A phase I/II study of nintedanib and capecitabine for refractory metastatic colorectal cancer.

BACKGROUND: Nintedanib is a tyrosine kinase inhibitor with efficacy in bevacizumab-resistant colorectal cancer models. This phase I/II study evaluated the recommended phase II dose and efficacy of nintedanib and capecitabine in refractory metastatic colorectal cancer. METHODS: Key eligibility criteria included refractory metastatic colorectal cancer and ECOG performance status of 1 or lower. The primary endpoint was 18-week progression-free survival (PFS). A 1-sided binomial test (at α = .1) compared the observed 18-week PFS with a historic control of .25. RESULTS: Forty-two patients were enrolled, including 39 at the recommended phase II dose. The recommended phase II dose was established to be nintedanib 200 mg by mouth twice daily and capecitabine 1000 mg/m2 by mouth twice daily. The protocol was evaluated for efficacy in 36 patients. The 18-week PFS was 42% (15/36 patients; P = .0209). Median PFS was 3.4 mo. Median overall survival was 8.9 mo. Sixteen (44%) patients experienced a grade 3/4 adverse event, most commonly fatigue (8%), palmoplantar erythrodysesthesia (8%), aspartate aminotransferase elevation (6%), asthenia (6%), pulmonary embolus (6%), and dehydration (6%). Osteopontin levels at cycle 1, day 1 and cycle 3, day 1 as well as ΔCCL2 levels correlated to disease control at 18 weeks. CONCLUSIONS: The combination of nintedanib and capecitabine is well tolerated. Clinical efficacy appears to be superior to regorafenib or tipiracil hydrochloride monotherapy. Further investigation of similar combinations is warranted. CLINICALTRIALS.GOV IDENTIFIER: NCT02393755.

Evaluation of ChatGPT and Gemini large language models for pharmacometrics with NONMEM.

To assess ChatGPT 4.0 (ChatGPT) and Gemini Ultra 1.0 (Gemini) large language models on NONMEM coding tasks relevant to pharmacometrics and clinical pharmacology. ChatGPT and Gemini were assessed on tasks mimicking real-world applications of NONMEM. The tasks ranged from providing a curriculum for learning NONMEM, an overview of NONMEM code structure to generating code. Prompts in lay language to elicit NONMEM code for a linear pharmacokinetic (PK) model with oral administration and a more complex model with two parallel first-order absorption mechanisms were investigated. Reproducibility and the impact of "temperature" hyperparameter settings were assessed. The code was reviewed by two NONMEM experts. ChatGPT and Gemini provided NONMEM curriculum structures combining foundational knowledge with advanced concepts (e.g., covariate modeling and Bayesian approaches) and practical skills including NONMEM code structure and syntax. ChatGPT provided an informative summary of the NONMEM control stream structure and outlined the key NONMEM Translator (NM-TRAN) records needed. ChatGPT and Gemini were able to generate code blocks for the NONMEM control stream from the lay language prompts for the two coding tasks. The control streams contained focal structural and syntax errors that required revision before they could be executed without errors and warnings. The code output from ChatGPT and Gemini was not reproducible, and varying the temperature hyperparameter did not reduce the errors and omissions substantively. Large language models may be useful in pharmacometrics for efficiently generating an initial coding template for modeling projects. However, the output can contain errors and omissions that require correction.

Model-Based Prediction of Irinotecan-Induced Grade 4 Neutropenia in Cancer Patients: Influence of Incorporating Germline Genetic Factors in the Model.

Neutropenia is the major dose-limiting toxicity of irinotecan-based therapy. The objective of this study was to assess whether inclusion of germline genetic variants into a population pharmacokinetic/pharmacodynamic model can improve prediction of irinotecan-induced grade 4 neutropenia and identify novel variants of clinical value. A semimechanistic population pharmacokinetic/pharmacodynamic model was used to predict neutrophil response over time in 197 patients receiving irinotecan. Covariate analysis was performed for demographic/clinical factors and 4,781 genetic variants in 84 drug response- and toxicity-related genes to identify covariates associated with neutrophil response. We evaluated the predictive value of the model for grade 4 neutropenia reflecting different clinical scenarios of available data on identified demographic/clinical covariates, baseline and post-treatment absolute neutrophil counts (ANCs), individual pharmacokinetics, and germline genetic variation. Adding 8 genetic identified covariates (rs10929302 (UGT1A1), rs1042482 (DPYD), rs2859101 (HLA-DQB3), rs61754806 (NR3C1), rs9266271 (HLA-B), rs7294 (VKORC1), rs1051713 (ALOX5), and ABCB1 rare variant burden) to a model using only baseline ANCs improved prediction of irinotecan-induced grade 4 neutropenia from area under the receiver operating characteristic curve (AUC-ROC) of 50-64% (95% confidence interval (CI), 54-74%). Individual pharmacokinetics further improved the prediction to 74% (95% CI, 64-84%). When weekly ANC was available, the identified covariates and individual pharmacokinetics yielded no additional contribution to the prediction. The model including only ANCs at baseline and at week 1 achieved an AUC-ROC of 78% (95% CI, 69-88%). Germline DNA genetic variants may contribute to the prediction of irinotecan-induced grade 4 neutropenia when incorporated into a population pharmacokinetic/pharmacodynamic model. This approach is generalizable to drugs that induce neutropenia and ultimately allows for personalized intervention to enhance patient safety.

Unraveling the Influence of Age, IQ, Education, and Negative Symptoms on Neurocognitive Performance in Schizophrenia: A Conditional Inference Tree Analysis.

INTRODUCTION: The complex nature of neurocognitive impairment in schizophrenia has been discussed in light of the mixed effects of antipsychotic drugs, psychotic symptoms, dopamine D2 receptor blockade, and intelligence quotient (IQ). These factors have not been thoroughly examined before. METHODS: This study conducted a comprehensive re-analysis of the CATIE data using machine learning techniques, in particular Conditional Inference Tree (CTREE) analysis, to investigate associations between neurocognitive functions and moderating factors such as estimated trough dopamine D2 receptor blockade with risperidone, olanzapine, or ziprasidone, Positive and Negative Syndrome Scale (PANSS), and baseline IQ in 573 patients with schizophrenia. RESULTS: The study reveals that IQ, age, and education consistently emerge as significant predictors across all neurocognitive domains. Furthermore, higher severity of PANSS-negative symptoms was associated with lower cognitive performance scores in several domains. CTREE analysis, in combination with a genetic algorithm approach, has been identified as particularly insightful for illustrating complex interactions between variables. Lower neurocognitive function was associated with factors such as age>52 years, IQ<94/95,<12/13 education years, and more pronounced negative symptoms (score<26). CONCLUSIONS: These findings emphasize the multifaceted nature of neurocognitive functioning in patients with schizophrenia, with the PANSS-negative score being an important predictor. This gives rise to a role in addressing negative symptoms as a therapeutic objective for enhancing cognitive impairments in these patients. Further research must examine nonlinear relationships among various moderating factors identified in this work, especially the role of D2 occupancy.

pyDarwin: A Machine Learning Enhanced Automated Nonlinear Mixed-Effect Model Selection Toolbox.

pyDarwin is an open-source Python package for nonlinear mixed-effect model selection. pyDarwin combines machine-learning algorithms and NONMEM to perform a global search for the optimal model in a user-defined model search space. Compared with traditional stepwise search, pyDarwin provides an efficient platform for conducting an objective, robust, less labor-intensive model selection process without compromising model interpretability. In this tutorial, we will begin by introducing the essential components and concepts within the package. Subsequently, we will provide an overview of the pyDarwin modeling workflow and the necessary files needed for model selection. To illustrate the entire process, we will conclude with an example utilizing quetiapine clinical data.

Modelling the progression of illicit substance use patterns from real-world evidence.

AIMS: To investigate an innovative pharmacometrics approach that addresses the challenges of using real-world evidence to model the progression of illicit substance use. METHODS: The modelling strategy analysed real-world data from the National Longitudinal Study of Adolescent to Adult Health (AddHealth) survey using survival analyses and differential equations. Respondents were categorized into drug-naïve, active users and nonusers. The transitions between categories were modelled using interval-censored parametric survival analysis. The resulting hazard rate functions were used as time-dependent rate constants in a differential equation system. Covariate models for sex and depression status were assessed. RESULTS: AddHealth enrolled 6504 American teenagers (median age 16 years, range 11-21 years); this cohort was followed with five interviews over a 22-year period; the median age at the last interview was 38 years (range 34-45 years). The percentages of illicit drug users at Interviews 1-5 were 7.7%, 5.9%, 15.8%, 21.4% and 0.98%, respectively. The generalized gamma distribution emerged as the preferred model for the survival functions for transitions between categories. Age-dependent prevalence was obtained from the differential equation system. Active drug use was more prevalent in males, increased in adolescence and college years, peaked at 24 years, and decreased to low levels by 35 years. Depression, which was more frequent in females, increased the drug-naïve-active user transition rates but not the active user-nonuser and nonuser-active user transition rates. The evidence did not support an interaction between sex and depression. CONCLUSIONS: The model provided a satisfactory approximation for the age-dependent progression of illicit substance use from preadolescence to early middle age.

Clinical trial simulation to evaluate tenofovir disoproxil fumarate/emtricitabine HIV pre-exposure prophylaxis dosing during pregnancy.

Objective: To evaluate upward-adjustment of tenofovir disoproxil fumarate (TDF)/emtricitabine (FTC) pre-exposure prophylaxis (PrEP) dosing during pregnancy in order to maintain target plasma concentrations associated with HIV protection. Design: Population pharmacokinetic (PK) modeling and clinical trial simulation (CTS). Material and methods: We developed population pharmacokinetic models for TFV and FTC using data from the Partners Demonstration Project and a PK study of TDF/FTC among cisgender women by Coleman et al., and performed an in-silico simulation. Pregnancy-trimester was identified as a significant covariate on apparent clearance in the optimized final model. We simulated 1,000 pregnant individuals starting standard daily oral TDF/FTC (300 mg/200 mg) prior to pregnancy. Upon becoming pregnant, simulated patients were split into two study arms: one continuing standard-dose and the other receiving double standard-dose throughout pregnancy. Results: Standard-dose trough TFV concentrations were significantly lower in pregnancy compared to pre-pregnancy, with 34.0%, 43.8%, and 65.1% of trough plasma concentrations below the lower bound of expected trough concentrations presumed to be the protective threshold in the 1st, 2nd, and 3rd trimesters, respectively. By comparison, in the simulated double-dose group, 10.7%, 14.4%, and 27.8% of trough concentrations fell below the estimated protective thresholds in the 1st, 2nd, and 3rd trimesters, respectively. The FTC trough plasma concentration during pregnancy was also lower than pre-pregnancy, with 45.2% of the steady-state trough concentrations below the estimated protective trough concentrations of FTC. In the pregnancy-adjusted double-dose group, 24.1% of trough plasma concentrations were lower than protective levels. Conclusions: Our simulation shows >50% of research participants on standard dosing would have 3rd trimester trough plasma TFV concentrations below levels associated with protection. This simulation provides the quantitative basis for the design of prospective TDF/FTC studies during pregnancy to evaluate the safety and appropriateness of pregnancy-adjusted dosing.

Deep Learning Methods Applied to Drug Concentration Prediction of Olanzapine.

Pharmacometrics and the utilization of population pharmacokinetics play an integral role in model-informed drug discovery and development (MIDD). Recently, there has been a growth in the application of deep learning approaches to aid in areas within MIDD. In this study, a deep learning model, LSTM-ANN, was developed to predict olanzapine drug concentrations from the CATIE study. A total of 1527 olanzapine drug concentrations from 523 individuals along with 11 patient-specific covariates were used in model development. The hyperparameters of the LSTM-ANN model were optimized through a Bayesian optimization algorithm. A population pharmacokinetic model using the NONMEM model was constructed as a reference to compare to the performance of the LSTM-ANN model. The RMSE of the LSTM-ANN model was 29.566 in the validation set, while the RMSE of the NONMEM model was 31.129. Permutation importance revealed that age, sex, and smoking were highly influential covariates in the LSTM-ANN model. The LSTM-ANN model showed potential in the application of drug concentration predictions as it was able to capture the relationships within a sparsely sampled pharmacokinetic dataset and perform comparably to the NONMEM model.

Mechanistic modeling of ophthalmic, nasal, injectable, and implant generic drug products: A workshop summary report.

For approval, a proposed generic drug product must demonstrate it is bioequivalent (BE) to the reference listed drug product. For locally acting drug products, conventional BE approaches may not be feasible because measurements in local tissues at the sites of action are often impractical, unethical, or cost-prohibitive. Mechanistic modeling approaches, such as physiologically-based pharmacokinetic (PBPK) modeling, may integrate information from drug product properties and human physiology to predict drug concentrations in these local tissues. This may allow clinical relevance determination of critical drug product attributes for BE assessment during the development of generic drug products. In this regard, the Office of Generic Drugs of the US Food and Drug Administration has recently established scientific research programs to accelerate the development and assessment of generic products by utilizing model-integrated alternative BE approaches. This report summarizes the presentations and panel discussion from a public workshop that provided research updates and information on the current state of the use of PBPK modeling approaches to support generic product development for ophthalmic, injectable, nasal, and implant drug products.

The MPRINT Hub Data, Model, Knowledge and Research Coordination Center: Bridging the gap in maternal-pediatric therapeutics research through data integration and pharmacometrics.

Maternal and pediatric populations have historically been considered "therapeutic orphans" due to their limited inclusion in clinical trials. Physiologic changes during pregnancy and lactation and growth and maturation of children alter pharmacokinetics (PK) and pharmacodynamics (PD) of drugs. Precision therapy in these populations requires knowledge of these effects. Efforts to enhance maternal and pediatric participation in clinical studies have increased over the past few decades. However, studies supporting precision therapeutics in these populations are often small and, in isolation, may have limited impact. Integration of data from various studies, for example through physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling or bioinformatics approaches, can augment the value of data from these studies, and help identify gaps in understanding. To catalyze research in maternal and pediatric precision therapeutics, the Obstetric and Pediatric Pharmacology and Therapeutics Branch of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) established the Maternal and Pediatric Precision in Therapeutics (MPRINT) Hub. Herein, we provide an overview of the status of maternal-pediatric therapeutics research and introduce the Indiana University-Ohio State University MPRINT Hub Data, Model, Knowledge and Research Coordination Center (DMKRCC), which aims to facilitate research in maternal and pediatric precision therapeutics through the integration and assessment of existing knowledge, supporting pharmacometrics and clinical trials design, development of new real-world evidence resources, educational initiatives, and building collaborations among public and private partners, including other NICHD-funded networks. By fostering use of existing data and resources, the DMKRCC will identify critical gaps in knowledge and support efforts to overcome these gaps to enhance maternal-pediatric precision therapeutics.

Systemic exposure to hydroxychloroquine and its relationship with outcome in severely ill COVID-19 patients in New York City.

AIM: To investigate the relationship between systemic exposure to hydroxychloroquine (HCQ) and its metabolite desethylhydroxychloroquine (DHCQ) and clinical outcome in severely ill patients treated with a standard oral dose regimen of HCQ during the first wave of COVID-19 in New York City. METHODS: We correlated retrospective clinical data with drug exposure prospectively assessed from convenience samples using population pharmacokinetics and Bayesian estimation. Systemic exposure was assessed in 215 patients admitted to ICU or COVID-ward for whom an interleukin-6 level was requested and who were still alive 24 hours after the last dose of HCQ. Patients received oral HCQ 600 mg twice daily on day 1 followed by 4 days of 400 mg daily. RESULTS: Fifty-three precent of the patients were intubated at 5.4 ± 6.4 days after admission and 26.5% died at an average of 32.2 ± 19.1 days. QTc at admission was 448 ± 34 ms. Systemic exposure to HCQ and DHCQ demonstrated substantial variability. Cumulative area under the serum concentration-time curve up to infinity for HCQ was 71.4 ± 19.3 h mg/L and for DHCQ 56.5 ± 28.3 h mg/L. Variability in systemic exposure was not clearly explained by renal function, liver function or inflammatory state. In turn, systemic exposure did not correlate with intubation status, survival or QTc prolongation. CONCLUSION: This study in severely ill patients was not able to find any relationship between systemic exposure to HCQ and DHCQ and clinical outcome at a routine dose regimen and adds to the growing body of evidence that oral HCQ does not alter the course of disease in COVID-19 patients.

A Physiologically-Based Pharmacokinetic Model of the Brain Considering Regional Lipid Variance.

Modeling and simulation of the central nervous system provides a tool for understanding and predicting the distribution of small molecules throughout the brain tissue and cerebral spinal fluid (CSF), and these efforts often rely on empirical data to make predictions of distributions to move toward a better mechanistic understanding. A physiologically based pharmacokinetic model presented here incorporates multiple means of drug distribution to assemble a model for understanding potential factors that may determine the distribution of drugs across various regions of the brain, including both intra- and extracellular regions. Two classes of parameters are presented. The first concerns regional gross anatomic variability of the brain; the second concerns estimation of unbound fractions of drugs using know membrane phospholipid heterogeneity derived from regional lipid content. The model was then tested by comparing its outcomes to data from published human pharmacokinetic studies of acetaminophen, morphine, and phenytoin. The alignment of model predictions in the plasma, CSF, and tissue concentrations with the published data from studies of those three drugs suggests that the model can be a template for identifying drug localization in the brain. Clearly, knowledge of differentiated drug distribution in the brain is a requisite step in postulating pharmacodynamic and certain disease mechanisms. SIGNIFICANCE STATEMENT: This study concerns the application of heterogenous lipid distribution in brain tissue to predict regional variations in drug distribution in the brain via a mathematical model, thus expanding upon the current understanding of mechanisms of drug distribution in the central nervous system.

A population pharmacokinetic model based on HPTN 077 of long-acting injectable cabotegravir for HIV PrEP.

AIMS: Cabotegravir delivered as a long-acting intramuscular injection has shown superior efficacy to oral tenofovir-emtricitabine as pre-exposure prophylaxis (PrEP) for HIV. Cabotegravir pharmacokinetics (PK), like those of other long-acting depot preparations, exhibit variability between individuals and between injection occasions. The aim of this study is to describe the population pharmacokinetics of long-acting cabotegravir (CAB-LA). METHODS: Using available PK measurements from 133 participants in the HIV Prevention Trials Network (HPTN) 077 trial, we analysed CAB-LA PK data using nonlinear mixed-effects modelling to develop a population PK model. RESULTS: A two-compartment model with first order absorption best described the CAB-LA PK. The analysis identified between-occasion variability (BOV, i.e., differences in PK within one individual from one injection to the next) as a significant covariate affecting the absorption rate, with an estimated contribution of BOV to PK variability on the absorption rate (ka ) of 38.5%. Sex and body weight were identified as significant covariates influencing the absorption rate and apparent clearance of CAB-LA after intramuscular injection at various doses and frequencies. Male participants had 67% higher ka than female participants. Serially adding to the model body weight on clearance, sex on ka , and BOV on ka led to a decrease in the objective function value (OFV) of 24.4, 36 and 321.4, respectively. CONCLUSION: The public availability of this model will facilitate and enable a wide variety of future clinically relevant simulations to inform the optimal use of CAB-LA.

Model-Based Prediction of Irinotecan-Induced Grade 4 Neutropenia in Advanced Cancer Patients: Influence of Demographic and Clinical Factors.

Severe neutropenia is the major dose-liming toxicity of irinotecan-based chemotherapy. The objective was to assess to what extent a population pharmacokinetic/pharmacodynamic model including patient-specific demographic/clinical characteristics, individual pharmacokinetics, and absolute neutrophil counts (ANCs) can predict irinotecan-induced grade 4 neutropenia. A semimechanistic population pharmacokinetic/pharmacodynamic model was developed to describe neutrophil response over time in 197 patients with cancer receiving irinotecan. For covariate analysis, sex, race, age, pretreatment total bilirubin, and body surface area were evaluated to identify significant covariates on system-related parameters (mean transit time (MTT) and É£) and sensitivity to neutropenia effects of irinotecan and SN-38 (SLOPE). The model-based simulation was performed to assess the contribution of the identified covariates, individual pharmacokinetics, and baseline ANC alone or with incremental addition of weekly ANC up to 3 weeks on predicting irinotecan-induced grade 4 neutropenia. The time course of neutrophil response was described using the model assuming that irinotecan and SN-38 have toxic effects on bone marrow proliferating cells. Sex and pretreatment total bilirubin explained 10.5% of interindividual variability in MTT. No covariates were identified for SLOPE and γ. Incorporating sex and pretreatment total bilirubin (area under the receiver operating characteristic curve (AUC-ROC): 50%, 95% CI 50-50%) or with the addition of individual pharmacokinetics (AUC-ROC: 62%, 95% CI 53-71%) in the model did not result in accurate prediction of grade 4 neutropenia. However, incorporating ANC only at baseline and week 1 in the model achieved a good prediction (AUC-ROC: 78%, 95% CI 69-88%). These results demonstrate the potential applicability of a model-based approach to predict irinotecan-induced neutropenia, which ultimately allows for personalized intervention to maximize treatment outcomes.

MAP Bayesian modelling combining striatal dopamine receptor occupancy and plasma concentrations to optimize antipsychotic dose regimens in individual patients.

AIMS: Develop a robust and user-friendly software tool for the prediction of dopamine D2 receptor occupancy (RO) in patients with schizophrenia treated with either olanzapine or risperidone, in order to facilitate clinician exploration of the impact of treatment strategies on RO using sparse plasma concentration measurements. METHODS: Previously developed population pharmacokinetic models for olanzapine and risperidone were combined with a pharmacodynamic model for D2 RO and implemented in the R programming language. Maximum a posteriori Bayesian estimation was used to provide predictions of plasma concentration and RO based on sparse concentration sampling. These predictions were then compared to observed plasma concentration and RO. RESULTS: The average (standard deviation) response times of the tools, defined as the time required for the application to predict parameter values and display the output, were 2.8 (3.1) and 5.3 (4.3) seconds for olanzapine and risperidone, respectively. The mean error (95% confidence interval) and root mean squared error (95% confidence interval) of predicted vs. observed concentrations were 3.73 ng/mL (-2.42-9.87) and 10.816 ng/mL (6.71-14.93) for olanzapine, and 0.46 ng/mL (-4.56-5.47) and 6.68 ng/mL (3.57-9.78) for risperidone and its active metabolite (9-OH risperidone). Mean error and root mean squared error of RO were -1.47% (-4.65-1.69) and 5.80% (3.89-7.72) for olanzapine and -0.91% (-7.68-5.85) and 8.87% (4.56-13.17) for risperidone. CONCLUSION: Our monitoring software predicts concentration-time profiles and the corresponding D2 RO from sparsely sampled concentration measurements in an accessible and accurate form.

Integration of DNA sequencing with population pharmacokinetics to improve the prediction of irinotecan exposure in cancer patients.

BACKGROUND: Irinotecan (CPT-11) is an anticancer agent widely used to treat adult solid tumours. Large interindividual variability in the clearance of irinotecan and SN-38, its active and toxic metabolite, results in highly unpredictable toxicity. METHODS: In 217 cancer patients treated with intravenous irinotecan single agent or in combination, germline DNA was used to interrogate the variation in 84 genes by next-generation sequencing. A stepwise analytical framework including a population pharmacokinetic model with SNP- and gene-based testing was used to identify demographic/clinical/genetic factors that influence the clearance of irinotecan and SN-38. RESULTS: Irinotecan clearance was influenced by rs4149057 in SLCO1B1, body surface area, and co-administration of 5-fluorouracil/leucovorin/bevacizumab. SN-38 clearance was influenced by rs887829 in UGT1A1, pre-treatment total bilirubin, and EGFR rare variant burden. Within each UGT1A1 genotype group, elevated pre-treatment total bilirubin and/or presence of at least one rare variant in EGFR resulted in significantly lower SN-38 clearance. The model reduced the interindividual variability in irinotecan clearance from 38 to 34% and SN-38 clearance from 49 to 32%. CONCLUSIONS: This new model significantly reduced the interindividual variability in the clearance of irinotecan and SN-38. New genetic factors of variability in clearance have been identified.

Development of a genetic algorithm and NONMEM workbench for automating and improving population pharmacokinetic/pharmacodynamic model selection.

The current approach to selection of a population PK/PD model is inherently flawed as it fails to account for interactions between structural, covariate, and statistical parameters. Further, the current approach requires significant manual and redundant model modifications that heavily lend themselves to automation. Within the discipline of numerical optimization it falls into the "local search" category. Genetic algorithms are a class of algorithms inspired by the mathematics of evolution. GAs are general, powerful, robust algorithms and can be used to find global optimal solutions for difficult problems even in the presence of non-differentiable functions, as is the case in the discrete nature of including/excluding model components in search of the best performing mixed-effects PK/PD model. A genetic algorithm implemented in an R-based NONMEM workbench for identification of near optimal models is presented. In addition to the GA capabilities, the workbench supports modeling efforts by: (1) Organizing and displaying models in tabular format, allowing the user to sort, filter, edit, create, and delete models seamlessly, (2) displaying run results, parameter estimates and precisions, (3) integrating xpose4 and PsN to facilitate generation of model diagnostic plots and run PsN scripts, (4) running regression models between post-hoc parameter estimates and covariates. This approach will further facilitate the scientist to shift efforts to focus on model evaluation, hypotheses generation, and interpretation and applications of resulting models.

Machine learning in pharmacometrics: Opportunities and challenges.

The explosive growth in medical devices, imaging and diagnostics, computing, and communication and information technologies in drug development and healthcare has created an ever-expanding data landscape that the pharmacometrics (PMX) research community must now traverse. The tools of machine learning (ML) have emerged as a powerful computational approach in other data-rich disciplines but its effective utilization in the pharmaceutical sciences and PMX modelling is in its infancy. ML-based methods can complement PMX modelling by enabling the information in diverse sources of big data, e.g. population-based public databases and disease-specific clinical registries, to be harnessed because they are capable of efficiently identifying salient variables associated with outcomes and delineating their interdependencies. ML algorithms are computationally efficient, have strong predictive capabilities and can enable learning in the big data setting. ML algorithms can be viewed as providing a computational bridge from big data to complement PMX modelling. This review provides an overview of the strengths and weaknesses of ML approaches vis-à-vis population methods, assesses current research into ML applications in the pharmaceutical sciences and provides perspective for potential opportunities and strategies for the successful integration and utilization of ML in PMX.

Clozapine Once-Daily Versus Divided Dosing Regimen: A Cross-sectional Study in Japan.

OBJECTIVE: Clozapine is generally recommended to be prescribed in a divided dosing regimen based on its relatively short plasma half-life. However, there has been little evidence to support the superiority of divided dosing of clozapine over once-daily dosing. To our knowledge, there have been no studies examining differences in actual plasma concentrations or adverse effects between the 2 dosing strategies of clozapine. We aimed to compare actual plasma concentrations of clozapine between once-daily and divided dosing regimens, and to examine the relationships of these regimens with psychiatric symptoms and adverse effects of clozapine. METHODS: We analyzed data from 108 participants of a previous study conducted in 2 hospitals in Japan. A population pharmacokinetic model was used to estimate the peak and trough plasma concentrations of clozapine based on actual plasma concentrations. We evaluated psychiatric symptoms with the Brief Evaluation of Psychosis Symptom Domains and adverse effects of clozapine with the Glasgow Antipsychotic Side-effects Scale for Clozapine. RESULTS: The estimated peak and trough plasma concentrations of clozapine did not differ significantly between once-daily and divided dosing regimens. There were no significant differences in psychiatric symptoms except for depression/anxiety or subjective adverse effects of clozapine between the 2 dosing strategies. CONCLUSIONS: Our findings tentatively support the feasibility and clinical utility of once-daily dosing of clozapine in clinical practice. Further studies are needed to replicate these findings and determine causality between dosing strategies and clinical outcomes.