Abstract
BACKGROUND AND AIMS: The progression of alcohol-associated liver disease (ALD) in its early precirrhotic stages can be a silent process. Serum keratin 18 levels (K18-M65) predict severe events and mortality in advanced stages of ALD, but data on this biomarker in early stages are scarce. We evaluated the diagnostic accuracy of K18-M65 levels in identifying early forms of ALD. METHODS: We prospectively evaluated two cohorts of actively drinking patients with alcohol use disorder (AUD) following a rehabilitation program (training (n = 162) and validation (n = 78)) and matched healthy controls (n = 21). Clinical, laboratory, and imaging data were used to distinguish AUD patients with simple steatosis (minimal ALD) and steatohepatitis/fibrosis (early ALD). We measured serum K18-M65 levels and assessed their ability to predict early ALD. RESULTS: High levels of K18-M65 characterized AUD patients with early ALD, while levels in the minimal ALD group were similar to those in healthy controls. K18-M65 levels distinguished minimal liver disease from early ALD (AUROC = 0.8704; p < 0.0001) with an optimal cutoff at 265.9 U/L. K18-M65 levels strongly correlated with transaminases and predicted early ALD (OR 25.81; 95% CI 3.166-336.1; p < 0.0001), controlled attenuation parameter, and liver stiffness independently from transaminases and other potential confounders. K18-M65 levels did not discriminate between fibrosis and steatohepatitis but correlated with histological signs of hepatocellular injury and inflammation (all p < 0.05). The K18-M65 cutoff detected early ALD in the validation cohort with high accuracy (sensitivity 86.67%, specificity 96.67%) and a very high positive likelihood ratio (28.6; 95% CI 4.14-197.73). CONCLUSIONS: Serum K18-M65 levels can be used as a biomarker to detect early ALD stages with excellent predictive value.