Redox-senescence function of PON1 in hepatocellular carcinoma and its non-invasive assessment using super-resolution radiomics: a multi-center study
摘要
Risk stratification in hepatocellular carcinoma (HCC) is limited by the lack of robust biomarkers reflecting tumor biology. The antioxidant enzyme Paraoxonase-1 (PON1) shows prognostic potential, yet its role in tumor tissues and the feasibility of non-invasive assessment remain unclear.
MethodsSenescence-related pathways and prognostic candidates were screened using transcriptomic data from TCGA-LIHC and GTEx. PON1 expression was validated in multicenter cohorts through qPCR, immunohistochemistry, and Western blotting. Functional assays in PON1 knockdown and overexpression models evaluated oxidative stress, glutathione balance, mitochondrial dysfunction, and senescence markers. We developed a radiomics model based on contrast-enhanced CT scans with super-resolution reconstruction to predict tumoral PON1 expression. This radiomics signature was then integrated with clinical variables to build a combined model, which was evaluated across training, validation, test, and external cohorts.
ResultsPON1 was identified as a downregulated senescence-related prognostic gene. Low PON1 expression was associated with poorer overall and progression-free survival across independent clinical cohorts. PON1 depletion increased intracellular and mitochondrial ROS, lowered the GSH/GSSG ratio, impaired mitochondrial membrane potential, and induced senescence phenotypes, while its restoration mitigated these effects. SR-enhanced radiomics improved prediction of tumoral PON1 expression across all cohorts. Integration of radiomics signatures with clinical variables further improved discrimination, achieving the highest accuracy and net clinical benefit.
ConclusionPON1 downregulation contributes to oxidative stress–driven senescence and unfavorable clinical outcomes in HCC. SR-enhanced radiomics provides an accurate, non-invasive method for estimating tumoral PON1 expression, demonstrating potential value for radiogenomic profiling and preoperative risk stratification.
Graphical Abstract