Compromised lymphocytes infiltrate hepatocellular carcinoma: The role of T-regulatory cells


  • See Editorial on Page 700.

  • Potential conflict of interest: Nothing to report.


Hepatocellular carcinoma (HCC) has a poor prognosis with limited therapeutic options. We propose that local immune responses in patients with HCC are held in check by tumor-infiltrating CD4+CD25+ T-regulatory lymphocytes (Treg cells), which suppress the activity and proliferation of effector CD4+ and CD8+ T cells. The phenotype and cell cycle status of tumor-infiltrating lymphocytes (TILs) in HCC were analyzed via immunohistochemistry of sections from patients undergoing surgery for HCC and via flow cytometry of peripheral blood mononuclear cells and TILs isolated from patients with HCC. Circulating and tumor-infiltrating T-cell function and activation status were assessed via proliferation and flow cytometry. More than 96% of TILs were quiescent as measured via Mcm-2 or Ki-67 expression, while less than 10% of CD8+ T cells expressed perforin or granzyme B. CD4+CD25+ Treg cells comprised 8.7% (1.4–13.8) of TILs and always exceeded the proportion in distant nontumor tissue (2.4% [1.5–5.6]; P = .014). Treg cells isolated from HCC suppressed proliferation of autologous circulating CD4+CD25 cells and perforin expression and proliferation of autologous CD8+ T cells. The proportion of circulating Treg cells in patients with HCC was similar in healthy controls (7.2% [1.2–23.3] and 9.2% [1.6–30.2], respectively), but the proportion of circulating Treg cells that were also transforming growth factor β1+ was elevated in HCC compared with controls (55.5% [8.2–73.9] and 2.0% [0–4.9], respectively; P = .003). In conclusion, TILs are compromised and contain a subpopulation of suppressive CD4+CD25+Foxp3+ Treg cells. Functional deletion of tumor-infiltrating Treg cells could enhance tumor-specific immunotherapy. (HEPATOLOGY 2005;41:722–730.)

Hepatocellular carcinoma (HCC), the fifth most common cancer worldwide with 564,000 new cases each year, is also the third most common cause of cancer-related death.1 HCC arises most frequently in males with cirrhosis, which is most often a consequence of chronic hepatitis infection or alcohol abuse.2 Recent reports from different countries suggest that the incidence of HCC is increasing, probably as a consequence of the increased prevalence of hepatitis C virus (HCV) infection, although increased alcohol consumption may be significant.3

The only effective approaches for patients with HCC are resection or liver transplantation. Following transplantation, there is an 83% 4-year recurrence-free survival in highly selected patients (single tumor <5 cm in diameter or fewer than three tumors <3 cm in diameter).4 However, the majority of patients do not meet such strict criteria or have other contraindications. Local therapies such as percutaneous ethanol injection, thermal ablation, and intra-arterial chemoembolization are less successful,5 and less than 10% of patients with moderate disease (Okuda stage 2) survive for 3 years.6

Immunological mechanisms are important in the surveillance of malignancy and control of tumor progression. Cytotoxic CD8+ lymphocytes (CTLs) and natural killer cells are potential effector cells in the control of tumor growth, although both require CD4+ T helper 1 immune responses for optimal function.7 Tumor-infiltrating lymphocytes (TILs) have been described in HCC,8 and extensive infiltration has been associated with reduced tumor recurrence following resection.9, 10

Continued tumor growth—despite the presence of a lymphocytic infiltration including tumor-specific T cells within and surrounding tumors—suggests a failure of immune control. Many mechanisms have been proposed for an attenuated immune response to tumors; these include partial antigen masking, failure of antigen processing, inadequate costimulation, and direct suppression of effector cells.11

T-regulatory lymphocytes (Treg cells) are a subgroup of CD4+ T cells with suppressor function characterized by expression of the interleukin 2 receptor α chain (CD25) in the resting state. Treg cells suppress activity and proliferation of both CD4+CD25 and CD8+ T cells12 in a contact-dependent manner through inhibition of interleukin 2 production,13 express high level intracellular CTLA-4, and do not proliferate in vitro following T-cell receptor stimulation.14 Recent murine studies indicate that Foxp3 is a specific marker of Treg cells,15, 16 and although human studies suggest that Foxp3 induction can be induced in other CD4+ cells, these studies confirm that Foxp3 induction is associated with suppressive activity.17 Treg cells secrete transforming growth factor (TGF)-β1 and interleukin 1018, 19 and may express membrane-bound TGF-β1, the levels of which are increased upon stimulation in vitro.20–23

Treg cells confer protection against autoimmunity and mediate transplant tolerance24 and may play a role in the regulation of tumor immunity.25 Increased Treg cell numbers are present within TILs of non–small cell lung, ovarian, breast, and pancreatic cancer.26 Treg cells inhibit autologous T-cell proliferation and interferon γ production in patients with lung cancer.27 Recent human studies suggest that interleukin 10 and TGF-β1 production by tumor Treg cells may mediate these effects.28 Murine studies have demonstrated that the rejection of early tumors is induced by the deletion of Treg cells.29–31

We propose that tumor-specific Treg cells accumulate locally in HCC and inhibit protective nonspecific and tumor-specific immune responses.


HCC, hepatocellular carcinoma; Treg cell, T-regulatory lymphocyte; TIL, tumor-infiltrating lymphocyte; HCV, hepatitis C virus; CTL, cytotoxic CD8+ lymphocyte; TGF, transforming growth factor; PBC, primary biliary cirrhosis; HBV, hepatitis B virus; PBMC, peripheral blood mononuclear cell; LAP, latency-associated protein.

Patients and Methods


Fresh tissue was obtained prospectively from 12 patients undergoing liver resection or liver transplantation for hepatocellular carcinoma. Eleven patients had cirrhosis, which was a consequence of HCV infection in 5, alcohol abuse in 4, primary biliary cirrhosis (PBC) in 1, and hemochromatosis in 1. The remaining patient had hepatitis B virus (HBV) infection (HBV DNA–positive) but without evidence of hepatic fibrosis on liver biopsy. Immunocytochemistry was undertaken on paraffin-embedded, formalin-fixed specimens taken from these patients. Immunohistochemistry for Mcm-2 was performed on a further 15 specimens taken from patients who had undergone liver transplantation for cirrhosis complicated by HCC.

Peripheral blood samples were obtained from 25 additional patients known to have HCC. The diagnosis of HCC was made using characteristic imaging or biopsy. Of these, 24 patients had cirrhosis, which was a consequence of HCV in 7, HBV in 2, alcohol in 9, PBC in 1, and hemochromatosis in 2. Cirrhosis was cryptogenic in 3 patients. The remaining patient had HBV infection without cirrhosis. Control blood samples were also taken before donation from 48 healthy blood donors attending the blood transfusion center, all of whom were negative for antibodies to HCV, HBV, HIV, and syphilis. Control tissues were obtained from patients with PBC or acute rejection after liver transplantation.

All samples were collected with approval of the Addenbrooke's Hospital Research Ethics Committee.

Culture Media, Reagents, and Antibodies.

For all in vitro assays, complete medium was used (RPMI 1640 medium [Invitrogen, Paisley, UK] supplemented with L-glutamine [2 mmol/L], penicillin/streptomycin [100 IU/mL and 100 μg/mL, respectively; Sigma-Aldrich, Gillingham, UK] and Hepes buffer [Sigma-Aldrich], with 10% [final concentration] human AB serum [First Link UK, Birmingham, UK]). Cells were incubated at 37°C with 5% CO2 and 95% air. FITC or Cychrome (PE-Cy-5)-conjugated anti-CD4, PE-Cy-5 and PE-conjugated anti-CD25 and anti-CD8, PE-conjugated anti-CD28, PE-Cy-5 conjugated anti-CD45RO and anti-CTLA-4 and FITC-conjugated 62L, TGF-β receptor II, anti-CD27, anti-CCR7, perforin and Ki-67 (Pharmingen, Cowley, Oxford, UK) were used in flow cytometry. Active TGF-β1 (R&D Systems, Abingdon, Oxon, UK) was biotinylated and labeled with streptavidin-PE (Pharmingen). For intracellular staining, cells were fixed and permeabilized with a BD cytofix/cytoperm kit (Pharmingen) according to the manufacturer's instructions. Flow cytometry data was acquired on FACSCaliber (Becton Dickinson, Cowley, Oxford, UK) and analyzed using WinMDI software.

Isolation of Peripheral Blood Mononuclear Cells.

Circulating peripheral blood mononuclear cells (PBMCs) were isolated from blood via density gradient centrifugation over lymphoprep (Axis Shield, Dundee, UK). Buffy coat lymphocytes were washed twice in phosphate-buffered saline and resuspended in complete medium. Magnetic bead separation was used to isolate CD25 or CD25+ cells. Cells were incubated with anti-CD25 microbeads for 15 minutes at 4°C before being passed through a magnetic MS column, following the manufacturer's instructions (Miltenyi Biotec, Bisley, Surrey, UK). Depletion achieved 92% to 99% purity measured by flow cytometry. The CD25+-enriched population was 60% to 70% pure.

Isolation of Lymphocytes From Liver Tissue.

Liver biopsy specimens suspended in complete medium were disaggregated using 21-gauge green needles before being filtered through a 70-μm nylon mesh. The final sample was layered on lymphoprep and centrifuged at 625g for 20 minutes. Buffy coat lymphocytes were washed in phosphate-buffered saline and resuspended in complete medium for use in FACS analysis or further experiments. Cell viability always exceeded 95% by trypan blue (Sigma-Aldrich).

Lymphocyte Proliferation.

Using 96-well plates, 2 to 5 × 104 responder cells (CD25 or CD8+ lymphocytes) were cultured alone or together with 2 to 5 × 104 CD25+ lymphocytes per well (final volume: 200 μL/well) in a 1:1 ratio. Cells were stimulated with soluble anti-CD3 at a final concentration of 1 μg/mL in the presence of 5 × 103 autologous irradiated PBMCs (or soluble anti-CD28 (1 μg/mL) in the case of CD8+ lymphocytes). After 60 hours in culture, 0.5 μCi-3[H]-thymidine was added to each well, and the cells were cultured for a further 6 hours before 3[H]-thymidine incorporation was assessed via liquid scintillation spectrometry. All experiments were performed in triplicate.

Immunohistochemical Staining of Paraffin-Embedded Tissue.

Paraffin-embedded, formalin-fixed liver tissue, obtained from patients undergoing orthotopic liver transplantation for HCC and anonymized in accordance with local ethical guidelines, was cut into 5-μm sections and placed on polylysine-coated slides. Slides were processed for immunohistochemistry as previously described.32 Antigen retrieval was achieved via pressure-cooking for 3 minutes in citrate buffer (pH 6.0). The following antibodies were used: mouse monoclonal; CD4, CD8, CD57, CD25, granzyme B, perforin, Fas, Fas ligand (all from Novo Castra, Newcastle-upon-Tyne, UK); Mcm-233; goat polyclonal Foxp3 (Abcam, UK). Biotinylated goat anti-mouse immunoglobulin was applied as a secondary antibody where mouse monoclonal antibodies were used and rabbit anti-goat immunoglobulin where goat polyclonal antibodies were used. Negative controls were performed by omitting the primary antibody. For assessment of CD8, perforin and granzyme B staining, liver biopsy tissue taken from patients with PBC was used as a positive control because PBC is an autoimmune disease associated with a vigorous CD8+ portal tract infiltrate.

TGF-β1, TGF-β1LAP, TGF-βRI, and TGF-βRII Expression.

TGF-β1 is secreted in an inactive form bound to protein (latency-associated protein [LAP]). After undergoing conformational change, TGF-β1 exerts its effects via membrane receptors TGF-β receptor I (TGF-βRI) and TGF-β receptor II (TGF-βRII). These combine to form a ligand-receptor complex and initiate a signaling cascade involving SMAD proteins that results ultimately in growth inhibition of target cells.34 An antibody to TGF-βRII was not available for use in immunohistochemistry, so antibody to TGF-βRI (Novo Castra) was used as an alternative. Antibody to TGF-β1 bound to LAP (TGF-β1LAP) was used in immunohistochemistry (R&D Systems).

Assessment of Proliferation.

Evidence of proliferation was sought using Mcm-2 (immunohistochemistry). An antibody to Mcm-2 was not available for use in flow cytometry; for these experiments an antibody to the conventional marker, Ki-67, was used. Mcm-2 is one of a closely related family of proteins (the mini-chromosome maintenance [Mcm] proteins) that form a prereplicative complex essential for DNA replication. Mcm proteins are present throughout the cell cycle but are lost from the cell during quiescence.32 The function of the Ki-67 antigen is unknown, but expression can be altered by other factors such as nutrient deprivation.35

Double-Labeling Studies.

Double-labeling for Mcm-2 and CD25 was performed in a subset of HCC sections (n = 3). For comparison, we used biopsies taken from patients with confirmed acute rejection as an example of a fully activated immune response (n = 5). The primary antibody was applied and incubated overnight. After washing, the sections were incubated with Alexa Fluro goat antimouse 488 (Molecular Probes, Leiden, The Netherlands), followed by a blocking step with F(ab)2 goat antimouse immunoglobulin G fragments (Jackson Immuno Research Laboratories, Soham Cambridgeshire, UK). A further washing step was then performed before incubation with the second primary antibody. Following a final washing step, the sections were incubated with Alexa Fluro goat anti-mouse 546 (Molecular Probes). Slides were counterstained using 4,6-diamidino-2-phenylindole (Sigma-Aldrich), washed, and mounted in fluorescent mounting media. Single-stained slides and primary antibodies applied in reverse order were included as controls.

Images were viewed and assessed using a Leica confocal microscope and a Zeiss Axioplan 2 confocal microscope (Heidelberg, Germany) at wavelengths of 488 and 546 nm.

Assessment of Sections.

For the enumeration of Mcm-2 positive lymphocytes, lymphocytes were counted in three high-powered fields (×40) by two independent observers. For each patient, the mean percentage of positive cells was taken. Results are expressed as the median and range of 15 patients.


The Mann-Whitney U test was used for the statistical analysis of nonparametric variables and the paired Student t test was used to analyze paired variables. Values of P < .05 were regarded as significant. Unless stated otherwise, all results are expressed as the median and range.


Lymphocytes Infiltrating Hepatocellular Carcinoma Proliferate Rarely.

Mcm-2 was used to assess proliferation of TILs in situ (n = 15). Despite a significant lymphocytic infiltrate, 97.9% (91.7–100) of TILs in HCC had not entered the cell cycle (Fig. 1A). Using Ki-67 to assess the proliferation of freshly isolated TILs, flow cytometry findings were similar; a minority of infiltrating CD4+ lymphocytes were in cell cycle (4.0% [3.2–14, n = 3]). The majority of TILs were CD4+ and CD8+ lymphocytes, although the CD4+CD8+ cell ratio varied substantially between samples. We used CD57 as a natrual killer cell marker, although it is known that CD57 may also be expressed on other cell types, including activated T cells. In all cases, CD57+ cells represented less than 10% of infiltrating lymphocytes (Fig. 1B–D). This low level of CD57 staining within our HCC samples suggests that neither CD57+ natural killer cells nor any other CD57+ cell populations are present in significant amounts in HCC.

Figure 1.

Immunohistochemistry of tumor-infiltrating lymphocytes. Adjacent sections of paraffin-embedded HCC tissue (original magnification ×60) were stained with antibodies to (A) Mcm-2, (B) CD4, (C) CD8, or (D) CD57. Positive lymphocytes are stained brown. The majority of lymphocytes are not proliferating, with blue nuclei. A rare proliferating lymphocyte is seen (black arrow). A proliferating malignant hepatocyte serves as a positive internal control (white arrow).

Functional Compromise of CD8+ Lymphocytes Infiltrating HCC.

Adjacent sections from 10 patients with a significant CD8+ infiltrate were stained for granzyme B and perforin, both markers of cytotoxic potential; less than 10% of CD8+ cells were positive for either granzyme B or perforin (Fig. 2A–C). Sections from patients with PBC (positive controls) revealed perforin expression in 65% (50%–80%) of CD8+ lymphocytes (Fig. 2E) and granzyme B expression in 30% (25%–50%) (Fig. 2F). Flow cytometry of isolated TILs confirmed that intracellular perforin was seen in a minority of CD8+ cells (10% [8%–24%]). Infiltrating CD8+ lymphocytes were also CD45RO+, CD27+, CD28+, and CCR7+, suggesting a lack of terminal differentiation (Fig. 3).36 Similar findings have been reported in melanoma.37 The Fas/FasL pathway appeared intact. FasL was expressed on the surface of infiltrating lymphocytes (> 90%). Fas antigen was seen on the membrane of malignant hepatocytes and on infiltrating lymphocytes (30% [10%–40%]) (Fig. 2G–H).

Figure 2.

CD8+ lymphocytes lack markers of cytotoxicity. (A–C) Adjacent sections of HCC tissue (original magnification ×40) were stained with antibodies to (A) CD8, (B) perforin, or (C) granzyme B. The majority of lymphocytes are negative for granzyme B or perforin (blue). PBC tissue was used as a positive control for (D) CD8, (E) perforin, or (F) granzyme B. Sections of HCC tissue were also stained for (G) Fas ligand and (h) Fas antigen. Fas antigen is seen on malignant hepatocytes (white arrow). The majority of lymphocytes express FasL and a proportion also express Fas antigen. In all cases, black arrows identify positive brown staining lymphocytes.

Figure 3.

Phenotype of CD8+ lymphocytes infiltrating HCC. (A) Flow cytometry of HCC-isolated lymphocytes with CD8+ lymphocytes gated as shown. The gated CD8+ population was assessed for the intra-cellular expression of perforin (B) and surface expression of (C) CD45RO, (D) CD28, (E) CD27 or (F) CCR7. Representative histograms are shown in red and in each case the isotype control is shown in black.

CD4+CD25+ Treg Cells Are Concentrated Within HCC.

Fresh TILs were isolated from liver tumors in 12 individuals undergoing surgery or liver transplantation and characterized via flow cytometry. Liver sections from the same patients were examined subsequently via immunohistochemistry for comparison.

Flow cytometry analysis revealed a population of cells with the Treg cell phenotype; they were membrane-positive for CD4, CD25, and CD45RO, with high expression of intracellular CTLA-4. Populations of both CD62Lhigh (60%) and CD62Llow (40%) CD25+ cells were seen (Fig. 4). It has been suggested that the CD62Lhigh population has greater suppressive capacity.38 The number of isolated TILs was small, so TGF-β1 expression was studied in only 3 patients. A proportion of CD4+CD25+ cells expressed active TGF-β1 on the membrane (35.6%, 8%–63%; n = 3; see Fig. 4). Treg cells represented 8.7% (1.4%–13.8%) of all isolated TILs. The numbers were similar regardless of the cause of liver disease. However, in 6 patients studied further (3 with alchoholic liver disease, 1 with PBC, 1 with haemochromatosis, and 1 with cryptogenic cirrhosis), the proportion of Treg cells isolated from tumor tissue always exceeded that in distant nontumor liver tissue (8.0% [6.0%–12.2%] and 2.4% [1.5%–5.6%] of all isolated lymphocytes, respectively; P = .014).

Figure 4.

Flow cytometric analysis of HCC-infiltrating lymphocytes. Representative histograms are shown. (A) CD4+CD25+ Treg cells are represented in the red box. The gated CD4+CD25+ population (red box) was assessed for the surface expression of (B) CD45RO, (C) CD62L, or (D) TGF-β1. Permeabilized cells were analyzed for intracellular (E) CTLA-4 or (F) Ki-67. The black line represents the isotype control. TGF-β, transforming growth factor β.

Immunohistochemistry confirmed that CD25+ lymphocytes were infiltrating the tumor diffusely and in small clusters (Fig. 5B). Their distribution was similar to that of CD4+ cells (Fig. 5A) but represented a far smaller proportion. A distinct population of lymphocytes expressing TGF-βLAP could also be seen infiltrating the tumor (Fig. 5C). Sinusoidal lining cells also demonstrate positive staining with this antibody and serve as an internal positive control. Membrane expression of TGF-βRI was evident on the majority of TILs (Fig. 5D). Foxp3+ lymphocytes infiltrated HCC tissue in a diffuse manner (Fig. 5E).

Figure 5.

Immunohistochemistry of TILs. Adjacent paraffin sections of HCC (original magnification ×40) were stained with antibodies to (A) CD4 or (B) CD25. (C) Sections were stained with antibody to TGF-βLAP. Membranous staining is seen in a proportion of infiltrating lymphocytes (black arrow). Sinusoidal lining cells served as an internal positive control. (D) Sections stained with antibody to TGF-βRI. The vast majority of infiltrating lymphocytes demonstrate positive membrane staining (black arrow). (E) Foxp3+ lymphocytes, with surface and cytoplasmic staining (black arrow), are infiltrating HCC diffusely.

Proliferative Status of CD25+ Lymphocytes Distinguishes Activated T Cells From Treg Cells.

Treg cells share some phenotypic features with activated CD4+ T lymphocytes. Intracellular expression of Ki-67, a nuclear marker of proliferation, in Treg cells and activated lymphocytes was assessed via flow cytometry. There was clear distinction between the groups. The majority of activated (CD25+) T cells expressed Ki-67. In contrast, tumor-derived Treg cells expressed Ki-67 rarely (Fig. 6). Double-labeling studies for Mcm-2 and CD25 in three cases revealed significant CD25+ lymphocyte infiltration with almost total absence of Mcm-2 (Fig. 7D). Mcm-2 and CD25 expression were seen in 90% of portal tract lymphocytes in the livers of patients with acute rejection (Fig. 7A–B). Double-labeling studies confirmed the colocalization of Mcm-2 with CD25 in these patients (Fig. 7C).

Figure 6.

Ki-67 staining distinguishes activated CD4+ cells from Treg cells. Isolated PBMCs were depleted of CD25+ cells before stimulation with soluble anti-CD3 for 5 days in the presence of irradiated antigen-presenting cells. (A) Before stimulation, depleted CD4+ cells were 98% pure. (B) Following stimulation, the proportion of CD4+ T cells expressing CD25 rose (red box). (C) Histogram showing high expression of nuclear Ki-67 in activated CD4+CD25+ cells (red). Black overlay demonstrates low Ki-67 expression on unstimulated CD4+CD25+ Treg cells.

Figure 7.

Mcm-2 and CD25 staining in acute liver graft rejection. Ninety percent of portal tract lymphocytes are positive for (A) CD25 and (B) Mcm-2. (C) Double-labeling confocal microscopy for Mcm-2 (red) and CD25 (green) confirms that these markers are expressed together. (D) Confocal microscopy for Mcm-2 and CD25 in HCC. No double-positive lymphocytes are seen. A malignant Mcm-2–positive hepatocyte serves as a positive control.

Circulating Treg Cells in Patients With HCC and in Healthy Controls.

Flow cytometry was performed on PBMCs from 25 patients with HCC and from 48 healthy blood donors. Treg cells represented 7.2% (1.2%–23.3%) of the total lymphocyte population in the peripheral blood of patients with HCC. There was a similar number in blood from healthy individuals (9.2% [1.6%–30.2%]; P = .895). There was no difference in the percentage of circulating Treg cells between patients with chronic HBV infection or chronic HCV infection when compared with healthy individuals (Rushbrook, Unitt, and Alexander, personal communication, 2004). Unstimulated circulating Treg cells from patients and healthy controls expressed membranous CD4, CD25, CD45RO, and CD62L. Treg cells also demonstrated high expression of intracellular CTLA-4 but did not proliferate upon stimulation of the T-cell receptor with anti-CD3. In patients with advanced HCC, 55.5% (11%–74%; n = 7) of circulating unstimulated CD25+ cells expressed membranous TGF-β1 (all patients had cirrhosis, related to HBV in 2, alcoholic liver disease in 3, hemochromatosis in 1, and cryptogenic in 1). In comparison, only 2.0% (0%–4.9%) of circulating CD25+ cells from healthy controls expressed TGF-β1 (n = 5; P = .003). TGF-β1 was not expressed on CD25 cells from patients or healthy blood donors; however, the proportion of CD25+ Treg cells that expressed active TGF-β1 rose to a similar level in patients and healthy controls following nonspecific stimulation with anti-CD3 in vitro (Fig. 8). Membrane expression of TGF-βRII was not detected on circulating PBMCs from patients or healthy controls but was detected in both groups on CD25+ and CD25 cells after nonspecific stimulation with anti-CD3 in vitro (see Fig. 8). Circulating CD25+ Treg cells from both patients and controls suppressed proliferation of CD4+CD25 T cells after nonspecific stimulation with anti-CD3 in vitro in coculture experiments (n = 3) (data not shown).

Figure 8.

TGF-β1 and TGF-βRII expression. (A,B) CD4+CD25+ (red box) and CD4+CD25 (black box) cell populations were gated as shown. Histograms demonstrate expression of membranous TGF-β1 on CD4+CD25+ cells (C,D) and TGF-βRII on CD4+CD25+ (E,F) and CD4+CD25 cells (G,H) from healthy individuals before and after stimulation with soluble anti-CD3 (72 hours). (I) Membranous TGF-β1 on in vitro unstimulated CD4+CD25+ cells and (J) after stimulation from a patient with HCC. In each case, the respective CD25+ or CD25 populations are represented in red, with the black overlay representing the isotype control.

Tumor Treg Cells Suppress Autologous Circulating CD4+ and CD8+ T Cells.

Coculture of tumor Treg cells with autologous circulating lymphocytes in a 1:1 ratio confirmed their ability to suppress proliferation of CD4+CD25 cells measured by both nuclear Ki-67 expression and 3[H]-thymidine incorporation after nonspecific stimulation with anti-CD3 in vitro. Sixty-five percent (42%–87%) suppression was seen (n = 4). Flow cytometry on day 5 demonstrated that tumor Treg cells prevented upregulation of CD25 on autologous circulating CD4+CD25 lymphocytes. Treg cells also suppressed proliferation (suppression 52.6% [49%–56%]; n = 3), perforin expression (95.5% vs. 77.8%; P = .01), and interferon γ secretion (568.5 pg/mL vs. 276.27 pg/mL; P = .031) of autologous CD8+ cells. Serial dilution study confirmed that a Treg/CD8+ ratio of 1:10 resulted in 40% suppression of proliferation.


This study revealed that tumor-infiltrating T cells were quiescent because they expressed neither nuclear Ki-67 nor Mcm-2 as measured using two techniques. In addition, a majority of infiltrating CD8+ lymphocytes do not express either perforin or granzyme B, which suggests they are incapable of cytotoxic function. This anergy could be mediated by CD4+CD25+Foxp3+ Treg cells, which are concentrated in tumor tissue compared with cirrhosis tissue. Their effects may be mediated through the TGF-β receptor, which is expressed on the majority of tumor-infiltrating CD25 T cells and is induced on circulating CD25 T cells through stimulation of the T-cell receptor. Our findings suggest that the Fas/FasL pathway remains intact and may indicate that Treg cells have no effect on this pathway. We are not aware of any data suggesting that this pathway is affected by Treg cells.

CD4+CD25 Treg cells play a critical role in protection against autoreactivity. Because tumor antigens are often derived from self, it has been suggested that CD4+CD25+ Treg cells may hinder tumor-specific immune responses. Several groups have demonstrated the presence of Treg cells among TILs in various malignancies,26–28 but we are unaware of any publication demonstrating tumor-infiltrating Treg cells in HCC.

Other groups have suggested that the number or proportion of circulating Treg cells is increased in patients with cancer.25, 39 However, no increase in the proportion of circulating Treg cells was observed in HCC patients when compared with healthy individuals in this series. In contrast, the proportion of circulating Treg cells expressing surface TGF-β1 was markedly elevated in patients with HCC compared with healthy controls. The consistent gradient between the proportion of Treg cells in cirrhosis tissue and that within HCC, which is unrelated to the cause of cirrhosis, suggests a preferential accumulation of Treg cells to HCC, possibly mediated by specific chemokines.40 The local tumor cytokine milieu, which is rich in TGF-β1, may also favor accumulation or induction of a regulatory phenotype.

It has been suggested that immature or tolerogenic dendritic cells induce a Treg cell phenotype preferentially. It is noteworthy that Chen et al.41 could not find CD83+ dendritic cells in HCC nodules, suggesting that the immature phenotype of dendritic cell predominates in HCC. Some HCC cell lines secrete MIP3-α, a potent chemoattractant for immature dendritic cells, and it has been reported that serum levels of MIP3-α are elevated in patients with HCC.42 Other reports indicate that TGF-β1 may prevent the final maturation of dendritic cells.43

A TGF-β1–rich tumor environment would be ideal for the maintenance of tolerogenic dendritic cells and induction of Treg cells. Giannelli et al.44 reported that TGF-β1 stimulates α3-integrin expression at a transcriptional level in noninvasive HCC, causing transformation into an invasive phenotype. The secretion of TGF-β1 by Treg cells may therefore potentiate HCC invasiveness through induction of integrins.

Treg cells represent approximately 10% of TILs, and this in vitro study indicates that they are effective at these ratios and that tumor Treg cells suppress immune responses of circulating lymphocytes to nonspecific stimuli. Although a tumor-specific antigen would be a more suitable substrate, none has been identified reliably. We therefore relied upon nonspecific but conventional stimuli for induction of lymphocyte proliferation. Virus-specific activation of Treg cells with hepatitis B and hepatitis C viral proteins has been previously demonstrated (Rushbrook, Unitt, and Alexander, personal communication, 2004). In this study, a higher proportion of Treg cells from patients with HCC expressed membranous TGF-β1 than healthy controls. However, similar expression could be induced in Treg cells from healthy volunteers following nonspecific activation in vitro. Other groups have also demonstrated in vitro that TGF-β expression on Treg cells rises upon stimulation.21, 22 Based on this premise, our findings suggest that circulating Treg cells may be “activated” in patients with HCC. The expression of membranous TGF-β1 on Treg cells within a tumor suggests antigen-specific activation in situ, or migration of activated Treg cells. Circulating CD4+CD25+TGF-β1+ cells have also been reported in a group of patients with a variety of malignant epithelial neoplasms, although this did not include HCC.45

Previous studies using peripheral blood have demonstrated clearly that regulatory function is abrogated by depletion/deletion of Treg cells. In contrast, we found that depletion of Treg cells from tumor tissue alone did not enhance CD25 T-cell proliferation (data not shown). This was disappointing from the viewpoint of future therapies based on immunomodulation of Treg cells. There is evidence that CD4+CD25+ cells can induce suppressor capability in CD4+CD25 cells ex vivo,46 which may help explain our findings. Camara et al.47 have shown that CD4+CD25+–induced CD8+ suppression was sustained in vitro for 48 hours.

In conclusion, these experimental data suggest that Treg cells play a role in controlling the immune response to HCC. Whether depletion of Treg cells from TILs in HCC can restore the proliferative and cytokine secretion responses of tumor-infiltrating CD4+CD25 T helper 1 cells and CTLs remains to be determined, but the presence of a significant Treg infiltrate is likely to reduce efficacy of other immunotherapeutic approaches. We recommend that Treg depletion play a part in the design of future trials of immunotherapy.