# Drug-induced gastrointestinal toxicity and barrier integrity: cytoskeleton-mediated impairment in a clinically relevant human intestinal epithelium model

## TL;DR

A new human intestinal cell model better predicts drug-induced gut toxicity than traditional methods, offering a more accurate preclinical tool.

## Contribution

A novel TEER-based assay using nontransformed human intestinal epithelial cells improves GI toxicity prediction compared to Caco-2 and viability assays.

## Key findings

- The hIEC TEER assay achieved an AUC of 0.96, outperforming Caco-2 TEER (0.72) and cell viability assays (≤0.69).
- Transcriptomic analysis revealed cytoskeleton-related pathways are downregulated by microtubule-targeting chemotherapeutics.
- The hIEC TEER assay showed 92% sensitivity, 100% specificity, and 94% accuracy in predicting GI toxicity.

## Abstract

Drug-induced gastrointestinal (GI) toxicity is common, dose-limiting and difficult to predict using conventional Caco-2-based assays that lack physiological relevance. Here we evaluate a transepithelial electrical resistance (TEER) assay using nontransformed human intestinal epithelial cells (hIECs), derived from human pluripotent stem cells, which superiorly recapitulated epithelial diversity and polarity as well as intestinal barrier function. Across 17 clinically relevant compounds (cell cycle inhibitors, tyrosine kinase inhibitors and nonsteroidal anti-inflammatory drugs), the hIEC TEER assay outperformed ATP cell viability assays, and the Caco-2 TEER assay (AUC of 0.96 for hIEC TEER, 0.72 for Caco-2 TEER and ≤0.69 for cell viability assays) correlated with integrated GI toxicity scores using a ≥50% TEER reduction cutoff (sensitivity 92%, specificity 100% and accuracy 94%). Drug exposure was quantified by calculating the margin of safety (IC15:Cmax) and a lumen–surrogate margin of safety for oral agents. For mechanistic insight, transcriptomic analysis using representative chemotherapeutics (paclitaxel and docetaxel) showed the downregulation of cytoskeleton-related pathways, including cytoskeleton in muscle cells, cell adhesion molecules and extracellular matrix–receptor interaction, linking microtubule-targeting chemotherapy to intestinal barrier impairment. This platform provides a robust tool that combines predictive accuracy with the evaluation of cytoskeleton-mediated barrier impairment, enabling the early identification of drug-induced GI toxicity.

Clinical trials often fail owing to drug toxicity causing symptoms such as diarrhea and abdominal pain. This study addresses the need for better methods to predict gastrointestinal (GI) toxicity. Researchers developed a new test using human intestinal epithelial cells derived from stem cells. They compared this with the widely used Caco-2 model. The researchers used a method called transepithelial electrical resistance to measure how well the intestinal barrier functioned after drug exposure. This method is important because it can detect early signs of barrier disruption, a key indicator of GI toxicity, often before standard viability assays show damage. The results showed that the human intestinal epithelial cell-based transepithelial electrical resistance assay was more accurate in predicting drug-induced GI toxicity than traditional methods. The study concludes that this new model could strengthen preclinical safety testing and improve drug development by reducing clinical trial failures. This summary was initially drafted using artificial intelligence, then revised and fact-checked by the author.