Editorial
June 16, 2023
Do Prolonged Infusions of β-Lactam Antibiotics Improve Outcomes in Critically Ill Patients With Sepsis?
Claire N. Shappell, Michael Klompas, Chanu Rhee
JAMA. Published online June 16, 2023. doi:10.1001/jama.2023.6483
There has been a concerted effort over the past 2 decades to improve sepsis outcomes by enhancing recognition of sepsis and decreasing the time to administration of appropriate antibiotics.1 Sepsis mortality rates remain high, however, and the increasing use of broad-spectrum antibiotics may accelerate the emergence of antibiotic-resistant bacteria.2,3 A tantalizing goal is to identify strategies for antibiotic use that can simultaneously improve patient outcomes and decrease antibiotic resistance.
One such potential strategy is prolonged infusion of β-lactam antibiotics. β-Lactam antibiotics demonstrate time-dependent killing, meaning that they are most effective when their serum concentration is above the target pathogen’s minimum inhibitory concentration.4 Conversely, decreases in serum concentrations below the minimum inhibitory concentration are associated with less bacterial killing and may facilitate the development of antibiotic resistance. Standard intermittent infusions (given over 30-60 minutes) create peaks and valleys in drug concentrations, whereas prolonged infusions (either extended dosing over 2-4 hours or continuous administration) can provide more consistent serum levels above the target pathogen’s minimum inhibitory concentration.5Prolonged infusions are particularly attractive when treating patients infected with gram-negative organisms that have higher minimum inhibitory concentrations within the susceptible range, and when treating critically ill patients in whom pharmacokinetics are often variable due to alterations in drug clearance, volume of distribution, fluid balance, and protein binding.
The compelling pharmacological rationale for prolonged infusions of β-lactam antibiotics has been assessed in numerous observational studies and in dozens of clinical trials enrolling variable patient populations with variable infectious syndromes with mixed results. However, several pooled analyses suggest that prolonged infusions improve mortality, clinical cure rates, or both, with the greatest benefit seen in severely ill patients.6 A systematic review and meta-analysis7 combining data from 1876 patients with sepsis from 22 randomized clinical trials (RCTs) reported a risk ratio of 0.70 (95% CI, 0.56-0.87) for all-cause mortality with prolonged infusion vs intermittent infusion of any antipseudomonal β-lactam antibiotic, with low heterogeneity and no evidence of publication bias. Another systematic review and meta-analysis8 combining data from 3401 patients with severe illness from 18 retrospective and prospective studies reported that prolonged infusion of piperacillin-tazobactam was associated with an odds ratio for mortality of 0.69 (95% CI, 0.56-0.84). Importantly, these systematic reviews and meta-analyses7,8 did not find any evidence of increased toxicity or other harms from prolonged infusions.
Given this evidence in favor of prolonged infusions, many hospitals have already implemented prolonged infusion of β-lactam antibiotics for patients being treated in high-risk units, for specific infectious syndromes, or even as a default hospital-wide dosing strategy. In 2021, the Surviving Sepsis Campaign updated its management guidelines to include a “weak” recommendation for prolonged infusion of β-lactam antibiotics (after an initial loading dose to avoid delays in achieving effective drug concentrations) over standard intermittent infusion based on “moderate-quality” evidence.9 However, questions remain about the true clinical benefit of prolonged infusions because most prior RCTs have been limited by small sample sizes, poor study quality (including open-label designs), and heterogeneous patient populations. Furthermore, the largest RCT to date enrolled 432 critically ill patients with sepsis and found no difference in outcomes with continuous infusion vs intermittent infusion.10
In this issue of JAMA, Monti and colleagues11 provide welcome new data from the Continuous Infusion vs Intermittent Administration of Meropenem in Critically Ill Patients (MERCY) trial that help address these gaps. The investigators performed a multinational, double-blind RCT that enrolled 607 patients admitted to an intensive care unit with sepsis or septic shock who were prescribed meropenem by their treating clinicians and randomized to continuous administration vs intermittent administration after a loading dose. Recruited patients were, on average, aged 60 years or older, white, predominantly male, and had a relatively low prevalence of diabetes, chronic kidney disease, and active cancer. The most common site of infection was the respiratory tract and nearly 70% of patients had a causative organism identified; Klebsiella species, Pseudomonas species, Escherichia coli, and Acinetobacter species were the most common gram-negative organisms.
Overall, there was no difference in the continuous administration group vs the intermittent administration group for the primary composite outcome of all-cause mortality and emergence of pandrug-resistant or extensively drug-resistant bacteria at day 28 (47% vs 49%, respectively; P = .60). Furthermore, there were no significant differences in any of the secondary outcomes examined, including 90-day mortality (42% in both groups), all-cause mortality at 28 days (30% in the continuous administration group vs 33% in the intermittent administration group; P = .50), or emergence of pandrug-resistant or extensively drug-resistant bacteria at day 28 (24% vs 25%, respectively; P = .70). The preplanned sensitivity analyses did not identify any subgroups in whom the primary outcome differed, including in patients with gram-negative organisms with high minimum inhibitory concentrations to carbapenem antibiotics and high vs low severity of illness.
The study was a well-conducted RCT that adds to the mixed and controversial literature on routine prolonged infusion of β-lactam antibiotics in patients with sepsis. Notable strengths include its rigorous double-blind design, multinational setting, and large sample size compared with prior trials. Furthermore, the trial enrolled severely ill patients (61% with septic shock, median Sequential Organ Failure Assessment score of 9) who, by virtue of altered pharmacokinetics, should stand to benefit the most from prolonged infusions. In addition, a sizeable proportion of patients had organisms with high minimum inhibitory concentrations to carbapenem antibiotics or intrinsically resistant pathogens like Pseudomonas and Acinetobacter, which is the patient population presumed to be most likely to benefit from continuous dosing. The study’s inclusion of new drug-resistant organisms as an outcome is also important given the theoretical but largely unproven possibility that prolonged infusions may decrease the emergence of antibiotic resistance.
At the same time, there are several aspects of the trial that still leave the door open to the possibility that prolonged infusion of β-lactam antibiotics may benefit some patients with sepsis. First, patients were hospitalized for a median of 9 days (continuous administration group) and 8 days (intermittent administration group) and in the intensive care unit for a median of 5 days (in both groups) before randomization, and most were already on antibiotics before a progressive or new infection triggered their treating teams to prescribe meropenem. Thus, in contrast to most sepsis trials that enroll patients with community-onset sepsis, many of these patients were admitted for other conditions that may have contributed to their poor prognoses and were less likely to be altered by the antibiotics given at the time of randomization.
Second, the results of this RCT cannot be extrapolated to prolonged infusion for all β-lactam antibiotics. Indeed, at least 1 meta-analysis12 suggests that there may be an agent-specific benefit for prolonged infusion, with the benefit limited to piperacillin-tazobactam but not to cephalosporin or carbapenem antibiotics.
Third, although this is the largest RCT to date on this topic, it may have been underpowered to detect small but still clinically meaningful results, particularly in the subgroup analyses. Fourth, although most patients in the trial had positive cultures during their hospitalization, only 10% had confirmed bloodstream infections. This is relevant given that the predictive value of positive cultures outside the bloodstream, particularly respiratory cultures, is relatively limited.13 Large, adequately powered trials focusing specifically on patients with definitive invasive infection (eg, gram-negative bacteremia) would be especially welcome.
Given these considerations, it is unclear whether the MERCY trial11 will influence clinical guidelines, particularly because there was no evidence of harm with prolonged infusion. With little to lose in terms of safety or cost, both clinicians and guideline authors will be hard-pressed to ignore the potential mortality reductions reported in previous meta-analyses. Indeed, the absolute 28-day mortality rate in the MERCY trial11 was lower in the continuous administration group vs the intermittent administration group, and thus adding this trial to an updated meta-analysis will likely bolster rather than refute the previously reported mortality benefits. Furthermore, the logistical barriers to prolonged infusions, such as limited intravenous access or nursing resources, are generally less of a concern in the intensive care unit setting.
In addition, the MERCY trial11 is unlikely to be the definitive trial on prolonged vs intermittent infusion of β-lactam antibiotics. The β-Lactam Infusion Group14 study (BLING III), currently nearing completion, aims to compare continuous piperacillin-tazobactam or meropenem vs intermittent infusion on 90-day all-cause mortality in 7000 critically ill patients with sepsis. Its inclusion of patients receiving piperacillin-tazobactam, which is more often used as first-line therapy in patients with sepsis, may facilitate enrolling patients admitted for sepsis earlier during the hospitalization, whose disease courses may be more modifiable through optimized use of antibiotics, and the large target sample size will afford superior power to detect small differences in the outcomes.
In summary, Monti and colleagues11 are to be commended for their well-designed and well-executed contribution to the controversial literature assessing dosing of β-lactam antibiotics in critically ill patients. Notwithstanding the negative results of this trial, guidelines and clinical practice may continue to favor prolonged dosing of β-lactam antibiotics when circumstances allow given the ongoing possibility of benefit and absence of harm. Meanwhile, there remains an urgent need to identify new interventions to realize the shared aspiration to reduce sepsis mortality and antimicrobial resistance.
