Application of In Silico, In Vitro and Preclinical Pharmacokinetic Data for the Effective and Efficient Prediction of Human Pharmacokinetics

The first-in-class pyrazole-based dual InhA-VEGFR inhibitors towards integrated antitubercular host-directed therapy

Several facets of the host response to tuberculosis have been tapped for clinical investigation, especially targeting angiogenesis mediated by VEGF signaling from infected macrophages. Herein, we rationalized combining the antiangiogenic effects of VEGFR-2 blockade with direct antitubercular InhA inhibition in single hybrid dual inhibitors as advantageous alternatives to the multidrug regimens. Inspired by expanded triclosans, the ether ligation of triclosan was replaced by rationalized linkers to assemble the VEGFR-2 inhibitors thematic scaffold. Accordingly, new series of 3-(p-chlorophenyl)-1-phenylpyrazole derivatives tethered to substituted ureas and their isosteres were synthesized, evaluated against Mycobacterium tuberculosis virulent cell line H37Rv, and assessed for their InhA inhibitory activities. The urea derivatives 8d and 8g exhibited the most promising antitubercular activity (MIC = 6.25 µg/mL) surpassing triclosan (MIC = 20 µg/mL) with potential InhA inhibition, thus identified as the study hits. Interestingly, both compounds inhibited VEGFR-2 at nanomolar IC50 (15.27 and 24.12 nM, respectively). Docking and molecular dynamics simulations presumed that 8d and 8g could bind to their molecular targets InhA and VEGFR-2 posing essential stable interactions shared by the reference inhibitors triclosan and sorafenib. Finally, practical LogP, Lipinski's parameters and in silico ADMET calculations highlighted their drug-likeness as novel leads in the arsenal against TB.

Virtual Screening of Different Subclasses of Lignans with Anticancer Potential and Based on Genetic Profile

Cancer is a multifactorial disease that continues to increase. Lignans are known to be important anticancer agents. However, due to the structural diversity of lignans, it is difficult to associate anticancer activity with a particular subclass. Therefore, the present study sought to evaluate the association of lignan subclasses with antitumor activity, considering the genetic profile of the variants of the selected targets. To do so, predictive models were built against the targets tyrosine-protein kinase ABL (ABL), epidermal growth factor receptor erbB1 (EGFR), histone deacetylase (HDAC), serine/threonine-protein kinase mTOR (mTOR) and poly [ADP-ribose] polymerase-1 (PARP1). Then, single nucleotide polymorphisms were mapped, target mutations were designed, and molecular docking was performed with the lignans with the best predicted biological activity. The results showed more anticancer activity in the dibenzocyclooctadiene, furofuran and aryltetralin subclasses. The lignans with the best predictive values of biological activity showed varying binding energy results in the presence of certain genetic variants.

Unravelling the Hepatic Elimination Mechanisms of Colistin

PURPOSE

Colistin is an antibiotic which is increasingly used as a last-resort therapy in critically-ill patients with multidrug resistant Gram-negative infections. The purpose of this study was to evaluate the mechanisms underlying colistin's pharmacokinetic (PK) behavior and to characterize its hepatic metabolism.

METHODS

In vitro incubations were performed using colistin sulfate with rat liver microsomes (RLM) and with rat and human hepatocytes (RH and HH) in suspension. The uptake of colistin in RH/HH and thefraction of unbound colistin in HH (fu,hep) was determined. In vitro to in vivo extrapolation (IVIVE) was employed to predict the hepatic clearance (CLh) of colistin.

RESULTS

Slow metabolism was detected in RH/HH, with intrinsic clearance (CLint) values of 9.34± 0.50 and 3.25 ± 0.27 mL/min/kg, respectively. Assuming the well-stirred model for hepatic drug elimination, the predicted rat CLh was 3.64± 0.22 mL/min/kg which could explain almost 70% of the reported non-renal in vivo clearance. The predicted human CLh was 91.5 ± 8.83 mL/min, which was within two-fold of the reported plasma clearance in healthy volunteers. When colistin was incubated together with the multidrug resistance-associated protein (MRP/Mrp) inhibitor benzbromarone, the intracellular accumulation of colistin in RH/HH increased significantly.

CONCLUSION

These findings indicate the major role of hepatic metabolism in the non-renal clearance of colistin, while MRP/Mrp-mediated efflux is involved in the hepatic disposition of colistin. Our data provide detailed quantitative insights into the hereto unknown mechanisms responsible for non-renal elimination of colistin.

Human pharmacokinetics prediction with an in vitro-in vivo correction factor approach and in vitro drug-drug interaction profile of bictegravir, a potent integrase-strand transfer inhibitor component in approved biktarvy® for the treatment of HIV-1 infection

Bictegravir (BIC) is a potent small-molecule integrase strand-transfer inhibitor (INSTI) and a component of Biktarvy^®, a single-tablet combination regimen that is currently approved for the treatment of human immunodeficiency virus type 1 (HIV-1) infection. The in vitro properties, pharmacokinetics (PK), and drug-drug interaction (DDI) profile of BIC were characterised in vitro and in vivo.BIC is a weakly acidic, ionisable, lipophilic, highly plasma protein-bound BCS class 2 molecule, which makes it difficult to predict human PK using standard methods. Its systemic plasma clearance is low, and the volume of distribution is approximately the volume of extracellular water in nonclinical species. BIC metabolism is predominantly mediated by cytochrome P450 enzyme (CYP) 3A and UDP-glucuronosyltransferase 1A1. BIC shows a low potential to perpetrate clinically meaningful DDIs via known drug metabolising enzymes or transporters.The human PK of BIC was predicted using a combination of bioavailability and volume of distribution scaled from nonclinical species and a modified in vitro-in vivo correlation (IVIVC) correction for clearance. Phase 1 studies in healthy subjects largely bore out the prediction and supported the methods used. The approach presented herein could be useful for other drug molecules where standard projections are not sufficiently accurate. .

Characterization of intravenous pharmacokinetics in Göttingen minipig and clearance prediction using established in vitro to in vivo extrapolation methodologies

1. The use of the Göttingen minipig as an animal model for drug safety testing and prediction of human pharmacokinetics (PK) continues to gain momentum in pharmaceutical research and development. The aim of this study was to evaluate in vitro to in vivo extrapolation (IVIVE) methodologies for prediction of hepatic, metabolic clearance (CL_hep,met) in Göttingen minipig, using a comprehensive set of compounds.2. In vivo clearance was determined in Göttingen minipig by intravenous cassette dosing and hepatocyte intrinsic clearance, plasma protein binding and non-specific incubation binding were determined in vitro. Prediction of CL_hep,met was performed by IVIVE using conventional and adapted formats of the well-stirred liver model.3. The best prediction of in vivo CL_hep,met from scaled in vitro kinetic data was achieved using an empirical correction factor based on a 'regression offset' of the IVIV relationship.4. In summary, these results expand the in vitro and in vivo PK knowledge in Göttingen minipig. We show regression corrected IVIVE provides superior prediction of in vivo CL_hep,met in minipig offering a practical, unified scaling approach to address systematic under-predictions. Finally, we propose a reference set for researchers to establish their own 'lab-specific' regression correction for IVIVE in minipig.

Non-specific binding of compounds in in vitro metabolism assays: a comparison of microsomal and hepatocyte binding in different species and an assessment of the accuracy of prediction models

Non-specific binding in in vitro metabolism systems leads to an underestimation of the true intrinsic metabolic clearance of compounds being studied. Therefore in vitro binding needs to be accounted for when extrapolating in vitro data to predict the in vivo metabolic clearance of a compound. While techniques exist for experimentally determining the fraction of a compound unbound in in vitro metabolism systems, early in drug discovery programmes computational approaches are often used to estimate the binding in the in vitro system.Experimental fraction unbound data (n = 60) were generated in liver microsomes (fu_mic) from five commonly used pre-clinical species (rat, mouse, dog, minipig, monkey) and humans. Unbound fraction in incubations with mouse, rat or human hepatocytes was determined for the same 60 compounds. These data were analysed to determine the relationship between experimentally determined binding in the different matrices and across different species. In hepatocytes there was a good correlation between fraction unbound in human and rat (r²=0.86) or mouse (r²=0.82) hepatocytes. Similar correlations were observed between binding in human liver microsomes and microsomes from rat, mouse, dog, Göttingen minipig or monkey liver microsomes (r² of >0.89, n = 51 - 52 measurements in different species). Physicochemical parameters (logP, pKa and logD) were predicted for all evaluated compounds. In addition, logP and/or logD were measured for a subset of compounds.Binding to human hepatocytes predicted using 5 different methods was compared to the measured data for a set of 59 compounds. The best methods evaluated used measured microsomal binding in human liver microsomes to predict hepatocyte binding. The collated physicochemical data were used to predict the human fu_mic using four different in silico models for a set of 53-60 compounds. The correlation (r²) and root mean square error between predicted and observed microsomal binding was 0.69 & 0.20, 0.47 & 0.23, 0.56 & 0.21 and 0.54 & 0.26 for the Turner-Simcyp, Austin, Hallifax-Houston and Poulin models, respectively. These analyses were extended to include measured literature values for binding in human liver microsomes for a larger set of compounds (n=697). For the larger dataset of compounds, microsomal binding was well predicted for neutral compounds (r²=0.67 - 0.70) using the Poulin, Austin, or Turner-Simcyp methods but not for acidic or basic compounds (r²<0.5) using any of the models. While the lipophilicity-based models can be used, the in vitro binding should be measured for compounds where more certainty is needed, using appropriately calibrated assays and possibly established weak, moderate, and strong binders as reference compounds to allow comparison across databases.

Novel QSAR Approach for a Regression Model of Clearance That Combines DeepSnap-Deep Learning and Conventional Machine Learning

The toxicity, absorption, distribution, metabolism, and excretion properties of some targets are difficult to predict by quantitative structure-activity relationship analysis. Therefore, there is a need for a new prediction method that performs well for these targets. The aim of this study was to develop a new regression model of rat clearance (CL). We constructed a regression model using 1545 in-house compounds for which we had rat CL data. Molecular descriptors were calculated using molecular operating environment, alvaDesc, and ADMET Predictor software. The classification model of DeepSnap and Deep Learning (DeepSnap-DL) with images of the three-dimensional chemical structures of compounds as features was constructed, and the prediction probabilities for each compound were calculated. For molecular descriptor-based methods that use molecular descriptors and conventional machine learning algorithms selected by DataRobot, the correlation coefficient (R ²) and root mean square error (RMSE) were 0.625-0.669 and 0.295-0.318, respectively. We combined molecular descriptors and prediction probability of DeepSnap-DL as features and developed a novel regression method we called the combination model. In the combination model with these two types of features and conventional algorithms selected by DataRobot, R ² and RMSE were 0.710-0.769 and 0.247-0.278, respectively. This finding shows that the combination model performed better than molecular descriptor-based methods. Our combination model will contribute to the design of more rational compounds for drug discovery. This method may be applicable not only to rat CL but also to other pharmacokinetic and pharmacological activity and toxicity parameters; therefore, applying it to other parameters may help to accelerate drug discovery.

Possible Extraction of Drugs from Lung Tissue During Broncho-alveolar Lavage Suggest Uncertainty in the Procedure's Utility for Quantitative Assessment of Airway Drug Exposure

Following inhaled dosing, broncho-alveolar lavage (BAL) is often used for sampling epithelial lining fluid (ELF) to determine drug concentration in the lungs. This study aimed to explore the technique's suitability. Urea is typically used to estimate the dilution factor between the BAL fluid and physiological ELF, since it readily permeates through all fluids in the body. As representatives of permeable small molecule drugs with high, medium and low tissue distribution properties, propranolol, diazepam, indomethacin and AZD4721 were infused intravenously to steady state to ensure equal unbound drug concentrations throughout the body. The results showed that propranolol had higher unbound concentrations in the ELF compared to the plasma whilst this was not the case for the other compounds. Experiments with different BAL volumes and repeated lavaging indicated that the amount of drug extracted is very sensitive to experimental procedure. In addition, the results show that the unbound concentrations in ELF compared to plasma differs dependent on molecule class and tissue distribution properties. Overall data suggests that lavaging can remove drug from lung tissue in addition to ELF and highlights significant uncertainty in the robustness of the procedure for determining ELF drug concentrations.

Active Learning for Drug Design: A Case Study on the Plasma Exposure of Orally Administered Drugs

The success of artificial intelligence (AI) models has been limited by the requirement of large amounts of high-quality training data, which is just the opposite of the situation in most drug discovery pipelines. Active learning (AL) is a subfield of AI that focuses on algorithms that select the data they need to improve their models. Here, we propose a two-phase AL pipeline and apply it to the prediction of drug oral plasma exposure. In phase I, the AL-based model demonstrated a remarkable capability to sample informative data from a noisy data set, which used only 30% of the training data to yield a prediction capability with an accuracy of 0.856 on an independent test set. In phase II, the AL-based model explored a large diverse chemical space (855K samples) for experimental testing and feedback. Improved accuracy and new highly confident predictions (50K samples) were observed, which suggest that the model's applicability domain has been significantly expanded.

An integrated exposure and pharmacokinetic modeling framework for assessing population-scale risks of phthalates and their substitutes

To effectively incorporate in vitro-in silico-based methods into the regulation of consumer product safety, a quantitative connection between product phthalate concentrations and in vitro bioactivity data must be established for the general population. We developed, evaluated, and demonstrated a modeling framework that integrates exposure and pharmacokinetic models to convert product phthalate concentrations into population-scale risks for phthalates and their substitutes. A probabilistic exposure model was developed to generate the distribution of multi-route exposures based on product phthalate concentrations, chemical properties, and human activities. Pharmacokinetic models were developed to simulate population toxicokinetics using Bayesian analysis via the Markov chain Monte Carlo method. Both exposure and pharmacokinetic models demonstrated good predictive capability when compared with worldwide studies. The distributions of exposures and pharmacokinetics were integrated to predict the population distributions of internal dosimetry. The predicted distributions showed reasonable agreement with the U.S. biomonitoring surveys of urinary metabolites. The "source-to-outcome" local sensitivity analysis revealed that food contact materials had the greatest impact on body burden for di(2-ethylhexyl) adipate (DEHA), di-2-ethylhexyl phthalate (DEHP), di(isononyl) cyclohexane-1,2-dicarboxylate (DINCH), and di(2-propylheptyl) phthalate (DPHP), whereas the body burden of diethyl phthalate (DEP) was most sensitive to the concentration in personal care products. The upper bounds of predicted plasma concentrations showed no overlap with ToxCast in vitro bioactivity values. Compared with the in vitro-to-in vivo extrapolation (IVIVE) approach, the integrated modeling framework has significant advantages in mapping product phthalate concentrations to multi-route risks, and thus is of great significance for regulatory use with a relatively low input requirement. Further integration with new approach methodologies will facilitate these in vitro-in silico-based risk assessments for a broad range of products containing an equally broad range of chemicals.

Anti-COVID-19 activity of some benzofused 1,2,3-triazolesulfonamide hybrids using in silico and in vitro analyses

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a pandemic fatal infection with no known treatment. The severity of the disease and the fast viral mutations forced the scientific community to search for potential solution. Here in the present manuscript, some benzofused1,2,3triazolesulfonamide hybrids were synthesized and evaluated for their anti- SARS-CoV-2 activity using in silico prediction then the most potent compounds were assessed using in-Vitro analysis. The in-Silico study was assessed against RNA dependent RNA polymerase, Spike protein S1, Main protease (3CLpro) and 2'-O-methyltransferase (nsp16). It was found that 4b and 4c showed high binding scores against RNA dependent RNA polymerase reached -8.40 and -8.75 ​kcal/mol, respectively compared to the approved antiviral (remdesivir -6.77 ​kcal/mol). Upon testing the binding score with SARS-CoV-2 Spike protein it was revealed that 4c exhibited the highest score (-7.22 ​kcal/mol) compared to the reference antibacterial drug Ceftazidime (-6.36 ​kcal/mol). Surprisingly, the two compounds 4b and 4c showed the highest binding scores against SARS-CoV-2 3CLpro (-8.75, -8.48 ​kcal/mol, respectively) and nsp16 (- 8.84 and - 8.89 ​kcal/mol, respectively) displaying many types of interaction with all the enzymes binding sites. The derivatives 4b and 4c were examined in vitro for their potential anti-SARS-CoV-2 and it was revealed that 4c was the most promising compound with IC50 reached 758.8108 ​mM and complete (100%) inhibition of the binding of SARS-CoV-2 virus to human ACE2 can be accomplished by using 0.01 ​mg.

AZD0284, a Potent, Selective, and Orally Bioavailable Inverse Agonist of Retinoic Acid Receptor-Related Orphan Receptor C2

Inverse agonists of the nuclear receptor RORC2 have been widely pursued as a potential treatment for a variety of autoimmune diseases. We have discovered a novel series of isoindoline-based inverse agonists of the nuclear receptor RORC2, derived from our recently disclosed RORC2 inverse agonist 2. Extensive structure-activity relationship (SAR) studies resulted in AZD0284 (20), which combined potent inhibition of IL-17A secretion from primary human TH17 cells with excellent metabolic stability and good PK in preclinical species. In two preclinical in vivo studies, compound 20 reduced thymocyte numbers in mice and showed dose-dependent reduction of IL-17A containing γδ-T cells and of IL-17A and IL-22 RNA in the imiquimod induced inflammation model. Based on these data and a favorable safety profile, 20 was progressed to phase 1 clinical studies.

Prediction Model of Clearance by a Novel Quantitative Structure-Activity Relationship Approach, Combination DeepSnap-Deep Learning and Conventional Machine Learning

Some targets predicted by machine learning (ML) in drug discovery remain a challenge because of poor prediction. In this study, a new prediction model was developed and rat clearance (CL) was selected as a target because it is difficult to predict. A classification model was constructed using 1545 in-house compounds with rat CL data. The molecular descriptors calculated by Molecular Operating Environment (MOE), alvaDesc, and ADMET Predictor software were used to construct the prediction model. In conventional ML using 100 descriptors and random forest selected by DataRobot, the area under the curve (AUC) and accuracy (ACC) were 0.883 and 0.825, respectively. Conversely, the prediction model using DeepSnap and Deep Learning (DeepSnap-DL) with compound features as images had AUC and ACC of 0.905 and 0.832, respectively. We combined the two models (conventional ML and DeepSnap-DL) to develop a novel prediction model. Using the ensemble model with the mean of the predicted probabilities from each model improved the evaluation metrics (AUC = 0.943 and ACC = 0.874). In addition, a consensus model using the results of the agreement between classifications had an increased ACC (0.959). These combination models with a high level of predictive performance can be applied to rat CL as well as other pharmacokinetic parameters, pharmacological activity, and toxicity prediction. Therefore, these models will aid in the design of more rational compounds for the development of drugs.

An update on the importance of plasma protein binding in drug discovery and development

Introduction: Plasma protein binding (PPB) remains a controversial topic in drug discovery and development. Fraction unbound (f_u) is a critical parameter that needs to be measured accurately, because it has significant impacts on the predictions of drug-drug interactions (DDI), estimations of therapeutic indices (TI), and developments of PK/PD relationships.  However, it is generally not advisable to change PPB through structural modifications, because PPB on its own has little relevance for in vivo efficacy.Areas covered: PPB fundamentals are discussed including the three main classes of drug binding proteins (i.e., albumin, alpha1-acid glycoprotein, and lipoproteins) and their physicochemical properties, in vivo half-life, and synthesis rate.  State-of-the-art methodologies for PPB are highlighted. Applications of PPB in drug discovery and development are presented.Expert opinion: PPB is an old topic in pharmacokinetics, but there are still many misconceptions. Improving the accuracy of PPB for highly bound compounds is an ongoing effort in the field with high priority. As the field continues to generate high quality data, the regulatory agencies will increase their confidence in our ability to accurately measure PPB of highly bound compounds, and experimental fu values below 0.01 will more likely be used for DDI predictions in the future.

Recent developments in in vitro and in vivo models for improved translation of preclinical pharmacokinetics and pharmacodynamics data

Improved pharmacokinetics/pharmacodynamics (PK/PD) prediction in the early stages of drug development is essential to inform lead optimization strategies and reduce attrition rates. Recently, there have been significant advancements in the development of new in vitro and in vivo strategies to better characterize pharmacokinetic properties and efficacy of drug leads. Herein, we review advances in experimental and mathematical models for clearance predictions, advancements in developing novel tools to capture slowly metabolized drugs, in vivo model developments to capture human etiology for supporting drug development, limitations and gaps in these efforts, and a perspective on the future in the field.

Model-based Target Pharmacology Assessment (mTPA): An Approach Using PBPK/PD Modeling and Machine Learning to Design Medicinal Chemistry and DMPK Strategies in Early Drug Discovery

The optimal pharmacokinetic (PK) required for a drug candidate to elicit efficacy is highly dependent on the targeted pharmacology, a relationship that is often not well characterized during early phases of drug discovery. Generic assumptions around PK and potency risk misguiding screening and compound design toward nonoptimal absorption, distribution, metabolism, and excretion (ADME) or molecular properties and ultimately may increase attrition as well as hit-to-lead and lead optimization timelines. The present work introduces model-based target pharmacology assessment (mTPA), a computational approach combining physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling, sensitivity analysis, and machine learning (ML) to elucidate the optimal combination of PK, potency, and ADME specific for the targeted pharmacology. Examples using frequently encountered PK/PD relationships are presented to illustrate its application, and the utility and benefits of deploying such an approach to guide early discovery efforts are discussed.

Improving the Accuracy of Predicted Human Pharmacokinetics: Lessons Learned from the AstraZeneca Drug Pipeline Over Two Decades

During drug discovery and prior to the first human dose of a novel candidate drug, the pharmacokinetic (PK) behavior of the drug in humans is predicted from preclinical data. This helps to inform the likelihood of achieving therapeutic exposures in early clinical development. Once clinical data are available, the observed human PK are compared with predictions, providing an opportunity to assess and refine prediction methods. Application of best practice in experimental data generation and predictive methodologies, and a focus on robust mechanistic understanding of the candidate drug disposition properties before nomination to clinical development, have led to maximizing the probability of successful PK predictions so that 83% of AstraZeneca drug development projects progress in the clinic with no PK issues; and 71% of key PK parameter predictions [64% of area under the curve (AUC) predictions; 78% of maximum concentration (C_max) predictions; and 70% of half-life predictions] are accurate to within twofold. Here, we discuss methods to predict human PK used by AstraZeneca, how these predictions are assessed and what can be learned from evaluating the predictions for 116 candidate drugs.

Further Considerations Towards an Effective and Efficient Oncology Drug Discovery DMPK Strategy

BACKGROUND

DMPK data and knowledge are critical in maximising the probability of developing successful drugs via the application of in silico, in vitro and in vivo approaches in drug discovery.

METHODS

The evaluation, optimisation and prediction of human pharmacokinetics is now a mainstay within drug discovery. These elements are at the heart of the 'right tissue' component of AstraZeneca's '5Rs framework' which, since its adoption, has resulted in increased success of Phase III clinical trials. With the plethora of DMPK related assays and models available, there is a need to continually refine and improve the effectiveness and efficiency of approaches best to facilitate the progression of quality compounds for human clinical testing.

RESULTS

This article builds on previously published strategies from our laboratories, highlighting recent discoveries and successes, that brings our AstraZeneca Oncology DMPK strategy up to date. We review the core aspects of DMPK in Oncology drug discovery and highlight data recently generated in our laboratories that have influenced our screening cascade and experimental design. We present data and our experiences of employing cassette animal PK, as well as re-evaluating in vitro assay design for metabolic stability assessments and expanding our use of freshly excised animal and human tissue to best inform first time in human dosing and dose escalation studies.

CONCLUSION

Application of our updated drug-drug interaction and central nervous system drug exposure strategies are exemplified, as is the impact of physiologically based pharmacokinetic and pharmacokinetic-pharmacodynamic modelling for human predictions.

Accelerated drug discovery by rapid candidate drug identification

The eventual candidate drug (CD) is often already synthesized during early drug discovery but not nominated until much later. To facilitate the rapid identification of a potential CD, a thoroughly worked-out CD target profile (CDTP) with criteria acceptable for the disease target product profile (TPP) is required at the start of lead generation (LG). In addition to driving the compound property optimization, the preclinical project team has to understand the ultimate goal to be able to rapidly identify and progress a potential CD. A screening cascade with meaningful and well-balanced progression criteria based on the CDTP is required to rapidly filter out unwanted compounds and to progress a potential CD through the cascade to candidate selection.

Development and Characterization of a Human Hepatocyte Low Intrinsic Clearance Assay for Use in Drug Discovery

Progression of new chemical entities is a multiparametric process involving a balance of potency; absorption, distribution, metabolism, and excretion; and safety properties. To accurately predict human pharmacokinetics and estimate human efficacious dose, the use of in vitro measures of clearance is often essential. Low metabolic clearance is often targeted to facilitate in vivo exposure and achieve appropriate half-life. Suspension primary human hepatocytes (PHHs) have been successfully used in predictions of clearance. However, incubation times are limited, hindering the limit of quantification. The aims herein were to evaluate the ability of a novel PHH media supplement, HepExtend, in order to maintain cell function, increase culture times, and define the clearance of stable compounds. Cell activity was analyzed with a range of cytochrome P450 (P450) and UDP-glucuronosyltransferase (UGT) substrates, and the mRNA expression of drug disposition and toxicity marker genes was determined. HepExtend and Geltrex were essential to maintain cell activity and viability for 5 days (N = 3 donors). In comparison with CM4000 ± Geltrex, HepExtend + Geltrex displayed a higher level of gene expression on day 1, particularly for the P450s, nuclear receptors, and UGTs. The novel medium, HepExtend + Geltrex, was robust and reproducible in generating statistically significant intrinsic clearance values at 0.1 µl/min/10⁶ cells over a 30-hour period (P < 0.05), which was lower than previously demonstrated. Following regression correction, human hepatic in vivo clearance was predicted within 3-fold for 83% of compounds tested for three human donors, with an average fold error of 2.2. The novel PHH medium, HepExtend, with matrix overlay offers significant improvement in determining compounds with low intrinsic clearance when compared with alternative approaches.

Impact of Physicochemical Properties on Dose and Hepatotoxicity of Oral Drugs

A database containing maximum daily doses of 1841 marketed oral drugs was used to examine the influence of physicochemical properties on dose and hepatotoxicity (drug induced liver injury, DILI). Drugs in the highest ∼20% dose range had significantly reduced mean lipophilicity and molecular weight, increased fractional surface area, increased % of acids, and decreased % of bases versus drugs in the lower ∼60% dose range. Drugs in the ∼20-40% dose range had intermediate mean properties, similar to the mean values for the full drug set. Drugs that are both large and highly lipophilic almost invariably do not have doses in the upper ∼20% range. The results show that oral druglike physicochemical properties are different according to these dose ranges, and this is consistent with maintenance of acceptable safety profiles as efficacious exposure increases. Verified DILI annotations from a compilation of >1000 approved drugs (Chen, M.; et al. Drug Discov. Today, 2016, 21, 648 ) were used. The drugs classified as "No DILI" ( n = 163) had significantly lower dose and lipophilicity, and higher Fsp3 (fraction of carbon atoms that are sp³ hybridized) versus the "Most DILI" ( n = 163) drugs. The percentages of acids were reduced and bases increased in the "No DILI" versus the "Most DILI" groups. Drugs classified as "Less DILI" or "Ambiguous DILI" had intermediate mean values of dose, lipophilicity, Fsp3, and % acids and bases. The impact of lipophilicity and Fsp3 on DILI increases in the upper 20% versus the lower 80% dose range, and a simple decision tree model predicted "No DILI" versus "Most DILI" outcomes with 82% accuracy. The model correctly classified 19 of 22 drugs (86%) that failed in development due to human hepatotoxicity. Because many oral drugs lacking DILI annotations are predicted to be "Most DILI", the model is best used preclinically in conjunction with experimental DILI mitigation.

Discovery of AZD-2098 and AZD-1678, Two Potent and Bioavailable CCR4 Receptor Antagonists

N-(5-Bromo-3-methoxypyrazin-2-yl)-5-chlorothiophene-2-sulfonamide 1 was identified as a hit in a CCR4 receptor antagonist high-throughput screen (HTS) of a subset of the AstraZeneca compound bank. As a hit with a lead-like profile, it was an excellent starting point for a CCR4 receptor antagonist program and enabled the rapid progression through the Lead Identification and Lead Optimization phases resulting in the discovery of two bioavailable CCR4 receptor antagonist candidate drugs.

An Investigation into the Prediction of in Vivo Clearance for a Range of Flavin-containing Monooxygenase Substrates

Flavin-containing monooxygenases (FMO) are metabolic enzymes mediating the oxygenation of nucleophilic atoms such as nitrogen, sulfur, phosphorus, and selenium. These enzymes share similar properties to the cytochrome P450 system but can be differentiated through heat inactivation and selective substrate inhibition by methimazole. This study investigated 10 compounds with varying degrees of FMO involvement to determine the nature of the correlation between human in vitro and in vivo unbound intrinsic clearance. To confirm and quantify the extent of FMO involvement six of the compounds were investigated in human liver microsomal (HLM) in vitro assays using heat inactivation and methimazole substrate inhibition. Under these conditions FMO contribution varied from 21% (imipramine) to 96% (itopride). Human hepatocyte and HLM intrinsic clearance (CL_int) data were scaled using standard methods to determine the predicted unbound intrinsic clearance (predicted CL_{int u}) for each compound. This was compared with observed unbound intrinsic clearance (observed CL_{int u}) values back calculated from human pharmacokinetic studies. A good correlation was observed between the predicted and observed CL_{int u} using hepatocytes (R² = 0.69), with 8 of the 10 compounds investigated within or close to a factor of 2. For HLM the in vitro-in vivo correlation was maintained (R² = 0.84) but the accuracy was reduced with only 3 out of 10 compounds falling within, or close to, twofold. This study demonstrates that human hepatocytes and HLM can be used with standard scaling approaches to predict the human in vivo clearance for FMO substrates.

Six global and local QSPR models of aqueous solubility at pH = 7.4 based on structural similarity and physicochemical descriptors

Aqueous solubility at pH = 7.4 is a very important property for medicinal chemists because this is the pH value of physiological media. The present work describes the application of three different methods (support vector machine (SVM), random forest (RF) and multiple linear regression (MLR)) and three local quantitative structure-property relationship (QSPR) models (regression corrected by nearest neighbours (RCNN), arithmetic mean property (AMP) and local regression property (LoReP)) to construct stable QSPRs with clear mechanistic interpretation. Our data set contained experimental values of aqueous solubility at pH = 7.4 of 387 chemicals (349 in the training set and 38 in the test set including 16 own measurements). The initial descriptor pool contained 210 physicochemical descriptors, calculated from the HYBOT, DRAGON, SYBYL and VolSurf+ programs. Six QSPRs with good statistics based on fundamentals of aqueous solubility and optimization of descriptor space were obtained. Those models have an RMSE close to experimental error (0.70), and are amenable to physical interpretation. The QSPR models developed in this study may be useful for medicinal chemists. Global MLR, RF and SVM models may be valuable for consideration of common factors that influence solubility. The RCNN, AMP and LoReP local models may be helpful for the optimization of aqueous solubility in small sets of related chemicals.

Designing safer oral drugs

Designing an oral drug such that its estimated dose to humans is both efficacious and safe is challenging. During the early design stage, where in vitro or preclinical species in vivo safety data are limited, heuristic-based criteria often related to physicochemical properties are used for guidance. The causal relationship between a compound's log P and its human in vivo toxicity is considered. With respect to designing efficacious oral drugs that potentially have reduced toxicity liabilities, an alternative heuristic-based criterion is proposed based on the amount of compound in the body at steady state. In humans, a threshold for the amount of compound in the body at steady state of 0.5 mg kg-1 is suggested. The criterion is based on the minimum toxic blood-plasma concentration that produces clinically relevant side effects or symptoms in humans for 242 oral drugs. It can be used to estimate a therapeutic window against which a compound's estimated in vivo plasma levels for a particular dose size and frequency can be assessed. The relationship between this criterion and acceptable oral dose sizes for different charge types with different in vivo plasma clearances is discussed.

Intrinsic Clearance Assay Incubational Binding: A Method Comparison

The fraction of unbound drug (fu_inc) in in vitro intrinsic clearance (CL_int) incubation is an important parameter in the pursuit of accurate clearance predictions and is often predicted using algorithms based on drug lipophilicity measures. However, analysis of an AstraZeneca database suggests that simple lipophilicity alone is a relatively poor predictor of fu_inc measured using equilibrium dialysis. He fu_inc value can also be measured directly in CL_int assays using multiple concentrations of hepatocytes or microsomal protein. Since this approach informs of the unbound drug concentration in the assay used to predict in vivo clearance, it should be considered the gold standard method. As a starting point for building better predictive algorithms we aimed to determine if equilibrium dialysis really is an appropriate assay for assessing fu_inc Employing a large number of compounds with a wide range of lipophilicities, experiments were performed to measure fu_inc using rat hepatocytes (RH) and human liver microsomes (HLM) in both assay formats. A high percentage (94% and 93% for HLM and RH, respectively) of the fu_inc values were within 2-fold when the compound distribution coefficient describing the ratio of compound concentration in octanol and pH 7.4 buffer when the test system is at equilibrium (lipophilicity measure) (logD_7.4) values were less than 3.5. However, with logD_7.4 values greater than these, the agreement was considerably worse. Additional experimental data generated indicated that this discrepancy was likely due to failings in the direct method when drug binding is high. Thus, we conclude that unbound CL_int can be indeed calculated indirectly by incorporating equilibrium dialysis data with measured CL_int but that simple lipophilicity descriptors alone may be inadequate for predicting fu_inc.

Mass Spectrometry Imaging in Oncology Drug Discovery

Over the last decade mass spectrometry imaging (MSI) has been integrated in to many areas of drug discovery and development. It can have significant impact in oncology drug discovery as it allows efficacy and safety of compounds to be assessed against the backdrop of the complex tumour microenvironment. We will discuss the roles of MSI in investigating compound and metabolite biodistribution and defining pharmacokinetic -pharmacodynamic relationships, analysis that is applicable to all drug discovery projects. We will then look more specifically at how MSI can be used to understand tumour metabolism and other applications specific to oncology research. This will all be described alongside the challenges of applying MSI to industry research with increased use of metrology for MSI.

Molecular inflation, attrition and the rule of five

Physicochemical properties underlie all aspects of drug action and are critical for solubility, permeability and successful formulation. Specific physicochemical properties shown to be relevant to oral drugs are size, lipophilicity, ionisation, hydrogen bonding, polarity, aromaticity and shape. The rule of 5 (Ro5) and subsequent studies have raised awareness of the importance of compound quality amongst bioactive molecules. Lipophilicity, probably the most important physical property of oral drugs, has on average changed little over time in oral drugs, until increases in drugs published after 1990. In contrast other molecular properties such as average size have increased significantly. Factors influencing property inflation include the targets pursued, where antivirals frequently violate the Ro5, risk/benefit considerations, and variable drug discovery practices. The compounds published in patents from the pharmaceutical industry are on average larger, more lipophilic and less complex than marketed oral drugs. The variation between individual companies' patented compounds is due to different practices and not to the targets pursued. Overall, there is demonstrable physical property attrition in moving from patents to candidate drugs to marketed drugs. The pharmaceutical industry's recent poor productivity has been due, in part, to progression of molecules that are unable to unambiguously test clinical efficacy, and attrition can therefore be improved by ensuring candidate drug quality is 'fit for purpose.' The combined ligand efficiency (LE) and lipophilic ligand efficiency (LLE) values of many marketed drugs are optimised relative to other molecules acting at the same target. Application of LLE in optimisation can help identify improved leads, even with challenging targets that seem to require lipophilic ligands. Because of their targets, some projects may need to pursue 'beyond Ro5' physicochemical space; such projects will require non-standard lead generation and optimisation and should not dominate in a well-balanced portfolio. Compound quality is controllable by lead selection and optimisation and should not be a cause of clinical failure.

Gut Wall Metabolism. Application of Pre-Clinical Models for the Prediction of Human Drug Absorption and First-Pass Elimination

Quantifying the multiple processes which control and modulate the extent of oral bioavailability for drug candidates is critical to accurate projection of human pharmacokinetics (PK). Understanding how gut wall metabolism and hepatic elimination factor into first-pass clearance of drugs has improved enormously. Typically, the cytochrome P450s, uridine 5'-diphosphate-glucuronosyltransferases and sulfotransferases, are the main enzyme classes responsible for drug metabolism. Knowledge of the isoforms functionally expressed within organs of first-pass clearance, their anatomical topology (e.g. zonal distribution), protein homology and relative abundances and how these differ across species is important for building models of human metabolic extraction. The focus of this manuscript is to explore the parameters influencing bioavailability and to consider how well these are predicted in human from animal models or from in vitro to in vivo extrapolation. A unique retrospective analysis of three AstraZeneca molecules progressed to first in human PK studies is used to highlight the impact that species differences in gut wall metabolism can have on predicted human PK. Compared to the liver, pharmaceutical research has further to go in terms of adopting a common approach for characterisation and quantitative prediction of intestinal metabolism. A broad strategy is needed to integrate assessment of intestinal metabolism in the context of typical DMPK activities ongoing within drug discovery programmes up until candidate drug nomination.

Determination of Human Hepatocyte Intrinsic Clearance for Slowly Metabolized Compounds: Comparison of a Primary Hepatocyte/Stromal Cell Co-culture with Plated Primary Hepatocytes and HepaRG

A key requirement in drug discovery is to accurately define intrinsic clearance (CL(int)) values of less than 1 µl/min/10(6) hepatocytes, which requires assays that allow for longer incubation time as a complement to suspended hepatocytes. This study assessed the effectiveness of plated HepaRG cells, plated primary human hepatocytes (PHHs), and the HµREL human hepatocyte/stromal cell co-cultures for determination of low CL(int) values. The investigation demonstrated that the systems were capable of providing statistically significant CL(int) estimations down to 0.2 µl/min/10(6) cells. The HµREL assay provided a higher level of reproducibility, with repeat significant CL(int) values being defined in a minimum of triplicate consecutive assays for six of seven of the low CL(int) compounds compared with four of seven for PHHs and two of seven for HepaRG. The assays were also compared with a suspension assay using drugs with higher CL(int) values and diverse enzymology. The CL(int) values from the PHH and HµREL assays were similar to those defined by a hepatocyte suspension assay, indicating that they can be used interchangeably alongside a standard assay. Finally, data from these two assays could also predict in vivo hepatic metabolic CL(int) to within 3-fold for greater than 70% of the compounds tested, with average fold errors (AFE) of 1.6 and 2.3, respectively, whereas the HepaRG data were predictive to within 3-fold for only 50% of compounds (AFE 2.9). In summary, all systems have utility for low CL(int) determination, but the HµREL co-culture appears slightly superior regarding overall assay performance.

Validation of Early Human Dose Prediction: A Key Metric for Compound Progression in Drug Discovery

Human dose prediction is increasingly recognized as an important parameter in Drug Discovery. Validation of a method using only in vitro and predicted parameters incorporated into a PK model was undertaken by predicting human dose and free Cmax for a number of marketed drugs and AZ Development compounds. Doses were compared to those most relevant to marketed drugs or to clinically administered doses of AZ compounds normalized either to predicted Cmin or Cmax values. Average (AFE) and absolute average (AAFE) fold-error analysis showed that best predictions were obtained using a QSAR model as the source of Vss, with Fabs set to 1 for acids and 0.5 for all other ion classes; for clearance prediction no binding correction to the well stirred model (WSM) was used for bases, while it was set to Fup/Fup(0.5) for all other ion classes. Using this combination of methods, predicted doses for 45 to 68% of the Cmin- and Cmax-normalized and marketed drug data sets were within 3-fold of the observed values, while 82 to 92% of these data sets were within 10-fold. This method for early human dose prediction is able to rank, identify, and flag risks or optimization opportunities for future development compounds within 10 days of first synthesis.

Pharmacokinetics in Drug Discovery: An Exposure-Centred Approach to Optimising and Predicting Drug Efficacy and Safety

The role of pharmacokinetics (PK) in drug discovery is to support the optimisation of the absorption, distribution, metabolism and excretion (ADME) properties of lead compounds with the ultimate goal to attain a clinical candidate which achieves a concentration-time profile in the body that is adequate for the desired efficacy and safety profile. A thorough characterisation of the lead compounds aiming at the identification of the inherent PK liabilities also includes an early generation of PK/PD relationships linking in vitro potency and target exposure/engagement with expression of pharmacological activity (mode-of-action) and efficacy in animal studies. The chapter describes an exposure-centred approach to lead generation, lead optimisation and candidate selection and profiling that focuses on a stepwise generation of an understanding between PK/exposure and PD/efficacy relationships by capturing target exposure or surrogates thereof and cellular mode-of-action readouts in vivo. Once robust PK/PD relationship in animal PD models has been constructed, it is translated to anticipate the pharmacologically active plasma concentrations in patients and the human therapeutic dose and dosing schedule which is also based on the prediction of the PK behaviour in human as described herein. The chapter outlines how the level of confidence in the predictions increases with the level of understanding of both the PK and the PK/PD of the new chemical entities (NCE) in relation to the disease hypothesis and the ability to propose safe and efficacious doses and dosing schedules in responsive patient populations. A sound identification of potential drug metabolism and pharmacokinetics (DMPK)-related development risks allows proposing of an effective de-risking strategy for the progression of the project that is able to reduce uncertainties and to increase the probability of success during preclinical and clinical development.

Oral drug-likeness criteria in preclinical species

This study considers how the estimated in vivo free plasma concentrations for compounds tends to vary between human, dog and rat and proposes empirical-based criteria to aid drug design.

Profiling the estimated plasma concentrations of 215 marketed oral drugs

The human pharmacokinetic parameters of 215 marketed oral drugs have been collated and their estimated plasma concentrations (following repeat dosing) profiled against time using a one-compartment model.

Montelukast Disposition: No Indication of Transporter-Mediated Uptake in OATP2B1 and OATP1B1 Expressing HEK293 Cells

Clinical studies with montelukast show variability in effect and polymorphic OATP2B1-dependent absorption has previously been implicated as a possible cause. This claim has been challenged with conflicting data and here we used OATP2B1-transfected HEK293 cells to clarify the mechanisms involved. For montelukast, no significant difference in cell uptake between HEK-OATP2B1 and empty vector cell lines was observed at pH 6.5 or pH 7.4, and no concentration-dependent uptake was detected. Montelukast is a carboxylic acid, a relatively potent inhibitor of OATP1B1, OATP1B3, and OATP2B1, and has previously been postulated to be actively transported into human hepatocytes. Using OATP1B1-transfected HEK293 cells and primary human hepatocytes in the presence of OATP inhibitors we demonstrate for the first time that active OATP-dependent transport is unlikely to play a significant role in the human disposition of montelukast.

A new paradigm for navigating compound property related drug attrition

Improving the efficiency of drug discovery remains a major focus for the pharmaceutical industry. Toxicity accounts for 90% of withdrawals and major early-stage terminations relate to suboptimal efficacy and safety. Traditional oral drug space is well defined with respect to physicochemical properties and ADMET risks but increased focus on ligand-lipophilicity efficiency, maximizing enthalpy contributions and new target classes challenge this paradigm. A hybrid space has been identified that combines physical properties and key interactions attributable to drug transporters. A novel algorithm is proposed that incorporates drug-transporter interactions and its utility evaluated against popular ligand efficiency indices. Simply reducing the bulk properties of compounds can exchange one problem for another and creates high-risk areas that challenge the successful delivery from a balanced portfolio.

In silico Prediction of Aqueous Solubility: a Comparative Study of Local and Global Predictive Models

32 Quantitative Structure-Property Relationship (QSPR) models were constructed for prediction of aqueous intrinsic solubility of liquid and crystalline chemicals. Data sets contained 1022 liquid and 2615 crystalline compounds. Multiple Linear Regression (MLR), Support Vector Machine (SVM) and Random Forest (RF) methods were used to construct global models, and k-nearest neighbour (kNN), Arithmetic Mean Property (AMP) and Local Regression Property (LoReP) were used to construct local models. A set of the best QSPR models was obtained: for liquid chemicals with RMSE (root mean square error) of prediction in the range 0.50-0.60 log unit; for crystalline chemicals 0.80-0.90 log unit. In the case of global models the large number of descriptors makes mechanistic interpretation difficult. The local models use only one or two descriptors, so that a medicinal chemist working with sets of structurally-related chemicals can readily estimate their solubility. However, construction of stable local models requires the presence of closely related neighbours for each chemical considered. It is probable that a consensus of global and local QSPR models will be the optimal approach for construction of stable predictive QSPR models with mechanistic interpretation.

Pharmacoinformatics approach for investigation of alternative potential hepatitis C virus nonstructural protein 5B inhibitors

Hepatitis C virus (HCV) is one of the major viruses affecting the world today. It is a highly variable virus, having a rapid reproduction and evolution rate. The variability of genomes is due to hasty replication catalyzed by nonstructural protein 5B (NS5B) which is also a potential target site for the development of anti-HCV agents. Recently, the US Food and Drug Administration approved sofosbuvir as a novel oral NS5B inhibitor for the treatment of HCV. Unfortunately, it is much highlighted for its pricing issues. Hence, there is an urgent need to scrutinize alternate therapies against HCV that are available at affordable price and do not have associated side effects. Such a need is crucial especially in underdeveloped countries. The search for various new bioactive compounds from plants is a key part of pharmaceutical research. In the current study, we applied a pharmacoinformatics-based approach for the identification of active plant-derived compounds against NS5B. The results were compared to docking results of sofosbuvir. The lead compounds with high-binding ligands were further analyzed for pharmacokinetic and pharmacodynamic parameters based on in silico absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile. The results showed the potential alternative lead compounds that can be developed into commercial drugs having high binding energy and promising ADMET properties.

Calculation of aqueous solubility of crystalline un-ionized organic chemicals and drugs based on structural similarity and physicochemical descriptors

Solubilities of crystalline organic compounds calculated according to AMP (arithmetic mean property) and LoReP (local one-parameter regression) models based on structural and physicochemical similarities are presented. We used data on water solubility of 2615 compounds in un-ionized form measured at 25±5 °C. The calculation results were compared with the equation based on the experimental data for lipophilicity and melting point. According to statistical criteria, the model based on structural and physicochemical similarities showed a better fit with the experimental data. An additional advantage of this model is that it uses only theoretical descriptors, and this provides means for calculating water solubility for both existing and not yet synthesized compounds.