Association of cerebrospinal fluid protein biomarkers with outcomes in patients with traumatic and non-traumatic acute brain injury: systematic review of the literature

Background Acute brain injuries are associated with high mortality rates and poor long-term functional outcomes. Measurement of cerebrospinal fluid (CSF) biomarkers in patients with acute brain injuries may help elucidate some of the pathophysiological pathways involved in the prognosis of these patients. Methods We performed a systematic search and descriptive review using the MEDLINE database and the PubMed interface from inception up to June 29, 2021, to retrieve observational studies in which the relationship between CSF concentrations of protein biomarkers and neurological outcomes was reported in patients with acute brain injury [traumatic brain injury, subarachnoid hemorrhage, acute ischemic stroke, status epilepticus or post-cardiac arrest]. We classified the studies according to whether or not biomarker concentrations were associated with neurological outcomes. The methodological quality of the studies was evaluated using the Newcastle–Ottawa quality assessment scale. Results Of the 39 studies that met our criteria, 30 reported that the biomarker concentration was associated with neurological outcome and 9 reported no association. In TBI, increased extracellular concentrations of biomarkers related to neuronal cytoskeletal disruption, apoptosis and inflammation were associated with the severity of acute brain injury, early mortality and worse long-term functional outcome. Reduced concentrations of protein biomarkers related to impaired redox function were associated with increased risk of neurological deficit. In non-traumatic acute brain injury, concentrations of CSF protein biomarkers related to dysregulated inflammation and apoptosis were associated with a greater risk of vasospasm and a larger volume of brain ischemia. There was a high risk of bias across the studies. Conclusion In patients with acute brain injury, altered CSF concentrations of protein biomarkers related to cytoskeletal damage, inflammation, apoptosis and oxidative stress may be predictive of worse neurological outcomes.

brain and are usually considered non-reversible. Secondary brain injuries arise from insults to the brain parenchyma that occur after the initial injury (e.g., as a result of hypoxemia and/or hypotension) and increase the overall area of damaged brain tissue [2,3]. After an acute brain injury, intrathecal expression of proteins related to brain inflammation, apoptosis and oxidative stress induces production and migration of chemotactic factors, which ultimately lead to blood-brain barrier (BBB) dysfunction, brain edema formation and intracranial hypertension [4]. This cellular response may render the brain more susceptible to secondary injuries in cases of decreased cerebral perfusion pressure and may increase the volume of nonviable tissue.
In humans, the cerebrospinal fluid (CSF) acts as a highly specific repository of cellular by-products, neurotransmitters and protein fragments as it is in close contact with the brain parenchyma and other products of neural origin [5]. Concentrations of protein biomarkers in the intrathecal space may therefore reflect the presence or severity of primary and/or secondary brain injuries. For example, in patients with traumatic brain injury (TBI), increased CSF concentrations of protein biomarkers from damaged neurons may serve as indicators of ongoing cellular damage [6], and, in patients with subarachnoid hemorrhage (SAH), higher concentrations of CSF protein biomarkers may be associated with increased risk of vasospasm and delayed cerebral ischemia [7]. CSF protein biomarkers may reflect the pathophysiological pathways involved in acute brain injuries that could be susceptible to interventions, and thus help in the development of therapies or to guide earlier intervention to improve long-term functional outcomes.
We therefore performed a systematic review to identify observational studies that have evaluated the relationship between CSF protein biomarkers in patients with acute brain injuries and neurological outcomes.

Data sources
Following protocol submission to the Prospero International Prospective Register of Systematic Reviews (ID 114294), we conducted a systematic search of the literature using the MEDLINE database and the PubMed interface from inception until June 29, 2021, to identify all observational studies that evaluated CSF protein biomarkers (proteins were defined as those with at least 50 amino acids or a molecular weight greater than 4000 Da) in patients with severe acute brain injury (as a result of TBI, SAH, acute ischemic stroke, status epilepticus or post-cardiac arrest syndrome) and that reported any neurological outcome. We used the We also searched the references of included articles for studies that had been missed in the initial search. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [8].

Study selection and data abstraction
Three of the authors (AB, LARG and CDC) performed the literature search and selected the studies. We excluded studies on descriptive proteomics; those evaluating metabolites (e.g., lactate, lactate/pyruvate, glucose, glutamate, glycerol, etc.), hormones or cytokines/ chemokines; those in patients with chronic degenerative or chronic traumatic injuries (e.g., multiple sclerosis, Alzheimer and Parkinson diseases, sport-related injuries, chronic traumatic encephalopathy); those in patients with autoimmune conditions (e.g., Guillain-Barré); pediatric studies; postmortem populations; studies with only physiological outcomes; and animal studies. Data abstraction regarding type of acute brain injury, source of CSF (ventricular or lumbar), number of included subjects, method used by the author to quantify the specific biomarker and neurological outcomes was performed by the same three reviewers (AB, LARG and CDC) in an independent blinded manner by completing predefined tables. Studies were classified according to whether or not the measured biomarker was associated with neurological outcome (as defined in the original study) and were grouped according to whether the brain injury was traumatic or non-traumatic. The methodological quality of the observational studies was evaluated using the Newcastle-Ottawa quality assessment scale [9]. Discrepancies in the assessment of methodologic quality and final classification of the selected studies were resolved by the involvement of a fourth author (CAS).

Observational trials reporting biomarker associations with neurological outcome
Of the 30 trials that reported a biomarker associated with outcome, 18 included patients with TBI (n = 1345), 6 included patients with SAH (n = 258), 5 included patients with acute ischemic stroke (n = 422), and one included a mixed population (TBI and SAH) (n = 102). The main biological functions reflected by the biomarkers were related to primary brain injury (neuron cell cytoskeleton) and secondary brain injury, e.g., increased apoptosis, inflammation and energy metabolism, reduced redox response to oxidative stress and increased neurodegeneration. Specifically, concentrations of the CSF biomarkers ubiquitin carboxy-terminal hydrolase L1 (UCH-L1), microtubule-associated protein (MAP)-2, alpha-synuclein and peroxiredoxin VI were associated with a lower Glasgow Coma Scale (GCS) score on admission, worse long-term functional outcome and increased mortality. In patients with SAH, NLRP1, ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain), caspase-1 and 3, α-2 spectrin and SBDP (spectrin breakdown products), apolipoprotein-E, S-100β, H-FABP (heart-type fatty acid binding protein) and tau protein were associated with an increased risk of vasospasm, late cerebral ischemia and worse functional outcome at 3-6 months. These findings were consistent when the CSF was collected from a mixed cohort of TBI and SAH patients. In patients with acute ischemic stroke, proteins related to cytoskeleton disruption and energy metabolism were consistently associated with the size of brain infarction and clinical status (see Table 1).

Observational trials reporting no association of the biomarker with neurological outcome
Of the 9 trials that reported no association of the biomarker with neurological outcome [39][40][41][42][43][44][45][46][47], 8 included patients with TBI (n = 254) and 1 had a mixed population of patients with hemorrhagic or ischemic stroke (n = 51). The main biological functions assessed by the studied biomarkers included inflammation, neuronal cytoskeleton components, apoptosis, energy metabolism and neurodegeneration (Table 2).

Methodological analysis
The risk of bias among the included studies was high according to the Newcastle-Ottawa scale [9] (Tables 1  and 2). In addition, different CSF sources were used for assessment of protein biomarker concentrations (ventricular CSF, lumbar CSF, serum, biobanks) across different studies and most control group patients also had neurological conditions that may have influenced biomarker concentrations (e.g., normal pressure hydrocephalus). The studies of patients with acute ischemic stroke were the only ones in which the source of CSF was always the same in the intervention and the control group (lumbar CSF).

Discussion
Our results suggest that CSF concentrations of protein biomarkers associated with the pathophysiological pathways involved in acute brain injuries may be predictive of increased morbidity and mortality after traumatic and non-traumatic acute brain injury. CSF proteomic expression may be altered by many factors including genetic background, the severity of the primary brain injury and secondary insults, such as hypoxemia and hypotension [48,49]. In patients with a traumatic origin of the acute brain injury, cytoskeletal damage was associated with an increased risk of cerebral hemorrhage, intracranial hypertension and early mortality rates, suggesting severe primary brain injuries. After the initial phase of acute brain injury, the expression of proteins involved in re-establishing normal homeostasis is altered [50]. If this response is dysregulated, it may overwhelm counter-regulatory measures initiated by the body to reduce tissue injury, increasing the risk of secondary brain injuries [51]. Moreover, impairment of normal biological functions (e.g., redox function capability, dysregulated inflammation, increased apoptosis) after a primary acute brain injury may render the brain more susceptible to secondary injuries. This seems to be the case in patients with SAH in whom CSF concentrations of C-reactive protein [31], α-2 spectrin and SBDP [33], apolipoprotein E [7], H-FABP and tau protein [23] were associated with an elevated risk of vasospasm and delayed cerebral ischemia. Interestingly, in a mixed population of patients with traumatic and non-traumatic acute brain injuries, concentrations of the structural protein S-100β were higher in patients with lower Glasgow Outcome Scale (GOS) scores [14], suggesting a common pathophysiological pathway for these two types of injury.
Consequences such as acute brain edema, vasospasm or non-convulsive status epilepticus are of crucial importance in patients with acute brain injury because they may affect long-and short-term outcomes. Jha et al. [39] evaluated the ability of the protein biomarker sulfonylurea receptor-1 (Sur1) to predict the risk of cerebral edema in patients with severe TBI. Patients with evidence of edema on computed tomography (CT) had higher concentrations of Sur1 with statistically significant differences in mean (p = 0.023) and peak (p = 0.019) concentrations in patients with and without edema. Although there were no differences in functional outcome, as assessed using the 3-month GOS score, in patients with higher Sur1 concentrations, prediction of cerebral edema may indicate the need for more aggressive therapeutic measures.
It is difficult to imagine that a single biomarker could explain the complex cascades of events following acute brain injury that may be related to worse long-term outcomes. A single CSF protein biomarker may indicate derangement of a specific biological function but may not be involved in other pathophysiological pathways. Moreover, the time point at which the biomarker is measured may reflect different stages of acute brain injury (e.g., primary vs secondary injury). Thus, earlier sampling of CSF biomarkers after initial injury may provide information about the severity of the initial injury (e.g., increased risk of early mortality, extent of brain tissue involvement, risk of severe intracranial pressure), whereas more delayed measurements could provide information on risk of chronic degenerative encephalopathy or longer-term outcomes. This could be an interesting area for future study.
Our review has several limitations. First, the search strategy was based solely on the MEDLINE database, and more studies may have been identified if other databases (e.g., Embase) had been used. Second, because of insufficient data we could only provide descriptive data. We were unable to determine which protein biomarker was most associated with worse short-or long-term outcomes. Also, there was a high risk of bias among the included studies because of trials without a control group, a control group with CSF-derived from patients with other neurological conditions (e.g., with normal pressure hydrocephalus) or studies comparing lumbar and ventricular CSF without taking into account the craniocaudal gradient [52]. Finally, some studies used frozen biobank samples, which may have lower protein concentrations because of proteolysis induced by freezethaw and contamination. Future studies should report in a more standardized fashion to enable comparison across different studies.

Conclusions
Changes to the CSF proteome in patients with acute brain injury reflecting the pathophysiological pathways involved may be indicative of the severity of the injury and predictive of worse neurological outcomes. However, there are currently insufficient data available to recommend the routine measurement of any CSF biomarker in these patients.