Abstract
Objectives
To describe the population pharmacokinetics of cefotaxime and desacetylcefotaxime in critically ill paediatric patients and provide dosing recommendations. We also sought to evaluate the use of capillary microsampling to facilitate data-rich blood sampling.
Methods
Patients were recruited into a pharmacokinetic study, with cefotaxime and desacetylcefotaxime concentrations from plasma samples collected at 0, 0.5, 2, 4 and 6 h used to develop a population pharmacokinetic model using Pmetrics. Monte Carlo dosing simulations were tested using a range of estimated glomerular filtration rates (60, 100, 170 and 200 mL/min/1.73 m2) and body weights (4, 10, 15, 20 and 40 kg) to achieve pharmacokinetic/pharmacodynamic (PK/PD) targets, including 100% ƒT>MIC with an MIC breakpoint of 1 mg/L.
Results
Thirty-six patients (0.2–12 years) provided 160 conventional samples for inclusion in the model. The pharmacokinetics of cefotaxime and desacetylcefotaxime were best described using one-compartmental model with first-order elimination. The clearance and volume of distribution for cefotaxime were 12.8 L/h and 39.4 L, respectively. The clearance for desacetylcefotaxime was 10.5 L/h. Standard dosing of 50 mg/kg q6h was only able to achieve the PK/PD target of 100% ƒT>MIC in patients >10 kg and with impaired renal function or patients of 40 kg with normal renal function.
Conclusions
Dosing recommendations support the use of extended or continuous infusion to achieve cefotaxime exposure suitable for bacterial killing in critically ill paediatric patients, including those with severe or deep-seated infection. An external validation of capillary microsampling demonstrated skin-prick sampling can facilitate data-rich pharmacokinetic studies.
Introduction
Severe infection can have long-term health consequences for paediatric patients, including impaired neurodevelopment and chronic disability.1,2 Effective antimicrobial dosing is one of the cornerstones of care to ensure therapeutic success in the treatment of severe infection. However, critical illness can manifest as extreme physiological derangements and this has the potential to impact on drug exposure, leading to treatment failure and/or antimicrobial resistance.3
Cefotaxime—a semi-synthetic, third-generation cephalosporin—is one of the most prescribed antimicrobials used to treat severe infections in critically ill paediatric patients.4–6 Approximately 43% of cefotaxime is bound to plasma proteins and it exhibits good penetration into body fluids and tissues.7,8 Cefotaxime is a hydrophilic drug with approximately 50%–60% eliminated by the kidneys by glomerular filtration followed by tubular secretion.9 Cefotaxime is metabolized by enzymatic hydrolysis of the O-acetyl group by an acetyl esterase in the liver to a pharmacologically active metabolite, desacetylcefotaxime.10,11 The metabolite is estimated as being between 0.5 and 10 times less microbiologically active than the parent compound, cefotaxime.8
Optimal cefotaxime dosing regimens target concentrations above the MIC throughout the dosing interval [pharmacokinetic/pharmacodynamic (PK/PD) ƒT>MIC], with targets of ≥60% ƒT>MIC and ≥100% ƒT>MIC for critically ill patients,12,13 and ≥100% ƒT > 4×MIC for critically ill patients with severe or deep-seated infection.14 Cefotaxime can be used to treat infections caused by Gram-positive and Gram-negative organisms, including meningitis caused by Escherichia coli, Neisseria meningitidis, Haemophilus influenzae and Streptococcus pneumoniae.14 Of the pathogens treated with cefotaxime, the reported MIC value according to the EUCAST is 1 mg/L for meningitis and indications other than meningitis caused by E. coli.15 Additionally, an MIC non-species-related breakpoint of 1 mg/L is commonly used for the treatment of a susceptible pathogen, with an MIC of ≥2 mg/L indicating a resistant pathogen.15 Current cefotaxime dosing regimens of 50 mg/kg every 6 h, with a maximum dose of 2 g (a total daily dose of up to 8 g), are commonly used for critically ill paediatric patients (>1 month old of life).16–18
The primary aims of this study were: (i) to describe the population pharmacokinetics of cefotaxime and desacetylcefotaxime in critically ill paediatric patients and to provide dosing recommendations for this special patient population; and (ii) to describe the suitability of using capillary microsampling for blood sampling compared with samples collected from an indwelling arterial or venous cannula (conventional sampling) by performing an external validation.
Patients and methods
Study design
A prospective, open-label, pharmacokinetic study was conducted at the paediatric ICU at the Queensland Children’s Hospital, Brisbane, Australia between March 2019 and September 2021. Critically ill patients between the ages of 1 month and 12 years and receiving intravenous cefotaxime, as prescribed by the treating physician, were included. Patients receiving extracorporeal membrane oxygenation, renal replacement therapy and peritoneal dialysis were excluded from this study. The research was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the Human Research & Ethics Committee of the Queensland Children’s Hospital (HREC/17/QRCH/45). Written informed consent was obtained from the parents or legal guardians prior to commencement of the study. Clinical characteristics were collected for patients including sex, age, height, weight, total bilirubin, haemoglobin, albumin, platelet count, white cell count, serum creatinine, alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, gamma-glutamyl transferase, prothrombin time, activated partial thromboplastin time, C-reactive protein, urinary creatinine, paediatric logistic organ dysfunction-2 (PELOD-2) score. For each patient, an estimated glomerular filtration rate (eGFR) was calculated using the bedside Schwartz equation (mL/min/1.73 m2).19,20
Conventional blood sampling and capillary microsampling
Paired blood samples using conventional sampling (from an arterial or venous line) and capillary microsamples21 (from a finger or heel prick) were simultaneously collected at five pre-defined timepoints: prior to administration of the cefotaxime dose (time 0), and then after the end of infusion at approximately 0.5, 2, 4 and 6 h. For capillary microsamples, the patient’s finger was cleaned with alcohol and punctured using a lancet device (either Haemolance Plus®, low flow 25G × 1.4 mm or BD microtainer Quikheel Infant Lancet, 1 mm × 2.5 mm). The finger was gently massaged and held below the heart of the patient until approximately 50 μL of blood was collected into a heparinized plastic capillary tube. The capillary microsample was centrifuged at 2000 g for 10 min to obtain plasma. The capillary tube was then snipped with scissors to isolate the plasma. For conventional plasma samples, approximately 0.6 mL of blood was obtained and collected into a heparinized 3 mL blood collection tube and centrifuged at 1500 g for 10 min to obtain plasma. After centrifugation, all plasma samples were transferred into screw-capped 2 mL polypropylene tubes and stored at –80°C until analysis.
Analysis of samples
Cefotaxime and desacetylcefotaxime concentrations were measured using a validated ultra-high-performance liquid chromatography tandem mass spectrometry (UHPLC-MS/MS) bioanalytical method22 in accordance with the guidelines provided by the EMA23 and the U.S. FDA.24 The linear concentration range was 0.5–500 mg/L and 0.2–10 mg/L for cefotaxime and desacetylcefotaxime, respectively. All intra-assay and inter-assay accuracy and precision were within 15% of acceptance criteria.
Pharmacokinetic model
Pmetrics version 1.5.0 (Laboratory of Applied Pharmacokinetics and Bioinformatics, Los Angeles, CA, USA) in RStudio (version 0.99.9.3) as a wrapper for R (version 3.3.1), Xcode (version 2.6.2) and the Intel Parallel Studio Fortran Compiler XE 2017 was used to develop a population pharmacokinetic model. One- and two-compartment models were constructed using non-parametric adaptive grid (NPAG) algorithms with total plasma cefotaxime and desacetylcefotaxime concentrations. A stepwise approach was followed to build the model to establish: (i) the structural base model, (ii) the best-fit error model, and (iii) development of a covariate model. Elimination from the central compartment and the rate of formation of the metabolite were modelled as first-order processes, and rate of formation of the metabolite was also tested for Michaelis–Menten kinetics. Lambda (additive) and gamma (multiplicative) error models were evaluated using a polynomial equation for SD as a function of observed concentration with observation weighting performed as error = SD × gamma or error = (SD2 + lambda2)0.5.
Pharmacokinetic model evaluation
Models were evaluated by the combination of diagnostic goodness-of-fit plots and statistical analysis. Diagnostic plots included scatter plots of observed-versus-predicted concentrations, visual predictive check plots. Statistical evaluation to compare different models was based on the regression coefficient r2, bias and imprecision, the log-likelihood ratio (–2*LL) and Akaike information criterion (AIC). The bias was measured using the mean weighted predicted – observed error. Imprecision was measured by using bias-adjusted and the mean weighted squared predicted – observed error. The percentage of shrinkage was measured using the total variation in the probability of each model.
Covariate screening
Covariate model building was performed using sequential assessment of biologically plausible clinical characteristics. Covariates were tested individually against the primary pharmacokinetic parameters and rate of metabolite formation, with inclusion based upon a statistically significant improvement in the AIC and –2*LL. The covariates evaluated, using allometric or linear scaling, against pharmacokinetic parameters were the clinical characteristics obtained for each patient and are listed in the Supplementary material (Table S1, available as Supplementary data at JAC Online).
External validation
An external validation was performed to describe the correlation between the measurement of cefotaxime concentrations obtained from capillary microsamples compared with the concentrations obtained using conventional sampling. For the external validation, the model developed using conventional sampling was used as a prior and Bayesian posterior simulations were calculated for each subject. A linear regression, the goodness-of-fit and the coefficient of determination were used to assess the correlation between the observed and predicted concentrations. Prediction errors were evaluated to describe bias (calculated as mean weighted prediction error, MWPE) and precision (Root Mean Square Predication error, RMSE) using Pmetrics. The acceptance criteria to establish validity were set to a bias of 20%, which has also been applied in a study by Guo et al. (2019).25 Scatter plots and Bland–Altman plots were used to visually inspect the predicted (model-simulated or measured conventional sampling) and observed (capillary microsampling) cefotaxime and desacetylcefotaxime concentrations for systematic bias.
Dosing simulations
Cefotaxime dosing regimens administered as a bolus dose every 4 or 6 h, as a 2 or 3 h extended infusion (EI), or as a continuous infusion (CI) across a range of eGFR (60, 100, 170 and 200 mL/min/1.73 m2) and a range of body weights (4, 10, 15, 20 and 40 kg) were evaluated using Monte Carlo simulations (n = 1000) in Pmetrics. Cefotaxime protein binding at 40% was used to calculate the probability of target attainment (PTA).26 For each dosing regimen, the PTA was calculated as the percentage of patients achieving a ≥60% ƒT>MIC, ≥100% ƒT>MIC or ≥100% ƒT > 4×MIC with MIC non-species-related breakpoint of 1 mg/L15 targeting success at 90%.
Results
A total of 36 critically ill paediatric patients [median age: 30.4 months (IQR age: 8.2–65.8 months)] with 160 conventional samples were included in the model development. Two plasma samples were below the lower limit of quantification for both cefotaxime and desacetylcefotaxime and these missing values have been simulated by Pmetrics during the analysis. Five samples were excluded from comparative paired analysis as the capillary microsample was haemolysed.27 In accordance with the clinical protocol, cefotaxime was administered to 34 patients as an intermittent infusion (duration mean, range: 0.19 h, 0.02–0.65 h) of 50 mg/kg every 6 h, two patients (weights 52.0 and 60.2 kg) received the maximum 2 g dose, with one of these patients receiving their dose as an EI over 3.75 h. Actual cefotaxime doses, including total daily doses, are reported in Table 1.
Table 1.
Clinical characteristics and patient information
| Demographic data | Mediana |
|---|---|
| Total patients | 36 |
| Sex, female/male, n (%) | 14/22 (39/61) |
| Age (months) | 30.4 (8.2–65.8) |
| Height (cm) | 86.5 (68.5–109) |
| Weight (kg) | 11.7 (8.1–18.2) |
| BSA (m2) | 0.5 (0.4–0.7) |
| Albumin (g/L) | 30 (24–33) |
| Bilirubin (μmol/L) | 6.0 (3.5–9.5) |
| Haemoglobin (g/L) | 106 (97–116) |
| Platelet count (×109/L) | 251 (189–306) |
| Serum creatinine (mL/min) | 24.0 (17.5–28.5) |
| eGFR (mL/min/1.73 m2) | 143 (109–259) |
| Illness severity score PELOD-2 score (on admission to the ICU) | 4 (2–6) |
| Patient mechanically ventilated at the time of dosing, n (%) | 23 (64) |
| Invasive ventilation at the time of dosing, n (%) | 22 (61) |
| Vasopressors/Inotropes at the time of dosing, n (%) | 8 (22) |
| Number of doses prior to PK sampling intervalb | 4 (2–11) |
| Dose prior to PK sampling interval (mg)b | 605 (404–910) |
| Daily total dose (mg)b | 2420 (1615–3640) |
PELOD-2 score, Paediatric Logistic Organ Dysfunction; eGFR, estimated glomerular filtration rate (indexed to BSA 1.73, calculated using bedside Schwartz equation).
Data displayed as mean with IQR (Q1–Q3) or n (%) as appropriate.
Data displayed as mean (minimum – maximum).
From the total study cohort, 58% of the patients (n = 21) had augmented renal function (eGFR >130 mL/min/1.73 m2), 33% (n = 12) of the patients had normal renal function (eGFR ranging between 80 and 130 mL/min/1.73 m2), while 8% (n = 3) of the patients had impaired renal function (eGFR values <80 mL/min/1.73 m2). Of the patients recruited, 39% (n = 14) weighed less than 10 kg, 47% of the patients weighed between 10 and 30 kg (n = 17) and 14% of the patients (n = 5) had a body weight above 30 kg. The baseline clinical characteristics and patient information are presented in Table 1.
Plasma–concentration data were best described using a one-compartmental model with first-order elimination for both cefotaxime and desacetylcefotaxime. For the model, empirical inclusion of weight normalized to 70 kg with allometric scaling (0.75) on clearance and linear scaling on volume of distribution was used on cefotaxime. As the volume of distribution for desacetylcefotaxime could not be estimated, it was assumed to be equal to the volume of distribution of cefotaxime.28 The inclusion of the patient population mean-adjusted eGFR (eGFR/150) was accepted as a covariate on cefotaxime clearance (CL1) as it resulted in a decrease in log-likelihood of 11.0. The inclusion of normalized body surface area (BSA/1.73 m2) as a covariate on desacetylcefotaxime clearance decreased the log-likelihood by 15.0. The goodness-of-fit of the final models were confirmed with the diagnostic plots shown in Figure 1. The final Pmetrics model and the results obtained for the log-likelihood and AIC during the model development are provided in the Supplementary material (Tables S1 and S2, respectively). The support points of the final covariate model are provided in the Supplementary material (Table S3).
Figure 1.
Diagnostic plots for the final covariate model for plasma concentrations (mg/L) of cefotaxime (top) and desacetylcefotaxime (bottom). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
The primary pharmacokinetic parameters are summarized in Table 2 and the visual predicted check plots for cefotaxime and desacetylcefotaxime are provided in the Supplementary material (Figure S1). Based on the visual predictive check, 94.7% of the observations for cefotaxime and 96.6% of the observations for desacetylcefotaxime were within the 5th and 95th of simulated percentiles. Individual plots are presented in the Supplementary material (Figures S2–S5). A dose selection flow chart is provided in Figure 2. PTA values for cefotaxime are presented across a range of patient weight and eGFR based on the PK/PD targets of ≥60% ƒT>MIC (Table 3), ≥100% ƒT>MIC (Table 4) and ≥100% ƒT>4×MIC (Table 5).
Table 2.
Population pharmacokinetic primary parameters of cefotaxime and desacetylcefotaxime concentrations of critically ill paediatric patients
| Parameter | Mean | SD | CV (%) | Median | Shrink (%) |
|---|---|---|---|---|---|
| CL1 (L/h) | 12.8 | 6.17 | 48.3 | 11.7 | 0.312 |
| CL2 (L/h) | 10.5 | 6.91 | 65.9 | 9.75 | 0.985 |
| V1 (L) | 39.4 | 20.7 | 52.6 | 34.0 | 1.08 |
| K12 (h–1) | 0.199 | 0.155 | 77.7 | 0.169 | 0.686 |
CL1, cefotaxime clearance; CL2, desacetylcefotaxime clearance; V1, central volume of cefotaxime; K12, rate of formation of desacetylcefotaxime; CV, coefficient of variation; Shrink %, model shrinkage.
Clearance and volume of distribution are standardized for an adult patient body weight of 70 kg
Figure 2.
Flow chart to support dosing recommendations in Tables 3–5. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Table 3.
Dose simulations (with PTA results, %) for pathogens susceptible to cefotaxime (MIC target of 1 mg/L) to achieve a PK/PD target of 60% ƒT>MIC
| Dosing regimen | eGFR | ||||
|---|---|---|---|---|---|
| WT | 60 | 100 | 170 | 200 | |
| 50 mg/kg q6h | 4 | 98.4 | 91.8 | 69.3 | 56.7 |
| 10 | 100 | 96.3 | 84.3 | 74.7 | |
| 15 | 100 | 97.5 | 87.5 | 81.3 | |
| 20 | 100 | 98.0 | 89.1 | 84.5 | |
| 40 | 100 | 98.7 | 92.3 | 89.3 | |
| 50 mg/kg q4h | 4 | 100 | 100 | 90.1 | 85.3 |
| 10 | 100 | 100 | 96.4 | 91.3 | |
| 15 | 100 | 100 | 99.5 | 94.4 | |
| 20 | 100 | 100 | 99.8 | 96.7 | |
| 40 | 100 | 100 | 100 | 99.8 | |
| 50 mg/kg EI q6h | 4 | 100 | 100 | 100 | 100 |
| 10 | 100 | 100 | 100 | 100 | |
| 15 | 100 | 100 | 100 | 100 | |
| 20 | 100 | 100 | 100 | 100 | |
| 40 | 100 | 100 | 100 | 100 | |
| 50 mg/kg EI q4h | 4 | 100 | 100 | 100 | 100 |
| 10 | 100 | 100 | 100 | 100 | |
| 15 | 100 | 100 | 100 | 100 | |
| 20 | 100 | 100 | 100 | 100 | |
| 40 | 100 | 100 | 100 | 100 | |
| 50 mg/kg CI | 4 | 99.7 | 98.7 | 96.5 | 96.4 |
| 10 | 100 | 98.7 | 97.2 | 96.5 | |
| 15 | 100 | 99.2 | 98.7 | 96.8 | |
| 20 | 100 | 100 | 99.4 | 97.5 | |
| 40 | 100 | 99.9 | 98.7 | 98.7 | |
ƒT >MIC, fraction of time (ƒT) where the drug exceeds the MIC; WT, weight (kg); q6h, dosed every 6 h; q4h, dosed every 4 h; EI, extended infusion for half the total dosing interval; CI, continuous infusion with dose calculated as total daily dose.
PK/PD targets ≥90% ƒT>MIC have been highlighted in bold.
Table 4.
Dose simulations (with PTA results, %) for pathogens susceptible to cefotaxime (MIC target of 1 mg/L) to achieve a PK/PD target of 100% ƒT>MIC
| Dosing regimen | eGFR | ||||
|---|---|---|---|---|---|
| WT | 60 | 100 | 170 | 200 | |
| 50 mg/kg q6h | 4 | 89.7 | 55.8 | 29.1 | 18.4 |
| 10 | 93.0 | 74.0 | 41.6 | 33.0 | |
| 15 | 94.9 | 81.3 | 46.5 | 38.2 | |
| 20 | 95.9 | 86.4 | 49.2 | 42.2 | |
| 40 | 97.2 | 91.4 | 58.6 | 49.7 | |
| 50 mg/kg q4h | 4 | 97.2 | 88.1 | 59.5 | 51.6 |
| 10 | 98.1 | 91.4 | 76.0 | 64.2 | |
| 15 | 99.3 | 94.9 | 81.7 | 71.2 | |
| 20 | 99.8 | 96.5 | 84.7 | 76.3 | |
| 40 | 100 | 97.5 | 88.8 | 85.0 | |
| 50 mg/kg EI q6h | 4 | 97.3 | 89.0 | 56.3 | 47.6 |
| 10 | 97.6 | 92.8 | 75.9 | 60.5 | |
| 15 | 97.9 | 95.4 | 81.9 | 69.1 | |
| 20 | 98.2 | 96.2 | 84.7 | 76.2 | |
| 40 | 99.1 | 97.4 | 90.0 | 85.4 | |
| 50 mg/kg EI q4h | 4 | 100 | 99.7 | 85.8 | 79.9 |
| 10 | 100 | 100 | 91.5 | 87.9 | |
| 15 | 100 | 100 | 94.8 | 90.4 | |
| 20 | 100 | 100 | 96.7 | 91.5 | |
| 40 | 100 | 100 | 99.8 | 96.9 | |
| 50 mg/kg CI | 4 | 99.7 | 98.7 | 96.5 | 96.4 |
| 10 | 100 | 98.7 | 97.2 | 96.5 | |
| 15 | 100 | 99.2 | 98.7 | 96.8 | |
| 20 | 100 | 100 | 99.4 | 97.5 | |
| 40 | 100 | 99.9 | 98.7 | 98.7 | |
ƒT >MIC, fraction of time (ƒT) where the drug exceeds the MIC; WT, weight (kg); q6h, dosed every 6 h; q4h, dosed every 4 h; EI, extended infusion for half the total dosing interval; CI, continuous infusion with dose calculated as total daily dose.
PK/PD targets ≥90% ƒT>MIC have been highlighted in bold.
Table 5.
Dose simulations (with PTA results, %) for pathogens susceptible to cefotaxime (MIC target of 1 mg/L) to achieve a PK/PD target of 100% ƒT>4×MIC
| Dosing regimen | eGFR | ||||
|---|---|---|---|---|---|
| WT | 60 | 100 | 170 | 200 | |
| 50 mg/kg q6h | 4 | 48.0 | 30.8 | 3.7 | 5.0 |
| 10 | 59.9 | 39.1 | 13.9 | 6.7 | |
| 15 | 67.9 | 42.0 | 19.0 | 10.7 | |
| 20 | 72.6 | 43.7 | 23.5 | 14.6 | |
| 40 | 83.6 | 51.5 | 32.3 | 23.9 | |
| 50 mg/kg q4h | 4 | 89.1 | 58.0 | 30.3 | 20.9 |
| 10 | 92.3 | 75.6 | 43.4 | 34.1 | |
| 15 | 94.5 | 83.4 | 48.3 | 40.6 | |
| 20 | 96.1 | 86.1 | 51.3 | 43.9 | |
| 40 | 97.0 | 90.1 | 60.5 | 51.4 | |
| 50 mg/kg EI q6h | 4 | 86.4 | 51.1 | 24.4 | 14.3 |
| 10 | 92.7 | 66.4 | 37.8 | 28.5 | |
| 15 | 93.9 | 73.6 | 43.3 | 34.7 | |
| 20 | 95.3 | 79.9 | 45.9 | 38.2 | |
| 40 | 97.2 | 88.7 | 53.4 | 46.5 | |
| 50 mg/kg EI q4h | 4 | 96.9 | 88.0 | 56.0 | 47.9 |
| 10 | 97.0 | 91.7 | 70.3 | 59.3 | |
| 15 | 97.0 | 94.3 | 79.0 | 66.6 | |
| 20 | 97.0 | 96.6 | 82.9 | 71.6 | |
| 40 | 97.4 | 97.0 | 89.4 | 83.3 | |
| 50 mg/kg CI | 4 | 80.1 | 46.9 | 20.7 | 12.4 |
| 10 | 85.7 | 65.4 | 30.2 | 23.5 | |
| 15 | 88.3 | 72.0 | 35.1 | 27.7 | |
| 20 | 89.2 | 75.7 | 39.1 | 30.9 | |
| 40 | 90.0 | 81.9 | 49.6 | 39.9 | |
| 100 mg/kg CI | 4 | 95.5 | 90.4 | 75.3 | 60.5 |
| 10 | 96.4 | 93.0 | 85.1 | 80.1 | |
| 15 | 96.4 | 94.2 | 86.2 | 84.3 | |
| 20 | 96.4 | 94.7 | 87.9 | 85.1 | |
| 40 | 96.5 | 95.9 | 90.7 | 88.3 | |
| 200 mg/kg CI | 4 | 99.7 | 98.7 | 96.5 | 96.4 |
| 10 | 100 | 98.7 | 97.2 | 96.5 | |
| 15 | 100 | 99.2 | 98.7 | 96.8 | |
| 20 | 100 | 99.4 | 98.7 | 97.5 | |
| 40 | 100 | 99.9 | 98.7 | 98.7 | |
ƒT >MIC, fraction of time (ƒT) where the drug exceeds the MIC; WT, weight (kg); q6h, dosed every 6 h; q4h, dosed every 4 h; EI, extended infusion for half the total dosing interval; CI, continuous infusion with dose calculated as total daily dose.
PK/PD targets ≥90% ƒT > 4×MIC have been highlighted in bold.
A linear regression of the model-predicted (using conventional sampling) versus observed (capillary microsampling) cefotaxime and desacetylcefotaxime concentrations are presented in Figure 3. The 95% CI for the intercept for cefotaxime and desacetylcefotaxime are –1.8 to 0.45 and –0.04 to 0.87 mg/L, respectively, and the slope of the regression line is close to 1 for both cefotaxime (95% CI 1.03 to 1.09) and desacetylcefotaxime (95% CI 0.927 to 0.999). The regression line crosses the line of equality for both cefotaxime and desacetylcefotaxime. The results of the external validation found for cefotaxime there was a bias (MWPE) of –0.137 mg/L (P = 0.1129, different than 0) and a precision (RMSE) of 14.6% and for desacetylcefotaxime there was a bias (MWPE) of –0.024 mg/L (P = 0.0967, different than 0) and a precision (RMSE) of 12.9%, when comparing the observed concentrations (capillary microsampling) with the model-predicted concentrations (using conventional sampling). Bland–Altman weighted residual plots of the observed concentrations (capillary microsampling) to the model-predicted concentrations (using conventional sampling) are presented in Figure 4. Scatter plots and Bland–Altman plots of the measured cefotaxime and desacetylcefotaxime concentrations from paired conventional samples and capillary microsamples are presented in Figure 5.
Figure 3.
External validation linear regression plots of cefotaxime (top) and desacetylcefotaxime (bottom) comparing observed concentrations (capillary microsampling) with model-predicted concentrations (using conventional sampling). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Figure 4.
Bland–Altman weighted residual plots for external validation of cefotaxime (top) and desacetylcefotaxime (bottom) comparing observed concentrations (capillary microsampling) with model-predicted concentrations (using conventional sampling). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Figure 5.
Scatter plots (top) and Bland–Altman plots (bottom) of conventional sampling and capillary microsampling (CMS) of cefotaxime (left) and desacetylcefotaxime (right). ULoA, upper 95% limit of agreement; LLoA, lower 95% limit of agreement.
Discussion
This study enhances our understanding of the pharmacokinetics of cefotaxime and its active metabolite, desacetylcefotaxime, and optimized dosing in critically ill paediatric patients,28–30 through the use of rich blood sampling to build the pharmacokinetic profiles (n = 5 samples/patient, range 3 to 5). Based on this study design, we have developed a model that supports the inclusion of eGFR on the clearance of cefotaxime and BSA on the clearance of desacetylcefotaxime. Additionally, our blood sampling strategy has demonstrated that the application of capillary microsampling to obtain blood from a finger or heel prick correlates with concentrations obtained using conventional sampling techniques.
A one-compartmental model with first-order elimination best fitted the data to describe the pharmacokinetics of cefotaxime and desacetylcefotaxime. In paediatric patients only one other study has used a population pharmacokinetic approach, which used a similar approach applied here, with a one-compartmental model and setting the volume of distribution of desacetylcefotaxime to equal that of cefotaxime.28 Studies of β-lactam antimicrobials in critically ill adults have described the pharmacokinetics using a two-compartment model31–34 and this difference may be due to the use of a single dosing event in the paediatric studies.
Cefotaxime clearance was similar to that previously reported in critically ill paediatric patients, where clearance ranged from 6.9 to 13.7 L/h.28–30 All studies report variable cefotaxime clearance in critically ill patients. Desacetylcefotaxime clearance was higher in our study, compared with the study by Beranger et al.,28 which reported a clearance of 4.2 L/h. Both studies had patients with a similar median renal function, so this may be a result of the increased definition allowed through the use of 2–4 samples per patient during the elimination phase in our study. Both our study and the study by Beranger et al.28 have found the clearance of the active metabolite was highly variable. Cefotaxime and desacetylcefotaxime are eliminated by the kidneys and estimated creatinine clearance was able to be included on the clearance of cefotaxime. This finding concords with data from critically ill adult patients that have shown clearance to be proportional to estimated creatinine clearance.35 The use of eGFR calculated with the bedside Schwartz equation in this paediatric model is advantageous as it includes a factor for age as a descriptor for renal maturation. No other studies have found an association between BSA and clearance of desacetylcefotaxime, although a relationship between BSA and liver volume has been recently identified in children36 and this may account for the relationship identified for the metabolite in our study.
The volume of distribution for cefotaxime was variable in our patient cohort, but similar to other studies in critically ill paediatric patients, which have reported it ranging from 21.4 to 96 L.28,30 Studies in critically ill patients have found the volume of distribution of hydrophilic antimicrobials, such as cefotaxime, can be highly altered due to a distribution of fluids into the interstitial space.37 This may occur in patients suffering from capillary leak syndrome caused by severe sepsis or critically ill patients requiring extensive fluid resuscitation.12
Based on a PK/PD target of ≥100% ƒT>MIC, with a non-species-related breakpoint MIC of 1 mg/L for susceptible organisms,38 14% of patients (n = 5) failed to achieve a target of 1 mg/L for cefotaxime and 39% of patients (n = 14) failed to achieve a target of 4 mg/L across the dosing interval. In our study cohort, 58% of patients had augmented renal clearance (defined as eGFR values above 130 mL/min/1.73 m2).39 This result concords with other studies of both critically ill adults and paediatric patients.39–41
Dosing simulations, using a range of weights and eGFR, support the use of shorter dosing intervals to achieve a PK/PD target of ≥60% ƒT>MIC. For critically ill paediatric patients with normal or impaired renal function, to achieve a PK/PD target of ≥100% ƒT>MIC a 4 hourly dosing interval or an EI with a 6 hourly interval was able to provide sufficient cefotaxime coverage (using) for most patient weight and eGFR ranges. However, critically ill patients with augmented renal clearance, or neonatal patients with normal renal clearance required both a 4 hourly dosing interval combined with a 2 h EI to achieve the PK/PD target of ≥100% ƒT>MIC. More aggressive PK/PD targets (≥100% ƒT > 4×MIC) that may be suitable for critically ill patients with severe or deep-seated infection were not achieved using standard dosing of 50 mg/kg every 6 h. For critically ill paediatric patients with normal or impaired renal function a 4 hourly dosing interval combined with a 2 h EI achieved target in all patient weight ranges, except for neonatal patients with normal renal function. For critically ill patients with augmented renal clearance or neonatal patients with normal renal clearance, an CI with a total daily dose of 100–200 mg/kg was required to achieve this PK/PD target. Previous studies have demonstrated the challenge of achieving effective PK/PD targets for cefotaxime28 and other β-lactam antimicrobials42–45 in critically ill paediatric patients with higher eGFR. The study by Beranger et al.,28 targeting ≥100% ƒT>MIC and ≥100% ƒT > 4×MIC for pathogens with an MIC of 0.5 mg/L, recommended the use of CI to achieve PK/PD targets in a similar patient population.
From the external validation, there is no systematic bias evident when comparing the concentration results of cefotaxime or desacetylcefotaxime obtained by conventional sampling to samples obtained by finger or heel prick using capillary microsampling and the calculated bias met the pre-established acceptance criteria, both when performed using the external validation methodology and as demonstrated in the Bland–Altman analysis of the measured concentrations. While the Bland–Altman plots of weighted residual error over the predicted concentration range show a greater imprecision at low cefotaxime and desacetylcefotaxime concentrations, the histograms show that, overall, there is a normal distribution of bias for both cefotaxime and desacetylcefotaxime across the predicted concentration range.
This study has several limitations. We measured total cefotaxime and desacetylcefotaxime concentrations in plasma samples and have not quantified the unbound concentrations. A consequence of this is that we have been unable to calculate protein binding for our patients and have set cefotaxime protein binding to 40% for the purpose of performing the PTA calculations and this may impact on the accuracy of the resultant dosing recommendations. Additionally, we did not collect and isolate the pathogens that caused the infections in the patients enrolled in the study and have therefore applied the PK/PD non-species-related breakpoints from the EUCAST38 as targets to derive suitable dosing recommendations.
The strengths of this study are that it is the first known pharmacokinetic study describing cefotaxime and desacetylcefotaxime in critically ill paediatric patients using rich sampling for blood collection. Additionally, we demonstrate that the use of capillary microsampling can be used perform pharmacokinetic studies and there is the potential for this to facilitate more studies in neonatal and paediatric patients.46
Standard dosing of 50 mg/kg every 6 h was only able to achieve the PK/PD target commonly used in intensive care of 100% ƒT>MIC in patients >10 kg and with impaired renal function or patients of 40 kg with normal renal function. Dosing recommendations support the use of shorter intervals or EI or CI to achieve cefotaxime exposure suitable for bacterial killing in critically ill paediatric patients, including patients with severe or deep-seated infection. Further research is required to confirm the suitability of these dosing recommendations. If implemented, we would recommend supporting patient care with therapeutic drug monitoring. Capillary microsampling for blood collection was externally validated and demonstrated the application of a finger/heel prick sample can facilitate data-rich pharmacokinetic studies.
Supplementary Material
Acknowledgements
We would like to acknowledge the critically ill children who participated in the study, their parents or legal guardians and the research nurses of the PICU at the Queensland Children’s Hospital, Brisbane, Australia for their support and assistance with sample collection and other relevant tasks for this study. We also acknowledge the assistance of the staff of the Mass Spectrometry Facility of UQ Centre for Clinical Research, Brisbane, Australia.
Contributor Information
Yarmarly C Guerra Valero, UQ Centre for Clinical Research, The University of Queensland, Brisbane, Australia.
Tavey Dorofaeff, UQ Centre for Clinical Research, The University of Queensland, Brisbane, Australia; Paediatric Intensive Care, Queensland Children’s Hospital, Brisbane, Australia.
Mark G Coulthard, Paediatric Intensive Care, Queensland Children’s Hospital, Brisbane, Australia; Mayne Academy of Paediatrics, Faculty of Medicine, The University of Queensland, Brisbane, Australia.
Louise Sparkes, Paediatric Intensive Care, Queensland Children’s Hospital, Brisbane, Australia.
Jeffrey Lipman, UQ Centre for Clinical Research, The University of Queensland, Brisbane, Australia; Department of Intensive Care Medicine, Royal Brisbane & Women’s Hospital, Brisbane, Australia; Jamieson Trauma Institute, Royal Brisbane & Women’s Hospital, Herston, QLD 4029, Australia.
Steven C Wallis, UQ Centre for Clinical Research, The University of Queensland, Brisbane, Australia.
Jason A Roberts, UQ Centre for Clinical Research, The University of Queensland, Brisbane, Australia; Department of Intensive Care Medicine, Royal Brisbane & Women’s Hospital, Brisbane, Australia; Department of Pharmacy, Royal Brisbane & Women’s Hospital, Brisbane, Australia.
Suzanne L Parker, UQ Centre for Clinical Research, The University of Queensland, Brisbane, Australia.
Funding
This study was funded by a grant from the Children’s Hospital Foundation, Queensland. Y.C.G.V. is a recipient of a Research Training Scholarship from The University of Queensland. S.L.P. is a recipient of an Early Career Research Fellowship from the Australian National Health and Medical Research Council (APP1142757). J.A.R. is a recipient of an Australian National Health and Medical Research Council Fellowship (APP1048652).
Transparency declarations
None to declare.
Supplementary data
Tables S1 to S3 and Figures S1 to S5 are available as Supplementary data at JAC Online.
References
- 1. Schlapbach LJ, Straney L, Alexander Jet al. Mortality related to invasive infections, sepsis, and septic shock in critically ill children in Australia and New Zealand 2002-13: a multicentre retrospective cohort study. Lancet Infect Dis 2015; 15: 46–54. [DOI] [PubMed] [Google Scholar]
- 2. Jablensky A, Johnson R, Bunney Wet al. Neurological, Psychiatric, and Developmental Disorders; Meeting the Challenge in the Developing World. National Academy Press, 2001. [PubMed] [Google Scholar]
- 3. Roberts JA, Lipman J. Pharmacokinetic issues for antibiotics in the critically ill patient. Crit Care Med 2009; 37: 840–51. [DOI] [PubMed] [Google Scholar]
- 4. Osowicki J, Gwee A, Noronha Jet al. Australia-wide point prevalence survey of antimicrobial prescribing in neonatal units: how much and how good? Pediatr Infect Dis J 2015; 34: e185–90. [DOI] [PubMed] [Google Scholar]
- 5. Shein SL, Kong M, McKee Bet al. Antibiotic prescription in young children with respiratory syncytial virus-associated respiratory failure and associated outcomes. Pediatr Crit Care Med 2019; 20: 101–9. [DOI] [PubMed] [Google Scholar]
- 6. Versporten A, Bielicki J, Drapier Net al. The Worldwide Antibiotic Resistance and Prescribing in European Children (ARPEC) point prevalence survey: developing hospital-quality indicators of antibiotic prescribing for children. J Antimicrob Chemother 2016; 71: 1106–17. [DOI] [PubMed] [Google Scholar]
- 7. Novick WJ Jr. Levels of cefotaxime in body fluids and tissues: a review. Rev Infect Dis 1982; 4Suppl 2: S346–53. [DOI] [PubMed] [Google Scholar]
- 8. Wise R, Wills PJ, Andrews JMet al. Activity of the cefotaxime (HR756) desacetyl metabolite compared with those of cefotaxime and other cephalosporins. Antimicrob Agents Chemother 1980; 17: 84–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. LeFrock JL, Prince RA, Left RD. Mechanism of action, antimicrobial activity, pharmacology, adverse effects, and clinical efficacy of cefotaxime. Pharmacotherapy 1982; 2: 174–84. [DOI] [PubMed] [Google Scholar]
- 10. Welch WD, Bawdon RE. Cefotaxime metabolism by hemolyzed blood: quantitation and inhibition of the deacetylation reaction. Diagn Microbiol Infect Dis 1986; 4: 119–24. [DOI] [PubMed] [Google Scholar]
- 11. Reeves DS, White LO, Holt HAet al. Human metabolism of cefotaxime. J Antimicrob Chemother 1980; 6: 93–101. [DOI] [PubMed] [Google Scholar]
- 12. Roberts JA, Abdul-Aziz MH, Lipman Jet al. Individualised antibiotic dosing for patients who are critically ill: challenges and potential solutions. Lancet Infect Dis 2014; 14: 498–509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Guilhaumou R, Benaboud S, Bennis Yet al. Optimization of the treatment with β-lactam antibiotics in critically ill patients—guidelines from the French Society of Pharmacology and Therapeutics (Société Française de Pharmacologie et Thérapeutique—SFPT) and the French Society of Anaesthesia and Intensive Care Medicine (Société Française d’Anesthésie et Réanimation—SFAR). Crit Care 2019; 23: 104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Huttner A, Harbarth S, Hope WWet al. Therapeutic drug monitoring of the β-lactam antibiotics: what is the evidence and which patients should we be using it for? J Antimicrob Chemother 2015; 70: 3178–83. [DOI] [PubMed] [Google Scholar]
- 15. Mouton JW, Brown D, Apfalter Pet al. The role of pharmacokinetics/pharmacodynamics in setting clinical MIC breakpoints: the EUCAST approach. Clin Microbiol Infect 2012; 18: E37–45. [DOI] [PubMed] [Google Scholar]
- 16. Hartman SJF, Boeddha NP, Ekinci Eet al. Target attainment of cefotaxime in critically ill children with meningococcal septic shock as a model for cefotaxime dosing in severe pediatric sepsis. Eur J Clin Microbiol 2019; 38: 1255–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Nau R, Prange H, Muth Pet al. Passage of cefotaxime and ceftriaxone into cerebrospinal fluid of patients with uninflamed meninges. Antimicrob Agents Chemother 1993; 37: 1518–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. The Royal Children’s Hospital Melbourne . Cefotaxime. Paediatric Infant Perinatal Emergency Retrieval [Internet]. https://www.rch.org.au/piper/neonatal_medication_guidelines/Cefotaxime/.
- 19. Schwartz G, Haycock G, Edelmann C Jret al. A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics 1976; 58: 259–63. [PubMed] [Google Scholar]
- 20. Arant BS Jr, Edelmann CM Jr, Spitzer A. The congruence of creatinine and inulin clearances in children: use of the technicon auto analyzer. J Pediatr 1972; 81: 559–61. [DOI] [PubMed] [Google Scholar]
- 21. Valero YCG, Roberts JA, Lipman Jet al. Analysis of capillary microsamples obtained from a skin-prick to measure vancomycin concentrations as a valid alternative to conventional sampling: a bridging study. J Pharm Biomed Anal 2019; 169: 288–92. [DOI] [PubMed] [Google Scholar]
- 22. Guerra Valero YC, Dorofaeff T, Roberts JAet al. Development and validation of a UHPLC-MS/MS method to measure cefotaxime and metabolite desacetylcefotaxime in blood plasma: a pilot study suitable for capillary microsampling in critically ill children. Anal Bioanal Chem 2021; 413: 4483–91. [DOI] [PubMed] [Google Scholar]
- 23. EMA . Guideline on bioanalytical method validation [Internet]. 2011. https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-bioanalytical-method-validation_en.pdf.
- 24. FDA . Guidance for Industry: Bioanalytical Method Validation. 2018. https://www.fda.gov/files/drugs/published/Bioanalytical-Method-Validation-Guidance-for-Industry.pdf.
- 25. Guo T, van Hest RM, Roggeveen LFet al. External evaluation of population pharmacokinetic models of vancomycin in large cohorts of intensive care unit patients. Antimicrob Agents Chemother 2019; 63: e02543-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Bergan T. Pharmacokinetic properties of the cephalosporins. Drugs 1987; 34: 89–104. [DOI] [PubMed] [Google Scholar]
- 27. Ingelse B, Barroso B, Gray Net al. European Bioanalysis Forum: recommendation on dealing with hemolyzed and hyperlipidemic matrices. Bioanalysis 2014; 6: 3113–20. [DOI] [PubMed] [Google Scholar]
- 28. Beranger A, Oualha M, Urien Set al. Population pharmacokinetic model to optimize cefotaxime dosing regimen in critically ill children. Clin Pharmacokinet 2018; 57: 867–75. [DOI] [PubMed] [Google Scholar]
- 29. Bertels RA, Semmekrot BA, Gerrits GPet al. Serum concentrations of cefotaxime and its metabolite desacetyl-cefotaxime in infants and children during continuous infusion. Infection 2008; 36: 415–20. [DOI] [PubMed] [Google Scholar]
- 30. Kafetzis DA, Brater DC, Kanarios Jet al. Clinical pharmacology of cefotaxime in pediatric patients. Antimicrob Agents Chemother 1981; 20: 487–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Cheng V, Abdul-Aziz MH, Burrows Fet al. Population pharmacokinetics and dosing simulations of ceftriaxone in critically ill patients receiving extracorporeal membrane oxygenation (An ASAP ECMO Study). Clin Pharmacokinet 2022. 10.1007/s40262-021-01106-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Tsai D, Stewart P, Goud Ret al. Optimising meropenem dosing in critically ill Australian Indigenous patients with severe sepsis. Int J Antimicrob Agents 2016; 48: 542–6. [DOI] [PubMed] [Google Scholar]
- 33. Jager NG, Van Hest RM, Xie Jet al. Optimization of flucloxacillin dosing regimens in critically ill patients using population pharmacokinetic modelling of total and unbound concentrations. J Antimicrob Chemother 2020; 75: 2641–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Al-Shaer MH, Neely MN, Liu Jet al. Population pharmacokinetics and target attainment of cefepime in critically ill patients and guidance for initial dosing. Antimicrob Agents Chemother 2020; 64: e00745-20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Swartling M, Smekal A-K, Furebring Met al. Population pharmacokinetics of cefotaxime in intensive care patients. Eur J Clin Pharmacol 2021; 78: 251–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Hosey-Cojocari C, Chan SS, Friesen CSet al. Are body surface area (BSA) based estimates of liver volume applicable to children with overweight or obesity? An in-vivo validation study. Clin Transl Sci 2021; 14: 2008–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Roberts JA, Kruger P, Paterson DLet al. Antibiotic resistance—what’s dosing got to do with it? Crit Care Med 2008; 36: 2433–40. [DOI] [PubMed] [Google Scholar]
- 38. EUCAST . Clinical breakpoints. Breakpoints and guidance [Internet]. 2021. https://www.eucast.org/clinical_breakpoints/.
- 39. Avedissian SN, Bradley E, Zhang Det al. Augmented renal clearance using population-based pharmacokinetic modeling in critically ill pediatric patients. Pediatr Crit Care Med 2017; 18: e388–94. [DOI] [PubMed] [Google Scholar]
- 40. Udy AA, Varghese JM, Altukroni Met al. Subtherapeutic initial β-lactam concentrations in select critically ill patients: association between augmented renal clearance and low trough drug concentrations. Chest 2012; 142: 30–9. [DOI] [PubMed] [Google Scholar]
- 41. Dhont E, Van Der Heggen T, De Jaeger Aet al. Augmented renal clearance in pediatric intensive care: are we undertreating our sickest patients? Pediatr Nephrol 2020; 35: 25–39. [DOI] [PubMed] [Google Scholar]
- 42. Cies JJ, Moore WS, Enache Aet al. Population pharmacokinetics and pharmacodynamic target attainment of meropenem in critically ill young children. J Pediatr Pharmacol Ther 2017; 22: 276–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Cies JJ, Moore WS, Enache Aet al. Ceftaroline for suspected or confirmed invasive methicillin-resistant Staphylococcus aureus: a pharmacokinetic case series. Pediatr Crit Care Med 2018; 19: e292–9. [DOI] [PubMed] [Google Scholar]
- 44. Cies JJ, Shankar V, Schlichting Cet al. Population pharmacokinetics of piperacillin/tazobactam in critically ill young children. Pediatr Infect Dis J 2014; 33: 168–73. [DOI] [PubMed] [Google Scholar]
- 45. De Cock PA, Van Dijkman SC, De Jaeger Aet al. Dose optimization of piperacillin/tazobactam in critically ill children. J Antimicrob Chemother 2017; 72: 2002–11. [DOI] [PubMed] [Google Scholar]
- 46. Dorofaeff T, Bandini RM, Lipman Jet al. Uncertainty in antibiotic dosing in critically ill neonate and pediatric patients: can microsampling provide the answers? Clin Ther 2016; 38: 1961–75. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.