Case reports describing the neuropathology of CTE initially appeared in the 1950s and 1960s [24, 41, 72, 87, 99]. In 1973, Corsellis, Bruton, and Freeman-Browne described the neuropathological features in a series of 15 former male boxers, ranging in age from 57 to 91 years [25]. Corsellis et al. described macroscopic changes of cerebral atrophy, enlargement of the lateral and third ventricles, thinning of the corpus callosum, cavum septum pellucidum with fenestrations, and scarring of the cerebellar tonsils. Microscopically, using the histological methods available at the time, namely cresyl violet, von Braunmühl’s silver stain, and Kings amyloid stain, they reported sparse argyrophilic neurofibrillary tangles (NFTs) in the cerebral cortex and substantia nigra along with senile plaques in 20% of cases. Subsequent re-examination of Corsellis’ original series of boxers using beta-amyloid (Aβ) immunohistochemistry determined that 95% of cases had diffuse Aβ deposits [117]. Re-examination of the original cases in 2018 using hyperphosphorylated tau (p-tau) immunohistochemistry and applying 2016 NINDS–NIBIB consensus criteria found that 50% met criteria for CTE; however, prolonged formalin fixation and limited tissue availability might have affected the findings [39].
In 1991, Hof et al. described the neuropathology of a young autistic patient with repetitive head-banging behaviors using thioflavin and Gallyas silver methods, a silver technique more sensitive to argyrophilic inclusions than other silver methods [47, 60]. They described perivascular clusters of thioflavin and Gallyas positive NFTs and neurites at the depths of the sulci in the inferior temporal, entorhinal, and perirhinal cortices, in the absence of Aβ plaques [47]. They also quantitatively demonstrated the superficial distribution of the NFTs in layers II and III, a laminar predilection not found in Alzheimer’s disease (AD) [46]. In 1996, using AT8 immunohistochemistry, Geddes et al. described patchy, perivascular NFTs in the brain of a 23-year-old boxer [37]. They later compared the immunohistochemical findings of five young men, ranging from 23 to 28 years old, including the young boxer [36]. The men were exposed to RHI from head banging, poorly controlled epilepsy, rugby, and boxing. They described perivascular, p-tau-positive cortical NFTs and neuropil threads and noted that the pathology principally involved the depths of sulci [36]. No Aβ deposits were evident. Of the 21 age-matched controls they also examined, none showed a similar pathology.
Omalu et al. were the first to report the neuropathological findings of an American National Football League (NFL) player who had died at age 50 after experiencing cognitive impairment, mood changes, and parkinsonian symptoms [97]. The authors described mild, non-specific, tau pathology, consisting of sparse neocortical and locus coeruleus (LC) NFTs alongside diffuse Aβ plaques. In their second NFL case, sparse to frequent NFTs were found in the frontal and temporal cortices, diencephalon, and brainstem [96]. There was no mention of a perivascular or superficial pattern to the tau pathology, nor predilection for sulcal depths, features considered characteristic of CTE by Hof and Geddes. Although not highlighted by the authors, photomicrographs of the case showed clusters of perivascular p-tau pathology now considered pathognomonic for CTE [74]. P-tau pathology was also reported in a 40-year-old professional wrestler who died by suicide [98]. Sparse to frequent NFTs and neuropil threads in the cortex, subcortical ganglia, and brainstem nuclei were interpreted as CTE.
In 2009, McKee et al. detailed the neuropathological findings of two former boxers and a former NFL player and conducted a systematic review of all 48 previous neuropathologically verified cases of CTE in the world’s literature [75]. Using large format 50 µm sections and CP-13 and AT8 immunostaining to illustrate the regional p-tau pathology, the authors emphasized the distinctive p-tau pathology of CTE including cellular and regional abnormalities not previously detailed. The authors stressed the irregular, patchy distribution of NFTs, the prominent perivascular pattern, and the preferential involvement of cortical laminae II and III of the cortex. The authors also noted the predilection for the depths of the cortical sulci in the frontal, temporal, parietal, insular, and septal cortices with sparing of primary visual cortex. In addition, the authors described astrocytic p-tau inclusions, "astrocytic tangles" in subpial regions and around small blood vessels, and dot-like and spindle-shaped neurites. Dense NFTs, ghost tangles, and neurites were present in the hippocampus, entorhinal and transentorhinal cortices, amygdala, nucleus basalis of Meynert, hypothalamic nuclei, mammillary bodies, olfactory bulb, thalamus, substantia nigra pars compacta, dorsal and median raphe nuclei, and LC. The subcortical white matter, especially the subcortical U-fibers, corpus callosum, internal, external, and extreme capsules, fornix, and mammillothalamic tracts contained p-tau neurites, although the white matter was less affected than gray matter [75].
The neuropathological findings of 17 athletes ranging in age at death from 18 to 50 years were reported by Omalu et al. in 2011 [94]. CTE was diagnosed in seven of eight (88%) football players, two of four (50%) professional wrestlers, and one boxer. The p-tau pathology was described as sparse, moderate, or frequent, and NFTs ranged from band shaped or flame shaped to globose. Diffuse amyloid plaques were found in two cases. Omalu proposed pathological criteria for the diagnosis of CTE as four “emerging phenotypes" based on the presence or absence of NFTs and neuritic threads in the cerebral cortex, subcortical nuclei, and brainstem, with or without diffuse Aβ plaques. In the description of the phenotypes, there was no mention of specific morphological, cellular, or regional features to distinguish CTE p-tau pathology from AD, primary age-related tauopathy (PART), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), or other tauopathies.
The first military veteran with CTE was reported in 2011 [95]. Omalu et al. described a 27-year-old Iraqi war veteran who experienced combat and was exposed to improvised explosive device blasts. Neuropathological examination revealed multifocal, cortical, and subcortical NFTs and neuritic threads, accentuated in the frontal cortex and at the depths of the sulci. In 2012, Goldstein et al. described the neuropathological features found in four male military veterans with known blast exposure or concussive injury (ages 22–45 years; mean, 32.3 years), three amateur American football players and a professional wrestler (ages 17–27 years; mean, 20.8 years), and four normal controls without a history of RHI (ages 18–24 years; mean, 20.5 years) [40]. In the young military veterans and athletes, perivascular foci of p-tau NFTs were found in the frontal cortices with a predilection for sulcal depths. None of the controls demonstrated p-tau pathology.
In 2013, McKee et al. reported the findings in a series of 85 subjects, all but one male, ranging in age from 17 to 98 years, mean 59.5 years, with a history of repetitive head impacts (RHI) from American football, ice hockey, boxing, military service, and head-banging behaviors, and compared them to 18 age- and gender-matched controls without a history of RHI [79]. They described a distinctive pattern of p-tau pathology in 68 men that they considered diagnostic for CTE and proposed the McKee criteria for the pathological diagnosis of CTE. The diagnosis of CTE required: (1) the presence of perivascular foci of p-tau NFTs and astrocytic tangles; (2) an irregular cortical distribution of p-tau NFTs and astrocytic tangles with a predilection for the depth of cerebral sulci; (3) clusters of subpial and periventricular astrocytic tangles in the cerebral cortex, diencephalon, basal ganglia, and brainstem; and (4) NFTs in the cerebral cortex located preferentially in the superficial layers.
Among those diagnosed with CTE, the spectrum of p-tau pathology ranged in severity from sparse perivascular foci of NFTs in the cortex, usually the frontal cortex, to a severe, widespread tauopathy affecting the medial temporal lobe, thalamus, hypothalamus, mammillary bodies, basal ganglia, brainstem, cerebellum, and white matter tracts. The p-tau pathology followed an ordered, hierarchical progression of severity, prompting the authors to propose a staging scheme: McKee stages I–IV. Stage I CTE was characterized by isolated cortical clusters of perivascular NFTs, astrocytic tangles, and dot-like neurites (i.e., “CTE lesions”), most prominent at the depths of the sulci, typically affecting the dorsolateral frontal cortices. In stage II CTE, multiple CTE lesions were found in the frontal, temporal, and parietal cortices with sparse NFTs in the superficial cortical layers, LC, and nucleus basalis of Meynert. Stage III CTE was characterized by multiple focal cortical CTE lesions, NFTs in the superficial cortical laminae and NFTs in widespread cortical areas, hippocampus, entorhinal cortex, amygdala, nucleus basalis of Meynert, substantia nigra, dorsal and median raphe, LC, and olfactory bulbs. In stage IV CTE, NFTs and CTE lesions were densely distributed throughout the cerebral cortex, with severe neurofibrillary degeneration and ghost tangles in the medial temporal lobe structures, and NFTs in the thalamus, mammillary bodies, nucleus basalis of Meynert, substantia nigra, dorsal and median raphe, LC, basis pontis and cerebellar dentate nucleus, often with neuronal loss and gliosis. The p-tau pathology consisted of three-repeat (3R) and four-repeat (4R) tau; the astrocytic tangles were predominantly 4R tau.
The stages of CTE were associated with progressive increases in macroscopic abnormalities: typically mild dilatation of the frontal horn of the lateral ventricles or third ventricle with cavum septum pellucidum in stages I and II, evolving to mild cerebral atrophy, depigmentation of the LC and substantia nigra in stage III, and severe atrophy of the frontal, temporal, and medial temporal lobes, a sharply concave contour of the third ventricle, cavum septum pellucidum and septal perforations, thalamic atrophy, a sharply convex contour of the medial thalamus ("thalamic notch"), thinning of the hypothalamic floor, atrophy of mammillary bodies, depigmentation of the LC and substantia nigra, and thinning of the posterior corpus callosum in stage IV.
Comorbid neurodegenerative disease was present in 25 of the 68 CTE cases (36.8%) including AD, Lewy body disease (LBD), frontotemporal lobar degeneration (FTLD)-TAR DNA-binding protein (TDP), progressive supranuclear palsy (PSP) and Pick’s disease. Aβ deposition was found in 44.1%, primarily as diffuse plaques, and was significantly associated with age at death.
Although not well characterized at the time, McKee et al. described a form of astrocytic p-tau pathology in CTE, now termed age-related tau astrogliopathy (ARTAG), consisting of 4R immunoreactive ‘thorn-shaped astrocytes’ (TSA) in the subpial, periventricular, and perivascular white matter, and less frequently in the gray matter. TSA had been previously described in the aging brain [48, 59, 61], although at the time of the 2013 publication, it was unclear whether the subpial and periventricular TSA and astrocytic tangles found in CTE were unique to CTE or a form of ARTAG. Subsequently, an international group of neuropathologists led by Gabor Kovacs defined the many p-tau astrocytic morphologies of ARTAG, resulting in improved discrimination between pathologies considered typical of ARTAG and those specific to CTE [58].
In 2013, Hazrati et al. reported the clinical and pathological findings in six former Canadian Football League players [44]. Three of the six (50%) cases had neuropathological findings diagnostic for CTE, defined as NFTs and astrocytic tangles in a patchy, perivascular distribution, localized to the depths of sulci, subpial areas, and the superficial cortical layers (layers II/III). In 2015, Bieniek et al. reviewed available medical records of 1,721 men in the Mayo Clinic neurodegenerative disease-focused brain bank for evidence of a history of traumatic brain injury (TBI) or participation in contact sports [12]. They identified 66 cases with a documented history of sports exposure and subsequently processed frontal and parietal cortical tissue samples for p-tau immunohistochemistry. Tissue samples in 198 age- and disease-matched men and women without contact sports exposure were similarly processed. Of the 66 men with exposure to contact sports, 21 (32%) had p-tau pathology consistent with CTE characterized by perivascular foci of TSA, NFTs, and threads at the depths of cortical sulci. Seven cases were classified as CTE stage I, seven as CTE stage II, five as CTE stage III, and two as CTE stage IV. The 198 controls without contact sports exposure showed no CTE pathology, including 33 individuals with single-incident TBI. The authors also drew attention to ARTAG pathology, most marked in two boxers, and commented that ARTAG must be considered in the differential diagnosis of CTE, especially in those with advanced age.
In 2014, the NINDS–NIBIB organized a consensus meeting to establish the neuropathological criteria for the diagnosis of CTE using the McKee criteria for CTE as a starting point [74, 79]. By the time the consensus panel met in November 2015, ARTAG was recognized as an age-related p-tau pathology characterized by multiple patterns of astrocytic p-tau, including subpial TSA that are commonly found in CTE. Several members of the consensus panel were involved in developing the harmonization criteria for ARTAG [58]. Recognizing that ARTAG pathology is frequently found in CTE but is non-diagnostic, the criteria for the diagnosis of CTE used by the consensus panel were modified to include: perivascular foci of p-tau NFTs and astrocytic tangles in the cortex; irregular clusters of p-tau NFTs and astrocytic tangles found preferentially at the sulcal depths; NFTs in the cerebral cortex located primarily in the superficial layers. Subpial TSA, or ARTAG, were considered only supportive, non-diagnostic, features of CTE. With these modified CTE criteria, a panel of expert neuropathologists evaluated 25 cases of various tauopathies blinded to all clinical, demographic, and gross neuropathological information. The 25 cases included 10 cases of suspected CTE, and 15 other cases, including AD, PSP, PART, argyrophilic grain disease, CBD, and parkinsonism dementia complex of Guam. A single laboratory processed all cases uniformly and the resulting slides were scanned into digital images that were provided to neuropathologists blinded to all other information. The neuropathologists submitted their independent evaluations prior to meeting in person. The panel found that the criteria reliably distinguished CTE from the other tauopathies and established the preliminary NINDS–NIBIB criteria for the pathological diagnosis of CTE. Importantly, the panel defined a pathognomonic lesion of CTE as the "accumulation of abnormally phosphorylated tau in neurons and astroglia distributed around small blood vessels at the depths of cortical sulci and in an irregular pattern." (Fig. 1) They further observed that the p-tau neurites in CTE were often dot-like, that TDP-43-immunoreactive inclusions in CTE were distinctive, and the pattern of hippocampal neurofibrillary degeneration was unlike AD. They also made recommendations for the diagnosis and evaluation process of potential CTE cases [74].
Fig. 1 The pathognomonic lesion of CTE and the staging schemes of pathological severity (adapted with permission from [83]). Representative images of p-tau pathology at Low and High chronic traumatic encephalopathy (CTE) pathological stage using the abbreviated staging scheme recommended by the second NINDS/NIBIB consensus panel (low–high) [11] and the McKee staging scheme (I–IV) [4, 79]. Low CTE is characterized by p-tau pathology restricted to focal cortical lesions. High CTE shows widespread p-tau pathology in the medial temporal lobe structures and diencephalon in addition to focal cortical lesions. McKee Stage I CTE is characterized by one or two isolated CTE lesions at the depths of the cortical sulci. In stage II, three or more cortical CTE lesions are found. In stage III CTE, there are multiple CTE lesions and diffuse NFTs in the medial temporal lobe. In stage IV CTE, CTE lesions and NFTs are widely distributed throughout the cerebral cortex, diencephalon, and brainstem. Top row: hemispheric 50-µm tissue sections immunostained with CP-13, directed against phosphoserine 202 of tau (courtesy of Peter Davies, Ph.D., Feinstein Institute for Medical Research; 1:200); positive p-tau immunostaining appears dark brown. Bottom row: 10-µm paraffin-embedded tissue sections immunostained for phosphorylated tau (AT8) (Pierce Endogen). Positive p-tau immunostaining appears dark red, hematoxylin counterstain. I A 26-year-old former college football player with stage I CTE (Low). Two perivascular p-tau CTE lesions are evident at the sulcal depths of the frontal cortex; there is no neurofibrillary degeneration in the medial temporal lobe. II A 49-year-old former NFL player with stage II CTE (Low). There are multiple perivascular p-tau CTE lesions at depths of sulci of the frontal cortex; there is no neurofibrillary degeneration in the amygdala or entorhinal cortex. III A 53-year-old former NFL player with stage III CTE (High). There are multiple CTE lesions in the frontal cortex and insula; there is diffuse neurofibrillary degeneration of hippocampus and entorhinal cortex (asterisk). IV A 62-year-old former NFL player with stage IV CTE (High). There are multiple CTE lesions in the cerebral cortex with widespread neurofibrillary degeneration. There is also extensive neurofibrillary degeneration of the amygdala and entorhinal cortex (asterisk). a Pathognomonic CTE lesion in stage I CTE. AT8 immunopositive neurofibrillary tangles, dot-like and threadlike neurites encircle a small blood vessel. b Pathognomonic CTE lesion in stage II CTE. A cluster of AT8 immunopositive neurofibrillary tangles and dense dot-like neurites surround several small blood vessels, c pathognomonic CTE lesion in stage III CTE. A large collection of AT8 immunopositive neurofibrillary tangles and dense dot-like neurites enclose several small blood vessels. With increasing age, AT8 immunoreactive astrocytes are increasingly evident within the pathognomonic lesion (open triangle). d Pathognomonic CTE lesion in stage IV CTE. A large accumulation of AT8 immunopositive neurofibrillary tangles, most of them ghost tangles, encompass several small blood vessels. With increasing age, AT8 immunoreactive astrocytes are increasingly prominent (open triangles) and there may be evidence of neuronal loss. a–d All magnification × 200. P-tau phosphorylated tau, CTE chronic traumatic encephalopathy, NFL National Football League Full size image
During the years 2013–2016, when ARTAG was emerging as an age-related tau pathology but not fully recognized by all neuropathologists, several publications reported the prevalence of CTE in brain bank collections and autopsy series that might have misinterpreted ARTAG for diagnostic CTE pathology (Fig. 2) [66, 92, 102]. In 2015, Ling et al. screened a consecutive series of brain donations to the Queen Square Brain Bank for Neurological Disorders for CTE using the proposed McKee criteria for CTE [66, 79]. Their total sample included 221 elderly individuals with neuropathological evidence of a neurodegenerative disease including PSP, LBD, multiple system atrophy, CBD, FTLD, and AD, and 47 controls (> 60 years of age). They identified 32 (11.9%) cases as having neuropathology consistent with CTE, mean age at death 81.0 years, including 13 females (40.6%). The prevalence of CTE in controls (12.8%) was essentially the same as in those with neurodegenerative disorders (11.8%). In hindsight, although we have not had the opportunity to review the histology, it is possible the authors might have considered subpial ARTAG as diagnostic of CTE in some of their cases, as the images they provide depicting characteristic CTE pathology (e.g., Ling et al. [66], Fig. 1) appear to show purely astrocytic perivascular p-tau pathology that would not meet current NINDS–NIBIB diagnostic criteria for CTE [11, 74]. In addition, 15 of their 32 CTE cases (47%) had diagnostic pathology restricted to the midbrain. Isolated midbrain p-tau pathology would not meet current criteria for CTE; at a minimum, the diagnosis of CTE requires at least one pathognomonic cortical lesion [11]. In addition, the advanced age (81.0 years) and female gender (47%) of their cases would be unusual for early-stage CTE (all cases were CTE stage I or II), but typical for ARTAG. Nevertheless, postmortem interviews with the next of kin indicated that 93.8% of their CTE cases had some form of exposure to RHI. Noy et al. prospectively examined 111 brains in a non-selected community-based neuropathology service in Winnipeg, Manitoba, Canada, for the presence of CTE pathology [92]. They excluded brains from individuals over age 60 or with any histological evidence of AD. They identified five cases that met full criteria for CTE based on the NINDS–NIBIB consensus criteria: three cases of stage I and two cases of stage II CTE. They also identified 34 cases (30.6%) that showed "CTE < 1", a category defined as small p-tau deposits in perivascular regions at the depths of cortical sulci. It is unclear how many of their CTE < 1 cases might have represented ARTAG. Among a separate group, they also identified four cases that met full criteria for CTE. All four cases with CTE had a history of head trauma, although detailed sports histories were not available. Puvenna et al. examined 6 pathologically verified cases of CTE (mean age at death, 73.3 years), 6 age-matched control samples, and 19 surgically resected brain specimens from individuals with temporal lobe epilepsy (TLE) (ranging in age from 4 months to 58 years; mean 27.6 years) [102]. Puvenna et al. reported that p-tau pathology in the TLE specimens was identical to that in the CTE specimens. Although we were not able to examine the TLE brain tissue, the CTE brain specimens originated from the Boston University Understanding Neurological Injury and Traumatic Encephalopathy (UNITE) brain bank, and we were unable to appreciate any diagnostic CTE pathology in the representative figures supplied in the manuscript. Later studies would report a non-specific increase in p-tau pathology in surgical specimens from epilepsy patients, unrelated to CTE [109].
Fig. 2 Differential diagnosis between mild CTE and ARTAG. 10-µm paraffin-embedded tissue sections immunostained for phosphorylated tau (AT8) (Pierce Endogen). Positive p-tau immunostaining appears dark red, hematoxylin counterstain. a–c p-tau-immunoreactive thorn-shaped astrocytes are present at the glial limitans at the depths of the sulcus, a form of ARTAG (a, b) magnification × 200, c magnification × 400). The depth of the sulcus is marked by an asterisk. d–f Clusters of p-tau neurons and dot-like neurites surrounding small blood vessels in deep cortical laminae at the depth of the sulci are representative of the diagnostic CTE lesion, all magnifications × 40. The diagnostic lesions are indicated by red circles and located deeper than subpial ARTAG, marked by an asterisks. g–i CTE lesions consist of p-tau-immunoreactive neurons, dot-like neurites, and variably astrocytes, surrounding small blood vessels, all magnifications × 200 Full size image
In 2016, the consensus panel met again to review and refine the preliminary pathological criteria for CTE provided by the first NINDS–NIBIB consensus conference using a second blinded sample of CTE cases representing all severities of disease: mild to severe [11]. The panel sought to further distinguish CTE from ARTAG and from PART [26], to address the minimum threshold for diagnosis, and to determine whether the McKee staging scheme was reliable using only a limited number of paraffin-embedded slides instead of large format 50 µm sections [79]. Eight neuropathologists evaluated 27 cases of tauopathies (17 CTE) and were highly accurate in CTE diagnosis using the preliminary NINDS–NIBIB criteria. In the first round, blinded to clinical and demographic data and gross neuropathological features, 88% of the responses correctly indicated a diagnosis of CTE, which rose to 97% after the clinical data and gross neuropathological features were supplied. Generalized estimating equation analyses showed a statistically significant association, with large effect sizes, between the raters’ assessment and the cases submitted as CTE for both the blinded [odds ratio (OR) 72, 95% confidence interval (CI) 19–267] and unblinded rounds (OR 257, 95% CI 64–1559). The panel concluded by refining the definition of the pathognomonic lesion to emphasize that the perivascular p-tau aggregates necessarily involve neurons, with or without p-tau in astrocytes, and are found in deeper cortical layers than the subpial and superficial regions (Fig. 2, Tables 1, 2). In addition, the panel confirmed that subpial TSA and purely astrocytic perivascular p-tau pathology represented ARTAG and did not meet the minimum criteria for CTE.
Table 1 Recommended protocol for evaluation for CTE [11, 74] Full size table
Table 2 NINDS-NIBIB criteria for the pathological diagnosis of CTE [11] Full size table
At this second consensus meeting, applying the McKee staging system to the limited series of paraffin-embedded slides proved to be inconsistent. P-tau pathology in CTE is patchy and widely dispersed in mild cases and might not be captured using restricted sampling. To aid in the assessment of CTE pathological severity when only a limited number of paraffin-embedded slides are available, the panel proposed a simplified working protocol. The algorithm considers cases as diagnostic for CTE if a single pathognomonic lesion is present, then classifies the case as either “Low CTE” or “High CTE,” depending on whether NFTs are present in the thalamus, mammillary bodies, hippocampus, amygdala, and entorhinal cortex. The designation “Low CTE” roughly equates to McKee CTE stages I and II, and “High CTE” to McKee CTE stages III and IV (Fig. 1, Table 3). The designations high and low CTE were not intended to replace the McKee staging scheme but were designed to facilitate evaluating disease severity by neuropathologists in routine practice.
Table 3 Systems to evaluate pathological severity in CTE Full size table
It is notable that while the pathognomonic lesion of CTE is cortical, brainstem p-tau pathology is an essential aspect of CTE. The first and second consensus panels recommended that the midbrain, pons, and medulla be sampled and evaluated [11, 74], as brainstem p-tau pathology is a supportive feature of CTE. In CTE, p-tau NFT are often found in the isodendritic core (nucleus basalis of Meynert, raphe nuclei, substantia nigra, and LC), with the LC affected at all stages, including stage I.