SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Background

The diagnosis of non-alcoholic fatty liver disease (NAFLD), non-alcoholic steatohepatitis (NASH) and fibrosis relies on liver biopsy. Non-invasive assessments are urgently needed.

Aim

To evaluate cell apoptotic marker cytokeratin-18 M30 and total cell death markers cytokeratin-18 M65/M65ED for the assessment and monitoring of NAFLD.

Methods

A cohort of 147 patients with biopsy-proven NAFLD and 73 controls were enroled, including 51 patients who received paired liver biopsies 36 months apart. Biomarkers were determined by enzyme-linked immunosorbent assay.

Results

M30, M65 and M65ED increased in a stepwise fashion in control subjects, patients with non-NASH, NAFLD and NASH (all < 0.001). All biomarkers had similarly high accuracy over 0.9 in predicting NAFLD and moderate accuracy around 0.7 in predicting NASH. Among patients with paired liver biopsies, changes in M30, M65 and M65ED positively correlated with disease progression (rho = 0.42, 0.32 and 0.39; = 0.002, 0.023 and 0.005 respectively), and only changes in M65 and M65ED correlated with fibrosis progression (rho = 0.29, 0.34; = 0.038, 0.015 respectively). Both M30 and M65 had area under receiver-operating characteristics curve above 0.8 in predicting disease progression. At cut-off of 236 U/L, changes of M65ED had 88% NPV and 59% PPV to exclude and predict fibrosis progression.

Conclusions

Cytokeratin-18 M30 and M65/M65ED have moderate accuracy in detecting non-alcoholic steatohepatitis. Changes in the biomarkers also correlate with histological progression. However, development of new biomarkers is still required to improve the diagnostic accuracy.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Non-alcoholic fatty liver disease (NAFLD) is one of the most common chronic liver diseases worldwide.[1-3] NAFLD encompasses a wide disease spectrum from simple steatosis to non-alcoholic steatohepatitis (NASH). Although simple steatosis is largely benign, patients with NASH have active liver injury and may progress to cirrhosis, liver failure and hepatocellular carcinoma.[4-6] Increased overall and liver-specific mortality is also observed mainly in NASH patients.[7, 8] Thus, it is important to distinguish between simple steatosis and NASH.

Liver biopsy is the gold standard for diagnosing NAFLD and NASH. However, it is an invasive procedure which might result in major complications like haemorrhage, and may not be acceptable to some patients. It is also not a suitable method for repeated assessments. Reliable non-invasive tests of NASH are urgently needed.

Several biomarkers have been developed to differentiate NASH from simple steatosis.[9-16] Among them, blood cytokeratin-18 (CK-18) fragment, CK18Asp396 neo-epitope (M30) has been validated in different cohorts.[9, 17-20] M30 is a fragment which is cleaved by caspases during cell apoptosis from CK-18, a major intermediate filament protein in hepatocytes. Detecting serum M30 level reflects the degree of hepatocellular apoptosis, a characteristic feature of NASH.[21, 22] Our previous study revealed that M30 has high overall accuracy in differentiating NAFLD from control subjects and moderate accuracy in differentiating NASH from simple steatosis.[20]

Recent studies suggest that other cell death markers may also be useful in the prediction of NASH. M65 and M65ED, which detect both caspase-cleaved and uncleaved CK-18, reflect total cell death including apoptosis and necrosis.[23] Joka and colleagues suggested that M65/M65ED may have superior performance to M30 in detecting mild fibrosis and steatosis.[24] However, that study was limited by the inclusion of different liver diseases and the small number of NAFLD patients.[25] As such, the performance of M65 and M65ED in detecting NASH compared with M30 is currently unknown. In addition, limited by cross-sectional design, previous studies could not address whether these biomarkers can be used for serial monitoring.

In this study, we aimed to evaluate the performance of serum CK-18 M30, M65 and M65ED in assessing NAFLD in a large histological cohort. We also aimed to test the performance of these biomarkers in predicting histological changes in 36 months.

Patients and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Study subjects

In this study, cell death biomarkers were measured in archived serum samples of a prospectively recruited cohort.[20] We recruited consecutive adult patients who underwent liver biopsy for suspected NAFLD at Prince of Wales Hospital, Hong Kong. Control subjects were recruited from a population screening project for NAFLD.[2] Subjects without underlying liver and metabolic diseases and intrahepatic triglyceride content (IHTG) less than 5% as estimated by proton-magnetic resonance spectroscopy (1H-MRS)[26, 27] were included as control subjects. Among patients with biopsy-proven NAFLD, 51 patients had prospectively planned follow-up liver biopsies 36 months later to study disease progression.[4] This subgroup served to validate the performance of serum biomarkers and test the utility of the biomarkers in predicting disease progression. The study protocol was approved by the Clinical Research Ethics Committee of The Chinese University of Hong Kong. All subjects gave informed written consents.

Clinical and pathological assessment

The medical history, anthropometric tests, blood tests and liver biopsy were evaluated and recorded.[20] Metabolic syndrome was defined according to the ethnic-specific criteria of the joint statement of the International Diabetes Federation; National Heart, Lung and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society and International Association for the Study of Obesity.[28] Percutaneous liver biopsy was performed using a 16-gauge Temno needle.[29] Histological slides were read by two experienced pathologists (PCLC, AWHC) who were blinded to the clinical data. In case of discrepant interpretation, the two pathologists came up with a consensus after reviewing the slides together. Liver histology was reported by both semiquantitative scoring according to the NASH Clinical Research Network system[30] and the pathologists’ global assessment, which was modified from the original description by Matteoni and colleagues.[31, 32] NAFLD activity score (NAS) was the sum of steatosis, lobular inflammation and hepatocellular ballooning scores. Fibrosis was staged from 0 to 4, with stage 0 = no fibrosis, 1 = perisinusoidal or periportal fibrosis, 2 = perisinusoidal and portal/periportal fibrosis, 3 = bridging fibrosis and 4 = cirrhosis. NASH was diagnosed for specimens with fatty liver, lobular inflammation and hepatocytes ballooning. NAFLD patients not fulfilling the criteria of NASH were labelled as non-NASH.[20] Among patients with paired liver biopsy 36 months apart, those who had non-NASH at baseline and NASH upon follow-up were classified to have disease progression. Patients were also classified according to NAS into three groups: NAS <3, 3–4, and ≥5.[30]

Proton-magnetic resonance spectroscopy

1H-MRS was performed to measure IHTG within 8 weeks from the clinical assessment. Whole-body 3.0T scanner with a single voxel point-resolved spectroscopy sequence and an echo time of 40 ms and repetition time of 5000 ms was used. Fatty liver was defined as IHTG≥5%.

Serum CK-18 M30, M65 and M65ED

During each clinic visit, part of the patients’ serum samples was stored at −80°C, and biomarker testing was performed in one batch afterwards. Serum level of CK-18 M30, M65 and M65ED was measured by the M30 Apoptosense enzyme-linked immunosorbent assay (ELISA) kit (PEVIVA, Bromma, Sweden), M65 ELISA kit (PEVIVA) and M65 EpiDeath ELISA kit (PEVIVA) respectively. All tests were performed in a single session by one investigator (J. S.).

Statistical analysis

This study included 73 control subjects and 147 NAFLD patients, among whom 69 had NASH. This sample size had the power to evaluate the performance of serum biomarkers in detecting NAFLD with standard errors of the area under the receiver-operating characteristics curve (AUROC) between 0.02 and 0.04. The cohort also had the power to evaluate the performance of biomarkers in detecting NASH with standard errors between 0.03 and 0.05.

Continuous variables were expressed as mean ± s.d. or median [interquartile range (IQR)] as appropriate. Categorical clinical data between groups were compared by chi-squared test; quantitative variables were analysed using t-test and one-way analysis of variance for normal distributional data, or Mann–Whitney U-test and Kruskal–Wallis test for highly skewed data. Spearman's correlation coefficient was used to estimate the association of serum biomarkers’ levels and several factors of interest, while multiple linear regression was used to determine the independent factors associated with levels of biomarkers. Receiver-operating characteristics curve analysis was conducted to assess the performance of biomarkers and in the prediction of NAFLD/NASH. Delong's test was used to compare the diagnostic performance between biomarkers. For each biomarker, three optimal cut-off values were selected based on high sensitivity around 90%, high specificity around 90% and the best combined sensitivity and specificity according to the Youden's index. All statistical tests were performed using the Statistical Package for Social Sciences version 16.0 (SPSS Inc., Chicago, IL, USA) and Analyse-it Method Evaluation Edition version 2.26 (Analyse-it, Leeds, UK). A two-tailed P-value of <0.05 was considered statistically significant.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Patient characteristics

From 2004 to 2010, 147 patients with biopsy-proven NAFLD were recruited and 73 controls were selected. Details of the case selection process were described previously.[20] Compared with the last report, one additional patient was reclassified into the NAFLD (non-NASH) group. That patient had simple steatosis at baseline and complete resolution of NAFLD 36 months later.

Clinical and pathological characteristics were shown in Table 1 and Table S1. The NAFLD patients and control subjects were well matched in age and gender. NAFLD patients were more obese than control subjects and had a higher prevalence of diabetes and hypertension. Three quarters of NAFLD patients had metabolic syndrome, compared with only 11% in control subjects. ALT levels were significantly higher in NAFLD patients. Sixty-nine (47%) NAFLD patients had NASH by global pathological assessment; 32 (22%) patients had NAS ≥5. NASH patients had higher body mass index and were more likely to have metabolic syndrome than non-NASH NAFLD patients, while the other clinical features were similar.

Table 1. Clinical characteristics of all patients’ population
 ControlNAFLDNon-NASHNASHPatients with paired liver biopsies
BaselineMonth 36
  1. ALT, alanine aminotransferase; BMI, body mass index; HbA1c, glycated haemoglobin; LDL, low density lipoprotein; NASH, non-alcoholic steatohepatitis; TG, triglycerides.

  2. Numbers in parentheses are percentage for categorical data or standard deviation for numerical data.

  3. a

     Significant at < 0.01, between control and NAFLD patients.

  4. b

    Significant at < 0.05.

  5. c

    Significant at < 0.01, between non-NASH and NASH patients.

  6. d

    Significant at < 0.05.

  7. e

    Significant at < 0.01, between baseline and month 36 in patients received paired liver biopsies.

All73147786951 
Male gender39 (53.4)82 (55.8)46 (59.0)36 (52.2)34 (66.7) 
Age (years)e47.4 (10.3)47.7 (9.7)47.8 (9.0)47.7 (10.5)44.2 (8.9)47.2 (9.0)
Body Weight (kg)a63.0 (8.4)74.5 (14.7)72.9 (14.4)76.2 (14.9)75.8 (12.8)75.6 (12.7)
BMIa,b22.5 (2.7)27.4 (3.9)26.7 (3.7)28.2 (4.0)27.5 (3.7)27.4 (3.8)
Waist (cm)b81 (8)94 (11)93 (11)95 (11)93 (9)91 (10)
Systolic blood pressure (mmHg)a126 (15)135 (16)135 (17)135 (16)133 (16)135 (18)
Diastolic blood pressure (mmHg)d81 (9)81 (11)79 (11)82 (10)78 (9)81 (12)
Diabetesa1 (1.4)70 (47.6)34 (43.6)36 (52.2)26 (51.0) 
Hypertensiona0 (0)63 (42.9)30 (38.5)33 (47.8)26 (51.0) 
Metabolic syndromea,b8 (11.0)110 (74.8)53 (67.9)57 (82.6)35 (68.6) 
ALT (IU/L)a,e28 (26)73 (45)66 (40)80 (49)78 (54)58 (30)
Fasting glucose (mmol/L)a4.9 (0.4)6.5 (2.4)6.3 (2.1)6.8 (2.7)6.7 (2.9)6.5 (2.1)
HbA1c (%)a5.3 (0.4)6.2 (1.3)6.0 (1.3)6.4 (1.3)6.5 (1.5)6.3 (1.2)
LDL-cholesterol (mol/L)3.0 (0.9)3.1 (0.9)3.1 (1.0)3.0 (0.7)3.0 (1.1)3.1 (0.8)
Total cholesterol (mol/L)5.2 (1.2)5.2 (1.0)5.3 (1.2)5.1 (0.7)5.3 (1.3)5.2 (1.2)
Triglyceride (mmol/L)a1.3 (1.2)2.2 (1.2)2.1 (1.3)2.3 (1.1)2.2 (1.3)2.6 (5.6)
Liver TG content (%)1.7 (1.2)     
Biopsy lengthc,e 1.9 (0.6)1.7 (0.5)2.1 (0.5)1.5 (0.4)1.8 (0.4)
Steatosis grade 1/2/3c,e 53/52/4238/28/1215/24/300/29/15/74/14/20/13
Lobular inflammation 0/1/2c,e 46/95/646/30/20/65/415/34/232/16/3
Ballooning 0/1/2c,e 56/82/956/21/10/61/835/15/112/38/1
Fibrosis 0/1/2/3/4d 

59/51/

16/10/11

46/21/

9/0/2

13/30/

7/10/9

26/16/7/1/128/14/2/4/3
NASH    14(27.5)18(35.3)

Prediction of NAFLD

The serum levels of M30, M65 and M65ED were significantly higher in NAFLD patients than controls (Figure 1a). The median M30 level was 354 (interquartile range: 221–529) U/L in NAFLD patients and 103 (80–138) U/L in controls (< 0.001). The median M65 level was 770 (539–1010) U/L in NAFLD patients and 309 (249–411) U/L in controls (< 0.001). The median M65ED level was 443 (202–801) U/L in NAFLD patients and 47 (30–92) U/L in controls (< 0.001). All three biomarkers had similar accuracy in detecting NAFLD with AUROC larger than 0.9 (Table 2 and Figure S1a). By spearman correlation test, all three biomarkers were highly correlated with each other and NAFLD (Table 3).

image

Figure 1. Serum level of CK-18 M30, M65 and M65ED in control and NAFLD patients. (a) Comparison between control and NAFLD patients. (b) Comparison between patients with non-NASH NAFLD and NASH. *Significant at < 0.05; **Significant at < 0.01; N.S., not statistically significant.

Download figure to PowerPoint

Table 2. Accuracy of biomarkers
 BiomarkerAUROCCut-off (U/L)Sensitivity (%)Specificity (%)PPV (%)NPV (%)LR+LR−
  1. AUROC, area under receiver-operating characteristics curve; LR, likelihood ratio; NPV, negative predictive value; PPV, positive predictive value.

  2. Cut-offs with high sensitivity, highest overall accuracy and high specificity were presented.

  3. a

    Represent the performance for discriminating NAFLD from control cases.

  4. b

    Represent the performance for discriminating NASH from Non-NASH.

NAFLDaM300.92(0.87–0.96)11095.257.581.985.62.20.08
   18084.490.494.774.28.80.17
   31057.194.595.452.210.40.45
 M650.92(0.89–0.96)36094.664.484.385.62.70.08
   52376.995.997.467.318.80.24
   52376.995.997.467.318.80.24
 M65ED0.94(0.92–0.97)8095.274.088.188.43.70.06
   10593.279.590.285.34.50.09
   23761.495.996.855.215.00.40
NASHbM300.66(0.57–0.75)20389.932.153.978.21.30.33
   33866.760.359.867.21.70.55
   67024.689.767.957.42.40.84
 M650.71(0.62–0.79)50191.334.655.381.81.40.25
   79062.370.565.167.92.10.53
   118331.989.773.359.83.10.76
 M65ED0.70(0.62–0.79)14391.317.949.669.91.10.48
   30979.757.762.576.31.90.35
   100027.591.073.058.73.10.80
Table 3. Correlations within M30, M65, M65ED and NAFLD/NASH prediction
 M30M65M65ED
rhoP-valuearhoP-valuearhoP-valuea
  1. a

     P-value corresponds to Ho: rho = 0.

All patients
M30  0.86<0.0010.80<0.001
M650.86<0.001  0.94<0.001
M65ED0.80<0.0010.94<0.001  
NAFLD0.68<0.0010.69<0.0010.73<0.001
NAFLD
M30  0.86<0.0010.80<0.001
M650.86<0.001  0.94<0.001
M65ED0.80<0.0010.94<0.001  
NASH0.270.0010.36<0.0010.35<0.001
NAS0.41<0.0010.50<0.0010.54<0.001

Prediction of NASH

All three biomarkers increased in a stepwise fashion in different steatosis, lobular inflammation, ballooning and fibrosis levels (Figure S2). While comparing between minimal and moderate diseases, M65 and M65ED were able to differentiate grade 2 from grade 1 steatosis (= 0.008 and 0.001 respectively), while M30 failed (= 0.190). All biomarkers could distinguish patients with grade 1 from grade 0 lobular inflammation and ballooning, or with stage 2–3 from stage 0–1 fibrosis.

Among NAFLD patients, M30, M65 and M65ED levels correlated highly with each other, and each was also moderately associated with NAS (Table 3). By multiple linear regression, higher ALT level and more severe lobular inflammation were independently associated with all three biomarkers (Table 4). High steatosis grade and fibrosis stage were independently associated with M65 and M65ED but not M30. Fasting glucose level was also independently associated with M65ED.

Table 4. Multivariable analysis for independent factors associated with M30, M65 and M65ED in NAFLD patients
  BetaP-value
M30ALT0.516<0.001
 Lobular inflammation0.260<0.001
M65ALT0.581<0.001
 Steatosis0.1960.002
 Lobular inflammation0.1720.007
 Fibrosis0.2170.001
M65EDALT0.443<0.001
 Gucose0.2450.034
 Steatosis0.2370.001
 Lobular inflammation0.1420.041
 Fibrosis0.263<0.001

The serum levels of M30, M65 and M65ED were significantly higher in NASH patients (Figure 1b). The median M30 levels in patients with non-NASH and NASH were 277 (186–472) U/L and 397 (264–657) U/L respectively (= 0.001). The median M65 levels in the two groups were 637 (457–886) U/L and 877 (671–1469) U/L respectively (< 0.001). The median M65ED levels in the two groups were 271 (187–579) U/L and 572 (328–1070) U/L respectively (< 0.001). The AUROC of these three biomarkers in differentiating NASH were 0.66 (95% CI: 0.57–0.75), 0.71 (0.62–0.79) and 0.70 (0.62–0.79), respectively, and were not significantly different by Delong's test (Table 2 and Figure S1b). The performances of three biomarkers in differentiating NAFLD patients with NAS ≥5 were similar to which in differentiating NASH (Table S2; Figure S1c and S3).

Assessment of disease progression

The clinical-pathological characteristics of the 51 patients who received paired liver biopsies were shown in Table 1. One patient was on pioglitazone during follow-up, and none was treated with vitamin E. Twenty-five patients had increased NAS in 36 months. Ten patients progressed from non-NASH to NASH, and 14 patients had fibrosis progression for at least 1 stage.

The changes in M30, M65 and M65ED were all associated with NAS change and disease status change; however, only changes of M65 and M65ED were associated with changes of fibrosis stage (Figure 2; Table S3 and Figure S4). Table 5 summarised the predictive performance and cut-off values with highest overall accuracy for changes of biomarkers. Delong's test revealed no significant difference among three biomarkers. At a single cut-off of 35 U/L, change in M30 had both sensitivity and specificity above 80% in predicting disease progression from non-NASH to NASH. When M65ED increased for not more than 62 U/L, the chance to have disease progression was only 10%. At a cut-off of 236 U/L, change in M65ED had sensitivity and specificity of 71.4% and 81.1% in predicting fibrosis progression, with NPV of 88.2% and PPV of 58.8%. On the other hand, the baseline levels of biomarkers could not predict disease progression.

image

Figure 2. Individual changes of serum level of CK-18 M30, M65 and M65ED and disease progression in 51 patients received paired liver biopsies. Individual changes of patients with or without (a) NAFLD activity score worsened, (b) disease progression from non-NASH to NASH (15 patients who were diagnosed as NASH at both time points were excluded) and (c) fibrosis progression. *Significant at < 0.05; **Significant at < 0.01; N.S., not statistically significant.

Download figure to PowerPoint

Table 5. Prediction of disease progression using baseline or changes of biomarkers
  BiomarkerAUROCP-valueaCutoff (U/L)SensitivitySpecificity
  1. a

    Compared to AUROC = 0.50.

  2. b

    For prediction of increased NAFLD activity score.

  3. c

    For prediction of disease progression, 14 patients who were diagnosed as NASH at both baseline and month 36 were excluded.

  4. d

    For prediction of increased fibrosis stage.

BaselineNASbM300.51 (0.34–0.67)0.940   
  M650.48 (0.31–0.64)0.777   
  M65ED0.47 (0.31–0.63)0.720   
 NASHcM300.57 (0.35–0.78)0.538   
  M650.61 (0.39–0.84)0.297   
  M65ED0.62 (0.39–0.85)0.274   
 FibrosisdM300.53 (0.36–0.71)0.720   
  M650.54 (0.35–0.72)0.688   
  M65ED0.46 (0.29–0.63)0.658   
ChangesNASbM300.75 (0.61–0.89)0.002368.076.9
  M650.72 (0.58–0.86)0.00613968.073.1
  M65ED0.72 (0.58–0.86)0.0075668.069.2
 NASHcM300.82 (0.65–0.99)0.0033580.081.5
  M650.74 (0.56–0.93)0.02418280.070.4
  M65ED0.80 (0.62–0.98)0.0066290.059.3
 FibrosisdM300.68 (0.52–0.84)0.05012342.989.2
  M650.72 (0.57–0.87)0.016178.645.9
  M65ED0.77 (0.64–0.91)0.00323671.481.1

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

In this study, we demonstrated that apoptosis biomarker CK-18 M30 and total cell death biomarkers M65/M65ED had similar accuracy in predicting NAFLD and NASH. M65 and M65ED were superior in differentiating minimal and moderate steatosis grade. Changes of M30 and M65ED were associated with disease progression. Changes of M65ED may also predict fibrosis progression.

Apoptosis and necrosis are both important modes of cell death in liver disease. While necrosis is the typical consequence of acute injuries such as ischemia/reperfusion or acute drug-induced hepatotoxicity, apoptosis represents programmed cell death.[33] Apoptosis was thought to be a characteristic feature of NASH.[21, 22] However, Joka and colleagues suggested recently that necrosis might also contribute to the liver damage in NAFLD and NASH by showing biomarkers detecting both apoptosis and necrosis might be superior to detecting apoptosis alone.[24] However, this study included patients with different liver diseases, and NAFLD patients were underrepresented. The role of different cell death markers was also discussed in other liver diseases, such as HBV infection and acute liver failure.[34, 35]. A study in US suggested that the overall diagnostic accuracy of M65 for NASH was higher than that of M30,[36] but that was not confirmed by another study in Turkey.[19] Nevertheless, the correlation between the biomarkers and individual histological features of NAFLD, in particular the evaluation of mild disease, was not reported. In the current study, we confirmed that M30, M65 and M65ED had similar overall accuracy in predicting NASH. M65 and M65ED had higher discriminating power in detecting mild steatosis and fibrosis. As studies on biopsy-proven NAFLD usually include patients with risk factors of advanced disease such as metabolic syndrome, the proportion of patients with NASH is relatively high. When the biomarkers are applied to primary care setting, the NPV in excluding NASH will be even higher.

Our study was unique in including a subgroup of patients with paired liver biopsies 36 months apart. We were therefore able to evaluate the performance of the biomarkers as a monitoring tool for NAFLD patients. The changes in all three biomarkers correlated well with changes in NAS and could be used to predict progression to NASH. Changes in M65 and M65ED were also associated with progression of liver fibrosis. Notably, the AUROCs for predicting disease progression were both higher than 0.8 for M30 and M65ED. These results could be promising as non-invasive tests are more acceptable for long-term and repeated monitoring of disease progression. Importantly, while the changes in CK-18 correlated with histological changes, baseline CK-18 level alone failed to predict disease progression. This indicates that NASH is a dynamic process. Change in disease activity is possible with lifestyle modifications.[4, 37, 38] Therefore, serially performing these biomarkers may be helpful to monitor disease progression.

While the cell death markers are elevated in NAFLD and particularly in NASH patients, it should be noted that they are nonspecific markers that may be elevated in other situations such as chronic hepatitis B and acute liver failure.[34, 35] However, the diagnosis of NAFLD is usually clinically apparent by the time these cell death markers are applied in real-life practice. In addition, the diagnostic accuracy of these biomarkers for NASH remains modest. Future work is needed to develop new biomarkers and evaluate the possibility of using multiple biomarkers as in the case of fibrosis testing.[39, 40]

Our study had several limitations. First, we used liver biopsy as the reference standard, which might suffer from sampling bias. Future studies are required to test these biomarkers against other reference standards such as clinical outcomes. Second, no liver biopsy was performed in control subjects. However, biopsy on healthy people is unethical. Instead, liver disease was excluded in controls stringently by history, laboratory tests and 1H-MRS. Third, some biopsy samples were shorter than 2 cm. However, after excluding samples with short biopsy length, the AUROCs of M30, M65 and M65ED in predicting NASH were 0.67 (0.54–0.81), 0.75 (0.62–0.87) and 0.73 (0.60–0.86) respectively. The performance was similar to the whole study population. Fourth, the paired liver biopsies cohort was relatively small, this resulted the wide statistical variations; a larger prospective longitudinal study should be conducted to further clarify the utility of cell death biomarkers in disease progression prediction. Finally, CK-18 ELISA assays from different companies have not been systematically compared. Our findings cannot be directly extrapolated to other assays. However, we chose the assays that were used in recent similar studies to facilitate comparison across studies.[9, 24]

In conclusion, apoptosis marker CK-18 M30 and total cell death marker M65/M65ED have similar high diagnostic performance in NAFLD and similar moderate accuracy in predicting NASH. Changes in the biomarkers also correlate with histological progression. However, development of new biomarkers is still required to improve the diagnostic accuracy.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Declaration of personal interests: Henry Chan is an advisory board member of Novartis, Merck, Bristol-Myers Squibb, Roche and Abbott Diagnostics. Grace Wong has served as a speaker for Echosens. Vincent Wong is an advisory board member of Novartis, Roche, Gilead and Otsuka. He has served as a speaker for Novartis, Roche, Bristol-Myers Squibb, Echosens and Abbott Diagnostics. Declaration of funding interests: The study was supported by the General Research Fund of the Research Grant Council, Hong Kong (Project reference 477710).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
apt12091-sup-0001-FigS1.tifimage/tif920KFigure S1. ROC curves of CK-18 M30, M65 and M65ED in predicting NAFLD, NASH and NAS ≥5. ROC curves in (a) distinguishing NAFLD patients from control subjects, (b) distinguishing NASH from non-NASH NAFLD patients and (c) distinguishing NAFLD patients with NAS ≥5.
apt12091-sup-0002-FigS2.tifimage/tif902KFigure S2. Serum level of CK-18 M30, M65 and M65ED with different histological features in NAFLD patients. Comparison among different (a) steatosis, (b) lobular inflammation, (c) ballooning and (d) fibrosis grades. *Significant at P < 0.05; **Significant at P < 0.01; N.S., not statistically significant.
apt12091-sup-0003-FigS3.tifimage/tif196KFigure S3. Serum level of CK-18 M30, M65 and M65ED in NAFLD patients according to NAS subgroups. M30 in NAS subgroups: (1–2): 265 (180–418) U/L; (3–4): 366 (223–548) U/L; (5–8): 476 (326–988) U/L; M65 in NAS subgroups: (1–2): 534 (411–795) U/L; (3–4): 765 (572–1038) U/L; (5–8): 954 (759–1762) U/L; M65ED in NAS subgroups: (1–2): 196 (135–374) U/L; (3–4): 466 (253–802) U/L; (5–8): 740 (449–1264) U/L. *Significant at < 0.05; **Significant at < 0.01; N.S., not statistically significant.
apt12091-sup-0004-FigS4.tifimage/tif688KFigure S4. Changes of serum level of CK-18 M30, M65 and M65ED and disease progression in 51 patients received paired live biopsies. Comparison between patients with or without (a) NAFLD activity score worsened, (b) disease progression from non-NASH to NASH (15 patients who were diagnosed as NASH at both time points were excluded) and (c) fibrosis progression. *Significant at < 0.05; **Significant at < 0.01; N.S., not statistically significant.
apt12091-sup-0005-FigS5.tifimage/tif901KFigure S5. ROC curves changes of CK-18 M30, M65 and M65ED in predicting disease progression. ROC curves in (a) distinguishing patients with NAFLD activity score worsened, (b) distinguishing patients with disease progression from non-NASH to NASH and (c) distinguishing NAFLD patients with fibrosis progression.
apt12091-sup-0006-TableS1-S3.docxWord document17K

Table S1. Clinical characteristics of NAFLD patients’ population according to NAS.

Table S2. Accuracy of biomarkers.

Table S3. Correlations between change of biomarkers and disease progression in 51 patients with paired liver biopsy.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.