Abstract
Onco-fetal (OF) reprogramming of the tumor ecosystem induces fetal-like characteristics, leading to tumor growth and malignant progression. Here, we develop a robust computational pipeline to systematically identify OF cells, using single-cell transcriptomic data from corresponding fetal and tumor tissues across multiple cancer types. OF cells are consistently detected across diverse malignancies and exhibit pronounced stem-like properties. Core transcription factors involved in OF reprogramming are identified, including PTMA, ATF5, TCF4, and FOS, whose coordinated activities may promote tumor progression. Strategically positioned OF cells orchestrate tumor microenvironment cross talk, specifically impairing B cell function. OF cells are further verified in additional cancers, and we find that high OF cell abundance contributes to enhanced drug resistance and cancer recurrence. Critically, pancancer analyses establish that OF cell markers universally portend adverse prognoses. Overall, our study uncovers a clinically consequential cancer cell subpopulation and highlights its potential as a target for precision oncology.