Antibiotic Target Attainment: What Are We Aiming for?*

In the 1979–2016 worldwide literature, sepsis was a leading cause of healthcare utilization, morbidity, and death in children, with an estimated 1.2 million cases of pediatric sepsis annually (1). Several years ago, the World Health Organization recognized pediatric sepsis as a global threat and set a goal to decrease mortality from sepsis by targeting improving prevention, diagnosis, and management strategies (2). Retrospective data from over 2700 critically ill adults with septic shock (1989–2004) showed that each hour delay in antimicrobial administration was associated with an average decrease in survival of 7.6% (3). As sepsis care has evolved since this 2006 publication, time to appropriate antibiotic administration remains a significant explanatory factor in models of survival for both adult and pediatric patients (4–6). More recently, however, the definition of “appropriate” antibiotic administration has begun to include not only matching the antimicrobial spectrum with suspected or confirmed pathogens but also sufficient dosing to achieve a pharmacologic response. The fact that critically ill patients experience alterations in population-based pharmacokinetic estimates and pharmacodynamic interactions has been known for at least 20 years (7). Extremes in volume of distribution and clearance, utilization of extracorporeal and drainage devices, and the increasing prevalence of obesity and drug–drug interactions are just some of the influences on the therapeutic response a patient may have to pharmacologic therapy (8). Until recently, clinicians had very limited ability to objectively evaluate the adequacy of their antimicrobial management. Indirect estimates of renal function reliant upon serum creatinine, matching antimicrobial selection to an in vitro susceptibility report, dosing recommendations likely developed in a dissimilar population, and following clinical response formed the backbone of antimicrobial management. A small number of antibiotics, namely vancomycin, and aminoglycosides, are routinely able to have therapeutic drug monitoring readily available, but these antimicrobials constitute a small percentage of overall antimicrobial utilization. Laboratories have begun to expand offerings for antibiotic concentration assays, and opportunities exist for many more antimicrobial agents, including beta-lactams and antifungal agents. Although the assays themselves are increasingly more available, they generally remain limited to a few laboratories as a “send out,” which can result in delays in reporting and being able to act on the information gained. In the article by Paice et al (9) in this issue of Pediatric Critical Care Medicine, the authors conducted a retrospective analysis of patients admitted to the PICU with severe sepsis who were prescribed meropenem and had blood available for opportunistic sampling to assay meropenem serum concentrations. Concentration-time profiles were then analyzed using modeling software and Bayesian estimation. The authors used time above four times the minimum inhibitory concentration (T > 4× MIC), with the MIC susceptibility breakpoint set to 1 µg/mL, as defined by the Clinical and Laboratory Standards Institute as susceptible for Enterobacteriaceae, and defined target attainment (TA) as 40% T > MIC 4 µg/mL for the first 24 hours of therapy. Individual clearance and volume of distribution estimates were calculated and compared between patients who were target attainers and nontarget attainers, in addition to fluid status and severe sepsis outcomes. The authors found that 8 of 29 enrolled patients were nontarget attainers and that meropenem clearance was the only association between TA and non-TA, but the analysis failed to identify an association with the estimated glomerular filtration rate (eGFR). Additionally, even though this study was not powered to adequately compare outcomes, the authors failed to identify differences in outcomes between TA and not. This study is small, but important as it highlights the complexities associated with defining antimicrobial therapy as “appropriate” and speaks to the pharmacokinetic variability seen between patients within the critical first 24 hours of sepsis treatment. Even defining TA is difficult, as there is a lack of data to support evidence-based outcomes. For example, the 2020 position paper on antimicrobial drug monitoring in critically ill adults (from various intensive care and infectious disease societies), noted that the optimal beta-lactam activity is 40–70% free (f) drug T > MIC, and that critically ill patients may benefit from longer (i.e., 100% fT > MIC) or higher (i.e., 2–5 × MIC) exposures (10). Ranges as broad as these may reflect the fact that we regularly use plasma serum concentrations as a substitute for concentrations at the true site of infection, noting that not all antimicrobials distribute evenly throughout the body. This makes research difficult, particularly in a heterogenous critically ill population, as the TA groups may have a target defined anywhere from 40% T > MIC to 100% T > 5 × MIC. In a review of common meropenem therapeutic drug monitoring practices, the most commonly adopted targets are 100% fT > MIC and 100% fT > 4 × MIC (11), which are different from the TA defined within the study by Paice et al (9). Studies describing outcomes for patients where therapeutic drug monitoring is performed have differing results, with some studies showing improvements in outcomes such as length of ICU stay in patients who achieve TA, whereas other studies have failed to find benefits (12). Interestingly, Cies et al (13) published a study where all 23 patients in an infection cohort failed to reach a beta-lactam TA of 40% fT > 4–6 × MIC due to subtherapeutic levels, and only four patients reached a TA of 40% fT > MIC, yet all patients in this infection cohort achieved a microbiologic response and 22 of 23 patients had a positive clinical response. These vast differences in outcome variables could reflect any number of differences between patient populations and care given, but could in part also be due to differences in defined TA goals. With the limitations most institutions have in obtaining antimicrobial drug concentrations, investigations such as that reported by Paice et al (9) are important for identifying our shortcomings when using clinical surrogates for dosing. The majority of meropenem clearance is via renal excretion as an unchanged drug (~70%), with a smaller percentage eliminated via hepatic metabolism (~28%) and feces (~2%), and while the investigators found that a higher clearance was associated with non-TA, the authors likely had too small a sample (or some other reason) to identify a correlated with eGFR. Of note, the study used serum creatinine as a biomarker for calculating the eGFR in this study. Serum creatinine has long been a gold standard biomarker for eGFR calculations as creatinine is freely filtered by the glomerulus with minimal tubular secretion, however, serum creatinine is reliant on muscle mass, and critically ill patients have not been included when developing creatinine-based eGFR calculations. More recently, serum creatinine has been found to be a poor estimate of eGFR in critically ill patients when compared with cystatin C or direct iohexol clearance (14). It would be important to then understand whether the increased clearance observed with non-TA was due to non-renal clearance, or if the use of serum creatinine-based eGFR is inaccurate enough in the critically ill population to precisely estimate renal function for the purposes of medication dosing. It can be hard to justify expanding laboratory, pharmacy, and physician resources to bring an increasing number of antimicrobial assays to local hospitals in the face of limited and conflicting data. However, there is mounting evidence that when therapeutic drug monitoring is employed in the critically ill, we often fail to reach TA. The impact of non-TA has yet to be fully explored and developed, and we will not be able to deepen that understanding until more clinicians have access to these assays in real-time. The complexity of the physiologic and pharmacokinetic changes occurring daily in the critically ill population should make most practitioners pause when assuming that medication therapy is adequate for the entirety of the treatment course. Frequently, an antibiotic is started and the course completed without ever changing the dose, yet the patient may have undergone fluid resuscitation, hemodynamic support, procedures, transfusions, de-escalation of resuscitation therapies, intubation/extubation, medication additions and subtractions, continuous tube feeds, etc. Generally, antibiotic dosing is adjusted only when the clinician suspects alterations in renal function, likely based on an inaccurate biomarker or the patient is placed on an extracorporeal therapy, where evidence has been published to suggest alternative dosing, or there is the ability to perform serum concentration monitoring. As sepsis management strategies evolve, clinicians need to be able to not only see the target but also understand which target to aim for.