Biofluid biomarkers of traumatic brain injury

ABSTRACT

Primary objective: The purpose of this paper is to review the clinical and research utility and applications of blood, cerebrospinal fluid (CSF), and cerebral microdialysis biomarkers in traumatic brain injury (TBI). Research design: Not applicable. Methods and procedures: A selective review was performed on these biofluid biomarkers in TBI. Main outcome and results: Neurofilament heavy chain protein (NF-H), glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase-L1 (UCHL1), neuron-specific enolase (NSE), myelin basic protein (MBP), tau, and s100β blood biomarkers are elevated during the acute phase of severe head trauma but have key limitations in their research and clinical applications to mild TBI (mTBI). CSF biomarkers currently provide the best reflection of the central nervous system (CNS) pathobiological processes in TBI. Both animal and human studies of TBI have demonstrated the importance of serial sampling of biofluids and suggest that CSF biomarkers may be better equipped to characterize both TBI severity and temporal profiles. Conclusions: The identification of biofluid biomarkers could play a vital role in identifying, diagnosing, and treating the underlying individual pathobiological changes of TBI. CNS-derived exosomes analyzed by ultra-high sensitivity detection methods have the potential to identify blood biomarkers for the range of TBI severity and time course.

KEYWORDS:

• (that are not already included in title): mild traumatic brain injury (mTBI)
• cerebrospinal fluid (CSF)

Introduction

Traumatic brain injury (TBI) results from a range of physical forces, including blows to the head from participation in amateur or professional contact sports; head and neck injuries suffered during falls, motor-vehicle accidents, or domestic violence; and shock wave caused by explosive blast sustained by military service members from improvised explosive devices (1–4). TBI is not a single disease entity, as these physical forces are of varying mechanisms and intensity, and their effects on the brain accordingly trigger a breadth of functional, cellular, and molecular changes (5,6). TBI is thus best described as a spectrum disorder with severity ranging from mild TBI (mTBI), which is synonymous with the term concussion, to severe TBI.

TBI is also a dynamic condition. The injury process following an insult to the head consists of two major phases: a primary phase and a secondary phase (5). The primary phase is the direct and immediate consequence of the physical forces, including the distortion of axons and blood vessels and injury to cell membranes, whereas the secondary phase represents the body’s attempt to limit and repair these consequences and to restore structural and functional integrity (7,8). The secondary phase may offer opportunities for pharmacological interventions that can limit damage and improve functional recovery, but future research is needed to identify the pathobiological processes and their time course to inform interventions during the appropriate therapeutic windows.

Depending on the type and severity of TBI, the secondary injury phase consists of several partly overlapping pathobiological processes; metabolic (9,10), axonal (11–13), and vascular (14,15) changes; and inflammation (16–18). Experimental evidence has shown that each of these processes has its own temporal profile and that the temporal pattern of these pathobiologies is especially dynamic during the acute to subacute post-injury phases (7,19). Thus, in addition to concentrating on the severity of TBI, researchers and clinicians should also address the temporal distinctions (20–23). These temporal distinctions are especially important for identifying and/or classifying the chronic effects of TBI as a separate disease entity.

Biofluid biomarkers can play a critical role in characterizing both TBI severity and temporal profiles. They hold promise for monitoring progression, predicting clinical outcomes, and providing molecular-level information about the ongoing pathobiological changes of TBI (24). Gathering this kind of information is critical for informing evidence-based therapeutic interventions and for assessing individual patient responses to such interventions. This is particularly important for patients with repetitive mTBI who may be at risk for chronic traumatic encephalopathy (CTE) or other tauopathy-associated neurodegenerative disorders (25,26).

Blood biomarkers in TBI in animal models

Animal models provide the most homogeneous and reproducible method of studying TBI, as animal models of experimental TBI avoid such confounding factors as age, medication use, comorbidity, and polytrauma (27). Thus, by using appropriate sham controls, changes in the serum or plasma levels of biochemical markers can be attributed solely to the injury.

Swine offer a high-fidelity biological model for TBI due to a similarity in physiology and size to humans that has enabled researchers to perform clinically relevant physiological monitoring (28–30). Moreover, swine have gyrencephalic brains with a cellular architecture that is closer to humans than to rodents (31). In a swine model of blast-induced TBI, changes in the serum concentrations of neurofilament heavy chain protein (NF-H) and the time course of those changes predicted injury severity and the likelihood of favourable vs. unfavourable outcomes (32). In that study, the temporal profiles of four blood protein biomarkers, S100β, neuron-specific enolase (NSE), myelin basic protein (MBP), and NF-H, demonstrated distinct temporal profiles over two weeks post-injury, but only the temporal profile of NF-H correlated with outcome. In contrast to animals with favourable outcomes, those with unfavourable outcomes had serum levels of NF-H reach maximum concentration at 6 hours post-injury, return to normal preinjury levels by three days post-injury, and remain at the pre-injury level at the two-week termination point.

Other studies using rat and mouse models of blast-induced TBI have demonstrated that serum concentrations of biomarkers are associated with dynamic changes in metabolism, cell adhesion, extracellular matrix, vascular function, neuronal and glial damage, axonal injury, and inflammation over time (7). Moreover, numerous individual markers have been linked to highly complex and dynamically changing temporal patterns up to 30 days after injury. These and other studies have demonstrated the importance of serial sampling of biofluids: a single post-injury sample may not capture dynamically changing concentrations of biofluid biomarkers. Single time-point biomarker values may therefore, in fact, be misleading. Without serial samples, underlying pathobiological processes, indicators of disease progression, and optimum therapeutic windows may be missed. These problems are even greater in TBI in humans, in whom TBI is often associated with damage to extracranial organ systems and other confounding factors (33).

Blood biomarkers in TBI in humans

Hundreds of millions of dollars have been invested by the National Institutes of Health, the Departments of Defense and Veterans Affairs, and industry in the search for reliable, sensitive, and specific blood biomarkers for both acute TBI and the chronic postconcussive state in humans. Although blood-based biochemical markers (i.e., biomarkers in serum and/or plasma) are easily obtained and have been successfully used in the diagnosis and monitoring of several other diseases, these studies have only had limited success in TBI. That said, a handful of plasma biomarkers—NF-H, glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase–L1 (UCHL1), NSE, MBP, tau, and s100β—have been consistently demonstrated to be elevated during the acute phase of severe head trauma and to be informative with respect to prognosis (reviewed in 34–36).

Several studies have attempted to translate the plasma NF-H work in animal models of TBI to humans with TBI, although some of this work has concentrated on the heavily phosphorylated axonal form of NF-H (pNF-H). For example, two studies found that NF-H and/or pNF-H levels increased significantly faster in samples of children who had worse Glasgow Outcome Scale scores and/or more deaths (37,38). Gatson JW et al. then extended this work to adults with TBI, suggesting that pNF-H may be a useful biomarker for more severe TBI at one and three days post-injury (39). However, no studies have shown the reliability of NF-H in adults outside of acute TBI settings.

GFAP, an intermediate filament protein of the astrocytic cytoskeleton, has been used successfully, especially in combination with UCHL1, to assess injury severity and predict outcomes in the acute TBI setting (40). In this context, elevated serum GFAP levels also correlate with injury severity across the TBI severity spectrum, as there is some evidence that elevated serum levels of GFAP and its breakdown products are elevated in mTBI (41,42). Nevertheless, despite some promising studies, the specificity and predictive value of serum GFAP in mTBI/concussion remains unclear.

UCHL1 is a neuron-specific cytoplasmic enzyme involved in protein ubiquitination and elimination via the ATP-dependent proteasome pathway. Elevated serum levels of UCHL1 during the acute phase following brain injury have been correlated with injury severity (43). As we have discussed, serum UCHL1 may be especially helpful when combined with serum GFAP; several studies have shown that both the predictive value of serum UCHL1 and its correlation with injury severity increase substantially when combined with serum GFAP. Moreover, determining the glial:neuronal ratio by measuring injury-induced changes in serum levels of GFAP and UCHL1 has been proposed as a potential aid in estimating intracranial pathology after severe TBI (44). Importantly, UCHL1 is not central nervous system (CNS) specific; it is also expressed in the neurons of the peripheral nervous system, various cells of the endocrine system, endothelial cells, smooth muscle cells, and certain tumours (45–47). This limits the utility of blood UCHL1 as a TBI biomarker.

The gamma-gamma isoform of NSE is a neuron-specific glycolytic enzyme that is released into the extracellular space upon neuronal damage and death (48). In the acute setting, serum NSE is a good predictor of the extent of neuronal damage, but it also has serious drawbacks, as NSE lacks neuronal specificity and is expressed in erythrocytes (and other cell types), rendering blood-based NSE measurements difficult to interpret. Only one study has shown success in these efforts—a study of boxers following a bout—and that study showed that elevated serum NSE persisted for two months (49). However, no plasma or serum biomarkers have previously been identified that are persistently elevated in mTBI long after exposure.

MBP is a small (18.5 kDa), positively charged extrinsic-membrane protein that constitutes ~30% of the total protein of the CNS myelin (50,51). MBP is released into the CSF and the peripheral blood in both myelin disorders and at the time of a TBI. Attempts to use MBP as a blood-based biomarker in the assessment of injury severity and outcome have led to mixed results. Clinical studies during the acute phase of severe TBI have found positive correlations between injury severity and MBP serum levels (50,52). However, MBP is not specific to the CNS, as the myelin of peripheral nerves also express MBP, and severe clinical TBI almost always involves injury to the peripheral nerves. This means that the diagnostic and predictive value of MBP serum levels in TBI is limited. Moreover, MBP transcripts are also present in the bone marrow and immune system (53).

The microtubule-associated protein tau is abundant in axons, but tau is also expressed by hepatocytes, as well as by cells in the kidneys and testis (54,55). ‘Big’ tau, the larger isoform of tau that shares epitopes with CNS-specific tau, is expressed by peripheral nerves and muscle. Ultra-sensitive methods, such as the Quanterix digital Simoa assay platform, enable single-molecule detection and have enabled measurement of plasma tau concentrations. A study in professional hockey players using the Quanterix Simoa platform demonstrated an elevation in plasma total tau acutely following a hockey game compared to pre-season baseline levels (56). A recent study of active-duty soldiers in the chronic postconcussive state (i.e., a mean of 18 months and at least three months following last exposure to repetitive mTBI) demonstrated similar elevations of plasma total tau using the Simoa platform (57). These researchers observed a dose-dependent effect in relation to the number of mTBI exposures, with increased plasma tau concentrations in service members who had more than three mTBIs. They also found that plasma tau concentrations were correlated with the severity of self-reported postconcussive symptoms, providing evidence that elevated plasma tau concentrations are clinically meaningful. However, these studies also pose several caveats: the differences in plasma tau concentrations between the TBI and control groups were very small and overlapped substantially; numerous samples were below the level of detection, even when using the ultra-sensitive Simoa assay; and most importantly, current antibody-based assays of tau do not distinguish between CNS-derived and non-CNS-derived sources of tau, including ‘big’ tau found in muscle. These caveats substantially limit the utility of blood-based tau levels as indicators of mTBI and the chronic postconcussive state.

Over 300 studies have used S100β, a calcium-binding protein that is primarily expressed in astroglia, to assess the extent of injury in acute care clinical settings (58,59). High initial serum S100β levels are known to predict poor outcomes, especially if there is a second rise in serum S100β levels during the more subacute phase (60–62). The appearance of the second S100β peak suggests ongoing damage to astroglia that is most likely due to excitotoxicity and/or inflammation, whereas low initial serum S100β levels and the lack of a second S100β peak (along with other clinical measures) indicate mTBI and a likelihood of good functional recovery. Nonetheless, the utility of serum S100β in mTBI/concussion is limited, as serum S100β can also be derived from extracranial sources: moderate increases in serum S100β concentration can therefore be the result of muscle injury, particularly in the setting of military operations and contact sports (59,63). This means that the moderate serum S100β elevations observed in the acute mTBI setting may be nonspecific. Given these complications, the predictive value of elevated blood levels of S100β for chronic sequelae of mTBI (chronic postconcussive symptoms and cognitive impairment) remains inconclusive.

Clinical studies analysing the temporal profile of S100β, however, have had encouraging findings, as patients with TBI have been shown to have a secondary increase in serum levels of S100β 48 hours after severe TBI, and these increases have strongly correlated with the development of clinically significant secondary radiological findings (62). A large retrospective study involving more than 150 patients with TBI highlighted the importance of determining the temporal profile of changes in serum biomarkers, especially S100β, by demonstrating a strong correlation between S100β and post-TBI outcomes (64).

Determining the temporal pattern of changes is a key use for TBI biomarkers in humans, as understanding these changes can help clinicians to objectively identify a safe timeline for patients to return to duty or to play. To this end, a multicenter prospective cohort study in ice hockey players monitored serum and plasma levels of S100β, NSE, and total tau in the pre-season and then after the athletes experienced concussions (65). This study demonstrated substantial time-dependent changes up to five days after injury in the blood levels of these biomarkers. They also found that plasma total tau and serum S100β levels correlated best with concussion and that the return of total tau and S100β to pre-injury levels correlated well with successful rehabilitation and safe return to play. If blood-based biomarkers are to be useful in the clinical management of TBI, these are the kinds of approaches that will be necessary.

Limitations of blood biomarker approaches in TBI

The levels of currently used biomarkers in the blood reflect the extent of neuronal and glial damage and loss. This means that these markers may be useful in severe TBI, especially in the acute post-trauma phase, for demonstrating severity of injury and in predicting clinical outcomes. However, as discussed above, these biomarkers are not exclusively CNS-specific; extracranial sources may contribute to circulating plasma or serum concentrations, confounding their diagnostic and prognostic values.

Moreover, these biomarkers do not reflect important pathologies, such as metabolic changes, vascular pathologies, and inflammation during the secondary injury phase that may be responsible for causing and maintaining progressive neuronal and glial damage following TBI. Identifying such pathobiological processes is critical for the rational design of evidence-based therapeutic interventions. Indeed, the current lack of biofluid biomarkers for downstream effects following acute TBI may have contributed to the current 100% failure rate of experimental pharmacotherapies for TBI.

Other factors also limit the utility of blood biomarkers. In seeking biomarkers for TBI, researchers must contend with systems that clear the brain’s interstitial space of molecules, particularly the proteins released from damaged cells. This process is complicated by the involvement of several overlapping systems, including the blood–brain barrier (BBB), interstitial fluid (ISF) bulk flow, and cerebrospinal fluid (CSF) absorption into the circulatory and peripheral lymphatic systems (66). One obstacle is that the mechanisms by which brain interstitial fluid molecules are transported into peripheral circulation are highly complex and not well understood, and another is that the transportation of proteins through the BBB depends on the solubility, molecular weight, and size of proteins. The blood biomarkers of TBI that we have described vary substantially in each of these parameters, which likely affects their clearance from the interstitial space in the absence of BBB dysintegrity. Depending on the severity and type of injury, the BBB becomes more permeable, allowing cellular degradation products and other molecules to enter more freely into the systemic circulation. The endothelial cells of the BBB, which contain specialized systems for the transport of specific molecules, are connected by tight junctions involving proteins such as claudin-5. Indeed, injury-induced increases in CSF proteins, such as claudin-5, indicate a disruption of the endothelial tight junction and a potential opening of the BBB (67). An intact BBB impedes CNS protein diffusion into blood; dilution of brain-specific proteins in the large volume of blood and extracellular fluid inconsistently affects the low concentrations in blood (68). Additionally, there is rapid degradation by blood proteases and removal by hepatic metabolism and renal clearance (69).

Another mechanism by which relevant TBI molecules may be cleared is the lymphatic drainage system (70,71). This system, which is driven by the recently discovered dural venous sinus lymphatics (72), may also be related to an even more important CNS-clearance mechanism, which has been termed the glial lymphatic or ‘glymphatic’ system (73). This glymphatic system comprises the bulk flow of solutes from CNS arterioles, across the interstitial space via astroglial aquaporin-4 channels, and ultimately into venous circulation. It has been demonstrated in animal models of TBI that alteration of the localization of astroglial aquaporin-4 channels is an early post-TBI event (74). Indeed, the glymphatic system may modify the concentration of blood-based biomarkers in ways that function separately from TBI itself and may thereby impede the delivery of brain injury markers to peripheral blood (75).

An important unknown in the context of blood biomarkers is the effect of pressure gradients on the movement of proteins and protein fragments from the CSF into the peripheral circulation (68,76). More severe TBIs are known to alter intracranial, arterial, and venous pressure, yet the precise mechanisms of the numerous highly complex and well-regulated intracranial pressure (ICP) gradients are not well understood. It is likely that in cases in which TBI alters the balances between various pressure gradients, the clearance of biomarkers from the brain interstitium into the systemic circulation is also altered (77,78).

Finally, most, if not all, CNS-derived markers are in extremely low concentration when measured in plasma or serum. This of course poses additional challenges for the use of blood biomarkers in TBI. Together, these limitations mean that despite an investment of hundreds of millions of dollars, there still are no validated blood-based biochemical markers for mTBI, which accounts for ~85% of all individuals with TBI. Similarly, there are no blood biochemical markers for the chronic postconcussive state that affects millions of persons in the USA alone.

Cerebrospinal fluid biomarkers in TBI in humans and animals

Given the poor performance of blood biomarkers in the detection of acute mTBI and abnormalities in the chronic postconcussive state, as well as their lack of CNS specificity, investigators are beginning to look to the example of Alzheimer’s disease (AD) biomarker research. An easily accessed biofluid, like blood, would be ideal for clinicians and researchers, but in AD, the long search for such a blood biomarker has been unsuccessful. In contrast, the utility of cerebrospinal fluid (CSF) Aβ₄₂, total tau, and phosphorylated tau for the clinical diagnosis of AD, the prediction of progression from mild cognitive impairment to AD, and the diagnosis of preclinical AD is well documented (79). For example, a recent study demonstrated the superiority of CSF biomarkers to plasma biomarkers in AD: proteomic analysis of serum from nearly 400 participants from the Texas Alzheimer’s Research Consortium and the Alzheimer’s Disease Neuroimaging Initiative used random forest analysis to show that CSF total tau/Aβ₄₂ ratio was far superior in its sensitivity and specificity for identifying AD than a combination of 11 plasma proteins (tau was not one of the identified proteins) (80).

CSF biomarkers have also been explored in the determination of the temporal changes of mTBI and the chronic post-TBI state. Neselius et al. conducted a 36-week longitudinal case study of CSF and serum biomarkers in a boxer following a knockout concussion with samples collected at 2, 9, 18, 28, and 36 weeks (81). CSF neurofilament light chain protein (NF-L), total tau, phosphorylated tau, GFAP, and Aβ were analysed at each time point and then compared to the results from a normal participant. The CSF:serum albumin ratio remained unaffected in the boxer, indicating that the BBB was intact. Of the five CSF biomarkers measured, NF-L levels were increased 10-fold over control values at 2 weeks post-concussion and remained elevated up to 28 weeks post-injury, indicating ongoing, subacute axonal injury processes. However, despite this evidence of underlying pathobiology, the athlete was symptom-free after one week, which according to typical protocols would have meant that he would have been deemed fit to return to the boxing ring. Neselius et al. performed another study in 30 amateur boxers; 80% of the boxers had elevated CSF levels of NF-L 1–6 days post-bout, even without knock-outs, whereas extensive neurobehavioral testing showed no abnormalities (82). These findings strongly argue for the serial collection of CSF samples following concussion, as well as for the consideration of CSF biomarkers in safe return-to-play guidelines.

Another example of the role that CSF biomarkers could play in monitoring TBI is demonstrated by a study of blast-induced TBI in swine. In that study, CSF biomarker levels of neuronal and glial injury/loss and vascular pathologies changed substantially over time (19). CSF NF-H, NSE, GFAP, claudin-5, and vascular endothelial growth factor displayed biphasic temporal profiles with an initial peak at 6 hours post-injury, likely reflecting the primary injury process. NF-H and NSE values then returned to normal pre-injury levels at 24 hours but peaked again at 72 hours and remained highly elevated at 2 weeks post-injury. Claudin-5 and vascular endothelial growth factor, which are markers of vascular integrity and vascular injury, were also elevated 72 hours post-injury, reflecting the secondary injury process that likely involves vascular pathologies which contribute to the ongoing neuronal and glial damage/loss. Analysis of simultaneously collected serum biomarkers (also referred to above) showed a different temporal profile: serum NF-H, NSE, and S100β were elevated at 24 hours post-injury compared to the baseline preinjury levels but had returned to normal at 2 weeks (32). These examples underline the importance of both serial sampling and the use of CSF as the preferred biofluid for monitoring injury-induced biofluid biomarker changes.

Cerebral microdialysis biomarker studies in severe TBI in humans

Cerebral microdialysis (cMD) is a process in which brain extracellular fluid (bECF) is collected (83,84). This technique, which was first developed by Ungerstedt, requires pumping perfusion fluid at a very low rate through a long catheter that is covered by a porous membrane at its distal end (84). Molecules in the extracellular space then pass through the membrane and diffuse into the catheter. The method was originally used to monitor changes in the extracellular concentration of small molecules, including lactate, pyruvate, glutamate, and glucose. For TBI this is relevant because extracellular levels of these molecules reflect the metabolic state of the injured brain, which is especially important in monitoring brain metabolism following trauma (85,86). Moreover, with the introduction of larger pore-sized membranes—up to 100 kD—it is now possible to sample the brain for injury-induced changes in the concentration of larger molecules, including proteins (86).

Serial sampling and analyses of bECF can thus identify the temporal pattern of pathobiologies and the response to interventions. cMD studies have shown dramatic time-dependent changes in the levels of biomarkers, including glycerol, glucose, total tau, and Aβ (87), and multiple cMD studies have demonstrated the dynamic nature of pathobiological processes following TBI. For instance, in studies of patients with severe TBI who are in the acute phase of the injury, researchers have used cMD to detect substantial time-dependent changes in the CSF concentrations of inflammatory markers and alpha-II-Spectrin breakdown products (88,89).

Although this method provides the ideal biofluid for the identification of intraparenchymal pathologies following TBI, there are several factors that limit the usefulness of cMD as a TBI biomarker. Most importantly, cMD can only be performed in neurointensive care units that are equipped to monitor acute severe TBI (86). More particularly, in some forms of severe TBI, multiple catheters are necessary to perform cMD, and there is a limitation in the number of catheters that can be inserted. Moreover, it is unclear how local changes may translate into global changes that affect the entire cerebrum. The usefulness of cMD is also dependent on the type of analytes measured. Small molecules (e.g., glucose, lactose, pyruvate) can diffuse faster in the extracellular space, and thus sampling from several catheters can provide a reasonably clear understanding of the metabolic state of the brain. However, larger molecules (e.g., proteins) diffuse much slower or not at all, and accordingly, their local concentrations can vary greatly (90).

In summary, the current clinical use of bECF as a valuable biomaterial for biomarker analysis highlights the dynamic nature of TBI as well as the potential of non-blood biomarkers to offer important clinical applications in the management of acute, severe TBI. However, important questions remain concerning the ways in which injury-induced changes in the bECF levels of biomarkers may be reflected in the CSF levels of these same biomarkers.

Clinical and research gaps and recommendations

About three-quarters of individuals who experience a single mild-impact TBI recover completely over time; the remaining one-quarter recover either very slowly or never fully recover (91–93). Athletes who experience multiple impact mTBIs and active-duty service members who experience multiple blast and/or impact mTBIs are even less likely to recover fully (94–96). The identification of biofluid biomarkers could play a vital role in identifying underlying individual pathobiological changes and could thereby assist in early diagnosis and treatment of TBI. This is particularly important for determining whether it is safe for individuals to return to duty or play.

Although progress has been made in the identification of a handful of blood biomarkers that are well-replicated in acute and more severe TBI, inadequate attention has been directed towards the dynamic alterations that occur during the hours, days, and weeks of the acute and subacute stage. There remains no convincing evidence of the utility of plasma or serum biomarkers for either acute mTBI or for detecting neurodegeneration in the chronic postconcussive state. Elucidation of these dynamic changes could both provide a better understanding of the pathobiologies that emerge from acute TBI and inform clinicians about the appropriate windows of opportunity for therapeutic interventions.

To that end, CNS-derived plasma exosomes show promise for identifying blood-based biomarkers that are more specific to CNS pathobiology (97,98). Exosomes are stable cell-derived nanovesicles that mirror the features of the parent cell, cross the BBB, and are present in all biological fluids, including blood, CSF, saliva, and urine. Given the rapid advancement of biochemical tools, it is becoming increasingly feasible to isolate plasma exosomes of CNS derivation by targeting CNS-specific cellular markers, such as the neural adhesion protein. By utilizing cell-specific noncoding RNA, it may even be possible to identify plasma exosomes that are from neurons versus glia. Although these efforts remain a work in progress, ultra-high sensitivity platforms have increased the promise of detecting CNS-derived exosomes in plasma that may serve as better biomarkers of TBI.

As researchers begin to recognize the importance of finding TBI biofluid biomarkers that can reflect time-dependent changes, including in their ongoing work with CNS-derived plasma exosomes, the half-life of those protein biomarkers in blood and/or CSF should also be considered. Indeed, both medications and medical comorbidities, such as polytraumatic organ and muscle injury, may modify the half-life of protein biomarkers, but researchers currently lack the ability to account for these potentially confounding factors. Moreover, the potential pitfall of biomarker half-life is further highlighted by the fact that the overwhelming majority of assays are antibody based and by the finding that cleavage or degradation which affects the epitope can also confound the actual concentrations of protein biomarkers in the serum and/or CSF.

Finally, researchers should abandon single time-point assays because the information value of such snapshots, which may not even be acquired at identical post-injury time-points, is extremely limited. Rather, long-term longitudinal sampling that can better evaluate time-dependent changes in CSF biomarkers from individuals in the acute and subacute phase of TBI should be pursued. Such efforts will be especially helpful for the identification of individuals who are likely to have slow or poor recovery. In repetitive mTBI, particularly in athletes, active-duty service members, and Veterans, long-term longitudinal sampling of CSF biomarkers will be essential to determine the risk of developing progressive neurodegenerative diseases such as CTE. Longitudinal studies of CSF biomarkers have the potential to assist with differential diagnosis, to identify chronic brain pathology in persons with persistent postconcussive symptoms, and to detect the presence of progressive neurodegeneration. Such longitudinal studies have the potential to inform therapeutic targets for the amelioration of chronic postconcussive symptoms, monitor the usefulness of future therapeutic agents, and, ultimately, to prevent neurodegeneration and dementia. With respect to the current state of the art in biofluid biomarkers for TBI, serial brain microdialysis measurements comprise the best approach to using biomarkers in acute, severe TBI when in the setting of a neurointensive care unit. For the acute and chronic effects of single and repetitive mild TBI, serial collections of CSF are the most useful approach in both clinical and research settings. However, the search for informative blood biomarkers remains elusive.

Declaration of Interest statement

The authors report no declarations of interest.

Acknowledgments

This material is the result of work supported with resources and the use of facilities at the VA Puget Sound Health Care System.