Establishment And Verification Of Non-invasive Diagnostic Model Of Liver Fibrosis In Primary Biliary Cholangitis

Primary biliary cholangitis(PBC)is an autoimmune liver disease characterized by chronic cholestasis and non-suppurative bile duct destruction in the hepatic lobule.It can eventually develop into fibrosis and cirrhosis without treatment.Early diagnosis of liver fibrosis has important clinical significance for the treatment and prognosis of PBC.Liver biopsy is considered the gold standard for the diagnosis of liver fibrosis.However,It is difficult to accept liver biopsy,which is an invasive examination and is expensive.Therefore,the study of non-invasive serological diagnostic methods to predict liver fibrosis in patients with PBC is of importance.Objective:The optimal index was selected from the clinically used serological indicators of PBC patients,and the PBC non-invasive liver fibrosis diagnosis model was constructed to provide a simple,inexpensive and accessible non-invasive liver fibrosis diagnostic model,which may reduce or possibly replace liver puncture.Method:A total of 157 PBC patients who had undergone a liver biopsy between January 1st,2008 and March 31 th,2018 in the First Hospital of Jilin University were enrolled in our study.All common parameters and liver pathological results were analyzed.A total of 118 subjects were included in the study,excluding data loss,viral hepatitis,autoimmune hepatitis,and alcohol-drinking history,and were randomized into a model group(78 patients)and a validation group(40 patients).The patients' histological stages were based on the classifications of the Scheuer's stage.The patients were divided into early fibrosis stage(Stage I and II)and advanced fibrosis stage(Stage IIIand IV)according to the Scheuer's stage.Predictive factors of clinically used indicators for early liver fibrosis and advanced liver fibrosis were screened by univariate analysis and binary logistic regression analysis in the modeling group.Based on this,non-invasive serological diagnostic model of PBC were constructed and then we simply the binary logistic regression model.Receiver's operating characteristic curve(ROC)was drawn from the liver pathological results.The diagnostic performance and verification of the new model,APRI,FIB-4 and RPR were evaluated by ROC curve in the validation group.Results:1.Comparison of baseline characteristics of modeling group and verification groupThere were no statistically significant differences between the model group and the validation group(P=0.808),gender(P=0.538),and histological stage(P=0.928).2.Selecting variates about liver fibrosisIn the model group,Univariate analysis revealed differences in PLT,PDW,TBIL,TBA,and DBIL between early liver fibrosis and advanced liver fibrosis(P=0.000,0.036,0.020,0.006,0.018 and 0.010),indicating that compared with early liver fibrosis PBC,patients with advanced liver fibrosis PBC have lower PLT and CHE,while PDW,TBA,TBIL and DBIL are higher.3.Establishme nt of non-invasive liver fibrosis logistic regression mode l in PBCBinary logistic regression analysis of PLT,PDW,CHE,TBA,DBIL and TBIL(introduction standard is P ? 0.05,rejection criterion is P > 0.10,introduction method is LR stepwise forward method)to establish liver fibrosis diagnostic model A,the final indicators that entered the model A were TBA and PLT.4.Simplifying the binary logistic regression modelThe OR values of TBA and PLT were 1.01 and 0.989,respectively,indicating that TBA is a risk factor for advanced liver fibrosis,but PLT is a protective factor for advanced liver fibrosis.we can simplify the logistic regression model named TPR index(bile acid to platelet ratio),which was established to predict advanced liver fibrosis in PBC.5.Evaluation of the diagnostic value of each model for predicting advanced liver fibrosisThe AUROCs of model A,TPR index,APRI index,FIB-4 index and RPR index for predicting liver fibrosis by ROC curve evaluation were 0.789,0.751,0.666,0.670 and 0.672,respectively,the model A and TPR index have higher a nd better ability predicting advanced liver fibrosis in PBC.6.Verification of TPR index for predicting advanced liver fibrosis in PBCIn the validation group,the TPR index predicts advanced liver fibrosis with an accuracy of 77.5%(31/40),a sensitivity of 0.727,and a specificity of 0.793.According to the consistency test,the Kappa value of the pathological gold standard is 0.48>0.4,that is,the TPR index has good consistency with the pathological results.The APROCs of the TPR index,the APRI index,the FIB-4 index,and the RPR index for predicting advanced liver fibrosis were 0.771,0.715,0.618 and 0.517 respectively.The AUROC of the TPR index was higher than other non-invasive serological models.Conclusions:As a simple,inexpensive and easily accessible non-invasive liver fibrosis diagnostic model,the TPR index can help to reduce liver puncture in some patients and be expected a larger sample size for verification.