Development of a novel RNA modification-based risk model to predict prognosis and immunotherapy response in melanoma using machine learning
Related Experiment Video
Updated: Nov 6, 2026
Predictive Immune Modeling of Solid Tumors
Published on: February 25, 2020
Bailu Wu1, Minghui Ge2, Qin Zhang2
1Interventional Radiology Department, The First Affiliated Hospital of Zhengzhou University, No. 1 Longhu Zhonghuan Road, Jinshui District, Zhengzhou, 450003, China.
Background:
Melanoma is a highly malignant cancer that poses a significant global health burden, yet there is limited ability to predict prognosis or immunotherapy efficacy for melanomas. Emerging evidence highlights the role of RNA modification regulators (m6A, m1A, m5C, and m7G) in modulating oncogenesis and immune responses.
Methods:
We comprehensively profiled 84 regulators of RNA modification (m6A/m1A/m5C/m7G) across multiple melanoma cohorts. Unsupervised clustering was performed to identify RNA modification-based subtypes, followed by WGCNA to reveal key modules. We developed the RMODscore prognostic model using LASSO and multivariate Cox regression and validated the model in several independent melanoma immunotherapy cohorts. Finally, we explored the association of RMODscore with immune signatures and drug sensitivity.
Results:
We found significant differences in survival among three different RNA modification subtypes and identified the ME13 module as significantly associated with these subtypes using the WGCNA method. Our newly constructed 10-gene RMODscore model showed strong predictive performance in the TCGA cohort and four independent immunotherapy datasets. A high RMODscore was associated with poor survival, immune-cold phenotypes, and resistance to checkpoint blockade. Functional analysis revealed that tumors with high RMODscore values exhibited upregulation of proliferation-related pathways, whereas tumors with low RMODscore values exhibited greater immune activation. In addition, the RMODscore predicted sensitivity to ERK and JNK inhibitors, suggesting potential drug repurposing strategies.
Conclusion:
This study presents a novel RNA modification-based prognostic risk score as a robust prognostic tool for melanoma which may provide potential prognostic stratification value. These findings highlight the potential of RNA modification signatures for prognostic stratification and may provide preliminary insights for future therapeutic strategy exploration in patients with melanoma.