Serum cytokeratin 18 fragment level as a noninvasive biomarker for non-alcoholic fatty liver disease

Background and Aim: We evaluated the usefulness of serum cytokeratin 18 fragment (CK18-F) as a non-invasive biomarker in differentiating nonalcoholic steatohepatitis (NASH) from nonalcoholic fatty liver (NAFL) since the prognosis of the 2 diseases differ. Methods: 116 Japanese patients with nonalcoholic fatty liver disease (NAFLD) proven by liver biopsy were studied. Histological findings were classified according to the NAFLD activity score (NAS) proposed by the Nonalcoholic Steatohepatitis Clinical Research Network. The correlation between histological findings and serum CK18-F levels was investigated. Results: Serum CK18-F levels showed a positive correlation with histologic steatosis (rho = 0.271, P = 0.0033), inflammation (rho = 0.353, P = 0.0005), ballooning (rho = 0.372, P = 0.0001), and the total NAS (rho = 0.474, P = 2.68 x 10-7). The serum CK18-F level was significantly lower for NAFL (NAS = 2) than for borderline NASH (NAS of 3-4) or definite NASH (NAS >= 5) (P = 0.0294, P = 1.163 x 10-5, respectively). The serum CK18-F level was significantly higher for definite NASH than for borderline NASH (P = 0.0002). The area under the receiver operating characteristic curve of serum CK18-F to predict the presence of NAFL and definite NASH was 0.762 and 0.757, respectively. The optimal cut-off point of serum CK18-F for NAFL and definite NASH was 230 and 270 U/L, respectively. The sensitivity, specificity, positive predict value, and negative predict value of serum CK18-F for NAFL were 0.89, 0.65, 0.34, and 0.97, and those for definite NASH were 0.64, 0.76, 0.72, and 0.67, respectively. Accuracies of diagnosis for both NAFL and definite NASH were 0.70. Conclusions: Serum CK18-F could be a clinically useful biomarker to discriminate between NAFL and NASH.