Baseline characteristics of the study population
To study viral and immune correlations in the different CHB disease phases, we carefully selected a homogeneous cohort of untreated chronic HBV infected patients without any other comorbidities, attending our outpatient clinic. To rule out the impact of advanced liver fibrosis on any identified immune parameter, patients with advanced fibrosis (F2 fibrosis or higher) were excluded. As is typical for the natural history of CHB patients, IT patients were youngest among the patient cohort. Owing to the stringent definition criteria, differences in age, ALT and HBV DNA levels were observed between clinical phases. Unlike some recent reports, the qHBsAg level in this study was higher in GZ patients than in IT patients [22, 23] (Table 1).
Cytokine profiles in CHB patients with different stages of disease
To investigate whether CHB patients beyond current treatment criteria are characterized by a state of defective antiviral response, we analyzed the expression profiles of three major effector cytokines, IFN-γ, TNF-α and IL-2, produced by innate and adaptive immunity. The representative dot plots and gating strategies of flow cytometry analysis for T cell-, NK cell- and NKT cell-derived cytokines are shown in Supplementary Figure 1. We first analyzed their T cell-derived cytokine profiles and compared them with those in healthy controls. The frequency of IFN-γ+ CD4+ T cells was significantly higher in the IA, IC and GZ groups than in the IT group. The frequency of IFN-γ+ CD8+ T cells was significantly higher in the IA, IC, GZ and HC group than in the IT group. Moreover, the frequency of TNF-α+ CD8+ T cells was higher in the GZ and IA groups than in the IT group while frequency of TNF-α+ CD8+ T cells by HC group was higher than that in the GZ, IC, IA and IT groups. Both the frequencies of TNF-α+ and IL-2+ CD4+ T cells were significantly higher in IA patients than in IT patients (Figure 1A).
We also measured the frequencies of cytokines produced by NK and NKT cells in the current CHB cohort. As expected, statistically significant differences were observed in the frequencies of NK and NKT cells secreting IFN-γ, with progressive decreasing levels from IA, IC, GZ, and IT patients. The differences of the frequencies of IFN-γ produced by NKT cells, IFN-γ and TNF-α produced by NKT and NK cells, respectively, between the patients in IA and GZ phases were not statistically significant. However, the frequencies of IFN-γ produced by NKT and NK cells and TNF-α produced by NK cells of patients in the IA phase were all higher than those of patients in the IT phase (P = 0.004 and 0.0008 for IFN-γ+ NKT and NK cells in IA vs IT; P = 0.004 for TNF-α+ NK cells in IA vs IT, respectively, Figure 1B).
Taken together, these results indicate that a certain number of CHB patients beyond the current treatment guidelines, particularly, patients in the GZ phase, still produce antiviral cytokines.
Distribution of distinct cytokine profiles in CHB patients with different disease phases
Because clinical-virological features from patients with CHB were associated with TNF-α+, IFN-γ+ and IL-2+ T cells, and TNF-α+ and IFN-γ+ NK cells, we then assessed whether their combined evaluation could be used to identify the maturation of an efficient antiviral response to therapy in individual treatment-naïve CHB patients. We first investigated the correlation among the current 3 pairs of T-cell subset cytokines and 2 pairs of NK and NKT cell cytokines. The overall correlation among these 10 cytokines is shown in Figure 2A. After correlation analysis, the expression levels of 6 cytokines (CD4_IFN-γ+, CD4_IL-2+, CD8_TNF-α+, CD8_IL-2+, NK_IFN-γ+, NKT_TNF-α+) were selected to construct an IA-similar cytokine profiles. The assumption was that acquisition of IA-similar cytokine profiles could reflect a vigorous response to antiviral therapy. A threshold was thus established as shown by the mean value found in IA patients plus one standard deviation for the above selected parameters, and calculation of their expressions in individual patients was conducted to compare with each matched threshold. Individual cytokine distribution profiles were distinguished according to the altered number of applicable parameters beyond the threshold for all patients. A profile with all the applicable parameters altered was assumed to reflect an active immune response to therapy whereas a profile with no applicable parameters altered was predicted to be associated with an awakening response to therapy. IA and IC with an active immune response to HBV showed a prevalent expression of more inflammatory patterns with 6 and 4 altered applicable parameters respectively. In contrast, IT patients showed an immune depletive pattern with only 2 altered application parameters. GZ patients instead showed an intermediate behavior with 5 altered applicable parameters, as a likely result of the transition from an immune depletive to an inflammatory pattern of typical IA patients (Figure 2B). Based on Spearman’s rank correlation analysis, CD4_TNF-α+, CD8_IL-2+ and NK_IFN-γ+ were selected as the representative cytokines in the current CHB cohort for principal component analysis, which further confirmed that most IT patients significantly differed from IA and IC patients (red, blue and black circles, respectively), both of whom clustered homogeneously and an intermediate distribution was observed for GZ patients, who were widely scattered (green circles, Figure 2C).
Association among T cell-secreted cytokines and correlation with clinical-virological characteristics
Linear regression analysis was used to examine the association between T cell-produced cytokines and clinical-virological parameters. Univariate analysis revealed that a positive HBeAg result, and elevated levels of ALT and HBV DNA were associated with increased levels of CD4+ T cell-secreted TNF-α. Older age and higher ALT levels were associated with more proportion of IFN-γ+ CD4+ T cells, while IL-2+ CD4+T cells were linked to the increased HBV genotypes. After adjusting for other confounding factors, multivariate analysis revealed that both higher ALT levels and older age were significantly associated with increased IFN-γ+ and TNF-α+ CD4+ T cells (Table 2)
Univariate analysis of the relationship between CD8+ T cell-derived cytokines and clinical-virological factors showed that age and ALT were associated with both TNF-α+ and IFN-γ+ CD8+ T cells, respectively. Multivariate analysis indicated that older age and high ALT were still associated with increased IFN-γ content, while only ALT was significantly related to increased TNF-α+ CD8+ T cell proportions. There was no statistically significant association between IL-2+ CD8+ T cells and viral parameters (Table 2).
Therefore, either IFN-γ+ CD4+ or CD8+ T cells were significantly associated with older age and higher ALT, while TNF-α+ from these T cell subsets was associated with ALT.
Association among NK and NKT cell-secreted cytokines and correlation with clinical-virological characteristics
Similarly, the linear regression analysis was used to examine the association between clinical-virological factors and NK or NKT cell-expressed cytokines in the current CHB cohort. Univariate analysis of the NK cell cytokine profiles showed that elevated frequencies of IFN-γ+ and TNF-α+ cells were correlated with higher ALT levels, while only TNF-α+ cells were also associated with HBV DNA. Multivariate analysis also showed similar results regarding the association between cytokines and ALT (Table 3). We also detected cytokines produced by a subset of T cells that express NK cell markers, NKT cells. The HBeAg, ALT, and HBV genotypes were found to be associated with NKT cell-secreted TNF-α via univariate analysis (Table 4).
In summary, multivariate analysis of 10 clinical-virological parameters and 10 cytokines implied that ALT was significantly associated with 8 cytokines from T cells and NK cells. Age and HBV DNA were associated with 2 cytokines, while sex was correlated with only one cytokine frequency (Supplementary Figure 2).