This study investigated the landscape of RNA modifications in the peripheral blood of patients with hepatocellular carcinoma (HCC), liver cirrhosis, and healthy controls. Using liquid chromatography-tandem mass spectrometry, 55 RNA modifications were quantified, revealing 11 significantly altered modifications in HCC patients, including those at stage I. N2,N2-dimethylguanosine (m2,2G) emerged as the most profoundly dysregulated. A diagnostic model based on a panel of modifications including m2,2G, combined with IFI27 and CCR2 gene expression, demonstrated promising diagnostic performance for HCC, especially for early stages, with high AUCs. These findings suggest the potential for novel diagnostic biomarkers for HCC.
This study introduces DNAmBERT, a Transformer-based deep learning model for non-invasive cancer diagnosis. It jointly models DNA sequence context and read-level methylation haplotype structure from cfDNA, addressing limitations of current computational methods. Evaluated across various platforms and cancer types, DNAmBERT achieved high performance with AUCs up to 0.99-1.00. The model supports multi-cancer and stage-aware classification, including early-stage disease, and demonstrates effective cross-cancer transfer learning.
This study characterized biallelic inactivation of the RB1 tumor suppressor gene (RB1-Bi) in hepatocellular carcinoma (HCC), finding it in 14.6% of patients and associating it with significantly reduced overall survival. A deep learning model (FR-MIL) was developed to accurately predict RB1-Bi status from routine histopathology slides. High-throughput drug screenings revealed that RB1-Bi HCC cells are particularly sensitive to inhibitors targeting mitotic regulators and PARP inhibitors. These synthetic lethal interactions were validated in vitro and in vivo, suggesting RB1-Bi is an actionable biomarker for precision medicine in HCC.
This multi-omics and experimental study identifies Ubiquitin-conjugating enzyme E2 C (UBE2C) as a central hub gene and a key prognostic biomarker in hepatocellular carcinoma (HCC). UBE2C is significantly upregulated in HCC, predominantly in hepatocytes, and its knockdown suppresses tumor proliferation and growth. The study reveals that UBE2C-high regions are associated with an immunosuppressive microenvironment, characterized by TGFB1 enrichment and cytotoxic T-cell exclusion. A nomogram integrating UBE2C expression, T stage, and tumor stage was developed to predict patient survival and stratify immunotherapy response.