Editorial 

October 25, 2023

β-Blockers in Patients With Sepsis: Putting the Puzzle Together, Piece by Piece

Steven M. Hollenberg

JAMA. Published online October 25, 2023. doi:10.1001/jama.2023.20455

Challenging conventional wisdom is important and frequently useful. Septic shock is a form of distributive shock in which hypotension results from vasodilation.¹ Although myocardial performance may not always be entirely normal, a phenomenon termed septic cardiomyopathy,² cardiac output is usually preserved, in part by increased heart rate. Therapy of septic shock refractory to fluid administration entails administration of vasopressor agents; norepinephrine is generally preferred as an initial agent.³ In those cases in which cardiac output is felt to be low enough to compromise perfusion, inotropic agents may be used.³ In this context, use of β-blockers may well be regarded as counterintuitive inasmuch as their hemodynamic effects would tend to decrease arterial pressure and cardiac output. Despite this, investigators have challenged conventional wisdom and evaluated the effects of β-blockade for treating sepsis.

β-Blockers have been shown to improve survival in some small animal models of septic shock.⁴ Despite decreased heart rate, cardiac output is often preserved in these models, and some studies have suggested that a decrease in myocardial workload may be associated with improved myocardial energy efficiency.⁵ Not all animal studies have shown benefits with β-blockade, however. In a sheep model, esmolol was associated with deterioration of kidney and microvascular perfusion.⁶

Among patients with sepsis, a retrospective study reported that chronic use of β-blockers prior to hospital admission was associated with improved outcomes.⁷ Clinical studies of administration of β-blockers for sepsis have used the short-acting intravenous agents, esmolol and landiolol. Early studies showed that these agents could usually be given without marked deterioration of hemodynamics. The largest randomized trial of β-blockade in sepsis was a single-center, open-label study that examined the safety of reducing heart rate to a target range of 80/min to 94/min among 154 patients with sepsis receiving norepinephrine infusion randomized to receive esmolol or placebo.⁸ The heart rate end point was met, stroke volume was increased, and cardiac output was unchanged with esmolol. The adjusted hazard ratio for mortality was 0.61, with a decrease from 80.5% to 49.4% with esmolol, but this was a nonprimary outcome. Another challenge with this trial was that mortality in the control group was much higher than most other trials involving patients with septic shock.⁸ This trial, although not regarded as definitive, spurred interest in this strategy.

Landiolol is an ultrashort-acting β-blocker that is 8 times more selective for the β₁-receptor than esmolol and as such was hypothesized to provide more myocardial protection and immunomodulation while providing a margin of safety due to its very short half-life. Whitehead and the STRESS-L investigators⁹ report the results of a pragmatic randomized trial that compared landiolol with placebo. Patients were eligible for randomization if they met Sepsis-3 criteria,¹⁰ had been adequately fluid resuscitated, were receiving norepinephrine at a dose of 0.1 µg/kg/min for more than 24 hours and less than 72 hours, and had a heart rate greater than 95/min. Open-label landiolol was given to achieve a heart rate of 80/min to 94/min via a specified protocol with titration every 15 minutes. The primary end point was mean SOFA score averaged over the first 14 days of the trial and while patients remained in the ICU, an end point previously used in the Levosimendan for the Prevention of Acute Organ Dysfunction in Sepsis (LeoPARDs)¹¹ trial. Secondary outcomes included mortality at 28 and 90 days, length of stay, mean arterial pressure, and norepinephrine doses over the first 5 days, lactate concentrations, and fluid balance. Safety outcomes included bradycardia and heart block, with and without hypotension.

There were no formal stopping rules, but the trial was stopped early by the data monitoring committee after enrollment of 126 of a planned 340 patients on the basis that landiolol was unlikely to demonstrate benefit even if recruitment had continued to the study’s full sample size and also because there was a signal of possible harm in relation to mortality in the intervention group. The mean (SD) SOFA score over the 14 days was 8.8 (3.9) for the landiolol group compared with 8.1 (3.2) for the standard care group (P = .24). Mortality at day 28 was 37.1% in the landiolol group and 25.4% in the standard care group (P = .16) and at 90 days was 43.5% for the landiolol group compared with 28.6% for the standard care group (P = .08). The landiolol group had lower mean arterial pressure, greater norepinephrine doses, and numerically higher lactate levels. A higher proportion of patients receiving landiolol experienced serious adverse events. The subgroup analyses, although underpowered due to the entirely appropriate decision of the data monitoring committee to terminate the trial given the early signal of potential harm, showed consistent results across subgroups of norepinephrine dose, presence of acute respiratory distress syndrome, and use of β-blockers prior to randomization.

Why did landiolol fail to show benefit in this trial? One possibility is a hemodynamic effect. The current study started β-blockade after 24 hours in the face of data from a pilot trial showing instances of hypotension and decreased cardiac output with early β-blockade¹²; such patients may have septic cardiomyopathy and may not tolerate β-blockade. Cardiac output monitoring was left to the discretion of the investigators for pragmatic reasons, and this is an important limitation of the trial. It is hard to exclude the possibility that at least some of the patients in the landiolol group may have had decreased cardiac output, with or without a vasodilatory effect, a possibility that is supported by decreased blood pressure, elevated lactate levels, and increased norepinephrine requirements in this group. It is conceivable that closer hemodynamic monitoring, at least in selected patients, would have allowed more careful titration of β-blockers so as to minimize decreases in cardiac output. In this respect, differences between the STRESS-L trial and the trial by Morelli et al⁸ included the use of hemodynamic monitoring, a requirement of a mixed-venous oxygen saturation of more than 65%, and use of levosimendan in the trial by Morelli et al, all of which may have helped ensure that cardiac output was maintained.

One of the postulated mechanisms of benefit from β-blockade is decreased myocardial oxygen requirement at a lower heart rate. There is little evidence for myocardial ischemia as a cause of septic cardiomyopathy, however.²In addition, to the extent that energetic failure in sepsis may result from dysfunction in mitochondrial oxygen use rather than perfusion failure, this is likely to be independent of hemodynamic status and also not especially responsive to β-blockers.

Another possibility is that potential benefits of β-blockade in sepsis are independent of systemic hemodynamic effects. Although it is hard to imagine titrating β-blockade to anything other than heart rate, if only for purposes of safety, those effects might require dosages different from those needed to control heart rate.

It is also possible if not likely that β-blockers have the potential for both beneficial and deleterious effects, both within and among patients. Sepsis is nothing if not complex. β-Blockade may have beneficial effects in some areas, such as inflammation and metabolism, but potentially deleterious effects in others. The balances between various effects may well diverge in different patients and at different times. In this context, further exploration of mechanisms of action in patients with sepsis could help guide future trials, in terms of patient selection, or dosing, or timing, or all of the above.

The authors should be commended for conducting a rigorous randomized trial with a clinical end point, something that had been previously lacking despite promising findings from other investigations with surrogates as primary end points. Does this signal the end for investigation of β-blockade for sepsis? One would hope not. Further elucidation of mechanisms by which β-blockers may be beneficial or harmful might lend insight into the pathophysiology of septic shock and potentially allow for identification of patient subsets that might be selected for β-blockade. There is still much more to be learned.