Pharmacokinetic Research in Pediatric Extracorporeal Therapies: Current State and Future Directions Free

Extracorporeal life support (ECLS), which includes extracorporeal membrane oxygenation (ECMO) and continuous renal replacement therapy (CRRT), can be lifesaving in children with refractory organ failure. Despite technological advances, mortality remains high in children receiving either of these therapies [1, 2]. One contributor to the high mortality in this population may be altered drug disposition. Pharmacokinetics (PK) is defined as what the body does to a drug via absorption, distribution, metabolism, and elimination processes [3]. Drug PK in children receiving ECLS is altered due to drug-circuit interactions, physiologic changes associated with human-circuit interactions, and underlying critical illness [4]. Despite broadly known PK changes in this patient group, optimal dosing for most drugs is unknown in this population.

ECMO is a potentially lifesaving therapy in children with severe respiratory and/or cardiac failure. It is a voluminous device, minimally including a gas exchange device (oxygenator), blood pump, and heat exchanger connected by circuit tubing. Patients receiving ECMO support experience PK alterations due to the device itself as well as the underlying illness. Adsorption of drugs to the circuit is commonly seen with highly lipophilic and protein-bound medications such as fentanyl and propofol, leading to an increase in apparent volume of distribution (Vd) [5]. The apparent Vd may also increase secondary to the inflammatory reaction and capillary leak triggered by ECMO. This effect may be further accentuated by resuscitation fluids or if repeated blood product transfusions are administered. Due to this resulting increase in apparent Vd, patients may subsequently require higher drug doses to achieve the desired effect. In contrast, inflammation contributes to hypoalbuminemia [6]. Lower albumin values translate into a larger free fraction of drug and may lead to greater effect or toxicity, particularly for highly protein-bound medications. Acute kidney injury (AKI) during ECMO is common and may require CRRT, further altering drug disposition [7].

CRRT is the renal support method recommended for clinically unstable patients due to its relative hemodynamic stability compared to intermittent methods [8]. The circuit consists of a blood pump, filter, and may include different combinations of dialysis and replacement fluids. Multiple modalities exist under the umbrella of CRRT, including continuous veno-venous hemofiltration, continuous veno-venous hemodialysis, and continuous veno-venous hemodiafiltration, with each of these modalities providing solute clearance via diffusive and/or convective mechanisms. Drug clearance is the PK parameter primarily affected by CRRT. Hydrophilic drugs (i.e., smaller Vd) with lower molecular weights and minimal protein binding are most likely to be efficiently cleared by CRRT [9, 10]. CRRT circuit characteristics may also affect drug PK, with effluent flow rate primarily affecting efficiency of drug clearance [11]. Additionally, neonates and small infants may have an increased Vd due to circuit priming volumes in excess of 10–15% of their total blood volume [12]. Finally, while generally not to the extent seen with ECMO, drug adsorption to CRRT membranes may also occur and be clinically significant depending on the drug-filter combination [13].

While outside the scope of this position paper focusing on ECMO and CRRT, other ECLS therapies, such as extracorporeal liver support, are also subject to PK alterations. PK alterations vary according to the ECLS method employed and cannot be extrapolated from one method to another. Drug PK during ECLS therapies other than ECMO and CRRT should be considered in clinical practice and better characterized in future research.

The Pediatric Paracorporeal and Extracorporeal Therapies Summit (PPETS) was organized with the goal of bringing together an international group of experts in the fields of ECMO, CRRT, extracorporeal liver support, ventricular assist devices, and apheresis. The multinational multidisciplinary group of adult and pediatric experts planned to address the existing gaps in standardizing practices, conducting research and quality improvement initiatives, and fostering innovation across different clinical disciplines. The summit aimed to transform these gaps into opportunities for knowledge and research by discussing the current state of these therapies, disseminating innovative support strategies, facilitating the exchange of experiences, and encouraging future collaborations. This meeting was hosted by the Baylor College of Medicine and supported by a grant from the Eunice Kennedy Shriver National Institute of Child Health & Human Development (1R13HD104433-01). There were six domains identified: outcomes, standardization and quality, PK, ethics, innovation, and combined circuits. In this article, we review the proceedings of the PK domain of PPETS 2022 and recommend a path forward to guide PK research in pediatric ECMO and CRRT.

Children on ECMO are exposed to many different drugs. In a recent retrospective single-center study, 254 children received 179 unique medications during their ECMO course [14]. Cumulative drug exposure increased with longer ECMO duration, with a median exposure to 31 different drugs in the 5 weeks following cannulation. Unfortunately, few studies focus on drug PK during pediatric ECMO. Of the 50 most received medications in the above study, only 40% had been studied in children on ECMO. Sutiman et al. [15] performed a systematic review highlighting the current evidence regarding PK alterations during pediatric ECMO. A total of 41 studies evaluating 23 unique medications were included in the analysis. Antimicrobials (32%) were the most frequently studied drugs, with infants <1 year old heavily represented (68% of studies) and significant PK alterations identified in 70% of the drugs evaluated. Most studies showed increased Vd, while the effect of ECMO on clearance was inconsistent with a general trend toward reduced clearance in renally excreted medications and increased clearance in highly metabolized medications. While decreased clearance of renally excreted drugs is predictable considering the high prevalence of AKI in ECMO patients, increased clearance of metabolized drugs is not as intuitive. It has not been well characterized and could be explained by a combination of factors, notably drug adsorption to the circuit, as these are often lipophilic drugs that tend to adhere to the circuit components [5, 16, 17]. It is essential to note that the heterogeneity of published studies and frequent lack of a non-ECMO comparator group limit the external validity of these findings. Population and ECMO characteristics such as patient age, diagnosis, ECMO modality, circuit componentry, and circuit priming impact PK and vary across studies. Comparison with historical cohorts, which may not represent comparable critically ill patients, impedes our ability to identify changes specifically attributed to ECMO. These limitations, in addition to the small number of studied drugs, restrict our capacity to predict medication PK at the bedside adequately and adjust doses accordingly.

Much like those receiving ECMO support, children on CRRT are exposed to numerous drugs. A retrospective cohort study utilizing the Pediatric Health Information System (PHIS) database analyzed 2,738 patients with AKI requiring renal replacement therapy from 2007 to 2011. In this study including both intermittent and CRRT, patients were exposed to a median of 18–23 unique drugs on day 1 of renal replacement therapy, while the cumulative number reached 41–52 unique drugs by day 15 [18]. Another retrospective cohort study also utilized the PHIS database to quantify commonly used antimicrobials in pediatric CRRT. From July 2018 through June 2021, the authors identified 77 unique antimicrobial agents administered during 812 admissions requiring CRRT [19]. In a recent systematic review of published studies from 1990 to 2021 focusing on drugs administered during pediatric CRRT, Dubinsky et al. [20] identified 45 relevant studies containing data on 33 unique medications, with 29 (64%) studies providing dosing recommendations. Antibiotics represented the most frequently studied drug class (42%), followed by antivirals (12%), antifungals (12%), vasoactives (12%), and anticonvulsants (6%). Most studies were case reports/series (42%), followed by prospective studies (36%). Similar Vd and total drug clearance were most frequently reported between children who were and were not receiving CRRT. However, studies including a comparator found an increase in total clearance, particularly in patients with residual renal function. More than half (53%) of the studies included at least 1 patient receiving ECMO support, highlighting the frequent overlap of extracorporeal therapies.

Optimizing dosing during pediatric ECMO and CRRT is challenging. Limited data are available to guide clinicians and more studies are needed. However, studying every possible combination of medications, specific populations, available circuits, and ECMO/CRRT flow rates is nearly impossible. Therefore, we hereby propose a rational approach to address the current gaps in research and improve application of knowledge to clinical care.

First, we must prioritize the drugs that need more in-depth studying. Based on a combination of frequency of use, potential for negative outcomes if dosed inappropriately, existing PK data, and lack of rapidly measurable clinical effect (e.g., change in blood pressure, clinically implemented therapeutic drug monitoring), we have proposed lists of medications for which research should be prioritized in the pediatric ECMO and CRRT populations (Tables 1, 2). Usage data for medications in the ECMO and CRRT populations most often come from single-center studies, so consideration should also be given to drugs in similar classes as those included in the prioritized lists (e.g., additional aminoglycosides and third-generation cephalosporins). The PK of individual constituent drugs in combination products may not be affected equally by ECMO or CRRT, so it is important that the PK of all components of combination products be studied. Additionally, medications used in subpopulations such as patients with oncologic or autoimmune diseases are often large molecules that have not been studied in ECLS, so potential drug-circuit interactions are unknown. Thus, special consideration should be given to including these agents in PK studies.

Second, we must reconsider our study designs. Single-center population PK studies are time-consuming and limited by on-site resources, with potentially ungeneralizable findings. Multicenter studies allow sharing of resources, facilitate recruitment of more participants, and introduce essential sources of variability to improve generalizability. The ASAP ECMO study is an example of successful collaboration with a single protocol leading to adult PK models for 18 drugs [21]. Efforts to facilitate recruitment in pediatric PK research should also include assays using minimal blood volume. Strategies may include opportunistic sampling, scavenged samples, microsampling, sampling drug concentrations in spent CRRT effluent to avoid the need for phlebotomy, and even novel measurement techniques such as minimally invasive biosensors and noninvasive wearable electroactive pharmaceutical monitors [22‒26]. Physiologically based PK is an alternative mechanistic modeling approach particularly well suited to studying extracorporeal PK as it allows ex vivo data to inform the ECMO or CRRT compartment, can predict drug concentration at the target site, and has the flexibility to adapt to different ages and disease states [27].

Third, we must seek better translation from models to the bedside and integrate validated models into clinical practice. Implementation of model-informed precision dosing, where patients’ characteristics and dosing history are utilized in prediction models to determine optimal dosing regimens, can be facilitated by an intuitive interface built into the institutional electronic health record [28]. Finally, once a model is implemented in clinical practice, automated continuous learning approaches can be used to improve the model [29].

Several lessons may be drawn from adult experiences to address PK knowledge gaps in pediatric ECLS. First, it is important not to assume that the PK of medications in patients with acute and chronic renal failure is the same. Mueller et al. [30] demonstrated that imipenem clearance in patients with AKI is nearly double that of patients with chronic kidney disease (95 ± 13.8 vs. 51.1 ± 10.5 mL/min, p < 0.02). This difference represents a change in nonrenal clearance in patients with AKI, a phenomenon also demonstrated with meropenem and vancomycin [31]. Second, when performing ECLS studies, there can be many potential sampling sites on the circuit, each with its own advantages and disadvantages and likely demonstrating different drug concentrations at any given time. Careful pre-study planning to precisely define the study’s goals while considering ethical aspects in a population where limited phlebotomies may be allowed is key. Finally, accounting for ECLS circuit age is important as the degree to which they alter PK parameters may change over time [32‒34]. This may be due to saturable adsorption of drugs within the circuit elements, a well-known issue in ECMO that has also been demonstrated for clinically relevant drugs in CRRT [13], as well as the formation of proteinaceous second membranes on the CRRT filter itself which limits solute clearance [32].

Children receiving extracorporeal support represent a uniquely vulnerable and understudied population in which PK research has the potential to make an outsized impact on clinical outcomes. A multifaceted approach to studying key medications in this population is vital to tackle this issue. Here, we have set forth a proposed list of medications in which to focus early efforts and key methodological considerations to maximize the impact of future research.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

G.S. was supported by a T32 training grant from NIGMS/NICHD (T32GM008562). PPETS was supported by an R13 scientific conference grant from NICHD (1R13HD104433-01).

G.S. and C.T. contributed to study design and co-wrote the manuscript. B.A.M., A.A.A., and K.M.W. contributed to study design and provided important intellectual contributions to the manuscript. J.J.C. and J.M.D. provided critical comments on the manuscript and revised it accordingly.