Development of a pediatric physiologically-based pharmacokinetic model to support recommended dosing of atezolizumab in children with solid tumors.

Background: Atezolizumab has been studied in multiple indications for both pediatric and adult patient populations. Generally, clinical studies enrolling pediatric patients may not collect sufficient pharmacokinetic data to characterize the drug exposure and disposition because of operational, ethical, and logistical challenges including burden to children and blood sample volume limitations. Therefore, mechanistic modeling and simulation may serve as a tool to predict and understand the drug exposure in pediatric patients. Objective: To use mechanistic physiologically-based pharmacokinetic (PBPK) modeling to predict atezolizumab exposure at a dose of 15 mg/kg (max 1,200 mg) in pediatric patients to support dose rationalization and label recommendations. Methods: A minimal mechanistic PBPK model was used which incorporated age-dependent changes in physiology and biochemistry that are related to atezolizumab disposition such as endogenous IgG concentration and lymph flow. The PBPK model was developed using both in vitro data and clinically observed data in adults and was verified across dose levels obtained from a phase I and multiple phase III studies in both pediatric patients and adults. The verified model was then used to generate PK predictions for pediatric and adult subjects ranging from 2- to 29-year-old. Results: Individualized verification in children and in adults showed that the simulated concentrations of atezolizumab were comparable (76% within two-fold and 90% within three-fold, respectively) to the observed data with no bias for either over- or under-prediction. Applying the verified model, the predicted exposure metrics including Cmin, Cmax, and AUCtau were consistent between pediatric and adult patients with a geometric mean of pediatric exposure metrics between 0.8- to 1.25-fold of the values in adults. Conclusion: The results show that a 15 mg/kg (max 1,200 mg) atezolizumab dose administered intravenously in pediatric patients provides comparable atezolizumab exposure to a dose of 1,200 mg in adults. This suggests that a dose of 15 mg/kg will provide adequate and effective atezolizumab exposure in pediatric patients from 2- to 18-year-old.

Application of physiologically based pharmacokinetic models for therapeutic proteins and other novel modalities.

The past two decades have seen diversification of drug development pipelines and approvals from traditional small molecule therapies to alternative modalities including monoclonal antibodies, engineered proteins, antibody drug conjugates (ADCs), oligonucleotides and gene therapies. At the same time, physiologically based pharmacokinetic (PBPK) models for small molecules have seen increased industry and regulatory acceptance.This review focusses on the current status of the application of PBPK models to these newer modalities and give a perspective on the successes, challenges and future directions of this field.There is greatest experience in the development of PBPK models for therapeutic proteins, and PBPK models for ADCs benefit from prior experience for both therapeutic proteins and small molecules. For other modalities, the application of PBPK models is in its infancy.Challenges are discussed and a common theme is lack of availability of physiological and experimental data to characterise systems and drug parameters to enable a priori prediction of pharmacokinetics. Furthermore, sufficient clinical data are required to build confidence in developed models.The PBPK modelling approach provides a quantitative framework for integrating knowledge and data from multiple sources and can be built on as more data becomes available.

Physiologically Based Pharmacokinetic Modeling To Predict Drug-Biologic Interactions with Cytokine Modulators: Are These Relevant and Is Interleukin-6 Enough?

Drugs that modulate cytokine levels are often used for the treatment of cancer as well as inflammatory or immunologic disorders. Pharmacokinetic drug-biologic interactions (DBIs) may arise from suppression or elevation of cytochrome P450 (P450) enzymes caused by the increase or decrease in cytokine levels after administration of these therapies. There is in vitro and in vivo evidence that demonstrates a clear link between raised interleukin (IL)-6 levels and P450 suppression, in particular CYP3A4. However, despite this, the changes in IL-6 levels in vivo rarely lead to significant drug interactions (area under the curve and Cmax ratios < 2-fold). The clinical significance of such interactions therefore remains questionable and is dependent on the therapeutic index of the small molecule therapy. Physiologically based pharmacokinetic (PBPK) modeling has been used successfully to predict the impact of raised IL-6 on P450 activities. Beyond IL-6, published data show little evidence that IL-8, IL-10, and IL-17 suppress P450 enzymes. In vitro data suggest that IL-1ß, IL-2, tumor necrosis factor (TNF)-α, and interferon (IFN)-γ can cause suppression of P450 enzymes. Despite in vivo there being a link between IL-6 levels and P450 suppression, the evidence to support a direct effect of IL-2, IL-8, IL-10, IL-17, IFN-γ, TNF-α, or vascular endothelial growth factor on P450 activity is inconclusive. This commentary will discuss the relevance of such drug-biologic interactions and whether current PBPK models considering only IL-6 are sufficient. SIGNIFICANCE STATEMENT: This commentary summarizes the current in vitro and in vivo literature regarding cytokine-mediated cytochrome P450 suppression and compares the relative suppressive potential of different cytokines in reference to interleukin (IL)-6. It also discusses the relevance of drug-biologic interactions to therapeutic use of small molecule drugs and whether current physiologically based pharmacokinetic models considering only IL-6 are sufficient to predict the extent of drug-biologic interactions.

Opportunities and Challenges for PBPK Model of mAbs in Paediatrics and Pregnancy.

New drugs may in some cases need to be tested in paediatric and pregnant patients. However, it is difficult to recruit such patients and there are many ethical issues around their inclusion in clinical trials. Modelling and simulation can help to plan well-designed clinical trials with a reduced number of participants and to bridge gaps where recruitment is difficult. Physiologically based pharmacokinetic (PBPK) models for small molecule drugs have been used to aid study design and dose adjustments in paediatrics and pregnancy, with several publications in the literature. However, published PBPK models for monoclonal antibodies (mAb) in these populations are scarce. Here, the current status of mAb PBPK models in paediatrics and pregnancy is discussed. Seven mAb PBPK models published for paediatrics were found, which report good prediction accuracy across a wide age range. No mAb PBPK models for pregnant women have been published to date. Current challenges to the development of such PBPK models are discussed, including gaps in our knowledge of relevant physiological processes and availability of clinical data to verify models. As the availability of such data increases, it will help to improve our confidence in the PBPK model predictive ability. Advantages for using PBPK models to predict mAb PK in paediatrics and pregnancy are discussed. For example, the ability to incorporate ontogeny and gestational changes in physiology, prediction of maternal, placental and foetal exposure and the ability to make predictions from in vitro and preclinical data prior to clinical data being available.

Development and Application of a Physiologically-Based Pharmacokinetic Model to Predict the Pharmacokinetics of Therapeutic Proteins from Full-term Neonates to Adolescents.

Physiologically-based pharmacokinetic (PBPK) modelling provides an integrated framework to predict the disposition of small molecule drugs in children and is increasingly being used for dose recommendation and optimal design of paediatric studies and in regulatory submissions. Existing paediatric PBPK models can be adopted to describe the disposition of therapeutic proteins (TPs) in children by incorporating information on age-related changes of additional physiological and biological parameters (e.g. endogenous IgG, neonatal Fc receptor, lymph flow). In this study, physiological parameters were collated from literature and evaluated for any age-dependent trends. The age-dependent physiological parameters were used to construct a paediatric PBPK model for TPs. The model was then used to predict the pharmacokinetics of recombinant human erythropoietin (EPO), infliximab, etanercept, basiliximab, anakinra and enfuvirtide in paediatric subjects. The developed paediatric PBPK model predicted the drug concentration-time profiles reasonably well in full-term neonates (clinical PK data only available for EPO), infants, children and adolescents with the ratios of predicted over observed clearance values within 1.5-fold and 25 out of 26 clearance predictions were within 0.8- to 1.25-fold of the observed values. The clinically reported data are required to further assess the predictive accuracy of PK for Fc-containing proteins in term-born children younger than 2 months. This study demonstrates the ability of PBPK models accounting for age-dependent changes in relevant parameters to predict the pharmacokinetics of different types of TPs in paediatrics. The information gained from the PBPK models described here can facilitate our understanding of the complexities of TPs' disposition during growth and development.

Simulating the Impact of Elevated Levels of Interleukin-6 on the Pharmacokinetics of Various CYP450 Substrates in Patients with Neuromyelitis Optica or Neuromyelitis Optica Spectrum Disorders in Different Ethnic Populations.

A physiologically based pharmacokinetic (PBPK) model was used to simulate the impact of elevated levels of interleukin (IL)-6 on the exposure of several orally administered cytochrome P450 (CYP) probe substrates (caffeine, S-warfarin, omeprazole, dextromethorphan, midazolam, and simvastatin). The changes in exposure of these substrates in subjects with rheumatoid arthritis (and hence elevated IL-6 levels) compared with healthy subjects were predicted with a reasonable degree of accuracy. The PBPK model was then used to simulate the change in oral exposure of the probe substrates in North European Caucasian, Chinese, and Japanese population of patients with neuromyelitis optica (NMO) or NMO spectrum disorder with elevated plasma IL-6 levels (up to 100 pg/mL). Moderate interactions [mean AUC fold change, ≤ 2.08 (midazolam) or 2.36 (simvastatin)] was predicted for CYP3A4 probe substrates and weak interactions (mean AUC fold change, ≤ 1.29-1.97) were predicted for CYP2C19, CYP2C9, and CYP2D6 substrates. No notable interaction was predicted with CYP1A2. Although ethnic differences led to differences in simulated exposure for some of the probe substrates, there were no marked differences in the predicted magnitude of the change in exposure following IL-6-mediated suppression of CYPs. Decreased levels of serum albumin (as reported in NMO patients) had little impact on the magnitude of the simulated IL-6-mediated drug interactions. This PBPK modeling approach allowed us to leverage knowledge from different disease and ethnic populations to make predictions of cytokine-related DDIs in a rare disease population where actual clinical studies would otherwise be difficult to conduct.

Abundance of Hepatic Transporters in Caucasians: A Meta-Analysis.

This study aimed to derive quantitative abundance values for key hepatic transporters suitable for in vitro-in vivo extrapolation within a physiologically based pharmacokinetic modeling framework. A meta-analysis was performed whereby data on abundance measurements, sample preparation methods, and donor demography were collated from the literature. To define values for a healthy Caucasian population, a subdatabase was created whereby exclusion criteria were applied to remove samples from non-Caucasian individuals, those with underlying disease, or those with subcellular fractions other than crude membrane. Where a clinically relevant active genotype was known, only samples from individuals with an extensive transporter phenotype were included. Authors were contacted directly when additional information was required. After removing duplicated samples, the weighted mean, geometric mean, standard deviation, coefficient of variation, and between-study homogeneity of transporter abundances were determined. From the complete database containing 24 transporters, suitable abundance data were available for 11 hepatic transporters from nine studies after exclusion criteria were applied. Organic anion transporting polypeptides OATP1B1 and OATP1B3 showed the highest population abundance in healthy adult Caucasians. For several transporters, the variability in abundance was reduced significantly once the exclusion criteria were applied. The highest variability was observed for OATP1B3 > OATP1B1 > multidrug resistance protein 2 > multidrug resistance gene 1. No relationship was found between transporter expression and donor age. To our knowledge, this study provides the first in-depth analysis of current quantitative abundance data for a wide range of hepatic transporters, with the aim of using these data for in vitro-in vivo extrapolation, and highlights the significance of investigating the background of tissue(s) used in quantitative transporter proteomic studies. Similar studies are now warranted for other ethnicities.

Potential Sources of Inter-Subject Variability in Monoclonal Antibody Pharmacokinetics.

Understanding inter-subject variability in drug pharmacokinetics and pharmacodynamics is important to ensure that all patients attain suitable drug exposure to achieve efficacy and avoid toxicity. Inter-subject variability in the pharmacokinetics of therapeutic monoclonal antibodies (mAbs) is generally moderate to high; however, the factors responsible for the high inter-subject variability have not been comprehensively reviewed. In this review, the extent of inter-subject variability for mAb pharmacokinetics is presented and potential factors contributing to this variability are explored and summarised. Disease status, age, sex, ethnicity, body size, genetic polymorphisms, concomitant medication, co-morbidities, immune status and multiple other patient-specific details have been considered. The inter-subject variability for mAb pharmacokinetics most likely depends on the complex interplay of multiple factors. However, studies aimed at investigating the reasons for the inter-subject variability are sparse. Population pharmacokinetic models and physiologically based pharmacokinetic models are useful tools to identify important covariates, aiding in the understanding of factors contributing to inter-subject variability. Further understanding of inter-subject variability in pharmacokinetics should aid in development of dosing regimens that are more appropriate.

A Bottom-Up Whole-Body Physiologically Based Pharmacokinetic Model to Mechanistically Predict Tissue Distribution and the Rate of Subcutaneous Absorption of Therapeutic Proteins.

The ability to predict subcutaneous (SC) absorption rate and tissue distribution of therapeutic proteins (TPs) using a bottom-up approach is highly desirable early in the drug development process prior to clinical data being available. A whole-body physiologically based pharmacokinetic (PBPK) model, requiring only a few drug parameters, to predict plasma and interstitial fluid concentrations of TPs in humans after intravenous and subcutaneous dosing has been developed. Movement of TPs between vascular and interstitial spaces was described by considering both convection and diffusion processes using a 2-pore framework. The model was optimised using a variety of literature sources, such as tissue lymph/plasma concentration ratios in humans and animals, information on the percentage of dose absorbed following SC dosing via lymph in animals and data showing loss of radiolabelled IgG from the SC dosing site in humans. The resultant model was used to predict t max and plasma concentration profiles for 12 TPs (molecular weight 8-150 kDa) following SC dosing. The predicted plasma concentration profiles were generally comparable to observed data. t max was predicted within 3-fold of reported values, with one third of the predictions within 0.8-1.25-fold. There was no systematic bias in simulated C max values, although a general trend for underprediction of t max was observed. No clear trend between prediction accuracy of t max and TP isoelectric point or molecular size was apparent. The mechanistic whole-body PBPK model described here can be applied to predict absorption rate of TPs into blood and movement into target tissues following SC dosing.

Are Physiologically Based Pharmacokinetic Models Reporting the Right C(max)? Central Venous Versus Peripheral Sampling Site.

Physiologically based pharmacokinetic (PBPK) models can over-predict maximum plasma concentrations (C(max)) following intravenous administration. A proposed explanation is that invariably PBPK models report the concentration in the central venous compartment, rather than the site where the samples are drawn. The purpose of this study was to identify and validate potential corrective models based on anatomy and physiology governing the blood supply at the site of sampling and incorporate them into a PBPK platform. Four models were developed and scrutinised for their corrective potential. All assumed the peripheral sampling site concentration could be described by contributions from surrounding tissues and utilised tissue-specific concentration-time profiles reported from the full-PBPK model within the Simcyp Simulator. Predicted concentrations for the peripheral site were compared to the observed C(max). The models results were compared to clinical data for 15 studies over seven compounds (alprazolam, imipramine, metoprolol, midazolam, omeprazole, rosiglitazone and theophylline). The final model utilised tissue concentrations from adipose, skin, muscle and a contribution from artery. Predicted C(max) values considering the central venous compartment can over-predict the observed values up to 10-fold whereas the new sampling site predictions were within 2-fold of observed values. The model was particularly relevant for studies where traditional PBPK models over-predict early time point concentrations. A successful corrective model for C(max) prediction has been developed, subject to further validation. These models can be enrolled as built-up modules into PBPK platforms and potentially account for factors that may affect the initial mixing of the blood at the site of sampling.

Application of a physiologically based pharmacokinetic model to assess propofol hepatic and renal glucuronidation in isolation: utility of in vitro and in vivo data.

A physiologically based pharmacokinetic (PBPK) modeling approach was used to assess the prediction accuracy of propofol hepatic and extrahepatic metabolic clearance and to address previously reported underprediction of in vivo clearance based on static in vitro-in vivo extrapolation methods. The predictive capacity of propofol intrinsic clearance data (CLint) obtained in human hepatocytes and liver and kidney microsomes was assessed using the PBPK model developed in MATLAB software. Microsomal data obtained by both substrate depletion and metabolite formation methods and in the presence of 2% bovine serum albumin were considered in the analysis. Incorporation of hepatic and renal in vitro metabolic clearance in the PBPK model resulted in underprediction of propofol clearance regardless of the source of in vitro data; the predicted value did not exceed 35% of the observed clearance. Subsequently, propofol clinical data from three dose levels in intact patients and anhepatic subjects were used for the optimization of hepatic and renal CLint in a simultaneous fitting routine. Optimization process highlighted that renal glucuronidation clearance was underpredicted to a greater extent than liver clearance, requiring empirical scaling factors of 17 and 9, respectively. The use of optimized clearance parameters predicted hepatic and renal extraction ratios within 20% of the observed values, reported in an additional independent clinical study. This study highlights the complexity involved in assessing the contribution of extrahepatic clearance mechanisms and illustrates the application of PBPK modeling, in conjunction with clinical data, to assess prediction of clearance from in vitro data for each tissue individually.

Characterization of in vitro glucuronidation clearance of a range of drugs in human kidney microsomes: comparison with liver and intestinal glucuronidation and impact of albumin.

Previous studies have shown the importance of the addition of albumin for characterization of hepatic glucuronidation in vitro; however, no reports exist on the effects of albumin on renal or intestinal microsomal glucuronidation assays. This study characterized glucuronidation clearance (CL(int, UGT)) in human kidney, liver, and intestinal microsomes in the presence and absence of bovine serum albumin (BSA) for seven drugs with differential UDP-glucuronosyltransferase (UGT) 1A9 and UGT2B7 specificity, namely, diclofenac, ezetimibe, gemfibrozil, mycophenolic acid, naloxone, propofol, and telmisartan. The impact of renal CL(int, UGT) on accuracy of in vitro-in vivo extrapolation (IVIVE) of glucuronidation clearance was investigated. Inclusion of 1% BSA for acidic drugs and 2% for bases/neutral drugs in incubations was found to be suitable for characterization of CL(int, UGT) in different tissues. Although BSA increased CL(int, UGT) in all tissues, the extent was tissue- and drug-dependent. Scaled CL(int, UGT) in the presence of BSA ranged from 2.22 to 207, 0.439 to 24.4, and 0.292 to 23.8 ml · min(-1) · g tissue(-1) in liver, kidney, and intestinal microsomes. Renal CL(int, UGT) (per gram of tissue) was up to 2-fold higher in comparison with that for liver for UGT1A9 substrates; in contrast, CL(int, UGT) for UGT2B7 substrates represented approximately one-third of hepatic estimates. Scaled renal CL(int, UGT) (in the presence of BSA) was up to 30-fold higher than intestinal glucuronidation for the drugs investigated. Use of in vitro data obtained in the presence of BSA and inclusion of renal clearance improved the IVIVE of glucuronidation clearance, with 50% of drugs predicted within 2-fold of observed values. Characterization and consideration of kidney CL(int, UGT) is particularly important for UGT1A9 substrates.