Comparative Study Of Non-invasive Models For Diagnosing Liver Fibrosis In Chronic Hepatitis B Patients

Objective: Evaluation and comparison the diagnostic efficacy of fourteen noninvasive diagnostic models(AAR?APind?APRI?CDS?Doha score?FCI?FI?FIB-4? GP?GPR?King 's score?RPR?S index?transient elastography)and combined application in the diagnosis of significant liver fibrosis of chronic hepatitis B patients with normal liver function to explore the reliable methods of clinical diagnosis of significant liver fibrosis.Methods: A total of 67 chronic hepatitis B patients with normal liver function and complete medical records who had undergone a liver biopsy,between January 1st,2009 and March 31 th,2019 in the department of hepatobiliary and pancreatic medicine of the First Hospital of Jilin University were enrolled in our study.The basic characteristics,serum indexes,liver hardness and the pathological data were collected and recorded.According to the Scheuer scoring system(GS grade),the patients were divided into no significant liver fibrosis group(S0-S1)and significant liver fibrosis group(S2-S4).The general clinical characteristics of patients,the correlation of each index and liver fibrosis were analysised.The cut-off,the specificity,the sensitivity and area under receive-operating curve of fourteen noninvasive diagnostic models in the diagnosis of liver fibrosis were calculated.And we combined thirteen noninvasive serological models with transient elastography to calculate the above indexes again.And the data were analyzed and compared with each other by statistical methods.Results: 1.AAR,GP,GPR,S index and TE showed significant positive correlation with liver fibrosis(P<0.05),the correlation coefficients were 0.269,0.255,0.246,0.278 and 0.370,respectively.The strongest one was TE,followed by S index,AAR,GP and GPR.The other indexes had no significant correlation with liver fibrosis.2.The AUROC of AAR,GP,GPR,S index and TE were of statistical significance(P<0.05).The highest one was TE,AUROC was 0.683,followed by S index,AAR,GP and GPR,AUROC were 0.673,0.667,0.658 and 0.653,respectively.3.When AAR and TE were combined in series,the AUROC was highest among the serial groups which was up to 0.759.When AAR and TE were combined in parallel,the AUROC was highest among the parallel groups which was up to 0.707.Compared with the AUROC of the optimal diagnosis models of the three diagnostic methods(separate,series and parallel),we could know that AAR cascade TE>GPR parallel TE>TE,and the difference was not statistically significant.4.When King's score and TE were connected in series to used to diagnose significant liver fibrosis,its specificity was better,up to 95.7%,which was higher than that of single diagnosis.5.When FIB-4 parallel TE,GP parallel TE,GPR parallel TE and S index parallel TE were used to diagnose significant liver fibrosis,their sensitivity were good,up to 100%,greater than that of single diagnosis.Conclusions: 1.Compared the ability of thirteen serological models with TE in individual,serial,parallel diagnosis of significant liver fibrosis of chronic hepatitis B patients with normal liver function.TE,AAR connect TE in series,GPR parallel TE were the optimal diagnosis models of the three diagnostic methods,and AAR connect TE in series was the best one,but did not significantly higher than that of TE,GPR parallel TE,and the diagnosis was only moderately accurate,which could not completely replace liver biopsy.2.Serial models could improve the specificity of liver fibrosis diagnosis,while parallel models could improve the sensitivity of liver fibrosis diagnosis.When King's score and TE were connected in series to used to diagnose significant liver fibrosis,the specificity was 95.7%.When FIB-4 parallel TE,GP parallel TE,GPR parallel TE and S index parallel TE were used to diagnose significant liver fibrosis,their sensitivity could reach 100%.