Noninvasive prediction of occult pT3a upstaging in localized ccRCC with radiogenomic insights and prognostic relevance.
Authors
Affiliations (9)
Affiliations (9)
- Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
- MOE Key Laboratory for Biomedical Photonics, Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan, China.
- Department of Oncology, Tongji Hospital of Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
- Department of Radiology, Xiangyang Central Hospital, Affiliated Hospital of Hubei University of Arts and Science, Xiangyang, China.
- Shanxi Bethune Hospital, Shanxi Academy of Medical Sciences, Third Hospital of Shanxi Medical University, Tongji Shanxi Hospital, Taiyuan, China.
- Shanxi Bethune Hospital, Shanxi Academy of Medical Sciences, Third Hospital of Shanxi Medical University, Tongji Shanxi Hospital, Taiyuan, China. [email protected].
- Department of Radiology, the Affiliated Cancer Hospital of Zhengzhou University & Henan Cancer Hospital, Zhengzhou, China. [email protected].
- Department of Radiology, the First Affiliated Hospital of Nanjing Medical University, Nanjing, China. [email protected].
- Department of Radiology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China. [email protected].
Abstract
Occult pathological T3a (pT3a) upstaging in cT1b-T2a clear cell renal cell carcinoma (ccRCC) correlated with poor prognosis and necessitated modifications in surgical planning. However, predicting it preoperatively remains challenging. In this multicenter study involving 1661 patients across five institutions and the KiTS23 dataset, RENALNet, a 3D deep learning framework trained on nephrographic-phase CT, was developed and validated. RENALNet outperformed radiomics models, further enhancing diagnostic accuracy when combined with radiologists of varying experience. Grad-CAM visualizations concentrated on anatomically significant invasion sites, improving interpretability. Risk scores derived from RENALNet were found to correlate with Ki-67 proliferation indices and effectively stratified 5-year progression-free survival, demonstrating both biological and prognostic relevance. Transcriptomic analysis revealed that high RENALNet risk was associated with gene expression signatures enriched in pathways such as epithelial-mesenchymal transition, IL6-JAK-STAT3 signaling, and PI3K-Akt signaling, highlighting its link to tumor aggressiveness. RENALNet thus offers a biologically interpretable framework for risk stratification in ccRCC, supporting surgical decision-making and advancing the integration of radiogenomic deep learning into precision oncology.