Chronic traumatic encephalopathy: A devastating legacy of repetitive concussion Authors: Sunghyun Jun, Colleen Marie Arrasmith, Volha Liaudanskaya Published: 2026 ABSTRACT Repetitive concussive and subconcussive traumatic brain injury (TBI) is increasingly linked to chronic traumatic encephalopathy (CTE), yet a central challenge remains in connecting exposure to long-term neurodegeneration through a coherent mechanistic framework. Here, we synthesize evidence across epidemiology, neuropathology, and clinical studies to define the continuum from repetitive injury to disease. Primary injury initiates secondary cascades, including mitochondrial dysfunction, metabolic stress, neuroinflammation, and axonal injury across neuronal, glial, and vascular compartments, which, over time, promote protein misfolding and progressive pathology involving tau, amyloid precursor protein (APP), and TDP-43. CTE is defined by a distinct pattern of perivascular hyperphosphorylated tau accumulation at the depths of cortical sulci, linking injury-associated biomechanical strain and vascular vulnerability to spatially localized disease progression. These pathological processes give rise to heterogeneous clinical features that are only partially captured by current diagnostic frameworks and emerging imaging and fluid biomarkers, which remain limited in specificity. Experimental models, including in vivo systems and human 3D in vitro platforms, provide complementary insight into specific aspects of CTE pathobiology, but no single model fully recapitulates the disease trajectory. Together, this synthesis reframes CTE as a mechanistically linked continuum from exposure to neurodegeneration, highlights key gaps in diagnosis and modeling, and identifies priorities for advancing in-life detection and therapeutic development. Traumatic brain injury as a driver of progressive neurodegeneration Traumatic brain injury (TBI) Traumatic brain injury (TBI) is a major cause of death and long-term disability worldwide, with an estimated 69 million new cases each year (Thurman et al., 1999; Liaudanskaya et al., 2023; Dewan et al., 2019). The majority are classified as mild traumatic brain injuries (mTBI), which account for 70–90% of all reported cases (Cassidy et al., 2004). Hospital-based data suggest an incidence of 100–300 per 100,000 adults annually, but the true rate is likely far higher as many individuals with mild or transient symptoms never seek medical attention (Cassidy et al., 2004) (Fig. 1A). TBI is defined as an external mechanical force to the head that disrupts normal brain structure or function (Galgano et al., 2017). While the initial impact causes direct mechanical damage, this is only the starting point of a complex process. The primary injury rapidly evolves into secondary cascades that unfold over hours to weeks, involving mitochondrial dysfunction, metabolic disturbances, neuroinflammation, axonal injury, protein misfolding (such as amyloid precursor protein (APP), tau, and transactive response DNA binding protein of 43 (TDP-43)), and eventually cell death (Gulf War and Health Volume 7 Long-Term Consequen.pdf, 2009; Ladak et al., 2019). These processes are driven by dynamic interactions between neurons, glia, and vascular cells, which amplify and sustain damage long after the initial insult (Galgano et al., 2017). Clinical outcomes after TBI are highly variable. Some individuals recover fully, while others experience persistent cognitive, behavioral, or motor impairments. In many cases, the long-term effects progress to chronic conditions that resemble or overlap with neurodegenerative diseases (Stocchetti and Zanier, 2016). This variability likely reflects differences in both injury characteristics and biological response. Factors such as injury burden, frequency, and biomechanical profile interact with individual susceptibility, including the magnitude and persistence of neuroinflammation, the extent of axonal damage, the efficiency of mitochondrial recovery, and the capacity for protein clearance. Together, these processes may determine whether injury-induced cascades resolve or persist, ultimately influencing the transition from acute injury to chronic neurodegeneration. Despite this burden, there are no effective therapies that modify disease progression. Dissecting the cellular and molecular mechanisms of TBI is therefore critical, not only to understand the transition from acute injury to chronic pathology, but also to identify biomarkers and develop treatments that can alter the course of disease (Ladak et al., 2019). Repetitive mild brain trauma and the emergence of chronic traumatic encephalopathy (CTE) Repetitive mild brain impacts, including concussive and subconcussive injuries, have become increasingly recognized as a major public health concern. Early work from Talavage, Nauman, and colleagues was instrumental in defining subconcussive injury as a biologically meaningful phenomenon in adolescent athletes. In high school football cohorts, these studies demonstrated that players without clinically diagnosed concussion could nonetheless exhibit measurable abnormalities in neurocognitive performance, task-based fMRI activation, and resting-state functional connectivity, with later work linking cumulative head-impact exposure to white matter diffusivity changes on diffusion MRI (Talavage et al., 2014; Katherine Morigaki Breedlove et al., 2014; Abbas et al., 2015; Jang et al., 2019). Collectively, this body of work challenged the assumption that absence of symptoms implies absence of brain injury and introduced the concept of a “subclinical” injury state. Importantly, these findings support a dose-dependent model in which repetitive subconcussive impacts produce cumulative functional and structural brain alterations, providing a critical foundation for understanding how repeated low-level trauma may contribute to long-term neurodegenerative risk. Millions of such injuries occur annually in athletes and military personnel, often going unreported because symptoms can be subtle or delayed (Vaishnavi et al., 2009) (Fig. 1B). Biomechanical studies, advanced imaging, and forensic analyses all confirm that repeated low-level trauma to the brain is widespread in these populations (Petraglia et al., 2014). Epidemiological and pathological evidence support a link between traumatic brain injury (TBI) and the later development of neurodegenerative disorders. Among them, chronic traumatic encephalopathy (CTE) is most strongly associated with mild and repetitive injuries, whereas moderate and severe TBI are more often connected to Alzheimer's disease (AD) or Parkinson's disease (PD) (Galgano et al., 2017; Brett et al., 2022). This association between repetitive injury and long-term neurodegeneration has prompted growing interest in defining CTE as a distinct disease entity, both historically and mechanistically. CTE itself is not a new disorder. First described in 1928 as ‘punch-drunk syndrome’ or ‘dementia pugilistica’ in boxers, it remained largely confined to the forensic and neuropathology literature for decades. The transition from this observational era into the modern molecular understanding of CTE was catalyzed by the work of Dr. Bennett Omalu and colleagues, who in 2005 reported the first neuropathologically confirmed case of CTE in a former NFL player, identifying perivascular tau pathology and neurodegeneration in a professional American football player with no prior diagnosis (Omalu et al., 2006; Omalu et al., 2005). This landmark description shifted the field's attention from boxers to a broader population of contact-sport athletes and military veterans and established the molecular framework of tau pathology that now defines the disease. It is recognized as a progressive neurodegenerative disease diagnosed post-mortem in individuals with documented histories of repetitive mTBI or sub-concussive trauma. Despite decades of research, the true prevalence of CTE remains unknown, as a definitive diagnosis can only be made after death, leaving the field without reliable in-life biomarkers or clinical criteria (DeKosky et al., 2013; Stein et al., 2015; McKee et al., 2016a). While the historical progression established the clinical and pathological identity of CTE, understanding how different forms of brain injury lead to this disease requires consideration of the underlying biomechanics and injury-specific responses. The mechanisms and consequences of TBI vary considerably depending on the nature of the injurious force. Broadly, TBI can be categorized into blunt-force impact, acceleration-deceleration injury, and blast-induced neurotrauma (Bryden et al., 2019; Maas et al., 2022). Blunt-force injuries typically produce focal cortical-subcortical damage localized at the site of impact, whereas acceleration-deceleration forces, predominant in contact sports such as American football and boxing, induce mechanical axonal deformation through rotational shearing of white matter tracts and are particularly associated with strain concentration at sulcal depths and at the gray-white matter junctions (Alisafaei et al., 2020; Hajiaghamemar and Margulies, 2021; Turner et al., 2015a). Blast-induced TBI, predominant in military populations, involves a distinct mechanism in which oscillating acceleration-deceleration cycles imposed on the head, rather than direct shock wave transmission, appear to be the primary driver of closed-head brain injury (Cherry et al., 2020a; Daneshvar et al., 2023; Tagge et al., 2018; Stern et al., 2013; Goldstein et al., 2012). Critically, these injury subtypes may confer differential yet convergent susceptibilities to CTE. The majority of neuropathologically confirmed CTE cases have been identified among individuals with histories of repetitive acceleration-deceleration exposure (McKee et al., 2023). However, postmortem analyses of military veterans with documented blast exposure have revealed focal perivascular tau pathology that closely resembles the hallmark CTE lesions observed in contact-sport athletes, suggesting that despite mechanistic differences in the initiating force, both subtypes can trigger a common downstream pathological cascade (Goldstein et al., 2012). Whether these trajectories represent truly equivalent pathobiological endpoints or distinct disease variants with overlapping features remains an important open question in the field. Pathological studies demonstrate that repetitive mild trauma, even without a clinically recognized concussion, is associated with the neuropathology of CTE. Key features include the accumulation of hyperphosphorylated tau (p-tau), TDP-43 pathology, and genetic associations such as variants in transmembrane protein 106B [TMEM106B] (Stern et al., 2011) (Fig. 1C). CTE is defined neuropathologically by p-tau aggregates in neurons, astrocytes, and perivascular regions at the depths of cortical sulci. While tau pathology also characterizes other disorders such as AD and frontotemporal lobar degeneration (FTLD), CTE is distinct (McKee et al., 2023; Buckland et al., 2022). Cryo-electron microscopy has revealed that p-tau filaments in CTE exhibit distinct structural differences from those in AD, underscoring its classification as a unique tauopathy (McKee et al., 2023). Furthermore, the distribution of tau pathology is different: in CTE, aggregates localize primarily to cortical layers II/III and extend broadly across hippocampal subfields, whereas in AD (Katsumoto et al., 2019), they are concentrated in layers IV/V and largely confined to cornu ammonis 1 [CA1] (Turner et al., 2016). This distinct spatial distribution likely reflects the underlying biomechanics and cellular response to injury. Rotational acceleration-deceleration forces generate maximal shear strain at the depths of cortical sulci and at gray-white matter interfaces, leading to focal axonal injury and microvascular disruption in these regions (Alisafaei et al., 2020; Fagan et al., 2020; Lu et al., 2021). The perivascular localization of p-tau may further reflect impairment of clearance pathways, including glymphatic flow along perivascular spaces, resulting in local accumulation of pathological proteins. In addition, the selective vulnerability of superficial cortical layers may be driven by differences in neuronal connectivity, metabolic demand, and susceptibility to mechanical deformation. Together, these factors provide a mechanistic basis for the characteristic anatomical pattern that distinguishes CTE from other tauopathies such as AD. Given these distinct pathological patterns, it becomes important to connect them with how CTE presents clinically and how different frameworks attempt to define the disorder during life. Clinical features and proposed diagnostic frameworks for CTE The clinical presentation of CTE is broad and heterogeneous, encompassing cognitive, behavioral, mood-related, and motor disturbances that typically emerge in midlife, often years to decades after the period of active exposure. Critically, these symptoms are neither specific nor sufficient to confirm the diagnosis in life, and some individuals with no clinical symptoms are later found to have CTE at autopsy, underscoring the fundamental challenge of establishing reliable in-life diagnostic criteria. This disconnect between clinical presentation and underlying pathology reflects the current inability to directly link symptom profiles to specific molecular and anatomical disease features, complicating both diagnosis and disease staging. Several clinical and pathological classification frameworks have been proposed to address this gap and are discussed in detail in Section 2.2. These frameworks attempt to bridge clinical symptoms with underlying pathology, but vary in their ability to capture disease heterogeneity and to align with molecular burden. Sex differences in mTBI and CTE Biological sex represents an important but underexplored determinant of response to repetitive brain injury, influencing both injury susceptibility and long-term outcomes. Most research on mild traumatic brain injury (mTBI) and chronic traumatic encephalopathy (CTE) has focused on men, reflecting the predominance of male athletes in contact sports and male service members in military cohorts. As participation of women in contact sports and military roles increases, attention has shifted toward the importance of sex-based differences in injury outcomes. Evidence suggests that men and women may respond differently to repetitive mild head trauma. Human studies often report that women experience worse long-term functional outcomes after TBI compared with men (Gupte et al., 2019; Valera et al., 2021; Mikolic et al., 2021; Wagberg et al., 2023). By contrast, preclinical animal studies often demonstrate that females recover more effectively, particularly at the cellular and molecular levels (Xiong et al., 2013; Osier and Dixon, 2016). This discrepancy may reflect differences in study design: clinical studies typically emphasize functional or behavioral outcomes, whereas animal studies often focus on mechanistic endpoints (Gupte et al., 2019). However, biological factors beyond study design likely contribute to these differences. Sex-dependent variation in axonal structure, neck strength, and brain biomechanics may influence the magnitude and distribution of injury, while differences in immune response and mitochondrial function may alter the balance between injury resolution and persistent pathology. Together, these observations highlight the importance of understanding sex-specific biological and clinical responses to mTBI, which is critical for developing accurate models of CTE and for designing prevention and treatment strategies that account for these differences. At the mechanistic level, the relative cellular protection observed in female animals is thought to be mediated primarily by the neuroprotective actions of estrogen and progesterone. Estrogen exerts anti-inflammatory effects on microglia through estrogen receptor (ER) signaling, including ERα, ERβ, and the G-protein coupled receptor GPER1, suppressing proinflammatory cytokine release and shifting microglial phenotype toward a less reactive state following injury (Acosta-Martinez, 2020). Progesterone complements these effects by reducing cerebral edema, suppressing neuronal apoptosis, and modulating aquaporin-4 expression to limit post-injury fluid dysregulation (Brotfain et al., 2016). Together, these hormones attenuate the acute neuroinflammatory cascade that drives secondary injury progression, providing a plausible mechanistic basis for the superior cellular and molecular recovery observed in female preclinical models. However, this hormonal neuroprotection is highly context-dependent. Fluctuations across the menstrual cycle, hormonal changes associated with pregnancy, and the precipitous loss of estrogen and progesterone at menopause all create a dynamic biological landscape that differs fundamentally from the controlled hormonal environments of most animal studies, which typically use young adult females at peak reproductive hormone levels. This mismatch - between the stable, peak-hormone conditions of preclinical models and the heterogeneous, fluctuating hormonal milieu of human clinical populations - is likely a key contributor to the observed paradox whereby superior cellular recovery in female animals does not translate into better long-term functional outcomes in clinical studies of women with TBI. These sex-specific biological and clinical differences underscore the complexity of CTE as a disease entity and highlight the need for inclusive research frameworks. The following sections address the molecular and anatomical hallmarks that define CTE progression. Literature search and limitations This review was conducted through a systematic search of PubMed and Google Scholar using primary search terms including ‘chronic traumatic encephalopathy,’ ‘CTE AND biomarkers,’ ‘CTE AND 3D in vitro models,’ ‘TBI AND tau pathology,’ ‘TBI AND neuroinflammation,’ ‘CTE AND in vivo models,’ and ‘CTE AND diagnosis,’ with a literature date range extending through early 2026. The substantial 20- to 30-fold increase in CTE-related publications since 2011 informed the need for this synthesis (Qi et al., 2020; Guan et al., 2024). Topics were prioritized based on their direct relevance to the central goal of this review: to trace the pathobiological continuum from repetitive head trauma exposure to CTE, and to evaluate the technologies and platforms most likely to facilitate the transition from retrospective post-mortem diagnosis to in-life detection and therapeutic intervention. Accordingly, we emphasize emerging diagnostic modalities, fluid and imaging biomarkers, and advanced experimental model systems over traditional clinical descriptions. We acknowledge several limitations inherent to the current CTE literature that apply throughout this review. First, the majority of neuropathologically confirmed CTE cases derive from brain bank cohorts enriched for symptomatic individuals with known contact sport or military exposure histories, introducing significant ascertainment bias that limits prevalence estimation and generalizability (Schwab and Hazrati, 2018; LeClair et al., 2022). Second, many longitudinal biomarker and imaging studies have small cohort sizes, limiting statistical power and reproducibility (Bergauer et al., 2022; Sparks et al., 2020; Dallmeier et al., 2019). Third, the absence of validated in-life diagnostic criteria means that all mechanistic and clinical conclusions in this review are ultimately grounded in post-mortem data, thereby constraining causal inference (McKee et al., 2016a; McKee et al., 2023; McKee et al., 2016b; Katz et al., 2021). These gaps are discussed in context throughout the review and collectively underscore the critical need for the prospective, multimodal approaches outlined in the Future Directions section. Clinical, anatomical, and molecular features of CTE Clinical presentation of CTE Most of what is known about the clinical features of chronic traumatic encephalopathy (CTE) comes from post-mortem studies and retrospective analyses (Petraglia et al., 2014). The true incidence and prevalence remain uncertain, in part because diagnostic criteria for mild concussive and sub-concussive injuries are inconsistent, and many cases are not recognized during life. Known risk factors are still not well-defined, although repetitive blunt impacts, blast exposure, and acceleration-deceleration forces are all associated with an increased risk (Petraglia et al., 2014; Stern et al., 2013). This variability likely reflects differences in both exposure patterns and biological response, including the extent of axonal injury, persistence of neuroinflammation, and the efficiency of molecular repair and clearance mechanisms. Symptoms of CTE typically appear in midlife, often many years after the period of active exposure. In a large series, McKee and colleagues reported a mean age of onset of 44.3 years (range 16–83, n = 119), with symptoms beginning on average 14.5 years after retirement from athletic or military service (n = 104) (Stein et al., 2015; Stern et al., 2011). Importantly, the clinical syndrome differs from the prolonged effects of a single concussion or sub-concussive injury. Rather than representing a continuation of acute symptoms, CTE reflects progressive molecular and epigenetic alterations that accumulate over years, ultimately disrupting neuronal and glial function and giving rise to distinct clinical phenotypes (McKee et al., 2009). The clinical picture of CTE is broad and heterogeneous. Early features often include mood and behavioral disturbances such as irritability, poor judgment, risk-taking, and depression (Stern et al., 2011; Turner et al., 2016) (Fig. 2A). As the disease progresses, cognitive impairments emerge, with deficits in memory, attention, and executive function being most common (Stern et al., 2013). Motor disturbances, including gait instability, speech difficulties, and Parkinsonian features, may develop in later stages (Stein et al., 2015; Montenigro et al., 2015). These symptom domains likely reflect the regional progression of underlying pathology, with early mood and behavioral changes linked to frontal and limbic circuit involvement, followed by cognitive decline associated with broader cortical dysfunction, and later motor features corresponding to subcortical and brainstem involvement. Sleep disruption and neuropsychiatric symptoms frequently contribute to the overall burden (Stern et al., 2013; Stern et al., 2011). In older patients, CTE can coexist with other neurodegenerative diseases, including Lewy body disease (LBD) and amyotrophic lateral sclerosis (ALS), further complicating diagnosis and clinical management (McKee et al., 2013; McKee et al., 2015). To better interpret how these heterogeneous symptoms arise and evolve in relation to underlying pathology, it is critical to link the clinical syndromes to their underlying molecular and anatomical substrates. Accordingly, two primary frameworks have been proposed to classify the presentation of CTE, each with distinct strengths and limitations with respect to the underlying molecular pathology. The neuropathological staging system developed by McKee and colleagues categorizes CTE into four progressive stages (I-IV) based on the extent and distribution of hyperphosphorylated tau (p-tau) pathology (Mikolic et al., 2021). This framework offers the strongest alignment with the molecular burden of p-tau because each stage directly reflects the anatomical spread of tau aggregates. Furthermore, TDP-43 co-pathology is well documented to emerge and intensify in later stages (III-IV) (McKee et al., 2010), further supporting the utility of this system as a comprehensive, pathobiologically grounded tool. However, its critical limitation is that it relies exclusively on post-mortem tissue analysis. In contrast, symptom-based clinical frameworks, originally proposed by Stern and colleagues (Stern et al., 2013) and recently formalized as the Traumatic Encephalopathy Syndrome (TES) consensus criteria (Katz et al., 2021), offer the practical advantage of being applicable in life, providing a clinical entry point for longitudinal monitoring. However, these clinical frameworks do not directly map to specific molecular burdens. While TES criteria effectively capture cognitive and neurobehavioral decline associated with p-tau spread, they may currently lack the sensitivity to accurately reflect atypical phenotypes, such as motor deficits driven by TDP-43 co-pathology (Katz et al., 2021; McKee et al., 2010). Together, these two frameworks are complementary rather than competing: McKee staging provides the pathobiological gold standard, while the Stern/TES clinical framework offers a pragmatic diagnostic scaffold. Bridging these two systems, ideally through fluid or imaging biomarkers that accurately reflect both p-tau and TDP-43 burden, remains one of the central challenges in the field. Importantly, these frameworks do not align temporally or mechanistically, as neuropathological staging reflects cumulative molecular burden, whereas clinical criteria capture symptom expression that may lag, overlap, or diverge from underlying pathology. Anatomical changes in CTE Although many clinical manifestations have been described in individuals with a history of repetitive brain trauma, definitive diagnosis of chronic traumatic encephalopathy (CTE) still relies on post-mortem analysis. Histochemical and immunohistochemical studies reveal multifocal or diffuse hyperphosphorylated tau pathology in cortical and subcortical regions, often accompanied by fibrillary astrogliosis and microglial activation (Petraglia et al., 2014; McKee et al., 2013) (Fig. 2B). These structural and cellular changes are not uniformly distributed, but instead reflect region-specific vulnerability to repetitive mechanical stress, metabolic demand, and neuroinflammatory signaling. As pathology progresses, CTE is associated with widespread structural brain changes. These include reduced brain weight; atrophy of gray and white matter, particularly in the frontal and medial lobes; and degeneration of the thalamus, hypothalamus, and mammillary bodies. Ventricular enlargement, cavum septum pellucidum, septal fenestrations, thinning of the corpus callosum, and depigmentation of the locus coeruleus and substantia nigra are also frequently reported (McKee et al., 2015). The location of atrophy correlates with clinical symptoms. Frontal and temporal lobe degeneration contributes to irritability, impulsivity, aggression, and cognitive impairment, including memory loss and poor judgment (Alosco et al., 2021a; McKee et al., 2015). Damage to medial temporal and limbic structures is linked to depression, anxiety, and short-term memory deficits (Stern et al., 2011; Stein et al., 2014). Thalamic and hypothalamic atrophy impairs attention and information processing and can disrupt autonomic regulation of blood pressure, heart rate, and respiration. Loss of mammillary body volume further compromises memory formation, while degeneration of the substantia nigra and locus coeruleus may underlie Parkinsonian motor features (Stern et al., 2011; McKee et al., 2015). Thinning of the corpus callosum can impair interhemispheric communication and higher-order cognition. These anatomical changes develop gradually over years or decades. This progression likely reflects the cumulative effects of persistent tau pathology, ongoing neuroinflammation, and impaired cellular repair mechanisms, which together drive the spread of neurodegeneration from initially vulnerable regions to broader cortical and subcortical networks. Clinically, early stages are often characterized by mood and behavioral changes, with later progression to cognitive impairment and motor symptoms. Ultimately, widespread tau pathology and neurodegeneration may culminate in dementia (Montenigro et al., 2015; Stein et al., 2014). These anatomical changes are closely linked to underlying molecular alterations, particularly tau modifications and other proteinopathies, which provide mechanistic insight into disease progression. Molecular pathology underlying CTE progression The neuropathology of chronic traumatic encephalopathy (CTE) reflects the cumulative and progressive consequences of traumatic brain injury (TBI), arising from sustained disruption of cellular homeostasis and intercellular signaling, and includes tau modifications, amyloid-beta (Aβ) deposits, neuronal and axonal loss, neuroinflammation, and disruption of the blood-brain barrier (Johnson et al., 2012) (Fig. 2B). These processes do not occur in isolation, but interact to drive a self-sustaining cycle of injury in which neuroinflammation, metabolic dysfunction, and impaired protein clearance reinforce one another, promoting progressive accumulation of pathological proteins. Among these, abnormal phosphorylation of the microtubule-associated protein tau is considered the defining molecular feature of CTE (McKee et al., 2016b). Tau normally stabilizes microtubules and supports axonal transport. Aberrant phosphorylation reduces tau's affinity for microtubules, particularly at residues such as Ser262 and Thr231 (Katsumoto et al., 2019; Alonso et al., 2010), leading to microtubule destabilization. Detached tau proteins self-aggregate into insoluble paired helical filaments and higher-order assemblies that disrupt cellular integrity (von Bergen et al., 2005). In early-stage CTE, hyperphosphorylated tau is found in clusters of perivascular neurons and glia, concentrated at the depths of cortical sulci (McKee et al., 2023; McKee et al., 2016b). With disease progression, these initially patchy deposits spread to involve widespread cortical, medial temporal, and subcortical regions (McKee et al., 2023; McKee et al., 2016b). This propagation is thought to occur through activity-dependent release and uptake of pathological tau species, as well as through network-based vulnerability of connected neuronal circuits, although the precise mechanisms remain incompletely defined. Other post-translational modifications, including acetylation, ubiquitination, methylation, and glycosylation, are also implicated in tau pathology, although their contribution to the spread of disease remains unclear (McKee et al., 2023; Gupta et al., 2021). At the microscopic level, CTE is characterized by p-tau–positive neurofibrillary tangles (NFTs), neuropil threads (NTs), and glial tangles (GTs) within the frontal and temporal cortices. TDP-43 inclusions are also present, particularly in cases with advanced tau pathology, adding another layer of proteinopathy to the disease process. In contrast, Aβ deposition is inconsistent in CTE, whereas neuritic Aβ plaques are a central feature of Alzheimer's disease (AD) (McKee et al., 2015). In parallel, disruption of the blood-brain barrier may exacerbate disease progression by permitting peripheral immune cell infiltration and impairing clearance of toxic protein species, further amplifying local neuroinflammatory and degenerative processes. Although the tau isoforms found in CTE are indistinguishable from those in AD, their distribution is distinct. In CTE, lesions cluster at the depths of sulci, around small blood vessels, and within superficial cortical layers (Katsumoto et al., 2019; McKee et al., 2015). By comparison, in AD, tau aggregates typically localize to deeper cortical layers and are most prominent in CA1 of the hippocampus, whereas in CTE, aggregates span CA1-CA4 (Turner et al., 2016). Taken together, the co-occurrence of TDP-43 pathology, inconsistent Aβ deposition, and unique anatomical patterns of tau accumulation indicate that CTE is a distinct disease entity with a molecular profile that directly shapes its clinical presentation. While the anatomical correlates of gross structural atrophy in CTE are increasingly well-characterized (Section 2.2), a critical and underappreciated dimension is how these specific molecular burdens of p-tau and TDP-43 map onto the emergence of distinct clinical symptom profiles. In early-stage CTE (Stages I-II), the focal p-tau deposition at the depths of cortical sulci and around perivascular spaces involves early engagement of the locus coeruleus and amygdala. This anatomical pattern provides a plausible molecular basis for the early predominance of behavioral and mood disturbances, including irritability, impulsivity, and depression, as the locus coeruleus is a primary source of noradrenergic input regulating affect and arousal, and amygdala involvement directly modulates emotional reactivity. As the disease advances to Stages III-IV, p-tau spreads to involve the medial temporal lobe (including hippocampal subfields CA1-CA4) and progressively engages widespread neocortical and subcortical regions. A recent large-scale clinicopathological study of 364 autopsy-confirmed CTE cases demonstrated that higher regional p-tau burden was significantly associated with worse cognitive and functional outcomes (Alosco et al., 2024). Behavioral dysregulation and depression showed a weaker correlation with regional p-tau burden compared to cognitive symptoms, suggesting that mood symptoms may reflect early locus coeruleus and amygdala involvement that is less tightly coupled to overall tau spread (Alosco et al., 2024). TDP-43 co-pathology, which intensifies in Stages III-IV and is present in nearly all Stage IV cases, adds a distinct layer of clinical severity. Cases with prominent TDP-43 inclusions show greater cognitive decline and, in a subset of individuals, progression to ALS-like motor neuron disease with profound weakness, spasticity, and bulbar involvement. This suggests that TDP-43 co-aggregation represents a molecular tipping point that accelerates disease progression and broadens the clinical phenotype (McKee et al., 2010). Together, these observations suggest that the behavioral versus cognitive subgroup distinction proposed by Stern may correspond to fundamentally different anatomical and molecular stages of the same underlying pathological cascade, highlighting the critical need for next-generation in vivo biomarkers capable of differentiating regional p-tau from TDP-43 burden Defining CTE in life: biomarkers, genetic susceptibility, and diagnostic approaches Biomarkers Over the past decade, the search for biomarkers of chronic traumatic encephalopathy (CTE) has accelerated, driven by the need to define the disease in life despite its current reliance on post-mortem diagnosis. Early neuropathological observations linking repetitive mild injury to tau pathology highlighted the need for reliable diagnostic markers (Omalu, 2014). Since 2011, PubMed has reported a 20- to 30-fold increase in publications on CTE and related biomarker research. Despite rapid expansion, no biomarker has yet demonstrated sufficient sensitivity and specificity to distinguish CTE from other neurodegenerative conditions or to reliably track disease progression. Current approaches largely adapt biomarker strategies from other neurodegenerative diseases, including Alzheimer's disease (AD), frontotemporal dementia (FTD), and Lewy body disease, although their applicability to CTE remains uncertain. Imaging-based methods such as magnetic resonance imaging (MRI), functional MRI (fMRI), susceptibility-weighted imaging (SWI), diffusion tensor imaging (DTI), magnetic resonance spectroscopy (MRS), and positron emission tomography (PET) have been adapted to detect structural and functional brain changes after repetitive injury (Table 1). These imaging modalities provide valuable insight into structural and functional changes, but often lack specificity for CTE, as many of the detected alterations overlap with other neurodegenerative or injury-related conditions. In parallel, fluid-based markers measured in blood or cerebrospinal fluid (CSF), including proteins, nucleic acids, and metabolites, are under investigation, offering the advantage of potential use in field or acute care settings (Selvam and Ayyavoo, 2024). Fluid-based biomarkers offer advantages in accessibility and potential for longitudinal monitoring, but similarly face challenges in distinguishing CTE-specific pathology from broader neuroinflammatory or neurodegenerative processes. Imaging modalities Imaging biomarkers in CTE aim to capture complementary aspects of disease, ranging from functional network disruption and white matter integrity to metabolic alterations and molecular pathology, although each modality provides only a partial view of the underlying disease process. With respect to imaging biomarkers, functional MRI (fMRI) provides insight into functional alterations in addition to structural information from conventional MRI. fMRI has two key advantages: it can assess brain activity in real time, and it is sensitive to working memory deficits following sub-concussive trauma. However, its use requires preinjury baseline scans for meaningful interpretation and is limited by high cost and availability. Kim and colleagues used perfusion-based fMRI to evaluate cognitive dysfunction after traumatic brain injury (TBI) (Kim et al., 2012). Patients with chronic moderate-to-severe TBI demonstrated resting hypoperfusion and abnormal activation of sensory cortices, particularly the occipital and temporal regions, during working-memory tasks, whereas healthy controls exhibited the expected deactivation. The correlation between task performance and regional activation differed between groups, suggesting altered network organization in individuals with TBI (Kim et al., 2012). Together, these findings indicate that fMRI is particularly sensitive to functional network disruption, but its dependence on baseline comparison and variability across individuals limits its utility as a standalone diagnostic tool for CTE. Diffusion tensor imaging (DTI) has also been widely used to detect diffuse axonal injury by mapping water diffusion along white matter tracts. Radiologic and clinical deficits often correlate well in DTI, and preliminary work has shown associations between DTI measurements and concussion history. Its main limitation is that tractography, required to reconstruct axonal pathways, can be time-intensive. Holleran and colleagues examined superior frontal cortex samples from 10 patients with confirmed CTE and found that axonal disruption at sulcal depths co-localized with tau deposition, suggesting a spatial link between white matter injury and tau pathology. They also reported diffusion abnormalities associated with astroglial scarring and microglial activation, indicating that DTI may capture both axonal damage and secondary inflammatory processes that contribute to CTE pathology (Holleran et al., 2017). DTI therefore provides strong sensitivity to axonal injury and structural connectivity changes but cannot directly distinguish between primary mechanical damage and downstream neurodegenerative processes. Magnetic resonance spectroscopy (MRS) offers a non-invasive tool to assess metabolites associated with neuronal and axonal injury. MRS provides insight into underlying pathophysiology and can detect metabolic alterations that correlate with clinical symptoms, although distinguishing injury-related changes from those due to normal aging remains challenging. Lin and colleagues applied a two-dimensional localized correlated spectroscopy technique (L-COSY) to former professional athletes with a history of multiple concussions or repetitive sub-concussive trauma. At the time of evaluation, all participants exhibited neurological and behavioral symptoms consistent with CTE. Compared to controls, the athletes showed elevated glutamate/glutamine and choline levels, highlighting the potential of MRS to detect metabolic disturbances linked to CTE (Lin et al., 2015). While MRS offers insight into metabolic dysfunction, its specificity remains limited, as similar metabolic changes are observed in aging and other neurodegenerative conditions. Positron emission tomography (PET) provides another important modality, enabling the detection of regional metabolic abnormalities and the in vivo measurement of tau deposition using specific tracers. Its strengths include the ability to identify brain regions with decreased metabolism and quantify molecular pathology, although exposure to radioactive isotopes limits its repeatability. Stern and colleagues used flortaucipir PET (tau) and florbetapir PET (Aβ) to study former NFL players with cognitive and neuropsychiatric symptoms compared with asymptomatic controls (Stern et al., 2019). The players demonstrated increased tau deposition in the superior frontal, medial temporal, and left parietal cortices, regions typically implicated in CTE, while amyloid-β was largely absent. These findings support tau PET as a promising biomarker for CTE and underscore its distinct pathology compared to AD (Stern et al., 2019). Among imaging modalities, PET offers the most direct link to molecular pathology, particularly tau burden, but its interpretation in CTE remains complicated by off-target binding and the lack of tracers specifically validated for CTE pathology. Taken together, imaging modalities provide complementary but incomplete insight into CTE, with functional, structural, metabolic, and molecular readouts that must be integrated to approximate the underlying disease process. Fluid biomarkers In parallel with imaging approaches, fluid biomarkers have been explored (Table 2). Blood is a particularly attractive source given its accessibility and minimally invasive collection. Tau protein has been widely studied due to its central role in neuronal structure and function, but other blood markers are also relevant. Neurofilament light chain (NFL) reflects axonal injury, glial fibrillary acidic protein (GFAP) reflects astrocytic damage, S100B is released after astroglial injury, and ubiquitin C-terminal hydrolase-L1 (UCH-L1) serves as a neuronal marker (Hier et al., 2021). These blood-based markers provide sensitive indicators of acute injury and cellular damage, but their specificity for CTE is limited, as similar elevations are observed in other forms of brain injury and neurodegenerative disease. The major advantage of blood biomarkers is accessibility; however, their interpretation is complicated by the presence of abundant circulating proteins and systemic inflammatory signals that can obscure brain-derived changes (Hossain et al., 2024). Cerebrospinal fluid (CSF) offers a more direct link to the brain environment. The recent discovery of the glymphatic system has advanced understanding of how CSF and interstitial fluid (ISF) clear metabolites, including tau and amyloid-β, from the brain parenchyma through perivascular pathways regulated by astroglial aquaporin-4 (AQP4) water channels. TBI disrupts this clearance system: experimental evidence shows that glymphatic pathway function is reduced by approximately 60% following TBI, and this impairment persists for more than one month post-injury due to the loss of perivascular AQP4 polarization in reactive astrocytes (Iliff et al., 2014). Importantly, extracellular tau is normally cleared from the brain along these paravascular pathways, and genetic disruption of AQP4 following TBI exacerbates tau aggregation and promotes neurofibrillary pathology, providing a direct mechanistic link between post-traumatic glymphatic failure and the perivascular p-tau accumulation that defines CTE (Iliff et al., 2014). This evidence positions chronic glymphatic dysfunction as a key intermediary in the pathobiological chain linking repetitive head trauma to the hallmark perivascular tau lesions of CTE, and suggests that strategies to restore glymphatic function, such as sleep optimization or AQP4-targeted therapies, may offer a novel therapeutic avenue. CSF biomarker collection via lumbar puncture, while invasive and not always well tolerated, captures a downstream readout of this impaired clearance system. As a result, CSF biomarkers may more closely reflect central pathological processes than blood-based measures, particularly those related to impaired protein clearance and accumulation. Although many candidate CSF biomarkers have been studied, none has shown unique specificity for CTE. However, like blood-based markers, most CSF biomarkers lack disease specificity and overlap with profiles observed in other taupathies and neurodegenerative conditions. Recently, exosomal tau has attracted particular attention. Exosomes are vesicles secreted by neurons and glia that carry molecular cargo reflective of the central nervous system. Measuring tau in exosomes isolated from blood may overcome the challenge of low tau concentrations in plasma and provide a more accurate representation of underlying pathology, representing a promising but still emerging direction for CTE biomarker development, requiring further validation for specificity and reproducibility (Halicki et al., 2023). Taken together, fluid biomarkers provide accessible and potentially longitudinal measures of brain injury and pathology, but currently lack the specificity required to distinguish CTE from other neurodegenerative conditions, highlighting the need for multimodal integration with imaging and clinical data. Genetic factors Despite increasing recognition of chronic traumatic encephalopathy (CTE), its genetic contributions remain poorly defined and are thought to act primarily as modifiers of disease susceptibility and progression rather than direct causal drivers. While strong genetic associations exist for other neurodegenerative disorders, including Huntington's disease, Alzheimer's disease (AD), Parkinson's disease (PD), amyotrophic lateral sclerosis (ALS), and multiple sclerosis (MS), the influence of genetic background on CTE is still unclear. This raises a central question: what role does genetic susceptibility play in determining who develops CTE after repetitive head trauma? Current evidence suggests that genetic factors influence how individuals respond to repetitive injury, shaping vulnerability to neurodegeneration through pathways involving lipid metabolism, protein aggregation, immune activation, and cellular repair. A small set of candidate genes has been repeatedly implicated in CTE, although most overlap with those associated with other neurodegenerative diseases (Cao et al., 2017). One example is Apolipoprotein E [APOE]. The ε4 allele of APOE, a gene involved in cholesterol and lipid metabolism in the brain, has long been associated with AD and is now implicated in CTE as well (Eisenbaum et al., 2024; Atherton et al., 2022). APOE ε4 appears to exacerbate tau hyperphosphorylation and neuroinflammation, processes central to CTE pathology (Gandy and Dekosky, 2012). In athletes or veterans with repetitive head impacts, carriers of APOE ε4 may show increased susceptibility to disease, and mechanistic studies suggest that the allele impairs lipid transport and metabolism, which are essential for neuronal repair after trauma (Mannix and Whalen, 2012). This dysfunction can reduce recovery capacity and increase vulnerability to degeneration (Yang et al., 2023). These candidate genes point toward a common theme in which genetic variation influences the brain's capacity to repair injury, clear pathological proteins, and regulate inflammatory responses following trauma. In addition to APOE, variation in microtubule-associated protein tau [MAPT], which encodes the tau protein, further influences disease susceptibility (Colom-Cadena et al., 2013). Tau stabilizes microtubules in neurons, but variation in MAPT haplotypes influences tau aggregation (Leveille et al., 2021). The H1 haplotype is more often linked to tau aggregation, and studies report a higher prevalence of the H1/H1 genotype in CTE patients (Pedicone et al., 2024). This may influence both the regional deposition of abnormal tau and the severity of neurodegeneration (Han et al., 2025). TMEM106B is also emerging as an important genetic factor (Cherry et al., 2018). This gene regulates lysosomal function and modulates microglial activity. A study of 86 deceased male athletes with confirmed CTE showed that a minor allele of a coding single-nucleotide polymorphism [SNP] in TMEM106B (rs3173615) was associated with reduced phosphorylated tau, lower levels of inflammation, and decreased odds of ante-mortem dementia (Cherry et al., 2018). Variations in TMEM106B may therefore confer partial protection by enhancing lysosomal clearance of aggregates, whereas dysregulation of this gene could impair degradation and promote tau accumulation and cellular stress (Feng et al., 2021). Genome-wide approaches in post-mortem CTE brains have identified additional candidate pathways, although their interpretation is complicated by bias. Brain donors typically had symptoms and histories of repetitive trauma, excluding resilient individuals who never developed disease (Butler et al., 2025). Despite this limitation, genome-wide studies remain important for hypothesis-free discovery. One consistent finding has been upregulation of immune-related genes, particularly CCL2, which encodes chemokine ligand 2 (Cherry et al., 2020b). In postmortem tissue, CCL2 levels in the dorsolateral frontal cortex correlated with years of American football play, microglial and CD68-positive cell densities, and severity of CTE pathology (Cherry et al., 2020b). CCL2 appears to drive microglial recruitment and activation, reinforcing cycles of inflammation and tau deposition. Other transcriptomic analyses show evidence of mitochondrial dysfunction, with altered expression of genes involved in oxidative phosphorylation and the electron transport chain, suggesting that impaired energy metabolism may amplify cellular stress early after repetitive trauma (Labadorf et al., 2023). Region-specific studies have also revealed that the cortical sulcus, where CTE pathology is first detected, shows unique molecular changes. Genes such as MANEA (mannosidase endo-alpha-like) were altered in repetitive head injury cases, while others involved in immune regulation, phagocytosis, synaptic signaling, and extracellular matrix (ECM) organization, including collagens and integrins, were disrupted. Distinct signatures, such as altered expression of JMJD6 (Jumonji Domain-Containing Protein 6) and VAT1L (Vesicle Amine Transport 1-Like Protein), emphasize the unique vulnerability of sulcal regions compared to gyral crests (Cherry et al., 2021). Beyond CTE-specific datasets, broader transcriptomic studies of TBI have highlighted overlapping mechanisms. A study led by the Xue group in China analyzed datasets GSE89866 and GSE104687 and identified hub genes, including RPL27 (ribosomal protein (RP) L27), RPS4X (S4, X-linked), RPL23A, RPS15A, and RPL7A, through protein-protein interaction analysis (Menardi et al., 2022). These genes, largely involved in ribosomal function, oxidative stress, and immune signaling, were significantly downregulated in TBI brains (Menardi et al., 2022). Pathway analyses implicated disruption of neurodevelopmental processes, including hippocampal neuron development, as well as immune-inflammatory signaling mediated by interleukins (IL-9, IL-12, IL-27, IL-35) and interferon-γ. Metabolic pathways, including oxidative phosphorylation, mTOR signaling, pyruvate metabolism, and mitophagy, were also affected, alongside ribosome biogenesis and cell cycle regulation. These findings point to widespread cellular dysfunction that may contribute to long-term vulnerability to CTE. These transcriptomic changes suggest that repetitive injury does not simply induce isolated molecular alterations, but instead drives coordinated dysfunction across metabolic, immune, and structural pathways that may predispose the brain to progressive degeneration. Current evidence supports a model in which genetic background modulates the response to repetitive brain injury, influencing disease risk, regional vulnerability, and progression rather than determining disease onset independently. Candidate genes such as APOE, MAPT, and TMEM106B highlight mechanisms involving lipid metabolism, tau aggregation, and lysosomal function. Genome-wide and transcriptomic studies implicate immune responses such as CCL2 upregulation, mitochondrial dysfunction, ECM remodeling, and ribosomal stress. While most human data come from late-stage post-mortem brains, integrating genomic data from asymptomatic individuals, patients with mild repetitive trauma, and experimental models will be essential for disentangling causal drivers from secondary changes and for identifying new biomarkers and therapeutic targets. These genetic insights underscore the need to integrate molecular susceptibility with diagnostic frameworks and emerging biomarkers to enable earlier and more precise identification of CTE during life. Emerging imaging modalities for diagnostics of CTE Efforts to diagnose chronic traumatic encephalopathy (CTE) during life remain limited, largely due to the absence of biomarkers that can reliably distinguish CTE from other neurodegenerative and neuropsychiatric conditions (Table 1). One key approach is the use of clinical diagnostic criteria under the umbrella of traumatic encephalopathy syndrome (TES). These criteria require substantial exposure to repetitive head impacts, core clinical features such as cognitive impairment (most often affecting memory or executive function) and/or neurobehavioral dysregulation, a progressive course of symptoms, and exclusion of other conditions that could account for the presentation. These TES criteria were refined through a modified Delphi process by a multidisciplinary panel forming the NINDS consensus diagnostic criteria published in 2021 (Katz et al., 2021). While these criteria provide a structured clinical framework, they rely on symptom patterns that are not specific to CTE and do not directly reflect underlying molecular pathology. However, multiple proposed diagnostic schemas exist, and no consensus has yet been adopted universally (Ritter et al., 2023), and research continues to assess their validity and specificity (Iverson et al., 2021). Imaging techniques have been adapted to support diagnosis, but their clinical utility remains constrained by limited specificity and inconsistent correlation with pathology. Conventional MRI can identify structural abnormalities such as cavum septum pellucidum, which has been associated with CTE, and microhemorrhages that may reflect diffuse axonal injury, though these findings are not specific to CTE. Diffusion tensor imaging (DTI) is used to assess white matter integrity by measuring directional water diffusion. Indices such as decreased fractional anisotropy (FA), reduced axial diffusivity (AD), and increased radial diffusivity (RD) may reflect axonal injury or demyelination (Winklewski et al., 2018). Some studies have shown meaningful structural–clinical correlations, but others, including a study of retired football players, found no statistically significant association between DTI metrics and neuropsychological outcomes (Hart Jr. et al., 2013). These inconsistencies highlight the challenge of using structural measures as diagnostic markers, as similar abnormalities are observed across multiple forms of brain injury and neurodegeneration. Magnetic resonance spectroscopy (MRS) offers metabolic information by quantifying neurometabolites that may reflect neuronal injury, axonal disruption, or neuroinflammation (Tosh et al., 2023; Koerte et al., 2015). Although several investigations in athletes have reported metabolite abnormalities, the lack of pathological confirmation means it is unclear whether these represent true CTE-specific changes (Lin et al., 2015; Henry et al., 2011). This limits the ability of MRS to serve as a diagnostic tool, as metabolic alterations lack clear disease specificity. Susceptibility-weighted imaging (SWI), which highlights magnetic susceptibility contrasts (e.g., from microbleeds or iron deposition), has also been used. In one study of 45 retired National Football League players, microhemorrhages were observed in only ~9% of participants, and their presence was not correlated with neurocognitive or psychiatric symptoms, casting doubt on its utility for CTE diagnosis (Casson et al., 2014). These findings further underscore the limited diagnostic value of SWI in isolation. Positron emission tomography (PET) is of particular interest because it has the potential to detect molecular pathology in vivo. Early tracers such as FDDNP bind neurofibrillary tangles and related proteins, but lack specificity because they also bind amyloid-β and hyperphosphorylated tau seen in Alzheimer's disease (Chen et al., 2018). Flortaucipir is a more tau-focused tracer, though it may preferentially bind the paired helical filaments characteristic of Alzheimer's, raising concerns about false negatives in CTE. In another study, Chen et al. (2018) used FDDNP-PET to compare 7 military personnel with suspected CTE, 15 retired football players with histories of mTBI and neuropsychiatric symptoms, 24 Alzheimer's patients, and 28 controls (Chen et al., 2018). The military and athlete groups showed FDDNP binding patterns distinct from Alzheimer's and significantly different from controls, supporting the potential of PET imaging to detect CTE-related pathology in vivo (Chen et al., 2018). Despite this promise, variability in tracer binding and overlap with other tauopathies continue to limit its reliability as a definitive diagnostic modality. Additional work has shown that postmortem autopsy confirmation correlates with in vivo FDDNP signals in brain regions consistent with tau deposition in CTE (Chen et al., 2018) (Table 1). Taken together, current diagnostic approaches remain insufficient when used in isolation. Clinical criteria capture symptom expression but lack molecular specificity, while imaging modalities provide structural and molecular insight but cannot definitively distinguish CTE from other conditions. These limitations highlight the need for integrated, multimodal approaches that combine clinical, imaging, and fluid biomarkers to improve diagnostic accuracy in life. Promising approaches to studying CTE In vivo models For decades, the pathological gold standard for chronic traumatic encephalopathy (CTE) has been post-mortem human brain examination, which defined the disease's signature lesions, mapped their anatomical distribution, and established staging frameworks that anchor the field (McKee et al., 2013; McKee et al., 2015) (Fig. 3A). This work delineated perivascular, sulcal-depth accumulations of hyperphosphorylated tau, clarified co-pathologies, and linked regional vulnerability with clinical phenotypes, providing fundamental knowledge for further hypothesis generation (McKee et al., 2023; McKee et al., 2013; McKee et al., 2015). Building upon these insights, non-invasive imaging in living humans provides valuable information on structure, connectivity, metabolism, and molecular burden, although it remains limited in its ability to resolve causal mechanisms (Kim et al., 2012; Holleran et al., 2017; Lin et al., 2015; Stern et al., 2019). Despite these advances, critical limitations remain. Post-mortem research, by definition, captures disease's late stages and cannot resolve initiating events; brain banks are biased toward donors with known head-impact histories and symptoms; and even the best clinical imaging cannot routinely deliver causal interrogation of mechanisms. Across human studies, ethical constraints, high costs, and heterogeneity in exposure, genetics, and comorbidities further complicate inference and intervention (Petraglia et al., 2014; McKee et al., 2013; McKee et al., 2015; Kim et al., 2012; Holleran et al., 2017; Lin et al., 2015; Stern et al., 2019). In this context, experimental in vivo models emerged to address gaps that human work cannot readily fill (Fig. 3B). Rodent and large-animal platforms recreate defined biomechanics and timelines, enabling observation from the earliest minutes through progression and resolution, systematic control of injury dose and interval, and targeted pathway perturbation under reproducible conditions (Casson et al., 2014; Machado et al., 2023). These models are particularly valuable for defining causal relationships between injury parameters and downstream molecular and cellular responses, which cannot be directly tested in human studies. Widely used paradigms, including controlled cortical impact, weight-drop, lateral fluid percussion (LFP), and blast-overpressure exposures, allow investigators to probe how repeated low-level insults accumulate into persistent cellular and network dysfunction (Osier and Dixon, 2016; Machado et al., 2023; Namjoshi et al., 2014). They therefore serve as controlled platforms to dissect how injury dose, frequency, and biomechanics shape the trajectory from acute damage to chronic neurodegeneration. The LFP model, in particular, is among the most extensively validated rodent paradigms for TBI, producing a graded injury by applying fluid-pressure pulses to the intact dura. Repetitive LFP protocols have been shown to induce progressive tau phosphorylation, axonal injury, and neuroinflammation that partially recapitulate early-stage CTE pathology, making it a widely used platform for studying the transition from acute TBI to chronic neurodegeneration (Shultz et al., 2012; Fronczak et al., 2022). Across species, these approaches recapitulate elements of the human condition, such as progressive tau phosphorylation, white-matter injury, neuroinflammation, and behaviorally measurable deficits, with reports of corpus callosum involvement and neocortical p-tau after repetitive or blast injuries and severity-dependent cognitive and motor impairment in larger brains (Sugahara et al., 2025; Schwerin et al., 2021; Kinder et al., 2019). At the same time, these models have important constraints: species differences in cytoarchitecture, immune system, and lifespan limit direct translation; many models do not develop decades-scale neurodegeneration without genetic manipulation, risking divergence from native human pathobiology; and scaling to species with closer brain mechanics imposes substantial costs, infrastructure demands, and ethical burden (Casson et al., 2014; Machado et al., 2023; Kinder et al., 2019). Inducing injuries in animals remains an ethically challenging endeavor, even when scientifically justified as essential to understanding disease mechanisms. These realities motivate complementary strategies that retain experimental control and causal access while reducing reliance on high-burden animal cohorts. A pivotal question in experimental CTE research is not merely whether a given model induces brain injury, but whether it accurately replicates the specific pathological hallmarks that characterize CTE in humans (McKee et al., 2023; Turner et al., 2015b). In rodent models, repetitive closed-head impact and LFP paradigms consistently reproduce perivascular tau phosphorylation, white matter axonal injury, microglial activation, and memory and executive function deficits, features that correspond to early-stage CTE (Stages I-II) (Petraglia et al., 2014; Namjoshi et al., 2014; Sugahara et al., 2025). However, the characteristic sulcal-depth localization of p-tau lesions, a defining neuropathological feature of human CTE, is difficult to model in lissencephalic rodent brains, which lack the cortical folding that concentrates strain at the depths of the sulci (Fagan et al., 2020; Holleran et al., 2017). Large-animal models, particularly ferrets, pigs, and sheep, offer gyrencephalic (folded) brain architecture that more closely approximates the biomechanical environment of human TBI, enabling better recapitulation of sulcal-depth strain patterns and the perivascular lesion distribution that defines CTE pathology (Schwerin et al., 2021; Kinder et al., 2019). In vitro 3D human models address a distinct yet complementary set of features: they can recapitulate early post-injury mitochondrial dysfunction, neuroinflammatory signaling, tau phosphorylation at the cellular level, and neurovascular unit disruption within a human genetic background, and can do so longitudinally over extended periods (Liaudanskaya et al., 2023; Hinrichsen et al., 2026; Liaudanskaya et al., 2020; Rouleau et al., 2020; Stanton et al., 2025). Together, these model systems are not redundant but complementary, with each addressing distinct aspects of CTE pathobiology. Importantly, no single model recapitulates the full spectrum of CTE, reinforcing the need for integrative approaches that combine mechanistic insight, biomechanical relevance, and human-specific biology. In vitro models In vitro systems address key limitations of human and in vivo models by providing human-relevant biology in controllable and scalable experimental formats (Fig. 3C). Three-dimensional engineered tissues and advanced co-cultures permit direct access to the earliest post-injury minutes and hours, longitudinal tracking over weeks to months, and causal interrogation of pathways with genetic or pharmacological perturbations that are impractical in people and costly in animals (Liaudanskaya et al., 2023; Xie et al., 2024; Cairns et al., 2016; Coe et al., 2025; Kansakar et al., 2025; Hinrichsen et al., 2025). In vitro platforms allow precise control over cellular composition, enabling investigators to define which cell types are included and in what proportions. Neurons, astrocytes, microglia, oligodendrocytes, endothelial cells, and pericytes can be combined in defined proportions, with additional control over developmental stage and donor background, including the use of patient-derived induced pluripotent stem cells that capture specific risk alleles and sex-linked differences (Liaudanskaya et al., 2023; Stanton et al., 2025; Hinrichsen et al., 2025). The type of injury applied can also be tuned to reflect human exposures, including controlled indentation, uniaxial or biaxial stretch, pressure pulses, or blast-like overpressure. Readouts from these systems span multiple biological scales, from mitochondrial and cytoskeletal integrity at the organelle level to synaptic function and multicellular network activity (Liaudanskaya et al., 2023; Liaudanskaya et al., 2020; Rouleau et al., 2020; Xie et al., 2024; Cairns et al., 2016; Coe et al., 2025; Kansakar et al., 2025; Hinrichsen et al., 2025; Liaudanskaya et al., 2019; Snapper et al., 2023; Jgamadze et al., 2023; Lancaster et al., 2013; Lovett et al., 2020; Slanzi et al., 2020). Compared with animal work, these models reduce ethical burden, lower experimental costs, enable higher-throughput designs, and make it possible to test mechanisms iteratively and systematically. Despite these advantages, in vitro models have important limitations. They do not reproduce whole-body homeostasis, systemic immune responses, or lifestyle effects on epigenetic background. Some constructs are mechanically fragile and cannot withstand high-strain paradigms, and no in vitro system fully captures the variability of human exposures and comorbidities (Xie et al., 2024). Nonetheless, long-term cultures can support maturation and aging trajectories beyond what is possible in short-lived animal experiments, allowing analysis of disease initiation, progression, and partial recovery with resolution at the level of specific cell types (Liaudanskaya et al., 2023; Xie et al., 2024; Cairns et al., 2016; Coe et al., 2025; Kansakar et al., 2025; Hinrichsen et al., 2025). This capability uniquely positions in vitro systems to dissect early pathobiological events and cell-type-specific responses that are not accessible in human or animal studies. Critically, these models address several limitations of human and in vivo studies: they allow access to the earliest disease stages not captured in post-mortem tissue, they permit causal testing of pathways that cannot be interrogated through observational imaging, they incorporate human genetics without cross-species confounds, and they reduce reliance on large animal cohorts while retaining the ability to model repetitive, sub-concussive injuries (Liaudanskaya et al., 2023; Xie et al., 2024; Cairns et al., 2016; Coe et al., 2025; Kansakar et al., 2025; Hinrichsen et al., 2025; Snapper et al., 2023). In this way, in vitro systems complement human and animal research by linking exposure to cell-type-specific injury responses and potential biomarkers, thereby accelerating translational discovery without replacing other approaches. Among available platforms, only a subset can reliably withstand and reproduce traumatic injury paradigms (Lovett et al., 2020). Three-dimensional scaffold-based cultures currently provide the most stable environments for applying controlled injuries within multicellular human systems, whereas organoids, assembloids, and microfluidic chips hold strong potential for mechanistic discovery but require further optimization to ensure reproducibility under injury conditions. For example, recent work by our group using long-term, multicellular human 3D scaffold-based brain models has demonstrated the ability to reproducibly model key features of injury-induced neurodegeneration in a controlled human system (Jun et al., 2025). These platforms integrate neurons, astrocytes, microglia, and vascular components within a structurally stable scaffold, enabling the application of defined mechanical injury paradigms and longitudinal tracking of pathological progression. Following injury, these systems recapitulate early mitochondrial dysfunction, sustained neuroinflammatory signaling, and progressive tau phosphorylation, while also capturing neurovascular unit disruption and impaired intercellular communication. Importantly, the ability to maintain these cultures over extended periods allows interrogation of how acute injury evolves into chronic pathological states, providing a direct experimental bridge between repetitive trauma and neurodegenerative-like phenotypes in a human biological context. Moving forward, progress will depend on combining injury-capable engineered tissues with emerging organoid and chip technologies to create models that capture both the complexity and chronicity of human CTE. Taken together, in vivo and in vitro model systems are not ends in themselves, but essential components of a broader translational pipeline linking biomechanical exposure to molecular pathology and, ultimately, clinical disease. Rodent and large-animal models provide mechanistic access to the earliest post-injury cascades and the biomechanical determinants of sulcal-depth lesion formation, while human 3D in vitro systems offer genetic fidelity, cell-type resolution, and longitudinal access to pathobiological progression in a human biological background (Liaudanskaya et al., 2023; Hinrichsen et al., 2026; Rouleau et al., 2020; Stanton et al., 2025). The biomarkers emerging from these systems, including p-tau, NfL, GFAP, and markers of TDP-43 co-pathology, are the same analytes being validated in living patient cohorts through initiatives such as the DIAGNOSE CTE project (Alosco et al., 2021b). This convergence between experimental and clinical biomarker discovery represents one of the most promising paths toward establishing the in-life diagnostic tools that the field critically lacks. As 3D human platforms become increasingly injury-capable and longitudinally stable (Liaudanskaya et al., 2023; Hinrichsen et al., 2026; Rouleau et al., 2020; Stanton et al., 2025), their potential to serve as direct preclinical testing grounds for candidate biomarkers and therapeutics, before validation in costly animal studies or clinical trials, will only grow. Importantly, no single platform captures the full spectrum of CTE, and progress will depend on integrating mechanistic insight from controlled experimental systems with biomarker validation in human cohorts. Future directions In the absence of disease-modifying treatments and reliable in-life diagnostics, current research on chronic traumatic encephalopathy (CTE) has focused on building an integrated mechanistic framework that distinguishes it from other neurodegenerative disorders. Although the clinical profile of CTE is increasingly recognized, consensus diagnostic criteria remain unsettled, consensus criteria remain unsettled, largely because individuals with repetitive mild traumatic brain injury (mTBI) are typically not studied prospectively, and diagnosis is still confirmed only at autopsy. Without the ability to diagnose CTE in vivo and without disease-specific therapies, the field has prioritized the development of human-relevant three-dimensional models, standardized injury paradigms, and the validation of imaging and fluid biomarkers as a foundation for translation. Moving forward, progress will depend on integrated, data-driven approaches that combine biomechanics, imaging, and molecular markers to yield actionable diagnostic and therapeutic strategies. Machine learning represents one promising avenue to bridge existing diagnostic gaps. Computational models have the potential to recognize neuropathological patterns that are routinely identified in post-mortem brains with CTE and translate them into diagnostic markers for living patients. In a recent study, Louis et al. applied proton magnetic resonance spectroscopy (MRS) combined with machine learning to former NFL players, using five metabolites (creatine, choline, N-acetyl-aspartate, glutamate, and macromolecules) to classify impairment status. Their model achieved 75% accuracy, while a localized correlated spectroscopy (L-COSY) model improved accuracy to 87%, underscoring both the utility of imaging biomarkers and the potential of machine learning to enhance their diagnostic power (Louis et al., 2017). These findings highlight the potential of combining imaging biomarkers with computational approaches, although further validation across diverse cohorts is required. Another emerging direction centers on the use of exosomes as a diagnostic tool. Exosomes are extracellular vesicles secreted by living cells that carry proteins and nucleic acids reflective of their cellular origin. In the context of brain injury, neuron-derived exosomes (NDEs) are of particular interest, as they provide a minimally invasive window into central nervous system pathology. Candidate biomarkers identified in NDEs include CD81, total tau, and the inflammatory mediator IL-10, all of which have been proposed as indicators of CTE progression (Ge et al., 2022; Goetzl et al., 2019). Because exosomes can be isolated from peripheral blood, they may overcome the problem of low analyte concentration that has limited the specificity of traditional blood-based assays. However, their diagnostic specificity and reproducibility across populations remain to be established. In parallel, fluid biomarkers from cerebrospinal fluid (CSF), blood, and even saliva are being studied as potential diagnostic tools. Leading candidates include phosphorylated tau isoforms, amyloid beta, glial fibrillary acidic protein (GFAP), neurofilament light chain (NFL), and soluble TREM2 (sTREM2), each reflecting distinct aspects of injury-induced pathology. The DIAGNOSE CTE project, a major initiative, is prospectively collecting biofluid samples from living individuals with high exposure to repetitive head impacts and plans to compare these samples with post-mortem measures from the same individuals, representing one of the most systematic attempts to link fluid biomarkers with neuropathological confirmation (Alosco et al., 2021b). Building on this framework, a compelling future direction involves coupling fluid biomarker collection with real-time impact-monitoring technologies, such as sensor-equipped mouthguards and instrumented helmets that continuously quantify head acceleration, impact frequency, and cumulative mechanical dose during athletic competition and military operations. By integrating biomechanical exposure data with longitudinal fluid biomarker trajectories, including p-tau, NfL, and GFAP, it may be possible to establish a quantitative dose-response relationship between physical trauma and biological disease progression. Such an approach has the potential to transform CTE research from a retrospective, postmortem discipline into a prospective, longitudinally monitored field. A key emerging direction in CTE modeling is the integration of microfluidic technologies with advanced 3D human brain platforms. Current state-of-the-art multicellular systems, such as the miBrain model, which incorporates all six major brain cell types, including neurons, astrocytes, microglia, oligodendrocytes, pericytes, and brain microvascular endothelial cells derived from patient iPSCs, have demonstrated functional BBB properties, neuronal activity, and tau phosphorylation in a human genetic background (Hinrichsen et al., 2026; Stanton et al., 2025). However, these systems currently operate under static culture conditions, limiting their ability to model the dynamic hemodynamic and mechanotransductive forces that govern neurovascular unit (NVU) function in vivo. Integrating microfluidic perfusion into such platforms, creating true ‘Brain-on-a-Chip’ configurations, would enable simulation of cerebrovascular flow, shear stress, and the propagation of pressure waves through the NVU that occur during repetitive head impacts (Campisi et al., 2018). Such systems could provide a mechanistic bridge between biomechanical exposure and molecular pathology by modeling both cellular injury responses and disruption of perivascular clearance pathways. Ultimately, progress in CTE research will depend on establishing a unified framework that integrates biomechanical exposure, molecular pathology, and clinical presentation. Although substantial progress has been made in identifying imaging and fluid markers, advancing in vitro and in vivo models, and developing computational approaches, large gaps remain. Prospective longitudinal studies that integrate these findings across domains will be essential to transition the field from retrospective post-mortem recognition toward prevention, early diagnosis, and therapeutic intervention for those living with or at risk of CTE. Achieving this integration will require coordinated efforts across experimental models, biomarker development, and longitudinal human studies to enable early diagnosis and targeted therapeutic intervention.