Proposal of a predictive model for advanced fibrosis containing Wisteria floribunda agglutinin-positive Mac-2-binding protein in chronic hepatitis C.

OBJECTIVE

We aimed to construct a predictive model for advanced fibrosis containing Wisteria floribunda agglutinin-positive Mac-2-binding protein (WFA+ -M2BP) level in patients with chronic hepatitis C (CHC) and to validate its accuracy in an independent cohort.

METHODS

A total of 386 patients with CHC were retrospectively analyzed. For the purpose of this study, we formed a training set (n = 210) and a validation set (n = 176). In the training set, we investigated variables linked to the presence of advanced fibrosis using univariate and multivariate analyses. We constructed a formula for predicting advanced fibrosis and validated its accuracy in the validation cohort. Receiver operating characteristic curve (ROC) analysis was carried out for calculating the area under the ROC (AUROC).

RESULTS

In multivariate analyses, WFA+ -M2BP (P = 0.029) and prothrombin time (PT) (P = 0.018) were found to be significant predictive factors linked to the presence of advanced fibrosis; platelet count (P = 0.098) and hyaluronic acid (P = 0.078) showed borderline statistical significance for the presence of advanced fibrosis. Using these four variables (with the initials MPPH), we constructed the following formula: MPPH score = -3.584 - (0.275 × WFA+ -M2BP) + (0.068 × platelet count) + (0.042 × PT) - (0.005 × hyaluronic acid). In the training and validation sets, MPPH score yielded the highest AUROCs (0.87 and 0.83) for predicting advanced fibrosis among eight serum liver fibrosis markers. Similarly, in the training and validation sets, MPPH score had the highest diagnostic accuracies for predicting advanced fibrosis among eight serum variables (81.4% and 74.4%).

CONCLUSIONS

Our proposed MPPH scoring system can be useful for predicting advanced fibrosis in patients with CHC.