552 Background: RCC has multiple distinct molecular and histological subtypes. Prior studies suggest that clear cell RCC (ccRCC) and papillary RCC (pRCC) originate from the proximal tubule (PT), while chromophobe RCC (chRCC) arises from the distal tubule/collecting duct. Cell-of-origin (COI) analysis provides insight into lineage-specific dependencies and signaling pathways, which can uncover novel therapeutic targets. Using a single-cell RNA sequencing (scRNA-seq) dataset, we conducted COI, differential gene expression and signaling pathway analysis in multiple RCC subtypes. Methods: ccRCC, pRCC, chRCC and unclassified RCC (uRCC) samples were analyzed via scRNA-seq. A normal kidney scRNA-seq reference was used from the KPMP (V1.5). A logistic regression model was trained using KPMP epithelial clusters with a previously described gene set (Labaki, et.al. JCO, 2025). This model was applied to tumor clusters to quantify the probability of similarity to normal epithelial clusters. Differential gene and pathway enrichment analysis was conducted on tumor clusters and the most similar epithelial cluster. Results: 84,362 tumor cells from 28 RCC tumors were identified (chRCC = 5 patients, 35,897 cells; pRCC = 3, 929; ccRCC = 14, 18,418; uRCC = 4, 29,119). The reference dataset included 171,390 normal epithelial cells annotated into major nephron cell types. Applying the COI model, ccRCC and pRCC tumor clusters demonstrated the greatest similarity to PT cells (probability (prob)-0.63 and 0.83, respectively). In contrast, chRCC most resembled α-intercalated cells (ICA) (prob-0.88). uRCC tumors had interpatient heterogeneity, with one patient (uRCC-1) resembling ICA (prob-0.75) and three patients (uRCC-2) resembling PT (prob-0.58). Comparing tumor cell clusters with their COI revealed that chRCC and uRCC-1 shared several upregulated oncogenic pathways (mTOR and MAPK), and 305 upregulated genes, including NUPR1, KLK1/4, IGF1, and IRS2 . pRCC demonstrated enrichment of IL-15 signaling and antigen processing/presentation pathways, with upregulation of HLA-A, -B, -C, -E, and -F . ccRCC and uRCC-2 exhibited high transcriptional similarity, with 812 shared differentially expressed genes. Both subtypes showed upregulation of glycolytic programs and glycolytic genes ( SLC2A1 and HK1) , accompanied by downregulation of fatty acid metabolism. Conclusions: COI analysis in our pan-RCC scRNA-seq dataset confirmed distinct COIs for RCC subtypes and aided in classifying uRCC relative to other histologies. Subtype-specific pathway upregulation included mTOR signaling in chRCC/uRCC-1 and antigen-presentation in pRCC, supporting the potential for mTOR-targeted and immunotherapy approaches in these subtypes. These findings underscore the utility of COI-based approaches for refining RCC classification and uncovering therapeutic vulnerabilities.
Yochum et al. (Sun,) studied this question.