Department of Hepatology, Toranomon Hospital, Okinaka Memorial Institute for Medical Research, Tokyo, Japan; Okinaka Memorial Institute for Medical Research, Tokyo, Japan.
In order to evaluate and judge a fibrotic stage of patients with chronic hepatitis B, multivariate regression analysis was performed using multiple fibrosis markers.
A total of 227 patients from seven hepatology units and institutes were diagnosed by needle biopsy as having chronic liver disease caused by hepatitis B virus. Twenty-three variables and their natural logarithmic transformation were employed in the multivariate analysis. Multiple regression function was generated from data of 158 patients in one hospital, and validation was performed using the other data of 69 patients from six other hospitals.
After stepwise variable selection, multivariate regression analysis finally obtained the following function: z = 1.40 × ln (type IV collagen 7S) (ng/mL) - 0.017 × (platelet count) (×1000(3) /mm(3) ) + 1.24 × ln (tissue inhibitor of matrix metalloproteinase-2) (ng/mL) + 1.19 × ln (α-2-macroglobulin) (mg/dL) - 9.15. Median values of fibrosis scores of F1 (n = 73), F2 (n = 42), F3 (n = 31) and F4 stages (n = 12) were calculated as 0.95, 2.07, 2.98 and 3.63, respectively. Multiple regression coefficient and coefficient of determination were 0.646 and 0.418, respectively. Validation with patient data from other institutions demonstrated good reproducibility of fibrosis score for hepatitis B (FSB), showing 1.33 in F1 (n = 27), 2.20 in F2 (n = 20), 3.11 in F3 (n = 20) and 5.30 in F4 (n = 2), respectively.
A concise multiple regression function using four laboratory parameters successfully predicted pathological fibrosis stage of patients with hepatitis B virus infection.