Objectives: To evaluate the performance of a new mathematical model γ-glutamyl transpeptidase to cholinesterase and platelet ratio (GCPR) versus γ-glutamyl transpeptidase to platelet ratio (GPR) in predicting significant fibrosis and cirrhosis of chronic hepatitis B.
Methods: A complete cohort of 2343 patients was divided into early and late cohort depending on the time of liver biopsy. With reference to the Scheuer standard, liver pathologic stage 2 or higher and stage 4 or higher were defined as significant fibrosis and cirrhosis, respectively. Receiver operating characteristic (ROC) curve was used to evaluate the performance of investigated models.
Results: In the early cohort, the areas under ROC curves (AUROCs) of GCPR in predicting significant fibrosis of hepatitis B e antigen (HBeAg)-positive and HBeAg-negative patients (0.782 and 0.775) were both significantly greater than those of GPR (0.748 and 0.747) (Z = 8.198 and Z = 6.023, both p <0.0001); the AUROCs of GCPR in predicting cirrhosis of HBeAg-positive and HBeAg-negative patients (0.842 and 0.861) were both significantly greater than those of GPR (0.802 and 0.823) (Z = 6.686 and Z = 6.116, both p <0.0001). In early, late and complete cohorts, using a single cutoff of GCPR > 0.080, the specificities of GCPR in predicting significant fibrosis of HBeAg-positive patients were 83.3%, 88.2% and 85.0% and of HBeAg-negative patients were 87.6%, 87.4% and 87.6%, respectively; and the sensitivities of GCPR in predicting cirrhosis of HBeAg-positive patients were 81.9%, 88.7% and 84.2% and of HBeAg-negative patients were 83.1%, 82.1% and 82.7%, respectively.
Conclusions: GCPR has higher performance than GPR in predicting significant fibrosis and cirrhosis of chronic hepatitis B.
【저자키워드】 fibrosis, mathematical model, Chronic Hepatitis B, noninvasive diagnosis, γ-glutamyl transpeptidase to platelet ratio., γ-glutamyl transpeptidase to cholinesterase and platelet ratio,