PeerJ · 2025
Mendelian randomization and nomogram-based prediction of hepatocellular carcinoma risk in patients with hepatitis B cirrhosis.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Zheng Xiaolong, Hong Yiping, Wei Wei
Аннотация
BACKGROUND: To innovatively integrate genetic causality and multidimensional clinical indicators, we aimed to investigate causal relationships between metabolic-inflammatory biomarkers and hepatocellular carcinoma (HCC) risk in hepatitis B-related cirrhosis (HBV-C) using Mendelian randomization (MR), and develop a precision prediction model combining genetic evidence with nonlinear biochemical dynamics.
METHODS: Leveraging bidirectional approaches, we first performed two-sample MR analysis on GWAS datasets (UK Biobank, n = 456,348) to establish causality between low-density lipoprotein (LDL) and HCC. In a retrospective cohort of patients with HBV-related cirrhosis from our institution (n = 147; 2022-2024), we identified nonlinear LDL-HCC thresholds via restricted cubic splines (RCS) and engineered a novel "A-index" (a composite score derived from principal component analysis (PCA) integrating alpha-fetoprotein (AFP), aspartate aminotransferase (AST), and alanine aminotransferase (ALT)). Machine learning-driven logistic regression synthesized LDL, A-index, and clinical predictors into a nomogram, rigorously validated by area under the curve-receiver operating characteristic (AUC-ROC), calibration curves, and decision curve analysis (DCA).
RESULTS: MR analysis revealed a robust causal link between reduced LDL levels and elevated HCC risk (OR = 0.472, 95% CI [0.259-0.860]; P = 0.014), with RCS identifying a critical LDL threshold at 2.28 mmol/L-below which HCC risk escalated exponentially. The PCA-synthesized A-index outperformed individual biomarkers (AUC = 0.652 vs. AFP = 0.579). The final nomogram integrating LDL dynamics, A-index, age, sex, prothrombin time, and antiviral therapy achieved exceptional discrimination (AUC = 0.938) and clinical net benefit across risk thresholds.
CONCLUSION: This study introduces a novel causal inference-guided prediction model, addressing the long-standing debate on LDL's dual role in hepatocarcinogenesis. By integrating MR-validated genetic causality, nonlinear biochemical modeling, and PCA-driven dimensionality reduction, our model provides a transformative tool for personalized HCC risk stratification in HBV-C patients.
METHODS: Leveraging bidirectional approaches, we first performed two-sample MR analysis on GWAS datasets (UK Biobank, n = 456,348) to establish causality between low-density lipoprotein (LDL) and HCC. In a retrospective cohort of patients with HBV-related cirrhosis from our institution (n = 147; 2022-2024), we identified nonlinear LDL-HCC thresholds via restricted cubic splines (RCS) and engineered a novel "A-index" (a composite score derived from principal component analysis (PCA) integrating alpha-fetoprotein (AFP), aspartate aminotransferase (AST), and alanine aminotransferase (ALT)). Machine learning-driven logistic regression synthesized LDL, A-index, and clinical predictors into a nomogram, rigorously validated by area under the curve-receiver operating characteristic (AUC-ROC), calibration curves, and decision curve analysis (DCA).
RESULTS: MR analysis revealed a robust causal link between reduced LDL levels and elevated HCC risk (OR = 0.472, 95% CI [0.259-0.860]; P = 0.014), with RCS identifying a critical LDL threshold at 2.28 mmol/L-below which HCC risk escalated exponentially. The PCA-synthesized A-index outperformed individual biomarkers (AUC = 0.652 vs. AFP = 0.579). The final nomogram integrating LDL dynamics, A-index, age, sex, prothrombin time, and antiviral therapy achieved exceptional discrimination (AUC = 0.938) and clinical net benefit across risk thresholds.
CONCLUSION: This study introduces a novel causal inference-guided prediction model, addressing the long-standing debate on LDL's dual role in hepatocarcinogenesis. By integrating MR-validated genetic causality, nonlinear biochemical modeling, and PCA-driven dimensionality reduction, our model provides a transformative tool for personalized HCC risk stratification in HBV-C patients.