Hi. My name is Kelly Pace. I'm a third-year pediatric critical care and first-year pediatric clinical pharmacology T32 fellow at Cincinnati Children's. Today I'll be talking to you about the work of myself and my colleagues investigating muripenem target attainment in early severe sepsis in children. I have no conflicts of interest to report. So we'll be reviewing the differences in pharmacokinetic parameters for patients with severe sepsis physiology, exploring how opportunistic sampling, population-based PK modeling, and Bayesian estimation can allow for generation of concentration time profiles to assess time above MIC, and we'll consider patient and clinical factors that may affect antibiotic target attainment in early pediatric severe sepsis. Pediatric severe sepsis is a significant cause of morbidity and mortality in children. The 2005 consensus criteria for pediatric severe sepsis defines severe sepsis as sepsis plus organ dysfunction, specifically cardiovascular dysfunction, acute respiratory distress syndrome, or two or more organ system failures. As of 2015, the Sprout study showed that around the world at any given time, about 8 percent of PICU patients meet criteria for severe sepsis, and the diagnosis carries a mortality rate of around 25 percent. The Surviving Sepsis Campaign 2020 updates emphasize the two key pillars of pediatric severe sepsis therapy, early antibiotics, specifically broad-spectrum antibiotics targeted at the likely organisms, and appropriate fluid resuscitation. These are the same bread-and-butter therapies we've been using to treat severe sepsis for decades, and at the bedside of a critically ill child, often we're quite literally pushing these two therapies at the same time during our resuscitation. But if these are our only two therapies, how do we know that we're using them effectively? Are we achieving effective antibiotic concentrations soon enough for our patients with severe sepsis, or even at all? And if we're not, what are the patient factors or medical management factors that contribute to treatment failure? And furthermore, what even are the antibiotic targets that we should aim for to give our patients their best chance at a good outcome? The work I'm presenting today attempts to begin to answer some of these questions. So I'd like to start by reviewing what defines an effective dose of antibiotics. For beta-lactam antibiotics, which are the most commonly used type of antibiotic in the pediatric ICU, efficacy is dependent on the amount of time that antibiotic concentrations remain above a critical threshold across each dosing interval. This critical threshold is called the MIC, or Minimum Inhibitory Concentration, and it's unique to each bacterial species, even to specific colonies of bacteria. This is an example of a beta-lactam concentration time profile. The amount of time that the blue curve stays above the dotted black MIC line is the amount of time that the antibiotic is actually working to kill bacteria. Unfortunately, on a patient-by-patient basis, because we don't routinely measure levels of beta-lactams in the PICU, we don't know which patients are meeting our treatment goals, or conversely, which patients are grossly exceeding them and at risk for toxicities. And on a population level, because severe sepsis can alter the way these curves look from what might be expected in a healthy patient, there's further uncertainty. So the pharmacologic milieu of severe sepsis in the pediatric population is unique because of changes in pharmacokinetics. Pharmacokinetics describes how medications are absorbed, distributed, metabolized, and cleared from the body. The two key parameters of pharmacokinetics are volume of distribution and clearance. So volume of distribution refers to how a drug is spread out into the various compartments of the body, and clearance describes how the drug is eliminated from the body, often via the renal or hepatic systems. Physiologic changes inherent to severe sepsis change these parameters. Volume of distribution could be decreased by dehydration or increased by a capillary leak or aggressive volume resuscitation. And clearance could be impacted by acute kidney injury, hepatic injury, or alternatively by augmented renal clearance driven by increased cardiac output. As you can imagine, when antibiotics were initially studied to understand and characterize these parameters, pediatric patients with severe sepsis were not the clinical trial participants. But despite all this, when it comes to antibiotic dosing, we treat most pediatric patients the same way, day after day of an illness course, despite obvious changes in physiology and phase of disease. So with all these physiologic changes inherent to critical illness and severe sepsis, we wanted to address the following questions. How often do PICU patients with early severe sepsis attain target antibiotic concentrations? And how does target attainment or failure impact clinical outcomes? For this work, we've chosen to focus on muropenem as the antibiotic of interest, because it's often used for patients at higher risk of infection with antibiotic-resistant organisms, and is needed for some of our most complex patients who are quite vulnerable to severe sepsis. So there's still some debate about what is the percentage of time above MIC we should aim for for beta-lactam therapy. Targets between 40 percent and 100 percent have been cited in various literature. Since aiming for 100 percent time above MIC is recommended for muropenem use in critical illness, this was the percentage benchmark that we aimed for in this study. As to selecting the specific minimum inhibitory concentration target, since culprit bacteria in early severe sepsis is often unknown, we chose enterobacter as the empiric organism and set our plasma target threshold as the intermediate MIC cutoff of two micrograms per milliliter for enterobacter species at our institution, in order to consider coverage for the potential of antibiotic-resistant organisms. So the goal of our study was to assess how often critically ill PICU patients with severe sepsis attained plasma and muropenem concentrations of at least two micrograms per milliliter for 100 percent of the first three dosing intervals of therapy, representing approximately the first 24 hours of therapy. Our hypothesis was that not all patients would achieve target muropenem concentrations, and that non-target-attaining patients would be more likely to have negative clinical outcomes. To address this goal and hypothesis, we turned to data collected from a prospective study of patients admitted to the PICU at Cincinnati Children's between 2018 and 2021 who were treated with one of four beta-lactam antibiotics. For this analysis, patients newly starting muropenem were eligible if they had severe sepsis, defined as requirement for at least seven days of antibiotic therapy, or death before seven days, and requirement within 24 hours of muropenem start of one or more of the following therapies, vasopressor greater than or equal to 40 mL per kilo of bolus fluid, or positive pressure ventilation above baseline. Patients were excluded if they were on extracorporeal support, or if muropenem was started during surgery due to differences in dosing strategy in these environments. And finally, patients had to have at least one plasma total muropenem concentration from one point within the first four dosing intervals of muropenem collected via opportunistic sampling. So what is opportunistic sampling? Opportunistic sampling refers to the use of leftover blood already collected from the patient for standard of care labs and monitoring in place of blood intentionally collected at specified times, but only drawn for research purposes. This opportunistic sampling method and the beta-lactam assays used in this work was previously validated in a study by my mentor, Dr. Sonja Tangerdwood. So here's an example of opportunistically collected muropenem concentrations for a patient in this study. As you can see, these individual patient drug concentrations don't really create a concentration time profile like I showed you earlier. And unlike some antibiotics whose efficacy can be measured by peaks or troughs alone, beta-lactam efficacy depends on overall time above MIC. So full concentration time profiles are required to calculate the time above MIC, but how do we do that with a collection of unrelated data points? For that, we turn to population-based pharmacokinetic modeling to fill in the blanks of the rest of the patient's first 24 hours of muropenem therapy. So here again are those patient-specific data points I showed for a patient in the study, the red dots reflecting actual measured concentrations of muropenem. Now shown by the dotted red line is an example concentration time profile generated for the patient based solely on a population model. So a population pharmacokinetic or POPPK model is a formula that uses known characteristics about a patient population, the medication dose and timing, and what is known about the pharmacokinetic behaviors of a drug to predict a more complete concentration time profile. For this study, we used a simplified version of a previously published POPPK model by Saito and colleagues from 2021 that was created based upon data from a group of critically ill patients in a Japanese PICU. But as close as this model group might be to the population of patients that we care for at Cincinnati Children's, the population-based PK model doesn't always well describe the pharmacokinetic milieu of any one individual patient. In this case, we were specifically looking at patients with severe sepsis, which, as we already discussed, can significantly change pharmacokinetic parameters from what might be expected in an otherwise critically ill patient but one without severe sepsis. And this is where Bayesian estimation comes into play. So in the third graphic, the solid red line, is the fitted concentration time profile, taking into account both the population PK model and the individual patient concentrations using Bayesian estimation. Bayesian estimation in a software like MWPharm++, which is what we used for this analysis, is a statistical technique that allows us to combine the individual characteristics of a patient and their actual known drug concentrations, along with a reasonable POPPK model, to predict forward and create individualized concentration time predictions for those data points we were unable to collect. In this case, the patient-specific curve looks fairly similar to the model prediction, but as you can see, the points fit better with the Bayesian estimated curve. As you can imagine, there are patients in any ICU that are not quite like the others, and Bayesian estimation allows us to see how they might stray from the overall population's behavior. These are the concentration time profiles that we used to allow for the calculation of percent time above MIC for this study. And in this analysis, I looked at target attainment over the first three dosing intervals of newly started mirapenem. So 28 patients were included in this analysis. They had a median age of four years and a median weight of 16.3 kilograms. 64 percent were male and 36 percent were female. You can see here how patients met severe sepsis criteria for inclusion. These categories were not mutually exclusive, and patients could meet severe sepsis criteria by requiring multiple of these therapies. So after reviewing the modeled concentration time profiles, we found that 16 out of 28 patients, or 57 percent, achieved the goal of staying above two micrograms per milliliter for 100 percent of their first three mirapenem dosing intervals. And thus, we simultaneously found that 12 out of 28, or 43 percent of patients, did not meet mirapenem concentration goals. Next, we looked at the association of time above MIC target attainment with patient outcome measures via students T. Fisher's exact or a Mann-Whitney rank sum tests. Hospital and PICU length of stay were not associated with target attainment. Though there was no statistically significant difference in 28-day mortality between the attaining and non-attaining groups, all four patients who died in the cohort were in the target attaining group. This was a surprising finding, which ran counter to our initial hypothesis that insufficient antibiotic target attainment would be associated with worse outcomes. Next, we looked at the association of time above MIC with volume of distribution and allometric clearance via the same tests. We found that target attainment was associated with lower allometric clearance. Given that lower clearance would allow mirapenem to stay in the plasma for longer, allowing for a buildup of higher concentrations, this made clinical sense. Target attainment, however, was not associated with a difference in volume of distribution. And finally, the target attaining group had higher PRISM-3 scores than the patients who did not attain target levels, in line with their higher mortality rate and lower clearance. Those findings are perhaps reflective of this group's more significant overall state of illness and organ dysfunction. So, in conclusion, we found that early in therapy, children with severe sepsis did not consistently achieve mirapenem concentrations sufficient to treat resistant bacteria. And lower clearance and higher PRISM scores may reflect organ dysfunction, resulting in a higher likelihood of mirapenem target attainment for some patients. So, where do we go next in thinking about antibiotic pharmacokinetics and pharmacodynamics in early pediatric severe sepsis? Questions that myself and other members of our team are considering include, what is the impact of fluid resuscitation on early antibiotic target attainment in severe sepsis? What are the early targets that affect patient outcomes in severe sepsis? And is there a role for model-informed precision dosing for beta-lactam antibiotics for pediatric patients with severe sepsis? I'd like to take a moment to acknowledge my co-authors and collaborators for their help in this work, and the many others who have assisted me with my research development throughout fellowship. Thank you very much.

meropenem target attainment

severe sepsis in children

antibiotic therapy

pharmacokinetics

pediatric patients