Evaluation of Prediction Accuracy for Volume of Distribution in Rat and Human Using In Vitro, In Vivo, PBPK and QSAR Methods

Use of pharmacokinetic versus pharmacodynamic endpoints to support human dose predictions: implications for rational drug design and early clinical development

INTRODUCTION

The predicted human dose regimen of new chemical entities represents the most holistic and clinically relevant measure of drug-likeness upon which to base decisions in drug design and selection of candidate molecules for further development. Likewise, the predicted human dose regimen for efficacy and safety provides critical insight into clinical development planning. As such, human dose predictions are commonly generated in early stages of research and continually revisited as new data are generated through development.

AREAS COVERED

In this work, the authors illustrate scenarios where conventional approaches based on discrete pharmacokinetic metrics are inappropriate and propose a generalizable approach leveraging a predicted average pharmacodynamic effect rather than pharmacokinetic metrics. Preclinical and clinical data of a JAK inhibitor, tofacitinib, were used to illustrate the relative value of this approach to human dose prediction.

EXPERT OPINION

Due to the simplicity of implementation, pharmacokinetic-based approaches which target a discrete maximal, average, or minimum concentration have been widely used across the pharmaceutical industry. However, in emphasizing only one point on the overall exposure-time profile, such approaches can be misleading in terms of the expected pharmacodynamic effect. For future projections, the authors recommend using the average pharmacodynamic effect-based approach to calculate human efficacious dose.

Mode of action approach supports a lack of carcinogenic potential of six organic UV filters

Ultraviolet (UV) filters, the active ingredients in sunscreens, have been used for several decades to reduce the risk of acute and chronic damage to the skin from solar UV radiation, which can lead to skin cancer. Based on recent clinical studies showing that certain UV filters are absorbed systemically at low levels in humans, the US Food and Drug Administration (FDA) has requested supplementing existing safety data with preclinical studies including oral and dermal 2-year rodent carcinogenicity studies. Although the conduct of 2-year rodent carcinogenicity studies has been the standard approach for evaluating the carcinogenic potential of chemicals and new drugs for approximately 6 decades, there are multiple examples showing that such studies are not predictive of human cancer risk. Given these concerns with 2-year rodent carcinogenicity studies, we have developed and applied an alternative approach for supplementing existing data related to carcinogenic potential for six of the most commonly used UV filters in sunscreen products (i.e. avobenzone, ensulizole, homosalate, octinoxate, octisalate, and octocrylene). This approach evaluates their mode of action (MOA) based on in vivo, in vitro, and in silico data combined with an assessment of exposure margins. This approach is based on the substantial progress in understanding the MOAs that are responsible for tumor induction in humans. It is consistent with those being developed by the International Council for Harmonization (ICH) and other health authorities to replace 2-year carcinogenicity studies given their limitations and questionable biological relevance to humans. The available data for the six UV filters show that they are not genotoxic and show no evidence of biologically relevant carcinogenic MOAs. Furthermore, their systemic exposure levels in humans fall well below concentrations at which they have biologic activity. In conclusion, these data support the continued safe use of these six filters in sunscreen products.

A cross-species assessment of in silico prediction methods of steady-state volume of distribution using Simcyp simulators

Predicting steady-state volume of distribution (Vss) is a key component of pharmacokinetic predictions and often guided using preclinical data. However, when bottom-up prediction from physiologically-based pharmacokinetic (PBPK) models and observed Vss misalign in preclinical species, or predicted Vss from different models varies significantly, no consensus exists for selecting models or preclinical species to improve the prediction. Through systematic analysis of Vss prediction across rat, dog, monkey, and human, using common methods, a practical strategy for predicting human Vss, with or without integration of preclinical PK information is warranted. In this analysis, we curated a dataset of 57 diverse compounds with measured physicochemical and protein binding data, together with observed Vss in these species. Using a bottom-up approach, prediction performance was consistent across species for each method. Although no method consistently outperformed others for all compound types and across species, M2 (Rodgers-Rowland method) performed marginally better for acids. Comparable compound-specific global tissue Kp scalars were needed to match observed Vss for both, human and preclinical species. Consequently, application of geometric mean values of preclinical Kp scalar to human Vss prediction improved accuracy. We propose a decision tree for human Vss prediction using PBPK methods with or without integrating preclinical PK information.

Beyond CL and VSS: A comprehensive approach to human pharmacokinetic predictions

This article presents a comprehensive examination of processes related to the prediction of human pharmacokinetics (PK), a crucial task of clinical drug candidate selection. By systematically incorporating in vitro absorption, distribution, metabolism and excretion (ADME) and in vivo PK data with expert judgement, the study achieves high-quality human PK predictions for 40 orally administered compounds from Boehringer Ingelheim's new chemical entity (NCE) portfolio. Overall, the article provides a detailed evaluation of and guidance for a structured process to predict full concentration-time profiles beyond single-parameter predictions, using state-of-the-art methodologies. Furthermore, it discusses future challenges and improvements, and aims to provide valuable insights for scientists working in drug metabolism and PK (DMPK) or PK/pharmacodynamics (PK/PD) modelling.

Human risk assessment through development and application of a physiologically based toxicokinetic model for 4-tert-octylphenol

4-tert-octylphenol (4-tert-OP) is an ecologically hazardous substance, and exposure to it in the environment has been consistently reported in the past. Despite the hazards and widespread exposure to 4-tert-OP, tools for scientific assessment of 4-tert-OP exposure risk level in humans are lacking. The main purpose of this study was to develop a physiologically-based-toxicokinetic (PBTK) model for 4-tert-OP and to perform quantitative risk assessment of 4-tert-OP in various population groups using the established model. Based on the results of toxicokinetic experiments on male rats, the PBTK model for 4-tert-OP was established and verified, and this was converted to a model for humans through interspecies extrapolation. Based on the previously reported no-observed-adverse-effect-levels for rats, it was possible to estimate the 4-tert-OP reference dose in humans through reverse dosimetry using the model. Biomonitoring data derived from various population groups were applied to the human PBTK model to calculate external exposures and margin of safety for 4-tert-OP for each population group. The PBTK model established in this study adequately explained the toxicokinetic experimental values at acceptable levels and was able to quantitatively predict the 4-tert-OP exposure level in the testes related to male reproductive toxicity. In addition, the degree of external exposure to 4-tert-OP could be scientifically estimated based on biomonitoring values derived from various biological matrices. The reference doses for systemic and reproductive toxicity caused by 4-tert-OP in male humans were calculated to be 0.16 and 1.12 mg/kg/day, respectively. The mean external exposure to 4-tert-OP in each population group estimated based on plasma and urine biomonitoring data was 0.04-66.24 mg/kg/day, showing very large exposure diversity between groups. Exposure risks to 4-tert-OP in populations ranged from safe to risky, suggesting the need for continued monitoring and risk management of 4-tert-OP worldwide. This study provides valuable scientific insight regarding the 4-tert-OP human risk assessment.

Quantitative pulmonary pharmacokinetics of tetrandrine for SARS-CoV-2 repurposing: a physiologically based pharmacokinetic modeling approach

Tetrandrine (TET) has been traditionally used in China as a medication to treat silicosis and has recently demonstrated anti-SARS-CoV-2 potential in vitro. By recognizing the disparity between in vitro findings and in vivo performance, we aimed to estimate the free lung concentration of TET using a physiologically based pharmacokinetic (PBPK) model to link in vitro activity with in vivo efficacy. Comparative pharmacokinetic studies of TET were performed in rats and dogs to elucidate the pharmacokinetic mechanisms as well as discern interspecies variations. These insights facilitated the creation of an animal-specific PBPK model, which was subsequently translated to a human model following thorough validation. Following validation of the pharmacokinetic profile from a literature report on single oral dosing of TET in humans, the plasma and lung concentrations were predicted after TET administration at approved dosage levels. Finally, the antiviral efficacy of TET in humans was assessed from the free drug concentration in the lungs. Both in vivo and in vitro experiments thus confirmed that the systemic clearance of TET was primarily through hepatic metabolism. Additionally, the lysosomal capture of basic TET was identified as a pivotal factor in its vast distribution volume and heterogeneous tissue distribution, which could modulate the absorption dynamics of TET in the gastrointestinal tract. Notably, the PBPK-model-based unbound lung concentration of TET (1.67-1.74 μg/mL) at the recommended clinical dosage surpassed the in vitro threshold for anti-SARS-CoV-2 activity (EC90 = 1.52 μg/mL). Thus, a PBPK model was successfully developed to bridge the in vitro activity and in vivo target exposure of TET to facilitate its repurposing.

Physiologically Based Pharmacokinetic Model of OATP1B Substrates with a Nonlinear Mixed Effect Approach: Estimating Empirical In Vitro-to-In Vivo Scaling Factors

BACKGROUND AND OBJECTIVE

Physiologically based pharmacokinetic (PBPK) models are valuable for translating in vitro absorption, distribution, metabolism, and excretion (ADME) data to predict clinical pharmacokinetics, and can enable discovery and early clinical stages of pharmaceutical research. However, in predicting pharmacokinetics of organic anion transporting polypeptide (OATP) 1B substrates based on in vitro transport and metabolism data, PBPK models typically require additional empirical in vitro-to-in vivo scaling factors (ESFs) in order to accurately recapitulate observed clinical profiles. As model simulation is very sensitive to ESFs, a critical evaluation of ESF estimation is prudent. Previously studies have applied classic 'two-stage' and 'naïve pooled data' approaches in identifying a set of compound independent ESFs. However, the 'two-stage' approach has the parameter identification issue in separately fitting data for individual compounds, while the 'naïve pooled data' approach ignores interstudy variability, leading to potentially biased ESF estimates.

METHODS

In this study, we have applied a nonlinear mixed-effect approach in estimating ESF of the PBPK model and incorporated additional data from 86 runs of in vitro uptake assay and 49 clinical studies of 12 training compounds in model development to further enhance the translation of in vitro data to predict the pharmacokinetics of OATP1B substrate drugs. To test predication accuracy of the model, a 'leave-one-out' analysis has been performed.

RESULTS

The established model can reasonably describe the clinical observations, with both mean values and interstudy variabilities quantified for ESF and volume of distribution parameters. The mean estimates are largely consistent with values in the previous reports. The interstudy variabilities of these parameters are estimated to be at least 50% (as coefficient of variation). Most compounds can be reasonably predicted in the 'leave-one-out' analysis.

CONCLUSION

This study improves the confidence in predicting the pharmacokinetics of OATP1B substrates in individual studies of small sample sizes, and quantifies the variability associated with the prediction.

Development and application of the physiologically-based toxicokinetic (PBTK) model for ochratoxin A (OTA) in rats and humans

Ochratoxin A (OTA) is a common fungal toxin frequently detected in food and human plasma samples. Currently, the physiologically based toxicokinetic (PBTK) model plays an active role in dose translation and can improve and enhance the risk assessment of toxins. In this study, the PBTK model of OTA in rats and humans was established based on knowledge of OTA-specific absorption, distribution, metabolism, and excretion (ADME) in order to better explain the disposition of OTA in humans and the discrepancies with other species. The models were calibrated and optimized using the available kinetic and toxicokinetic (TK) data, and independent test datasets were used for model evaluation. Subsequently, sensitivity analyses and population simulations were performed to characterize the extent to which variations in physiological and specific chemical parameters affected the model output. Finally, the constructed models were used for dose extrapolation of OTA, including the rat-to-human dose adjustment factor (DAF) and the human exposure conversion factor (ECF). The results showed that the unbound fraction (Fup) of OTA in plasma of rat and human was 0.02-0.04% and 0.13-4.21%, respectively. In vitro experiments, the maximum enzyme velocity (Vmax) and Michaelis-Menten constant (Km) of OTA in rat and human liver microsomes were 3.86 and 78.17 μg/g min-1, 0.46 and 4.108 μg/mL, respectively. The predicted results of the model were in good agreement with the observed data, and the models in rats and humans were verified. The PBTK model derived a DAF of 0.1081 between rats and humans, whereas the ECF was 2.03. The established PBTK model can be used to estimate short- or long-term OTA exposure levels in rats and humans, with the capacity for dose translation of OTA to provide the underlying data for risk assessment of OTA.

A New Version of the Tissue Composition-Based Model for Improving the Mechanism-Based Prediction of Volume of Distribution at Steady-State for Neutral Drugs

In-vitro models are available in the literature for predicting the volume of distribution at steady-state (Vdss) of drugs. The mechanistic model refers to the tissue composition-based model (TCM), which includes important factors that govern Vdss such as drug physiochemistry and physiological data. The recognized TCM published by Rodgers and Rowland (TCM-RR) and a subsequent adjustment made by Simulations Plus Inc. (TCM-SP) have been shown to be generally less accurate with neutral compared to ionized drugs. Therefore, improving these models for neutral drugs becomes necessary. The objective of this study was to propose a new TCM for improving the prediction of Vdss for neutral drugs. The new TCM included two modifications of the published models (i) accentuate the effect of the blood-to-plasma ratio (BPR) that should cover permeated molecules across the biomembranes, which is lacking in these models for neutral compounds, and (ii) use a different approach to estimate the binding in tissues. The new TCM was validated with a large dataset of 202 commercial and proprietary compounds including preclinical and clinical data. All scenario datasets were predicted more accurately with the TCM-New, whereas all statistical parameters indicate that the TCM-New showed significant improvements in terms of accuracy over the TCM-RR and TCM-SP. Predictions of Vdss were frequently more accurate for the TCM-new with 83% within twofold error versus only 50% for the TCM-RR. And more than 95% of the predictions were within threefold error and patient interindividual differences can be predicted with the TCM-New, greatly exceeding the accuracy of the published models. Overall, the new TCM incorporating BPR significantly improved the Vdss predictions in animals and humans for neutral drugs, and, hence, has the potential to better support the drug discovery and facilitate the first-in-human predictions.

The Comparison of Machine Learning and Mechanistic In Vitro-In Vivo Extrapolation Models for the Prediction of Human Intrinsic Clearance

Accurate prediction of human pharmacokinetics (PK) remains one of the key objectives of drug metabolism and PK (DMPK) scientists in drug discovery projects. This is typically performed by using in vitro-in vivo extrapolation (IVIVE) based on mechanistic PK models. In recent years, machine learning (ML), with its ability to harness patterns from previous outcomes to predict future events, has gained increased popularity in application to absorption, distribution, metabolism, and excretion (ADME) sciences. This study compares the performance of various ML and mechanistic models for the prediction of human IV clearance for a large (645) set of diverse compounds with literature human IV PK data, as well as measured relevant in vitro end points. ML models were built using multiple approaches for the descriptors: (1) calculated physical properties and structural descriptors based on chemical structure alone (classical QSAR/QSPR); (2) in vitro measured inputs only with no structure-based descriptors (ML IVIVE); and (3) in silico ML IVIVE using in silico model predictions for the in vitro inputs. For the mechanistic models, well-stirred and parallel-tube liver models were considered with and without the use of empirical scaling factors and with and without renal clearance. The best ML model for the prediction of in vivo human intrinsic clearance (CLint) was an in vitro ML IVIVE model using only six in vitro inputs with an average absolute fold error (AAFE) of 2.5. The best mechanistic model used the parallel-tube liver model, with empirical scaling factors resulting in an AAFE of 2.8. The corresponding mechanistic model with full in silico inputs achieved an AAFE of 3.3. These relative performances of the models were confirmed with the prediction of 16 Pfizer drug candidates that were not part of the original data set. Results show that ML IVIVE models are comparable to or superior to their best mechanistic counterparts. We also show that ML IVIVE models can be used to derive insights into factors for the improvement of mechanistic PK prediction.

ADME characterization and PBK model development of 3 highly protein-bound UV filters through topical application

Estimating human exposure in the safety assessment of chemicals is crucial. Physiologically based kinetic (PBK) models which combine information on exposure, physiology, and chemical properties, describing the absorption, distribution, metabolism, and excretion (ADME) processes of a chemical, can be used to calculate internal exposure metrics such as maximum concentration and area under the concentration-time curve in plasma or tissues of a test chemical in next-generation risk assessment. This article demonstrates the development of PBK models for 3 UV filters, specifically octyl methoxycinnamate, octocrylene, and 4-methylbenzylidene camphor. The models were parameterized entirely based on data obtained from in vitro and/or in silico methods in a bottom-up modeling approach and then validated based on human dermal pharmacokinetic (PK) data. The 3 UV filters are "difficult to test" in in vitro test systems due to high lipophilicity, high binding affinity for proteins, and nonspecific binding, for example, toward plastic. This research work presents critical considerations in ADME data generation, interpretation, and parameterization to assure valid PBK model development to increase confidence in using PBK modeling to help make safety decisions in the absence of human PK data. The developed PBK models of the 3 chemicals successfully simulated the plasma concentration profiles of clinical PK data following dermal application, indicating the reliability of the ADME data generated and the parameters determined. The study also provides insights and lessons learned for characterizing ADME and developing PBK models for highly lipophilic and protein-bound chemicals in the future.

Shared learning from a physiologically based pharmacokinetic modeling strategy for human pharmacokinetics prediction through retrospective analysis of Genentech compounds

The quantitative prediction of human pharmacokinetics (PK) including the PK profile and key PK parameters are critical for early drug development decisions, successful phase I clinical trials, and the establishment of a range of doses to enable phase II clinical dose selection. Here, we describe an approach employing physiologically based pharmacokinetic (PBPK) modeling (Simcyp) to predict human PK and to validate its performance through retrospective analysis of 18 Genentech compounds for which clinical data are available. In short, physicochemical parameters and in vitro data for preclinical species were integrated using PBPK modeling to predict the in vivo PK observed in mouse, rat, dog, and cynomolgus monkey. Through this process, the in vitro to in vivo extrapolation (IVIVE) was determined and then incorporated into PBPK modeling in order to predict human PK. Overall, the prediction obtained using this PBPK-IVIVE approach captured the observed human PK profiles of the compounds from the dataset well. The predicted Cmax was within 2-fold of the observed Cmax for 94% of the compounds while the predicted area under the curve (AUC) was within 2-fold of the observed AUC for 72% of the compounds. Additionally, important IVIVE trends were revealed through this investigation, including application of scaling factors determined from preclinical IVIVE to human PK prediction for each molecule. Based upon the analysis, this PBPK-based approach now serves as a practical strategy for human PK prediction at the candidate selection stage at Genentech.

In Vitro-In Vivo Extrapolation and Scaling Factors for Clearance of Human and Preclinical Species with Liver Microsomes and Hepatocytes

In vitro-in vivo extrapolation ((IVIVE) and empirical scaling factors (SF) of human intrinsic clearance (CL_int) were developed using one of the largest dataset of 455 compounds with data from human liver microsomes (HLM) and human hepatocytes (HHEP). For extended clearance classification system (ECCS) class 2/4 compounds, linear SFs (SF_lin) are approximately 1, suggesting enzyme activities in HLM and HHEP are similar to those in vivo under physiological conditions. For ECCS class 1A/1B compounds, a unified set of SFs was developed for CL_int. These SFs contain both SF_lin and an exponential SF (SF_β) of fraction unbound in plasma (f_u,p). The unified SFs for class 1A/1B eliminate the need to identify the transporters involved prior to clearance prediction. The underlying mechanisms of these SFs are not entirely clear at this point, but they serve practical purposes to reduce biases and increase prediction accuracy. Similar SFs have also been developed for preclinical species. For HLM-HHEP disconnect (HLM > HHEP) ECCS class 2/4 compounds that are mainly metabolized by cytochrome P450s/FMO, HLM significantly overpredicted in vivo CL_int, while HHEP slightly underpredicted and geometric mean of HLM and HHEP slightly overpredicted in vivo CL_int. This observation is different than in rats, where rat liver microsomal CL_int correlates well with in vivo CL_int for compounds demonstrating permeability-limited metabolism. The good CL_int IVIVE developed using HLM and HHEP helps build confidence for prospective predictions of human clearance and supports the continued utilization of these assays to guide structure-activity relationships to improve metabolic stability.

Physiologically Based Pharmacokinetic Model of Brain Delivery of Plasma Protein Bound Drugs

INTRODUCTION

A physiologically based pharmacokinetic (PBPK) model is developed that focuses on the kinetic parameters of drug association and dissociation with albumin, alpha-1 acid glycoprotein (AGP), and brain tissue proteins, as well as drug permeability at the blood-brain barrier, drug metabolism, and brain blood flow.

GOAL

The model evaluates the extent to which plasma protein-mediated uptake (PMU) of drugs by brain influences the concentration of free drug both within the brain capillary compartment in vivo and the brain compartment. The model also studies the effect of drug binding to brain tissue proteins on the concentration of free drug in brain.

METHODS

The steady state and non-steady state PBPK models are comprised of 11-12 variables, and 18-23 parameters, respectively. Two model drugs are analyzed: propranolol, which undergoes modest PMU from the AGP-bound pool, and imipramine, which undergoes a high degree of PMU from both the albumin-bound and AGP-bound pools in plasma.

RESULTS

The free propranolol concentration in brain is under-estimated 2- to fourfold by in vitro measurements of free plasma propranolol, and the free imipramine concentration in brain is under-estimated by 18- to 31-fold by in vitro measurements of free imipramine in plasma. The free drug concentration in brain in vivo is independent of drug binding to brain tissue proteins.

CONCLUSIONS

In vitro measurement of free drug concentration in plasma under-estimates the free drug in brain in vivo if PMU in vivo from either the albumin and/or the AGP pools in plasma takes place at the BBB surface.

Computational Predictions of Nonclinical Pharmacokinetics at the Drug Design Stage

Although computational predictions of pharmacokinetics (PK) are desirable at the drug design stage, existing approaches are often limited by prediction accuracy and human interpretability. Using a discovery data set of mouse and rat PK studies at Roche (9,685 unique compounds), we performed a proof-of-concept study to predict key PK properties from chemical structure alone, including plasma clearance (CLp), volume of distribution at steady-state (Vss), and oral bioavailability (F). Ten machine learning (ML) models were evaluated, including Single-Task, Multitask, and transfer learning approaches (i.e., pretraining with in vitro data). In addition to prediction accuracy, we emphasized human interpretability of outcomes, especially the quantification of uncertainty, applicability domains, and explanations of predictions in terms of molecular features. Results show that intravenous (IV) PK properties (CLp and Vss) can be predicted with good precision (average absolute fold error, AAFE of 1.96-2.84 depending on data split) and low bias (average fold error, AFE of 0.98-1.36), with AutoGluon, Gaussian Process Regressor (GP), and ChemProp displaying the best performance. Driven by higher complexity of oral PK studies, predictions of F were more challenging, with the best AAFE values of 2.35-2.60 and higher overprediction bias (AFE of 1.45-1.62). Multi-Task approaches and pretraining of ChemProp neural networks with in vitro data showed similar precision to Single-Task models but helped reduce the bias and increase correlations between observations and predictions. A combination of GP-computed prediction variance, molecular clustering, and dimensionality-reduction provided valuable quantitative insights into prediction uncertainty and applicability domains. SHAPley Additive exPlanations (SHAPs) highlighted molecular features contributing to prediction outcomes of Vss, providing explanations that could aid drug design. Combined results show that computational predictions of PK are feasible at the drug design stage, with several ML technologies converging to successfully leverage historical PK data sets. Further studies are needed to unlock the full potential of this approach, especially with respect to data set sizes and quality, transfer learning between in vitro and in vivo data sets, model-independent quantification of uncertainty, and explainability of predictions.

Translational pharmacokinetics of a novel bispecific antibody against Ebola virus (MBS77E) from animal to human by PBPK modeling & simulation

The goal of this study was to construct a PBPK model to accelerate the translation of MBS77E, a humanized bispecific antibody against the Ebola virus. In-depth nonclinical pharmacokinetic studies in rats, monkeys, wild-type mice and transgenic mice were conducted. The pH-dependent affinities (K_D) of MBS77E to recombinant FcRn of different species were determined by surface plasmon resonance analysis. A mechanistic whole-body PBPK model of MBS77E was developed and validated in the assessment of PK profiles and tissue distributions in preclinical models. This PBPK model was finally used to predict human PK behaviors of MBS77E. Simulations from the PBPK model with measured and fitted parameters were able to yield good predictions of the serum and tissue pharmacokinetic parameters of MBS77E within 2-fold errors. The predicted serum concentration in humans was able to maintain a sufficiently high level for more than 14 days after 50 mg/kg i.v. administrating. This achievement unlocks that PBPK modeling is a powerful tool to gain insights into the properties of antibody drugs. It guided experimental efforts to obtain necessary information before entry into humans.

Utility of preclinical species for uncertainty assessment and correction of prediction of human volume of distribution using the Rodgers-Lukacova model

Prediction of rat, dog, monkey, and human volume of distribution (VD_ss) by Rodgers-Lukacova model was evaluated using a data set of more than 100 compounds.The prediction accuracy was best for humans followed by monkeys and dogs with 59, 52, and 41% of compounds within 2-fold, respectively.The accuracy of predictions in preclinical species was indicative of the human situation. This was particularly true for monkeys, where 87% of the compounds that were predicted within 2-fold in monkeys were also predicted within 2-fold in humans.The model's tendency to underestimate VD_ss was higher in rats and dogs compared to humans and monkeys for all ion classes but zwitterions. Hence, correction of human predictions using prediction errors in rats and dogs resulted in overestimation of VD_ss.The model had a similar degree of underestimation in humans and monkeys. Correction using monkeys improved the accuracy of the human estimate, especially for basic and zwitterion compounds.A strategy is proposed based on the accuracy of prediction in monkey and monkey scalars for prediction and prospective assessment of the accuracy of human VD_ss.