Deep learning predicts and in vitro experiments validates the synergistic anti-liver cancer effect of vincristine and lenvatinib: Mechanism involving apoptosis induction via the TNF-α/Caspase-8 pathway.

Resistance to lenvatinib has become a major obstacle in the clinical treatment of liver cancer, highlighting the significant research value and translational potential of developing synergistic drug combinations. In this study, deep learning models (MARSY and MatchMaker) were employed to predict potential synergistic partners for lenvatinib, with vincristine identified as a promising candidate. In vitro experiments confirmed that the combination synergistically inhibited the proliferation, migration, and clonogenic formation of liver cancer cells: CCK-8 and colony formation assays demonstrated a significant reduction in cell viability and clonogenic ability, while wound healing and Transwell assays indicated effective suppression of cell migration. The synergistic effect was quantitatively validated using the ZIP model. Furthermore, flow cytometry and Western blot analyses confirmed that the combination effectively induced apoptosis. Mechanistic studies revealed that the co-treatment led to excessive accumulation of intracellular reactive oxygen species (ROS), which activated the TNF-α/Caspase-8 signaling pathway, thereby inducing apoptosis in liver cancer cells. The cytotoxicity and pro-apoptotic effects were significantly attenuated by the ROS scavenger NAC. These findings provide a solid preclinical foundation for the further development of this combination therapy and underscore the importance of the "computational prediction-mechanistic validation" strategy in advancing cancer drug discovery.