Metabolism-driven in vitro/in vivo disconnect of an oral ERɑ VHL-PROTAC.

Targeting the estrogen receptor alpha (ERα) pathway is validated in the clinic as an effective means to treat ER+ breast cancers. Here we present the development of a VHL-targeting and orally bioavailable proteolysis-targeting chimera (PROTAC) degrader of ERα. In vitro studies with this PROTAC demonstrate excellent ERα degradation and ER antagonism in ER+ breast cancer cell lines. However, upon dosing the compound in vivo we observe an in vitro-in vivo disconnect. ERα degradation is lower in vivo than expected based on the in vitro data. Investigation into potential causes for the reduced maximal degradation reveals that metabolic instability of the PROTAC linker generates metabolites that compete for binding to ERα with the full PROTAC, limiting degradation. This observation highlights the requirement for metabolically stable PROTACs to ensure maximal efficacy and thus optimisation of the linker should be a key consideration when designing PROTACs.

Development of a Series of Pyrrolopyridone MAT2A Inhibitors.

The optimization of an allosteric fragment, discovered by differential scanning fluorimetry, to an in vivo MAT2a tool inhibitor is discussed. The structure-based drug discovery approach, aided by relative binding free energy calculations, resulted in AZ'9567 (21), a potent inhibitor in vitro with excellent preclinical pharmacokinetic properties. This tool showed a selective antiproliferative effect on methylthioadenosine phosphorylase (MTAP) KO cells, both in vitro and in vivo, providing further evidence to support the utility of MAT2a inhibitors as potential anticancer therapies for MTAP-deficient tumors.

Evaluating the performance of machine-learning regression models for pharmacokinetic drug-drug interactions.

Combination therapy or concomitant drug administration can be associated with pharmacokinetic drug-drug interactions, increasing the risk of adverse drug events and reduced drug efficacy. Thus far, machine-learning models have been developed that can classify drug-drug interactions. However, to enable quantification of the pharmacokinetic effects of a drug-drug interaction, regression-based machine learning should be explored. Therefore, this study investigated the use of regression-based machine learning to predict changes in drug exposure caused by pharmacokinetic drug-drug interactions. Fold changes in exposure relative to substrate drug monotherapy were collected from 120 clinical drug-drug interaction studies extracted from the Washington Drug Interaction Database and SimCYP compound library files. Drug characteristics (features) were collected such as structure, physicochemical properties, in vitro pharmacokinetic properties, cytochrome P450 metabolic activity, and population characteristics. Three different regression-based supervised machine-learning models were then applied to the prediction task: random forest, elastic net, and support vector regressor. Model performance was evaluated using fivefold cross-validation. Strongest performance was observed with support vector regression, with 78% of predictions within twofold of the observed exposure changes. The results show that changes in drug exposure can be predicted with reasonable accuracy using regression-based machine-learning models trained on data available early in drug discovery. This has potential applications in enabling earlier drug-drug interaction risk assessment for new drug candidates.

The role of the efflux transporter, P-glycoprotein, at the blood-brain barrier in drug discovery.

The blood-brain barrier (BBB) expresses a high abundance of transporters, particularly P-glycoprotein (P-gp), that regulate endogenous and exogenous molecule uptake and removal of waste. This review discusses key drug metabolism and pharmacokinetic considerations for the efflux transporter P-gp at the BBB in drug discovery and development. We highlight the differences in P-gp expression and protein levels across species but the limited observations of species-specific substrates. Given the impact of age and disease on BBB biology, we summarise the modulation of P-gp for several neurological disorders and ageing and exemplify several disease-specific hurdles or opportunities for drug exposure in the brain. Furthermore, the review includes observations of CNS-related drug-drug interactions due to the inhibition or induction of P-gp at the BBB in animal studies and humans and the need for continued evaluation especially for compounds with a narrow therapeutic window. This review focusses primarily on small molecules but also considers the impact of new chemical entities, particularly beyond Ro5 molecules and their potential to be recognised as P-gp substrates as well as advanced drug delivery systems which offer an alternative approach to achieve and sustain central nervous system exposure.

Use of selective substrates and inhibitors to rapidly characterise batches of cryopreserved primary human hepatocytes for assessment of active uptake liability in drug discovery and development.

The use of hepatocytes to predict human hepatic metabolic clearance is the gold standard approach. However whilst enzymes are well characterised, knowledge gaps remain for transporters. Furthermore, methods to study specific transporter involvement are often complicated by overlapping substrate specificity. Selective substrates and inhibitors would aid investigations into clinically relevant pharmacokinetic effects. However, to date no consensus has been reached.This work defines selective hepatic uptake transporter substrates and inhibitors for the six main human hepatocyte transporters (OATP1B1, OATP1B3, OATP2B1, NTCP, OAT2 & OCT1), and demonstrates their use to rapidly characterise batches of human hepatocytes for uptake transporter activity. Hepatic uptake was determined across a range of substrate concentrations, allowing the definition of kinetic parameters and hence active and passive components. Systematic investigations identified a specific substrate and inhibitor for each transporter, with no overlap between the specificity of substrate and inhibitor for any given transporter.Early characterisation of compound interactions with uptake transporters will aid in early risk assessment and chemistry design. Hence, this work further highlights the feasibility of a refined methodology for rapid compound characterisation for the application of static and dynamic models, for early clinical risk assessment and guidance for the clinical development plan.

Comparing the applications of machine learning, PBPK, and population pharmacokinetic models in pharmacokinetic drug-drug interaction prediction.

The gold-standard approach for modeling pharmacokinetic mediated drug-drug interactions is the use of physiologically-based pharmacokinetic modeling and population pharmacokinetics. However, these models require extensive amounts of drug-specific data generated from a wide variety of in vitro and in vivo models, which are later refined with clinical data and system-specific parameters. Machine learning has the potential to be utilized for the prediction of drug-drug interactions much earlier in the drug discovery cycle, using inputs derived from, among others, chemical structure. This could lead to refined chemical designs in early drug discovery. Machine-learning models have many advantages, such as the capacity to automate learning (increasing the speed and scalability of predictions), improved generalizability by learning from multicase historical data, and highlighting statistical and potentially clinically significant relationships between input variables. In contrast, the routinely used mechanistic models (physiologically-based pharmacokinetic models and population pharmacokinetics) are currently considered more interpretable, reliable, and require a smaller sample size of data, although insights differ on a case-by-case basis. Therefore, they may be appropriate for later stages of drug-drug interaction assessment when more in vivo and clinical data are available. A combined approach of using mechanistic models to highlight features that can be used for training machine-learning models may also be exploitable in the future to improve the performance of machine learning. In this review, we provide concepts, strategic considerations, and compare machine learning to mechanistic modeling for drug-drug interaction risk assessment across the stages of drug discovery and development.

Optimization of hERG and Pharmacokinetic Properties for Basic Dihydro-8H-purin-8-one Inhibitors of DNA-PK.

The DNA-PK complex is activated by double-strand DNA breaks and regulates the non-homologous end-joining repair pathway; thus, targeting DNA-PK by inhibiting the DNA-PK catalytic subunit (DNA-PKcs) is potentially a useful therapeutic approach for oncology. A previously reported series of neutral DNA-PKcs inhibitors were modified to incorporate a basic group, with the rationale that increasing the volume of distribution while maintaining good metabolic stability should increase the half-life. However, adding a basic group introduced hERG activity, and basic compounds with modest hERG activity (IC50 = 10-15 µM) prolonged QTc (time from the start of the Q wave to the end of the T wave, corrected by heart rate) in an anaesthetized guinea pig cardiovascular model. Further optimization was necessary, including modulation of pK a, to identify compound 18, which combines low hERG activity (IC50 = 75 µM) with excellent kinome selectivity and favorable pharmacokinetic properties.

Harnessing Clinical Trial and Real-World Data Towards an Understanding of Sex Effects on Drug Pharmacokinetics, Pharmacodynamics and Efficacy.

Increasing clinical data on sex-related differences in drug efficacy and toxicity has highlighted the importance of understanding the impact of sex on drug pharmacokinetics and pharmacodynamics. Intrinsic differences between males and females, such as different CYP enzyme activity, drug transporter expression or levels of sex hormones can all contribute to different responses to medications. However, most studies do not include sex-specific investigations, leading to lack of sex-disaggregated pharmacokinetic and pharmacodynamic data. Based available literature, the potential influence of sex on exposure-response relationship has not been fully explored for many drugs used in clinical practice, though population-based pharmacokinetic/pharmacodynamic modelling is well-placed to explore this effect. The aim of this review is to highlight existing knowledge gaps regarding the effect of sex on clinical outcomes, thereby proposing future research direction for the drugs with significant sex differences. Based on evaluated drugs encompassing all therapeutic areas, 25 drugs demonstrated a clinically meaningful sex differences in drug exposure (characterised by ≥ 50% change in drug exposure) and this altered PK was correlated with differential response.

Identification and Strategies to Mitigate High Total Clearance of Benzylamine-Substituted Biphenyl Ring Systems.

For most oral small-molecule projects within drug discovery, the extent and duration of the effect are influenced by the total clearance of the compound; hence, designing compounds with low clearance remains a key focus to help enable sufficient protein target engagement. Comprehensive understanding and accurate prediction of animal clearance and pharmacokinetics provides confidence that the same can be observed for human. During a MERTK inhibitor lead optimization project, a series containing a biphenyl ring system with benzylamine meta-substitution on one phenyl and nitrogen inclusion as the meta atom on the other ring demonstrated multiple routes of compound elimination in rats. Here, we describe the identification of a structural pharmacophore involving two key interactions observed for both the MERTK program and an additional internal project. Four strategies to mitigate these clearance liabilities were identified and systematically investigated. We provide evidence that disruption of at least one of the interactions led to a significant reduction in CL that was subsequently predicted from rat hepatocytes using in vitro/in vivo extrapolation and the well-stirred scaling method. These tactics will likely be of general utility to the medicinal chemistry and DMPK community during compound optimization when similar issues are encountered for biphenyl benzylamines.

Prediction of In Vivo Pharmacokinetic Parameters and Time-Exposure Curves in Rats Using Machine Learning from the Chemical Structure.

Animal pharmacokinetic (PK) data as well as human and animal in vitro systems are utilized in drug discovery to define the rate and route of drug elimination. Accurate prediction and mechanistic understanding of drug clearance and disposition in animals provide a degree of confidence for extrapolation to humans. In addition, prediction of in vivo properties can be used to improve design during drug discovery, help select compounds with better properties, and reduce the number of in vivo experiments. In this study, we generated machine learning models able to predict rat in vivo PK parameters and concentration-time PK profiles based on the molecular chemical structure and either measured or predicted in vitro parameters. The models were trained on internal in vivo rat PK data for over 3000 diverse compounds from multiple projects and therapeutic areas, and the predicted endpoints include clearance and oral bioavailability. We compared the performance of various traditional machine learning algorithms and deep learning approaches, including graph convolutional neural networks. The best models for PK parameters achieved R2 = 0.63 [root mean squared error (RMSE) = 0.26] for clearance and R2 = 0.55 (RMSE = 0.46) for bioavailability. The models provide a fast and cost-efficient way to guide the design of molecules with optimal PK profiles, to enable the prediction of virtual compounds at the point of design, and to drive prioritization of compounds for in vivo assays.

Pharmacological inhibition of MERTK induces in vivo retinal degeneration: a multimodal imaging ocular safety assessment.

The receptor tyrosine kinase, MERTK, plays an essential role in homeostasis of the retina via efferocytosis of shed outer nuclear segments of photoreceptors. The Royal College of Surgeons rat model of retinal degeneration has been linked to loss-of-function of MERTK, and together with the MERTK knock-out mouse, phenocopy retinitis pigmentosa in humans with MERTK mutations. Given recent efforts and interest in MERTK as a potential immuno-oncology target, development of a strategy to assess ocular safety at an early pre-clinical stage is critical. We have applied a state-of-the-art, multi-modal imaging platform to assess the in vivo effects of pharmacological inhibition of MERTK in mice. This involved the application of mass spectrometry imaging (MSI) to characterize the ocular spatial distribution of our highly selective MERTK inhibitor; AZ14145845, together with histopathology and transmission electron microscopy to characterize pathological and ultra-structural change in response to MERTK inhibition. In addition, we assessed the utility of a human retinal in vitro cell model to identify perturbation of phagocytosis post MERTK inhibition. We identified high localized total compound concentrations in the retinal pigment epithelium (RPE) and retinal lesions following 28 days of treatment with AZ14145845. These lesions were present in 4 of 8 treated animals, and were characterized by a thinning of the outer nuclear layer, loss of photoreceptors (PR) and accumulation of photoreceptor outer segments at the interface of the RPE and PRs. Furthermore, the lesions were very similar to that shown in the RCS rat and MERTK knock-out mouse, suggesting a MERTK-induced mechanism of PR cell death. This was further supported by the observation of reduced phagocytosis in the human retinal cell model following treatment with AZ14145845. Our study provides a viable, translational strategy to investigate the pre-clinical toxicity of MERTK inhibitors but is equally transferrable to novel chemotypes.

Diverse, Potent, and Efficacious Inhibitors That Target the EED Subunit of the Polycomb Repressive Complex 2 Methyltransferase.

Aberrant activity of the histone methyltransferase polycomb repressive complex 2 (PRC2) has been linked to several cancers, with small-molecule inhibitors of the catalytic subunit of the PRC2 enhancer of zeste homologue 2 (EZH2) being recently approved for the treatment of epithelioid sarcoma (ES) and follicular lymphoma (FL). Compounds binding to the EED subunit of PRC2 have recently emerged as allosteric inhibitors of PRC2 methyltransferase activity. In contrast to orthosteric inhibitors that target EZH2, small molecules that bind to EED retain their efficacy in EZH2 inhibitor-resistant cell lines. In this paper we disclose the discovery of potent and orally bioavailable EED ligands with good solubilities. The solubility of the EED ligands was optimized through a variety of design tactics, with the resulting compounds exhibiting in vivo efficacy in EZH2-driven tumors.

Machine Learning Models for Human In Vivo Pharmacokinetic Parameters with In-House Validation.

Prior to clinical development, a comprehensive pharmacokinetic characterization of a novel drug is required to understand its exposure at the site of action and elimination. Accordingly, in vitro assays and animal pharmacokinetic studies are regularly employed to predict drug exposure in humans, which is often costly and time-consuming. For this reason, the prediction of human pharmacokinetics at the point of design would be of high value for drug discovery. Therefore, we have established a comprehensive data curation protocol that enables machine learning evaluation of 12 human in vivo pharmacokinetic parameters using only chemical structure information and available doses for 1001 unique compounds. These machine learning models were thoroughly investigated and validated using both an independent hold-out test set and AstraZeneca clinical data. In addition, the availability of preclinical predictions for a subset of internal clinical candidates allowed us to compare our in silico approach with state-of-the-art pharmacokinetic predictions. Based on this evaluation, three fit-for-purpose models for AUC PO (Rtest2 = 0.63; RMSEtest = 0.76), Cmax PO (Rtest2 = 0.68; RMSEtest = 0.62), and Vdss IV (Rtest2 = 0.47; RMSEtest = 0.50) were identified. Based on the findings, our machine learning models have considerable potential for practical applications in drug discovery, such as influencing decision-making in drug discovery projects and progression of drug candidates toward the clinic.

Optimization of an Imidazo[1,2-a]pyridine Series to Afford Highly Selective Type I1/2 Dual Mer/Axl Kinase Inhibitors with In Vivo Efficacy.

Inhibition of Mer and Axl kinases has been implicated as a potential way to improve the efficacy of current immuno-oncology therapeutics by restoring the innate immune response in the tumor microenvironment. Highly selective dual Mer/Axl kinase inhibitors are required to validate this hypothesis. Starting from hits from a DNA-encoded library screen, we optimized an imidazo[1,2-a]pyridine series using structure-based compound design to improve potency and reduce lipophilicity, resulting in a highly selective in vivo probe compound 32. We demonstrated dose-dependent in vivo efficacy and target engagement in Mer- and Axl-dependent efficacy models using two structurally differentiated and selective dual Mer/Axl inhibitors. Additionally, in vivo efficacy was observed in a preclinical MC38 immuno-oncology model in combination with anti-PD1 antibodies and ionizing radiation.

Free energy perturbation in the design of EED ligands as inhibitors of polycomb repressive complex 2 (PRC2) methyltransferase.

Free Energy Perturbation (FEP) calculations can provide high-confidence predictions of the interaction strength between a ligand and its protein target. We sought to explore a series of triazolopyrimidines which bind to the EED subunit of the PRC2 complex as potential anticancer therapeutics, using FEP calculations to inform compound design. Combining FEP predictions with a late-stage functionalisation (LSF) inspired synthetic approach allowed us to rapidly evaluate structural modifications in a previously unexplored region of the EED binding site. This approach generated a series of novel triazolopyrimidine EED ligands with improved physicochemical properties and which inhibit PRC2 methyltransferase activity in a cancer-relevant G401 cell line.

Blood retinal barrier and ocular pharmacokinetics: Considerations for the development of oncology drugs.

Tyrosine kinase inhibitors (TKIs) are an example of targeted drug therapy to treat cancer while minimizing damage to healthy tissue. In contrast to traditional oncology drugs, the toxicity profile of targeted therapies is less well understood and can include severe ocular adverse events, which are among the most common toxicity reported by these therapeutics. Inhibition of Mer receptor tyrosine kinase (MERTK) promotes innate tumor immunity by decreasing M2-macrophage polarization and efferocytosis. This mechanism offers the opportunity for targeted immunotherapy to treat cancer; however, the ocular expression of MERTK increases the difficulty for developing a targeted drug due to toxicity concerns. In this article we review the pharmacokinetic (PK) parameters and in vitro absorption, distribution, metabolism, and excretion (ADME) assays available to evaluate ocular disposition and assess the relationship between clinical PK and reported ocular events for TKIs to allow backtranslation to preclinical models. Understanding the ocular disposition in the context of PK and safety remains an evolving area and is likely to be a key aspect of developing safe and efficacious oncology drugs, devoid of ocular toxicity.

A Novel Method for Preventing Non-specific Binding in Equilibrium Dialysis Assays Using Solutol® as an Additive.

Accurate determination of fraction unbound in plasma is required for the interpretation of pharmacology and toxicology data, in addition to predicting human pharmacokinetics, dose, and drug-drug interaction potential. A trend, largely driven by changing target space and new chemical modalities, has increased the occurrence of compounds beyond the traditional rule of 5 physicochemical property space, meaning many drugs under development have high lipophilicity. This can present challenges for ADME assays, including non-specific binding to labware, low dynamic range and solubility. When determining unbound fraction, low recovery, due to non-specific binding, makes bioanalytical sensitivity limiting and prevents determination of free fraction for highly bound compounds. Here, mitigation of non-specific binding through the addition of 0.01% v/v of the excipient Solutol® to an equilibrium dialysis assay has been explored. Solutol® prevented non-specific binding to the dialysis membrane and showed no significant binding to plasma proteins. A test set of compounds demonstrates that this method gives comparable values of fraction unbound. In conclusion, the use of Solutol® as an additive in equilibrium dialysis formats could provide a method of mitigating non-specific binding, enabling the determination of fraction unbound values for highly lipophilic compounds.

Evaluation of the Disconnect between Hepatocyte and Microsome Intrinsic Clearance and In Vitro In Vivo Extrapolation Performance.

The use of in vitro in vivo extrapolation (IVIVE) from human hepatocyte (HH) and human liver microsome (HLM) stability assays is a widely accepted predictive methodology for human metabolic clearance (CLmet). However, a systematic underprediction of CLmet from both matrices appears to be universally apparent, which can be corrected for via an empirical regression offset. After physiological scaling, intrinsic clearance (CLint) for compounds metabolized via the same enzymatic pathway should be equivalent for both matrices. Compounds demonstrating significantly higher HLM CLint relative to HH CLint have been encountered, raising questions regarding how to predict CLmet for such compounds. Here, we determined the HLM:HH CLint ratio for 140 marketed drugs/compounds, compared this ratio as a function of physiochemical properties and drug metabolism enzyme dependence, and examined methodologies to predict CLmet from both matrices. The majority (78%) of compounds displaying a high HLM:HH CLint ratio were CYP3A substrates. Using HH CLint for CYP3A substrates, the current IVIVE regression offset approach remains an appropriate strategy to predict CLmet (% compounds overpredicted/correctly predicted/underpredicted 27/62/11, respectively). However, using the same approach for HLM significantly overpredicts CLmet for CYP3A substrates (% compounds overpredicted/correctly predicted/underpredicted 56/33/11, respectively), highlighting that a different IVIVE offset is required for CYP3A substrates using HLM. This work furthers the understanding of compound properties associated with a disproportionately high HLM:HH CLint ratio and outlines a successful IVIVE approach for such compounds. SIGNIFICANCE STATEMENT: Oral drug discovery programs typically strive for low clearance compounds to ensure sufficient target engagement. Human liver microsomes and isolated human hepatocytes are used to optimize and predict human hepatic metabolic clearance. After physiological scaling, intrinsic clearance for compounds of the same metabolic pathway should be equivalent between matrices. However, a disconnect in intrinsic clearance is sometimes apparent. The work described attempts to further understand this phenomenon, and by achieving a mechanistic understanding, improvements in clearance predictions may be realized.

Optimising proteolysis-targeting chimeras (PROTACs) for oral drug delivery: a drug metabolism and pharmacokinetics perspective.

Proteolysis-targeting chimeras (PROTACs) are an emerging therapeutic modality with the potential to open target space not accessible to conventional small molecules via a degradation-based mechanism; however, their bifunctional nature can result in physicochemical properties that breach commonly accepted limits for small-molecule oral drugs. We offer a drug metabolism and pharmacokinetics (DMPK) perspective on the optimisation of oral PROTACs across a diverse set of projects within Oncology R&D at AstraZeneca, highlighting some of the challenges that they have presented to our established screening cascade. Furthermore, we challenge some of the perceptions and dogma surrounding the feasibility of oral PROTACS and demonstrate that acceptable oral PK properties for this modality can be regularly achievable despite the physicochemical property challenges they present.

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.