HBV‐specific CD8 T cells present higher TNF‐α expression but lower cytotoxicity in hepatocellular carcinoma

Summary

Tumor necrosis factor (TNF)‐α is largely regarded as a proinflammatory cytokine, but several recent researches have demonstrated that TNF‐α could possess immunoregulatory roles with potential to suppress anti‐tumor immunity. Chronic hepatitis B virus (HBV) infection is a major risk factor of hepatocellular carcinoma (HCC), and HBV‐specific CD8 T cells could exert anti‐tumor roles in HCC patients. Here, we found that HBV‐specific CD8 T cells, both in the peripheral blood and in the tumor microenvironment, were more enriched with TNF‐α‐expressing cells than interferon (IFN)‐γ‐expressing cells. Compared to IFN‐γ‐expressing HBV‐specific CD8 T cells, TNF‐α‐expressing HBV‐specific CD8 T cells presented lower expression of inhibitory checkpoint molecules, including programmed cell death (PD)‐1, T cell immunoglobulin mucin‐3 (TIM‐3) and cytotoxic T lymphocyte antigen (CTLA)‐4. HBV‐specific CD8 T cells could mediate the lysis of autologous primary tumor cells, and the inhibition of TNF‐α could further elevate their cytotoxic capacity. Subsequently, we demonstrated that TNF‐α inhibition in HBV‐specific CD8 T cells could significantly increase granzyme B (GZMB) and perforin 1 (PRF1) expression while having no effect towards granzyme A (GZMA) expression. The addition of exogenous TNF‐α at low levels had no consistent effect on the expression of GZMA, GZMB and PRF1, but at higher levels, exogenous TNF‐α significantly reduced GZMA, GZMB and PRF1 expression. Overall, these results suggested that TNF‐α‐expressing cells probably presented a deleterious role in HCC but were enriched in HBV‐specific CD8 T cells.

TNFα‐expressing HBV‐specific CD8 T cells presented lower expression of inhibitory checkpoint molecules than IFNγ‐expressing HBV‐specific CD8 T cells. HBV‐specific CD8 T cells could mediate the lysis of autologous primary tumor cells and the inhibition of TNFα could further elevate their cytotoxic capacity. TNFα inhibition in HBV‐specific CD8 T cells could significantly increase granzyme B and perforin 1 expression while having no effect toward granzyme A expression.

Introduction

Hepatitis B virus (HBV) is one of the major risk factors for hepatocellular carcinoma (HCC) [1]. In adults, HBV infection can produce an acute immune response that is able to clear the virus and provide lifelong immunity. This response is characterized by the generation of HBV‐specific CD8 T cells that are able to eliminate HBV‐infected hepatocytes and secrete anti‐viral cytokines interferon (IFN)‐γ and tumor necrosis factor (TNF)‐α [2, 3]. The T cell responses during acute, self‐limiting infections tend to be broad‐spectrum and target multiple viral proteins [4]. Also, antibodies against HBV antigens are produced to neutralize HBV in circulation [5]. However, these responses are not observed in individuals who were infected by HBV at birth or during early childhood. Instead, a low‐level immune response is observed, in which the HBV‐specific CD8 T cells are characterized by lower frequencies of HBV‐core and polymerase‐targeting cells [4], higher expression of suppressive checkpoint molecules [6, 7] and reduced levels of anti‐viral cytokine production [8]. These types of response fail to clear the virus but are thought to induce progressive liver damage, resulting in a chronic and persistent infection that results in hepatitis, liver fibrosis and cirrhosis and culminates in HCC later in life [9, 10].

Immunotherapy has experienced major advances in recent years and is rapidly becoming a novel treatment modality in solid and hematopoietic tumors [11, 12]. In HCC, antagonistic monoclonal antibodies targeting programmed cell death 1/programmed cell death ligand 1 (PD‐1/PD‐L1) and cytotoxic T lymphocyte antigen 4 (CTLA‐4) are either being investigated in Phase III trials or have received approval from the United States Food and Drug Administration [13, 14]. The main strategy of these therapeutic options is to revive T cell‐mediated anti‐tumor responses by inhibiting the signal transduction of suppressive checkpoint molecules [15]. The overall result has been encouraging, but many patients do not respond to checkpoint inhibitors. Moreover, these inhibitors are, in general, ineffective at inducing tumor regression [15]. More investigations are required to examine whether T cells in HCC suffer from additional suppressive mechanisms that limit their effector functions.

TNF‐α is a pleiotropic cytokine with complicated and contradictory roles in anti‐cancer immunity. It has been shown that in the complete absence of TNF‐α signaling, tumor immunosurveillance is severely impaired [16]. However, there are also studies that suggest the opposite, that TNF‐α may exert immunoregulatory functions by promoting the recruitment and differentiation of regulatory T cells (T_reg) cells, regulatory B cells (B_reg) cells, and/or myeloid‐derived suppressor cells [17, 18, 19]. In a murine melanoma model, TNF‐α promotes the growth of major histocompatibility complex (MHC)‐I^high, but not MHC‐I^low, tumor cells, while the administration of soluble human TNF‐α receptor 2 (TNF‐R2) is able to neutralize murine TNF‐α and lower the growth of MHC‐I^high tumor cells [20]. Additionally, TNF‐α may trigger activation‐induced cell death in T cells and impair CD8 T cell homeostasis in a manner that is dependent upon TNF‐R1 signaling [20, 21]. The role of TNF‐α in HCC T cells has not been fully elucidated.

In this study, we found that HBV‐specific CD8 T cells, compared to total CD8 T cells, were enriched with a TNF‐α‐expressing subset, both in the peripheral blood and in the tumor. Subsequently, we investigated the functional characteristics of this subset and discovered that TNF‐α at high levels may inhibit the cytotoxic response by CD8 T cells.

Methods

Participants

HCC patients were recruited from recently diagnosed primary HCC patients with history of chronic HBV infection, who presented resectable tumors. Healthy individuals were recruited from age‐ and gender‐matched healthy subjects who were HBV‐uninfected and did not have HCC. The characteristics of the study subjects are summarized in Table 1. Candidate participants with the following conditions, such as other forms of malignancy, chronic virus infection, autoimmune diseases, alcoholic liver diseases, renal diseases, diabetes, inflammatory bowel diseases and hereditary predisposition of cancer, were excluded from participation. All subjects provided written informed consent. Ethical approval was obtained from the ethics committee from Linyi People’s Hospital.

Table 1.

Characteristics of the study subjects

	Healthy subjects	HCC patients	P
n	19	19
Female, n (%)	7 (37)	8 (42)	> 0·05
Age in years, mean ± s.d.	57·2 ± 6·8	55·9 ± 7·2	> 0·05
AFP in ng/ml, mean ± s.d.	5·5 ± 1·5	81 ± 30·4	< 0·001
GGT in U/l, mean ± s.d.	20·5 ± 7·8	65 ± 20·3	< 0·001
Cirrhosis, Y (%)	0 (0)	15 (79)
Tumor size in cm, n (%)
≤ 5	0 (0)	15 (79)
> 5	0 (0)	4 (21)

HCC = hepatocellular carcinoma; AFP = alpha‐fetoprotein; GGT = gamma‐glutamyltranspeptidase; s.d. = standard deviation.

Sample collection

Peripheral blood mononuclear cells (PBMCs) were collected from untreated patient blood via Ficoll‐Paque PLUS (GE Healthcare, Chicago, IL, USA) via standard procedure. Tumor‐infiltrating lymphocytes (TILs) were harvested from enzymatically digested tumors. Briefly, the tumor mass from each individual was minced into small pieces and incubated in ×1 collagenase/hyaluronidase (Stemcell, Cambridge, MA, USA) in a 37˚C waterbath for 2 h. The digestion product was filtered using a sterile strainer with a 70‐μm mesh (Corning, New York, NY, USA) to obtain single cells.

CD8 T cell stimulation

Pan‐T cell activation was performed using Dynabeads human T activator CD3/CD28 at one bead per cell. HBV peptide stimulation was performed using 15‐mer HBV peptides with 11 amino acids overlapping (JPT Peptide Technologies, Berlin, Germany) at an overall concentration of 2 μg/ml. Cytokine trapping was performed by adding 5 μg/ml GolgiPlug and 5 μg/ml GolgiStop (BD Biosciences, San Jose, CA, USA). After 6‐h stimulation, cells were harvested and examined using flow cytometry.

Flow cytometry

Fluorophore‐labeled anti‐human antibodies against surface antigens, including CD3 [phycoerythrin/cyanin5 (PE/Cy5)‐conjugated HIT3a], CD8 (AF488‐conjugated SK1), CD45RO [peridinin chlorophyll (PerCP)‐conjugated UCHL1], PD‐1 (PE/Cy7‐conjugated A17188B), T cell immunoglobulin mucin‐3 (TIM‐3) [allophycocyanin (APC)/Cy7‐conjugated F38‐2E2] and CTLA‐4 (PE/Cy7‐conjugated BNI3), and against intracellular cytokines, including TNF‐α [APC‐conjugated TNF‐a monoclonal antibody (MAb11)] and IFN‐γ (APC‐conjugated 4S.B3), were all obtained from BioLegend (San Diego, CA, USA). All cells were first labeled with Fixable Aqua Dead Cell Stain (Invitrogen, Carlsbad, CA, USA), washed and labeled with surface antibodies. Subsequently, cells were fixed and permeabilized with CytoFix/CytoPerm buffer (BD Biosciences) and stained with intracellular antibodies. After thorough washing, cells were acquired in an LSR cytometer system (BD Biosciences) and the results were analyzed using the FlowJo system (BD Biosciences). A minimum of 5 × 10⁵ cells per sample were acquired.

Specific lysis

Primary tumor cells were seeded at 5 × 10⁴ cells per well in a 96‐well flat‐bottomed plate (Corning). CD8 T cells were isolated from PBMCs using a CD8⁺ T Cell Isolation Kit, human (Miltenyi Biotec, Cambridge, MA, USA) and then stimulated using HBV peptides for 6 h and added to seeded tumor cells at concentrations specified in the experiment. When indicated, TNF‐α‐neutralizing antibodies or isotype controls (R&D Systems, Minneapolis, MN, USA) were added at 10 μg/ml. After 4‐h incubation, the level of lactate dehydrogenase (LDH) release was measured using the CyQUANT LDH cytotoxicity assay (Invitrogen), according to the manufacturer’s instructions. Triplicate experiments were performed for each condition.

mRNA analysis

CD8 T cells (2 × 10⁴) were stimulated using HBV peptides for 6 h and the total RNA was harvested using the RNeasy mini kit (Qiagen, Valencia, CA, USA). The following reagents were added when indicated, including TNF‐α‐neutralizing antibodies or isotype controls at 10 μg/ml or recombinant human TNF‐α protein carrier‐free (R&D Systems) at concentrations specified in the experiment. cDNA was obtained from mRNA transcripts using SuperScript III reverse transcriptase (Invitrogen). The expression of GZMA, GZMB and PRF1 was then measured using TaqMan gene expression assays with IDs Hs00989184_m1, Hs00188051_m1 and Hs00169473_m1, respectively. The assay was run in the ABI PRISM 7000 sequence detection system (Thermo Fisher, Fremont, CA, USA) using the following program: 45 cycles of denaturation at 95˚C for 20 s, annealing at 55˚C for 20 s and extension at 72˚C for 45 s, followed by cooling for 1 min at 40˚C. All assays were performed in three replicate experiments.

Statistical analyses

Comparisons between two groups were performed using Student’s t‐test with Welch’s correction. Comparisons between more groups were performed using two‐way analysis of variance (ANOVA) followed by Tukey’s test. For matched experiments, comparisons were performed using repeated‐measures (RM) one‐way ANOVA followed by Holm–Sidak’s test. All tests were two‐tailed. P < 0·05 was considered significant.

Results

HBV‐specific CD8 T cell response was biased toward TNF‐α but not IFN‐γ

To analyze the characteristics of HBV‐specific CD8 T cell responses in HBV‐related HCC, circulating CD8 T cells were activated using the HBV peptide pool (HBV‐specific) or pan‐activated using anti‐CD3 and anti‐CD28 monoclonal antibodies (pan‐T). The frequency of CD8 T cells that expressed proinflammatory cytokines TNF‐α and IFN‐γ was detected using intracellular flow cytometry (Fig. 1a,b). The frequency of TNF‐α‐expressing cells and IFN‐γ‐expressing cells was measured in 19 HBV‐related HCC patients. The ratio of TNF‐α‐ to IFN‐γ‐expressing (TNF‐α/IFN‐γ) cells was compared (Fig. 1c). Following pan‐T cell activation, the ratio was 1·38 ± 0·45 (mean ± standard deviation), while following HBV‐specific activation, the ratio was significantly higher at 1·91 ± 0·75, indicating that compared to the overall T cell population, HBV‐specific CD8 T cells were slightly more enriched with TNF‐α‐expressing cells.

Fig. 1.

Tumor necrosis factor (TNF)‐α to interferon (IFN)‐γ ratio following pan‐T cell activation or hepatitis B virus (HBV)‐specific activation. (a) Representative CD8 T cell identification strategy. (b) Representative identification of TNF‐α‐expressing and IFN‐γ‐expressing cells in CD8 T cells, after pan‐T cell activation or HBV‐specific activation. (c) The ratio of TNF‐α‐expressing to IFN‐γ‐expressing cell frequency in 19 HBV– hepatocellular carcinoma (HCC) patients. Student’s t‐test with Welch’s correction; *P < 0·05.

Phenotyping of IFN‐γ‐ and TNF‐α‐expressing HBV‐specific CD8 T cells

Subsequently, we investigated the surface markers of HBV‐specific CD8 T cells that expressed IFN‐γ or TNF‐α (Fig. 2a). Both naive and memory CD8 T cells may express TNF‐α upon activation [22, 23]. In HBV‐specific TNF‐α‐expressing CD8 T cells, 86·9 ± 7·4 % were CD45RO⁺ memory cells, while in HBV‐specific IFN‐γ‐expressing CD8 T cells, 95·6 ± 3·9% were CD45RO⁺ memory cells (Fig. 2b); 64·7 ± 11·0 % HBV‐specific TNF‐α‐expressing CD8 T cells expressed PD‐1, while 79·5 ± 10·6 % HBV‐specific IFN‐γ‐expressing CD8 T cells expressed PD‐1 (Fig. 2c). The expression of TIM‐3 was found on 65·5 ± 10·9% of HBV‐specific TNF‐α‐expressing CD8 T cells and on 78·1 ± 8·5% HBV‐specific IFN‐γ‐expressing CD8 T cells (Fig. 2d). In addition, the expression of CTLA‐4 was found on 77·7 ± 9·8% HBV‐specific TNF‐α‐expressing CD8 T cells and on 86·7 ± 8·5% HBV‐specific IFN‐γ‐expressing CD8 T cells (Fig. 2e). Overall, these data indicated that the vast majority of HBV‐specific TNF‐α‐expressing CD8 T cells were memory cells and expressed high levels of inhibitory receptors, but compared to HBV‐specific IFN‐γ‐expressing CD8 T cells, HBV‐specific TNF‐α‐expressing CD8 T cells presented significantly lower levels of PD‐1, TIM‐3 and CTLA‐4.

Fig. 2.

Phenotyping of hepatitis B virus (HBV)‐specific tumor necrosis factor (TNF)‐α‐expressing CD8 T cells. (a) CD8 T cells were stimulated via pan‐T cell activation or HBV‐specific peptides. Expression of CD45RO, tumor necrosis factor receptor 1 (TNF‐R1) and TNF‐R2 was shown in one representative individual. (b) The frequency of CD45RO⁺ cells in TNF‐α‐expressing total CD8 T cells or hepatitis B virus (HBV)‐specific CD8 T cells. (c) The frequency of TNF‐R1⁺ cells in TNF‐α‐expressing total CD8 T cells or HBV‐specific CD8 T cells. (d) The frequency of TNF‐R2⁺ cells in TNF‐α‐expressing total CD8 T cells or HBV‐specific CD8 T cells. Student’s t‐test with Welch’s correction; **P < 0·01; ***P < 0·001.

TNF‐α expression hampered cytotoxicity by HBV‐specific CD8 T cells

It has been shown that TNF‐α may down‐regulate immune responses and promote tumorigenesis [18, 20]. To investigate whether TNF‐α over‐representation may affect the cytotoxicity of HBV‐specific CD8 T cells, we performed a cytotoxicity assay with or without TNF‐α inhibition. Primary tumor cells from HCC patients who had undergone surgical resection were used as target cells, while autologous CD8 T cells were activated using HBV peptides and used as effector cells. Unstimulated CD8 T cells presented little cytotoxic potential, while HBV peptide‐stimulated CD8 T cells could induce tumor cell death in a concentration‐dependent fashion (P < 0·001 compared to unstimulated CD8 T cells; Fig. 3). Interestingly, the addition of TNF‐α neutralization antibodies significantly increased the specific lysis of HBV‐stimulated CD8 T cells towards tumor cells (P < 0·01 compared to HBV‐stimulated CD8 T cells and P < 0·05 compared to HBV‐stimulated CD8 T cells + isotype control).

Fig. 3.

The cytotoxicity of CD8 T cells towards autologous primary cancer cells. Unstimulated CD8 T cells or hepatitis B virus (HBV)‐stimulated CD8 T cells were incubated with autologous tumor cells at ratios indicated in the x‐axis. Cytotoxicity was measured using lactate dehydrogenase (LDH) assay. When indicated, tumor necrosis factor (TNF)‐α neutralizing antibodies or isotype control antibodies were added at 5 μg/ml. Specific lysis was calculated using the formula specific lysis (%) = (CD8 T cell‐induced lysis – spontaneous lysis)/(maximum lysis – spontaneous lysis) × 100. Two‐way analysis of variance (ANOVA) followed by Tukey’s test.

TNF‐α could suppress GZM and perforin expression by HBV‐specific CD8 T cells

To investigate whether TNF‐α could directly lower CD8 T cell cytotoxicity, we examined the expression of GZMA, GZMB and PRF1 by HBV‐specific CD8 T cells with or without TNF‐α neutralization. GZMA expression by HBV‐specific CD8 T cells was not uniformly affected by TNF‐α inhibition (Fig. 4a). In contrast, GZMB and PRF1 expression levels by HBV‐specific CD8 T cells were significantly higher following TNF‐α inhibition (Fig. 4b,c).

Fig. 4.

The expression of cytotoxic molecules with or without tumor necrosis factor (TNF)‐α neutralization. CD8 T cells were incubated with TNF‐α‐neutralizing antibody or isotype control during hepatitis B virus (HBV)‐peptide stimulation. The CD8 T cells were then lyzed and the expression of (a) granzyme A (GZMA), (b) granzyme B (GZMB) and (c) perforin 1 (PRF1) was examined. RM one‐way analysis of variance (ANOVA) followed by Holm–Sidak’s test; n.s. = not significant; *P < 0·05; ***P < 0·001.

We also investigated the effect of exogenous TNF‐α on the expression of cytotoxic molecules, including GZMA, GZMB and PRF1, by HBV‐specific CD8 T cells. The expression of GZMA was not significantly effected by 0·1, 1 or 10 ng/ml of exogenous TN‐α but was significantly reduced in the presence of 100 ng/ml of exogenous TNF‐α (Fig. 5a). The expressions of GZMB and PRF1 were unaffected by 0·1 or 1 ng/ml of exogenous TNF‐α but were significantly reduced by 10 and 100 ng/ml of exogenous TNF‐α (Fig. 5b,c).

Fig. 5.

The expression of cytotoxic molecules with or without tumor necrosis factor (TNF)‐α neutralization. CD8 T cells were incubated with TNF‐α‐neutralizing antibody or isotype control during hepatitis B virus (HBV)‐peptide stimulation. The CD8 T cells were then lyzed and the expression of (a) granzyme A (GZMA), (b) granzyme B (GZMB) and (c) perforin 1 (PRF1) was examined. RM one‐way analysis of variance (ANOVA) followed by Holm–Sidak’s test; n.s. = not significant; *P < 0·05; ***P < 0·001.

TNF‐α bias was more pronounced in tumor‐infiltrating HBV‐specific CD8 T cells

To investigate the characteristics of HBV‐specific CD8 T cells in the tumor, we harvested tumor‐infiltrating CD8 T cells from HCC patients. The frequencies of TNF‐α‐ and IFN‐γ‐expressing CD8 T cells following pan‐T cell activation or HBV peptide activation were measured (Fig. 6a). Pan‐T cell activation was significantly more potent at mediating TNF‐α and IFN‐γ expression than HBV peptides (Fig. 6b,c). Following pan‐T cell activation the TNF‐α/IFN‐γ ratio was 1·33 ± 0·69, while following HBV‐specific activation the TNF‐α/IFN‐γ ratio was significantly higher at 2·63 ± 1·37, indicating that HBV‐specific tumor‐infiltrating CD8 T cells was markedly more enriched with TNF‐α‐expressing cells (Fig. 6d).

Fig. 6.

Tumor necrosis factor (TNF)‐α to interferon (IFN) ratio in tumor‐infiltrating CD8 T cells following pan‐T cell activation or HBV‐specific activation. (a) Representative identification of TNF‐α‐expressing and IFN‐γ‐expressing cells in tumor‐infiltrating CD8 T cells, after pan‐T cell activation or hepatitis B virus (HBV)‐specific activation. (b) The frequency of TNF‐α‐expressing CD8 T cells. (c) The frequency of IFN‐γ‐expressing CD8 T cells. (d) The ratio of TNF‐α‐expressing cell frequency to IFN‐γ‐expressing cell frequency in 19 HBV–hepatocellular carcinoma (HCC) patients. Student’s t‐test with Welch’s correction. ***P < 0·001.

Discussion

This study provides a description of HBV‐specific CD8 T cells with regard to TNF‐α expression and other associated characteristics. Compared to total CD8 T cells activated via the anti‐CD3/CD28 pathway the HBV‐specific CD8 T cells, activated via HBV peptides, presented a higher TNF‐α/IFN‐γ ratio, indicating a preferential enrichment in TNF‐α‐expressing cells. This phenomenon was observed both in circulating CD8 T cells and in tumor‐infiltrating CD8 T cells, thus warranting the rationale for further analyses.

We compared the expression of suppressive checkpoint molecules between IFN‐γ‐expressing HBV‐specific CD8 T cells and TNF‐α‐expressing HBV‐specific CD8 T cells. Interestingly, for all three molecules examined, including PD‐1, TIM‐3 and CTLA‐4, the expression in TNF‐α‐expressing HBV‐specific CD8 T cells was significantly higher than that in IFN‐γ‐expressing HBV‐specific CD8 T cells, potentially suggesting that these TNF‐α‐expressing HBV‐specific CD8 T cells were less susceptible to immune suppression mediated by those molecules, especially in HCC tumors, which are enriched with inhibitory ligands such as PD‐L1 [24]. Interestingly, it has been reported that TNF‐α and transforming growth factor (TGF)‐β regulate the expression of PD‐L1 in opposite directions, and while TGF‐β suppresses PD‐L1 expression in monocytes from patients with systemic lupus erythematosus (SLE), TNF‐α up‐regulates PD‐L1 expression in SLE patients [25]. Also, TNF‐α blockade significantly enhanced the effect of PD‐1 blockade in experimental melanoma [26]. Hence, TNF‐α‐expressing HBV‐specific CD8 T cells might contribute to PD‐L1 expression by HCC cells while limiting their own exposure to the PD‐L1/PD‐1 pathway with lower PD‐1 expression and contribute to tumor pathogenesis. This hypothesis should be investigated in future experiments.

Subsequently, using cytotoxicity assay, we found that HBV‐specific CD8 T cells could mediate lysis of autologous tumors, thus demonstrating cytotoxic capacity. However, the inhibition of TNF‐α could significantly elevate this cytotoxic capacity, which seemed to suggest that TNF‐α suppressed the cytotoxic response of CD8 T cells. Subsequently, we demonstrated that TNF‐α inhibition in HBV‐specific CD8 T cells could significantly increase GZMB and PRF1 expression while having no effect towards GZMA expression. The addition of exogenous TNF‐α at low levels had no consistent effect on the expression of GZMA, GZMB and PRF1, but at higher levels exogenous TNF‐α significantly reduced GZMA, GZMB and PRF1 expression. These results seemed to suggest that TNF‐α presented a deleterious role in CD8 T cell responses in HCC patients. One issue with these data is that the TNF‐α concentration needed to show that inhibitory effects were much higher than the physiological TNF‐α concentration in the serum of healthy subjects. However, it has been reported that TNF‐α could be significantly elevated in cancer patients and following radiation therapy [27, 28]. In addition, the local concentration of TNF‐α inside the tumor may be much higher than that in serum. Hence, the level of intratumoral TNF‐α in HCC patients should be investigated to determine whether or not TNF‐α has an effect on suppressing tumor‐infiltrating CD8 T cells. In addition, whether TNF‐α could act as an autocrine to increase its own production in tumor‐infiltrating CD8 T cells should be examined.

Overall, our data revealed an additional CD8 T cell‐related dysregulation in HCC. Several questions were raised by these data. First, the reason of TNF‐α enrichment among the HBV‐specific CD8 T cell population is unclear. Given that checkpoint molecules are up‐regulated following activation, TNF‐α‐expressing CD8 T cells may enhance their own survival by lowering activation status. The regulation of TNF‐α in chronic HBV infection and HCC should be examined. Secondly, the precise role of these TNF‐α‐expressing HBV‐specific CD8 T cells in the pathogenesis of HCC remains unknown. Further analyses regarding the frequency of TNF‐α‐expressing HBV‐specific CD8 T cells and the clinical characteristics of the HCC patients, including their response towards treatment, should be investigated. Thirdly, whether or not the inhibition of TNF‐α could directly support the antitumor cytotoxicity of CD8 T cells in HCC patients should be examined in in‐vivo systems.

Conflict of interest

None.