Published population pharmacokinetic models of valproic acid in adult patients: a systematic review and external validation in a Chinese sample of inpatients with bipolar disorder

Prediction Trough Concentrations of Valproic Acid Among Chinese Adult Patients with Epilepsy Using Machine Learning Techniques

OBJECTIVE

This study aimed to establish an optimal model based on machine learning (ML) to predict Valproic acid (VPA) trough concentrations in Chinese adult epilepsy patients.

METHODS

A single-center retrospective study was carried out at the Jinshan Hospital affiliated with Fudan University from January 2022 to December 2023. A total of 102 VPA trough concentrations were split into a derivation cohort and a validation cohort at a ratio of 8:2. Thirteen ML algorithms were developed using 27 variables in the derivation cohort and were filtered by the lowest mean absolute error (MAE) value. In addition, feature selection was applied to optimize the model.

RESULTS

Ultimately, the extra tree algorithm was chosen to establish the personalized VPA trough concentration prediction model due to its best performance (MAE = 13.08). The SHapley Additive exPlanations (SHAP) plots were used to visualize and rank the importance of features, providing insights into how each feature influences the model's predictions. After feature selection, we found that the model with the top 9 variables [including daily dose, last dose, uric acid (UA), platelet (PLT), combination, gender, weight, albumin (ALB), aspartate aminotransferase (AST)] outperformed the model with 27 variables, with MAE of 6.82, RMSE of 9.62, R2 value of 0.720, relative accuracy (±20%) of 61.90%, and absolute accuracy (±20 mg/L) of 90.48%.

CONCLUSION

In conclusion, the trough concentration prediction model for VPA in Chinese adult epileptic patients based on the extra tree algorithm demonstrated strong predictive ability which is valuable for clinicians in medication guidance.

Fixed parameters in the population pharmacokinetic modeling of valproic acid might not be suitable: external validation in Chinese adults with epilepsy or after neurosurgery

PURPOSE

This study aims to assess the predictive performance of published valproic acid (VPA) population pharmacokinetic (PPK) models using an external data set in Chinese adults with epilepsy or after neurosurgery.

METHODS

A total of 384 concentrations from 290 Chinese adults with epilepsy or after neurosurgery were used for external validation. Data on published VPA PPK models were extracted from the literature. Prediction-based diagnostics (such as F20 and F30), simulation-based diagnostics, and Bayesian forecasting were used to evaluate the predictability of models.

RESULTS

The results of prediction-based diagnostics of all models were unsatisfactory. Models B, F, and H showed the best prediction performance in simulation-based diagnostics and Bayesian forecasting, demonstrating superior precision and accuracy. Bayesian forecasting demonstrated significant improvements in the model predictability.

CONCLUSION

The published PPK models showed extensive variation in predictive performance for extrapolation among Chinese adults with epilepsy or after neurosurgery patients. Fixed parameters of Vd and Ka in the PPK modeling of VPA might be the reason for the unsatisfied predictive performance. Bayesian forecasting significantly improved model predictability and may help to individualize VPA dosing.

A Cross-sectional Comparative Analysis of Eleven Population Pharmacokinetic Models for Docetaxel in Chinese Breast Cancer Patients

OBJECTIVE

Various population pharmacokinetic (PPK) models have been established to help determine the appropriate dosage of docetaxel, however, no clear consensus on optimal dosing has been achieved. The purpose of this study is to perform an external evaluation of published models in order to test their predictive performance, and to find an appropriate PPK model for Chinese breast cancer patients.

METHODS

A systematic literature search of docetaxel PPK models was performed using PubMed, Web of Science, China National Knowledge Infrastructure, and WanFang databases. The predictive performance of eleven identified models was evaluated using prediction-based and simulation-based diagnostics on an independent dataset (112 docetaxel concentrations from 56 breast cancer patients). The -2×log (likelihood) and Akaike information criterion were also calculated to evaluate model fit.

RESULTS

The median prediction error of eight of the eleven models was less than 10%. The model fitting results showed that the three-compartment model of Bruno et al. had the best prediction performance and that the three compartment model of Wang et al. had the best simulation effect. Furthermore, although the covariates that significantly affect PK parameters were different between them, seven models demonstrated that docetaxel PK parameters were influenced by liver function.

CONCLUSIONS

Three compartment PPK models may be predictive of optimal docetaxel dosage for Chinese breast cancer patients. However, for patients with impaired liver function, the choice of which model to use to predict the blood concentration of docetaxel still requires great care.

Effects of Sex Differences and Combined Use of Clozapine on Initial Dosage Optimization of Valproic Acid in Patients with Bipolar Disorder

BACKGROUND

Due to the narrow therapeutic window and large pharmacokinetic variation of valproic acid (VPA), it is difficult to make an optimal dosage regimen. The present study aims to optimize the initial dosage of VPA in patients with bipolar disorder.

METHODS

A total of 126 patients with bipolar disorder treated by VPA were included to construct the VPA population pharmacokinetic model retrospectively. Sex differences and combined use of clozapine were found to significantly affect VPA clearance in patients with bipolar disorder. The initial dosage of VPA was further optimized in male patients without the combined use of clozapine, female patients without the combined use of clozapine, male patients with the combined use of clozapine, and female patients with the combined use of clozapine, respectively.

RESULTS

The CL/F and V/F of VPA in patients with bipolar disorder were 11.3 L/h and 36.4 L, respectively. It was found that sex differences and combined use of clozapine significantly affected VPA clearance in patients with bipolar disorder. At the same weight, the VPA clearance rates were 1.134, 1, 1.276884, and 1.126 in male patients without the combined use of clozapine, female patients without the combined use of clozapine, male patients with the combined use of clozapine, and female patients with the combined use of clozapine, respectively. This study further optimized the initial dosage of VPA in male patients without the combined use of clozapine, female patients without the combined use of clozapine, male patients with the combined use of clozapine, and female patients with the combined use of clozapine, respectively.

CONCLUSION

This study is the first to investigate the initial dosage optimization of VPA in patients with bipolar disorder based on sex differences and the combined use of clozapine. Male patients had higher clearance, and the recommended initial dose decreased with increasing weight, providing a reference for the precision drug use of VPA in clinical patients with bipolar disorder.

Modeling the protein binding non-linearity in population pharmacokinetic model of valproic acid in children with epilepsy: a systematic evaluation study

Background: Several studies have investigated the population pharmacokinetics (popPK) of valproic acid (VPA) in children with epilepsy. However, the predictive performance of these models in the extrapolation to other clinical environments has not been studied. Hence, this study evaluated the predictive abilities of pediatric popPK models of VPA and identified the potential effects of protein binding modeling strategies. Methods: A dataset of 255 trough concentrations in 202 children with epilepsy was analyzed to assess the predictive performance of qualified models, following literature review. The evaluation of external predictive ability was conducted by prediction- and simulation-based diagnostics as well as Bayesian forecasting. Furthermore, five popPK models with different protein binding modeling strategies were developed to investigate the discrepancy among the one-binding site model, Langmuir equation, dose-dependent maximum effect model, linear non-saturable binding equation and the simple exponent model on model predictive ability. Results: Ten popPK models were identified in the literature. Co-medication, body weight, daily dose, and age were the four most commonly involved covariates influencing VPA clearance. The model proposed by Serrano et al. showed the best performance with a median prediction error (MDPE) of 1.40%, median absolute prediction error (MAPE) of 17.38%, and percentages of PE within 20% (F20, 55.69%) and 30% (F30, 76.47%). However, all models performed inadequately in terms of the simulation-based normalized prediction distribution error, indicating unsatisfactory normality. Bayesian forecasting enhanced predictive performance, as prior observations were available. More prior observations are needed for model predictability to reach a stable state. The linear non-saturable binding equation had a higher predictive value than other protein binding models. Conclusion: The predictive abilities of most popPK models of VPA in children with epilepsy were unsatisfactory. The linear non-saturable binding equation is more suitable for modeling non-linearity. Moreover, Bayesian forecasting with prior observations improved model fitness.

Population Pharmacokinetics and Exposure-Safety of Lipophilic Conjugates Prodrug DP-VPA in Healthy Chinese Subjects for Dose Regime Exploring

Phospholipid-valproic acid (DP-VPA)is a prodrug for treating epilepsy. The present study explored the pharmacokinetics (PK) and exposure safety of DP-VPA to provide a basis for future studies exploring the safe dosage and therapeutic strategies for epilepsy. The study included a randomized placebo-controlled dose-escalation tolerance evaluation trial and a randomized triple crossover food-effect trial in healthy Chinese volunteers. A population pharmacokinetic (PopPK) model was established to analyze the PK of DP-VPA and active metabolite VPA. The exposure safety was assessed with the adverse drug reaction (ADR) in CNS. The PopPK of DP-VPA and metabolite VPA fitted a two-compartment model coupling one-compartment with Michaelis-Menten metabolite kinetics and first-order elimination. The absorption processes after single oral administration of DP-VPA tablet demonstrated nonlinear characteristics, including 0-order kinetic phase and time-dependent phase fitting Weibull distribution. The final model indicated that the DP-VPA PK was significantly affected by dosage and food. The exposure-safety relationship demonstrated a generalized linear regression; mild/moderate ADRs occurred in some subjects with 600 mg and all subjects with 1500 mg of DP-VPA, and no severe ADRs were reported up to 2400 mg. In conclusion, the study established a PopPK model describing the processing of DP-VPA and VPA in healthy Chinese subjects. DP-VPA showed good tolerance after a single dose of 600-2400 mg with nonlinear PK and was affected by dosage and food. Based on the association between neurological ADRs and higher exposure to DP-VPA by exposure-safety analysis, 900-1200 mg was recommended for subsequent study of safety and clinical effectiveness.

Machine learning advances the integration of covariates in population pharmacokinetic models: Valproic acid as an example

Background and Aim: Many studies associated with the combination of machine learning (ML) and pharmacometrics have appeared in recent years. ML can be used as an initial step for fast screening of covariates in population pharmacokinetic (popPK) models. The present study aimed to integrate covariates derived from different popPK models using ML.

Methods: Two published popPK models of valproic acid (VPA) in Chinese epileptic patients were used, where the population parameters were influenced by some covariates. Based on the covariates and a one-compartment model that describes the pharmacokinetics of VPA, a dataset was constructed using Monte Carlo simulation, to develop an XGBoost model to estimate the steady-state concentrations (Css) of VPA. We utilized SHapley Additive exPlanation (SHAP) values to interpret the prediction model, and calculated estimates of VPA exposure in four assumed scenarios involving different combinations of CYP2C19 genotypes and co-administered antiepileptic drugs. To develop an easy-to-use model in the clinic, we built a simplified model by using CYP2C19 genotypes and some noninvasive clinical parameters, and omitting several features that were infrequently measured or whose clinically available values were inaccurate, and verified it on our independent external dataset.

Results: After data preprocessing, the finally generated combined dataset was divided into a derivation cohort and a validation cohort (8:2). The XGBoost model was developed in the derivation cohort and yielded excellent performance in the validation cohort with a mean absolute error of 2.4 mg/L, root-mean-squared error of 3.3 mg/L, mean relative error of 0%, and percentages within ±20% of actual values of 98.85%. The SHAP analysis revealed that daily dose, time, CYP2C19*2 and/or *3 variants, albumin, body weight, single dose, and CYP2C19*1*1 genotype were the top seven confounding factors influencing the Css of VPA. Under the simulated dosage regimen of 500 mg/bid, the VPA exposure in patients who had CYP2C19*2 and/or *3 variants and no carbamazepine, phenytoin, or phenobarbital treatment, was approximately 1.74-fold compared to those with CYP2C19*1/*1 genotype and co-administered carbamazepine + phenytoin + phenobarbital. The feasibility of the simplified model was fully illustrated by its performance in our external dataset.

Conclusion: This study highlighted the bridging role of ML in big data and pharmacometrics, by integrating covariates derived from different popPK models.