After goodbye? Dead hepatocytes as a biomarker for fibrosis and steatohepatitis


  • Steve Choi M.D.,

    1. Division of Gastroenterology, Duke University School of Medicine, Durham, NC
    2. Section of Gastroenterology, Durham Veterans Affairs Medical Center, Durham, NC
    Search for more papers by this author
  • Anna Mae Diehl M.D.

    Corresponding author
    1. Division of Gastroenterology, Duke University School of Medicine, Durham, NC
    • Division of Gastroenterology, Duke University School of Medicine, 595 LaSalle Street, Suite 1073, DUMC 3256, Durham, NC 27710
    Search for more papers by this author
    • fax: 919-684-4183

  • Potential conflict of interest: Nothing to report.

See Article on Page 455


ALT, alanine aminotransferase; CK-18, cytokeratin 18; ELISA, enzyme-linked immunoassay; M65ED, M65 EpiDeath; NAFL, nonalcoholic fatty liver; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis.

To optimize the outcomes of patients with liver disease, physicians and researchers need better noninvasive tests for reliably detecting subclinical liver injuries and monitoring the progression/regression of liver damage. Joka et al.1 report exciting evidence showing that such tests are being perfected. The new assays quantify cytokeratin 18 (CK-18), an epithelial cytoskeletal protein that is released into the blood when hepatocytes die.

CK-18 is a substrate for caspases that are activated during apoptosis. Fragmented CK-18 accumulates in apoptosing cells and then is released into the blood.1 Because many liver injuries increase hepatocyte apoptosis, an enzyme-linked immunoassay (ELISA) for detecting caspase-cleaved CK-18 was developed with M30, an antibody that recognizes a specific caspase-cleaved CK-18 epitope.1 Several groups have examined the ability of the M30 ELISA to detect subclinical liver disease and measure disease severity, and there has been much focus on patients with fatty liver disease and particularly nonalcoholic fatty liver disease (NAFLD). The M30 ELISA is superior to alanine aminotransferase (ALT) for distinguishing nonalcoholic steatohepatitis (NASH) from simple steatosis [nonalcoholic fatty liver (NAFL)] because the levels of M30-reactive CK-18 correlate better with the histological stage of liver fibrosis than ALT, but it has similarly been found to have important limitations. It is not reliable for differentiating individuals with little hepatic fat from those with more significant steatosis or for stratifying patients according to the fibrosis stage.1

Accumulating evidence indicates that apoptosis is not the only mechanism of cell death in injured livers. Cellular necrosis, necroptosis (combined necrotic and apoptotic death), autophagy, and other mechanisms are involved.2 Because cytoskeletal proteins are released from dying hepatocytes, assaying for both cleaved and uncleaved (total) CK-18 might improve the detection of liver cell death and thereby refine the assessment of related phenomena, including steatosis and fibrosis. To evaluate this possibility, the performance of the M30 ELISA was compared with the performance of ELISAs based on antibodies that react with two different epitopes of CK-18 independently of the cleavage status. One of these total CK-18 ELISAs (M65) uses the M6 antibody to capture the CK-18 antigen and the M5 antibody to detect the bound CK-18 antigen; the other ELISA [M65 EpiDeath (M65ED)] uses the M5 antibody to capture CK-18 and the M6 antibody for detection. Earlier work has suggested that the binding specificity of the M65ED assay for CK-18 is superior to that of the M65 assay (with lower signals in healthy controls).

The present study compared the sensitivities and specificities of three assays for predicting the severity of steatosis and fibrosis in 121 patients with chronic liver disease. Viral hepatitis was the underlying cause of liver disease for more than half of the cohort (approximately 15% had NAFLD/NASH). In addition to ALT measurements and liver histology findings, elastography data were available for 107 of the 121 patients. Serum ELISA results were compared to findings from 18 healthy controls and 200 blood donors (the real-life cohort). The results demonstrated the advantages of assaying for total CK-18 (versus cleaved CK-18). First, total CK-18 proved to be the better liver fibrosis biomarker. Although the levels of cleaved CK-18 correlated with the fibrosis stage and the liver stiffness, a regression analysis demonstrated significantly stronger correlations with total CK-18 levels (particularly M65ED). A receiver operating characteristic plot analysis confirmed this finding: total CK-18 levels ≥ 353 U/L in the M65ED assay correctly predicted fibrosis stages ≥ F2 with a sensitivity of 74% and a specificity of 68%, whereas the optimal cutoff value for the M30 assay (157.5 U/L) was 64% sensitive and 61% specific. Although all the assays differentiated advanced fibrosis (Ishak stages F5-F6) from lower stage fibrosis, only M65ED distinguished between various stages of liver fibrosis and separated low-level fibrosis (F0-F1) from intermediate-level fibrosis (F2-F4) and intermediate-level fibrosis from high-level fibrosis (F5-F6). Second, total CK-18 levels were a better biomarker of NAFL/NASH than cleaved CK-18 levels. Only total CK-18 assays differentiated patients with minimal steatosis (<10% hepatic fat accumulation) from those with more hepatic fat (>10% steatosis). All three assays distinguished patients with steatosis from healthy controls. A selective analysis of the NAFLD subgroup demonstrated that the total CK-18 assays reliably differentiated patients with mild NAFL from healthy or real-life controls, whereas the M30 assay could not. The total CK-18 assays were more sensitive (100% versus 75%) and specific (80% versus 70%) than the M30 ELISA for discriminating between NAFL and NASH. The better classification of NAFL/NASH by the total CK-18 assays was not a byproduct of superior fibrosis staging because the fibrosis severity was similar (generally low) in these patients. Although the NASH patients tended to have higher ALT levels than the NAFL patients, a regression analysis revealed that total CK-18 levels predicted NASH independently of ALT levels, whereas cleaved CK-18 levels did not.

The aggregated data demonstrate that a noninvasive measure of various types of hepatocyte death (total CK-18) is a better biomarker of related liver pathology than a test that merely reflects apoptotic severity (cleaved CK-18). This finding has a number of fundamental implications. First, it supports the concept that dying liver epithelial cells provide key fibrosis stimuli. Completely dead hepatocytes have long been implicated in the pathogenesis of liver fibrosis because of increased fibrogenic activity in hepatic stellate cells that have phagocytosed apoptotic hepatocytes.3 More recent data show that dying (but viable) liver epithelial cells produce and release soluble factors that promote liver fibrosis. Diverse insults that sensitize hepatocytes to death (e.g., an infection with hepatitis C virus,4 an exposure to agents that induce endoplasmic reticulum stress,5 or an inhibition of autocrine viability factors6) induce them to generate damage-associated molecules. These include hedgehog ligands, which are morphogens stimulating wound-healing mechanisms that promote myofibroblastic cell outgrowth, immune cell infiltration, and the accumulation of liver epithelial progenitors (with fibrogenic activity).7, 8 It is thus not surprising that an assay with improved sensitivity and specificity for detecting hepatocyte death would provide better sensitivity and specificity for responses that are triggered by all dying hepatocytes (and not simply a subset of dead cells). Second, the latter concept raises the intriguing possibility that dying hepatocytes promote hepatic steatosis (rather than vice versa as currently believed). This possibility is supported by evidence showing that hepatic steatosis occurs transiently in remnant livers after partial hepatectomy, which is another process stimulating wound-healing responses to promote liver regeneration.8 Third, the fact that levels of circulating CK-18, which is a highly sensitive and specific biomarker of liver epithelial death, are less than perfect for predicting the severity of coincident fibrosis suggests the importance of mechanisms that control the response to liver epithelial cell death.

This study showed that the area under the receiver operating characteristic curve for the total CK-18 prediction of a liver fibrosis stage ≥ F2 was 0.73, which is similar to the values reported for various assays based on multiple putative fibrosis biomarkers.9 This proves that a single good biomarker of liver epithelial death (total CK-18) provides a fairly robust readout of liver fibrosis, which is a complex injury response presumably captured by the other assays. This suggests that an active liver injury (often subclinical) is the main force perpetuating liver fibrosis in most individuals and provides a reason for optimism because it supports the dynamic nature of fibrogenesis/fibrinolysis and the inherent reversibility of liver fibrosis. Like more complex assays, the total CK-18 ELISA reliably differentiates advanced fibrosis from no fibrosis, but it is less discriminating at lower fibrosis stages; this reflects the dynamic nature of fibrosis and the fact that no available assay captures both sides of the fibrosis equation. Total CK-18 assays measure the predominant liver fibrosis stimulus (i.e., liver epithelial cell death), whereas other assays largely reflect various aspects of the resultant wound-healing response. The use of both stimulus and response assays might provide complementary/additive information that could perfect the noninvasive staging of fibrosis. In other words, combining an assessment of the strength of the fibrosis stimulus (CK-18) with an estimate of the intensity of the fibrogenic response (fibrosis markers or elastography) might permit individuals with given levels of liver cell death to be stratified into groups of hyporesponders, normoresponders, and hyperresponders with respect to wound healing. Additional power might be obtained by the serial acquisition of such information from given individuals. Further research is needed to test and refine these concepts. Success offers exciting opportunities for personalizing liver disease management and facilitating the discovery of effective antifibrotic therapies.