Prospective biopsy-controlled evaluation of cell death biomarkers for prediction of liver fibrosis and nonalcoholic steatohepatitis


  • Potential conflict of interest: Nothing to report.

  • Supported by the Deutsche Forschungsgemeinschaft (grants BA-2092/9-1, SFB Transregio 77, SFB 685, and SFB 773).


Fibrosis and steatosis are major histopathological alterations in chronic liver diseases. Despite various shortcomings, disease severity is generally determined by liver biopsy, emphasizing the need for simple noninvasive methods for assessing disease activity. Because hepatocyte cell death is considered a crucial pathogenic factor, we prospectively evaluated the utility of serum biomarkers of cell death to predict different stages of fibrosis and steatosis in 121 patients with chronic liver disease. We compared the M30 enzyme-linked immunosorbent assay (ELISA), which detects a caspase-cleaved cytokeratin-18 (CK-18) fragment and thereby apoptotic cell death, with the M65 ELISA, which detects both caspase-cleaved and uncleaved CK-18 and thereby overall cell death. Both biomarkers significantly discriminated patients with different fibrosis stages from healthy controls. However, whereas both markers differentiated low or moderate from advanced fibrosis, only the M65 antigen could discriminate even lower stages of fibrosis. The M65 assay also performed better in distinguishing low (≤10%) and higher (>10%) grades of steatosis. In a subgroup of patients, we evaluated the biomarkers for their power to predict nonalcoholic steatohepatitis (NASH). Importantly, both markers accurately differentiated healthy controls or simple steatosis from NASH. However, only serum levels of M65 antigen could differentiate simple steatosis from healthy controls. Conclusion: Cell death biomarkers are potentially useful to predict fibrosis, steatosis, or NASH. Compared with the widely used apoptosis marker M30, the M65 assay had a better diagnostic performance and even differentiated between lower fibrosis stages as well as between healthy individuals and patients with simple steatosis. (HEPATOLOGY 2012)

Liver fibrosis, inflammation, and steatosis are major features of acute and chronic liver diseases, such as viral hepatitis, autoimmune or metabolic liver diseases, and alcoholic or nonalcoholic steatohepatitis (NASH). An increasingly common chronic disease is nonalcoholic fatty liver disease (NAFLD), ranging from nonalcoholic fatty liver (NAFL) or simple steatosis to progressive NASH and fibrosis.1 Although most patients with steatosis tend to have a benign clinical course, a significant proportion of those with NASH have a progressive disease with a risk of developing cirrhosis and hepatocellular carcinoma.2 The mechanisms of why some patients with simple steatosis progress to NASH, whereas others do not, are only poorly understood.

Prediction of fibrosis and steatosis is essential for the management of patients with chronic liver disease. Although liver biopsy remains the reference standard for disease staging, it is limited by sample errors and the risk of complications.3 Much attention has therefore been focused on whether noninvasive methods can detect clinically significant steatosis, fibrosis, or cirrhosis or can discriminate between simple steatosis and NASH in NAFLD patients.4-6 Several imaging techniques may be used to detect steatosis but are not sufficient to stage liver fibrosis. In addition, several markers including extracellular matrix components or enzymes involved in their degradation or synthesis have been described to predict the degree of fibrosis.7, 8 However, the utility of these markers as predictors of liver damage is limited and controversial. Clinical decision making often requires differentiation of minimal from intermediate stages of fibrosis. The inability of serological fibrosis markers to correctly identify patients with intermediate fibrosis stages has been suggested to be 30%-70%.9 For instance, a combination of different parameters involved in fibrogenesis was shown to accurately detect fibrosis, but the discriminative power between early fibrosis stages was limited.10, 11 Thus, there is an urgent need to develop simple, noninvasive tests that can identify the stage of liver disease and accurately distinguish NASH from simple steatosis.

Increasing evidence suggests an important role for hepatocyte apoptosis in the progression of NALFD and other liver diseases.12, 13 During apoptosis, caspases are activated and cleave various substrates, including cytokeratin-18 (CK-18), a major intermediate filament protein in hepatocytes.14, 15 Apoptosis of hepatocytes is further associated with the release of caspase-cleaved CK18 fragments in the bloodstream. CK-18 cleavage generates a neoepitope that can be detected by the monoclonal antibody M30 and therefore allows the assessment of apoptosis specifically of epithelial cells by an enzyme-linked immunosorbent assay (ELISA). In contrast, another assay, the M65 ELISA, detects both caspase-cleaved and uncleaved CK-18 and is therefore used as a marker of overall death including apoptosis and necrosis.

Using the M30 antibody, we initially demonstrated that CK-18 cleavage and apoptosis are increased in liver tissue of patients with various liver diseases.16, 17 Moreover, we could detect a caspase-generated CK-18 fragment in sera of patients with liver disease but not in healthy individuals.18 Subsequently, we and others demonstrated that caspase-generated CK-18 fragments are increased in sera of patients with various acute or chronic liver diseases.19-23 Furthermore, it was recently shown that the plasma concentration of the CK-18 fragments accurately differentiated NASH from NAFL.24-26 These results therefore suggest a potential use of CK18 fragments as a biomarker for the staging of chronic liver disease.

Whether apoptosis is the sole cell death mechanism involved in liver diseases is currently unknown. In addition to apoptosis, several other forms of cell death have been described including necrosis, necroptosis, autophagic cell death, and others.27 Interestingly, using the M65 and M30 assays and additional markers of cell death, we have recently shown that in acute liver failure, apoptosis and necrosis occur.21 Both apoptosis and necrosis have also been proposed to be responsible for the development and progression of liver fibrosis.28

In this biopsy-proven study we prospectively evaluated the M30 and M65 assay as well as an improved version of the M65 assay to predict clinically relevant stages of fibrosis and steatosis. Moreover, we analyzed which of the biomarkers shows the best performance for predicting NASH in NAFLD patients.


ALT, alanine aminotransferase; AUC, area under curve; BMI, body mass index; CK, cytokeratin; ELISA, enzyme-linked immunosorbent assay; HCV, hepatitis C virus; NAFL, nonalcoholic fatty liver; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NAS, NAFLD activity score; ROC, receiver operating characteristics.

Patients and Methods


We investigated sera from 121 patients (50.4% male, 18-75 years, mean age 46.5 ± 1.2) with chronic liver diseases (viral hepatitis, n = 66; autoimmune hepatitis, n = 15; Wilson's disease, n = 4; NAFLD/NASH, n = 22; unknown origin, n = 14). Sera from 18 healthy individuals (33.3% male, 25-40 years, mean age 28.7 ± 1.0) and from 200 blood donors defined as a “real-life cohort” served as controls (53.0% male, 18-67 years, mean age 44.2 ± 0.9). Sera were stored at −20°C. At the time of blood withdrawal, all patients obtained a liver biopsy. The fibrosis stage (F1-F6) was assessed according to Ishak et al.29 by the same pathologist. The diagnosis of NAFL (n = 10) versus NASH (n = 12) was based on histological examination. The NAFLD activity score (NAS) as the sum of steatosis, lobular inflammation, and hepatocellular ballooning scores was assessed according to Kleiner et al.30 by the same pathologist. In 107 of 121 patients, we performed transient elastography using the FibroScan (Echosens, Paris, France). All liver stiffness measurements were performed by a single experienced investigator (M.D.) as described.31 The result of liver stiffness determination was expressed in kPa and was the median of at least 10 individual measurements with a success rate of >60%. The study was approved by the Ethics Committee of Hannover Medical School.

Serological Detection of Caspase-Cleaved and Total Cytokeratin-18.

For quantitative measurement of the caspase-generated neoepitope of CK-18, we used the M30-Apoptosense ELISA according to the manufacturer's instructions (Peviva, Bromma, Sweden) and as described.18 We further used the M65 and M65 EpiDeath (M65ED) ELISA (Peviva), which quantifies both uncleaved and caspase-cleaved CK-18.21 The M65 assay is based on the capture (M6) and detection (M5) antibodies that are directed against two different epitopes of CK-18 and recognize total CK18. Inversely to the M65 ELISA, the M65ED ELISA uses the M6 antibody for detection and the M5 as capture antibody, resulting in improved binding specificity and lower signals in healthy controls.32 For all assays a cubic spline algorithm was employed for data interpolation.

Statistical Analyses.

Statistical analyses were computed using Graphpad Prism 5.0 and SPSS 19.0 software and confirmed by a professional statistician. All assays were performed in duplicate. Data are presented as box plot and whiskers analysis as well as means ± standard error of the mean (SEM). Different serum markers in patients and healthy controls were compared using Mann-Whitney's U test. Regression analysis was performed to calculate the Spearman rank correlation coefficient. Receiver operating characteristics (ROC) analysis was calculated. A multivariate logistic regression analysis was performed in order to adjust for variables found to be associated with fibrosis or with NASH. A P value < 0.05 was considered significant.


Noninvasive Assessment of Fibrosis Stages in Chronic Liver Diseases by Serum Biomarkers of Cell Death.

Because apoptosis has been implicated in liver fibrogenesis, we analyzed the ability of different cell death biomarkers to discriminate between different fibrosis stages in patients with chronic liver disease (n = 121). To this end, we compared the M30 ELISA, which selectively detects caspase-cleaved CK-18 and thereby apoptotic cell death, with the M65 ELISA that detects both caspase-cleaved and -uncleaved CK-18 and thereby overall cell death. In addition, the M65ED ELISA was employed as a modified version of the M65 ELISA. Initial regression analyses showed a significant correlation of each cell death biomarker with fibrosis stage and liver stiffness, revealing the best correlation for the M65ED assay (Table 1). In contrast, no significant differences among the different fibrosis stages were found for alanine aminotransferase (ALT) levels (Table 2). Despite a significantly (P < 0.05) higher liver steatosis in patients with moderate compared with low fibrosis stages, no significant difference in the percentage of steatosis was found between the groups of moderate and high or low and high fibrosis stages (Table 2).

Table 1. Correlation of Cell Death Biomarkers With Histological Fibrosis (Ishak Stage) or Transient Elastography (Liver Stiffness)
Cell Death MarkerIshak Fibrosis StageLiver Stiffness
  • **

    Correlation is significant at the 0.01 level (two-tailed).

M30r = 0.309**r = 0.392**
M65r = 0.335**r = 0.388**
M65EDr = 0.402**r = 0.443**
Table 2. Demographic and Clinical Characteristics of Patients With Different Stages of Fibrosis
 Ishak Fibrosis Stage
  • Abbreviations: ALT, alanine aminotransferase; BMI, body mass index.

  • *

    F0-F1 versus F2-F4, P < 0.05.

No. of patients793111
Mean age ± SEM, y45.2 ± 1.748.9 ± 1.749.1 ± 3.3
Sex, % male48.154.854.5
Steatosis, %16.1 ± 2.722.1 ± 3.6*14.3 ± 5.3
ALT, U/L64.0 ± 5.273.9 ± 7.192.7 ± 19.5
BMI25.0 ± 0.428.1 ± 1.5*24.2 ± 1.0
Biopsy length, mm24.0 ± 1.519.4 ± 1.722.1 ± 2.2

We then compared the cell death biomarkers for their ability to discriminate between different stages of fibrosis, including patients with low (F0-F1, n = 79), moderate (F2-F4, n = 31) or high (F5-F6, n = 11) fibrosis. All three biomarkers discriminated significantly (P < 0.01) between the patients with different fibrosis stages and either healthy control individuals (M30: mean 111.9 ± 7.9 U/L, M65: mean 234.5 ± 19.9 U/L, M65ED: mean 96.8 ± 10.1 U/L; n = 18) or individuals from the real-life cohort (M30: mean 128.2 ± 4.9 U/L; M65: mean 288.4 ± 9.2 U/L; M65ED mean 100.1 ± 8.1 U/L; n = 200). Whereas the M30 assay could significantly (P < 0.01) discriminate between low (mean 174.1 ± 12.4 U/L) and high fibrosis stages (mean 346.5 ± 54.2 U/L) and between moderate (mean 199.1 ± 18.3 U/L) and high fibrosis, no significant differences were found between low and moderate fibrosis stages (Fig. 1A). In contrast, using the M65 ELISA, we found significant (P < 0.05) differences between low (mean 503.2 ± 33.1 U/L) and moderate (mean 635.2 ± 65.1 U/L) fibrosis stages as well as between moderate and high (mean 988.0 ± 179.4 U/L) fibrosis (Fig. 1B). Similar results between moderate (mean 549.6 ± 73.3 U/L) and high (mean 1145.5 ± 224.7 U/L) fibrosis stages were found with the M65ED ELISA, which could discriminate low (mean 429.1 ± 52.4 U/L) and moderate fibrosis stages with an even better sensitivity (P < 0.01) compared with the M65 ELISA (Fig. 1C).

Figure 1.

Serological detection of a caspase-cleaved CK-18 neoepitope by the M30 ELISA (A) and of overall cell death (caspase-cleaved and uncleaved CK-18) by the M65 (B) or M65ED (C) ELISA in healthy controls and patients with chronic liver diseases and minimal (F0-F1), moderate (F2-F4), and high (F5-F6) stages of fibrosis. Data are presented as box plots including medians and 25th and 75th percentiles. In all three assays, cell death was significantly (P < 0.01) elevated in patients with different stages of fibrosis compared with healthy controls. Detection of caspase-cleaved CK-18 (A) was not able to significantly discriminate between low and moderate fibrosis stages, but between low and high or moderate and high stages of fibrosis. Measurement of overall cell death by the M65 (B) or M65ED ELISA (C) significantly discriminates between all stages of fibrosis. *, P < 0.05; **, P < 0.01; n.s., not significant.

Diagnostic Value of the Cell Death Biomarkers to Predict Clinically Relevant or Progressed Fibrosis.

We then calculated the cutoff values of the cell death assays to correctly predict relevant stages of fibrosis (≥F2) or progressed fibrosis/cirrhosis (≥F5) with the best compromise sensitivity/specificity. To this end, we performed a ROC plot analysis including all patients from different fibrosis stages (n = 121). Total CK-18 level detected by the M65ED ELISA above or below 353.0 U/L correctly predicted fibrosis stages ≥F2 with a sensitivity of 74% and a specificity of 68% [area under the curve (AUC) 0.73; confidence interval (CI) 95%: 0.64-0.82] (Fig. 2). Similar results (sensitivity 71%, specificity 67%; AUC 0.70, CI 95%: 0.60-0.79) were obtained with a cutoff value of 479.5 U/L detected by the M65 ELISA. However, compared with the M65 ELISAs, lower sensitivity (64%) and specificity (61%) were obtained with the M30 assay (cutoff 157.5 U/L; AUC 0.66, CI 95%: 0.56-0.76). To predict (pre)cirrhotic stages (≥F5), all three biomarkers showed similar discriminating power with good compromise sensitivity/specificity (M30: 82%/71%, AUC 0.81, CI 95% 0.65-0.97; M65: 73%/78%, AUC 0.78, CI 95% 0.64-0.92; M65ED: 82%/77%, AUC 0.82, CI 95% 0.68-0.96). However, the cutoff value for the M30 ELISA to predict ≥F2 (157.5 U/L) was close to the cutoff value to predict ≥F5 (205.5 U/L). In contrast, the cutoff values of the M65 assays for prediction of ≥F2 or ≥F5 fibrosis stages showed higher differences.

Figure 2.

Predictive discrimination of hepatocyte death as determined by ROC plot analysis. The ROC analyses indicate the threshold for M30 (A,D), M65 (B,E), or M65ED (C,F) biomarkers for the best compromise sensitivity/specificity to predict fibrosis stages of ≥F2 (A-C) or ≥F5 (D-F). AUC, area under the curve.

To exclude variables other than those biomarkers influencing prediction of fibrosis, we performed a multivariate logistic regression analysis. In this model, variables known to influence fibrosis severity, i.e., liver steatosis, cholestasis, and ALT levels, were studied as possible confounders of the cell death biomarkers to predict fibrosis stages ≥F2. This analysis confirmed that the biomarkers M30 (P < 0.05), M65 (P < 0.01), and M65 ED (P < 0.01) predict fibrosis stages ≥F2 independently of steatosis, cholestasis, or ALT levels.

Serological Detection of Liver Steatosis by Different Cell Death Biomarkers.

In addition to fibrosis, liver cell death has been implicated in steatosis-associated injuries.19 We therefore investigated whether the biomarkers can discriminate between healthy individuals and steatosis patients, and in particular between patients with minimal (≤10% of hepatocytes containing fat droplets) and higher grades of steatosis (>10%). Compared with patients with minimal steatosis (≤10%, mean 2.1 ± 0.4%, n = 69), patients with advanced steatosis (>10%, mean 37.7 ± 3.0%, n = 52) showed significantly elevated ALT levels but no significantly different stages of fibrosis (Table 3). Importantly, all cell death markers were able to discriminate (P < 0.01) between minimal steatosis (M30: mean 189.3 ± 16.9 U/L; M65: mean 528.9 ± 45.2 U/L; M65ED: mean 500.6 ± 73.1 U/L) and the healthy individuals (Fig. 3). Whereas the M30 marker did not significantly differentiate between minimal (mean 189.3 ± 16.9 U/L) and higher (mean 205.3 ± 14.2 U/L) grades of steatosis (Fig. 3A), results of both M65 assays showed significant (P < 0.01) differences between minimal (M65: mean 528.9 ± 45.2 U/L; M65ED: mean 500.6 ± 73.1 U/L), and higher (M65: mean 650.3 ± 49.9 U/L and M65ED: mean 557.7 ± 52.3 U/L) percentage of steatosis (Fig. 3B,C).

Figure 3.

Measurement of caspase-cleaved CK-18 (apoptosis) by the M30 ELISA (A) and of overall cell death (caspase-cleaved and uncleaved CK-18) by the M65 (B) or M65ED (C) ELISA in sera of patients with minimal (≤10%), relevant (>10%) liver steatosis or healthy individuals. Data are presented as box plots including medians and 25th and 75th percentiles. Cell death measured by all three assays was significantly (P < 0.01) elevated in patients with minimal or relevant steatosis compared with healthy controls. In contrast to the M65 or M65ED ELISA, which both significantly (P < 0.01) distinguished between minimal and relevant steatosis, no differences between the two patients groups were found by determination of caspase-cleaved CK-18. **, P < 0.01; n.s., not significant.

Table 3. Demographic and Clinical Characteristics of Patients With different Degrees of Steatosis
  1. Abbreviations: ALT, alanine aminotransferase; BMI, body mass index.

No. of patients6952 
Mean age ± SEM, y46.8 ± 1.746.1 ± 1.6 
Sex, % male39.165.4 
Steatosis, %2.1 ± 0.437.7 ± 3.0<0.01
Ishak fibrosis stage1.4 ± 0.21.6 ± 0.2n.s.
ALT, U/L53.7 ± 4.589.7 ± 7.0<0.01
BMI24.0 ± 0.428.0 ± 0.9<0.01
Biopsy length, mm23.8 ± 1.421.1 ± 1.7n.s.

We then selectively analyzed patients with NAFL (n = 10) and NASH (n = 12) from our cohort (Fig. 4A–C). Detection of apoptosis (M30) allowed for significant (P < 0.05) discrimination between NAFL (mean 138.0 ± 11.4 U/L) and NASH (mean 228.6 ± 29.8 U/L) and between NASH and healthy individuals (P < 0.01; Fig. 4A). However, the M30 ELISA did not significantly differentiate between patients with NAFL and healthy or real-life controls. In contrast, the M65 (Fig. 4B) and M65ED assays (Fig. 4C) allowed for a significant (P < 0.01) differentiation between NAFL and healthy controls as well as between NAFL (M65: mean 362.7 ± 34.7 U/L and M65ED: mean 216.9 ± 27.3 U/L) and NASH (M65: mean 725.1 ± 92.9 U/L and M65ED: mean 586.9 ± 99.4 U/L) patients. Compared with NAFL patients, NASH patients showed higher ALT levels and percentage of steatosis but similar low stages of fibrosis, indicating that the NASH patients in our cohort revealed early disease stages without progressed fibrosis (Table 4). The absence of advanced fibrosis therefore allowed analysis of the different cell death biomarkers to discriminate between NASH and NAFL without an additional influence from fibrosis.

Figure 4.

Detection of the cell death biomarkers M30 (A), M65 (B), and M65ED (C) in sera of NAFL or NASH patients compared with healthy controls. Data are presented as box plots including medians and 25th and 75th percentiles. Unlike the M65 biomarkers (B,C), the M30 marker (A) did not significantly discriminate between NAFL and healthy individuals. All three biomarkers allowed for significant discrimination between NAFL and NASH patients. *, P < 0.05; **, P < 0.01; n.s., not significant.

Table 4. Demographic and Clinical Characteristics of Patients With NAFL or NASH
  1. Abbreviations: ALT, alanine aminotransferase; BMI, body mass index; NAFL, non-alcoholic fatty liver; NAS, nonalcoholic steatosis; NASH, nonalcoholic steatohepatitis.

No. of patients1012 
Mean age ± SEM, y49.9 ± 3.245.6 ± 3.3 
Sex, % male7066.7 
Steatosis, %18.3 ± 4.160.4 ± 7.1<0.01
Ishak fibrosis stage0.3 ± 0.20.6 ± 0.2n.s.
ALT, U/L52.8 ± 8.494.4 ± 10.4<0.05
NAS2.6 ± 0.26.6 ± 0.4<0.01
BMI26.0 ± 0.927.8 ± 1.1n.s.
Biopsy length, mm28.3 ± 4.620.7 ± 4.1n.s.

Predictive Value of the Cell Death Biomarkers to Detect Steatosis >10% and to Discriminate Between NAFLD and NASH

The previous results indicated that, unlike the M30 marker, both M65 assays discriminate not only between NAFL and NASH, but also between NAFL patients and healthy individuals. To determine the predictive discriminating value of the biomarkers for detection of higher grades of steatosis (>10%) or NASH, we performed ROC analyses comparing patients with steatosis above or ≤10% (n = 121; Fig. 5A–C) or comparing patients with NASH or NAFL (n = 22; Fig. 5D–F). A cutoff value of 144 U/L of the M30 assay (Fig. 5A) correctly predicted steatosis >10% with a sensitivity of 64% and specificity of 59% (AUC 0.60, CI 95% 0.50-0.70). Compared with the M30 ELISA, the cutoff values of the M65 (469 U/L; Fig. 5B) or M65ED (310 U/L; Fig. 5C) ELISAs showed a higher sensitivity (65% and 73%, respectively) and similar specificity (61%; AUC 0.68, CI 95% 0.58-0.77 and AUC 0.67, CI 95% 0.57-0.77, respectively). Better sensitivity and specificity were obtained for all three biomarkers when we selectively analyzed patients with NALFD for the prediction of NASH (Fig. 5D–F).

Figure 5.

Prediction of steatosis >10% or NASH by cell death biomarkers. The ROC analyses indicate the threshold for M30 (A,D), M65 (B,E), or M65ED (C,F) biomarkers for the best compromise sensitivity/specificity to predict steatosis >10% (A-C) or NASH (D-F). AUC, area under the curve.

Compared with the M30 ELISA, which predicts NASH with sensitivity of 75% and specificity of 70% (cutoff value 149.5 U/L, AUC 0.77, CI 95% 0.57-0.97), we again found a better sensitivity (100%) and specificity (80%) for the discrimination between NAFL and NASH by the M65 or M65ED ELISAs (AUC 0.93 and 0.92, CI 95% 0.82-1.0 and 0.79-1.0, respectively). Multiple logistic regression analysis revealed that detection of total CK-18 by the M65 ELISAs predicted NASH independently (P < 0.05) of ALT levels. In contrast, measurement of caspase-cleaved CK-18 (M30) could not predict NASH independently of ALT levels.


Monitoring of disease progression represents a major goal in chronic liver diseases. Despite various shortcomings, liver biopsy is the gold standard for liver fibrosis staging. In addition to various laboratory methods, new noninvasive approaches such as transient elastography are currently being evaluated to replace liver biopsy. However, these methods are associated with limitations. Interobserver agreement may be reduced in patients with early-stage fibrosis, elevated steatosis, or increased body mass index.33, 34 Several biomarkers or approaches using proteomic and genomic technologies have been developed; however, so far no single assay has gained clinical validity. Furthermore, serological fibrosis markers only weakly discriminate between intermediate and minimal fibrosis, which is often required for clinical decision making.9 Thus, a noninvasive and reliable approach would constitute a major advance in the field.

Hepatocyte apoptosis has been recognized as a mechanism of liver injury that may also contribute to fibrogenesis. For instance, engulfment of apoptotic bodies by hepatic stellate cells stimulates their fibrogenic activity, whereas attenuation of hepatocyte apoptosis was shown to reduce fibrogenesis in experimental cholestasis.35, 36 In this study we analyzed sera from patients with chronic liver diseases (n = 121) and different fibrosis stages for caspase-cleaved CK-18 fragments or total CK-18. Both biomarkers were able to differentiate healthy individuals from patients with different stages of fibrosis. We found a significant correlation between the CK-18 levels and both liver stiffness and histological fibrosis. These findings extend our earlier results and subsequent findings by others, demonstrating a correlation of CK-18 fragment levels with different fibrosis stages in patients with chronic HCV infection.18, 20, 37 These findings are also in line with data obtained in NAFLD patients and alcoholic liver disease, indicating that cell death markers can predict severe fibrosis.25, 38

Interestingly, although both the M30 and M65 assays differentiated severe fibrosis from moderate or low fibrosis, measurement of caspase-cleaved CK-18 was unable to discriminate low from moderate fibrosis. In contrast, serum levels of total CK-18 significantly differentiated between low (F0-1) and moderate fibrosis stages (F2-4). In line with this observation, the M65 assays revealed a better diagnostic accuracy (compromise sensitivity/specificity) for the detection of clinically relevant fibrosis (≥F2) compared with the M30 ELISA, whereas all biomarkers showed similar sensitivity and specificity to detect progressed fibrosis (≥F5).

In addition to fibrosis, steatosis is increasingly recognized as a cofactor influencing the progression of liver injury. We have recently shown that serum levels of caspase-cleaved CK-18 correlate with the severity of liver steatosis in chronic HCV infection,19 a finding that was subsequently confirmed in pediatric HCV patients.39 In this study, we evaluated whether detection of caspase-cleaved or total CK-18 can discriminate between minimal (≤10%) and higher grades of steatosis (>10%) in 121 patients with chronic liver diseases including 52 HCV patients and 22 NAFLD patients. We found significantly higher levels of both biomarkers in patients with liver steatosis compared with healthy controls. Our results further revealed a better diagnostic performance of the M65 assays with improved AUC values to detect relevant steatosis compared with the M30 assay. In contrast to the M30 marker, detection of total CK-18 by both M65 ELISAs discriminated between minimal or relevant steatosis. Because the two patient cohorts showed no significant differences in fibrosis, total CK-18 levels reflect steatosis independently of liver fibrosis.

NAFLD is one of the most common causes of chronic liver disease, ranging from simple steatosis to NASH and cirrhosis. A variety of panel markers using the combination of different variables for NASH diagnosis has been proposed.6 Single markers such as aminotransferases are not suitable to distinguish between NAFL and NASH. In two studies, NASH was diagnosed in up to 59% of NAFLD patients despite normal ALT levels.40, 41 Intriguingly, Wieckowska et al.24 showed that plasma levels of CK-18 fragments might allow the discrimination between NASH and NAFL patients. Subsequent studies confirmed an increase of caspase-cleaved CK18 fragments in NASH patients, supporting the assertion that CK-18 may be useful for the diagnosis of NASH.25, 26, 42-45 In view of these findings we evaluated whether the M30 and M65 assays could distinguish NASH patients from those with simple steatosis.

Both assays could discriminate between NASH and NAFL and between NASH and healthy individuals. Surprisingly, unlike the M30 assay, only serum levels of total CK-18 significantly discriminated between NAFL patients and healthy controls. This differentiation is important because it was demonstrated that 58% of NAFL patients progress toward NASH and 28% among them show fibrosis progression within 3 years.46 Measurement of total CK-18 levels also revealed higher significance for distinguishing between NAFL and NASH compared with the M30 assay. In addition, compared with the M30 marker, both M65 assays revealed a better accuracy with improved AUC values to detect NASH. The fibrosis stages of NAFL and NASH patients were similar in our cohort and did therefore not influence the power of the biomarkers to discriminate between both groups.

In contrast to the M65 assays, the M30 biomarker was unable to predict NASH independently of ALT levels. This is important information, because NASH can also be observed in NAFLD patients with normal aminotransferases.40, 41 A composite model including both ALT and caspase-cleaved CK-18 revealed higher accuracy for prediction of NAFLD activity compared with detection of CK-18 fragments alone.47 Intriguingly, we found an even better diagnostic accuracy of the M65 marker to predict NASH compared with the diagnostic value of the M30 ELISA in a previous validation study.26 This observation has been supported in a small patient cohort, which already indicated a higher predictive value to diagnose NASH for the M65 compared with M30 biomarker.48 Thus, total CK-18 levels might be superior to discriminate between healthy and minimal and between minimal and significant disease conditions.

Whether the differential sensitivity of the M30 and M65 markers reflect different cell death modes remains to be investigated. So far, no appropriate biomarkers for the detection of necrosis or necroptosis have been established. Furthermore, it is clear that various intermediate forms of cell death exist. Moreover, different serum stabilities of the CK-18 forms might account for the different sensitivity of the assays. The lower values obtained with the M30 assay, especially in patients with mild liver disease, might further compromise the accuracy of this test. In summary, measurement of total CK-18 is superior to detection of caspase-cleaved CK-18 to determine relevant stages of fibrosis and steatosis, in particular at low disease stages. Further large cohort studies in NAFLD patients analyzing the mode of cell death and the value of cell death biomarkers to correctly predict NASH are warranted.