Pharmacokinetics of oral spironolactone in infants up to 2 years of age.

PURPOSE: Spironolactone is a potassium sparing diuretic used for decades. Until now, pharmacokinetic (PK) studies of spironolactone have not been conducted in infants and therefore pediatric dosing is based on expert opinion. We aimed to describe the PK profiles of spironolactone and its main metabolites (7alpha-thiomethylspironolactone (TMS) and canrenone (CAN)) in infants up to two years of age. METHODS: The PK of spironolactone and its main metabolites were evaluated following an oral administration of spironolactone (1 mg/kg/dose) to pediatric patients with chronic heart failure, ascites, and/or oedema. The plasma concentration of spironolactone and metabolites (TMS and CAN) was determined using an ultra-high performance liquid chromatography tandem mass spectrometry (UHPLC-MS/MS). Based on rich sampling PK data, the estimation of population PK parameters was performed using nonlinear mixed-effects modelling software Monolix 2018R2. RESULTS: A total of 150 spironolactone, 158 TMS, and 158 CAN concentrations from 23 patients (ages: 3 days-21 months; median weight 4.3 kg (2.2-12.6)) were available for PK analysis. A one-compartment model for spironolactone, TMS, and CAN best fitted the data. The median (range) of individual estimated apparent clearance values were 47.7 (11.9-138.1) L/h for spironolactone, 9.7 (1.5-66.9) L/h for TMS, and 1.0 (0.2-5.9) L/h for CAN. The disposition of spironolactone and metabolites was mainly affected by size of the patient: body weight explained 22% of inter-individual variability of spironolactone clearance. None of the undesirable effects of spironolactone was documented during the study period. CONCLUSION: The pharmacokinetics of spironolactone and its metabolites was highly variable between patients below 2 years of age. Body weight explained a significant part of this variability; this highlights the need to take it into account for dosing prescription in this population. (Clinical trial Registration Number 2013-001189-40).

Influence of early identification and therapy on long-term outcomes in early-onset MTHFR deficiency.

MTHFR deficiency is a severe inborn error of metabolism leading to impairment of the remethylation of homocysteine to methionine. Neonatal and early-onset patients mostly exhibit a life-threatening acute neurologic deterioration. Furthermore, data on early-onset patients' long-term outcomes are scarce. The aims of this study were (1) to study and describe the clinical and laboratory parameters of early-onset MTHFR-deficient patients (i.e., ≤3 months of age) and (2) to identify predictive factors for severe neurodevelopmental outcomes in a cohort with early and late onset MTHFR-deficient patients. To this end, we conducted a retrospective, multicentric, international cohort study on 72 patients with MTHFR deficiency from 32 international metabolic centres. Characteristics of the 32 patients with early-onset MTHFR deficiency were described at time of diagnosis and at the last follow-up visit. Logistic regression analysis was used to identify predictive factors of severe neurodevelopmental outcome in a broader set of patients with early and non-early-onset MTHFR deficiency. The majority of early-onset MTHFR-deficient patients (n = 32) exhibited neurologic symptoms (76%) and feeding difficulties (70%) at time of diagnosis. At the last follow-up visit (median follow-up time of 8.1 years), 76% of treated early-onset patients (n = 29) exhibited a severe neurodevelopmental outcome. Among the whole study population of 64 patients, pre-symptomatic diagnosis was independently associated with a significantly better neurodevelopmental outcome (adjusted OR 0.004, [0.002-0.232]; p = 0.003). This study provides evidence for benefits of pre-symptomatic diagnosis and appropriate therapeutic management, highlighting the need for systematic newborn screening for MTHFR deficiency and pre-symptomatic treatment that may improve outcome.

Modeling of levothyroxine in newborns and infants with congenital hypothyroidism: challenges and opportunities of a rare disease multi-center study.

Modeling of retrospectively collected multi-center data of a rare disease in pediatrics is challenging because laboratory data can stem from several decades measured with different assays. Here we present a retrospective pharmacometrics (PMX) based data analysis of the rare disease congenital hypothyroidism (CH) in newborns and infants. Our overall aim is to develop a model that can be applied to optimize dosing in this pediatric patient population since suboptimal treatment of CH during the first 2 years of life is associated with a reduced intelligence quotient between 10 and 14 years. The first goal is to describe a retrospectively collected dataset consisting of 61 newborns and infants with CH up to 2 years of age. Overall, 505 measurements of free thyroxine (FT4) and 510 measurements of thyrotropin or thyroid-stimulating hormone were available from patients receiving substitution treatment with levothyroxine (LT4). The second goal is to introduce a scale/location-scale normalization method to merge available FT4 measurements since 34 different postnatal age- and assay-specific laboratory reference ranges were applied. This method takes into account the change of the distribution of FT4 values over time, i.e. a transformation from right-skewed towards normality during LT4 treatment. The third goal is to develop a practical and useful PMX model for LT4 treatment to characterize FT4 measurements, which is applicable within a clinical setting. In summary, a time-dependent normalization method and a practical PMX model are presented. Since there is no on-going or planned development of new pharmacological approaches for CH, PMX based modeling and simulation can be leveraged to personalize dosing with the goal to enhance longer-term neurological outcome in children with the rare disease CH.

Body Surface Area-Based Dosing Regimen of Caspofungin in Children: a Population Pharmacokinetics Confirmatory Study.

We evaluated the population pharmacokinetics of caspofungin in children (2 to 12 years of age). The real-world data from 48 children were best fit by a two-compartment model with first-order elimination. Subsequent covariate analysis demonstrated that body surface area had a significant correlation with caspofungin pharmacokinetics, compared to body weight. The population pharmacokinetics of caspofungin confirmed that adjustment of caspofungin dosage based on body surface area is most appropriate for pediatric use.

Population Pharmacokinetics and Dosing Optimization of Amoxicillin in Neonates and Young Infants.

Amoxicillin is widely used to treat bacterial infections in neonates. However, considerable intercenter variability in dosage regimens of antibiotics exists in clinical practice. The pharmacokinetics of amoxicillin has been described in only a few preterm neonates. Thus, we aimed to evaluate the population pharmacokinetics of amoxicillin through a large sample size covering the entire age range of neonates and young infants and to establish evidence-based dosage regimens based on developmental pharmacokinetics-pharmacodynamics. This is a prospective, multicenter, pharmacokinetic study using an opportunistic sampling design. Amoxicillin plasma concentrations were determined using high-performance liquid chromatography. Population pharmacokinetic analysis was performed using NONMEM. A total of 224 pharmacokinetic samples from 187 newborns (postmenstrual age range, 28.4 to 46.3 weeks) were available for analysis. A two-compartment model with first-order elimination was used to describe population pharmacokinetics. Covariate analysis showed that current weight, postnatal age, and gestational age were significant covariates. The final model was further validated for predictive performance in an independent cohort of patients. Monte Carlo simulation demonstrated that for early-onset sepsis, the currently used dosage regimen (25 mg/kg twice daily [BID]) resulted in 99.0% of premature neonates and 87.3% of term neonates achieving the pharmacodynamic target (percent time above MIC), using a MIC breakpoint of 1 mg/liter. For late-onset sepsis, 86.1% of premature neonates treated with 25 mg/kg three times a day (TID) and 79.0% of term neonates receiving 25 mg/kg four times a day (QID) reached the pharmacodynamic target, using a MIC breakpoint of 2 mg/liter. The population pharmacokinetics of amoxicillin was assessed in neonates and young infants. A dosage regimen was established based on developmental pharmacokinetics-pharmacodynamics.

Population pharmacokinetic meta-analysis of individual data to design the first randomized efficacy trial of vancomycin in neonates and young infants.

OBJECTIVES: In the absence of consensus, the present meta-analysis was performed to determine an optimal dosing regimen of vancomycin for neonates. METHODS: A 'meta-model' with 4894 concentrations from 1631 neonates was built using NONMEM, and Monte Carlo simulations were performed to design an optimal intermittent infusion, aiming to reach a target AUC0-24 of 400 mg·h/L at steady-state in at least 80% of neonates. RESULTS: A two-compartment model best fitted the data. Current weight, postmenstrual age (PMA) and serum creatinine were the significant covariates for CL. After model validation, simulations showed that a loading dose (25 mg/kg) and a maintenance dose (15 mg/kg q12h if <35 weeks PMA and 15 mg/kg q8h if ≥35 weeks PMA) achieved the AUC0-24 target earlier than a standard 'Blue Book' dosage regimen in >89% of the treated patients. CONCLUSIONS: The results of a population meta-analysis of vancomycin data have been used to develop a new dosing regimen for neonatal use and to assist in the design of the model-based, multinational European trial, NeoVanc.

Pilot Study of Model-Based Dosage Individualization of Ganciclovir in Neonates and Young Infants with Congenital Cytomegalovirus Infection.

Newborns with congenital cytomegalovirus (CMV) infection are at high risk for developing permanent sequelae. Intravenous ganciclovir therapy is frequently used for the treatment of congenital CMV infection. A target area under the concentration-time curve from 0 to 24 h (AUC0-24) of 40 to 50 µg · h/ml is recommended. The standard dose has resulted in a large variability in ganciclovir exposure in newborns, indicating the unmet need of dosage individualization for this vulnerable population, but the implementation of this strategy remains challenging in clinical practice. We aim to evaluate the clinical utility of model-based dosage individualization of ganciclovir in newborns using an opportunistic sampling approach. The predictive performance of a published ganciclovir population pharmacokinetic model was evaluated using an independent patient cohort. The individual dose was adjusted based on the target AUC0-24 to ensure its efficacy. A total of 26 newborns with congenital CMV infection were included in the present study. Only 11 (42.3%) patients achieved the target AUC0-24 after being given the standard dose. For all the subtherapeutic patients (achieving <80% of the target AUC) (n = 5), a model-based dosage adjustment was performed using the Bayesian estimation method combined with the opportunistic sampling strategy. The adjusted doses were increased by 28.6% to 60.0% in these five patients, and all adapted AUC0-24 values achieved the target (range, 48.6 to 66.1 µg · h/ml). The clinical utility of model-based dosing individualization of ganciclovir was demonstrated in newborns with congenital CMV infection. The population pharmacokinetic model combined with the opportunistic sampling strategy provides a clinically feasible method to adapt the ganciclovir dose in neonatal clinical practice. (This study has been registered at ClinicalTrials.gov under registration no. NCT03113344.).

Penetration of Cefotaxime into Cerebrospinal Fluid in Neonates and Young Infants.

Cefotaxime is the first-line treatment for meningitis in neonates and young infants. However, limited data on cefotaxime cerebrospinal fluid (CSF) concentrations in neonates and young infants were available. The aim of the present study was to evaluate the penetration of cefotaxime into CSF in neonates and young infants. Blood and CSF samples were collected from neonates and young infants treated with cefotaxime using an opportunistic pharmacokinetic sampling strategy, and concentrations were quantified by high-performance liquid chromatography-tandem mass spectrometry. The analysis was performed using NONMEM and R software. Thirty neonates and young infants (postmenstrual age range, 25.4 to 47.4 weeks) were included. A total of 67 plasma samples and 30 CSF samples were available for analysis. Cefotaxime plasma and CSF concentrations ranged from 2.30 to 175.42 mg/liter and from 0.39 to 25.38 mg/liter, respectively. The median ratio of the CSF concentration to the plasma concentration was 0.28 (range, 0.06 to 0.76). Monte Carlo simulation demonstrated that 88.4% and 63.9% of hypothetical neonates treated with 50 mg/kg of body weight three times a day (TID) would reach the pharmacodynamic target (the percentage of the dosing interval that the free antimicrobial drug concentration remains above the MIC, 70%) using the standard EUCAST MIC susceptibility breakpoints of 2 mg/liter and 4 mg/liter, respectively. The penetration of cefotaxime into the CSF of neonates and young infants was evaluated using an opportunistic sampling approach. A dosage regimen of 50 mg/kg TID could cover the most causative pathogens with MICs of <2 mg/liter. Individual dosage adaptation was required for more resistant bacterial strains, such as Staphylococcus aureus.

Developmental pharmacogenetics of CYP2C19 in neonates and young infants: omeprazole as a probe drug.

AIMS: Although substantial progress has been made in understanding of ontogeny of drug metabolism, there is still a gap of knowledge in developmental pharmacogenetics in neonates. We hypothesized that both age and pharmacogenetics might explain the developmental pattern of CYP2C19. We conducted a population pharmacokinetic-pharmacogenetic study to quantify the developmental pharmacogenetics of CYP2C19 in neonates and young infants using omeprazole as a probe drug. METHODS: Pharmacokinetic samples were collected from 51 Caucasian neonates and young infants, who were receiving omeprazole treatment. Population pharmacokinetic-pharmacogenetic analysis of omeprazole and its metabolites was performed using NONMEM. RESULTS: Data fitted a one-compartment parent and metabolite model with first-order absorption and elimination. CYP2C19 and CYP3A4 are predominantly involved in the metabolism of omeprazole despite their relatively low activities compared to adults. The clearance of omeprazole converted to 5-hydroxy-omeprazole (CLOMZ-M1 ) increases with postnatal age. In CYP2C19 poor and intermediate metabolizers, model-predicted CLOMZ-M1 are 12.5% (5-95% percentile: 3-14.9%) and 44.9% (5-95% percentile: 29.9-72.6%) of the value in extensive/ultrarapid metabolizer, respectively. Model-predicted absorption rate constant of omeprazole is 6.93 (5-95% percentile: 3.01-14.61) times higher in ABCB1 homozygous mutant patients, 1.86 (5-95% percentile: 0.86-3.47) times higher in ABCB1 heterozygous patients than that in ABCB1 homozygous wild-type patients. CONCLUSIONS: Developmental pharmacogenetics of CYP2C19 was quantitatively described in neonates and young infants using omeprazole as a probe drug. Our findings emphasize the importance of semiphysiological developmental pharmacokinetic modelling approach when evaluating developmental pharmacogenetics of drugs with multiple routes of biotransformation.

Capacities and Competences for Drug Evaluation in European Neonatal Intensive Care Units: A Survey and Key Issues for Improvement.

BACKGROUND: Multicenter neonatal clinical trials aim to provide evidence-based drug evaluation, but recruiting neonates requires collaboration, standard procedures, and trained neonatologists. METHODS: A questionnaire based on a previous Delphi study was sent to European neonatal intensive care units (NICUs) to collect their research experience and identify areas for improvement. RESULTS: Of 247 NICUs,79 (32%) responded: 69 were level III units and 10 were level II units. In level III centers, 62% had medical staff dedicated to research and 65% conducted regular in-house audits. Similarities were observed in the median number of trials per year (level II: 2; level III: 5), Good Clinical Practice training (level II: 78%; level III: 66%), and standard operating procedures (level II: 63%; level III: 71%). Most NICUs had access to scientific advice for trial design, conduct, data management, and regulatory aspects. Involvement of patient advocacy groups was more common in level II units (level II: 75%; level III: 59%). A "quality" score of 34 "quality" research items was calculated for all centers (mean: 23.2 ± 6.2; range: 6-34). CONCLUSION: Research experience and processes vary across Europe. Harmonizing research practices and setting standards will allow building a European neonatal network for effective, safe, and quality neonatal drug development.

Drug evaluation in children 10years after the European pediatric regulation current challenges and perspectives.

The European pediatric regulation, that entered into force in June 2007 with the objectives to improve the health of children in Europe, dramatically changed the regulatory environment of paediatric drug evaluation in Europe. The recent 10years European medicines agency (EMA) report showed that the number of paediatric trials increased and that 238 new medicines and indications for use in children were authorised in the EU. However, results remain constrated and futur developments require european collaborations beween all experts in developmental pharmacology, drug evaluation and trial conduct, training, all aspects already considered in different EU paediatric programs.

Principles and applications of pharmacometrics in drug evaluation in children.

Drug evaluation in children is difficult for many well-identified reasons and many drugs are still used off-label. Innovative approaches are particularly adapted to the paediatric and neonatal populations, as clinical trials are difficult to conduct, need adapted designs in order to define the optimal dosage regimen in many diseases and therapeutic areas. Population approaches to define pharmacokinetics and pharmacokinetic/pharmacodynamics are now more currently used to define dosing regimens, adapted to the different paediatric and neonatal age groups, that allow to increase efficacy and reduce toxicity, by taking into account factors explaining variability in drug response. Such approaches are presented and the evaluation of vancomycin in neonates is detailed as different steps allowed validation of the optimal strategy to administer vancomycin in neonates.

A Population and Developmental Pharmacokinetic Analysis To Evaluate and Optimize Cefotaxime Dosing Regimen in Neonates and Young Infants.

Cefotaxime is one of the most frequently prescribed antibiotics for the treatment of Gram-negative bacterial sepsis in neonates. However, the dosing regimens routinely used in clinical practice vary considerably. The objective of the present study was to conduct a population pharmacokinetic study of cefotaxime in neonates and young infants in order to evaluate and optimize the dosing regimen. An opportunistic sampling strategy combined with population pharmacokinetic analysis using NONMEM software was performed. Cefotaxime concentrations were measured by high-performance liquid chromatography-tandem mass spectrometry. Developmental pharmacokinetics-pharmacodynamics, the microbiological pathogens, and safety aspects were taken into account to optimize the dose. The pharmacokinetic data from 100 neonates (gestational age [GA] range, 23 to 42 weeks) were modeled with an allometric two-compartment model with first-order elimination. The median values for clearance and the volume of distribution at steady state were 0.12 liter/h/kg of body weight and 0.64 liter/kg, respectively. The covariate analysis showed that current weight, GA, and postnatal age (PNA) had significant impacts on cefotaxime pharmacokinetics. Monte Carlo simulations demonstrated that the current dose recommendations underdosed older newborns. A model-based dosing regimen of 50 mg/kg twice a day to four times a day, according to GA and PNA, was established. The associated risk of overdose for the proposed dosing regimen was 0.01%. We determined the population pharmacokinetics of cefotaxime and established a model-based dosing regimen to optimize treatment for neonates and young infants.

Clinical Utility and Safety of a Model-Based Patient-Tailored Dose of Vancomycin in Neonates.

Pharmacokinetic modeling has often been applied to evaluate vancomycin pharmacokinetics in neonates. However, clinical application of the model-based personalized vancomycin therapy is still limited. The objective of the present study was to evaluate the clinical utility and safety of a model-based patient-tailored dose of vancomycin in neonates. A model-based vancomycin dosing calculator, developed from a population pharmacokinetic study, has been integrated into the routine clinical care in 3 neonatal intensive care units (Robert Debré, Cochin Port Royal, and Clocheville hospitals) between 2012 and 2014. The target attainment rate, defined as the percentage of patients with a first therapeutic drug monitoring serum vancomycin concentration achieving the target window of 15 to 25 mg/liter, was selected as an endpoint for evaluating the clinical utility. The safety evaluation was focused on nephrotoxicity. The clinical application of the model-based patient-tailored dose of vancomycin has been demonstrated in 190 neonates. The mean (standard deviation) gestational and postnatal ages of the study population were 31.1 (4.9) weeks and 16.7 (21.7) days, respectively. The target attainment rate increased from 41% to 72% without any case of vancomycin-related nephrotoxicity. This proof-of-concept study provides evidence for integrating model-based antimicrobial therapy in neonatal routine care.

Use of Machine Learning for Dosage Individualization of Vancomycin in Neonates.

BACKGROUND AND OBJECTIVE: High variability in vancomycin exposure in neonates requires advanced individualized dosing regimens. Achieving steady-state trough concentration (C0) and steady-state area-under-curve (AUC0-24) targets is important to optimize treatment. The objective was to evaluate whether machine learning (ML) can be used to predict these treatment targets to calculate optimal individual dosing regimens under intermittent administration conditions. METHODS: C0 were retrieved from a large neonatal vancomycin dataset. Individual estimates of AUC0-24 were obtained from Bayesian post hoc estimation. Various ML algorithms were used for model building to C0 and AUC0-24. An external dataset was used for predictive performance evaluation. RESULTS: Before starting treatment, C0 can be predicted a priori using the Catboost-based C0-ML model combined with dosing regimen and nine covariates. External validation results showed a 42.5% improvement in prediction accuracy by using the ML model compared with the population pharmacokinetic model. The virtual trial showed that using the ML optimized dose; 80.3% of the virtual neonates achieved the pharmacodynamic target (C0 in the range of 10-20 mg/L), much higher than the international standard dose (37.7-61.5%). Once therapeutic drug monitoring (TDM) measurements (C0) in patients have been obtained, AUC0-24 can be further predicted using the Catboost-based AUC-ML model combined with C0 and nine covariates. External validation results showed that the AUC-ML model can achieve an prediction accuracy of 80.3%. CONCLUSION: C0-based and AUC0-24-based ML models were developed accurately and precisely. These can be used for individual dose recommendations of vancomycin in neonates before treatment and dose revision after the first TDM result is obtained, respectively.

Contribution of Population Pharmacokinetics of Glycopeptides and Antifungals to Dosage Adaptation in Paediatric Onco-hematological Malignancies: A Review.

The response to medications in children differs not only in comparison to adults but also between children of the different age groups and according to the disease. This is true for anti-infectives that are widely prescribed in children with malignancy. In the absence of pharmacokinetic/pharmacodynamic paediatric studies, dosage is frequently based on protocols adapted to adults. After a short presentation of the drugs, we reviewed the population pharmacokinetic studies available for glycopeptides (vancomycin and teicoplanin, n = 5) and antifungals (voriconazole, posaconazole, and amphotericin B, n = 9) currently administered in children with onco-hematological malignancies. For each of them, we reported the main study characteristics including identified covariates affecting pharmacokinetics and proposed paediatric dosage recommendations. This review highlighted the very limited amount of data available, the lack of consensus regarding PK/PD targets used for dosing optimization and regarding dosage recommendations when available. Additional PK studies are urgently needed in this specific patient population. In addition to pharmacokinetics, efficacy may be altered in immunocompromised patients and prospective clinical evaluation of new dosage regimen should be provided as they are missing in most cases.

Serum Creatinine and Serum Cystatin C are Both Relevant Renal Markers to Estimate Vancomycin Clearance in Critically Ill Neonates.

Purpose: Serum creatinine (SCr) is used as a marker of kidney function to guide dosing of renally eliminated drugs. Serum Cystatin C (S-CysC) has been suggested as a more reliable kidney marker than SCr in adults and children. Purpose of this study was to investigate S-CysC as alternative renal marker to SCr for estimating vancomycin clearance in neonates undergoing intensive care. Methods: Vancomycin pharmacokinetics (PK), SCr and S-CysC data were collected in patients undergoing vancomycin treatment in the neonatal intensive care unit of Robert Debré Hospital - Paris. A population PK analysis was performed utilizing routine therapeutic drug monitoring samples. S-CysC and SCr were compared as covariates on vancomycin clearance using stepwise covariate modeling (forward inclusion [p < 0.05] and backward elimination [p < 0.01]). Model performance was evaluated by graphical and statistical criteria. Results: A total of 108 vancomycin concentrations from 66 patients (postmenstrual age [PMA] of 26-46 weeks) were modeled with an allometric one-compartment model. The median (range) values for SCr and S-CysC were 41 (12-153) µmol/l and 1.43 (0.95-2.83) mg/l, respectively. Following stepwise covariate model building, SCr was retained as single marker of kidney function (after accounting for weight and PMA) in the final model. Compared to the final model based on SCr, the alternative model based on S-CysC showed very similar performance (e.g. BIC of 578.3 vs. 576.4) but included one additional covariate: impact of mechanical ventilation on vancomycin clearance, in addition to the effects of size and maturation. Conclusion: ill neonates. However, if using S-CysC for this purpose mechanical ventilation needs to be taken into account.

Diagnosis of Neonatal Late-Onset Infection in Very Preterm Infant: Inter-Observer Agreement and International Classifications.

BACKGROUND: The definition of late-onset bacterial sepsis (LOS) in very preterm infants is not unified. The objective was to assess the concordance of LOS diagnosis between experts in neonatal infection and international classifications and to evaluate the potential impact on heart rate variability and rate of "bronchopulmonary dysplasia or death". METHODS: A retrospective (2017-2020) multicenter study including hospitalized infants born before 31 weeks of gestation with intention to treat at least 5-days with antibiotics was performed. LOS was classified as "certain or probable" or "doubtful" independently by five experts and according to four international classifications with concordance assessed by Fleiss's kappa test. RESULTS: LOS was suspected at seven days (IQR: 5-11) of life in 48 infants. Following expert classification, 36 of them (75%) were considered as "certain or probable" (kappa = 0.41). Following international classification, this number varied from 13 to 46 (kappa = -0.08). Using the expert classification, "bronchopulmonary dysplasia or death" occurred less frequently in the doubtful group (25% vs. 78%, p < 0.001). Differences existed in HRV changes between the two groups. CONCLUSION: The definition of LOS is not consensual with a low international and moderate inter-observer agreement. This affects the evaluation of associated organ dysfunction and prognosis.

Drug Clearance in Neonates: A Combination of Population Pharmacokinetic Modelling and Machine Learning Approaches to Improve Individual Prediction.

BACKGROUND: Population pharmacokinetic evaluations have been widely used in neonatal pharmacokinetic studies, while machine learning has become a popular approach to solving complex problems in the current era of big data. OBJECTIVE: The aim of this proof-of-concept study was to evaluate whether combining population pharmacokinetic and machine learning approaches could provide a more accurate prediction of the clearance of renally eliminated drugs in individual neonates. METHODS: Six drugs that are primarily eliminated by the kidneys were selected (vancomycin, latamoxef, cefepime, azlocillin, ceftazidime, and amoxicillin) as 'proof of concept' compounds. Individual estimates of clearance obtained from population pharmacokinetic models were used as reference clearances, and diverse machine learning methods and nested cross-validation were adopted and evaluated against these reference clearances. The predictive performance of these combined methods was compared with the performance of two other predictive methods: a covariate-based maturation model and a postmenstrual age and body weight scaling model. Relative error was used to evaluate the different methods. RESULTS: The extra tree regressor was selected as the best-fit machine learning method. Using the combined method, more than 95% of predictions for all six drugs had a relative error of < 50% and the mean relative error was reduced by an average of 44.3% and 71.3% compared with the other two predictive methods. CONCLUSION: A combined population pharmacokinetic and machine learning approach provided improved predictions of individual clearances of renally cleared drugs in neonates. For a new patient treated in clinical practice, individual clearance can be predicted a priori using our model code combined with demographic data.