Chinese researchers have developed a powerful machine learning model that will help determine whether patients with nasopharyngeal cancer (NPC) are more likely to respond well to radiation therapy. It is a common treatment for this type of cancer. The study was conducted by scientists from Zhujiang Hospital and Nanfang Hospital at Nanbu Medical University, and introduces a prediction tool known as NPC-RSS (natopharyngeal cancer radiotherapy sensitivity score).
Using a rigorous machine learning framework that evaluates the combination of transcriptome data and 113 algorithms, the team identified 18 gene signatures that could predict patient radiation sensitivity. This model showed impressive accuracy for both internal datasets and external validation sets.
Radiation therapy is the primary treatment for NPCs, but up to 30% of patients recur due to radiation resistance. Our model helps to solve this problem by identifying the patients most likely to benefit from radiation therapy, allowing for more customized and effective treatment strategies. ”
Lead author, Dr. Jiang Chan
We found that the core genes of the model, such as SMARCA2, DMC1, and CD9, influence tumor immune invasion and important signaling pathways such as Wnt/β-catenin and Jak-stat. In particular, the radiation-sensitive group exhibits higher levels of immune cell activity, suggesting an intimate link between radioactivity and immunodynamics.
The predictive power of NPC-RSS was confirmed using cell lines and single cell sequencing, indicating that radiosensitivity tumors have an abundant immune environment compared to resistant tumors. According to co-author Dr. Hui Meng, “Our findings suggest that integrating gene scores with immune profiles could be a game-changer in NPC care.”
The team believes that models could be a clinical tool to guide treatment decisions, minimize unnecessary radiation exposure, and optimize treatment outcomes. They are currently working to expand sample sizes and work with international partners to further validate and refine the model.
sauce:
Journal Reference:
Li, K. , et al. (2025). A multigene prediction model for the radiation susceptibility of nasopharyngeal cancer based on machine learning. Elif. doi.org/10.7554/elife.99849.3.
