Hepatitis B virus (HBV) is a global public health burden that plays an essential role in the development of cirrhosis and liver cancer (1). An estimated 2 billion people worldwide have been infected with HBV, and approximately 250 million are chronic hepatitis B (CHB). Globally, the proportion of hepatocellular carcinoma (HCC) patients induced by HBV infection is 10%–25% (2). The situation is even more severe in the developing world particularly in Asia and Africa since approximately 80% of HCC is related to HBV infection (3). Among different HBV genotypes, genotype C is the most common in South Asia (4), and it is a higher risk for the occurrence of HCC (5, 6).

Hepatitis B surface antibody (HBsAb) is normally considered as an indicator of immune protective status against HBV infection (7). Theoretically, the simultaneous presence of both positive hepatitis B surface antigen (HBsAg) and HBsAb in circulation should not appear in the same patient. However, this understanding has been challenged by several studies indicating the HBsAg+/HBsAb+ in chronic HBV patients from 2.63% to 8.9% (7–11). The mechanism underlying the simultaneous presence of HBsAg and HBsAb remains unclear and even controversial. Moreover, concurrent HBsAg and HBsAb also increased the risk for the development of HCC in CHB (8, 12). Therefore, it is necessary to investigate the clinical as well as biological implications of HBsAg+/HBsAb+ serological profile, which might help in the precise clinical management such as proper therapeutic interventions and disease monitoring in HBV-related diseases (13).

The HBV genome encompasses four overlapping open reading frames (ORFs). S-ORF can be subdivided further into three coding regions (preS1, preS2, and S) encoding three different lengths of HBV surface antigen. HBsAg is the shortest, which is 226 amino acids (aa) long containing several important antigenic epitopes. In particular, the “a” determinant in HBsAg spanning the region of aa124-147 within the major hydrophilic region (MHR; aa99-169) is the major conformational epitope exposed on the external surface of HBV for the antigen–antibody reaction (14). The antigenicity alteration of HBsAg could influence the recognition and neutralization (NT) between HBsAg and HBsAb (15, 16). Several previous studies have attributed the HBsAg+/HBsAb+ serological patterns to aa substitutions that occurred in the “a” determinant of HBsAg resulting in immune evasion (10, 17). The mutation of the HBV genome acquired during the disease process led to HBsAb being incapable of binding mutated HBsAg (18–20). Moreover, viral quasispecies refers to a viral population consisting of extremely large numbers of mutations, termed mutant spectra or mutant clouds (21). HBV strains in infected patients were considered to be a mixture of different quasispecies including mutated HBV and wild-type isolates (22, 23). The existence and expansion of mutated quasispecies might cause the coexistence of HBsAg and anti-HBs antibodies leading to the selection of immune escape (24).

Most studies investigated HBsAg+/HBsAb+-associated mutations on individual sequences derived by the Sanger sequencing or PCR. Due to the technical limitation, rare variants and the vital information within these non-dominant sequences might be ignored. Next-generation sequencing (NGS) is a powerful technique to study quasispecies diversity in viral strains (25). In particular, its high coverage makes its sensitivity much superior to that of traditional methods and offers a powerful method for detecting minor variants as low as 0.25%–5% (26, 27). But till now, only a few studies have utilized NGS towards this special HBV serological status (28–30). Even for those studies, the interpretation of sequence datasets and bioinformatics methods applied to handle such high-dimensional NGS data remains insufficient (31). In addition, the similarity and difference in this special serological pattern for CHB and HCC patients in a large cohort have not been explored.

The big data from NGS have substantial advantage to improve our understanding of the coexistence of HBsAg and HBsAb more comprehensively. Therefore, this study aimed to conduct high-throughput parallel sequencing for a systematic characterization of the HBV s gene between HBsAg+/HBsAb+ and HBsAg+/HBsAb−. To our knowledge, this is the first study to utilize HBV quasispecies features interpreting the underlying mechanism for patients with concurrent presence of positive HBsAg and HBsAb in both CHB and HCC patients.

This study performed a comprehensive exploration of characteristics and diversities of HBV s region in the CHB/HCC patients with the coexistence of HBsAg and HBsAb. The systematic analysis includes high frequency of mutations, quasispecies patterns, antigenicity/immunogenicity, physical properties, and clinical characterization. This research for the first time investigated the similarity and difference of HBV s gene features between DP and SP in both the CHB and HCC patients. More importantly, it further revealed the selection of HBsAg immune escape mechanism from the HBsAb based on a large-representative multicenter NGS cohort and powerful bioinformatics analysis.

The study showed that the DP group accumulated more variants than the SP group in HBsAg. It was observed that higher mutations are harbored in the s region, which was in line with the previous researches (44, 45). This study discovered two mutations, L104F and M133T, in both the CHB_DP and HCC_DP patients. The DP profile in CHB is associated with accumulative HBsAg mutations P62L, L77R, Y100S, Q101H, M103L, I126S, and G145R, whereas P46L, T118P, I126T, T125P, T131N, M133S, and I195M are harbored in HCC. Moreover, the presence of mutations in and around antigenic epitopes, such as I126S, I126T, and I195M, may impact the conformational structure of this region, hence altering the antigenicity/immunogenicity of HBsAg and leading to immune escape in the end ( Figures 4C–E ) (46). This study identified that the DP profile showed different mutation patterns of s126 in patients between CHB and HCC. Identified mutation I126S in CHB and I126T in HCC were both correlated with the alteration of HBsAg antigenicity ( Figure 5 ). Moreover, the NT experiment was performed to validate the impact of the mutation frequency of I126S in CHB_DP or I126T in HCC_DP on the NT rate. The result showed that the NT rate has a reduced trend in the DP patients with I126S or I126T ( Figure S4 ). In addition, due to the overlap between HBsAg and rt gene, I195M, corresponding to rt changes of M204V/I, is considered as a lamivudine-resistant mutation, which may also raise the possibility of drug-selected HBsAg and the potential to escape NT by HBsAb (47).

Besides missense mutations, the percentage of stop codon mutations identified by NGS was also found much higher in the DP patients than in the SP patients. The proportion of W199* in the DP groups was found to be significantly greater than that in the SP groups. Moreover, the average mutation frequency of C69* was much higher in both the CHB_DP and HCC_DP patients. Higher stop codon mutations could not only abrogate full-length wild-type HBsAg quantification to affect the proper diagnosis of disease severity (48) but also reduce the efficient recognition and NT by HBsAb. Both two mutations have been described to contribute to the progression of HCC (49, 50). Besides, W182* rate in the CHB_DP group was significantly higher than the rates in the CHB_SP group, whereas no obvious difference was observed in patients with HCC. It was reported that premature stop at codon 182 may provide an important contribution to the progression of HCC (51), which might explain the differences of W182* between the CHB_DP and HCC_DP groups.

Our results suggest that viral quasispecies features in the s gene play an important role in the development of DP profile. The DP profile was partially associated with the emergence of HBV escape mutations ( Figure 2 ). Meanwhile, there were yet some patients who failed to be clustered together by several high frequencies of amino acid substitutions in HBsAg. Interestingly, when NGS was applied to characterize HBV quasispecies, it was found that five quasispecies features patterns can well classify those patients who were misclassified based on differential mutations ( Figure 3 ). This result suggests that quasispecies features provide a better indication of HBsAg+/HBsAb+ than mutations. Entirely, the distribution of feature patterns was complex in the DP patients but relatively simple for SP patients, which may be attributed to the diversity of HBV quasispecies ( Figure 1 ).

The advantage of quasispecies features in interpreting the immune escape in the DP patients was further studied. Data from NGS revealed that the genetic distance, as well as the ratio of dN/dS, varied between the DP and SP groups ( Figure 1 ). The mean viral genetic distance was greater in the DP group (p = 0.0002) combined with a higher dN/dS ratio (p = 0.0006), indicating stronger positive selection acting on HBsAg to enhance viral fitness (52). These findings provided clues for differences in selection pressures between the SP and DP patients. Besides, decreased immunogenicity of CTL epitopes in DP might also contribute to immune escape (53). As MHR A epitope and CTL epitopes situated on the surface of HBV have been described to be potential NT domains, we can speculate that increased genetic diversity may lead to altered T-cell and B-cell epitopes, which can change their immunogenicity leading to escape from host immune response (54). Thus, we further investigated the change of antigenicity/immunogenicity in these regions to study the coexistence mechanism of HBsAg+/HBsAb+. Notably, our findings demonstrated that the percentages of antigenicity for MHR and immunogenicity for CTL epitope Se ( Figures 4A, B ) were both reduced in the DP patients. To study whether the accumulation of hotspot mutations in the DP patients can contribute to immune evasion, we then compared the change of dp.AS between high accumulation of hotspot mutations (defined as the total mutation rate of differential high-frequency mutation sites) and low accumulation of hotspot mutations in the DP samples. We found that a considerably reduced dp.AS in the DP patients was shown in the MHR of HBV in the low accumulation of hotspot mutations group, compared with the high accumulation of hotspot mutation group ( Figure S5 , p = 0.0128). For the CHB_DP and HCC_DP subgroups, there is a consistent tendency that patients in the CHB_DP and HCC_DP groups were found to have a greater dp.AS in high accumulation of hotspot mutation groups ( Figure S5 ).

More interestingly, our study demonstrated that entropy features are also associated with the clinical indicators of HBV DNA duplication and liver inflammation for the DP samples. Previous studies (55–57) have observed a negative relationship between viral diversity (Shannon entropy) and HBV DNA level in s region. Our study further indicated this inverse relationship firstly in the DP samples. One possible explanation could be that viruses adapted to host immune pressure at the cost of viral replication, which imperils the efficient HBV DNA replication (55). We speculated that strong host immune stress suppressed viral replication. Meanwhile, it also increased the selection of mutations in HBsAg, which as a consequence increases the viral diversity. This might explain the reason why with the accompanying high features of s region in the DP patients, the viremia status was lower. Moreover, we also analyzed the HBsAb level with mutation rates, HBV-DNA, ALT/AST, and entropy features. We performed the correlation analysis between the mutation rate of differential high-frequency mutation sites and HBsAb levels in CHB_DP and HCC_DP, respectively. Among those high-frequency mutation sites, there was a statistically significant correlation between the mutation frequency of site 125 and HBsAb levels in the HCC_DP samples (R = −0.38, p = 0.04) ( Figure S6A ). We further analyzed the correlation between the total mutation feature value of s gene and the HBsAb levels. In the HCC_DP samples, there was a tendency that increased s gene mutation feature values were associated with the decreased HBsAb levels ( Figure S6B ). While the current data can only show the tendency, more experiments are needed to validate whether a higher s gene mutation contributes to the decline of HBsAb in the future. Subsequently, we analyzed the difference between the HBsAb level and HBV-DNA, ALT/AST, or entropy features in the DP patients. Although we did not observe the obvious impact of the change of HBsAb level on entropy features and clinical routine tests ( Figure S7 ), lower entropy features were associated with higher HBV DNA titer and higher ALT and AST values ( Figure 6 ). We further research the HBsAb loss in HBsAg+/HBsAb+ patients. During the follow-up of our study (n = 73), there were only two DP samples with HBsAb loss: case 3 (CHB_DP) and case 6 (HCC_DP). The HBsAb level of case 3 decreased from 10.74 to <0.2 IU/L after 101 days. Likewise, in 2016, the HBsAb level of case 6 was 89.8 IU/L, which decreased to 4.56 IU/L in 2018. Both of the cases were also accompanied by the decreasing HBV DNA (case 3: dropped from 382,000 to <50 IU/ml; case 6: dropped from 15,700 to <50 IU/ml). An analysis of the hot mutations and entropy feature was performed. Several mutants in the MHR were identified, and a low total entropy feature of the MHR was also observed in these two cases.

We also acknowledge certain limitations in this study. Firstly, insertions and deletions were not included due to technical limitations, which may also result in the change of the conformational structure and antigenicity of HBsAg. Secondly, since the HBV genome is compact and contains overlapping genes, mutations in HBsAg can also influence the functions of other encoding regions. The potential of concomitant mutations in these regions might be missed in this study. Therefore, how insertions/deletions and other HBV regions have an impact on the immunogenicity of HBsAg should be evaluated in our further research.

In conclusion, utilizing HBV s region sequence analyses, mathematical models, and immune characterization, this study demonstrated a comprehensive characteristic of viral quasispecies between the SP and DP patients by NGS from a large-representative multicenter cohort. Integration of sequence feature patterns is capable of revealing the controversial underlying mechanism of a mixed viral population in the DP patients, and it contributes to better interpretation of the coexistence of HBsAg and HBsAb and benefits to precise clinical management.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm.nih.gov/sra/?term=PRJNA765447.

The studies involving human participants were reviewed and approved by the Institutional Ethics Committee of the leading medical center (Shanghai Eastern Hepatobiliary Surgery Hospital, EHBHKY2015-01-004). The patients/participants provided their written informed consent to participate in this study.

YW, XX, and SC performed the investigation, methodology, and formal analysis and drafted the manuscript. CH, JZ, and ZG performed data curation and technique assistance. ED, YL, XH, and CW provided the clinical samples. CG obtained the funding, designed and coordinated the overall study, and revised the manuscript with MF. All authors contributed to the article and approved the submitted version.

This work was supported by the Innovation Group Project of Shanghai Municipal Health Commission [2019CXJQ03], the National Natural Science Foundation of China [81572072], the Science and Technology Commission of Shanghai Municipality [17JC1404500 and 16441907200], and the Shanghai “Rising Stars of Medical Talent” Youth Development Program [2019-72].

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.