Validation of multiple models to predict progression to hepatocellular carcinoma in patients with HBV on antiviral therapy
Objective To evaluate the performance of 4 common models for predicting the risk of progression to hep-atocellular carcinoma(HCC)in patients with HBV during antiviral therapy.Methods Four common models were ret-rospectively included from Jan.2013 to Jun.2017.A total of 1 376 patients with HBV treated with antiviral therapy at the First Affiliated Hospital of Xi'an Jiaotong University from Jan.2013 to Jun.2017 were included.The patients were divided into 117 cases(8.50%)in the trial group and 1 259 cases(91.50%)in the control group according to whether they had secondary HCC at 5 years of follow-up.The clinical data of all patients were collected by EMR system and CAMD,PAGE-B,APA-B and REAL-B scores were calculated.Risk factors for HCC were analyzed using multivariate Cox regression method.The ROC curve was used to assess the 4 models for predicting HCC in terms of zone.The ROC curves were used to assess the 4 models to predict the degree of HCC.Results Univariate analysis showed that age,di-abetes mellitus,cirrhosis,platelets,erythrocyte width of distribution,alpha-fetoprotein levels and CAMD,PAGE-B,APA-B,REAL-B scores were statistically significant(P<0.05).Multivariate Cox regression analysis showed that the differences in the levels of alpha-fetoprotein,cirrhosis,CAMD,PAGE-B,APA-B,REAL-B were independent risk fac-tors for HCC.The ROC curves showed that the CAMD,PAGE-B,APA-B and REAL-B models predicted the risk of HBV patients' overtreatment with antiviral therapy.The AUCs for progression to HCC during antiviral therapy were 0.719,0.710,0.758 and 0.879,respectively.Conclusion The 4 models have a certain predictive ability for the long-term occurrence of HCC in HBV infected patients treated with antiviral therapy,and REAL-B model has the best predictive effect.
Hepatitis BAntiviral therapyHepatocellular carcinomaPredictive model
NETL
NSTL
万方数据
维普
