HBV-specific T cell reactivity in HBV-infected patients with different clinical profiles

A total of 294 patients with HBV infection and distinct clinical profiles were detected for the numbers of reactive HBV-specific T cells (SFUs) in PBMCs using the in-house ELISpot assay. As shown in Fig. 1a, acute resolved patients (median 83 SFUs) displayed the highest numbers of reactive HBV-specific T cells while CHB patients (median 46 SFUs) showed the lowest ones among the R, CHB, LC and HCC groups, and an increasing trend of SFU numbers was found from CHB to LC and HCC groups. Meanwhile, the trends of HBsAg-, HBpol-, HBx- or HBeAg-specific T cells across the four disease stages were similar to that of total HBV-specific T cells (Fig. 1b). Moreover, multivariate linear regression analysis was performed for the 294 HBV-infected patients. The SFUs (assigned: continuous variable) were used as the dependent variable, while age, disease stage and sero-virological parameters (all as classification variables) were used as independent variables. The regression equation was tested as F = 4.973, p = 0.0009, indicating that the regression model was qualified. The collinearity was evaluated by variance inflation factor (VIF) (ALT: 1.10; DNA: 1.51; HBsAg: 1.42; stages: 1.03). Finally, the number of reactive HBV-specific T cells in PBMCs was significantly correlated with ALT level, HBsAg level, and disease stage (R, CHB, LC and HCC) (Table 1).

Fig. 1

HBV-specific T cell reactivity in 294 HBV-infected patients at different disease stages. Reactive HBV-specific T cells in PBMCs were detected using ex vivo IFN-γ ELISpot assay and 103 validated T-cell epitope peptides. A Total HBV-specific T cells (SFUs) in HBV-infected patients at different disease stages (R, n = 13; CHB, n = 203; LC, n = 52; HCC, n = 26). B Deconvolution of HBV-specific T cells from total antigens into the indicated HBV protein (HBsAg, HBpol, HBx, HBeAg) in HBV-infected patients. C Total HBV-specific T cells (SFUs) in CHB patients at different clinical phases (IA, n = 23; IT, n = 24; IC, n = 44). D Deconvolution of HBV-specific T cells from total antigens into the indicated HBV protein (HBsAg, HBpol, HBx, HBeAg) in CHB patients. Medians (interquartile range) were presented and statistical analyses were performed using Kruskal–Wallis test (K–W) across multiple groups and Mann–Whitney test (M–W) between two groups

Furthermore, the numbers of reactive HBV-specific T cells in PBMCs were compared across the clinical phases of CHB patients. In total, 114 of 203 CHB patients can be grouped into IA, IT or IC phases. A significantly increasing trend of SFU numbers was observed from IA group (median 37 SFUs), IT group (median 53 SFUs) to IC group (median 56 SFUs) with a p-value of 0.022 (IA vs. IT) and 0.0093 (IA vs. IC) (Fig. 1c). The trends of HBsAg-, HBx- or HBeAg-specific T cells across the three CHB subgroups were similar to that of total HBV-specific T cells, but HBpol-specific T cells displayed no significant differences across the CHB subgroups (Fig. 1d). In each CHB phase, the numbers of reactive HBV-specific T cells induced by different HBV proteins showed no significant difference (Additional file 1: Fig. S1). The spot plots reactive to each peptide pool in the in-house ELISpot assay were presented for five representative CHB subjects (Additional file 1: Fig. S2).

HBV-specific T cell reactivity in CHB patients with different sero-virological profiles

Stratified analyses manifested that the numbers of reactive HBV-specific T cells in PBMCs of CHB patients presented a significantly declined trend when the serum HBV DNA load, HBsAg, HBeAg or ALT level gradually increased (Fig. 2a). Considering different treatments may bring impact on the HBV-specific T cell reactivity, the stratified analyses were repeated in a cohort of CHB patients undergoing NUCs monotherapy (n = 167). The results (Additional file 1: Fig. S3) were in accordance with those from 203 CHB patients who were undergoing different therapies (NUCs alone or combined with IFN-α) (Fig. 2a). Furthermore, the numbers of HBsAg-, HBpol-, HBx-, or HBeAg-specific T cells also showed a decreasing trend similar to that of total HBV-specific T cells when HBV DNA, HBsAg, HBeAg and ALT levels gradually increased (Additional file 1: Fig. S4).

Fig. 2

Association of HBV-specific T cell reactivity with sero-virological parameters in CHB patients. A Stratified analyses of HBV-specific T cells (SFUs) in CHB patients grouped by HBV DNA load (< 3.0, n = 70; 3.0–5.0, n = 27; > 5.0, n = 23), HBsAg level (< 1000, n = 82; 1000–10000, n = 64; > 10,000, n = 34), HBeAg level (< 1, n = 53; 1–100, n = 45; 100–1000, n = 13; > 1000, n = 15) and ALT level (< 40, n = 146; > 40, n = 56). B Stratified analyses of sero-virological parameters in CHB patients grouped by HBV-specific T cell reactivity (0–24 SFUs for 25% of the cohort; 25–90 SFUs for 50% of the cohort; 91–622 SFUs for 25% of the cohort). For HBV DNA load, HBsAg level, HBeAg level, and ALT level analyses, 0–24 group, n = 42, 27, 41, 50, respectively; 25–90 group, n = 93, 59, 77,101, respectively; 91–622 group, n = 45, 24, 52, 51, respectively. C Spearman correlation tests between HBV-specific T cells (SFUs) and HBV DNA, HBsAg, HBeAg or ALT levels

To further assess the association of HBV-specific T cell reactivity with sero-virological parameters, CHB patients (n = 203) were divided into three subgroups according to SFU levels by using interquartile method: 25% of the cohort who presented low SFU level (0–24 SFUs/2 × 105 PBMCs), 50% of the cohort who presented intermediate SFU level (25–90 SFUs/2 × 105 PBMCs), and 25% of the cohort who presented high SFU level (91–622 SFUs/2 × 105 PBMCs). As shown in Table 3 and Fig. 2b, serum HBV-DNA, HBsAg, HBeAg and ALT levels were the highest in the subgroups with low SFU level, and the lowest in the subgroup with high SFU level.

Table 3 Stratification analyses of HBV-specific T cell reactivity for 203 CHB patientsCorrelation between HBV-specific T cell reactivity and sero-virological parameters in CHB patients

Multivariate linear regression analysis was performed for the 203 CHB patients (F = 3.384, p = 0.021). VIF was applied for multicollinearity (DNA load: 1.74; HBsAg: 1.43; ALT: 1.28) and showed a weaker degree of multicollinearity in the model. The p values of DNA, ALT, and HBsAg items were 0.168, 0.346 and 0.007, respectively (Table 3). That means the numbers of reactive HBV-specific T cells in PBMCs of CHB patients only significantly correlated with HBsAg level. Moreover, age, HBeAg, different phases of CHB (IT, IA and IC) were also used as independent variables in different regression models, but no correlation with SFU levels was found.

Furthermore, Spearman correlation tests between the numbers of reactive HBV-specific T cells in PBMCs and each sero-virological parameter were conducted for the 203 CHB patients. The SFU numbers were weakly and negatively correlated with serum HBV DNA (r = − 0.21), HBsAg (r = − 0.21) and HBeAg (r = − 0.27) levels, but not ALT level (r = − 0.079) (Fig. 2c). Meanwhile, HBsAg-, HBpol-, HBx-, or HBeAg-specific T cells also negatively correlated with serum HBV DNA, HBsAg and HBeAg levels with a very low coefficient (r = − 0.12 to − 0.26), and not correlated with ALT levels (r = − 0.027 to − 0.12) (Fig. S5). In addition, the correlations tests were repeated for the IA-phase, IT-phase and IC-phase patients, respectively, and no significant correlations between the SFU numbers and each sero-virological parameter was found in each subgroup (Additional file 1: Fig. S6).

Association of HBV-specific T cell reactivity with anti-virus therapy

HBV-specific T cells were further investigated and compared between the NUCs treatment group and NUCs/IFN-α combination group for CHB patients. The numbers of reactive HBV-specific T cells in NUCs/IFN-α group (median 103 SFUs) were obviously higher than those in the untreated (median 53 SFUs) or NUCs monotherapy (median 45 SFUs) groups (Fig. 3a). HBsAg-, HBpol-, HBx- or HBeAg-specific T cells also displayed the trends similar to total HBV-specific T cells across groups (Fig. 3b). Obviously, NUCs/IFN-α combination led to much more reactive HBV-specific T cells than NUCs monotherapy in CHB patients. In different treatment duration subgroups, the NUCs treatment more than 4 years presented more reactive HBV-specific T cells than the NUCs treatment less than 1 year, especially TMF treatment (Fig. 3c), but different NUCs in the same duration of treatment did not bring different reactivity of HBV-specific T cells (Fig. 3d).

Fig. 3

Association of HBV-specific T cell reactivity with anti-virus therapy in CHB patients. A HBV-specific T cells (SFUs) in CHB patients with different treatments. B Specific T cells (SFUs) reactive to each HBV protein (HBsAg, HBpol, HBx, HBeAg) in different treatment groups. Untreated group, n = 15; NUCs monotherapy, n = 167; NUCs/IFN combination therapy, n = 21. C HBV-specific T cells (SFUs) after NUCs treatment in different durations of treatment. D HBV-specific T cells (SFUs) after different NUCs treatment in the same treatment duration. Medians (interquartile range) were presented and statistical analyses were performed using Kruskal–Wallis test (K–W) across multiple groups and Mann–Whitney test (M–W) between two groups

Dynamic reactivity of HBV-specific T cells in CHB patients with different sero-virological courses

HBV-specific T cells were detected three times at an interval of 3–5 months for 33 CHB patients who undergoing NUCs monotherapy or NUCs/IFN-α combination therapy, and sero-virological data were synchronously collected at the same time point. The treatment regimens and duration before the first test were presented in Additional file 1: Table S3. The numbers of reactive HBV-specific T cells were obviously and gradually increasing during the three tests (Fig. 4a). HBsAg-, HBpol-, HBx-, and HBeAg-specific T cell responses were consistent with the total HBV-specific T cell responses (Fig. 4b). Meanwhile, serum HBV DNA loads, HBsAg, HBeAg and ALT/AST levels displayed a continuously declined tendency during the observation period (Fig. 4c).

Fig. 4

Dynamic changes of HBV-specific T cells and sero-virological parameters in CHB patients. 33 CHB patients undergoing routine treatment were followed by HBV-specific T cell detection and sero-virological parameters collections for three times at an interval of 3–5 months. A, B Dynamic changes of total HBV-specific T cells and the specific T cells reactive to each HBV protein in 33 CHB patients. C Dynamic changes of HBV DNA (n = 18), HBsAg (n = 27), HBeAg (n = 19), ALT (n = 32), and AST (n = 32) levels. Then, the dynamic changes of HBV-specific T cells in CHB patients with different fluctuation courses of D HBV DNA load (decrease, n = 7; no alternation, n = 3; increase, n = 2), E HBsAg level (decrease, n = 18; no alternation, n = 10), F HBeAg level (seroconversion, n = 4; retained, n = 13), and G ALT level (normal, n = 14; decrease, n = 12; increase, n = 7) were presented. The patients who achieved DNA fluctuations (increase or decrease) > 30% were defined as the DNA-increase or DNA-decrease group, and the other patients were defined as DNA-no alternations group. HBsAg-decrease was defined as an amplitude decrement of more than 30%. CHB patients who experienced a positive HBeAg serology (HBeAg COI > 1) at first and seroconverted (HBeAg COI < 1) later were defined as the HBeAg-seroconversion group. ALT-decrease was defined as a decline to the normal range (< 40 IU/L) or decreased more than 30%, while ALT that rose more than 30% or beyond 40 IU/L was defined as ALT-increase. The paired, two-tailed Student’s t tests between two groups and Kruskal–Wallis test (K–W) across more than two groups were performed

Subsequently, the 33 CHB patients were divided into the different subgroups with an increasing, no-alteration or declining sero-virological course, and stratified analysis were further performed. As displayed, the numbers of reactive HBV-specific T cells continuously increased from baseline (first test) to third test in the DNA load decreasing subgroup, but no significant increase in the no-alteration subgroup or increasing subgroup (Fig. 4d). Similarly, the SFU numbers also continuously increased in the HBsAg level decreasing subgroup, while no significant increase in the no-alteration subgroup (Fig. 4e). Notably, in the HBeAg-retained subgroup, SFU numbers exhibited a trend of initially decreasing and subsequently increasing, while no meaningful pattern was discernible in the HBeAg seroconversion group (Fig. 4f). Meanwhile, the SFU numbers exhibited a significant increase in ALT normal or decreased subgroups (Fig. 4g).

To confirm the dynamic trends in the absence of IFN-α treatment, the longitudinal data were collected from 28 CHB patients who only undergoing NUCs monotherapy, then the dynamic tendency and conclusion were obtained (Additional file 1: Fig. S7) and as same as that in Fig. 4.

Dynamic tendencies of sero-virological parameters in CHB patients with different HBV-specific T cell courses

Here, the 33 CHB patients was categorized into several subgroups according to the different courses of HBV-specific T cell reactivity during the three tests. The fluctuation (increase or decrease) of about 50% of SFU numbers was defined as an obvious change between two tests, thus 5 subgroups were obtained: ascending, ascending/descending, stationary, descending, and descending/ascending. The dynamic spot plots of ELISpot assay for representative ascending patients and ascending/descending patient were presented in Additional file 1: Fig. S8. As expected, in the SFU numbers ascending subgroup (n = 22), a continuous decrease of HBV-DNA loads, HBsAg levels, HBeAg levels, or ALT levels was found in most cases (Fig. 5a). In the SFU numbers ascending/descending subgroup (n = 7), HBsAg levels and ALT levels retained stable or slight decline in most cases (Fig. 5b). The dynamic courses in other subgroups cannot be concluded due to small cohorts.

Fig. 5

Dynamic changes of sero-virological parameters during different fluctuation courses of HBV-specific T cells. 33 CHB patients were followed by HBV-specific T cell detection and sero-virological parameters collections for three times at an interval of 3–5 months. According to the dynamic courses of HBV-specific T cells (SFUs), patients were categorized as ascending (A), ascending/descending (B), stationary, descending, or descending/ascending groups, then HBV DNA load, HBsAg, HBeAg, HBx, HBpol and ALT levels were longitudinally analyzed. The SFU numbers of reactive HBV-specific T cells which increased or decreased more than 50% than last test were defined as ascending or descending during the follow-up period

Predictive power of cross-sectional and longitudinal numbers of reactive HBV-specific T cells for liver function progression in CHB patients

Firstly, the predictive power of cross-sectional numbers of reactive HBV-specific T cells from CHB patients were evaluated for the hepatitis progression at test time and 6 months later. As shown in Fig. 6, HBV-specific T cells presented 0.600 AUC, 21.5 (SFUs) cut-off value, 30.0% specificity and 86.0% sensitivity for the liver function at test time, and 0.604 AUC, 31 (SFUs) cut-off value, 68.6% specificity, and 52.6% sensitivity for the liver function 6 months later. Comparably, HBV DNA load presented a higher predictive value. Further, ROC curves for the combined two factors conferred a further increase in predictive accuracy (0.748 AUC, 57.5% specificity, 90.0% sensitivity) at 6 months after T-cell test. These data suggest that the cross-sectional number of reactive HBV-specific T cells, as a parameter of host adaptive immunity, is also a valued predictor for liver function progression in CHB patients undergoing routine treatment, especially when combined with the viral DNA load. Then, the data of HBV-specific T cells (SFUs/4 × 105 PBMCs) and sero-virological parameters were collected from another 176 CHB patients (beyond the CHB patients in Fig. 6), and were used to confirm the predictive power by ROC analysis. As shown in Additional file 1: Fig. S9, the AUC values of SFU numbers were much higher than that in Fig. 6 (0.750 vs. 0.600 at test time; 0.662 vs. 0.604 at 6 months later). Meanwhile, viral DNA load did not present a higher predictive value than HBV-specific T cells and the combined two factors only presented a slightly increased predictive accuracy (0.722 vs. 0.662 AUC at 6 months after T cell test).

Fig. 6

Predictive power of cross-sectional reactivity of HBV-specific T cells for hepatitis progression in CHB patients. CHB patients were divided into normal (ALT < 40 IU/L) group and abnormal (ALT > 40 IU/L) group of liver function at the test time of HBV-specific T cells or 6 months later after the test. ROC curve analyses of DNA load (IU/mL), HBV-specific T cells (SFUs/2 × 105 PBMCs), and a combination were performed to predict hepatitis progression at the test time of HBV-specific T cells (A) and 6 months later after the test (B), using R package pROC, and summarized in table. ROC, receiver operating characteristic; AUC, area under the curve; PPV, positive predictive value; NPV, negative predictive value. The p1 values represent the significance of model. The p2 values represent the significance of difference between the AUC of combined markers (DNA load + Specific T cell) and single predictor

Additionally, ROC curve analyses were also performed on the longitudinal data of reactive HBV-specific T cells from the 33 CHB patients who were tested three times at an interval of 3–5 months. When compared to when either was analyzed independently, the combination of first, second and third test led to the highest AUC value to predict liver function progression at 6 months or 12 months after the last test. The AUC was 0.735 and 0.747, specificity was 80.0% and 81.5%, and sensitivity was 62.5% and 66.7%, respectively (Fig. 7). These data imply that the longitudinal monitoring of HBV-specific T cell reactivity possesses much higher predictive power for hepatitis progression than the cross-sectional detection.

Fig. 7

Predictive power of longitudinal reactivity of HBV-specific T cells for liver hepatitis progression in CHB patients. 33 CHB patients undergoing NUCs or NUCs/INF-α treatment were divided into normal (ALT < 40 IU/L) group and abnormal (ALT > 40 IU/L) group of liver function at 6 months (normal/abnormal: 8/25) or 12 months (normal/abnormal: 6/27) after the last test of HBV-specific T cells. ROC curve analyses of single test and combined tests of HBV-specific T cells (SFUs/2 × 105 PBMCs) were performed to predict hepatitis progression at 6 months (A) and 12 months after the last test (B) of HBV-specific T cells, using R package pROC, and summarized in table. The p1 values represent the significance of model. The p2 values represent the significance of difference between the AUC of combined markers (First + Second + Third) and other predictors