Localized cortical chronic traumatic encephalopathy pathology after single, severe axonal injury in human brain

In 2010 an estimated 2.5 million persons suffered from traumatic brain injury (TBI) in the United States, of whom 50,000 died from the injury and another 2,80,000 required hospitalization [36]. For survivors, TBI is a known environmental risk factor for the development of dementia later in life. Multiple studies provide evidence that single moderate or severe TBI increases the likelihood of eventual presentation with clinical signs and symptoms of neurodegeneration, such as Alzheimer Disease (AD) [7, 29, 35]. Such studies typically did not include postmortem examinations of brain tissues. Rather, the medical literature scantily addressed chronic neuropathological sequelae after TBI. One recent autopsy study of 39 patients with long-term survival (1–47 years) after single moderate or severe TBI, however, showed widespread β-amyloid plaques and neurofibrillary tangles (NFTs) in the brains of approximately one-third of patients [17]. Whether single mild TBI (concussion) similarly confers increased risk of developing dementia with neuropathological abnormalities in susceptible individuals remains undetermined.

For contact sport athletes with repetitive mild TBIs or subconcussive events, such as boxers and American football players, studies of postmortem brains from symptomatic patients revealed a progressive neurodegenerative disorder with distinctive neuropathological features, called chronic traumatic encephalopathy (CTE) [4, 27]. Often presenting with symptoms years after cessation of contact sport participation, CTE patients tend to exhibit initially behavioral and mood abnormalities, cognitive impairments and chronic headache, whereas in advanced stages, these patients deteriorate to frank dementia [28, 33, 45]. Like other neurodegenerative tauopathies, disease progression relates to the burgeoning spread of abnormally hyperphosphorylated tau protein (p-tau) [28]. In the first consensus meeting to define neuropathological criteria for CTE, the participants established the pathognomonic lesion as p-tau accumulation in neurons (NFTs), astrocytes and cell processes, with predilection around small blood vessels, at sulcal depths in irregular cortical spatial patterns [26]. The biological mechanism underlying this observation is unknown. In addition, a subset of reported CTE cases demonstrated diffuse, or more rarely neuritic, β-amyloid plaques [4, 26, 39, 44].

TBIs commonly involve axonal injury, a prominent factor contributing to symptomatology and one of multiple possible lesions in the human brain consequent to head impact [15, 34, 46]. Therefore, we identified a human model that isolates the axonal damage component of TBI, that is, surgical leucotomy, and then tested the hypothesis that chronic axonal injury can result in neurodegenerative changes, such as NFTs and β-amyloid plaques. We analyzed postmortem brain tissues from five institutionalized patients with schizophrenia, who had undergone bilateral prefrontal leucotomy as therapy prior to 1953 and then lived at least another 40 years (Table 1). Since leucotomy involves severing axons in the prefrontal cortex, this surgical procedure represents a single iatrogenic TBI with severe axonal damage and no external cortical impact. We examined cortical tissues at the leucotomy site, prefrontal cortex rostral and frontal cortex caudal to the leucotomy site, and hippocampus for evidence of p-tau and β-amyloid accumulations. We also examined postmortem cortical brain tissues of five patients with schizophrenia from the same institution, who had not undergone leucotomies in their lifetimes, matched in gender and approximate age (±4 years) to the leucotomy patients (Table 1). Moreover, we determined the apolipoprotein E (APOE) genotype for all ten patients. We found p-tau accumulations in the gray matter at the axotomy site in all five leucotomy patients, with four cases showing the pathognomonic p-tau pattern of CTE. Scattered β-amyloid plaques also occupied the gray matter at the lesion site, but in the three leucotomy patients with APOE ε4 haplotype. We observed minimal p-tau and β-amyloid pathology in the cortical samples rostral and caudal to the leucotomy site and in all cortical samples from the non-leucotomized schizophrenia patients.

All brain tissues for this study were obtained from the brain bank at the Pilgrim Psychiatric Center in West Brentwood, NY. The Pilgrim Psychiatric Center opened in 1931 as a site for care and treatment of patients with chronic psychiatric conditions. The patient population continually expanded such that the institution cared for almost 14, 000 patients in 1954. Many resident patients suffered from chronic schizophrenia. In the late 1940s until 1953, surgeons commonly performed bilateral prefrontal leucotomy as therapy for intractable symptoms. Currently available clinical records do not detail the surgical procedure, but based on autopsy findings and medical literature, the surgeons most likely used bilateral burr holes at the coronal suture between the frontal and parietal bones to access the brain [5]. In recent years, many patients have been discharged to the care of local community mental health centers. Several hundred patients, however, were too impaired for community living and thus remained as inpatients. Drs. Kenneth Davis and Vahram Haroutunian established the brain bank for the Pilgrim Psychiatric Center, with Dr. Daniel Perl serving as neuropathologist.

We sampled brain specimens fixed in formalin, embedded these tissues in paraffin and then sectioned the tissue blocks at 5 μm thickness. For the leucotomy postmortem brain tissues, the samples comprised cortical tissues cut in the coronal plane at the leucotomy site, prefrontal cortex rostral and frontal cortex caudal to the leucotomy site, and hippocampus; for the non-leucotomized postmortem brain tissues, the samples included equivalent cortical neuroanatomical sites. We conducted hematoxylin and eosin (H&E) stains on tissue sections for general morphology/structure and immunohistochemistry with antibodies detecting glial fibrillary acidic protein (GFAP, astrocyte marker, mouse anti-human monoclonal antibody GA5, Leica Biosystems, PA0026), abnormally hyperphosphorylated tau (AT8, pS199/S202/T205; mouse anti-human monoclonal antibody, Thermo Scientific, MN1020), β-amyloid (mouse anti-human monoclonal antibody, clone 6F/3D, Leica Biosystems, NCL-B-Amyloid) and antigen CD68 (marker of macrophages and activated microglia, mouse anti-human monoclonal antibody clone 514H12, Leica Biosystems, PA0273). Immunohistochemistry for these antibodies was performed on a Leica Bond III automated immunostainer with a diaminobenzidine (DAB) chromogen detection system (Leica Biosystems, DS9800, Buffalo Grove, IL).

We also performed immunohistochemistry experiments with two other antibodies against abnormally hyperphosphorylated tau (CP13, pS202; PHF1, pS396/S404). Briefly, tissue sections were first deparaffinized and rehydrated. For antigen retrieval, tissue sections were immersed in sodium citrate buffer (pH 6) at boiling temperature for 6 min and then cooled to room temperature. To quench endogenous peroxidase activity, tissue sections were placed in a solution of 0.5% H₂O₂ in methanol for 1 h. Tissue sections were incubated in blocking solution, comprised of 5% bovine serum albumin in phosphate-buffered saline (pH 7.4), for 1 h at room temperature. The primary antibodies were diluted in block solution (CP13, mouse anti-human monoclonal antibody, 1:1000; PHF1, mouse anti-human monoclonal antibody, 1:1000; gifts from Dr. Peter Davies, Feinstein Institute for Medical Research, Manhasset, NY). The tissue sections were incubated with primary antibody solution in a humidified chamber at 4 °C overnight. Tissue sections were then incubated with secondary antibody, followed by avidin–biotin-complex incubation (Vectastain Elite ABC Kit, mouse IgG, Vector Laboratories, PK-6102) and visualization with DAB (ImmPACT DAB, Vector Laboratories, SK-4105, Burlingame, CA), according to manufacturer instructions. The sections were counterstained with hematoxylin, dehydrated and cleared with xylene before application of mounting media and coverslip.

Tissue samples also underwent modified Bielschowsky silver staining for detection of NFTs and neuritic plaques [8]. We extracted DNA from formalin-fixed brain tissues, according to manufacturer’s instructions (DNeasy Blood & Tissue Kit, QIAGEN, 69504, Valencia, CA). APOE genotyping was performed with real-time polymerase chain reaction (RT-PCR) [2].

Two neuropathologists (DI, DPP) conducted blinded evaluations of the stained and immunostained brain tissue samples. This protocol received Institutional Review Board approval prior to initiation of study.

CTE is a distinct neurodegenerative disorder linked to previous repetitive mild impact TBI and subconcussive events, classically described in retired boxers [4, 24]. The recent publication establishing initial diagnostic criteria for CTE points to p-tau accumulations in neurons, astrocytes and cell processes, favoring ensheathment of blood vessels at sulcal depths in irregular cortical patterns, as the pathognomonic lesion, thus distinguishing CTE from other neurodegenerative tauopathies and also aging-related tau astrogliopathy (ARTAG) [19, 26]. The underlying pathophysiology leading to this specific neuroanatomical distribution of p-tau aggregates, especially encompassing blood vessels, remains unanswered. There are two main hypotheses: that underlying axonal damage may be related to these pathognomonic lesions or that mechanical stress at sulcal depths during the traumatic event may render them susceptible to eventual p-tau deposition. Due to lack of extensive clinical history, we cannot comment upon head injury for an individual patient in this study. Nevertheless, because we observed p-tau immunoreactivity patterns similar to CTE only in the schizophrenia patients with leucotomy, which results in severe axonal injury without head impact, the data in this study support the hypothesis that selective p-tau accumulation at sulcal depths relates to underlying axonal damage.

Evidence exists to support the connection between traumatic axonal damage, which can persist for years, and accumulation of protein aggregates typically associated with neurodegenerative disease, with particular research focus on β-amyloid [14, 16, 43]. Pointedly, among those patients who had died or experienced temporal cortex surgical resection shortly after severe TBI, a subset showed β-amyloid plaques, including young people [10, 40, 41]. In diffuse axonal injury resultant of impact TBI, disruption of protein transport at sites of damage can lead to accumulation of proteins, including amyloid precursor protein (APP), in axonal varicosities and bulbs [16]. With axonal damage, proteins can abnormally congregate in the same space, either within the damaged axon or extracellular domain. As such, noxious molecular interactions potentially can occur, for example, β-secretase and presenilin-1 cleaving APP to generate β-amyloid that can then aggregate to form plaques [3, 16, 43, 49]. Likewise, indigenous tau proteins normally reside within axons, stabilizing neuronal microtubules to aid with axonal transport of essential components necessary for cell survival and function. With tauopathic neurodegenerative disease, tau proteins become hyperphosphorylated, disassociate from microtubules and aggregate in insoluble form. One postmortem study of human brain tissues, from persons who had died between 4 h and 5 weeks after TBI, showed PHF1 immunopositivity in axonal varicosities and bulbs, neuronal cell bodies and reactive astrocytes within the cerebral cortex in two of 18 cases [49]. In the TBI study of surgically resected temporal cortices, tissue samples in 12 of 18 cases displayed dystrophic axons in white matter with positive PHF1 immunoreactivity; three cases showed PHF1 immunopositive glia, and four cases revealed PHF1 immunopositive neurons [10]. For the schizophrenia patients who had survived at least 40 years after leucotomy in this study, we similarly noted NFTs, astrocytic tangles and p-tau immunoreactive cell processes in neuroanatomical areas near the axotomy site, with three p-tau antibodies including PHF1. In addition, we observed β-amyloid plaques scattered in the gray matter within proximity of the massive axonal damage in a subset of the leucotomized cases (APOE ε4 haplotype). Given these data and literature, we surmise that damaged axons may inadvertently release proteins at the lesion site into the extracellular brain parenchyma, perhaps over long periods of time, thereby promoting aberrant molecular reactions and deleterious aggregation of particular proteins.

One possible explanation for the p-tau accumulation pattern in CTE, and the leucotomized patients in this study, lies in interstitial solute clearance in the human brain through local diffusion and fluid bulk flow, a process first proposed about 100 years ago with recent scientific breakthroughs [12]. The human brain lacks characteristic lymphatic vessels, utilized by the rest of the body, to provide protein and fluid homeostasis. In contemporary experiments involving rodents, data show that cerebrospinal fluid (CSF) in the subarachnoid space influxes through periarterial spaces and exits through the aquaporin 4 channels of astrocytic endfeet, mixes with interstitial fluid to flow by convection through the interstitial space clearing macromolecules and finally passes through aquaporin 4 channels of astrocytic endfeet into perivenous spaces—termed the glymphatic pathway for the dependence upon glial cells to function as the surrogate lymphatic system in the brain [30]. The perivenous fluid effluxes to contiguous CSF that filtrates through arachnoid granulations to dural sinuses for venous drainage. Two recent papers provide evidence that CSF, with its soluble and cellular contents, flows into newly discovered lymphatic vessels in meninges lining dural sinuses that drain into cervical lymph nodes [1, 23], in addition to lymphatic vessels of nasal mucosa, cranial nerve sheathes and particular arteries and veins exiting the cranium [1]. In another paper employing mouse models, the authors showed impaired function of the glymphatic system after TBI, with prominent astrogliosis, loss of perivascular aquaporin 4 polarization and axonal damage and then specifically identified diminished interstitial tau clearance with accompanying hyperphosphorylated tau increase [11]. One common observation reported in CTE, and also in the leucotomy patients of this study, is p-tau accumulation encompassing blood vessels and in perivascular astrocytes, which is consistent with clearance data in rodents. Considered a biomarker for axonal injury, total tau levels increase in CSF and blood serum after TBI in humans, including contact sports athletes shortly after mild head injuries [31, 42, 50]. Nevertheless, since most experiments involve rodents, with lissencephalic brain structures, the specific mechanisms of clearance within the human brain, particularly its relation to sulcal depths, remain poorly understood. We suspect that the sulcal depth serves as an exodus for tau clearance after axonal damage in humans. If true, then axonal injury, which occurs with TBI ranging from mild to severe, may account for the pathognomonic p-tau signature pattern of CTE [26]. Finally, brain tissues from long-term survivors of single moderate to severe TBI, with presumable axonal injuries that can persist for years, also display NFTs and p-tau immunoreactive astrocytes in cortical sulcal depths [14, 17]. Taken together, these data and literature not only bolster the link between head trauma and CTE, but also attenuate the notion of the pathognomonic CTE lesion being limited to multiple mild TBIs and/or subconcussive events.

CTE is classified as a tauopathy, yet a subset of reported cases displays diffuse or neuritic β-amyloid plaques in cortical gray matter, which instigated the debate decades ago concerning the pathophysiologic connections among head injury, neuropathological neurodegenerative indicators such as p-tau and β-amyloid protein accumulations and augmented development of neurodegenerative diseases. In 1973, after extensive examination of brain tissues from 15 deceased boxers, Corsellis, et al. reported only one case with neuritic β-amyloid plaques, using classic silver staining techniques [4]. Interestingly, Roberts, et al. analyzed brain tissues from 14 of these same cases plus six others, but additionally employed immunohistochemistry experiments for β-amyloid detection, to discover numerous diffuse plaques in most cases [39]. Moreover, several papers provide data suggesting an association between the APOE ε4 genotype and predilection for β-amyloid deposition in the brains of patients dying shortly after head injury [9, 32] and also in CTE patients especially with advancing years [44]. In this present study, all three leucotomized patients with the APOE ε4 haplotype displayed diffuse and cored β-amyloid plaques scattered throughout the gray matter adjacent to the lesion site, in contrast to the leucotomized patients with the APOE ε3/3 genotype. The oldest leucotomized patient, who had died at 87 years with an APOE ε3/3 genotype, showed a focus of cored β-amyloid plaques at one sulcal depth, which complements a recent study reporting predilection of β-amyloid aggregates at cortical sulcal depths in CTE patients [44]. All five schizophrenia patients without leucotomies showed minimal or no β-amyloid immunoreactivity in cortical gray matter, including three patients with APOE ε4 haplotype. The molecular interaction between ApoE protein and β-amyloid warrants further clarification; however, ApoE can bind β-amyloid to enhance its clearance through multiple mechanisms including proteolysis, with ApoE E₄ as the least effective isoform [22]. In mouse models, experimental data demonstrated that murine brains clear soluble β-amyloid along glymphatic paravenous drainage pathways as well and that deletion of the Aqp4 gene attenuates this process [13]. In addition to the glymphatic system for clearance of β-amyloid, efflux transporters, such as the low-density lipoprotein receptor-related protein-1 (LRP1), facilitate transendothelial transport across the blood–brain barrier [51, 52]. Human brain studies revealed cerebrovascular β-amyloid deposits in three former boxers and increased frequency of CAA in APOE ε4 patients with severe acute TBI [21, 48]. In this study, we observed minimal CAA or perivascular β-amyloid aggregation in all ten schizophrenia patients. Rather, brain tissues of the APOE ε4 haplotype patients with leucotomy showed β-amyloid plaques throughout gray matter overlying the axotomy site.

This study has several limitations in design and interpretation. Foremost, we lacked detailed antemortem clinical data, including head injury history, which curtailed the extent of neuropathological analysis. Additionally, we assessed brain tissues from two small cohorts: five schizophrenia patients with leucotomy and for comparison, five schizophrenia patients without leucotomy who had also lived at the Pilgrim Psychiatric Center during approximately the same time period. Most likely aging in the schizophrenic brain diverges from persons without this condition, as evidenced by schizophrenia patients in advanced years more frequently suffering from severe cognitive impairments than the general population [37, 38]. A neuropathological study of 100 consecutive autopsy patients (age range, 52–101 years; mean, 77 years), however, shows equivalent amounts of β-amyloid plaques and NFTs in neocortex and hippocampus of patients with schizophrenia, as compared to 47 patients with other psychiatric disorders and 50 patients without psychiatric or dementing diagnoses, when matched by age [37]. The study authors concluded that the percentage of elderly schizophrenia patients with AD, as determined by neuropathological criteria, basically equates to the general elderly population with corresponding ages. Since schizophrenia involves neurodevelopmental abnormalities, the possibility exists that clearance of p-tau, β-amyloid and other neurodegenerative proteins may differ from persons without schizophrenia. Due to the small number of patients in each cohort, uncertainty of tissue integrity equivalence from archived brain specimens and tendency for focality and irregularity patterns of especially the p-tau immunoreactivity, we did not numerically quantify pathological results. Also, because we obtained specimens previously dissected for other studies, we were unable to analyze hemibrain sections similar to contemporary CTE studies to attain more thorough neuroanatomical distribution assessment of p-tau aggregates in sulcal depths. As such, we could not determine potential cortical spatial patterns for sulcal tauopathy, which currently are described as irregular in CTE. Furthermore, given these limiting circumstances, we may have serendipitously missed sulcal tauopathy in the sampling of brain tissues from case 1. Other possibilities for lack of sulcal p-tau pathology in case 1 include that this individual responded differently than the other four patients after the leucotomy procedure, or that since this patient died at the youngest age (67 years), with more prolonged life this patient may have accumulated p-tau in sulcal depths similar to the other older leucotomized patients in the cohort.

In summary, we studied a human model of severe axonal injury, analyzing brain tissues specifically for chronic neurodegenerative changes. We found p-tau pathology in patterns resembling CTE in gray matter adjacent to the leucotomy site. We also observed a subset of cases with β-amyloid plaques, similarly reported in CTE, scattered in gray matter at the leucotomy site, but in patients with APOE ε4 haplotype, which is associated with worse clinical outcomes for TBI patients [6, 25, 47], including boxers and American football players [18, 20]. These data suggest that axonal injury, common to all severities of TBI, contributes to the distinct pathology of CTE over time.