Evaluation Of Five Noninvasive Models In The Diagnosis Of Hepatic Fibrosis In Chronic Hepatitis B Virus Infection

Objective: To evaluate the diagnostic value of noninvasive diagnostic models APRI,FIB-4,Forns,FibroIndex and GP in the diagnosis of hepatic fibrosis stages in chronic HBV infection,and to provide scientific basis for clinical application.Materials and methods:The data of all cases of chronic HBV infection who received liver biopsy in the Department of Infectious Diseases of our hospital from October 2013 to December 2017 were collected.After screening according to the inclusion and exclusion criteria,432 cases of patient data were finally obtained.The corresponding values of each patient were calculated according to the five diagnostic models of liver fibrosis(APRI,FIB-4,Forns,FibroIndex,GP).The study population was divided into HBeAg-positive group and HBeAg-negative group according to HBeAg status,and divided into ALT normal group,ULN<ALT≤2×ULN group,and 2×ULN<ALT group according to ALT levels.Then the pathological diagnosis of liver fibrosis was used as the gold standard to evaluate the value of five non-invasive hepatic fibrosis models in the diagnosis of liver fibrosis in the study population and five subgroups.The pathological staging criteria for liver fibrosis were based on the Metavir scoring system-fibrosis staging score,F0-F1 was defined as non-significant fibrosis and F2-F4 was defined as significant fibrosis.The diagnostic value of non-invasive models was evaluated using ROC curve analysis,diagnostic test evaluation indicators(sensitivity,specificity,negative predictive value,negative likelihood ratio,positive predictive value,positive likelihood ratio,coincidence rate).The area under the ROC curve(AUROC)of 0.5 to 0.7 is a low diagnostic accuracy,AUROC of 0.7 to 0.9 is a medium diagnostic accuracy,and AUROC>0.9 is a high diagnostic accuracy(P< 0.05).The cut-off values for the diagnosis of significant liver fibrosis in the five models was defined by the ROC curve.The value corresponding to the maximum Youden index was taken as the cut-off value.The statistical software package uses spss20.0 and MedCalc for statistical analysis of data.Result:1.Among the stupulatdy poion(432 patients),males and females were respectively 260 and 172 cases,accounted for 60.19% and 39.18%;the range of age was 13-68 years(median 37,interquartile range 16).2.The distribution of liver fibrosis’ s pathological stage in the study population: F0,F1,F2,F3 and F4 are respectively 30,281,85,30 and 6 cases,accounted for 6.94%,65.05%,19.68%,6.94% and 1.39%.There were non-significant fibrosis and significant fibrosis respectively 311 and 121 cases,accounted for 71.99% and 28.01%.3.Among the study population,HBeAg positive and HBeAg negative patients were respectively 202 and 270 cases,accounted for respectively 46.76%,53.24%.normal ALT,ULN<ALT≤2×ULN and 2×ULN>ALT of were respectively 270,103,59 cases,accounted for 62.50%,23.84% and 13.66%.4.Evaluation of the value of five models in the diagnosis of liver fibrosis in the overall study populationThe corrected AUROC of model APRI,FIB-4,Forns,FibroIndex,and GP in the overall study population was 0.788,0.733,0.729,0.791,and 0.740,respectively(P<0.05).Comparison of the AUROCs of each model revealed that APRI has higher diagnostic accuracy than FIB-4,and FibroIndex has higher diagnostic accuracy than Forns and FIB-4.The sensitivity of the five models to diagnose the patients with significant hepatic fibrosis was highest with FibroIndex being 69.42%;the specificity was highest with Forns being 76.53%;the positive predictive value was highest with APRI being 42.68%;the negative predictive value was highest with FibroIndex It is 83.70%;the compliance rate is highest for Forns,which is 66.90%.5.Evaluation of the value of five models in the diagnosis of liver fibrosis in HBeAg-positive patientsThe corrected AUROC for models of APRI,FIB-4,Forns,FibroIndex,and GP for the diagnosis of HBeAg-positive patients with liver fibrosis were 0.813,0.785,0.760,0.823,and 0.762,respectively(P<0.05).A pairwise comparison of the AUROCs of the five models revealed that FibroIndex had higher diagnostic accuracy than Forns.The sensitivity of fibrosis in the five model diagnosis researchers was highest with FibroIndex being 80.33%;specificity was highest with GP,80.85%;positive predictive value was highest with GP,50.91%;negative predictive value was highest with FibroIndex.87.37;The highest GP rate was 70.30%.6.Evaluation of the value of five models in the diagnosis of liver fibrosis in HBeAg-negative patientsThe AUROC of the model FIB-4 in the diagnosis of hepatic fibrosis in the HBeAg-negative group was 0.584(P>0.05),suggesting no diagnostic value.The corrected AUROC for model APRI,Forns,FibroIndex,and GP diagnosed HBeAg-negative patients were 0.758,0.710,0.755,and 0.720,respectively(P<0.05).There was no significant difference in the AUROC of each model.The sensitivity of the four models in the diagnosis of significant hepatic fibrosis was 80% for GP,80.8% for APRI,43.33% for positive predictive value,and highest for FibreIndex with negative predictive value.It is 85.34%;the highest rate of compliance is APRI,which is 70.43%.7.Evaluation of five models in the diagnosis of liver fibrosis in patients with normal ALTThe corrected AUROC for model APRI,FIB-4,Forns,FibroIndex,and GP for diagnosis of ALT in the normal ALT group were 0.792,0.763,0.755,0.787,and 0.779,respectively(P<0.05),and the AUROC of each model was compared.No significant difference was found.The sensitivity of the five models in diagnosing significant hepatic fibrosis was highest in FIB-4(67.19%);the highest in specificity was GP(83.98%);the highest positive predictive value was in GP(46.77%);the negative predictive value was FibroIndex.The highest was 85.90%;the highest GP rate was 74.81%.8.Evaluation of five models in the diagnosis of liver fibrosis in patients with ULN<ALT≤2×ULNThe AUROC of patients with significant liver fibrosis in the models of APRI,FIB-4,Forns and GP diagnosed in the ULN<ALT≤2×ULN group was 0.576,0.584,0.600,and 0.586,respectively(P>0.05),indicating no diagnostic ability.The accuracy of model FibroIndex in the diagnosis of ULN<ALT≤2×ULN patients with significant hepatic fibrosis was 0.779(P<0.05),sensitivity was 80.00%,specificity was 55.41%,negative predictive value was 87.23%,positive predictive value 42.11%,the compliance rate is 62.50%.9.Evaluation of the value of five models in the diagnosis of hepatic fibrosis in patients with ALT>2×ULNThe AUROC of the model APRI,FIB-4,Forns,FibroIndex,and GP in the diagnosis of significant liver fibrosis in the ALT>2×ULN group were 0.484,0.531,0.506,0.559,and 0.504(P>0.05),suggesting no diagnostic value.Conclusion:There are differences in the diagnostic value of the five models in the diagnosis of patients with or without significant hepatic fibrosis in different groups of patients.Therefore,we should make the right choices in clinical application in order to make the non-invasive model play the biggest role.1.In patients with chronic HBV infection,the diagnostic accuracy of the results of comprehensive APRI and FibroIndex diagnosis of patients with or without significant fibrosis can be high.2.In the HBeAg-positive group,the diagnosis of fibrotic patients with or without FibroIndex and GP diagnoses has a significant diagnostic value.In the HBeAg-negative group,the combined APRI,FibroIndex,and GP have better diagnostic results.3.In patients with normal ALT,patients with comprehensive FIB-4,FibroIndex,and GP diagnosed had significant liver fibrosis;in ULN<ALT≤2×ULN patients,only FibroIndex had diagnostic value;In the ALT>2×ULN group,the five models had no significant diagnostic value for patients with or without significant hepatic fibrosis.