Association of Serum Autotaxin Levels with Liver Fibrosis in Patients with Chronic Hepatitis C

Metabolized by liver sinusoidal endothelial cells, autotaxin (ATX) is a secreted enzyme considered to be associated with liver damage. We sought to clarify the diagnostic ability of ATX for liver fibrosis in 593 biopsy-confirmed hepatitis C virus (HCV)-infected patients. The diagnostic accuracy of ATX was compared with clinical parameters and the established fibrosis biomarkers Wisteria floribunda agglutinin-positive Mac-2-binding protein, FIB-4 index, AST-to-platelet ratio, and Forn’s index. Median ATX levels were consistently higher in female controls and patients than in their male counterparts (P < 0.01). Serum ATX concentration increased significantly according to liver fibrosis stage in overall and both genders (P < 0.001). The cutoff values of ATX for prediction of fibrosis stages ≥F1, ≥F2, ≥F3, and F4 were 0.8, 1.1, 1.3, and 1.7 mg/L, respectively, in male patients and 0.9, 1.7, 1.8, and 2.0 mg/L, respectively, in female patients. The area under the receiver operating characteristic curve for ATX to diagnose fibrosis of ≥F2 (0.861) in male patients was superior to those of FIB-4 index and Forn’s index (P < 0.001), while that in female patients (0.801) was comparable with those of the other markers. ATX therefore represents a novel non-invasive biomarker for liver fibrosis in HCV-infected patients.

ATX levels reportedly vary significantly between genders in healthy subjects 26 , they nonetheless represent a novel serum marker of liver fibrosis 27 . However, the clinical features of ATX are inconclusive due to small cohort sizes.
This study aimed to clarify the clinical characteristics of ATX in patients chronically infected with HCV based on histological assessment evidence. We also compared the fibrosis predictive ability of ATX with that of other serum liver fibrosis markers.

Results
Baseline clinical characteristics. The baseline clinical characteristics of our cohort are summarized in Table 1. Of the 593 patients enrolled, 292 (49.2%) were male. Median age was 58 years. The vast majority of patients were seen for histological assessment prior to antiviral treatment with interferon-based therapy or DAAs. Fibrosis stage was F0 for 8 cases (1.3%), F1 for 284 cases (47.9%), F2 for 144 cases (24.3%), F3 for 103 cases (17.4%), and F4 for 54 cases (9.1%). Comparisons of baseline clinical characteristics between genders revealed several significant differences (Table 1), although the distributions of fibrosis stage and activity grade were comparable (P = 0.888 and P = 0.171, respectively).
Correlation between ATX and liver fibrosis stage. We excluded the 8 patients (6 male and 2 female) who were diagnosed as having F0 stage from the analysis to evaluate the association between ATX and histologically significant fibrosis. Correlations between ATX and fibrosis stage (Fig. 2) and activity grade (Fig. 3) of the liver were investigated for all subjects and according to gender. Overall, median ATX in healthy controls and patients with fibrosis stage F1, F2, F3, and F4 was 0.76, 1.05, 1.64, 1.93, and 2.16, respectively. Median ATX levels in male healthy controls and male patients with fibrosis stage F1, F2, F3, and F4 were 0.70, 0.90, 1.33, 1.56, and 2.17 mg/L, respectively, while those in women were 0.82, 1.33, 1.96, 2.21, and 2.16 mg/L, respectively. Serum ATX increased significantly according to liver fibrosis progression (r = 0.72, P < 0.001 overall, r = 0.77, P < 0.001 in men, and r = 0.73, and P < 0.001 in women) and activity (r = 0.46, P < 0.001 overall, r = 0.69, P < 0.001 in men, and r = 0.66, P < 0.001 in women). Female ATX levels at each fibrosis stage were significantly higher than corresponding values in men (F1: P < 0.001, F2: P < 0.001, F3: P < 0.005) apart from F4 (P = 0.708).
Correlations between ATX and fibrosis markers. The correlation coefficients between ATX and other clinical markers (albumin, AST, alanine aminotransferase (ALT), γ -glutamyltransferase (GGT), alpha-fetoprotein (AFP), and platelet count) and non-invasive fibrosis markers (WFA + -M2BP, APRI, FIB-4 index, and Forn's index) for male and female patients after analysis by Spearman's rank correlation coefficient test are shown in Table 2. ATX was significantly positively correlated to AST, ALT, GGT, and AFP and significantly negatively correlated to albumin and platelet count. We observed the strongest correlation between ATX and WFA + -M2BP in total (r = 0.75, P < 0.001), male (r = 0.83, P < 0.001), and female (r = 0.71, P < 0.001) patients in addition to remarkable correlations with other fibrosis markers (Fig. 4).

Comparison of fibrosis markers by AUC.
The ROC curves of ATX for predicting early fibrosis (≥ F1 vs. controls), significant fibrosis (≥ F2), severe fibrosis (≥ F3), and cirrhosis (F4) in total, male, and female patients are presented in Fig. 5 and comparisons of AUC between ATX and other fibrosis markers in predicting fibrosis stage are summarized in Table 4

Discussion
This investigation confirmed that ATX levels were significantly higher in women than in men in both healthy controls and patients with HCV, as earlier reported 26,28 . Accordingly, we compared ATX values with those of  clinical markers according to gender in the study, and would advocate the same approach for assessing liver fibrosis stage with this biomarker. Tokumura et al. found serum ATX activity to be increased in normal pregnant women 19 , possibly since it also stabilizes blood vessels and is essential for blood vessel formation during fetal development 29,30 . However, the precise reason for this discrepancy remains under debate. ATX may participate in female reproductive biology, which is an important issue that requires investigation in future studies comparing ATX by gender. Serum levels of ATX were verified to correlate both with liver fibrosis stage as well as with other clinical and non-invasive fibrosis markers in our study cohort of 593 patients that was substantially larger than a previous report of 74 patients 27 . Notably, ATX levels were validated by biopsy-determined histological scores in all patients. During the progression of liver fibrosis, liver sinusoidal endothelial cells are known to exhibit a loss of several receptors and sinusoidal endothelial fenestrae, causing the capillarization of sinusoids and uptake impairment of various substances 31 . ATX is generally considered to be high in the serum despite being cleared from the circulation and degraded in liver sinusoidal endothelial cells within minutes 32 . The phenotypic changes occurring in the liver during liver fibrosis may lead to reduced ATX clearance, thereby increasing circulating ATX concentration. Pleli et al. reported that ATX levels correlated closely with Child-Pugh stage and model of end stage liver disease (MELD) score, suggesting that ATX was an indicator of liver injury severity 28 . Although we could not address their findings since patients with hepatic failure were excluded from our cohort, our data confirmed that ATX level was highest in cirrhosis (F4).
ROC assessment of ATX confirmed the enzyme to be a comparably reliable marker for detecting fibrosis stage, especially F1 and F2. DAAs are currently the standard of care in chronic HCV infection as these IFN-free regimens achieve high rates of a sustained virological response with few side effects. Patients who are elderly or who exhibit persistently normal ALT levels are particularly well suited for DAA therapy. It is becoming difficult to perform liver biopsy in such individuals due to refusal and contraindications that include anticoagulant and antiplatelet agents. Moreover, most patients with normal ALT levels have none to early fibrosis 33 . ATX may therefore be useful to accurately determine early to significant fibrosis stage prior to initiating DAA treatment in these patients.
Interestingly, ATX was positively correlated with ALT and AFP as well as with histological activity grade. Elevated ALT and AFP reflect liver inflammation and regeneration. A recent study showed that ATX might be causally linked to immune activation during HCV infection and uncovered a decline in ATX to levels comparable to those of uninfected participants within 24 weeks of DAA therapy 34 . ATX was also correlated with transaminase and APRI, which was compatible with our own data. Hence, serum ATX level may be associated not only with liver fibrosis, but also with liver inflammation and regeneration.
A longitudinal investigation of patients with HCV with respect to ATX and clinical features, including long-term prognosis and liver carcinogenesis, is needed. The clinical utility of ATX should be assessed in other chronic liver diseases as well. In addition, the latest EASL NAFLD guidelines have established alcohol consumption of 20 g ethanol daily for females and 30 g daily for males to exclude alcoholic liver disease 35,36 . We had originally set this limit at 60 g ethanol/day in 1982 when the first liver biopsies were performed and therefore only had patient data for ≤ or > 60 g/day. Such differences in alcohol intake may have influenced fibrosis differently in chronic hepatitis C and added bias to the results of this study. Lastly, we cannot exclude the possibility that ATX levels may have been underestimated in patients with liver cirrhosis since liver biopsy is sometimes contraindicated in such individuals in the clinical setting.
In conclusion, ATX represents an accurate, non-invasive biomarker for liver fibrosis estimation in patients with chronic HCV infection and merits more comprehensive establishment based on gender.

Subjects.
A total of 593 patients chronically infected with HCV, all of whom having received liver biopsy at Shinshu University Hospital in Matsumoto, Japan, between 1982 and 2015, were enrolled in this cross-sectional study. Their racial background was uniformly Japanese. The diagnosis of chronic hepatitis C was based on previously reported criteria as the presence of serum HCV antibodies and detectable HCV RNA 37 . One hundred and sixty subjects (80 male and 80 female) whose liver function tests were within normal levels were recruited as healthy controls. Age distribution was equally distributed among male and female controls (twenties: 20 subjects, thirties: 20 subjects, forties: 20 subjects, fifties: 20 subjects). All patients and controls were negative for hepatitis B surface antigen and antibodies to hepatitis B core antigen and the human immunodeficiency virus. No patients complicated with HCC were included. Patients who were diagnosed as having alcoholic liver disease, defined as an average daily consumption of > 60 g of ethanol, were excluded. Patients who exhibited evidence of other liver disease, such as non-alcoholic liver disease, primary biliary cholangitis, or autoimmune hepatitis, were excluded as well. This study was reviewed and approved by the Institutional Review Board of Shinshu University Hospital (Matsumoto, Japan) (approval number: 3244), and written informed consent was obtained from all participating subjects. The investigation was conducted according to the principals of the Declaration of Helsinki.
Laboratory testing. All laboratory data were obtained on the same day or within 14 days after liver biopsy.
ALT, AST, GGT, AFP, and other relevant biochemical tests were performed using standard methods.

Measurement of ATX.
Blood samples were obtained on the same day or within 14 days after liver biopsy.
Separated serum samples were immediately stored at − 20 °C until testing. Serum ATX antigen concentration was determined using a specific 2-site enzyme immunoassay. For preparation of the immunoassay, R10.23 was digested with pepsin, and the purified F(ab) 2 form was isolated using phenyl-5PW hydrophobic column chromatography (Tosoh Co., Tokyo, Japan) to avoid the nonspecific binding of human antibodies to animal IgG types, such as human anti-mouse antibodies. In-house magnetic beads were coated with R10.23 F(ab) 2 and placed in a reaction cup, and then 35 ng of alkaline phosphatase-labeled R10.21 in assay buffer (5% BSA, 5% sucrose, 10 mmol/l Tris-HCl, 10 mmol/l MgCl 2 , pH 7.4) was added. The ATX assay reagent was prepared by immediate freeze-drying of the reaction cup for subsequent use in a commercial automated immunoassay analyzer AIA-system (Tosoh Co.). The AIA-system process included automated specimen dispensation, incubation of the reaction cup, bound/free washing procedure, 4-methylumbelliferyl phosphate substrate dispensation, fluorometric detection, and result report. Antigen-antibody reaction time was 10 min and the first result was reported within 22 min. The throughput of the system was 60 samples/h using the AIA-600 II system 26 . Liver biopsy and histological evaluation. Liver biopsies were performed on all patients by percutaneous sampling of the right lobe with a 14-gauge ACECUT needle (TSK LABORATORY Co., Hyogo, Japan) or by laparoscopic liver biopsy with a 13-gauge Silverman needle (Olympus Co., Tokyo, Japan). All biopsy specimens were 1.5 cm or more in length. Formalin-fixed, paraffin-embedded specimens were prepared and used for histopathological studies. Sections of 4 μ m in thickness were cut from each paraffin block and stained with hematoxylin and eosin, periodic acid-Schiff after diastase digestion, or Azan-Mallory staining. All liver biopsy samples were independently evaluated by 2 investigators (AM and ET) who were blinded to the clinical data. Fibrosis stage and activity grade were assessed independently on each histological section by both investigators. In the case of a discrepancy, histological sections were simultaneously reviewed using a multi-pipe microscope to reach a consensus. Liver fibrosis stage (F0-4) and activity grade (A0-3) were determined according to the METAVIR scoring system 38 , as follows: F0 = no fibrosis, F1 = portal fibrosis without septa, F2 = portal fibrosis with few septa, F3 = numerous septa without cirrhosis, and F4 = cirrhosis. Significant fibrosis was defined as ≥ F2, severe fibrosis as ≥ F3, and cirrhosis as F4. Statistical analysis. Statistical analysis and data visualization were carried out using IBM SPSS Statistics version 23.0 (IBM, Chicago, IL) and StatFlex version 6.0 (Artech Co., Ltd., Osaka, Japan) software. Data are presented as median ± interquartile range (IQR) for continuous variables. Groups were compared using the chi-square test for categorical variables. Correlations between fibrosis stage and serum ATX or other fibrosis markers were analyzed by means of Spearman's rank test. Diagnostic accuracy was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC). Comparisons of paired AUCs and 95% confidence intervals (CIs) were carried out using the nonparametric Delong test 39 . Cutoff values were identified by the Youden index, and the nearest clinically applicable value to the cutoff was considered as the optimal threshold for clinical convenience. All statistical tests were two-sided and evaluated at the 0.05 level of significance.  Table 4. Comparisons of AUC between ATX and other fibrosis markers in predicting fibrosis stage for all, male, and female patients. Abbreviations: AUC, area under the curve; ATX, autotoxin; WFA + -M2BP, Wisteria floribunda agglutinin-positive Mac-2-binding protein; APRI, AST-to-platelet ratio; ND, not determined; F, fibrosis. a P < 0.01 vs. ATX; b P < 0.05 vs. ATX.