A CpG-based prediction model for the diagnosis of hepatocellular carcinoma patients

Abstract

Abstract Objective: Hepatocellular carcinoma(HCC),the most prevalent form of liver cancer, owns high morbidity and mortality. Early diagnosis for HCC is critical for the treatment and prognosis. Early diagnosis plays an important role in the improvement of HCC prognosis. Methods: All clinical characteristics of 233 participants from multicenter were collected, including 115 HCC patients, 103 patients with cirrhosis, and 15 samples from healthy individuals. We identified several indicators significantly associated with HCC morbidity through logistic analysis to develop the prediction model. Further analysis revealed the independent predictive capacity of the predictive model. A nomogram comprising the predictive model was established, and data on 133 patients was utilized for the development of the model and on 100 patients was utilized for the validation. Furthermore, dozens of patients with tumors smaller than 2cm were collected for additional validation. The receiver operating characteristic (ROC) curve analysis confirmed the good performance of the predictive model. Results: As a result, we identified five indicators that were significantly associated with HCC morbidity through univariate analysis and multivariate analysis. The predictive model was consist of age, drinking status and blood indicators, including AFP(alpha-fetoprotein),HBV(hepatitis B virus)infection status and a differential methylation CpG site. All the factors above were incorporated into the nomogramand the application of the nomogram gave good discrimination and good calibration. Calibration curves showed a favorable consistency between the predicted probabilities. ROC curve analysis showed that the nomogram had good discrimination, with AUC of 0.852 and 0.857 in the training group and validation group, respectively. Moreover, decision curve analysis has been implemented to evaluate and compare prediction nomogram. Conclusion: The study provides a novel model for early diagnosis HCC, better than traditional screening and diagnostic indicators.