Deep Visual Multi-Omics Profiling Links Morphology and Molecular Programs in Clear Cell Renal Cell Carcinoma

Clear cell renal cell carcinoma exhibits striking intra-tumoral heterogeneity at morphological and genetic levels, complicating treatment and contributing to disease progression. CcRCCs with rhabdoid differentiation are highly aggressive tumors characterized by distinct histopathologies. However, the relationship between morphology, underlying molecular alterations, and tumor behavior remains largely unclear. Here, we present Deep Visual Multi-Omics, an approach integrating digital pathology, morphology-guided single-cell isolation, and ultra-sensitive multi-omics profiling to link cell morphologies to their molecular underpinnings. Across five tumors, we profiled 40,000 AI-classified and expert-curated cells. We identified progressive molecular dysregulation across cells with increasing histopathological grade coexisting within heterogeneous tumors as well as distinct molecular alterations associated with aggressive rhabdoid ccRCC cells, including signatures consistent with enhanced FOXM1-driven proliferation, altered cell-matrix interactions, and a putative immunomodulatory phenotype. Notably, rhabdoid cells exhibited elevated expression of IFN-beta, PD-L1, CD38, ITGB2, and integrin signaling, suggesting that they themselves may act as a source of signals influencing the local immune microenvironment. Besides providing new insights into the biology of ccRCC and highlighting avenues for future translational studies, this illustrates the potential of Deep Visual Multi-omics to dissect cancer heterogeneity and characterize high-risk cell populations. Intra-tumor heterogeneity drives tumor progression and therapy resistance. In ccRCC for example, distinct histopathologies coexist and are associated with patient outcome, yet their molecular basis is poorly understood. Deep Visual Multi-Omics links cell morphology to the underlying molecular state. Intra-tumor heterogeneity drives tumor progression and therapy resistance. In ccRCC for example, distinct histopathologies coexist and are associated with patient outcome, yet their molecular basis is poorly understood. Deep Visual Multi-Omics links cell morphology to the underlying molecular state.