Background The aim was to describe intravenous paracetamol pharmacokinetics, determine major covariates and suggest a dosing regimen for (pre)term neonates.
Methods A population pharmacokinetic analysis of 943 paracetamol observations from 158 neonates (27–45 weeks' postmenstrual age (PMA)) was undertaken using non-linear mixed effects models. Data from three published studies were pooled with newly collected time–concentration points during repeated intravenous paracetamol administration.
Results A two-compartment (central, peripheral) linear disposition model was used. Population parameter estimates (between-subject variability, %) were central volume 51.9 l/70 kg (21.6%), peripheral volume of distribution 22.7 l/70 kg, clearance 5 l/h/70 kg (40%) and intercompartment clearance 16.2 l/h/70 kg. Covariate information predicts 60.9% of clearance variance. Weight was used to predict patient size and was the major covariate contributing 57.5% of variance. Clearance expressed as mg/kg/h increases only slightly with PMA (0.138 l/kg/h at 28 weeks' PMA to 0.167 l/kg/h at 44 weeks' PMA) and contributes only 2.2% of variance. High unconjugated bilirubin levels contributed an additional 1.2%.
Conclusions Patient size (predicted by weight) is the major covariate of clearance variance in neonates. Using these estimates, a mean paracetamol serum concentration of 11 mg/l is predicted in neonates of 32–44 weeks' PMA given a standard dose of intravenous paracetamol of 10 mg/kg every 6 h. Safety data for this drug are limited in neonates. Continued surveillance therefore remains essential.
Statistics from Altmetric.com
In general, neonates have an overall low clearance capacity.1 Between-subject variability (BSV) is explained by covariates such as size, weight, organ function, coadministration of drugs, genetic polymorphisms, growth restriction or disease characteristics.2,–,6 Paracetamol is an antipyretic and analgesic agent, commonly prescribed in infants, including neonates. Doses in neonates vary between practitioners and industry recommendations, potentially because the impact of covariates on dosing is poorly defined.7,–,9 A maturational trend with an increase in urinary glucuronide metabolite elimination has been described, in part explained by postnatal age (PNA), postmenstrual age (PMA) or repeated administration.10 Unconjugated hyperbilirubinaemia ≥150 µmol/l was also associated with a 40% reduction in paracetamol clearance.11 Other studies provided in vivo observations on the impact of growth restriction on hepatic metabolism.12
What is already known on the topic
▶ Pharmacokinetic data and the impact of covariate information on intravenous paracetamol in neonates are limited.
▶ The dosing regimens used in clinical practice vary from the manufacturers' recommendations.
What this study adds
▶ An increased volume of distribution supports the use of a loading dose in neonates.
▶ Size (described by patient weight) is the major covariate contributing to paracetamol clearance variance in neonates. Paracetamol clearance (mg/kg/h) increases marginally throughout neonatal life.
▶ Consequently, we suggest a loading dose of 20 mg/kg followed by 10 mg/kg every 6 h within the age range evaluated (32–44 weeks' postmenstrual age)
We pooled all available time–concentration points in neonates to explore the covariate effects of intravenous paracetamol clearance variability. Age (PMA, PNA), weight, growth restriction and unconjugated bilirubin concentration were available from three published investigations and were pooled with new observations from 60 neonates to describe covariate effects. Through improved understanding of covariate information, we aimed to formulate dosing suggestions for intravenous administration in (pre)term neonates.
Materials and methods
Four studies were performed at two institutions in neonates (27–45 weeks' PMA). Table 1 describes the number of patients, the demographics and the number of plasma samples taken for each study. Term neonates were defined as more than 36 weeks' PMA, preterm neonates as up to 36 weeks' PMA and extreme preterm neonates as less than 32 weeks' PMA. The medication was infused over 15 min in all studies. Further details can be found within individual references, including analytical methods used, procedures for ethical approval and informed consent.
Study 1: single dose intravenous propacetamol on day 113
Neonates admitted within the first 24 h of life and with an arterial cannula were considered for the study if propacetamol (Prodafalgan; Bristol-Myers Squibb Pharmaceuticals, Braine l'Alleud, Belgium) was administered. Propacetamol was administered for minor, painful procedures (ie, insertion of peripheral arterial, venous or central venous line, chest tube placement) or as adjuvant therapy when receiving opioids. Exclusion criteria were major congenital malformations or severe birth asphyxia (Apgar <4 at 5 min). Thirty neonates and 213 concentration–time points were available. Propacetamol dosing was 20 mg/kg (10 mg/kg paracetamol) for the first 15 neonates and 40 mg/kg (20 mg/kg paracetamol) for the next 15 neonates. Propacetamol was prepared from a 1 g propacetamol (500 mg paracetamol) powder-containing phial diluted in 50 ml normal saline.
Cut-off values for growth restriction were based on birthweight reference charts of a Flemish cohort of neonates to classify neonates as small for gestational age (SGA) (ie, <10th percentile) or appropriate for gestational age.14 Paracetamol plasma concentrations were determined using fluorescence polarisation immunoassay (Adx system; Abbott Laboratories, North Chicago, Illinois, USA). The lower limit of quantification (LLOQ) was 1 mg/l and the precision was 7%.
Study 2: repeated dose intravenous propacetamol in neonates15
The inclusion criteria and propacetamol preparation were similar to those used in study 1.13 The decision to prescribe propacetamol was made by the neonatologist based on a standardised evaluation of pain and an analgesic algorithm after surgery or during specific medical conditions. Eighteen neonates received multiple doses of propacetamol (up to 48 h) and 107 time–concentration points were collected. The dosing regimen was based on a loading dose (20 mg/kg paracetamol), with PMA-based maintenance dosing of 10 mg/kg/dose every 12, 8 or 6 h for extreme preterm, preterm and term neonates respectively.
Blood samples for paracetamol assay were taken from an arterial line during the first 48 h after the first propacetamol dose. Plasma samples were analysed using high-performance liquid chromatography (HPLC). The linearity of the calibration curve in plasma was found to be in the range of 0.078–40 μg/ml (y=0.003+0.707x, r=0.99). The LLOQ was 0.08 mg/l, which was the lowest concentration of the standard curve with a coefficient of variation below 20%. Paracetamol analytical recovery was 62.6% (SD 6.6%) and for the internal standard p-propionamidophenol was 77.0% (SD 8.3%). Variation coefficients in intraday and interday accuracy and precision were below 15%.
Study 3: repeated dose intravenous paracetamol in neonates11
Fifty neonates were studied who received multiple doses of intravenous paracetamol (Perfalgan; Bristol-Myers Squibb Pharmaceuticals: 500 mg in 50 ml vials). The mean number of doses administered was 15 (range 5–26), mostly for postoperative analgesia. Dosing was PMA based: 10, 12.5 and 15 mg/kg every 6 h (without a loading dose) for extreme preterm, preterm and term neonates respectively. Blood samples were taken through the venous, capillary or arterial route after at least two doses were administered.
Paracetamol assays (n=189) were performed on fresh plasma using quantitative colorimetry-Vitros paracetamol slides and the Vitros-950 analyser (Ortho-Clinical Diagnostics, Johnson and Johnson, Rochester, USA). This system has high correlation with the fluorescent polarisation immunoassay (FPIA 0.991) and HPLC assay techniques. The determination limit is 6 mg/l and the precision is 0.8–2.1% over the concentration range of 19–166 mg/l. The paracetamol concentration is positively biased at high bilirubin concentrations; for example, for a bilirubin concentration of 256 mmol/l, a paracetamol concentration of 30 mg/l has a positive bias of 4.2 mg/l (14%). The raw data, uncorrected for bilirubin, were entered in the pooled analysis.
Study 4: repeated dose intravenous paracetamol in neonates
Time–concentration points were collected during repeated administration of intravenous paracetamol (Paracetamol Sintetica; Treviso, Italy) (n=60, 434 observations) (EUdraCT study register number: 2009-011243-39). Similar to earlier intravenous paracetamol studies performed at the University Hospitals Leuven, neonates were included after obtaining informed parental consent and approval by the local ethics committee.13 15 The paracetamol concentration of Paracetamol Synthetica is the same as that of Perfalgan (10 mg/ml). Indications for intravenous paracetamol administration were medical (traumatic delivery, necrotising enterocolitis, prostaglandin E2 administration, fever) or surgical (eg, cardiac, thoracic, abdominal). The dosing regimen consisted of a loading dose (20 mg/kg), followed by a maintenance dose of 5, 7.5 or 10 mg/kg every 6 h for extreme preterm, preterm and term neonates respectively. Samples were collected by arterial line in the first 48 h after the paracetamol loading dose. The HPLC paracetamol assay was the same as that described in study 2.15
Population parameter estimations
A two-compartment linear disposition model with first-order elimination was used to analyse time–concentration points. The model was parameterised in terms of clearance (CL), intercompartment clearance (Q), central volume (V1) and peripheral volume (V2). An additional depot compartment was used for propacetamol administration. Propacetamol was infused into the depot compartment and a rate constant (Ka, mimicking hydrolysis) was applied to describe movement into the central compartment. This parameter was described using an absorption half life (Tabs):
Population parameter estimates were obtained using non-linear mixed effects modelling (NONMEM VI; Globomax LLC, Hanover, Maryland, USA).16 This model accounts for population parameter variability (between and within subjects) and residual variability (random effects) as well as parameter differences predicted by covariates (fixed effects). The population parameter variability in model parameters was modelled by a proportional variance model. Residual unidentified variability was described using a combined proportional and additive residual error model for each observation prediction with random differences (ErrPROP, ErrADD). The variance of residual unidentified variability (ηRUV,i) was estimated.17 Population mean parameters, BSV and residual variance were estimated using the first-order conditional interaction estimate method (ADVAN4 TRANS4 of NONMEM VI). The convergence criterion was three significant figures. A Compaq Digital Fortran Version 6.6A compiler with Intel Celeron 333 MHz CPU (Intel, Santa Clara, California, USA) under MS Windows XP (Microsoft, Seattle, Washington, USA) was used to compile NONMEM.
Propacetamol is hydrolysed by plasma esterases so that 1 g of propacetamol is hydrolysed to 0.5 g paracetamol in adults.18,–,20 Propacetamol bioavailability was compared with paracetamol. A logit function was used to constrict the estimate of propacetamol bioavailability (F) between 0 and 1:
where POP_F is the population estimate of the bioavailability with its variability (var).
The population parameter variability is modelled in terms of random effect (η) variables. Each of these variables is assumed to have a mean of 0 and an estimated variance denoted by ω2. The covariance between two elements of η (eg, CL and V) is a measure of statistical association between these two variables. Their covariance is related to their correlation (R), that is:
The covariance of clearance and distribution volume variability was incorporated into the model.
The parameter values were estimated for a standard body weight of 70 kg using an allometric model2:
where Pi is the parameter in the ith individual, Wi is the weight in the ith individual and Pstd is the parameter in an individual with a weight Wstd of 70 kg. Standardisation allows neonatal parameter estimates to be compared with those reported for adults. The PWR exponent was 0.75 for clearances and 1 for distribution volumes.2
Exploration of a relationship between clearance and unconjugated bilirubin was explored with a scaling factor (Fbili) based on a dichotomous value depending on whether the unconjugated bilirubin was above (UBLI=1) or below (UBILI=0) a discrete concentration dictated by PNA (table 2). This approach was deliberately chosen to reflect normal postnatal trends in bilirubinaemia.11
Covariate analysis included a model investigating age-related changes for clearance and volume of distribution using an exponential function21:
where Vstd and CLstd are the population estimates for V and CL respectively, standardised to a 70 kg person using allometric models; PMA is given in weeks; SLPage is a slope parameter describing age-related changes in CL, referenced to 40 weeks' PMA.
Quality of fit
The quality of fit of the pharmacokinetic model to the data was assessed using NONMEM's objective function and by visual examination of plots of observed versus predicted concentrations. Models were nested and an improvement in the objective function was referred to χ2 distribution to assess significance; for example, an objective function change (OBJ) of 3.84 is significant at α=0.05.
Bootstrap methods, incorporated within the Wings for NONMEM program, provided a means to evaluate parameter uncertainty.22 A thousand replications were used to estimate parameter CIs. A visual predictive check (VPC),23 a modelling tool that estimates concentration prediction intervals and graphically superimposes these intervals on observed concentrations after a standardised dose, was used to evaluate how well the model predicted distribution of observed concentrations. Simulation was performed using 1000 subjects with characteristics taken from studied patients. This is an advanced internal method of evaluation and is considered more accurate than commonly used plots of observed versus predicted values.24 25 Because covariates such as dose, weight, height and sex are different for each patient, we used a prediction corrected VPC (PC-VPC).26 Observations and simulations are multiplied by the population baseline value divided by the individual estimated baseline.
The pooled analysis included 158 subjects and 943 paracetamol plasma concentrations. There were 58 preterm neonates, of whom 21 were extreme preterm, 19 had a birth weight lower than 1500 g and 31 neonates were SGA (table 1). A two-compartment model was superior to a one-compartment model (ΔOBJ=72.023). Parameter estimates for the two-compartment analysis are shown in table 3. Propacetamol bioavailability was 0.519 (CV 38%). The correlation of BSV for CL and V1 was 0.357. Weight, PMA and unconjugated bilirubin concentration reduced the objective function significantly (p<0.01). The addition of PNA to PMA had no effect on the objective function. SGA neonates did not affect clearance. There were no changes in distribution volumes with age in the range investigated (PMA 28–44 weeks). Figure 1 shows satisfactory PC-VPC plots for pharmacokinetic data.
The difference between BSV without covariates and with covariates is a measure of the predictable decrease in BSV due to covariates. The ω2 estimates for the different components contributing to clearance variability are shown in table 4. The ratio of the BSV predictable from covariates to the total population parameter variance obtained without covariate analysis indicates the relative importance of covariate information. The ratio of 0.609 achieved for clearance indicates that 60.9% of overall clearance variance is predictable from covariates.
Weight was used to predict size and was the major covariate contributing 57.5% of variance. PMA only contributed an additional 2.2% of variance and high unconjugated bilirubin contributed 1.2 %. Clearance (mg/kg/h) increases marginally with increasing PMA (from 0.138 l/kg/h at 28 weeks' PMA to 0.167 l/kg/h at 44 weeks' PMA) with minimal impact of bilirubin (figure 2). Using the clearance estimates obtained, time–concentration profiles were plotted with and without a loading dose (20 mg/kg) to achieve a mean plasma concentration of 11 mg/l in a typical neonate (figure 3).
Paracetamol clearance described using allometric scaling was one third (5 l/h/70 kg) of mature values reported in adults and children (16.2 l/h/70 kg).27 Paracetamol clearance maturation remains slow before 40 weeks' PMA and matures rapidly afterwards, with a maturation half time at 52 weeks' PMA to reach 90% of adult rates at 1 year, equal to 92 weeks' PMA.27 Consequently, size described using weight is the most important covariate for determining clearance in neonates during this initial slow maturation phase, irrespective of PMA. Age has a greater contribution after neonatal life so that both age and size contribute 91% of clearance variance throughout childhood.28
At 36 weeks' PMA mean clearance is 0.151 l/kg/h and only differs by 10% from this value at 28 weeks' PMA (10% lower) or 44 weeks' PMA (10% higher). Since typical bioequivalence criteria accept a bias up to 25% in the delivered dose from an equivalent formulation, this means that maintenance dosing in neonates can be based on weight only.29 However, after taking size (57.7%), age (2.2%) and bilirubin (1.2%) into account, unexplained variance in intravenous paracetamol clearance remains high (39.1%).
Additional contributors such as disease severity were not investigated in this pooled study. For example, clearance of morphine, a drug also eliminated by an individual isoform of glucuronosyl transferase (UGT2B7), is reduced in neonates presenting for extracorporeal membrane oxygenation or after cardiac surgery.6 30 The impact of such comorbidity on paracetamol clearance (UGT1A6) remains unmeasured. The extensive unexplained variability, lack of observations in extreme preterm neonates and the limited safety data in neonates restricted us from formulating dose recommendations in extreme preterm neonates and warrants continued surveillance in all exposed neonates. Finally, V1 estimates are increased in neonates (51.9 l/70 kg, CV 26.6%) compared with mature estimates (35.4 l/70 kg, CV 61.2%) while V2 estimates are similar (22.7 versus 27.8 l/70 kg), confirming earlier findings.11 15 31
In the absence of a robust target effect concentration in neonates, a target effect concentration reported in children after tonsillectomy was used (10 mg/l).32 This effect compartment concentration is assumed to be similar to the plasma concentration at steady state. To reach such a target concentration, a loading dose of intravenous paracetamol in neonates (PARANEO) should be considered. The increased volume of distribution in neonates supports the use of a larger initial dose when starting paracetamol therapy in neonates if one aims to attain a given threshold paracetamol concentration sooner. Using the clearance estimates obtained, the impact of a loading dose (20 mg/kg) on time–concentration profiles to achieve a mean plasma concentration of 11 mg/l is illustrated in figure 3.
To maintain a mean plasma concentration of 11 mg/l, we suggest that maintenance dosing in neonates of 32–44 weeks' PMA can be simplified.7 11 15 Six-hourly dosing of 10 mg/kg is predicted to sustain a mean plasma concentration of 11 mg/l. This dosing regimen is easily remembered and the dosing interval retains parsimony with the 6 h dosing of intravenous paracetamol suggested in children and adults.33 However, caution is required when extending these dosing suggestions to extreme preterm neonates and the current dosing regimen needs further pharmacodynamics and toxicity evaluation.
The pooled dataset allowed greater pharmacokinetic exploration than earlier analyses.11 13 15 A two-compartment model, rather than a one-compartment model, was realised and bioavailability of propacetamol was compared directly with intravenous PARANEO. Our data confirm a bioavailability of 0.5 for propacetamol to paracetamol with a narrower 90% CI of 0.465 to 0.557.18,–,20 There was considerable variability associated with this estimate (CV 38.6%) due to the use of adult drug preparations for administration in neonates. Both intravenous paracetamol formulations evaluated have a concentration of 10 mg/ml. Infusion volumes were between 0.5 and 2 ml/kg for intravenous paracetamol but propacetamol needed additional manipulation before administration. Dilution adds error to the volume of drug given as does the accuracy of dose volume measuring devices, which is reflected in variability associated with bioavailability. Formulations or vials at doses appropriate for use in children have the potential to reduce preventable systematic errors.34 The introduction of an amikacin paediatric phial (50 mg/ml) instead of the adult phial (250 mg/ml) reduced clearance variability by 53%.35
The parameter describing the hydrolysis half life of conversion from propacetamol into paracetamol by esterases was 30 s with a wide 90% CI (0.04 to 5.7 min), reflecting delayed timing of the first assay at 30 min. However, the short conversion half life and the confirmation of a bioavailability of 0.5 suggest that esterase activity is mature in neonates. The rapid clearance in young infants of remifentanil, an opioid also cleared by plasma esterases, is consistent with this.36 The lack of sampling also means that we were unable to estimate variability associated with intercompartment clearance or peripheral volume of distribution.
Finally, neither the dosing regimens used in individual studies1 11 13 15 nor the new dosing suggestion are consistent with manufacturers' recommendations.9 The dosing regimens were off label, although one intravenous paracetamol (Perfalgan) is registered for use in term neonates (7.5 mg/kg every 6 h, no loading dose). Similar discrepancies have been reported for oral paracetamol when the summary of product characteristics was compared with the British National Formulary, illustrating the complex interaction between manufacturers, authorities and clinicians.37
KA is supported by the Fund for Scientific Research, Flanders (Belgium) (F W O Vlaanderen) by a Fundamental Clinical Investigatorship (1800209N).
Competing interests None.
Ethics approval This study was conducted with the approval of the reanalysis of three already published studies. New data set (PARANEO study): approval by the Local Ethics Committee University Hospitals, Leuven, Belgium.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.