Zinc finger E-box binding homeobox 1 (ZEB1) is a member of the zinc finger homeodomain transcription factor family, with a pivotal role in regulating the epithelial to mesenchymal transition (EMT) process. Increasing evidence suggests that ZEB1 is overexpressed in liver tumors, including hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA), and it correlates with advanced disease features and reduced overall survival. Here, we examine ZEB1 molecular functions, regulatory networks and contribution to tumorigenesis. We also discuss the emerging therapeutic strategies and future research directions aimed at targeting the ZEB1 molecular network to improve the outcome of liver cancer patients.
Simultaneous evaluation of acute and chronic lesions is critical for staging renal disease, yet manual assessment suffers from high inter-observer variability and heavy workloads. While deep learning models excel at structural segmentation, none currently consolidate the concurrent quantification of key AKI-to-CKD transition markers in standard stains. We developed a modified U-Net convolutional neural network for the pixel-wise segmentation and quantification of inflammation-associated interstitial fibrosis and tubular atrophy (i-IFTA), focal inflammatory infiltrates and tubular dilation in haematoxylin-eosin-stained whole-slide images (WSIs) from rat models of renal injury. The model achieved an overall test accuracy of 90.21% (weighted F1-score: 90.22%). Per-class F1-scores ranged from 0.73 for focal infiltrates to 0.94 for unaltered tissue, mirroring biological class distributions. Expert pathologists validated the high quality of the boundaries, which was comparable to junior-level annotations. The end-to-end inference pipeline successfully generates full-slide visual overlays and automated quantitative area metrics in a mean time of 215 seconds per WSI on standard CPU hardware. This model addresses a critical gap by simultaneously quantifying key structural and inflammatory lesions in H&E tissue. By reducing evaluation times from hours to minutes, it offers a scalable, reproducible tool to standardize histopathological assessment.
Intrahepatic cholangiocarcinoma (iCCA) is an aggressive tumour from the biliary tree that is characterized by a prominent desmoplastic stroma, mainly composed of cancer-associated fibroblasts (CAF), and a poor prognosis due to its late clinical presentation and the lack of effective non-surgical treatments. Current therapies still include chemotherapeutic combinations of gemcitabine and cisplatin for the majority of the patients showing poor results due the apparition of resistance. This situation led us to interrogate the potential role in the development of chemoresistance of ZEB1, an EMT-inducing transcription factor (EMT-TF) that we previously identified as a pro-tumorigenic factor in tumour cells and CAF of iCCA. By analysing human CCA samples and sc-RNAseq public databases, we show here that ZEB1 is present in the tumour microenvironment of all iCCA patients tested and is prominently expressed by the myofibroblastic CAF subpopulation (myCAF). Using cellular models of CAF, we show that cells depleted for ZEB1 are more sensitive to gemcitabine and cisplatin, via a mechanism involving the direct regulation of the anti-apoptotic gene BCL2. Moreover, ZEB1 expressing CAF protect iCCA tumour cells against the toxicity of the chemotherapeutic drugs through the release of paracrine soluble factors, including CTGF, IL6 and IL8, an event that could be reversed by the BCL2 inhibitor venetoclax. Therefore, our results point to the use of BCL2 inhibitors as a way to improve the efficacy of current chemotherapeutic regimens in iCCA patients.
(1) Background: Simultaneous evaluation of acute and chronic lesions is critical for staging renal disease, yet manual assessment suffers from high inter-observer variability and heavy workloads. While deep learning models excel at structural segmentation, none currently consolidate the concurrent quantification of key AKI-to-CKD transition markers on standard-stained tissue slides. (2) Methods: We developed a U-Net-based convolutional neural network for the pixel-wise segmentation and quantification of inflammation-associated interstitial fibrosis and tubular atrophy (i-IFTA), focal inflammatory infiltrates and tubular dilation in haematoxylin–eosin-stained whole-slide images (WSIs) from rat models of renal injury. (3) Results: The model demonstrated robust pixel-wise segmentation in H&E-stained rodent renal tissue (overall accuracy: 88.26%, macro-averaged Dice coefficient: 0.843; per-class Dice ranging from 0.727 for focal infiltrates to 0.920 for unaltered tissue). While comparative evaluation against human annotators was performed on a limited patch subset, the model exhibited agreement trends close to expert pathologists and superior to junior annotators. The end-to-end WSI inference pipeline generates full-slide visual overlays and automated WSI-level quantitative area measurements in a mean time of 103.57 s per WSI on GPU hardware. (4) Conclusions: This model offers an objective and automated framework for quantifying key structural and inflammatory lesions in rodent renal tissue. By enabling rapid WSI-level quantitative area measurements, it establishes a solid foundation for preclinical histopathological assessment and provides high potential for future translation into human clinical pathology.