Total serum bile acids predict therapy for HBeAg-negative chronic hepatitis B patients with borderline ALT and high HBV DNA

Introduction: The introduction of antiviral therapy in chronic hepatitis B (CHB) infection depends on precise evaluation of hepatic lesions. Total serum bile acids (TSBAs) are highly sensitive in monitoring liver dysfunction. We evaluated the predictive role of TSBAs for hepatic lesions in CHB patients with borderline alanine aminotransferase (ALT) and high level of hepatitis B virus (HBV) DNA copies. Methodology: 328 CHB patients were enrolled, 241 were hepatitis B e antigen (HBeAg)-positive and 87 were HBeAg-negative. Patients were further divided into two entities according to inflammation/fibrosis evaluated by liver biopsy, low-grade (inflammation grade < 2 and fibrosis stage < 2) and high-grade (inflammation grade ≥ 2 or/and fibrosis stage ≥ 2) cohorts. TSBAs were compared with noninvasive tools including aspartate aminnotransferase (AST)-to-platelet ratio index (APRI), fibrosis-4 (FIB-4) and red cell distribution width (RDW)-to-platelet ratio (RPR) to predict high-grade hepatic lesions in CHB subgroups. Results: TSBAs, APRI, FIB-4 and RPR were statistically different between low- and high-grade patients in HBeAg-positive cohort. Only TSBAs showed significant difference between low and high grade in HBeAg-negative patients. Similarly, APRI, FIB-4 and RPR were correlated with different division of inflammation/fibrosis only in HBeAg-positive while TSBAs were correlated with inflammation/fibrosis levels in both HBeAg-positive and HBeAg-negative groups. Of the four indicators, the receiver operating characteristic (ROC) curve analysis showed that TSBAs have the maximum AUC (area under the curve) in HBeAg-negative group but the minimum in HBeAg-positive cohort. Conclusions: TSBAs can be used for predicting antiviral therapy in CHB patients with HBeAg-negative, borderline ALT and high HBV DNA.


Introduction
Hepatitis B virus (HBV) infection is the leading cause of liver cirrhosis and hepatocellular carcinoma (HCC). The estimated data from 120 countries showed that the global prevalence of hepatitis B surface antigen (HBsAg) in 2016 was 3.9% (95% uncertainty interval [UI] 3.4-4.6), corresponding to 291,992,000 (251,513,000-341,114,000) infections [1]. In China, the mortality rate of HBV-related HCC increased from 12.88 per 100,000 people in 1990 to 16.42 per 100,000 people in 2016 [2]. The benefit of timely and effective antiviral therapy has been demonstrated in reducing endpoints of cirrhosis and HCC. Thus, it is important to evaluate the indicators for HBV treatment since not all patients with chronic hepatitis B (CHB) progress to lifethreatening complications [3].
Currently, a panel of non-invasive indices and tools are used to evaluate hepatic lesions and categorize patients to determine who are eligible for treatment [4]. The American Association for the Study of Liver Diseases (AASLD) and Asian-Pacific Association for the Study of the Liver (APASL) guidelines provide good evidence that patients with borderline alanine aminotransferase (ALT) > 2 ULN should have therapy after 3-6 months of observation [5,6]. However, the indication is less clear for those with borderline ALT levels (≤ 2 ULN), even for those who are considered at high risk of fibrosis, i.e., hepatitis B e antigen (HBeAg)positive, high HBV DNA and normal ALT but > 40 years of age. Liver biopsy is recommended for evaluation of inflammation and fibrosis via histological grading, although it is invasive and has up to 30% sampling error [7]. On the other hand, biochemical and radiological methods such as aspartate aminotransferase (AST)-to-platelet ratio index (APRI), fibrosis-4 (FIB-4), red blood cell distribution width-toplatelet ratio (RPR) and FibroScan have considerably reduced the need for liver biopsy [4,8]. However, these non-invasive tools still remain to be validated, especially in those with HBeAg-negative CHB patients where they may not provide an accurate correlation with hepatic lesions.
ALT is an enzyme released from damaged liver cells and therefore is an indicator of hepatocellular integrity. However, CHB patients with normal ALT may have debates and clinical dilemmas in monitoring hepatic injury especially in the HBeAg-negative cohort [9]. Bile acids are synthesized in the liver, excreted into the bile, and passed into the intestine where they are reabsorbed and transported back to the liver by the portal system. The enterohepatic circulation of bile acids is a highly efficient process based on efficient reabsorption by the liver. The total bile acids leaked into the peripheral circulation constitute total serum bile acids (TSBAs), which therefore reflects the synthetic, excretory and re-absorptive function of the liver. Thus, TSBAs have early sensitivity in mirroring both the initial phase as well as the longitudinal development in liver diseases [10]. In this study, we aim to investigate the predictive value of TSBAs for hepatic injury in CHB patients with normal or slightly elevated ALT and high HBV DNA. We compared the predictive performance of TSBAs both in HBeAg-positive and HBeAg-negative CHB patients with traditional noninvasive methods including APRI, FIB-4 and RPR.

Patients
This is a retrospective cross-sectional study of diagnostic test accuracy. We included 328 CHB patients in the Second Hospital of Anhui Medical University between 2012 and 2019. This hospital is a nonprofit academic medical center focused on integrated health care, education, and research. It was first accredited by the Joint Commission International (JCI) in 2017, with the catchment area of Anhui Province and served more than 2 million patients in 2021. CHB was defined as persistent presence of serum HBsAg for > 6 months. According to the AASLD and APASL recommendations, the inclusion criteria were as follows: (1) patients with age ≥ 18 years and who were treatment-naive; (2) ALT ≤ 2 ULN; (3) HBeAgpositive patients with HBV DNA ≥ 20000 IU/mL; (4) HBeAg-negative patients with HBV DNA ≥ 2000 IU/mL. The exclusion criteria were as follows: (1) coinfection with other hepatotropic viruses (hepatitis A/C/D/E virus) or human immunodeficiency virus (HIV); (2) coexistence of autoimmune hepatitis, alcoholic liver disease, non-alcoholic fatty liver disease, drug hepatitis, primary biliary cirrhosis, HCC, or any other type of cancer. (3) patients who were pregnant, and (4) who had hematological diseases or other diseases that could interfere with liver function tests. The study was conducted according to the principles of the Declaration of Helsinki and was approved by the Ethics Committee for the Second Hospital of Anhui Medical University (NO. 2019SHAMU0013), and all patients had signed informed consent.

Liver biopsy
An ultrasonography-guided percutaneous liver biopsy was performed with a 16-gauge needle before antiviral therapy in all CHB patients. Biopsy specimens were fixed in formalin, embedded in paraffin and stained with hematoxylin-eosin and reticular fiber staining or Masson's staining on each section. These specimens were evaluated by two independent and experienced pathologists who were unaware of the patients' identity and clinical information. Histologic characteristics of liver inflammation grade (G) and fibrosis stage (S) were determined using the Batts-Ludwig system [11,12]. This system is simple and easily reproducible via a clear graphic demonstration of what is meant by each grade and stage, and is most appropriate for management of individual patients included in this study. Inflammation was graded according to the amount of portal/periportal and lobular activity (G0-G4, count whichever is greater). Fibrosis was staged as follows: stage 0 (S0), the absence of fibrosis; stage 1 (S1), fibrous portal expansion; stage 2 (S2), periportal or rare portal-portal septa; stage 3 (S3), fibrous septa with architectural distortion; and stage 4 (S4), cirrhosis. G0-1 and S0-1 were considered to be low-grade hepatic lesions (< G/S2), indicating no or mild inflammation and fibrosis, respectively; G2-4 and S2-4 were considered to be high-grade (≥ G/S2) which indicates moderate to severe inflammation and fibrosis, respectively.

Statistical analysis
The data were expressed as a median along with the range of lower-upper quartiles (between 25th and 75th percentiles). Mann-Whitney U test was used to compare age, ALT, AST, GGT, ALP, ALB, GLB, TSBAs, CHO, HBV DNA, platelet/neutrophil/lymphocyte counts, RDW and APRI/FIB-4/RPR indices between HBeAg-positive and HBeAg-negative groups, and between low-grade (< G/S2) and high-grade (≥ G/S2) cohorts. The sample rates of hierarchical G and S panels between HBeAg-positive and HBeAg-negative groups were analyzed by Pearson's χ2 test or Fisher's exact test. Spearman rank-order correlation was employed to test the association between TSBAs, APRI, FIB-4, RPR and inflammation grades or fibrosis stages. Receiver operating characteristic (ROC) analysis was performed for TSBAs, APRI, FIB-4 and RPR. AUC (area under the curve), sensitivity and specificity at defined points of the curve were determined. Statistical analyses were performed using the SPSS version 22.0 (Chicago, IL) and MedCalc version 19.4 (Ostend, Belgium). A twotailed p < 0.05 was considered statistically significant.

TSBAs have good performance in predicting ≥ G/S2 lesions in HBeAg-negative CHB.
The ROC curves of TSBAs and the other three noninvasive indices for predicting ≥ G/S2 hepatic lesions in HBeAg-positive and HBeAg-negative CHB were shown in Figure 2, and the diagnostic accuracy are compared in Table 3  . Accordingly, the sensitivity, specificity, PPV and NPV of TSBAs were 67.35%, 71.05%, 75.00% and 72.79% respectively, with an optimal cut-off value of 5.40 μmol/L. Thus, TSBAs have good performance in predicting high-grade (≥ G/S2) lesions in CHB patients with HBeAg-negative.

Discussion
Accurate assessment of hepatic lesions is pivotal for the introduction of antiviral therapy in CHB patients. Liver biopsy is undoubtedly the gold standard for determining liver injury based on inflammation and fibrosis grading, especially in CHB patients with borderline ALT levels. However, sampling limitations and complications of invasive procedure need to be balanced against the benefits of liver puncture [7]. Thus, the non-invasive methods that allow for a more rational, evidence-based approach to aid clinical decisions are preferred and are more acceptable. The classical algorithmic models, APRI, FIB-4 and RPR, have been widely used for hepatic diseases to estimate liver fibrosis, predict antiviral therapy and reflect adverse prognosis [16].
: In this study, we showed that HBeAg-positive patients have significantly higher platelet counts than HBeAg-negative cohort, whereas ALT, AST and RDW (%) levels showed no significant difference between the two groups. This led to significantly lower APRI, FIB-4 and RPR values in the HBeAg-positive group, since all the three indices were calculated using platelet counts as denominators. These results are consistent with the findings of previous study [17] that compared PLT and FIB-4 between HBeAg-positive and HBeAgnegative groups. However, APRI was higher in HBeAg-negative cohort than that of HBeAg-negative patients in the same study [17]. The inconsistency may result from sampling differences as our study focused on CHB patients with ALT ≤ 2 ULN. It is very likely that the eligibility criteria also led to the lack of statistical difference in other liver function tests including GGT, ALB, GLB, ALP and CHO between HBeAg-positive and HBeAg-negative subjects (Table  1). Nevertheless, both TSBAs and HBV DNA were significantly higher in HBeAg-positive than that in HBeAg-negative cohort. It is also to be noted that HBeAg-positive patients were significantly younger than HBeAg-negative, just as it was found in the study by Goyal et al. [18]. One possible explanation is that HBV infection in older adults is usually mild and selflimiting, which is therefore prone to develop into HBeAg negative inactive carriers [19].
Mounting evidence has suggested that bile acid metabolism and HBV infection are interlinked by the sodium taurocholate co-transporting polypeptide (NTCP), which is responsible for sodium-dependent bile salts uptake by hepatocytes [20]. NTCP also functions as a cellular receptor for viral entry of HBV [20]. HBV infection may interfere with NTCPmediated enterohepatic circulation of bile acids, since the molecular determinants critical for HBV entry overlap with that for bile acid transport [20,21]. On the  other hand, in the study by Kim et al. and Xun et al., bile acids were shown to increase HBV gene expression and inhibit interferon-alpha activity [22,23]. This interaction may contribute to a better understanding of the increased TSBAs in CHB patients. Our data demonstrated that the percentage of CHB patients with no or mild fibrosis (S0-1) is significantly higher in HBeAg-positive group than the negative control (73.44% vs. 51.72%, p < 0.001). In contrast, the percentage of moderate fibrosis (S2) was observed to be higher in HBeAg-negative vs. HBeAg-positive subjects (34.48% vs. 17.01%, p = 0.001). Besides, no significant difference was observed in the distribution of S3, S4, and all inflammation grades (G0 to 4) between the two groups ( Table 1). These results are in agreement with our observation and that of other researchers that HBeAg seroconversion during the natural history of CHB is associated with a decline in serum HBV DNA levels and growth of liver fibrosis [24].
It is well documented that the non-invasive tools APRI, FIB-4 and RPR have the advantages of predicting liver fibrosis with high applicability (> 95%) and good interlaboratory reproducibility [25]. In this study, APRI, FIB-4 and RPR indices were statistically different between low-grade (< G/S2) and high-grade (≥ G/S2) subgroups, and were positively correlated with inflammation/fibrosis levels only in HBeAg-positive cohort. In contrast, TSBAs were observed to be significantly different between < G/S2 and ≥ G/S2, and correlated with inflammation/fibrosis levels both in HBeAg-positive and HBeAg-negative CHB patients. These results are consistent with metabonomic studies by Wang et al. [26] and Yin et al. [27] who showed that TSBAs were associated with pathological progression of CHB and therefore may be indicative of hepatitis Binduced cirrhosis. Interestingly, the Spearman's correlation coefficient of TSBAs was substantially higher in HBeAg-negative than that in HBeAg-positive. Furthermore, of the four indicators, TSBAs have the maximum AUC, PPV and NPV scores in the HBeAgnegative rather than HBeAg-positive subjects. It is to be noted that a higher NPV indicates that TSBAs can accurately rule out those who do not need treatment independent of liver biopsy. These data strongly suggest that TSBAs are superior to APRI, FIB-4 and RPR for predicting liver lesions in HBeAg-negative CHB patients.
This study has two limitations. Firstly, we studied a small number of CHB patients, especially the HBeAgnegative cohort. Secondly, this is a retrospective study which is limited to the data acquisition of TSBAs fluctuation after antiviral therapy. Thus, a prospective longitudinal study with sufficient sample size is needed to verify the indicative role of TSBAs in CHB-related hepatic lesions [28]. However, the prospect of this study is undoubtedly exciting in revealing that TSBAs are eligible for predicting ≥ G/S2 hepatic lesions in HBeAg-negative CHB patients with borderline ALT (≤ 2 ULN) and high HBV DNA. Early initiation of antiviral therapy is critical for CHB patients with these serological characteristics. Choi et al. have shown that untreated high viral load HBeAg-negative CHB patients without significant ALT elevation had higher risks of HCC and death/transplantation than treated active phase patients with elevated ALT [29].

Conclusions
Our results provide evidence that TSBAs have good performance in predicting high-grade (≥ G/S2) hepatic lesions in HBeAg-negative CHB patients with borderline ALT (≤ 2 ULN) and high HBV DNA. These findings can be used to optimize the antiviral therapy of CHB patients by concentrating on understanding TSBAs and traditional non-invasive methods such as APRI, FIB-4 and RPR.