Search

Background: Pharmacogenetic profiling and therapeutic drug monitoring (TDM) have both been proposed to manage inter-individual variability (IIV) in drug exposure. However, determining the most effective approach for estimating exposure for a particular drug remains a challenge. This study aimed to quantitatively assess the circumstances in which pharmacogenetic profiling may outperform TDM in estimating drug exposure, under three sources of variability (IIV, inter-occasion variability [IOV], and residual unexplained variability [RUV]). Methods: Pharmacokinetic models were selected from the literature corresponding to drugs for which pharmacogenetic profiling and TDM are both clinically considered approaches for dose individualization. The models were used to simulate relevant drug exposures (trough concentration or area under the curve [AUC]) under varying degrees of IIV, IOV, and RUV. Results: Six drug cases were selected from the literature. Model-based simulations demonstrated that the percentage of patients for whom pharmacogenetic exposure prediction is superior to TDM differs for each drug case: tacrolimus (11.0%), tamoxifen (12.7%), efavirenz (49.2%), vincristine (49.6%), risperidone (48.1%), and 5-fluorouracil (5-FU) (100%). Generally, in the presence of higher unexplained IIV in combination with lower RUV and IOV, exposure was best estimated by TDM, whereas, under lower unexplained IIV in combination with higher IOV or RUV, pharmacogenetic profiling was preferred. Conclusions: For the drugs with relatively low RUV and IOV (e.g., tamoxifen and tacrolimus), TDM estimated true exposure the best. Conversely, for drugs with similar or lower unexplained IIV (e.g., efavirenz or 5-FU, respectively) combined with relatively high RUV, pharmacogenetic profiling provided the most accurate estimate for most patients. However, genotype prevalence and the relative influence of genotypes on the PK, as well as the ability of TDM to accurately estimate AUC with a limited number of samples, had an impact. The results could be used to support clinical decision making when considering other factors, such as the probability for severe side effects. [ABSTRACT FROM AUTHOR]

Background: Vincristine-induced peripheral neuropathy (VIPN) is a common adverse effect of vincristine, a drug often used in pediatric oncology. Previous studies demonstrated large inter- and intrapatient variability in vincristine pharmacokinetics (PK). Model-informed precision dosing (MIPD) can be applied to calculate patient exposure and individualize dosing using therapeutic drug monitoring (TDM) measurements. This study set out to investigate the PK/pharmacodynamic (PKPD) relationship of VIPN and determine the utility of MIPD to support clinical decisions regarding dose selection and individualization. Methods: Data from 35 pediatric patients were utilized to quantify the relationship between vincristine dose, exposure and the development of VIPN. Measurements of vincristine exposure and VIPN (Common Terminology Criteria for Adverse Events [CTCAE]) were available at baseline and for each subsequent dosing occasions (1–5). A PK and PKPD analysis was performed to assess the inter- and intraindividual variability in vincristine exposure and VIPN over time. In silico trials were performed to portray the utility of vincristine MIPD in pediatric subpopulations with a certain age, weight and cytochrome P450 (CYP) 3A5 genotype distribution. Results: A two-compartmental model with linear PK provided a good description of the vincristine exposure data. Clearance and distribution parameters were related to bodyweight through allometric scaling. A proportional odds model with Markovian elements described the incidence of Grades 0, 1 and ≥ 2 VIPN overdosing occasions. Vincristine area under the curve (AUC) was the most significant exposure metric related to the development of VIPN, where an AUC of 50 ng⋅h/mL was estimated to be related to an average VIPN probability of 40% over five dosing occasions. The incidence of Grade ≥ 2 VIPN reduced from 62.1 to 53.9% for MIPD-based dosing compared with body surface area (BSA)-based dosing in patients. Dose decreases occurred in 81.4% of patients with MIPD (vs. 86.4% for standard dosing) and dose increments were performed in 33.4% of patients (no dose increments allowed for standard dosing). Conclusions: The PK and PKPD analysis supports the use of MIPD to guide clinical dose decisions and reduce the incidence of VIPN. The current work can be used to support decisions with respect to dose selection and dose individualization in children receiving vincristine. [ABSTRACT FROM AUTHOR]

Purpose: Flucloxacillin is a β-lactam penicillin commonly used in the treatment of bone and soft tissue infections. In a recent porcine study, we found surprisingly low time for which the free concentration was maintained above the minimal inhibitory concentration (fT>MIC) in bone and soft tissue, following flucloxacillin oral (PO) and intravenous (IV) administration at 1g every 6h (q6h). In addition to plasma, sampling was obtained from subcutaneous tissue, knee joint, cancellous bone and cortical bone, using microdialysis. To identify flucloxacillin dosing regimens that result in theoretically therapeutic concentrations, we developed a population pharmacokinetic (PK) model for the porcine data, and combined it with a human flucloxacillin population PK model for simulations. Methods: A four-compartment model was developed, and various dosing regimens and modes of administration were simulated. Predicted concentrations were compared to %fT>MIC (0.5 mg/L and 2 mg/L). Results: Continuous infusion (CI) resulted in higher %fT>MIC compared to intermittent administration. For intermittent IV dosing (4, 8 and 12g/24h), fT>MIC (0.5 mg/L) was ≥70% in plasma, and ranged between 42-96% in the sampled tissue in a typical individual. By applying CI, 4g/day was sufficient to achieve ≥98% fT>MIC (0.5 mg/L) in all sampled tissues. For MIC 2 mg/L, ≥50% fT>MIC was only achieved in plasma at CI 8 and 12g/24h and IV 3g q6h. Conclusions: To reach efficacious flucloxacillin bone and tissue concentrations, dose increment or continuous infusion needs to be considered. [ABSTRACT FROM AUTHOR]

Quantitative characterization of evolving tumor resistance under targeted treatment could help identify novel treatment schedules, which may improve the outcome of anti-cancer treatment. In this study, a mathematical model which considers various clonal populations and evolving treatment resistance was developed. With parameter values fitted to the data or informed by literature data, the model could capture previously reported tumor burden dynamics and mutant KRAS levels in circulating tumor DNA (ctDNA) of patients with metastatic colorectal cancer treated with panitumumab. Treatment schedules, including a continuous schedule, intermittent schedules incorporating treatment holidays, and adaptive schedules guided by ctDNA measurements were evaluated using simulations. Compared with the continuous regimen, the simulated intermittent regimen which consisted of 8-week treatment and 4-week suspension prolonged median progression-free survival (PFS) of the simulated population from 36 to 44 weeks. The median time period in which the tumor size stayed below the baseline level (TTS) was prolonged from 52 to 60 weeks. Extending the treatment holiday resulted in inferior outcomes. The simulated adaptive regimens showed to further prolong median PFS to 56–64 weeks and TTS to 114–132 weeks under different treatment designs. A prospective clinical study is required to validate the results and to confirm the added value of the suggested schedules. [ABSTRACT FROM AUTHOR]

PurposeAn item response theory (IRT) pharmacometric framework is presented to characterize Functional Assessment of Cancer Therapy-Breast (FACT-B) data in locally-advanced or metastatic breast cancer patients treated with ado-trastuzumab emtansine (T-DM1) or capecitabine-plus-lapatinib.MethodsIn the IRT model, four latent well-being variables, based on FACT-B general subscales, were used to describe the physical, social/family, emotional and functional well-being. Each breast cancer subscale item was reassigned to one of the other subscales. Longitudinal changes in FACT-B responses and covariate effects were investigated.ResultsThe IRT model could describe both item-level and subscale-level FACT-B data. Non-Asian patients showed better baseline social/family and functional well-being than Asian patients. Moreover, patients with Eastern Cooperative Oncology Group performance status of 0 had better baseline physical and functional well-being. Well-being was described as initially increasing or decreasing before reaching a steady-state, which varied substantially between patients and subscales. T-DM1 exposure was not related to any of the latent variables. Physical well-being worsening was identified in capecitabine-plus-lapatinib-treated patients, whereas T-DM1-treated patients typically stayed stable.ConclusionThe developed framework provides a thorough description of FACT-B longitudinal data. It acknowledges the multi-dimensional nature of the questionnaire and allows covariate and exposure effects to be evaluated on responses. [ABSTRACT FROM AUTHOR]

Purpose: To investigate whether a more frequent monitoring of the absolute neutrophil counts (ANC) during myelosuppressive chemotherapy, together with model-based predictions, can improve therapy management, compared to the limited clinical monitoring typically applied today.Methods: Daily ANC in chemotherapy-treated cancer patients were simulated from a previously published population model describing docetaxel-induced myelosuppression. The simulated values were used to generate predictions of the individual ANC time-courses, given the myelosuppression model. The accuracy of the predicted ANC was evaluated under a range of conditions with reduced amount of ANC measurements.Results: The predictions were most accurate when more data were available for generating the predictions and when making short forecasts. The inaccuracy of ANC predictions was highest around nadir, although a high sensitivity (≥90%) was demonstrated to forecast Grade 4 neutropenia before it occurred. The time for a patient to recover to baseline could be well forecasted 6 days (±1 day) before the typical value occurred on day 17.Conclusions: Daily monitoring of the ANC, together with model-based predictions, could improve anticancer drug treatment by identifying patients at risk for severe neutropenia and predicting when the next cycle could be initiated. [ABSTRACT FROM AUTHOR]

Correction to: I Scientific Reports i https://doi.org/10.1038/s41598-022-08012-7, published online 10 March 2022 The original version of this Article contained errors in Table 1 where the data was listed incorrectly in the column "M I KRAS i patients". [Extracted from the article]

Prolactin elevation is a side effect of all currently available D2 receptor antagonists used in the treatment of schizophrenia. Prolactin elevation is the result of a direct antagonistic D2 effect blocking the tonic inhibition of prolactin release by dopamine. The aims of this work were to assess the correlation between in vitro estimates of D2 receptor affinity and pharmacokinetic-pharmacodynamic model-based estimates obtained from analysis of clinical data using an agonist-antagonist interaction (AAI) model and to assess the value of such a correlation in early prediction of full prolactin time profiles. A population model describing longitudinal prolactin data was fitted to clinical data from 16 clinical phases 1 and 3 trials including five different compounds. Pharmacokinetic data were modeled for each compound and the prolactin model was both fitted in per-compound fits aswell as simultaneously to all prolactin data. Estimates of prolactin elevating potency were compared to corresponding in vitro values and their predictability was evaluated through model-based simulations. The model successfully described the prolactin time course for all compounds. Estimates derived from experimental preclinical data and the model fit of the clinical data were strongly correlated (p<0.001), and simulations adequately predicted the prolactin elevation in five out of six compounds. The AAI model has the potential to be used in drug development to predict prolactin response for a given exposure of D2 antagonists using routinely produced preclinical data. [ABSTRACT FROM AUTHOR]

A previously developed semi-physiological model of chemotherapy-induced myelosuppression has shown consistent system-related parameter and inter-individual variability (IIV) estimates across drugs. A requirement for dose individualization to be useful is relatively low variability between treatment courses (inter-occasion variability [IOV]) in relation to IIV. The objective of this study was to evaluate and compare magnitudes of IOV and IIV in myelosuppression model parameters across six different anti-cancer drug treatments. Neutrophil counts from several treatment courses following therapy with docetaxel, paclitaxel, epirubicin-docetaxel, 5-fluorouracil–epirubicin–cyclophosphamide, topotecan, and etoposide were included in the analysis. The myelosuppression model was fitted to the data using NONMEM VI. IOV in the model parameters baseline neutrophil counts (ANC0), mean transit time through the non-mitotic maturation chain (mean transit time [MTT]), and the parameter describing the concentration–effect relationship (slope), were evaluated for statistical significance ( P < 0.001). Inter-occasion variability in MTT was significant for all the investigated datasets, except for topotecan, and was of similar magnitude (8–16 CV%). IOV in slope was significant for docetaxel, topotecan, and etoposide (19–39 CV%). For all six investigated datasets, the IOV in myelosuppression parameters was lower than the IIV. There was no indication of systematic shifts in the system- or drug sensitivity-related parameters over time across datasets. This study indicates that the semi-physiological model of chemotherapy-induced myelosuppression has potential to be used for prediction of the time-course of myelosuppression in future courses and is, thereby, a valuable step towards individually tailored anticancer drug therapy. [ABSTRACT FROM AUTHOR]

Background and Objective: Preterm and term newborn infants show wide interindividual variability (IIV) in pharmacokinetic parameters of gentamicin. More extensive knowledge and use of predictive covariates could lead to faster attainment of therapeutic concentrations and a reduced need for concentration monitoring. This study was performed to characterize the population pharmacokinetics of gentamicin in preterm and term neonates and to identify and quantify relationships between patient characteristics and IIV. A secondary aim was to evaluate cystatin C as a marker for gentamicin clearance in this patient population.Methods: Data were collected in a prospective study performed in the Neonatal Intensive Care Unit at the University Children's Hospital, Uppsala, Sweden. Population pharmacokinetic modelling was performed using nonlinear mixed-effects modelling (NONMEM) software. Bodyweight was included as the primary covariate according to an allometric power model. Other evaluated covariates were age (postmenstrual age, gestational age [GA], postnatal age [PNA]), markers for renal function (serum creatinine, serum cystatin C) and concomitant medication with cefuroxime, vancomycin or indometacin. Covariate-parameter relationships were explored using a stepwise covariate model building procedure. The predictive performance of the developed model was evaluated using an independent external dataset for a similar patient population.Results: Sixty-one newborn infants (GA range 23.3-42.1 weeks, PNA range 0-45 days) were enrolled in the study. In total, 894 serum gentamicin samples were included in the analysis. The concentration-time profile was described using a three-compartment model. Gentamicin clearance increased with the GA and PNA (included in a nonlinear fashion). The GA was also identified as having a significant influence on the central volume of distribution, with a preterm neonate having a larger central volume of distribution per kilogram of bodyweight than a term neonate. Cystatin C and creatinine were not correlated with gentamicin clearance in this study population. The external dataset was well predicted by the developed model.Conclusion: Bodyweight and age (GA and PNA) were found to be major factors contributing to IIV in gentamicin clearance in neonates. Based on these data, cystatin C and serum creatinine were not correlated with gentamicin clearance and therefore not likely to be predictive markers of renal function in this patient population. Based on predictions from the developed model, preterm neonates do not reach targeted peak and trough gentamicin concentrations after a standard dosage regimen of 4 mg/kg given once daily, suggesting a need for higher loading doses and prolonged dosing intervals in this patient population. [ABSTRACT FROM AUTHOR]

OBJECTIVE: The objective of the study was to characterise the population pharmacokinetic  properties  of  itraconazole  and  its  active metabolite hydroxy-itraconazole in a representative paediatric population of cystic fibrosis and bone marrow transplant (BMT) patients and to identify patient characteristics influencing the pharmacokinetics of itraconazole. The ultimate goals were to determine the relative bioavailability between the two oral formulations (capsules vs oral solution) and to optimise dosing regimens in these patients. METHODS: All paediatric patients with cystic fibrosis or patients undergoing BMT at The Royal Children’s Hospital, Brisbane, QLD, Australia, who were prescribed oral itraconazole for the treatment of allergic bronchopulmonary aspergillosis (cystic fibrosis patients) or for prophylaxis of any fungal infection (BMT patients) were eligible for the study. Blood samples were taken from the recruited patients as per an empirical sampling design either during hospitalisation or during outpatient clinic visits. Itraconazole and hydroxy-itraconazole plasma concentrations were determined by a validated high-performance liquid chromatography assay with fluorometric detection. A nonlinear mixed-effect modelling approach using the NONMEM software to simultaneously describe the pharmacokinetics of itraconazole and its metabolite. RESULTS: A one-compartment model with first-order absorption described the itraconazole data, and the metabolism of the parent drug to hydroxy-itraconazole was described by a first-order rate constant. The metabolite data also showed one-compartment characteristics with linear elimination. For itraconazole the apparent clearance (CLitraconazole) was 35.5 L/hour, the apparent volume of distribution (Vd(itraconazole)) was 672L, the absorption rate constant for the capsule formulation was 0.0901 h and for the oral solution formulation was 0.96 h. The lag time was estimated to be 19.1 minutes and the relative bioavailability between capsules and oral solution (Frel) was 0.55. For the metabolite, volume of distribution, Vm/(F · fm), and clearance, CL/(F · fm), were 10.6L and 5.28 L/h, respectively. The influence of total bodyweight was significant, added as a covariate on CLitraconazole/F and Vd(itraconazole)/F (standardised to a 70kg person) using allometric three-quarter power scaling on CLitraconazole/F, which therefore reflected adult values. The unexplained between-subject variability (coefficient of variation %) was 68.7%, 75.8%, 73.4% and 61.1% for CLitraconazole/F, Vd(itraconazole)/F, CLm/(F · fm) and Frel, respectively. The correlation between random effects of CLitraconazole and Vd(itraconazole) was 0.69. CONCLUSION: The developed population pharmacokinetic model adequately described  the  pharmacokinetics  of  itraconazole and its active metabolite, hydroxy-itraconazole, in paediatric patients with either cystic fibrosis or undergoing BMT. More appropriate dosing schedules have been developed for the oral solution and the capsules to secure a minimum therapeutic trough plasma concentration of 0.5 mg/L for these patients. [ABSTRACT FROM AUTHOR]

Objective: To develop guidelines for the management of QT prolongation after citalopram overdose, including decontamination with single-dose activated charcoal (SDAC) and cardiac monitoring.Design: Simulation study using a previously developed pharmacokinetic-pharmacodynamic (PKPD) model which predicted the time-course of QT prolongation and the effect of citalopram dose and use of SDAC on QT prolongation.Main Measures and Results: The previously developed PKPD model was used to address the following in patients following citalopram overdose: (1) Above what dose should patients be decontaminated? (2) Above what dose should patients have cardiac monitoring? (3) For what period of time should patients be monitored? The primary outcome was QT,RR combinations above an abnormal threshold as a surrogate predictor of torsades de pointes. Simulations were performed using MATLAB for an overdose patient with typical demographics: 30-year-old female with a heart rate of 79 bpm taking citalopram therapeutically. The simulations showed: (1) There was significant benefit associated with the administration of SDAC to patients following citalopram overdose ingesting >600 mg; (2) With citalopram overdoses >1,000mg it was advisable to give SDAC and cardiac monitor the patient; (3) The risk of developing future abnormal QT,RR combinations was less than 1% in patients with normal QT,RR combinations up to 13 h post-dose, so the minimum monitoring time for citalopram overdoses >1,000mg should be 13h. Recommended dose levels for intervention should be lowered in older patients and patients with tachycardia, while men are less sensitive to QT prolongation.Conclusions: Guidelines for the management of QT prolongation after citalopram overdose were developed. We believe the model will help clinicians to decide which patients to decontaminate and monitor. [ABSTRACT FROM AUTHOR]

Purpose: Based on the previously published hollow-fibre assay mainly used for early in vivo anticancer drug screening, we wanted to develop an extended hollow-fibre model in which antitumour activity, haematological toxicity and pharmacokinetics could be studied in the same animal.Method: The breast cancer cell lines MDA-MB-231 and MCF-7 were cultured in semipermeable hollow fibres. The fibres were implanted subcutaneously into immunocompetent male Sprague Dawley rats, and the rats were treated with 5-fluorouracil (5-FU, 125 mg/kg), epirubicin (EPI, 10 mg/kg) or cyclophosphamide (CP, 120 mg/kg) intraperitoneally, the new cyanoguanidine CHS 828 (375 mg/kg or 75 mg/kg x 5) orally, or vehicle only. After 6 days the fibres were retrieved and the cell density was evaluated. Haematological parameters were monitored and two to four samples per animal were drawn to determine the pharmacokinetic parameters in NONMEM.Results: Drug treatment had generally low effects on the tumour cells. Of the standard drugs (5-FU, EPI and CP), only CP exerted a statistically significant antiproliferative effect. CHS 828 had only a minor effect as a single dose, but divided into five daily doses had a pronounced effect on both cell lines. 5-FU, EPI and CP all caused a marked decrease in leucocytes, platelets and haemoglobin, while CHS 828 did not seem to affect these parameters. The pharmacokinetics of 5-FU and EPI were in accordance with previously established pharmacokinetic models. The pharmacokinetics of CP and CHS 828 were both described by one-compartment models.Conclusions: This study illustrates the possibility of measuring antitumour effect, haematological toxicity and pharmacokinetics in the same animal using the hollow-fibre model. [ABSTRACT FROM AUTHOR]

A letter to the editor is presented in response to an article about application of pharmacokinetic-pharmacodynamic modeling in management of QT abnormalities after citalopram overdose.

A response by Isbister, Duffull and Friberg to a letter to the editor about their article on the modeling of cardiac toxicity for venlafaxine overdose in the November 2006 issue is presented.

Purpose: Model-based methods can predict pediatric exposure and support initial dose selection. The aim of this study was to evaluate the performance of allometric scaling of population pharmacokinetic (popPK) versus physiologically based pharmacokinetic (PBPK) models in predicting the exposure of tyrosine kinase inhibitors (TKIs) for pediatric patients (≥ 2 years), based on adult data. The drugs imatinib, sunitinib and pazopanib were selected as case studies due to their complex PK profiles including high inter-patient variability, active metabolites, time-varying clearances and non-linear absorption.Pediatric concentration measurements and adult popPK models were derived from the literature. Adult PBPK models were generated in PK-Sim® using available physicochemical properties, calibrated to adult data when needed. PBPK and popPK models for the pediatric populations were translated from the models for adults and were used to simulate concentration-time profiles that were compared to the observed values.Ten pediatric datasets were collected from the literature. While both types of models captured the concentration-time profiles of imatinib, its active metabolite, sunitinib and pazopanib, the PBPK models underestimated sunitinib metabolite concentrations. In contrast, allometrically scaled popPK simulations accurately predicted all concentration-time profiles. Trough concentration (Ctrough) predictions from the popPK model fell within a 2-fold range for all compounds, while 3 out of 5 PBPK predictions exceeded this range for the imatinib and sunitinib metabolite concentrations.Based on the identified case studies it appears that allometric scaling of popPK models is better suited to predict exposure of TKIs in pediatric patients ≥ 2 years. This advantage may be attributed to the stable enzyme expression patterns from 2 years old onwards, which can be easily related to adult levels through allometric scaling. In some instances, both methods performed comparably. Understanding where discrepancies between the model methods arise, can further inform model development and ultimately support pediatric dose selection.Methods: Model-based methods can predict pediatric exposure and support initial dose selection. The aim of this study was to evaluate the performance of allometric scaling of population pharmacokinetic (popPK) versus physiologically based pharmacokinetic (PBPK) models in predicting the exposure of tyrosine kinase inhibitors (TKIs) for pediatric patients (≥ 2 years), based on adult data. The drugs imatinib, sunitinib and pazopanib were selected as case studies due to their complex PK profiles including high inter-patient variability, active metabolites, time-varying clearances and non-linear absorption.Pediatric concentration measurements and adult popPK models were derived from the literature. Adult PBPK models were generated in PK-Sim® using available physicochemical properties, calibrated to adult data when needed. PBPK and popPK models for the pediatric populations were translated from the models for adults and were used to simulate concentration-time profiles that were compared to the observed values.Ten pediatric datasets were collected from the literature. While both types of models captured the concentration-time profiles of imatinib, its active metabolite, sunitinib and pazopanib, the PBPK models underestimated sunitinib metabolite concentrations. In contrast, allometrically scaled popPK simulations accurately predicted all concentration-time profiles. Trough concentration (Ctrough) predictions from the popPK model fell within a 2-fold range for all compounds, while 3 out of 5 PBPK predictions exceeded this range for the imatinib and sunitinib metabolite concentrations.Based on the identified case studies it appears that allometric scaling of popPK models is better suited to predict exposure of TKIs in pediatric patients ≥ 2 years. This advantage may be attributed to the stable enzyme expression patterns from 2 years old onwards, which can be easily related to adult levels through allometric scaling. In some instances, both methods performed comparably. Understanding where discrepancies between the model methods arise, can further inform model development and ultimately support pediatric dose selection.Results: Model-based methods can predict pediatric exposure and support initial dose selection. The aim of this study was to evaluate the performance of allometric scaling of population pharmacokinetic (popPK) versus physiologically based pharmacokinetic (PBPK) models in predicting the exposure of tyrosine kinase inhibitors (TKIs) for pediatric patients (≥ 2 years), based on adult data. The drugs imatinib, sunitinib and pazopanib were selected as case studies due to their complex PK profiles including high inter-patient variability, active metabolites, time-varying clearances and non-linear absorption.Pediatric concentration measurements and adult popPK models were derived from the literature. Adult PBPK models were generated in PK-Sim® using available physicochemical properties, calibrated to adult data when needed. PBPK and popPK models for the pediatric populations were translated from the models for adults and were used to simulate concentration-time profiles that were compared to the observed values.Ten pediatric datasets were collected from the literature. While both types of models captured the concentration-time profiles of imatinib, its active metabolite, sunitinib and pazopanib, the PBPK models underestimated sunitinib metabolite concentrations. In contrast, allometrically scaled popPK simulations accurately predicted all concentration-time profiles. Trough concentration (Ctrough) predictions from the popPK model fell within a 2-fold range for all compounds, while 3 out of 5 PBPK predictions exceeded this range for the imatinib and sunitinib metabolite concentrations.Based on the identified case studies it appears that allometric scaling of popPK models is better suited to predict exposure of TKIs in pediatric patients ≥ 2 years. This advantage may be attributed to the stable enzyme expression patterns from 2 years old onwards, which can be easily related to adult levels through allometric scaling. In some instances, both methods performed comparably. Understanding where discrepancies between the model methods arise, can further inform model development and ultimately support pediatric dose selection.Conclusion: Model-based methods can predict pediatric exposure and support initial dose selection. The aim of this study was to evaluate the performance of allometric scaling of population pharmacokinetic (popPK) versus physiologically based pharmacokinetic (PBPK) models in predicting the exposure of tyrosine kinase inhibitors (TKIs) for pediatric patients (≥ 2 years), based on adult data. The drugs imatinib, sunitinib and pazopanib were selected as case studies due to their complex PK profiles including high inter-patient variability, active metabolites, time-varying clearances and non-linear absorption.Pediatric concentration measurements and adult popPK models were derived from the literature. Adult PBPK models were generated in PK-Sim® using available physicochemical properties, calibrated to adult data when needed. PBPK and popPK models for the pediatric populations were translated from the models for adults and were used to simulate concentration-time profiles that were compared to the observed values.Ten pediatric datasets were collected from the literature. While both types of models captured the concentration-time profiles of imatinib, its active metabolite, sunitinib and pazopanib, the PBPK models underestimated sunitinib metabolite concentrations. In contrast, allometrically scaled popPK simulations accurately predicted all concentration-time profiles. Trough concentration (Ctrough) predictions from the popPK model fell within a 2-fold range for all compounds, while 3 out of 5 PBPK predictions exceeded this range for the imatinib and sunitinib metabolite concentrations.Based on the identified case studies it appears that allometric scaling of popPK models is better suited to predict exposure of TKIs in pediatric patients ≥ 2 years. This advantage may be attributed to the stable enzyme expression patterns from 2 years old onwards, which can be easily related to adult levels through allometric scaling. In some instances, both methods performed comparably. Understanding where discrepancies between the model methods arise, can further inform model development and ultimately support pediatric dose selection. [ABSTRACT FROM AUTHOR]