C . C . Honeycutt et al .: J Extra Corpor Technol 2023 , 55 , 159 – 166 163
Figure 3 . Recovery of meropenem in CRRT circuits and controls over 4 h after administration . Values shown are means ( n = 4 for circuit , n = 3 for controls ) with error bars representing standard deviations . CRRT – continuous renal replacement therapy .
not significantly different from the controls suggesting that the loss is due to degradation rather than adsorption by the ECMO circuit . These findings were expected based on the physicochemical properties of meropenem .
Meropenem is a hydrophilic ( logP �0.6 ) small molecule ( 383 Da ) with low protein binding (~ 2 %) and a low volume of distribution [ 17 ]. Prior studies have shown that these physicochemical properties predispose a drug to rapid clearance by CRRT dialysis and filtration and tend to limit adsorption to ECMO circuit materials [ 18 , 19 ]. For CRRT , meropenem ’ s low molecular weight , high hydrophilicity , low volume of distribution , and low protein binding allow for free hemodiafiltration [ 20 – 22 ]. This was supported by our data that showed high meropenem recovery in the hemofiltrate , reflected by the mean saturation coefficient ( SA ) consistently around one . These factors suggest that meropenem can be rapidly cleared by CRRT and that clearance is related to CRRT flow rates [ 37 ]. These findings align with in vivo studies of critically ill patients with sepsis receiving CRRT , which have found that CRRT causes significant clearance of meropenem , necessitating steady-state intravenous doses of 500 – 1000 mg every 6 – 8 h to maintain sufficient plasma concentrations [ 30 – 32 ]. This is also consistent with results across b-lactam antibiotics , which as a class have target attainments that are highly impacted by RRT [ 24 ].
In the ECMO system , the loss of meropenem was not significantly different from the controls suggesting that degradation plays a major role rather than interaction with the ECMO circuit . This conclusion is supported by meropenem ’ s very short half-life , approximately 1 h [ 38 ]. Additionally , there is evidence that meropenem undergoes plasma metabolism . Studies of patients with end-stage renal disease and bilateral nephrectomy found that meropenem undergoes extrarenal metabolism or degradation with one detectable metabolite , the ring-open lactam form [ 39 ]. It has also been shown that the conversion of meropenem to the ring-open lactam form occurs at physiological pH and temperature but does not occur to a great degree at room temperature [ 39 ]. These findings have been observed in other ex vivo ECMO studies . Previous ex vivo work by Shekar et al . ( 2012 ) in isolated ECMO circuits identified a similar pattern of substantial loss in both circuits and controls [ 29 ]. However , in the Shekar study and a study by Cies , et al . ( 2022 ), there was a small but significantly greater loss in the circuits over time compared to controls suggesting some degree of interaction between meropenem and the circuit materials [ 23 , 29 ]. Differences in equipment and PVC surface coatings used may help to explain the differences between our findings and the findings of these studies [ 23 , 40 ].
In vivo pharmacokinetic studies of adult patients on ECMO have produced conflicting results . Shekar , et al . ( 2013 ) describe higher clearance of meropenem in ECMO patients compared with critically ill patients not on ECMO [ 26 ]. In contrast , a subsequent study by Shekar , et al . ( 2014 ) demonstrated a higher volume of distribution but lower clearance of meropenem in ECMO patients compared with critically ill patients not on ECMO [ 27 ]. Finally , Gijsen , et al . ( 2021 ) and Donatello et al . ( 2015 ) do not find significant differences in serum meropenem concentrations or target attainment between ECMO and non-ECMO patients [ 25 , 28 ]. Given the minimal differences in ex vivo results , the conflicting in vivo results are likely due to patient-specific factors , such as differences in renal function between individual patients .
Our study has multiple limitations . First , while there are multiple different PVC surface coatings used currently in clinical practice , including heparin coating , we solely utilized phosphorylcholine-coated PVC tubing [ 23 , 40 ]. Second , differences in the type of pump and diameter of the tubing could impact findings and were not explored in our study [ 23 ]. Third , our study design does not allow us to interrogate the roles of various additional mechanisms implicated in drug extraction from ECLS , including the ability of hemolysis to provide additional adsorption binding sites and release drugs from the cytoplasm of erythrocytes [ 41 ]. Fourth , our sample sizes were limited to controls ( n = 3 ) and circuits ( n = 3forECMO , n = 4 for CRRT ). However , these sample sizes are consistent with previous ex vivo studies [ 9 , 18 , 29 , 42 ]. Fifth , for CRRT , we did not investigate different effluent flow rates ( Q EFF ). For drugs that are cleared by hemofiltration or hemodialysis , the flow rate will impact the rate of clearance [ 37 ]. Because meropenem is a hydrophilic small molecule drug with low protein binding , we expect it to be freely filtered and thus impacted