Abstract
Traumatic brain injury (TBI) is the leading cause of disability and death worldwide, affecting as many as 54,000,000–60,000,000 people annually. TBI is associated with significant impairments in brain function, impacting cognitive, emotional, behavioral, and physical functioning. Although much previous research has focused on the impairment immediately following injury, TBI may have much longer-lasting consequences, including neuropsychiatric disorders and cognitive impairment. TBI, even mild brain injury, has also been recognized as a significant risk factor for the later development of dementia and Alzheimer's disease. Although the link between TBI and dementia is currently unknown, several proposed mechanisms have been put forward, including alterations in glucose metabolism, excitotoxicity, calcium influx, mitochondrial dysfunction, oxidative stress, and neuroinflammation. A treatment for the devastating long-term consequences of TBI is desperately needed. Unfortunately, however, no such treatment is currently available, making this a major area of unmet medical need. Increasing the level of neurotrophic factor expression in key brain areas may be one potential therapeutic strategy. Of the neurotrophic factors, granulocyte-colony stimulating factor (G-CSF) may be particularly effective for preventing the emergence of long-term complications of TBI, including dementia, because of its ability to reduce apoptosis, stimulate neurogenesis, and increase neuroplasticity.
Introduction
T
Traumatic Brain Injury
TBI is a form of acquired brain injury in which an external force to the head causes a direct or indirect injury to the brain. 9 This leads to either diffuse injury, as seen in road accidents, characterized by diffuse axonal injury (DAI), edema, and vascular injury throughout the brain, 10 or focal TBI following direct impact to the head, which can present as lesions such as contusions and hemorrhages. 10 At times, both a diffuse and focal injury can occur simultaneously at the point of impact. 10 In addition to categorizing TBI on the type of injury, TBI can also be classified as either mild, moderate, or severe based on Glasgow Coma Scale (GCS) score acutely after injury. A GCS ≥13 represents mild TBI, a score of 9–12 represents moderate TBI, and a score <8 indicates a severe TBI. 11
Cognitive Impairment after TBI
Acutely following injury, TBI is associated with significant impairments in cognitive function, affecting language, visuospatial function, intelligence, and memory. 12,13 Memory impairment is the most common cognitive complaint following TBI. 14 It first presents itself as post-traumatic amnesia (PTA), either in retrograde form (loss of memory immediately preceding the TBI) or anterograde form (inability to form new memory), both of which progressively improve with recovery. 15 Cognitive impairments following TBI may be the result either of the primary injury itself, such as contusions, hemorrhages, and axonal shearing, 16 or of secondary injury factors (i.e., the cascade of biochemical, cellular, and molecular changes that are triggered by the primary injury). 16,17 Recently, it was proposed that DAI is the major cause of cognitive dysfunction, particularly in memory, with increasing severity of DAI associated with more severe cognitive impairment. 18 Other factors that may play a role include synaptic dysfunction and cell death, which are ongoing from the time of injury. 19
Depending on the severity of the primary injury, and its resultant secondary injury cascade, impairments in cognitive function may be transient, long-lasting, or permanent. In mild cases (i.e., 95% of TBI cases), 2 recovery typically begins immediately after injury, with recovery to baseline levels of cognitive functioning and a favorable long-term outcome within 1–3 months after injury. 20 In moderate or severe cases, however, recovery may be less robust, and full return to baseline function may not occur. 4 In large prospective studies, almost all patients with moderate or severe TBI still displayed detectable cognitive impairment 1 month post-injury, 21 and, by 1 year, 100% of severe and 50% of moderate TBI cases exhibited some degree of cognitive impairment. 3
TBI and Dementia Risk
In addition to persistent cognitive impairment, a growing body of evidence suggests that TBI may also be an important factor predisposing to the development of dementia later in life, often decades after recovery from the original injury. 22 A recent retrospective study by Gardner and colleagues utilizing administrative health data from emergency department visits in the state of California, looked at the link between TBI and dementia development in 164,661 individuals >55 years of age with a history of TBI, over a follow-up period of 5–7 years, and found that prior moderate or severe TBI can increase the risk of dementia with a minimum hazard ratio of 1.3. 23 Similarly, a recent retrospective cohort study in 188,764 United States military veterans ≥55 years of age suggested a 60% increase in the risk of dementia development over a 9 year follow-up period, even after accounting for potential confounding variables. 24 Some studies suggest that a dose-dependent relationship might exist, in which risk of dementia increases with TBI severity. 25 The risk of developing dementia increased at least twofold and fourfold in patients who had sustained moderate and severe TBI, respectively. 26 It should be noted, however, that the studies in this field are predominantly retrospective, and hence are limited by factors such as recall and uncertainty regarding premorbid functioning. Additionally, some studies do not see a link between TBI and later dementia risk. 27 Therefore, additional studies and meta-analyses investigating this link are critically needed.
Nonetheless, a previous history of TBI may lower the age of dementia onset. In support of this, in a 2014 cohort study of 811,622 Swedish men conscripted for mandatory military service, followed over a median period of 33 years, strong associations were found between TBI and young onset dementia (defined as dementia before 65 years of age). 28 This relationship appears to be dose dependent, with a hazard ratio of 1.5 for mild TBI and 2.3 for severe TBI, even after adjusting for covariates. 28 Interestingly, Nördstrom and colleagues relied on physician-generated diagnoses of TBI, eliminating the potential recall bias that, as noted, is major limitation of many of the previous studies in this area. The findings of this study have recently been corroborated by a 2016 study in the Alzheimer's Disease Neuroimaging Initiative (ADNI) medical history database. After controlling for potentially confounding factors, the age at onset for cognitive impairment in older adults with a past history of TBI was 2 years earlier than those without a previous TBI (68.2 years versus 70.9 years). 29 Although the mechanisms that may account for this are still unknown, a 2015 study using a predictive model of normal brain aging defined using machine learning in 1537 healthy individuals found that the brains of TBI patients appeared “older,” with a mean predicted age difference between chronological and estimated brain age of 4.66 years in the gray matter and 5.97 years in the white matter. 30 The discrepancy between chronological and estimated brain age increased with time since injury, 30 suggesting that TBI may accelerate the rate of brain atrophy and prematurely age the brain, predisposing to earlier onset dementia.
Age at occurrence of TBI may also influence the risk of dementia development, with older age known to be associated with significantly worsened outcomes following TBI. 31 –34 In the aforementioned 2014 prospective cohort study by Gardner and colleagues, although severe TBI was associated with increased risk of dementia across all ages, mild TBI was a more significant risk factor with increasing age, suggesting that older adults may be particularly susceptible to the effects of mild injury. 23 Similarly, in a 2015 study in 5486 Western Australian men 70–89 years of age, 17.4% had a history of TBI (as determined either by self-report or by analysis of medical history), and this history of TBI was associated with increased odds of cognitive impairment (Mini-Mental State Examination [MMSE] score <24 or medical diagnosis of dementia), with an odds ratio (OR) of 1.23. 35 Recently, unsupervised gene clustering and pathway analysis demonstrated that age predisposes the brain to an exacerbated inflammatory response to TBI, leading to the increased production of ligands that bind to both C-C chemokine receptor type (CCR)2 and CCR5, increasing the recruitment of peripheral macrophages into the injured brain. 36 This may result in chronic neurodegeneration following TBI, placing the older adult at increased risk of dementia development.
Link between TBI and Alzheimer's Disease (AD) and Chronic Traumatic Encephalopathy (CTE)
The type of dementia that one is predisposed to following TBI may depend on the initiating injury, with support for the hypothesis that a single moderate or severe TBI increases the risk of developing AD, whereas repetitive mild TBI is associated with an elevated risk of CTE. 37,38
CTE has been identified in populations exposed to repeated mild TBIs, including veterans of military service, professional National Football League (NFL) players and rugby players, among others. 37,39 –41 In a recent study of retired NFL players in the United States, higher rates of cognitive impairment were reported than would be expected in age-matched controls in the general population. 42 Similarly, some veterans of the Iran and Afghanistan war showed signs of CTE as a result of brain injuries and blast injuries sustained during combat. 43 CTE is characterized by cerebral atrophy, enlargement of the lateral and third ventricles, and thinning of white matter pathways. 44 Perhaps most strikingly, CTE has been shown to be associated with a distinctive pattern of distribution of neurofibrillary tangles (NFTs), composed of the abnormally phosphorylated microtubule (MT)-associated protein tau. 45 These NFTs are contained to layer II and the upper third of layer III of the cerebral cortex, 46 and tend to begin perivascularly, with concentration at the base of the sulci. 47,48
In contrast, a single TBI may be a risk factor for AD, 4 –8,49 with TBI suggested to be the most established environmental risk factor after age. 50 Although the findings of some epidemiological studies have challenged the link, 27,51,52 multiple studies have reported a positive association between brain trauma and AD, with the relative risk ranging from 253 to 14. 54 Two in-depth meta-analyses have been conducted on this issue, and have found that TBI increases the risk of developing AD by 58–82%. 8, 49 A more recent large scale meta-analysis by Perry and colleagues, which investigated all studies from January 1995 to February 2012, also reported that a history of TBI was associated with higher odds of both mild cognitive impairment (MCI) (OR = 2.69) and AD (OR = 1.40). 55 Recently, Gamberger and colleagues used a multi-layer clustering approach, a well-established machine learning methodology, to identify three distinct clusters of patients with either MCI or AD in the ADNI database, and found that, within the largest cluster, patients had mild signs of brain atrophy, as well as large ventricular, intracerebral, and whole brain volumes, indicative of prior brain injury. 56 This suggests that prior brain injury may increase the risk of AD, at least in a subset of individuals.
Pathologically, AD is characterized by extracellular amyloid-beta (Aβ) plaques, composed of aggregations of amyloid peptides, and intracellular NFTs, consisting of hyperphosphorylated tau protein. 57 The “amyloid cascade hypothesis,” currently the prevailing hypothesis regarding AD pathogenesis, states that Aβ peptides aggregate to form toxic Aβ oligomers and plaques, which then trigger a cascade of neuropathological events, including neuroinflammation, oxidative stress, tau hyperphosphorylation, and NFT formation, resulting, ultimately, in widespread neurodegeneration and dementia. 58,59 It has been proposed that Aβ may be only one player in a complex neural cascade. 57,60 For example, NFTs have been shown to correlate better with cognitive impairment than amyloid plaques. 61 Further, in recent years, another hypothesis – the “inflammatory hypothesis” – has been proposed, which posits that neuroinflammation may contribute to and exacerbate AD pathology. 60,62 –67 Activated microglia and reactive astrocytes have been shown to surround Aβ plaques in AD, initiating the release of pro-inflammatory cytokines and activation of the complement cascade. 68,69 Pro-inflammatory cytokines upregulate the cleavage of Aβ from amyloid precursor protein (APP), perpetuating the cycle. 70 Additionally, neuroinflammation has also been shown to be critical for tau production. 71,72 Although it is still intensely debated in the literature whether inflammation causes or is the result of pathological changes in AD, in recent years, growing evidence has suggested that neuroinflammation may play an important role in the pathogenesis of the disease. 73 In support of this, Wright and colleagues demonstrated that both neuroinflammation and neuronal loss preceded the deposition of Aβ plaques in the hAPP-J20 mouse model of AD. 74 Similarly, several animal studies have suggested that neuroinflammation may predate the emergence of tau pathology, 75 –77 with multiple pre-clinical studies suggesting that blocking neuroinflammation can ameliorate tau pathology and improve cognitive function. 71,72 Although a more detailed discussion of this hypothesis is outside the scope of the current work, an excellent review on the subject has recently been conducted by Heppner and colleagues. 78
TBI induces Accumulation of Pathological Proteins Associated with Dementia
Apart from epidemiological evidence, TBI has been shown to be associated with many of the neuropathological changes that are characteristic of dementia, as discussed. 79 –81 Clinical neuroimaging studies have demonstrated chronic neurodegenerative features, such as diffuse brain atrophy, many years following TBI, 82 –84 and tissue studies have demonstrated accumulation of the hallmark neurodegenerative proteins Aβ and tau post-TBI. 40,85 Aβ has been shown to increase within cell bodies and axons of injured neurons within hours to days following TBI. 86 –92 In a study of cortical tissue resected from a group of living TBI patients, intracellular Aβ staining was reported in 80% of cases within hours of injury. 86 This increased expression of Aβ may lead to the formation of amyloid plaques, a key pathological hallmark of AD, in a subset of individuals, post-TBI. 93 In a postmortem study of individuals who survived 4 h to 2.5 years after injury, the presence of Aβ plaques was reported in 30% of cases (20% in cases involving those <50 years of age and 60% in cases involving those 51–60 years of age). 88 A similar finding has also been reported in long-term survivors of TBI. In a follow-up study in individuals surviving 1–47 years following a single TBI, the presence of Aβ plaques was noted in 30% of long-term TBI survivors, and these individuals had a trend toward greater density of plaques than those who had not previously experienced a TBI. 94 More recently, a neuroimaging study using Pittsburgh Compound B (PIB) to image Aβ has supported the results of these previous postmortem studies, reporting an increased PIB signal in the cortex and striatum of TBI patients compared with controls when imaged <1 year post-injury. 95 Additionally, not only has TBI been shown to be associated with Aβ plaque formation, it may actually reduce the age at which such plaque formation occurs. 86,88 It should be noted, however, that significant Aβ deposits can be seen with normal aging, making it difficult to ascertain the significance of the presence of these plaques in these studies. 96
Hyperphosphorylated tau and NFTs are similarly upregulated following TBI. 89,90,97 –100 Within 24 h of injury, an increase in phosphorylated tau can be observed in the axons and white matter. 86 Tau has also been found to be present in glial cells in 20% of cases following a single, acute TBI. 99 In a study of individuals surviving 1–47 years following a single TBI, the presence of widespread NFTs was noted in 34% of long-term TBI survivors, compared with a more focal tau distribution (i.e., entorhinal cortex and hippocampus) in only 9% of those who had not previously experienced a TBI. 94 Additionally, consistent with the findings in CTE discussed prior, these NFTs were more likely to be located at the base of the sulci and in the superficial layers (i.e., layer II and the upper third of layer III) of the cortex. 94 Tau pathology has been particularly reported following repetitive mild TBI, such as that observed in contact sports or in former military personnel. 37,39 –41 This may suggest that repeat exposure to brain injury is critical for tau hyperphosphorylation and NFT deposition. An increase in oligomeric forms of tau, which accumulate rapidly following TBI, may be particularly detrimental for long-term cognitive outcomes. 101 In support of this, a recent study by Gerson and colleagues demonstrated that TBI-derived tau oligomers worsened cognition and accelerated pathology development when injected into the hippocampi of mice overexpressing human Tau (Htau). 102 Collectively, the pathological changes in Aβ and tau may represent a foundation for the long-term development of dementia. The mechanisms that may lead to this abnormal aggregation of neurodegenerative proteins and increase in neurodegenerative changes following TBI, however, are still uncertain. 37
Neural Mechanisms by which TBI May Lead to Dementia
Following TBI, cell death does not just result from the initial insult, but is ongoing because of the initiation of a number of secondary injury factors, such as glutamate excitotoxicity, oxidative stress, and chronic neuroinflammation. This can drive long-term neuronal damage. 103 –106 and accumulation of pathological proteins, such as Aβ and phosphorylated tau, and subsequently exacerbate cognitive deficits at the time of injury, as well as increasing the risk of later developing dementia. A better understanding of the role that these factors play in ongoing neuronal injury will allow for the development of more targeted treatment to improve outcome.
DAI
One of the most harmful consequences of TBI is the widespread or diffuse disruption and disconnection of axons, in DAI. 107 Axonal injury is currently considered to be principally a progressive event, which evolves from an initial focal perturbation of the axon to ultimate axonal disconnection (secondary axotomy). 108 –110 The initiating event is thought to be an alteration in axolemmal permeability involving mechanoporation of the axolemma evoked by the shearing forces of the injury. 111 –113 Mechanoporation allows local intra-axonal calcium accumulation, with subsequent activation of various calcium-dependent cysteine protease (calpain) pathways leading to disruption of the cytoskeletal network, with subsequent disruption in axonal transport and eventual detatchment. 114 Widespread deafferentiation of downstream targets is thought to cause many of the functional deficits observed post-TBI, including cognitive impairment. 115
This disruption of normal axonal transport following TBI may also promote the abnormal accumulation of multiple proteins in the injured axon, including APP. 116 Following axonal injury, the anterograde transport of APP is impeded, allowing it to accumulate in increased levels along the length of the injured axon in areas of DAI, where it co-localizes with enzymes that are critical for the cleavage of APP into Aβ, including β-secretase (BACE1) and the presinilin 1 (PS1) subunit of γ-secretase. 117 APP, BACE1, and PS1 have all been shown to be increased following TBI. 92,118 This may subsequently lead to an upregulation of Aβ expression, 116,119 particularly of Aβ42, 90 over time, which may set into motion the pathogenic amyloid cascade seen in AD. 58,59
Axonal damage during DAI may also help to explain the abnormal accumulation of hyperphosphorylated tau observed following TBI. Tau is an MT-associated protein, which, in its normal state, helps to stabilize and organize the MTs in a parallel arrangement along the axon, known as MT bundles. 120 –122 Under the high strain rate on the axon experienced in TBI, tau proteins behave more stiffly than usual, because of their intrinsic viscoelasticity, which may inhibit the ability of adjacent MTs to slide past each other during rapid axon stretch and result in the subsequent rupture of the MTs. 123 This may be associated with the loss of the bond between tau and the MT, causing the increased release of tau dimers from the MTs into the cytosol. 124 When the MT-binding domain of tau, which mediates its ability to bind to proteins, including tau itself, is freed by its dissociation from the MT, this can increase the self-aggregation of tau into oligomers and NFTs. 125 –127 Additionally, once tau is detached from the MT and is free in the cytosol, it is more prone to be phosphorylated at disease-associated sites. 126,128 –130 In support of this, a recent study by Ando and colleagues demonstrated that the stabilization of MT-unbound tau via tau phosphorylation by the protease-activated receptor (Par)-1/microtubule affinity regulating kinase (MARK) contributes to the augmentation of AD-related phosphorylation, and may be a critical first step of abnormal tau metabolism in the pathogenesis of AD. 131 As discussed earlier, increased deposition of hyperphosphorylated tau is also a hallmark pathological marker of CTE.
Glutamate excitotoxicity
Glutamate is an excitatory neurotransmitter that is abundantly released immediately after TBI, 132 –135 because of the loss of membrane potential, 136 and release from activated microglia and astrocytes. 137 Typically, glutamate is rapidly removed from the synapse via astrocytic glutamate transporters, such as excitatory amino acid transporter (EAAT)2, but, following TBI, its reuptake is decreased because of a reduction in EAAT2138 and impaired glutamate transporter activity. 137 It is of note that similarly reduced glutamate transport function is seen in AD, with a concomitant decrease in EAAT2 protein expression. 139
Acute glutamate excitotoxicity primarily leads to neuronal cell death and degradation by initiating and sustaining intracellular calcium influx.
103,140
–142
Glutamate acts on both ionotrophic glutamatergic receptors (i.e., N-methyl-
Excitotoxicity may also play a role in the accumulation of abnormal proteins associated with neurodegeneration and dementia. Prolonged activation of NMDA receptors alters APP processing and stimulates Aβ production in neurons, 146 in line with reports of a correlation between neuronal firing at excitatory synapses and Aβ production. 147,148 Further, although excitotoxicity, induced by a kainite model, was associated with a brief acute decrease in tau phosphorylation, this was followed by a prolonged period of increased tau phosphorylation. 149
Oxidative stress
Following TBI, there is an increase in the production of free radical, oxidative stress-inducing agents, such as reactive oxygen species (ROS), which include superoxide radicals and hydroxyl radicals, and reactive nitrogen species (RNS), which include peroxynitrite. 150,151 Oxidative stress ensues when their production overwhelms the antioxidant mechanisms within the brain. This is signified by the decrease seen in levels of antioxidants such as superoxide dismutase (SOD), 152 glutathione (GSH), and glutathione peroxidase (GPx) 106 after TBI. As ROS contain an unpaired electron in the outermost orbit, they are highly reactive, inducing tissue damage via peroxidation of cellular and vascular structures, protein oxidation, cleavage of DNA, and inhibition of the mitochondrial electron transport chain. 153 Membrane lipids are particularly vulnerable, with lipid peroxidation altering the structure, fluidity, and transport function of the membrane, ultimately causing membrane lysis. 154 Oxidative damage is linked to cell death pathway activation, axonal injury, and impaired synaptic plasticity, which can all contribute to cognitive deficits after TBI. 19
Importantly, the oxidative stress cascade set in motion by the initial injury may persist for years to decades. Mackay and colleagues found that levels of lipid peroxidation products within the serum of patients who had sustained a brain injury up to 35 years previously (range 1–35 years) were approximately sevenfold greater than levels in control subjects. 155 This may play a role in the later development of dementia, with a decrease in antioxidant activity, indicative of oxidative stress, shown to be related to age-related memory impairments. Evidence suggests that oxidative stress is a key player in the development of AD, 156 with cellular damage caused by oxidative stress such as protein oxidation, lipid oxidation, and DNA oxidation linked to the development of cognitive decline. 157
Inflammation
A growing body of evidence suggests that, of all the mechanisms discussed, neuroinflammation may provide the key link between TBI and later dementia. 158,159 It is well established that a robust neuroinflammatory response develops acutely post-TBI, contributing to the secondary injury cascade. 160 Neuroinflammation is an immune response to injury, triggered by alarmins binding to pattern-recognition receptors, such as the Toll-like receptor family, in order to activate resident microglia and recruit peripheral leukocytes and astrocytes to the site of injury, in hopes of injury recovery and repair. 161 –163 This is the acute phase of the neuroinflammatory response. At the site of the injury, glia, the resident inflammatory cells of the central nervous system (CNS), become activated within minutes of injury, 164 cleaning up damaged cell debris by phagocytosis, releasing anti-inflammatory cytokines and neurotrophic factors, thus serving neuroprotective functions immediately post-TBI. 163,165,166 These glial cells, however, also express pro-inflammatory cytokines, such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β and IL-6, promoting neuronal damage, as well as expression of activation markers, such as major histocompatibility complex class II (MHC II). 167,168
It has been hypothesized that, in a subset of TBI patients, there is incomplete resolution of this acute neuroinflammatory response. 37 A vicious cycle is initiated following the original insult, in which release of pro-inflammatory factors by resident glial cells promotes further glial activation, leading to a progressive, chronic cycle of neuroinflammation, 169 which can have neurotoxic effects on neurons through mechanisms such as oxidative stress, apoptosis, and excitotoxicity. 169,170 Multiple studies have demonstrated persistent neuroinflammation in the brain parenchyma, serum and, cerebrospinal fluid of TBI patients. 105,171 –173 In rodents, microglial activation has been demonstrated up to 1 year post-TBI, and was associated with progressive lesion expansion, hippocampal degeneration, myelin loss, and oxidative stress. 81 In humans, reactive microglia have been found in brain tissue up to 18 years following single, moderate-to-severe TBI. 174 Although most studies have relied on postmortem analyses, a small study by Ramlackhansingh and colleagues corroborated these findings by assessing the neuroinflammatory response to human TBI in vivo. 105 The study utilized positron emission tomography to demonstrate microglial activation persisting up to 17 years post-injury in multiple brain regions, including the thalami, putamen, occipital cortices, and posterior limb of the internal capsule. Interestingly, thalamic microglial activation correlated with degree of cognitive impairment. Over time, these persistently elevated levels of reactive glial cells post-TBI may result in increased release of cytokines, chemokines, and other neurotoxic chemicals, causing ongoing neuronal injury and thus promoting the emergence of cognitive impairment. 175 To date, however, most research on neuroinflammation after TBI has focused on the acute microglial response and the subsequent release of pro-inflammatory cytokines immediately following injury, 81,105,171,176,177 with further research needed into the effects of persistent neuroinflammation following TBI. This ongoing neuroinflammatory state is of particular interest given the fact that other conditions that promote chronic inflammation, such as a history of systemic infection, obesity, and reduced physical activity, are also known risk factors for AD. 178 –180
Neurotrophic Factors as a Treatment Strategy for Long-Term Cognitive Deficits after TBI
Given the knowledge that the secondary injury cascade following TBI plays an integral role in cognitive deficits experienced immediately post-injury, as well as delayed onset symptoms, timely treatment may halt ongoing neuronal injury, with a resultant improvement in cognition. To date, no treatments for TBI have successfully translated from animal to clinical studies, 181 and the reasons underlying this lack of translation have been reviewed extensively elsewhere, including lack of full pre-clinical evaluation with appropriate dose/response relationships and pharmacokinetic dynamics, as well as difficulty in classifying TBI. 182 A recent Cochrane systematic review showed that pharmacological agents for chronic cognitive impairment post-TBI had insufficient evidence for effectiveness when compared with placebo. 183 One such treatment, methylphenidate, despite looking promising in initial clinical trials, has now been reported to be associated with long-term adverse effects associated with the prolonged use required to achieve cognitive benefits post-TBI. 184 This highlights the need to continue research into this area to allow a better understanding of the underlying pathophysiology driving the development of cognitive deficits following TBI to allow for the development of better treatments.
One potential treatment strategy may be to enhance the body's natural neurotrophic factors. Neurotrophic factors (neurotrophins) are proteins that are important for neuronal cell growth, stability and plasticity. 185,186 Examples of neurotrophins include nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), neurotrophin-4/5 (NT 4/5), glial-derived neurotrophic factor (GDNF), and granulocyte colony stimulating factor (G-CSF). 186,187 Recently, it has become evident that the dysregulation of neurotrophic factors in the brain exacerbates functional impairment post-TBI. 188 Therefore, it can be argued that the acute administration of exogenous neurotrophic factors after TBI may target key mediators in the secondary injury cascade, while promoting repair, thereby not only having potential benefits for the acute effects of TBI, but also potentially preventing long-term cognitive impairment and decreasing the risk of later developing dementia (Fig. 1).
In addition to the primary injury, traumatic brain injury (TBI) is associated with a variety of secondary injury cascades, including excitotoxicity, changes in glucose metabolism, mitochondrial dysfunction, neuroinflammation, and oxidative stress. Together, these cascades result in diffuse axonal injury, synaptic loss, the deposition of neuropathological proteins, and neuronal cell death. Subsequently, cognitive impairment and dementia may develop. Neurotrophic factors (NTFs) may be capable of intervening to block these pathological secondary cascade mechanisms (left), while simultaneously stimulating beneficial processes, such as neurogenesis, neurite outgrowth, axonal sprouting, synaptic plasticity, and additional growth factor production (right). Therefore, it is reasonable to hypothesize that NTFs may be beneficial for preventing the development of cognitive impairment and dementia chronically following TBI. BDNF, brain-derived neurotrophic factor; G-CSF, granulocyte colony stimulating factor; GDNF, glial-derived neurotrophic factor; NGF, nerve growth factor; NT3, neurotrophin-3; NT4/5, neurotrophin-4/5. Color image is available online at
NGF
NGF was one of the first neurotrophic factors characterized, when it was discovered to stimulate neurite outgrowth in the peripheral nervous system by Viktor Hamberger, Rita Levi-Montalcini, and Stanley Cohen in the 1960s. 189 NGF is expressed on many cell types, including neurons and glia, 190 helping to induce the growth of neurons 191 and to maintain neuronal proliferation through its interaction with other growth factors. 192 It is translated as the unprocessed precursor proNGF, and cleaved either intracellularly by furin or extracellularly by plasmin and matrix metalloproteases (MMPs) into mature NGF. 193,194 A role in the secondary injury cascade following TBI has been suggested, with higher NGF expression associated with better neurological outcomes following TBI in children. 195,196 A number of mechanisms whereby NGF may prove beneficial have been proposed, with NGF protecting neurons from excitotoxicity, 197,198 and from cytotoxic agents, such as ROS 199 and Aβ. 200 NGF administration in a rodent model of TBI increased major antioxidant enzymes, such as SOD, GPx, and catalase, in the injured brain, while simultaneously decreasing concentrations of intracellular calcium. 201 It has also been shown to inhibit the nuclear factor (NF)-κB pathway and to reduce neuroinflammation, 202 a major mechanism of secondary injury post-TBI that may help to explain the link between TBI and dementia risk. It is of note that NGF also acts to preserve the cholinergic system within the basal forebrain, which has been shown to undergo significant dysfunction following TBI, 203 with this pathway critically involved in learning and memory. 204 Dixon and colleagues demonstrated that intraventricular NGF infusion immediately following injury was able to preserve forebrain cholinergic function and improve spatial memory performance at 1 week post-injury in a rodent model of TBI. 205 It is likely that these beneficial effects are the result of actions of NGF directly within the basal forebrain. In support of this, the engraftment of NGF-expressing human Ntera2/D1 neuron-like (NT2N) neurons directly into the medial septum, a key area in the production of acetylcholine (ACh), was effective at improving learning ability in mice at 1 month following TBI. 206
These beneficial effects at 1 month post-injury suggest that the utility of NGF as a treatment strategy could potentially be long lasting. Lin and colleagues have recently corroborated this by demonstrating that pseudo-lentivirus-delivered NGF gene administration directly into the hippocampus led to expression of NGF lasting >1 month, and was associated with increased neurite outgrowth and improved cognitive function following TBI in a rat model. 207 Unfortunately, the poor blood–brain barrier (BBB) permeability of NGF has limited its clinical utility to date. 208 More recent studies have sought to resolve this difficulty through the use of an intranasal formulation. Intranasal delivery of NGF in a rodent model of TBI has recently been demonstrated to attenuate aquaporin-4 induced edema, 202 reduce Aβ deposition, 209 and decrease tau phosphorylation. 210
To date, the long-term effects of NGF following TBI have not been evaluated, although NGF has been examined as a potential treatment for AD. The earliest small studies utilized an intracerebroventricular method of delivery in either a single case, 280 or in three patients, 281 and although positive effects were seen in some areas of cognitive functioning, side effects such as back pain, weight loss, and Schwann cell migration into the medulla and spinal cord led to the discontinuing of this particular route of delivery. More targeted gene therapy was employed by Tuszynski and colleagues in a small study of 10 AD patients, which was associated with axonal sprouting in degenerating cholinergic neurons for 10 years post-gene transfer with no reported side effects. 211 Therefore, NGF as a treatment remains in the earliest explorative phase; however, it may be worth investigating whether an acute period of administration following TBI could have long-lasting benefits on cognition as a result of its ability to intervene in a number of aspects of the secondary injury cascade, and whether this could extend to reducing the risk of later developing dementia.
BDNF
BDNF, which has been shown to have affinity for the tropomyosin receptor kinase B (TrkB) receptor, 212 has potent effects on plasticity as it promotes neurogenesis and synaptic growth and modulates synaptic plasticity. 213 Apart from these potential beneficial effects on the reparative process following TBI, BDNF has also been shown to target secondary injury processes that are known to occur following TBI that are linked with cognitive impairment. For example, BDNF protects against glutamate excitotoxicity, 214 and also modulates the inflammatory response by downregulating pro-inflammatory cytokine expression, while enhancing anti-inflammatory cytokine expression. 215,216 The potent effects of BDNF on cell survival have been demonstrated extensively in in vitro models, with BDNF preventing neuronal death induced by ischemia, 217 oxidative stress, 218 glutamate toxicity, 214,218 and Aβ. 219 This supports a study by Korley and colleagues that showed that patients with lower serum levels of BDNF had a higher risk of incomplete recovery after TBI, with serum BDNF concentrations inversely proportional to TBI severity. 220 Similarly, a decrease in BDNF mRNA and protein expression in the adult rat hippocampus post-TBI was associated with cognitive decline. 221 –224 This may be a factor in the later development of dementia, with BDNF mRNA and protein levels reduced in key areas related to cognitive function including the hippocampus, entorhinal cortex, and parietal lobes in AD patients. 225, 226 Patients with AD have a decrease in the levels of BDNF within the cerebrospinal fluid (CSF) compared with age-matched controls, 227,228 and BDNF CSF levels are a predictor of progression from minor cognitive impairment to AD. 228
Treatments that enhance endogenous BDNF, such as exercise and simvastatin, have been shown to have beneficial effects pre-clinically post-TBI, maintaining synaptic integrity within the hippocampus and improving Morris Water Maze (MWM) performance, both acutely and more chronically (3 months post-injury). 229,230 Despite this, exogenous application of BDNF has been less successful. A study by Blaha and colleagues showed that a 14 day continuous administration of exogenous BDNF directly into the brain, starting at 4 h after a lateral fluid percussion TBI, had no significant neuroprotective effect, with similar levels of neuronal damage and cognitive impairment on the MWM. 231 This was not the result of a downregulation of the BDNF receptor TrκB, as transgenic mice overexpressing the receptor still demonstrated no improvement following TBI after BDNF treatment. 232 Coupled with the difficulty in administering BDNF, given its inability to cross the BBB and poor bioavailability, 233 treatment strategies targeting this neurotrophic factor after TBI may be best focused on enhancing endogenous BDNF activity.
NT-3 and NT-4/5
Unlike the neurotrophic factors discussed previously, studies on the effects of NT-3 and NT-4/5 have received less attention. Like BDNF, NT-4/5 can bind to TrκB, resulting in its homodimerization and autophosphorylation, and activating downstream signaling via the MAPK pathway, which plays a pivotal role in cell survival and synaptic plasticity. 234 Royo and his group not only showed that NT-4/5 was upregulated immediately post-injury as a neuroprotective measure in the hippocampus, but also demonstrated that exogenous administration of NT-4/5 via delivery directly into the injured left primary somatosensory cortex in the brain prevented neuronal cell death in the hippocampal CA region following lateral fluid percussion injury. 235 However, in a follow up study, they concluded that the neuroprotective effects of NT-4/5 in the hippocampus did not result in functional improvement in the MWM or contextual fear paradigm. 236 Further, NT 4/5 has also shown toxicity effects in previous studies, indicating that it may be a poor choice for long-term treatment after TBI. 237, 238
Unlike NT 4/5, NT-3 has the highest affinity to TrκC, and has been shown to prevent glutamate excitotoxicity in vitro 239 and to reduce apoptotic cell death following ischemic stroke when applied directly to the cortical surface. 240 NT-3 mRNA levels are slightly decreased within the hippocampus in the immediate aftermath following TBI; 241 however, no changes in mRNA or protein levels within the brain have been reported in AD, 242,243 suggesting that other neurotrophic factors may provide better therapeutic targets.
GDNF
GDNF is a member of the transforming growth factor superfamily. Although its primary function is as a potent trophic factor for dopaminergic neurons, it has also been shown to be neuroprotective in a number of paradigms including TBI, 244,245 as well as ischemia 246 and excitotoxic injury. 247 GDNF may also reduce inflammation by downregulating the expression of pro-inflammatory cytokines 248 and may protect against apoptotic cell death by maintaining mitochondrial integrity in neurons. 249
It also plays a role in cognitive function, as it promotes survival and axonal sprouting of noradrenergic neurons within the locus coeruleus, which has projections to the frontal cortex and hippocampus. 250,251 When astrocytes within the hippocampus were induced to overexpress GDNF levels, aged rats demonstrated improvements in cognition, spending more time searching in the correct quadrant of the MWM. 252 Brain-injured rats treated with neural progenitors engineered to secrete GDNF also showed improved outcome, with decreased latency to find the hidden platform on the MWM, although cortical lesion volume was not affected. 244 Whether GDNF treatment following TBI would lead to lasting changes, including reducing the risk of developing dementia, has yet to be determined. GDNF concentrations within the CSF are increased in patients with AD, 252 although reports have suggested that overall GDNF signaling may be decreased because of a reduction in its receptor, GDNF family receptor alpha-1 (GFRα1). 253 Future studies are needed to explore whether GDNF treatment is capable of influencing long-term outcome post-TBI.
G-CSF
Instead of administration of exogenous neurotrophic factors, another possible approach to treating cognitive deficits post-TBI is to stimulate the production and maintain the stability of endogenous neurotrophic factors. Recently, researchers have proposed the usage of G-CSF, a naturally occurring potent growth factor, as a possible treatment for TBI, as it has the ability to promote and maintain endogenous neurotrophic factors such as BDNF, without major side effects. 254
G-CSF is a hematopoietic growth factor that is widely expressed in the endothelium, macrophages, fibroblasts, and epithelial cells together with its receptor (G-CSFR). 255, 256 G-CSF is currently used clinically as a treatment for post-chemotherapy neutropenia, 257 –259 but as it can cross the BBB, it may also prove beneficial for disorders of the CNS. Of particular interest in the treatment of TBI, G-CSF has been shown to target a number of the secondary injury factors, with evidence showing that it can suppress production of pro-inflammatory cytokines, activate anti-apoptotic pathways (phosphatidylinositol 3-kinase [PI3K]/Akt, STAT3, and extracellular signal–regulated kinase [ERK]) and increase the production of BDNF. 187,260 –263 Of particular relevance for the treatment of long-term cognitive consequences of TBI, the G-CSFR is known to be expressed on adult neural stem cell and to induce neuronal differentiation and neurogenesis in the hippocampus. 262 Excitingly, G-CSF is also capable of mobilizing bone marrow mesenchymal stem cells into the peripheral blood, from which they can integrate into the injured brain and differentiate into neuronal cells. 264 This generation of new neurons in key brain areas may be critical for both the treatment of persistent cognitive deficits and the prevention of dementia following TBI.
The acute effects of G-CSF in injury models have been extensively studied. 187,265,266 Subcutaneous administration of G-CSF resulted in improved cognitive performance on the MWM up to 1 week post-injury in an experimental model of TBI. 265 G-CSF has also been shown to enhance acute cellular proliferation and motor recovery following TBI, 267 as well as inducing repair and regeneration of neurons in animal models of spinal cord injury. 263,268 Interestingly, a 2004 study by Sheibani and colleagues showed that mice administered G-CSF twice daily for 1 week post-TBI showed only modest improvements on cognitive performance on the MWM, and no improvement in motor performance, compared with saline-treated controls. 266 Sakowitz and colleagues reported a similar lack of efficacy, noting that G-CSF did not reduce either contusion size, brain edema, or CSF glutamate levels in the controlled cortical impact model. 269 It is important to note, however, that this study employed only a single dose of G-CSF, 30 min after injury induction. Conversely, a more recent study by Song showed that G-CSF treatment for 3 consecutive days post-injury in mice resulted in significant motor and cognitive improvements up to 2 weeks post-TBI, which were associated with increased neurogenesis in the hippocampus, increased recruitment of microglia and astrocytes, and upregulation of the neurotrophic factors BDNF and GDNF. 187 Even delayed G-CSF treatment has been shown to be effective in models of injury. In a severe contusive spinal cord injury model, serial subcutaneous injections on days 9–13 post-injury were able to improve functional outcomes and promote neural recovery. 270 Recently, several studies have suggested that the beneficial effects of G-CSF may be further augmented by co-administration with umbilical cord blood. 271 –273
Although the positive effects of G-CSF on cognition acutely following TBI have now been well characterized, the chronic effects of G-CSF remain unknown. Because G-CSF is highly effective in the CNS, can be administered peripherally, and has shown no irreversible adverse side, even at high doses, 254,274 it may be a novel and ideal treatment for the prevention of dementia development post-TBI. In support of this, G-CSF has been shown to be potentially beneficial as a treatment strategy for AD. G-CSF plasma levels are known to be reduced in patients with early AD. 275 Subcutaneous injection of G-CSF was able to significantly improve cognitive function and increase ACh levels in two different Aβ-induced mouse models of AD. 276 Similarly, in a mouse model of AD, treatment with G-CSF decreased brain amyloid burden and reversed cognitive impairment. 277 These beneficial effects may be the result of a stimulation of neurogenesis, an accumulation of hematopoietic stem cells from the bone marrow, which G-CSF is able to mobilize into the peripheral blood, 276 or an attenuation of chronic neuroinflammation. 277,278 A recent small pilot double-blind placebo-controlled crossover design clinical study with eight patients found that there was a small improvement in a hippocampal-dependent task in Alzheimer's disease patients. 254 G-CSF treatment was not associated with serious adverse events in this small trial; the most common side effects were transient increases in white blood cells, myalgias, and diffuse aching that improved with nonsteroidal anti-inflammatory medications. Despite this, G-CSF may still lead to side effects in some patients. In a trial with coronary artery disease patients, G-CSF of 10 mg/day for 5 days led to a doubling of plasma levels of C-reactive protein, most patients reported musculoskeletal pain, and white blood cells increased nearly fivefold. 279 Further, two patients experienced serious adverse events, including one myocardial infarction (MI) 8 h after the fifth dose of G-CSF, and one MI, resulting in death, 17 days after treatment. Further investigation is therefore needed to confirm whether G-CSF may represent a safe and viable treatment option.
Conclusion
TBI remains a major cause of disability and mortality worldwide. It is important to recognize, however, that the consequences of TBI go far beyond the biomechanical injury itself. Because of the complex secondary injury cascade, significant impairments in brain function may persist for years, or even decades, following the acute phase of injury. 3,4 This ongoing secondary injury cascade may explain the increased number of persons with TBI who develop dementia, even individuals who have seemingly recovered from their injury decades earlier. 4 –8 Despite the growing awareness of the chronic consequences of TBI for cognition, however, this is still a greatly under-studied area. The majority of experimental studies investigating the consequences of TBI only follow the animals for up to a month post-injury. Because of this significant gap in the literature, the specific neural mechanisms that may underlie the link among TBI, persistent cognitive impairment, and dementia risk remain unknown. This has significantly limited the development of treatment strategies to date. Pharmacological agents for chronic cognitive impairment post-TBI have insufficient evidence for effectiveness when compared with placebo, 183 and these treatments may even potentially be harmful. 184
Given the ineffectiveness of current treatments, a novel therapeutic strategy to treat long-term cognitive deficits and prevent the development of dementia is an area of major unmet clinical need. As discussed in this review, because of their multi-faceted effects on multiple aspects of the secondary injury cascade, neurotrophic factors may represent just such an ideal novel treatment strategy. Although several of these factors have been shown to be effective in treating the acute cognitive consequences of TBI, their chronic effects remain unknown. Longer-term, longitudinal studies are needed to truly understand the full spectrum of potential benefits associated with neurotrophic factor administration. Further, any potential side effects, such as increased risk of epilepsy relating to neuronal sprouting, will need to be evaluated. Additionally, it remains to be investigated whether a combination of neurotrophic factors, administered simultaneously, may have additional benefits beyond treatment with one factor alone. Finally, future research should more fully investigate additional neural mechanisms that may be affected by neurotrophic factors. Such studies may be critical in identifying additional potential therapeutic targets, which could provide hope for the millions of people currently living with a past history TBI and worrying about their future risk of dementia.
Footnotes
Acknowledgments
This work was supported by a grant from the Neurosurgical Research Foundation.
Author Disclosure Statement
No competing financial interests exist.