Background: Ramucirumab is an anti-VEGFR2 monoclonal antibody used in gastric and gastroesophageal junction cancers. A pharmacokinetic-pharmacodynamic (PK-PD) relationship has been highlighted in these cancers (Kim 2018, Tabernero 2017) as well as non-small cell lung cancer (Akagi 2023), using Kaplan-Meier and Cox proportional hazards analyses. However, these approaches are subject to confounding factors as disease activity can also influence drug levels (Lobet 2023).

Aims: This study aimed to assess the relationship between ramucirumab pharmacokinetics, target occupancy and overall survival (OS) using joint population target-mediated drug disposition (TMDD) and parametric survival modeling.

Methods: Data were obtained from a phase II clinical trial (SOCRATE, EudraCT 2024-513515-27-00), including 70 patients receiving ramucirumab every 2 weeks up to progression or grade 3 toxicity. Ramucirumab concentrations were measured using an ELISA method (Desvignes 2021). Ramucirumab PK and hazard function for survival were described using a TMDD and a log-logistic model, respectively.

Results: Estimated parameters were volumes of distribution (V1=2.03L, V2=2.89L), clearances (CL=0.23L/d, Q=2.01L/d), baseline target level (R0=30.2nM), immune complex elimination rate (kout=0.91d-1) and dissociation constant (KSS=0.47nM). Overall survival significantly increased with ramucirumab exposure, using either concentrations or target occupancy in the model (Δ-2LL = -42.46 and -45.10, respectively).

Conclusions: This study is the first to simultaneously depict the relationship between ramucirumab PK, target occupancy and OS, and demonstrated target-mediated nonlinear PK and an association between ramucirumab exposure and clinical efficacy. A nonlinear PK was previously suspected but never described (O’Brien 2017). These results show that TDM of ramucirumab could be promising.

Key words: Ramucirumab, oncology, pharmacokinetics

Background: Isoniazid (INH), a first-line drug to treat tuberculosis (TB) for more than 50 years. However, large inter-individual variability was found in its pharmacokinetics, and effects of non-adherence to INH treatment and optimal remedial dosing remains unclear.

Aims: This study aimed to develop a population pharmacokinetics (PPK) model of INH in Chinese patients with TB, to provide model-informed precision dosing and explore appropriate remedial dosing regimens for non-adherent patients.

Methods: 1012 INH therapeutic drug monitoring (TDM) results from 736 TB patients were included. A nonlinear mixed-effects modeling was used to analyze the PPK of INH. Using Monte Carlo simulations to determine optimal dosage regimens and design remedial dosing regimens.

Results: A two-compartmental model, including first-order absorption and elimination with allometric scaling, was found to best describe the PK characteristics of INH. A mixture model was used to characterize dual rates of INH elimination. Estimates of apparent clearance in fast and slow eliminators were 28.0 and 11.2 L/h, respectively. Monte Carlo simulations determined optimal dosage regimens for slow and fast eliminators with different body weight. For remedial dosing regimens, the missed dose should be taken as soon as possible when the delay does not exceed 12 h, and an additional dose is not needed. On delaying a INH dose exceed 12 h, only need to take the next single dose normally.

Conclusions: TDM and PPK modeling and simulation provide valid evidence on the precision dosing and remedial dosing regimen of INH.

Keywords: Isoniazid; PPK; model-based precision dosing; remedial dosing; tuberculosis; TDM

Background: Cefuroxime is a widely prescribed beta-lactam antibiotic, particularly in pediatric cardiac and medical-surgical intensive care units.

Aims: To describe intravenous cefuroxime disposition in critically ill pediatric patients. Moreover, target attainment of currently applied dosing regimens were evaluated and suggestions for improvement of these dosing regimens were provided.

Methods: This pooled population pharmacokinetic (popPK) study combined two datasets for analysis in NONMEM v7.5. Cefuroxime exposure was simulated using the percentage of time above the minimum inhibitory concentration (%T>MIC). PK profiles of different dosage regimens (intermittent/continuous) were simulated with the developed popPK model. A target of 90% of cases 100%T>(4x)MIC8mg/L was used.

Results: The cohort consisted of 45 pediatric patient’s with a median (range) age of 391 days (0-6131), and bodyweight of 9.0 kg (2.8-70.0). A two-compartment popPK model with first-order elimination and allometric scaling best described cefuroxime disposition. Postnatal age and creatinine clearance were the best descriptive covariates for cefuroxime clearance. Simulations evaluating the current cefuroxime dosing regimens stratified for eGFR levels illustrated moderate (eGFR <30 and 30-80 mL/min./1.73m2) and poor (eGFR 80-120 and >120 mL/min/1.73m2) cefuroxime target attainment across the entire age range. Alternative dosing regimens, including four times daily schedules and continuous infusion, improved target attainment, particularly in older children and those with augmented renal clearance.

Conclusions: While applying the current dosing regimens, underexposure due to enhance renal functions is possible. Future research should focus on individualized dosing strategies to optimize cefuroxime exposure and efficacy in pediatric populations.

Key Words: Cefuroxime, Pediatrics, Population pharmacokinetics, PICU

Background: PopPK models are drug and population specific and combining them for model informed personalized dosing (MIPD) has thus far been limited to averaging dose recommendations from different models made for one drug in similar populations.

Aims: Use reinforcement learning (RL) to combine diverse popPK models for optimized dosing

Methods: We developed non-parametric population models for theophylline in adults and voriconazole in children and adult and from these models simulated 50 patients and concentration-time profiles for each drug after standard dosing, splitting the data into training (60%) and validation (40%) subsets. With one simulated concentration and set of covariates (drug, weight, age for voriconazole, drug for theophylline), we trained a deep RL agent to predict the optimal loading and maintenance doses to achieve specified concentration targets within a therapeutic window.

Results: In the validation data, the agent found loading and maintenance doses that achieved therapeutic concentrations in 90.5% of episodes overall. Specifically, therapeutic concentrations were achieved in 84% of episodes for voriconazole and 98% for theophylline.

Conclusions: We present a completely novel approach to MIPD that learns across drugs and populations, with subsequent creation of a single, generalized model where the drug itself becomes a covariate. Our results suggest that the method is feasible from typical popPK models and can be extended to more drugs and populations.

Keywords: Dose optimization, patient individualization, reinforcement learning.

Background: Invasive candidiasis, particularly candidaemia, is a life-threatening complication in critically ill patients. Fluconazole is the recommended step-down therapy from echinocandins for fluconazole-susceptible Candida spp. However, standard fluconazole dosing does not achieve adequate exposure in critically ill patients.

Aims: This study aimed to identify factors that impact fluconazole target attainment and provide a dosing regimen ensuring adequate pharmacokinetic–pharmacodynamic (PKPD) target attainment in critically ill patients.

Methods: An individual patient data meta-analysis was conducted, combining fluconazole concentration data from eight published studies. We developed a population pharmacokinetics (popPK) model and used multiple imputation to handle missing covariate data. We performed Monte Carlo simulations to identify a dosing strategy with at least a 90% probability of PKPD target attainment (PTA) in every patient.

Results: Data from 177 critically ill patients were included. A two-compartment popPK model with linear elimination best described the data. Continuous renal replacement therapy (CRRT) status, estimated glomerular filtration rate, and total body weight (BW) were statistically significant covariates. However, with standard dosing, only CRRT status and BW were clinically relevant with the PTA dropping below 90% for all patients on CRRT, and for patients not undergoing CRRT weighing more than 60 kg. An optimised dosing regimen considering the patient’s CRRT status and BW is proposed.

Conclusion: We have developed a fluconazole dosing regimen that could achieve adequate population-wide PKPD target attainment in critically ill patients. Confirmation that the proposed dosing regimen achieves optimal fluconazole exposure is required.

Keywords: Population pharmacokinetics, Monte Carlo simulation, Individual patient data meta-analysis

Background: Sepsis affects approximately 8% of pediatric intensive care unit (PICU) admissions in high-income countries. Ceftriaxone, a broad-spectrum beta-lactam antibiotic, is widely used for treating severe infections and bacterial meningitis in children. Despite its frequent use, limited studies address the population pharmacokinetic (popPK) of ceftriaxone in pediatrics.

Aims: To externally evaluate pediatric ceftriaxone popPK model peformance, enabling selection of the model best suited to this population.

Methods: This study used data from the EXPAT Kids study, a prospective PK/PD study. Included popPK models were implemented in NONMEM, with diagnostic goodness-of-fit and visual predictive check analyses performed to assess model accuracy. Predictive performance was evaluated using the relative prediction error, relative root mean square error, and mean (absolute) percentage error.

Results: The predictive performance of the evaluated models varied widely. Included models showed only modest performance and generally seem to overpredict ceftriaxone concentrations. Unbound ceftriaxone popPK models did not perform adequately. None of the models met all the predefined thresholds for accuracy and precision.

Conclusion: Our external dataset comprised high ceftriaxone trough concentrations, indicating re-evaluation of current ceftriaxone dosing regimens to minimize the risk of overdosing and prevent toxicity. Future research should focus on the fine dosing balance for ceftriaxone (especially in meningitis patients) considering adequate exposure but preventing high trough concentrations. Model informed precision dosing may enhance the use of the optimal individual dosage for critically ill children. However, our findings highlight the importance of externally evaluating ceftriaxone popPK models in the PICU population.

Key words: Ceftriaxone, Population Pharmacokinetics, Pediatrics

Background: Therapeutic drug monitoring (TDM) is essential for optimizing clozapine treatment. Current guidelines recommend blood sampling 12 (±2) hours after intake for once-daily dosing. However, this timing is primarily based on twice-daily dosing regimens and its suitability for once daily dosing remains uncertain.

Aims: This study aims to evaluate the validity of current TDM practice for once-daily clozapine dosing.

Methods: This retrospective, single-center study includes patients with treatment-resistant schizophrenia spectrum disorders on stable once-daily clozapine, with TDM performed as part of standard care. Bayesian analysis was used to predict individual concentration-time profiles. Predicted clozapine concentrations at 10, 11, 13, and 14 hours post-intake were compared to the 12-hour reference value, with deviations exceeding 20% considered clinically significant.

Results: Data were collected from seven patients (86% male, 14% non-smoker) with a clozapine dose of 246±109 mg (mean±SD). A total of 22 clozapine samples were analyzed, of which two outliers were excluded. The mean clozapine concentration at t=12 was 608±255 μg/L. The mean percentage differences from the 12-hour reference value were 6.4±2.0%, 3.1±1.0%, -2.9±1.0%, and -5.7±1.8% at t=10, 11, 13 and 14 hours, respectively.

Conclusions: These preliminary findings suggest that the current sampling window of 12 (±2) hours post-intake is appropriate for the clinical interpretation of clozapine levels in once-daily dosing. However, given the limited dataset and absence of the consideration of inter-occasion PK variability (IOV), additional data and further analysis is required to confirm the sampling window for once-daily clozapine dosing.

Keywords: Clozapine, therapeutic drug monitoring, population pharmacokinetics, sampling time, schizophrenia