Moving beyond Binary Predictions of Human Drug-Induced Liver Injury (DILI) toward Contrasting Relative Risk Potential

Cyanide Trapping of Iminium Ion Reactive Metabolites: Implications for Clinical Hepatotoxicity

Reactive metabolite formation is a major mechanism of hepatotoxicity. Although reactive electrophiles can be soft or hard in nature, screening strategies have generally focused on the use of glutathione trapping assays to screen for soft electrophiles, with many data sets available to support their use. The use of a similar assay for hard electrophiles using cyanide as the trapping agent is far less common, and there is a lack of studies with sufficient supporting data. Using a set of 260 compounds with a defined hepatotoxicity status by the FDA, a comprehensive literature search yielded cyanide trapping data on an unbalanced set of 20 compounds that were all clinically hepatotoxic. Thus, a further set of 19 compounds was selected to generate cyanide trapping data, resulting in a more balanced data set of 39 compounds. Analysis of the data demonstrated that the cyanide trapping assay had high specificity (92%) and a positive predictive value (83%) such that hepatotoxic compounds would be confidently flagged. Structural analysis of the adducts formed revealed artifactual methylated cyanide adducts to also occur, highlighting the importance of full structural identification to confirm the nature of the adduct formed. The assay was demonstrated to add the most value for compounds containing typical structural alerts for hard electrophile formation: half of the severe hepatotoxins with these structural alerts formed cyanide adducts, while none of the severe hepatotoxins with no relevant structural alerts formed adducts. The assay conditions used included cytosolic enzymes (e.g., aldehyde oxidase) and an optimized cyanide concentration to minimize the inhibition of cytochrome P450 enzymes by cyanide. Based on the demonstrated added value of this assay, it is to be initiated for use at GSK as part of the integrated hepatotoxicity strategy, with its performance being reviewed periodically as more data is generated.

Deep Learning Algorithm Based on Molecular Fingerprint for Prediction of Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is one the rare adverse drug reaction (ADR) and multifactorial endpoints. Current preclinical animal models struggle to anticipate it, and in silico methods have emerged as a way with significant potential for doing so. In this study, a high-quality dataset of 1573 compounds was assembled. The 48 classification models, which depended on six different molecular fingerprints, were built via deep neural network (DNN) and seven machine learning algorithms. Comparing the results of the DNN and machine learning models, the optional performing model was found as the one developed based on the DNN with ECFP_6 as input, which achieved the area under the receiver operating characteristic curve (AUC) of 0.713, balanced accuracy (BA) of 0.680, and F1 of 0.753. In addition, we used the SHapley Additive exPlanations (SHAP) algorithm to interpret the models, identified the crucial structural fragments related to DILI risk, and selected the top ten substructures with the highest contribution rankings to serve as warning indicators for subsequent drug hepatotoxicity screening studies. The study demonstrates that the DNN models developed based on molecular fingerprints can be a trustworthy and efficient tool for determining the risk of DILI during the pre-development of novel medications.

Hepatic Bile Acid Transporters and Drug-induced Hepatotoxicity

Drug-induced liver injury (DILI) remains a major concern in drug development from a patient safety perspective because it is the leading cause of acute liver failure. One mechanism of DILI is altered bile acid homeostasis and involves several hepatic bile acid transporters. Functional impairment of some hepatic bile acid transporters by drugs, disease, or genetic mutations may lead to toxic accumulation of bile acids within hepatocytes and increase DILI susceptibility. This review focuses on the role of hepatic bile acid transporters in DILI. Model systems, primarily in vitro and modeling tools, such as DILIsym, used in assessing transporter-mediated DILI are discussed. Due to species differences in bile acid homeostasis and drug-transporter interactions, key aspects and challenges associated with the use of preclinical animal models for DILI assessment are emphasized. Learnings are highlighted from three case studies of hepatotoxic drugs: troglitazone, tolvaptan, and tyrosine kinase inhibitors (dasatinib, pazopanib, and sorafenib). The development of advanced in vitro models and novel biomarkers that can reliably predict DILI is critical and remains an important focus of ongoing investigations to minimize patient risk for liver-related adverse reactions associated with medication use.

New in vitro screening system to detect drug-induced liver injury using a culture plate with low drug sorption and high oxygen permeability

Drug-induced liver injury (DILI) is a major factor underlying drug withdrawal from the market. Therefore, it is important to predict DILI during the early phase of drug discovery. Metabolic activation and mitochondrial toxicity are good indicators of the potential for DILI. However, hepatocyte function, including drug-metabolizing enzyme activity and mitochondrial function, reportedly decreases under conventional culture conditions; therefore, these conditions fail to precisely detect metabolic activation and mitochondrial toxicity-induced cell death. To resolve this issue, we employed a newly developed cell culture plate with high oxygen permeability and low drug sorption (4-polymethyl-1-pentene [PMP] plate). Under PMP plate conditions, cytochrome P450 (CYP) activity and mitochondrial function were increased in primary rat hepatocytes. Following l-buthionine-sulfoximine-induced glutathione depletion, acetaminophen-induced cell death significantly increased under PMP plate conditions. Additionally, 1-aminobenzotriazole reduced cell death. Moreover, mitochondrial toxicity due to mitochondrial complex inhibitors (ketoconazole, metformin, and phenformin) increased under PMP plate conditions. In summary, PMP plate conditions could improve CYP activity and mitochondrial function in primary rat hepatocytes and potentially detect metabolic activation and mitochondrial toxicity.

Chemistry-Based Modeling on Phenotype-Based Drug-Induced Liver Injury Annotation: From Public to Proprietary Data

Drug-induced liver injury (DILI) is an important safety concern and a major reason to remove a drug from the market. Advancements in recent machine learning methods have led to a wide range of in silico models for DILI predictive methods based on molecule chemical structures (fingerprints). Existing publicly available DILI data sets used for model building are based on the interpretation of drug labels or patient case reports, resulting in a typical binary clinical DILI annotation. We developed a novel phenotype-based annotation to process hepatotoxicity information extracted from repeated dose in vivo preclinical toxicology studies using INHAND annotation to provide a more informative and reliable data set for machine learning algorithms. This work resulted in a data set of 430 unique compounds covering diverse liver pathology findings which were utilized to develop multiple DILI prediction models trained on the publicly available data (TG-GATEs) using the compound's fingerprint. We demonstrate that the TG-GATEs compounds DILI labels can be predicted well and how the differences between TG-GATEs and the external test compounds (Johnson & Johnson) impact the model generalization performance.

Leveraging Cell Painting Images to Expand the Applicability Domain and Actively Improve Deep Learning Quantitative Structure-Activity Relationship Models

The search for chemical hit material is a lengthy and increasingly expensive drug discovery process. To improve it, ligand-based quantitative structure-activity relationship models have been broadly applied to optimize primary and secondary compound properties. Although these models can be deployed as early as the stage of molecule design, they have a limited applicability domain─if the structures of interest differ substantially from the chemical space on which the model was trained, a reliable prediction will not be possible. Image-informed ligand-based models partly solve this shortcoming by focusing on the phenotype of a cell caused by small molecules, rather than on their structure. While this enables chemical diversity expansion, it limits the application to compounds physically available and imaged. Here, we employ an active learning approach to capitalize on both of these methods' strengths and boost the model performance of a mitochondrial toxicity assay (Glu/Gal). Specifically, we used a phenotypic Cell Painting screen to build a chemistry-independent model and adopted the results as the main factor in selecting compounds for experimental testing. With the additional Glu/Gal annotation for selected compounds we were able to dramatically improve the chemistry-informed ligand-based model with respect to the increased recognition of compounds from a 10% broader chemical space.

AI/ML Models to Predict the Severity of Drug-Induced Liver Injury for Small Molecules

Drug-induced liver injury (DILI), believed to be a multifactorial toxicity, has been a leading cause of attrition of small molecules during discovery, clinical development, and postmarketing. Identification of DILI risk early reduces the costs and cycle times associated with drug development. In recent years, several groups have reported predictive models that use physicochemical properties or in vitro and in vivo assay endpoints; however, these approaches have not accounted for liver-expressed proteins and drug molecules. To address this gap, we have developed an integrated artificial intelligence/machine learning (AI/ML) model to predict DILI severity for small molecules using a combination of physicochemical properties and off-target interactions predicted in silico. We compiled a data set of 603 diverse compounds from public databases. Among them, 164 were categorized as Most DILI (M-DILI), 245 as Less DILI (L-DILI), and 194 as No DILI (N-DILI) by the FDA. Six machine learning methods were used to create a consensus model for predicting the DILI potential. These methods include k-nearest neighbor (k-NN), support vector machine (SVM), random forest (RF), Naïve Bayes (NB), artificial neural network (ANN), logistic regression (LR), weighted average ensemble learning (WA) and penalized logistic regression (PLR). Among the analyzed ML methods, SVM, RF, LR, WA, and PLR identified M-DILI and N-DILI compounds, achieving a receiver operating characteristic area under the curve of 0.88, sensitivity of 0.73, and specificity of 0.9. Approximately 43 off-targets, along with physicochemical properties (fsp³, log S, basicity, reactive functional groups, and predicted metabolites), were identified as significant factors in distinguishing between M-DILI and N-DILI compounds. The key off-targets that we identified include: PTGS1, PTGS2, SLC22A12, PPARγ, RXRA, CYP2C9, AKR1C3, MGLL, RET, AR, and ABCC4. The present AI/ML computational approach therefore demonstrates that the integration of physicochemical properties and predicted on- and off-target biological interactions can significantly improve DILI predictivity compared to chemical properties alone.

Role of Hepatocyte Transporters in Drug-Induced Liver Injury (DILI)-In Vitro Testing

Bile acids and bile salts (BA/BS) are substrates of both influx and efflux transporters on hepatocytes. Canalicular efflux transporters, such as BSEP and MRP2, are crucial for the removal of BA/BS to the bile. Basolateral influx transporters, such as NTCP, OATP1B1/1B3, and OSTα/β, cooperate with canalicular transporters in the transcellular vectorial flux of BA/BS from the sinusoids to the bile. The blockage of canalicular transporters not only impairs the bile flow but also causes the intracellular accumulation of BA/BS in hepatocytes that contributes to, or even triggers, liver injury. In the case of BA/BS overload, the efflux of these toxic substances back to the blood via MRP3, MRP4, and OST α/β is considered a relief function. FXR, a key regulator of defense against BA/BS toxicity suppresses de novo bile acid synthesis and bile acid uptake, and promotes bile acid removal via increased efflux. In drug development, the early testing of the inhibition of these transporters, BSEP in particular, is important to flag compounds that could potentially inflict drug-induced liver injury (DILI). In vitro test systems for efflux transporters employ membrane vesicles, whereas those for influx transporters employ whole cells. Additional in vitro pharmaceutical testing panels usually include cellular toxicity tests using hepatocytes, as well as assessments of the mitochondrial toxicity and accumulation of reactive oxygen species (ROS). Primary hepatocytes are the cells of choice for toxicity testing, with HepaRG cells emerging as an alternative. Inhibition of the FXR function is also included in some testing panels. The molecular weight and hydrophobicity of the drug, as well as the steady-state total plasma levels, may positively correlate with the DILI potential. Depending on the phase of drug development, the physicochemical properties, dosing, and cut-off values of BSEP IC₅₀ ≤ 25-50 µM or total C_ss,plasma/BSEP IC₅₀ ≥ 0.1 may be an indication for further testing to minimize the risk of DILI liability.

Nonclinical Safety Signals in PharmaPendium Improve the Predictability of Human Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is a leading cause of candidate attrition during drug development in the pharmaceutical industry. This study evaluated liver toxicity signals for 249 approved drugs (114 of "most-DILI concern" and 135 of "no-DILI concern") using PharmaPendium and assessed the association between nonclinical and clinical injuries using contingency table analysis. All animal liver findings were combined into eight toxicity categories based on nature and severity. Together, these analyses revealed that cholestasis [odds ratio (OR): 5.02; 95% confidence interval (CI) 1.04-24.03] or liver aminotransferase increases (OR: 1.86; 95% CI 1.09-3.09) in rats and steatosis (OR-1.9; 95% CI 1.03-3.49) or liver aminotransferase increases (OR-2.57; 95% CI 1.4-4.7) in dogs were significant predictors of human liver injury. The predictive value further improved when the liver injury categories were combined into less severe (steatosis, cholestasis, liver aminotransferase increase, hyperbilirubinemia, or jaundice) and more-severe (liver necrosis, acute liver failure, or hepatotoxicity) injuries. In particular, less-severe liver injuries in the following pairs of species predicted human hepatotoxicity {[dog and mouse] (OR: 2.70; 95% CI 1.25-5.84), [dog and rat] (OR-2.61; 95% CI 1.48-4.59), [monkey and mouse] (OR-4.22; 95% CI 1.33-13.32), and [monkey and rat] (OR-2.45; 95% CI 1.15-5.21)} were predictive of human hepatotoxicity. Meanwhile, severe liver injuries in both [dog and rat] (OR-1.9; 95% CI 1.04-3.49) were significant predictors of human liver toxicity. Therefore, we concluded that the occurrence of DILI in humans is highly likely if liver injuries are observed in one rodent and one nonrodent species and that liver aminotransferase increases in dogs and rats can predict DILI in humans. Together, these findings indicate that the liver safety signals observed in animal toxicity studies indicate potential DILI risk in humans and could therefore be used to prioritize small molecules with less potential to cause DILI in humans.

Early Drug-Induced Liver Injury (DILI) Risk Screening: "Free", as good as it gets

For all the promise of and need for clinical drug-induced liver injury (DILI) risk screening systems, demonstrating the predictive value of these systems versus readily available physicochemical properties and inherent dosing information has not been thoroughly evaluated. Therefore, we utilized a systematic approach to evaluate the predictive value of in vitro safety assays including Bile Salt Export Pump (BSEP) transporter inhibition and cytotoxicity in HepG2 and transformed human liver epithelial (THLE) along with physicochemical properties. We also evaluated the predictive value of in vitro ADME assays including hepatic partition coefficient (Kp) and its unbound counterpart since they provide insight on hepatic accumulation potential. The datasets comprised of 569 marketed drugs with FDA DILIrank annotation (Most vs Less/None), dose and physicochemical information, 384 drugs with Kp and plasma protein binding data, and 279 drugs with safety assay data. For each dataset and combination of input parameters, we developed random forest machine learning models and measured model performance using the receiver operator characteristic area-under-the-curve (ROC AUC). The median ROC AUC across the various data and parameters sets ranged from 0.67 to 0.77 with little evidence of additive predictivity when including safety or ADME assay data. Subsequent machine learning models consistently demonstrated daily dose, fraction sp3 or ionization, and cLogP/D inputs produced the best, simplest model for predicting clinical DILI risk with an ROC AUC of 0.75. This systematic framework should be used for future assay predictive value assessments and highlights the need for continued improvements to clinical DILI risk annotation.

Highly Sensitive Two-Photon Lipid Droplet Tracker for In Vivo Screening of Drug Induced Liver Injury

Lipid droplets (LDs) are lipid-abundant organelles found in most cell lines and primarily consist of neutral lipids. They serve as a repository of various lipids and are associated with many cellular metabolic processes, including energy storage, membrane synthesis, and protein homeostasis. LDs are prominent in a variety of diseases related to lipid regulation, including obesity, fatty liver disease, diabetes, and atherosclerosis. To monitor LD dynamics in live samples, we developed a highly selective two-photon fluorescent tracker for LDs (LD1). It exhibited outstanding sensitivity with a remarkable two-photon-action cross section (Φδ_max > 600 GM), photostability, and low cytotoxicity. In human hepatocytes and in vivo mouse liver tissue imaging, LD1 showed very bright fluorescence with high LD selectivity and minimized background signal to evaluate the stages of nonalcoholic fatty liver disease. Interestingly, we demonstrated that the liver sinusoid morphology became narrower with increasing LD size and visualized the dynamics including fusion of the LDs in vivo. Moreover, real-time and dual-color TPM imaging with LD1 and a two-photon lysosome tracker could be a useful predictive screening tool in the drug development process to monitor impending drug-induced liver injury inducing drug candidates.

In vitro cell-based models of drug-induced hepatotoxicity screening: progress and limitation

Drug-induced liver injury (DILI) is one of the major causes of post-approval withdrawal of therapeutics. As a result, there is an increasing need for accurate predictive in vitro assays that reliably detect hepatotoxic drug candidates while reducing drug discovery time, costs, and the number of animal experiments. In vitro hepatocyte-based research has led to an improved comprehension of the underlying mechanisms of chemical toxicity and can assist the prioritization of therapeutic choices with low hepatotoxicity risk. Therefore, several in vitro systems have been generated over the last few decades. This review aims to comprehensively present the development and validation of 2D (two-dimensional) and 3D (three-dimensional) culture approaches on hepatotoxicity screening of compounds and highlight the main factors affecting predictive power of experiments. To this end, we first summarize some of the recognized hepatotoxicity mechanisms and related assays used to appraise DILI mechanisms and then discuss the challenges and limitations of in vitro models.

Joint Decision-Making Model Based on Consensus Modeling Technology for the Prediction of Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is the major cause of clinical trial failure and postmarketing withdrawals of approved drugs. It is very expensive and time-consuming to evaluate hepatotoxicity using animal or cell-based experiments in the early stage of drug development. In this study, an in silico model based on the joint decision-making strategy was developed for DILI assessment using a relatively large dataset of 2608 compounds. Five consensus models were developed with PaDEL descriptors and PubChem, Substructure, Estate, and Klekota–Roth fingerprints, respectively. Submodels for each consensus model were obtained through joint optimization. The parameters and features of each submodel were optimized jointly based on the hybrid quantum particle swarm optimization (HQPSO) algorithm. The application domain (AD) based on the frequency-weighted and distance (FWD)-based method and Tanimoto similarity index showed the wide AD of the qualified consensus models. A joint decision-making model was integrated by the qualified consensus models, and the overwhelming majority principle was used to improve the performance of consensus models. The application scope narrowing caused by the overwhelming majority principle was successfully solved by joint decision-making. The proposed model successfully predicted 99.2% of the compounds in the test set, with an accuracy of 80.0%, a sensitivity of 83.9, and a specificity of 73.3%. For an external validation set containing 390 compounds collected from DILIrank, 98.2% of the compounds were successfully predicted with an accuracy of 79.9%, a sensitivity of 97.1%, and a specificity of 66.0%. Furthermore, 25 privileged substructures responsible for DILI were identified from Substructure, PubChem, and Klekota–Roth fingerprints. These privileged substructures can be regarded as structural alerts in hepatotoxicity evaluation. Compared with the main published studies, our method exhibits certain advantage in data size, transparency, and standardization of the modeling process and accuracy and credibility of prediction results. It is a promising tool for virtual screening in the early stage of drug development.

Hepatotoxicity reports in the FDA adverse event reporting system database: A comparison of drugs that cause injury via mitochondrial or other mechanisms

Drug-induced liver injury (DILI) is a leading reason for preclinical safety attrition and post-market drug withdrawals. Drug-induced mitochondrial toxicity has been shown to play an essential role in various forms of DILI, especially in idiosyncratic liver injury. This study examined liver injury reports submitted to the Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) for drugs associated with hepatotoxicity via mitochondrial mechanisms compared with non-mitochondrial mechanisms of toxicity. The frequency of hepatotoxicity was determined at a group level and individual drug level. A reporting odds ratio (ROR) was calculated as the measure of effect. Between the two DILI groups, reports for DILI involving mitochondrial mechanisms of toxicity had a 1.43 (95% CI 1.42-1.45; P < 0.0001) times higher odds compared to drugs associated with non-mitochondrial mechanisms of toxicity. Antineoplastic, antiviral, analgesic, antibiotic, and antimycobacterial drugs were the top five drug classes with the highest ROR values. Although the top 20 drugs with the highest ROR values included drugs with both mitochondrial and non-mitochondrial injury mechanisms, the top four drugs (ROR values > 18: benzbromarone, troglitazone, isoniazid, rifampin) were associated with mitochondrial mechanisms of toxicity. The major demographic influence for DILI risk was also examined. There was a higher mean patient age among reports for drugs that were associated with mitochondrial mechanisms of toxicity [56.1 ± 18.33 (SD)] compared to non-mitochondrial mechanisms [48 ± 19.53 (SD)] (P < 0.0001), suggesting that age may play a role in susceptibility to DILI via mitochondrial mechanisms of toxicity. Univariate logistic regression analysis showed that reports of liver injury were 2.2 (odds ratio: 2.2, 95% CI 2.12-2.26) times more likely to be associated with older patient age, as compared with reports involving patients less than 65 years of age. Compared to males, female patients were 37% less likely (odds ratio: 0.63, 95% CI 0.61-0.64) to be subjects of liver injury reports for drugs associated with mitochondrial toxicity mechanisms. Given the higher proportion of severe liver injury reports among drugs associated with mitochondrial mechanisms of toxicity, it is essential to understand if a drug causes mitochondrial toxicity during preclinical drug development when drug design alternatives, more clinically relevant animal models, and better clinical biomarkers may provide a better translation of drug-induced mitochondrial toxicity risk assessment from animals to humans. Our findings from this study align with mitochondrial mechanisms of toxicity being an important cause of DILI, and this should be further investigated in real-world studies with robust designs.

Preclinical models of idiosyncratic drug-induced liver injury (iDILI): Moving towards prediction

Idiosyncratic drug-induced liver injury (iDILI) encompasses the unexpected harms that prescription and non-prescription drugs, herbal and dietary supplements can cause to the liver. iDILI remains a major public health problem and a major cause of drug attrition. Given the lack of biomarkers for iDILI prediction, diagnosis and prognosis, searching new models to predict and study mechanisms of iDILI is necessary. One of the major limitations of iDILI preclinical assessment has been the lack of correlation between the markers of hepatotoxicity in animal toxicological studies and clinically significant iDILI. Thus, major advances in the understanding of iDILI susceptibility and pathogenesis have come from the study of well-phenotyped iDILI patients. However, there are many gaps for explaining all the complexity of iDILI susceptibility and mechanisms. Therefore, there is a need to optimize preclinical human in vitro models to reduce the risk of iDILI during drug development. Here, the current experimental models and the future directions in iDILI modelling are thoroughly discussed, focusing on the human cellular models available to study the pathophysiological mechanisms of the disease and the most used in vivo animal iDILI models. We also comment about in silico approaches and the increasing relevance of patient-derived cellular models.

Modeling approaches for reducing safety-related attrition in drug discovery and development: a review on myelotoxicity, immunotoxicity, cardiovascular toxicity, and liver toxicity

Introduction:Safety and tolerability is a critical area where improvements are needed to decrease the attrition rates during development of new drug candidates. Modeling approaches, when smartly implemented, can contribute to this aim.Areas covered:The focus of this review was on modeling approaches applied to four kinds of drug-induced toxicities: hematological, immunological, cardiovascular (CV) and liver toxicity. Papers, mainly published in the last 10 years, reporting models in three main methodological categories - computational models (e.g., quantitative structure-property relationships, machine learning approaches, neural networks, etc.), pharmacokinetic-pharmacodynamic (PK-PD) models, and quantitative system pharmacology (QSP) models - have been considered.Expert opinion:The picture observed in the four examined toxicity areas appears heterogeneous. Computational models are typically used in all areas as screening tools in the early stages of development for hematological, cardiovascular and liver toxicity, with accuracies in the range of 70-90%. A limited number of computational models, based on the analysis of drug protein sequence, was instead proposed for immunotoxicity. In the later stages of development, toxicities are quantitatively predicted with reasonably good accuracy using either semi-mechanistic PK-PD models (hematological and cardiovascular toxicity), or fully exploited QSP models (immuno-toxicity and liver toxicity).

A property-response perspective on modern toxicity assessment and drug toxicity index (DTI)

Toxicity related failures in drug discovery and clinical development have motivated scientists and regulators to develop a wide range of in-vitro, in-silico tools coupled with data science methods. Older drug discovery rules are being constantly modified to churn out any hidden predictive value. Nonetheless, the dose-response concepts remain central to all these methods. Over the last 2 decades medicinal chemists, and pharmacologists have observed that different physicochemical, and pharmacological properties capture trends in toxic responses. We propose that these observations should be viewed in a comprehensive property-response framework where dose is only a factor that modifies the inherent toxicity potential. We then introduce the recently proposed "Drug Toxicity Index (DTI)" and briefly summarize its applications. A webserver is available to calculate DTI values (https://all-tool-kit.github.io/Web-Tool.html).

The Promise of AI for DILI Prediction

Drug-induced liver injury (DILI) is a common reason for the withdrawal of a drug from the market. Early assessment of DILI risk is an essential part of drug development, but it is rendered challenging prior to clinical trials by the complex factors that give rise to liver damage. Artificial intelligence (AI) approaches, particularly those building on machine learning, range from random forests to more recent techniques such as deep learning, and provide tools that can analyze chemical compounds and accurately predict some of their properties based purely on their structure. This article reviews existing AI approaches to predicting DILI and elaborates on the challenges that arise from the as yet limited availability of data. Future directions are discussed focusing on rich data modalities, such as 3D spheroids, and the slow but steady increase in drugs annotated with DILI risk labels.

Deep Graph Learning with Property Augmentation for Predicting Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is a crucial factor in determining the qualification of potential drugs. However, the DILI property is excessively difficult to obtain due to the complex testing process. Consequently, an in silico screening in the early stage of drug discovery would help to reduce the total development cost by filtering those drug candidates with a high risk to cause DILI. To serve the screening goal, we apply several computational techniques to predict the DILI property, including traditional machine learning methods and graph-based deep learning techniques. While deep learning models require large training data to tune huge model parameters, the DILI data set only contains a few hundred annotated molecules. To alleviate the data scarcity problem, we propose a property augmentation strategy to include massive training data with other property information. Extensive experiments demonstrate that our proposed method significantly outperforms all existing baselines on the DILI data set by obtaining a 81.4% accuracy using cross-validation with random splitting, 78.7% using leave-one-out cross-validation, and 76.5% using cross-validation with scaffold splitting.

A history of the roles of cytochrome P450 enzymes in the toxicity of drugs

The history of drug metabolism began in the 19th Century and developed slowly. In the mid-20th Century the relationship between drug metabolism and toxicity became appreciated, and the roles of cytochrome P450 (P450) enzymes began to be defined in the 1960s. Today we understand much about the metabolism of drugs and many aspects of safety assessment in the context of a relatively small number of human P450s. P450s affect drug toxicity mainly by either reducing exposure to the parent molecule or, in some cases, by converting the drug into a toxic entity. Some of the factors involved are enzyme induction, enzyme inhibition (both reversible and irreversible), and pharmacogenetics. Issues related to drug toxicity include drug-drug interactions, drug-food interactions, and the roles of chemical moieties of drug candidates in drug discovery and development. The maturation of the field of P450 and drug toxicity has been facilitated by advances in analytical chemistry, computational capability, biochemistry and enzymology, and molecular and cell biology. Problems still arise with P450s and drug toxicity in drug discovery and development, and in the pharmaceutical industry the interaction of scientists in medicinal chemistry, drug metabolism, and safety assessment is critical for success.

Improving drug safety predictions by reducing poor analytical practices

Predicting the safety of a drug from preclinical data is a major challenge in drug discovery, and progressing an unsafe compound into the clinic puts patients at risk and wastes resources. In drug safety pharmacology and related fields, methods and analytical decisions known to provide poor predictions are common and include creating arbitrary thresholds, binning continuous values, giving all assays equal weight, and multiple reuse of information. In addition, the metrics used to evaluate models often omit important criteria and models’ performance on new data are often not assessed rigorously. Prediction models with these problems are unlikely to perform well, and published models suffer from many of these issues. We describe these problems in detail, demonstrate their negative consequences, and propose simple solutions that are standard in other disciplines where predictive modelling is used.

Deep Learning on High-Throughput Transcriptomics to Predict Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is one of the most cited reasons for the high drug attrition rate and drug withdrawal from the market. The accumulated large amount of high throughput transcriptomic profiles and advances in deep learning provide an unprecedented opportunity to improve the suboptimal performance of DILI prediction. In this study, we developed an eight-layer Deep Neural Network (DNN) model for DILI prediction using transcriptomic profiles of human cell lines (LINCS L1000 dataset) with the current largest binary DILI annotation data [i.e., DILI severity and toxicity (DILIst)]. The developed models were evaluated by Monte Carlo cross-validation (MCCV), permutation test, and an independent validation (IV) set. The developed DNN model achieved the area under the receiver operating characteristic curve (AUC) of 0.802 and 0.798, and balanced accuracy of 0.741 and 0.721 for training and an IV set, respectively, outperforming the conventional machine learning algorithms, including K-nearest neighbors (KNN), Support Vector Machine (SVM), and Random Forest (RF). Moreover, the developed DNN model provided a more balanced sensitivity of 0.839 and specificity of 0.603. Besides, we found the developed DNN model had a superior predictive performance for oncology drugs. Also, the functional and network analysis of genes driving the predictions revealed their relevance to the underlying mechanisms of DILI. The proposed DNN model could be a promising tool for early detection of DILI potential in the pre-clinical setting.

[The trends in predicting drug-induced liver injury]

Drug-induced liver injury (DILI) is the major reason for the discontinuation of new drug development and the withdrawal of drugs from the market. Hence, the evaluation systems which predict the onset of DILI in the pre-clinical stage are needed. To date, many researchers have conducted the mechanism of DILI, but the DILI prediction is poor because of the complexity of DILI. In this regard, based on the information obtained from basic research and clinical case, several pharmaceutical companies have been developed DILI prediction methods with high sensitivity and specificity by combining multiple targets. Another reason for low predictability is derived from the conventional culture method which causes a rapid decrease in hepatocyte function. To overcome these problems, the construction of a high-level in vitro evaluation system has been developed and applied to DILI evaluation. On the other hand, these in vitro evaluation methods require a lot of labor and cost so, in silico prediction methods have also been constructed in recent years. Based on this point, this article reviews the trends in DILI prediction systems in the non-clinical stage.

An in vitro coculture system of human peripheral blood mononuclear cells with hepatocellular carcinoma-derived cells for predicting drug-induced liver injury

Preventing clinical drug-induced liver injury (DILI) remains a major challenge, because DILI develops via multifactorial mechanisms. Immune and inflammatory reactions are considered important mechanisms of DILI; however, biomarkers from in vitro systems using immune cells have not been comprehensively studied. The aims of this study were (1) to identify promising biomarker genes for predicting DILI in an in vitro coculture model of peripheral blood mononuclear cells (PBMCs) with a human liver cell line, and (2) to evaluate these genes as predictors of DILI using a panel of drugs with different clinical DILI risk. Transcriptome-wide analysis of PBMCs cocultured with HepG2 or differentiated HepaRG cells that were treated with several drugs revealed an appropriate separation of DILI-positive and DILI-negative drugs, from which 12 putative biomarker genes were selected. To evaluate the predictive performance of these genes, PBMCs cocultured with HepG2 cells were exposed to 77 different drugs, and gene expression levels in PBMCs were determined. The MET proto-oncogene receptor tyrosine kinase (MET) showed the highest area under the receiver-operating characteristic curve (AUC) value of 0.81 among the 12 genes with a high sensitivity/specificity (85/66%). However, a stepwise logistic regression model using the 12 identified genes showed the highest AUC value of 0.94 with a high sensitivity/specificity (93/86%). Taken together, we established a coculture system using PBMCs and HepG2 cells and selected biomarkers that can predict DILI risk. The established model would be useful in detecting the DILI potential of compounds, in particular those that involve an immune mechanism.

In vitro assessment of farnesoid X receptor antagonism to predict drug-induced liver injury risk

Drug-induced liver injury (DILI) continues to be a major cause of drug attrition and restrictive labeling. Given the importance of farnesoid X receptor (FXR) in bile acid homeostasis, drug-related FXR antagonism may be an important mechanism of DILI. However, a comprehensive assessment of this phenomenon broadly in the context of DILI is lacking. As such, we used an orthogonal approach comprising a FXR target gene assay in primary human hepatocytes and a commercially available FXR reporter assay to investigate the potential FXR antagonistic effects of an extensive test set of 159 compounds with and without association with clinical DILI. Data were omitted from analysis based on the presence of cytotoxicity to minimize false positive assay signals and other complications in data interpretation. Based on the experimental approaches employed and corresponding data, the prevalence of FXR antagonism was relatively low across this broad DILI test set, with 16-24% prevalence based on individual assay results or combined signals in both assays. Moreover, FXR antagonism was not highly predictive for identifying clinically relevant hepatotoxicants retrospectively, where FXR antagonist classification alone had minimal to moderate predictive value as represented by positive and negative likelihood ratios of 2.24-3.84 and 0.72-0.85, respectively. The predictivity did not increase significantly when considering only compounds with high clinical exposure (maximal or efficacious plasma exposures > 1.0 μM). In contrast, modest gains in predictive value of FXR antagonism were observed considering compounds that also inhibit bile salt export pump. In addition, we have identified novel FXR antagonistic effects of well-studied hepatotoxic drugs, including bosentan, tolcapone and ritonavir. In conclusion, this work represents a comprehensive evaluation of FXR antagonism in the context of DILI, including its overall predictivity and challenges associated with detecting this phenomenon in vitro.

Drug Induced Liver Injury (DILI). Mechanisms and Medicinal Chemistry Avoidance/Mitigation Strategies

Adverse drug reactions (ADRs) are a common cause of attrition in drug discovery and development and drug-induced liver injury (DILI) is a leading cause of preclinical and clinical drug terminations. This perspective outlines many of the known DILI mechanisms and assessment methods used to evaluate and mitigate DILI risk. Literature assessments and retrospective analyses using verified DILI-associated drugs from the Liver Tox Knowledge Base (LTKB) have been used to derive the predictive value of each end point, along with combination approaches of multiple methods. In vitro assays to assess inhibition of the bile salt export pump (BSEP), mitotoxicity, reactive metabolite (RM) formation, and hepatocyte cytolethality, along with physicochemical properties and clinical dose provide useful DILI predictivity. This Perspective also highlights some of the strategies used by medicinal chemists to reduce DILI risk during the optimization of drug candidates.

Comparing Machine Learning Algorithms for Predicting Drug-Induced Liver Injury (DILI)

Drug-induced liver injury (DILI) is one the most unpredictable adverse reactions to xenobiotics in humans and the leading cause of postmarketing withdrawals of approved drugs. To date, these drugs have been collated by the FDA to form the DILIRank database, which classifies DILI severity and potential. These classifications have been used by various research groups in generating computational predictions for this type of liver injury. Recently, groups from Pfizer and AstraZeneca have collated DILI in vitro data and physicochemical properties for compounds that can be used along with data from the FDA to build machine learning models for DILI. In this study, we have used these data sets, as well as the Biopharmaceutics Drug Disposition Classification System data set, to generate Bayesian machine learning models with our in-house software, Assay Central. The performance of all machine learning models was assessed through both the internal 5-fold cross-validation metrics and prediction accuracy of an external test set of compounds with known hepatotoxicity. The best-performing Bayesian model was based on the DILI-concern category from the DILIRank database with an ROC of 0.814, a sensitivity of 0.741, a specificity of 0.755, and an accuracy of 0.746. A comparison of alternative machine learning algorithms, such as k-nearest neighbors, support vector classification, AdaBoosted decision trees, and deep learning methods, produced similar statistics to those generated with the Bayesian algorithm in Assay Central. This study demonstrates machine learning models grouped in a tool called MegaTox that can be used to predict early-stage clinical compounds, as well as recent FDA-approved drugs, to identify potential DILI.

Ensemble Models Based on QuBiLS-MAS Features and Shallow Learning for the Prediction of Drug-Induced Liver Toxicity: Improving Deep Learning and Traditional Approaches

Drug-induced liver injury (DILI) is a key safety issue in the drug discovery pipeline and a regulatory concern. Thus, many in silico tools have been proposed to improve the hepatotoxicity prediction of organic-type chemicals. Here, classifiers for the prediction of DILI were developed by using QuBiLS-MAS 0-2.5D molecular descriptors and shallow machine learning techniques, on a training set composed of 1075 molecules. The best ensemble model build, E13, was obtained with good statistical parameters for the learning series, namely, the following: accuracy = 0.840, sensibility = 0.890, specificity = 0.761, Matthew's correlation coefficient = 0.660, and area under the ROC curve = 0.904. The model was also satisfactorily evaluated with Y-scrambling test, and repeated k-fold cross-validation and repeated k-holdout validation. In addition, an exhaustive external validation was also carried out by using two test sets and five external test sets, with an average accuracy value equal to 0.854 (±0.062) and a coverage equal to 98.4% according to its applicability domain. A statistical comparison of the performance of the E13 model, with regard to results and tools (e.g., Padel DDPredictor Software, Deep Learning DILIserver, and Vslead) reported in the literature, was also performed. In general, E13 presented the best global performance in all experiments. The sum of the ranking differences procedure provided a very similar grouping pattern to that of the M-ANOVA statistical analysis, where E13 was identified as the best model for DILI predictions. A noncommercial and fully cross-platform software for the DILI prediction was also developed, which is freely available at http://tomocomd.com/apps/ptoxra. This software was used for the screening of seven data sets, containing natural products, leads, toxic materials, and FDA approved drugs, to assess the usefulness of the QSAR models in the DILI labeling of organic substances; it was found that 50-92% of the evaluated molecules are positive-DILI compounds. All in all, it can be stated that the E13 model is a relevant method for the prediction of DILI risk in humans, as it shows the best results among all of the methods analyzed.

Most Influential Physicochemical and In Vitro Assay Descriptors for Hepatotoxicity and Nephrotoxicity Prediction

Drug-induced organ injury is a major reason for drug candidate attrition in preclinical and clinical drug development. The liver, kidneys, and heart have been recognized as the most common organ systems affected in safety-related attrition or the subject of black box warnings and postmarket drug withdrawals. In silico physicochemical property calculations and in vitro assays have been utilized separately in the early stages of the drug discovery and development process to predict drug safety. In this study, we combined physicochemical properties and in vitro cytotoxicity assays including mitochondrial dysfunction to build organ-specific univariate and multivariable logistic regression models to achieve odds ratios for the prediction of clinical hepatotoxicity, nephrotoxicity, and cardiotoxicity using 215 marketed drugs. The multivariable hepatotoxic predictive model showed an odds ratio of 6.2 (95% confidence interval (CI) 1.7-22.8) or 7.5 (95% CI 3.2-17.8) for mitochondrial inhibition or drug plasma C_max >1 μM for drugs associated with liver injury, respectively. The multivariable nephrotoxicity predictive model showed an odds ratio of 5.8 (95% CI 2.0-16.9), 6.4 (95% CI 1.1-39.3), or 15.9 (95% CI 2.8-89.0) for drug plasma C_max >1 μM, mitochondrial inhibition, or hydrogen-bond-acceptor atoms >7 for drugs associated with kidney injury, respectively. Conversely, drugs with a total polar surface area ≥75 Å were 79% (odds ratio 0.21, 95% CI 0.061-0.74) less likely to be associated with kidney injury. Drugs belonging to the extended clearance classification system (ECCS) class 4, where renal secretion is the primary clearance mechanism (low permeability drugs that are bases/neutrals), were 4 (95% CI 1.8-9.5) times more likely to to be associated with kidney injury with this data set. Alternatively, ECCS class 2 drugs, where hepatic metabolism is the primary clearance (high permeability drugs that are bases/neutrals) were 77% less likely (odds ratio 0.23 95% CI 0.095-0.54) to to be associated with kidney injury. A cardiotoxicity model was poorly defined using any of these drug physicochemical attributes. Combining in silico physicochemical properties descriptors along with in vitro toxicity assays can be used to build predictive toxicity models to select small molecule therapeutics with less potential to cause liver and kidney organ toxicity.

The evolution of strategies to minimise the risk of human drug-induced liver injury (DILI) in drug discovery and development

Early identification of toxicity associated with new chemical entities (NCEs) is critical in preventing late-stage drug development attrition. Liver injury remains a leading cause of drug failures in clinical trials and post-approval withdrawals reflecting the poor translation between traditional preclinical animal models and human clinical outcomes. For this reason, preclinical strategies have evolved over recent years to incorporate more sophisticated human in vitro cell-based models with multi-parametric endpoints. This review aims to highlight the evolution of the strategies adopted to improve human hepatotoxicity prediction in drug discovery and compares/contrasts these with recent activities in our lab. The key role of human exposure and hepatic drug uptake transporters (e.g. OATPs, OAT2) is also elaborated.

Metabolism and Bioactivation: It's Time to Expect the Unexpected

Improvements in in vitro ADME tools and pharmacokinetic prediction models have helped to shift attrition rates in early clinical trials from poor exposure to drug safety concerns, such as drug-induced liver injury (DILI). Assessing a new chemical entity's potential for liver toxicity is an important consideration for the likely success of new drug candidates. Reactive intermediates produced during drug metabolism have been implicated as a cause of DILI, and their formation has been correlated to the addition of a black box warning on a drug label. In this work, we will present contemporary examples of the bioactivation of atypical structures usually regarded as benign and often used by medicinal chemists when attempting to avoid bioactivation. Medicinal chemistry strategies used to derisk bioactivation will be discussed, and an emphasis will be placed on the necessity of a multidisciplinary approach.

A Novel Lipidomics-Based Approach to Evaluating the Risk of Clinical Hepatotoxicity Potential of Drugs in 3D Human Microtissues

The importance of adsorption, distribution, metabolism, excretion, and toxicity (ADMET) analysis is expected to grow substantially due to recent failures in detecting severe toxicity issues of new chemical entities during preclinical/clinical development. Traditionally, safety risk assessment studies for humans have been conducted in animals during advanced preclinical or clinical phase of drug development. However, potential drug toxicity in humans now needs to be detected in the drug discovery process as soon as possible without reliance on animal studies. The "omics", such as genomics, proteomics, and metabolomics, have recently entered pharmaceutical research in both drug discovery and drug development, but to the best of our knowledge, no applications in high-throughput safety risk assessment have been attempted so far. This paper reports an innovative method to anticipate adverse drug effects in an early discovery phase based on lipid fingerprints using human three-dimensional microtissues. The risk of clinical hepatotoxicity potential was evaluated for a data set of 22 drugs belonging to five different therapeutic chemical classes and with various drug-induced liver injury effect. The treatment of microtissues with repeated doses of each drug allowed collecting lipid fingerprints for five time points (2, 4, 7, 9, and 11 days), and multivariate statistical analysis was applied to search for correlations with the hepatotoxic effect. The method allowed clustering of the drugs based on their hepatotoxic effect, and the observed lipid impairments for a number of drugs was confirmed by literature sources. Compared to traditional screening methods, here multiple interconnected variables (lipids) are measured simultaneously, providing a snapshot of the cellular status from the lipid perspective at a molecular level. Applied here to hepatotoxicity, the proposed workflow can be applied to several tissues, being tridimensional microtissues from various origins.

Metabolite Profiling and Reaction Phenotyping for the in Vitro Assessment of the Bioactivation of Bromfenac †

Bromfenac is a nonsteroidal anti-inflammatory drug that was approved and subsequently withdrawn from the market because of reported cases of acute hepatotoxicity. Recently, in vitro studies have revealed that bromfenac requires UDPGA and alamethicin supplemented human liver microsomes (HLM) to form a major metabolite, bromfenac indolinone (BI). Bromfenac and BI form thioether adducts through a bioactivation pathway in HLM and hepatocytes. [J. P. Driscoll et al., Chem. Res. Toxicol. 2018, 31, 223-230.] Here, Cytochrome P450 (CYP) and UDP-glucuronosyltransferase (UGT) reaction phenotyping experiments using recombinant enzymes were performed on bromfenac and BI to identify the CYP and UGT enzymes responsible for bromfenac's metabolism and bioactivation. It was determined that UGT2B7 converts bromfenac to BI, and that while CYP2C8, CYP2C9, and CYP2C19 catalyze the hydroxylation of bromfenac, only CYP2C9 forms thioether adducts when incubated with NAC or GSH as trapping agents. Although CYP2C9 was shown to form a reactive intermediate, no inhibition of CYP2C9 was observed when an IC₅₀ shift assay was performed. Reaction phenotyping experiments with BI and recombinant CYP enzymes indicated that CYPs 1A2, 2B6, 2C8, 2C9, 2C19, 2D6, and 3A4 were responsible for the formation of an aliphatic hydroxylated metabolite. An aromatic hydroxylation on the indolinone moiety was also formed by CYP1A2 and CYP3A4. The aromatic hydroxylated BI is a precursor to the quinone methide and quinone imine intermediates in the proposed bioactivation pathway. Through time-dependent inhibition (TDI) experiments, it was revealed that BI can cause an IC₅₀ shift in CYP1A2 and CYP3A4. However, BI does not inhibit the other isoforms that were also responsible for the formation of the aliphatic hydroxylation, an alternative biotransformation that does not undergo further downstream bioactivation. The results of these metabolism studies with bromfenac and BI add to our understanding of the relationship between biotransformation, reactive intermediate generation, and a potential mechanistic link to the hepatotoxicity of this compound.