Ceftriaxone Pharmacokinetics and Pharmacodynamics in Two Pediatric Patients on Extracorporeal Membrane Oxygenation Therapy

Abstract

Background:

Critically ill patients with cardiac or respiratory failure may require extracorporeal membrane oxygenation (ECMO). Antibiotics are frequently administered when the suspected cause of organ failure is an infection. Ceftriaxone, a β-lactam antibiotic, is commonly used in patients who are critically ill. While studies in adults on ECMO have suggested minimal impact on ceftriaxone pharmacokinetics (PK), limited research exists on ceftriaxone PK/PD in pediatric ECMO patients. We report the PK profiles and target attainment of two pediatric patients on ECMO who received ceftriaxone.

Methods:

Ceftriaxone concentrations were measured in two pediatric patients on ECMO using scavenged opportunistic sampling. PK profiles were generated and individual PK parameters were estimated using measured free ceftriaxone concentrations and a published population PK model in children who are critically ill, using Bayesian estimation.

Results:

Patient 1, an 11-year-old boy on venovenous ECMO for respiratory failure received two doses of 52 mg/kg ceftriaxone 12 h apart while on ECMO and additional doses every 12 h off ECMO. On ECMO, ceftriaxone clearance was 13.0 L/h/70 kg compared to 7.6 L/h/70 kg off ECMO, whereas the model-predicted mean clearance in children who are critically ill without ECMO support was 6.54 L/h/70 kg. Patient 2, a 2-year-old boy on venoarterial ECMO due to cardiac arrest received 50 mg/kg ceftriaxone every 12 h while on ECMO for > 7 d. Only clearance while on ECMO could be estimated (9.1 L/h/70 kg). Trough concentrations in both patients were > 1 mg/L (the breakpoint for Streptococcus pneumoniae) while on ECMO.

Conclusion:

ECMO increased ceftriaxone clearance above the model-predicted clearances in the two pediatric patients studied. Twelve-hour dosing allowed concentrations to remain above the breakpoint for commonly targeted bacteria but not four times the breakpoint in one patient, suggesting that precision dosing may be beneficial to ensure target attainment in children on ECMO.

Introduction:

Beta-lactam antibiotics, including cephalosporins, penicillins, and carbapenems, are routinely prescribed to treat serious systemic infections in patients who are critically ill. The effectiveness of these antibiotics is related to the duration of time during which free antibiotic concentrations remain above the minimum inhibitory concentration of the targeted bacteria (fT>MIC).¹ Antibiotic dosing regimens are frequently derived from studies in patients who are not critically ill. In current practice, clinicians often administer the same standard antibiotic drug dosing regimen to both non-critically ill and critically ill patients.¹ However, patients who are critically ill experience pathophysiological changes that alter drug pharmacokinetics (PK) and pharmacodynamics (PD). Specifically, clearance may increase or decrease depending on the patient’s condition. Increased cardiac output due to sepsis-induced systemic vasodilation can lead to increased clearance,^2,3 whereas end-organ dysfunction may decrease clearance and potentially result in elevated antibiotic concentrations and toxicity.^1,4

These PK changes affect PD target attainment, and therefore, antibiotic effectiveness. For β-lactam antibiotics, failure to achieve the PD target of free antibiotic concentrations greater than four times the bacterial minimum inhibitory concentration (MIC) for 100% of the dosing interval (100% fT>_4xMIC) may increase morbidity and mortality.⁵ In adults, traditional β-lactam dosing has been reported to result in low target attainment, with only 63% of the patients achieving a PD target (PD) of 100%ƒT > _1xMIC and only 36% of the patients achieving 100%ƒT >_4xMIC.⁶

Some patients who are critically ill may require Extracorporeal Membrane Oxygenation (ECMO) due to severe cardiac or respiratory failure. Case reports in adults suggest that ECMO has minimal impact on ceftriaxone PK and PD and that standard dosing regimens for non-ECMO ICU patients should suffice during ECMO.⁷ A recent study on population ceftriaxone PK modeling in adults on ECMO supports the adequacy of conventional dosing regimens for target attainment.⁸ To our knowledge, the effects of ECMO on ceftriaxone PK or PD in pediatric patients who are critically ill have not been studied. This report describes the PK parameters and PD target attainment in two pediatric patients receiving ECMO.

Materials and Methods

Both patients in this report were part of a prospective β-lactam antibiotic PK/PD study in the pediatric ICU (PICU) at Cincinnati Children’s Hospital Medical Center. The Institutional Review Board (IRB) granted a waiver of consent to obtain scavenged opportunistic samples⁹ for measuring β-lactam antibiotics.¹⁰ Standard of care was provided by the clinical team.

Patient 1 was an 11-year-old male weighing 25 kg. The patient had respiratory failure due to airway obstruction from a foreign object. On presentation, his creatinine was 0.43 mg/dL (creatinine clearance by bedside Schwartz equation: 132.5 mL/min/1.73 m²), and his Pediatric Risk Mortality (PRISM) III Score was 10. Day 1 albumin level was 1.5 g/dL, which increased over the subsequent days (Figure 1). The patient received venovenous ECMO (Cardiohelp, Maquet Medical Systems, Wayne, NJ, USA) (initial flow rate of 100 mL/kg/min), and two doses of 52 mg/kg ceftriaxone administered approximately 12 h apart for pneumonia. He developed cardiac arrest on day 1 of ECMO, likely due to air introduction in the circuit, and required three minutes of cardiopulmonary resuscitation (CPR). Despite developing acute kidney injury (AKI; Figure 1)) on ECMO, the patient did not require continuous renal replacement therapy. After two doses of ceftriaxone, he was transitioned to piperacillin/tazobactam to provide better coverage of the oral flora and anaerobes, given the inciting aspiration event. He was on ECMO for 42 h, after which he was switched back to 52 mg/kg ceftriaxone q12h (Figure 1). Creatinine levels remained high when ECMO was discontinued but started to improve. On day 8, the patient was switched to ampicillin/sulbactam for a complete antibiotic course of 14 days. The culture results remained negative. The patient was discharged on day 14.

Figure 1: Concentration vs. time profiles for Patient 1 while on (blue panel) and off ECMO (orange panel).

Ceftriaxone PK profiles on and off ECMO were simulated separately but included in the same figure with a continuous time scale for comparison. Red dotted line: population model-predicted profile based on patient covariates, except for creatinine clearance. Red circles: measured free ceftriaxone concentrations. Red solid line: Bayesian-estimated fitted profile using the population pharmacokinetic model and measured concentrations. Red dashed line: 4 mg/L (4× MIC); black dashed line: 1 mg/L (1× MIC). All albumin levels measured during the course are shown and displayed at the time when they were measured. The daily maximum creatinine levels are shown and displayed at the time they were measured. ECMO: extracorporeal membrane oxygenation; MIC: minimum inhibitory concentration.

Patient 2 was a 2-year-old male weighing 13 kg who experienced a non-fatal submersion and hypothermia (27 °C). The patient experienced pulseless electrical activity and developed ventricular fibrillation on arrival. He underwent 90 min of CPR and extracorporeal CPR (eCPR) cannulation for veno-arterial ECMO (initial flow rate of 150 mL/kg/min) to facilitate rewarming. Before starting ECMO, his creatinine was 0.38 mg/dL (creatinine clearance by bedside Schwartz equation: 101.1 mL/min/1.73 m²) and his PRISM III score was 36. Creatinine fluctuated between 0.32 mg/dL and 0.49 mg/dL over the following seven days. His albumin level was 3.9 g/dL on day 1, decreasing to its lowest on day 4 (2.7 g/dL). He received 50 mg/kg ceftriaxone q12h for seven days while on ECMO (Figure 2). The patient’s body temperature had been normalized to 36 °C at the time of ceftriaxone initiation. After decannulation from ECMO, the patient was administered vancomycin and cefepime on day 8 because of a new fever and completed an antibiotic course of 14 days. The culture results remained negative. The patient had a prolonged hospital stay and rehabilitation course due to severe anoxic brain injury, resulting in hypoxic-ischemic encephalopathy.

Figure 2: Concentration vs. time profiles for Patient 2 while on ECMO.

Red dotted line: population model-predicted profile based on patient covariates, except for creatinine clearance. Red circles: measured free ceftriaxone concentrations. Red solid line: Bayesian-estimated fitted profile using the population pharmacokinetic model and measured concentrations. Red dashed line: 4 mg/L (4× MIC); black dashed line: 1 mg/L (1× MIC). Daily minimum albumin levels and maximum creatinine levels are shown and displayed at the time they were measured. ECMO: extracorporeal membrane oxygenation; MIC: minimum inhibitory concentration.

As part of the IRB-approved β-lactam antibiotic parent study, residual blood from clinical samples were obtained from clinical laboratory storage using scavenged opportunistic sampling.^9,10 Samples were initially stored at 4 °C for up to seven days, centrifuged (2060 x g, 10–20 °C, 10 min), and the supernatant stored at −80 °C. Our previous study¹¹ showed that ceftriaxone is stable in scavenged samples for up to 7 days, and the samples from this study were handled similarly. Free ceftriaxone concentrations were measured using a validated high-performance liquid chromatography (HPLC) assay.¹¹ Briefly, to measure free ceftriaxone concentrations, ultrafiltration using centrifugation was used, and the filtrate was mixed with an internal standard for measurement of free ceftriaxone concentrations by HPLC.

The free ceftriaxone concentration vs. time profiles were generated for both patients with MwPharm++ (Mediware, Prague, Czech Republic), which can account for variable doses over time, time on and off ECMO, sampling times, and drug concentrations.¹² Using our previously published two-compartment population PK model of free ceftriaxone in children who were critically ill but not on any extracorporeal support devices,¹¹ individual PK parameters were estimated with the measured free ceftriaxone concentrations using Bayesian estimation.^13,14 In our population PK model, clearance (CL) and inter-compartmental clearance (Q) were allometrically scaled, and central and peripheral volumes of distribution (V1, V2) were linearly scaled to 70 kg, according to standard practice.¹⁵ Factors impacting CL of free ceftriaxone include maturation effect, PRISM III score, highest daily body temperature, and creatinine clearance. The reference patient weighs 70 kg with no significant maturation effect (i.e., >>2 years old) and has an estimated creatinine clearance of 149.5 mL/min/1.73 m², PRISM III score of 0, and no fever, as reported previously.¹¹ The model was also adapted to exclude creatinine clearance for model-based prediction as creatinine may not be a reliable marker of kidney function when on extracorporeal support devices. PK estimates were based on the observed ceftriaxone concentrations and either the original or the adapted PK model, which served as the Bayesian prior. Samples were primarily collected during the elimination phase rather than around the peak (Figures 1 and 2). Therefore, the central and peripheral volume of distribution and inter-compartmental clearance were fixed to the population mean (25.4 L/70 kg, 19.6 L/70 kg, and 4.26 L/h/70 kg, respectively) and only clearance was estimated. The assay error was set to 5%. For Patient 1, the ceftriaxone clearance during on- and off-ECMO was estimated. We used only the concentrations during the ECMO phase with the settings described above to estimate individual ceftriaxone clearance during ECMO. We repeated the process with only concentrations while off ECMO to generate individual ceftriaxone clearances. For patient 2, only ceftriaxone clearance while on ECMO could be estimated owing to the lack of concentration data while off ECMO, and all concentrations obtained for patient 2 were used to generate an individual ceftriaxone clearance estimate on ECMO. From the estimated parameters, the modeling software, MwPharm++ (Mediware, Czech Republic), predicted concentration-time profiles for patient 1 while on and off ECMO and for patient 2 while on ECMO. We also evaluated the percentage of dosing intervals in which free ceftriaxone concentrations was above the MIC and four times the MIC for the entire dosing interval (100% fT>_1XMIC, 100% fT>_4xMIC), two of the most stringent PK/PD targets for children who are critically ill. As no positive cultures were obtained, the Clinical & Laboratory Standards Institute ceftriaxone breakpoint for Streptococcus pneumoniae (1 µg/mL) was used as MIC.¹⁶ S. pneumoniae was chosen given the concern for pneumonia in both patients, and as S. pneumoniae is the typical bacteria clinicians empirically cover when initiating ceftriaxone.

Results

Figures 1 and 2 depict the estimated concentration-time profiles and ceftriaxone clearances during ECMO for both patients using the adapted model without creatinine clearance as the Bayesian prior. Supplemental Table 1 presents the clearance estimates obtained from the original and adapted models. We present the adapted model estimates, as ECMO may have an effect on creatinine values that is not linked to true alterations in a patient’s renal function. Patient 1 had an allometrically scaled, normalized ceftriaxone clearance of 13.0 L/h/70 kg (denormalized: 5.2 L/h) on ECMO and 7.6 L/h/70 kg (3.0 L/h) off ECMO. ECMO contributed to 71% additional clearance of ceftriaxone. In both scenarios, the clearance exceeded the model-predicted clearance of 6.5 L/h/70 kg, with a nearly double of the expected clearance during ECMO. Patient 1 maintained concentrations above 1 mg/L for all dosing intervals while on and off ECMO. However, for a target MIC of 4 mg/L, concentrations were only above the target for 86% of the dosing intervals while on ECMO (trough ranges: 2.8–3.1 mg/L) and for 100% of intervals while off ECMO.

For patient 2, the allometrically scaled ceftriaxone clearance on ECMO was 9.1 L/h/70 kg (denormalized: 1.5 L/h), higher than the model-predicted clearance of 6.54 L/h/70 kg. As no concentrations were measured while off ECMO, ceftriaxone clearance could not be determined. Patient 2 maintained concentrations above 1 mg/L and 4 mg/L for all dosing intervals, meeting both stringent targets.

Discussion

To the best of our knowledge, this is the first report on ceftriaxone PK/PD in children who are critically ill and undergoing ECMO. For patient 1, ceftriaxone clearance significantly increased while on ECMO compared to the time off ECMO. However, this may have been confounded as the patient likely sustained AKI after ECMO was discontinued, leading to lower ceftriaxone clearance while off ECMO. For Patient 2, clearance on ECMO was higher than that predicted by the free ceftriaxone model in children who are critically ill without ECMO. For both patients, 12-hour dosing allowed concentrations to remain above 1 mg/L (1× MIC); however, for patient 1, the concentrations fell below 4 mg/L (4× MIC) for less than 15% of the dosing intervals.

The first reports of ceftriaxone PK/PD in adults on ECMO were published in 2020 by Gijsen et al.,⁷ who demonstrated that ECMO did not significantly affect the unbound ceftriaxone PK and PK/PD target attainment of ceftriaxone. They suggested using the same dosing recommendations as in non-ECMO ICU patients, with adjustments based on renal function. The minimal impact of ECMO on ceftriaxone PK was also demonstrated in a recent population pharmacokinetic study of ceftriaxone in 14 adult patients who were critically ill and on ECMO.⁸ The findings that total and unbound ceftriaxone PK in adults is not significantly changed on ECMO are in contrast with our findings in which free ceftriaxone clearance in the two patients on ECMO described in this study was significantly higher than the population median clearance among children in the ICU not on ECMO in our previous study.¹¹ Further, for patient 1, clearance on ECMO was nearly double of that observed when off ECMO. It is important to note that while Patient 1 was on ECMO and receiving piperacillin/tazobactam, creatinine levels doubled, suggesting the development of AKI. This possible injury was persistent but improved when the patient was off ECMO and receiving ceftriaxone. Furthermore, it may have led to lower ceftriaxone clearance, and thus, a higher difference between on- and off-ECMO ceftriaxone clearance. One reason for the difference in findings between adults and children is that the case reports of adult patients on ECMO had more intensive sampling during only one dosing interval on ECMO and used non-compartmental analysis to generate PK parameters.⁷ In contrast, we utilized sparse sampling over multiple dosing intervals and used Bayesian estimation to estimate clearance only. In addition, the adult case reports did not compare intrapatient clearances on and off ECMO, as we were able to do for one patient. Instead, the case reports compared two patients on ECMO with a different cohort of non-ECMO ICU patients. The population PK analysis of ceftriaxone in adults on ECMO by Cheng et al. ⁸ demonstrated that every 12-hour dosing in patients on ECMO with normal kidney function, especially with low albumin concentrations, achieved concentrations above 4 mg/mL, which supports our practice of administering ceftriaxone every 12 h in children on ECMO.

With a higher estimated glomerular filtration rate based on creatinine clearance at presentation and prior to placement on ECMO, patient 1 had a higher ceftriaxone clearance on ECMO than patient 2 (13.0 L/h/70 kg vs. 9.1 L/h/70 kg). However, the higher ceftriaxone clearance in patient 1 led to the failure to attain the most stringent target. A lack of consensus remains with respect to the appropriate pharmacodynamic targets for children who are critically ill, especially when under extracorporeal life support. We evaluated two of the most stringent targets using a CLSI breakpoint (100% fT>_MIC and 100% fT>_4xMIC) because no bacteria were cultured from either patient. As CLSI breakpoints are often higher than the MIC of cultured bacteria (representing a worst-case scenario), these stringent targets may not be necessary. Interestingly, the first patient, for whom the most stringent target was not attained, was de-escalated to a narrower spectrum of antibiotics, whereas the second patient who met both targets had antibiotics broadened owing to new fevers. It is also important to note that dosing ceftriaxone every 12 h, as was done for both patients, is not standard practice for patients who are critically ill in other PICUs, where 24-hour dosing is more commonly used (personal communication). A longer dosing interval would likely lead to failed PK/PD target attainment while on ECMO, even for less stringent targets. This needs to be considered.

Conclusion

We present the first report on ceftriaxone PK/PD in two pediatric patients on ECMO and demonstrate that ECMO increases the clearance of free ceftriaxone. Further studies with more patients to develop population PK models of ceftriaxone in children who are critically ill and undergoing ECMO are warranted to better understand the effect of ECMO on ceftriaxone clearance and determine the appropriate ceftriaxone dosing regimen while on ECMO.