Cancer stem cells and where to find them: Their trajectories and implications in cancer evolution

Cancer cells are heterogeneous within and among biopsies and can be perceived as a hierarchical structure defined by growth potential, in which more stem-like cells are at the apex. Cancer stem cells (CSCs) thus represent a distinct subset of cancer cells with stemness properties that play a pivotal role in tumour growth, recurrence, and metastasis. How and where CSCs originate remain somewhat controversial as differentiated cells may also regain stemness-like properties through dedifferentiation, and CSC phenotypes can vary among patients and cancer types. To systematically identify and trace potential CSC populations in lung adenocarcinoma (LUAD), we implemented an integrative computational framework that leverages both genomic and transcriptomic features from single-cell RNA sequencing (scRNA-seq) datasets from LUAD patients. By reconstructing developmental trajectories, we defined differentiation states within malignant compartments and detected rare cancer cell subsets as candidate CSCs in each biopsy. We then inferred copy number variations (CNVs) for individual malignant cells and generated phylogenetic trees based on CNV patterns to reconstruct clonal evolution. Our analyses revealed that CSC-like cells not only emerged as early ancestral clones but also arose later during tumour evolution, exhibiting similar stemness programs but undergoing dedifferentiation. We also explored their spatial organization with Visium spatial transcriptomics, and we observed that CSC-like states localized to spatially variable regions and were found near immune and stromal niches, suggesting microenvironmental influences on stemness. Finally, through deconvoluting bulk RNA-seq data and performing survival analysis, we derived a stemness signature capable of stratifying patient disease outcomes. In summary, this integrated multi-modal analysis provides insights into the identity, evolutionary dynamics, spatial distribution, and clinical relevance of CSCs in LUAD. Our approach offers a foundation for identifying CSC-driven mechanisms of tumour progression in LUAD and other cancers, and new biomarkers for prognosis and therapy response.