Pharmacometric in silico studies used to facilitate a national dose standardisation process in neonatology – application to amikacin

DOI: https://doi.org/https://doi.org/10.57187/s.3632

Introduction

Neonatal dosing with anti-infective drugs is highly variable at both national [1, 2] and international levels [3] and across neonatal treatment guidelines [4]. In Switzerland, the Swiss database for dosing medicinal products in paediatrics (SwissPedDose, https://www.swisspeddose.ch/database) aims to standardise drug dosing in paediatrics at a national level. Recommendations consider the currently applied dosing regimens, the latest scientific evidence, clinical experience, and expert opinion [5, 6]. Pharmacometric modelling and simulation is a recognised approach to guiding dosage decisions in the development of new medicines and post-marketing drug optimisation [7, 8]. Modelling may also provide a scientific rationale for neonatal and paediatric dosing approaches in areas of uncertainty [9–12]. We present a motivational case study to illustrate and discuss the potential broader usefulness, prerequisites, and implementation in the context of a national dose standardisation effort.

Motivational case study

In Switzerland, amikacin is a frequently used aminoglycoside antibiotic with activity against gentamicin-resistant bacteria [13, 14]. Amikacin is mostly used in combination with a β-lactam antibiotic as a first-line empirical treatment for suspected neonatal sepsis [15]. During the national standardisation process, a dosing approach according to post-menstrual age (PMA, <30/30–35/35–44 weeks, defined as gestational age plus chronological post-natal age) and post-natal age (PNA, 0–14/≥14 days) was proposed by SwissPedDose after a literature review, defining six subpopulations (table 1) [6]. In the preterm subpopulation with a post-menstrual age of 30–35 weeks, a 24-hour vs 36-hour dosing interval was discussed, with the decision of a 24 hour dosing interval disregarding post-natal age for practical reasons. Clinical concerns regarding this approach were raised by neonatologists for the preterm subpopulation #3 (table 1) defined by a post-menstrual age of 30–35 weeks and a post-natal age <14 days (hereafter referred to as the subpopulation of clinical concern), as elevated trough concentrations (C_trough) were observed in a significant proportion of patients. For this reason, clinicians contacted SwissPedDose and requested an evaluation of extended dosing intervals such as 36 hours and 48 hours to mitigate the risk of elevated C_trough values associated with potential nephro-/ototoxicity, with a defined safety threshold as C_trough <5 mg/l [6, 16].

Challenges addressed by pharmacometric in silico studies

To the best of our knowledge, a dosing approach based on post-menstrual age and post-natal age has not been formally evaluated for amikacin regarding the suitability to achieve commonly used exposure targets [13, 17, 18], including safety targets associated with low risks of oto-/nephrotoxicity [17]. Although post-menstrual age and post-natal age have been shown to influence amikacin clearance in various population pharmacokinetic analyses, none of the reported models provides a description of clearance based on the combination of post-menstrual age and post-natal age [18, 19]. This is in line with characterisations of kidney function maturation that determines amikacin clearance: although post-menstrual age and post-natal age are important determinates of post-natal kidney function compared to other demographic factors, they have not been used in combination to describe the maturation of renal amikacin clearance in preterm neonates [20–23]. Pharmacometric in silico studies can systematically predict pharmacological expectations of amikacin clearance as a function of birth weight and post-natal age to clinical exposure expectations. They can be performed (figure 1) leveraging a suitable model combined with a large representative neonatal demographic dataset (regarding the distribution of and correlation between weight, birth weight, post-natal age, and post-menstrual age). Various dosing approaches, including extended dose intervals, can be simulated, and the percentage of neonatal patients achieving the clinical target of interest derived for each subpopulation of interest. Model-informed drug development (MIDD) has become a standard approach in the pharmaceutical and biotech industries and is expected by health authorities such as the Food and Drug Administration (FDA) and European Medicines Agency (EMA), particularly in the context of development, approval, and utilisation of new medicines in paediatrics [24, 25]. In addition, applying a “model-informed dosing” approach to support efforts to optimise and standardise drug dosing in clinical paediatric practice has gained interest in other countries [26]. The pharmacometric approach used to support the Swiss neonatal dose standardisation of amikacin, our motivational case example, will be described in more detail in the following sections, which may serve as a proof-of-concept for further cases.

Figure 1Implementation of pharmacometric in silico studies in neonatal dose standardisation. The general prerequisites for pharmacometric guided dose evaluation are (1) the availability of at least one suitable pharmacometric model for the population of interest (e.g., a population pharmacokinetic or pharmacokinetic/pharmacodynamic model), (2) the possibility of defining a representative patient population with relevant covariate information (i.e., for the pharmacometric model employed, and considered for different dosing approaches), and (3) the definition of a quantitative target or reference outcome (e.g., therapeutic area-under the curve [AUC], trough concentration [C_trough], peak concentration [C_max], or general adult reference exposure). The illustrated example represents dose-exposure simulations (as shown for the motivational example of amikacin), but these may be extended to dose-exposure-response simulations in the case of available pharmacodynamic models. Proposed in silico studies are not meant to be unidirectional, but are ideally part of a “learn, confirm, and update” cycle and should be weighted by expert opinion. PNA: post-natal age; GA: gestational age; CL: clearance.

Methods

We applied pharmacometric approaches to assess overall expected exposure target achievement (percentage of neonates with C_trough <5 mg/l) under the initially proposed recommended dosage and to evaluate extended dosing intervals (36 hours and 48 hours instead of 24 hours) to quantify the improvement in target achievement in the neonatal subpopulation of concern.

As this was a virtual experiment, there was no clinical trial registration, and the study was deemed exempt from ethics approval.

Neonatal real-world dataset

Demographic covariate data were taken from the Antibiotic Resistance and Prescribing in European Children (ARPEC) point prevalence study [27, 28], an anonymous population of paediatric and neonatal in-patients treated with antibiotics at 8 A.M. in a one-day cross-sectional international web-based survey. For this study, only neonates with post-natal age <28 days and weight <5 kg (to exclude potential erroneous outliers) were included. Post-menstrual age (weeks) was calculated as gestational age (weeks) + post-natal age (days) / 7. The total selected neonatal population was stratified into six subpopulations according to post-menstrual age (<30 weeks, 30–35 weeks, 35–44 weeks) and post-natal age (0–14 days, and ≥14 days), in line with current dosage recommendation of SwissPedDose (table 1). A total of 500 individuals were randomly sampled with replacement from each neonatal subpopulation.

The primary pharmacometric analysis included only patients with complete information for relevant covariates (post-natal age, gestational age, weight, and birth weight). Sensitivity analyses were performed including all patients after imputation of missing covariates (birth weight, gestational age) as described below.

Pharmacometric model

An extensive externally evaluated and updated two-compartmental pharmacokinetic model [16] describing amikacin pharmacokinetics in neonates was used in which typical pharmacokinetic parameters were set as a function of demographic covariates: typical clearance (CL_typ) as a function of birth weight and post-natal age, and the typical central and peripheral volumes of the distribution (V1_typ, V2_typ) as a function of weight. The model also accounted for clearance decreasing during whole-body cooling (therapeutic hypothermia [TH], binary with 1 = yes, 0 = no) and ibuprofen (IBU, binary with 1 = yes, 0 = no) treatment [16]:

CL_typ [l/h] = CL_pop [l/h] × (birthweight [kg]/1.75 kg)^1.34 × (1 + 0.22 post-natal age [days]/2 days) × 0.838^IBU × 0.594^TH, with CL_pop = 0.0495 l/h

V1_typ [l] = V1_pop [l] × (weight [kg]/1.75 kg)^0.93 = V2_typ, with V1_pop = 0.832 l

Inter-compartmental clearance (Q_typ) was defined as proportional to CL_typ (Q_typ = 0.45 × CL_typ). The model further described the effects of ibuprofen and therapeutic hypothermia on amikacin clearance (associated with 16% and 41% decreases in clearance, respectively). These treatment-associated variables were not used in the primary analysis, as it was assumed that the population of interest generally does not receive ibuprofen or therapeutic hypothermia (i.e., IBU = 0 and TH = 0). The remaining random inter-individual variability (between-subject variability) was 0.32 (standard deviation on a log-scale) in clearance, and not quantified for other pharmacokinetic parameters.

Demographic characteristics of the neonatal populations on which the model was developed and externally evaluated are summarised in table S1 in the appendix. Briefly, an initial model [29] was developed using a total of 874 neonates [30, 31] and externally evaluated on a total of 239 neonates [32, 33]. The respective model parameter estimates were similar when re-estimated from another population of 573 neonates [34]. The structure of the initial model [29] was further refined by the inclusion of additional therapeutic drug monitoring data of 56 neonates (the model applied herein), which allowed consideration of the effect of therapeutic hypothermia on amikacin clearance [16], again yielding almost identical model parameter estimates as previously reported [29].

Pharmacometric in silico studies: dose-exposure simulations

Individual concentration-time profiles were generated by Monte Carlo simulations, given the covariates and random inter-individual variability in amikacin clearance. The individual predicted C_trough before the third dose was then extracted (the usual sampling time in clinical practice), and the percentage (%) of neonates with C_trough <5 mg/l was calculated for each of the six subpopulations (the goal being at least 80% with a desirable C_trough in all neonatal subpopulations).

Individually predicted C_trough values were plotted against patient demographics to evaluate their potential correlation with elevated C_trough >5 mg/l.

Initially proposed standardised dosing approach

A total of 500 neonates were simulated for each of the six neonatal populations to calculate the percentage of neonates with C_trough before the third dose with the initially proposed approach.

Evaluation of alternative dosing intervals

Two additional simulations were performed for the subpopulation of clinical concern to calculate the percentage of neonates with C_trough before the third dose with alternative dosing intervals of 36 h or 48 h.

Sensitivity analyses

The following sensitivity analyses were performed: (a) pharmacometric simulations using a dataset with missing covariates of birth weight and gestational age imputed and obtained by three methods: (a1) linear regression, and multiple imputation by (a2) predictive mean matching or (a3) random forest (supplemental methods in the appendix); (b) pharmacometric simulations, including residual intra-individual variability; (c) simulations adding hypothetical inter-individual variability in the distribution (standard deviation = 0.1 for log-transformed parameters) and a small correlation of 0.3 between individual random effects of the distribution and clearance; (d) pharmacometric simulations for the same population treated with therapeutic hypothermia (TH = 1) to decrease the neurological sequelae of perinatal asphyxia; (e) simulations for the same population receiving ibuprofen (IBU = 1) treatment for a patent ductus arteriosus.

Software packages

In silico studies were performed using the software Simulx (Version 2021R2, Lixoft SAS, a Simulations Plus company). Data preparation and further statistical computing or figure creation were performed in R (version 4.2.1, R Foundation for Statistical Computing, https://www.R-project.org/).

Results

Neonatal real-world dataset

From the available demographic ARPEC database (comprising 3844 patients), 2590 neonates were eligible (post-natal age <28 days, post-menstrual age <44 weeks, weight <5 kg); among these, 1563 had complete demographic data with respect to gestational age/post-menstrual age and birth weight and were included in the primary analysis (table 1 and figure 2; overall median [IQR] post-natal age: 6 [3–13 days], weight: 2.3 [1.39, 3.15] kg). In most other patients with missing covariates (n = 1027, median [IQR] post-natal age: 8 [4–16], weight: 2.80 [1.80–3.40] kg), both gestational age and birth weight were missing (n = 918).

Figure 2Illustration of real-life patient population demographic data used as a basis for simulations: birth weight versus postnatal age (PNA) as clearance-relevant covariates depicted by postmenstrual age (PMA) according to the SwissPedDose recommendations. Post-natal age values are slightly jittered for easier visual assessment. Black dots: complete data. Grey dots: imputed data with respect to post-menstrual age and birth weight (method a1).

Pharmacometric model

Simulated individual pharmacokinetic parameters for each subgroup are depicted in figure S3 in the appendix. The corresponding simulated mean half-lives were 6.6 h and 5.3 h in subpopulations 1 and 3, respectively, and 2.2–3.7 h in the other subpopulations (initial exponential decline), suggesting that C_trough sampling before the third dose likely corresponds to a steady-state measurement in the plasma.

Figure 3Illustration of a simulated amikacin concentration-time profile distribution for each of the six subpopulations of the primary analysis (where complete covariate data were available), following two administrations of the dose according to the initial national SwissPedDose recommendations. Shaded areas delimit the 50% prediction interval (percentile 25–75, dark grey) and 90% prediction interval (percentile 5–95, light grey). Orange line: C_trough target of <5 mg/l. PMA: post-menstrual age; PNA: post-natal age.

Pharmacometric dose-exposure simulations

The distribution of predicted amikacin exposure under the initially agreed harmonised dosing approach is shown in figure 3. The corresponding predicted distribution of C_trough before the third dose is shown in figure 4 and table 2.

Figure 4Summary of the C_trough distribution before the third dose, and the predicted proportion of patients with C_trough <5 mg/l (i.e., the violet shaded area, with the proportions summarised by numbers).

Initially agreed harmonised dosing approach

The simulated proportion of neonates with C_trough <5 mg/l before the third dose was 59% (95%CI: 55–64%) in the subpopulation of clinical concern, and 79% (75–82%) to 99% (98–100%) in the other subpopulations. The correlations between elevated C_trough and patient demographics in the subpopulation of clinical concern are depicted in figure 5.

Figure 5Inspection of the predicted Ctrough distribution in the subpopulation of clinical concern following administration of one dose every 24 h (preterm neonates post-menstrual age (PMA) 30–35 weeks, post-natal age (PNA) <14d) according to patient demographics. Violet shaded area: C_trough <5 mg/l. Yellow shaded area: C_trough ≥5 mg/l.

Evaluation of alternative dosing intervals

When extending the dosing interval in the subpopulation of clinical concern from 24 h to 36 h or 48 h, the simulated percentages of neonates with a desirable C_trough increased to 89% (86–91%) and 97% (95–98%), respectively.

Sensitivity analyses

The proportion of patients with predicted individual C_trough values <5 mg/l before the third dose are summarised in table S2 (a–c, see appendix) and table 2 (d–e). Briefly, predicted proportions from sensitivity analyses a, b, and c were similar to those of the primary analysis. In sensitivity analysis d (therapeutic hypothermia), all proportions of target achievement were lower, with only the subpopulation of late preterm or term neonates outside the newborn period (post-menstrual age 35–44 weeks, post-natal age ≥14 days) achieving C_trough <5 mg/l in at least 80% of neonates (89.9% versus 15.6–59.8% in the other subpopulations under the initially proposed dosing regimen). In sensitivity analysis e (ibuprofen treatment), all proportions of target achievement were lower, but most subpopulations still achieved C_trough <5 mg/l in at least 80% of neonates.

Discussion

We applied pharmacometric in silico approaches in the context of a national dose standardisation process in neonatology to illustrate their utility and prerequisites. For our motivational example of amikacin, all three prerequisites for performing a pharmacometric simulation study were met (figure 1). First, a suitable, externally evaluated population pharmacokinetic model was available for the drug and population of interest [16]. Second, representative demographic patient population data could be taken from a large real-world data set for neonates treated with antibiotics [27]. Third, the clinical target exposure of interest was quantitatively defined (C_trough <5 mg/l), as was the outcome of interest (percentage of neonatal patients achieving the target: ≥80%).

In our case, this approach confirmed the clinically observed risk of amikacin accumulation with elevated C_trough (>5 mg/l prior to the third dose) in a neonatal subpopulation of clinical concern (preterm neonates with post-menstrual age 30–35 weeks, post-natal age 0–14 days) following the introduction of a new nationally standardised dose recommendation (i.e., a 24 h-interval with a dosage of 15 mg/kg). Simulations provided a scientific, quantitative rationale indicating that extension of the dosing interval to 36 h would be sufficient for similar target achievement as in the other neonatal subpopulations for whom no safety concerns were raised. An extended dosing interval had already been discussed for a subgroup of preterm neonates during the SwissPedDose standardisation process but was finally set to 24 hours for practical reasons. To simplify dosing across different subpopulations and to guarantee sufficient peak plasma concentration (C_max), which is considered relevant for aminoglycoside efficacy [16], a reduction of the dose was not discussed. Figure 3 demonstrates that simulated C_max remained nearly constant despite the prolongation of the dosing interval. The results presented herein show that performing such pharmacometric simulations can facilitate scientific discussions by formulating and providing model-based pharmacological predictions. This can help to better define optimal dosing recommendations among clinical experts, in particular where study data are sparse or missing or when discussions are controversial or have not reached a consensus. Our simulations suggest that elevated C_trough values were indeed highly associated with post-natal age <7 days (<50% achieving C_trough <5 mg/l versus >80% for 7–14 days, figure 5) facilitating dose standardisation towards this cutoff, in line with the approach chosen for several other renally eliminated anti-infectives such as gentamicin and beta-lactam antibiotics [6].

As a result of developmental and maturational processes, neonatal dosing is a complex process. On one hand, optimised dosing approaches in neonatology should consider not only safety and efficacy aspects but also quickly developing physiology and pharmacokinetics. On the other hand, it is crucial that dosing be practical and simple to mitigate the risk of medication errors in this vulnerable patient population [35–37]. In the development of new medicines, similar pharmacometric simulation approaches are regularly applied to translate initial complex approaches to dosing (e.g., based on body surface area) into simpler and more practical (e.g., stratified fixed dose) recommendations [7, 38, 39]. As our understanding of neonatal pharmacology is constantly increasing for many anti-infective drugs [40–43], it is time to bring such quantitative pharmacological knowledge into clinical practice, particularly in the context of national dose standardisation initiatives such as SwissPedDose.

Neonatal real-world dataset

Especially in the neonatal population, generating a virtual patient population with a representative multidimensional covariate distribution may be a challenge. Whereas the general neonatal population would show a uniform post-natal age distribution that can easily be simulated, our real-live demographic data show that post-natal age distribution in neonates treated with antibiotics is skewed to the left (i.e., an over-representation of neonates with post-natal age <7 days). Given the rapid improvement in kidney function and renal drug clearance with post-natal age, the predicted percentage of patients with amikacin C_trough <5 mg/l (the defined safety target) would increase if more neonates with older post-natal age were included. Utilizing a large real-world demographic dataset of a representative population has the advantage of fewer assumptions concerning a complex covariate distribution and correlations but may require the handling of missing demographic data [44]. In our case, neonates with missing information on birth weight and gestational age most likely represented neonates that were treated on a general paediatric ward (older gestational age and post-natal age, higher body weight as compared to “complete-information” neonates). Optimal dosing approaches may hence differ between neonates treated on general paediatric versus neonatal wards as well as in patients receiving additional potentially nephrotoxic drugs or therapeutic hypothermia (sensitivity analyses d–e).

In our simulations, we assumed that the covariates did not change over time. Our simulations may hence be considered conservative, as renal clearance improves with each day of life and increasing weight in neonates. A combination with neonatal weight-prediction models [45] could be of interest in other studies regarding questions as to when to adapt the dosing strategy in an individual patient.

Pharmacometric model

A suitable pharmacometric model may not be available for all dosing questions, or several candidate models may be available and compared regarding their predictions [46]. In cases where no suitable model is available for the drug and population of interest, physiology-based pharmacokinetic models may be considered to formulate semi-quantitative exposure comparisons for different neonatal or paediatric subgroups, which then may be updated quantitatively upon the availability of actual pharmacokinetic data by population pharmacokinetic modelling [8]. Ideally, not only dose-exposure but also exposure-outcome simulations can be performed, such as for clinical cure, bacterial killing, and resistance development [47]. Pharmacodynamic studies in neonates, however, remain scarce due to practical and ethical difficulties [8].

In our case, we did not compare simulations from other models [18, 19], but the model we used represents by far the most extensively evaluated neonatal model, having additionally demonstrated favourable predictive performance regarding post-natal renal function maturation [20]. A large number of trough and peak amikacin concentration measurements were used to develop the model, suggesting its suitability for C_trough prediction.

Pharmacometric in silico studies

In general, many different exposure questions can be addressed, but clinical limitations with respect to formulated targets need to be kept in mind for final dosing decisions, as illustrated in figure 1 by expert opinion relevance (e.g., in our case uncertainty concerning the actual risk of oto-/nephrotoxicity associated with short-term high C_trough, an optimal C_trough cutoff to define “elevated” C_trough values under varying dosing intervals, and duration of post-antibiotic effect [48–50]). Simulations may be easily adjusted and updated if new evidence emerges regarding exposure targets, such as increasing minimal inhibitory concentration (MIC), as has been observed for gentamicin [12, 51]. Improved infrastructures allowing the collection and use of real-world data for scientific purposes [52] will further facilitate the development and evaluation of such model-informed dosing approaches and their implementation in a learn-and-confirm cycle (figure 1).

In our case, simulations may also be used to evaluate amikacin C_max, a key parameter related to aminoglycoside efficacy [16]; C_max >20 mg/l was achieved in the majority of patients (minimum efficacy target used in other studies [40]). As no random inter-individual variability was incorporated in the model applied, a factor that might be expected, this outcome may require a more cautious interpretation. Interestingly, however, the proportion of patients achieving C_max >20 mg/l was least in the “oldest” subpopulation (i.e., post-menstrual age 35–44 weeks, post-natal age 14–28 days) treated with single doses of 15 mg/mg, supporting the use of single doses up to 20 mg/kg in this subpopulation (table 1). Antibiotic coverage for efficacy (normally with a target ratio of C_max / minimal inhibitory concentration >8–10) may be more important in the first days of treatment compared to the risk of short-term drug accumulation [53].

Conclusion

In conclusion, pharmacometric in silico studies based on high-quality real-world demographic datasets can provide a quantitative scientific rationale and rating for dose optimisation and facilitate national dose standardisation, particularly in neonatology. In the present motivational example, this approach allowed us to translate pharmacological expectations for amikacin based on birth weight and post-natal age to exposure predictions stratified by post-natal age and post-menstrual age for various dosing intervals. Further, computer simulations indicated that a post-natal age stratification ≤7 or >7 days may be considered in the future, in line with dosing recommendations for other renally eliminated anti-infective drugs employed in neonatology. Implementation of pharmacometrics into the decision-making process of neonatal dose standardisation should not be unidirectional, but rather part of a continuous learn-and-confirm process combining scientific evidence, clinical experience, and expert opinion (figure 1). In addition to facilitating the standardisation of existing neonatal dose recommendations, such an iterative approach will increase clinical acceptance of fine-tuned dose recommendations and support their implementation.

Data availability statement

All demographic and simulated data are presented in the manuscript/supplemental material. Further inquiries can be directed to the corresponding authors.

Acknowledgments

This project was realised thanks to close collaboration between the team establishing the Swiss database for dosing medicinal products in paediatrics (SwissPedDose; www.swisspeddose.ch) and the Swiss Research Network of Clinical Pediatric Hubs (SwissPedNet). SwissPedDose is supported by the Swiss Federal Office of Public Health (FOPH). We thank Herman Goossens and Ann Versporten, University of Antwerp and ARPEC project group, for providing the ARPEC data set. We thank clinical SwissPedDose and SwissPedNet experts, in particular Matteo Fontana (harmonisation expert neonatology) for initiating discussions on the presented case example.

Author contributions: Conception and Design: Clinical question of case example formulated by SwissPedDose experts: EG (SwissPedDose harmonisation specialist, neonatology), MB and CB (SwissPedDose harmonisation experts paediatric infectious diseases). Design of scientific approach: VG, PP, CC, MP (SwissPedDose/SwissPedNet collaboration expert team). Data acquisition: JB. Analysis: VG, DB. Interpretation of the data and approach: all. Drafting the article: VG with JB, PP, CC, MP. Revising the manuscript: all.

References

1 Leroux S, Zhao W, Bétrémieux P, Pladys P, Saliba E, Jacqz-Aigrain E; French Society of Neonatology. Therapeutic guidelines for prescribing antibiotics in neonates should be evidence-based: a French national survey. Arch Dis Child. 2015 Apr;100(4):394–8. 10.1136/ARCHDISCHILD-2014-306873 10.1136/archdischild-2014-306873 

2 Kadambari S, Heath PT, Sharland M, Lewis S, Nichols A, Turner MA. Variation in gentamicin and vancomycin dosage and monitoring in UK neonatal units. J Antimicrob Chemother. 2011 Nov;66(11):2647–50. 10.1093/jac/dkr351 

3 Metsvaht T, Nellis G, Varendi H, Nunn AJ, Graham S, Rieutord A, et al. High variability in the dosing of commonly used antibiotics revealed by a Europe-wide point prevalence study: implications for research and dissemination. BMC Pediatr. 2015 Apr;15(1):41. 10.1186/s12887-015-0359-y 

4 Liem TB, Slob EM, Termote JU, Wolfs TF, Egberts AC, Rademaker CM. Comparison of antibiotic dosing recommendations for neonatal sepsis from established reference sources. Int J Clin Pharm. 2018 Apr;40(2):436–43. 10.1007/s11096-018-0589-9 

5 Tilen R, Panis D, Aeschbacher S, Sabine T, Meyer Zu Schwabedissen HE, Berger C. Development of the Swiss Database for dosing medicinal products in pediatrics. Eur J Pediatr. 2022 Mar;181(3):1221–31. 10.1007/S00431-021-04304-8 10.1007/s00431-021-04304-8 

6 SwissPedDose Association SwissPedDose. National pediatric drug doses. https://swisspeddose.ch/ 

7 Mehrotra N, Bhattaram A, Earp JC, Florian J, Krudys K, Lee JE, et al. Role of Quantitative Clinical Pharmacology in Pediatric Approval and Labeling. Drug Metab Dispos. 2016 Jul;44(7):924–33. 10.1124/DMD.116.069559 10.1124/dmd.116.069559 

8 Vinks AA, Emoto C, Fukuda T. Modeling and simulation in pediatric drug therapy: application of pharmacometrics to define the right dose for children. Clin Pharmacol Ther. 2015 Sep;98(3):298–308. 10.1002/cpt.169 

9 Pfiffner M, Berger-Olah E, Vonbach P, Pfister M, Gotta V. Pharmacometric Analysis of Intranasal and Intravenous Nalbuphine to Optimize Pain Management in Infants. Front Pediatr. 2022 Mar;10:837492. 10.3389/FPED.2022.837492 10.3389/fped.2022.837492 

10 Pfiffner M, Gotta V, Pfister M, et al (2022) Pharmacokinetics and tolerability of intranasal or intravenous administration of nalbuphine in infants. Arch Dis Child archdischild-2022-323807. https://doi.org/10.1136/archdischild-2022-323807 

11 van Donge T, Samiee-Zafarghandy S, Pfister M, Koch G, Kalani M, Bordbar A, et al. Methadone dosing strategies in preterm neonates can be simplified. Br J Clin Pharmacol. 2019 Jun;85(6):1348–56. 10.1111/BCP.13906 10.1111/bcp.13906 

12 van Donge T, Pfister M, Bielicki J, Csajka C, Rodieux F, van den Anker J, et al. Quantitative analysis of gentamicin exposure in neonates and infants calls into question its current dosing recommendations. Antimicrob Agents Chemother. 2018 Mar;62(4):e02004-17. 10.1128/AAC.02004-17 

13 Darlow CA, da Costa RM, Ellis S, Franceschi F, Sharland M, Piddock L, et al. Potential Antibiotics for the Treatment of Neonatal Sepsis Caused by Multidrug-Resistant Bacteria. Paediatr Drugs. 2021 Sep;23(5):465–84. 10.1007/S40272-021-00465-Z 10.1007/s40272-021-00465-z 

14 Ramirez MS, Tolmasky ME. Amikacin: Uses, Resistance, and Prospects for Inhibition. Molecules. 2017 Dec;22(12):2267. 10.3390/MOLECULES22122267 10.3390/molecules22122267 

15 Shane AL, Sánchez PJ, Stoll BJ. Neonatal sepsis. Lancet. 2017 Oct;390(10104):1770–80. 10.1016/S0140-6736(17)31002-4 

16 Cristea S, Smits A, Kulo A, Knibbe CA, van Weissenbruch M, Krekels EH, et al. Amikacin Pharmacokinetics To Optimize Dosing in Neonates with Perinatal Asphyxia Treated with Hypothermia. Antimicrob Agents Chemother. 2017 Nov;61(12):e01282–17. 10.1128/AAC.01282-17 

17 Jenkins A, Thomson AH, Brown NM, Semple Y, Sluman C, MacGowan A, et al.; BSAC Working Party on Therapeutic Drug Monitoring. Amikacin use and therapeutic drug monitoring in adults: do dose regimens and drug exposures affect either outcome or adverse events? A systematic review. J Antimicrob Chemother. 2016 Oct;71(10):2754–9. 10.1093/JAC/DKW250 10.1093/jac/dkw250 

18 Wilbaux M, Fuchs A, Samardzic J, Rodieux F, Csajka C, Allegaert K, et al. Pharmacometric Approaches to Personalize Use of Primarily Renally Eliminated Antibiotics in Preterm and Term Neonates. J Clin Pharmacol. 2016 Aug;56(8):909–35. 10.1002/jcph.705 

19 Illamola SM, Sherwin CM, van Hasselt JG. Clinical Pharmacokinetics of Amikacin in Pediatric Patients: A Comprehensive Review of Population Pharmacokinetic Analyses. Clin Pharmacokinet. 2018 Oct;57(10):1217–28. 10.1007/s40262-018-0641-x 

20 Wu Y, Allegaert K, Flint RB, Simons SH, Krekels EH, Knibbe CA, et al. Prediction of glomerular filtration rate maturation across preterm and term neonates and young infants using inulin as marker. AAPS J. 2022 Feb;24(2):38. 10.1208/S12248-022-00688-Z/FIGURES/4 10.1208/s12248-022-00688-z 

21 Salem F, Johnson TN, Hodgkinson AB, Ogungbenro K, Rostami-Hodjegan A. Does “Birth” as an Event impact maturation trajectory of renal clearance via glomerular filtration? Reexamining data in preterm and full-term neonates by avoiding the creatinine bias. J Clin Pharmacol. 2021 Feb;61(2):159–71. 10.1002/jcph.1725 

22 Rhodin MM, Anderson BJ, Peters AM, Coulthard MG, Wilkins B, Cole M, et al. Human renal function maturation: a quantitative description using weight and postmenstrual age. Pediatr Nephrol. 2009 Jan;24(1):67–76. 10.1007/s00467-008-0997-5 

23 van Donge T, Allegaert K, Gotta V, Smits A, Levtchenko E, Mekahli D, et al. Characterizing dynamics of serum creatinine and creatinine clearance in extremely low birth weight neonates during the first 6 weeks of life. Pediatr Nephrol. 2021 Mar;36(3):649–59. 10.1007/S00467-020-04749-3 10.1007/s00467-020-04749-3 

24 Bi Y, Liu J, Li L, Yu J, Bhattaram A, Bewernitz M, et al. Role of Model-Informed Drug Development in Pediatric Drug Development, Regulatory Evaluation, and Labeling. J Clin Pharmacol. 2019 Sep;59(S1 Suppl 1):S104–11. 10.1002/jcph.1478 

25 Madabushi R, Seo P, Zhao L, Tegenge M, Zhu H. Review: Role of Model-Informed Drug Development Approaches in the Lifecycle of Drug Development and Regulatory Decision-Making. Pharm Res. 2022 Aug;39(8):1669–80. 10.1007/s11095-022-03288-w 

26 Hartman SJ, Swaving JG, van Beek SW, van Groen BD, de Hoop M, van der Zanden TM, et al. A New Framework to Implement Model-Informed Dosing in Clinical Guidelines: Piperacillin and Amikacin as Proof of Concept. Front Pharmacol. 2020 Dec;11:592204. 10.3389/FPHAR.2020.592204 10.3389/fphar.2020.592204 

27 Versporten A, Bielicki J, Drapier N, Sharland M, Goossens H; ARPEC project group. 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 Apr;71(4):1106–17. 10.1093/jac/dkv418 

28 Versporten A, Sharland M, Bielicki J, Drapier N, Vankerckhoven V, Goossens H; ARPEC Project Group Members. The antibiotic resistance and prescribing in European Children project: a neonatal and pediatric antimicrobial web-based point prevalence survey in 73 hospitals worldwide. Pediatr Infect Dis J. 2013 Jun;32(6):e242–53. 10.1097/INF.0b013e318286c612 

29 De Cock RF, Allegaert K, Schreuder MF, Sherwin CM, de Hoog M, van den Anker JN, et al. Maturation of the glomerular filtration rate in neonates, as reflected by amikacin clearance. Clin Pharmacokinet. 2012 Feb;51(2):105–17. 10.2165/11595640-000000000-00000 

30 Allegaert K, Scheers I, Cossey V, Anderson BJ. Covariates of amikacin clearance in neonates: the impact of postnatal age on predictability. Drug Metab Lett. 2008 Dec;2(4):286–9. 10.2174/187231208786734157 

31 Allegaert K, Anderson BJ, Cossey V, Holford NH. Limited predictability of amikacin clearance in extreme premature neonates at birth. Br J Clin Pharmacol. 2006 Jan;61(1):39–48. 10.1111/j.1365-2125.2005.02530.x 

32 Sherwin CM, Svahn S, Van der Linden A, Broadbent RS, Medlicott NJ, Reith DM. Individualised dosing of amikacin in neonates: a pharmacokinetic/pharmacodynamic analysis. Eur J Clin Pharmacol. 2009 Jul;65(7):705–13. 10.1007/S00228-009-0637-4 10.1007/s00228-009-0637-4 

33 Schreuder MF, Wilhelm AJ, Bökenkamp A, Timmermans SM, Delemarre-van de Waal HA, van Wijk JA. Impact of gestational age and birth weight on amikacin clearance on day 1 of life. Clin J Am Soc Nephrol. 2009 Nov;4(11):1774–8. 10.2215/CJN.02230409 

34 Smits A, De Cock RF, Allegaert K, Vanhaesebrouck S, Danhof M, Knibbe CA. Prospective Evaluation of a Model-Based Dosing Regimen for Amikacin in Preterm and Term Neonates in Clinical Practice. Antimicrob Agents Chemother. 2015 Oct;59(10):6344–51. 10.1128/AAC.01157-15 

35 Gotta V, van den Anker J, Pfister M. [Understanding and reducing the risk of adverse drug reactions in pediatric patients]. Ther Umsch. 2015 Dec;72(11-12):679–86. 10.1024/0040-5930/a000737 

36 Kaushal R, Bates DW, Landrigan C, McKenna KJ, Clapp MD, Federico F, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA. 2001 Apr;285(16):2114–20. 10.1001/jama.285.16.2114 

37 Kaushal R, Goldmann DA, Keohane CA, Abramson EL, Woolf S, Yoon C, et al. Medication errors in paediatric outpatients. Qual Saf Health Care. 2010 Dec;19(6):e30. 10.1136/qshc.2008.031179 

38 Vinks AA, Emoto C, Fukuda T. Modeling and simulation in pediatric drug therapy: application of pharmacometrics to define the right dose for children. Clin Pharmacol Ther. 2015 Sep;98(3):298–308. 10.1002/cpt.169 

39 Hong Y, Kowalski KG, Zhang J, Zhu L, Horga M, Bertz R, et al. Model-based approach for optimization of atazanavir dose recommendations for HIV-infected pediatric patients. Antimicrob Agents Chemother. 2011 Dec;55(12):5746–52. 10.1128/AAC.00554-11 

40 Wilbaux M, Fuchs A, Samardzic J, Rodieux F, Csajka C, Allegaert K, et al. Pharmacometric Approaches to Personalize Use of Primarily Renally Eliminated Antibiotics in Preterm and Term Neonates. J Clin Pharmacol. 2016 Aug;56(8):909–35. 10.1002/JCPH.705 10.1002/jcph.705 

41 Dao K, Fuchs A, André P, Giannoni E, Decosterd LA, Marchetti O, et al. Dosing strategies of imipenem in neonates based on pharmacometric modelling and simulation. J Antimicrob Chemother. 2022 Feb;77(2):457–65. 10.1093/JAC/DKAB394 10.1093/jac/dkab394 

42 Dao K, Guidi M, André P, Giannoni E, Basterrechea S, Zhao W, et al. Optimisation of vancomycin exposure in neonates based on the best level of evidence. Pharmacol Res. 2020 Apr;154:104278. 10.1016/J.PHRS.2019.104278 10.1016/j.phrs.2019.104278 

43 van Donge T, Fuchs A, Leroux S, Pfister M, Rodieux F, Atkinson A, et al. Amoxicillin Dosing Regimens for the Treatment of Neonatal Sepsis: Balancing Efficacy and Neurotoxicity. Neonatology. 2020;117(5):619–27. 10.1159/000509751 

44 Bräm DS, Nahum U, Atkinson A, Koch G, Pfister M. Evaluation of machine learning methods for covariate data imputation in pharmacometrics. CPT Pharmacometrics Syst Pharmacol. 2022 Dec;11(12):1638–48. 10.1002/PSP4.12874 10.1002/psp4.12874 

45 Wilbaux M, Kasser S, Wellmann S, Lapaire O, van den Anker JN, Pfister M. Characterizing and Forecasting Individual Weight Changes in Term Neonates. J Pediatr. 2016 Jun;173:101–107.e10. 10.1016/J.JPEDS.2016.02.044 10.1016/j.jpeds.2016.02.044 

46 Gotta V, Buclin T, Csajka C, Widmer N. Systematic review of population pharmacokinetic analyses of imatinib and relationships with treatment outcomes. Ther Drug Monit. 2013 Apr;35(2):150–67. 10.1097/FTD.0b013e318284ef11 

47 Alhadab AA, Ahmed MA, Brundage RC. Amikacin Pharmacokinetic-Pharmacodynamic Analysis in Pediatric Cancer Patients. Antimicrob Agents Chemother. 2018 Mar;62(4):e01781-17. 10.1128/AAC.01781-17 

48 Rao SC, Srinivasjois R, Moon K. One dose per day compared to multiple doses per day of gentamicin for treatment of suspected or proven sepsis in neonates. Cochrane Database Syst Rev. 2016 Dec;12(12):CD005091. 10.1002/14651858.CD005091.PUB4 10.1002/14651858.CD005091.pub4 

49 Hanberger H, Edlund C, Furebring M, G Giske C, Melhus A, Nilsson LE, et al.; Swedish Reference Group for Antibiotics. Rational use of aminoglycosides—review and recommendations by the Swedish Reference Group for Antibiotics (SRGA). Scand J Infect Dis. 2013 Mar;45(3):161–75. 10.3109/00365548.2012.747694 

50 Craig WA. Optimizing aminoglycoside use. Crit Care Clin. 2011 Jan;27(1):107–21. 10.1016/J.CCC.2010.11.006 10.1016/j.ccc.2010.11.006 

51 Paioni P, Jäggi VF, Tilen R, Seiler M, Baumann P, Bräm DS, et al. Gentamicin population pharmacokinetics in pediatric patients—a prospective study with data analysis using the saemix package in r. Pharmaceutics. 2021 Oct;13(10):1596. 10.3390/PHARMACEUTICS13101596/S1 10.3390/pharmaceutics13101596 

52 Goers R, Coman Schmid D, Jäggi VF, Paioni P, Okoniewski MJ, Parker A, et al. SwissPKcdw - A clinical data warehouse for the optimization of pediatric dosing regimens. CPT Pharmacometrics Syst Pharmacol. 2021 Dec;10(12):1578–87. 10.1002/PSP4.12723 10.1002/psp4.12723 

53 van Donge T, Bielicki JA, van den Anker J, Pfister M. Key Components for Antibiotic Dose Optimization of Sepsis in Neonates and Infants. Front Pediatr. 2018 Oct;6:325. 10.3389/FPED.2018.00325 10.3389/fped.2018.00325 

Appendix

The appendix is available in the pdf version of the article at https://doi.org/10.57187/s.3632.