Abstract
Objective: To establish and evaluate diagnostic efficacy and applicability of serum Golgi protein (GP) 73 based non-invasive diagnostic model with other conventional serological indicators for compensated stage hepatitis B cirrhosis. Methods: 666 cases with chronic hepatitis B (CHB) who had visited to the Fifth Medical Center of People's Liberation Army General Hospital from January 2010 to December 2017 were selected as the study subjects, and were classified according to compensated stage cirrhosis into clinical and pathological diagnosis group based on whether or not the liver histological examination was performed. A diagnostic model of compensated stage hepatitis B cirrhosis in the clinical diagnosis group was established. The current clinically used diagnostic model of liver cirrhosis, aspartate aminotransferase/platelet ratio index (APRI), fibrosis index (FIB)-4 and liver stiffness measurement (LSM) were compared. Eventually, the diagnostic model was verified step by step by pathological diagnosis group. Results: The area under the receiver operating characteristic curve (AUC) of GP73 and APRI, FIB-4, and LSM for cirrhosis patients in the clinical diagnosis group were 0.842, 0.857, 0.864, and 0.832, respectively. The diagnostic efficiency of the four indicators were of similar (P value > 0.05). A diagnostic model of compensated stage hepatitis B cirrhosis (GAPA) using logistic regression analysis was established: LogitP = 1/ [1 + exp (1.614-0.054 × GP73-0.045 × Age + 0.030 × PLT-0.015 × ALP)]. The AUC of the model was as high as 0.940 and the optimal cut-off value were 0.41. The corresponding diagnostic sensitivity and specificity were 0.92 and 0.82, respectively. The diagnostic efficiency was better than that of APRI, FIB-4, LSM and GP73 alone (P < 0.05). The AUC of GAPA was 0.877 in the pathological diagnosis group, which was similar to the diagnostic efficacy of LSM (0.891) and FIB-4 (0.847) (P > 0.1), but still superior to that of APRI (0.811) and GP73 alone (0.780) (P < 0.001). Conclusion: GAPA, a diagnostic model for compensated stage hepatitis B cirrhosis established in this study, has a good diagnostic efficacy in both the clinical and pathological diagnosis group, and has certain auxiliary diagnostic value in the areas where resources are relatively scarce or where LSM has not been developed.