive linemen (14%), linebackers (14%), quarterbacks (6%), defensive backs (6%), tight ends (6%), and wide receivers (6%). The authors concluded a positive correlation between years of football played and stage of CTE; yet number of concussions, years of education, and lifetime steroid use, all obtained secondarily from family, were not significantly associated with CTE stage. Moreover, none of the secondary analyses controlled for confounding variables, such as learning disorder, attention-deficit hyperactivity disorder, family history of neurocognitive disorders, or alcohol/substance abuse. Importantly, 17/50 had concomitant neurodegenerative diseases (AD, LBD, frontotemporal lobar degeneration, motor neurone disease, Pick’s disease, and PSP). The authors concluded that repetitive TBI “might trigger molecular pathways that result in the overproduction and aggregation of other proteins prone to pathological accumulation in neurodegenerative disease such as those listed above,” yet no justification was provided. The authors also cited literature showing that trauma was a risk factor for dementia, AD, ALS, and PD, yet these studies defined head injury as a hospitalized, closed head injury suffered during World War II, [82] lifetime head injury resulting in loss of consciousness and hospitalization, [83, 84] and one study assessed any traumatic incident, where only 7% (76/1131 participants) were concussions [85]. Perhaps most important, the study by Plassman et al. [82] of World War II veterans showed that mild TBI, irrespective of APOE ɛ4 allele status, was not a risk factor for dementia. Overall, these preliminary reports of CTE are exploratory in nature, and causal associations prove difficult to conclude given their nature as convenience samples, data collection methods, and comorbid pathologies.

In 2011, Omalu and colleagues [9] reported that 10 of 14 (71%) of professional male athletes (ages 18–52) were positive for CTE, including 7 of 8 professional football players and 1 of 3 high school football players. The single professional football player who tested negative for CTE was 24 years old. All three high school footballers died of acute brain and spine injuries suffered while playing. Importantly, the coexistence of severe TBI that resulted in these deaths has the potential to markedly affect pathological examination of these brains and diminish generalizability. No NFTs or neuropil threads (NTs) were identified in the 16- and 17-year-olds, but very sparse (1 to several) NFTs and NTs were seen in the cerebral cortex, subcortical nuclei/basal ganglia, hippocampus, and brainstem of the 18-year-old, which was interpreted as “incipient” CTE (abnormal tau protein). This individual had been playing football for approximately 6 years, but no other biopsychosocial or demographic information was provided. Though not separated into football players only, the authors noted that alcohol- and drug-related deaths were overrepresented in the CTE cohort, and AD-type atrophy was absent in all cases. The authors also cautioned against diagnosing CTE in the elderly (>65 years) to avoid confusing CTE changes with AD, normal aging, and chronic ischemic changes due to small vessel disease, in addition to the fact that amyloid may be found in cognitive normal elderly individuals [86]. Interestingly, the authors concluded by postulating four emerging and recurring histomorphologic CTE phenotypes based on the presence or absence of NFTs, NTs, and diffuse amyloid plaques in the cerebral cortex, subcortical nuclei/basal ganglia, hippocampus, and cerebellum, as well as the topographic distribution and predominance of NFTs, NTs, and amyloid plaques in those locations [9]. As stated above, this description of CTE differed from that proposed by McKee in 2013 [76] and the subsequent NINDS classification [44]. Omalu [9] did not find accumulation of tau-immunoreactive astrocytes, and the authors proposed that this difference may be due to difference in sport played (boxing versus football).

In 2013, Hazarati et al. [87] reported on a convenience sample of six retired Canadian Football League players, all of whom manifested progressive neurodegeneration prior to death. Three of 6 (50%) tested positive for CTE, while the remaining three were diagnosed with AD, ALS, and PD, respectively. The 3 CTE positive cases had comorbid pathology indicative of cancer, AD, and vascular disease. The authors concluded that the comorbid pathological findings may have contributed to the neurologic decline and that caution should be utilized in the diagnosis of CTE when other neurodegenerative diseases are present. The study by Hazrati and colleagues supports the notion that not all athletes with a history of repetitive head trauma developed CTE, and CTE alone is not responsible for any neurologic decline.

The largest study of football players and CTE was published in 2017 by Mez and colleagues [2] and was the most read article of JAMA in 2017 [88]. In a study of 202 former football players whose next of kin voluntarily donated their brains to the BU brain bank, 177 brains (87%) were positive for CTE, including 0/2 pre-high school, 3/14 high school (21%), 48/53 college (91%), 9/14 semi-professional (64%), 7/8 CFL (88%), and 110/111 NFL (99%) players. The diagnosis of CTE was divided into two principle categories, each made up of two stages: Mild CTE (n = 44), including Stage I (1 or 2 lesions) and Stage II (3 or more lesions). Severe CTE (n = 133) included Stage III (“multiple” lesions) and Stage IV (“densely distributed” lesions). The authors acknowledged their convenience sample of deceased football players, and postulated that it was unclear whether concussions or subconcussive impacts were responsible for the high proportion of CTE.

Several aspects of the study limit interpretation of the findings. First, ascertainment bias (families of players voluntarily donated their brains to comprise the study population) resulted in a biased sample that inaccurately estimates the prevalence of CTE. Quite accurately, the authors mentioned that 99% of this convenience sample of former NFL players had CTE. However, the conclusion was apparently misinterpreted by media outlets as, “99% of NFL players had CTE.” Second, all NFL concussion and cognitive data were collected through family and next of kin on a postmortem basis. Recall bias has been shown to affect an individual athlete’s ability to recall head injury information at the youth level, let alone in retired athletes [89]. Asking family members about the careers of retired players that took place 20–30 years prior may falsely represent specific facts about one’s football playing career, as well as the onset, degree, and progression of clinical presentation/symptoms. Third, the study population from the BU brain bank was not representative of the American football population, limiting generalizability of the results. Fourth, substance use disorders were noted in 2/3 of the study population with mild or severe CTE. Not only can substance use cause neurologic injury and decline, but opiate abuse can also cause abnormal p-tau deposition, [90–92] and opiate abuse has been noted to be common among NFL retirees [93]. Fifth, whereas previous studies have utilized comparison groups with good control selection, no such comparison of all individuals exposed to football was present.

Sixth, and perhaps most important, the minimum threshold for CTE diagnosis consisted of a single p-tau lesion at the depths of a sulcus in the cerebral cortex, reportedly distinct from the lesions of aging-related tau astrogliopathy. Although this criterion is the solitary one postulated by the 2016 consensus criteria for a neuropathologic diagnosis of CTE, this low threshold maximizes sensitivity at the cost of specificity [94–96]. Whereas known neuropathologic diagnoses such as AD and LBD require many lesions to be present per a certain area, no such criteria exists for CTE. Moreover, in the Severe CTE group, accumulations of amyloid-β, a-synuclein, and TDP-43 were common, in addition to other comorbid neurodegenerative diseases, including AD, LBD, or frontotemporal dementia. It is possible that the p-tau seen were associated with these comorbid neuropathologies, and not CTE.

Reviews

Three systematic reviews on CTE in sports have been published; two exclusively on CTE in sports [45, 97]. Though many excellent narrative reviews have been published, for the purposes of this chapter, only systematic reviews will be discussed.

In 2014, Gardner, Iverson, and McCrory summarized the 85 cases of CTE published to date, [9, 76] drawing from 158 autopsy cases examined for CTE. First, the authors described the previously mentioned dichotomy of older, ‘classic’ description of CTE, which differed significantly from the ‘modern’ syndrome in the age of onset, natural history, clinical symptoms, and pathologic diagnostic criteria. The specific characteristics of modern CTE appeared to be the location of the neuropathology, in the gray matter and perivascular space at the depths of sulci, rather than the specific protein or lesion type [45]. Of the 85 cases of autopsies that have been performed in athletes over the past 10 years, 20% had ‘pure’ neuropathology consistent with CTE only, 52% had CTE plus other neuropathology, and 29% had no neuropathology or other neuropathology. The authors concluded that the strongly held causal assumptions relating to concussive and subconcussive brain impact exposure derived from the case studies were, “scientifically premature,” especially given the absence of cross-sectional, epidemiological, prospective studies on the topic [45]. Furthermore, they summarized five important methodologic concerns when interpreting this literature: 1) the neuropathologic impact of drug/steroid abuse, alcohol abuse, psychiatric problems, cardiovascular/cerebrovascular disease is unknown; 2) the degree to which neuropathologic findings contribute to the clinical features witnessed; 3) genetic assessment of these individuals remains incomplete; 4) moderating and confounding variables between neuropathologic findings and neurologic symptoms are not known; and 5) a denominator of those at risk for CTE to accurately assess prevalence has not been adequately determined.

The second systematic review of CTE was conducted by Maroon and colleagues [97], which included a total of 153 neuropathologically confirmed cases of CTE in contact sports, 63 of which were football players. Of the 63 football players, substance abuse was seen in 9 players (14%). The authors concluded that the true incidence of CTE is unknown due to inconsistent definitions, a lack of longitudinal studies, and significant overlapping symptoms with common neurodegenerative disease. Furthermore, while McKee et al. [76] postulated that pathological findings in CTE were correlated with the numbers of years of football, three of seven high school football players in their review had CTE prior to the age of 20. It was also noted that a lower proportion of substance abuse was seen in football players compared to non-football players (14% versus 21%); however, the impact of substance abuse, namely opioid abuse in former NFL players, could not be determined as many studies did not assess these data points. The authors concluded that given the many limitations in the literature, conclusions regarding the prevalence of CTE among football and other contact sports cannot yet be determined.

Lastly, as part of the 2016 Concussion In Sport Group meeting in Berlin, Manley and colleagues [48] produced a comprehensive systematic review on potential long-term effects of SRC, of which CTE was discussed extensively. The authors correctly mentioned that prior to 2015, no agreed upon neuropathological criteria existed for identifying CTE, [44] and that the new consensus criteria excludes tauopathies associated with aging, such as age-related tauopathy [98] and age-related tau astrogliopathy [99]. Furthermore, tauopathy, amyloid-β, alpha-synuclein, and TDP-43 positive immunoreactivity occurs with normal aging, other neurodegenerative diseases, and in those with normal cognition [100]. After the exhaustive review, the international experts concluded, “A cause and effect relationship between CTE and concussions or exposure to contact sports has not been established,” and additional case-control, cohort, and longitudinal studies are needed to understand the incidence, prevalence, extent to which the neuropathological findings cause specific clinical symptoms, the progressive nature of the disease, clinical diagnostic criteria, and other risk or protective factors.

Epidemiological studies

Though professional football players have been well studied, these elite athletes are not representative of the general population. Several epidemiologic investigations of high school football players and the subsequent development of neurodegenerative disorders have yielded important results, the most notable of which will be discussed here [79, 101–103].

In the largest population-based neuropathologic study with an outcome of pathologically confirmed CTE, Bieniek et al. [79] reported brain autopsy findings in 66 former athletes. In these former athletes, 21/66 (32%) had CTE, and only 1/66 (5%) had “pure” CTE, whereas the remaining 20 displayed mixed neuropathology with concomitant neurodegenerative diseases. Taken from the Mayo Clinic brain bank in Jacksonville, FL, inclusion criteria were presence of paraffin-embedded tissue, at least minimal medical documentation, and male sex, and subjects were excluded if they had a known neuropathologic diagnosis of PSP, corticobasal degeneration, Pick’s disease, and a mutation in the microtubule associated protein tau gene (MAPT). Of note, cases of comorbid dementia, AD, or LBD were not excluded. Of the 1721 male remaining cases, 66 had a history of exposure to contact sports, and controls were men without documented exposure to contact sport. Within the contact sport exposure group, a total of 21/66 individuals (32%) tested positive for CTE, with 7 cases were classified as CTE Stage I, 7 CTE Stage II, 5 CTE Stage III, and 2 CTE Stage IV. Of the 43 patients with exposure to contact sports through football (football only and multiple sports), 16 (37%) had CTE pathology. Of these, 6 played up to high school level, 7 played to college, and 1 played professional football. In the 27 former football players without evidence of CTE pathology, 15 played up to the high school, 7 played up to college, and none played professional football. There were no significant differences in the proportion of football players with CTE pathology with respect to level of involvement (high school, college, professional; p = 0.175). On the 198 disease- and age-matched controls, no CTE pathology was seen, even in 33 controls with documented head trauma not related to contact spots. In terms of comorbid pathology, 20 of 21 (95%) confirmed CTE diagnoses met criteria for other neurodegenerative conditions. Overall, this study is important in that it serves as one of the few large scale, epidemiologic efforts to assess the prevalence of CTE. It should be noted that the rate of contact sports was exceedingly low, at 66/1721 (3.8%) male subjects, which raises concerns regarding the representativeness of the sample and results. A 2010 report concluded that more than 7.6 million students played high school sports, approximately 55.5% of all high school students [104]. Though the generalizability of these results is greater than the aforementioned studies of elite football players and athletes, the sample for the study was derived from brain bank predominantly dedicated to the collection of brains diagnosed with neurodegenerative disorders in the living, and in mostly Caucasians living in the Southeastern U.S. Furthermore, any descriptive assessment of head trauma or presentation/symptoms in the affected individuals could not be evaluated, creating ambiguity regarding the clinical meaningfulness of pathological findings.

In a study of high school football players from Rochester, MN from 1946–1956, Savica et al. [102] compared football players to a control group comprised of male students in the band, glee club, or choir. Using the records-linkage system of the Rochester Epidemiology Project from 2010-2011, the authors recorded later development of dementia, PD, or ALS and compared disease frequencies to the general population of Olmsted County, MN. The authors reported no increased risk of dementia, PD, or ALS among the 438 football players compared with the 140 non-football-playing male classmates. Compared to the general population, only PD was significantly increased; however, this was true for both groups, with a 2.4x greater risk for the football playing group and 5x greater for controls. Five years later, a follow-up study over a 40-year time period by Janssen et al. [101] of 296 high school football players compared to 90 swimmers, wrestlers, and basketball players from 1956–1970, assessing for the presence of dementia/mild cognitive impairment, ALS, and PD. The results indicated that varsity high school football players did not have an increased risk of neurodegenerative diseases compared with athletes engaged in other varsity sports.

The third long-term epidemiologic study estimated the association of playing high school football with cognitive impairment and depression at 65 years of age. After exclusions for missing data, Desphande et al. [103] studied 2,692 men, 834 (31%) of whom played football compared to 1,858 (69%) who did not. Two primary outcomes included a measure of cognitive impairment with a composite cognition measure of Letter Fluency and Delayed Word Recall. Depression was measured with the WLS-modified Center for Epidemiological Studies Depression Scale. After an extensive 1:1 matching process of baseline covariates, including adolescent IQ, family background, education level, and three different control conditions (all controls, non-collision sports, and no sports), playing football was not statistically associated with a reduced composite cognition score or increased depression score. Furthermore, after adjustment for multiple statistical tests, playing football did not have a significant adverse association with any of the secondary outcomes, such as the likelihood of heavy alcohol use at 65 years of age. Though other well-known neurodegenerative disorders were studied rather than CTE, which is understandable given its infancy and brain autopsy is the only means of diagnosis, these well done, prospective epidemiologic studies suggest no increased risk of neurodegenerative disorders or clinicopathological symptoms (e.g., cognitive impairment and depression) supposedly associated with CTE among high school football players compared to controls and the general population.

Additional studies

Though not specifically related to CTE or neurodegenerative disorders, four additional studies of American football players deserve mention [13, 105–107]. Baron studied 334 deceased former NFL players from 1959–1988 and concluded the rates of psychiatric illness and suicide were lower for former NFL players compared to the general population, and though the rate of neurodegenerative diseases was slightly higher than the general population (3.6% versus 2.9%), this was not a statistically significant difference. Using the same dataset of 334 deceased NFL players, Lehman et al. [106] found that overall player mortality compared to that of the US population was reduced, but neurodegenerative mortality was increased, including ALS and AD (NFL players were 3 times higher than that of the general US population). Two years later, the authors reported on 537 deceased NFL players and revealed that suicide among NFL players was significantly less than the general population (2.2% versus 4.8%), with no difference between speed and non-speed position players. Further, on the topic of suicide, Webner and Iverson [108] summarized 26 professional football players who committed suicide, with most deaths occurring in the last 15 years. The authors noted that most of the men suffered from multiple life stressors prior to their deaths, including retirement, loss of income, divorce, failed business ventures, family member estrangement, and substance abuse disorders, highlighting the multi-factorial nature of suicide in former NFL athletes.

Most recently in 2018, Venkataramani and colleagues [105] retrospectively studied mortality risk in 2,933 regular NFL players and 879 replacement players, who debuted between 1982–1992. A total of 144 NFL players (4.9%) and 37 replacement players (4.2%) died, and no statistically significant increase in mortality was associated with career players versus replacement players (HR 1.38, 95% CI, 0.95 to 1.99; p = 0.09). Neurodegenerative disorders were responsible for only 7% of deaths in the career players, which ranked 7th behind cardiometabolic (35%), transportation injuries (14%), unintentional injuries (10%), and others. Of note, all 7 deaths were due to ALS, and none due to CTE.

How common is p-tau?

Recent focus has been directed toward the question of whether p-tau is unique to CTE. Multiple studies have previously demonstrated “CTE-like” p-tau pathology co-morbidly in a number of individuals without a history of head trauma, which include temporal lobe epilepsy [109, 110], ALS [111], the general population [112], and multiple systems atrophy [113]. Ultimately, this creates uncertainty around the specificity of the diagnostic consensus criteria put forth by the NINDS/NIBIB, which includes p-tau aggregates in neurons, astrocytes, and cell processes around small vessels in an irregular pattern at the depths of the cortical sulci [44]. Specificity of the criteria is called into question further based studies that have demonstrated head injury was not an independent predictor of p-tau pathology, when accounting for other comorbid pathologies or processes, such as ALS [80]. This is supported further by Noy et al. [112], who demonstrated that only a combination of head injury and alcohol and/or drug abuse, as well as age, were significant predictors of p-tau pathology in autopsies performed prospectively at a routine neuropathology service.

Lack of sociodemographic information

A crucial feature in studies on CTE to date is the paucity information regarding past psychiatric, socioeconomic, substance use, medical and family histories of the individuals examined. Whereas these demographic factors are routinely controlled in studies of SRC and TBI outcomes, they are largely missing due to the rarity and small samples of CTE studies. It has been empirically established that multiple factors have the potential to impact the development and presentation of CTE. Without rigorously controlling for such crucial biopsychosocial information, including age [114, 115], substance use [90, 117], psychiatric history [115], cardiovascular/cerebrovascular disease, or other health conditions, interpretation of any findings may be seriously compromised. In their systematic review, Maroon et al. [97] noted that very few studies took into account drug and substance use/abuse histories. Furthermore, study populations were homogenous and suffered from low numbers, making it difficult to draw generalizable information applicable to other populations. That said, some studies have made an effort to measure and control for confounding variables. In largest study of CTE to date, substance use disorders, suicidality, and family history of psychiatric illness were seen in 32 (67%), 22 (47%), and 23 (49%) cases, respectively [2]. Outside of this, few reports have considered demographics and comorbidities. Several authors have cautioned against over concluding without consideration of these important modifiers. Iverson and colleagues [46] contrasted CTE to AD, that when low levels of AD neuropathology are seen with major cognitive impairment, it is possible that other comorbid diseases may substantially contribute to the clinical decline.

The importance of comorbid pathology

A common theme among the previously mentioned studies is the concomitant neuropathology that often accompanies a CTE diagnosis, including LBD and AD, among others. In some of the more notable studies, the following rates of comorbid pathology were found: 32 of 111 in Mez et al. [2], 20 of 21 by Bienek et al. [79], and 3 of 3 by Hazrati et al. [87] The review by Gardner and colleagues concluded that 20% of reports had “pure” CTE and 52% had CTE plus other neuropathology. Given the prevalence of comorbid diagnoses, what can we conclude about the high rate of comorbid pathology?

One interpretation is that the relationship between contact sport and neurodegeneration is strengthened by multiple neuropathologic entities seen together. A cascade of neuropathologic abnormalities has been noted, contributing to a neurologic decline and possibly premature death. However, an alternative explanation questions the existence and importance of CTE— is it possible that the p-tau in cortical sulci depths is just an artifact of normal aging and/or part of the other comorbid neuropathologic diseases, and not unique to CTE? The truth remains elusive, but several authors have weighed in. In 2011, Omalu et al. [9] acknowledged the coexistence of CTE and other pathologic findings, and recommended that, “these additional findings be enumerated as separate diagnoses accompanying CTE.” The same authors go on to caution the diagnosis of CTE in those older than 65 years, “to avoid confusing CTE changes with AD pathology, normal age related, changes in the brain, and/or chronic ischemic changes/small vessel disease changes in the brain,” and that NFTs and amyloid plaque, “may be found in low densities … in cognitively normal elderly individuals [9]. Hazrati et al. [87] and Gardner et al. [45] emphasized that the comorbid pathology may have contributed to the clinical signs and symptoms witnessed prior to death, and that differentiation of which pathologic diagnoses was responsible for clinical decline cannot be determined. As with most controversial aspects of CTE, research remains ongoing.