Impact of extracorporeal membrane oxygenation on antimicrobial exposure in critically ill patients: a prospective observational study

Abstract

Background

Antimicrobial pharmacokinetics (PK) in patients receiving extracorporeal membrane oxygenation (ECMO) remains poorly defined. Drug sequestration within the ECMO circuit has been reported; however, existing evidence is primarily based on ex vivo studies and small clinical series, resulting in conflicting findings with limited clinical applicability. This study aimed to characterize antimicrobial sequestration within different ECMO circuit components and to describe the PK of meropenem, piperacillin/tazobactam, linezolid, and ceftaroline in critically ill patients receiving ECMO support.

Methods

We conducted a prospective, single-center observational study in critically ill patients receiving ECMO (veno-venous (VV) or veno-arterial (VA)) and treated with meropenem, piperacillin/tazobactam, linezolid, or ceftaroline. Serial blood samples were collected simultaneously from the patient’s arterial line and from pre- and post-membrane oxygenator sampling sites. Non-compartmental analysis was used to derive PK measures, and concentration differences across sampling sites were analyzed to estimate circuit-related drug sequestration. Antimicrobial target attainment was assessed according to established PKPD targets.

Results

A total of 237 samples from 8 patients (4 VV, 4 VA) were analyzed. Limited, component-specific sequestration was observed in the membrane oxygenator for meropenem (9.28% ± 21.6%, p = 0.046) and ceftaroline (14.6% ± 17.5%, p = 0.010), for linezolid it was also identified (10.1% ± 12.3%, p = 0.005), but in tubing and connectors. Piperacillin/tazobactam showed negligible retention. However, when considering the ECMO circuit as a whole, no statistically significant reductions in systemic concentrations were observed for any antimicrobial. PK measures were generally consistent across sampling sites. In 20% of treatments, predefined PKPD efficacy targets were not achieved.

Conclusions

Antimicrobial sequestration within the ECMO circuit is detectable and drug- and component-specific. However, overall impact on systemic antimicrobial exposure appears limited. Inadequate PKPD target attainment in a subset of antimicrobial treatments likely reflects the marked PK variability of critical illness rather than ECMO-related drug loss alone. These exploratory results support the use of therapeutic drug monitoring to individualize antimicrobial therapy in ECMO-supported patients and the need to conduct further population PK studies to better characterize determinants of antimicrobial exposure and to clarify the clinical relevance of circuit-related drug sequestration.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-026-06076-9.

Background

Extracorporeal membrane oxygenation (ECMO) is an advanced form of mechanical cardiopulmonary support used in critically ill patients with refractory cardiac and/or respiratory failure [1]. Data from the 2022 Extracorporeal Life Support Organization (ELSO) Registry confirm a steady global rise in ECMO utilization [2], underscoring the urgent need to better understand its clinical and pharmacological implications.

Although life-saving, ECMO introduces substantial complexity to the pharmacological management of critically ill patients. The ECMO circuit comprises several components—including cannulas, a blood pump, an oxygenator, and tubing—all of which add extracorporeal volume and provide a large surface area where drugs can be sequestered or adsorbed, potentially affecting the drug pharmacokinetics (PK) [3–5]. The oxygenator, which enables gas exchange and is typically composed of a polymethylpentene (PMP) membrane, represents a major site for drug adsorption, particularly for lipophilic or highly protein-bound drugs [3–5]. Tubing, generally made of polyvinyl chloride (PVC) and often coated with heparin or other biocompatible materials to reduce thrombogenicity and immunogenicity, may also affect drug binding and fluid dynamics [3, 5]. Additionally, circuit-related factors—including circuit age, tubing design, and pump type—can further modulate the extent of drug sequestration [3, 5]. Collectively, these circuit characteristics, combined with drug-specific properties such as lipophilicity—commonly expressed as the octanol–water partition coefficient (log P), with values typically ≥ 1— and high protein-binding (generally > 80%) [6], can significantly alter drug PK, leading to increased apparent volume of distribution (Vd), reduced systemic clearance (CL), and prolonged elimination half-life (t_1/2) [7].

Of particular concern is the impact of ECMO on antimicrobial PK. Antimicrobial therapy is guided by nonspecific surrogate markers, and inadequate dosing can compromise the achievement of therapeutic targets while promoting antimicrobial resistance. Infections are prevalent in patients receiving ECMO, especially in those supported with veno-venous (VV) ECMO, representing both a frequent indication for support and a major source of complications during treatment. Notably, a systematic review and individual patient data meta-analysis—including the CESAR and EOLIA trials, both conducted in patients with VV-ECMO—reported that pneumonia accounted for over 60% of cases [3, 8].

Given the strong association between subtherapeutic antimicrobial exposure and increased mortality in critically ill patients [9], optimizing dosing regimens remains critical. A recent expert consensus has identified both PK/pharmacodynamics (PD) reporting and antimicrobial dosing as key research priorities in ECMO pharmacology, underscoring the need for high-quality clinical PK studies in this field [10]. Several systematic reviews have synthesized the available data and proposed dosing strategies in ECMO patients. However, current evidence is largely limited to small observational studies and in vitro orex vivo studies, often based on single-dose experiments in closed circuits that do not account for the metabolic and physiological complexities of critical illness [4, 11, 12]. In addition, substantial methodological heterogeneity across studies—including differences in study design, sampling strategies, and patient populations—further limits the external validity of the findings [13]. Consequently, the available evidence remains inconsistent and of limited clinical applicability.

Accordingly, there is currently a lack of robust prospective studies evaluating clinical outcomes, as well as validated, evidence-based antimicrobial dosing guidelines for patients receiving ECMO.

Beyond circuit-related effects, critical illness is associated with profound pathophysiological changes that independently alter drug PK, including decreased plasma protein-binding, increased Vd, and altered drug CL [14]. The combined impact of these patient-specific changes and ECMO circuit factors can markedly compromise drug exposure, increasing the risk of subtherapeutic concentrations and treatment failure. Consistently, significant PK alterations have been reported for analgesics, antibiotics, antiepileptics, and sedatives in patients receiving ECMO support [15].

Therefore, this study aimed to characterize antimicrobial sequestration in the membrane oxygenator and to describe the PK of meropenem, piperacillin/tazobactam, linezolid, and ceftaroline in critically ill patients undergoing ECMO support. Secondary objectives were to evaluate overall drug loss within the entire ECMO circuit and to characterize antimicrobial sequestration in other ECMO components, such as tubing and connectors. The results of this exploratory analysis would provide preliminary real-world in vivo PK data on multiple antimicrobials in critically ill patients receiving ECMO.

Materials and methods

Study population

We conducted a prospective, observational, single-center study in five different Intensive Care Units (ICUs) within the Hospital Clínic of Barcelona, a Spanish tertiary university hospital. All ICUs are geographically co-located within the same hospital and follow standardized institutional protocols, ensuring harmonized clinical management across units, including ECMO support and related therapies. The study was approved by the local Ethics Committee (HCB/2024/0574) and was performed in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients, or their legal representatives when applicable, prior to study inclusion, in accordance with local ethical regulations.

Inclusion criteria comprised critically ill patients aged ≥ 16 years admitted to the ICU, receiving ECMO support, either VV or veno-arterial (VA), and treated with at least one of the study antimicrobials: meropenem, piperacillin/tazobactam, linezolid or ceftaroline. Exclusion criteria included pregnancy, hypersensitivity to any of the study drugs or their excipients, and patients who had undergone massive blood transfusion (> 50% of total blood volume within the previous 8 h) and/or plasma exchange in the 24 h prior to PK sampling.

Patients could contribute more than once to sample collection, according to the antimicrobial received while on ECMO. The minimum sample size was set at 3 patients for each antibiotic, based on the limited number of patients meeting all inclusion criteria at our center and the requirement to obtain 8–12 samples per patient to ensure sufficient data for analysis.

Demographic and clinical data

Patients’ baseline characteristics were obtained from the hospital’s electronic clinical records. Data included: (i) demographic characteristics (age, sex, total body weight, height, body mass index [BMI]); (ii) clinical biochemistry measurements such as serum creatinine, creatinine clearance (CLCR) estimated with the Cockcroft–Gault formula [16], and serum albumin; (iii) use of renal replacement therapy (RRT), including intermittent hemodialysis or continuous renal replacement therapy (CRRT); (iv) ECMO-related parameters, including ECMO configuration [VV or VA], device type (Rotaflow or CardioHelp), fraction of delivered oxygen (FDO₂), sweep gas flow, blood flow rate, centrifugal pump settings, days under ECMO support, oxygenator membrane age (days in use at sampling), and presence of fibrin or clots at the time of sampling; (v) antimicrobial therapy details, including administered dose, dosing frequency, and time of infusion; (vi) infection-related information, including infection type and microbiological isolates; and (vii) clinical outcomes, such as ICU length of stay, total hospitalization duration, and mortality at discharge.

Antimicrobial administration and blood sampling

Eligible subjects received meropenem (administered either as a 1-hour or 4-hour infusion), piperacillin/tazobactam (30-minute or 3-hour infusion), linezolid (1-hour infusion) or ceftaroline (1-hour infusion), all of them administered via a central venous catheter. Antibiotics were prescribed based on the clinical judgment of the attending physician, with dose adjustments according to renal function or RRT, following the Summary of Product Characteristics (SmPC) [17–20] and expert recommendations [21–23], including current antimicrobial treatment guidelines [24], when appropriate. Serial blood samples (4 mL collected in EDTA tubes) were obtained simultaneously, within a time difference of less than 5 min, from three sites at each sampling time: pre- and post-membrane oxygenator and patient’s arterial line (Fig. 1). This short time interval between the three sampling sites reflects routine clinical practice, as all samples were obtained sequentially by the same healthcare professional. Regarding the sampling sequence, pre-membrane samples were consistently collected before post-membrane samples. Arterial line samples were collected either before the pre-membrane sample or after the post-membrane sample, depending on the clinical workflow. Sampling times were precisely recorded. Samples were collected over a complete dosing interval under steady-state conditions, defined as the PK state in which drug input and elimination are balanced, so that plasma concentrations remain consistent between dosing intervals, and is typically achieved after approximately five drug half-lives. Sampling included a pre-dose sample—taken immediately before drug administration—and multiple post-dose timepoints scheduled according to the dosing regimen and mode of administration of each antimicrobial, with an average of 8–12 samples per patient for each antimicrobial treatment. Blood samples were immediately centrifuged, and plasma was stored at − 80 °C until analysis.

Fig. 1.

Sampling points used in the study: (1) pre-membrane oxygenator; (2) post-membrane oxygenator; and (3) patient’s arterial line. Created with BioRender (https://biorender.com)

For intermittent infusions (0.5–1 h), samples were typically collected at pre-dose (0 min), at the end of infusion, and at different additional post-infusion time points (2–12 h) to adequately characterize both the distribution and elimination phases. Sampling was adapted according to the dosing interval of each antimicrobial.

For extended infusions (3–4 h), sampling included pre-dose, end-of-infusion, and additional post-infusion time points (5–12 h) to capture both distribution and elimination phases. Sampling schedules were similarly tailored to the dosing interval of each drug.

Analytical method

Antimicrobial plasma total concentrations were determined in the Laboratory of Pharmacology and Toxicology, CDB, Hospital Clinic, using a validated ultra-high-performance liquid chromatography method coupled with tandem mass-spectrometry (UHPLC-MS/MS) [25] following the manufacturer’s instructions. Calibration was performed using the ClinMass^® TDM Platform (RECIPE). The calibration range was 1.7–114 mg/L for meropenem; 1.52–102 mg/L for piperacillin; 0.48–28.6 mg/L for tazobactam; 0.49–28.6 mg/L for linezolid; and 1–80 mg/L for ceftaroline. The lower limit of quantification (LLOQ) was 1.7 mg/L for meropenem; 1.52 mg/L for piperacillin; 0.48 mg/L for tazobactam; 0.49 mg/L for linezolid; and 1 mg/L for ceftaroline. According to the manufacturer’s performance data, intra- and inter-assay imprecision for meropenem, piperacillin, tazobactam and linezolid was ≤ 14%, expressed as coefficient of variation (CV), across three quality control (QC) levels.

Pharmacokinetic analysis

PK measures were derived using non-compartmental analysis (NCA), considering the time of infusion. Calculated PK measures included maximum plasma concentration (Cmax), time to maximum concentration (Tmax), area under the plasma concentration–time curve from 0 to 24 h (AUC_0–24), elimination rate constant (kel), total body CL, Vd, and t_1/2, based on plasma concentrations measured at the three different sample points.

To evaluate drug sequestration within the ECMO circuit, estimated absolute differences in plasma concentrations between sampling sites were analyzed. Specifically (i) the difference between pre-membrane and post-membrane concentrations (Cpre – Cpost) was used to estimate drug sequestration by the membrane oxygenator; (ii) the difference between post-membrane and arterial line concentrations (Cpost – Cart) was used to estimate drug retention or loss in other ECMO circuit components, such as tubing and connectors; and (iii) the difference between pre-membrane and arterial line concentrations (Cpre – Cart) was considered a proxy for overall drug loss across the entire ECMO circuit.

In addition, the percentage of estimated drug sequestration was calculated for each comparison to normalize absolute concentration differences relative to the corresponding reference sampling site, thereby providing a quantitative estimate of the proportion of drug loss or retention within specific ECMO circuit components and across the entire circuit. Percentage sequestration was calculated as the absolute concentration difference divided by the reference concentration, multiplied by 100.

Dosing adequacy and PKPD targets

Antimicrobial dosing regimens’ adequacy was assessed by evaluating PKPD target attainment based on current clinical evidence [13]. For β-lactams, including meropenem, piperacillin, and ceftaroline, efficacy was achieved when free drug concentrations remained above the minimum inhibitory concentration (MIC) for 100% of the dosing interval (%fT > MIC). For tazobactam, administered in combination with piperacillin, efficacy was evaluated using the β-lactamase inhibitor target of 80–85% fT > critical threshold concentration (CT), applying a CT of 2 mg/L for Enterobacteriaceae isolates [26–28]. Free drug concentrations were estimated from total plasma concentrations using plasma protein-binding values reported in the Summary of Product Characteristics for each antimicrobial [17–19]: meropenem 2%, piperacillin 30%, tazobactam 30% and ceftaroline 20%. Linezolid efficacy was determined by total trough concentrations (Cmin) exceeding 2 mg/L and achieving ≥ 85% T > MIC, and an AUC₀₋₂₄/MIC ratio between 80 and 120. When isolate-specific MIC values were unavailable or in patients without microbiological confirmation, the EUCAST clinical breakpoints were used as surrogate thresholds [29]. Accordingly, we applied the Pseudomonas aeruginosa MIC of 2 mg/L for meropenem and 16 mg/L for piperacillin, and the Staphylococcus aureus MIC of 1 mg/L for ceftaroline and 2 mg/L for linezolid [29].

Clinical PKPD toxicity thresholds were defined as follows: meropenem, free Cmin > 44.5 mg/L; piperacillin, free Cmin > 361 mg/L; ceftaroline, free Cmin > 20 mg/L; and linezolid, total Cmin > 7 mg/L [13]. No validated PKPD toxicity threshold has been established for tazobactam.

Statistics

Descriptive and PK analyses were performed using R statistical software (version 4.4.2; R Core Team, 2025). NCA was conducted using the PKNCA package in R (version 0.12.1; Denney et al., 2025). Categorical variables are presented as counts (percentage), while continuous data are expressed as mean ± standard deviation (SD) or median with interquartile range (IQR), as appropriate.

Comparisons of continuous data across three or more groups were performed using ANOVA or the Kruskal–Wallis test, depending on data distribution. Paired comparisons of pre-membrane, post-membrane, and arterial line concentrations were assessed using a paired t-test or the Wilcoxon signed-rank test, as appropriate. A p-value < 0.05 was considered statistically significant in all analyses.

Results

Baseline clinical characteristics

A total of 237 blood samples were collected from 8 critically ill patients undergoing ECMO support, corresponding to 15 antimicrobial treatments. The median age was 58 years (47–69), and 62.5% (5/8) were male. Four patients (50%) received VV-ECMO, all for acute respiratory distress syndrome (ARDS), and four patients (50%) received VA-ECMO, primarily for cardiac arrest (75%, 3/4) and cardiogenic shock (25%, 1/4). Patients’ median SOFA score was 11(7–15), and the median APACHE II score was 27 (18–37). At the time of sample collection, the median duration of ECMO support was 3 days (3–10), and the median duration of oxygenator membrane was 3 days (3–5). Pneumonia was the primary indication for antimicrobial therapy, accounting for 87.5% (7/8) of cases. Additionally, 3 patients (37.5%) were also under CRRT; in all three cases, CRRT was integrated into the ECMO circuit in series, functioning as an additional extracorporeal clearance step. One patient (33.3%) received continuous venovenous hemodiafiltration (CVVHDF) and two patients (66.7%) received continuous venovenous hemodialysis (CVVHD). The median effluent flow rate was 32.5 mL/kg/h (22.5–32.5). CRRT was maintained continuously without interruption during the PK sampling period.

A total of 93 meropenem plasma concentrations were obtained from six patients. In addition, 45 piperacillin/tazobactam, 51 linezolid, and 48 ceftaroline plasma concentrations were analyzed, corresponding to three patients for each antimicrobial. One patient received three antibiotics, five patients received two antibiotics, and two patients received a single antibiotic, corresponding to 15 antimicrobial treatments. Patients’ demographic and clinical characteristics, stratified by antimicrobial agent, are summarized in Table 1.

Table 1.

Demographic and baseline clinical characteristics, ECMO parameters and patient outcomes stratified by antimicrobial. One patient received three antibiotics, five patients received two antibiotics concomitantly, and two patients received a single antibiotic

	Meropenem n = 6	Piperacillin/ Tazobactam n = 3	Linezolid n = 3	Ceftaroline n = 3
Demographics and clinical characteristics
Age (years)	53 (41–63)	53 (41–65)	53 (16–63)	72 (41–73)
Male	4 (66.7)	3 (100)	1 (33.3)	2 (66.7)
Weight (kg)	92.5 (80–100)	55 (49.5–95)	100 (90–110)	48 (40–80)
Height (cm)	170.5 (163–180)	175 (155–184)	164 (163–177)	155(154–180)
BMI (kg/m2)	30.0 (24.7–33.9)	20.6 (18.0-28.1)	33.9 (31.9–40.9)	20.2 (16.6–24.7)
CRRT	2 (33.3)	0 (0)	1 (33.3)	2 (66.7)
Serum creatinine (mg/dL)*	0.64 (0.46–1.3)	1.28 (0.42–1.66)	0.46 (0.43–0.49)	1.41
CLCR (CG) (mL/min)*	157.1 (81.3-227.5)	69.2 (44.8- 162.1)	227.5 (215.0-240.1)	26.9
Serum albumin (g/L)	28 (26–35)	30 (29–35)	29 (27–35)	27 (26–29)
Vasopressors	3 (50)	2 (66.7)	1 (33.3)	2 (66.7)
SOFA	11 (7–16)	14 (10–16)	7 (6–12)	16 (5–17)
APACHE II	27 (19–36)	37 (21–42)	19 (10–33)	36 (17–42)
ECMO parameters
Type
VV  VA	3 (50) 3 (50)	1 (33.3) 2 (66.7)	2 (66.7) 1 (33.3)	2 (66.7) 1 (33.3)
Device type
Rotaflow  CardioHelp	2 (33.3) 4 (66.7)	3 (100)	3 (100)	2 (66.7) 1 (33.3)
FDO2 (%)†	80 (60–90)	50 (40–100)	70 (50–100)	90 (40–100)
Sweep gas (L/min)†	3 (1.5–3.5)	1 (0-3.5)	2.5 (1.5-4)	2 (2-3.5)
Blood flow (L/min)†	3.3 (2.6–4.05)	3.4 (3-3.55)	4.05 (3.3–5.5)	3.3 (3-4.2)
Centrifugal pump (rpm)†	3150 (3000–3455)	3100 (3000–3200)	3455 (3000–3555)	3200 (2880–3300)
Days on ECMO†	4 (2–10)	4 (3–25)	3 (2–10)	3 (2–3)
Days of ECMO membrane†	3.5 (2–5)	4 (3–11)	3 (2–4)	3 (2–3)
Clots/fibrin in ECMO membrane†	1 (16.7)	1 (33.3)	0 (0)	0 (0)
Antimicrobial therapy and clinical outcomes
Dose / Frequency	1 g/8 h: 2 (33.3) 2 g/12 h: 1 (16.7) 2 g/8 h: 3 (50)	4/0.5 g/6 h: 2 (66.7) 4/0.5 g/8 h: 1 (33.3)	0,6 g/12 h: 3 (100)	300 mg/8 h: 1 (33.3) 400 mg/8 h: 1 (33.3) 600 mg/8 h: 1 (33.3)
Mode of administration	4-h EI: 5 (83.3) 1 h: 1 (16.7)	3-h EI: 1 (33.3) 0.5 h: 2 (66.7)	1 h: 3 (100)	1 h: 3 (100)
Infection type
Pneumonia  Surgical site infection	5 (83.3) 1 (16.7)	3 (100)	2 (66.7) 1 (33.3)	3 (100)
Microbiological isolates	Pseudomonas aeruginosa: 1 (16.7) (MIC ≤ 0.125 mg/L) Burkholderia cepacia: 1 (16.7) (MIC 12 mg/L) No pathogens isolated: 4 (66.7)	Pseudomonas aeruginosa: 1 (33.3) (MIC 8/4 mg/L) No pathogens isolated: 2 (66.7)	No pathogens isolated: 3 (100)	Streptococcus pneumoniae: 3 (100) (MIC < 0.03–0.75 mg/L; 1 MIC unavailable)
ICU length of stay (days)	64.5 (29–86)	6 (5–49)	80 (29–86)	49 (28–108)
Duration of hospitalization (days)	112 (29–120)	6 (5-105)	120 (29–155)	105 (32–119)
Mortality	3 (50)	2 (66.7)	1 (33.3)	1 (33.3)

Categorical variables are presented as counts (percentage). Continuous data are expressed as median with interquartile range (IQR)

*Calculated in patients without CRRT

^†Measured at the time of the first sample collection

BMI: Body Mass Index; CG: Cockcroft–Gault; CLCR: creatinine clearance; CRRT: Continuous Renal Replacement Therapy; EI: extended infusion; FDO₂: fraction of delivered oxygen; ICU: Intensive Care Unit; MIC: minimum inhibitory concentration; VA: Veno-arterial ECMO; VV: Veno-venous ECMO

Non-compartmental pharmacokinetic analysis

PK measures (Cmax, Tmax, AUC_0–24, Kel, t_1/2, CL, and Vd) derived from pre-membrane, post-membrane, and arterial line samples were generally consistent across all antimicrobials and infusion regimens, with no statistically significant differences observed between sampling sites at the same sampling time. Detailed PK measures, stratified by sampling site are summarized in Tables 2, 3, 4 and 5 and Table S1.

Table 2.

Pharmacokinetic measures of meropenem according to sampling site

Antimicrobial	Infusion time	Blood samples type	Cmax (mg/L)	Tmax (h)	AUC0−24 (mg · h/L)	Kel (h− 1)	T1/2 (h)	CL (L/h)	Vd (L)
Meropenem	4 h	Pre-M [n = 24]	49.9 ± 33.1	3.44 ± 1.10	533 ± 270	0.336 ± 0.0746	2.15 ± 0.520	11.1 ± 3.89	35.0 ± 6.16
		Post-M [n = 24]	46.0 ± 26.9	3.46 ± 1.11	530 ± 268	0.329 ± 0.123	2.45 ± 1.25	11.2 ± 3.97	38.5 ± 7.25
		AL [n = 29]	43.6 ± 29.0	3.48 ± 1.11	434 ± 174	0.302 ± 0.106	2.57 ± 1.06	12.7 ± 3.39	43.8 ± 8.08
		p-value	0.946	0.925	0.809	0.884	0.828	0.795	0.270

All PK measures are expressed as mean ± SD. Values in brackets [n = x] indicate the number of concentration-time points used for the non-compartmental analysis

AL: arterial line. AUC_0–24: area under the plasma concentration–time curve from 0 to 24 h; CL: clearance; Cmax: maximum plasma concentration; Kel: elimination rate constant. Post-M: post-membrane concentrations; Pre-M: pre-membrane concentrations; T_1/2: half-life; Tmax: time to maximum concentration; Vd: volume of distribution

Table 3.

Pharmacokinetic measures of piperacillin/tazobactam according to sampling site

Antimicrobial	Infusion time	Blood samples type	Cmax (mg/L)	Tmax (h)	AUC0−24 (mg · h/L)	Kel (h− 1)	T1/2 (h)	CL (L/h)	Vd (L)
Piperacillin	0.5 h	Pre-M [n = 10]	223 ± 107	0.525 ± 0.0118	2208 ± 1516	0.353 ± 0.211	2.39 ± 1.43	7.72 ± 4.02	30.8 ± 29.8
		Post-M [n = 10]	226 ± 131	0.550 ± 0.000	2187 ± 1576	0.342 ± 0.222	2.57 ± 1.67	8.02 ± 4.48	35.1 ± 35.9
		AL [n = 12]	297 ± 60.9	0.558 ± 0.0354	2488 ± 1697	0.333 ± 0.190	2.48 ± 1.41	6.83 ± 3.52	28.0 ± 26.6
Tazobactam	0.5 h	Pre-M [n = 10]	20.6 ± 10.0	0.525 ± 0.0118	198 ± 111	0.336 ± 0.108	2.18 ± 0.702	9.88 ± 3.74	32.9 ± 21.8
		Post-M [n = 10]	19.2 ± 9.12	0.550 ± 0.000	190 ± 110	0.311 ± 0.112	2.38 ± 0.855	10.5 ± 4.21	38.5 ± 27.3
		AL [n = 12]	27.0 ± 2.12	0.558 ± 0.0354	227 ± 120	0.323 ± 0.118	2.30 ± 0.837	8.51 ± 2.96	30.0 ± 20.1

All PK measures are expressed as mean ± SD. Values in brackets [n = x] indicate the number of concentration-time points used for the non-compartmental analysis. Statistical comparisons were not performed for piperacillin/tazobactam due to the small sample size; results are presented descriptively

AL: arterial line. AUC_0–24: area under the plasma concentration–time curve from 0 to 24 h; CL: clearance; Cmax: maximum plasma concentration; Kel: elimination rate constant. Post-M: post-membrane concentrations; Pre-M: pre-membrane concentrations; T_1/2: half-life; Tmax: time to maximum concentration; Vd: volume of distribution

Table 4.

Pharmacokinetic measures of linezolid according to sampling site

Infusion time	Blood samples type	Cmax (mg/L)	Tmax (h)	AUC0−24 (mg · h/L)	Kel (h− 1)	T1/2 (h)	CL (L/h)	Vd (L)
1 h	Pre-M [n = 16]	14.9 ± 4.81	1.24 ± 0.150	152 ± 62.3	0.170 ± 0.0713	4.79 ± 2.53	9.05 ± 4.21	58.6 ± 28.0
	Post-M [n = 16]	14.5 ± 5.33	1.27 ± 0.150	154 ± 64.2	0.169 ± 0.0622	4.49 ± 1.59	8.92 ± 4.14	53.5 ± 16.6
	AL [n = 19]	13.6 ± 2.07	1.29 ± 0.167	130 ± 42.0	0.185 ± 0.0735	4.17 ± 1.68	10.1 ± 3.93	56.3 ± 14.8
	p-value	0.932	0.926	0.850	0.946	0.930	0.927	0.956

All PK measures are expressed as mean ± SD. Values in brackets [n = x] indicate the number of concentration-time points used for the non-compartmental analysis

AL: arterial line. AUC_0–24: area under the plasma concentration–time curve from 0 to 24 h; CL: clearance; Cmax: maximum plasma concentration; Kel: elimination rate constant. Post-M: post-membrane concentrations; Pre-M: pre-membrane concentrations; T_1/2: half-life; Tmax: time to maximum concentration; Vd: volume of distribution

Table 5.

Pharmacokinetic measures of ceftaroline according to sampling site

Infusion time	Blood samples type	Cmax (mg/L)	Tmax (h)	AUC0−24 (mg · h/L)	Kel (h− 1)	T1/2 (h)	CL (L/h)	Vd (L)
1 h	Pre-M [n = 15]	32.6 ± 16.1	0.950 ± 0.275	339 ± 165	0.142 ± 0.0234	4.98 ± 0.894	4.34 ± 2.15	29.5 ± 10.7
	Post-M [n = 15]	28.3 ± 16.5	0.989 ± 0.231	297 ± 178	0.137 ± 0.0118	5.07 ± 0.448	5.30 ± 2.45	37.7± 15.7
	AL [n = 18]	30.9 ± 13.8	0.978 ± 0.300	313 ± 150	0.200 ± 0.0840	3.83 ± 1.30	4.71 ± 2.49	23.1 ± 5.92
	p-value	0.943	0.984	0.952	0.202	0.281	0.886	0.359

All PK measures are expressed as mean ± SD. Values in brackets [n = x] indicate the number of concentration-time points used for the non-compartmental analysis

AL: arterial line. AUC_0–24: area under the plasma concentration–time curve from 0 to 24 h; CL: clearance; Cmax: maximum plasma concentration; Kel: elimination rate constant. Post-M: post-membrane concentrations; Pre-M: pre-membrane concentrations; T_1/2: half-life; Tmax: time to maximum concentration; Vd: volume of distribution

Meropenem

For the 4-h extended infusion, the most notable differences were observed in Cmax and AUC_0–24. Both measures were lower in arterial line samples (Cmax 43.6 ± 29.0 mg/L; AUC_0–24 434 ± 174 mg·h/L) compared with pre-membrane (Cmax 49.9 ± 33.1 mg/L; AUC_0–24 533 ± 270 mg·h/L) and post-membrane samples (Cmax 46.0 ± 26.9 mg/L; AUC_0–24 530 ± 268 mg·h/L), without statistically significant differences. One-hour infusions were only studied in a single patient, showing greater variability in Cmax and AUC_0–24, with higher values observed in arterial line samples compared with pre- and post-membrane samples.

Piperacillin and tazobactam

Similar PK profiles were observed across infusion regimens. In the 0.5-h infusion, arterial line samples consistently exhibited the highest exposure compared with pre-membrane and post-membrane samples, with Cmax and AUC_0–24 values of 297 ± 60.9 mg/L and 2488 ± 1697 mg·h/L for piperacillin, and 27.0 ± 2.12 mg/L and 227 ± 120 mg·h/L for tazobactam, respectively. A similar pattern was observed in the 3-h infusion, where arterial Cmax reached 103 mg/L for piperacillin and 12.1 mg/L for tazobactam, with corresponding AUC_0–24 values of 625 mg·h/L and 68.9 mg·h/L.

Linezolid

PK measures were generally consistent across pre-membrane, post-membrane, and arterial sites. Arterial line samples showed non-significant slightly lower Cmax and AUC_0–24 values (Cmax 13.6 ± 2.07 mg/L; AUC_0–24 130 ± 42 mg·h/L) compared with pre-membrane (Cmax 14.9 ± 4.81 mg/L; AUC_0–24 152 ± 62.3 mg·h/L) and post-membrane samples (Cmax 14.5 ± 5.33 mg/L; AUC_0–24 154 ± 64.2 mg·h/L), without statistically significant difference. Elimination and distribution PK measures were similar across pre-membrane, post-membrane, and arterial line samples.

Ceftaroline

Post-membrane samples exhibited non-significant lower Cmax and AUC_0–24 values (Cmax 28.3 ± 16.5 mg/L; AUC_0–24 297 ± 178 mg·h/L) compared with pre-membrane values (Cmax 32.6 ± 16.1 mg/L; AUC_0–24 339 ± 165 mg·h/L). Arterial line measurements (Cmax 30.9 ± 13.8 mg/L; AUC_0–24 313 ± 150 mg·h/L) fell between pre- and post-membrane values.

Impact of ECMO circuit on antimicrobial concentrations

The analysis of antimicrobial concentrations between the three sampling sites (arterial line, pre-membrane and post-membrane) identified limited and component-specific differences, with no significant sequestration observed across the entire ECMO circuit (Cpre–Cart) for any of the antimicrobials evaluated. Results are presented as mean absolute difference (mg/L) and relative difference (%) in antibiotic concentrations across sampling points (Cpre–Cpost, Cpost–Cart, and Cpre–Cart), with detailed data summarized in Table 6. Individual plasma concentration–time profiles for each antimicrobial, stratified by sampling site, are illustrated in Fig. 2.

Table 6.

Differences in antimicrobial concentrations between ECMO sampling sites

Antimicrobial	Cpre – Cpost		Cpre-Cart		Cpost-Cart
	Absolute difference (mg/L)	Relative difference (%)	Absolute difference (mg/L)	Relative difference (%)	Absolute difference (mg/L)	Relative difference (%)
Meropenem	3.23 ± 7.95	9.28 ± 21.6 p = 0.046	2.11 ± 7.40	6.39 ± 18.0 p = 0.066	-1.11 ± 11.2	-14.0 ± 54.7 p = 0.933
Piperacillin	1.06 ± 7.41	1.10 ± 5.19 p = 0.443	-19.2 ± 28.3	-17.0 ± 18.2 p = 0.000	-20.3 ± 31.1	-19.0 ± 23.1 p = 0.000
Tazobactam	0.507 ± 0.997	3.50 ± 5.58 p = 0.036	-1.91 ± 3.10	-17.9 ± 22.1 p = 0.003	-2.41 ± 3.42	-22.7 ± 24.9 p = 0.002
Linezolid	-0.075 ± 1.08	-5.09 ± 14.9 p = 0.191	0.95 ± 1.81	5.50 ± 19.7 p = 0.094	1.03 ± 1.73	10.1 ± 12.3 p = 0.005
Ceftaroline	2.16 ± 2.44	14.6 ± 17.5 p = 0.010	0.86 ± 1.47	4.92 ± 10.1 p = 0.064	-1.3 ± 2.81	-17.0 ± 32.0 p = 0.103

All measures are expressed as mean ± SD. Cart: arterial line concentrations; Cpost: post-membrane concentrations; Cpre: pre-membrane concentrations

Fig. 2.

Individual plasma concentration–time profiles of study antimicrobials, stratified by sampling site. Y-axis scales differ between panels depending on the range of antimicrobial concentrations

Meropenem showed limited oxygenator-related sequestration, with a mean sequestration of 9.28% ± 21.6%, corresponding to a mean Cpre–Cpost difference of 3.23 ± 7.95 mg/L, which reached statistical significance (p = 0.046). Pre-membrane concentrations were higher than arterial concentrations (6.39% ± 18.0%; Cpre–Cart of 2.11 ± 7.40 mg/L), whereas post-membrane concentrations were lower (-14.0% ± 54.7%; Cpost–Cart of − 1.11 ± 11.2 mg/L); however, these differences were not statistically significant.

Piperacillin did not demonstrate significant oxygenator-related sequestration, with mean estimated sequestration of 1.10% ± 5.19% (p = 0.443) and a mean Cpre–Cpost difference of 1.06 ± 7.41 mg/L. In contrast, tazobactam exhibited small but statistically detectable oxygenator-related sequestration, with a mean sequestration of 3.50% ± 5.58% (p = 0.036), corresponding to a mean Cpre–Cpost of 0.507 ± 0.997 mg/L. For both compounds, arterial concentrations were higher than pre and post-membrane concentrations, with statistically significant differences observed for both Cpre–Cart and Cpost–Cart.

Linezolid concentrations remained generally stable, with no statistically significant differences observed for Cpre–Cpost or Cpre–Cart comparisons. Post-membrane concentrations were higher than arterial line concentrations, resulting in a small but statistically significant post-membrane-to-arterial gradient (mean relative difference: 10.1% ± 12.3%; mean Cpost–Cart difference: 1.03 ± 1.73 mg/L; p = 0.005).

Ceftaroline exhibited larger oxygenator-related concentration differences compared with the other antimicrobials evaluated, with a mean estimated sequestration of 14.6% ± 17.5% (mean Cpre–Cpost: 2.16 ± 2.44 mg/L), which reached statistical significance (p = 0.010). However, the estimated concentration difference across the entire ECMO circuit was smaller (4.92% ± 10.1%; Cpre–Cart difference: 0.86 ± 1.47 mg/L) and did not reach statistical significance (p = 0.064). Differences between Cpost–Cart were not statistically significant.

Target attainment and clinical outcomes

For all antimicrobials, Cmin values were comparable across sampling sites (arterial line, pre-membrane, and post-membrane), as summarized in Table 7. Overall, 12 of 15 antimicrobial treatments (80%) achieved the predefined PKPD efficacy target, and none exceeded the established toxicity thresholds. Specifically, target attainment stratified by antimicrobial was 83.3% (5/6) for meropenem, 66.7% (2/3) for piperacillin, 66.7% (2/3) for tazobactam, 66.7% (2/3) for linezolid, and 100% (3/3) for ceftaroline.

Table 7.

Trough concentrations of study antimicrobials stratified by sampling site

Antimicrobial	Cmin arterial line (mg/L)	Cmin pre membrane (mg/L)	Cmin post membrane (mg/L)
Meropenem	12.67 ± 7.50 (n = 6)	12.15 ± 6.50 (n = 6)	11.43 ± 7.35 (n = 6)
Piperacillin	47.93 ± 53.34 (n = 3)	42.47 ± 46.30 (n = 3)	42.23 ± 45.86 (n = 3)
Tazobactam	4.13 ± 4.09 (n = 3)	3.80 ± 3.63 (n = 3)	3.87 ± 3.88 (n = 3)
Linezolid	2.10 ± 1.01 (n = 3)	2.03 ± 1.38 (n = 3)	2.37 ± 1.36 (n = 3)
Ceftaroline	7.17 ± 3.38 (n = 3)	7.57 ± 3.32 (n = 3)	6.73 ± 3.78 (n = 3)

All PK measures are expressed as mean ± SD. Cmin: total trough plasma concentration. Sampling points: arterial line, pre-membrane, and post-membrane

Regarding clinical outcomes, the median duration of hospitalization was 68.5 days (31–130) for VV-ECMO patients compared to 62.5 days (6-120) for VA-ECMO patients, while the median ICU length of stay was 39 days (29–68) for VV-ECMO and 43 days (6–94) for VA-ECMO. Overall mortality was 50% (4/8), including one patient receiving VV-ECMO (1/4; 25%) and three patients receiving VA-ECMO (3/4; 75%). Among the three patients who had at least one antimicrobial exposure not achieving the PKPD efficacy target, two died (66.7%), whereas two of the five patients who achieved the PKPD target were non-survivors (40%). All non-survivors in the target attainment group were supported with VA-ECMO. Importantly, no definitive attribution of mortality can be made based on these data. Causes of death were heterogeneous and, in the available clinical assessment, were not clearly attributed to uncontrolled infection or antimicrobial treatment failure. Outcomes stratified by antimicrobial are detailed in Table 1.

Discussion

To our knowledge, this is the first in vivo study that simultaneously evaluates meropenem, piperacillin/tazobactam, linezolid and ceftaroline concentrations at three distinct sampling sites of the ECMO circuit—arterial line, pre-membrane, and post-membrane—in order to assess drug sequestration within it, and to characterize the PK of commonly used antimicrobials in critically ill patients receiving extracorporeal support. This prospective observational study provides novel insights into the impact of the ECMO circuit on antimicrobial PK. Although exploratory, these preliminary findings may help inform antimicrobial dosing strategies in this complex population.

Overall, our results suggest a limited impact of ECMO circuit on systemic antimicrobial exposure. Most standard dosing regimens of meropenem, piperacillin/tazobactam, linezolid, and ceftaroline achieved PKPD efficacy targets, with no cases exceeding toxicity thresholds. Nevertheless, despite all patients receiving dosing regimens optimized for critically ill patients and individually adjusted for renal function or RRT, 20% of treatments failed to reach efficacy targets. Mortality was higher in this group; however, given the small sample size, the observed differences in outcomes should be interpreted cautiously. No robust conclusions can be drawn regarding factors associated with mortality, including PKPD target attainment, underlying disease severity, or ECMO type.

In more than half of the patients treated with meropenem and piperacillin/tazobactam, antimicrobial therapy was initiated empirically. Consequently, PKPD target attainment for both agents was assessed using EUCAST clinical breakpoints values for Pseudomonas aeruginosa, applying targets of 100% fT > 2 mg/L for meropenem and 100% fT > 16 mg/L for piperacillin. Similarly, all patients receiving linezolid were treated empirically, and PKPD targets were evaluated using the EUCAST clinical breakpoints for Staphylococcus aureus, with efficacy defined as achieving ≥ 85% T > 2 mg/L and an AUC₀₋₂₄/2 ratio between 80 and 120. Although this approach does not capture isolate-specific susceptibility, it provides a conservative and clinically meaningful framework for evaluating antimicrobial exposure in this high-risk population.

Conversely, all patients receiving ceftaroline—each with microbiologically confirmed Streptococcus pneumoniae pneumonia—achieved the predefined PKPD efficacy targets. In addition, all observed ceftaroline concentrations met the PKPD efficacy target for Staphylococcus aureus (100% fT > 1 mg/L), suggesting that exposures were sufficient for both pathogens; however, this inference is limited by the small sample size and should be interpreted cautiously.

Drug-specific analyses revealed distinct patterns of circuit-related effects. Meropenem and ceftaroline showed limited, oxygenator-related concentration differences, which reached statistical significance. Pre-membrane concentrations for both antibiotics were higher than post-membrane and arterial line concentrations, suggesting that the oxygenator may represent a relevant site of circuit-related adsorption. Arterial line concentrations also showed slightly lower Cmax and AUC_0–24 values, reflecting a subtle impact of the ECMO circuit on systemic drug exposure. These observations are consistent with previous ex vivo studies reporting drug sequestration in the ECMO circuit for both meropenem [12, 30] and ceftaroline [31]. Nevertheless, while ceftaroline showed the largest apparent Cpre–Cpost difference, this finding should be interpreted cautiously, as no statistically significant overall circuit-related sequestration was observed.

Piperacillin/tazobactam exhibited negligible retention within the ECMO circuit, with arterial line concentrations exceeding both pre- and post-membrane concentrations. This finding is most likely explained by minor temporal differences between sampling sites during blood collection, in the context of the rapid distribution and elimination kinetics of piperacillin/tazobactam, and is suggestive of negligible drug sequestration within the circuit. Although tazobactam presented a statistically significant retention across the membrane oxygenator, arterial concentrations were higher than pre-membrane values, indicating that systemic exposure was not compromised. Nevertheless, one patient who received the drug as a short infusion at a dose of 4/0.5 g every 8 h failed to achieve the predefined PKPD efficacy target. This observation aligns with previous prospective studies reporting suboptimal target attainment in critically ill ECMO patients, in which piperacillin/tazobactam was administered at daily doses ranging from 9 g to 18 g, adjusted for renal function and delivered via intermittent or continuous infusion [32–35]. These results emphasize the importance of optimizing administration strategies such as extended and continuous infusions to maximize exposure.

Linezolid did not show significant sequestration in the membrane oxygenator as pre- and post-membrane concentrations were comparable. However, post-membrane concentrations were significantly higher than those measured at the arterial line, suggesting minimal retention in other circuit components, such as tubing and connectors. Published data on linezolid PK during ECMO remain limited and inconsistent. While ex vivo studies indicate minimal losses [36], clinical reports have described variable exposures and high rates of PKPD target failure, even in patients receiving daily doses of 1200–1800 mg via intermittent or continuous infusion [32, 37, 38]. In our study, one of three patients—a morbidly obese individual (BMI > 40 kg/m²) with community-acquired pneumonia and a CLCR > 90mL/min—failed to achieve the predefined efficacy target. This patient received linezolid 600 mg every 12 h via intermittent infusion, underscoring the risk of subtherapeutic exposure despite standard dosing.

Negative sequestration values observed for some antimicrobials are likely explained by the minimal time differences between sampling points. Although blood samples were collected quasi-simultaneously, a maximum time interval of up to 5 min between sampling sites was unavoidable in clinical practice. Given the rapid distribution and elimination kinetics of several antimicrobials, even small temporal differences may result in measurable concentration changes, potentially leading to apparent negative sequestration values. Therefore, these findings should be interpreted with caution, as they are more likely attributable to sampling-related variability and may be considered indicative of the absence of drug sequestration.

Although lipophilic drugs with high plasma protein-binding have been reported as the most susceptible to ECMO-related sequestration, our results reveal measurable circuit-related adsorption of hydrophilic, low protein-binding antimicrobials such as meropenem and ceftaroline, as well as of a moderately lipophilic, low protein-binding agent like linezolid. These findings suggest that ECMO-related drug loss cannot be predicted solely based on physicochemical properties of the drug and emphasize the critical role of circuit-related factors—including circuit age, tubing design, pump type, and the composition and surface properties of circuit components—in determining the extent of drug retention within the extracorporeal system.

The PK measures observed for all studied antimicrobials were consistent with previously reported values in critically ill patients not receiving ECMO [39–43]. Cmax, AUC_0–24, t_1/2, CL, and Vd values fell within the ranges described in the literature for ICU patients receiving comparable dosing regimens, including patients with similar clinical characteristics, such as renal impairment or concomitant renal replacement therapy [39–43]. These findings support a limited impact of ECMO-related sequestration on systemic drug exposure.

Concomitant CRRT represents a potential confounding factor in antimicrobial PK assessment during ECMO, as it is integrated in series into the extracorporeal circuit. In our cohort, arterial antimicrobial concentrations in patients receiving CRRT were not significantly lower than those measured at pre- and post-membrane sites. However, the independent contribution of CRRT to antimicrobial clearance could not be quantified and could not be fully disentangled from the effects of ECMO. Nevertheless, the presence of CRRT introduces variability that should be considered when interpreting PK data and designing future studies aimed at optimizing dosing strategies in ECMO-supported patients, particularly in those receiving combined extracorporeal support.

In summary, our study identifies that antimicrobial sequestration occurs not only within the membrane oxygenator but also in other ECMO circuit components, such as tubing and connectors, particularly for meropenem, ceftaroline, and linezolid. Nevertheless, overall sequestration across the entire ECMO circuit remained statistically non-significant for all antimicrobials evaluated. Quantitative analysis showed that sequestration was highest in the membrane oxygenator for β-lactams such as meropenem and ceftaroline, and was also evident in tubing and connectors for linezolid; however, these localized losses did not translate into clinically meaningful reductions in systemic concentrations, as most treatments achieved the desired PKPD target. This was further supported by the absence of significant differences in PK measures across sampling sites.

These results underscore that, while circuit-related adsorption exists, its clinical relevance appears limited. The development of population pharmacokinetic (popPK) models will be critical to comprehensively characterize inter- and intra-individual variability, quantify the impact of ECMO support, and identify covariates influencing antimicrobial exposure. Such investigations will be crucial for advancing the clinical implementation of model-informed precision dosing (MIPD) strategies in critically ill patients receiving ECMO support and to generate robust evidence to guide optimal dosing recommendations in this complex population.

Importantly, our findings suggest that inadequate antimicrobial exposure in ECMO-supported patients cannot be explained solely by circuit-related drug sequestration, which appeared limited at the systemic level. Instead, suboptimal PKPD target attainment likely reflects the profound PK variability characteristic of critically ill patients receiving ECMO, driven by factors such as altered Vd, augmented or fluctuating renal CL, obesity, concomitant RRT, and the mode of drug administration. In this context, therapeutic drug monitoring (TDM) should be considered a key component of antimicrobial management in ECMO-supported patients, not to compensate for circuit-related drug loss per se, but to address inter- and intra-individual variability and to ensure timely attainment of PKPD efficacy targets.

This study has several limitations. First, the single-center design, relatively small sample size, and heterogeneity of the cohort—including differences in underlying conditions, ECMO configuration, and concomitant therapies—limit the generalizability of our findings and precluded meaningful subgroup analyses. Second, drug sequestration within the ECMO circuit was assessed using indirect estimations based on concentration differences between sampling sites rather than direct mass balance measurements. Comparisons involving arterial concentrations may therefore be influenced by patient-related factors such as drug distribution, metabolism, and clearance, particularly given the inclusion of both VV and VA ECMO configurations. In VV-ECMO, post-oxygenator blood re-enters the venous circulation and undergoes pulmonary and potentially hepatic transit before arterial sampling, and may include a component of recirculation within the ECMO circuit, while in VA-ECMO arterial concentrations may reflect variable admixture of native cardiac output and ECMO flow. Importantly, the small sample size and observational design further limit the ability to distinguish circuit-related effects from patient-specific PK variability. Third, blood samples were collected quasi-simultaneously, with a maximum interval of up to 5 min between sampling sites, which may have introduced additional variability in measured concentrations, particularly for drugs with rapid distribution and elimination kinetics. In addition, arterial line samples were not collected at a fixed point within the sampling sequence, and could be obtained either before pre-membrane samples or after post-membrane samples; this may have introduced additional variability in concentration measurements. Fourth, a subset of patients received CRRT integrated in series into the ECMO circuit, which may have introduced additional drug clearance and adsorption that could not be separately quantified. Fifth, percentage estimates of sequestration, while useful for comparing antimicrobials with different exposure ranges, remain dependent on the selected reference concentration and should therefore be interpreted alongside absolute concentration differences. Sixth, due to the limited sample size, no subgroup analysis was performed to evaluate potential differences in PK between VV- and VA-ECMO configurations. Seventh, for β-lactam antibiotics, free drug concentrations were estimated using protein binding values reported in the literature. This approach does not account for potential alterations in critically ill patients, particularly in the context of ECMO, hypoalbuminaemia, or systemic inflammation, where protein binding may be significantly affected. Finally, no conclusions can be drawn regarding association between PKPD target attainment and clinical outcomes, as mortality could not be directly associated to uncontrolled infection or antimicrobial treatment failure.

Despite these limitations, the collection of multiple samples per patient across three sampling points provides a detailed characterization of antimicrobial PK and drug sequestration, offering valuable insights to guide clinical practice. Future multicenter studies are needed to validate these findings and refine evidence-based dosing recommendations.

Conclusions

Our findings suggest that antimicrobial sequestration occurs in the membrane oxygenator for meropenem and ceftaroline, and in other ECMO circuit components for linezolid, while piperacillin/tazobactam exhibits minimal retention. Nevertheless, the overall impact of ECMO on systemic antimicrobial exposure appears limited and of minimal clinical relevance, as most antibiotic regimens achieved the predefined PKPD efficacy targets.

However, 20% of treatments failed to achieve PKPD efficacy targets, underscoring the marked PK variability of critically ill patients receiving ECMO and the importance of TDM and individualized dosing strategies in this population. These preliminary results provide a foundation for optimizing antimicrobial therapy in ECMO-supported patients and emphasize the need for further population PK studies to better characterize determinants of antimicrobial exposure, refine evidence-based dosing recommendations, and clarify the clinical relevance of circuit-related drug sequestration.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Abbreviations

ARDS

Acute respiratory distress syndrome

AUC_0–24

Area under the plasma concentration–time curve from 0 to 24 h

BMI

Body mass index

Cart

Arterial line concentrations

CL

Clearance

CLCR

Creatinine clearance

Cmax

Maximum plasma concentration

Cmin

Trough concentrations

Cpost

Post-membrane concentrations

Cpre

Pre-membrane concentrations

CRRT

Continuous renal replacement therapy

CT

Critical threshold concentration

CV

Coefficient of variation

ECMO

Extracorporeal membrane oxygenation

EUCAST

European committee on antimicrobial susceptibility testing

ELSO

Extracorporeal life support organization

FDO₂

Fraction of delivered oxygen

fT > MIC

Time that free drug concentration remains above the minimum inhibitory concentration

ICU

Intensive care unit

IQR

Interquartile range

Kel

Elimination rate constant

LLOQ

Limit of quantification

MIC

Minimum inhibitory concentration

MIPD

Model-informed precision dosing

NCA

Non-compartmental analysis

PD

Pharmacodynamics

PK

Pharmacokinetics

PMP

Polymethylpentene

PopPK

Population pharmacokinetic

PVC

Polyvinyl chloride

QC

Quality control

RRT

Renal replacement therapy

SD

Standard deviation

T_1/2

Half-life

TDM

Therapeutic drug monitoring

Tmax

Time to maximum concentration

UHPLC-MS/MS

Ultra-high-performance liquid chromatography method coupled with tandem mass-spectrometry

VA

Veno-arterial

Vd

Volume of distribution

VV

Veno-venous

Author contributions

A.E.R. was responsible for the conception of the work, study design, data collection, data interpretation, drafting the original manuscript, and reviewing the final manuscript. E.S., A.T.S., and P.C. were responsible for data interpretation and reviewing the final manuscript. R.R.B., S.H., M.H.M., A.C., J.I.S.M., M.B.S., J.M., and C.R. reviewed the final manuscript. D.S. and C.B. were responsible for the conception of the work, study design, data interpretation, and reviewing the final manuscript. Subsequently, all authors provided critical revisions until the final manuscript was completed. All authors read and approved the final manuscript and ensured that questions related to the accuracy or integrity of any part of the work were investigated and resolved.

Funding

This research was funded by the Joaquim Bonal grant 2024–2026, awarded by the Catalan Society of Clinical Pharmacy (SCFC).

Data availability

Data available on request. The data underlying this article will be shared on reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Helsinki declaration and approved by the ethics committee of Hospital Clínic of Barcelona. Its trial approval number is HCB/2024/0574.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.