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.

Human liver cell-based assays for the prediction of hepatic bile acid efflux transporter inhibition by drugs

INTRODUCTION

Drug-mediated inhibition of bile salt efflux transporters may cause liver injury. In vitro prediction of drug effects toward canalicular and/or sinusoidal efflux of bile salts from human hepatocytes is therefore a major issue, which can be addressed using liver cell-based assays.

AREA COVERED

This review, based on a thorough literature search in the scientific databases PubMed and Web of Science, provides key information about hepatic transporters implicated in bile salt efflux, the human liver cell models available for investigating functional inhibition of bile salt efflux, the different methodologies used for this purpose, and the modes of expression of the results. Applications of the assays to drugs are summarized, with special emphasis to the performance values of some assays for predicting hepatotoxicity/cholestatic effects of drugs.

EXPERT OPINION

Human liver cell-based assays for evaluating drug-mediated inhibition of bile acid efflux transporters face various limitations, such as the lack of method standardization and validation, the present poor adaptability to high throughput approaches, and some pitfalls with respect to interpretation of bile acid biliary excretion indexes. Hepatotoxicity of drugs is additionally likely multifactorial, highlighting that inhibition of hepatic bile salt efflux by drugs provides important, but not full, information about potential drug hepatotoxicity.

Mathematical models of cystic fibrosis as a systemic disease

Cystic fibrosis (CF) is widely known as a disease of the lung, even though it is in truth a systemic disease, whose symptoms typically manifest in gastrointestinal dysfunction first. CF ultimately impairs not only the pancreas and intestine but also the lungs, gonads, liver, kidneys, bones, and the cardiovascular system. It is caused by one of several mutations in the gene of the epithelial ion channel protein CFTR. Intense research and improved antimicrobial treatments during the past eight decades have steadily increased the predicted life expectancy of a person with CF (pwCF) from a few weeks to over 50 years. Moreover, several drugs ameliorating the sequelae of the disease have become available in recent years, and notable treatments of the root cause of the disease have recently generated substantial improvements in health for some but not all pwCF. Yet, numerous fundamental questions remain unanswered. Complicating CF, for instance in the lung, is the fact that the associated insufficient chloride secretion typically perturbs the electrochemical balance across epithelia and, in the airways, leads to the accumulation of thick, viscous mucus and mucus plaques that cannot be cleared effectively and provide a rich breeding ground for a spectrum of bacterial and fungal communities. The subsequent infections often become chronic and respond poorly to antibiotic treatments, with outcomes sometimes only weakly correlated with the drug susceptibility of the target pathogen. Furthermore, in contrast to rapidly resolved acute infections with a single target pathogen, chronic infections commonly involve multi-species bacterial communities, called "infection microbiomes," that develop their own ecological and evolutionary dynamics. It is presently impossible to devise mathematical models of CF in its entirety, but it is feasible to design models for many of the distinct drivers of the disease. Building upon these growing yet isolated modeling efforts, we discuss in the following the feasibility of a multi-scale modeling framework, known as template-and-anchor modeling, that allows the gradual integration of refined sub-models with different granularity. The article first reviews the most important biomedical aspects of CF and subsequently describes mathematical modeling approaches that already exist or have the potential to deepen our understanding of the multitude aspects of the disease and their interrelationships. The conceptual ideas behind the approaches proposed here do not only pertain to CF but are translatable to other systemic diseases. This article is categorized under: Congenital Diseases > Computational Models.

Review of Predicting Synergistic Drug Combinations

The prediction of drug combinations is of great clinical significance. In many diseases, such as high blood pressure, diabetes, and stomach ulcers, the simultaneous use of two or more drugs has shown clear efficacy. It has greatly reduced the progression of drug resistance. This review presents the latest applications of methods for predicting the effects of drug combinations and the bioactivity databases commonly used in drug combination prediction. These studies have played a significant role in developing precision therapy. We first describe the concept of synergy. we study various publicly available databases for drug combination prediction tasks. Next, we introduce five algorithms applied to drug combinatorial prediction, which include traditional machine learning methods, deep learning methods, mathematical methods, systems biology methods and search algorithms. In the end, we sum up the difficulties encountered in prediction models.

Measuring Apoptosis and Necrosis in Cholestatic Liver Injury

Cholestasis can be induced by obstruction of bile ducts or intrahepatic toxicity of drugs and chemicals. However, the mode of cell death during cholestasis, i.e., apoptosis or necrosis, has been controversial. There are fundamental reasons for the controversies, both of which are discussed here, namely the design of experiments and the use of parameters with limited specificity for a certain mode of cell death. Based on the assumption that cholestatic liver injury is caused by accumulation of bile acids, rodent (mainly rat) hepatocytes have been exposed to hydrophobic, glycine-conjugated bile acids, which resulted in apoptotic cell death. The problems with this experimental design are that in rodents bile acids are predominantly taurine conjugated and rodent hepatocytes are never exposed to these levels of glycine-conjugated bile acids. In contrast, taurine-conjugated bile acids trigger inflammatory gene activation in rodent hepatocytes and a necro-inflammatory injury in vivo. On the other hand, human hepatocytes are more resistant to glycine-conjugated bile acids and die by necrosis when exposed to high biliary levels of these bile acids. In this chapter, we describe multiple assays including the caspase activity assay, which is specific for apoptosis, and the general cell death assays alanine aminotransferase or lactate dehydrogenase activities in cell culture medium or plasma. An increase in these enzyme activities without caspase activity indicates necrotic cell death. Thus, both the experimental design and the selection of cell death parameters are critical for the relevance of the experiments for the human pathophysiology.