Multi-omics analysis of the lactylation-driven microenvironment and non-invasive clinical translation in hepatocellular carcinoma

Authors:

Abstract
Background Through integrated multi-omics strategy, this study aims to investigate the clinical value of a prognostic model derived from lysine lactylation (KLa) in hepatocellular carcinoma (HCC), thereby promoting its clinical translation. Methods The key features for constructing lactylation-associated risk model were identified across the levels of KLa, protein and transcriptomic by integrating local cohort and public HCC cohorts. Immunohistochemistry and western blotting were performed to confirm the consistency between lactylation-associated risk and KLa modification level. Clinical and biological significance of lactylation-associated risk stratification was further characterized leveraging local cohort with public HCC cohort and single-cell sequencing. Finally, we explored the potential of ultrasound-based radiomics for non-invasive prediction of lactylation-associated risk. Results Lactylation-associated risk model effectively stratified HCC patients based on recurrence-free survival and identified individuals with early postoperative recurrence. High-risk patients demonstrated elevated KLa modification level and clinical characteristics indicative of poor prognosis: microvascular invasion, tumor thrombus, high Ki-67, advanced TNM and BCLC stages. At single-cell resolution, the predominantly immunosuppressive tumor immune microenvironment and KLa-driven proliferative tumor phenotype in high-risk patients elucidate the underlying biological mechanisms of HCC progression. The integration of ultrasound-based radiomics and clinical features demonstrated superior potential for lactylation-associated risk non-invasive prediction compared to single clinical model. Conclusion Lactylation-associated risk model not only provides an effective assessment of recurrence risk in HCC patients but also mirrors the intratumoral KLa modification level, holding significant clinical and tumor biological relevance. Furthermore, the joint model integrating ultrasound-based radiomics and clinical features demonstrates potential for non-invasive prediction of lactylation-associated risk. Graphical abstract
Funding Information
  • National Natural Science Foundation of China (82160350)
  • National Natural Science Foundation of China (82160336)
  • China Postdoctoral Science Foundation (2023MD744191)
  • Joint Project on Regional High-Incidence Diseases Research of Guangxi Natural Science Foundation (2024GXNSFBA999019)
  • Innovation Project of Guangxi Graduate Education (YCSW2024238)

This publication has 29 references indexed in Scilit: