Search Results (191)

Interindividual variability in drug response remains a major challenge in clinical pharmacology despite substantial advances in therapeutic drug monitoring (TDM), pharmacogenomics, pharmacokinetics/pharmacodynamics (PK/PD), and model-informed precision dosing (MIPD). Although these approaches have improved individualized therapy, clinically important variability in efficacy and toxicity persists because drug response is determined not only by systemic exposure but also by target engagement, disease biology, compensatory pathways, organ function, immune status, and dynamic patient-specific molecular states. Recent advances in multiomics, systems pharmacology, and artificial intelligence (AI) provide an opportunity to integrate these complementary biological and clinical dimensions within more comprehensive precision pharmacotherapy frameworks. This narrative review examines the evolving integration of PK, PD, TDM, pharmacometrics, multiomic technologies, mechanistic AI, and systems pharmacology across drug development and clinical care. Particular emphasis is placed on the limitations of exposure-based dosing alone, the biological determinants of interindividual variability, the transition from conventional TDM toward adaptive model-informed monitoring, and emerging approaches for integrating molecular and clinical data to support individualized therapeutic decision-making. Operon™ is discussed as an illustrative example of an internally operated mechanistic systems biology platform to demonstrate how biologically informed computational frameworks may integrate pharmacological and multiomic information within drug development workflows. The review further examines applications in polypharmacy, drug–drug interaction assessment, clinical trial enrichment, regulatory science, and adaptive dosing, while emphasizing that analytical validity, clinical validity, clinical utility, prospective validation, transparency, and clearly defined contexts of use remain essential prerequisites for clinical implementation. Collectively, these developments support a transition from concentration-guided dosing toward mechanism-informed precision pharmacotherapy that integrates drug exposure with biological response and clinical outcomes while maintaining rigorous standards for validation and regulatory acceptance. Full article

Background: Identifying influential individuals is a critical step in nonlinear mixed-effects modeling, yet commonly used diagnostics primarily reflect local model fit or parameter perturbation and may fail to capture the full impact of individual data on model estimation. Methods: We conducted a simulation study based on a one-compartment pharmacokinetic model with first-order absorption. Four scenarios were investigated: a reference scenario without induced influence, a residual outlier scenario, a structurally influential individual with enriched sampling, and a latent subpopulation scenario. For each scenario, 25 datasets of 100 individuals were simulated, and individual influence was assessed using a leave-one-out jackknife approach. Influence metrics included the change in objective function value (ΔOFV) and parameter perturbation measures. Detection performance was evaluated at both subject and dataset levels. Results: All jackknife runs were successfully completed and analyzable. Residual outliers were consistently identified by both ΔOFV and parameter-based metrics. In contrast, structurally influential individuals were reliably detected by ΔOFV (median rank = 1) but not by parameter-based metrics (median rank ≈ 25). Across scenarios, the association between ΔOFV and parameter perturbation was weak, and nearly absent in the structurally influential scenario. In the latent subpopulation scenario, individual-level detection was limited, but dataset-level detection remained effective, with at least one subpopulation member frequently identified among top-ranked individuals. Conclusions: Individual influence in nonlinear mixed-effects models is strongly mechanism-dependent. Jackknife-based ΔOFV provides a direct and general measure of individual influence, capable of detecting both residual outliers and structurally influential individuals. In contrast, parameter-based metrics quantify parameter sensitivity rather than individual influence and may therefore overlook influential subjects in specific contexts. These findings support the use of jackknife deletion as a reference approach for influence assessment in pharmacometric workflows. Full article

Background/Objectives: Pemetrexed–platinum chemotherapy is a key treatment option for non-squamous non-small cell lung cancer (NSCLC); however, its use is often limited by hematologic toxicity, particularly neutropenia. We aimed to develop a model-informed framework for assessing neutrophil dynamics using routinely available clinical data and pemetrexed exposure. Methods: This real-world investigation included 86 patients with NSCLC who received pemetrexed–platinum chemotherapy for model development, and 83 patients who received the same chemotherapy plus pembrolizumab or bevacizumab for validation. We developed a nonlinear mixed-effects model to predict neutrophil dynamics during the first cycle following pemetrexed–platinum chemotherapy, using patient-specific clinical data collected before chemotherapy initiation and pemetrexed pharmacokinetic parameters derived from physiologically based pharmacokinetic (PBPK) modeling. Results: The final model suggested that the area under the curve (AUC)_0–24 >175 μg·h/mL for pemetrexed, blood urea nitrogen, and concomitant use of renin–angiotensin system inhibitors influenced neutrophil suppression and delayed recovery. The receiver operating characteristic curve (AUROC) for identifying patients with a neutrophil count <1500/μL immediately before the anticipated next treatment cycle was 0.768 (95% CI: 0.639–0.898) in the development cohort, and 0.718 (95% CI: 0.545–0.891) in the validation cohort. Conclusions: This model-informed framework, based on PBPK-derived pemetrexed exposure and routinely available clinical factors, may help identify patients at risk of clinically relevant neutropenia that could delay the initiation of the next treatment cycle. Full article

Background/Objectives: Selecting appropriate doses for pediatric patients remains one of the most complex challenges in drug development because developmental changes in physiology, metabolism, organ function, and pharmacodynamics substantially influence drug exposure and response. This review summarizes current evidence-based approaches to pediatric dose selection across the developmental continuum and evaluates contemporary model-informed strategies for individualized dosing. Methods: A narrative review of the literature was conducted focusing on pediatric dose-scaling methodologies, developmental pharmacology, physiologically based pharmacokinetic (PBPK) modeling, population pharmacokinetic (PopPK) approaches, exposure–response analysis, therapeutic drug monitoring, and regulatory extrapolation frameworks. Special populations and clinical scenarios relevant to pediatric dose optimization were also evaluated. Results: Simple body weight-based scaling from adult doses inadequately accounts for developmental changes in drug disposition and response. Allometric scaling combined with maturation functions provides improved dose prediction in neonates and infants, while PBPK and PopPK modeling support mechanistic and data-driven dose optimization across pediatric age groups. Fat-free-mass (FFM)-based scaling is preferred over total body weight for many drugs in children with obesity. Additional considerations including obesity, biologics, formulation and excipient safety, pharmacogenomics, critical illness, therapeutic hypothermia, extracorporeal support, therapeutic drug monitoring, and drug–drug interactions substantially influence pediatric dosing strategies. Regulatory frameworks including ICH E11A increasingly support model-informed pediatric extrapolation and precision dosing approaches. Conclusions: Pediatric dose selection has evolved from empirical weight-based dosing toward integrated model-informed strategies incorporating developmental physiology, pharmacometrics, and regulatory science. Allometry, maturation functions, FFM-based scaling, PBPK, PopPK, and therapeutic drug monitoring provide complementary tools for rational pediatric dose optimization, although drug- and pathway-specific validation remains essential, particularly in neonates and critically ill children. Full article

Background: Hyperuricemia is an adverse effect frequently observed during favipiravir treatment. The time course, from uric acid elevation to recovery, and quantitative relationship between drug exposure and changes in serum uric acid levels remain insufficiently characterized. We investigated the pharmacodynamic mechanism of uric acid elevation and described its time course by population pharmacokinetic and pharmacodynamic modelling. Methods: Patients who received favipiravir for coronavirus disease 2019 or severe fever with thrombocytopenia syndrome were retrospectively evaluated. The pharmacokinetics of favipiravir were described by a one-compartment model with first-order absorption and elimination. Metabolite concentrations were predicted based on previously reported values. Changes in serum uric acid levels were described by a turnover model with zero-order production and first-order elimination. The drug effect was implemented as inhibition of the uric acid elimination process. Simulations based on the final model were performed for 10 consecutive days after the clinical regimen, with a 21-day follow-up. Results: The final model supported the inhibition of uric acid elimination by favipiravir and its metabolite. Regarding simulations, serum uric acid levels reached a median peak of 6.93 mg/dL at 6.7 days after treatment initiation and returned to pre-treatment levels within 4.0 days after treatment discontinuation. Conclusions: This combined population pharmacokinetic and pharmacodynamic turnover model quantified favipiravir-associated increases in serum uric acid levels and showed a transient profile with rapid recovery after drug discontinuation. These findings underscore the need for monitoring serum uric acid levels during favipiravir treatment, particularly in patients at a higher risk of gout. Full article

Background/Objectives: Special populations, including pediatric, geriatric, and organ-impaired patients, are consistently underrepresented in randomized controlled trials (RCTs), resulting in limited evidence for safe and effective dosing. Off-label use is common, and variability in drug exposure and response increases the risk of adverse drug reactions (ADRs). This review aims to examine how integrating pharmacometrics (PMX) with real-world data (RWD) can address evidence gaps by supporting dose optimization, population expansion, and safety evaluation in these vulnerable groups. Methods: A narrative literature review was conducted using PubMed, Embase, and Web of Science (January 2000–November 2025). Using Boolean combinations of PMX and RWD-related search terms, approximately 200–300 records were identified across the three databases; approximately 30 full-text articles were reviewed, and representative case studies were selected based on population diversity, methodological variation, and regulatory or clinical impact. Results: RWD–PMX integration has been applied across three domains: (i) dosing optimization through therapeutic drug monitoring (TDM)-informed PopPK modeling and model external validation in pediatric and neonatal populations; (ii) population expansion supporting dose extrapolation and regulatory decision-making for unapproved groups; and (iii) safety evaluation enabling identification of exposure–toxicity risk factors in vulnerable cohorts. Conclusions: Integrating PMX with RWD provides a practical and mechanistically grounded framework for evaluating dosing, treatment eligibility, and safety in populations insufficiently represented in clinical trials. Accumulating evidence indicates that RWD–PMX methodologies can complement traditional clinical research and inform regulatory decision-making. Continued refinement of data quality standards, validation practices, and guidance frameworks will be essential for broader adoption. Full article

Methods: This study explored key study design factors that could impact the precision of pediatric pharmacokinetics (PK) estimation. A virtual pediatric population was constructed, incorporating diverse body weight distribution sourced from the U.S. Centers for Disease Control and Prevention (CDC) growth charts. These generated weights were aggregated based on age ranges (2–5, 6–11, 12–17, and 2–17 y.o.), and different sample sizes were randomly selected to simulate PK concentrations over an approximately 5 half-lives period for a hypothetical monoclonal antibody. Throughout the simulations, the “true” allometric scaling exponents for the apparent volume of distribution and apparent clearance were consistently assumed to be 1.0 and 0.75, respectively, consistent with physiological and pharmacological knowledge for monoclonal antibodies. The impact of various pediatric study design factors on the model estimates of allometric exponents was then investigated by assuming the generated PK data as observed, with unknown PK parameters and allometric scaling exponent values. The data were subsequently fitted with population PK models, and estimated parameters were compared to “true” values to assess precision. Precision in estimated allometric exponents served as a marker for evaluating how effectively data from various study designs can inform the pediatric PK estimation. Results: Generally, estimates of allometric scaling exponents were more accurate with a larger sample size, proper PK sampling scheme, inclusion of densely sampled adult data, and a broader range of age. Conclusions: Considering the limitations in designing most pediatric studies, these findings support recent regulatory recommendations that standard allometric exponents should be considered in pediatric PK analysis for monoclonal antibodies in general. Full article

The usefulness of ChatGPT, a large language model, has recently been explored in medical research. However, no studies have examined its reproducibility or applicability to therapeutic drug monitoring (TDM), a core task of clinical pharmacists. In this simulation study, we evaluated the feasibility of using ChatGPT for vancomycin (VCM) TDM based on Bayesian estimation. A total of 1000 virtual patients were generated by Monte Carlo simulations using a population pharmacokinetic model of VCM. Bayesian-estimated pharmacokinetic parameters and predicted concentrations were input into ChatGPT, and dosage regimens were compared among the three conditions, using temperature as a hyperparameter (T = 0.1, 0.5, and 1.0). Reproducibility was evaluated using the mode percentage in repeated runs. The reproducibility of the ChatGPT output was higher at T = 0.1 than at T = 0.5 and T = 1.0. When ChatGPT simulated the mode-recommended regimen (T = 0.1), the target attainment rate of the area under the serum concentration (AUC) (400–600 mg·h/L) improved from 25.5% (pre-optimization AUC (fixed-dose regimen)) to 71.5% (post-optimization AUC (ChatGPT-guided regimen)). These findings demonstrate that ChatGPT-based TDM using Bayesian estimation can enhance dose optimization. Adjusting the hyperparameter temperature to 0.1 improved reproducibility, suggesting that a reliable ChatGPT-assisted TDM support system may be clinically useful. Full article

Background/Objectives: The integration of machine learning (ML) within model-informed drug development (MIDD) represents a rapidly evolving paradigm in pharmacometrics, enabling improved prediction, optimization, and regulatory decision-making across drug development pipelines. However, the extent to which ML methods are explicitly integrated into regulatory decision-making remains limited and unevenly characterized. This study aims to systematically map the ML-MIDD scholarly landscape, identify core sources and contributors, assess thematic evolution, and summarize key methodological advancements through an integrative bibliometric and systematic review. Methods: A comprehensive literature search was conducted across Web of Science, Scopus, and PubMed (2015–2025), followed by metadata harmonization and deduplication. Bibliometric analysis was performed using Bibliometrix, VOSviewer, and PRISMA guidelines to characterize publication trends, collaboration patterns, thematic structures, and representative methodological contributions. Results: A total of 770 records were initially retrieved, with Scopus contributing the largest share (n = 343; 44.5%), followed by Web of Science (n = 322; 41.8%) and PubMed (n = 105; 13.6%). After deduplication, 607 unique publications remained (78.8% of total), and 560 were included in the final systematic review (97.6% of full texts). Publications spanned 269 sources, with core journals accounting for 28% of output. The United States led in volume (n = 665; 20.8%) and international collaboration (16.47%). Thematic evolution revealed transitions from foundational PK/PD methods (2016–2018) to applied ML-driven precision pharmacology (2022–2025). Conclusions: Emerging methods included deep learning, reinforcement learning, and hybrid mechanistic–ML models. ML-MIDD is a rapidly maturing interdisciplinary field, evidenced by expanding methodological diversity and increasing use of ML-enabled components within regulatory-relevant modeling workflows, rather than formal regulatory endorsement of ML-MIDD as a standalone methodology, indicating growing translational relevance but continued need for validation and regulatory clarity. Full article

Geriatric pharmacotherapy is usually challenged by physiological senescence. For instance, progressive declines in organ function and alterations in body composition can complicate drug disposition. However, conventional pharmacometrics models commonly have limited capacity to map these high-dimensional, nonlinear relationships. In this review, we are examining the recent shift toward integrating machine learning (ML) with mechanistic pharmacokinetic (PK)/pharmacodynamic (PD) models to improve the accuracy and precision of dosing. Machine learning approaches like Random Forest and XGBoost consistently provided more accurate exposure predictions and significantly more efficient computational workflows than conventional methods. Nevertheless, concerns such as “black box” transparency and the potential of algorithmic bias toward specific patient demographics are challenging. It is important to incorporate explainability tools like SHAP, and adopting FAIR data principles is crucial for achieving professional trust and ensuring site-specific generalizability. Full article

Background/Objectives: Population pharmacokinetic (PopPK) modeling in NONMEM requires iterative, expertise-dependent workflows. Naïve zero-shot prompting of general-purpose large language models (LLMs) typically produces NONMEM code that fails to execute. This study introduces PKGPT, a recursive agentic LLM system designed to automate NONMEM-based PopPK model development and benchmarks its performance against human expert models. Methods: PKGPT, powered by Google’s Gemini 3.0 Flash, embeds pharmacometrics expertise into phase-specific expert-agent prompts orchestrated across five sequential phases: base model establishment, structural diagnostics, overfitting reduction, random-effects optimization, and covariate analysis. The system recursively executes NONMEM, parses outputs, and iteratively refines control streams. PKGPT was evaluated on three public datasets (warfarin, theophylline, and tobramycin) and benchmarked against independently developed human expert models. Results: PKGPT consistently produced executable, converging NONMEM models across all three datasets. In warfarin, both PKGPT and the human expert selected a one-compartment oral structure (ADVAN2), but the expert achieved a lower OFV (294.41 vs. 484.43) via covariate scaling. In theophylline, PKGPT produced parameter estimates close to the expert solution (Ka = 1.59 vs. 1.46 h⁻¹; CL = 0.0399 vs. 0.0404 L/h/kg). In tobramycin, PKGPT correctly identified a two-compartment structure but produced physiologically implausible peripheral volume estimates (V2 = 149 L vs. expert’s 13.2 L). Across datasets, PKGPT did not identify clinically established covariates, and run-to-run reproducibility was variable. Conclusions: PKGPT substantially improves the robustness and usability of LLM-generated NONMEM code compared with naïve zero-shot prompting, accelerating model drafting and iterative refinement, but physiological plausibility and clinical interpretability still require a human-in-the-loop oversight. Full article

Continuous, in vivo drug and biomarker measurements could transform healthcare, enabling both the high-precision personalization of drug dosing and the real-time monitoring of health status. A practical realization of this vision, however, requires an improved understanding of the relationship between concentrations measured in the easily accessible dermal interstitial fluid (ISF) that correlate with the plasma concentrations that guide clinical decision making. As a preliminary step towards this goal, here we have used electrochemical, aptamer-based (EAB) sensors to perform seconds-resolved vancomycin measurements in the plasma and subcutaneous ISF of live rats. Concentrations of the antibiotic in the ISF vary rather little between different subcutaneous sites and, after the very rapid initial distribution phase is complete, they are well correlated with the plasma concentrations (mean R² = 0.88). Likewise, a simple, two-compartment, two-parameter model describes our six paired plasma and ISF drug time courses quantitatively. Together, these findings provide further evidence of the viability of the drug concentration measurements performed in the subcutaneous or dermal ISF as a less invasive approach to real-time drug monitoring in individual patients. Full article

Antibody–drug conjugates (ADCs), which integrate a cytotoxic drug known as the payload into a tumor-targeting monoclonal antibody via a linker, have emerged as promising candidates for cancer therapy and are a new avenue for targeted cancer therapy. The pharmacokinetic (PK) profiles of ADCs are distinctive due to their unique distribution, catabolism, and elimination. Their deconjugation in circulation and variations in the drug-to-antibody ratio increase the complexity of their PK profiles. Pharmacometric models depicting the PK properties and exposure-response (E-R) relationships of ADCs are important for optimizing dosing regimens and supporting decisions during ADC development. This review considers the PK profiles of ADCs, physiologically based PK models, semi-mechanistic and mechanistic PK models, population PK models, and E-R analyses for dose optimization. The prospects and challenges for ADCs, especially the urgent need for advanced analytical technology and modeling approaches, are also outlined. Full article

Background: Antiviral therapies are widely used during pregnancy and are generally considered safe, pregnancy-specific severe safety signals continue to be observed in post-marketing pharmacovigilance data. These signals are rarely interpreted within an integrated mechanistic framework. Methods: We analysed pregnancy-related EudraVigilance reports (2015–2025) using a previously network-based pharmacovigilance framework. Established ADR clusters were treated as fixed phenotypes and integrated with in silico ADMET liabilities, literature-derived pregnancy pharmacokinetic/pharmacodynamic (PK/PD) parameters, polypharmacy and co-medication network metrics, and exploratory statistical, machine-learning, and exposure–liability analyses for mechanistic prioritisation. Results: Phenotype membership explained 22.3% of the variance in composite ADMET risk (intraclass correlation coefficient = 0.223; p < 0.001), and all tested ADMET parameters differed significantly across phenotypes (FDR-adjusted p < 10⁻¹⁰). One phenotype showed pronounced enrichment, with 13 antivirals over-represented. Polypharmacy strongly modified seriousness, with odds of serious outcomes increasing by ~5% per additional co-reported active drug (OR 1.05, 95% CI 1.04–1.05). A composite mechanistic vulnerability index showed moderate concordance with empirical burden (Spearman’s ρ = 0.65), while regimen-level prioritisation of drug–drug interactions (DDIs) identified no high-priority combinations. Conclusions: Pregnancy-related antiviral ADRs cluster into reproducible phenotypes driven by mechanistic liability and system-level complexity, supporting mechanistically informed prioritisation and targeted pharmacometric follow-up. Full article

Background: Vancomycin dosing during continuous renal replacement therapy (CRRT) remains challenging due to profound pharmacokinetic (PK) variability and lack of standardized guidance. Population pharmacokinetic (PopPK) models provide a quantitative framework to identify covariates affecting drug disposition and support individualized dosing. Methods: This systematic review, registered in PROSPERO (CRD420250655157), comprehensively identified PopPK studies evaluating vancomycin in critically ill adults undergoing CRRT. PubMed, Embase, Web of Science, and Cochrane Library were searched from inception to 28 February 2025. Eligible studies reported PopPK analyses or simulations providing PK parameters and/or dosing recommendations. Data were extracted on study characteristics, CRRT settings, PK findings, and dose optimization strategies. In addition, covariates were categorized based on whether they were merely explored or statistically confirmed within the respective PopPK models, as commonly reported in similar pharmacometrics studies. Due to the methodological nature of population pharmacokinetic model development studies, no standardized risk-of-bias tool was applied; instead, a structured descriptive methodological appraisal was performed. Results were synthesized narratively given the heterogeneity in structural models, covariate strategies, and CRRT modalities. Results: Twelve PopPK studies published between 2013 and 2023 met the inclusion criteria. Considerable heterogeneity was observed across study designs, CRRT modalities, and dosing strategies. Reported vancomycin clearance ranged from 0.7 to 3.0 L/h, and volume of distribution from 0.8 L/kg to >100 L. Effluent rate consistently emerged as the primary determinant of clearance, while residual diuresis, albumin concentration, and vasopressor use acted as relevant covariates. Loading doses of 25–30 mg/kg (up to 35 mg/kg at high effluent rates) and effluent-adjusted maintenance regimens achieved therapeutic AUC24/MIC targets more consistently when supported by early and repeated therapeutic drug monitoring (TDM). Conclusions: Vancomycin PK during CRRT is highly variable and driven by effluent intensity and patient-specific factors. Fixed regimens are inadequate; individualized dosing guided by effluent flow, renal function, and TDM is essential for optimal exposure. Prospective, multicenter PopPK studies integrating pharmacodynamic targets and clinical outcomes are warranted to refine and validate CRRT-specific dosing strategies. Full article