Excitatory amino acid inhibitors in adults with acute moderate to severe traumatic brain injury: a systematic review and meta-analysis

Abstract

Objective

To evaluate the effectiveness of early administration of excitatory amino acid (EAA) inhibitors on long-term neurological outcomes in adults with acute moderate to severe traumatic brain injury (TBI).

Design

Systematic review and meta-analysis of randomised controlled trials (RCTs).

Data sources

MEDLINE, EMBASE, Cochrane CENTRAL, Web of Science, WHO International Clinical Trials Registry Platform and ClinicalTrials.gov from inception to January 2026.

Study selection

RCTs comparing EAA inhibitors with placebo, standard care or any other interventions were included. Trials enrolled adult patients (≥18 years) with moderate to severe TBI (Glasgow Coma Scale score ≤12) receiving the intervention within the acute phase of care (first week).

Data extraction and synthesis

Pairs of reviewers independently screened trials, extracted data, assessed the risk of bias (RoB) with the Cochrane RoB tool 2 and graded the certainty of evidence using the Grades of Recommendation, Assessment, Development and Evaluation approach. Random effects models were used for all effect measures and trial sequential analyses (TSA) were performed for each outcome.

Main outcome measures

The primary outcome was long-term neurological function at 6 months (or the nearest earlier time point), assessed with the Glasgow Outcome Scale (GOS) or extended version (GOS-E), using the classical definitions of an unfavourable outcome (GOS 1–3 or GOS-E 1–4).

Results

28 trials enrolling 4238 patients were included. Early administration of EAA inhibitors was not associated with reduced unfavourable neurological outcomes (relative risk 0.93 (95% CI (0.84 to 1.03); I²=40%; 15 trials, n=3613, moderate certainty). No statistically significant difference was observed based on EAA inhibitor type, timing or duration of administration, RoB or TBI severity. Mortality, intensive care unit lengths of stay and mean intracranial pressure were not statistically different between groups, but hospital length of stay was reduced in the EAA inhibitors group. The early use of EAA inhibitors was not associated with adverse events (low certainty). TSA showed insufficient power for the primary outcome.

Conclusions

In adults with moderate to severe TBI, the early administration of EAA inhibitors was not associated with a reduction of unfavourable neurological outcomes. Further high-quality and adequately powered RCTs are required to clarify their role in TBI management.

PROSPERO registration number

CRD42025635527.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Our systematic review and meta-analysis included 28 randomised controlled trials enrolling more than 4000 patients, providing the most comprehensive synthesis on this topic.

• Our review was conducted following methodological guidance from the Cochrane Handbook and was reported in compliance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses 2020 statement.

• A comprehensive search strategy was used across multiple databases without language or date restrictions.

• Trial sequential analyses were used to evaluate the conclusiveness and robustness of the cumulative evidence.

• Heterogeneity in Glasgow Outcome Scale (GOS)/GOS extended version reporting, particularly when treated as continuous variables, limited their inclusion in meta-analyses.

Introduction

Traumatic brain injury (TBI) is a leading cause of mortality and long-term disability worldwide.¹ In the immediate aftermath of TBI, excitotoxicity is a key mechanism of brain damage, triggered by an early surge in excitatory amino acid (EAA) neurotransmitters, primarily glutamate, which drives neuronal injury.^{2 3} This early process involves overstimulation of both ionotropic (iGluRs) and metabotropic (mGluRs) glutamate receptors, resulting in excessive calcium influx and subsequent neuronal death.⁴ Among iGluRs, overstimulation of N-methyl-D-aspartate (NMDA) receptors plays a critical role in disrupting cellular homeostasis and triggering neurotoxic cascades,⁵ while overstimulation of mGluRs contributes to damage by modulating intracellular signalling pathways.⁶

Given the association between excitotoxicity and poor neurological outcomes,² inhibiting the EAA pathway is considered a promising therapeutic approach in TBI. Nevertheless, its efficacy remains controversial due to mixed evidence,⁷ and these agents are currently not recommended by guidelines. Several strategies to inhibit the glutamate pathway have been explored, from NMDA receptor antagonists like ketamine⁸ to agents like magnesium sulfate, which inhibits presynaptic excitatory neurotransmitter release and blocks voltage-gated calcium channels.⁹,¹¹ A Cochrane systematic review published in 2003 showed no beneficial effect of EAA inhibitors on neurological outcome with only two trials included.¹² The review was conducted in a mixed population of mild to severe TBI, further limiting the strength of its conclusions.¹² More recent reviews have focused on specific EAA inhibitors rather than on the class of drugs itself.¹³,¹⁶ In recent years, interest in EAA inhibition for TBI patients has grown, especially in more severe forms, driven by renewed focus on ketamine and its derivatives.^{17 18} Despite promising preclinical results,¹⁹ the impact of EAA inhibitors on neurological recovery in adult TBI patients still remains uncertain. This systematic review and meta-analysis of randomised controlled trials (RCTs) aims to evaluate whether the early administration of EAA inhibitors improves long-term neurological outcomes in adults with acute moderate to severe TBI.

Methods

Study design

Using a prior published protocol (PROSPERO #CRD42025635527), we conducted our systematic review using methodological approaches outlined in the Cochrane Handbook for Systematic Reviews of Interventions.²⁰ We reported our results according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses criteria.²¹

Eligibility criteria

We included RCTs comparing EAA inhibitors (eg, iGluR or mGluR blockers, or treatments inhibiting excitatory neurotransmission) with placebo, standard of care or any other intervention. We included adults (≥18 years) with acute moderate to severe TBI, defined by a Glasgow Coma Scale (GCS) score of 3–12 (full scale range, 3–15; lower scores indicate greater impairment of consciousness).²² To be eligible, trials must have evaluated the administration of EAA inhibitors initiated during the acute phase of care (first week after injury) and assessed at least one of our outcomes of interest. In the case of mixed populations, trials were considered eligible if at least 80% of the study population were adults with moderate or severe TBI. All routes of administration were considered. We applied no date or language restrictions.

Outcomes

Our primary outcome was long-term neurological function, assessed using the Glasgow Outcome Scale (GOS) or its extended version (GOS-E) at 6 months.^{23 24} The nearest earlier time point was considered when the assessment was not provided at 6 months. The GOS-E scale is an 8-point ordinal scale ranging from 1 (death) to 8 (upper good recovery) with intermediate levels including neurovegetative state, lower and upper severe disability, lower and upper moderate disability, and lower good recovery. We dichotomised the outcome into ‘unfavourable’ (GOS 1–3; GOS-E 1–4) and ‘favourable’ (GOS 4–5; GOS-E 5–8) neurological outcomes which are the most commonly used definitions.^{25 26}

Our secondary outcome measures included all-cause mortality, length of stay in the intensive care unit (ICU) and hospital, quality of life (assessed using any scale), mean intracranial pressure (ICP) and adverse events.

Search strategy

We searched MEDLINE (Ovid), EMBASE (Ovid), Cochrane Central Register of Controlled Trials (CENTRAL), Web of Science, WHO International Clinical Trials Registry Platform and ClinicalTrials.gov database from their inception to 3 January 2026. The search strategy was developed with an information specialist and validated according to the Peer Review of Electronic Search Strategies 2015 guidelines.²⁷ Full electronic search strategies for all databases and trial registers, with all applied filters, are detailed in online supplemental appendix 1.

Study selection and data extraction

We managed citations and removed duplicates with Zotero (Corporation for Digital Scholarship, Virginia, Zotero 6) and Covidence (Cochrane, Melbourne, 2020 version). Pairs of reviewers (MB, MV, THT, AO and M-AG) independently screened citations for eligibility and extracted data, with disagreements resolved by a third reviewer (AFT). Authors were contacted for missing information. Extracted trial characteristics included author, year, duration, country and sample size. Participant data included demographics, TBI severity, clinical data (eg, GCS at admission), type of injury. Intervention details included drug name, dosing, timing and duration. Outcome data (GOS/GOS-E), ICU and hospital lengths of stay, ICP, quality of life and adverse events, follow-up duration) were recorded. Data from diagrams or graphs were extracted using WebPlotDigitizer V.4.1.²⁸

Risk of bias assessment

Pairs of reviewers (MB, MV, THT, AO and M-AG) independently assessed the risk of bias for each of the included trials and each outcome using the Cochrane Risk of Bias 2 tool.²⁹ A third reviewer was consulted in case of disagreement (AFT). This tool consists of five domains representing potential sources of bias, which reviewers categorised as ‘high’, ‘low’ or ‘some concerns’. We assessed the risk of bias of each trial based on the worst assessment across the five domains.

Statistical analysis plan

For continuous data (ICU and hospital length of stay, ICP, quality of life), we extracted means and SDs and when necessary, medians and ranges were converted into means and SDs using standardised formulas.³⁰ Continuous outcomes were reported as mean differences. For dichotomous data (neurological outcomes, mortality), we collected the number of events and total participants in each group and expressed the results as risk ratios (RRs) with 95% CIs. We performed analyses according to the Der Simonian & Laird random-effects models using the inverse variance method. Summary effect measures were calculated using Review Manager Web, V.8.11.0 (RevMan, The Cochrane Collaboration, Oxford, UK). Statistical heterogeneity was assessed with the I² index.³¹ Values greater than 50% were considered as substantial heterogeneity. We evaluated the presence of publication bias using funnel plots for outcomes reported in more than ten trials.²⁰ To test the robustness of our findings and explore potential sources of heterogeneity, we performed prespecified subgroup analyses.³² These included the type of EAA inhibitor (iGluR, mGluR, others), TBI severity (moderate vs severe vs moderate and severe), risk of bias (low vs high/some concerns); timing of intervention (≤ or >24 hours after TBI), and duration of intervention (≤ or >24 hours).

Strength of evidence

We evaluated the strength of evidence for each outcome according to the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) working group statement.³³ The GRADE approach adjudicates the strength of the evidence on a continuum from high, moderate, low or very low for each outcome based on a structured approach. The grading was performed in duplicate by two independent reviewers (MB and THT). To determine whether the meta-analyses sample sizes reach the required information size, we conducted a trial sequential analysis (TSA) for each outcome. This analysis was performed using TSA software V.0.9.5.10 Beta (Copenhagen Trial Unit, Center for Clinical Intervention Research, Copenhagen, Denmark, 2011).³⁴ The data for each outcome were used to calculate the required information size and apply the O’Brien-Fleming alpha-spending boundaries function to compute the cumulative Z-score. The calculations were conducted with a two-tailed test, assuming an alpha level of 5% and a power of 80%.

Results

Our search yielded 15 762 citations. After full-text screening of 278 articles, 28 trials that enrolled 4238 patients were included (figure 1). Four trials were reported only as conference abstracts.³⁵,³⁸ Data from a large unpublished trial³⁹ from the 1990s were included based on primary outcome data published in a book chapter,⁴⁰ verified with the authors, and incorporated into the IMPACT data set.⁴¹ Data from two other large trials could not be included because they were unpublished and inaccessible (O Ianotti, unpublished data).⁴² Characteristics of included trials are detailed in online supplemental eTable 1 and those of the 18 ongoing trials are listed in online supplemental eTable 2.

Figure 1. Study flow diagram.

Characteristics of trials

Of the 28 trials included, three trials involved multiple countries,^{36 43 44} five were conducted in the USA,^{3739 45},⁴⁷ three in France,⁴⁸,⁵⁰ seven in Iran,⁵¹,⁵⁷ three in China,⁵⁸,⁶⁰ three in India,^{35 61 62} two in Egypt^{63 64} and one in Israel.⁶⁵ Nine different EAA inhibitors were evaluated, including eight iGluR antagonists (amantadine,³⁵⁵¹,54 57 62 63 aptiganel,^{36 39} dexanabinol,^{43 65} gacyclidine,⁴⁸ ketamine,^{49 50} memantine,⁵⁵ selfotel (CGS 19755)^{37 44} and traxoprodil⁴⁵) and magnesium sulfate.⁹,1138 46 47 56 58 Seven trials assessed EAA inhibitors in a mixed severity population (moderate and severe, n=1688),^{38 43 47 51 54 55 64} while 19 (n=2482)³⁵,^{3739 44} focused on severe TBI and one on moderate TBI.⁶³ One trial was classified as conducted in ‘mixed severity of TBI’ because it was impossible to rule out the inclusion of moderate TBI, despite being labelled as severe by the authors.⁴³ The EAA inhibitors were initiated beyond 24 hours in only three trials (n=117).^{49 52 63} Most trials (20 trials, n=3831) used a placebo as the control, five used the standard of care^{35 51 58 59 61} (n=258), two trials^{49 50} compared ketamine with sufentanil, one⁴⁶ compared magnesium sulfate combined with hypothermia to hypothermia alone, one compared ketamine to propofol.³⁵

Risk of bias

The risk of bias assessment for the main outcome of each trial is reported in online supplemental eTable 3. The risk of bias for the primary outcome was low in eight trials^{43 45 47 51 52 55 62 64} and high in five trials,^{35 46 54 61 65} primarily due to missing data and selective reporting of results. The risks of bias for secondary outcomes and the overall risk of bias by domain are detailed in online supplemental eFigures 1 and 2. We found no evidence of publication bias for either the primary outcome or mortality based on visual inspection of the funnel plot (online supplemental material eFigures 3 and 4).

Primary outcome

15 trials (n=3613)³⁹⁴³,45 47 48 50 51 58 were included in the meta-analysis of the primary outcome. Functional neurological outcome was assessed at 6 months in 8 trials (n=3016),³⁹⁴³,45 47 50 59 65 at 3 months in 6 trials (n=537),^{4851 58 60},⁶² and at 2 months in one trial (n=60)⁶⁴ (figure 2). An unfavourable neurological outcome occurred in 46.7% of patients in the EAA inhibitors group compared with 48.3% in the control group (RR 0.93, 95% CI 0.84 to 1.03), I²=40%) (figure 2) (moderate certainty, table 1). No statistically significant differences were observed across subgroup analyses (figure 3, online supplemental material eFigure 5). With an estimated relative risk reduction of 10%, the TSA showed that the required information size to draw definitive conclusions would be 7534 patients, suggesting that the pooled analysis of available trials is likely prone to a type II error (online supplemental material eFigure 6). Six trials⁴⁶⁵²,^{55 57} (n=270) reported neurological outcome as continuous variables rather than ordinal scales, and thus could not be included in the meta-analyses. None of these trials reported a statistically significant difference in neurological outcomes between the intervention and control groups.

Figure 2. Meta-analysis for unfavourable neurological outcome. Unfavourable neurological outcome was defined by Glasgow Outcome Scale (GOS) from 1 to 3 or GOS-Extended (GOS-E) from 1 to 4. ¹GOS at 6 months. ²GOS-Extended at 3 months. ³GOS at 3 months. ⁴GOS-Extended at 6 months. ⁵GOS at 2 months. EAA, excitatory amino acid.

Table 1. Summary of findings.

Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with control group	Risk with EAA inhibitors
Unfavourable neurological outcome Follow-up: 6 months (or nearest available)	48 per 100	45 per 100 (41 to 50)	RR 0.93 (0.84 to 1.03)	3613 (15 RCTs)	⨁⨁⨁◯ Moderate†	EAA inhibitors probably result in little to no difference in unfavourable neurological outcome.
Mortality Follow-up: 6 months (or nearest available)	19 per 100	17 per 100 (14 to 21)	RR 0.90 (0.73 to 1.10)	3427 (20 RCTs)	⨁⨁⨁◯ Moderate†	EAA inhibitors probably result in little to no difference in mortality.
Mean intracranial pressure mm Hg	The mean intracranial pressure was 16.7 mm Hg	MD 0.45 mm Hg lower (1.67 lower to 0.77 higher)	–	924 (6 RCTs)	⨁⨁⨁◯ Moderate‡	EAA inhibitors probably result in little to no difference in mean intracranial pressure.
Mean intensive care unit length of stay days	The mean intensive care unit length of stay was 16 days	MD 2.47 days lower (5.53 lower to 0.59 higher)	–	187 (2 RCTs)	⨁⨁◯ Low†,§	EAA inhibitors may not reduce mean intensive care unit length of stay.
Mean hospital length of stay days	The mean hospital length of stay was 23.2 days	MD 2.86 days lower (5.58 lower to 0.15 lower)	–	169 (3 RCTs)	⨁⨁◯ Low†§	EAA inhibitors may not reduce mean hospital length of stay.
Quality of life at 6 months SF-36 scale	The studies showed no significant differences between EAA inhibitors and the control group in SF-36 outcomes at 6 months.			1307 (2 RCTs)	⨁⨁⨁◯ Moderate¶	EAA inhibitors probably result in little to no difference in quality of life at 6 months.
Adverse events follow-up: range 1 day to 6 months	A total of 64 adverse effects were reported, with none documented in more than six trials. The studies reported no significant difference between EAA inhibitors and the control group regarding overall adverse events or severe adverse events.			3019 (17 RCTs)	⨁⨁◯ Low**

GRADE Working Group grades of evidence.

High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.

Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

^†Downgraded for imprecision based on trial sequential analysis results.

^‡Downgraded for risk of bias: three of the six RCTs evaluating ICP are at high risk of bias.

^§Downgraded for imprecision: large CIs caused by small number of participants.

^¶Downgraded for indirectness because only one quality of life scale (SF36) was estimated.

^**Downgraded for imprecision because to the high number of adverse events identified, the majority of which had low event counts.

EAA, excitatory amino acid; GRADE, Grades of Recommendation, Assessment, Development and Evaluation; ICP, intracranial pressure; MD, mean difference; RCTs, randomised controlled trials; RR, risk ratio; SF 36, Short Form 36.

Figure 3. Subgroup analyses of unfavourable neurological outcome. Unfavourable neurological outcome was defined by Glasgow Outcome Scale (GOS) from 1 to 3 or GOS-Extended (GOS-E) from 1 to 4. EAA, excitatory amino acid; iGluR, ionotropic glutamate receptors; NMDA-R, N-Methyl-D-Aspartate receptor; TBI, traumatic brain injury.

Secondary outcomes

Mortality

The timing of mortality assessment varied across studies, ranging from hospital discharge to 6 months (detailed timepoints of mortality assessment are provided in online supplemental eFigure 7). The use of early EAA inhibitors was not associated with reduced mortality (RR 0.90, 95% CI (0.73 to 1.10), I²=35%, 20 trials,³⁵⁴³,45 47 48 50 n=3427) (figure 4, online supplemental material eFigure 7) (moderate certainty, table 1). These results were consistent regardless of the type of EAA inhibitor, timing or duration of administration, or risk of bias in the included trials (online supplemental material eFigure 8). TSA showed that our analysis was underpowered (online supplemental material eFigure 9).

Figure 4. Secondary outcomes for (A) dichotomous values and (B) continuous values. EAA, excitatory amino acid inhibitor; ICU, intensive care unit.

Intracranial pressure

No significant difference in the mean ICP was observed (mean difference MD, –0.45 mm Hg (−1.67 to 0.77 I²=0%, 6 trials,^{45 47 49 50 61 65} n=924)) (figure 4, online supplemental material eFigure 10) with moderate certainty evidence (table 1). The results were consistent across all subgroups (online supplemental material eFigure 11). One trial⁴³ reported the percentage of time with ICP >20 mm Hg, while two trials (n=733)^{44 54} provided a narrative description of ICP and reported no difference between groups. TSA showed that the cumulative evidence crossed the futility boundary (online supplemental material eFigure 12).

ICU and hospital length of stay

We observed no difference in ICU length of stay (MD, −2.47 days; (−5.53 to 0.59) I²=84%, 5 trials,^{51 55 60 62 63} n=461) (figure 4 and online supplemental material eFigure 13) but we observed a significant reduction of hospital length of stay (MD, −2.86 days (−5.58 to −0.15) I²=39%, 6 trials,^{53 55 57} n=443) (figure 4 and online supplemental file 1material eFigures 14–16). The certainty of evidence was low (table 1). TSA showed that our analysis was underpowered for ICU length of stay but not for hospital length of stay (online supplemental material eFigures 17 and 18).

Quality of life

Quality of life was assessed in two trials^{43 47} (n=1345). Both trials evaluated quality of life using the Short Form 36 (SF-36) scale.⁶⁶ No beneficial effects were observed with the use of dexanabinol on the SF-36 median physical composite score⁴³ (MD −0.4 (−2.53 to 1.73) or on the mental composite score (MD −0.8 (−3.23 to 1.63). Similarly, no improvement in SF-36 scores was observed with magnesium sulfate,⁴⁷ both in the physical composite score (MD 0.23 (−2.73 to 3.19)) and the mental composite score (MD −1.33 (−4.17 to 1.51)).

Adverse events

19 trials (n=3289)³⁶,³⁸⁴³ assessed adverse events, with reporting ranging from detailed data to brief narratives. Two trials^{45 65} (n=471) observed comparable adverse event rates between groups, and five trials^{39 52 56 57 59} (n=725) described the drugs as ‘safe’ or ‘well-tolerated.’ Two small trials^{36 38} reported no serious adverse event, while a large trial noted serious adverse events in 41.8% of intervention patients vs 39% in controls.⁴³ Adverse events were categorised by body system in six trials (n=3984),³⁶⁴³,^{45 58 65} affecting haematologic⁴³,⁴⁵⁶⁵ (mainly anaemia), cardiovascular⁴³,⁴⁵⁶⁵ (mainly rhythm disorders), neurologic³⁶⁴³,^{45 65} and gastrointestinal systems^{43 58 65} with no differences between intervention and control. In total, 64 adverse effects were reported, none in more than six trials. Despite wide CIs for many estimates, no significant differences between groups were observed. The certainty of evidence was low (table 1).

Discussion

In this systematic review and meta-analysis of adults with moderate to severe TBI, early administration of EAA inhibitors was not associated with reduced risk of unfavourable neurological outcomes at 6 months. Early EAA inhibitor administration was similarly not associated with improvements in mortality, ICU length of stay, quality of life or adverse events. Hospital length of stay was reduced in the EAA inhibitor group.

Although our results align with those of previous systematic reviews, these reviews have primarily focused on specific subclasses of EAA inhibitors rather than the broader EAA inhibitors class. No significant effect on mortality or long-term functional outcomes has been observed with magnesium sulfate in TBI patients.^{13 14} Likewise, evidence on ketamine is limited and with a low level of evidence, especially concerning its impact on neurological outcomes, mortality and ICP.^{67 68} A recent systematic review and meta-analysis, which included patients with both TBI and subarachnoid haemorrhage, confirmed these findings and reported no significant harm associated with ketamine use.¹⁵ Other meta-analyses evaluated the effects of amantadine and memantine in TBI,^{16 69 70} suggesting that these treatments may enhance cognitive performance and reduce irritability, with good tolerance and a low risk of severe adverse events. However, their use has been investigated primarily in the later stages of neurological recovery. In our review, we adopted a broader perspective by evaluating the entire class of EAA inhibitors in the acute phase of moderate to severe TBI. Our findings suggested that only two classes of EAA inhibitors (iGluR antagonists targeting NMDA receptors and magnesium sulfate) have been investigated in this setting. Notably, ketamine has been evaluated in only three trials, with a limited number of patients, making it difficult to draw definitive conclusions on this specific drug. Beyond these, other types of EAA inhibitor used in epilepsy, including other iGluR inhibitors (such as AMPA receptor antagonist) and mGluR antagonists,⁷¹ have not yet been evaluated in humans in the context of TBI. After showing promise in preclinical TBI studies,^{72 73} their potential clinical application warrants further exploration through RCTs.

EAA inhibitors have also been evaluated in other neurocritically ill patient populations, including stroke^{74 75} and subarachnoid haemorrhage.⁷⁶,⁷⁹ A previous systematic review assessing the entire class of EAA inhibitors in stroke found no significant evidence of benefit or harm associated with drugs modulating EAA activity on death or dependence.⁷⁵ This review concluded that further clinical trials of EAA inhibitors are warranted, as CIs around effect estimates remain wide, preventing definitive conclusions regarding potential benefits. Another systematic review on ischaemic or non-ischaemic strokes has examined specific subclasses of EAA inhibitors, such as intravenous magnesium sulfate, and did not show an improvement in functional outcomes or 90-day mortality.⁷⁴ In subarachnoid haemorrhage, findings on intravenous magnesium have been inconsistent. While one study suggested that magnesium sulfate might reduce the risk of delayed cerebral ischaemia and poor outcomes,⁷⁶ others found no beneficial effect on neurological recovery.⁷⁷,⁷⁹ In our review, a subgroup analysis of magnesium sulfate, based on six studies including more than 800 patients does not exclude a potential beneficial effect on neurological recovery. While these findings should be interpreted with caution, they suggest that the potential neuroprotective role of magnesium sulfate in TBI remains uncertain and may warrant further investigation.

From a pathophysiological perspective, TBI is characterised by an early diffuse excitotoxic response driven by mechanical tissue deformation and widespread glutamate release, extending beyond focal lesions and propagated by spreading depolarisations that amplify glutamate excitotoxicity and metabolic crisis.⁸⁰ Spreading depolarisations are clinically relevant, as their occurrence after acute TBI has been associated with unfavourable neurological outcomes.⁸¹ Accordingly, targeting spreading depolarisations through EAA pathway inhibition may represent a promising and complementary therapeutic avenue in TBI patients, supported by evidence that NMDA receptor antagonists such as ketamine^{82 83} and memantine⁸⁴ can suppress spreading depolarisations.

Strengths and limitations

Our review has several strengths. First, we focused on a more severely ill population of TBI (moderate and severe) that is more likely to benefit from such an intervention considering the acute excitotoxicity induced by the TBI aligned with the underlying pathophysiological mechanisms. Second, we focused on the whole class of EAA inhibitors, which allowed us to consider a broader variety of molecules. Third, we conducted a TSA, which showed that the current body of evidence remains underpowered to detect a potential effect on neurological outcomes, with only 45% of the required information size accrued to demonstrate a 10% relative risk reduction. We identified 18 ongoing trials investigating EAA inhibitors in moderate to severe TBI, 12 of which will assess neurological function. However, all are small-scale RCTs, limiting their ability to provide conclusive evidence.

Our study also has limitations. First, most included trials studied NMDA receptor antagonists, with significant variability in administration timing and duration of treatment. Notably, among the trials contributing to our primary outcome, EAA inhibitors were administered for less than 24 hours in four of them, representing more than 1000 patients.^{43 48 61 65} This limited exposure may be suboptimal, as excitotoxicity and secondary brain injury processes are most prominent during the first days following TBI, potentially contributing to neutral effects on long-term neurological outcomes. Second, several trials reported neurological outcome scales (GOS or GOS-E) as continuous variables rather than ordinal scales, which prevented their inclusion in the meta-analysis for the primary outcome. However, these trials represented a small proportion of the overall population (7%), and none reported a statistically significant difference in neurological outcomes between groups, supporting the robustness of our findings.

Conclusions

In this systematic review, we did not observe a reduction in long-term unfavourable neurological outcomes with the early administration of EAA inhibitors in patients with moderate to severe TBI. TSA underscores the need for well-designed and adequately powered RCTs assessing the effect of EAA inhibitors in this patient population.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.