1 INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is a common chronic liver disease that globally poses a huge public health burden. The continuous necroinflammatory process may lead to a severe form of NAFLD, named steatohepatitis, which may further progress to fibrosis, cirrhosis, and hepatocellular carcinoma (HCC).1, 2 Therefore, assessment of disease severity is necessary for a clinical setting. Liver biopsy is currently the gold standard method for diagnosing severe NAFLD, named nonalcoholic steatohepatitis (NASH). In addition, liver fibrosis assessment is mainly dependent on liver biopsy results.3 However, liver biopsy is an invasive procedure that carries the risk of complications, limiting its application in daily practice. Therefore, several noninvasive approaches, including imaging-based and serum-based methods, have been developed over the past decade for the noninvasive evaluation of liver fibrosis.

In histopathology, liver fibrosis is associated with increased and deposited extracellular matrix (ECM).4 Hepatic stellate cells (HSCs) are activated simultaneously into myofibroblast-like cells during fibrogenesis. The severity of liver fibrosis may be related to the release of direct or indirect products in parallel with the fibrogenesis process.5 Wisteria floribunda agglutinin-positive Mac-2 binding protein (WFA+-M2BP) is a by-product of fibrogenesis and has been demonstrated to be a promising noninvasive method for liver fibrosis assessment.6 Previous studies have demonstrated that serum WFA+-M2BP measured by glycan-based immunoassays reflects liver fibrosis progression. It is also a good predictor of liver fibrosis in chronic hepatitis B (CHB) and chronic hepatitis C.7, 8 Novel serum biomarkers play a role in the prediction of liver disease outcomes and the development of HCC.9, 10 The application of the serum-based method has recently been extended to patients with NAFLD. However, the results are inconsistent in terms of fibrosis prediction.11, 12 Therefore, the prediction performance of WFA+-M2BP remains to be determined, as NAFLD has become the major liver disease globally.

Therefore, this study aimed to elucidate the diagnostic performance of the WFA+-M2BP assay for liver fibrosis assessment in NASH patients in Taiwan.

2 METHODS 2.1 Patients

Patients with biopsy-proven NASH were consecutively enrolled from a tertiary referral medical center in Southern Taiwan. The eligible NASH patients were Taiwanese patients aged ≥18 years. Patients with CHB, CHC, HIV infection, autoimmune hepatitis, primary biliary cirrhosis, sclerosing cholangitis, Wilson's disease, α1-antitrypsin deficiency, current alcohol misuse or history of alcohol misuse (>20 g/day), previous liver transplantation, or evidence of HCC were excluded from the study. The study was conducted in accordance with the Declaration of Helsinki. The Institutional Review Board of Kaohsiung Medical University Hospital approved this study before it began. Written informed consent for the interview, anthropomorphic measurements, blood sampling, and medical record review were obtained from patients before the enrollment.

2.2 Laboratory analyses

WFA+-M2BP was quantified using a lectin-antibody sandwich immunoassay with a fully automatic HISCL-5000 immune analyzer (Sysmex Co., Hyogo, Japan). Body mass index (BMI) and other anthropomorphic items were measured using the current standard methods. Serum biochemical variables, including aspartate aminotransferase,13 alanine aminotransferase,7 gamma-glutamyltransferase, alkaline phosphatase, total cholesterol, triglyceride, high-density lipoprotein cholesterol, fasting plasma glucose, albumin,7 and platelet count were measured using a conventional automated blood analyzer. The fibrosis-4 (FIB-4) index was calculated using the following formula: age (years) × AST [U/L]/platelets [109/L] × ALT [U/L])1/2. The aminotransferase-to-platelet ratio index (APRI) was calculated using the following formula: AST [U/L] × 100/platelets [109/L] × ALT [upper limit of normal] [U/L]). The NAFLD fibrosis score (NFS): −1.675 + 0.037 × Age (years) + 0.094 × BMI (kg/m2) + 1.13 × impaired glucose tolerance/diabetes (yes = 1, no = 0) + 0.99 × (AST/ALT ratio) − 0.013 × platelets (×109/L) − 0.66 × serum albumin (g/dl).14 The laboratory tests were performed at the time of liver biopsy.

2.3 Histological analyses

For each recruited NASH patient, a liver biopsy specimen was collected via the percutaneous route using a 16-G biopsy needle. At least 2 cm length was taken and fixed in 10% formalin buffer, and biopsy samples were stained with hematoxylin–eosin. The results were reported by an experienced pathologist and concurred with a peer review of the slides. Liver histology was staged as F0-4 according to the Metavir scoring system.15 Advanced fibrosis was defined as fibrotic stage ≥3 based on pathology (bridging fibrosis or cirrhosis).

2.4 Statistical analyses

Frequency was compared between groups using the χ2 test, with the Yates correction or Fisher's exact test. Results are expressed as mean values ± standard deviation and were compared between groups using analysis of variance and the Student's t test or nonparametric Mann–Whitney U test when appropriate. The area under the curve was compared using receiver operating characteristic analysis to determine the cut-off value of the WFA+-M2BP level in predicting the liver fibrosis stage. The Cochran–Armitage test was used to evaluate trends in serum WFA+-M2BP levels among patients with different liver disease severities. Linear regression analysis was used to test the variables correlated with the WFA+-M2BP levels. Stepwise logistic regression analysis was performed to analyze the factors associated with fibrosis. The strength of each association is presented as a regression coefficient with a 95% confidence interval (CI) and p-value. Statistical significance was set at p < 0.05. Quality control procedures, database processing, and analyses were performed using SPSS 20 statistical package (SPSS Inc., Chicago, IL).

3 RESULTS 3.1 Patient characteristics

Eighty patients with biopsy-proven NASH were recruited consecutively. Their basic characteristics are shown in Table 1. The mean age of the patients was 46.4 ± 14.1 years, and 51 (63.8%) were men. There were 26 (32.5%) patients with diabetes mellitus and 27 (33.8%) patients had a history of hypertension. The average BMI, AST, and ALT were 29.0 kg/m2, 60.3, and 102.9 IU/L, respectively. The NFS of the patients was −2.4, whereas the APRI and the FIB-4 value were 0.7 and 1.3, respectively. The cut-off index (COI) of the WFA+-M2BP level was 1.1. Twelve (15%) patients had advanced fibrosis, defined as Metavir fibrosis stage 3 or more.

TABLE 1. Basic characteristics of the biopsy-proven NASH patients Total (N = 80) Age (years, mean [SD]) 46.4 (14.1) Male gender, n (%) 51 (63.8) DM, n/N (%) 26 (32.5) Hypertension, n/N (%) 27 (33.8) BMI (kg/m2, mean [SD]) 29.0 (5.7) Platelet count (×103 U/L, mean [SD]) 241.6 (61.8) r-GT (U/L, mean [SD]) 78.2 (87.7) AST (IU/L, mean [SD]) 60.3 (28.2) ALT (IU/L, mean [SD]) 102.9 (46.6) α-fetoprotein (ng/ml, mean [SD]) 3.8 (2.3) Total cholesterol (mg/dl, mean [SD]) 212.0 (38.0) Triglycerides (mg/dl, mean [SD]) 164.7 (107.4) HDL-C (mg/dl, mean [SD]) 46.2 (24.4) Albumin (g/dl, mean [SD]) 4.4 (0.3) HbA1C (mg/dl, mean [SD]) 6.3 (1.0) Fasting plasma glucose (mg/dl, mean [SD]) 110.4 (27.2) APRI (mean [SD]) 0.7 (0.5) FIB-4 (mean [SD]) 1.3 (1.1) NAFLD fibrosis score (mean [SD]) −2.4 (1.4) WFA+-M2BP (COI, mean [SD]) 1.1 (0.8) Fibrosis stage 0/1/2/3/4 (n) 22/29/17/11/1 Abbreviations: ALT, alanine aminotransferase; APRI, aspartate aminotransferase-to-platelet ratio index; AST, aspartate aminotransferase; BMI, body mass index; COI, cut-off index; DM, diabetes mellitus; FIB-4, fibrosis-4; HDL-C, high-density lipoprotein cholesterol; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; r-GT, r-glutamyl transferase; SD, standard deviation; WFA+-M2BP, Wisteria floribunda agglutinin-positive Mac-2 binding protein. 3.2 The performance of WFA+-M2BP in assessing fibrosis stage

Figure 1 shows a boxplot of WFA+-M2BP at different fibrotic stages. The mean values of WFA+-M2BP were 1.0 (0.5), 1.0 (0.5), 0.8 (0.5), and 2.2 (1.3) in fibrotic stage 0, 1, 2, and 3–4, respectively (linear trend p = 0.005). The optimal cutoff value of fibrotic stage ≥1 and 2 were 1.50 COI (p = 0.56; sensitivity = 25.9%, specificity = 90.0%, and accuracy = 43.8%) and 1.63 COI (p = 0.35; sensitivity = 31.0%, specificity = 88.2%, and accuracy = 67.4%), respectively (Table 2). The optimal cut-off value of WFA+-M2BP in predicting advanced fibrosis (F3-4) was 1.37 COI, yielding a sensitivity of 75.0%, specificity of 79.4%, positive predictive value (PPV) of 39.1%, negative predictive value (NPV) of 94.7%, and accuracy of 78.7%, respectively (p < 0.001) (Table 2).

Correlation of serum Wisteria floribunda agglutinin-positive Mac-2 binding protein (WFA+-M2BP) with liver fibrosis

TABLE 2. WFA+-M2BP values for assessment of liver fibrosis Optimal cutoff (COI) AUC p Sensitivity (%) Specificity (%) PPV (%) NPV (%) Accuracy (%) F ≧ 1 1.50 0.54 0.56 25.9 90.9 88.2 31.7 43.8 F ≧ 2 1.63 0.56 0.35 31.0 88.2 60.0 69.2 67.4 F ≧ 3 1.37 0.83 <0.001 75.0 79.4 39.1 94.7 78.7 Abbreviations: COI, cut-off index; NPV, negative predictive value; PPV, positive predictive value; WFA+-M2BP, Wisteria floribunda agglutinin-positive Mac-2 binding protein. 3.3 Other noninvasive tools have been used for advanced fibrosis assessment

Furthermore, we validated the performance with other commonly used noninvasive tools. The optimal cut-off value of FIB-4 for predicting advanced fibrosis was 2.80, yielding a sensitivity of 58.3%, specificity of 97.1%, PPV of 77.8%, NPV of 93.0%, and accuracy of 91.3%, respectively (p = 0.001) (Supplementary Table 1). Moreover, the optimal cut-off value of NFS for predicting advanced fibrosis was −1.65, yielding a sensitivity of 70.0%, specificity of 81.2%, PPV of 38.6%, NPV of 94.5%, and accuracy of 79.8%, respectively (p = 0.02) (Supplementary Table 2). The area under the ROC curve for predicting advanced fibrosis of WFA+-M2BP, FIB-4, and NFS were 0.83, 0.80, and 0.73, respectively.

3.4 Combined FIB-4 to increase the performance of WFA+-M2BP

Thereafter, we used FIB-4 to increase the performance of WFA+-M2BP in predicting advanced fibrosis. Compared to the 1.37 COI of WFA+-M2BP, the 2.80 value of FIB-4 had higher specificity (97 vs. 79%) and accuracy (91 vs. 79%) than WFA+-M2BP. However, WFA+-M2BP had a higher sensitivity (75 vs. 58%) and negative predictive value (95 vs. 93%) than FIB-4. An algorithm using two sequential steps was designed to improve the performance (Figure 2). Combining WFA+-M2BP with FIB-4 significantly increased the diagnostic performance for advanced fibrosis, yielding a specificity, PPV, and accuracy of 100, 100, and 93%, respectively (Table 3).

An algorithm by using sequential two steps of fibrosis-4 (FIB-4) and Wisteria floribunda agglutinin-positive Mac-2 binding protein (WFA+-M2BP)

TABLE 3. Accuracy of WFA+-M2BP and FIB-4 in predicting advanced fibrosis F012 (n = 68) F34 (n = 12) p-Value SEN SPE PPV NPV ACC n (%) n (%) % % % % % WFA+-M2BP ≥ 1.37 14 (20.6) 9 (75.0) <0.001 75 79 39 95 79 FIB-4 ≥ 2.80 2 (2.9) 7 (58.3) <0.001 58 97 78 93 91 WFA+-M2BP ≥ 1.37 and FIB-4 ≥ 2.80 0 (0) 6 (50.0) <0.001 50 100 100 92 93 Abbreviations: FIB-4, fibrosis-4; NPV, negative predictive value; PPV, positive predictive value; WFA+-M2BP, Wisteria floribunda agglutinin-positive Mac-2 binding protein. 3.5 The factors predicted advanced liver fibrosis

Compared to patients without advanced liver fibrosis, those with advanced liver fibrosis were older (54.5 Â years vs. 43.9 years, p = 0.01), had a lower proportion of men (25.0 vs. 70.6%, p = 0.01), lower platelet count (216.3 × 103 U/L vs. 246.2 × 103 U/L, p = 0.01), and a higher proportion of WFA+-M2BP ≥ 1.37 COI (75.0 vs. 20.6%, p < 0.001), NFS ≥ −1.65 (58.3 vs. 17.6%, p = 0.002) and FIB-4 ≥ 2.80 (58.3 vs. 2.9%, p < 0.001) (Table 4). Multivariate regression analysis demonstrated that WFA+-M2BP ≥ 1.37 COI (OR/CI: 9.49/1.63–55.21, p = 0.01) and FIB-4 ≥ 2.80 (OR/CI: 38.18/4.89–297.93, p = 0.001) were significant predictors of advanced liver fibrosis. NFS no longer played an independent role to be associated liver disease severity.

TABLE 4. Factors associated with advance liver fibrosis F012 (n = 68) F34 (n = 12) p-Value Logistic regression OR 95% CI p-Value Age (years, mean [SD]) 43.9 (13.3) 54.5 (14.3) 0.01 Male gender, n (%) 48 (70.6) 3 (25.0) 0.01 BMI (kg/m2, mean [SD]) 29.2 (5.8) 27.7 (5.6) 0.25 Diabetes, n/N (%) 22/68 (32.4) 4/11 (36.4) 1.00 Sugar (AC, mean [SD]) 110.8 (27.9) 108.4 (25.2) 0.97 Cholesterol (mg/dl, mean [SD]) 210.6 (31.9) 220.5 (67.1) 0.80 Triglyceride (mg/dl, mean [SD]) 168.9 (113.7) 140.3 (60.6) 0.38 Platelet count (×103 U/L, mean [SD]) 246.2 (59.8) 216.3 (72.8) 0.01 AST (IU/L, mean [SD]) 56.4 (25.0) 85.4 (35.7) 0.01 ALT (IU/L, mean [SD]) 101.5 (46.4) 116.2 (47.3) 0.30 GGT (IU/L, mean [SD]) 71.4 (63.5) 122.9 (174.0) 0.60 APRI 0.6 (0.5) 1.1 (0.5) 0.002 FIB-4 ≥ 2.80, n (%) 12 (17.6) 7 (58.3) <0.001 38.18 4.89–297.93 0.001 NAFLD fibrosis score ≥ −1.65 0.6 (0.5) 7 (58.3) 0.002 WFA+-M2BP ≥ 1.37, n (%) 14 (20.6) 9 (75.0) <0.001 9.49 1.63–55.21 0.01 Abbreviations: ALT, alanine aminotransferase; APRI, aspartate aminotransferase-to-platelet ratio index; AST, aspartate aminotransferase; BMI, body mass index; DM, diabetes mellitus; FIB-4, fibrosis-4; NAFLD, nonalcoholic fatty liver disease; r-GT, r-glutamyl transferase; SD, standard deviation; WFA+-M2BP Wisteria floribunda agglutinin-positive Mac-2 binding protein. 4 DISCUSSION

Accurate and easy-to-access noninvasive fibrosis assessment is challenging in a clinical setting. In a biopsy-proven NASH study, we demonstrated that serum WFA+-M2BP performed well in liver fibrosis assessment, particularly in patients with advanced liver fibrosis. Serum WFA+-M2BP ≥ 1.37 COI could predict advanced fibrosis with high accuracy in our patients with NASH. Our results demonstrated that the accuracy increased from 78.7 to 93%, while the PPV was 100% when adding FIB-4 for advanced fibrosis assessment. Our study showed the validation evidence for WFA+-M2BP measurement and explored the potential increasing performance of the two methods in the noninvasive assessment of liver fibrosis.

In 2013, WFA+-M2BP was initially identified in Japan and has been widely investigated in other regions. It is a glycoprotein in the ECM and secreted from HSCs. WFA+-M2BP activates Mac-2 expression in Kupffer cells (KCs), which accelerates the fibrogenesis of HSCs.16 KCs have profibrotic functions, and it has been reported using in vivo models of hepatic injury in transgenic mice. The serum level of WFA+-M2BP increased with the progression of fibrosis. Our previous studies demonstrated that WFA+-M2BP reflects the fibrosis status in patients with CHB and CHC.17, 18 Further studies also extended the application showing that WFA+-M2BP predicted HCC development and liver-related events in patients with chronic liver diseases.17-19

NAFLD is the most prevalent liver disease in Asia, with a prevalence of 25%. It reaches 40–50% in Taiwan and is currently a huge health burden.20, 21 NAFLD might progress to liver fibrosis or cirrhosis and develop HCC.22 Fibrosis is a major determining factor for disease outcomes in NAFLD.23 Several noninvasive methods to assess the severity of NAFLD have been developed in the past decade. WFA+-M2BP levels were significantly correlated with the severity of liver fibrosis in patients with NAFLD.24 WFA+-M2BP level could also predict fibrosis progression and HCC development in patients with NAFLD.12, 25 Jang et al. showed that serum WFA+-M2BP level had a moderate correlation with fibrosis grade in 113 biopsy-proven NASH patients. Logistic regression analysis showed that WFA+-M2BP level was the only independent factor associated with F > 2 and F > 3.12 However, Kamada et al. demonstrated that during the early stages of fibrosis (F1, F2), WFA+-M2BP levels did not increase significantly in their study by recruiting 510 biopsy-proven NASH patients.11 Our results are consistent with a previous observation indicating that WFA+-M2BP was competent in predicting advanced fibrosis in NAFLD. Although progression to liver-related events and HCC is not commonly seen in NAFLD, our study validated a competent biomarker in the prediction of advanced fibrosis, which is a foundation toward grave outcomes in a clinical setting.

The cut-off values of WFA+-M2BP in advanced fibrosis were similar in CHB, CHC, and NAFLD according to our previous and current studies (CHB 1.54, CHC 1.42, NAFLD 1.37 COI). However, the performance for predicting advanced fibrosis needs to be addressed because of the more subtle progression of metabolic liver disease. FIB-4 has been widely used as a biomarker for predicting fibrosis in chronic liver disease.17, 18 Our study further demonstrated that the combination of WFA+-M2BP and FIB-4 could predict fibrosis more precisely, which echoes the previous study showing that combined FIB-4 with WFA+-M2BP had a high PPV and specificity in patients with NAFLD.26 The sequential diagram using FIB-4 as the first step followed byWFA-M2BP further provided the clinical utility in distinguishing liver fibrosis. For F34 patients (n = 5) misjudged asFIB-4 ≤ 2.8, three patients (60%) could be identified by having high serum WTA-M2BP level. By contrast, for the two F0-2 patients misjudged as FIB-4 ≥ 2.8, both of them had low WTA-M2BP levels. Taken collectively, WTA-M2BP may play a complementary role in addition to FIB-4 in predicting NASH related liver fibrosis. Further studies should assess the performance of the combination in predicting outcomes and liver-related events among patients with NASH in a long-term manner.

This study has several limitations. Our patients had a lower baseline fibrosis score, and the number of patients with advanced fibrosis was limited. However, the histopathological diagnosis was made by a well-experienced pathologist who was blinded to the clinical profile of the patients. Therefore, our patients provided a whole spectrum of real-world data to validate their performance in a clinical setting. It is a challenging task in a clinical setting with reluctance of receiving liver biopsy at present. Future multicenter collaborative study may overcome the limited patient numbers, particularly those of advanced fibrosis stage. Second, we did not use imaging-based or other serum-based noninvasive tools for fibrosis validation.27 A well-designed algorithm with sequential noninvasive tools, either serum-based or imaging-based, is mandatory for a useful clinical application.

In conclusion, serum WFA+-M2BP levels could assess liver fibrosis in NASH, particularly in patients with advanced liver fibrosis. WFA+-M2BP combined with FIB-4 increased the prediction of advanced liver fibrosis in Taiwanese patients.

CONFLICT OF INTEREST

The authors declare no conflict of interest.