Toxicogenomics of drug induced liver injury - from mechanistic understanding to early prediction

Evaluating the synergistic use of advanced liver models and AI for the prediction of drug-induced liver injury

INTRODUCTION

Drug-induced liver injury (DILI) is a leading cause of acute liver failure. Hepatotoxicity typically occurs only in a subset of individuals after prolonged exposure and constitutes a major risk factor for the termination of drug development projects.

AREAS COVERED

We provide an overview of available human liver models for DILI research and discuss how they have been used to aid in early risk assessments and to mitigate the risk of project closures due to DILI in clinical stages. We summarize the different data that can be provided by such models and illustrate how these diverse data types can be interfaced with machine learning strategies to improve predictions of liver safety liabilities.

EXPERT OPINION

Advanced human liver models closely mimic human liver phenotypes and functions for many weeks, allowing for the recapitulation of hepatotoxicity events in vitro. Integration of the biochemical, histological, and toxicogenomic output data from these models with physicochemical compound properties using different machine learning architectures holds promise to enhance preclinical DILI predictions. However, to realize this aim, it is important to benchmark the available liver models on test sets of DILI positive and negative compounds and to carefully annotate and share the resulting data.

OCT1 (SLC22A1) transporter kinetics and regulation in primary human hepatocyte 3D spheroids

3D spheroids of primary human hepatocytes (3D PHH) retain a differentiated phenotype with largely conserved metabolic function and proteomic fingerprint over weeks in culture. As a result, 3D PHH are gaining importance as a model for mechanistic liver homeostasis studies and in vitro to in vivo extrapolation (IVIVE) in drug discovery. However, the kinetics and regulation of drug transporters have not yet been assessed in 3D PHH. Here, we used organic cation transporter 1 (OCT1/SLC22A1) as a model to study both transport kinetics and the long-term regulation of transporter activity via relevant signalling pathways. The kinetics of the OCT1 transporter was studied using the fluorescent model substrate 4-(4-(dimethylamino)styryl)-N-methylpyridinium (ASP+) and known OCT1 inhibitors in individual 3D PHH. For long-term studies, 3D PHH were treated with xenobiotics for seven days, after which protein expression and OCT1 function were assessed. Global proteomic analysis was used to track hepatic phenotypes as well as prototypical changes in other regulated proteins, such as P-glycoprotein and Cytochrome P450 3A4. ASP+ kinetics indicated a fully functional OCT1 transporter with a Km value of 14 ± 4.0µM as the mean from three donors. Co-incubation with known OCT1 inhibitors decreased the uptake of ASP+ in the 3D PHH spheroids by 35-52%. The long-term exposure studies showed that OCT1 is relatively stable upon activation of nuclear receptor signalling or exposure to compounds that could induce inflammation, steatosis or liver injury. Our results demonstrate that 3D PHH spheroids express physiologically relevant levels of fully active OCT1 and that its transporter kinetics can be accurately studied in the 3D PHH configuration. We also confirm that OCT1 remains stable and functional during the activation of key metabolic pathways that alter the expression and function of other drug transporters and drug-metabolizing enzymes. These results will expand the range of studies that can be performed using 3D PHH.

Compound Absorption in Polymer Devices Impairs the Translatability of Preclinical Safety Assessments

Organotypic and microphysiological systems (MPS) that can emulate the molecular phenotype and function of human tissues, such as liver, are increasingly used in preclinical drug development. However, despite their improved predictivity, drug development success rates have remained low with most compounds failing in clinical phases despite promising preclinical data. Here, it is tested whether absorption of small molecules to polymers commonly used for MPS fabrication can impact preclinical pharmacological and toxicological assessments and contribute to the high clinical failure rates. To this end, identical devices are fabricated from eight different MPS polymers and absorption of prototypic compounds with different physicochemical properties are analyzed. It is found that overall absorption is primarily driven by compound hydrophobicity and the number of rotatable bonds. However, absorption can differ by >1000-fold between polymers with polydimethyl siloxane (PDMS) being most absorptive, whereas polytetrafluoroethylene (PTFE) and thiol-ene epoxy (TEE) absorbed the least. Strikingly, organotypic primary human liver cultures successfully flagged hydrophobic hepatotoxins in lowly absorbing TEE devices at therapeutically relevant concentrations, whereas isogenic cultures in PDMS devices are resistant, resulting in false negative safety signals. Combined, these results can guide the selection of MPS materials and facilitate the development of preclinical assays with improved translatability.

High-Throughput Transcriptomics Differentiates Toxic versus Non-Toxic Chemical Exposures Using a Rat Liver Model

To address the challenge of limited throughput with traditional toxicity testing, a newly developed high-throughput transcriptomics (HTT) platform, together with a 5-day in vivo rat model, offers an alternative approach to estimate chemical exposures and provide reasonable estimates of toxicological endpoints. This study contains an HTT analysis of 18 environmental chemicals with known liver toxicity. They were evaluated using male Sprague Dawley rats exposed to various concentrations daily for five consecutive days via oral gavage, with data collected on the sixth day. Here, we further explored the 5-day rat model to identify potential gene signatures that can differentiate between toxic and non-toxic liver responses and provide us with a potential histopathological endpoint of chemical exposure. We identified a distinct gene expression pattern that differentiated non-hepatotoxic compounds from hepatotoxic compounds in a dose-dependent manner, and an analysis of the significantly altered common genes indicated that toxic chemicals predominantly upregulated most of the genes and several pathways in amino acid and lipid metabolism. Finally, our liver injury module analysis revealed that several liver-toxic compounds showed similarities in the key injury phenotypes of cellular inflammation and proliferation, indicating potential molecular initiating processes that may lead to a specific end-stage liver disease.

Assessment of Drug-Induced Liver Injury through Cell Morphology and Gene Expression Analysis

Drug-induced liver injury (DILI) is a significant concern in drug development, often leading to drug withdrawal. Although many studies aim to identify biomarkers and gene/pathway signatures related to liver toxicity and aim to predict DILI compounds, this remains a challenge in drug discovery. With a strong development of high-content screening/imaging (HCS/HCI) for phenotypic screening, we explored the morphological cell perturbations induced by DILI compounds. In the first step, cell morphological signatures were associated with two datasets of DILI chemicals (DILIRank and eTox). The mechanisms of action were then analyzed for chemicals having transcriptomics data and sharing similar morphological perturbations. Signaling pathways associated with liver toxicity (cell cycle, cell growth, apoptosis, ...) were then captured, and a hypothetical relation between cell morphological perturbations and gene deregulation was illustrated within our analysis. Finally, using the cell morphological signatures, machine learning approaches were developed to predict chemicals with a potential risk of DILI. Some models showed relevant performance with validation set balanced accuracies between 0.645 and 0.739. Overall, our findings demonstrate the utility of combining HCI with transcriptomics data to identify the morphological and gene expression signatures related to DILI chemicals. Moreover, our protocol could be extended to other toxicity end points, offering a promising avenue for comprehensive toxicity assessment in drug discovery.

In silico modeling-based new alternative methods to predict drug and herb-induced liver injury: A review

New approach methods (NAMs) have been developed to predict a wide range of toxicities through innovative technologies. Liver injury is one of the most extensively studied endpoints due to its severity and frequency, occurring among populations that consume drugs or dietary supplements. In this review, we focus on recent developments of in silico modeling for liver injury prediction using deep learning and in vitro data based on adverse outcome pathways (AOPs). Despite these models being mainly developed using datasets generated from drug-like molecules, they were also applied to the prediction of hepatotoxicity caused by herbal products. As deep learning has achieved great success in many different fields, advanced machine learning algorithms have been actively applied to improve the accuracy of in silico models. Additionally, the development of liver AOPs, combined with big data in toxicology, has been valuable in developing in silico models with enhanced predictive performance and interpretability. Specifically, one approach involves developing structure-based models for predicting molecular initiating events of liver AOPs, while others use in vitro data with structure information as model inputs for making predictions. Even though liver injury remains a difficult endpoint to predict, advancements in machine learning algorithms and the expansion of in vitro databases with relevant biological knowledge have made a huge impact on improving in silico modeling for drug-induced liver injury prediction.

Artificial Intelligence in Drug Toxicity Prediction: Recent Advances, Challenges, and Future Perspectives

Toxicity prediction is a critical step in the drug discovery process that helps identify and prioritize compounds with the greatest potential for safe and effective use in humans, while also reducing the risk of costly late-stage failures. It is estimated that over 30% of drug candidates are discarded owing to toxicity. Recently, artificial intelligence (AI) has been used to improve drug toxicity prediction as it provides more accurate and efficient methods for identifying the potentially toxic effects of new compounds before they are tested in human clinical trials, thus saving time and money. In this review, we present an overview of recent advances in AI-based drug toxicity prediction, including the use of various machine learning algorithms and deep learning architectures, of six major toxicity properties and Tox21 assay end points. Additionally, we provide a list of public data sources and useful toxicity prediction tools for the research community and highlight the challenges that must be addressed to enhance model performance. Finally, we discuss future perspectives for AI-based drug toxicity prediction. This review can aid researchers in understanding toxicity prediction and pave the way for new methods of drug discovery.

The native liver as inspiration to create superior in vitro hepatic models

Drug induced liver injury (DILI) is one of the major reasons of drug withdrawal during the different phases of drug development. The later in the drug development a drug is discovered to be toxic, the higher the economical as well as the ethical impact will be. In vitro models for early detection of drug liver toxicity are under constant development, however to date a superior model of the liver is still lacking. Ideally, a highly reliable model should be established to maintain the different hepatic cell functionalities to the greatest extent possible, during a period of time long enough to allow for tracking of the toxicity of compounds. In the case of DILI, toxicity can appear even after months of exposure. To reach this goal, an in vitro model should be developed that mimics the in vivo liver environment, function and response to external stimuli. The different approaches for the development of liver models currently used in the field of tissue engineering will be described in this review. Combining different technologies, leading to optimal materials, cells and 3D-constructs will ultimately lead to an ideal superior model that fully recapitulates the liver.

Human Liver Organoid Models for Assessment of Drug Toxicity at the Preclinical Stage

The hepatotoxicity of drugs is one of the leading causes of drug withdrawal from the pharmaceutical market and high drug attrition rates. Currently, the commonly used hepatocyte models include conventional hepatic cell lines and animal models, which cannot mimic human drug-induced liver injury (DILI) due to poorly defined dose-response relationships and/or lack of human-specific mechanisms of toxicity. In comparison to 2D culture systems from different cell sources such as primary human hepatocytes and hepatomas, 3D organoids derived from an inducible pluripotent stem cell (iPSC) or adult stem cells are promising accurate models to mimic organ behavior with a higher level of complexity and functionality owing to their ability to self-renewal. Meanwhile, the heterogeneous cell composition of the organoids enables metabolic and functional zonation of hepatic lobule important in drug detoxification and has the ability to mimic idiosyncratic DILI as well. Organoids having higher drug-metabolizing enzyme capacities can culture long-term and be combined with microfluidic-based technologies such as organ-on-chips for a more precise representation of human susceptibility to drug response in a high-throughput manner. However, there are numerous limitations to be considered about this technology, such as enough maturation, differences between protocols and high cost. Herein, we first reviewed the current preclinical DILI assessment tools and looked at the organoid technology with respect to in vitro detoxification capacities. Then we discussed the clinically applicable DILI assessment markers and the importance of liver zonation in the next generation organoid- based DILI models.

Omics technologies to understand drug toxicity mechanisms

Drug side effects are an important study subject in pharmacology. Recent omics technologies provide a range of omics data and help to understand the biological mechanisms involved in drug effects. These modern technologies provide significant support to all biological disciplines, including drug toxicology. In this review, we provide an overview the use of omics applications to understand drug side effects at the molecular level. We discuss by available omics technologies, their possible uses, as well as their advantages and limitations.

Organotypic and Microphysiological Human Tissue Models for Drug Discovery and Development-Current State-of-the-Art and Future Perspectives

The number of successful drug development projects has been stagnant for decades despite major breakthroughs in chemistry, molecular biology, and genetics. Unreliable target identification and poor translatability of preclinical models have been identified as major causes of failure. To improve predictions of clinical efficacy and safety, interest has shifted to three-dimensional culture methods in which human cells can retain many physiologically and functionally relevant phenotypes for extended periods of time. Here, we review the state of the art of available organotypic culture techniques and critically review emerging models of human tissues with key importance for pharmacokinetics, pharmacodynamics, and toxicity. In addition, developments in bioprinting and microfluidic multiorgan cultures to emulate systemic drug disposition are summarized. We close by highlighting important trends regarding the fabrication of organotypic culture platforms and the choice of platform material to limit drug absorption and polymer leaching while supporting the phenotypic maintenance of cultured cells and allowing for scalable device fabrication. We conclude that organotypic and microphysiological human tissue models constitute promising systems to promote drug discovery and development by facilitating drug target identification and improving the preclinical evaluation of drug toxicity and pharmacokinetics. There is, however, a critical need for further validation, benchmarking, and consolidation efforts ideally conducted in intersectoral multicenter settings to accelerate acceptance of these novel models as reliable tools for translational pharmacology and toxicology. SIGNIFICANCE STATEMENT: Organotypic and microphysiological culture of human cells has emerged as a promising tool for preclinical drug discovery and development that might be able to narrow the translation gap. This review discusses recent technological and methodological advancements and the use of these systems for hit discovery and the evaluation of toxicity, clearance, and absorption of lead compounds.

3D human liver spheroids for translational pharmacology and toxicology

Drug development is a failure-prone endeavour, and more than 85% of drugs fail during clinical development, showcasing that current preclinical systems for compound selection are clearly inadequate. Liver toxicity remains a major reason for safety failures. Furthermore, all efforts to develop pharmacological therapies for a variety of chronic liver diseases, such as non-alcoholic steatohepatitis (NASH) and fibrosis, remain unsuccessful. Considering the time and expense of clinical trials, as well as the substantial burden on patients, new strategies are thus of paramount importance to increase clinical success rates. To this end, human liver spheroids are becoming increasingly utilized as they allow to preserve patient-specific phenotypes and functions for multiple weeks in culture. We here review the recent application of such systems for i) predictive and mechanistic analyses of drug hepatotoxicity, ii) the evaluation of hepatic disposition and metabolite formation of low clearance drugs and iii) the development of drugs for metabolic and infectious liver diseases, including NASH, fibrosis, malaria and viral hepatitis. We envision that with increasing dissemination, liver spheroids might become the new gold standard for such applications in translational pharmacology and toxicology.

Idiosyncratic Drug Induced Liver Injury, Cytochrome P450, Metabolic Risk Factors and Lipophilicity: Highlights and Controversies

Progress in understanding the mechanisms of the idiosyncratic drug induced liver injury (iDILI) was highlighted in a scientometric investigation on the knowledge mapping of iDILI throughout the world, but uncertainty remained on metabolic risk factors of iDILI, the focus of the present review article. For the first time, a quantitative analysis of 3312 cases of iDILI assessed for causality with RUCAM (Roussel Uclaf Causality Assessment Method) showed that most drugs (61.1%) were metabolized by cytochrome P450 (CYP) isoforms: 49.6% by CYP 3A4/5, 24.6% by CYP 2C9, 13.2% by CYP 2E1, 7.3% by CYP 2C19, 3.5% by CYP 1A2 and 1.8% by CYP 2D6. Other studies showed high OR (odds ratio) for drugs metabolized by unspecified CYPs but the iDILI cases were not assessed for causality with RUCAM, a major shortcoming. In addition to critical comments on methodological flaws, several risk factors of iDILI were identified such as high but yet recommended daily drug doses, actual daily drug doses taken by the patients, hepatic drug metabolism and drug lipophilicity. These risk factors are subject to controversies by many experts seen critically also by others who outlined that none of these medication characteristics is able to predict iDILI with high confidence, leading to the statement of an outstanding caveat. It was also argued that all previous studies lacked comprehensive data because the number of examined drugs was relatively small as compared to the number of approved new molecular entities or currently used oral prescription drugs. In conclusion, trends are evident that some metabolic parameters are likely risk factors of iDILI but strong evidence can only be achieved when methodological issues will be successfully met.