Introduction

At present, chronic hepatitis B (CHB) remains a serious global health concern, with approximately 296 million people living with CHB infection worldwide1. Given that the virus is not directly cytopathic in itself, it is the conflict intensity between the hepatitis B virus (HBV) and the host immune response2,3 that determines the control of HBV infection, the development of liver damage (fibrosis, cirrhosis or hepatocellular carcinoma [HCC]) and the progressive selection of HBV variants. The typical course of HBV infection tends to involve a hepatitis B e antigen (HBeAg)-positive phase with high HBV DNA replication. This is followed by an HBeAg seroconversion phase with clinical remission and a decline in HBV viral load4. Nevertheless, during the HBeAg seroconversion phase, immune pressure can select a number of genetic variants in the basal core promoter (BCP) and precore (PC) regions that downregulate or abolish HBeAg production with active DNA replication5,6. Chronic HBeAg-negative hepatitis B has become the predominant type of CHB infection in France7 and worldwide. However, the impact of these mutations on the natural course of infection and on the severity of liver damage has not yet been clearly established for some of these mutants. The G1896A mutation in codon 28 of the PC region is the most frequently described mutation in this region and creates a premature stop codon that stops HBeAg synthesis8. Some studies reported a relationship between the G1896A mutation and severe liver hepatitis, whereas others reported no associations between the mutation and clinical outcomes9,10. In the BCP region, the A1762T/G1764A double mutant has been found to downregulate precore mRNA production, resulting in reduced HBeAg secretion11. An association between the A1762T/G1764A double mutant and severe hepatic lesions or increased risk for HCC has been frequently described12,13,14,15,51.

Detection of basal core promoter and precore mutations

The BCP (nucleotide position 1742–1849) and PC (nucleotide position 1814–1900) regions of the HBV genome were amplified by nested PCR as previously described17,52.

Nucleotide mutations were defined by their differences from the consensus sequence, and dual signals (mixed type) were considered a mutant type. The dominant viral strain was determined by Sanger sequencing and defined as > 50% of the virus.

Fibrosis marker analysis

The secreted fibrosis markers were measured in the serum samples of patients by multiplex fluorescent-bead-based technology (Luminex 200, Austin, TX, USA) using two commercial Luminex screening assay kits: a customized Luminex Assay kit from R&D Systems (Lilles, France) for MMP-1, resistin, Collagen IA1, PDGF-BB and TIMP-1 and the Bio-Plex Pro TGF-β assay from Bio–Rad Laboratories (Marnes-la-Coquette, France).

In brief, the samples were diluted before incubation with specific antibody-coated fluorescent beads according to the manufacturer’s recommendations. After washing, 50 beads were analyzed with the Luminex 200™ analyzer and Bio-Plex Manager software version 6 (Bio–Rad Laboratories), and the analyte concentrations of the samples were estimated through the serial dilution of cytokine standards (MMP-1 sensitivity < 3 pg/mL; resistin sensitivity < 20 ng/mL; Col IA1 sensitivity < 100 pg/mL; PDGF-BB sensitivity < 50 pg/mL; TIMP-1 sensitivity < 800 pg/mL; and TGF-β1 sensitivity < 15 pg/mL).

Statistical analysis

Quantitative variables are described as the means (standard error) or median (interquartile range) when appropriate. Categorical variables are described as total numbers (percentages).

Univariate analyses were performed to compare fibrosis markers in the serum samples according to fibrosis stage using Wilcoxon tests53 or according to the profile of mutation using Kruskal–Wallis tests54 and Steel–Dwass55 tests for post-hoc pairwise comparison. Proportions were compared using the chi-square test of independence. The data distributions were visualized in violin plots.

A first exploratory multivariate analysis was performed using FAMD, with the R package FactoMineR56 to analyze the variability of the data taking into account both the mutation profile and serum marker levels. Logistic regressions were then used to identify factors associated with advanced fibrosis. The candidate variables were MMP-1, resistin, Col IA1, PDGF-BB, TIMP-1, TGF- β1 and mutation profiles. The variables identified by univariate regressions (p value < 0.20) were then introduced in a multivariate analysis (stepwise backward logistic regression using Akaike Information Criterion for model selection). The regression formula used for this purpose also allowed us to identify factors independently associated with advanced fibrosis. Finally, we assessed the Spearman correlations between the serum levels of liver markers according to the mutation profile using the R package corrplot.

A p value lower than 0.05 was considered statistically significant. Statistical analyses were performed using R version 3.6.2.