Abstract
INTRODUCTION: Hepatoblastoma (HB) with hepatocellular carcinoma (HCC) features (HBHF) is a rare liver malignancy. Due to its rarity and diverse histological presentations, the prognosis of HBHF remains controversial, and diagnostic differentiation poses significant challenges. To enable more accurate outcome evaluation and targeted therapeutic strategies, rapid, comprehensive, and cost-effective methods are needed to complement histopathological evaluation. METHODS: In this study, we conducted transcriptomic profiling on an HBHF cohort from our center and developed a machine-learning algorithm to quantify HCC-like expression features in HB tumors. Given overlapping histopathological and molecular charateristicss between HBHF and HCC, we further investigated shared risk factors associated with HBHF prognosis. RESULTS: Significantly poorer outcomes in HBHF patients suggest fundamental biological distinctions from classical HB. Transcriptomic analysis revealed comparable somatic mutation profiles between HB and HBHF cohorts but identified inflammation activation, rather than specific mutations, as a key high-risk factor in HBHF. Clinical outcomes aligned with risk stratification generated by our quantification model. CONCLUSIONS: HBHF represents a distinct transitional entity between HB and HCC, exhibiting markedly worse clinical outcomes than HB. Our transcriptome-based computational model effectively discriminates HBHF and predicts its prognostic risk. Importantly, inflammatory activation emerges as a critical driver of tumor aggressiveness in this subtype.