We read with great interest the article by Feldstein et al. reporting the potential usefulness of cytokeratin-18 (CK-18) fragment as a noninvasive serum biomarker for diagnosing nonalcoholic steatohepatitis.1 We would like to draw attention to similar studies on serum CK-18 fragment in the differentiation of alcoholic steatohepatitis from healthy controls with no liver disease. Cytokeratins (CKs) are normal constituents of the epithelial cell cytoskeleton.2 Serum CK-18 level is a clinical tool useful as a tumor marker in epithelial malignancies.3 However, serum CK-18 level has also been found to increase in nonmalignant diseases, such as alcoholic hepatitis, which limits its specificity as a tumor marker.4–6 Thus, CK-18 fragment level may be a noninvasive biomarker for early detection of alcoholic steatohepatitis.
We tested the serum CK-18 fragment levels in patients with alcoholic hepatitis (50), hepatocellular carcinoma (50), heavy drinkers (50), and healthy controls (50). Our results showed that serum levels of CK-18 fragment in patients with alcoholic hepatitis were higher than those of healthy controls and heavy drinkers, and even tended to be higher than those of patients with malignancy. Serum CK-18 fragment levels were median 27 U/L (range = 11-72 U/L) in controls with no liver disease, median 759 U/L (range = 152-4739 U/L) in heavy drinkers, median 1598 U/L (range = 531-4237 U/L) in patients with alcoholic hepatitis, and median 449 U/L (range = 21-17,326 U/L) in those with hepatocellular carcinoma (P < 0.001 for alcoholic hepatitis versus healthy controls). However, serum CK-18 fragment levels in heavy drinkers were similar or even higher than those observed in patients with advanced malignancy, which is further evidence for the belief that the diagnostic value of serum CK-18 fragment as a tumor marker is limited by those heavy drinkers. Our results are very similar to Prof. Fayetteville's previous report.
In summary, our data highlight three points. First, we show that serum CK-18 fragment is a better noninvasive biomarker for alcoholic steatohepatitis, and is not only limited to nonalcoholic steatohepatitis. Second, the sample size is not large enough and this did not allow us to establish the performance of CK-18 fragment according to the etiology of underlying liver disease. Third, considering that the genotypes of different geographic or ethnic groups may have a significant impact on the serum CK-18 fragment levels, more multicenter cohorts of validation are still needed before this marker can be applied clinically.