Abstract
Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality. It is often diagnosed at advanced stages and characterized by high recurrence rates. Although chronic liver inflammation and metabolic dysfunction are established contributors to tumorigenesis, the molecular mechanisms that link microenvironmental stress to malignant transformation remain poorly understood. MYCN, a proto-oncogenic transcription factor, has emerged as a potential biomarker of cancer stemness. However, its role in hepatocarcinogenesis remains unclear. In this study, we elucidated the oncogenic role of MYCN and its dynamic transcriptional regulation during liver tumorigenesis. Using a hydrodynamic tail vein injection-based transposon system in mice, we demonstrated that MYCN overexpression synergizes with AKT activation to promote liver tumorigenesis. Transcriptomic profiling revealed that MYCN-driven tumors exhibited features of human HCC subtypes enriched in stress-adaptive transcriptional programs. Time-resolved spatial transcriptomics further uncovered a MYCN-enriched niche characterized by epithelial-mesenchymal transition (EMT) and Wnt/β-catenin signaling, which expanded during tumor progression and was spatially proximate to transformed malignant cells. To translate these findings to human HCC, we developed a machine learning-based MYCN niche score and validated its clinical relevance across multiple human HCC cohorts. This score reliably predicted recurrence risk and identified EMT-prone microenvironments, with stronger predictive performance in nontumor tissues, suggesting its potential in detecting precancerous niches predisposed to de novo tumorigenesis. Collectively, our findings establish MYCN as a functional driver and spatial marker of tumor-promoting microenvironments in liver tumorigenesis; additionally, we propose a clinically actionable strategy to identify high-risk patients through transcriptomic profiling of nontumor liver tissue.