Liver Steatosis Replaced With Non-Invasive Viral and Host Parametars Can Serve as Negative Predictive Model in Patients with Chronic Hepatitis-C

Ana Višnjić, Zvonimir Ostojić, Irena Hrstić, Marijana Ćorić, Marina Premužić

Abstract


Almost 70% of chronic hepatitis C (CHC) patients will have concomitant hepatic steatosis (HS) usually determined with invasive method. HS serve as negative predictive factor for lower sustaind viral responce (SVR) in CHC patients treated with standard of care (SOC) (PEG-IFN and Rib). Retrospective analysis of biochemical, virological and histological data in CHC patients treated with PEG-IFN and Ribavarin. Statistical analysis was carried out by Biometrika Healthcare Research. Level of significance was set to 95% (p<0.05). 72 patients (43 M; 29 F; median age 41y) with CHC (60 G1; 12 G3) with no concomitant metabolic syndrome were analyzed. HS ranged from 5 to 30% (median 15%). Overall accuracy of prediction of SVR based on the levels of HS was AUC=0.71 (95% CI=0.58-0.84; P=0.005). When HS was split regarding cut-off value of 5% significant difference was found between responders and non-responders to treatment (χ2 = 10.025; df=1; P=0.002). Overall sensitivity was 48% and specificity 91%. Conventional predictive variables (gender, age, fibrosis and genotype) where combined with HS (>5%) and all together achieved Nagelkerke R squared of 34.0% in prediction of SVR, with accuracy rate of 75.0%. Further, invasive variables (fibrosis and HS) where replaced with viremia and body mass index (BMI). All noninvasive variables together achieved Nagelkerke R squared of 26.5% in prediction of SVR with 74% accuracy rate of the logistic regression model. Very low HS (<5%) is negative predictor of SVR and can be replaced with noninvasive variables (gender, age, viremia and BMI) with same accuracy rate of the logistic regression model.


Keywords*


Chronic hepatitis C, Hepatic steatosis, Hepatitis C Chronic/pathology, Noninvasive parameters, Adult

Full Text:

PDF


Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.