Another string to your bow: machine learning prediction of the pharmacokinetic properties of small molecules.

INTRODUCTION: Prediction of pharmacokinetic (PK) properties is crucial for drug discovery and development. Machine-learning (ML) models, which use statistical pattern recognition to learn correlations between input features (such as chemical structures) and target variables (such as PK parameters), are being increasingly used for this purpose. To embed ML models for PK prediction into workflows and to guide future development, a solid understanding of their applicability, advantages, limitations, and synergies with other approaches is necessary. AREAS COVERED: This narrative review discusses the design and application of ML models to predict PK parameters of small molecules, especially in light of established approaches including in vitro-in vivo extrapolation (IVIVE) and physiologically based pharmacokinetic (PBPK) models. The authors illustrate scenarios in which the three approaches are used and emphasize how they enhance and complement each other. In particular, they highlight achievements, the state of the art and potentials of applying machine learning for PK prediction through a comphrehensive literature review. EXPERT OPINION: ML models, when carefully crafted, regularly updated, and appropriately used, empower users to prioritize molecules with favorable PK properties. Informed practitioners can leverage these models to improve the efficiency of drug discovery and development process.

Physiologically based pharmacokinetic modeling to predict the impact of liver cirrhosis on glucuronidation via UGT1A4 & UGT2B7/2B4 - a case study with midazolam.

Hepatic impairment, due to liver cirrhosis, decreases the activity of cytochrome P450 enzymes (CYPs). The use of physiologically-based pharmacokinetic (PBPK) modeling to predict this effect for CYP substrates has been well-established, but the effect of cirrhosis on uridine-glucuronosyltransferase (UGT) activities is less studied and few PBPK models have been reported. UGT enzymes are involved in primary N-glucuronidation of midazolam and glucuronidation of 1'-OH-midazolam following CYP3A hydroxylation. In this study Simcyp® was used to establish PBPK models for midazolam, its primary metabolites midazolam-N-glucuronide (UGT1A4) and 1'-OH midazolam (CYP3A4/3A5) and the secondary metabolite 1'-OH-midazolam-O-glucuronide (UGT2B7/2B4), allowing to simulate the impact of liver cirrhosis on the primary and secondary glucuronidation of midazolam. The model was verified in non-cirrhotic subjects before extrapolation to cirrhotic patients of Child-Pugh (CP) classes A, B, and C. Our model successfully predicted the exposures of midazolam and its metabolites in non-cirrhotic and cirrhotic patients, with 86% of observed plasma concentrations within 5th-95th percentiles of predictions and observed geometrical mean of AUCinf and Cmax within 0.7-1.43-fold of predictions. The simulated metabolic ratio (AUCglucuronide/AUCparent, MR), was calculated for midazolam-N-glucuronide to midazolam (indicative of UGT1A4 activity) and decreased by 40% (CP A), 48% (CP B), and 75% (CP C). For 1'-OH-midazolam-O-glucuronide to 1'-OH-midazolam the MR (indicative of UGT2B7/2B4 activity) dropped by 35% (CP A), 51% (CP B), and 64% (CP C). These predicted MRs were corroborated by the observed data. This work thus increases confidence in Simcyp® predictions of the effect of liver cirrhosis on the pharmacokinetics of UGT1A4 and UGT2B7/UGT2B4 substrates. Significance Statement This paper presents a PBPK model for midazolam and its metabolites and verifies the accurate simulation of pharmacokinetic profiles when using the Simcyp® hepatic impairment population models. Exposure changes of midazolam-N-glucuronide and 1'-OH-midazolam-O-glucuronide reflect the impact of decreases in UGT1A4 and UGT2B7/2B4 glucuronidation activity in cirrhotic patients. The approach used in this study may be extended to verify the modeling of other UGT enzymes affected by liver cirrhosis.

Mechanistic Modeling of In Vitro Biopharmaceutic Data for a Weak Acid Drug: A Pathway Towards Deriving Fundamental Parameters for Physiologically Based Biopharmaceutic Modeling.

Mechanistic modeling of in vitro experiments using metabolic enzyme systems enables the extrapolation of metabolic clearance for in vitro-in vivo predictions. This is particularly important for successful clearance predictions using physiologically based pharmacokinetic (PBPK) modeling. The concept of mechanistic modeling can also be extended to biopharmaceutics, where in vitro data is used to predict the in vivo pharmacokinetic profile of the drug. This approach further allows for the identification of parameters that are critical for oral drug absorption in vivo. However, the routine use of this analysis approach has been hindered by the lack of an integrated analysis workflow. The objective of this tutorial is to (1) review processes and parameters contributing to oral drug absorption in increasing levels of complexity, (2) outline a general physiologically based biopharmaceutic modeling workflow for weak acids, and (3) illustrate the outlined concepts via an ibuprofen (i.e., a weak, poorly soluble acid) case example in order to provide practical guidance on how to integrate biopharmaceutic and physiological data to better understand oral drug absorption. In the future, we plan to explore the usefulness of this tutorial/roadmap to inform the development of PBPK models for BCS 2 weak bases, by expanding the stepwise modeling approach to accommodate more intricate scenarios, including the presence of diprotic basic compounds and acidifying agents within the formulation.

In silico modeling and simulation of organ-on-a-chip systems to support data analysis and a priori experimental design.

Organ-on-a-chip (OoC) systems are a promising new class of in vitro devices that can combine various tissues, cultured in different compartments, linked by media flow. The properties of these novel in vitro systems linked to increased physiological relevance of culture conditions may lead to more in vivo-relevant cell phenotypes, enabling better in vitro pharmacology and toxicology assessment. Improved cell activities combined with longer lasting cultures offer opportunities to improve the characterization of absorption, distribution, metabolism, and excretion (ADME) processes, potentially leading to more accurate prediction of human pharmacokinetics (PKs). The inclusion of barrier tissue elements and metabolically competent tissue types results in complex concentration-time profiles (in vitro PK) for test drugs and their metabolites that require appropriate mathematical modeling of in vitro data for parameter estimation. In particular, modeling is critical to estimate in vitro ADME parameters when multiple different tissues are combined in a single device. Therefore, sophisticated in silico data analysis and a priori experimental design are highly recommended for OoC experiments in a manner not needed with standard ADME screening. The design of the experiment should be optimized based on an investigation of the structural characteristics of the in vitro system, the ADME features of the test compound and any available knowledge of cell phenotypes. This tutorial aims to provide such a modeling framework to inform experimental design and refine parameter estimation in a Gut-Liver OoC (the most studied multi-organ systems to predict the oral drug PKs) to improve translatability of data generated in such complex cellular systems.

Evaluation of the potential impact on pharmacokinetics of various cytochrome P450 substrates of increasing IL-6 levels following administration of the T-cell bispecific engager glofitamab.

Glofitamab is a novel T cell bispecific antibody developed for treatment of relapsed-refractory diffuse large B cell lymphoma and other non-Hodgkin's lymphoma indications. By simultaneously binding human CD20-expressing tumor cells and CD3 on T cells, glofitamab induces tumor cell lysis, in addition to T-cell activation, proliferation, and cytokine release. Here, we describe physiologically-based pharmacokinetic (PBPK) modeling performed to assess the impact of glofitamab-associated transient increases in interleukin 6 (IL-6) on the pharmacokinetics of several cytochrome P450 (CYP) substrates. By refinement of a previously described IL-6 model and inclusion of in vitro CYP suppression data for CYP3A4, CYP1A2, and 2C9, a PBPK model was established in Simcyp to capture the induced IL-6 levels seen when glofitamab is administered at the intended dose and dosing regimen. Following model qualification, the PBPK model was used to predict the potential impact of CYP suppression on exposures of various CYP probe substrates. PBPK analysis predicted that, in the worst-case, the transient elevation of IL-6 would increase exposures of CYP3A4, CYP2C9, and CYP1A2 substrates by less than or equal to twofold. Increases for CYP3A4, CYP2C9, and CYP1A2 substrates were projected to be 1.75, 1.19, and 1.09-fold following the first administration and 2.08, 1.28, and 1.49-fold following repeated administrations. It is recommended that there are no restrictions on concomitant treatment with any other drugs. Consideration may be given for potential drug-drug interaction during the first cycle in patients who are receiving concomitant CYP substrates with a narrow therapeutic index via monitoring for toxicity or for drug concentrations.

Development and application of neonatal physiology-based pharmacokinetic models of amikacin and fosfomycin to assess pharmacodynamic target attainment.

Antimicrobial resistance increasingly complicates neonatal sepsis in a global context. Fosfomycin and amikacin are two agents being tested in an ongoing multicenter neonatal sepsis trial. Although neonatal pharmacokinetics (PKs) have been described for these drugs, the physiological variability within neonatal populations makes population PKs in this group uncertain. Physiologically-based pharmacokinetic (PBPK) models were developed in Simcyp for fosfomycin and amikacin sequentially for adult, pediatric, and neonatal populations, with visual and quantitative validation compared to observed data at each stage. Simulations were performed using the final validated neonatal models to determine drug exposures for each drug across a demographic range, with probability of target attainment (PTA) assessments. Successfully validated neonatal PBPK models were developed for both fosfomycin and amikacin. PTA analysis demonstrated high probability of target attainment for amikacin 15 mg/kg i.v. q24h and fosfomycin 100 mg/kg (in neonates aged 0-7 days) or 150 mg/kg (in neonates aged 7-28 days) i.v. q12h for Enterobacterales with fosfomycin and amikacin minimum inhibitory concentrations at the adult breakpoints. Repeat analysis in premature populations demonstrated the same result. PTA analysis for a proposed combination fosfomycin-amikacin target was also performed. The simulated regimens, tested in a neonatal sepsis trial, are likely to be adequate for neonates across different postnatal ages and gestational age. This work demonstrates a template for determining target attainment for antimicrobials (alone or in combination) in special populations without sufficient available PK data to otherwise assess with traditional pharmacometric methods.

PBPK Modelling for Drugs Cleared by Non-CYP Enzymes: State-of-the-Art and Future Perspectives.

Physiologically-based pharmacokinetic (PBPK) modeling has become the established method for predicting human pharmacokinetics (PK) and drug-drug interactions (DDI). The number of drugs cleared by non-CYP enzyme metabolism has increased steadily and to date, there is no consolidated overview of PBPK modeling for drugs cleared by non-CYP enzymes. This review aims to describe the state-of-the-art for PBPK modeling for drugs cleared via non-CYP enzymes, to identify successful strategies, to describe gaps and to provide suggestion to overcome them. To this end, we conducted a detailed literature search and found 58 articles published before the 1st of January 2023 containing 95 examples of clinical PBPK models for 62 non-CYP enzyme substrates. Reviewed articles covered the drug clearance by uridine 5'-diphospho-glucuronosyltransferases (UGTs), aldehyde oxidase (AO), flavin-containing monooxygenases (FMOs), sulfotransferases (SULTs) and carboxylesterases (CES), with UGT2B7, UGT1A9, CES1, FMO3 and AO being the enzymes most frequently involved. In vitro-in vivo extrapolation (IVIVE) of intrinsic clearance and the bottom-up PBPK modeling involving non-CYP enzymes remains challenging. We observed that the middle-out modeling approach was applied in 80% of the cases, with metabolism parameters optimized in 73% of the models. Our review could not identify a standardized approach used for model optimization based on clinical data, with manual optimization employed most frequently. Successful development of models for UGT2B7, UGT1A9, CES1, and FMO3 substrates provides a foundation for other drugs metabolized by these enzymes and guides the way forward in creating PBPK models for other enzymes in these families. Significance Statement Our review charts the rise of PBPK modeling for drugs cleared by non-CYP enzymes. Analyzing 58 articles and 62 non-CYP enzyme substrates, we found that UGTs, AO, FMOs, SULTs, and CES were the main enzyme families involved and that UGT2B7, UGT1A9, CES1, FMO3 and AO are the individual enzymes with the strongest PBPK modeling precedents. Approaches established for these enzymes can now be extended to additional substrates and to drugs metabolized by enzymes that are similarly well characterized.

Investigating Tacrolimus Disposition in Paediatric Patients with a Physiologically Based Pharmacokinetic Model Incorporating CYP3A4 Ontogeny, Mechanistic Absorption and Red Blood Cell Binding.

Tacrolimus is a crucial immunosuppressant for organ transplant patients, requiring therapeutic drug monitoring due to its variable exposure after oral intake. Physiologically based pharmacokinetic (PBPK) modelling has provided insights into tacrolimus disposition in adults but has limited application in paediatrics. This study investigated age dependency in tacrolimus exposure at the levels of absorption, metabolism, and distribution. Based on the literature data, a PBPK model was developed to predict tacrolimus exposure in adults after intravenous and oral administration. This model was then extrapolated to the paediatric population, using a unique reference dataset of kidney transplant patients. Selecting adequate ontogeny profiles for hepatic and intestinal CYP3A4 appeared critical to using the model in children. The best model performance was achieved by using the Upreti ontogeny in both the liver and intestines. To mechanistically evaluate the impact of absorption on tacrolimus exposure, biorelevant in vitro solubility and dissolution data were obtained. A relatively fast and complete release of tacrolimus from its amorphous formulation was observed when mimicking adult or paediatric dissolution conditions (dose, fluid volume). In both the adult and paediatric PBPK models, the in vitro dissolution profiles could be adequately substituted by diffusion-layer-based dissolution modelling. At the level of distribution, sensitivity analysis suggested that differences in blood plasma partitioning of tacrolimus may contribute to the variability in exposure in paediatric patients.

Correction: Van der Veken et al. Gastrointestinal Fluid Volumes in Pediatrics: A Retrospective MRI Study. Pharmaceutics 2022, 14, 1935.

In the original publication [...].

A Machine Learning Framework to Improve Rat Clearance Predictions and Inform Physiologically Based Pharmacokinetic Modeling.

During drug discovery and development, achieving appropriate pharmacokinetics is key to establishment of the efficacy and safety of new drugs. Physiologically based pharmacokinetic (PBPK) models integrating in vitro-to-in vivo extrapolation have become an essential in silico tool to achieve this goal. In this context, the most important and probably most challenging pharmacokinetic parameter to estimate is the clearance. Recent work on high-throughput PBPK modeling during drug discovery has shown that a good estimate of the unbound intrinsic clearance (CLint,u,) is the key factor for useful PBPK application. In this work, three different machine learning-based strategies were explored to predict the rat CLint,u as the input into PBPK. Therefore, in vivo and in vitro data was collected for a total of 2639 proprietary compounds. The strategies were compared to the standard in vitro bottom-up approach. Using the well-stirred liver model to back-calculate in vivo CLint,u from in vivo rat clearance and then training a machine learning model on this CLint,u led to more accurate clearance predictions (absolute average fold error (AAFE) 3.1 in temporal cross-validation) than the bottom-up approach (AAFE 3.6-16, depending on the scaling method) and has the advantage that no experimental in vitro data is needed. However, building a machine learning model on the bias between the back-calculated in vivo CLint,u and the bottom-up scaled in vitro CLint,u also performed well. For example, using unbound hepatocyte scaling, adding the bias prediction improved the AAFE in the temporal cross-validation from 16 for bottom-up to 2.9 together with the bias prediction. Similarly, the log Pearson r2 improved from 0.1 to 0.29. Although it would still require in vitro measurement of CLint,u., using unbound scaling for the bottom-up approach, the need for correction of the fu,inc by fu,p data is circumvented. While the above-described ML models were built on all data points available per approach, it is discussed that evaluation comparison across all approaches could only be performed on a subset because ca. 75% of the molecules had missing or unquantifiable measurements of the fraction unbound in plasma or in vitro unbound intrinsic clearance, or they dropped out due to the blood-flow limitation assumed by the well-stirred model. Advantageously, by predicting CLint,u as the input into PBPK, existing workflows can be reused and the prediction of the in vivo clearance and other PK parameters can be improved.

Streamlining Food Effect Assessment - Are Repeated Food Effect Studies Needed? An IQ Analysis.

Current regulatory guidelines on drug-food interactions recommend an early assessment of food effect to inform clinical dosing instructions, as well as a pivotal food effect study on the to-be-marketed formulation if different from that used in earlier trials. Study waivers are currently only granted for BCS class 1 drugs. Thus, repeated food effect studies are prevalent in clinical development, with the initial evaluation conducted as early as the first-in-human studies. Information on repeated food effect studies is not common in the public domain. The goal of the work presented in this manuscript from the Food Effect PBPK IQ Working Group was to compile a dataset on these studies across pharmaceutical companies and provide recommendations on their conduct. Based on 54 studies collected, we report that most of the repeat food effect studies do not result in meaningful differences in the assessment of the food effect. Seldom changes observed were more than twofold. There was no clear relationship between the change in food effect and the formulation change, indicating that in most cases, once a compound is formulated appropriately within a specific formulation technology, the food effect is primarily driven by inherent compound properties. Representative examples of PBPK models demonstrate that following appropriate validation of the model with the initial food effect study, the models can be applied to future formulations. We recommend that repeat food effect studies should be approached on a case-by-case basis taking into account the totality of the evidence including the use of PBPK modeling.

The Activity of Members of the UDP-Glucuronosyltransferase Subfamilies UGT1A and UGT2B is Impaired in Patients with Liver Cirrhosis.

BACKGROUND AND OBJECTIVE: The impact of liver cirrhosis on the activity of UDP-glucuronosyltransferases (UGTs) is currently not well characterized. We investigated the glucuronidation capacity and glucuronide accumulation in patients with liver cirrhosis. METHODS: We administered the Basel phenotyping cocktail (caffeine, efavirenz, flurbiprofen, omeprazole, metoprolol, midazolam) to patients with liver cirrhosis (n = 16 Child A, n = 15 Child B, n = 5 Child C) and n = 12 control subjects and obtained pharmacokinetic profiles of substrates and primary metabolites and their glucuronides. RESULTS: Caffeine and its metabolite paraxanthine were only slightly glucuronidated. The metabolic ratio (AUCglucuronide/AUCparent, MR) was not affected for caffeine but decreased by 60% for paraxanthine glucuronide formation in Child C patients. Efavirenz was not glucuronidated whereas 8-hydroxyefavirenz was efficiently glucuronidated. The MR of 8-hydroxyefavirenz-glucuronide formation increased three-fold in Child C patients and was negatively correlated with the glomerular filtration rate. Flurbiprofen and omeprazole were not glucuronidated. 4-Hydroxyflurbiprofen and 5-hydroxyomeprazole were both glucuronidated but the corresponding MRs for glucuronide formation were not affected by liver cirrhosis. Metoprolol, but not α-hydroxymetoprolol, was glucuronidated, and the MR for metoprolol-glucuronide formation dropped by 60% in Child C patients. Both midazolam and its metabolite 1'-hydroxymidazolam underwent glucuronidation, and the corresponding MRs for glucuronide formation dropped by approximately 80% in Child C patients. No relevant glucuronide accumulation occurred in patients with liver cirrhosis. CONCLUSIONS: Detailed analysis revealed that liver cirrhosis may affect the activity of UGTs of the UGT1A and UGT2B subfamilies according to liver function. Clinically significant glucuronide accumulation did not occur in the population investigated. CLINICAL TRIAL REGISTRATION: NCT03337945.

Drug Design and Success of Prospective Mouse In Vitro-In Vivo Extrapolation (IVIVE) for Predictions of Plasma Clearance (CLp) from Hepatocyte Intrinsic Clearance (CLint).

Hepatocyte intrinsic clearance (CLint) and methods of in vitro-in vivo extrapolation (IVIVE) are often used to predict plasma clearance (CLp) in drug discovery. While the prediction success of this approach is dependent on the chemotype, specific molecular properties and drug design features that govern these outcomes are poorly understood. To address this challenge, we investigated the success of prospective mouse CLp IVIVE across 2142 chemically diverse compounds. Dilution scaling, which assumes that the free fraction in hepatocyte incubations (fu,inc) is governed by binding to the 10% of serum in the incubation medium, was used as our default CLp IVIVE approach. Results show that predictions of CLp are better for smaller (molecular weight (MW) < 500 Da), less polar (total polar surface area (TPSA) < 100 Å2, hydrogen bond donor (HBD) ≤1, hydrogen bond acceptor (HBA) ≤ 6), lipophilic (log D > 3), and neutral compounds, with low HBD count playing the key role. If compounds are classified according to their chemical space, predictions were good for compounds resembling central nervous system (CNS) drugs [average absolute fold error (AAFE) of 2.05, average fold error (AFE) of 0.90], moderate for classical druglike compounds (according to Lipinski, Veber, and Ghose guidelines; AAFE of 2.55; AFE of 0.68), and poor for nonclassical "beyond the rule of 5" compounds (AAFE of 3.31; AFE of 0.41). From the perspective of measured druglike properties, predictions of CLp were better for compounds with moderate-to-high hepatocyte CLint (>10 µL/min/106 cells), high passive cellular permeability (Papp > 100 nm/s), and moderate observed CLp (5-50 mL/min/kg). Influences of plasma protein binding (fu,p) and P-glycoprotein (Pgp) apical efflux ratio (AP-ER) were less pronounced. If the extended clearance classification system (ECCS) is applied, predictions were good for class 2 (Papp > 50 nm/s; neutral or basic; AAFE of 2.35; AFE of 0.70) and acceptable for class 1A compounds (AAFE of 2.98; AFE of 0.70). Classes 1B, 3 A/B, and 4 showed poor outcomes (AAFE > 3.80; AFE < 0.60). Functional groups trending toward weaker CLp IVIVE were esters, carbamates, sulfonamides, carboxylic acids, ketones, primary and secondary amines, primary alcohols, oxetanes, and compounds liable to aldehyde oxidase metabolism, likely due to multifactorial reasons. Multivariate analysis showed that multiple properties are relevant, combining together to define the overall success of CLp IVIVE. Our results indicate that the current practice of prospective CLp IVIVE is suitable only for CNS-like compounds and well-behaved classical druglike space (e.g., high permeability or ECCS class 2) without challenging functional groups. Unfortunately, based on existing mouse data, prospective CLp IVIVE for complex and nonclassical chemotypes is poor and hardly better than random guessing. This is likely due to complexities such as extrahepatic metabolism and transporter-mediated disposition which are poorly captured by this methodology. With small-molecule drug discovery increasingly evolving toward nonclassical and complex chemotypes, existing CLp IVIVE methodology will require improvement. While empirical correction factors may bridge the gap in the near future, improved and new in vitro assays, data integration models, and machine learning (ML) methods are increasingly needed to address this challenge and reduce the number of nonclinical pharmacokinetic (PK) studies.

Dose Adjustment in Patients with Liver Cirrhosis - Comparison of Two Different Modeling Approaches.

Failure to perform adequate dose adjustment in patients with liver cirrhosis may be associated with increased toxicity. We compared the prediction of area under the curve (AUC) and clearance for the six compounds of the Basel phenotyping cocktail (caffeine, efavirenz, flurbiprofen, omeprazole, metoprolol, and midazolam) using a well-known physiology-based pharmacokinetic approach (physiologically-based pharmacokinetic [PBPK] approach, Simcyp) and a novel top-down method based on the systemic clearance in healthy volunteers adjusted for markers of liver and renal dysfunction ("top-down approach"). With few exceptions, plasma concentration-time curves were accurately predicted by the PBPK approach. In comparison to the measured AUC and clearance of these drugs in patients with liver cirrhosis and healthy controls, except for efavirenz, the estimates of both approaches were within two standard deviations of the mean for total and free drug concentrations. For both approaches, a correction factor for dose adjustment in patients with liver cirrhosis could be calculated for the drugs administered. AUCs calculated using the adjusted doses were comparable to the AUCs measured in control subjects, with slightly more accurate predictions generated by the PBPK approach. For drugs with a free fraction < 50%, predictions using free drug concentrations were more accurate than with total drug concentrations. In conclusion, both methods provided good qualitative predictions of the changes by liver cirrhosis in the pharmacokinetics of the six compounds investigated. The top-down approach is easier to implement but the PBPK approach predicted changes in drug exposure more accurately than the top-down approach and provided reliable estimates for plasma concentrations.

Entrectinib dose confirmation in pediatric oncology patients: pharmacokinetic considerations.

PURPOSE: Entrectinib is a central nervous system-active potent inhibitor of tropomyosin receptor kinase (TRK), with anti-tumor activity against neurotrophic NTRK gene fusion-positive tumors. This study investigates the pharmacokinetics of entrectinib and its active metabolite (M5) in pediatric patients and aims to understand whether the pediatric dose of 300 mg/m2 once daily (QD) provides an exposure that is consistent with the approved adult dose (600 mg QD). METHODS: Forty-three patients aged from birth to 22 years were administered entrectinib (250-750 mg/m2 QD) orally with food in 4-week cycles. Entrectinib formulations included capsules without acidulant (F1) and capsules with acidulant (F2B and F06). RESULTS: Although there was interpatient variability with F1, entrectinib and M5 exposures increased dose dependently. Lower systemic exposures were observed in pediatric patients receiving 400 mg/m2 QD entrectinib (F1) versus adults receiving either the same dose/formulation or the recommended flat dose of 600 mg QD (~ 300 mg/m2 for a 70 kg adult) due to suboptimal F1 performance in the pediatric study. The observed pediatric exposures following 300 mg/m2 QD entrectinib (F06) were comparable to those in adults receiving 600 mg QD. CONCLUSIONS: Overall, the F1 formulation of entrectinib was associated with lower systemic exposure in pediatric patients compared with the commercial acidulant formulation (F06). Systemic exposures achieved in pediatric patients with the F06 recommended dose (300 mg/m2) were within the known efficacious range in adults, confirming the adequacy of the recommended dose regimen with the commercial formulation.

Regulatory utility of physiologically based pharmacokinetic modeling for assessing food impact in bioequivalence studies: A workshop summary report.

This workshop report summarizes the presentations and panel discussion related to the use of physiologically based pharmacokinetic (PBPK) modeling approaches for food effect assessment, collected from Session 2 of Day 2 of the workshop titled "Regulatory Utility of Mechanistic Modeling to Support Alternative Bioequivalence Approaches." The US Food and Drug Administration in collaboration with the Center for Research on Complex Generics organized this workshop where this particular session titled "Oral PBPK for Evaluating the Impact of Food on BE" presented successful cases of PBPK modeling approaches for food effect assessment. Recently, PBPK modeling has started to gain popularity among academia, industries, and regulatory agencies for its potential utility during bioavailability (BA) and/or bioequivalence (BE) studies of new and generic drug products to assess the impact of food on BA/BE. Considering the promises of PBPK modeling in generic drug development, the aim of this workshop session was to facilitate knowledge sharing among academia, industries, and regulatory agencies to understand the knowledge gap and guide the path forward. This report collects and summarizes the information presented and discussed during this session to disseminate the information into a broader audience for further advancement in this area.

Physiologically Based Biopharmaceutics Modeling of Food Effect for Basmisanil: A Retrospective Case Study of the Utility for Formulation Bridging.

Basmisanil, is a lipophilic drug substance, exhibiting poor solubility and good permeability (BCS class 2). A validated physiologically based biopharmaceutics model (PBBM) has been previously described for tablets dosed in the fed state. The PBBM captured the less than proportional increases in exposure at higher doses well and indicated that absorption was dissolution rate-limited below 200 mg while solubility was limiting for higher doses. In this study, a model for dosing in the fasted state is described and is verified for simulation of the food effect where exposures were ~1.5 fold higher when a 660 mg tablet was given with food. The model is then applied to simulate the food effect for a granules formulation given at a lower dose (120 mg). The food effect at the lower dose was reasonably simulated with a ratio of simulated/observed food effect of 1.35 for Cmax and 0.83 for AUC. Sensitivity analysis was carried out for uncertain model parameters to confirm that the model could predict the magnitude of the positive food effect with moderate to high confidence. This study suggests that a verified PBBM can provide a useful alternative to a repeat food effect study when formulation changes are minor. However, there is need for further evaluation of the approach and a definition of what formulation changes are minor in this context. In addition, this work highlights some uncertainties in the handling of solubility in PBBM, in particular around temperature dependency of solubility and the parameterization of bile salt solubilization using measurements in biorelevant media.

Quantitative Cytochrome P450 3A4 Induction Risk Assessment Using Human Hepatocytes Complemented with Pregnane X Receptor-Activating Profiles.

Reliable in vitro to in vivo translation of cytochrome P450 (CYP) 3A4 induction potential is essential to support risk mitigation for compounds during pharmaceutical discovery and development. In this study, a linear correlation of CYP3A4 mRNA induction potential in human hepatocytes with the respective pregnane-X receptor (PXR) activation in a reporter gene assay using DPX2 cells was successfully demonstrated for 13 clinically used drugs. Based on this correlation, using rifampicin as a positive control, the magnitude of CYP3A4 mRNA induction for 71 internal compounds at several concentrations up to 10 µM (n = 90) was predicted within 2-fold error for 64% of cases with only a few false positives (19%). Furthermore, the in vivo area under the curve reduction of probe CYP substrates was reasonably predicted for eight marketed drugs (carbamazepine, dexamethasone, enzalutamide, nevirapine, phenobarbital, phenytoin, rifampicin, and rufinamide) using the static net effect model using both the PXR activation and CYP3A4 mRNA induction data. The liver exit concentrations were used for the model in place of the inlet concentrations to avoid false positive predictions and the concentration achieving twofold induction (F2) was used to compensate for the lack of full induction kinetics due to cytotoxicity and solubility limitations in vitro. These findings can complement the currently available induction risk mitigation strategy and potentially influence the drug interaction modeling work conducted at clinical stages. SIGNIFICANCE STATEMENT: The established correlation of CYP3A4 mRNA in human hepatocytes to PXR activation provides a clear cut-off to identify a compound showing an in vitro induction risk, complementing current regulatory guidance. Also, the demonstrated in vitro-in vivo translation of induction data strongly supports a clinical development program although limitations remain for drug candidates showing complex disposition pathways, such as involvement of auto-inhibition/induction, active transport and high protein binding.

Can We Predict Clinical Pharmacokinetics of Highly Lipophilic Compounds by Integration of Machine Learning or In Vitro Data into Physiologically Based Models? A Feasibility Study Based on 12 Development Compounds.

While high lipophilicity tends to improve potency, its effects on pharmacokinetics (PK) are complex and often unfavorable. To predict clinical PK in early drug discovery, we built human physiologically based PK (PBPK) models integrating either (i) machine learning (ML)-predicted properties or (ii) discovery stage in vitro data. Our test set was composed of 12 challenging development compounds with high lipophilicity (mean calculated log P 4.2), low plasma-free fraction (50% of compounds with fu,p < 1%), and low aqueous solubility. Predictions focused on key human PK parameters, including plasma clearance (CL), volume of distribution at steady state (Vss), and oral bioavailability (%F). For predictions of CL, the ML inputs showed acceptable accuracy and slight underprediction bias [an average absolute fold error (AAFE) of 3.55; an average fold error (AFE) of 0.95]. Surprisingly, use of measured data only slightly improved accuracy but introduced an overprediction bias (AAFE = 3.35; AFE = 2.63). Predictions of Vss were more successful, with both ML (AAFE = 2.21; AFE = 0.90) and in vitro (AAFE = 2.24; AFE = 1.72) inputs showing good accuracy and moderate bias. The %F was poorly predicted using ML inputs [average absolute prediction error (AAPE) of 45%], and use of measured data for solubility and permeability improved this to 34%. Sensitivity analysis showed that predictions of CL limited the overall accuracy of human PK predictions, partly due to high nonspecific binding of lipophilic compounds, leading to uncertainty of unbound clearance. For accurate predictions of %F, solubility was the key factor. Despite current limitations, this work encourages further development of ML models and integration of their results within PBPK models to enable human PK prediction at the drug design stage, even before compounds are synthesized. Further evaluation of this approach with more diverse chemical types is warranted.