Tri-antennary tri-sialylated mono-fucosylated glycan of alpha-1 antitrypsin as a non-invasive biomarker for non-alcoholic steatohepatitis: a novel glycobiomarker for non-alcoholic steatohepatitis

Non-alcoholic steatohepatitis (NASH) is a progressive form of non-alcoholic fatty liver disease (NAFLD) that may lead to liver cirrhosis or hepatocellular carcinoma. Here, we examined the diagnostic utility of tri-antennary tri-sialylated mono-fucosylated glycan of alpha-1 antitrypsin (AAT-A3F), a non-invasive glycobiomarker identified in a previous study of NASH diagnosis. This study included 131 biopsy-proven Japanese patients with NAFLD. We evaluated the utility of AAT-A3F in NASH diagnosis, and conducted genetic analysis to analyse the mechanism of AAT-A3F elevation in NASH. Serum AAT-A3F concentrations were significantly higher in NASH patients than in NAFL patients, and in patients with fibrosis, lobular inflammation, and ballooning. Hepatic FUT6 gene expression was significantly higher in NASH than in NAFL. IL-6 expression levels were significantly higher in NASH than in NAFL and showed a positive correlation with FUT6 expression levels. The serum-AAT-A3F levels strongly correlated with hepatic FUT6 expression levels. AAT-A3F levels increased with fibrosis, pathological inflammation, and ballooning in patients with NAFLD and may be useful for non-invasive diagnosis of NASH from the early stages of fibrosis.


Characteristics of the NAFLD patients. A total of 131 NAFLD patients (57 with NAFL, 74 with NASH)
were enrolled in this study. Patients were divided into NAFL and NASH groups based on the pathological appearance of their liver biopsy specimens. The NASH group had a higher proportion of females than the NAFL group. Age, AST, HbA1c, FBS, and HOMA-IR were significantly higher in the NASH group than in the NAFL group. In contrast, albumin and LDL-C were significantly lower in the NASH group than in the NAFL group. BMI, Plt, ALT, AFP, and TG were not significantly different between the NAFL and NASH groups. Serum cCK18 and M2BPGi levels were significantly higher in the NASH group than in the NAFL group. Both fibrosis-prediction equations (FIB-4 index and APRI) were significantly higher in the NASH group than in the NAFL group. As shown in Table 1, the NAFLD activity scores (NAS) in the NAFL group for steatosis were 0-3 (n = 0, 23, 22, and 12, respectively); those for lobular inflammation were 0-3 (n = 24, 30, 3, and 0, respectively); and those for hepatocyte ballooning were 0-2 (n = 54, 2, and 1, respectively). The NAS (NAFLD activity score) in NASH group for steatosis were 0-3 (n = 0, 23, 30, and 21, respectively); those for lobular inflammation were 0-3 (n = 0, 38, 31, and 5, respectively); and those for hepatocyte ballooning were 0-2 (n = 1, 46, and 27, respectively). The Brunt classifications of the fibrosis stage in the NAFL group were 0-4 (n = 30, 27, 0, 0, and 0, respectively), while those in the NASH group were 0-4 (n = 2, 37, 10, 23, and 2, respectively).
The cut-off value of serum AAT-A3F for diagnosis of early NASH (Brunt stages 0-1), was set to 7.9 μM in the ROC analysis. The AUROC of AAT-A3F was 0.696, and the sensitivity, specificity, PPV, NPV of AAT-A3F were 79%, 58%, 56%, and 80%, respectively. The cut-off value of other markers for early NASH diagnosis was set to 977 for cCK18, 0.79 for M2BPGi, 1.44 for FIB4 index, and 0.86 for APRI, similar to NASH diagnosis. The AUROC values of early NASH were 0.665 for cCK18, 0.673 for M2BPGi, 0.624 for the FIB4 index, and 0.648 for the APRI (Table 4, Fig. 3b).

Discussion
In patients with NAFLD, NAFL shows a non-progressive clinical course, whereas NASH is a serious disease with a high risk of both overall and liver-related morbidity and mortality [2][3][4] . Therefore, differentiating between NAFL and NASH is extremely important in the clinical management of NAFLD patients. Although liver biopsy is still considered a gold standard for differentiating between NASH and NAFL 5 , a non-invasive and reliable diagnostic biomarker is required.
Here, we validated the diagnostic utility of AAT-A3F identified in our previous study 15 . Serum AAT levels were not different between patients with NASH and NAFL, but AAT-A3F was significantly elevated in patients with NASH versus patients with NAFL. In addition, AAT-A3F levels increased significantly not only with hepatic fibrosis, but also with intrahepatic inflammation and hepatocyte ballooning. AAT is a major, liver-derived circulating protein that functions as a natural inhibitor of various serine proteases and is a component of the acute-phase response 16 . During inflammation, the liver synthesises acute phase proteins (APPs), including AAT, which are thought to contribute to the inflammatory response, but also play a central role in limiting local and systemic inflammation 17 . APPs play a dual role in inflammation, and AAT can induce the production of both pro-inflammatory interleukin 1 (IL-1) and anti-inflammatory IL-1 receptor antagonist (IL-1Ra) in peripheral blood mononuclear cells. Other studies have shown that AAT inhibits neutrophil superoxide production 18 , prevents hepatocyte apoptosis 19 , and functions as an endogenous inhibitor of pro-inflammatory cytokine production in whole blood 20 . Thus, AAT is thought to be related to intrahepatic inflammation.  www.nature.com/scientificreports www.nature.com/scientificreports/ To clarify the mechanism underlying AAT-A3F elevation in NASH, we evaluated gene expression levels in liver biopsy tissues. The expression level of the FUT6 gene, whose protein product is responsible for outer arm fucosylation of glycoproteins secreted from the liver 21 , was significantly higher in NASH than in NAFL. It was previously reported that FUT6 expression was positively regulated by IL-6 22 , and hence, we determined the hepatic expression level of IL-6. IL-6 expression level was significantly higher in NASH than in NAFL, as previously described 23 . A positive and significant correlation between the expression levels of IL-6 and FUT6 was observed, indicating that inflammatory response accompanying IL-6 production in NASH liver can lead to FUT6 upregulation. The serum level of AAT-A3F was strongly correlated with the hepatic FUT6 expression level. Therefore, we conclude that hepatic inflammation caused the elevation in FUT6 expression, resulting in an increase in outer arm fucosylation of AAT produced in the liver.
In this study, the AUROC for NASH diagnosis of AAT-A3F was 0.687, which was higher than cCK18. The apoptosis marker cCK18 was reported to be useful for distinguishing between healthy subjects and those with NAFL, and further between patients with NAFL versus those with NASH 24 . In this study, the diagnostic utility of AAT-A3F was found to be higher than that of cCK18 in our cohort. The AUROC for NASH diagnosis of AAT-A3F was lower than that of M2BPGi and FIB-4 index, however, the AUROC values of AAT-A3F for diagnosing early NASH was higher than that of biomarkers tested in this study. Furthermore, while the cut-off values of cCK18, M2BPGi, FIB-4 index, and APRI for diagnosing NASH and early NASH were very close, the cut-off values in AAT-A3F were nearly doubled. This suggests that AAT-A3F can stratify NASH by the degree of fibrosis. These results may perhaps be because AAT-A3F also reflects liver inflammation and fibrosis. Although liver fibrosis was considered the most important factor for predicting the prognosis of NASH 25 , the mortality rate is reported to increase since mild fibrosis 26,27 . In addition, lobular inflammation has been reported as a risk factor for the progression of liver fibrosis, compared to fatty liver alone 28 . In general, the prognosis of patients with NAFLD is often lower than that of the general population 29 ; therefore, it is necessary to intervene at an early stage with heathier diet and exercise. Identifying biomarkers that indicate lobular inflammation and hepatocyte ballooning as well as fibrosis such as AAT-A3F can lead to early diagnosis and treatment of NASH, which helps reduce mortality.
In the previous study, we comprehensively analysed N-glycans of serum glycoproteins in NAFLD patients using the FPG method and found that AAT-A3F could be a useful biomarker for NASH diagnosis. FPG is a useful method for biomarker discovery, but it is not suitable for examinations of many samples, because it requires many steps. Therefore, we have developed a simplified IPG method suitable for the measurement of a large number of   www.nature.com/scientificreports www.nature.com/scientificreports/ samples. Using the IPG method, we verified the usefulness of AAT-A3F in this study. Because the IPG method can be used for any proteins, this method will be applicable to the analysis of glycobiomarkers related to various diseases.
The primary limitation of our study was that the number of patients included was relatively small, and our NAFLD patients were enrolled from a regional liver disease hospital. Therefore, the potential for referral bias cannot be ruled out in our study. Thus, our findings for liver biopsy-proven NAFLD might not represent patients with NAFLD in the general population. Therefore, larger multicentre studies are required to validate our results. The secondary limitation is that we used liver biopsy as the gold standard for assessing the utility of AAT-A3F. This technique has serious limitations including those related to sampling errors and pathological assessment. Therefore, liver biopsy specimens from various facilities were also centrally evaluated by a hepatopathologist in this study. The third limitation is the small sample size that was subjected to gene expression analyses. Although in a multi-centre study, the number of the liver tissue storage are limited, we analysed all possible cases.
In conclusion, this multicentre study revealed that AAT-A3F was especially increased in not only fibrosis, but also pathological inflammation and hepatocyte ballooning in patients with NAFLD, and AAT-A3F was considered useful for non-invasive diagnosis of early NASH. Furthermore, it was beneficial for the diagnosis of NASH by combining AAT-A3F with fibrosis markers such as M2BPGi and FIB4 index depending of progression stage of NASH (Fig. 5).

Study populations.
In this study, we included 131 biopsy-proven NAFLD patients from the Hokkaido University Hospital and other affiliated hospitals belonging to NORTE study group. These patients were examined between 2007 and 2017 and underwent a percutaneous liver needle biopsy. Biopsied liver samples were embedded in paraffin blocks following standard procedures and stained with haematoxylin and eosin, Masson's www.nature.com/scientificreports www.nature.com/scientificreports/ trichrome stain, and Gitter stain. All biopsy specimens were evaluated by a hepatopathologist blinded to the clinical data. Samples were investigated and quantified based on NAFLD-activity scoring 30 for steatosis (0-3), lobular inflammation (0-3), and hepatocyte ballooning (0-2). NASH was diagnosed when steatosis, ballooning, lobular inflammation and hepatocyte ballooning of any grade were all present at the same time 31,32 . Individual fibrosis parameters were scored based on the fibrosis stage of the Brunt classification 33 . Early NASH was defined as NASH with no or mild fibrosis (Brunt stages S0-1). Advanced fibrosis was classified as a Brunt stage of 3-4 (S3-4).
The exclusion criteria for this study included daily alcohol consumption (>30 g for men or >20 g for women) and another hepatic disease such as hepatitis B, hepatitis C, hepatocellular carcinoma, autoimmune hepatitis, primary biliary cholangitis, primary sclerosing cholangitis, hemochromatosis, α1-antitrypsin deficiency, Wilson's disease, or congestive liver disease, according to our previous study 15 . The study protocol complied with the ethical guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board of Hokkaido University Hospital and each participating hospital. Written informed consent to participate in this study was obtained from each patient. This study was registered in the UMIN Clinical Trials Registry as UMIN000025644. Insulin resistance was evaluated based on the homeostasis model assessment as an index of insulin resistance (HOMA-IR) value using the following equation: HOMA-IR value = fasting insulin (µU/mL) × fasting glucose (mg/dL)/405. Caspase-cleaved cytokeratin 18 (cCK18) and M2BPGi were measured as markers of hepatocellular apoptosis and liver fibrosis, respectively. The formula used for predicting liver fibrosis from non-invasively obtained data was as follows: fibrosis 4 (FIB4) index = age × AST (IU/L) × Plt ( × 10 9 /L) −1 × √ALT (IU/L) −1 , as reported previously 34  Measurement of tri-sialylated mono-fucosylated tri-antennary glycan (A3F) bound to alpha-1 antitrypsin (AAT). Serum concentrations of AAT-A3F were determined as previously described 14 . Briefly, AAT was purified from 50 μL of serum using commercially available immune-affinity beads (Alpha-1 Antitrypsin Select, GE Healthcare). The serum AAT concentration was determined by ELISA (Human Serpin A1 DuoSet ELISA, R&D Systems, Minneapolis, MN) with a standard AAT protein (Sigma-Aldrich, A9024). After tryptic digestion of AAT, N-glycans were released with N-glycosidase F. Released glycans were purified and derivatised with N α -((aminooxy)acetyl) tryptophanylarginine methyl ester using a glycoblotting procedure 36 and analysed with an Ultraflex II TOF/TOF mass spectrometer. The amount of A3F was quantified by the area ratio using ethyl-esterified A3 (3.8 pmol) as an external glycan. Serum concentrations of AAT-A3F were calculated by dividing the amount of A3F by AAT concentrations of each patient.
Real-time quantitative pcR analysis. We evaluated gene expression levels in liver tissues from 8 patients with NAFL and 16 patients with NASH. Total RNA was extracted from biopsied liver samples using the AllPrep DNA/RNA/Protein Mini Kit (Qiagen, Venlo, Netherlands). The quantity and quality of the isolated RNA were determined using an Agilent 2100 Bioanalyzer with the RNA 6000 Nano Kit (Agilent Technologies). Total RNA (500 ng) was reverse transcribed using PrimeScript RT Master Mix (Perfect Real Time) (Takara Bio, Shiga, Japan). Real-time quantitative PCR was performed using the Applied Biosystems 7500 Real-Time PCR System. Gene-specific primers were purchased from Takara Bio. Gene expression values were normalised to that of the reference gene, peptidylprolyl isomerase A, and calculated based on the ΔΔCt method 37 . www.nature.com/scientificreports www.nature.com/scientificreports/ Statistical analysis. Statistical analysis was conducted using R software (version 3.5.1). Results are presented as the mean ± standard deviation (SD) for continuous data, and as numbers (percentages) for categorical data. Statistical analysis included descriptive statistics, analysis of variance, Mann-Whitney U test, Fischer's exact test, and Spearman R correlations. All P values were two-tailed, and the threshold of statistical significance was set at P < 0.05. The diagnostic performance of the markers was assessed by analysing receiver operating characteristic (ROC) curves. The probabilities of true positives (sensitivity), true negatives (specificity), positive-predictive values (PPVs), and negative-predictive values (NPVs) were determined for selected cut-off values, and the AUROC was calculated for each index. Cut-off points were determined based on the optimum sum of the sensitivity and specificity.