Establishment and validation of a simple noninvasive model to predict significant liver fibrosis in patients with chronic hepatitis B.

BACKGROUND

There have been still few valuable noninvasive models that can be used as indirect markers of liver fibrosis in chronic hepatitis B (CHB) infection.

METHODS

In 374 patients with chronic hepatitis B virus infection, the correlation between the conventional parameters and significant fibrosis confirmed by liver biopsy was assessed using univariate analysis and logistic regression. A model was established and assessed by the receiver operating characteristic (ROC) curves. Then it was validated in 108 prospectively enrolled patients. A part of the patients were followed up with cirrhosis as the end point, using survival analysis to assess the prognostic value of the model.

RESULTS

A model named AIAG was constructed consisting of age, international normalized ratio, albumin, and gamma-glutamyltransferase which could discriminate between CHB patients with and without significant fibrosis. The area under ROC curves was 0.842 (95% CI, 0.795-0.888) for the training group (n = 250) and 0.806 (95% CI, 0.730-0.882) for the validation group (n = 124). In the training group, using a cut-off score of <0.32, the presence of significant fibrosis could be excluded with high accuracy (90% negative predictive value); similarly, applying a cut-off score of >0.72, the presence of significant fibrosis could be correctly identified with high accuracy (93% positive predictive value). Similar results have been shown in the internal and external validation groups. In the follow-up study, we found that the AIAG score may have good prognostic values to predict the progression of clinically overt cirrhosis in CHB patients.

CONCLUSIONS

AIAG, a simple marker panel consisting of conventional parameters, could easily predict significant fibrosis with a high degree of accuracy.