Simulations Plus (United States)

Accurate prediction of oral drug absorption is essential for anticipating variability in bioavailability and guiding clinical development. Physiological factors, such as gastric pH, gastric emptying, and intestinal transit time, influenced by food intake and acid-reducing agents (ARAs), play critical roles in the dissolution and absorption of drugs with pH-dependent solubility. Intestinal transporters and metabolizing enzymes also affect systemic exposure by regulating drug absorption, metabolism, and excretion in the intestine. Species differences in gastrointestinal physiology, including pH profiles, transporter expression, and motility patterns, present challenges when translating nonclinical data to humans. Physiologically based pharmacokinetic (PBPK) modeling offers a mechanistic framework to address these variations by integrating drug- and species-specific parameters. By calibrating in vitro and in vivo data, PBPK models can simulate human gastrointestinal conditions and more accurately predict oral absorption dynamics, enhancing translational relevance. These models facilitate early evaluation of food and ARA impacts, transporter-related absorption variabilities, and drug-drug interactions (DDIs), supporting the rational development of formulation and dose selection. Verifying model performance in nonclinical species, commonly in dogs, can further enhance confidence in human predictions. This review highlights drug- and species-specific factors and illustrates how PBPK modeling can address uncertainties regarding drug absorption in humans using nonclinical in vitro and in vivo data. Despite challenges such as parameter uncertainty and interspecies differences, PBPK modeling remains a valuable tool for estimating drug exposure and informing biopharmaceutical and clinical pharmacology strategies.

9 Introduction: There is a global need to provide rifamycin-based tuberculosis treatments in people who may be receiving lenacapavir (LEN), a potent capsid inhibitor approved for the treatment of multidrug-resistant HIV-1 and with high efficacy as HIV-1 pre-exposure prophylaxis (PrEP). LEN is eliminated by CYP3A, UGT1A1 and P-glycoprotein (P-gp) efflux, which are induced by rifamycins. We developed a physiologically based pharmacokinetic (PBPK) model to propose dosing recommendations for LEN when rifampin (or strong CYP3A inducer) or rifabutin (or moderate CYP3A inducer) is co-administered with subcutaneous (SC) LEN. Methods: A full-body LEN PBPK model was developed in GastroPlus® (version 9.8.3). Systemic distribution was captured with a perfusion-limited model for all tissues. Pathways for LEN elimination include efflux via direct intestinal secretion and liver uptake leading to biliary secretion and potential recirculation via P-gp, in addition to hepatic and intestinal metabolism by CYP3A4, UGT1A1 and minor contribution from non-specific clearance in liver. Kinetic parameters for LEN interactions with enzymes and P-gp were fitted against in vivo human data. The contribution of different pathways to LEN elimination were validated by comparing simulations to observed data from Phase 1 study participants receiving oral LEN in combination with voriconazole (strong CYP3A inhibitor), cobicistat (strong CYP3A and P-gp inhibitor), efavirenz (moderate CYP3A and P-gp inducer) and rifampin (strong CYP3A4, UGT1A1 and P-gp inducer, and P-gp inhibitor). We simulated different scenarios of a LEN dosing regimen (oral LEN 600 mg on Days 1 and 2 plus SC LEN administered on Day 1 and every 6 months thereafter) that could maintain plasma concentrations above the inhibitory quotient-4 (fourfold greater than the in vitro protein-adjusted 95% effective concentrations derived from MT-4 cells; 15.5 ng/mL) throughout the every-six-months interval, regardless of when rifampin/rifabutin (moderate CYP3A/P-gp inducer) treatment was initiated. Conclusions: With supplemental dosage, LEN can be co-administered with strong or moderate CYP3A inducers including rifampin and rifabutin for tuberculosis treatment.

Background: Accurately determining pediatric dosing is essential prior to initiating clinical trials or administering medications in routine clinical settings. In children, ethical considerations demand careful evaluation of both safety and effectiveness. Typically, dosing recommendations for therapeutic proteins, such as monoclonal antibodies (mAbs), are derived from adult dosages using body weight as a scaling factor. However, this method overlooks key physiological and biochemical distinctions between pediatric and adult patients. Therefore, this could lead to the underexposure of mAbs, limiting their efficacy in this population. Additional methods are necessary to predict pediatric doses mechanistically. For small molecules, physiologically based pharmacokinetic (PBPK) models have been extensively used to predict pediatric doses based on physiological age-related changes and enzymes/transporters ontogeny. This study aims to evaluate the ability of PBPK models to predict mAbs’ pediatric exposure. Methods: Three mAbs were used for model development and validation: bevacizumab, infliximab, and atezolizumab. The PBPK models were built using GastroPlus© Biologics module. For each mAb, the PBPK model was developed based on observed data in healthy and/or patient adults. Then, the physiological parameters were scaled to describe the pediatric physiology to predict exposure to the pediatric populations. Predicted plasma concentration–time courses were overlaid with reported observed data to assess the ability of the PBPK model to predict pediatric exposure. Results: Results showed that PBPK models accurately predicted pediatric pharmacokinetics for mAbs. Conclusions: This research marks a significant step in validating mechanistic extrapolation methods for biologics exposure prediction in children using PBPK models.

Chaturbedi et al.1 used the validated translational model of synthetic opioid overdose from Mann et al.2 to evaluate the effectiveness of two FDA-approved intranasal opioid antagonists, nalmefene and naloxone, in a community setting. Using formulations of intranasal nalmefene and naloxone with different pharmacokinetic profiles,1 the authors emphasize the criticality of rapid absorption for an antagonist to reverse an overdose produced by potentially lethal doses of fentanyl and carfentanil. They also conclude that intranasal 4 mg naloxone hydrochloride and 3 mg nalmefene hydrochloride (2.7 mg base) are similarly effective based on predicted rates of cardiac arrest.1 These findings differ substantially from previous work3 conducted using the same model2 and demonstrating large and clinically meaningful reductions in the incidence of cardiac arrest with intranasal nalmefene compared to naloxone. This apparent discrepancy is explained by the source of pharmacokinetic data used to evaluate the FDA-approved formulation of nalmefene. While previous work3 utilized nalmefene plasma concentrations from subjects breathing room air, Chaturbedi et al.1 used data (figure 2C) measured in subjects breathing a hypercapnic gas mixture during a pharmacodynamic study.4 Breathing a hypercapnic gas mixture markedly impacted the pharmacokinetics of intranasal nalmefene, with a 35% lower absorption rate3 manifested in a delayed and 50% lower peak concentration compared to subjects breathing room air4, 5 (Table 1). The pharmacokinetic profile in subjects breathing room air was replicated in independent study cohorts5 and constitutes the reference pharmacokinetic profile of intranasal nalmefene in the product label. These data were used to construct a robust population pharmacokinetic model of intranasal nalmefene in 153 subjects.3 Applying this pharmacokinetic model to the model of synthetic opioid overdose developed by Mann et al.,2 large and clinically meaningful differences in the effectiveness of intranasal naloxone and nalmefene were observed across all dosing scenarios.3 For example, following a 1.63-mg intravenous fentanyl dose resulting in a cardiac arrest in 52.1% (95% confidence interval, 47.3–56.8) of simulated subjects absent intervention, intranasal naloxone, and nalmefene reduced this percentage to 19.2% (15.5–23.3) and 2.2% (1.0–3.8), respectively.3 The robust nature of these simulations is underscored by an incidence of cardiac arrest following intranasal naloxone3 similar to the value reported by Chaturbedi et al.1 using an independent pharmacokinetic dataset. Overall, these findings raise issues about the interpretation and validity of results using pharmacokinetic data for intranasal nalmefene that do not mirror the conditions encountered in a “real world” overdose.

Model Master Files (MMFs) offer a much needed approach to integrating computational modelling into drug development and regulatory frameworks, supporting the growth of quantitative medicine. By acting as confidential repositories for validated models, MMFs enable streamlined submissions, model reuse, and context-specific reviews while safeguarding intellectual property. For technology companies (such as software providers), MMFs provide a structured pathway to engage with the FDA, align innovations with regulatory standards, and expand the use of models across diverse applications. A challenge with the current framework is the need to provide the same validation and verification information to multiple drug companies each time the submit an application. With an MMF in place, drug companies can refer to this same document which the technology provider can add to over time. However, challenges persist, including limited direct interaction with the FDA outside (A)NDA submissions and the need for consistent model validation and version management. Addressing these issues through enhanced collaboration and clear guidelines will maximize the potential of MMFs, fostering broader adoption of modelling and simulation in drug development and advancing personalized medicine.

Abstract Want to be published in Significance? Our data visualisation guide, created by Andreas Krause and Nicola Rennie, offers contributors insights, advice, and examples (with code) to make data outputs more readable, accessible, and impactful. Over the next few issues, we will publish it section by section until you have a complete guide to creating the best data visualisation for the needs of both your article and Significance magazine

Quantitative in silico tools may be leveraged to mechanistically predict the dermato-pharmacokinetics of compounds delivered from topical and transdermal formulations by integrating systems of rate equations that describe permeation through the formulation and layers of skin and pilo-sebaceous unit, and exchange with systemic circulation via local blood flow. Delivery of clobetasol-17 propionate (CP) from DermovateTM cream was simulated using the Transdermal Compartmental Absorption & Transit (TCATTM) Model in GastroPlus®. The cream was treated as an oil-in-water emulsion, with model input parameters estimated from publicly available information and quantitative structure-permeation relationships. From the ranges of values available for model input parameters, a set of parameters was selected by comparing model outputs to CP dermis concentration-time profiles measured by dermal open-flow microperfusion (Bodenlenz et al. Pharm Res. 33(9):2229–38, 2016). Predictions of unbound dermis CP concentrations were reasonably accurate with respect to time and skin depth. Parameter sensitivity analyses revealed considerable dependence of dermis CP concentration profiles on drug solubility in the emulsion, relatively less dependence on dispersed phase volume fraction and CP effective diffusivity in the continuous phase of the emulsion, and negligible dependence on dispersed phase droplet size. Effects of evaporative water loss from the cream and corticosteroid-induced vasoconstriction were also assessed. This work illustrates the applicability of computational modeling to predict sensitivity of dermato-pharmacokinetics to changes in thermodynamic and transport properties of a compound in a topical formulation, particularly in relation to rate-limiting steps in skin permeation. Where these properties can be related to formulation composition and processing, such a computational approach may support the design of topically applied formulations.

This study was designed to verify a virtual population representing patients with nonalcoholic fatty liver disease (NAFLD) to support the implementation of a physiologically based pharmacokinetic (PBPK) modeling approach for prediction of disease-related changes in drug pharmacokinetics. A virtual NAFLD patient population was developed in GastroPlus (v.9.8.2) by accounting for pathophysiological changes associated with the disease and proteomics-informed alterations in the abundance of metabolizing enzymes and transporters pertinent to drug disposition. The NAFLD population model was verified using exemplar drugs where elimination is influenced predominantly by cytochrome P450 (CYP) enzymes (chlorzoxazone, caffeine, midazolam, pioglitazone) or by transporters (rosuvastatin, 11C-metformin, morphine and the glucuronide metabolite of morphine). PBPK model predictions of plasma concentrations of all the selected drugs and hepatic radioactivity levels of 11C-metformin were consistent with the clinically-observed data. Importantly, the PBPK simulations using the virtual NAFLD population model provided reliable estimates of the extent of changes in key pharmacokinetic parameters for the exemplar drugs, with mean predicted ratios (NAFLD patients divided by healthy individuals) within 0.80- to 1.25-fold of the clinically-reported values, except for midazolam (prediction-fold difference of 0.72). A virtual NAFLD population model within the PBPK framework was successfully developed with good predictive capability of estimating disease-related changes in drug pharmacokinetics. This supports the use of a PBPK modeling approach for prediction of the pharmacokinetics of new investigational or repurposed drugs in patients with NAFLD and may help inform dose adjustments for drugs commonly used to treat comorbidities in this patient population.