Explore the vast range of titles available.

We utilized an integrated approach for model‐informed dose selection to predict the recommended phase 2 dose (RP2D) of iruplinalkib, a selective oral ALK and ROS1 tyrosine kinase inhibitor. The efficacy and pharmacokinetics data obtained from ROS1 or ALK‐overexpressing cell‐derived xenograft models were used for pharmacokinetics pharmacodynamics (PKPD) modeling and calculating human iruplinalkib tumor static concentration (TSC). The plasma concentration‐time profile based on pooled clinical data was included in population PK (PopPK) analysis. The steady‐state plasma concentration‐time profile of iruplinalkib was predicted based on 1000 simulated replicates of the analysis dataset overlaid with data from 54 patients who received iruplinalkib at 120, 180, or 240 mg QD. A two‐compartment PopPK model with first‐order absorption and linear elimination successfully delineated iruplinalkib PK characteristics in mice, with good precision (relative standard error [RSE] < 30%). TSC in humans, estimated using a modified Simeoni model, was 98 and 78 ng/mL for ROS1‐positive and ALK‐positive tumors, respectively. A two‐compartment PopPK model with first‐order absorption and first‐order elimination was established based on data collected from previous clinical studies, and the model described iruplinalkib PK properties well (RSE < 30%). Iruplinalkib 180 mg QD was predicted to benefit over 90% of the population and recommended as the RP2D. This dose regimen was further validated by results of advanced clinical trials and ultimately incorporated into the prescribing information as the recommended dosage. A translational model‐based approach using integrated preclinical PK/PD and PopPK modeling in patients with non‐small cell lung cancer is a reliable method to predict RP2D. Trial Registration: ChiCTR.org.cn number: ChiCTR20170871; ClinicalTrials.gov identifier: NCT03389815; ChinaDrugTrials.org.cn number: CTR20190737

This analysis was conducted to assess the risk of anemia in schizophrenia patients during long-term treatment with the glycine transporter-1 (GlyT1) inhibitor iclepertin. A population pharmacokinetic-pharmacodynamic (popPKPD) analysis to characterize the impact of iclepertin exposure on hemoglobin levels was performed using a sequential nonlinear mixed effects modeling approach. The effects of patient characteristics were investigated in a covariate analysis to identify vulnerable patient subgroups, and population simulations were conducted to evaluate different treatment scenarios. Simulations predicted a new, decreased hemoglobin steady state under chronic iclepertin treatment, reached after approximately 120 days. For a typical patient, the intended therapeutic dose of 10 mg iclepertin daily led to a 2% decrease of hemoglobin levels. In a potential extreme scenario of iclepertin exposure fivefold higher than the average exposure following a 10 mg dose (e.g., due to co-administration of a strong CYP3A4 inhibitor), a 7.6% decrease of hemoglobin levels was found. In both scenarios more than 97.5% of the virtual patients stayed above the drug discontinuation safety threshold of 100 g/L hemoglobin (Phase III trial drug discontinuation threshold defined by the patient safety team). Sex, race, age, body mass index and alanine transaminase levels were found to correlate with changes in hemoglobin levels. No correlation with kidney function could be identified. None of the investigated covariate effects were strong enough to raise any safety concerns during chronic treatment with 10 mg iclepertin daily. This work provides a generalizable modeling and simulation framework to assess the anemia risk in patients and vulnerable patient subgroups during chronic iclepertin treatment. The results of this analysis suggest that iclepertin drug effects on patient hemoglobin levels are small, reversible, and of limited significance, even under long-term treatment. This is the first model of the relationship between iclepertin exposure and hemoglobin levels, and, to our best knowledge, the first model to characterize a drug effect on hemoglobin levels over time that has been validated with clinical data that extend beyond the erythrocyte life span of ∼126 days.

This overview summarizes the history and advancements of the modeling and simulation programs utilized in drug development and regulatory assessment, including the FDA’s Model-Informed Drug Development (MIDD) Paired Meeting Program and the Model-Integrated Evidence (MIE) Meeting Pilot Between FDA and Generic Drug Applicants. The U.S. Food and Drug Administration’s (FDA) recent notice concerning the use of the Type V Drug Master File (DMF) for Model Master File (MMF) submissions to support abbreviated new drug applications (ANDAs) encourages and facilitates model-sharing and model-reusability in drug development, supporting MIE programs using a broad range of quantitative models, including, but not limited to physiologically based pharmacokinetic (PBPK), population pharmacokinetics (PPK) and computational fluid dynamics (CFD) modeling. This overview also introduces the considerations and representative mock examples of MMFs discussed in the workshop titled “Considerations and Potential Regulatory Applications for a Model Master File (MMF)” co-hosted by the U.S. Food and Drug Administration (FDA) and the Center for Research on Complex Generics (CRCG) on May 2–3, 2024. MMFs promote modeling and simulation approaches by reducing the burden of resources in developing this type of approaches for the pharmaceutical industry while increasing consistency and efficiency in regulatory assessments.

During drug development, the design, interpretation and risk assessment of drug-drug interaction (DDI) are generally performed with physiologically-based pharmacokinetic (PBPK) modelling. Critical parameters are the hepatic metabolic fraction (fm) and intestinal availability (F G ) which are commonly informed by clinical data. In this study, two methods for the simultaneous estimation of these parameters are proposed which utilize the distinctive changes in substrate’s plasma concentration profiles in response to inhibition of intestinal and hepatic enzymes. The two-dimensional DDI (2D-DDI) method estimates the fm and F G values directly from the ratios of area-under-curve (AUCR) and maximum concentration (C max R), while the population PBPK method utilizes the full concentration–time data of a substrate without or with an inhibitor. The utility of both methods was demonstrated for a broad range of > 50,000 virtual and six actual CYP3A substrates. The 2D-DDI method is fast, reliable, and does not require a priori PBPK model development. The population PBPK method can estimate the population parameters and inter-individual variabilities of fm and F G and is applicable to more complex DDIs (e.g., multiple pathways/dynamic inhibitor concentration–time profiles) without the need for IV data. Like other approaches, both methods show an increasing uncertainty for substrates with high hepatic extraction and sensitivity to the assumed degree of enzyme inhibition. While both methods were evaluated for CYP3A substrates, the methodology equally applies to other enzymes. Additionally, this study provides guidance for clinical DDI study design to facilitate robust DDI extrapolation necessary to inform drug labels on concomitant medications in lieu of clinical trials. Graphical Abstract

Quantitative Systems Pharmacology (QSP) has emerged as a cornerstone of modern drug development, providing a robust framework to integrate data from preclinical and clinical studies, enhance decision-making, and optimize therapeutic strategies. By modeling biological systems and drug interactions, QSP enables predictions of outcomes, optimization of dosing regimens, and personalized medicine applications. Recent advancements in artificial intelligence (AI) and machine learning (ML) hold the potential to significantly transform QSP by enabling enhanced data extraction, fostering the development of hybrid mechanistic ML models, and supporting the introduction of surrogate models and digital twins. This manuscript explores the transformative role of AI and ML in reshaping QSP modeling workflows. AI/ML tools now enable automated literature mining, the generation of dynamic models from data, and the creation of hybrid frameworks that blend mechanistic insights with data-driven approaches. Large Language Models (LLMs) further revolutionize the field by transitioning AI/ML from merely a tool to becoming an active partner in QSP modeling. By facilitating interdisciplinary collaboration, lowering barriers to entry, and democratizing QSP workflows, LLMs empower researchers without deep coding expertise to engage in complex modeling tasks. Additionally, the integration of Artificial General Intelligence (AGI) holds the potential to autonomously propose, refine, and validate models, further accelerating innovation across multiscale biological processes. Key challenges remain in integrating AI/ML into QSP workflows, particularly in ensuring rigorous validation pipelines, addressing ethical considerations, and establishing robust regulatory frameworks to address the reliability and reproducibility of AI-assisted models. Moreover, the complexity of multiscale biological integration, effective data management, and fostering interdisciplinary collaboration present ongoing hurdles. Despite these challenges, the potential of AI/ML to enhance hybrid model development, improve model interpretability, and democratize QSP modeling offers an exciting opportunity to revolutionize drug development and therapeutic innovation. This work highlights a pathway toward a transformative era for QSP, leveraging advancements in AI and ML to address these challenges and drive innovation in the field.

Background: Glucagon-like peptide-1 receptor agonists (GLP-1RAs) play an important role in the treatment of type 2 diabetes (T2D) and obesity. The relationship between efficacy and dosing regimen has been studied extensively for this class of molecules. However, a comprehensive analysis of the translation of in vitro data to in vivo efficacious exposure is still lacking. Methods: We collected clinical pharmacokinetics for five approved GLP-1RAs to enable the simulation of exposure profiles and compared published clinical efficacy endpoints (HbA1c and body weight) with in-house in vitro potency values generated in different cell-based assays. Additionally, we investigated the correlation with target coverage, expressed as a ratio between the steady state drug exposure and unbound potency, body weight, or HbA1c reduction in patients with T2D. Results: We found that the best correlation with in vivo efficacy was seen for in vitro potency data generated in cellular assays performed in the absence of any serum albumin or using ovalbumin. Residual variability was larger using in vitro potency data generated in endogenous cell lines or in the presence of human serum albumin. For the human receptor assay with no albumin, exposures above 100-fold in vitro EC50 resulted in >1.5% point HbA1c reduction, while a 5% BW reduction was related to approximately 3× higher exposures. A similar relationship was seen in the ovalbumin assay. Conclusions: Overall, the relationship established for in vitro potency and in vivo efficacy will help to increase confidence in human dose prediction and trial design for new GLP-1RAs in the discovery and early clinical phases.

B cell-mediated immunity plays a crucial role in long-term humoral protection. However, dysregulation of B cell development and differentiation may lead to the persistence of autoreactive clones, contributing to autoimmune diseases. Despite a number of therapeutic agents in preclinical and clinical development targeting B cell biology, several challenges still limit their successful translation into clinical use. To better understand B cell-targeting mechanisms of action quantitatively and mechanistically, we developed an integrative systems pharmacology model that describes T cell-dependent B cell response to antigen exposure in mouse. The model includes 20 ordinary differential equations representing key biological processes involved in the B lymphocyte response: B cell activation in secondary lymphoid organs; antibody-secreting cell (ASC) generation; migration to the bone marrow; and redistribution to peripheral tissues. The model adequately described ASC dynamics across tissues and IgG time profiles in plasma. Local and global sensitivity analyses identified the ASC production rate as the main contributor to cell counts in the spleen and in lymph nodes, while ASC levels in the bone marrow were primarily controlled by their influx rate, reflecting survival niche availability. Moderate variations of this influx rate parameter allowed the model to capture high inter-study variability in bone marrow ASC levels, explaining the observed heterogeneity in the functional immune response. The model can be further used as a quantitative tool to study B cell responses and their dysregulation in autoimmunity. It can be extended by integrating plasma-cell biology and autoantibody production, ultimately supporting the development of new therapeutic strategies for autoimmune diseases.

FLT3L-Fc is a half-life extended, effectorless Fc-fusion of the native human FLT3-ligand. In cynomolgus monkeys, treatment with FLT3L-Fc leads to a complex pharmacokinetic/pharmacodynamic (PK/PD) relationship, with observed nonlinear PK and expansion of different immune cell types across different dose levels. A minimal physiologically based PK/PD model with expansion-enhanced target-mediated drug disposition (TMDD) was developed to integrate the molecule’s mechanism of action, as well as the complex preclinical and clinical PK/PD data, to support the preclinical-to-clinical translation of FLT3L-Fc. In addition to the preclinical PK data of FLT3L-Fc in cynomolgus monkeys, clinical PK and PD data from other FLT3-agonist molecules (GS-3583 and CDX-301) were used to inform the model and project the expansion profiles of conventional DC1s (cDC1s) and total DCs in peripheral blood. This work constitutes an essential part of our model-informed drug development (MIDD) strategy for clinical development of FLT3L-Fc by projecting PK/PD in healthy volunteers, determining the first-in-human (FIH) dose, and informing the efficacious dose in clinical settings. Model-generated results were incorporated in regulatory filings to support the rationale for the FIH dose selection.