Altered antibiotic pharmacokinetics during extracorporeal membrane oxygenation: cause for concern?

Sir, Extracorporeal membrane oxygenation (ECMO) is a supportive therapy and definitive management often relies on adequate drug therapy aimed at reversing the underlying cause of cardiac and/or respiratory failure. Despite this, a paucity of data exists describing the impact of ECMO on pharmacokinetics (PK) and antibiotic dosing requirements. By default, clinicians prescribe standard doses to these complicated patients, thereby risking therapeutic failure or toxicity. We report two patients to highlight our concerns with regard to antibiotic therapy in the presence of ECMO. Ethics approval was granted by the local Human Research Ethics Committee (HREC/11/QPCH/121). The first patient with pneumonia and respiratory failure was supported with venovenous ECMO. Meropenem (1 g every 8 h) was commenced for Enterobacter septicaemia. Serial blood samples (2 mL) were collected to assay meropenem plasma concentrations (Figure 1). The meropenem clearance (CL) was 20.4 L/h, which is higher than that for critically ill patients either without renal dysfunction (13.6 L/h) or those with renal failure receiving high-volume continuous venovenous haemofiltration (6.0 L/h). The calculated volume of distribution (V) was also high at 0.56 L/kg. Whilst a minimum target of 40% for the time that the meropenem concentration was maintained above the MIC (40% T.MIC) was achieved, the more aggressive target (4× MIC that is used clinically, 8 mg/L) was only achieved for 30% of the dosing interval. The second patient received venoarterial ECMO and extended daily diafiltration (EDD-f) for Pseudomonas aeruginosa pneumonia and multiorgan failure. Guided by the data from Patient 1, meropenem was administered as a high-dose infusion (6.5 g every 24 h) and serial blood samples (2 mL) were collected for measurement of meropenem plasma concentrations (Figure 1). The meropenem CL was once again high at 20.8 L/h. However, this elevated dosing regimen successfully maintained meropenem plasma concentrations above the target level (4× MIC, where the P. aeruginosa MIC is 2 mg/L). The ECMO circuitry in both patients comprised bioline tubing, a centrifugal pump and a polymethyl pentene oxygenator (pump, Jostra RotaflowTM and oxygenator, Quadrox DTM; Maquet, Germany). Dosing meropenem as per standard practice resulted in underdosing in the first patient with isolated respiratory failure. A higher dose administered by continuous infusion was necessary to maintain optimal meropenem concentrations in the second patient with multiple organ failure, who received EDD-f. Both patients demonstrated a high meropenem CL despite significant differences in their severity-of-illness sequential organ failure assessment score of 4 versus 17 on the day of PK sampling. The meropenem CL seen in these two patients was substantially higher than is seen in critically ill patients not on ECMO and may be due in part to circuit sequestration. A recent ex vivo study demonstrated significant sequestration of meropenem in adult ECMO circuits. On the contrary, Wildschut et al., in their in vitro study using neonatal circuits, demonstrated no significant meropenem loss in the circuit beyond 60 min. However, it should be noted that meropenem is unstable at 378C and ongoing exteriorization of blood during ECMO may lead to a degree of spontaneous degradation, which can be erroneously interpreted as increased CL. Whether there is saturation of the circuit over time for meropenem remains to be determined in future studies. Although there are no meropenem CL data in neonates on ECMO for comparison, it is likely to be low given the organ immaturity and physiological differences. Studies in neonates on ECMO have consistently shown reduced CL for several antibiotics. Factors affecting PK during ECMO include sequestration in the circuit, increased V and decreased CL. Most of the available PK data are from neonates and relate to sedative and antibiotic drugs. Emerging data from adult patients on ECMO suggest escalating sedation requirements over time. However, unlike sedation, there are no real-time pharmacodynamic endpoints for