Ketamine administration after traumatic brain injury was not associated with increased mortality or greater disability despite being given to more severely injured patients. Ketamine exposure correlated with fewer episodes of elevated intracranial pressure and smaller rises in TBI protein biomarkers, but a higher incidence of seizures and, in the intracranial haemorrhage subgroup, more cardiac events.
Purpose
Ketamine has historically been contraindicated in traumatic brain injury (TBI) due to concern for raising intracranial pressure. However, it is increasingly being used in TBI due to the favorable respiratory and hemodynamic properties. To date, no studies have evaluated whether ketamine administered in subjects with TBI is associated with patient survival or disability.
Methods
We performed a retrospective analysis of data from the multicenter Prehospital Tranexamic Acid Use for Traumatic Brain Injury trial, comparing ketamine-exposed and ketamine-unexposed TBI subjects to determine whether an association exists between ketamine administration and mortality, as well as secondary outcome measures.
Results
We analyzed 841 eligible subjects from the original study, of which 131 (15.5%) received ketamine. Ketamine-exposed subjects were younger (37.3±16.9 years vs. 42.0±18.6 years, P=0.037), had a worse initial Glasgow Coma Scale score (7±3 vs. 8±4, P=0.003), and were more likely to be intubated than ketamine-unexposed subjects (88.5% vs. 44.2%, P<0.001). Overall, there was no difference in mortality (12.2% vs. 15.5%, P=0.391) or disability measures between groups. Ketamine-exposed subjects had significantly fewer instances of elevated intracranial pressure (ICP) compared to ketamine-unexposed subjects (56.3% vs. 82.3%, P=0.048). In the very rare outcomes of cardiac events and seizure activity, seizure activity was statistically more likely in ketamine-exposed subjects (3.1% vs. 1.0%, P=0.010). In the intracranial hemorrhage subgroup, cardiac events were more likely in ketamine-exposed subjects (2.3% vs. 0.2%, P=0.025). Ketamine exposure was associated with a smaller increase in TBI protein biomarker concentrations.
Conclusions
Ketamine administration was not associated with worse survival or disability despite being administered to more severely injured subjects. Ketamine exposure was associated with reduced elevations of ICP, more instances of seizure activity, and lower concentrations of TBI protein biomarkers.
The authors contextualise ketamine as a dissociative anaesthetic historically avoided in traumatic brain injury (TBI) because early reports suggested it could raise intracranial pressure (ICP), cerebral blood flow and metabolic rate. More recent clinical practice has increasingly used ketamine in trauma because of its haemodynamic stability, analgesic properties and respiratory-sparing effects. Small-scale prospective studies in TBI have reported either no increase or a reduction in ICP after ketamine, and preclinical mechanisms (e.g. reduction of glutamate toxicity and inhibition of cortical spreading depression) have suggested possible neuroprotective effects. However, earlier work did not examine patient-centred outcomes such as survival or disability, leaving uncertainty about the overall safety and outcome implications of ketamine use after TBI. Using the dataset from the multicentre Prehospital Tranexamic Acid Use for Traumatic Brain Injury (TXA for TBI) trial, the researchers set out to examine whether prehospital ketamine administration is associated with mortality, disability and physiologic responses in subjects with moderate-to-severe TBI. They also investigated ketamine's association with longitudinal trajectories of circulating TBI protein biomarkers (GFAP, MAP2, UCHL1) and performed a predefined subgroup analysis in patients with intracranial haemorrhage (ICH). The a priori hypothesis was that ketamine exposure would not be associated with worse survival or disability and would not produce different physiologic or biomarker responses compared with no ketamine exposure.
This study is a retrospective analysis of prospectively collected data from the TXA for TBI trial, a multicentre, North American randomised controlled trial that enrolled subjects from May 2015 to November 2017. The original trial randomised subjects with moderate-to-severe TBI to different tranexamic acid (TXA) regimens; the present analysis used that trial's dataset to compare subjects with and without recorded prehospital ketamine administration. Ketamine exposure was defined as documentation of ketamine given by emergency medical services in the prehospital setting; exact dosing and timing were not available. Subjects without ketamine exposure data were excluded, leaving 841 subjects for analysis (131 ketamine-exposed, 710 unexposed). Demographic and clinical variables extracted included age, sex, BMI, weight, initial Glasgow Coma Scale (GCS), Injury Severity Score (ISS), history of seizure, intubation status, presence of radiographic ICH and TXA study arm. The primary outcome was death within 6 months of injury. Secondary outcomes comprised Glasgow Outcome Scale Extended (GOSE) and Disability Rating Scale (DRS) at discharge and 6 months, incidence of seizure activity, cardiac events (cardiac arrest or heart failure), and surgical interventions. Physiologic measures recorded during the first 24 hours after hospital presentation included systolic blood pressure (SBP), heart rate, temperature, arterial oxygen pressure (PaO2), and ICP. ICP elevation was defined as ICP > 20 cmH2O. Definitions for bradycardia, tachycardia, hypoxia, hypotension, hypertension, hypothermia and hyperthermia were prespecified. Longitudinal serum measurements of GFAP, MAP2 and UCHL1 were available at admission and at 6, 12, 24 and 48 hours; biomarker analyses used change from admission and absolute concentrations. Statistical analysis used multivariate mixed-effects logistic regression to evaluate associations between ketamine exposure and outcomes, with covariates including age, sex, race, BMI, seizure history, intubation status and TXA study assignment. Because cardiac events and seizures were rare, simplified models with reduced covariates were specified for those outcomes (details of which covariates were removed are reported). Biomarker concentrations were log-transformed and analysed with mixed-effects models using the same covariates; p-values for biomarker models were adjusted for multiple comparisons using Bonferroni correction. Stratifications and score cut-offs (e.g. GCS, ISS, GOSE, DRS) are reported, and analyses were performed using Stata v17.0. The ICH subgroup (radiographically confirmed intracranial haemorrhage) was analysed separately given prior findings in the parent trial.
Of 966 subjects in the original trial, 910 had ketamine data and 841 had complete ketamine exposure data for this analysis; 131 subjects (15.6%) received prehospital ketamine and 710 (84.4%) did not. Ketamine-exposed subjects were younger (mean age 37.3 ± 16.9 years vs. 42.0 ± 18.6 years; P = 0.037), had a worse initial GCS (mean 7 ± 3 vs. 8 ± 4; P = 0.003) and were more frequently intubated (88.5% vs. 44.2%; P < 0.001). ISS and ICH prevalence were similar between groups. Primary outcome: death within 6 months occurred in 12.2% of ketamine-exposed subjects versus 15.5% of unexposed subjects (P = 0.391), indicating no statistically significant difference in mortality. Secondary outcomes: there were no significant differences in disability measures (GOSE at discharge or at 6 months; DRS) or in the need for surgical interventions (6.9% ketamine-exposed vs. 11.1% unexposed; P = 0.113). Using simplified models due to low event counts, seizure activity was more frequent in ketamine-exposed subjects (3.1% vs. 1.0%; P = 0.010). Cardiac events were rare and not associated with ketamine exposure in the whole cohort (1.5% vs. 0.8%; P = 0.961). Physiologic responses in the first 24 hours: instances of elevated ICP (ICP > 20 cmH2O) were less frequent in the ketamine-exposed group (56.3% vs. 82.3%; P = 0.048). Other extremes (hypoxia, heart rate, SBP, temperature) showed no significant differences between groups. Biomarkers: ketamine exposure was associated with smaller increases from admission in GFAP at 12, 24 and 48 hours, and in MAP2 at 12 and 24 hours. Changes in UCHL1 from baseline did not differ significantly between groups. Absolute biomarker concentrations were not different between groups after correction for multiple comparisons, except that in the ICH subgroup GFAP concentration at 48 hours was lower in ketamine-exposed subjects (2,261 ± 4,727 pg/mL vs. 4,497 ± 10,508 pg/mL; P = 0.003). ICH subgroup (n = 502 with radiographic ICH; 86 ketamine-exposed): demographic patterns mirrored the full cohort (ketamine-exposed younger and more often intubated). Mortality was higher overall in the ICH subgroup but did not differ by ketamine exposure (18.6% vs. 23.3%; P = 0.177). Ketamine exposure remained associated with increased seizure activity in the ICH subgroup (3.5% vs. 1.0%; P = 0.009) and, in this subgroup, cardiac events were more frequent with ketamine (2.3% vs. 0.2%; P = 0.025). The difference in ICP elevation observed in the full cohort did not reach statistical significance in the ICH subgroup (56.3% vs. 82.0%; P = 0.055). Biomarker trajectories in the ICH subgroup paralleled the overall findings (smaller rises in GFAP and MAP2 with ketamine, no difference in UCHL1).
The authors interpret these retrospective findings as indicating that prehospital ketamine administration was relatively common (15.5%) in the TXA for TBI trial and, despite being administered to subjects with a worse initial neurologic profile (lower GCS, higher intubation rates), ketamine exposure was not associated with increased mortality or disability. They note that the ketamine-exposed group was on average younger, which could favour outcomes, but emphasise that no prior studies have examined ketamine's association with morbidity and mortality after TBI and consider these results reassuring given ketamine's frequent use. The authors also highlight safety signals: seizure activity was significantly more common among ketamine-exposed subjects in both the full cohort and the ICH subgroup, and cardiac events were associated with ketamine in the ICH subgroup. They caution that these events were rare and that it is unclear whether the associations reflect ketamine effects or residual confounding related to the worse injury profile of the ketamine group. The authors remind readers that ketamine has been associated with cardiovascular collapse in other populations and recommend continued caution regarding cardiovascular and neurological side effects when administering ketamine. Contrary to historical concerns that ketamine raises ICP, the authors found fewer instances of elevated ICP in ketamine-exposed subjects. They situate this result with prior small clinical studies in TBI that showed no increase or reductions in ICP with ketamine, and they note limitations of the early studies (e.g. healthy volunteers, uncontrolled respiratory rate) that may have contributed to earlier findings. The authors acknowledge that their analysis captured instances of ICP extremes rather than duration or magnitude of ICP elevation. Regarding biomarkers, the authors report that ketamine exposure was associated with smaller rises in GFAP and MAP2 over time, and a significantly lower GFAP concentration at 48 hours in the ICH subgroup. They suggest this could indicate a suppressive effect of ketamine on these circulating TBI biomarkers and propose that longitudinal biomarker trajectories might be useful to monitor pharmacologic effects and therapeutic responses to ketamine. The authors conclude by noting that these associations are hypothesis-generating and warrant further investigation. Limitations acknowledged by the authors include the retrospective nature of the analysis (precluding causal inference), ICP measurements being available in fewer than 20% of subjects (limiting power to draw conclusions about ICP), absence of detailed ketamine dosing and timing data, and physiologic analyses being limited to instances of extremes rather than continuous measures of severity or duration.
The TXA for TBI (Prehospital Tranexamic Acid Use for Traumatic Brain Injury) trial was a large, multinational, multicenter, randomized controlled trial that provided the opportunity to evaluate an extensive sample of TBI subjects and associated outcomes. Using the dataset from this study, we examined the characteristics of ketamine administration in the TBI population to analyze ketamine's association with morbidity and mortality, as well as physiologic responses including ICP and longitudinal TBI biomarker trajectories. We hypothesized that ketamine exposure would not be associated with worse survival or disability after TBI, and that the physiologic responses and TBI-related biomarker responses would not be different between ketamine-exposed and ketamine-unexposed TBI subjects in the TXA for TBI trial population.
The original studywas approved by the Institutional Review Board for the study's associated Resuscitation Outcomes Consortium Clinical Trials Center at the University of Washington. In-formed consent was not possible for all participants initially and therefore enrollment was conducted under US regulations for Exception from Informed Consent Requirements for Emergency Research as well as the Canadian Tri-County Policy Statement 2, and informed consent was obtained as soon as feasible. The original study is registered on ClinicalTrials.gov (identifier: NCT01990768).
The TXA for TBI trial was a multicenter, North American randomized, controlled trial that enrolled subjects from May 2015 until November 2017. Subjects with moderate-to-severe TBI were randomized to placebo bolus prehospital and 8-hour placebo infusion in-hospital, a 2-g tranexamic acid (TXA) prehospital bolus followed by an 8-hour placebo infusion in the hospital, or a 1-g prehospital bolus followed by a 1-g in-hospital 8-hour infusion. In the primary trial, there were no significant differences in mortality or morbidity between groups overall, however, in the subset of subjects with intracranial hemorrhage (ICH), prehospital administration of a 2-g bolus of TXA bolus was associated with decreased mortality.
We performed a retrospective analysis of the TXA for TBI trial dataset for ketamine exposure to examine the associations between ketamine and clinical and laboratory outcomes in subjects with TBI. Ketamine exposure was defined as recorded ketamine administration by emergency medical services in the prehospital setting; exact dosing and timing of the ketamine administration was not available. Subjects without ketamine exposure data were excluded.
Demographics variables included age, sex, body mass index (BMI), weight, initial Glasgow Coma Scale (GCS) score, initial Injury Severity Score (ISS), history of seizure, intubation status, presence of ICH, and TXA study group allocation.
The primary outcome assessed was death within 6 months of injury; secondary outcomes included the Glasgow Outcome Scale Extended (GOSE) and Disability Rating Scale (DRS) scores at discharge and 6 months postinjury, as well as incidences of seizure activity, cardiac events (cardiac arrest or heart failure), and surgical interventions.
Physiologic outcome measures included the following: systolic blood pressure (SBP), heart rate and temperature, PaO 2 , and ICP. Measurements were taken from the first 24 hours following hospital presentation. ICP was measured with either a Camino (Natus Medical Inc) intracranial pressure and temperature monitor or an external ventricular drain pressure monitor. ICP elevation was defined as ICP > 20 cmH 2 O; bradycardia was defined as heart rate < 60 beats per minute (bpm) and tachycardia as heart rate > 100 bpm; hypoxia was defined as PaO 2 < 80 mmHg; hypotension was defined as SBP < 90 mmHg and hypertension defined as SBP > 180 mmHg; hypothermia was defined as < 35 °C and hyperthermia as > 38 °C.
The TXA for TBI trial performed longitudinal biomarker measurements of the TBI-related biomarkers glial fibrillary acidic protein (GFAP), microtubule-associated protein 2 (MAP2), and ubiquitin C-terminal hydrolase L1 (UCHL1). Serum concentration measurements of GFAP, MAP2, and UCHL1 were compared between the ketamine-exposed and ketamine-unexposed groups at admission, 6-, 12-, 24-, and 48-hours following hospital admission. Biomarkers were compared using the change in measured serum concentration from admission to each subsequent time point, as well as a comparison of biomarker concentrations at each time point.
Based on the decreased mortality observed in the original TXA for TBI trialin the ICH subgroup of the 2-g TXA arm, we performed a subgroup analysis of ICH subjects comparing ketamine exposure groups.
Demographic variables were assessed for their association with outcomes including our primary outcome of death within 6 months of injury. The P-value for significance of variables was set at P < 0.05. A multivariate mixed-effects logistic regression model was created using the variables age, sex, race, BMI, seizure history, intubation status, and TXA study assignment to evaluate the association of ketamine administration on morbidity, mortality, and vital signs outcomes, for both the entire cohort and for the ICH subgroup analysis. Because there were very few instances of cardiac events and seizure activity, a simplified statistical model was created to analyze both outcomes in the entire cohort and the ICH subgroup. For cardiac events in the entire cohort, sex and race were removed from the model; for seizure activity in this same cohort, only race was removed. For ICH subgroup cardiac events, only BMI and intubation status were used as covariates; for seizure activity in the ICH subgroup analysis, sex, BMI, seizure history, and intubation status were used as covariates. To evaluate the association of ketamine exposure with TBI-related protein biomarkers, biomarker data were log-transformed, and a mixed-effects model was created using the same demographic variables (age, sex, race, BMI, seizure history, intubation status, and TXA study arm); final P-values in this model were adjusted for multiple comparisons using Bonferroni correction. All results are provided as mean ± standard deviation unless otherwise specified. TBI was stratified into severe (GCS ≤ 8), moderate (GCS 9-12) and mild (GCS 13-15); ISS was stratified into severe ( > 25), moderate (16-25), and mild ( ≤ 15); GOSE was stratified into poor recovery (1-4) and good recovery (5-8); DRS was stratified into mild (0-1), moderate (2-6), severe, and vegetative or worse ( ≥ 12). Statistical analyses were completed using Stata ver. 17.0 (Stata Corp).
Of the 966 subjects included in the primary analysis in the original TXA for TBI trial, 910 had complete data collected and were analyzed for ketamine exposure; 69 subjects lacked ketamine administration information and were excluded, leaving 841 subjects for our analysis. Of these subjects, 131 (15.6%) received ketamine and 710 (84.4%) did not receive ketamine (Fig.). Demographic and clinical characteristics of the ketamine-exposed and ketamine-unexposed subjects are listed in Table. Overall, the ketamine-exposed group was younger (37.3 ± 16.9 years vs. 42.0 ± 18.6 years, P = 0.037); the groups otherwise had similar sex-and race-distribution, BMI, and TXA study allocations. Ketamine-exposed subjects had a worse initial GCS (7 ± 3 vs. 8 ± 4, P = 0.003) and were more likely to be intubated (88.5% vs. 44.2%, P < 0.001); initial ISS and ICH status were not different between groups.
In our primary outcome of death at any time, there was no difference between the ketamine-exposed and the ketamine-unexposed groups (mortality rate: ketamine exposed, 12.2% vs. ketamine unexposed, 15.5%; P = 0.391) (Table). For our secondary outcomes, there was similarly no difference in disability scores for GOSE at discharge (ketamine exposed, or at 6-months postinjury (ketamine exposed, 5.8 ± 9.9 vs. ketamine unexposed, 7.0 ± 11.2; P = 0.151). There was no difference in the need for surgical intervention between groups (ketamine exposed, 6.9% vs. ketamine unexposed, 11.1%; P = 0.113). Using a simplified statistical model due to very few instances, ketamine exposure was significantly more likely to be associated with seizure activity (3.1% vs. 1.0%, P = 0.010); ketamine exposure was not associated with a higher rate of cardiac events (1.5% vs. 0.8%, P = 0.961) (Table).
Elevated intracranial pressure was significantly less frequent in the ketamine-exposed group (instances of ICP elevation, 56.3% vs. 82.3%; P =0.048). Instances of hypoxia, and extremes of heart rate, SBP, and temperature were similar between groups (Table).
Ketamine exposure was associated with significantly smaller changes in concentration for GFAP at 12-, 24-, and 48-hours following admission and for MAP2 at 12-and 24-hours following admission. UCHL1 changes from baseline were not significantly different between groups at any time point (Fig.). There were no differences in biomarker concentrations when compared. Association of ketamine administration on biomarker trajectories following traumatic brain injury (TBI). TBI-related biomarkers ubiquitin C-terminal hydrolase L1 (UCHL1), glial fibrillary acidic protein (GFAP), and microtubule-associated protein 2 (MAP2) were measured at multiple time points after admission. Ketamine exposure was associated with a smaller increase in GFAP concentration from admission to 12, 24, and 48 hours later. MAP2 concentrations were associated with a smaller increase in concentration from admission to 12 and 24 hours later. Means and standard errors are indicated with points and error bars, respectively. For ease of readability, the scale of the y-axis was increased for biomarker concentrations below 500 pg/mL. Statistically significant differences in the change of biomarker concentration from admission are indicated with asterisks. against each other at any time point after correcting for multiple comparisons (Table).
Of 841 subjects with ketamine exposure data, 502 subjects (59.7%) had a radiographically confirmed ICH, 86 of 131 (65.6%) in the ketamine-exposed group and 416 of 710 (58.6%) in the ketamine-unexposed group (Fig.). Demographic characteristics and TXA study group distribution were similar to the entire cohort, including a significant age difference (ketamine exposed, 38.1 ± 17.5 years vs. ketamine unexposed, 42.0 ± 18.8 years; P = 0.030) and a significantly higher percentage of intubated subjects in the ketamine-exposed group (90.7% vs. 53.8%, P < 0.001). GCS was not significantly different in the ICH subgroup (ketamine exposed, 6 ± 3 vs. ketamine unexposed, 7 ± 3; P = 0.086); all other demographic variables were similar to the entire cohort (Table). Mortality in the ICH subgroup was higher than in the entire cohort but there was again no difference between ketamine exposure groups (ketamine exposed, 18.6% vs. ketamine unexposed, 23.3%; P = 0.177). There were no differences in GOSE or DRS measurements between groups, and no difference in surgical intervention (Table). Using a simplified model, ketamine exposure continued to be associated with increased seizure activity (3.5% vs. 1.0%, P = 0.009); cardiac events were also associated with ketamine exposure (2.3% vs. 0.2%, P = 0.025). Physiologic responses were similar to the entire cohort as well, but no statically significant difference in ICP was observed in this subgroup analysis (ketamine exposed, 56.3% vs. ketamine unexposed, 82.0% with an instance of elevated ICP; P = 0.055) (Table). Biomarker trajectories in the ICH subgroup were similar to the entire cohort, including a smaller increase from baseline for the biomarkers GFAP and MAP2 at similar time points, and no differences in UCHL1 between groups (Fig.). At the 48-hour time point, GFAP concentration was lower in the ketamine-exposed group (2,261 ± 4,727 pg/mL vs. 4,497 ± 10,508 pg/mL, P = 0.003) (Table).
In this retrospective analysis of a large, multinational, multicenter trial, ketamine was administered to 15.5% of all subjects sustaining TBI. It is notable that a significant number of subjects with TBI received ketamine in this large clinical trial involving 12 centers and almost 40 emergency medical services agencies, despite ketamine's traditional avoidance in TBI. In our analysis, ketamine-exposed subjects had a more severe head injury profile based on initial GCS and intubation status yet had no differences in survival or disability. An important qualification of these data is that the ketamine-unexposed group was on average 5 years older than the ketamine-exposed subjects, and older age is associated with worse outcomes after TBI. However, no prior studies have examined ketamine's association with morbidity and mortality after TBI, and these findings, especially taken in context of the higher injury severity, are reassuring given the frequency of use of ketamine in patients with TBI. There remain important safety considerations in the use of ketamine in patients with TBI given our findings associating its use with cardiovascular events and seizure activity. Seizure activity was significantly associated with ketamine administration in both the entire cohort as well as in the ICH subgroup, and ketamine administration was significantly associated with cardiac events in the ICH subgroup. However, these were rare events, and at this time it remains unclear whether the association between seizure activity and ketamine administration is due to the worse injury profile of the ketamine-exposed group or can be attributed to the administration of ketamine itself. Notably, ketamine has been shown to be associated with cardiovascular collapse in other patient populations. Any administration of ketamine should continue to consider its cardiovascular and neurological side effects. In this study, we found that ketamine administration was associated with a decrease in the frequency of elevated ICP, which is in contrast to the traditionally accepted association of ketamine and increased ICP. Early studies examining ketamine's effects on ICP had notable limitations that may explain their findings: study subjects were healthy volunteers without TBI, and respiratory rate was not controlled, which is notable as ketamine does reduce respiratory rate, which has the potential to impact ICP. Another important consideration is that we only analyzed instances of extremes in ICP and not duration or severity of the ICP response. Our findings associating ketamine administration with reduced instances of elevated ICP are consistent with the few clinical studies that have explored ketamine's effects directly in subjects with TBI, which have shown that ketamine either had no effect on ICPor reduced ICP. Finally, the longitudinal circulating biomarkers of brain injury examined in the original TXA for TBI trial presented a unique opportunity to evaluate the effect that ketamine had on the trajectory of these biomarkers. By examining the changes in the concentrations at multiple time points, we observed a potential suppressive effect from ketamine administration in that ketamine exposure was associated with a reduced rise in both GFAP and MAP2 concentrations at all time points, and a significantly reduced GFAP concentration at 48-hours postinjury in the ICH subgroup. Based on our previous observations that these biomarker concentrations are positively associated with injury severity, this suppressive effect associated with ketamine exposure warrants further investigation. Furthermore, these data support the potential utility of using longitudinal TBI biomarker trajectories to monitor pharmacologic effect and therapeutic responses associated with ketamine administration.
The primary limitation of this study is the retrospective collection of these prospectively collected data which therefore limits our findings to associations between ketamine exposure and outcomes rather than establishing a causal relationship. Furthermore, ICP was measured in fewer than 20% of subjects which limits our power to draw conclusions. Finally, complete details regarding the dose and timing of ketamine administration were not available and physiologic responses were limited to instances of extremes.
Create a free account to open full-text PDFs.