Ceftaroline pharmacokinetics on ECMO, a pediatric case report

Abstract

In-vitro studies suggest that patients on extracorporeal membrane oxygenation (ECMO) may struggle to achieve therapeutic ceftaroline exposure to treat infections due to antibiotic sequestration within the ECMO oxygenator; however, there are no published data regarding ceftaroline concentrations from patients supported by ECMO.

Here we present an analysis of ceftaroline pharmacokinetics in a 5-year-old patient supported on ECMO due to influenza-associated necrotizing pneumonia. A total of 9 concentrations were analyzed, including 3 concentrations prior to cannulation and 6 on ECMO. ECMO cannulation was associated with a relatively small change in the volume of distribution (0.17 L/kg to 0.20 L/kg) and clearance (1.00 mL/min/kg to 0.86 mL/min/kg). This patient continued to achieve 100% time above MIC while on ECMO. Substantial adsorption or sequestration were not observed in this circuit using a Nautilus oxygenator.

This case report demonstrates that achieving therapeutic ceftaroline concentrations for patients on ECMO may be feasible using current oxygenators. Clinicians could consider continuing ceftaroline therapy following ECMO cannulation for patients achieving good microbiologic response.

Introduction:

Ceftaroline is a fifth-generation cephalosporin with activity against gram-positive and gram-negative bacteria, including methicillin-resistant Staphylococcus aureus (MRSA). Ceftaroline was approved for pediatric use in 2016 for treatment of community acquired bacterial pneumonia.¹ It is administered intravenously as the prodrug ceftaroline fosamil, which is then converted to active ceftaroline by plasma phosphatases. It is not a significant substrate of hepatic cytochrome p450 metabolism and is primarily excreted unchanged or as hydrolyzed metabolites in the urine.¹ It is hydrophilic with a volume of distribution approximating the estimated extracellular fluid volume in adults.¹ Dose adjustment is required for adults with decreased kidney function and an estimated GFR (eGFR) ≤50 mL/min/1.73m²; however, there are no approved dosage adjustments for pediatric patients with eGFR ≤50 mL/min/1.73m².

Despite the importance of antibiotic use in critically ill patients, data supporting dose adjustments for patients who require mechanical organ support such as extracorporeal membrane oxygenation (ECMO) are lacking.² One ex-vivo study using an isolated ECMO circuit with a Quadrox oxygenator attached to a blood reservoir demonstrated up to 80% ceftaroline loss attributed to the oxygenator component,³ raising concern that patients requiring ECMO support might be unable to attain therapeutic exposure to this crucial medication. An adult case report describes successful treatment of MRSA pneumonia with ceftaroline in an ECMO patient who had previously failed linezolid,⁴ but was unable to evaluate ceftaroline pharmacokinetics. Currently, there are no case reports describing ceftaroline pharmacokinetics in ECMO patients of any age.

Patient presentation:

The patient is a 17.5 kg 5-year-old male with a history of 28-week prematurity and asthma who presented with influenza B and community-acquired pneumonia. He presented on the day of admission with worsening work of breathing, cough, and lethargy following an influenza B diagnosis the day prior. In the emergency room, he was hypoxic and hypotensive requiring fluid resuscitation and noninvasive-positive pressure ventilation prior to transfer to the pediatric intensive care unit (PICU). Presenting labs were notable for leukopenia, thrombocytopenia, acute kidney injury (AKI), elevated transaminases, and lactic acidosis (Supplemental Table 1). Blood cultures collected in the emergency room grew Streptococcus pneumoniae and Staphylococcus aureus at 5 hours (Supplemental Table 2). He was transitioned from initial empiric antibiotics to ceftaroline on hospital day 1.

The patient’s respiratory failure continued to progress following PICU admission, requiring intubation on hospital day 1 and cannulation to veno-venous (VV) ECMO with a 3/8 inch circuit and a Nautilus membrane on hospital day 2. He had evidence of mild right heart dysfunction prior to cannulation, which worsened following ECMO cannulation and required conversion to veno-arterial (VA) ECMO early on hospital day 3. Continuous renal replacement therapy (CRRT) was added for fluid overload at the end of hospital day 3 using an HF1000 filter and clearance of 1000 mL/m². A visual timeline of events is provided in Supplemental Figure 1. Ceftaroline was administered at a dose of 15 mg/kg via a 1-hour intravenous infusion every 8 hours. The dose and time of ceftaroline administrations were recorded from the medication administration record and there were no dose changes during his course. Plasma samples for pharmacokinetic analysis were obtained from scavenged blood, which was initially obtained and separated in lithium-heparin tubes for clinical electrolyte analysis, stored in the clinical lab at 4°C for 48h, and subsequently relinquished for research purposes after the completion of clinically ordered laboratory testing. Scavenged plasma samples were frozen at −80°C and shipped to the reference laboratory. Ceftaroline concentrations were determined by validated liquid chromatography tandem mass spectrometry at Atlantic Diagnostic Laboratories (Bensalem, PA), with a lower limit of quantification at 0.5 mcg/mL.

A total of 10 pharmacokinetic samples were obtained (Table 1). One sample was collected during a ceftaroline infusion and was excluded from the analysis. Pharmacokinetic analysis was performed with all remaining samples (n=9), as well as subsets of samples prior to CRRT initiation (n=7), prior to ECMO (n=4), and on ECMO alone (n=3) (Figure 1). Using the mapbayr R package⁵, maximum a posteriori Bayesian estimation was performed to obtain individual pharmacokinetic parameters (half-life [t_1/2], volume of distribution [Vd], and clearance [CL]) by individually modeling each of the four clinical contexts. This approach combines information about a drug’s pharmacokinetic parameters (e.g., population pharmacokinetic model) with one or more measured drug concentrations to estimate the most likely parameters for an individual patient. A published two-compartment ceftaroline population pharmacokinetic model was used as the model prior.⁶ This model was developed by pooling pharmacokinetic data from five pediatric studies (birth to <18 years old) combined with adult pharmacokinetic data.

Table 1:

Ceftaroline concentrations from scavenged blood samples.

Time since admit (h)	Sample #	Ceftaroline concentration (mcg/mL)	Time since end of ceftaroline infusion (h)	Sample source	Creatinine (mg/dL)	eGFR (mL/min/1.73m2)	Event
27	1	53.6	1.3	unk	0.63	74
39	2	61.5	1.5	A-line	0.47	99
50	3	22.6	4.0	A-line	0.6	78
52							VV ECMO canulation*
58	4	51.2	2.5	unk	0.69	68
61							VV → VA conversion
63	5	19.4	7.7	CVL	0.65	72
75	6	20.5	6.5	unk	0.55	85
80	7	54.7	1.9	unk	0.72	65
82							CRRT start#
87	8	40.7	2.0	unk	0.77	61
93	9^	77.4	during infusion	unk	0.74	63
99	10	1.2	5.3	unk	0.65	72

Time of blood draw and time of most recent ceftaroline dose were assessed from the EMR. Time since ceftaroline is calculated from the end of a 60 minute infusion. Blood draw source listed if documented in the EMR, otherwise listed as unknown (unk) if no source is listed. Creatinine reported from results in the EMR from electrolyte panels corresponding with the scavenged sample. eGFR calculated using bedside Schwartz equation. A-line: arterial line. CVL: central venous line. ECMO: extracorporeal membrane oxygenation. VV: veno-venous. VA: veno-arterial. CRRT: continuous renal replacement therapy. eGFR: estimated glomerular filtration rate

^*3/8 inch circuit, Nautilus membrane, blood primed. Circuit volume 437 mL, membrane volume 226 mL

^#CVVHDF, HF1000 filter, blood primed, clearance 1000 mL/m²

^{^}Sample excluded from pharmacokinetic analysis due to collection during ceftaroline infusion

Figure 1: Ceftaroline concentration time profiles for each clinical context.

Each panel represents a different clinical context, where the solid lines represent the individual predicted ceftaroline concentrations after Bayesian estimation and the circles represent observed concentrations used for that modeled context. The dotted horizontal line at 0.5 mg/L represents the Staphylococcus aureus minimum inhibitory concentration.

Analysis of all nine concentrations revealed a t_1/2 of 3.4 h, Vd of 0.20 L/kg, and CL of 0.78 mL/min/kg (Table 2). Only considering samples prior to the initiation of CRRT produced a Vd of 0.19 L/kg, clearance of 0.78 mL/min/kg, half-life of 3.3 h. The period following ECMO cannulation but prior to CRRT initiation (ECMO alone) is not dramatically different than the pre-ECMO period. The fT>MIC was 100% in all contexts.

Table 2:

Individual Parameter Estimates.

Clinical context	PK Samples	Vda	Vd scaleda	CLb	CL scaledb	T1/2b	fT>MIC
		L	L/kg	L/h	ml/min/kg	h	%
All data	9	3.4	0.20	0.82 (0.77–0.85)	0.78 (0.73–0.83)	3.4 (3.2–3.6)	100
Prior to CRRT	7	3.4	0.19	0.82 (0.77–0.85)	0.78 (0.73–0.83)	3.3 (3.2–3.5)	100
Pre ECMO	3	3.1	0.17	1.05 (1.02–1.06)	1.00 (0.97–1.01)	2.4 (2.4–2.4)	100
ECMO alone	4	3.5	0.20	0.90 (0.86–0.95)	0.86 (0.82–0.90)	3.2 (3.0–3.3)	100

The concentration collected during ceftaroline infusion (sample 9) was excluded from all analyses. Vd: volume of distribution. CL: clearance.

T1/2: half life. fT>MIC: fraction of dosing interval with time greater than S. aureus minimum inhibitory concentration.

^aThe Vd was calculated as the sum of the central and peripheral ceftaroline compartments based on the two-compartment structure from Riccobene et al. 2017.

^bParameters summarized as median (minimum-maximum) because CL is a function of time-varying creatinine clearance. There were 8, 6, 3, and 5 unique parameter values for the scenarios with All Samples, Prior to CRRT, Pre ECMO, and ECMO alone, respectively.

Discussion:

This is the first description of the pharmacokinetics of ceftaroline in a patient supported by ECMO. Concentrations collected on ECMO were relatively similar to those observed prior to cannulation. This patient demonstrated notably higher than expected ceftaroline concentrations throughout the dosing interval, both prior to and following ECMO cannulation. According to the United States Food and Drug Administration package insert, the expected maximum concentration (C_max) following repeated dosing of 600 mg in adult patients is 21.3 mcg/mL,¹ while a population pharmacokinetic model based on pediatric licensing data predicted C_max of 27.1 mcg/mL in non-critically ill children treated with doses of 12 mg/kg.⁷

It is most probable that the high concentrations detected in this patient indicate a greater degree of kidney dysfunction than appreciated based on creatinine-based GFR, in combination with higher ceftaroline dosing used in the setting of severe infection. Our patient never had creatinine-based eGFR ≤50 mL/min/1.73 m², but he had a severe AKI with oliguria and the limitations of creatinine-based eGFR alone in critically ill patients are well recognized.^8–11 The patient did not have a cystatin C measured during his hospitalization for use of alternate, potentially improved, methods for eGFR calculation.¹² He did have a neutrophil gelatinase-associated lipocalin (NGAL) of >1500 ng/mL on hospital day 1 and 2, which has shown an association with antibiotic clearance beyond that predicted by eGFR alone,^13,14 but is not currently incorporated in any dosing adjustment recommendations.

Existing pediatric literature about the effects of critical illness on ceftaroline pharmacokinetics consists of a case series of seven critically ill children treated with ceftaroline. ¹⁵ No children in this case series had significant renal impairment, and there were no extracorporeal therapies in use. This case series demonstrated that the PICU population had increased Vd (range: 0.17–0.84 L/kg) and CL (range: 1.6–6.1 mL/min/kg) compared to less ill children of similar ages. Compared to patients in that study, our patient had lower CL and had Vd at the low end of the reported range. While the slower CL in the setting of AKI were expected, the low Vd in our patient compared to PICU peers is more surprising. ECMO would be expected to increase the Vd from the increase in circulating blood volume and potentially due to adsorption and sequestration within circuit components.

There was no evidence of extreme sequestration within this circuit using a Nautilus oxygenator. It is possible that the extraction observed ex-vivo was specific to the Quadrox oxygenators used in that report.³ While the total surface area of Nautilus and Quadrox adult oxygenators are similar and both use polymethylpentane hollow fibers, they differ in coating proteins and charge,¹⁶ which could affect antibiotic adsorption. Alternately, it may be that medication behavior in-vivo differs significantly from that observed in the isolated circuit, a pattern that has been observed for several other antibiotics in adults.¹⁷ Polypharmacy in critically ill patients could affect the saturation of circuit adsorption sites leading to less sequestration of each individual medication compared to laboratory circumstances where only one exogenous medication is present. While ceftaroline exhibits only modest protein binding (~20%),¹ it is also possible that the content and characteristics of plasma proteins in critically ill patients differ in ways that affect medication adsorption and binding to ECMO circuit components.

This patient had only two samples collected following CRRT initiation, and the impact of CRRT on his CL should be interpreted with caution. It is notable that both concentrations measured following initiation of CRRT are lower than concentrations at similar timepoints prior to CRRT initiation, suggesting increased extracorporeal medication CL. The concentration measured in sample 10 was approximately 20-fold lower than other concentrations at similar timepoints, but there are no similar timepoints while on CRRT to validate this observation. While it is expected that CRRT would contribute to ceftaroline CL thereby decreasing concentrations, without more samples to validate the effect of extracorporeal clearance it is also possible that sample 10 is an outlier given the magnitude of difference in concentration compared to the remainder of measurements. An adult case series found that among critically ill adults on CRRT, non-renal (i.e., CRRT) CL contributed to >50% of total medication CL with variability related to dialysis prescription and residual renal function, with predicted C_min on CRRT of 0.9–4.5 mcg/mL. ¹⁸

There are limitations intrinsic to the scavenged samples used for this case report. The opportunistic sampling scheme across different dosing intervals reduces the ability to distinguish the pharmacokinetic effects attributable any single machine or physiologic change, including effects attributable exclusively to the ECMO circuit. The population pharmacokinetic model used for Bayesian estimation was developed for children with skin and soft tissue infections or community acquired pneumonia in licensing studies⁶, who were less acutely ill than our patient and who had normal renal function. Using a model developed in pediatric patients with severe illnesses and that incorporates covariates that account for extracorporeal therapy may improve estimations but, to our knowledge, no models specific to the critically ill pediatric population have been published. However, estimates derived using this approach were similar to those obtained using a noncompartmental approach with pooled samples for each clinical context (supplemental table 3). The source used for clinical blood draws is not always recorded and it is possible that one or more samples were collected from lines that had recently infused ceftaroline. Collection from a “contaminated” line could contribute to elevated concentrations in a sample which do not truly represent concentrations in plasma. However, at least two samples were documented as being collected from an arterial line—which would not have been exposed to antibiotic infusion—and these concentrations are similar to the others from unknown sources. Finally, while ceftaroline has excellent stability at 4°C prior to human administration¹⁹ and is stable in human plasma at room temperature for at least 24h³, the stability in human plasma at 4°C for ≥48h as required for scavenged sampling has not been directly reported. We cannot exclude the possibility of sample degradation during this storage period, although the high measured concentrations lead us to believe substantial degradation is less likely. The use of scavenged samples could affect the ability to compare directly to studies that obtained samples intended for concentration measurement.

The findings of this single case report may not be generalizable to all critically ill patients, especially because the physiologic effects of critical illness are heterogeneous and produce profound and varied effects on drug disposition. For example, while this patient experienced acute kidney injury resulting in decreased CL, other patients such as those reported in the 2018 PICU case series¹⁵ have augmented renal CL, a recognized critical illness phenomenon related to factors such as vasodilation and increased cardiac output.²⁰ Profound anasarca, massive blood loss/replacement, and/or sequestered sites of infection may further contribute to the pharmacokinetic complexities among critically ill patients, which differ from their peers and the general population. This patient experienced impaired cardiac output leading up to his VA ECMO cannulation and fluid overload leading up to CRRT initiation, but the impact of each of these overlapping physiologic changes cannot be quantified in this single patient analysis. Larger studies are needed to understand the spectrum of pharmacokinetic and pharmacodynamic changes in patients on ECMO, including frequency of alterations in renal function and the impact of different oxygenators.

Given the high intra- and inter-patient variability during critical illness, PICU patients would be ideal candidates for therapeutic drug monitoring to optimize beta-lactam therapy. Therapeutic drug monitoring could be particularly impactful during events which may rapidly alter dose-exposure relationships, such as initiation or discontinuation of extracorporeal therapies, major surgeries, and changing organ function. For patients where there is concern about inadequate response to therapy, therapeutic drug monitoring could be helpful to evaluate whether patients are meeting pharmacodynamic goals (fT>MIC), and to recommend earlier intervention, such as changes to dosing schedule or extended infusions. Conversely, demonstration of adequate antibiotic exposure could be used to empower clinicians to continue current agents rather than switching classes, potentially impacting antibiotic stewardship.

Patient outcome:

The patient’s ECMO course was complicated by necrotizing pneumonia with bronchopleural fistulae requiring bilateral chest tubes as well as persistent evidence of right heart dysfunction requiring sildenafil. For treatment of necrotizing pneumonia, the patient continued ceftaroline for 4 days, then was transitioned to ceftriaxone/linezolid, and subsequently clindamycin to complete a 28-day antibiotic course. All subsequent blood cultures as well as endotracheal and broncho-alveolar lavage cultures following ECMO cannulation had no growth. He had no other infectious complications.

CRRT was discontinued on hospital day 9, and renal function normalized for the remainder of hospitalization. The patient was decannulated from ECMO after 17 days (hospital day 19) and transferred out of the PICU on hospital day 26. Following ECMO decannulation, his heart function normalized without evidence of persistent pulmonary hypertension. He was discharged home after 40 days of hospitalization with no long-term oxygen support.

Conclusion:

In this case report, ECMO cannulation produced modest alterations to ceftaroline pharmacokinetics. Importantly, there was no evidence of profound antibiotic sequestration in the Nautilus oxygenator. While the relatively high concentrations and 100% fT> MIC demonstrated by our patient were partially driven by his concurrent kidney injury, it follows that other patients who are achieving therapeutic ceftaroline exposure and adequate microbiologic response prior to ECMO cannulation may continue to be effectively treated with ceftaroline following ECMO cannulation. Patients dependent on extracorporeal devices may be a population for whom therapeutic drug monitoring would assist in understanding individual patient pharmacokinetics.