Characterisation of data resources for in silico modelling: benchmark datasets for ADME properties

Introduction: The cost of in vivo and in vitro screening of ADME properties of compounds has motivated efforts to develop a range of in silico models. At the heart of the development of any computational model are the data; high quality data are essential for developing robust and accurate models. The characteristics of a dataset, such as its availability, size, format and type of chemical identifiers used, influence the modelability of the data.

Areas covered: This review explores the usefulness of publicly available ADME datasets for researchers to use in the development of predictive models. More than 140 ADME datasets were collated from publicly available resources and the modelability of 31 selected datasets were assessed using specific criteria derived in this study.

Expert opinion: Publicly available datasets differ significantly in information content and presentation. From a modelling perspective, datasets should be of adequate size, available in a user-friendly format with all chemical structures associated with one or more chemical identifiers suitable for automated processing (e.g. CAS number, SMILES string or InChIKey). Recommendations for assessing dataset suitability for modelling and publishing data in an appropriate format are discussed.     

In silico techniques for the study and prediction of xenobiotic metabolism: A review

Knowledge about metabolism is very important to understand the health risks posed by chemicals. The biochemical process of metabolism causes activation, inactivation, toxification, detoxification as well as changes in the physicochemical properties of a chemical. The long time consumption and high costs associated with animal tests and the challenges faced by traditional quantitative structure–activity relationship (QSAR) models in dealing with situations wherein parent chemical structures are less relevant to the ultimate effects have led to the development of in silico techniques for the prediction of xenobiotic metabolism. The strengths and limitations of some of the most commonly used in silico expert systems, and their application in studying metabolism of xenobiotic chemicals, have been reviewed. The in silico metabolism simulators possessed several distinguishing features imparted in part by the nature of knowledge rules (algorithms) encoded within them and in part by the integration of QSAR libraries and computational engines.     

In silico methods combined with expert knowledge rule out mutagenic potential of pharmaceutical impurities: An industry survey

With the increasing emphasis on identification and low level control of potentially genotoxic impurities (GTIs), there has been increased use of structure-based assessments including application of computerized models. To date many publications have focused on the ability of computational models, either individually or in combination, to accurately predict the mutagenic effects of a chemical in the Ames assay. Typically, these investigations take large numbers of compounds and use in silico tools to predict their activity with no human interpretation being made. However, this does not reflect how these assessments are conducted in practice across the pharmaceutical industry. Current guidelines indicate that a structural assessment is sufficient to conclude that an impurity is non-mutagenic. To assess how confident we can be in identifying non-mutagenic structures, eight companies were surveyed for their success rate. The Negative Predictive Value (NPV) of the in silico approaches was 94%. When human interpretation of in silico model predictions was conducted, the NPV increased substantially to 99%. The survey illustrates the importance of expert interpretation of in silico predictions. The survey also suggests the use of multiple computational models is not a significant factor in the success of these approaches with respect to NPV.     

The Acquisition and Application of Absorption,Distribution, Metabolism,and Excretion (ADME) Data in Agricultural Chemical Safety Assessments

A proposal has been developed by the Agricultural Chemical Safety Assessment (ACSA) Technical Committee of the ILSI Health and Environmental Sciences Institute (HESI) for an improved approach to assessing the safety of crop protection chemicals. The goal is to ensure that studies are scientifically appropriate and necessary without being redundant, and that tests emphasize toxicological endpoints and exposure durations that are relevant for risk assessment. Incorporation of pharmacokinetic studies describing absorption, distribution, metabolism, and excretion is an essential tool for improving the design and interpretation of toxicity studies and their application for safety assessment. A tiered approach is described in which basic pharmacokinetic studies, similar to those for pharmaceuticals, are conducted for regulatory submission. Subsequent tiers provide additional information in an iterative manner, depending on pharmacokinetic properties, toxicity study results, and the intended uses of the compound.     

Molecular approaches to the identification of biomarkers of exposure and effect--report of an expert meeting organized by COST Action B15. November 28, 2003

In the past, the term biomarker has been used with several meanings when used in human and environmental toxicology as compared to pharmaceutical development. However, with the advent of molecular approaches and their application in the field of drug development and toxicology, the concept of biomarkers has to be newly defined. In the meeting, the experts found consent in defining the term and described the application of biomarkers in toxicology, drug development and clinical diagnostics. Molecular approaches to biomarker identification and selection lead to a large amount of data. Hence, the statistical analysis is challenging and special statistical problems have to be solved in biomarker characterization, of particular interest are attempts aiming at class discovery and prediction. Reliability and biological relevance are to be demonstrated for biomarkers of exposure and effect which is also true for biomarkers of susceptibility. It is envisaged that the application of biomarkers will expand from current use in pre-clinical toxicology to the risk characterization and risk assessment of chemicals and from early clinical phases of drug development to later phases and even into daily clinical use in diagnostics and disease classification.     

In silico prediction of SARS protease inhibitors by virtual high throughput screening

A structure-based in silico virtual drug discovery procedure was assessed with severe acute respiratory syndrome coronavirus main protease serving as a case study. First, potential compounds were extracted from protein-ligand complexes selected from Protein Data Bank database based on structural similarity to the target protein. Later, the set of compounds was ranked by docking scores using a Electronic High-Throughput Screening flexible docking procedure to select the most promising molecules. The set of best performing compounds was then used for similarity search over the 1 million entries in the Ligand.Info Meta-Database. Selected molecules having close structural relationship to a 2-methyl-2,4-pentanediol may provide candidate lead compounds toward the development of novel allosteric severe acute respiratory syndrome protease inhibitors.     

In vivo pharmacokinetics of levodopa and 3-O-methyldopa in muscle

Summary In the present study in vivo microdialysis sampling coupled to high-performance liquid chromatography with electrochemical detection, was used to study the pharmacokinetics of levodopa and 3-O-methyldopa in skeletal muscle in dog, after intravenous administration of levodopa. For comparison, the pharmacokinetic parameters of both compounds were simultaneously determined in plasma using blood collection. Muscle microdialysis samples and blood were continuously collected for 4 h after the administration of levodopa (25 mg/kg). Pharmacokinetic profiles of levodopa in plasma and muscle were different. The mean Tmax value of levodopa in plasma and muscle was 0.16 h and 1.0 h, respectively.The AUC_0inf for levodopa in plasma was nearly 18-fold higher in plasma than in muscle. The 3-O-methyldopa concentration increased very rapidly after the administration of levodopa, to reach a plateau after 2.5 h and 3 h in plasma and muscle, respectively. The AUC₀₄ for 3-O-methyldopa was 3.6-fold higher in plasma than in muscle. The ratio levodopa/3-O-methyldopa, reflecting the metabolic rate of levodopa, was 3.5 times higher in plasma than in muscle, at the peak value of levodopa, and then rapidly declined to values lower than 1, one hour after administration of the drug. We compared our results with literature data from postmortem studies done in rat experiments.We concluded that levodopa is not accumulating in muscle as such, but is converted to 3-O-methyldopa probably before leaving the plasma compartment. Send offprint requests to D. Deleu at the above address     

Use of constraint-based modeling for the prediction and validation of antimicrobial targets

The overall process of antimicrobial drug discovery and development seems simple, to cure infectious disease by identifying suitable antibiotic drugs. However, this goal has been difficult to fulfill in recent years. Despite the promise of the high-throughput innovations sparked by the genomics revolution, discovery, and development of new antibiotics has lagged in recent years exacerbating the already serious problem of evolution of antibiotic resistance. Therefore, both new antimicrobials are desperately needed as are improvements to speed up or improve nearly all steps in the process of discovering novel antibiotics and bringing these to clinical use. Another product of the genomic revolution is the modeling of metabolism using computational methodologies. Genomic-scale networks of metabolic reactions based on stoichiometry, thermodynamics and other physico-chemical constraints that emulate microbial metabolism have been developed into valuable research tools in metabolic engineering and other fields. This constraint-based modeling is predictive in identifying critical reactions, metabolites, and genes in metabolism. This is extremely useful in determining and rationalizing cellular metabolic requirements. In turn, these methods can be used to predict potential metabolic targets for antimicrobial research especially if used to increase the confidence in prioritization of metabolic targets. The many different capacities of constraint-based modeling also enable prediction of cellular response to specific inhibitors such as antibiotics and this may, ultimately find a role in drug discovery and development. Herein, we describe the principles of metabolic modeling and how they might initially be applied to antimicrobial research.     

Predicting drug metabolism and pharmacokinetics features of in-house compounds by a hybrid machine-learning model

We constructed machine learning-based pharmacokinetic prediction models with very high performance. The models were trained on 26138 and 16613 compounds involved in metabolic stability and cytochrome P450 inhibition, respectively. Because the compound features largely differed between the publicly available and in-house compounds, the models learned on the public data could not predict the in-house compounds, suggesting that outside of a certain applicability domain (AD), the prediction results are unreliable and can mislead the design of novel compounds. To exclude the uncertain prediction results, we constructed another machine learning model that determines whether the newly designed compound is inside or outside the AD. The AD was evaluated multi-dimensionally with some explanatory variables: The structural similarities and the probability obtained from the pharmacokinetic prediction model. The accuracy of predicting metabolic stability was 79.9% on the test set, increasing significantly to 93.6% after excluding the low-reliability compounds. The model properly classified the reliability of the compounds. After learning on the in-house compounds, the reliability model classified almost all (90%) of the public compounds as low reliability, indicating that the AD was properly determined by the model.     

The single-point method of dosage prediction: Pharmacokinetic basis and method optimization

Based on the principle of superposition an expression has been established relating a drug concentration at steady-state to a concentration after a single dose. This relationship applies for drugs with linear pharmacokinetics given at equal dosage intervals and it is independent of the route of administration. The relationship provides theoretical justification for the single-point method of dosage prediction reported in the literature. The test conditions in the method can therefore be optimized and the limits of the method defined. The expression can also be used for individualized prediction of maintenance dose after estimation of drug half-life in the patient with no limit to half-life values to which it can be applied.     

In silico studies and evaluation of antiparasitic role of a novel pyruvate phosphate dikinase inhibitor in Leishmania donovani infected macrophages

The present work deals with the identification and characterization of a novel inhibitor Z220582104, specific to pyruvate phosphate dikinase, for leishmanicidal activities against free promastigotes and intracellular amastigotes. We have used structure-based drug designing approaches and performed homology modelling, virtual screening and molecular dynamics studies. Primary mouse macrophages and macrophage cell line J774A1 were infected with promastigotes of Leishmania donovani. Both promastigotes and infected macrophages were subjected to treatment with the varying concentrations of Z220582104 or miltefosine for assessment of leishmanicidal activity. The novel inhibitor Z220582104 demonstrated growth inhibitory potential and reduced the viability of the free promastigotes in a concentration- and time-dependent manner. Z220582104 was also effective against the intracellular form of the parasites and reduced the number of amastigotes in macrophages and also lowered the parasite index, compared with the untreated infected macrophages. Although less effective compared with the miltefosine, Z220582104 is well tolerated by the dividing cells and normal human lymphocytes and monocytes with no adverse effects on the growth kinetics or viability. Our in silico and in vitro studies suggested that Leishmania donovani pyruvate phosphate dikinase could be a potential new drug target.     

Estimation of nizatidine gastric nitrosatability and product toxicity via an integrated approach combining HILIC,in silico toxicology,and molecular docking

The liability of the H₂-receptor antagonist nizatidine (NZ) to nitrosation in simulated gastric juice (SGJ) and under WHO-suggested conditions was investigated for the first time. For monitoring the nitrosatability of NZ, a hydrophilic interaction liquid chromatography (HILIC) method was optimized and validated according to FDA guidance. A Cosmosil HILIC® column and a mobile phase composed of acetonitrile: 0.04 M acetate buffer pH 6.0 (92:8, v/v) were used for the separation of NZ and its N-nitroso derivative (NZ-NO) within 6 min with LODs of 0.02 and 0.1 μg/mL, respectively. NZ was found highly susceptible to nitrosation in SGJ reaching 100% nitrosation in 10 min, while only 18% nitrosation was observed after 160 min under the WHO-suggested conditions. The chemical structure of NZ-NO was clarified by ESI⁺/MS. In silico toxicology study confirmed the mutagenicity and toxicity of NZ-NO. Experiments evidenced that ascorbic acid strongly suppresses the nitrosation of NZ suggesting their co-administration for protection from potential risks. In addition, the impacts of the HILIC method on safety, health, and environment were favorably evaluated by three green analytical chemistry metrics and it was proved that, unlike the popular impression, HILIC methods could be green to the environment.     

Early assessment of proarrhythmic risk of drugs using the in vitro data and single-cell-based in silico models: proof of concept

Background and purpose: To determine the predictive performance of in silico models using drug-specific preclinical cardiac electrophysiology data to investigate drug-induced arrhythmia risk (e.g. Torsade de pointes (TdP)) in virtual human subjects.

Experimental approach: To assess drug proarrhythmic risk, we used a set of in vitro electrophysiological measurements describing ion channel inhibition triggered by the investigated drugs. The Cardiac Safety Simulator version 2.0 (CSS; Simcyp, Sheffield, UK) platform was used to simulate human left ventricular cardiac myocyte action potential models.

Results: This study shows the impact of drug concentration changes on particular ionic currents by using available experimental data. The simulation results display safety threshold according to drug concentration threshold and log (threshold concentration/ effective therapeutic plasma concentration (ETPC)).

Conclusion and implications: We reproduced the underlying biophysical characteristics of cardiac cells resulted in effects of drugs associated with cardiac arrhythmias (action potential duration (APD) and QT prolongation and TdP) which were observed in published 3D simulations, yet with much less computational burden.     

Cytotoxic activity of quinolinequinones in cancer: In vitro studies,molecular docking,and ADME/PK profiling

Lead molecules containing 1,4-quinone moiety are intriguing novel compounds that can be utilized to treat cancer owing to their antiproliferative activities. Nine previously reported quinolinequinones ( AQQ1-9 ) were studied to better understand their inhibitory profile to produce potent and possibly safe lead molecules. The National Cancer Institute (NCI) of Bethesda chose all quinolinequinones ( AQQ1-9 ) based on the NCI Developmental Therapeutics Program and tested them against a panel of 60 cancer cell lines. At a single dose and five further doses, AQQ7 significantly inhibited the proliferation of all leukemia cell lines and some breast cancer cell lines. We investigated the in vitro cytotoxic activities of the most promising compounds, AQQ2 and AQQ7 , in MCF7 and T-47D breast cancer cells, DU-145 prostate cancer cells, HCT-116 and COLO 205 colon cancer cell lines, and HaCaT human keratinocytes using the MTT assay. AQQ7 showed particularly high cytotoxicity against MCF7 cells. Further analysis showed that AQQ7 exhibits anticancer activity through the induction of apoptosis without causing cell cycle arrest or oxidative stress. Molecular docking simulations for AQQ2 and AQQ7 were conducted against the COX, PTEN, and EGFR proteins, which are commonly overexpressed in breast, cervical, and prostate cancers. The in vitro ADME and in vivo PK profiling of these compounds have also been reported.     

In vitro - in vivo - in silico approach in the development of inhaled drug products: Nanocrystal-based formulations with budesonide as a model drug

This study aims to understand the absorption patterns of three different kinds of inhaled formulations via in silico modeling using budesonide (BUD) as a model drug. The formulations investigated in this study are: (i) commercially available micronized BUD mixed with lactose (BUD-PT), (ii) BUD nanocrystal suspension (BUD-NC), (iii) BUD nanocrystals embedded hyaluronic acid microparticles (BUD-NEM). The deposition patterns of the three inhaled formulations in the rats’ lungs were determined in vivo and in silico predicted, which were used as inputs in GastroPlus™ software to predict drug absorption following aerosolization of the tested formulations. BUD pharmacokinetics, estimated based on intravenous data in rats, was used to establish a drug-specific in silico absorption model. The BUD-specific in silico model revealed that drug pulmonary solubility and absorption rate constant were the key factors affecting pulmonary absorption of BUD-NC and BUD-NEM, respectively. In the case of BUD-PT, the in silico model revealed significant gastrointestinal absorption of BUD, which could be overlooked by traditional in vivo experimental observation. This study demonstrated that in vitro-in vivo-in silico approach was able to identify the key factors that influence the absorption of different inhaled formulations, which may facilitate the development of orally inhaled formulations with different drug release/absorption rates.     

Pharmacokinetics of morphine and its surrogates. XI: Effect of simultaneously administered naltrexone and morphine on the pharmacokinetics and pharmacodynamics of each in the dog

There were no dramatic modifications of the pharmacokinetics in the dog of i.v. bolus doses of 0.5, 2.7 and 5 mg kg-1 morphine by coadministering i.v. 5 mg kg-1 naltrexone as bolus injections over 15-20s and 12.3 mg kg-1 by continuous infusion. Morphine's terminal half-life, clearances, apparent volumes of distribution (except for that of the central compartment), percentages of drug and conjugated metabolite excreted in urine and bile did not differ significantly by paired t-test (probability (p) greater than 0.05 for rejection of the null hypothesis of no difference) when naltrexone was coadministered. There were no statistically significant (by t-test) modifications of the plasma pharmacokinetics in the dog of i.v. bolus doses of 5 mg kg-1 naltrexone with and without morphine coadministration except for the coefficient of the second (or terminal) exponential of the sum that fitted the plasma concentration-time data of naltrexone. Although morphine coadministration did not significantly affect the terminal half-life of naltrexone, its clearances or apparent volumes of distribution by t-test of the differences between averages (with each dog equally weighted), drug coadministration did significantly (by t-test) affect the fraction of naltrexone dose secreted into bile as conjugate (fB), the fraction of the dose excreted as conjugate in urine, and the fraction excreted elsewhere (f'B). Although naltrexone reversed the central action of morphine in affecting monitored pupil diameters, it did not antagonize the peripheral effects of morphine in perturbing renal and biliary flow rates. This led to a larger fraction of the naltrexone dose being metabolized to conjugate on morphine coadministration. Since less naltrexone conjugate was renally and biliary excreted initially, due to morphine inhibition of the initial renal and biliary processes, naltrexone conjugate plasma concentrations were higher when morphine was coadministered.     

In vitro prediction of gastrointestinal absorption and bioavailability: an experts' meeting report

The most convenient route of drug administration is peroral. To reach their target, drug molecules must be absorbed from the gastrointestinal tract and enter the systemic circulation in sufficient quantities. For this reason, understanding and anticipating the mechanisms and factors affecting gastrointestinal absorption and metabolism are of the utmost importance in developing new drugs. In contrast to drugs, which are administered intentionally for therapeutic reasons, chemical residues in food and other matrices enter the body unintentionally. Hence, in this case, a low systemic availability would be advantageous. For many reasons, but particularly because of financial and ethical (reduced used of animals) considerations, in vitro and ex vivo approaches to this problem have been pursued over the last few years. The use of in vitro methods, however, inherently creates questions about the validity of extrapolation to the in vivo situation. The purpose of this report is to review the current status of the field and to identify major gaps in our knowledge. Currently, there are a number of in silico, in vitro, cultured cell-based and ex vivo approaches available to predict the cell permeation, absorption and gastrointestinal metabolism of molecules. Some strengths and weaknesses of these approaches are presented, together with a discussion of genetic, environmental, physiological and pathological factors responsible for interspecies and inter-individual variability in these processes. Recent advances in our understanding of active processes such as gut epithelial transporters, involved in absorption, and drug-metabolising enzymes, responsible for intestinal presystemic metabolism, are highlighted. Some major research priorities are identified, including the need for high-quality, information-rich databases against which testing methods being developed can be prevalidated and validated. Preclinical drug development is changing rapidly, and the role of in vitro and ex vivo approaches in this process is becoming increasingly more important. Methods available now are very useful in the drug discovery and development process, including lead compound selection and optimisation and in the design of very early clinical studies, but whether any of them will eventually obviate the need for clinical trials of bioavailability is still very debatable and will require their full validation. It is clear, however, that the results from such in vitro tests are important in shaping drug discovery and the early preclinical drug development process. For other environmental, industrial and household chemicals to which humans are exposed, in particular new chemicals, results from in vitro studies might be the only source of information concerning systemic availability.     

In silico screening of chemicals for bacterial mutagenicity using electrotopological E-state indices and MDL QSAR software

Quantitative structure–activity relationship (QSAR) software offers a rapid, cost effective means of prioritizing the mutagenic potential of chemicals. MDL QSAR models were developed using atom-type E-state indices and non-parametric discriminant analysis. Models were developed for Salmonella typhimurium gene mutation, combining results from strains TA97, TA98, TA100, TA1535, TA1536, TA1537, and TA1538 (n = 3228), and Escherichia coli gene mutation tests WP2, WP100, and polA (n = 472). Composite microbial mutation models (n = 3338) were developed combining all Salmonella, E. coli, and the Bacillus subtilis rec spot test study results. The datasets contained 74% non-pharmaceuticals and 26% pharmaceuticals. Salmonella and microbial mutagenesis external validation studies included a total of 1444 and 1485 compounds, respectively. The average specificity, sensitivity, positive predictivity, concordance, and coverage of Salmonella models was 76, 81, 73, 78, and 98%, respectively, with similar performance for the microbial mutagenesis models. MDL QSAR and discriminant analysis provides rapid and highly automated mutagenicity screening software with good specificity, sensitivity, and coverage that is simpler and requires less user intervention than other similar software. MDL QSAR modules for microbial mutagenicity can provide efficient and cost effective large scale screening of compounds for mutagenic potential for the chemical and pharmaceutical industry.