White Matter and Cognition in Traumatic Brain Injury

INTRODUCTION

Traumatic brain injury (TBI) affects some 10 million people worldwide every year [1], and in the United States a recent estimate holds that some 42% of the population suffers at least one TBI during their lifetime [2]. Falls and road traffic accidents account for more than 50% of all cases, with peak TBI incidence found in adults more than 50 years old, but military service members represent a unique at-risk group because of TBI related to combat and training exposures [1]. Men tend to be at higher risk of TBI than women, yet women may endure more prolonged post-TBI symptoms [1]. Children and young adolescents appear to be at risk for slower recovery than adults, and early participation in contact sports may result in a higher likelihood of long-term cognitive sequelae; similarly, older people with TBI likely have a higher risk of dementia than younger adults [1]. Cognitive dysfunction is a central clinical feature of all TBI, contributing significantly to both acute and chronic disability. Moreover, emerging associations between TBI and neurodegenerative disease, both international public health concerns, underscore the special urgency of this problem. Despite these observations, however, major questions remain unanswered regarding pathophysiology, long-term sequelae, and treatment. Typically classified as mild, moderate, or severe based on neurological assessment and neuroimaging, TBI may result from penetrating wounds, or, far more commonly, closed head injury related to blows to the cranium, whiplash mechanism injuries, or blast injury sustained in military settings [1, 3]. Whereas moderate and severe TBI often present as neurosurgical emergencies and usually produce clinically obvious cognitive sequelae, mild TBI, which accounts for 80–90% of all cases, also leads to cognitive dysfunction, albeit more subtle and often overlooked [1]. In this regard, the prognosis of mild TBI has been regarded as favorable, with full recovery within days to weeks, but recent data have questioned the veracity of this view, and these patients may experience cognitive symptoms lasting far longer. Most concerning is the potential for progressive cognitive dysfunction, and even dementia, as a result of single TBI, or repetitive mild TBI sustained in military service and various forms of contact sports [1].

In this review, we consider evidence for neuropathology in the white matter of the brain as the primum movens of cognitive dysfunction generated by TBI [1, 4]. While neuropathological changes in this brain component have been recognized for many years, concomitant neuropathology in the gray matter, particularly the cerebral cortex and hippocampus, is also well-known in TBI and naturally attracts much neuroscientific interest. However, converging evidence increasingly suggests that white matter neuropathology is most crucial in the pathogenesis of cognitive dysfunction across the entire range of TBI severity, both acutely and in the long term [1, 4]. Relevant terms used in this review are explained in the Table 1.

WHITE MATTER NEUROBIOLOGY

About half of the brain is comprised of white matter, the term used to refer to grossly visible collections of myelinated tracts coursing throughout the central nervous system, together with blood vessels and associated glial cells: myelin-producing oligodendrocytes, astrocytes, microglia, and oligodendrocyte progenitor cells (OPCs). White matter tracts are bundles of axons, many of which are surrounded by myelin, an insulating substance that confers much greater axonal conduction velocity than in unmyelinated axons, and concomitant increase in cognitive processing speed [5]. Most critical for cognition are the association and commissural tracts, which connect gray matter areas within and between the cerebral hemispheres, and other tracts link the hemispheres with deep structures including the thalamus, basal ganglia, brainstem, and cerebellum; white matter thus participates in all neural networks subserving cognition [5].

The operations of cognition rely heavily on the activity of the modulatory neurotransmitter acetylcholine, which is widely distributed within cerebral white matter tracts [6] and serves an essential role in the facilitation or inhibition of neuronal responses to other neurotransmitters including glutamate, γ-aminobutyric acid, and the monoamines [7]. All sectors of the cerebral cortex and the hippocampus receive dense cholinergic input, which arises from 8 distinct nuclear groups in the basal forebrain (Ch1-Ch8) from which cholinergic axons project to more rostral cerebral regions [7]. Most noteworthy for this review are the medial and lateral pathways coursing from Ch4 (the nucleus basalis of Meynert) to the cerebral cortex [8], and the fornix traveling from Ch1 and Ch2 to the hippocampus [7]. Cerebral cholinergic projections are particularly vulnerable to TBI, resulting in acute and chronic dysfunction, the latter being especially critical because cholinergic function remains chronically reduced despite the relative normalization of other neurotransmitters such as glutamate, catecholamines, and serotonin [9].

An important observation is that cerebral white matter has actually expanded more in evolution than cortical gray matter, implying a crucial contribution to the unique cognitive functions of homo sapiens. Also noteworthy is that white matter undergoes developmental changes such that myelination is a slow process not complete until mid-life, and then, in late life, a decline in white matter volume appears to occur, implying differential vulnerability to acquired neuropathology as a function of age [11]. White matter is susceptible to a wide range of neuropathological insults beyond TBI, and in all clinical examples available for detailed study, neurobehavioral dysfunction of some kind— broadly considered, cognitive or emotional— has been observed [12]. In the tradition of Norman Geschwind and the legacy of his work on disconnection in the brain [13], white matter lesions of all types can be conceptualized as disrupting distributed neural networks mediating cognition, and the study of white matter in health and disease has direct relevance to modern ideas on brain connectivity as well as clinical practice in behavioral neurology and allied disciplines [5, 12].

DIFFUSE AXONAL INJURY

The concept of diffuse axonal injury (DAI) was introduced in 1982 when neuropathological studies of humans [14] and monkeys [15] with non-penetrating TBI documented widespread focal areas of damage in the cerebral white matter. Some years later, the term traumatic axonal injury was proposed in an effort to be more specific about etiology [16], but DAI remains widely popular in discussions of this phenomenon. DAI is thought to result from rotational and translational injury to the brain exerted by the mechanical force imparted by head impacts, and while DAI denotes injury to axons, myelin damage is also prominent [14, 15]. The effects of DAI are most apparent in the cerebral hemispheres near gray-white junctions, corpus callosum, fornices, and rostral brainstem, and associated findings include microhemorrhages, axonal varicosities with amyloid-β protein precursor (AβPP), microtubular disruption, and activated microglia [1]. Rotational force is considered more injurious to white matter than translational force, and a direct relationship has been shown between the amount of rotational head movement and the severity of DAI, the duration of coma, and the level of neurological impairment [15]. DAI was a novel term when first proposed, but similar observations were made in postmortem human brains years before the modern terminology was introduced; in 1968, Strich documented “shearing injury” of white matter in association with dementia after severe TBI [17], and in 1968 Oppenheimer described “shearing” and “stretching” of white matter fibers after mild TBI [18]. Today, the neuropathological lesion of DAI is accepted as a core neuropathological lesion in TBI that is present with or without other traumatically-induced lesions [1, 19–21].

A critical issue in the study of TBI, however, is the heterogeneity of neuropathology, a problem considered one of the most important barriers to finding effective therapeutic strategies [22]. A recent review pointedly commented that TBI “remains one of the most complex diseases known in the most complex of all organs in the body” [23]. In this light, it is appropriate to consider the degree to which other injuries that can occur in addition to DAI— focal cortical contusions, intracranial hematomas (epidural, subdural, and parenchymal), and subarachnoid hemorrhage— contribute to cognitive dysfunction and dementia following TBI [22]. This question remains under study and without a definitive answer, but as clinical, neuroimaging, and neuropathological techniques continue to improve, a more precise understanding of the cognitive sequelae of various TBI lesions is emerging. This review focuses on DAI, for which neuropsychological [24–26], neuroimaging [27], and neuropathological [23, 29] evidence exists to support the generation of cognitive deficits even when DAI is unaccompanied by any other TBI-related brain lesion. Although the cognitive importance of other lesions cannot be neglected, a plausible case can be made that DAI is universal in TBI, and that levels of post-TBI cognitive loss can be securely linked with the degree of DAI sustained [1, 23–29]. As will be discussed in detail below, DAI can also be conceptualized as a lesion that is crucial for the pathogenesis of acute and chronic cognitive dysfunction [1, 19–21].

CONCUSSION

The idea of concussion has existed in medical thinking for centuries and is understood as a general concept in clinical practice, but a secure and widely accepted definition has proven remarkably elusive. Current consensus opinion holds that concussion is a short-lived TBI caused by biomechanical forces, with or without loss of consciousness, that may involve neuropathological changes or a functional disturbance of brain tissue [30]. A distinction between concussion and mild TBI has often been suggested, but as no differentiation has been conclusively established [30], the two terms will be used interchangeably herein. Concussion is presently characterized as featuring loss of consciousness for 0–30 minutes, confusion for less than 24 hours, post-traumatic amnesia for up to one day, normal structural neuroimaging, and a Glasgow Coma Scale score of 13–15 within 24 hours of the injury [1]. Cognitive symptoms are typical, including poor concentration and impaired memory, but disrupted sleep, headache, nausea, vomiting, dizziness, anxiety, and irritability are very common [1]. A favorable prognosis of concussion within days to weeks has long been taught, and remains largely supportable in most cases, but it has become clear that prolonged symptoms and signs may plague some individuals for extended periods [31]. In view of evidence of DAI in the brains of concussed patients as demonstrated by autopsy studies [18, 32], white matter involvement appears to be clinically significant in concussion. In recent years, diffusion tensor imaging (DTI) has added valuable new microstructural information by allowing the study of tract integrity with the measures of fractional anisotropy (FA) and mean diffusivity (MD), and white matter damage can also be shown by this technique in concussion [33, 34]. Thus, even at what has been considered the mildest end of the TBI spectrum, the potential importance of white matter injury in the pathogenesis of lasting neurobehavioral dysfunction assumes increasing importance.

COMPLICATED MILD TBI

A subgroup of patients with concussion presents with a trauma-related intracranial abnormality on computed tomography (CT) of the head, or brain magnetic resonance imaging (MRI), along with clinical evidence of the injury. This group has been designated as having sustained complicated mild TBI [35], and typical lesions include microhemorrhages on susceptibility-weighted imaging, T2 hyperintensities, and small intracerebral or extra-axial hemorrhages. The prevalence of this condition varies widely, with a reported range of 5–40% among all mild TBI cases [36]. Not unexpectedly, white matter is damaged in these patients as in those with uncomplicated mild TBI, but may be more severely injured, as one DTI study demonstrated [36]. Currently, the presence of an intracranial CT or MRI lesion after mild TBI is clinically concerning and may be associated with more widespread microstructural white matter damage but does necessarily imply a worse prognosis [36]. Further study will be required to understand the implications of this added insult to the brain as a consequence of concussion.

BLAST INJURY

The recent military conflicts in the Middle East have brought to light a new form of TBI known as blast injury. Soldiers in these wars are relatively well protected from fatal wounds by body armor and helmets, but the head is nevertheless left susceptible to injuries from the detonation of nearby improvised explosive devices (IEDs) that produce a high-velocity air wave with associated shrapnel, sand, and dust [37]. In addition to the blast injury itself, individuals affected may sustain additional TBI as a result of being thrown by the explosion [37]. More than 60% of the U.S. combat casualties in Iraq and Afghanistan have in fact been attributed to explosive blasts related to IEDs [37]. Similar to blunt head trauma, blast injury is classified as mild, moderate, and severe, with most injuries being mild, and a post-concussion syndrome (see below) after blast injury is also recognized [37]. Blast injury is thought by many to be similar to concussion in its clinical effects and pathophysiology, although the frequent co-morbid state of post-traumatic stress disorder (PTSD) in military personnel complicates analysis [37]. Whereas previous postmortem reports of individuals with blast exposure disclosed DAI and petechial hemorrhages in the white matter, intriguing new data have recently been presented from five cases that were found to have distinctive astroglial scarring at gray-white junctions as well as U fiber DAI [38]. The involvement of U fibers at the gray-white matter junction in the cerebral hemispheres may underlie cognitive dysfunction, while astroglial scarring in the orbitofrontal cortex, amygdala, and hippocampus may explain PTSD symptoms [38]. Thus, in contrast to concussion from blunt impact, white matter appears to be variably damaged in blast injury, suggesting a distinct neuropathology that may have unique pathophysiological origins. Evidence is also accumulating that that the effects of this kind of injury may be lasting, as shown by DTI studies that found white matter changes in veterans more than one year after blast injury [39], and a dose-response relationship between the number of blast exposures and decreased white matter integrity [40].

POST-CONCUSSION SYNDROME

As mentioned above, a small proportion of mild TBI patients do not experience a prompt recovery, and instead develop the post-concussion syndrome (PCS), currently defined as the presence of symptoms persisting for more than 3 months after the traumatic injury [41]. Some 10–15% of mild TBI patients are thought to go on to PCS [41], which features a variety of symptoms including poor concentration, memory loss, sleep disturbance, fatigue, headache, dizziness, anxiety, irritability, and depression [1]. The pathogenesis of the PCS is poorly understood, and psychological factors complicate the study of this disorder [41], at times leading to much litigation and psychosocial distress. A neurobiological basis for PCS finds support in a controlled DTI study of mild TBI patients which disclosed microstructural injury in several association tracts and the corpus callosum that correlated with severity of PCS symptoms [42]. Of note, the individuals with mild TBI had a mean Mini-Mental State Examination [43] score of 27 and no evidence of no macrostructural white matter changes on conventional MRI, suggesting that, while the PCS can be disabling, it may reflect subtle white matter damage that often escapes detection by standard clinical and neuroradiological assessment. It is also of interest that the symptoms of PCS overlap considerably with those of more prolonged cognitive disorders, most often associated with repetitive mild TBI. Thus the cognitive dysfunction of PCS can be interpreted as a harbinger of more lasting dysfunction that may result from repeated mild TBI. White matter neuropathology may play a critical role in the pathophysiology of PCS and may exert an influence on whether any further cognitive dysfunction occurs in the months or years that follow the onset of the syndrome.

REPETITIVE MILD TBI

In military service and in sports, mild TBI can often be repetitive, and affected individuals may sustain hundreds or even thousands of injuries. Figures such as these suggest that the brain may simply not possess the capacity to recover fully from such frequent injury. One of the early advances in repetitive mild TBI associated with sports was the recognition of the second impact syndrome with a fatal outcome [44], and effective return-to-play guidelines in sport are critical to prevent severe or catastrophic sequelae [45, 46]. It is clear, however, that the accumulation of these injuries over time may have much more gradual effects on cognitive function, as will be considered below. As the discussion proceeds, some epidemiological data will be helpful to keep in mind. Repetitive mild TBI occurs most often in military service with combat exposure, and in sporting contests involving the potential of frequent head blows such as American football, ice hockey, and combat sports such as boxing [47]. It has been difficult to quantitate how many impacts are sustained and over what time periods, but recent studies from high school and collegiate athletics have shed some light on these questions. The use of helmet-based accelerometers has permitted estimates of 170–652 impacts per season in football and hockey players, with a maximum of 2,235 impacts recorded in football [47]. These impacts result in some 300,000 sport-related concussions in the U.S. each year, and prospective study has shown not only that players with a history of concussion are more likely to have future concussions, but also that previous concussions may lead to slower recovery [48]. As might be expected, DAI has been implicated in repetitive mild TBI, and cognitive consequences can result. Evidence for white matter injury has been adduced from study of Olympic boxers within six days of a bout, of whom more than 80% had elevated cerebrospinal fluid biomarkers suggestive of subcortical myelinated axon injury (neurofilament light) and cortical axonal damage (total-tau) [49]. As for cognitive sequelae of repetitive mild TBI, a DTI investigation of 80 collegiate football and ice hockey players over a single season found a relationship between the number of head impacts, white matter diffusion measures, and cognitive function [50]. Intriguingly, concussion was not diagnosed in these subjects, raising the possibility that blows to the head can be cognitively significant even without inducing what is regarded as mild TBI [50]. These kinds of blows, however, remain controversial, as the next section will discuss.

SUBCONCUSSIVE BLOWS

A particularly worrisome phenomenon that has recently been observed is that blows to the head may be harmful to the brain even when concussion is not produced. These impacts have been called subconcussive blows [52]. Also called repetitive subclinical brain trauma [52], a subconcussion is defined as a cranial impact that does not result in a concussion as typically diagnosed [51]. One of the most alarming implications of subconcussion is the suggestion that this kind of blow— which may occur many times more often than those impacts sufficient to induce concussion and be clinically inapparent— may lead to long-term detrimental effects on cognition, including neurodegenerative diseases [51]. Indeed, preliminary neuropathological evidence exists for cerebral white matter injury in former professional athletes who had no history of concussion [51]. However, the concept of subconcussion has come under significant criticism based on its imprecision (how does a subconcussive blow differ from a more severe concussive blow, and from any trivial blow to the head?), and from the absence of any clinical data that disclose short-term effects in exposed individuals [53]. Similarly, with respect to white matter involvement, the data are inconclusive. Some DTI evidence, for example, exists for lasting changes in white matter after impacts that do not lead to a concussion diagnosis [54], but other DTI work has shown that athletes with subconcussion have FA changes in both directions, that is, indicating both decreased and increased white matter integrity [53]. At this point in neuroscientific understanding, the idea of subconcussion is best considered imperfect [53] as the entity lacks a secure definition and is not well correlated with either clinical features or neuroimaging data [53]. Attention has also been called to the need for caution in presenting information to the general public about the risk of neurodegenerative disease from subconcussions that is as yet unwarranted [53]. Still, the notion of subconcussion deserves further study with prospective, controlled studies using a clear definition and validated outcome measures to determine its utility in the study of TBI. If more solid data emerge from such investigation, the role of white matter injury associated with these blows surely merits careful consideration.

NEUROPSYCHOLOGICAL OVERVIEW OF LONG-TERM COGNITIVE SEQUELAE

The long-term cognitive effects of mild, moderate, and severe TBI have received increasing scrutiny, and studies have examined both single incidents and repetitive injuries. We now turn to these long-term sequelae, beginning with a brief consideration of what is known about cognitive outcome without regard for its neuropathological basis, and then proceeding to the role of white matter involvement within specific categories of TBI.

Whereas most people with mild TBI make a full cognitive recovery, this outcome can be expected in just 60% of those moderately injured and in 15–20% of those with severe TBI [55]. Cognitive deficits in all groups typically become most evident in the domains of executive function, processing speed, attention, and episodic memory [56]. Whereas other comorbidities such as depression, PTSD, epilepsy, and physical injury can impact overall outcome, systematic review of the neuropsychological literature has found that severity of the inciting injury is a good predictor of cognitive outcome after TBI, as a strong dose-response relationship exists between TBI severity and the extent of neuropsychological impairment 6 months or longer after injury [56]. Other work has extended cognitive follow-up to 1-2 years with similar results [56], but it is important to note that observation of cognitive outcome over many years or decades is limited. Whereas many case-control studies are available, the relationship of TBI to later cognitive decline or dementia has rarely been investigated in a prospective longitudinal study. One of the most urgent questions facing the research communities in both TBI and dementia— in what way does early TBI affect subsequent cognitive decline— must be approached by more indirect investigation using methods presently available.

This discussion leads to a consideration of what is meant by the term dementia. This question is critical, as much of the medical and neuroscientific literature tends to confine the scope of dementia to the common neurodegenerative diseases of the cortical and deep gray matter while neglecting the many other causes of the syndrome. Dementia, however, simply considered as an acquired disorder of cognition that affects usual social and occupational activities [57], carries no implications about etiology, pathogenesis, or natural history, and TBI can certainly produce cognitive dysfunction of sufficient severity to qualify as dementia [17, 59]. This syndrome has naturally been better appreciated after moderate and especially severe TBI, but concern is rapidly growing with respect to dementia after repetitive mild TBI. The recent adoption of the term “major neurocognitive disorder” by the widely influential DSM-5 [60] may offer a more suitable descriptor for the concept of dementia, helping avert the tendency to assume that all demented individuals harbor a recognized neurodegenerative disease, but the more familiar term will be used here. The specific types of dementia associated with TBI are unclear [1], and much controversy pervades research addressing this question [58, 59], but the issue highlights the pressing public health implications of both problems, and considering dementia in its most fundamental sense as a clinical syndrome is vital for maintaining an appropriate framework with which to consider its association with TBI. As will be discussed below, dementia syndromes that follow TBI may indeed implicate neurodegeneration in the cerebral cortex, but a broadly inclusive conceptualization of dementia leads to the recognition that a prominent contribution of white matter is increasingly compelling in the understanding of how dementia evolves from TBI.

CONCUSSION, DEMENTIA, AND WHITE MATTER

When TBI occurs, it is most often mild, and the effects of single incident concussion have been most often studied. The commonly assumed dictum that most patients with concussion recover promptly and without sequelae remains appropriate in clinical practice, but new data have begun to introduce the notion that dementia may be a long-term outcome in some individuals. A recent retrospective Swedish study that included all individuals in that country aged 50 and older (well over 3,000,000 people) found that a single mild TBI increased the risk of dementia, including Alzheimer’s disease (AD), vascular dementia, and unspecified dementia [61]. The risk was highest in the first year after injury but was still evident more than 30 years later [61]. It should be recognized that dementia was diagnosed clinically in this study, and the specific cause was not confirmed by autopsy, highlighting the problem of diagnostic uncertainty that besets the literature on this topic in general. Another issue is that identification of dementia may have been erroneous because the individuals were experiencing PCS, which is presumably transient, or because of reverse causality (i.e., that dementia caused the TBI rather than the opposite). However, the results from Sweden showing increased risk over more than 30 years cannot be explained by these factors, and the Swedish data are consistent with prior work suggesting a similar small increase in risk of dementia, including AD, after concussion [62]. This arresting notion requires further study, as clinical experience would suggest that mild TBI is a self-limited disorder with no chronic sequelae. If verified, however, this increased risk of dementia may be related to persisting white matter injury. As mentioned above, concussion has been associated with white matter injury in both autopsy [18, 32] and DTI [33, 34] studies. More recently, moreover, a multiparametric MRI study of adolescent hockey players with a single concussion disclosed abnormalities in multiple white matter tracts after all clinical scores had returned to normal and the players had been cleared to play [63]. While much remains to be learned, the idea has been raised that even a single concussion may lead to ongoing white matter injury that later contributes to dementia.

These concerns are worrisome indeed, but it should be noted that the great majority of work on the long-term cognitive sequelae of TBI has been done on repetitive mild TBI and single incident moderate or severe TBI. Three dementia syndromes with prominent white matter neuropathology emerge from a consideration of these forms of TBI, and to begin, we consider a disorder seldom recognized clinically but potentially of great importance.

POST-TRAUMATIC LEUKOENCEPHALOPATHY

Leukoencephalopathy refers to a disorder of white matter, and whereas this term is generally used to describe white matter disease irrespective of etiology, a growing body of evidence has appeared to support the existence of a syndrome of static or even progressive leukoencephalopathy in the years that follow TBI [64]. It is important to emphasize that this syndrome differs from the more familiar dementias being studied as late sequelae of TBI. As discussed above, these syndromes are widely considered neurodegenerative diseases that are presumed to involve the cerebral cortex primarily. Post-traumatic leukoencephalopathy, in contrast, is proposed to be a disorder in which the white matter is selectively damaged without concomitant gray matter neuropathology of comparable significance. Some background information will help introduce this idea.

Despite Geschwind’s seminal work on disconnection in 1965 establishing the relevance of white matter to cognition in the context of behavioral neurology [13], myelinated systems of the brain have not been as readily incorporated into the thinking of the dementia research community. Yet it is clear that dementia does result from disorders of white matter as surely as from those that affect the cortical or subcortical gray matter [65–68]. The term “white matter dementia” has been introduced to all attention to the capacity of selective white matter involvement to produce often devastating dementia syndromes [65–68]. TBI is a plausible example of this principle, as it may clearly lead to dementia, an acquired state of cognitive dysfunction that interferes with usual activities, and manifests equally clear white matter neuropathology [65–68]. Although work has not yet advanced to support an unequivocal link between post-TBI dementia and white matter involvement, the evidence supporting a syndrome we term post-traumatic leukoencephalopathy is increasingly intriguing. We will consider this topic by developing a synopsis of selective, chronic white matter involvement in TBI, as demonstrated by neuropathology, experimental studies, and neuroimaging, together with correlations of these data with cognitive dysfunction.

In 1956, Sabina Strich published a seminal neuropathological report in which five patients with severe TBI had “extreme dementia” that was unequivocally attributed to diffuse degeneration of the white matter [17]. Cortical involvement was minimal, and in language presaging modern connectionist thinking, Strich described “… severe diffuse degeneration of the white matter in the cerebral hemispheres, which had interrupted… connexions between different areas of cortex,” adding that “… it is not enough to examine the brain-stem reticular formation or the cerebral cortex when investigating cases of severe disturbance of consciousness after head injury” [17]. This remarkable work was eventually followed by more extensive studies finding similar results in larger numbers of severe TBI in humans [14] and experimental animals [15], and in concussion [18, 32].

What Strich found in 1956 was most likely DAI, and this lesion has since become widely accepted as a consistent feature of all TBI [17–29, 69–71]. DAI produces widespread damage to white matter [17–29, 69–71], and unmyelinated fibers within white matter tracts may be more vulnerable to this kind of injury than those fibers invested with myelin [72]. Recent information from animal and human studies of concussion has identified a neurometabolic cascade after DAI involving ionic flux, glutamate release, oxidative stress, mitochondrial dysfunction, cytoskeletal damage, altered neurotransmission, and inflammation, all of which contribute to axonal dysfunction that disrupts the function of multiple neural networks subserving cognition [73]. Damage to both myelinated and unmyelinated axons leads to cognitive impairment in TBI, but network efficiency is particularly compromised by DAI impacting myelinated fibers, as these lesions dramatically reduce conduction velocity to produce slowed cognition and related neurobehavioral deficits [73]. The ultimate result of white matter neuropathology is therefore network disruption with multiple cognitive domains affected.

With the advent of modern neuroimaging, researchers have increasingly turned to in vivo studies for the assessment of white matter neuropathology. MRI has been and remains the standard modality for imaging of macrostructural TBI white matter lesions, the most important of which, for our purposes, is volume loss. As MRI is insensitive to microstructural injuries, however, other techniques, most notably DTI, have been exploited to examine white matter integrity in regions that appear normal on MRI. We will consider studies using MRI and DTI in turn, and conclude this section with a discussion of how post-traumatic leukoencephalopathy may be a progressive disorder in some patients.

MRI has primarily been employed in severe TBI patients, in whom gross volume reductions detectable with volumetric techniques can be predicted. Studies of severe TBI have shown prominent volume reductions in white matter structures including the corpus callosum, fornix, cerebellar peduncles, internal capsule, external capsule, inferior longitudinal fasciculus, superior longitudinal fasciculus, and corona radiata [74–76]. Whereas these studies all found reduced cerebral volumes that included concomitant gray matter atrophy, white matter changes tended to be more critical for cognitive impairment, and in one study, lower global white matter volume was correlated with impaired performance on tests of psychomotor speed, attention, memory, and executive function, while gray matter volume loss produced no such effect [76].

In contrast to the macrostructural measurements of white matter volume suitable for MRI, the microstructure of white matter is accessible to DTI, a technique with the capacity to assess the normal-appearing white matter. Whereas DTI is very sensitive to subtle changes in white matter, some caution is warranted in data interpretation as studies of acute mild TBI have shown either increased or decreased white matter integrity, likely related to issues such as timing after injury, patient age, injury severity, regions analyzed, and specifics of the method used [73]. As more data are reported, however, DTI evidence is accumulating to support the notion that white matter involvement with cognitive decline is prominent at all levels of TBI severity.

DTI has been conducted in combination with neuropsychological evaluation, and all levels of TBI severity have been analyzed [42, 76–80]. In general, microstructural changes as indexed by decreased FA and increased MD have been found in a variety of white matter tracts, including the corpus callosum, corona radiata, inferior longitudinal fasciculus, superior longitudinal fasciculus, sagittal striatum, uncinate fasciculus, cingulum, and fornix [42, 76–80]. Cognitive dysfunction has been clearly associated with DTI white matter involvement, and the domains of attention, memory, executive function, and processing speed tend to be most strongly correlated [42, 76–80]. In most studies, aggregate measures of white matter disruption and neuropsychological dysfunction have been used to detect correlations, and both of these parameters worsen as injury severity level increases [42, 76–80]. One study found that the severity of PCS symptoms after mild TBI also correlated with widespread microstructural white matter changes [42]. Specific correlations between damage to a single tract and impairment in one cognitive domain are uncommon, but in one study, disruption in the fornix predicted declines in learning and memory while damage to frontal lobe white matter connections predicted executive dysfunction [78], consistent with what is known with respect to white matter-behavior relationships [5, 68]. Injury to the fornix may even be the primary lesion leading to the often-observed hippocampal atrophy that occurs in TBI, as the fornix is the major tract associated with this structure [64].

This account of post-traumatic leukoencephalopathy would not be complete without a discussion of longitudinal white matter changes that have been identified years after TBI [64]. Whereas cross-sectional data, reviewed above, showing loss of white matter concur with the plausible clinical assumption that white matter changes after TBI would simply represent static leukoencephalopathy, the possibility of progressive white matter neurodegeneration is being raised by neuroimaging data from patients whose injury occurred many years before. The recognition that ongoing damage to white matter tracts may occur over time suggests a locus for neurodegeneration specific to TBI that does not involve the cerebral cortex or deep gray matter. Gray matter damage, when it does occur, may be secondary to the tract injury, as has been suggested in the case of hippocampal atrophy deriving from primary DAI in the fornix [64].

One of the most notable long-term studies in this context reported neuropathological evidence of ongoing corpus callosum atrophy for up to 18 years after a single moderate or severe TBI [71]. In this autopsy study, white matter deterioration was thought to be driven primarily by ongoing neuroinflammation [71]. This conclusion has found support from neuroimaging studies, using either MRI volumetrics [81, 82] or DTI [83–86], for long-term degradation of white matter after moderate to severe TBI. Some studies have also found correlations between ongoing white matter injury and declines in cognition, typically in memory, processing speed, and executive function [83–86]. The corpus callosum has thus far been the tract most often identified to undergo progressive volume loss, and in the longest follow-up TBI study available using structural neuroimaging, the corpus callosum showed continued atrophy over 8 years while the hippocampus did not [82]. The predilection of the corpus callosum for this kind of degeneration may be explained by the presence of extensive, densely packed reactive microglia in this structure that strongly suggests a chronic inflammatory state as a potential causative mechanism [71]. This leukoencephalopathy may not occur in all TBI patients, as only about 50% of moderate to severe pediatric TBI patients in one study were found to have impaired callosal organization 13–19 months post-TBI [86]. These data suggest that as yet unidentified factors— most notably genetic variation— may be protective in some individuals with TBI, resulting in variable outcome with respect to leukoencephalopathy. Taken together, these studies suggest that a progressive inflammatory post-traumatic leukoencephalopathy can develop in some patients after moderate to severe TBI.

CHRONIC TRAUMATIC ENCEPHALOPATHY

The previous section considered the evidence supporting a unique leukoencephalopathy that follows TBI, and may even be progressive over time, but a far more familiar cognitive sequel of TBI is the disorder known as chronic traumatic encephalopathy (CTE) [87]. This disease has been widely publicized in recent years because of alarming suggestions that repetitive mild TBI in athletes and military combatants may lead to a progressive dementia with possibly high prevalence in the general population. Whereas concern is appropriate with respect to any dementia syndrome that may have its origin in environmental insults, undue public anxiety is unwarranted given how much is still unknown in this area. We will approach CTE in the context of how TBI and its signature lesion of DAI may help inform the many areas of uncertainty surrounding this topic.

The history of the disorder now referred to as CTE deserves brief review to highlight major gaps in our knowledge as well as what is known [88]. The first report of a disease suggestive of CTE dates from 1928, when Martland described the “Punch Drunk” syndrome in boxers [89]. Nine years later, Millspaugh reported on similar patients, using the somewhat less derisive term “dementia pugilistica” [90]. A treatise on progressive dementia in boxers appeared in 1949, when Critchley used the term CTE [91]. In the late 1960 s, due to concern about how prevalent this disorder might be, the Royal College of Physicians in London sponsored a cross-sectional clinical follow-up study of British boxers who had competed several decades previously [88]. This project, which remains the only clinical outcome study of TBI in boxers over many years, found that 17% of these athletes had central nervous system lesions attributable to boxing, although only 6% had a clinical picture consistent with dementia pugilistica [92]. Then, in 1973, Corsellis and colleagues presented a detailed neuropathological study of 15 boxers who died many years after their careers, identifying neurofibrillary tangles with limited numbers of neuritic plaques [93]. The modern introduction of CTE occurred in 2005, when Omalu and colleagues described the autopsy of a former National Football League player [94], and together with another case in the following year [95], CTE came to be conceptualized as a tauopathy distinct from AD mainly because of sparse or absent amyloid neuropathology in CTE brains. Since then, interest in CTE rapidly accelerated as the neuropathology was found in many other former contact sports athletes and in military service members who had been exposed to combat [87].

From the perspective of this review, it is noteworthy that, in addition to the tauopathy, all four neuropathological categories of CTE involve white matter changes consistent with DAI [87]. Whereas the prominence of tau in its unique distribution surely deserves attention, the importance of DAI should not be overlooked, as the significance of tau in TBI is still uncertain. Tau is a normal component of axons, promoting microtubule assembly and stabilization, and while this protein may prove to be toxic, it may alternatively represent normal physiology or even play a protective role [96]. However, tau deposition is interpreted, the centrality of DAI in CTE is apparent, suggesting that white matter damage may be the inciting event in this disorder, with all the clinical implications implied by such a perspective.

Recent neuropathological studies have added support to the role of white matter damage as a direct precursor of CTE. A novel approach involved the post-mortem study of chronic schizophrenic patients (mean age 78 years) who had been subjected to prefrontal leukotomy during the era of psychosurgery [97]. This procedure, now long abandoned, was intended to treat psychosis by disconnecting the frontal lobes from the remainder of the brain, and, as the target was frontal lobe white matter, prefrontal leukotomy generated a useful clinical model for examining the effects of DAI over time [97]. Remarkably, there was a clear predilection of hyperphosphorylated tau in the depths of sulci, but only in areas adjacent to the site of the leukotomy, implying a direct effect of DAI on the development of tauopathy [98]. As this leukotomy-induced lesion is comparable to that described in CTE, repetitive mild TBI can be postulated to have a similar effect. In support of this idea is an autopsy study of four teenage athletes with repeated mTBI and subconcussive blows who died from other reasons and were found to have phosphorylated tauopathy – one case even meeting criteria for early-stage CTE – in association with myelinated axonopathy [98].

ALZHEIMER’S DISEASE

AD is the prototype cortical dementia, and its pathogenesis is widely held to involve degeneration of the cerebral cortex and hippocampus in association with the deposition of amyloid-β (Aβ) in extracellular neuritic plaques and hyperphosphorylated tau within intracellular neurofibrillary tangles [99]. Yet this pathogenetic formulation has never been indisputably established [100], and all efforts to treat the disease by targeting its signature proteins have thus far failed [101]. This discouraging reality raises the question whether Aβ and hyperphosphorylated tau are invariably toxic to the brain, and indeed, it is well known that their precursors— AβPP and intra-axonal tau— are normal brain constituents whose role in neuropathology remains obscure. In this light, the myelin model of AD has been proposed [11], which proposes that the initial neuropathology takes hold in white matter, preceding the appearance of plaques and tangles. This innovative model posits that late-life decline in white matter volume, a process often termed retrogenesis [102], is already underway in older people, and acquired insults hasten this decline and lead to the deposition of cortical proteins as secondary phenomena [11]. One of the insults to the brain is thought to be TBI (clinically apparent or silent) that involves white matter lesions disrupting distributed neural networks and may lead ultimately to AD as well as to other dementias [11, 103].

The relationship of TBI to AD remains somewhat uncertain because while many studies document an association, some do not [58]. A large prospective historical study, however, merits special mention in the regard, as World War II veterans studied 50 years after combat showed a 2.3-fold increased risk of AD following single-incident moderate TBI, and a 4.5-fold increased risk after severe TBI [104]. These data serve to help build the argument for white matter involvement preceding the late-life neurodegeneration of AD. The presumed mechanism of this development is thought to begin with DAI as the inciting insult, which then leads to the deposition of protein aggregates in the brain parenchyma [52, 105]. Within a few hours of single-incident severe TBI, for example, AβPP is visible in damaged axons and cell bodies, and whereas this change resolves over time in many individuals, some 30% of severely-injured TBI patients develop diffuse Aβ plaques [52]. DAI and parenchymal Aβ can be putatively linked by postulating that TBI produces a large pool of intra-axonal Aβ that is released by lysis or leakage of damaged axons into the cerebrospinal fluid, where high levels have been described [105]; this process would enable Aβ to gain access to remote brain areas where it is deposited in the parenchyma [105] and detected as Aβ plaques many years later [106]. Tauopathy also develops as a consequence of DAI, as the shear-strain force from TBI leads to hyperphosphorylation, misfolding, and tau accumulation, thus completing the standard microscopic neuropathology of AD [52, 99].

Whereas the still dominant amyloid cascade hypothesis of AD has undergone significant criticism in view of the frequent presence of AβPP and tau in normal older brains, and the failure of all anti-amyloid and anti-tau drugs to treat AD dementia [101], the relationship of DAI to Aβ and hyperphosphorylated tau merits consideration. As mentioned above, both proteins may represent salutary host responses in some individuals [96, 100], and if so, they only become toxic in those individuals in whom the proteins persist and accumulate [67, 68]. As the common feature of all cases, DAI in the white matter may be the initiating event leading to cognitive impairment in AD, with a potential additive component of amyloid and tau neuropathology as time progresses and neurodegeneration develops [67, 68]. Thus, the focus on white matter as the initial site of TBI-related dementia offers a more nuanced approach to amyloid and tau in AD pathogenesis.

PATHOPHYSIOLOGY: THE ROLE OF INFLAMMATION

In an earlier section, neuropathological data were cited showing ongoing white matter inflammation and corpus callosum atrophy for up to 18 years after a single moderate or severe TBI [71]. This study echoes others in which ongoing inflammation within white matter has been observed many years after TBI [107, 108]. This inflammatory process likely implicates activated microglia, which are normal microglia modified by the external stimulus of TBI to undergo morphological change that results in cytokine release and a sustained inflammatory response [108]. Some evidence exists from study of normal brains that microglia are significantly more numerous in white than gray matter [109], which may be a factor explaining why DAI may lead to greater microglial activation in white matter tracts [110]. The inflammatory cascade facilitated by activated microglia and cytokine release may then lead to the neurodegenerative dementias of CTE and AD [1, 67].

Although this review is focused on clinical aspects of TBI, the human data on TBI-related white matter inflammation [71, 108] finds critical support in an intriguing experimental study demonstrating long-standing inflammation and atrophy in the corpus callosum, in association with cognitive dysfunction, after repetitive mild TBI [110]. Of particular interest in this study is that neither Aβ nor hyperphosphorylated tau were present in the cerebral cortex or hippocampus of the animals, and that the white matter neuropathology was considered responsible for the observed neurobehavioral deficits [110]. A chronic traumatic inflammatory encephalopathy is thus a plausible pathophysiological mechanism that contributes to the series of events leading from TBI to dementia [111], and this sequence may proceed from TBI to one of the three major outcomes discussed above: post-traumatic leukoencephalopathy, CTE, or AD. The specific cognitive outcome of TBI in any given individual is unknown, but presumably results from the effects of as yet undetermined factors unique to each injured person.

THERAPEUTIC IMPLICATIONS

The preceding sections have constructed the case for white matter injury after TBI as the common denominator of a wide range of cognitive deficits that may follow. The next task is to consider how this insight, if accurate or even partially so, can inform attempts at treatment. The lack of successful therapy for DAI thus far [112] should not deter further treatment efforts, as it is clear that white matter is a biologically active component of the brain, likely under the variable influence of ongoing inflammation after TBI, that may well be amenable to various disease-modifying and rehabilitative modalities. After some general comments on white matter as a target of treatment, some medical interventions centered on inflammation and the concept of white matter plasticity will be considered.

Two principles of white matter structure and function are pertinent to a discussion of possible treatment of DAI. First, the prognosis of white matter injury depends to a large extent on the relative degree of injury sustained by axons and myelin. In central nervous system disorders such as TBI that include a prominent degree of white matter damage, such as spinal cord injury, multiple sclerosis, leukodystrophies, central pontine myelinolysis, acquired immunodeficiency syndrome (AIDS), and subcortical ischemic disease, the prognosis is clearly more favorable if axons are preserved [113, 114]. That is, if the axonal scaffolding on which myelin can regain its integrity is preserved, tract structure and function have a far greater chance of recovering so that network efficiency and cognitive function may be restored. In the case of TBI with its typical DAI, the prompt treatment of inflammation would presumably have the potential to preserve both myelin and axons before more significant damage is sustained.

The second principle is that white matter possesses the capacity of plasticity, which is most simply conceptualized as the modification of brain structure by experience [115]. Plasticity is not confined to the gray matter, where synaptic function has long been known to manifest physiological and structural changes associated, for example, with learning and memory, but can readily be observed in white matter as well [116, 117]. White matter is dynamic tissue no different in this respect than its gray matter counterpart, and, to illustrate, recent data have shown that neuronal activity can produce plastic changes in white matter via activity-dependent myelination [117] and enhanced oligodendrogenesis [118]. In addition to the restoration of myelin implied by these data, another means of plasticity may be axonal sprouting, which has not been demonstrated in brain white matter but which does occur in spinal cord white matter after injury [116]. Cellular events following changes in experience thus indicate that many possibilities for restoring white matter structure and function after TBI may be found useful.

A wide range of pharmacological interventions are potentially conceivable to suppress inflammation after TBI [112, 119], the details of which are beyond the scope of this review. However, the most familiar agents with anti-inflammatory effects— the hormones with glucocorticoid actions widely known as corticosteroids— deserve some attention in view of their very common clinical use. Although corticosteroids have not been shown effective in the acute treatment of moderate and severe TBI, patients with these injuries are critically ill and typically have many other issues, such as cerebral edema, that complicate the pathophysiology [120]. With respect to white matter neuropathology, encouraging experimental evidence shows that estrogen and progesterone may have effects on activated microglia after TBI [121]. Moreover, the neurosteroid allopregnanolone, under study for AD, appears to act in the white matter, upstream of Aβ generation, to increase proliferation of OPCs and axonal myelination [122]. Theoretically, while such treatment is underway, the use of cholinesterase inhibitors such as donepezil could help potentiate the function of still viable axons coursing within the injured white matter [7, 123].

A related line of investigation relevant to future neurodegeneration pertains to potential immunotherapy focused on the prevention of tauopathy. In experimental studies, the tau epitope cis P-tau is robustly expressed in axons acutely after TBI, and treatment with a monoclonal antibody against cis P-tau blocks the development and spread of tauopathy, while restoring behavioral deficits and long-term potentiation, and preventing brain atrophy [124]. As tauopathy is a central feature of both CTE and AD, any intervention that could abort the accumulation of tau after TBI has much appeal [124].

Another therapeutic possibility for which supportive evidence is rapidly accumulating involves the exploitation of white matter plasticity. In contrast to medical intervention intended to reverse microstructural dysfunction such as inflammatory degradation, or prevent the development and spread of tauopathy, this approach begins with the notion that white matter is damaged but not destroyed, particularly if axons are preserved, and thus has the capacity to repair itself under the influence of appropriately designed therapeutic interventions. As white matter tracts are essential components of distributed neural networks subserving cognition [5, 65–68], this strategy fundamentally involves the restoration of connectivity and resultant improvement in cognition [125].

An appealing and readily available therapeutic approach is the traditional triad of physiatric treatments: physical therapy, occupational therapy, and speech therapy. These familiar treatments may exert significant effects on white matter structure and function by stimulating plasticity within injured tracts. To illustrate, a recent meta-analysis of 29 studies of aging individuals reported significant correlation between physical activity and white matter structure in aging [126], and in patients with Broca’s aphasia who had left frontal lesions, melodic intonation therapy not only improved language output but was also associated with increased size of the right arcuate fasciculus [127]. Related to physiatric treatment is the emerging field of non-invasive transcranial brain stimulation, which aims to exploit modalities such as transcranial magnetic stimulation, transcranial direct current stimulation, and transcranial alternating current stimulation. Whereas this kind of stimulation is thought to promote brain plasticity through its influence on synaptic function in the cerebral cortex [128], effects on the white matter subjacent to the cortex may also be possible. In a study of patients with chronic left hemisphere cerebral infarcts, for example, transcranial direct current stimulation was found to be associated with increased FA in ipsilesional white matter of the internal capsule and upper brainstem [129]. Data such as these indicate that it may become feasible with advancing technology to influence plasticity in white matter tracts devoted to cognitive function. Last, a more remote but still intriguing possibility is the use of cell-based therapy to regenerate white matter. This intervention could involve stimulation of OPCs to form functional oligodendrocytes [118], or the use of stem cells (endogenous or exogenous) to replenish oligodendrocytes and enhance remyelination [130].

In sum, the options for treating white matter damage after TBI may involve a wide range of therapeutic modalities, all of which show promise and need not be mutually exclusive. The goals of reducing inflammation, preventing tauopathy, and restoring myelination are all under consideration, and other ideas may also lead to treatment options. It hardly need be stated that, if the primacy of white matter involvement in TBI is unequivocally confirmed, the implementation of these methods would ideally occur as soon as possible after the injury is sustained.

CONCLUSION

The purpose of this review has been to call attention to the prominence of white matter injury in TBI with the intention of focusing clinical and basic research on a component of the brain not always thought to be important for cognition. While we readily acknowledge the presence of neuropathology elsewhere in the brain that must be considered, the notion that DAI is a fundamental TBI lesion from which major acute and chronic cognitive effects develop has much support and warrants further detailed investigation. Much more study is needed to establish pathophysiological mechanisms, but a powerful hypothesis emerges from this perspective: a common neuropathology links all severity of TBI, leads to post-concussion symptoms and, in perhaps many cases, dementia, and offers a potentially useful avenue toward early intervention to restore brain health and avert long-term cognitive decline. We readily acknowledge that proposing a fundamental role of DAI in the causation of lasting or progressive cognitive dysfunction— including CTE and even AD— is fraught with uncertainty. Nevertheless, the public health importance of TBI and dementia, combined with major gaps in information that can serve to guide clinical practice and public policy, suggest that this point of view may be transformative. Clinical and basic science methods are becoming ever more effective in evaluating these ideas, and such work may offer untold benefits to the many individuals currently coping with TBI, as well as to the many more who will undoubtedly follow.

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/18-0287r2).