Immunotherapeutic strategies for hepatocellular carcinoma

GASTROENTEROLOGY 2004;127:S232–S241

Immunotherapeutic Strategies for Hepatocellular Carcinoma LISA H. BUTTERFIELD Departments of Medicine, Surgery, and Immunology, University of Pittsburgh Medical School, Pittsburgh, Pennsylvania

There is a continuing need for innovative, alternative therapies for hepatocellular carcinoma (HCC). Immunotherapy for cancer is attractive because of the exquisite specificity of the immune response. Activation of an HCC-specific response can be accomplished by strategies targeting tumor-associated self-antigens (for example, ␣-fetoprotein [AFP]). Gene array studies have added to the list of HCC-specific gene products that can be targeted. Alternatively, the immune response can be targeted against viral antigens in those patients infected with hepatitis B or C virus. Uncharacterized and mutated antigens can also be targeted with whole tumor cell or tumor lysate– based immunization strategies or with vectors coding for genes that make the tumor immunogenic, allowing the immune system to naturally evolve specificity against immunogenic target antigens. Strategies being investigated in animal models include increasing tumor immunogenicity by targeting cytokines or costimulatory molecules to tumor; immunization with tumor cells fused with antigen-presenting cells; adoptive transfer of viral antigen-specific T cells; and targeting AFP-expressing HCC cells by DNA, adenovirus, peptide, and dendritic cell (DC) strategies. Strategies that have been tested in human clinical trials include adoptive transfer of lymphocytes and autologous tumor-pulsed DC as well as 2 AFP-based strategies: AFP-derived peptides in Montanide and AFP peptides pulsed onto autologous DC. These trials, testing novel immune-based interventions in HCC subjects, have resulted in immunologic responses and have impacted recurrence and survival in HCC subjects.

here is a pressing need for novel strategies to impact the development and recurrence of hepatocellular carcinoma (HCC). The goal of immune-based therapies is to harness the sensitivity, specificity, and self-regulation of the immune system to find and eradicate tumor cells wherever they reside. To activate an immune response against a cell, the immune system must first detect that cell. CD8⫹ effector T cells identify targets by detecting the 8 –11 amino acid peptide for which their T-cell receptor is specific. Proteins in the cytoplasm of cells are processed into small peptides by the proteosome complex, which are transported by transporter associated with antigen processing

T

(TAP) proteins into the endoplasmic reticulum, where they can associate with major histocompatibility complex (MHC) class I molecules and be transported to the cell surface (Figure 1). It is there that these peptides, presented in the context of self MHC, are seen by CD8⫹ T cells, the cells that are most critical to antitumor responses. Exogenous proteins taken up by cells can also be presented to CD8⫹ effector cells, especially by professional antigen-presenting cells. To point the immune system in the right direction, a tumor-specific target is needed. Many proteins have been identified with differential expression in HCC cells, and strategies directed toward uncharacterized tumor-specific proteins are actively pursued. This review will describe potential targets for immunotherapy, the wide variety of strategies being used in in vitro studies and animal models to activate effective tumor-specific immune responses, and summarize results from clinical trials testing immunotherapy in subjects with HCC.

HCC Targets Serum ␣-fetoprotein (AFP) is a diagnostic marker for the presence of HCC and an obvious choice of a protein target made by HCC cells (Table 1). AFP is the major serum protein expressed by the fetus, in which serum levels are at a maximum of 3 mg/mL at 10 –13 weeks of development, the levels drop at birth to 30 – 100 ␮g/mL, and the level of AFP expression in normal adults is quite low, 1–3 ng/mL. Importantly, 50%– 80% of HCC reactivate AFP expression, secreting up to 1 mg/mL in serum. Targeting AFP is not a new idea; it Abbreviations used in this paper: AFP, ␣-fetoprotein; APC, antigenpresenting cells; DC, dendritic cells; DTH, delayed-type hypersensitivity; ELISPOT, enzyme-linked immunoSPOT; GM-CSF, granulocyte-macrophage– colony stimulating factor; HCC, hepatocellular carcinoma; IFN-␥, interferon-␥; IL, interleukin; KLH, keyhole limpet hemocyanin; LAK, lymphocyte-activated killer cells; MHC, major histocompatibility complex; PBL, peripheral blood lymphocytes; TAP, transporter associated with antigen processing; TIL, tumor-infiltrating lymphocytes; TNF-␣, tumor necrosis factor ␣. © 2004 by the American Gastroenterological Association 0016-5085/04/$30.00 doi:10.1053/j.gastro.2004.09.038

November Supplement 2004

Figure 1. Peptide processing. Proteins in the cytoplasm of cells are degraded by the proteosome complex, yielding short peptide fragments. A subset of these peptides is transported from the cytoplasm to the endoplasmic reticulum (ER) by transporter-associated protein (TAP). Once in the ER, the peptides can associate with major histocompatibility complex (MHC) class I molecules and move to the cell surface. Once on the cell surface, these complexes are scanned by CD8 T cells. Each CD8 T cell bears a T-cell receptor able to recognize specific peptide/MHC complexes. CTLp, cytotoxic lymphocyte precursor.

was intensely studied in the 1970s and 1980s as a target for antibody-based therapies, but these efforts were largely unsuccessful. Although AFP is secreted, its peptides are processed by the cell and can be presented to both CD8⫹ and CD4⫹ T cells. Antigens originally characterized by their melanoma tumor-specific expression have also been found to be expressed in HCC, for example, MAGE-A family genes and NY-ESO, members of tumor-specific “cancer-testes” gene families. These genes are on the X chromosome and are upregulated in tumors via decrease of promoter methylation. Table 2 shows the reported percentage of tumors that express one of the members of this gene family.1– 8 Although the percentages vary widely with the study and sample numbers, the MAGE family of genes is clearly overexpressed in a number of HCC tumors. Despite being self antigens, the MAGE proteins have been shown to be immunogenic in melanoma patients and are being pursued as targets in HCC. Hepatitis viral antigens are other potential targets for T-cell responses. They have the additional benefit of being foreign proteins, not seen as “self” by the immune system. Viral genes from HBV and HCV, like the HBx protein (which is believed to modulate p53 expression), are potential targets. When considering viral targets, it is important to consider directing the immune system to the nonmalignant infected instead of only malignant cells. Importantly, immune responses targeting viral gene products have been implicated in the pathogenesis of liver cell injury that

IMMUNOTHERAPY OF HCC

S233

leads to HCC; therefore it may be preferred to target molecules more specific to the tumor only. Many HCC express increased levels of oncogenes myc, fos, and ras, whose expression can be critical to the unregulated growth of the tumor. These molecules are also critical to normal growth patterns of noncancerous cells. Similarly, tumor suppressor genes p53 and pRB are mutated and/or dysregulated in HCC, but also are critical to normal cells. Targeting of these molecules for immunotherapy relies on a threshold effect, below which the molecules are not detected and above which the molecule becomes immunologically detectable. There is also a plethora of uncharacterized tumorspecific antigens. Some of these are “private” or only expressed by a single patient’s tumor, others are “public” or shared and are more applicable to generalizable immunization strategies. Several recent articles have taken strides toward better characterization of HCC gene expression profiles with gene array technology studies, comparing genes upregulated and downregulated in HCC tumors, compared with adjacent normal tissue9 –13 (Table 1).

Strategies In cases in which one wishes to target an overexpressed gene product, the strategy is to induce an immune response to that protein to eliminate all cells expressing that target. For proteins with reduced expression in tumors, the strategy would be to increase its level or replace the nonfunctional gene. This would be a gene therapy type of strategy, as would replacement of a mutated gene’s function with a good copy of the gene to add back the function. Strategies requiring genetic engineering of entire tumors have huge technological hurdles to overcome. Addition of normal p53 to HCC tumors by plasmid DNA injection has been tested in humans.14 An exquisitely tumor-specific way to target mutated genes is to direct the immune response to the mutation. From a practical standpoint, this would require that the targeted mutation be in a part of the protein that is processed and presented to the immune system bound to Table 1. Molecular Changes in HCC Change

Gene product (percent of HCC tested)

Increased

RAR␣, p53 (22%–46%), tetraspanin CO-029 (26%), cyclin D1, TGF␤, TGF␣, IGFII, N-ras, c-myc, c-fos, survivin, MAGE-A family (86%), AFP (50%–80%) CCR5, IFN-␥, IGFBPs, tetraspanins CD9 and CD82 p53 (30%–50%), pRB (20%–25%), ␤-catenin HBV and HCV gene expression (HBV X, HBV preS1/2; HCV E2 glycoprotein, NS5B)

Decreased Mutated Viral

S234

LISA H. BUTTERFIELD

GASTROENTEROLOGY Vol. 127, No. 5

Table 2. Tumor-Specific Cancer–Testes Antigen Expression in HCCa Study

MAGE-1

Luo, 2002 Chen, 2003 Yamashita, 1996 Kariyama, 1999 Tahara, 1999 Chen, 2001 Mou, 2002 Korangy, 2004

19 66 80 78 68

aPercent

70

MAGE-2

30

MAGE-3

MAGE-4

24 70

4 20

42 68

MAGE-10

MAGE-12

SSX-1

NY-ESO

38

0 40

80

36

36

30

30

53 24

of tumor samples tested scoring positive for mRNA.

MHC molecules. It would also be important that the mutation be in a common “hot spot” of the protein carried by the tumors of many patients. This makes specific immunologic targeting of mutated proteins difficult and largely patient-specific. The most practical way to target mutations as well as any patient-specific tumor gene regulation changes is to immunize with the patient’s own tumor. There are many strategies commonly used to direct the immune system to the tumor: making the tumor itself more immunogenic, altering the immunosuppressive tumor microenvironment, and generalized immune activation as well as antigen-specific immune activation. Despite the wide variety of genetic changes and gene regulation changes found in HCC cells, these tumors are not immunogenic. A central issue is that HCC cells do not seem to activate the immune system—they may express MHC class I, presenting peptides derived from proteins expressed internally, but they do not generally express MHC class II to activate helper cells, nor do they express other adhesion and costimulatory molecules to give strong positive signals to effector cells (Figure 2A). Antigen-presenting cells (APC), like B cells, macrophages, and especially dendritic cells (DC), do express these molecules and are specifically designed to stimulate immunity (Figure 2B).

Heat shock or hyperthermia treatment of cells enhances the expression of heat-shock proteins, which have been shown to be chaperones of proteins and peptides as well as immunogenic delivery vehicles for tumor-derived

In Vitro Studies A variety of recent in vitro studies have tested some of these targets and strategies, including strategies to make the tumor sufficiently immunogenic to directly activate an immune response. The bacterial superantigen SEA has been linked to a transmembrane sequence and used to transfect the HCC line HepG2.15 The foreign bacterial protein is hypothesized to create an immunogenic danger signal to the immune system. The combination of SEA antigen with tumor was superior to HepG2 cells alone in stimulating HepG2-cell-specific cytotoxic T lymphocytes.

Figure 2. (A) Tumor cells do not activate CD8 T cells. Tumor cells may or may not express MHC class I/peptide complexes on their surface, enabling recognition by previously activated T cells. They do not express the optimal arrangement of costimulatory and adhesion molecules required for activation of naive T cells. They may also secrete inhibitory cytokines such as IL-10 and TGF␤. ICAM, intercellular adhesion molecule; TCR, T-cell receptor. (B) APC activate T cells. APC, like B cells, macrophages, and DC, express an array of costimulatory molecules in addition to high levels of MHC class I and II. This enables APC to strongly activate T cells, assisted by additional molecules such as adhesion molecules and synthesis of cytokines such as IL-12 p70.

November Supplement 2004

antigens. Heat shock creates a stress and “danger” environment. Tumor lysate from heat-shocked HepG2 cells was shown to be more immunogenic when pulsed onto APC than lysate from non– heat-shocked HepG2 cells.16 DC have been the subject of extremely active research for the last decade. These bone-marrow-derived cells have been identified as the most potent immune-stimulatory cells known, specialized for the initiation of and shaping of immune responses of all kinds.17 Defects in the function of DC have been identified in HCC patients and in those infected with either HBV or HCV. For example, the balance of different types of DC, myeloid and plasmacytoid DC, have been found to be skewed in subjects with HBV infection,18 and generalized impairment in these cells’ ability to function has been reported in both HBV-positive and HCV-positive subjects.19 Lastly, T cells specific for HCC have been isolated and analyzed in vitro. These preclinical studies have been directed at identifying tumor antigen-specific cells or nonspecific effector cells with antitumor activity. Methods of isolating and expanding functional tumor-infiltrating lymphocytes (TIL) from HCC subjects have been performed,20 MAGE antigen-specific T cells have been found in TIL from HCC subjects,8 and AFP-specific T cells have been identified in patients with cirrhosis and HCC, with and without immunization.21,22

Animal Models Because immunotherapy strategies are systemic therapies that are designed to harness the entire immune system, a model with intact primary and secondary lymphoid structures will yield vital information on induction and regulation of the resultant immune response as well as the potential for autoimmune responses coincident with targeting of self tumor antigens. When drawing conclusions from murine models, caution must be used because there are key differences between the murine and human immune systems. For example, human white blood cells comprise ⬃60% neutrophils and ⬃40% lymphocytes, whereas murine white cells are only ⬃15% neutrophils and as high as ⬃80% lymphocytes. Conversely, delayed-type hypersensitivity (DTH) lesions in humans (used to assess successful immunization) are lymphocyte-rich, whereas murine DTH lesions are neutrophil-rich. Critical regulatory molecules such as Tolllike-receptor-9 are only expressed in plasmacytoid DC in humans, whereas this immunostimulatory DNA response element is expressed in both myeloid DC and plasmacytoid DC in mice, making responses to DNAbased vaccines more potent in mice. Despite these dif-

IMMUNOTHERAPY OF HCC

S235

Figure 3. DC-based strategies for immunotherapy. DC can be used to present tumor antigens to the immune system in several ways. (1) Tumor cell lysate or purified proteins can be mixed with immature DC so that the DC will process and present multiple peptides in MHC class I and II. (2) Mild acid-eluted mixed peptides from tumor cells or large concentrations of synthetic peptides of known epitopes can be pulsed onto the DC surface. (3) DC can be engineered with plasmid DNA or viruses to express specific gene products. (4) DC can be fused with entire tumor cells via polyethylene glycol or electrofusion.

ferences, preclinical studies in mice yield important data that are often subsequently confirmed in human trials. In an effort to increase the immunogenicity of tumors, mice implanted with tumor cells that had been previously engineered to express immunostimulatory genes CD40L and/or Flt-3L via adenoviral vectors showed slowed tumor growth compared with parental tumors, primarily because of activation of natural killer cells.23 However, when first implanted with parental tumors and then treated in vivo with these vectors, a more clinically relevant strategy, no antitumor response was detected. This is a common observation in murine tumor models when prevaccination strategies are compared with established tumor treatment, possibly because of the faster kinetics of transplantable murine tumors or the weakness of the therapeutic strategy being tested. Engineering an entire tumor is quite challenging, and there are many technical hurdles, such as efficient in vivo transfection, that need to be addressed. Many strategies using the immune-activating ability of professional APC are being pursued. There are many ways to use the potent DC to present tumor-derived antigens to the immune system (Figure 3). Adding tumor lysate or purified proteins to DC takes advantage of the active macropinocytosis (and other uptake mechanisms) of the immature DC to load them with either known or unknown protein. This is subsequently processed into MHC class I and II to activate both CD8⫹ and CD4⫹ T cells. One can coat DC with either acid-

S236

LISA H. BUTTERFIELD

eluted mixed tumor peptides or high levels of synthetic peptides. Antigen engineering of DC with plasmids (which is extremely inefficient) or with viral vectors such as adenovirus24 allows the DC to express entire antigens. Fusion of DC and tumor cells via PEG, or more recently, electrofusion, transfers any and all tumor proteins to a cell that also expresses necessary immune-activating molecules. A report from 1994 in which rats were cured of established tumors by treatment with a vaccine of tumor cells fused to activated B cells generated much excitement.25 Tumor-APC fusion is a very controversial area (for technical reasons regarding the fusion process). Nevertheless, other groups inspired to test this technology were not able to fully reproduce the results. Recent reports observed only prevention of tumor growth by prevaccination or a slowing rather than cure of tumor growth from treatment strategies.26 However, tumorlysate–loaded DC have been used successfully to treat small tumors,27 and addition of interleukin (IL)-12 to lysate-pulsed DC has improved antitumor responses.28 These animal data support further testing of tumorlysate–loaded DC, especially in the setting of low tumor burden.

Human Clinical Trials Immunotherapy trials for HCC to date fall into several categories: cytokine injection to nonspecifically activate subsets of immune cells and/or make the tumor more immunogenic; addition of lymphokine-activated killer (LAK) cells to therapy; infusion of TIL, or activated peripheral blood lymphocytes (PBL); infusion of antigenpresenting cells, and 1 recent report of an autologous tumor-based vaccine. The results of these different trials are mixed, but many have shown biological activity and a subset have shown clinical effects (Table 3). Cytokines In a trial with 20 stage III–IV patients from 1995, transarterial chemotherapy was combined with interferon gamma (IFN-␥, which would increase MHC expression for improved antigen presentation and possibly skew the immune response toward activation of cytotoxic T cells) and IL-2 (which activates and acts as a growth factor for T cells).29 Fourteen subjects had some level of tumor size decrease and reduced serum AFP levels, indicating a clear positive biological effect of this treatment combination. In a more recent trial in 15 patients with advanced, inoperable HCC, subcutaneous injection of IFN-␥ and granulocyte-macrophage colony stimulating factor (GM-CSF, which acts as a growth

GASTROENTEROLOGY Vol. 127, No. 5

factor for antigen-presenting cells such as DC that reside in the skin) was tested.30 However, no clinical responses were observed. Effector Cells Several trials have been reported testing infusion of different types of cytotoxic lymphocytes. Two separate but similar randomized trials were performed in 199131 and 199532 combining chemotherapy (Adriamycin) with LAK cells generated from splenocytes postresection. The first study showed a decrease in rate of recurrence, and the second showed no benefit overall. The second study did show a benefit for the subset of subjects with a negative margin of ⱖ1 cm postresection. The reason for this discrepancy is unknown. One issue with LAK cells is their lack of tumor antigen specificity. TIL, because they are derived from T cells infiltrating the tumor, have been shown to contain tumor antigen– reactive T cells. Also in 1991, a pilot trafficking study of indium111labeled TIL found that these cells (activated by IL-2 and CD3 receptor triggering), delivered via intrahepatic artery infusion, trafficked to sites of disease preferentially in all of 3 subjects.33 One subject had HCC, and the other 2 had liver metastases from colorectal cancer. In addition, 2 of the 3 subjects (1 HCC) had partial tumor responses lasting 5–10 months. In a study from 1997, 10 subjects received TIL (activated by IL-2 and LAK cell supernatant).34 Five subjects had multinodular tumors, the others were solitary, and the tumors ranged in diameter from 2.7 to 15 cm. Treated subjects had improved recurrence rates compared with historical controls. In by far the largest trial of its kind, 150 patients were randomized to receive either IL-2 and anti-CD3–activated PBL or observation after what was believed to be curative resection.35 This trial found statistically significant improvements in risk of recurrence, time to recurrence, and recurrence-free survival. Overall survival fell short of statistical significance at P ⫽ .09. Given the randomized nature of this large trial, its results offer more objective support for the potential for immunotherapy, especially in the adjuvant setting. APC Several small pilot studies have been performed testing infusions of APC. In each case, the APC were mixed with tumor or tumor lysate to prepare a patientspecific vaccine to activate an immune response to any and all antigens in the tumor, both tumor-specific and normal.

November Supplement 2004

IMMUNOTHERAPY OF HCC

S237

Table 3. Immunotherapy Clinical Trials Strategy

Year, Author

Patients

Settinga

Responses 14/20 tumor size decrease No survival benefit versus historical controls Improved recurrence rate with LAK

Chemotherapy ⫹ IFN-␥ ⫹ IL-2 IFN-␥ ⫹ GM-CSF

1995, Lygidakis et al.

20 stage III–IV

Transarterial infusion

2002, Reinisch et al.

15 inoperable stage III–IV

Cytokines sc.

Chemotherapy ⫾ LAK

1991, Une et al.

Randomized 12 per arm

Chemotherapy ⫾ LAK

1995, Kawata et al.

Randomized 12 per arm

TIL

1991, Takayama et al.

TIL

1997, Wang et al.

3 (1 HCC, 2 colorectal mets) 10

Adriamycin ⫾ IL-2– activated spleen LAK ia Adriamycin ⫾ IL-2– activated spleen LAK ia ln111-labeled TIL (IL-2/␣CD3) ⫹ chemotherapy IL-2/LAK supernatant activated TIL

PBL

2000, Takayama et al.

150 randomized postcurative resection

PBL ⫹ IL-2/␣CD3, iv

Activated B cells

1995, Wu et al.

11

Dendritic cells

2002, Ladhams et al.

2 metastatic

Tonsil-activated B cells PEG—fused to hepatoma cells, sc GM/IL-4 DC ⫹ tumor

Dendritic cells

2003, Iwashita et al.

10 unresectable

Dendritic cells

2003, Stift et al.

2 HCC of 20 total

Autologous tumor ⫹ cytokines

2004, Kuang et al.

41 randomized, 19 received vaccine

AFP peptides

2003, Butterfield et al.

6 stage IVa and IVb

GM/IL-4 DC ⫹ tumor lysate ⫹ TNF␣ ⫹ K LH, ln GM/IL-4 DC ⫹ tumor lysate ⫹ TNF␣ ⫹ IL-2, ln Formalin-fixed autologous tumor ⫹ GM-CSF ⫹ IL-2 ⫹ BCG, id AFP peptides in Montanide adjuvant id

Comments 14/20 reduced serum AFP

No differences in outcomes 2/3 PR

Cells trafficked to liver

Improved recurrence rates versus historical controls Significant improvements in risk of recurrence, time to recurrence, recurrence-free survival 3 PR

AACR abstract

1 patient slowed tumor growth 1/10 MR

7/10 KLH DTH⫹

Overall survival ⫽ 0.09, NS

No PR or CR

Statistically significant improvements in risk of recurrence, time to recurrence, recurrencefree survival No PR or CR

Improved overall survival ⫽ 0.01

Increased frequency of AFP T cells in blood

mets, metastases; PR, partial tumor response; CR, complete tumor response; AACR, American Association for Cancer Research. aRoutes of administration: sc, subcutaneous; ia, intra-arterial; iv, intravenous; ln, intralymphatically; id, intradermal.

In a trial based on the rat model of activated B cells fused to syngeneic tumor via PEG, the same group performed a trial with activated B cells from tonsils fused via PEG to tumor cells (Wu MC, Shen F, Che XY, Wang XN, Tykocinski M, Si MS, Guo YJ. Proc Am Assoc Cancer Res, 1995 AACR abstract). They treated 11 patients and reported 3 partial tumor responses. Further details were not available in this abstract. Three publications have reported the use of DC generated ex vivo from peripheral blood, pulsed with tumor or tumor lysate. In 1 report, 2 patients with metastatic HCC were treated with immature DC,36 grown in GMCSF (growth factor) and IL-4 (to block macrophage development from monocyte precursors), the most com-

mon and straightforward method for generation of DC. Of the 2 patients, 1 had slowed tumor growth compared with their pretreatment status. In general, immature DC are more effective at antigen acquisition than at T-cell stimulation. Immature DC are, therefore, optimal for mixing with tumor or tumor lysate to allow uptake and processing of large amounts of tumor antigens. However, more recent data indicate that potent T-cell activation is the result of a maturation step of DC (after antigen uptake), which reduces phagocytosis activity and upregulates T-cell activation cell surface molecules, improves trafficking to lymph nodes, and increases production of T-cell-activating cytokines such as IL-12p70. The degree of maturation required for optimal T-cell activa-

S238

LISA H. BUTTERFIELD

tion and the extent to which maturation may occur in vivo after injection of DC are unknown. A second trial used DC pretreated with GM-CSF and IL-4, then loaded with tumor lysate, stimulated with TNF-␣, and mixed with keyhole limpet hemocyanin (KLH) before injection.37 The KLH was added as a foreign protein that activates non–antigen-specific helper T cells and serves as a marker of successful immunization by subsequent DTH testing. In this trial, in which 10 subjects with unresectable HCC were treated, positive DTH responses to the KLH (indicating successful vaccination) developed in 7 patients, and 1 subject had a mixed tumor response. A third trial used tumor-lysate–loaded DC that were matured with TNF-␣, but also mixed with IL-2. A mixed population of subjects were treated, 2 of whom had HCC.38 There were no tumor responses in any treated subjects. Taken together, in the 25 HCC subjects treated with APC published to date, there have been 2 minor responses and 3 partial responses. These disappointing results may be related to the defects in APC function found in HCC subjects. The defects observed in APC function have been seen both in directly isolated DC as well as in DC cultured ex vivo from progenitors. Thus, use of DC in vaccination against tumors may be too inefficient to significantly impact the advanced stages of disease tested to date. Autologous Tumor Recently, a randomized phase II trial was published in which formalin-fixed autologous tumor (as a source of tumor antigens) was mixed with GM-CSF (as an APC growth factor), IL-2 (to activate T cells), and bacille Calmette-Guérin as a foreign immune stimulus.39 Forty-one stage I–IIIa subjects, post– curative resection, were enrolled and randomized; 19 received the tumor vaccine. Most subjects were HBV positive. Overall, treated patients had statistically significant improvements in risk of recurrence, time to recurrence, and recurrence-free survival. In this trial, overall survival was also improved at P ⫽ .01. These encouraging data again support the need to enroll subjects with low tumor burden in immunotherapy trials.

AFP-Based Immunotherapy To generate tumor antigen–specific immune responses to HCC, AFP has been targeted by our group and others. To induce an immune response to this self antigen, many strategies have been used, including plasmid DNA, AFP-derived peptides, peptide-pulsed DC,

GASTROENTEROLOGY Vol. 127, No. 5

antigen-engineered DC, prime-boost, and AFP-based prevention strategies, in animal models, in vitro T-cell cultures, and clinical trials. In our studies, we wanted to first determine whether peripheral tolerance to AFP could be broken and whether AFP-specific T cells capable of recognizing HCC could be activated. We first investigated whether the murine T cell repertoire would recognize murine AFP. This was tested in an immunocompetent mouse model in which C57BL/6 mice were immunized with syngeneic DC that were transduced with an adenovirus that expressed murine AFP. These murine AFP–vaccinated mice were then challenged with the syngeneic tumor line, the EL4 lymphoma, stably transfected with murine AFP, or its parental line, or the naturally AFP-expressing HCC line BWIC3. When these mice were challenged with the AFP-expressing tumor, they showed either significantly decreased tumor growth or complete protection, whereas tumors in unimmunized mice or mice immunized with a control adenovirus-transduced DC grew progressively.40 This effect was antigen specific: EL4 parental tumors grew identically in immunized and unimmunized mice, and growth was accompanied by the expansion of AFPspecific splenocytes that secreted IFN-␥ and killed AFPexpressing targets. The effects were dependent on both CD4⫹ and CD8⫹ T cells. These antitumor effects were also seen when the immunized mice were challenged with the AFP-expressing murine HCC line, BWIC3. Similar but weaker effects were seen with plasmid AFPbased immunization.40,41 Notably, AFP levels are higher in normal adult mice than humans, so breaking tolerance in mice may be more difficult than in healthy adult humans. We next investigated whether the human T-cell repertoire would respond to the human AFP self antigen and what peptide epitopes from this molecule were presented to autologous T cells. To find candidate AFP epitopes, the amino acid sequence was screened for peptides that might be restricted by the common HLA type, HLAA2.1, which is expressed by 30%–50% of white subjects, and is common in other ethnicities as well. A general motif of preferred residues is present in positions 2 and 9 (hydrophobic residues L, V, I, and M)42 of peptides that bind to HLA A2.1. An analysis was performed of the entire AFP protein sequence searching for peptides with at least 1 preferred anchor residue and not more than 1 destabilizing residue. Seventy-four peptides were identified, synthesized, and screened by a variety of strategies using both in vitro human T-cell cultures and HLAA2.1/Kb transgenic mice. A total of 4 immunodominant and 10 subdominant epitopes were identified.43,44

November Supplement 2004

Having identified these AFP-derived peptides that might serve as targets for immunotherapy patients with HCC that expresses AFP, a pilot clinical trial was initiated in which patients were to receive the 4 immunodominant peptides emulsified in Montanide ISA-51 adjuvant.22 Patient blood samples were tested before, during, and after each vaccination for the presence of AFP-specific cells using MHC tetramer and IFN-␥ ELISPOT assays. This was a phase I, dose-escalation trial in which patients received either 100 ␮g, 500 ␮g, or 1 mg of each of the 4 peptides. The patients enrolled had received multiple chemotherapies, had presented with stage IVa and IVb disease, and had average pretreatment serum AFP levels of 25,517 ng/mL (range, 162–105,617 ng/mL). The cumulative MHC tetramer data for the first 6 patients treated in this peptide trial showed that 5 patients had expansion of AFP peptide-specific T cells to 1, 2, or 3 of the peptides (no tetramer to the AFP542 peptide was available because of lack of stability of the molecule). Pre-existing AFP-reactive T cells were detected in several subjects for 1–3 of the peptide epitopes, indicating some level of spontaneous immune response to this antigen. Importantly, the tetramer assay enumerates these cells in peripheral blood, but does not provide assessment of T-cell function. The cumulative IFN-␥ ELISPOT results for the 6 patients showed development of functional cytokine responses to all 4 peptides after vaccination. Most of the pretreatment AFP-specific cells detected by MHC tetramer assays did not synthesize IFN-␥ until after vaccination. Many of these responses were transient in the peripheral blood, similar to data from our group and others in melanoma peptide trials. The fate of these T cells is unknown; they may traffic to tumor deposits, they may undergo apoptosis. In these advanced-stage HCC patients, no clinical responses were detected, nor did serum AFP levels decrease. The overall survival posttreatment ranged from 2 to 17 months. The immunologic responses in these first 6 patients allowed us to show that AFP peptide epitopes were immunogenic in vivo and were able to stimulate antigen-specific T cells in HCC patients with very high serum levels of AFP. Because murine models and in vitro T-cell cultures suggest that peptides on DC have superior immunogenicity, the follow-up trial is using AFP peptide-pulsed DC. Enrollment in this study is ongoing. One issue in generating DC from the peripheral blood of patients with HCC is that the phenotype of the DC is quite variable. The AFP peptide-pulsed DC are required to be HLA-DR– and B7-2–positive, indicating an APC phenotype for use as a vaccine. Unlike the DC of mela-

IMMUNOTHERAPY OF HCC

S239

noma patients, DC from HCC patients were more variable for phenotypic markers and, for unknown reasons, cells from some patients did not generate DC with the GM-CSF and IL-4 protocol that was successful in most patients. The immunologic responses of the first 8 fully treated and evaluated patients to date (as assessed by IFN-␥ ELISPOT assay) indicate that the peptide-pulsed DC vaccine induced an increase in frequency of functional peptide-specific cells to 3 of the 4 peptides in each patient, showing successful vaccination. Pre-existing AFP-specific cells were detected by MHC tetramer assay, but few made IFN-␥. These results indicate that there is a low level of spontaneous AFP-specific immunologic reactivity in these subjects. Using adenovirally transduced DC as a vaccine may be a more potent strategy for initiation and expansion of AFP-specific effector cells. When compared in transgenic mice and in in vitro human T-cell cultures, adenovirustransduced DC are superior to peptide-pulsed DC. This could be because of long-term antigen expression driven from the vector, presentation of multiple dominant and subdominant epitopes (which may activate a broader range of T cells, including subdominant epitopes, which may be of higher avidity), MHC class II presentation to antigen-specific CD4 helper T cells (which would have a positive effect on CD8⫹ effector function), or the positive biological effects of transduction on the DC.45 A reality that must be considered is that most HCC patients are in parts of the world where an expensive, individual, cell-based vaccine is not feasible. Therefore, cell-free strategies of immunization have been investigated by our group and others. In murine models, priming with plasmids expressing both AFP antigen and GM-CSF adjuvant and subsequent boosting with adenovirus AFP yields superior antitumor protection than either plasmid DNA or adenovirus alone, giving response levels that approach those of adenovirus-transduced DC.46 Strategies relying on DNA immunization alone to activate an immune response to self antigens have generally yielded weak results. Nevertheless, the use of DNA vaccination (in particular, with multiple booster injections) is a hypothetical means of expanding high-avidity T cells. The boost with an immunogenic vector (such as adenovirus or a pox virus) expressing the same tumor antigen could then act as a secondary stimulus to expand these cells. Another area in which DNA-based immunization may make an impact is in prevention of HCC in predisposed populations, such as those already infected with HBV or HCV. This setting may be ideal for activating tumor-specific T cells and inducing immunologic

S240

LISA H. BUTTERFIELD

memory before a tumor can exert negative effects on general immune function. This strategy would also allow for vaccination in a more immunogenic way before HCC could present tumor antigens in an immunosuppressive milieu, potentially inactivating the T cells capable of reacting to tumor.

Conclusion The promise of immunotherapy is the specific and systemic elimination of tumor, based on the expression of specific proteins by the tumor. To date, several different strategies for immune activation have shown biological activity in HCC subjects, and a subset of those have shown antitumor efficacy. The published data show that adoptive transfer of activated effector cells can impact the recurrence and survival of HCC subjects. The optimal time for use of immune-based therapies is, not unexpectedly, postresection in the adjuvant setting.

References 1. Luo G, Huang S, Xie X, Stockert E, Chen YT, Kubuschok B, Pfreundschuh M. Expression of cancer-testis genes in human hepatocellular carcinomas. Cancer Immun 2002;2:11. 2. Chen HS, Qin LL, Cong X, Wang Y, Fei R, Jiang D, Cao J, Su JJ, Wei L, Chen WF. [Expression of tumor-specific cancer/testis antigens in hepatocellular carcinoma]. Zhonghua Gan Zang Bing Za Zhi 2003;11:145–148. 3. Yamashita N, Ishibashi H, Hayashida K, Kudo J, Takenaka K, Itoh K, Niho Y. High frequency of the MAGE-1 gene expression in hepatocellular carcinoma. Hepatology 1996;24:1437–1440. 4. Kariyama K, Higashi T, Kobayashi Y, Nouso K, Nakatsukasa H, Yamano T, Ishizaki M, Kaneyoshi T, Toshikuni N, Ohnishi T, Fujiwara K, Nakayama E, Terracciano L, Spagnoli GC, Tsuji T. Expression of MAGE-1 and -3 genes and gene products in human hepatocellular carcinoma. Br J Cancer 1999;81:1080 –1087. 5. Tahara K, Mori M, Sadanaga N, Sakamoto Y, Kitano S, Makuuchi M. Expression of the MAGE gene family in human hepatocellular carcinoma. Cancer 1999;85:1234 –1240. 6. Chen CH, Chen GJ, Lee HS, Huang GT, Yang PM, Tsai LJ, Chen DS, Sheu JC. Expressions of cancer-testis antigens in human hepatocellular carcinomas. Cancer Lett 2001;164:189 –195. 7. Mou DC, Cai SL, Peng JR, Wang Y, Chen HS, Pang XW, Leng XS, Chen WF. Evaluation of MAGE-1 and MAGE-3 as tumour-specific markers to detect blood dissemination of hepatocellular carcinoma cells. Br J Cancer 2002;86:110 –116. 8. Zerbini A, Pilli M, Soliani P, Ziegler S, Pelosi G, Orlandini A, Cavallo C, Uggeri J, Scandroglio R, Crafa P, Spagnoli GC, Ferrari C, Missale G. Ex vivo characterization of tumor-derived melanoma antigen encoding gene-specific CD8⫹ cells in patients with hepatocellular carcinoma. J Hepatol 2004;40:102–109. 9. Iizuka N, Oka M, Yamada-Okabe H, Mori N, Tamesa T, Okada T, Takemoto N, Tangoku A, Hamada K, Nakayama H, Miyamoto T, Uchimura S, Hamamoto Y. Comparison of gene expression profiles between hepatitis B virus– and hepatitis C virus–infected hepatocellular carcinoma by oligonucleotide microarray data on the basis of a supervised learning method. Cancer Res 2002; 62:3939 –3944. 10. Wang Y, Han KJ, Pang XW, Vaughan HA, Qu W, Dong XY, Peng JR, Zhao HT, Rui JA, Leng XS, Cebon J, Burgess AW, Chen WF. Large scale identification of human hepatocellular carcinoma–associated antigens by autoantibodies. J Immunol 2002;169:1102–1109.

GASTROENTEROLOGY Vol. 127, No. 5

11. Kato T, Satoh S, Okabe H, Kitahara O, Ono K, Kihara C, Tanaka T, Tsunoda T, Yamaoka Y, Nakamura Y, Furukawa Y. Isolation of a novel human gene, MARKL1, homologous to MARK3 and its involvement in hepatocellular carcinogenesis. Neoplasia 2001;3:4 –9. 12. Okabe H, Satoh S, Kato T, Kitahara O, Yanagawa R, Yamaoka Y, Tsunoda T, Furukawa Y, Nakamura Y. Genome-wide analysis of gene expression in human hepatocellular carcinomas using cDNA microarray: identification of genes involved in viral carcinogenesis and tumor progression. Cancer Res 2001;61:2129 –2137. 13. Cheung ST, Chen X, Guan XY, Wong SY, Tai LS, Ng IO, So S, Fan ST. Identify metastasis-associated genes in hepatocellular carcinoma through clonality delineation for multinodular tumor. Cancer Res 2002;62:4711– 4721. 14. Habib NA, Ding SF, el-Masry R, Mitry RR, Honda K, Michail NE, Dalla Serra G, Izzi G, Greco L, Bassyouni M, el-Toukhy M, AbdelGaffar Y. Preliminary report: the short-term effects of direct p53 DNA injection in primary hepatocellular carcinomas. Cancer Detect Prev 1996;20:103–107. 15. Lu SY, Sui YF, Li ZS, Ye J, Dong HL, Qu P, Zhang XM, Wang WY, Li YS. Superantigen-SEA gene modified tumor vaccine for hepatocellular carcinoma: an in vitro study. World J Gastroenterol 2004;10:53–57. 16. Schueller G, Stift A, Friedl J, Dubsky P, Bachleitner-Hofmann T, Benkoe T, Jakesz R, Gnant M. Hyperthermia improves cellular immune response to human hepatocellular carcinoma subsequent to co-culture with tumor lysate pulsed dendritic cells. Int J Oncol 2003;22:1397–1402. 17. Banchereau J, Steinman RM. Dendritic cells and the control of immunity. Nature 1998;392:245–252. 18. Beckebaum S, Cicinnati VR, Dworacki G, Muller-Berghaus J, Stolz D, Harnaha J, Whiteside TL, Thomson AW, Lu L, Fung JJ, Bonham CA. Reduction in the circulating pDC1/pDC2 ratio and impaired function of ex vivo– generated DC1 in chronic hepatitis B infection. Clin Immunol 2002;104:138 –150. 19. Ninomiya T, Akbar SM, Masumoto T, Horiike N, Onji M. Dendritic cells with immature phenotype and defective function in the peripheral blood from patients with hepatocellular carcinoma. J Hepatol 1999;31:323–331. 20. Takagi S, Chen K, Schwarz R, Iwatsuki S, Herberman RB, Whiteside TL. Functional and phenotypic analysis of tumor-infiltrating lymphocytes isolated from human primary and metastatic liver tumors and cultured in recombinant interleukin-2. Cancer 1989; 63:102–111. 21. Hanke P, Rabe C, Serwe M, Bohm S, Pagenstecher C, Sauerbruch T, Caselmann WH. Cirrhotic patients with or without hepatocellular carcinoma harbour AFP-specific T-lymphocytes that can be activated in vitro by human alpha-fetoprotein. Scand J Gastroenterol 2002;37:949 –955. 22. Butterfield LH, Ribas A, Meng WS, Dissette VB, Amarnani S, Vu HT, Seja E, Todd K, Glaspy JA, McBride WH, Economou JS. T-cell responses to HLA-A2.1 immunodominant peptides derived from alpha-fetoprotein in patients with hepatocellular cancer. Clin Cancer Res 2003;9:5902–5908. 23. Yanagi K, Nagayama Y, Nakao K, Saeki A, Matsumoto K, Ichikawa T, Ishikawa H, Hamasaki K, Ishii N, Eguchi K. Immunogene therapy with adenoviruses expressing fms-like tyrosine kinase 3 ligand and CD40 ligand for mouse hepatoma cells in vivo. Int J Oncol 2003;22:345–351. 24. Arthur JF, Butterfield LH, Roth MD, Bui LA, Kiertscher SM, Lau R, Dubinett S, Glaspy J, McBride WH, Economou JS. A comparison of gene transfer methods in human dendritic cells. Cancer Gene Ther 1997;4:17–25. 25. Guo Y, Wu M, Chen H, Wang X, Liu G, Li G, Ma J, Sy MS. Effective tumor vaccine generated by fusion of hepatoma cells with activated B cells. Science 1994;263:518 –520. 26. Homma S, Toda G, Gong J, Kufe D, Ohno T. Preventive antitumor activity against hepatocellular carcinoma (HCC) induced by im-

November Supplement 2004

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

munization with fusions of dendritic cells and HCC cells in mice. J Gastroenterol 2001;36:764 –771. Lee WC, Wang HC, Jeng LB, Chiang YJ, Lia CR, Huang PF, Chen MF, Qian S, Lu L. Effective treatment of small murine hepatocellular carcinoma by dendritic cells. Hepatology 2001;34:896 –905. Tatsumi T, Takehara T, Kanto T, Miyagi T, Kuzushita N, Sugimoto Y, Jinushi M, Kasahara A, Sasaki Y, Hori M, Hayashi N. Administration of interleukin-12 enhances the therapeutic efficacy of dendritic cell-based tumor vaccines in mouse hepatocellular carcinoma. Cancer Res 2001;61:7563–7567. Lygidakis NJ, Kosmidis P, Ziras N, Parissis J, Kyparidou E. Combined transarterial targeting locoregional immunotherapy-chemotherapy for patients with unresectable hepatocellular carcinoma: a new alternative for an old problem. J Interferon Cytokine Res 1995;15:467– 472. Reinisch W, Holub M, Katz A, Herneth A, Lichtenberger C, SchonigerHekele M, Waldhoer T, Oberhuber G, Ferenci P, Gangl A, Mueller C. Prospective pilot study of recombinant granulocyte-macrophage colony-stimulating factor and interferon-gamma in patients with inoperable hepatocellular carcinoma. J Immunother 2002;25:489 – 499. Une Y, Kawata A,Uchino J. [Adopted immunochemotherapy using IL-2 and spleen LAK cell—randomized study]. Nippon Geka Gakkai Zasshi 1991;92:1330 –1333. Kawata A, Une Y, Hosokawa M, Wakizaka Y, Namieno T, Uchino J, Kobayashi H. Adjuvant chemoimmunotherapy for hepatocellular carcinoma patients. Adriamycin, interleukin-2, and lymphokine-activated killer cells versus Adriamycin alone. Am J Clin Oncol 1995;18:257–262. Takayama T, Makuuchi M, Sekine T, Terui S, Shiraiwa H, Kosuge T, Yamazaki S, Hasegawa H, Suzuki K, Yamagata M, Fujii M, Tanaka T, Kakizoe T. Distribution and therapeutic effect of intraarterially transferred tumor-infiltrating lymphocytes in hepatic malignancies. A preliminary report. Cancer 1991;68:2391–2396. Wang Y, Chen H, Wu M, Bao J, Cong W, Wang H. Postoperative immunotherapy for patients with hepatocarcinoma using tumorinfiltrating lymphocytes. Chin Med J (Engl) 1997;110:114 –117. Takayama T, Sekine T, Makuuchi M, Yamasaki S, Kosuge T, Yamamoto J, Shimada K, Sakamoto M, Hirohashi S, Ohashi Y, Kakizoe T. Adoptive immunotherapy to lower postsurgical recurrence rates of hepatocellular carcinoma: a randomised trial. Lancet 2000;356:802– 807. Ladhams A, Schmidt C, Sing G, Butterworth L, Fielding G, Tesar P, Strong R, Leggett B, Powell L, Maddern G, Ellem K, Cooksley G. Treatment of non-resectable hepatocellular carcinoma with autologous tumor-pulsed dendritic cells. J Gastroenterol Hepatol 2002;17:889 – 896. Iwashita Y, Tahara K, Goto S, Sasaki A, Kai S, Seike M, Chen CL, Kawano K, Kitano S. A phase I study of autologous dendritic

IMMUNOTHERAPY OF HCC

38.

39.

40.

41.

42. 43.

44.

45.

46.

S241

cell-based immunotherapy for patients with unresectable primary liver cancer. Cancer Immunol Immunother 2003;52:155–161. Stift A, Friedl J, Dubsky P, Bachleitner-Hofmann T, Schueller G, Zontsich T, Benkoe T, Radelbauer K, Brostjan C, Jakesz R, Gnant M. Dendritic cell-based vaccination in solid cancer. J Clin Oncol 2003;21:135–142. Kuang M, Peng BG, Lu MD, Liang LJ, Huang JF, He Q, Hua YP, Totsuka S, Liu SQ, Leong KW, Ohno T. Phase II randomized trial of autologous formalin-fixed tumor vaccine for postsurgical recurrence of hepatocellular carcinoma. Clin Cancer Res 2004;10: 1574 –1579. Vollmer CM Jr, Eilber FC, Butterfield LH, Ribas A, Dissette VB, Koh A, Montejo LD, Lee MC, Andrews KJ, McBride WH, Glaspy JA, Economou JS. Alpha-fetoprotein-specific genetic immunotherapy for hepatocellular carcinoma. Cancer Res 1999;59:3064 –3067. Grimm CF, Ortmann D, Mohr L, Michalak S, Krohne TU, Meckel S, Eislele S, Encke J, Blum HE, Geissler M. Mouse alpha-fetoprotein-specific DNA-based immunotherapy of hepatocellular carcinoma leads to tumor regression in mice. Gastroenterology 2000; 119:1104 –1112. Rammensee HG, Falk K, Rotzschke O. MHC molecules as peptide receptors. Curr Opin Immunol 1993;5:35– 44. Butterfield LH, Koh A, Meng W, Vollmer CM, Ribas A, Dissette V, Lee E, Glaspy JA, McBride WH, Economou JS. Generation of human T-cell responses to an HLA-A2.1–restricted peptide epitope derived from alpha-fetoprotein. Cancer Res 1999;59: 3134 –3142. Butterfield LH, Meng WS, Koh A, Vollmer CM, Ribas A, Dissette VB, Faull K, Glaspy JA, McBride WH, Economou JS. T cell responses to HLA-A*0201–restricted peptides derived from human alpha fetoprotein. J Immunol 2001;166:5300 –5308. Schumacher LY, Ribas A, Dissette VB, Glaspy JA, McBride WH, Economou JS, Butterfield LH. Human dendritic cell maturation by adenovirus transduction enhances tumor antigen–specific T cell responses. J Immunother 2004;27:191–200. Meng WS, Butterfield LH, Ribas A, Dissette VB, Heller JB, Miranda GA, Glaspy JA, McBride WH, Economou JS. Alpha-fetoprotein-specific tumor immunity induced by plasmid prime-adenovirus boost genetic vaccination. Cancer Res 2001;61:8782– 8786.

Address requests for reprints to: Lisa H. Butterfield, PhD, Departments of Medicine, Surgery and Immunology, University of Pittsburgh, Hillman Cancer Center, Research Pavilion, Room 1.19, 5117 Centre Avenue, Pittsburgh, Pennsylvania 15213. e-mail: butterfi[email protected]; fax: (412) 623-1415. L.H.B. is supported in part by a Scientist Development Grant from the American Heart Association (0330102N), NIH/NCI (RO1 CA104524), the University of Pittsburgh Cancer Institute, and the Henry L. Hillman Foundation.