A Systematic Review of Published Physiologically-based Kinetic Models and an Assessment of their Chemical Space Coverage

Courtney V. Thompson; James W. Firman; Michael R. Goldsmith; Christopher M. Grulke; Yu-Mei Tan; Alicia Paini; Peter E. Penson; Risa R. Sayre; Steven Webb; Judith C. Madden

doi:10.1177/02611929211060264

A Systematic Review of Published Physiologically-based Kinetic Models and an Assessment of their Chemical Space Coverage

Thompson, Courtney V.; Firman, James W.; Goldsmith, Michael R.; Grulke, Christopher M.; Tan, Yu-Mei; Paini, Alicia; Penson, Peter E.; Sayre, Risa R.; Webb, Steven; Madden, Judith C. 2021-09-01 00:00:00 Across multiple sectors, including food, cosmetics and pharmaceutical industries, there is a need to predict the potential effects of xenobiotics. These effects are determined by the intrinsic ability of the substance, or its derivatives, to interact with the biological system, and its concentration–time proﬁle at the target site. Physiologically-based kinetic (PBK) models can predict organ-level concentration–time proﬁles, however, the models are time and resource intensive to generate de novo. Read-across is an approach used to reduce or replace animal testing, wherein information from a data-rich chemical is used to make predictions for a data-poor chemical. The recent increase in published PBK models presents the opportunity to use a read-across approach for PBK modelling, that is, to use PBK model information from one chemical to inform the development or evaluation of a PBK model for a similar chemical. Essential to this process, is identifying the chemicals for which a PBK model already exists. Herein, the results of a systematic review of existing PBK models, compliant with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) format, are presented. Model information, including species, sex, life-stage, route of administration, software platform used and the availability of model equations, was captured for 7541 PBK models. Chemical information (identiﬁers and physico-chemical properties) has also been recorded for 1150 unique chemicals associated with these models. This PBK model data set has been made readily accessible, as a Microsoft Excel spreadsheet, providing a valuable resource for those developing, using or evaluating PBK models in industry, academia and the regulatory sectors. Keywords PBK, PBPK, PBTK, systematic review, pharmacokinetic modelling, read-across Introduction Humans, like other animals, are exposed daily to a multitude School of Pharmacy and Biomolecular Sciences, Liverpool John Moores of chemicals of anthropogenic origin, including pharma- University, Liverpool, UK ceuticals, food additives, pesticides, consumer goods and Ofﬁce of Research and Development, Center for Computational Toxicology cosmetic ingredients. The safety assessment of chemicals is and Exposure, Chemical Characterization and Exposure Division, US Environmental Protection Agency, Research Triangle Park, NC, USA a legal requirement that is essential to ensure their safe use Ofﬁce of Pesticide Programs, Health Effects Division, US Environmental by workers and consumers, and to ensure the protection of Protection Agency, Research Triangle Park, NC, USA domestic/farm animals and environmental species. How- European Commission Joint Research Centre (JRC), Ispra, Italy ever, for the majority of chemicals, there is a lack of Syngenta, Product Safety, Early Stage Research, Jealott’s Hill International available data for safety assessment — hence predictive Research Centre, Bracknell, UK models are essential. Predicting toxicity requires knowledge Corresponding author: of both the intrinsic activity of the chemical (or its deriv- Judith C Madden, School of Pharmacy and Biomolecular Sciences, atives) and the extent to which the organism is exposed. Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK. Whilst external exposure, or dose, has traditionally been Email: j.c.madden@ljmu.ac.uk 198 Alternatives to Laboratory Animals 49(5) Figure 1. The key characteristics of PBK models and the data captured in the PBK model data set. used in assessments, it is recognised that the dose at the combination with chemical-speciﬁc information (such as target site (i.e. organ-level exposure) is the more relevant solubility and partitioning behaviour) to predict the measure, being causally linked to observed toxicity. This concentration–time proﬁle of the chemical in tissues, cel- reasoning has long been applied in drug design, where the lular compartments or sub-compartments. Differential internal exposure level can be linked more reliably to the equations are used to describe the rate of change of con- desirable, therapeutic effect. As discussed by Pistollato centration of the chemical in each compartment, as sum- et al. in terms of legislation, kinetic data are a speciﬁc marised in Figure 1. Detailed information on how to requirement for plant protection and biocidal product safety construct and validate PBK models, their applications in assessment, and, whilst not formally required, the incor- different sectors and tools available to support PBK mod- 1,5–9 poration of such data is widely recommended in other elling have been well-reported previously. Of particular regulations such as Classiﬁcation, Labelling and Packaging note is the recent Organisation for Economic Co-operation (CLP) and the Registration, Evaluation, Authorisation and and Development (OECD) Guidance on the character- Restriction of Chemicals (REACH). Guidance documents isation, validation and reporting of Physiologically Based 3 10 from the European Chemicals Agency (ECHA) and the Kinetic (PBK) models for regulatory purposes. This Scientiﬁc Committee on Consumer Safety recommend document builds on the principles described in the World making use of all available data (including kinetic data) to Health Organisation (WHO) report of 2010, but focuses on support decision-making. Whilst general information re- the use of alternative approaches (in silico and in vitro) for garding absorption, distribution, metabolism or excretion parameterising PBK models. The potential for applying new (ADME) may be useful, more accurate prediction requires approach methodologies (NAMs) and next generation (NG) organ-level concentration–time proﬁles. Physiologically- methods to support the development and use of PBK models based kinetic (PBK) models (synonymous with physio- in safety assessment, was also promulgated by Paini and logically-based pharmacokinetic, toxicokinetic or bio- colleagues. PBK models can assimilate new information kinetic (PBPK, PBTK or PBBK) models) are employed in as it becomes available to increase predictive capacity; these numerous industries to provide such predictions. models provide an advantage over traditional one or two In a PBK model, the body is represented as a series of compartment kinetic models. compartments (e.g. individual organs) connected by blood For environmental chemicals, the numerous applications ﬂow. The models use knowledge of physiology and anat- of these models include: determining the dose at target omy (such as organ volumes and cardiac output), in tissues following external exposure; route-to-route Thompson et al. 199 extrapolation; dose extrapolation; inter-species and intra- gender, life-stage, route of administration, compartments species extrapolation (accounting for species, population or and PubMed ID for the source of the models. genetic variability through adaptation of physiological and An enriched version of this PBK Knowledgebase was anatomical parameters); in vitro-to-in-vivo extrapolation recently used as a proof-of-principle, to demonstrate that (IVIVE); ascertaining safe levels based on tissue dosimetry; information from an existing PBK model could be used, in a estimating chemical exposure from biomonitoring or epide- read-across approach, to inform safety assessment. In the miological data (by using reverse dosimetry); and assessing analysis, methyleugenol was considered as a target chem- potential for bioaccumulation. These applications comple- ical, with estragole and safrole being identiﬁed as suitable ment the traditional role of PBPK modelling of drugs where source chemicals (with respect to structural similarity). This they can be utilised to predict ﬁrst dose in man, potential for approach was also successful, exemplifying how infor- drug–drug interactions or the inﬂuence of health status (e.g. mation from an existing PBK model could assist the de- hepatic impairment) on kinetics. velopment of a model for a similar chemical. Making best The ECHA reports that read-across is the most commonly use of existing data and in particular the application of the used alternative method to reduce or replace animal testing in read-across approach are recognised as important tools in 14 24 safety assessment. In this approach, information from a data- reducing animal testing. In order to facilitate the appli- rich (source) chemical is used to predict information for a data- cation of this approach, it is essential to identify chemicals poor (target) substance that is considered similar. Kinetic for which PBK models are available. As ‘similarity’ is often information plays a key role in supporting read-across considered in relation to structure or physico-chemical 3,15 predictions and recent efforts have aimed to increase the properties, it is also important to ascertain the nature of accessibility of such data. Sayre and co-workers published a the chemicals for which models are available, comparing database of time-series concentration data, extracted from an their characteristics to existing chemical data sets. Having extensive search of the literature, and Pawar and co-workers information regarding the chemicals and the models in a identiﬁed 38 databases containing a range of ADME-relevant readily accessible and updateable resource would be a data, as part of their overall review of resources to support signiﬁcant asset for researchers, industry and regulators, read-across and in silico model development. PBK models with the potential to reduce the number of animals used in provide an additional opportunity to derive data to support drug development and chemical safety assessment. read-across. Data may be acquired either from a PBK model Several key features (which are represented in Figure 1) for the chemical under investigation (the target chemical) or characterise an individual PBK model and include species, from a PBK model for an existing chemical considered similar sex, life-stage, route of administration and the compartments to the target (a source chemical). This latter approach — required to accurately describe the time-course of the whereinanexistingPBK modelfor asourcechemicalisused chemical. In some models, key organs (such as the liver, as atemplatefor atargetchemical — is contingent upon the lungs, etc.) are incorporated individually as compartments; in identiﬁcation of existing, suitable PBK models. others, these are further divided into constituent sub- Over the past 30 years, the number of published PBK compartments (for example, considering histopathological models and their applications has increased signiﬁ- regions or explicitly including lymph or interstitial/vascular cantly. In 2016, Lu and co-workers published a PBK space) giving higher-level, more complex models. In other Knowledgebase, comprising 307 chemicals for which scenarios, organs are grouped together (referred to as PBK models were available from papers published be- ‘lumping’) to create simpler models, for example, all poorly tween1977and 2014. In their report, the authors de- perfused organs are considered as one compartment and all scribed two case studies wherein PBK models from the highly-perfused organs are considered as another. In addition Knowledgebase were used to inform the development of to the physiological and anatomical information required, PBK models for ‘similar’ chemicals. In their study, chemical-speciﬁc data are also a prerequisite. A substance chemical analogues were identiﬁed based on similarity of may be identiﬁed using common names or chemical iden- physico-chemical properties, although it is recognised tiﬁers such as the Chemical Abstracts Service (CAS) Registry that there is no consensus as to the best method to de- Number, a Simpliﬁed Molecular Input Line Entry System 20 21 termine similarity. Ellison and Wu successfully (SMILES) string or the International Chemical Identiﬁer demonstrated an analogous approach wherein a PBK Key (InChiKey). Ideally, multiple identiﬁers should be model for a target chemical was evaluated by using in- incorporated in the data set to avoid ambiguity. Model formation from source chemicals identiﬁed as structural development can be performed with a range of software, or functional analogues. In order to assist researchers in and the equations employed may be speciﬁed within the identifying existing PBK models, a spreadsheet of those publication itself or as part of the supplementary infor- collated from the literature by the US Environmental mation accompanying the article. Within this systematic Protection Agency (EPA), was made available via Fig- review, key model characteristics, such as species, sex, life- share. This resource included information on species, stage, route of administration, compartments, availability of 200 Alternatives to Laboratory Animals 49(5) model equations and chemical identiﬁers, were captured and keywords of papers within each database, across all years within the PBK model data set, as summarised in Figure 1. available. The systematic review management tool Covi- The second part of the analysis relates to the assessment of dence was used for processing papers for the review (https:// the chemical space coverage of the PBK model data set. www.covidence.org/; accessed May 2021). A total of 14,803 There is no simple process by which a chemical can be papers were initially identiﬁed; however, following auto- designated as being a particular ‘type’— for example, mated removal of duplicates in Covidence, 6771 remained. cosmetic ingredients may also be food additives, botanicals All abstracts were screened independently by two researchers may have pharmaceutical properties, etc. Consequently, in with all conﬂicts being resolved by discussion. The inclusion order to assess the nature of the chemicals in the PBK model criteria encompassed PBK models for all routes of admin- data set, key physico-chemical properties were generated and istration for chemical, biological and carrier systems, in- compared to those of chemicals appearing in other data sets. cluding cases where normal physiology was altered or The data sets studied were: botanicals, pesticides, pharma- interactions between administered substances were investi- ceuticals, food, cosmetic ingredients and REACH chemicals. gated. Models that could not be associated with a speciﬁc The number of chemicals in the PBK model data set that also substance (such as generic models applicable to large groups appeared in each of the other data sets was ascertained. of compounds) were excluded. Where an abstract was as- In summary, the aim of this systematic review was the sociated with a paper that had subsequently been retracted, it curation of a data resource for existing PBK models. was ensured that this model was excluded from the data set. Relevant information for the models (species, sex, life- Although standard practice in other systematic reviews, as- stage, substance identity, software used, etc.) was cap- sessment of the quality of the reported models and risk of bias tured in a ﬂexible spreadsheet format. The chemical space in reporting was considered unnecessary for this review. Our occupied by the PBK models (in terms of physico-chemical intention here was to document all available models, enabling properties) was compared to that of other chemical types, by interested researchers to rapidly identify potentially useful using six existing data sets. This resource has been created models to assist with future model development. The as- to assist the development and evaluation of PBK models sessment of PBK model quality (aside from fundamental based on existing data, thereby reducing the need to gen- considerations relating to good modelling practice) needs to be erate new data from animal studies. considered in terms of ﬁtness for a given purpose. Therefore, it is context dependent and remains the decision of the model user. Following abstract screening, 3120 abstracts were re- Methods tained for full text screening. PBK model data were extracted from 1649 of these papers, resulting in 7541 individual models Systematic review being captured. Note that, if oral and intravenous dosing were This systematic review was prospectively registered on used for both male and female subjects for the same chemical, PROSPERO, the National Institute for Health Research’s this would be extracted as four individual models, hence there international prospective registration system with the re- are many more models than individual chemicals. Reasons for view question stipulated as: “For which substances are the exclusion of papers during full text screening included: physiologically based kinetic (PBK) models available and PBK model not being reported in the article; the article ref- which species, genders, life-stages and routes of adminis- erenced a previously published model with no adaptations tration have been investigated for these substances? This (information on the PBK model was extracted from the will include determining the chemical space coverage of the original publication); and full article not being available in models and the availability of the associated model equa- English or not being reasonably accessible. tions within the literature”. The review complies with the PRISMA reporting standards; the PRISMA checklist is available as Supplementary Material S(i). Extraction of data from available Brieﬂy, following a scoping study of potentially useful physiologically-based kinetic models databases and search terms, Scopus (https://www.scopus.com/), PubMed (https://pubmed.ncbi.nlm.nih.gov/)and Webof Data were manually extracted from these 1649 papers by one Science (https://www.webofknowledge.com) were selected reviewer, with information being acquired from text, tables, as the most appropriate databases for identifying published ﬁgures and supplementary information. The data were en- papers on PBK models. The search of these databases was tered into a Microsoft Excel spreadsheet that captured completed in October 2020. The search terms (“pbpk” OR details of the chemical: parent and metabolites (speciﬁed “pbk” OR “pbbk” OR “pbtk” OR “pbpd” OR “pbpm” OR where appropriate), species (with sub-category where rele- “physiologically based”) AND (“pharmacokinetic” OR vant), sex, life-stage, route of administration, literature ref- “toxicokinetic” OR “biokinetic” OR “pharmacodynamics” erence for the model (with DOI), compartments considered in OR “biopharmaceutical”) were used to search abstracts, titles the model, the software employed and the availability of PBK Thompson et al. 201 Table 1. The number (and percentage) of chemicals with existing PBK models that are present in the six comparative data sets investigated. Number of chemicals in data set (with Number of chemicals in data set also Percentage of chemicals in data set also Data set unique, identiﬁable structure) present in PBK model data set present in PBK model data set PBK 1187 N/A N/A model Botanical 899 24 2.67 Pesticide 945 43 4.55 COSMOS 5105 125 2.45 Food 2615 89 3.40 DrugBank 6587 392 5.95 REACH 73,192 633 0.86 model equations within the article. Where possible, con- comprise: botanicals, pesticides, cosmetic ingredients (obtained trolled vocabulary was used to ensure consistency of data from COSMOS db version 2 (https://cosmosdb.eu; accessed extraction and to enable the resulting spreadsheet to be April 2021), food additives (obtained from http://foodb.ca; readily ﬁltered and searched for speciﬁc types of models, that accessed April 2021), pharmaceuticals (obtained from www. is, controlled vocabulary was used for species, sex, life-stage, drugbank.ca; accessed April 2021) and REACH chemicals (as route of administration, availability of equations and software summarised in Table 1). used. The vocabulary was empirically derived, to enable the Canonical SMILES for all chemicals in these data sets most efﬁcient searching — the full rationale is given in the were generated by using OpenBabel (v.3.0.0; http:// Supplementary Material S(ii). For example, life-stages openbabel.org/wiki/Main_Page; accessed April 2021). From can be reported in multiple ways — number of weeks, thePBK modeldataset,1150 uniqueSMILES were months, years of age, young adult, adult, neonate, young identiﬁed with 1187 unique InChiKeys (note that chemical child, juvenile, etc. Therefore, for consistency, this informa- isomers may have the same SMILES string but different tion was allocated to the more generic categories of: pre- InChiKeys). In order to determine how many chemicals birth or pre-hatch; from birth or hatch up to adult; adult; with PBK models were present in each of the other six data pregnant; old age (if speciﬁed); as well as a generic category sets, the InChiKeys were compared. for health-compromised (excluding old age) individuals. The SMILES strings for all data sets were inputted into Chemical identiﬁer information was obtained by manually the RDKit (v. 2020.03.6; www.rdkit.org) Descriptor Node, inputting the chemical name (as given in the publication) into accessed through KNIME software (v. 4.3.1; www.knime. PubChem (https://pubchem.ncbi.nlm.nih.gov/; last accessed com), in order to obtain the physico-chemical properties for May 2021) and extracting the molecular weight, canonical all chemicals. The properties included molecular weight, SMILES, isomeric SMILES, InChiKey and CAS number. The number of hydrogen bond donors/acceptors, predicted CAS Registry Number from PubChem was used as input for logarithm of the octanol:water partition coefﬁcient (SlogP) the COSMOS database, version 2 (https://cosmosdb.eu;ac- and the topological polar surface area (TPSA); the number cessed April 2021). Where available, the CAS Registry of Lipinski rule violations were calculated from this in- Number and chemical name, as recorded in COSMOS, were formation. Whilst it is possible to generate thousands of extracted to conﬁrm the identity of the chemical; the COS- physico-chemical properties, here only a few readily cal- MOS ID was also extracted. This information was captured by culable properties were selected, representing those most one reviewer. An assessment of the reliability of screening and often used to broadly characterise chemicals in terms of data extraction was undertaken and is reported below. size, polarity and partitioning behaviour. These simple properties were also used to determine Lipinski rule viola- tions (frequently used to indicate potential oral absorption — Assessment of the chemical space coverage of the a common route of administration for these models). The physiologically-based kinetic model data set in minimum, maximum, mean and median values, and the relation to other chemical data sets interquartile ranges of these properties, were calculated by using Minitab version 19.2 for all data sets. Histograms were In order to identify the nature of the chemicals in the PBK model also generated with Microsoft Excel to enable a visual data set and to compare the chemical space coverage (in terms of comparison of the property ranges between the different data key physico-chemical properties) six existing data sets were sets. The results of the statistical analysis are available as investigated. These have been compiled in-house at Liverpool Supplementary Material S(iii). John Moores University (Liverpool, United Kingdom) and 202 Alternatives to Laboratory Animals 49(5) Table 2. The number of models associated with different species. Species Number of models Percentage (%) Human 3676 48.8 Rat 2348 31.1 Mouse unmodiﬁed 839 11.1 Non-human primate 145 1.9 Dog 103 1.4 Porcine 94 1.2 Aquatic 82 1.1 Rabbit 70 0.9 Hamster 36 0.5 Mouse modiﬁed 36 0.5 Guinea-pig 29 <0.5 Bovine 26 <0.5 Bird 21 <0.5 Gerbil 10 <0.5 Horse 8 <0.5 Sheep 8 <0.5 Cat 4 <0.5 Figure 2. A summary of the papers considered at each stage of Goat 4 <0.5 the review process, and the total number of models extracted. Oyster 1 <0.5 Rodent combined 1 <0.5 Structural feature analysis models may be available in one paper. For example, if male In order to determine the relative frequency of the occur- and female subjects were used and doses given intrave- rence of speciﬁc structural features in the chemicals com- nously and by oral dosing, then this would be classed as four prising the PBK model data set, the SMILES strings were models. This approach enables more ﬂexible searching — entered into the Chemotyper software (version 1.0; Molec- searches can be conducted, not just by chemical names, but ular Networks, Erlangen, Germany; https://chemotyper.org/). also by species or route of administration, etc. A total of A table which indicated the presence or absence of 1889 chemical names are present in the PBK model data set. structural features identiﬁed by using ToxPrints (https:// Some represent biological entities (such as monoclonal toxprint.org/) was created and exported into Microsoft antibodies) or are not associated with a speciﬁc structure, Excel. In brief, ToxPrints represent a collection of 729 resulting in 1187 unique InChiKeys, that is, unique generic structural fragments covering (amongst others) chemicals with identiﬁable structures. Information con- commonly occurring functional groups, cyclic units and cerning the PBK models extracted is shown in Tables 2 and biomolecular substituents. 3: Table 2 shows the breakdown of models by species investigated and Table 3 shows the breakdown of models by Results route of administration. Systematic review Assessment of the chemical space coverage of the Figure 2 summarises the number of papers considered at physiologically-based kinetic model data set in each stage of the review process and the ﬁnal number of relation to other chemical data sets models extracted in the PBK model spreadsheet. Of the 6771 of papers initially identiﬁed, 3120 remained after The results for the comparison of InChiKeys for chemicals abstract screening and data were extracted from 1649 of in the PBK model data set to those for the six comparative these. data sets are shown in Table 1. Table 4 shows the results of the statistical analysis of the key physico-chemical properties (molecular weight, Extraction of data from available number of hydrogen bond donors and acceptors, logarithm physiologically-based kinetic models of the octanol:water partition coefﬁcient, topological sur- face area and number of Lipinski rule violations for the A total of 7541 individual models were identiﬁed and chemicals in the PBK model data set. extracted. Note that, for an individual substance, multiple Thompson et al. 203 The Supplementary Material S(iii) showsthe resultsofa Material S(ii) Sheet ﬁve; the complete output from Tox- similar statistical analysis of the physico-chemical properties for Prints for chemicals of the PBK model data set is available the six data sets to which the PBK model data set was compared. from the authors on request. Figures 3(a–f) shows the comparison of these key physico-chemical properties across the seven data sets. Assessment of screening and data extraction reliability Structural feature analysis After screening all 6771 abstracts in duplicate, 3120 were Analysis using ToxPrints showed that of the 729 chemo- taken forward to full text screening; of these, 1362 papers types, 458 were present at least once in chemicals of the were rejected at this stage. In addition, 109 papers could PBK model data set. Table 5 summarises the prevalence of not be readily obtained or were not in English. Therefore, 32 chemotypes (manually selected as illustrative examples data were extracted from 1649 papers, resulting in 7541 of common functional groups and other features of interest). models. The resultant spreadsheet comprises over 150,000 The prevalence of all chemotypes is given in Supplementary individual entries, as for each model, the species, sex, life- stage, route of administration, availability of equations, Table 3. The number of models associated with different routes compartments, references and chemical identiﬁers were of administration. captured. It is expected that errors will arise when as- Route of administration Number of models Percentage (%) sessing the suitability of papers for inclusion and per- forming extensive manual processing, hence a quality Metabolism from parent 2138 28.4 assessment exercise was undertaken. As part of this pro- Oral bolus 1903 25.2 cess, a representative sample from each of three stages of Inhalation 1193 15.8 the screening and data extraction process was assessed by a Intravenous bolus injection 1049 13.9 second investigator: Oral feed/water 381 5.1 Intravenous infusion 360 4.8 Dermal topical 181 2.4 — 5% of the papers that had been excluded at the full Dermal injection 82 1.1 text screening phase were reviewed; Intramuscular 59 0.8 — 5% of PBK model data extracted from the papers Intraperitoneal 58 0.8 (chemical information, species data (primary and Gills 36 0.5 secondary categories), sex, life-stage, route of In utero 35 0.5 administration, reference (DOI and PubMed ID if Unspeciﬁed 21 <0.5 available), compartments investigated, availability Intratracheal 17 <0.5 of equations and simulation software were Intraarterial 7 <0.5 checked; and Nasal 7 <0.5 — 5% of the chemical identiﬁer information from Buccal 6 <0.5 PubChem and COSMOS (chemical name, CAS Intramammary 4 <0.5 Registry Number, molecular weight, canonical Intraocular 2 <0.5 SMILES, isomeric SMILES, InChiKey and COSMOS Intrathecal 1 <0.5 ID) was obtained again from these sources and Intravaginal 1 <0.5 compared to the values in the spreadsheet. Table 4. Statistical analysis of the physico-chemical properties of the chemicals in the PBK model data set. Variable Mean Minimum Q1 Median Q3 Maximum Range IQR MW 325.61 6.94 163.10 292.28 410.67 6496.26 6489.26 247.57 No. HBD 1.79 0.00 0.00 1.00 2.00 100.00 100.00 2.00 No. HBA 4.17 0.00 1.00 3.00 6.00 120.00 120.00 5.00 SlogP 2.25 45.03 0.80 2.08 3.73 11.10 56.13 2.93 TPSA 73.07 0.00 23.47 56.93 93.00 2536.36 2536.36 69.53 nViolations 0.37 0.00 0.00 0.00 0.00 4.00 4.00 0.00 The variables are: MW = molecular weight; No. HBD = number of hydrogen bond donors; No. HBA = number of hydrogen bond acceptors; SlogP = predicted logarithm of the octanol:water partition coefﬁcient; TPSA = topological polar surface area; nViolations = number of violations of the Lipinski Rule of Five (Lipinski et al. ). The extreme values here are for vistarem , a magnetic resonance imaging contrast agent with large hydrophilic chains. 204 Alternatives to Laboratory Animals 49(5) Figure 3. A comparison of the ranges of physico-chemical properties across the seven data sets investigated. Thompson et al. 205 Table 5. The percentage of chemicals within the PBK model data set that contain the speciﬁed chemotypes identiﬁed by using ToxPrints. % % ToxPrint chemotype Prevalence ToxPrint chemotype Prevalence 6-Membered heterocycle 34.3 Organic sulphide/thiol 5.2 Carboxamide 27.7 Urea 4.8 Alcohol (aliphatic) 24.0 Ether (aromatic) 2.9 Amine (aliphatic) 22.6 Nitrile 2.7 Ether (aliphatic-aromatic) 21.9 Carboxylic acid (aromatic) 2.3 Organohalide (aromatic) 20.5 3-Membered heterocycle (e.g. epoxide) 2.2 Ether (aliphatic) 15.8 Transition metal 1.8 5-Membered heterocycle (single heteroatom, e.g. 13.6 4-Membered heterocycle (single heteroatom, 1.8 pyrrole) e.g. azetidine) Amine (aromatic) 12.6 Metalloid 1.7 Carboxylic acid (alkyl) 12.4 Imine 1.7 Organohalide (aliphatic) 11.2 Organophosphate (P = O) 1.6 Alcohol (aromatic) 10.6 Group I/II metal 1.4 5-Membered heterocycle (heteroatoms at 1,3- 10.6 Aldehyde 1.1 positions, e.g. imidazole) Ketone 10.5 Thiocarbonyl 0.8 Carboxylic acid ester (alkyl) 6.8 Alkyl chain (C length ≥8) 0.5 Sulphonyl 5.8 Organic azo 0.0 Note that only 32 selected chemotypes from the total of 458 identiﬁed within the data set are given in the table; the full list is given in Supplementary Material S(ii). The greatest source of ‘error’ was determined to be the Discussion exclusion of papers that were considered as potentially In this systematic review, information concerning over relevant by a second investigator, that is, 6% of excluded 7500 PBK models were extracted from 1649 papers. The papers. In terms of the systematic review, this is not con- models encompassed 18 species (including rat, human, sidered a highly signiﬁcant problem. PBK models are mouse, cow and guinea-pig) at various life-stages (e.g. continually being published, hence there can never be a juvenile, adult, pregnant and health-compromised) across ﬁnalised set of models. It is the intention to make this re- 21 administration routes (e.g. oral, inhalation and in utero). source available in its current form, as a tool to assist re- The information has been distilled into a Microsoft Excel searchers in ﬁnding relevant PBK models, and to update the spreadsheet that was constructed using controlled vo- resource in the future capturing models previously not cabulary to enable users to search by using different cri- identiﬁed or erroneously excluded. teria (e.g. to allow the selection of models by species or For PBK model data, manually extracted from the pa- pers, an error was detected in the information captured for routes of administration, etc.). It is anticipated that re- 3.5% of the substances. This does not equate to 3.5% of the searchers or regulatory scientists can use this information total information being incorrect, as this may indicate an to assist in the building or evaluation of new models, or as error in only one (or possibly more) of the 13 columns that a resource from which to extract relevant pharmacokinetic relate to the PBK model information. or toxicokinetic data. An error was detected in the data for 2.4% of the Although this is the largest collation of PBK models that chemicals in relation to the identiﬁer information. As above, the authors are aware of, it is not a complete list. As this does not equate to 2.4% of the total information being identiﬁed in the quality assessment exercise, some of the incorrect, but that for 2.4% of chemicals an error was de- historic models were omitted. In addition, as this is such a tected in one (or more) of the seven columns associated with dynamic area of research, the generation of a ﬁnite list of all chemical identiﬁer information. models would not be possible. The publication of new models has shown a rapidly increasing trajectory in recent The authors welcome any feedback from users regarding years. However, the current data set serves as a basis for errors, omissions of existing models or updates for new the continuing curation of existing models, which will models (note that models require a minimum level of in- provide an increasingly rich source of information for formation and novelty to be included); please email the modellers in the future. corresponding author. 206 Alternatives to Laboratory Animals 49(5) exceptions to this); here a similar pattern to the range of Trends in model availability: Coverage of values is generally observed for log P and molecular chemical types weight. For both properties, for the majority of chemicals Chemicals can be used for a variety of purposes, and often it the values fall within a narrow range; however, there are is not feasible to allocate a chemical to a unique group (e.g. also extreme values for a few chemicals. Pesticide and there is a signiﬁcant cross-over between chemicals used as botanical data sets have a greater percentage of chemicals food additives and as cosmetic ingredients, hence the same in the log P ranges 3–4 (43% and 47%, respectively), chemicals may appear in more than one of the different data whereas the PBK data set only has 28% of chemicals in sets). It is therefore difﬁcult to determine the chemical this range. ‘types’ for which there are the most PBK models; however, Pesticide and botanical data sets comprise fewer mole- some trends are discernible from the analysis undertaken. cules capable of carrying a charge (associated with in- Unsurprisingly, given that PBK modelling evolved in drug creased hydrophilicity) — hence, on average, they have development, the greatest proportion of models correspond higher log P values. This is signiﬁcant, as partitioning to chemicals in the DrugBank data set. Pesticides are behaviour (often estimated by using log P) is a key element generally well studied and data-rich; therefore, the second in building PBK models. Whilst the extreme values for log most common type of chemical with PBK models are the P, calculated by the software used here, may be unrealistic pesticides. For food additives and cosmetic ingredients, (and therefore unsuitable for model building), when used for where there are often chemicals in common, similar pro- comparison purposes they are still useful for demonstrating portions of chemicals have PBK models. Due to the size and the trends in the data. The range in values for all of the generality of the REACH data set, it would be anticipated physico-chemical properties of pesticides, is narrower than that relatively few chemicals would have existing PBK for the other chemical types, indicating the more restrictive models. The results conﬁrm the paucity of PBK models chemical properties required for these chemicals. available in relation to different areas of chemical space, and Botanicals generally show a wider range of values for underline the importance of using existing PBK models to each of the physico-chemical properties (in particular, more help ﬁll data gaps. chemicals show properties at the upper extremes of the ranges). A signiﬁcant number of compounds within this set are large and complex. Whilst in other data sets, molecules Trends in model availability: Coverage of tend to be designed for a speciﬁc purpose, and those out- physico-chemical properties with the given property ranges are ﬁltered out, the same Figure 3(a) shows the distribution of molecular weight exclusions would not be applicable to this data set. The across the seven data sets. As expected, the majority of the diversity of structural features present in the chemicals with chemicals fall within the range of 100–600 Da, but notable existing PBK models is demonstrated by the ToxPrints differences exist between the data sets. There are a relatively analysis, with 458 chemotypes identiﬁed as being present high number of chemicals in the PBK data set with low within this data set. The diversity of these chemicals, in molecular weight — these will include the volatile chem- terms of their molecular complexity, is demonstrated by the icals for which respiratory uptake has been extensively number of chemotypes identiﬁed — individual chemicals studied. Food additives and cosmetic ingredients (which were shown to contain between one and 69 chemotypes have chemicals in common) show a relatively high within their structure. proportion of low molecular weight chemicals. Chemicals that are designed to be biologically active, such as drugs Conclusion and pesticides, tend to be developed in accordance with guidelines relating to preferred physico-chemical prop- Understanding the kinetic behaviour of a chemical within erties. For example, the Lipinski Rule of Five stipulates the body, particularly its concentration–time proﬁle at a that drugs with poor oral absorption are associated with target site, is essential to accurately determine its potential chemicals with: molecular weight > 500 Da, log P > 5, effect. For the majority of chemicals, there is a lack of data and > 10 hydrogen bond acceptors or > 5 hydrogen bond concerning toxicity and kinetics. However, generating donors. Other research has also suggested that a to- such information de novo would require excessive use of pological polar surface area (TPSA) higher than 140 A is animals and is legally, ethically and ﬁnancially con- also unfavourable for oral absorption. Consequently, strained. Hence there is a need to leverage existing certain chemical types are designed to fall within nar- knowledge in order to obtain as much information as rower property ranges and such trends are evident in the possible to assist decision-making. Read-across is the most property ranges mentioned in this article. A correlation common method by which information from data-rich between molecular weight and log P is often observed chemicals is used to predict information for data-poor amongst groups of chemicals (although there are many chemicals. Herein is presented a comprehensive Thompson et al. 207 collation of existing PBK models that can be searched by Declaration of Conﬂicting Interests using multiple criteria. The physico-chemical space of The authors declared no potential conﬂicts of interest with respect PBK models has been mapped against that of other to the research, authorship, and/or publication of this article. chemical types — that is, food additives, cosmetic in- gredients, drugs, REACH chemicals, botanicals and pes- Funding ticides. Organising the current state of knowledge of The authors disclosed receipt of the following ﬁnancial support for existing PBK models provides a valuable resource for the research, authorship, and/or publication of this article: This those working in the area to identify models for chemicals work was supported by the European Partnership for Alternative of interest, or analogues, that can be used to assist the Approaches to animal testing (EPAA). development or evaluation of new PBK models. The concept of using existing PBK model information in a ORCID iDs read-across approach to develop a new PBK model for an analogous chemical has been demonstrated successfully in Courtney V. Thompson https://orcid.org/0000-0003-2243-2974 18,22 recent publications. Such an approach is recom- Peter E. Penson  https://orcid.org/0000-0001-6763-1489 mended in the recent OECD guidance on PBK modelling, which focuses on the use of alternative methods in PBK Supplementary Material model development. The PBK model data set described Supplementary material for this article is available online. herein enables researchers to readily gain insight into available PBK models across multiple species, life-stages References and routes of administration, such that the structure and parameterisation of PBK models for different chemicals is 1. International Programme on Chemical Safety and Inter- more accessible. This ensures maximum use of existing organisation Programme for the Sound Management of knowledge on PBK modelling, and reduces the time and Chemicals (2010). Characterization And Application of Physi- cost associated with the development of new PBK models. ologically Based Phamacokinetic Models in Risk Assessment. Continuing effort is required to curate existing PBK Geneva, Switzerland: World Health Organisation, 2010, 92 pp. models. The ability to extract relevant data from models and 2. Pistollato F, Madia F, Corvi R, et al. Current EU regulatory reproduce those published in the literature would be fa- requirements for the assessment of chemicals and cosmetic cilitated by researchers embracing the use of systematic products: Challenges and opportunities for introducing methods to record PBK models. Consistent formats for new approach methodologies. Arch Toxicol 2021; 95: 28,29 publishing PBK models have been proposed previously, 1867–1897. with the template proposed in the recent OECD guidance 3. ECHA. Read-Across Assessment Framework (RAAF). Lux- drawing on previous recommendations, most signiﬁcantly embourg, UK: EU Publications Ofﬁce, 2017, 60 pp. the reporting format proposed by Tan and co-workers. The 4. Scientiﬁc Committee on Consumer Safety. The SCCS notes of use of consistent reporting formats is strongly encouraged, as guidance for the testing of cosmetic ingredients and their this assists other researchers in re-using or re-purposing safety evaluation, 10th revision, 24–25 October 2018, https:// existing models. ec.europa.eu/health/sites/default/ﬁles/scientiﬁc_committees/ Future effort is required to curate available models, consumer_safety/docs/sccs_o_224.pdf, (2018, accessed 22 identify appropriate similarity metrics to assist in the November 2021). identiﬁcation of PBK models for analogues to chemicals of 5. US EPA. Approaches for the Application of Physiologically interest, and, ideally, make such resources widely available, Based Pharmacokinetic (PBPK) Models and Supporting for example, by incorporation into free webtools. Such Data in Risk Assessment (Final Report). Washington, DC: US endeavours will help to leverage the wealth of information Environmental Protection Agency, 2006. available in existing PBK models. 6. Kuepfer L, Niederalt C, Wendl T, et al. Applied concepts in PBPK modeling: How to build a PBPK/PD model. CPT Acknowledgments Pharmacometrics Syst Pharmacol 2016; 5: 516–531. JCM and CVT gratefully acknowledge the funding and scientiﬁcadvice 7. Upton RN, Foster DJ and Abuhelwa AY. An introduction to of the European Partnership for Alternative Approaches to animal physiologically-based pharmacokinetic models. Paediatr testing (EPAA) with respect to the funding of a studentship for CVT. Anaesth 2016; 26: 1036–1046. 8. Laroche C, Aggarwal M, Bender H, et al. Finding synergies for Author’s Note 3Rs — toxicokinetics and read-across: Report from an EPAA The views expressed in this article are those of the authors and do partners’ forum. Regul Toxicol Pharmacol 2018; 99: 5–21. not necessarily represent the views or the policies of the US 9. Madden JC, Pawar G, Cronin MTD, et al. In silico re- Environmental Protection Agency. sources to assist in the development and evaluation of 208 Alternatives to Laboratory Animals 49(5) physiologically-based kinetic models. Comput Toxicol 2019; 21. Ellison CA and Wu S. Application of structural and functional 11: 33–49. pharmacokinetic analogs for physiologically based pharma- 10. OECD. OECD Series on Testing and Assessment, No. 331, cokinetic model development and evaluation. Regul Toxicol Environment, Health and Safety, Environment Directorate. Pharmacol 2020; 114: 104667. Paris, France: Organisation for Economic Cooperation and 22. Sayre RR, Grulke C, Tan Y-M, et al. Existing physiologically- Development, 2021, 103 pp. based kinetic (PBK) models, identiﬁed via pubMed searches (with 11. Paini A, Leonard JA, Joossens E, et al. Next generation associated endnote library), https://ﬁgshare.com/articles/dataset/ physiologically based kinetic (NG-PBK) models in support of Existing_physiologically-based_kinetic_PBK_models_identiﬁed_ regulatory decision making. Comput Toxicol 2019; 9: 61–72. via_PubMed_searches_with_associated_EndNote_library_/ 12. Rowland M, Lesko LJ and Rostami-Hodjegan A. Physiolog- 10075574/1 (2019, accessed 21 September 2021). ically based pharmacokinetics is impacting drug development 23. Paini A, Worth A, Kulkarni S, et al. Assessment of the pre- and regulatory decision making. CPT Pharmacometrics Syst dictive capacity of a physiologically based kinetic model using Pharmacol 2015;4:313–315. a read-across approach. Comput Toxicol 2021; 18: 100159. 13. Perry C, Davis G, Conner TM, et al. Utilization of physio- 24. Madden JC, Enoch SJ, Paini AP, et al. A review of in silico logically based pharmacokinetic modeling in clinical phar- tools as alternatives to animal testing: Principles, resources macology and therapeutics: An overview. Curr Pharmacol and applications. Altern Lab Anim 2020; 48: 146–172. Rep 2020; 6: 71–84. 25. Thompson C, Madden J and Penson P. Systematic Review to 14. ECHA. The Use of Alternatives to Testing on Animals for the Determine the Chemical Space of Existing Physiologically- REACH Regulation. Fourth report under Article 117(3) of the Based Kinetic (PBK) Models, https://www.crd.york.ac.uk/ REACH Regulation. Helsinki, Finland: European Chemicals prospero/display_record.php?ID=CRD42020171130 (2020, Agency, 2020, 85 pp. accessed 21 September 2021). 15. Schultz TW, Amcoff P, Berggren E, et al. A strategy for 26. Paini A, Leonard JA, Kliment T, et al. Investigating the state structuring and reporting a read-across prediction of toxicity. of physiologically based kinetic modelling practices and Regul Toxicol Pharmacol 2015; 72: 586–601. challenges associated with gaining regulatory acceptance of 16. Sayre RR, Wambaugh JF and Grulke C. Database of phar- model applications. Regul Toxicol Pharm 2017; 90: 104–115. macokinetic time-series data and parameters for 144 envi- 27. Lipinski CA, Lombardo F, Dominy BW, et al. Experimental ronmental chemicals. Sci Data 2020; 7: 122. and computational approaches to estimate solubility and 17. Pawar G, Madden JC, Ebbrell D, et al. In silico toxicology permeability in drug discovery and development settings. Adv data resources to support read-across and (Q)SAR. Front Drug Deliver Rev 1997; 23: 3–25. Pharmacol 2019; 10: 561. 28. EMA. Guideline on the Reporting of Physiologically Based 18. Tan Y-M, Worley RR, Leonard JA, et al. Challenges asso- Pharmacokinetic (PBPK) Modelling and Simulation, https:// ciated with applying physiologically based pharmacokinetic www.ema.europa.eu/en/reporting-physiologically-based- modeling for public health decision-making. Toxicol Sci pharmacokinetic-pbpk-modelling-simulation (2018, accessed 2018; 162: 341–348. 21 September 2021). 19. Lu J, Goldsmith MR, Grulke C, et al. Developing a 29. US FDA. Physiologically Based Pharmacokinetic Analyses — physiologically-based pharmacokinetic model knowledgebase Format and Content, Guidance for Industry, https://www.fda. in support of provisional model construction. PLoS Comput gov/ﬁles/drugs/published/Physiologically-Based-Pharmacokinetic- Biol 2016; 12: e1004495. Analyses-%E2%80%94-Format-and-Content-Guidance- 20. Mellor CL, Marchese Robinson RL, Benigni R, et al. Mo- for-Industry.pdf (2018, accessed 21 September 2021). lecular ﬁngerprint-derived similarity measures for toxico- 30. Tan Y-M, Chan M, Chukwudebe A, et al. PBPK Model re- logical read-across: Recommendations for optimal use. Regul porting template for chemical risk assessment applications. Toxicol Pharmacol 2019; 101: 121–134. Regul Toxicol Pharmacol 2020; 115: 104691. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Alternative to Laboratory Animals SAGE http://www.deepdyve.com/lp/sage/a-systematic-review-of-published-physiologically-based-kinetic-models-v9aEgVLVom

Loading next page...

References (45)

J. Madden, S. Enoch, A. Paini, M. Cronin (2020)
A Review of In Silico Tools as Alternatives to Animal Testing: Principles, Resources and Applications
Alternatives to Laboratory Animals, 48
Gopal Pawar, J. Madden, D. Ebbrell, J. Firman, M. Cronin (2019)
In Silico Toxicology Data Resources to Support Read-Across and (Q)SAR
Frontiers in Pharmacology, 10
European Chemicals Agency (ECHA, 2017) Read-across assessment framework (RAAF)
(2021)
Systematic Review to Determine the Chemical Space of Existing PhysiologicallyBased Kinetic (PBK) Models, https://www.crd.york.ac.uk/ prospero/display_record.php?ID=CRD42020171130 (2020
(2021)
Physiologically Based Pharmacokinetic Analyses — Format and Content, Guidance for Industry, https://www.fda
Charles Laroche, M. Aggarwal, H. Bender, P. Benndorf, B. Birk, J. Crozier, G. Negro, Federica Gaetano, C. Desaintes, I. Gardner, B. Hubesch, A. Irizar, David John, Vikas Kumar, A. Lostia, I. Manou, M. Monshouwer, B. Müller, A. Paini, Kirsty Reid, T. Rowan, M. Sachana, Katrin Schutte, C. Stirling, R. Taalman, L. Aerts, Renate Weissenhorn, U. Sauer (2018)
Finding synergies for 3Rs – Toxicokinetics and read‐across: Report from an EPAA partners' Forum
Regulatory Toxicology and Pharmacology, 99
R. Upton, D. Foster, A. Abuhelwa (2016)
An introduction to physiologically‐based pharmacokinetic models
Pediatric Anesthesia, 26
(2006)
for the Application of Physiologically Based Pharmacokinetic (PBPK) Models and Supporting Data in Risk Assessment (Final Report)
(2017)
Read-Across Assessment Framework (RAAF)
A. Paini, J. Leonard, T. Kliment, Y. Tan, Andrew Worth (2017)
Investigating the state of physiologically based kinetic modelling practices and challenges associated with gaining regulatory acceptance of model applications
Regulatory Toxicology and Pharmacology, 90
T. Schultz, P. Amcoff, Elisabet Berggren, F. Gautier, M. Klarić, D. Knight, C. Mahony, Michael Schwarz, Anne White, M. Cronin (2015)
A strategy for structuring and reporting a read-across prediction of toxicity.
Regulatory toxicology and pharmacology : RTP, 72 3
(2019)
In silico resources to assist in the development and evaluation of Thompson et al. 207 physiologically-based kinetic models
(2017)
Framework (RAAF)
Opinion β-arbutin, U. Bernauer, Q. Chaudhry, P. Coenraads, Gisela Degen, M. Dusinska, Werner Lilienblum, Elsa Nielsen, T. Platzek, S. Rastogi, C. Rousselle, J. Benthem (2010)
Scientific Committee on Consumer Safety SCCS
M. Hughes, J. Pfeffer, M. Martig, N. Bastian, R. Crain, J. Kruijssen, M. Reina-Campos (2018)
Fossil stellar streams and their globular cluster populations in the E-MOSAICS simulations
Monthly Notices of the Royal Astronomical Society
M. Rowland, L. Lesko, A. Rostami-Hodjegan (2015)
Physiologically Based Pharmacokinetics Is Impacting Drug Development and Regulatory Decision Making
CPT: Pharmacometrics & Systems Pharmacology, 4
(2021)
Guidance document on the characterisation, validation and reporting of Physiologically Based Kinetic (PBK) models for regulatory purposes
(2018)
Physiologically based pharmacokinetic analyses -Format and content Guidance for Industry
C. Lipinski, F. Lombardo, B. Dominy, P. Feeney (2001)
Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings.
Advanced drug delivery reviews, 46 1-3
Existing physiologicallybased kinetic (PBK) models, identified via pubMed searches
L. Kuepfer, Christoph Niederalt, Thomas Wendl, Jan‐Frederik Schlender, S. Willmann, J. Lippert, M. Block, T. Eissing, D. Teutonico (2016)
Applied Concepts in PBPK Modeling: How to Build a PBPK/PD Model
CPT: Pharmacometrics & Systems Pharmacology, 5
A. Paini, Andrew Worth, S. Kulkarni, D. Ebbrell, J. Madden (2021)
Assessment of the predictive capacity of a physiologically based kinetic model using a read-across approach
Computational Toxicology (Amsterdam, Netherlands), 18
(2014)
THE USE OF ALTERNATIVES TO TESTING ON ANIMALS FOR THE REACH REGULATION
(2018)
Guideline on the reporting of physiologically based pharmacokinetic (PBPK) modelling and simulation
Analyses-%E2%80%94-Format-and-Content-Guidance-for-Industry.pdf
Scientific Committee on Consumer Safety. The SCCS notes of guidance for the testing of cosmetic ingredients and their safety evaluation, 10th revision
Courtney Perry, Grace Davis, T. Conner, Tao Zhang (2020)
Utilization of Physiologically Based Pharmacokinetic Modeling in Clinical Pharmacology and Therapeutics: an Overview
Current Pharmacology Reports, 6
(2010)
International Programme on Chemical Safety and Interorganisation Programme for the Sound Management of Chemicals
Y. Tan, M. Chan, A. Chukwudebe, J. Domoradzki, J. Fisher, C. Hack, P. Hinderliter, Kota Hirasawa, J. Leonard, A. Lumen, A. Paini, H. Qian, P. Ruiz, J. Wambaugh, Fagen Zhang, M. Embry (2020)
PBPK model reporting template for chemical risk assessment applications
Regulatory toxicology and pharmacology : RTP, 115
H. Buist (2010)
Characterization and application of physiologically based pharmacokinetic models in risk assessment
F. Pistollato, F. Madia, R. Corvi, S. Munn, E. Grignard, A. Paini, Andrew Worth, A. Bal-Price, P. Prieto, S. Casati, Elisabet Berggren, S. Bopp, V. Zuang (2021)
Current EU regulatory requirements for the assessment of chemicals and cosmetic products: challenges and opportunities for introducing new approach methodologies
Archives of Toxicology, 95
C. Ellison, Shengde Wu (2020)
Application of structural and functional pharmacokinetic analogs for physiologically based pharmacokinetic model development and evaluation.
Regulatory toxicology and pharmacology : RTP
Guideline on the reporting of physiologically based pharmacokinetic (PBPK) modelling and simulation EMA/CHMP/458101/2016
(2021)
on Testing and Assessment, No
J. Madden, Gopal Pawar, M. Cronin, S. Webb, Y. Tan, A. Paini (2019)
In silico resources to assist in the development and evaluation of physiologically-based kinetic models
Computational Toxicology
Jingtao Lu, M. Goldsmith, C. Grulke, D. Chang, R. Brooks, J. Leonard, M. Phillips, Ethan Hypes, Matthew Fair, R. Tornero-Velez, Jeffre Johnson, C. Dary, Y. Tan (2016)
Developing a Physiologically-Based Pharmacokinetic Model Knowledgebase in Support of Provisional Model Construction
PLoS Computational Biology, 12
(2021)
Research Protocol Registration: Systematic review to determine the chemical space of existing physiologically-based kinetic (PBK) models PROSPERO 2020 CRD42020171130
(2020)
of Alternatives to Testing on Animals for the REACH Regulation
Risa Sayre, J. Wambaugh, C. Grulke (2020)
Database of pharmacokinetic time-series data and parameters for 144 environmental chemicals
Scientific Data, 7
Y. Tan, R. Worley, J. Leonard, J. Fisher (2018)
Challenges Associated With Applying Physiologically Based Pharmacokinetic Modeling for Public Health Decision-Making
Toxicological Sciences, 162
(2020)
Existing physiologically - based kinetic ( PBK ) models , identi fi ed via pubMed searches ( with associated endnote library )
Systematic Review to Determine the Chemical Space of Existing Physiologically-Based Kinetic (PBK) Models
(2021)
Environment, Health and Safety, Environment Directorate
A. Paini, J. Leonard, E. Joossens, J. Bessems, A. Desalegn, J. Dorne, J. Gosling, M. Heringa, M. Klarić, T. Kliment, N. Kramer, G. Loizou, J. Louisse, A. Lumen, J. Madden, E. Patterson, S. Proença, A. Punt, R. Setzer, N. Suciu, J. Troutman, M. Yoon, Andrew Worth, Y.M. Tan (2019)
Next generation physiologically based kinetic (NG-PBK) models in support of regulatory decision making
Computational Toxicology (Amsterdam, Netherlands), 9
(2006)
Approaches for the application of physiologically based pharmacokinetic (PBPK) models and supporting data in risk assessment (final report)

Publisher: SAGE
Copyright: © The Author(s) 2021
ISSN: 0261-1929
eISSN: 2632-3559
DOI: 10.1177/02611929211060264
Publisher site: See Article on Publisher Site

Abstract

Across multiple sectors, including food, cosmetics and pharmaceutical industries, there is a need to predict the potential effects of xenobiotics. These effects are determined by the intrinsic ability of the substance, or its derivatives, to interact with the biological system, and its concentration–time proﬁle at the target site. Physiologically-based kinetic (PBK) models can predict organ-level concentration–time proﬁles, however, the models are time and resource intensive to generate de novo. Read-across is an approach used to reduce or replace animal testing, wherein information from a data-rich chemical is used to make predictions for a data-poor chemical. The recent increase in published PBK models presents the opportunity to use a read-across approach for PBK modelling, that is, to use PBK model information from one chemical to inform the development or evaluation of a PBK model for a similar chemical. Essential to this process, is identifying the chemicals for which a PBK model already exists. Herein, the results of a systematic review of existing PBK models, compliant with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) format, are presented. Model information, including species, sex, life-stage, route of administration, software platform used and the availability of model equations, was captured for 7541 PBK models. Chemical information (identiﬁers and physico-chemical properties) has also been recorded for 1150 unique chemicals associated with these models. This PBK model data set has been made readily accessible, as a Microsoft Excel spreadsheet, providing a valuable resource for those developing, using or evaluating PBK models in industry, academia and the regulatory sectors. Keywords PBK, PBPK, PBTK, systematic review, pharmacokinetic modelling, read-across Introduction Humans, like other animals, are exposed daily to a multitude School of Pharmacy and Biomolecular Sciences, Liverpool John Moores of chemicals of anthropogenic origin, including pharma- University, Liverpool, UK ceuticals, food additives, pesticides, consumer goods and Ofﬁce of Research and Development, Center for Computational Toxicology cosmetic ingredients. The safety assessment of chemicals is and Exposure, Chemical Characterization and Exposure Division, US Environmental Protection Agency, Research Triangle Park, NC, USA a legal requirement that is essential to ensure their safe use Ofﬁce of Pesticide Programs, Health Effects Division, US Environmental by workers and consumers, and to ensure the protection of Protection Agency, Research Triangle Park, NC, USA domestic/farm animals and environmental species. How- European Commission Joint Research Centre (JRC), Ispra, Italy ever, for the majority of chemicals, there is a lack of Syngenta, Product Safety, Early Stage Research, Jealott’s Hill International available data for safety assessment — hence predictive Research Centre, Bracknell, UK models are essential. Predicting toxicity requires knowledge Corresponding author: of both the intrinsic activity of the chemical (or its deriv- Judith C Madden, School of Pharmacy and Biomolecular Sciences, atives) and the extent to which the organism is exposed. Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK. Whilst external exposure, or dose, has traditionally been Email: j.c.madden@ljmu.ac.uk 198 Alternatives to Laboratory Animals 49(5) Figure 1. The key characteristics of PBK models and the data captured in the PBK model data set. used in assessments, it is recognised that the dose at the combination with chemical-speciﬁc information (such as target site (i.e. organ-level exposure) is the more relevant solubility and partitioning behaviour) to predict the measure, being causally linked to observed toxicity. This concentration–time proﬁle of the chemical in tissues, cel- reasoning has long been applied in drug design, where the lular compartments or sub-compartments. Differential internal exposure level can be linked more reliably to the equations are used to describe the rate of change of con- desirable, therapeutic effect. As discussed by Pistollato centration of the chemical in each compartment, as sum- et al. in terms of legislation, kinetic data are a speciﬁc marised in Figure 1. Detailed information on how to requirement for plant protection and biocidal product safety construct and validate PBK models, their applications in assessment, and, whilst not formally required, the incor- different sectors and tools available to support PBK mod- 1,5–9 poration of such data is widely recommended in other elling have been well-reported previously. Of particular regulations such as Classiﬁcation, Labelling and Packaging note is the recent Organisation for Economic Co-operation (CLP) and the Registration, Evaluation, Authorisation and and Development (OECD) Guidance on the character- Restriction of Chemicals (REACH). Guidance documents isation, validation and reporting of Physiologically Based 3 10 from the European Chemicals Agency (ECHA) and the Kinetic (PBK) models for regulatory purposes. This Scientiﬁc Committee on Consumer Safety recommend document builds on the principles described in the World making use of all available data (including kinetic data) to Health Organisation (WHO) report of 2010, but focuses on support decision-making. Whilst general information re- the use of alternative approaches (in silico and in vitro) for garding absorption, distribution, metabolism or excretion parameterising PBK models. The potential for applying new (ADME) may be useful, more accurate prediction requires approach methodologies (NAMs) and next generation (NG) organ-level concentration–time proﬁles. Physiologically- methods to support the development and use of PBK models based kinetic (PBK) models (synonymous with physio- in safety assessment, was also promulgated by Paini and logically-based pharmacokinetic, toxicokinetic or bio- colleagues. PBK models can assimilate new information kinetic (PBPK, PBTK or PBBK) models) are employed in as it becomes available to increase predictive capacity; these numerous industries to provide such predictions. models provide an advantage over traditional one or two In a PBK model, the body is represented as a series of compartment kinetic models. compartments (e.g. individual organs) connected by blood For environmental chemicals, the numerous applications ﬂow. The models use knowledge of physiology and anat- of these models include: determining the dose at target omy (such as organ volumes and cardiac output), in tissues following external exposure; route-to-route Thompson et al. 199 extrapolation; dose extrapolation; inter-species and intra- gender, life-stage, route of administration, compartments species extrapolation (accounting for species, population or and PubMed ID for the source of the models. genetic variability through adaptation of physiological and An enriched version of this PBK Knowledgebase was anatomical parameters); in vitro-to-in-vivo extrapolation recently used as a proof-of-principle, to demonstrate that (IVIVE); ascertaining safe levels based on tissue dosimetry; information from an existing PBK model could be used, in a estimating chemical exposure from biomonitoring or epide- read-across approach, to inform safety assessment. In the miological data (by using reverse dosimetry); and assessing analysis, methyleugenol was considered as a target chem- potential for bioaccumulation. These applications comple- ical, with estragole and safrole being identiﬁed as suitable ment the traditional role of PBPK modelling of drugs where source chemicals (with respect to structural similarity). This they can be utilised to predict ﬁrst dose in man, potential for approach was also successful, exemplifying how infor- drug–drug interactions or the inﬂuence of health status (e.g. mation from an existing PBK model could assist the de- hepatic impairment) on kinetics. velopment of a model for a similar chemical. Making best The ECHA reports that read-across is the most commonly use of existing data and in particular the application of the used alternative method to reduce or replace animal testing in read-across approach are recognised as important tools in 14 24 safety assessment. In this approach, information from a data- reducing animal testing. In order to facilitate the appli- rich (source) chemical is used to predict information for a data- cation of this approach, it is essential to identify chemicals poor (target) substance that is considered similar. Kinetic for which PBK models are available. As ‘similarity’ is often information plays a key role in supporting read-across considered in relation to structure or physico-chemical 3,15 predictions and recent efforts have aimed to increase the properties, it is also important to ascertain the nature of accessibility of such data. Sayre and co-workers published a the chemicals for which models are available, comparing database of time-series concentration data, extracted from an their characteristics to existing chemical data sets. Having extensive search of the literature, and Pawar and co-workers information regarding the chemicals and the models in a identiﬁed 38 databases containing a range of ADME-relevant readily accessible and updateable resource would be a data, as part of their overall review of resources to support signiﬁcant asset for researchers, industry and regulators, read-across and in silico model development. PBK models with the potential to reduce the number of animals used in provide an additional opportunity to derive data to support drug development and chemical safety assessment. read-across. Data may be acquired either from a PBK model Several key features (which are represented in Figure 1) for the chemical under investigation (the target chemical) or characterise an individual PBK model and include species, from a PBK model for an existing chemical considered similar sex, life-stage, route of administration and the compartments to the target (a source chemical). This latter approach — required to accurately describe the time-course of the whereinanexistingPBK modelfor asourcechemicalisused chemical. In some models, key organs (such as the liver, as atemplatefor atargetchemical — is contingent upon the lungs, etc.) are incorporated individually as compartments; in identiﬁcation of existing, suitable PBK models. others, these are further divided into constituent sub- Over the past 30 years, the number of published PBK compartments (for example, considering histopathological models and their applications has increased signiﬁ- regions or explicitly including lymph or interstitial/vascular cantly. In 2016, Lu and co-workers published a PBK space) giving higher-level, more complex models. In other Knowledgebase, comprising 307 chemicals for which scenarios, organs are grouped together (referred to as PBK models were available from papers published be- ‘lumping’) to create simpler models, for example, all poorly tween1977and 2014. In their report, the authors de- perfused organs are considered as one compartment and all scribed two case studies wherein PBK models from the highly-perfused organs are considered as another. In addition Knowledgebase were used to inform the development of to the physiological and anatomical information required, PBK models for ‘similar’ chemicals. In their study, chemical-speciﬁc data are also a prerequisite. A substance chemical analogues were identiﬁed based on similarity of may be identiﬁed using common names or chemical iden- physico-chemical properties, although it is recognised tiﬁers such as the Chemical Abstracts Service (CAS) Registry that there is no consensus as to the best method to de- Number, a Simpliﬁed Molecular Input Line Entry System 20 21 termine similarity. Ellison and Wu successfully (SMILES) string or the International Chemical Identiﬁer demonstrated an analogous approach wherein a PBK Key (InChiKey). Ideally, multiple identiﬁers should be model for a target chemical was evaluated by using in- incorporated in the data set to avoid ambiguity. Model formation from source chemicals identiﬁed as structural development can be performed with a range of software, or functional analogues. In order to assist researchers in and the equations employed may be speciﬁed within the identifying existing PBK models, a spreadsheet of those publication itself or as part of the supplementary infor- collated from the literature by the US Environmental mation accompanying the article. Within this systematic Protection Agency (EPA), was made available via Fig- review, key model characteristics, such as species, sex, life- share. This resource included information on species, stage, route of administration, compartments, availability of 200 Alternatives to Laboratory Animals 49(5) model equations and chemical identiﬁers, were captured and keywords of papers within each database, across all years within the PBK model data set, as summarised in Figure 1. available. The systematic review management tool Covi- The second part of the analysis relates to the assessment of dence was used for processing papers for the review (https:// the chemical space coverage of the PBK model data set. www.covidence.org/; accessed May 2021). A total of 14,803 There is no simple process by which a chemical can be papers were initially identiﬁed; however, following auto- designated as being a particular ‘type’— for example, mated removal of duplicates in Covidence, 6771 remained. cosmetic ingredients may also be food additives, botanicals All abstracts were screened independently by two researchers may have pharmaceutical properties, etc. Consequently, in with all conﬂicts being resolved by discussion. The inclusion order to assess the nature of the chemicals in the PBK model criteria encompassed PBK models for all routes of admin- data set, key physico-chemical properties were generated and istration for chemical, biological and carrier systems, in- compared to those of chemicals appearing in other data sets. cluding cases where normal physiology was altered or The data sets studied were: botanicals, pesticides, pharma- interactions between administered substances were investi- ceuticals, food, cosmetic ingredients and REACH chemicals. gated. Models that could not be associated with a speciﬁc The number of chemicals in the PBK model data set that also substance (such as generic models applicable to large groups appeared in each of the other data sets was ascertained. of compounds) were excluded. Where an abstract was as- In summary, the aim of this systematic review was the sociated with a paper that had subsequently been retracted, it curation of a data resource for existing PBK models. was ensured that this model was excluded from the data set. Relevant information for the models (species, sex, life- Although standard practice in other systematic reviews, as- stage, substance identity, software used, etc.) was cap- sessment of the quality of the reported models and risk of bias tured in a ﬂexible spreadsheet format. The chemical space in reporting was considered unnecessary for this review. Our occupied by the PBK models (in terms of physico-chemical intention here was to document all available models, enabling properties) was compared to that of other chemical types, by interested researchers to rapidly identify potentially useful using six existing data sets. This resource has been created models to assist with future model development. The as- to assist the development and evaluation of PBK models sessment of PBK model quality (aside from fundamental based on existing data, thereby reducing the need to gen- considerations relating to good modelling practice) needs to be erate new data from animal studies. considered in terms of ﬁtness for a given purpose. Therefore, it is context dependent and remains the decision of the model user. Following abstract screening, 3120 abstracts were re- Methods tained for full text screening. PBK model data were extracted from 1649 of these papers, resulting in 7541 individual models Systematic review being captured. Note that, if oral and intravenous dosing were This systematic review was prospectively registered on used for both male and female subjects for the same chemical, PROSPERO, the National Institute for Health Research’s this would be extracted as four individual models, hence there international prospective registration system with the re- are many more models than individual chemicals. Reasons for view question stipulated as: “For which substances are the exclusion of papers during full text screening included: physiologically based kinetic (PBK) models available and PBK model not being reported in the article; the article ref- which species, genders, life-stages and routes of adminis- erenced a previously published model with no adaptations tration have been investigated for these substances? This (information on the PBK model was extracted from the will include determining the chemical space coverage of the original publication); and full article not being available in models and the availability of the associated model equa- English or not being reasonably accessible. tions within the literature”. The review complies with the PRISMA reporting standards; the PRISMA checklist is available as Supplementary Material S(i). Extraction of data from available Brieﬂy, following a scoping study of potentially useful physiologically-based kinetic models databases and search terms, Scopus (https://www.scopus.com/), PubMed (https://pubmed.ncbi.nlm.nih.gov/)and Webof Data were manually extracted from these 1649 papers by one Science (https://www.webofknowledge.com) were selected reviewer, with information being acquired from text, tables, as the most appropriate databases for identifying published ﬁgures and supplementary information. The data were en- papers on PBK models. The search of these databases was tered into a Microsoft Excel spreadsheet that captured completed in October 2020. The search terms (“pbpk” OR details of the chemical: parent and metabolites (speciﬁed “pbk” OR “pbbk” OR “pbtk” OR “pbpd” OR “pbpm” OR where appropriate), species (with sub-category where rele- “physiologically based”) AND (“pharmacokinetic” OR vant), sex, life-stage, route of administration, literature ref- “toxicokinetic” OR “biokinetic” OR “pharmacodynamics” erence for the model (with DOI), compartments considered in OR “biopharmaceutical”) were used to search abstracts, titles the model, the software employed and the availability of PBK Thompson et al. 201 Table 1. The number (and percentage) of chemicals with existing PBK models that are present in the six comparative data sets investigated. Number of chemicals in data set (with Number of chemicals in data set also Percentage of chemicals in data set also Data set unique, identiﬁable structure) present in PBK model data set present in PBK model data set PBK 1187 N/A N/A model Botanical 899 24 2.67 Pesticide 945 43 4.55 COSMOS 5105 125 2.45 Food 2615 89 3.40 DrugBank 6587 392 5.95 REACH 73,192 633 0.86 model equations within the article. Where possible, con- comprise: botanicals, pesticides, cosmetic ingredients (obtained trolled vocabulary was used to ensure consistency of data from COSMOS db version 2 (https://cosmosdb.eu; accessed extraction and to enable the resulting spreadsheet to be April 2021), food additives (obtained from http://foodb.ca; readily ﬁltered and searched for speciﬁc types of models, that accessed April 2021), pharmaceuticals (obtained from www. is, controlled vocabulary was used for species, sex, life-stage, drugbank.ca; accessed April 2021) and REACH chemicals (as route of administration, availability of equations and software summarised in Table 1). used. The vocabulary was empirically derived, to enable the Canonical SMILES for all chemicals in these data sets most efﬁcient searching — the full rationale is given in the were generated by using OpenBabel (v.3.0.0; http:// Supplementary Material S(ii). For example, life-stages openbabel.org/wiki/Main_Page; accessed April 2021). From can be reported in multiple ways — number of weeks, thePBK modeldataset,1150 uniqueSMILES were months, years of age, young adult, adult, neonate, young identiﬁed with 1187 unique InChiKeys (note that chemical child, juvenile, etc. Therefore, for consistency, this informa- isomers may have the same SMILES string but different tion was allocated to the more generic categories of: pre- InChiKeys). In order to determine how many chemicals birth or pre-hatch; from birth or hatch up to adult; adult; with PBK models were present in each of the other six data pregnant; old age (if speciﬁed); as well as a generic category sets, the InChiKeys were compared. for health-compromised (excluding old age) individuals. The SMILES strings for all data sets were inputted into Chemical identiﬁer information was obtained by manually the RDKit (v. 2020.03.6; www.rdkit.org) Descriptor Node, inputting the chemical name (as given in the publication) into accessed through KNIME software (v. 4.3.1; www.knime. PubChem (https://pubchem.ncbi.nlm.nih.gov/; last accessed com), in order to obtain the physico-chemical properties for May 2021) and extracting the molecular weight, canonical all chemicals. The properties included molecular weight, SMILES, isomeric SMILES, InChiKey and CAS number. The number of hydrogen bond donors/acceptors, predicted CAS Registry Number from PubChem was used as input for logarithm of the octanol:water partition coefﬁcient (SlogP) the COSMOS database, version 2 (https://cosmosdb.eu;ac- and the topological polar surface area (TPSA); the number cessed April 2021). Where available, the CAS Registry of Lipinski rule violations were calculated from this in- Number and chemical name, as recorded in COSMOS, were formation. Whilst it is possible to generate thousands of extracted to conﬁrm the identity of the chemical; the COS- physico-chemical properties, here only a few readily cal- MOS ID was also extracted. This information was captured by culable properties were selected, representing those most one reviewer. An assessment of the reliability of screening and often used to broadly characterise chemicals in terms of data extraction was undertaken and is reported below. size, polarity and partitioning behaviour. These simple properties were also used to determine Lipinski rule viola- tions (frequently used to indicate potential oral absorption — Assessment of the chemical space coverage of the a common route of administration for these models). The physiologically-based kinetic model data set in minimum, maximum, mean and median values, and the relation to other chemical data sets interquartile ranges of these properties, were calculated by using Minitab version 19.2 for all data sets. Histograms were In order to identify the nature of the chemicals in the PBK model also generated with Microsoft Excel to enable a visual data set and to compare the chemical space coverage (in terms of comparison of the property ranges between the different data key physico-chemical properties) six existing data sets were sets. The results of the statistical analysis are available as investigated. These have been compiled in-house at Liverpool Supplementary Material S(iii). John Moores University (Liverpool, United Kingdom) and 202 Alternatives to Laboratory Animals 49(5) Table 2. The number of models associated with different species. Species Number of models Percentage (%) Human 3676 48.8 Rat 2348 31.1 Mouse unmodiﬁed 839 11.1 Non-human primate 145 1.9 Dog 103 1.4 Porcine 94 1.2 Aquatic 82 1.1 Rabbit 70 0.9 Hamster 36 0.5 Mouse modiﬁed 36 0.5 Guinea-pig 29 <0.5 Bovine 26 <0.5 Bird 21 <0.5 Gerbil 10 <0.5 Horse 8 <0.5 Sheep 8 <0.5 Cat 4 <0.5 Figure 2. A summary of the papers considered at each stage of Goat 4 <0.5 the review process, and the total number of models extracted. Oyster 1 <0.5 Rodent combined 1 <0.5 Structural feature analysis models may be available in one paper. For example, if male In order to determine the relative frequency of the occur- and female subjects were used and doses given intrave- rence of speciﬁc structural features in the chemicals com- nously and by oral dosing, then this would be classed as four prising the PBK model data set, the SMILES strings were models. This approach enables more ﬂexible searching — entered into the Chemotyper software (version 1.0; Molec- searches can be conducted, not just by chemical names, but ular Networks, Erlangen, Germany; https://chemotyper.org/). also by species or route of administration, etc. A total of A table which indicated the presence or absence of 1889 chemical names are present in the PBK model data set. structural features identiﬁed by using ToxPrints (https:// Some represent biological entities (such as monoclonal toxprint.org/) was created and exported into Microsoft antibodies) or are not associated with a speciﬁc structure, Excel. In brief, ToxPrints represent a collection of 729 resulting in 1187 unique InChiKeys, that is, unique generic structural fragments covering (amongst others) chemicals with identiﬁable structures. Information con- commonly occurring functional groups, cyclic units and cerning the PBK models extracted is shown in Tables 2 and biomolecular substituents. 3: Table 2 shows the breakdown of models by species investigated and Table 3 shows the breakdown of models by Results route of administration. Systematic review Assessment of the chemical space coverage of the Figure 2 summarises the number of papers considered at physiologically-based kinetic model data set in each stage of the review process and the ﬁnal number of relation to other chemical data sets models extracted in the PBK model spreadsheet. Of the 6771 of papers initially identiﬁed, 3120 remained after The results for the comparison of InChiKeys for chemicals abstract screening and data were extracted from 1649 of in the PBK model data set to those for the six comparative these. data sets are shown in Table 1. Table 4 shows the results of the statistical analysis of the key physico-chemical properties (molecular weight, Extraction of data from available number of hydrogen bond donors and acceptors, logarithm physiologically-based kinetic models of the octanol:water partition coefﬁcient, topological sur- face area and number of Lipinski rule violations for the A total of 7541 individual models were identiﬁed and chemicals in the PBK model data set. extracted. Note that, for an individual substance, multiple Thompson et al. 203 The Supplementary Material S(iii) showsthe resultsofa Material S(ii) Sheet ﬁve; the complete output from Tox- similar statistical analysis of the physico-chemical properties for Prints for chemicals of the PBK model data set is available the six data sets to which the PBK model data set was compared. from the authors on request. Figures 3(a–f) shows the comparison of these key physico-chemical properties across the seven data sets. Assessment of screening and data extraction reliability Structural feature analysis After screening all 6771 abstracts in duplicate, 3120 were Analysis using ToxPrints showed that of the 729 chemo- taken forward to full text screening; of these, 1362 papers types, 458 were present at least once in chemicals of the were rejected at this stage. In addition, 109 papers could PBK model data set. Table 5 summarises the prevalence of not be readily obtained or were not in English. Therefore, 32 chemotypes (manually selected as illustrative examples data were extracted from 1649 papers, resulting in 7541 of common functional groups and other features of interest). models. The resultant spreadsheet comprises over 150,000 The prevalence of all chemotypes is given in Supplementary individual entries, as for each model, the species, sex, life- stage, route of administration, availability of equations, Table 3. The number of models associated with different routes compartments, references and chemical identiﬁers were of administration. captured. It is expected that errors will arise when as- Route of administration Number of models Percentage (%) sessing the suitability of papers for inclusion and per- forming extensive manual processing, hence a quality Metabolism from parent 2138 28.4 assessment exercise was undertaken. As part of this pro- Oral bolus 1903 25.2 cess, a representative sample from each of three stages of Inhalation 1193 15.8 the screening and data extraction process was assessed by a Intravenous bolus injection 1049 13.9 second investigator: Oral feed/water 381 5.1 Intravenous infusion 360 4.8 Dermal topical 181 2.4 — 5% of the papers that had been excluded at the full Dermal injection 82 1.1 text screening phase were reviewed; Intramuscular 59 0.8 — 5% of PBK model data extracted from the papers Intraperitoneal 58 0.8 (chemical information, species data (primary and Gills 36 0.5 secondary categories), sex, life-stage, route of In utero 35 0.5 administration, reference (DOI and PubMed ID if Unspeciﬁed 21 <0.5 available), compartments investigated, availability Intratracheal 17 <0.5 of equations and simulation software were Intraarterial 7 <0.5 checked; and Nasal 7 <0.5 — 5% of the chemical identiﬁer information from Buccal 6 <0.5 PubChem and COSMOS (chemical name, CAS Intramammary 4 <0.5 Registry Number, molecular weight, canonical Intraocular 2 <0.5 SMILES, isomeric SMILES, InChiKey and COSMOS Intrathecal 1 <0.5 ID) was obtained again from these sources and Intravaginal 1 <0.5 compared to the values in the spreadsheet. Table 4. Statistical analysis of the physico-chemical properties of the chemicals in the PBK model data set. Variable Mean Minimum Q1 Median Q3 Maximum Range IQR MW 325.61 6.94 163.10 292.28 410.67 6496.26 6489.26 247.57 No. HBD 1.79 0.00 0.00 1.00 2.00 100.00 100.00 2.00 No. HBA 4.17 0.00 1.00 3.00 6.00 120.00 120.00 5.00 SlogP 2.25 45.03 0.80 2.08 3.73 11.10 56.13 2.93 TPSA 73.07 0.00 23.47 56.93 93.00 2536.36 2536.36 69.53 nViolations 0.37 0.00 0.00 0.00 0.00 4.00 4.00 0.00 The variables are: MW = molecular weight; No. HBD = number of hydrogen bond donors; No. HBA = number of hydrogen bond acceptors; SlogP = predicted logarithm of the octanol:water partition coefﬁcient; TPSA = topological polar surface area; nViolations = number of violations of the Lipinski Rule of Five (Lipinski et al. ). The extreme values here are for vistarem , a magnetic resonance imaging contrast agent with large hydrophilic chains. 204 Alternatives to Laboratory Animals 49(5) Figure 3. A comparison of the ranges of physico-chemical properties across the seven data sets investigated. Thompson et al. 205 Table 5. The percentage of chemicals within the PBK model data set that contain the speciﬁed chemotypes identiﬁed by using ToxPrints. % % ToxPrint chemotype Prevalence ToxPrint chemotype Prevalence 6-Membered heterocycle 34.3 Organic sulphide/thiol 5.2 Carboxamide 27.7 Urea 4.8 Alcohol (aliphatic) 24.0 Ether (aromatic) 2.9 Amine (aliphatic) 22.6 Nitrile 2.7 Ether (aliphatic-aromatic) 21.9 Carboxylic acid (aromatic) 2.3 Organohalide (aromatic) 20.5 3-Membered heterocycle (e.g. epoxide) 2.2 Ether (aliphatic) 15.8 Transition metal 1.8 5-Membered heterocycle (single heteroatom, e.g. 13.6 4-Membered heterocycle (single heteroatom, 1.8 pyrrole) e.g. azetidine) Amine (aromatic) 12.6 Metalloid 1.7 Carboxylic acid (alkyl) 12.4 Imine 1.7 Organohalide (aliphatic) 11.2 Organophosphate (P = O) 1.6 Alcohol (aromatic) 10.6 Group I/II metal 1.4 5-Membered heterocycle (heteroatoms at 1,3- 10.6 Aldehyde 1.1 positions, e.g. imidazole) Ketone 10.5 Thiocarbonyl 0.8 Carboxylic acid ester (alkyl) 6.8 Alkyl chain (C length ≥8) 0.5 Sulphonyl 5.8 Organic azo 0.0 Note that only 32 selected chemotypes from the total of 458 identiﬁed within the data set are given in the table; the full list is given in Supplementary Material S(ii). The greatest source of ‘error’ was determined to be the Discussion exclusion of papers that were considered as potentially In this systematic review, information concerning over relevant by a second investigator, that is, 6% of excluded 7500 PBK models were extracted from 1649 papers. The papers. In terms of the systematic review, this is not con- models encompassed 18 species (including rat, human, sidered a highly signiﬁcant problem. PBK models are mouse, cow and guinea-pig) at various life-stages (e.g. continually being published, hence there can never be a juvenile, adult, pregnant and health-compromised) across ﬁnalised set of models. It is the intention to make this re- 21 administration routes (e.g. oral, inhalation and in utero). source available in its current form, as a tool to assist re- The information has been distilled into a Microsoft Excel searchers in ﬁnding relevant PBK models, and to update the spreadsheet that was constructed using controlled vo- resource in the future capturing models previously not cabulary to enable users to search by using different cri- identiﬁed or erroneously excluded. teria (e.g. to allow the selection of models by species or For PBK model data, manually extracted from the pa- pers, an error was detected in the information captured for routes of administration, etc.). It is anticipated that re- 3.5% of the substances. This does not equate to 3.5% of the searchers or regulatory scientists can use this information total information being incorrect, as this may indicate an to assist in the building or evaluation of new models, or as error in only one (or possibly more) of the 13 columns that a resource from which to extract relevant pharmacokinetic relate to the PBK model information. or toxicokinetic data. An error was detected in the data for 2.4% of the Although this is the largest collation of PBK models that chemicals in relation to the identiﬁer information. As above, the authors are aware of, it is not a complete list. As this does not equate to 2.4% of the total information being identiﬁed in the quality assessment exercise, some of the incorrect, but that for 2.4% of chemicals an error was de- historic models were omitted. In addition, as this is such a tected in one (or more) of the seven columns associated with dynamic area of research, the generation of a ﬁnite list of all chemical identiﬁer information. models would not be possible. The publication of new models has shown a rapidly increasing trajectory in recent The authors welcome any feedback from users regarding years. However, the current data set serves as a basis for errors, omissions of existing models or updates for new the continuing curation of existing models, which will models (note that models require a minimum level of in- provide an increasingly rich source of information for formation and novelty to be included); please email the modellers in the future. corresponding author. 206 Alternatives to Laboratory Animals 49(5) exceptions to this); here a similar pattern to the range of Trends in model availability: Coverage of values is generally observed for log P and molecular chemical types weight. For both properties, for the majority of chemicals Chemicals can be used for a variety of purposes, and often it the values fall within a narrow range; however, there are is not feasible to allocate a chemical to a unique group (e.g. also extreme values for a few chemicals. Pesticide and there is a signiﬁcant cross-over between chemicals used as botanical data sets have a greater percentage of chemicals food additives and as cosmetic ingredients, hence the same in the log P ranges 3–4 (43% and 47%, respectively), chemicals may appear in more than one of the different data whereas the PBK data set only has 28% of chemicals in sets). It is therefore difﬁcult to determine the chemical this range. ‘types’ for which there are the most PBK models; however, Pesticide and botanical data sets comprise fewer mole- some trends are discernible from the analysis undertaken. cules capable of carrying a charge (associated with in- Unsurprisingly, given that PBK modelling evolved in drug creased hydrophilicity) — hence, on average, they have development, the greatest proportion of models correspond higher log P values. This is signiﬁcant, as partitioning to chemicals in the DrugBank data set. Pesticides are behaviour (often estimated by using log P) is a key element generally well studied and data-rich; therefore, the second in building PBK models. Whilst the extreme values for log most common type of chemical with PBK models are the P, calculated by the software used here, may be unrealistic pesticides. For food additives and cosmetic ingredients, (and therefore unsuitable for model building), when used for where there are often chemicals in common, similar pro- comparison purposes they are still useful for demonstrating portions of chemicals have PBK models. Due to the size and the trends in the data. The range in values for all of the generality of the REACH data set, it would be anticipated physico-chemical properties of pesticides, is narrower than that relatively few chemicals would have existing PBK for the other chemical types, indicating the more restrictive models. The results conﬁrm the paucity of PBK models chemical properties required for these chemicals. available in relation to different areas of chemical space, and Botanicals generally show a wider range of values for underline the importance of using existing PBK models to each of the physico-chemical properties (in particular, more help ﬁll data gaps. chemicals show properties at the upper extremes of the ranges). A signiﬁcant number of compounds within this set are large and complex. Whilst in other data sets, molecules Trends in model availability: Coverage of tend to be designed for a speciﬁc purpose, and those out- physico-chemical properties with the given property ranges are ﬁltered out, the same Figure 3(a) shows the distribution of molecular weight exclusions would not be applicable to this data set. The across the seven data sets. As expected, the majority of the diversity of structural features present in the chemicals with chemicals fall within the range of 100–600 Da, but notable existing PBK models is demonstrated by the ToxPrints differences exist between the data sets. There are a relatively analysis, with 458 chemotypes identiﬁed as being present high number of chemicals in the PBK data set with low within this data set. The diversity of these chemicals, in molecular weight — these will include the volatile chem- terms of their molecular complexity, is demonstrated by the icals for which respiratory uptake has been extensively number of chemotypes identiﬁed — individual chemicals studied. Food additives and cosmetic ingredients (which were shown to contain between one and 69 chemotypes have chemicals in common) show a relatively high within their structure. proportion of low molecular weight chemicals. Chemicals that are designed to be biologically active, such as drugs Conclusion and pesticides, tend to be developed in accordance with guidelines relating to preferred physico-chemical prop- Understanding the kinetic behaviour of a chemical within erties. For example, the Lipinski Rule of Five stipulates the body, particularly its concentration–time proﬁle at a that drugs with poor oral absorption are associated with target site, is essential to accurately determine its potential chemicals with: molecular weight > 500 Da, log P > 5, effect. For the majority of chemicals, there is a lack of data and > 10 hydrogen bond acceptors or > 5 hydrogen bond concerning toxicity and kinetics. However, generating donors. Other research has also suggested that a to- such information de novo would require excessive use of pological polar surface area (TPSA) higher than 140 A is animals and is legally, ethically and ﬁnancially con- also unfavourable for oral absorption. Consequently, strained. Hence there is a need to leverage existing certain chemical types are designed to fall within nar- knowledge in order to obtain as much information as rower property ranges and such trends are evident in the possible to assist decision-making. Read-across is the most property ranges mentioned in this article. A correlation common method by which information from data-rich between molecular weight and log P is often observed chemicals is used to predict information for data-poor amongst groups of chemicals (although there are many chemicals. Herein is presented a comprehensive Thompson et al. 207 collation of existing PBK models that can be searched by Declaration of Conﬂicting Interests using multiple criteria. The physico-chemical space of The authors declared no potential conﬂicts of interest with respect PBK models has been mapped against that of other to the research, authorship, and/or publication of this article. chemical types — that is, food additives, cosmetic in- gredients, drugs, REACH chemicals, botanicals and pes- Funding ticides. Organising the current state of knowledge of The authors disclosed receipt of the following ﬁnancial support for existing PBK models provides a valuable resource for the research, authorship, and/or publication of this article: This those working in the area to identify models for chemicals work was supported by the European Partnership for Alternative of interest, or analogues, that can be used to assist the Approaches to animal testing (EPAA). development or evaluation of new PBK models. The concept of using existing PBK model information in a ORCID iDs read-across approach to develop a new PBK model for an analogous chemical has been demonstrated successfully in Courtney V. Thompson https://orcid.org/0000-0003-2243-2974 18,22 recent publications. Such an approach is recom- Peter E. Penson  https://orcid.org/0000-0001-6763-1489 mended in the recent OECD guidance on PBK modelling, which focuses on the use of alternative methods in PBK Supplementary Material model development. The PBK model data set described Supplementary material for this article is available online. herein enables researchers to readily gain insight into available PBK models across multiple species, life-stages References and routes of administration, such that the structure and parameterisation of PBK models for different chemicals is 1. International Programme on Chemical Safety and Inter- more accessible. This ensures maximum use of existing organisation Programme for the Sound Management of knowledge on PBK modelling, and reduces the time and Chemicals (2010). Characterization And Application of Physi- cost associated with the development of new PBK models. ologically Based Phamacokinetic Models in Risk Assessment. Continuing effort is required to curate existing PBK Geneva, Switzerland: World Health Organisation, 2010, 92 pp. models. The ability to extract relevant data from models and 2. Pistollato F, Madia F, Corvi R, et al. Current EU regulatory reproduce those published in the literature would be fa- requirements for the assessment of chemicals and cosmetic cilitated by researchers embracing the use of systematic products: Challenges and opportunities for introducing methods to record PBK models. Consistent formats for new approach methodologies. Arch Toxicol 2021; 95: 28,29 publishing PBK models have been proposed previously, 1867–1897. with the template proposed in the recent OECD guidance 3. ECHA. Read-Across Assessment Framework (RAAF). Lux- drawing on previous recommendations, most signiﬁcantly embourg, UK: EU Publications Ofﬁce, 2017, 60 pp. the reporting format proposed by Tan and co-workers. The 4. Scientiﬁc Committee on Consumer Safety. The SCCS notes of use of consistent reporting formats is strongly encouraged, as guidance for the testing of cosmetic ingredients and their this assists other researchers in re-using or re-purposing safety evaluation, 10th revision, 24–25 October 2018, https:// existing models. ec.europa.eu/health/sites/default/ﬁles/scientiﬁc_committees/ Future effort is required to curate available models, consumer_safety/docs/sccs_o_224.pdf, (2018, accessed 22 identify appropriate similarity metrics to assist in the November 2021). identiﬁcation of PBK models for analogues to chemicals of 5. US EPA. Approaches for the Application of Physiologically interest, and, ideally, make such resources widely available, Based Pharmacokinetic (PBPK) Models and Supporting for example, by incorporation into free webtools. Such Data in Risk Assessment (Final Report). Washington, DC: US endeavours will help to leverage the wealth of information Environmental Protection Agency, 2006. available in existing PBK models. 6. Kuepfer L, Niederalt C, Wendl T, et al. Applied concepts in PBPK modeling: How to build a PBPK/PD model. CPT Acknowledgments Pharmacometrics Syst Pharmacol 2016; 5: 516–531. JCM and CVT gratefully acknowledge the funding and scientiﬁcadvice 7. Upton RN, Foster DJ and Abuhelwa AY. An introduction to of the European Partnership for Alternative Approaches to animal physiologically-based pharmacokinetic models. Paediatr testing (EPAA) with respect to the funding of a studentship for CVT. Anaesth 2016; 26: 1036–1046. 8. Laroche C, Aggarwal M, Bender H, et al. Finding synergies for Author’s Note 3Rs — toxicokinetics and read-across: Report from an EPAA The views expressed in this article are those of the authors and do partners’ forum. Regul Toxicol Pharmacol 2018; 99: 5–21. not necessarily represent the views or the policies of the US 9. Madden JC, Pawar G, Cronin MTD, et al. In silico re- Environmental Protection Agency. sources to assist in the development and evaluation of 208 Alternatives to Laboratory Animals 49(5) physiologically-based kinetic models. Comput Toxicol 2019; 21. Ellison CA and Wu S. Application of structural and functional 11: 33–49. pharmacokinetic analogs for physiologically based pharma- 10. OECD. OECD Series on Testing and Assessment, No. 331, cokinetic model development and evaluation. Regul Toxicol Environment, Health and Safety, Environment Directorate. Pharmacol 2020; 114: 104667. Paris, France: Organisation for Economic Cooperation and 22. Sayre RR, Grulke C, Tan Y-M, et al. Existing physiologically- Development, 2021, 103 pp. based kinetic (PBK) models, identiﬁed via pubMed searches (with 11. Paini A, Leonard JA, Joossens E, et al. Next generation associated endnote library), https://ﬁgshare.com/articles/dataset/ physiologically based kinetic (NG-PBK) models in support of Existing_physiologically-based_kinetic_PBK_models_identiﬁed_ regulatory decision making. Comput Toxicol 2019; 9: 61–72. via_PubMed_searches_with_associated_EndNote_library_/ 12. Rowland M, Lesko LJ and Rostami-Hodjegan A. Physiolog- 10075574/1 (2019, accessed 21 September 2021). ically based pharmacokinetics is impacting drug development 23. Paini A, Worth A, Kulkarni S, et al. Assessment of the pre- and regulatory decision making. CPT Pharmacometrics Syst dictive capacity of a physiologically based kinetic model using Pharmacol 2015;4:313–315. a read-across approach. Comput Toxicol 2021; 18: 100159. 13. Perry C, Davis G, Conner TM, et al. Utilization of physio- 24. Madden JC, Enoch SJ, Paini AP, et al. A review of in silico logically based pharmacokinetic modeling in clinical phar- tools as alternatives to animal testing: Principles, resources macology and therapeutics: An overview. Curr Pharmacol and applications. Altern Lab Anim 2020; 48: 146–172. Rep 2020; 6: 71–84. 25. Thompson C, Madden J and Penson P. Systematic Review to 14. ECHA. The Use of Alternatives to Testing on Animals for the Determine the Chemical Space of Existing Physiologically- REACH Regulation. Fourth report under Article 117(3) of the Based Kinetic (PBK) Models, https://www.crd.york.ac.uk/ REACH Regulation. Helsinki, Finland: European Chemicals prospero/display_record.php?ID=CRD42020171130 (2020, Agency, 2020, 85 pp. accessed 21 September 2021). 15. Schultz TW, Amcoff P, Berggren E, et al. A strategy for 26. Paini A, Leonard JA, Kliment T, et al. Investigating the state structuring and reporting a read-across prediction of toxicity. of physiologically based kinetic modelling practices and Regul Toxicol Pharmacol 2015; 72: 586–601. challenges associated with gaining regulatory acceptance of 16. Sayre RR, Wambaugh JF and Grulke C. Database of phar- model applications. Regul Toxicol Pharm 2017; 90: 104–115. macokinetic time-series data and parameters for 144 envi- 27. Lipinski CA, Lombardo F, Dominy BW, et al. Experimental ronmental chemicals. Sci Data 2020; 7: 122. and computational approaches to estimate solubility and 17. Pawar G, Madden JC, Ebbrell D, et al. In silico toxicology permeability in drug discovery and development settings. Adv data resources to support read-across and (Q)SAR. Front Drug Deliver Rev 1997; 23: 3–25. Pharmacol 2019; 10: 561. 28. EMA. Guideline on the Reporting of Physiologically Based 18. Tan Y-M, Worley RR, Leonard JA, et al. Challenges asso- Pharmacokinetic (PBPK) Modelling and Simulation, https:// ciated with applying physiologically based pharmacokinetic www.ema.europa.eu/en/reporting-physiologically-based- modeling for public health decision-making. Toxicol Sci pharmacokinetic-pbpk-modelling-simulation (2018, accessed 2018; 162: 341–348. 21 September 2021). 19. Lu J, Goldsmith MR, Grulke C, et al. Developing a 29. US FDA. Physiologically Based Pharmacokinetic Analyses — physiologically-based pharmacokinetic model knowledgebase Format and Content, Guidance for Industry, https://www.fda. in support of provisional model construction. PLoS Comput gov/ﬁles/drugs/published/Physiologically-Based-Pharmacokinetic- Biol 2016; 12: e1004495. Analyses-%E2%80%94-Format-and-Content-Guidance- 20. Mellor CL, Marchese Robinson RL, Benigni R, et al. Mo- for-Industry.pdf (2018, accessed 21 September 2021). lecular ﬁngerprint-derived similarity measures for toxico- 30. Tan Y-M, Chan M, Chukwudebe A, et al. PBPK Model re- logical read-across: Recommendations for optimal use. Regul porting template for chemical risk assessment applications. Toxicol Pharmacol 2019; 101: 121–134. Regul Toxicol Pharmacol 2020; 115: 104691.

Journal

Alternative to Laboratory Animals – SAGE

Published: Sep 1, 2021

Keywords: PBK; PBPK; PBTK; systematic review; pharmacokinetic modelling; read-across

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A Systematic Review of Published Physiologically-based Kinetic Models and an Assessment of their Chemical Space Coverage

A Systematic Review of Published Physiologically-based Kinetic Models and an Assessment of their Chemical Space Coverage

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A Systematic Review of Published Physiologically-based Kinetic Models and an Assessment of their Chemical Space Coverage

A Systematic Review of Published Physiologically-based Kinetic Models and an Assessment of their Chemical Space Coverage

References (45)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies