Related Experiment Video For Programmed cell death
Updated: Jun 22, 2026
Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018
A Multimodal Diagnostic Model for Hepatocellular Carcinoma Integrating Biomarkers Related to Programmed Cell Death
Lixing Lei1, Nian Liu1, Lingling Tang1
1Department of Radiology, Affiliated Hospital of North Sichuan Medical College, Nanchong, 637000, China.
Introduction:
Hepatocellular carcinoma (HCC) is characterized by its insidious onset and rapid progression. Investigating diagnostic and therapeutic strategies targeting programmed cell death (PCD) represents a promising research direction.
Methods:
PCD pathway activation in HCC was assessed via single-sample GSEA (ss- GSEA). Differentially expressed genes (DEGs) were screened by WGCNA, FindMarkers, and the limma package from bulk and single-cell data. Intersecting DEGs were refined by LASSO regression in the glmnet package, and diagnostic performance was validated using ROC analysis. Functional enrichment analysis was conducted with the clusterProfiler package, and drug prediction was performed with Enrichr package. Py- MOL, AutoDockTools, and AutoDock Vina software were employed to perform molecular docking simulations. A radiomics-driven LASSO model was applied to construct a diagnostic nomogram.
Results:
PCD plays a crucial part in the development of HCC. Three genes (CAPG, MS4A6A, and TREM2) were identified as the PCD-related diagnostic genes for HCC. The reliability of the three genes was confirmed by ROC analysis, with AUC values exceeding 0.7. Drug prediction screened 46 candidate compounds, from which Tamibarotene and Vorinostat were selected for molecular docking. Finally, a nomogram was established based on the radiomics features of the CAPG gene, reaching an AUC above 0.8.
Discussion:
Using interpretable machine learning to integrate transcriptomic, single-- cell, and radiomic data, we revealed systemic dysregulation of PCD-related pathways in HCC and validated the diagnostic value of CAPG, MS4A6A, and TREM2 across datasets. A CAPG-based radiomics nomogram showed favorable discrimination and calibration, suggesting translational potential.
Conclusion:
We developed a reusable multimodal diagnostic framework with candidate biomarkers and a radiomics tool to facilitate the early detection and risk stratification of HCC.
