A noninvasive model to predict liver histology for antiviral therapy decision in chronic hepatitis B with alanine aminotransferase < 2 upper limit of normal

Hepatitis B virus (HBV) infection is still a major public health problem in the world. It is one of the most serious and common health problems, affecting more than 2 billion people [1]. Once HBV chronic infection, the covalently closed circular DNA (ccc-DNA) was difficult to be removed from hepatocytes.

Studies have shown that HBV replication and subsequent immune-mediated liver damage are the main key factors for disease progression. The inevitable progression of HBV infection to persistent or intermittent liver inflammation and fibrosis greatly increases the risk of progression to cirrhosis, Hepatocellular carcinoma (HCC) and end-stage liver disease [2]. Early and timely antiviral treatment can reverse liver fibrosis and early cirrhosis, thereby reducing the incidence of HCC [3‐5].Therefore, the main purpose of chronic hepatitis B virus (CHB) treatment is to inhibit HBV replication, reduce the occurrence of liver decompensation and prevent the progress of disease as soon as possible [1]. The recent American Association for the Liver Diseases (AASLD) guidelines recommend antiviral therapy when alanine aminotransferase (ALT) ≥ 2 upper limit of normal (ULN), moderate or severe inflammation or significant fibrosis and HBV-DNA elevation [6]. European Association for the Study of the Liver (EASL) guidelines recommend that patients with ALT > ULN, moderate or severe inflammation or obvious fibrosis and HBV-DNA elevation should receive antiviral treatment [7].

The level of ALT is able to reflect the severity of liver injury and is an important biochemical index to evaluate liver injury. However, due to its effectiveness was limited, not all patients with HBV infection will continue to increase of ALT level [8, 9]. About one fifth of patients with CHB with normal ALT level may have significant liver fibrosis [10]. Patients with high ALT level are more likely to have liver inflammation or fibrosis, but those with long-term ALT level lower than 2 ULN value of HBV infection, liver biopsy is recommended. However, due to the invasive nature of liver biopsy with serious complications, short-term non-repeatability, sampling errors and differences between observer and within the observer, the examination was rejected by some patients [11, 12]. Therefore, more and more attention has been paid to the prediction of liver histology by noninvasive diagnostic methods.

In this study, we aimed to constructed a noninvasive model based on serum markers to accurately identify moderate and severe inflammation or significant fibrosis related to CHB, which is helpful to the decision-making of antiviral treatment in patients with chronic hepatitis B with ALT < 2 ULN and to reduce the need for liver biopsy as much as possible.

This study retrospectively analyzed 577 patients with CHB who underwent liver biopsy from October 2008 to November 2019 in the Department of infectious diseases of Zhejiang Provincial People's Hospital. According to the guidelines of Helsinki declaration in 1975 and the approval of the local research and ethics committee of Zhejiang Provincial People's Hospital, all patients were given written informed consent for analysis. HBV infection was defined as HBsAg positive for at least 6 months [13]. The inclusion criteria include: the diagnosis of CHB infection, the level of ALT less than 2 ULN (ULN = 40U/L), detectable HBV-DNA with a level > 10² IU/ml. Exclusion criteria include: such as hepatitis C virus (HCV), hepatitis D virus (HDV), hepatitis E virus or human immunodeficiency virus (HIV) Co-infection and other causes of liver disease, alcoholic liver disease, decompensated cirrhosis, HCC, nonalcoholic fatty liver disease (NAFLD). Autoimmune liver disease, insufficient liver biopsy samples and incomplete clinical data.

In this study, we established a noninvasive model to predict liver histology to determine moderate or severe inflammation or significant fibrosis, and to guide the decision-making of antiviral treatment in patients with chronic hepatitis B with ALT < 2 ULN. The results showed that AST, anti-HBC and GGT were independent risk factors for antiviral therapy. The combination of AST, anti- HBC and GGT has better diagnostic performance than single variable.

In patients with chronic hepatitis B, persistent liver inflammation is a major risk factor for the development of cirrhosis, HCC and end-stage liver disease [1, 12]. The latest guidelines suggest that early and timely control of liver inflammation to prevent disease progression to end-stage liver disease is the primary task of liver fibrosis detection [15]. The latest EASL guidelines point out that antiviral therapy should be performed when ALT > ULN [7]. The AASLD guidelines indicate that patients with ALT > ULN but < 2 ULN need to consider the severity of liver histological changes [6]. Therefore, it is very important to evaluate the progress of liver inflammation and fibrosis in the early stage for the treatment and prognosis in the later stage. However, liver biopsy is an invasive operation with related complications, and its clinical application is limited [16]. At present, most assessments of liver fibrosis staging focus on non-invasive diagnostic methods, such as transient elastography (TE). But TE itself has certain limitations, such as TE requires specialized centers, so it is less readily available, limited applicability in obese and ascites patients [17], and the results of TE are also affected by liver inflammation and activity [18]. Therefore, our research is to avoid the clinical need of liver biopsy by establishing noninvasive, repeatable and rapid indicators of antiviral treatment decision-making.

Serum ALT level is one of the main biochemical markers to evaluate liver inflammation. A study in a Korean population showed that some people with high viral load and ALT < 2 ULN had significant liver necrosis inflammation and fibrosis [19]. About 37% of CHB patients with normal or near normal ALT level, but had significant liver histologic changes [20‐22]. The results showed that ALT level was not enough to assess the severity of liver injury [22]. Some studies have shown that AST has better diagnostic performance than ALT in predicting hepatic necrotizing inflammation [19]. As an independent factor to predict significant fibrosis, AST level increased with the aggravation of liver fibrosis [23]. Some of the delay clearance in AST was related to ALT level [24], and some of the advanced liver fibrosis was related to mitochondrial damage [25]. Another study suggested that the increase of AST level was related to the advanced fibrosis [26]. Our study shows that AST level can better predict moderate and severe inflammation and significant fibrosis. Therefore, we should be closely monitored and evaluated AST level regularly.

Some studies have suggested that anti-HBC in serum biomarkers is a sensitive marker in the history of chronic HBV infection. After the appearance of HBsAg in the early stage of HBV infection, anti-HBC antibodies will be detected in serum soon [27, 28]. Serum anti-HBC level is closely related to the state of HBsAg and the activity of liver inflammation [28]. In patients with chronic hepatitis B whose ALT level is less than 2 ULN, the serum anti-HBC level has a better diagnostic performance in predicting moderate and severe liver inflammation, even in patients with normal or slightly elevated ALT level [22, 23]. Li et al. founded that the anti-HBC level increased with gradually severity of hepatic inflammation and fibrosis stage. Anti-HBC showed a high diagnostic accuracy for identifying moderate or severe inflammation in all patients and patients with ALT lower than 64 IU/L (AUROC = 0.768 and 0.767, respectively) in a study of 469 treatment-naïve CHB patients [22]. In our study, Anti-HBC is a major predictor of liver fibrosis, which has good diagnostic performance in predicting liver inflammation and necrosis.

In patients with chronic hepatitis B, GGT as an independent variable has a higher diagnostic performance for significant fibrosis [29]. GGT is more stable in predicting liver inflammation than ALT and AST. GGT is an important predictor of noninvasive fibrosis in patients with HBV infection [30‐32]. Therefore, GGT is a reliable predictor of liver histological changes in routine laboratory indexes. We should regularly monitor the dynamic changes of this index and give drug treatment when necessary.

The nomogram not only reflects the predicted value of each variable, but also reflects the complex interaction with other variables [33]. Through constructing the model, clinicians can more accurately assess the incidence of liver fibrosis, so as to provide guidance for better monitoring and treatment of patients with chronic hepatitis B.

According to logistic analysis, we set up a new combined model, which combines AST, anti-HBC and GGT. This combined model is simple and easy to use, and has a good effect on the decision-making of antiviral therapy. Through ROC curve analysis, the combined index is 0.700 and 0.742 in training set and verification set respectively, which is higher than AST, anti-HBC and GGT. If AGH index is higher than the high critical value, antiviral therapy is recommended; if AGH index is lower than the low critical value, regular follow-up is recommended; if AGH index is between the low and high critical value, liver biopsy is recommended.

The treatment of chronic hepatitis B was mainly based on serum HBV-DNA level, serum ALT level and hepatic histological severity [34]. However, in our study, single-factor and multi-factor logistic regression analysis showed that HBV-DNA level was not an independent factor influencing antiviral treatment. Therefore, the variable HBV-DNA level was not included in AGH model. Similarly, through logistic regression analysis, we found that HBeAg was not an independent factor in antiviral treatment decision. Previous studies have shown that HBeAg-positive and HBeAg-negative chronic hepatitis B patients have different factors related to liver histology [35]. AST, HBsAg, platelet (PLT) and albumin (ALB) were independent factors in HBeAg-positive patients [35]. Age, AST, HBV-DNA and PLT were independent factors in HBeAg-negative patients [36].

Our research has its own advantages. AGH model is a new index, which is composed of clinical routine laboratory tests and easy to obtain. It is suitable for most patients with chronic hepatitis B virus infection as well as be used to predict liver histology to avoid the need of liver biopsy. Our model used only serum markers which made it suitable for most medical facilities. We further evaluated the performance of other non-invasive models in antiviral treatment decision-making. As other models like Forn’s model, AST/ALT ratio (AAR), AST to PLT ratio Index (APRI), Fibrosis Index, FIB-4, Hui’s model, Goteborg University Cirrhosis Index (GUCI) and Zeng’s model [37‐44]. Generally, the diagnostic performance of most non-invasive models is lower than AGH model, but not as good as that of APRI and FIB-4, probably because of the different choices of participants.

This study also has some defects in this model. Firstly, the number of patients included is small, the clinical application should be included in a large number of patients for multicenter cohort study. Secondly, we constructed the model in the training set and randomly select another group in the same center to verify, but we insist that the credibility of the model can be enhanced by the validation in the multicenter study.

Our study shows that AST, Anti-HBC and GGT are independent variables in the decision-making of antiviral therapy for chronic hepatitis B. AGH model which is a new combined index, has a good diagnosis performance. It can provide a noninvasive prediction model for CHB patients to choose antiviral therapy, liver biopsy or regular follow-up, and help to reduce the clinical needs of liver biopsy.