In adults, administration of esmolol within 24 h of severe TBI at a starting dosage of 10 μg/kg/min with increments of 5 μg/kg/min titrated to a heart rate reduction of 15% from baseline is associated with a low probability (0.07) of dose-limiting toxicity necessitating withdrawal of esmolol. The infrequent hemodynamic adverse effects can be managed using standard vasopressor titration with cerebral perfusion pressure defended for 90% of the infusion time. An internal review of the single dose-limiting toxicity event suggested that this risk could be minimized with consistent adherence to study flowcharts. Defining toxicity using a single metric, although necessary for the CRM, is an oversimplification of the complex effects of beta-blockade in critically ill patients. Nevertheless, our findings support the safety and feasibility of early beta-blockade with esmolol within 24 h of severe TBI in adults.

The observed 6-month mortality of 12.5% is less than that predicted by the Helsinki CT score in our cohort [39, 40]. It is also lower than the ICU stratum of Collaborative European NeuroTrauma Effectiveness Research in TBI (CENTER-TBI; 15% in a cohort of patients with both moderate and severe TBI) and lower than that shown in the Trauma Audit and Research Network analysis of TBI in England (40.4% for severe TBI) [41, 42]. In the absence of a control group, we are unable to attribute this to esmolol. It may reflect study bias or nonprotocolized aspects of care in our center. It is however consistent with findings in a meta-analysis of beta-blockade in TBI and supportive of further research to investigate a potential benefit of esmolol.

This is only the third study reporting the use of esmolol in adults after severe TBI. In an observational study in Iran, 12 patients exposed to a loading dose (0.5 mg/kg) and then a 24-h fixed dosage infusion of esmolol (50 mg/kg/min) had lower ICP during that period than a contemporary control group, with no differences in heart rate or mean arterial pressure [20]. We have shown that a longer duration of infusion is feasible, as is dosage adjustment to heart rate as a simple biomarker of the stress response. Our exclusion criteria are less restrictive, and our starting point is better defined, which, together with the difference in clinical setting, extends the generalizability of the intervention.

In a larger observational study from the United States, esmolol use was reported in 7% of the 1120 patients (from a cohort of 2337) receiving any beta-blocker for any usual indication rather than for neuroprotection during an ICU stay. Only 38% of patients had severe head injury (GCS score ≤ 8). No esmolol-specific outcomes were reported, although both beta-1 selective and all beta-blocker use were associated with reduced 30-day mortality [43]. Our study adds data on longer-term mortality associated with esmolol in a more severely injured cohort.

Early administration of beta-blockade with titration to heart rate was tested in a randomized controlled trial of metoprolol, another beta-1 selective beta-blocker, in 60 patients in Egypt [18]. In this study, reductions in mortality (seen in patients > 40 years old only) and ICU length of stay and improvement in Glasgow Outcome Scale scores at 1 month were shown in the intervention group. Outcomes were better in those achieving the fixed heart rate target of 60–70 beats per minute early, though this was not further defined. Dose titration was made on a 6-h cycle. Baseline heart rates, time taken to achieve target, and other hemodynamic data are not available in the study report. We have shown that use of an intravenous drug with titration every 30 min allows rapid target attainment. Whether this translates into greater benefit and whether a relative or an absolute heart rate target is preferable requires further research.

As with these prior studies, we chose not to protocolize all aspects of management, particularly those such as analgesia and sedation or fluid resuscitation that could have a significant bearing on heart rate early after major trauma. Indeed, given the reducing number of patients requiring esmolol each study day, it is likely other factors, such as the natural time course of the stress response, were contributing to heart rate reduction over time. The advantages of this protocol approach are simplicity in terms of intervention delivery—there is only one infusion and one target—and pragmatism when scaling up to effectiveness trials. The disadvantages are somewhat mitigated by consistent patient management practice in a single center. A key next step in the EBB-TBI program is the trial of the intervention at more sites.

Another consideration for future research is the wide interindividual variability in the patient population in terms of stress response (evidenced by baseline heart rate), dosage of esmolol, and duration of infusion. This likely reflects the heterogeneity of the TBI population, arising from patient, mechanism, intracranial and extracranial injury pattern, and interventions received. In future work, we plan to undertake assessment of biomarkers, including of brain, endothelial, immune, and sympathetic function, to try to define subphenotypes for subsequent study enrichment. Cardiac troponin remains an attractive and simple biomarker that could define a group most likely to benefit [44]. In its absence, titration of esmolol to a proportional heart rate reduction as a biomarker of pharmacodynamic engagement indicating a level of beta-blockade that might be sufficient to provide neuroprotection seems a rational approach. The simplicity and universal availability of heart rate balances the fact that it is at best a crude approximation of the stress response. Our target of a 15% reduction from baseline is convenient and may be safe but is not proven to be the ideal sole marker for posttraumatic neuroprotection [37]. Whether any particular modality of advanced neuromonitoring would add to safety (by identifying low cerebral blood flow or suboptimal perfusion pressure, for example) or to effect (by identifying groups likely to benefit) is as yet unknown.

We acknowledge the apparent contradiction in use of beta-blockade with concurrent use of catecholaminergic vasopressors. The resulting in vivo balance between alpha-1, beta-1, and beta-2 activation and the consequent effect on pathophysiology is difficult to predict. The stable SOFA scores and lactate values we observed argue for the maintenance of hemodynamics and vital organ perfusion. In terms of secondary brain injury, there is evidence to show that phenylephrine (a pure peripheral alpha-agonist) is associated with lower in-hospital mortality after severe TBI [45]. This is consistent with harm from excessive catecholamines being mediated via beta-receptors, which may potentially be mitigated with beta-blockade.

Propranolol is a nonselective beta-blocker that is frequently used after TBI and favored by some, but not all, as the first-line neuroprotective beta-blocker [16, 18, 43, 46, 47]. Prospective randomized trials comparing selective against nonselective beta-blockers in this setting have not been undertaken. A previous review favoring propranolol uses a heterogenous comparator of all other selective and nonselective beta-blockers together with those with additional actions (e.g., labetalol, sotalol) [43]. In our opinion, the theoretical and practical advantages of esmolol make it deserving of further research, including potentially in direct comparison with propranolol in the ICU setting.

There are still other potential strategies (such as dexmedetomidine, a centrally acting alpha-agonist that reduces sympathetic outflow) that could mitigate harm associated with excess catecholamines [48]. Whichever strategy is chosen, underlying mechanisms are complex and extend beyond simple hemodynamics to regulation of fundamental metabolic, immune, and inflammatory pathways [49, 50]. Further research is needed to determine the ideal therapeutic approach.

The strengths of our study include the pragmatic approach, broad inclusion criteria (especially compared to other trials of beta-blockade in TBI) [15, 17,18,19], and the application of an established methodology to a novel setting and research question. Our study also has limitations. It represents practice in a single center, so intervention delivery and effect may not generalize to other units. A high proportion of the patients admitted with TBI did not meet the inclusion criteria, further limiting generalizability. In part, this reflects local practice in patient selection for ICP monitoring. In addition, a lack of out-of-hours research staff contributed to the number of patients excluded because they were identified more than 24 h after injury. This also prevents deeper analysis of the patient population to try to identify subgroups more likely to receive benefit or suffer harm. We cannot differentiate the relative contribution of esmolol and other interventions (such as fluid or sedation) on heart rate control or clinical outcomes. We chose a conservative intervention period focused on the point of greatest physiological instability and prior to the typical development of cerebral edema and cannot determine whether this is the ideal duration of infusion. Our sample is small, meaning there is potential for uncertainty in our estimate of maximum tolerated dosage. This is mitigated by titration to effect for each patient. We had a high loss to longer-term follow-up, perhaps relating to the large catchment area and cognitive impairment or psychological distress in survivors, although this did not compromise analysis of the primary outcome.

This study adds to the small number of esmolol-treated patients with severe TBI reported in the literature. Our results are consistent with the positive findings of other studies with esmolol or alternative beta-1 selective blockers, which supports the argument for further research into early beta-1 selective blockade. We have identified some key areas of uncertainty, among which we consider the need to identify severe TBI populations most likely to benefit, the choice of beta-1 or nonselective beta-blockade, and the influence of other interventions (such as sedation or surgery), as the most important. Given these uncertainties and the very small number of esmolol-treated patients (28, including in this study) in only two centers worldwide with surrogate primary outcomes, it is too early to recommend esmolol for widespread use or even for a randomized trial of effectiveness. Further study in multiple centers and larger cohorts with analysis of a range of biomarkers reflecting both the complexity of the stress response and severity of cranial and noncranial injury is necessary, along with data on relevant patient-centered outcomes [51]. With better understanding of pathophysiology and the links to outcomes, better therapy can be identified and personalized [52].