ALBERT

Description: Evaluating liver fibrosis is crucial for disease severity assessment, treatment decisions, and hepatocarcinogenic risk prediction among patients with chronic hepatitis C. In this retrospective multicenter study, we aimed to construct a novel model formula to predict cirrhosis. A total of 749 patients were randomly allocated to training and validation sets at a ratio of 2:1. Liver stiffness measurement (LSM) was made via transient elastography using FibroScan. Patients with LSM ≥12.5 kPa were regarded as having cirrhosis. The best model formula for predicting cirrhosis was constructed based on factors significantly and independently associated with LSM (≥12.5 kPa) using multivariate regression analysis. Among the 749 patients, 198 (26.4%) had LSM ≥12.5 kPa. In the training set, multivariate analysis identified logarithm natural (ln) type IV collagen 7S, ln hyaluronic acid, and ln Wisteria floribunda agglutinin positive Mac-2-binding protein (WFA+-Mac-2 BP) as the factors that were significantly and independently associated with LSM ≥12.5 kPa. Thus, the formula was constructed as follows: score = −6.154 + 1.166 × ln type IV collagen 7S + 0.526 × ln hyaluronic acid + 1.069 × WFA+-Mac-2 BP. The novel formula yielded the highest area under the curve (0.882; optimal cutoff, −0.381), specificity (81.5%), positive predictive values (62.6%), and predictive accuracy (81.6%) for predicting LSM ≥12.5 kPa among fibrosis markers and indices. These results were almost similar to those in the validated set, indicating the reproducibility and validity of the novel formula. The novel formula scores were significantly, strongly, and positively correlated with LSM values in both the training and validation data sets (correlation coefficient, 0.721 and 0.762; p = 2.67 × 10−81 and 1.88 × 10−48, respectively). In conclusion, the novel formula was highly capable of diagnosing cirrhosis in patients with chronic hepatitis C and exhibited better diagnostic performance compared to conventional fibrosis markers and indices.