Impact of continuous-infusion meropenem degradation and infusion bag changes on bacterial killing of Pseudomonas aeruginosa based on model-informed translation

Background: Continuous infusion of meropenem has been proposed to increase target attainment in critically ill patients, although stability might limit its practical use. This study investigated the impact of meropenem degradation and infusion bag changes on the concentration-time proﬁles and bacterial growth and killing of P. aeruginosa given different continuous-infusion solutions. Methods: A semi-mechanistic pharmacokinetic-pharmacodynamic (PK-PD) model quantifying meropenem concentrations (C MEM ) and bacterial counts of a resistant P. aeruginosa strain (ARU552, MIC = 16 mg/L) over 24 h was used to translate in vitro antibiotic effects to patients with severe infections. Concentration-dependent drug degradation of saline infusion solutions was considered using an additional compartment in the population PK model. C MEM , f T > MIC (time that concentrations exceed the MIC) and total bacterial load (B TOT ) after 24 h were simulated for different scenarios ( n = 144), considering low-and high-dose regimens (30 0 0/60 0 0 mg/day ± loading dose), clinically relevant infusion solutions (20/40/50 mg/mL), different intervals of infusion bag changes (every 8/24 h, q8/24 h), and varied renal function (creatinine clearance 40/80/120 mL/min) and MIC values (8/16 mg/L). Results: Highest deviations between changing infusion bags q8h and q24h were observed for 50 mg/mL solutions and scenarios with C MEM_24 h close to the MIC, with differences ( (cid:2) ) in C MEM_24 h up to 4.9 mg/L, (cid:2) f T > MIC ≤ 65.7%, and (cid:2) B TOT_24h ≤ 1.1 log10 CFU/mL, thus affecting conclusions on whether bacteriostasis was reached. Conclusions: In summary, this study indicated that for continuous infusion of meropenem, eight-hourly infusion bag changes improved PK/PD target attainment and might be beneﬁcial particularly for high meropenem concentrations of saline infusion solutions and for plasma concentrations in close proximity to the MIC.


Introduction
Effective antibiotic exposure is crucial amidst increasing antimicrobial resistance, driven by insufficient drug concentrations [ 1 ].Prudent antibiotic use is imperative, especially with priority pathogens like carbapenem-resistant Pseudomonas aeruginosa [ 2 ].Prolonged infusion durations have been shown to improve the attainment of pharmacokinetic/pharmacodynamic (PK/PD) f T > MIC (time that unbound concentrations exceed the minimum inhibitory concentration) targets for beta-lactams and have thus been com-mon practice in the treatment of severe infections [ 3 ].Potential meropenem degradation in the infusion bag thereby merits attention but its impact on antibiotic plasma concentrations and bacterial killing remains largely unclear.
Antibiotic stability in infusion devices is crucial for continuous administration, potentially limiting its practical use.The stability of meropenem has been studied under diverse experimental conditions.For example, concentration-dependent drug degradation was shown for 0.9% sodium chloride solutions (C sol = 10-50 mg/mL), with decreasing stability at higher concentrations.Stability (defined as ≥90% of the initial concentration, unchanged colour, absent turbidity) persisted for 24 h (C sol = 10 mg/mL), 18 h (C sol = 20 mg/mL), 15 h (C sol = 30 mg/mL) and 8 h (C sol = 50 mg/mL) https://doi.org/10.1016/j.ijantimicag.2024.1072360924-8579/© 2024 The Author(s).Published by Elsevier Ltd.This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ) An overview of the investigated administration forms together with the respective degradation rates is provided in Table S1.[ 4 , 5 ].Previous stability studies have, however, not covered a potential impact of meropenem degradation on its PK or PD profile.
Published clinical PK studies and model-based analyses assessing meropenem exposure for different continuous-infusion regimens commonly do not consider the potential instability of meropenem in infusion solutions, potentially leading to lower doses delivered.The same is true for simulations with PKPD models that describe the continuous time course of drug concentrations and bacterial load of specific pathogen strains (e.g., observed in in vitro time-kill experiments) and allow to identify potentially efficacious dosing regimens [ 6 ].If patient-based PK models are used to drive bacterial growth and killing, PKPD models enable in vitro-in vivo translation, illustrating the impact of clinically relevant drug administration forms and dosing regimens on human PK and bacterial PD profiles [ 6 ].
This study aimed to assess the impact of drug degradation and infusion bag changes during continuous infusion on meropenem concentration-time profiles and resulting bacterial (re)growth and killing of P. aeruginosa using a semi-mechanistic PKPD model.

Methods
A PKPD model describing the concentrations and in vitro antibiotic effects of meropenem against a resistant P. aeruginosa strain (ARU552, MIC = 16 mg/L) over 24 h was extended to mimic drug degradation in the infusion bag during continuous infusion [ 7 ].To translate microbiological effects from an in vitro setting to humans, simulations were performed assuming bacterial growth and killing to be driven by the PK profiles of a patient population with severe infections [ 8 , 9 ].Unbound concentration-time profiles of meropenem (unbound fraction 98%) and colony forming units (CFU) were simulated for 24 h since start of treatment [ 10 ].
To account for drug degradation in the infusion bag during continuous infusion, the two-compartment PK model was extended with a dosing compartment ( Fig. 1 ), in which the amount of meropenem (AMT MEM ) decreased over time according to previously performed stability tests in saline (NaCl) solutions (20-50 mg/mL) at ICU conditions (23 °C, see [ 4 ] (supplement), [ 5 ]).
For each investigated scenario ( n = 144), meropenem concentrations (C MEM ) and total bacterial load (B TOT ) were simulated over time for 10 0 0 patients considering interindividual PK variability.Particular attention was paid to C MEM , f T > MIC and B TOT of median PK and PD profiles, as well as of the 95 th percentile (P 0.95 ) of the population at 24 h.Moreover, the proportion of patients without bacterial net growth after 24 h and those reaching the target f T > MIC = 50% were determined.Details of all modelling and simulation activities as well as on the stability data underlying the study are presented in the Supplementary Material.

Results
The impact of infusion bag changes was first studied for meropenem concentrations and the resulting PK/PD index f T > MIC .Further analyses investigated how these antibiotic concentrations (potentially reduced due to meropenem instability) would translate to the total bacterial load for different scenarios and (highdose and low-dose) dosing regimens.Detailed results for all scenarios are provided as tables in the Supplementary Material (Table S2: population with CLCR = 80 mL/min and bacterial strain with (default) MIC = 16 mg/L; Table S3: more susceptible bacterial strain with MIC = 8 mg/L; Table S4: population with lower renal function, CLCR = 40 mL/min, MIC = 16 mg/L; Table S5: population with high renal function, CLCR = 120 mL/min).Key scenarios are depicted in Figs. 2 and S1.

Impact of infusion bag changes on the PK/PD index f T > MIC
Differences in f T > MIC between q8h versus q24h infusion bag changes were marked merely when C MEM_24 h was close to the bacterial MIC.For example, at CLCR = 80 mL/min (median profiles), continuous-infusion regimens without loading dose (60 0 0 mg) and without bag changes resulted in f T > MIC = 0% for all three investigated infusion concentrations over 24 h, while q8h changes led to f T > MIC = 65.7/52.5/29.4% (C sol = 20/40/50 mg/mL; Fig. 2 , Fig. 3 , Table S2).Similar trends were observed for lower-dose infusions (30 0 0 mg) when assuming a more susceptible strain (EC 50 = 8.85 mg/L).

Impact of infusion bag changes on the total bacterial load
Infusion bag changes showed more evident benefits regarding bacterial growth suppression of the resistant P. aeruginosa strain (MIC = 16 mg/L) for high-dose continuous-infusion regimens and high infusion concentrations (60 0 0 mg/day, C sol = 40-50 mg/mL) than for low-dose regimens (30 0 0 mg/day) and C sol = 20 mg/mL.Assuming higher bacterial susceptibility (MIC = 8 mg/L), bag changes appeared most beneficial for the low-dose regimens (due to extensive bacterial suppression at high doses, Fig. S2, Table S3).
The difference in the proportion of the population without net bacterial regrowth at 24 h (stasis) between q8h and q24h infusion bag changes was at most 13% (e.g.observed for CLCR = 80 mL/min, B TOT = 1 log10 CFU/mL; see Table S2, Fig. 3 ).The impact of infusion bag changes on B TOT was best visible when meropenem concentrations at 24 h (C MEM_24 h ) were close to the MIC of the pathogen (see Figs. 2 + S1), as otherwise extensive bacterial regrowth or killing occurred for all scenarios investigated.

Discussion
The overall feasibility of continuous infusion for meropenem and changing infusion bags to mitigate drug degradation are subjects of ongoing discussion.The present study investigated the impact of meropenem degradation during continuous infusion on antibiotic concentrations, f T > MIC and bacterial load of a P. aeruginosa strain at 24 h after start of treatment, and revealed differences be- tween 24-h infusions and q8h infusion bag changes for diverse scenarios.
Changing infusion bags q8h rather than q24h resulted in more stable and relatively constant meropenem concentrations over 24 h, which often translated to a slightly more pronounced bacterial reduction at 24 h.For some scenarios, particularly for the highest infusion concentration (50 mg/mL), the difference in the bacterial count between q8h and q24h infusion bag changes amounted to 1 log10 CFU/mL (e.g.B TOT_median : CLCR = 80 mL/min, CI 60 0 0mg or B TOT_P0.95 : CLCR = 40 mL/min, off-label CI 60 0 0mg + LD50 0mg ).A magnitude of 1-log kill was previously considered clinically relevant, but this also depends on the type and acuity of an infection [ 12 ].A key question still remains, that is, how well the results of in vitro pharmacodynamic effects can be translated to antibacterial effects in the human body, in which multiple complex systems contribute to the elimination of pathogens.The main aim of this analysis, however, was to compare different dosing regimens and drug administration forms and to assess relative benefits of different clinically conceivable scenarios rather than absolute antibiotic effects.
The impact of changing infusion bags was most evident when meropenem concentrations 24 h after the start of infusion were close to the pathogen MIC.Bacterial suppression could be observed even for concentrations slightly below the MIC (14.9 mg/L, f T > MIC = 0), challenging the binary nature of PK/PD indices like f T > MIC , suggesting no killing for concentrations below the MIC and maximum killing for concentrations above the MIC.Values of the PK/PD index f T > MIC did not always reflect B TOT .Therefore, pronounced differences in B TOT_24 h between q8h and q24h bag changes were observed despite f T > MIC = 0% for both scenarios (e.g.B TOT = 1.03 log10 CFU/mL when CLCR = 40 mL/min, off-label CI 60 0 0mg , C sol = 50 mg/mL).The proportion of patients without net bacterial growth at 24 h was up to 13% higher (40.3%→ 53.1%) with q8h infusion bag changes versus one 24-h infusion, corresponding to a relative increase of > 30%.This difference might propagate further and become more significant if meropenem was administered for a longer duration.
In different clinical settings, different meropenem infusion solutions and concentrations are used.For example, 500 mg or 1 g meropenem can be reconstituted with 50 mL of 0.9% sodium chloride injection USP (i.e.resulting in 10 or 20 mg/mL solutions) [ 13 ].However, also higher concentrations up to 50 mg/mL can be used [ 10 ].
Previous product labels indicated varying storage durations for saline infusion solutions (prepared with 0.9% sodium chloride for concentrations up to 20 mg/mL), ranging from one to 4 h at 15-25 °C, depending on the type of preparation and container [ 10 , 13 , 14 ].Meropenem stability has further been studied under diverse experimental conditions.Our simulations were based on stability data investigated specifically for an ICU setting at a constant room temperature of 23 °C.Previous evidence supported these data: Venugopalan et al. [ 15 ] suggested that 94.8% of the initial meropenem concentration (92.8 mg/mL) would remain after 24 h, aligning closely with the stability data used in our analysis, that is, 91% of the original concentration after 24 h (10 mg/mL solution).Manning et al. [ 16 ] reported recoveries of 88% and 83% after 24 h for elastomeric infusion devices containing 1% or 2% meropenem at ambient temperature under outpatient parenteral antimicrobial therapy conditions.The analysis further reported a 95% maximum deliverable dose (MDD) of 1% meropenem and MDD = 87% at 24 h for a 2% meropenem concentration, which corresponded to the stability data underlying our study (87.5% after 24 h).However, temperature was not continuously monitored in the previous study.
Various studies have provided different findings on the stability of meropenem solutions under other study conditions, including variations in temperature, container types or product brands, which may have produced slightly different results compared to the present simulation study.For example, a NaCl solution of 42 mg/mL remained stable for 8 h in a polypropylen syringe at 25 °C, thus at warmer than ICU conditions [ 17 ].A 50 mg/ml saline solution of meropenem was found to be stable for 8 h at 21-26 °C in a glass container [ 18 ].In clinical practice, continuous infusions over 24 h are probably less common and expectedly administered solely for low infusion concentrations (e.g.C sol = 10 mg/mL).Meropenem stability in infusions with lower concentrations appears higher (e.g.> 24 h for 12 mg/mL water-NaCl solution [ 19 ]).
The present study aimed to illustrate the impact of meropenem degradation on selected what-if-including worst-case-scenarios rather than to cover all conceivable dosing regimens and infusion solutions.For example, previous studies investigated a q12h exchange of infusion, yielding similar median serum concentrations to q8h exchanges; however, rather stable 10 mg/mL solutions (1.5 g in 150 mL) were used in the study concerned.Other studies suggest changing infusion bags every 6 hours [ 20 ].

Conclusions
We present a simulation study illustrating how PKPD models allow to translate information on in vitro efficacy to clinical settings and to assess the potential impact of continuous-infusion meropenem degradation and infusion bag changes on bacterial dynamics.The study hints that infusion bag changes may be beneficial particularly for high meropenem concentrations of saline infusion solutions and for plasma concentrations near the MIC of the pathogen.Further investigations are necessary to assess the clinical relevance of this finding.

Fig. 1 .
Fig. 1.Scheme of the population PK model for meropenem (MEM) considering drug degradation during continuous infusion.AMT MEM : amount of meropenem (mg) in the infusion bag; AMT 0 : initial amount of meropenem at start of treatment (nominal dose), k deg : degradation rate, V C /V P : central/peripheral volume of distribution, Q: intercompartmental clearance; CL: clearance.The model included renal function (creatinine clearance according to Cockcroft and Gault) as a covariate on clearance.An overview of the investigated administration forms together with the respective degradation rates is provided in TableS1.

Fig. 2 .
Fig. 2. Concentration-time profiles of meropenem MEM (upper panels) and resulting bacterial load (lower panels) over time for a resistant Pseudomonas aeruginosa strain in an adult infected patient population with normal renal function (CLCR = 80 mL/min) given 60 0 0 mg/24 h dosing administered as continuous infusion.q8h: every 8 h; solid line: median; long-dashed blue line: profile representing P 0.95 in the population; horizontal dashed black line: minimum inhibitory concentration (16 mg/L); log 10 CFU/mL denotes the change of bacterial load over time (bacterial load at time 0h = 6 log 10 CFU/mL).

Fig. 3 .
Fig. 3. Impact of different continuous-infusion regimens of meropenem (with and without infusion bag changes) on the PK/PD index f T > MIC (time that meropenem concentrations exceed the minimum inhibitory concentration; left column) and on the total bacterial load after 24 h (right column) for different scenarios and median profiles in the population.Filled and open circles represent f T > MIC based on 24 h (left column) and the bacterial count at 24 h after start of treatment (right column; also see Graphical Abstract).log10 CFU/mL denotes the change of bacterial load from baseline (initial bacterial load at time 0h = 6 log10 CFU/mL).Percentages in brackets indicate proportions of patients reaching a target of f T > MIC = 50% (left column) or bacteriostasis (right column) when receiving one infusion over 24 h (non-bold letters) versus infusion bag changes q8h (bold letters).Q8h and q24h scenarios in the lower right panel are overlapping.