Drug-induced liver injury (DILI) is a critical safety issue in drug development, characterized by its idiosyncratic nature, complex mechanisms, and poor predictability in standard preclinical models. High-dimensional omics strategies, particularly toxicogenomics, have attracted increased interest in addressing the complexity of hepatotoxicity, especially in the context of emerging artificial intelligence (AI) technologies. This review traces the evolution of AI and machine learning (ML) within DILI-related omics research, highlighting toxicogenomics as a primary driver of advancement in this field. We first explore the early studies in computational toxicogenomics, which primarily focused on exploratory approaches, utilizing clustering, time-series, co-expression, and basic pathway analyses to identify molecular signatures indicative of nascent liver injury. We then examine how the adoption of supervised machine learning enabled robust predictive modeling, facilitating systematic feature selection, signature refinement, and rigorous validation. More recently, the field has been further transformed by deep learning, biologically informed network architectures, and generative artificial intelligence. Across these methodological eras, AI has enhanced mechanistic interpretation by identifying biologically relevant signatures, integrating multimodal evidence, and strengthening evidence for established DILI mechanisms, including oxidative stress, mitochondrial dysfunction, altered xenobiotic metabolism, inflammation, and cell death. Ultimately, the synergy of AI and DILI toxicogenomics has transitioned the discipline from descriptive profiling toward mechanism-driven predictive toxicology.

Drug induced liver injury (DILI) can have severe negative outcomes on patient survival of hepatotoxic adverse drug reactions (ADRs), attrition of new drug programs, and post-market withdrawal of approved drugs. Cytochrome P450 (CYP) enzyme metabolism has been associated with increased dose-dependent DILI risk. Additionally, studies have indicated other metabolizing enzymes, like UDP-Glucuronosyltransferase (UGT) enzymes, can induce DILI via the toxicity of reactive metabolites. Although metabolism of some noncytochrome P450 (non-CYP) enzymes have been associated with DILI induction, their relationship with DILI development has not been systematically characterized.The in-vitro drug-enzyme interactions of 317 drugs with 42 non-CYP enzymes acting as substrates, inducers, and/or inhibitors were retrieved from historical regulatory documents using PharmaPendium. Bioactivity data was obtained from Reaxys. PharmaPendium is a translational tool that makes regulatory grade documents from the EMA and FDA fully text searchable for pharmacokinetic, metabolism, efficacy, and drug safety data. Reaxys contains chemistry data and bioactivity data from journal articles and patents. The drugs selected for the data set and their DILI concern classification were obtained by the Liver Toxicity Knowledge Base (LTKB) data set developed by the US Food and Drug Administration (FDA)’s National Center for Toxicological Research (NCTR). The relationship between DILI concern (high versus low) and drug-enzyme interactions with non-CYP enzymes were examined binomially. Multivariate linear regression was used to assess associations between DILI concern and enzymatic relationship.Examining the in vitro drug-enzyme interactions for potential DILI concern revealed that non-CYP enzymes are highly associated. Specifically, 31 drugs in the 317 drug data set interacted with enzymes in the UGT1 family with 21 of 31 (67%) having high DILI concern. Analysis also showed that a drug that acts as a dual UGT substrate-inhibitor is associated with high DILI concern as compared to a drug that is a pure UGT substrate.Drug interactions with UGT enzymes may independently predict DILI, and their combined use with the rule-of-two model further improves overall predictive performance. These findings could expand the currently available tools for assessing the potential for DILI in humans which can improve translational assessment and drug development.

The FDA's 'Roadmap to Reducing Animal Testing in Preclinical Safety Studies' supports the regulatory advancement of new approach methodologies (NAMs). Focusing on overlapping drugs, this review compared the performance of in vitro NAMs, animal studies, and microphysiological systems (MPS) in predicting drug-induced liver injury (DILI). We observed considerable variability among in vitro NAMs and their potential advantages over animal models. Moreover, limited MPS data hindered meaningful comparison. To enable objective and systematic assessment of NAMs, we propose DILIference, a curated reference drug list compiled from the literature to guide the development and benchmarking of DILI-predictive NAMs.

Hepatotoxicity can lead to the discontinuation of approved or investigational drugs. The evaluation of the potential hepatoxicity of drugs in development is challenging because current models assessing this adverse effect are not always predictive of the outcome in human beings. Cell lines are routinely used for early hepatotoxicity screening, but to improve the detection of potential hepatotoxicity, in vitro models that better reflect liver morphology and function are needed. One such promising model is human liver microtissues. These are spheroids made of primary human parenchymal and nonparenchymal liver cells, which are amenable to high throughput screening. To test the predictivity of this model, the cytotoxicity of 152 FDA (US Food & Drug Administration)-approved small molecule drugs was measured as per changes in ATP content in human liver microtissues incubated in 384-well microplates. The results were analyzed with respect to drug label information, drug-induced liver injury (DILI) concern class, and drug class. The threshold IC50ATP-to-Cmax ratio of 176 was used to discriminate between safe and hepatotoxic drugs. "vMost-DILI-concern" drugs were detected with a sensitivity of 72% and a specificity of 89%, and "vMost-DILI-concern" drugs affecting the nervous system were detected with a sensitivity of 92% and a specificity of 91%. The robustness and relevance of this evaluation were assessed using a 5-fold cross-validation. The good predictivity, together with the in vivo-like morphology of the liver microtissues and scalability to a 384-well microplate, makes this method a promising and practical in vitro alternative to 2D cell line cultures for the early hepatotoxicity screening of drug candidates.

Drug-induced liver injury (DILI) is of great concern in drug development and public health. DILIrank 1.0, a widely used public dataset that ranks FDA-approved drugs by their potential to cause DILI, has significantly enabled the development of new methods for improved DILI assessment. Here, we introduce DILIrank 2.0, an essential update of DILIrank 1.0, to capture new non-biologics drug and liver-related adverse event data generated since its inception 15 years ago. Using DILIrank 2.0, we also observe changes in the DILI profiles of approved drugs following the introduction of different FDA regulatory programs. DILIrank 2.0 is an up-to-date resource that offers opportunities for supporting DILI safety assessment, predictive model development, and evaluation of new approach methods (NAMs).

Background and Aims: Herbal and dietary supplements (HDS) have been associated with liver injury and severe liver disease, including acute liver failure (ALF), collectively referred to as HDS-induced ALF (HDS-ALF). With the global increase in HDS usage, reports of HDS-ALF cases have also risen. Methods: Genetic analysis was conducted using whole exome sequencing data from 23 HDS-ALF cases, with population controls sourced from the 1000 Genomes Project. Single-nucleotide polymorphisms (SNPs) with chi-square test P values less than 5 × 10−8 and stable allele frequencies across different ethnicities and healthy controls were considered statistically significant. Variants were also analyzed based on supplement type (herbal vs dietary) and sex (female vs male). Results: Thirty-three SNPs were statistically significant (P < 5 × 10−8), predominantly located in human leukocyte antigen (HLA) genes (HLA-A, HLA-B, HLA-DQA1, and HLA-DRB1) and estrogen-related receptor alpha (ESRRA). The top SNPs, including rs17879990 (HLA-A), rs1131500 (HLA-B), and rs1161801407 (HLA-DRB1), showed strong associations. In subgroup analyses, 4 SNPs were significant in the herbal product group and 3 in the dietary supplement group. Gender-based analyses revealed 15 significant SNPs in females and 8 in males, particularly 7 SNPs within HLA-B and HLA-DQA1, appeared female-specific. Conclusion: This study identifies specific HLA and ESRRA gene variants as genetic risk factors for HDS-ALF, with significant associations observed across HDS exposures. The findings suggest potential cross-gender risks, particularly for certain HLA and ESRRA SNPs, while highlighting female-specific vulnerabilities within HLA-B and HLA-DQA1 variants.

BACKGROUND AND AIMS:Acute liver failure (ALF) is a serious condition, typically in individuals without prior liver disease. Drug-induced ALF (DIALF) constitutes a major portion of ALF cases. Our research aimed to identify potential genetic predispositions to DIALF. METHODS:We analysed the potential genetic variants associated with DIALF using the whole exome sequencing data from 75 cases, including 40 non-Finnish European cases in the pilot study. Chi-square tests were performed for case-control analysis against the 1000 genomes project as the control. A replication study of 44 DIALF cases that included 24 non-Finnish Europeans was conducted to validate candidate variants. The association between clinical phenotype and genotypes was analysed using one-way analysis of variance. RESULTS:Eight variants (rs561037, rs561042, rs608339, rs655260, rs1142886, rs1142888, rs1142889 and rs1142890) in the guanylate binding protein 4 (GBP4) were significantly associated with DIALF in non-Finnish Europeans in the pilot study and confirmed in the replication study. Rs561037 and rs561042 were highly significant with the lowest allele frequencies in both pilot and replication studies. An association was also found between these variants and milder clinical outcomes, indicated by lower peak levels of ALT, AST and higher Karnofsky performance scores. CONCLUSION:Our study identified eight GBP4 missense variants linked to a lower risk of DIALF in the non-Finnish European population. The GBP4 protein, activated by interferon-gamma, plays a critical role in innate immunity. These findings suggest that GBP4 variants might influence immune and inflammatory responses in DIALF, though further studies are needed to elucidate the underlying mechanisms.

Drug-induced liver injury (DILI) is a complex and potentially severe adverse reaction to drugs, herbal products or dietary supplements. DILI can mimic other liver diseases clinical presentation, and currently lacks specific diagnostic biomarkers, which hinders its diagnosis. In some cases, DILI may progress to acute liver failure. Given its public health risk, novel methodologies to enhance the understanding of DILI are crucial. Recently, the increasing availability of larger datasets has highlighted artificial intelligence (AI) as a powerful tool to construct complex models. In this review, we summarise the evidence about the use of AI in DILI research, explaining fundamental AI concepts and its subfields. We present findings from AI-based approaches in DILI investigations for risk stratification, prognostic evaluation and causality assessment and discuss the adoption of natural language processing (NLP) and large language models (LLM) in the clinical setting. Finally, we explore future perspectives and challenges in utilising AI for DILI research.

BACKGROUND:Drug-induced liver injury (DILI) is a complex liver pathology modulated by multiple factors, most of which remain unknown. Previous studies have suggested that concomitant medications and patient characteristics play an important role as modulators of this disease. This study aimed to determine the most relevant concomitant medications and patient characteristics that influence the severity of idiosyncratic DILI. METHODS:Two clinical databases, discovery and validation, were analyzed to evaluate host and drug properties. Predictive algorithms, elastic net regression model and logistic regression model, were implemented using R, both achieving ROC AUC > 0.7. RESULTS:The findings revealed the existence of significant relationship between DILI severity and multiple factors. These factors included: hepatocellular injury, hydrophobic drugs with logP > 3 (octanol-water partition coefficient), and the use of concomitant medications containing halogen compounds or heterorings when taken together with culprit drugs with significant hepatic metabolism. CONCLUSIONS:These findings offer valuable insights into predicting the severity of DILI. By identifying the key factors that influence the severity of DILI, it would be possible for healthcare providers to predict the severity of damage in a patient with DILI. This enables early interventions in cases of DILI, thus subsequently reducing its negative effects.

INTRODUCTION:Drug-induced liver injury (DILI) poses a significant challenge to drug development and human healthcare. The complex mechanisms underlying DILI make it challenging to accurately predict its occurrence, often leading to substantial financial losses from failed drug development projects and drug withdrawals. Growing evidence suggests that drug-metabolizing enzymes and transporters (DMETs) play a critical role in the development of DILI. AREAS COVERED:In this review, we explore findings about the contributions of DMETs to DILI, with a focus on the studies examining genetic polymorphisms and their interactions with drugs. Additionally, we highlight the roles of DMETs in the development of predictive models for assessing DILI potential and in uncovering the mechanisms involved in DILI. EXPERT OPINION:As new approach methods (NAMs) for assessing and predicting drug toxicity gain more prominence, it is imperative to better understand the adverse outcome pathways (AOPs) that underpin these methods. DMETs largely play a pivotal role in the molecular initiating events of DILI-related AOPs. Further research is needed to characterize DILI-related AOP networks and enhance the predictive performance of NAMs for assessing DILI risk.

Drug-induced liver injury (DILI) is a major drug safety concern for clinicians, drug developers, and regulators. New tools and approaches to better predict DILI risk in humans, especially at the early stages of drug development, are still urgently needed. The development of predictive models requires a drug reference list with an accurate annotation of DILI risk in humans. Here, we summarized previously developed schema for annotating a drug's potential to cause DILI in humans. This effort utilized a drug labeling–based approach by weighing the causality assessment in literature. A large dataset, namely DILIrank, was published in which 1036 drugs were classified into three verified DILI groups (i.e., Most-, Less-, or No-DILI-concern) or a group of "ambiguous DILI" drugs without causality verification. This large dataset has been widely used to support the development of quantitative structure–activity relationships and other predictive models.

Drug-induced liver injury (DILI) is an adverse drug reaction that may be caused by overdose or occur as an unpredictable idiosyncratic event, the mechanism of which remains loosely defined. DILI is also one of the most common causes of drug withdrawal from the market. To derisk DILI at the early stage of drug discovery, many efforts have been invested to develop DILI predictive models, including quantitative structure–activity relationship (QSAR) models. In this chapter, we focus on how machine learning and deep learning can assist QSAR modeling, as well as the recent development of DILI predictive models in the past 3 years.

On-treatment excursions of liver laboratory test values in clinical trials involving subjects with underlying liver disease are relevant for the efficacy and safety assessment of drug products and biologics. Existing visualization and analysis tools do not efficiently provide an integrated view of these excursions when baseline liver tests are abnormal. The aim of this study was to develop a composite plot that enables visualization of on-treatment changes in liver test results both as multiples of the upper limit of normal defined by each laboratory’s reference population (×ULN) and multiples of the subjects’ baseline (×BLN) values. The composite plot approach combines biochemical evaluation for drug-induced severe hepatotoxicity (eDISH) plots sequentially applied to subjects’ baseline and peak on-treatment liver test results normalized by ULN and integrates them into a four-panel shift plot of peak on-treatment values normalized by BLN. The composite plot enabled efficient assessment of improvement in liver test values during treatment compared with pretreatment in subjects treated with the investigational drug (or the natural history of placebo-treated subjects) and identified outlier subjects for potential drug-induced liver injury. For studies in subjects with abnormal baseline values, the composite plot has potential application in the assessment of beneficial and concerning on-treatment modifications in liver test values in reference to the individual subject’s baseline and population threshold values.

DILI frequently contributes to the attrition of new drug candidates and is a common cause for the withdrawal of approved drugs from the market. Although some noncytochrome P450 (non-CYP) metabolism enzymes have been implicated in DILI development, their association with DILI outcomes has not been systematically evaluated. In this study, we analyzed a large data set comprising 317 drugs and their interactions in vitro with 42 non-CYP enzymes as substrates, inducers, and/or inhibitors retrieved from historical regulatory documents. We examined how these in vitro drug-enzyme interactions are correlated with the drugs’ potential for DILI concern, as classified in the Liver Toxicity Knowledge Base database. Our study revealed that drugs that inhibit non-CYP enzymes are significantly associated with high DILI concern. Particularly, interaction with UDP-glucuronosyltransferases (UGT) enzymes is an important predictor of DILI outcomes. Further analysis indicated that only pure UGT inhibitors and dual substrate inhibitors, but not pure UGT substrates, are significantly associated with high DILI concern. Notably, drug interactions with UGT enzymes may independently predict DILI, and their combined use with the rule-of-two model further improves overall predictive performance. These findings could expand the currently available tools for assessing the potential for DILI in humans.

There are a huge number of chemicals that have been introduced at the market and in the environment. Safety evaluation of chemical-containing products at the market and risk assessment of chemicals in the environment are necessary for protecting public health. However, experimentally evaluating toxicity potential for all chemicals is extremely difficult. Therefore, computational methods such as quantitative structure-activity relationship (QSAR) modeling have been used to facilitate safety evaluation and risk assessment. Machine learning algorithms are necessary for QSAR modeling. Similar to random forest, Decision forest is a consensus machine learning algorithm using decision trees as member models. It uses less but diverse decision trees and needs only a small number of independent variables. Thus, it is suitable for QSAR modeling. This chapter introduces the algorithm and features of decision forest. We also demonstrate its usefulness in applications of QSAR modeling by reviewing some cases of applications to both two-class and multiclass QSAR modeling.

IntroductionRegulatory agencies generate a vast amount of textual data in the review process. For example, drug labeling serves as a valuable resource for regulatory agencies, such as U.S. Food and Drug Administration (FDA) and Europe Medical Agency (EMA), to communicate drug safety and effectiveness information to healthcare professionals and patients. Drug labeling also serves as a resource for pharmacovigilance and drug safety research. Automated text classification would significantly improve the analysis of drug labeling documents and conserve reviewer resources.MethodsWe utilized artificial intelligence in this study to classify drug-induced liver injury (DILI)-related content from drug labeling documents based on FDA’s DILIrank dataset. We employed text mining and XGBoost models and utilized the Preferred Terms of Medical queries for adverse event standards to simplify the elimination of common words and phrases while retaining medical standard terms for FDA and EMA drug label datasets. Then, we constructed a document term matrix using weights computed by Term Frequency-Inverse Document Frequency (TF-IDF) for each included word/term/token.ResultsThe automatic text classification model exhibited robust performance in predicting DILI, achieving cross-validation AUC scores exceeding 0.90 for both drug labels from FDA and EMA and literature abstracts from the Critical Assessment of Massive Data Analysis (CAMDA).DiscussionMoreover, the text mining and XGBoost functions demonstrated in this study can be applied to other text processing and classification tasks.

Post-market medical device-associated failures and patient problems are reported in Medical Device Reports (MDRs) to the US Food and Drug Administration. Reports are accessible through Manufacturer and User Facility Device Experience (MAUDE), a database including both required and voluntary submissions. We present an overview of >10 million MDRs received from 2011 to 2021. Approximately 92% of reporting issues represent medical device physical or functional failures, categorized from 1704 codes related to medical device integrity or function. ∼8% were coded adverse events (AEs). Patient outcomes are reported via 998 patient codes in 19 medical specialties (cardiovascular, orthopedic, etc.). ∼40% of patient reports indicated “no health consequences”; however, a small number of devices had consistently high AE reports. While overall reports did not exhibit a sex-based dichotomy, ∼9% of the reported AEs occurred more frequently in females, many of which were related to immune effects. The analyses are subject to uncertainties and potential bias based on data available and data selected for analysis. However, such an overview of post-market MDR data, not previously published, fills a gap in understanding medical device issues and patient-based outcomes related to medical device use. Trends identified may be subjects of additional hypotheses, analysis, and research.

Drug-induced liver injury (DILI) poses a significant challenge for the pharmaceutical industry and regulatory bodies. Despite extensive toxicological research aimed at mitigating DILI risk, the effectiveness of these techniques in predicting DILI in humans remains limited. Consequently, researchers have explored novel approaches and procedures to enhance the accuracy of DILI risk prediction for drug candidates under development. In this study, we leveraged a large human dataset to develop machine learning models for assessing DILI risk. The performance of these prediction models was rigorously evaluated using a 10-fold cross-validation approach and an external test set. Notably, the random forest (RF) and multilayer perceptron (MLP) models emerged as the most effective in predicting DILI. During cross-validation, RF achieved an average prediction accuracy of 0.631, while MLP achieved the highest Matthews Correlation Coefficient (MCC) of 0.245. To validate the models externally, we applied them to a set of drug candidates that had failed in clinical development due to hepatotoxicity. Both RF and MLP accurately predicted the toxic drug candidates in this external validation. Our findings suggest that in silico machine learning approaches hold promise for identifying DILI liabilities associated with drug candidates during development.