Integrating single-cell multi-omics and machine learning to reveal triaptosis heterogeneity in clear cell renal cell carcinoma

Triaptosis, an emerging form of cell death, remains poorly characterized in terms of its heterogeneity within clear cell renal cell carcinoma (ccRCC). Utilizing single-cell transcriptomics, we delineate a landscape of triaptosis heterogeneity and identify monocytes and macrophages as exhibiting the highest triaptosis activity, which further increases upon terminal differentiation. These high-activity cells also demonstrate enhanced pro-angiogenic signaling toward endothelial cells. Within epithelial cells, subpopulations with the strongest triaptosis activity are located at the late differentiation stage and are closely associated with ccRCC traits. Spatial transcriptomic analysis reveals a decline in triaptosis activity with increasing distance from the tumor epithelial core. The epithelial cluster with the highest triaptosis activity showed reduced metabolic activity. In bulk transcriptome analysis, patients with high epithelial triaptosis activity infiltration exhibited improved prognosis, broader immune activation, and similarly suppressed metabolism. We subsequently developed a robust 4-gene prognostic signature based on module genes derived from high-triaptosis epithelial subpopulations. This model showed strong performance in prognostic stratification, immunotherapy guidance, and chemotherapy response prediction. Finally, we identified SLC25A37 as a core oncogenic gene within the signature and proposed Yohimbic acid among several potential molecularly targeted therapeutics.