[Prognostic value of a logistic regression model based on keratin 18 in patients with HBV-related liver failure]

Objective: To investigate the association between serum keratin 18 (K18) level and prognosis in patients with HBV-related acute-on-chronic liver failure (ACLF). Methods: A total of 120 patients who visited Department of Hepatology in Fuzhou Infectious Disease Hospital and were diagnosed with HBV-related ACLF from December 2012 to March 2014 were enrolled and followed up for 3 months. The patients were divided into death group and survival group. The serum levels of K18 fragments (M30 and M65) were measured and related laboratory data were collected to analyze the differences in M30, M65, M30/M65, and other laboratory markers. Binary logistic regression analysis was performed to screen out independent risk factors for death in patients with HBV-related ACLF, and the corresponding logistic regression model (LRM) was established. Another 51 patients with HBV-related ACLF from April to October, 2014 were enrolled; M30 and M65 were measured and related clinical data were collected to calculate LRM value and validate the diagnostic value of LRM. Results: The 120 patients with HBV-related ACLF were followed up for 3 months, and 40 of them died, resulting in a mortality rate of 33.3%. Compared with the survival group, the death group had significantly higher age, percentage of neutrophils, blood ammonia, international normalized ratio, Model for End-Stage Liver Disease (MELD) score, and M65, as well as significantly lower prothrombin time activity and alpha-fetoprotein level. The death group also had significantly higher incidence rates of underlying diseases and complications such as diabetes, liver cirrhosis, hepatic encephalopathy (HE), pulmonary infection, upper gastrointestinal bleeding, and hepatorenal syndrome than the survival group. Age, HE, upper gastrointestinal bleeding, direct bilirubin (DBil), and M30/M65 were independent risk factors for the prognosis of patients with HBV-related ACLF. The model established was LRM = 0.061 × age + 0.69 × HE + 4.11 × upper gastrointestinal bleeding + 3.201 × ln(DBil) - 3.875 × ln(M30/M65) - 24.248. The 51 patients with HBV-related ACLF were followed up for 3 months, and the LRM value and MELD score were calculated. The areas under the ROC curve for the LRM model and MELD score were 0.889 and 0.858, respectively, and there was no significant difference between them (Z = 0.417, P > 0.05). Conclusion: M30/M65 ratio has a high value in the diagnosis of HBV-related ACLF, and the LRM model containing M30/M65 ratio can well predict the short-term outcome of patients with HBV-related ACLF.