Drug toxicity prediction plays a crucial role in the drug research and development process, ensuring clinical drug safety. However, traditional methods are hampered by high cost, low throughput, and uncertainty of cross-species extrapolation, which has become a key bottleneck restricting the efficiency of new drug research and development. The breakthrough development of Artificial Intelligence (AI) technology, especially the application of deep learning and multimodal data fusion strategy, is reshaping the scientific paradigm of drug toxicology assessment. In this review, we focus on the application of AI in the field of drug toxicity prediction and systematically summarize the relevant literature and development status globally in the past years. The application of various toxicity databases in the prediction was elaborated in detail, and the research results and methods for the prediction of different toxicity endpoints were analyzed in depth, including acute toxicity, carcinogenicity, organ-specific toxicity, etc. Furthermore, this paper discusses the application progress of AI technologies (e.g., machine learning and deep learning model) in drug toxicity prediction, analyzes their advantages and challenges, and outlines the future development direction. It aims to provide a comprehensive and in-depth theoretical framework and actionable technical strategies for toxicity prediction in drug development.

This study integrates network toxicology with molecular docking technology to systematically elucidate the key molecular mechanisms and signaling pathways by which bisphenol A (BPA) induces obesity. By cross-referencing multiple databases—including the Comparative Toxicogenomics Database (CTD), SwissTarget prediction platform, and PharmMapper—potential BPA target genes were identified, yielding a total of 1326 candidate targets. Obesity-related genes were collected from GeneCards and OMIM databases, yielding 4570 disease-associated targets. Among these, 653 overlapping genes were identified as potential mediators linking BPA exposure to obesity. Protein interaction networks were constructed using STRING and Cytoscape, and the MCC algorithm identified five core hub genes: STAT3, MYC, TP53, IL6, and mTOR. Validation using random datasets demonstrated significant upregulation of these genes in the obesity group (p < 0.05), highlighting their potential central role in BPA-induced obesity effects. Functional enrichment analysis via GO and KEGG pathways indicated that BPA may promote obesity by interfering with endocrine signaling, activating lipid metabolism, and stimulating atherosclerosis pathways. Molecular docking analysis using CB-Dock2 confirmed strong binding affinity between BPA and core targets, providing structural evidence for their potential interactions. This study elucidates the potential biological mechanism by which BPA exacerbates obesity through endocrine disruption and metabolic reprogramming, employing a multidimensional approach encompassing cross-target analysis, pathway enrichment, and molecular interactions. It provides an innovative systems toxicology framework and empirical basis for assessing metabolic health risks induced by environmental pollutants.

Mesenchymal stem cells have made remarkable progress in recent years. Many studies have reported that human umbilical cord mesenchymal stem cells (hUC-MSCs) have no toxicity, but thromboembolism appeared in patients treated with hUC-MSCs. Therefore, people are still worried about the safety of clinical application. The study aims to determine the safety, potential toxic mechanism and biodistribution of hUC-MSCs. F344RG rats were given 5 or 50 million cells/kg of hUC-MSCs by single administration in compliance with Good Laboratory Practice standards. Standard toxicity was performed. RNA sequencing was then performed to explore the potential toxic mechanisms. In parallel, the biodistribution of hUC-MSCs was examined. The dose of 5 million cells/kg hUC-MSCs had no obvious toxicity on symptom, weight, food intake, hematology, serum biochemistry, urine biochemistry, cytokines, and histopathology. However, blood-tinged secretions in the urethral orifice and 20% mortality occurred at 50 million cells/kg. Disseminated intravascular coagulopathy (DIC) is the leading cause of death. hUC-MSCs significantly upregulated complement and coagulation cascade pathways gene expression, resulting in DIC. Besides, hUC-MSCs upregulated fibrinolytic system suppressor genes A2m, Serping1 and Serpinf2. hUC-MSCs survived in rats for less than 28 days, no hUC-MSC was detected in tissues outside the lungs. There was no toxicity in F344RG rats at 5 million cells/kg, but some toxicities were detected at 50 million cells/kg. hUC-MSCs significantly upregulated complement and coagulation cascade pathways, upregulated the expression of fibrinolytic system suppressor genes A2m, Serping1 and Serpinf2, to inhibit fibrinolytic system, caused DIC, which provided a new insight into the toxic mechanism of hUC-MSCs. Graphical Abstract

This study investigated the molecular mechanisms by which phthalates induce depression, utilizing network toxicology and molecular docking techniques. By integrating the TargetNet, GeneCards, and PharmMapper databases, 658 potential target genes of phthalates were identified. Additionally, 5433 depression-related targets were retrieved from the GeneCards and OMIM databases. Comparative analysis revealed 360 common targets implicated in both phthalate action and depression. A Protein-Protein Interaction (PPI) network was constructed using the STRING database. Subsequently, the CytoHubba plugin (employing the MCC algorithm) within Cytoscape was used to screen the network, identifying the top 20 hub genes. These core genes include AKT1, CASP3, TNF, TP53, BCL2, and IL6, among others. Validation on the GEO dataset (GSE23848) revealed that the expression of multiple core genes was significantly upregulated in patients with depression (p < 0.05). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses indicated that phthalates mainly regulate biological processes such as extracellular stimulus response, lipopolysaccharide metabolism, and chemical synaptic transmission. Depression is mediated by the AGE-RAGE signaling pathway (a complication of diabetes), lipids and atherosclerosis, Endocrine resistance, and the PI3K-Akt signaling pathway. Molecular docking confirmed that phthalates have strong binding activity with key targets (CASP3, TNF, TP53, BCL2, IL6). These findings present a novel paradigm for evaluating the health risks posed by environmental pollutants.

Aristolochic acid I (AAI) is widely recognized as a genotoxic and cytotoxic compound. To rationally propose detoxification strategies, it is essential to fully elucidate the in vivo disposition of AAI. Nevertheless, the toxicokinetic characteristics of AAI, particularly the possible involvement of the recirculation process, remain incompletely understood. In this research, toxicokinetics of AAI was studied following a single oral administration of AAI in Fisher rats (10, 30 and 100 mg/kg, n = 6). A method of ultra-performance liquid chromatography coupled with triple quadrupole tandem mass spectrometry (UPLC-QQQ-MS/MS) was developed to achieve the quantitation of AAI in rat plasma. Plasma concentration–time profiles and kinetic parameters were analyzed to characterize the toxicokinetic behavior of AAI. A secondary elevation was observed in the plasma concentration–time profiles of AAI, suggesting the existence of AAI reabsorption. The non-linear elimination kinetics of AAI might be attributed to capacity-limited excretion via bile. Additionally, the biliary excretion of AAI and several key metabolites was also explored through qualitative analysis of bile samples. For the first time, AAI-O-glucuronide was identified in bile, providing further support for enterohepatic recirculation (EHR)-mediated reabsorption of AAI. In conclusion, these findings provided solid evidence for EHR-mediated reabsorption of AAI in rats. The recirculation process might be a key mechanism responsible for the prolonged retention of AAI. In the future, detoxification strategies targeting the EHR process could be effective approaches to minimize the systemic exposure of AAI.

Objective: In non-clinical safety evaluation of drugs, pathological result is one of the gold standards for determining toxic effects. However, pathological diagnosis might be challenging and affected by pathologist expertise. In carcinogenicity studies, drug-induced squamous cell carcinoma (SCC) of the mouse stomach represents a diagnostic challenge for toxicopathologists. This study aims to establish a detection model for mouse gastric squamous cell carcinoma (GSCC) using deep learning algorithms, to improve the accuracy and consistency of pathological diagnoses.Methods: A total of 93 cases of drug-induced mouse GSCC and 56 cases of normal mouse stomach tissue from carcinogenicity studies were collected. After scanning into digital slides, semi-automated data annotation was performed. All images underwent preprocessing, including tissue extraction, artifact removal, and exclusion of normal epithelial regions. The images were then randomly divided into training, validation, and test sets in an 8:1:1 ratio. Five different convolutional neural networks (CNNs)-FCN, LR-ASPP, DeepLabv3+, U-Net, and DenseNet were applied to identify GSCC and non-GSCC regions. Tumor prediction images (algorithm results shown as overlays) derived from the slide images were compared, and the performance of the constructed models was evaluated using Precision, Recall, and F1-score.Results: The Precision, Recall, and F1-scores of DenseNet, U-Net, and DeepLabv3 + algorithms were all above 90%. Specifically, the DenseNet model achieved an overall Precision of 0.9044, Recall of 0.9291, and F1-score of 0.9157 in the test set. Compared to the other algorithms, DenseNet exhibited the highest F1-score and Recall, demonstrating superior generalization ability.Conclusion: The DenseNet algorithm model developed in this study shown promising application potential for assisting in the diagnosis of mouse GSCC. As artificial intelligence (AI) technology continues to advance in non-clinical safety evaluation of drugs, CNN-based toxicological pathology detection models will become essential tools to assist pathologists in precise diagnosis and consistency evaluation.