| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Review |
Markey Cancer Center, Lexington, Kentucky 40536-0093
Colorectal cancer represents 10% of all cancers and 10% of cancer deaths in the United States with 
130,000 new cases diagnosed annually (1)
. There have been important advances in our understanding of the biology and genetics of this disease, and if diagnosed early, colorectal cancer is highly curable. Patients with Dukes’ A colon cancer that invades only the submucosa and Dukes’ B1 disease with invasion into the muscularis propria are typically cured with surgery, and no additional therapy is recommended (2)
. Patients with Dukes’ B2 colon cancer with invasion through the muscularis propria or through the serosa, but without lymph node involvement, and Dukes’ C colon cancer which includes lymph node involvement, are treated with 5-FU2
with levamisole or leucovorin with 10–15% improved survival (3, 4, 5)
. Adjuvant immunotherapy with agents such as BCG or methanol extraction residue of BCG have not improved upon these results (discussed below). More recently, a study from Germany in Dukes’ C patients who were randomized to a monoclonal antibody, designated 171A, demonstrated improved survival by 30% over patients on the control arm (6)
. The opportunity to improve results in Dukes’ B2, C, and D patients with an immunotherapeutic approach in addition to chemotherapy is excellent.
Immune approaches to the therapy of colorectal cancer have substantially evolved over the past years, from treating patients with nonspecific immune stimulants to a focus on the use of TAAs either by passive immune therapy with antibodies targeted directly to tumor cells or by active immune therapy via vaccination with tumor cells, tumor cell lysates, peptides, carbohydrates, gene constructs encoding proteins, or anti-idiotype antibodies that mimic TAAs.
Passive Therapy with Unlabeled Monoclonal Antibodies
Numerous clinical studies using unlabeled (naked) murine monoclonal antibodies targeted to a variety of TAAs have been reported. In some cases, the antibodies have been combined with cytokines such as IL-2 (7) , M-CSF (8) , or GM-CSF (9) .
A murine monoclonal antibody that binds to the majority of colorectal cancers, 17-1A (IgG2A), has been studied in colorectal cancer patients with metastatic disease as well as in the postsurgical adjuvant setting. 17-1A targets to the nonsecreted Mr 40,000 glycoprotein CD17-1A antigen, which is overexpressed by most epithelial tumors. Rare clinical responses have been described after the i.v. infusion of unlabeled 17-1A antibody in patients with metastatic disease (10, 11, 12, 13) . Mechanisms remain unknown, but it has been hypothesized that responses are secondary to activation of the "idiotypic network" and/or ADCC. GM-CSF enhances ADCC activity in vitro and was combined with 17-1A in 20 patients with metastatic colorectal cancer (9) . Two complete responses and one minor response were reported. In another study, 24 patients with metastatic colorectal cancer were treated with i.v. infusions of 17-1A antibody (14) . Five clinical responses were correlated with proliferative T-cell responses that recognized the human anti-idiotypic antibody to 17-1A, suggesting that the responses were related to activation of the idiotypic network (discussed below). The most impressive and important clinical trial with 17-1A was reported in 189 patients with Dukes’ C colorectal cancer who were randomly assigned to an observation regimen or to postoperative treatment with 500 mg of 17-1A antibody, followed by four 100-mg monthly infusions (6, 7, 8, 9, 10, 11, 12, 13, 14, 15) . After a median follow-up of 7 years, the antibody-treated group had a reduced mortality of 32% and decreased recurrence rate of 23% (6) . Confirmatory studies are presently under way in the United States and Europe.
Murine monoclonal antibody L6 is an IgG2A antibody that binds to a poorly characterized antigen on adenocarcinoma cells (16 , 17) and is expressed on >90% of breast, colorectal, and non-small cell lung cancer. This antibody has been shown to mediate ADCC and complement-mediated cellular cytotoxicity. Fifteen patients with refractory breast, lung, or colorectal cancer were treated with 200 mg/m2 of L6 i.v. daily on days 1–7, followed by a 1-week rest period, then IL-2 was given at either 2, 3, or 4.5 x 106 units/m2 daily for 4 days (7) . They reported one partial response in a patient with colorectal cancer.
Monoclonal antibody D612 is a murine IgG2A antibody directed at a membrane glycoprotein expressed on gastrointestinal tumors (18 , 19) . This antibody was also shown to mediate ADCC, and its effect was enhanced in vitro when combined with M-CSF (20) . Phase I and II trials of D612 in patients with metastatic colorectal cancer have been reported (8 , 21) . In a Phase II trial, human M-CSF was given daily for 14 days, and D612 was given at 40 mg/m2 on days 4, 7, and 11. Several biological effects were noted, but without antitumor responses (8) .
In summary, although rare responses have been reported in advanced colorectal cancer patients treated with nascent murine monoclonal antibodies with or without other biological response modifiers, the results, overall, have not been promising. Murine antibodies are typically weak mediators of human effector functions and stimulate neutralizing human anti-mouse antibody responses in patients. Chimerizing or humanizing these antibodies may prove beneficial (discussed below). The positive results reported in the postsurgical adjuvant setting in patients with Dukes’ C colorectal cancer require confirmation (6 , 15) .
Radioimmunodiagnostics and Radioimmunotherapy
Two radioimmunodiagnostics are presently commercially available for colorectal cancer. The first is 111In-labeled B72.3 murine monoclonal antibody (22, 23, 24, 25) . This radioimmunoconjugate was demonstrated to improve the detection of intraabdominal sites of tumor over computed axial tomography of the abdomen and pelvis. In one study, the monoclonal antibody imaging identified tumors in 27% of patients in which computed axial tomography scan failed to identify sites of tumor. Liver metastasis required identification with single photon emission tomography because of a high radioactive background in the liver. The second radioimmunoconjugate commercially available for imaging is the Fab' antibody fragment of an anti-CEA, designated IMMU-4 labeled with 99mTc (26) . Similar to 111In-B72.3, 99mTc-IMMU-4 Fab' was superior to conventional diagnostic modalities in the extrahepatic abdomen and pelvis and complemented conventional modalities in the liver. The positive predictive value was significantly higher when both modalities were positive (98%) compared with each alone (68–70%). Interestingly, only 2 patients among 210 scanned developed human anti-mouse antibodies to 99mTc-IMMU-4 Fab' after a single injection, and none of 19 after two injections.
F19 is a murine monoclonal antibody that recognizes the fibroblast activation protein which is highly expressed on epithelial tumors. 131I-F19 demonstrated tumor images on planar and single photon emission tomography scans in 15 of 17 patients with hepatic metastasis, portal lymph nodes, and/or pelvic disease (27) . Lesions as small as 1 cm in diameter were visualized.
Although radioimmunodiagnostics have filled a niche in detection and disease assessment, the ultimate goal is radioimmunotherapy. This topic has been reviewed recently and is discussed briefly (28, 29, 30)
. There are a variety of radioisotopes potentially suitable for radioimmunotherapy (Table 1
; Ref. 29
). Gamma rays emitted by radionuclides such as 131I exit the body and allow for the use of external scintigraphic imaging to determine the biodistribution of the isotope. Beta particles, in contrast, travel only a few millimeters to centimeters in tissue and deposit their energy in the vicinity of the point of decay. Beta emissions from isotopes such as 90Y, 131I, and 67Cu that bind to antigen-positive cells can also kill nearby antigen-negative cells via the "crossfire effect." 131I is the gold standard for radioimmunotherapy because of its relatively straightforward radiochemistry, availability, and low cost. The major disadvantages are dehalogenation with the associated radiation safety concerns and enzymatic elimination from tumors before 131I can deposit all of its energy. 125I is unique in that its tissue penetration is only .02 mm as it degrades via electron capture. It has been typically used for radioimmunotherapy in combination with antibodies that modulate internally. Its disadvantages are a 60-day half life and lack of imaging capability. 90Y is another popular therapeutic isotope for radioimmunotherapy because of its high energy ß penetration, metal chemistry, and 3-day half life. The disadvantage of 90Y is very poor imaging, but this can be countered by using 111In-labeled antibody as a tracer for 90Y-labeled antibody. This topic has been reviewed and will be discussed briefly (28, 29, 30)
.
|
Numerous approaches have been used to increase the concentration of radiolabeled antibody in tumor. One approach is increasing the quantity of infused antibody. However, this may increase the concentration of the radiolabeled antibody in the tumor, but it also increases deposition in normal tissues. Another approach is the use of bifunctional or bispecific antibodies where there is a linkage of one antibody or fragment that binds to the tumor with another antibody or fragment that binds to a chelated hapten (35) . After injection of the unlabeled bifunctional antibody, the chelated isotope that reacts with the second antibody arm is injected following sufficient time to clear the bifunctional antibody from normal tissues. Another targeting strategy involves high affinity noncovalent binding between avidin and biotin. Twenty patients were injected with biotinylated anti-CEA monoclonal antibody i.v., and then 3 h later, cold avidin was administered to remove the nontumor bound antibodies (36) . A third injection of biotin derivative labeled with 111In was given 48 h later. Tumors were imaged within 3 h of the radioconjugate administration with high tumor:background ratios.
Neutralizing human anti-mouse antibody develops in nearly all immune-competent patients receiving murine monoclonal antibodies. This can be reduced or entirely eliminated by using chimeric or humanized antibodies. One approach is to replace the constant region of the mouse antibody with the human constant region by genetic engineering, resulting in a mouse human chimera (37) . Another approach is to graft the murine complementarity-determining regions onto human antibody (38 , 39) . Such approaches should decrease the immunogenicity of antibody in patients, without having a negative effect on the biological activity of the antibody. In one clinical trial, 131I labeled chimeric B72.3 was injected into patients with metastatic colorectal cancer, and the serum half-life was shown to be 6–8 times as long as the murine B72.3 (40) . However, it was determined that the chimeric B72.3 still generated immunity in many of the patients, often targeted to the "hinge" region of the mouse and human immunoglobulin. A potential therapeutic advantage to chimeric and humanized antibodies, is that the human Fc portion of the immunoglobulin binds to human effector cells better than murine Fc, thus mediating a more potent ADCC.
Vaccines
Specific active immunotherapy differs from nonspecific immune-based therapies such as BCG in that the goal is not general but rather specific activation of the immune system to eliminate tumor cells and not affect surrounding normal tissue. Theoretically, specific immunotherapy through vaccines activates a unique lymphocyte (B- and/or T-cell) response, which has an immediate antitumor effect as well as a memory response against future tumor challenge. The first and most obvious type of vaccines are autologous or allogeneic tumor cell preparations. Alternatively, membrane preparations from tumor cells have been used. In either instance, these vaccines have been combined with a variety of cytokines. More recently, with advances in molecular biological approaches, gene-modified tumor cells expressing antigens designed to increase immunogenicity or gene modified to secrete cytokines have been a valuable tool for vaccination. In addition, increase in our knowledge of TAA biology has led to the use of purified TAAs, DNA-encoding protein antigens, and/or protein-derived peptides. All of these approaches are being tested in the clinic.
Mechanistically, the ultimate aim of a vaccine is to activate a component of the immune system such as antibodies or lymphocytes against TAAs presented by the tumor. Antibodies must recognize antigens in the native protein state at the cell surface. Once bound, these molecules can mediate ADCC or complement-mediated cytotoxicity. T lymphocytes, on the other hand, recognize proteins as fragments or peptides of varying size, presented in the context of MHC antigens on the surface of the cells being recognized (41, 42, 43) . The proteins from which the peptides are derived may be cell surface or cytoplasmic proteins (44 , 45) . MHC antigens are highly polymorphic, and different alleles have distinct peptide binding capabilities. The sequencing of peptides derived from MHC molecules has led to the discovery of allele-specific motifs that correspond to anchor residues that fit into specific pockets on MHC class I or II molecules (46 , 47) .
There are two types of T lymphocytes, helper and cytotoxic, which recognize antigens through a specific TCR composed of both 
and ß subunits arranged in close conjunction to the CD3 molecule, which is responsible for signaling (Fig. 2)
. CD4 helper T cells secrete cytokines and lymphokines that enhance immunoglobulin production as well as activate CD8 CTLs. CD4 helper T cells are activated by binding via their TCR to class II molecules, which contain 14–25 amino acid (mer) peptides in their antigen binding cleft (48, 49, 50)
. Extracellular proteins are endocytosed and degraded (exogenous processing) into 14–25 mer peptides in endocytic compartments (acidified endosomes) and bind to newly synthesized MHC class II molecules. The MHC peptide complex is transported to the cell membrane, where it can be recognized by specific CD4 helper T cells. In most cases, the MHC class II antigen-containing peptide is presented to the CD4 helper T cells by a specialized cell called an APC. More specifically, a variety of cells are capable of processing and presenting exogenous antigen including B cells, monocytes, macrophages, and the bone marrow-derived DC. DCs are the most efficient APCs and express high levels of MHC class I and II molecules, costimulatory molecules such as CD80 and CD86, and specific markers such as CD83. After antigen uptake, DCs migrate peripherally to lymph nodes, where antigen presentation to CD4 helper T cells takes place (51
, 52)
.
|
CD8-positive CTLs are activated in most cases by peptides derived from intracellular proteins that are cleaved to 9–10 mer peptides in the cytosol of tumor cells or APCs by proteosomes. The peptides are then transported via specialized transporter molecules called TAP proteins to the endoplasmic reticulum, where they become associated with newly synthesized MHC class I molecules (54) . The complex is then transported via the Golgi apparatus to the cell surface membrane, where the complex is recognized by CD8 cytotoxic T cells via a specific TCR. Any endogenously processed protein can be presented to the immune system in this way. Several reports suggest that a subset of APCs can present exogenously processed proteins on MHC class I molecules to CTLs (55, 56, 57, 58, 59) .
Tumor Cells
The most straightforward means of immunization is the use of whole tumor cell preparations (either autologous or allogeneic tumor cells). The advantage to this approach is that all potential TAAs are presented to the immune system for processing and presentations to the appropriate T-cell precursors. The difficulty with this approach lies in the availability of fresh autologous tumor material and in the sparcity of well-characterized, long-term tumor cell lines, which are HLA typed and express high levels of MHC antigens. Regardless, whole tumor cell vaccines have been an area of intense interest.
In a prospective randomized trial, 98 patients with Dukes’ stage B2 through stage C3 colon or rectal cancer were treated by resection alone or resection plus active specific immunotherapy (60) . This study design was based on a highly successful guinea pig model (61, 62, 63, 64, 65) . Vaccine administration began 4–5 weeks after tumor resection, beginning with one intradermal vaccination per week for 2 weeks consisting of 107 viable irradiated autologous tumor cells and 107 viable BCG organisms. In the third week, patients received one vaccination of 107 irradiated tumor cells alone. Overall and disease-free survival did not show a statistically significant difference for the 80 eligible patients. However, the rectal cancer patients received postimmunotherapy radiation. When these patients were separated from the colon cancer patients, with a median follow-up of 93 months, a significant improvement in overall and disease-free survival was seen in the colon cancer patients who received active specific immunotherapy. Correlations with immune responses were not reported.
In another study, freshly thawed autologous colon cancer cells were inactivated with radiation and infected with Newcastle disease virus or mixed with BCG (66) . All patients had resected Dukes’ B or C colorectal cancer. The 2-year survival rate for patients treated with cells containing Newcastle disease virus was 98% versus 67% for those treated with cells mixed with BCG. Delayed-type hypersensitivity skin reactions to Newcastle virus-infected cells were reported in 68% of patients studied.
Genetically Modified Tumor Cells
Another approach to tumor cell vaccines is the introduction into tumor cells of foreign genes encoding cytokines such as IL-2, GM-CSF, tumor necrosis factor or IFN-
(67
, 68)
. Alternatively, molecules designed to increase tumor cell immunogenicity such as CD80 and CD86 have proven to be very effective in murine models and are showing promise in vitro in allowing the generation of tumor-specific CTLs (69, 70, 71, 72, 73)
. Gene transfer can be accomplished by transfection of plasmid constructs (electroporation, Lipofectamine) or transduction using a viral vehicle such as retroviruses or adenoviruses. Retroviruses have been most widely used for gene transfer into fresh human tumor cells. Retroviral vectors have a high efficiency of gene transfer as well as stable insertion and expression of the protein in the target cell (67
, 74) . However, because retroviral vectors require actively proliferating cells for stable gene transfer, their usefulness in human clinical trials has been hampered. In most cases, this approach is most successful using tumor cell lines because they more readily take up foreign DNA and express the protein product than do fresh tumor cells. Alternative gene delivery has been tested using other viral delivery systems such as adenovirus and poxviruses, where cell division is not a prerequisite for gene transfer; however, specificity of binding of the virus to the target cell becomes an issue. In addition, other possible adverse effects of these viruses include potential adverse effects on antigen presentation through the down-regulation of class I molecules, induction of antiviral responses that may limit subsequent immunization, and the safety concerns inherent in the use of attenuated viruses in human patients. Another option that has been tested for gene transfer is physical gene delivery in which plasmid or "naked" DNA is delivered directly into tumor cells. Liposomes can serve as gene carriers, use of a "gene gun," electroporation, and calcium phosphate-mediated gene transfer are all alternative methodologies that have been evaluated for the physical delivery of genes into tumor cells. The primary problem with a nonviral gene delivery system is gene expression in the transfected cells tends to be transient.
Overall, this approach is most interesting in that the vaccine can have a profound effect on the inflammatory infiltrate. The granulocytes and macrophages that are contained therein serve to begin the rejection and destruction of tumor cells. Macrophages and DC precursors contained in the infiltrate phagocytize tumor debris and begin the presentation to TAA-specific lymphocyte precursors (Fig. 1)
. All of these activities may be enhanced in the presence of the cytokine delivered by the tumor cells. Alternatively, tumor cell genes modified with lymphocyte costimulatory molecules (CD80/86) present TAAs directly to lymphocyte precursors. Ultimately, one looks for a localized antitumor response that, if properly propagated, develops into a potent systemic antitumor immunity.
|
Peptides, Mucins, and Carbohydrates
An alternative to vaccines described above is the use of purified molecules as immunogens. The molecules themselves code for the relevant TAAs. In most instances, these molecules can be manufactured in large quantities, and if delivered properly, can result in a potent antitumor immune response. Whole proteins, as opposed to peptides, can be processed and presented by a wider array of class I and II molecules. An extensive literature is beginning to amass on class I- and II-restricted protein antigens in melanoma (41 , 77, 78, 79, 80, 81, 82, 83, 84, 85) . Unfortunately, a less extensive literature exists for epithelial tumors including colon cancer (86, 87, 88, 89, 90, 91, 92, 93) . It has remained elusive to definitively describe a T cell-mediated immune response in this disease. Nevertheless, candidate proteins are being explored. An advantage to peptide vaccines is that they can be synthetically generated in a reproducible fashion. The major disadvantage of peptides is that they are restricted to a single HLA molecule and are not of themselves very immunogenic (94) . To increase their immunogenicity, peptides may be injected with adjuvants, cytokines, or liposomes or presented on DCs (95, 96, 97, 98, 99, 100, 101, 102, 103) . Whole proteins have the advantage over peptides in that they can be processed for a wider range of MHC class I and II antigens.
Mucins such as MUC1 are heavily glycosylated high molecular weight proteins abundantly expressed on human cancers of epithelial origin (104, 105, 106, 107, 108) . The MUC1 gene is overexpressed and aberrantly glycosylated on a variety of cancers including colorectal cancer. Much of the glycosylation is found within regions of tandemly repeated sequences of 20 amino acids/repeat (108, 109, 110) . Tumors derived from cells of epithelial origin often lose the carbohydrate side chains, exposing the tandemly repeated protein core, resulting in antigenically active epitopes exposed to the cell surface membrane (108) . Numerous mucin-specific antibodies have been generated following immunization of animals with epithelial cells (108 , 111 , 112) . MHC-restricted and unrestricted recognition of mucin by T cells has also been reported (113 , 114) . In one study, patients were vaccinated with a 105-amino acid polypeptide that included five repetitions of the entire conserved tandem repeat of the MUC1 peptide (115) . Sixty-three patients were vaccinated with 100 µg of the mucin peptide mixed with BCG. Two additional vaccinations were given at 3-week intervals. Toxicity included local ulceration at the site of the vaccination. Delayed-type hypersensitivity reactions were evaluated at 48 h, and intense T-cell infiltration was reported in the majority of patients. A limited number of patients had a 2–4-fold increase in mucin-specific CTL precursors in the peripheral blood after vaccination. A 9-mer peptide spanning the MUC1 tandem repeat with an HLA-A11 MHC class I restriction association has been identified. CTLs specific for this peptide have been identified from peripheral blood of HLA-A11 donors (116) .
The "Holy Grail" for successful tumor immunotherapy has been the induction of CTLs rather than the generation or use of antibodies. In a recent study, it was determined that humans immunized with MUC1 produce antibody responses with poor CTL responses. This is due to the fact that humans do not express Gal(1, 3)Gal on their tissues and, therefore, produce natural antibodies against exogenous Gal(1, 3)Gal present in bacteria and food. These antibodies cross-react with MUC1 (117)
, causing the CTL response to switch to an antibody response. Mice do not produce these natural antibodies because they express Gal(1, 3)Gal and thus generate CTLs in response to MUC1. Authors of this approach express disappointment in not being able to generate anti-MUC1 CTLs in patients. In an accompanying editorial, Houghton and Lloyd (118)
take issue with the general tone of negativism by these authors. They point out that the present era of "CTL chauvinism" is largely based on experiments in transplantable tumor models in mice, usually tumors produced by mutagens, which are rejected following CTL responses. They argue that although these models are extremely valuable for our understanding of tumor immunology in general, the direct relevance to slowly progressing cancers in humans is not clear. They point out that vaccines against infectious agents act through antibodies, not CTLs. These antibodies likely prevent blood-borne dissemination to compensate for limited efficacy against infection at tissue sites. They propose that antibodies may play an important role in preventing metastasis, which could be critical in the postsurgical adjuvant setting.
Immunization against tumor-associated carbohydrate antigen has also been attempted. Vaccine studies have been reported using the GM2 disialoganglioside primarily associated with melanoma, sarcoma, and neural-derived tumors (119, 120, 121) . Carbohydrate antigens typically bypass T-cell help for B-cell activation. Investigators have demonstrated that some carbohydrates may activate an alternative T-cell pathway (122, 123, 124, 125) . TF and sTn antigens are blood group-related disaccharides that are O-linked to serine and threonine residues of mucins on epithelial cancers including colorectal cancer (126, 127, 128, 129, 130, 131, 132) . In normal tissues, TF and sTn antigens are restricted to the luminal surface of secretory cells, which is largely inaccessible to the immune system. Similar to the case with MUC-1, altered glycosylation leads to exposure of these core structures in malignant tissues. TF and sTn are poor immunogens because they are carbohydrates and autoantigens. It has also been hypothesized that altered mucins shed by cancer cells induce a T-suppressor lymphocyte response (132) . Postsurgical patients who were disease free but at high risk for recurrence were immunized with synthetic TF and sTn covalently linked to keyhole limpet hemocyanin, without adjuvant or mixed with the adjuvants Detox or QS-21 (133) . The QS-21 mixture was most potent in inducing IgM and IgG titers against the respective synthetic disaccharide epitopes. However, the antibodies only weakly reacted against the natural antigens.
Recombinant Vaccines Expressing CEA
The CEA gene has been sequenced and is part of the human immunoglobulin supergene family located on chromosome 19 (134 , 135) . CEA is highly expressed on colorectal cancer and a variety of other epithelial tumors and is thought to be involved in cell-cell interactions. CEA is considered an adhesion molecule and may play an important role in the metastatic process by mediating attachment of tumor cells to normal cells (136 , 137) . For all of the above reasons, we and others have found CEA to be a very attractive target antigen for immunotherapy. The immunogenic nature of CEA in humans is unclear, and there has been no evidence of naturally occurring cell-mediated responses to CEA in humans. Copresentation of CEA with a strong immunogen such as the vaccinia virus would be a logical approach to induce an anti-CEA response. Vaccinia viruses are highly immunogenic and stimulate both humoral and cellular mediated responses. A recombinant vaccinia virus expressing human CEA (rV-CEA) stimulated T-cell responses in animal species, including nonhuman primates (138, 139, 140) . A variety of CEA peptides selected to conform to human HLA-A2 motifs was established, and one 9-mer peptide designated CAP-1 stimulated T-cell lines from the peripheral blood of patients’ vaccinated with rV-CEA (141) . T-cell lines were demonstrated to lyse HLA-A2-positive colon carcinoma cell lines.
This study was important for a number of reasons: (a) it was the first to demonstrate human CTL responses to specific CEA epitopes; (b) it demonstrated class I HLA-A2-restricted T-cell mediated lysis; and (c) it demonstrated the ability of human tumor cells to endogenously process CEA to present a specific CEA peptide in the context of a MHC for T cell-mediated lysis.
To enhance the induced immune response, low-dose IL-2 was administered in a murine tumor model with rV-CEA (142) . The addition of low-dose IL-2 enhanced immunity and resulted in complete tumor regression in the majority of animals. A DNA plasmid has also been constructed that encodes the full-length cDNA for CEA and can function as a polynucleotide vaccine (143) . After lingual injections in mice, this polynucleotide vaccine generated humoral and/or cellular immune responses specific for CEA. Clinical trials are in progress.
Anti-Idiotype Antibodies
The idiotype network hypothesis of Lindenmann (144)
and Jerne (145)
offers an elegant approach to transforming epitope structures into idiotypic determinants expressed on the surface of antibodies. According to the network concept, immunization with a given TAA will generate production of antibodies against this TAA, which are termed Ab1; Ab1 is then used to generate a series of anti-idiotype antibodies against the Ab1, termed Ab2. Some of these Ab2 molecules can effectively mimic the three-dimensional structure of the TAA identified by the Ab1. These particular antibodies, called Ab2ß, fit into the paratopes of Ab1 and express the internal image of the TAA. The Ab2ß can induce specific immune responses similar to those induced by the original TAA and, therefore, can be used as surrogate TAAs. Immunization with Ab2 can lead to the generation of anti-anti-idiotypic antibodies (Ab3) that recognize the corresponding original tumor-associated antigen identified by Ab1. Because of this Ab1-like reactivity, the Ab3 is also called Ab1' to indicate that it might differ in its other idiotopes from Ab1. The putative immune pathways for anti-idiotype vaccines are presented in Fig. 2
. The anti-idiotype antibody represents an exogenous protein that should be endocytosed by APCs and degraded to 14–25-mer peptides to be presented by class II antigens to activate CD4 helper T cells. Activated Th2 CD4 helper T cells secrete cytokines such as IL-4 that stimulate B cells that have been directly activated by the Ab2 to produce antibody (Ab1') that binds to the original antigen identified by the Ab1. In addition, activation of Th1 CD4 helper T cells secrete cytokines that activate T cells, macrophages, and natural killer cells that directly lyse tumor cells and, in addition, contribute to ADCC. Th1 cytokines such as IL-2 also contribute to the activation of a CD8 cytotoxic T-cell response. This represents a second putative pathway of endocytosed anti-idiotype antibody. The anti-idiotype antibody may be degraded to 9/10-mer peptides to present in the context of class I antigens to activate CD8 cytotoxic T cells (52, 53, 54, 55, 56)
, which are also stimulated by the IL-2 from Th1 CD4 helper T cells.
Several anti-idiotype antibodies that mimic TAAs on colorectal cancer cells have been reported. One such antibody was generated against the murine 17-1A antibody, described previously. After surgery for colorectal cancer, six patients were immunized with this human anti-idiotype antibody that mimicks the GA733-2 antigen (146) . All of the patients developed a long-lasting T-cell immunity against GA733-2, and five mounted a specific IgG antibody response against GA733. Another group, using a rat anti-idiotype antibody generated to the 17-1A antibody, immunized nine colorectal cancer patients with aluminum hydroxide precipitated 17-1A; none of the nine patients developed specific antibodies, although four patients developed delayed-type hypersensitivity (147) . Another group of investigators has developed both murine and human monoclonal anti-idiotype antibodies that mimic the gp72 antigen (148, 149, 150, 151) . They demonstrated delayed-type hypersensitivity reactions when murine anti-idiotype antibody was injected without adjuvant (151) . When the anti-idiotype was linked to keyhole limpet hemocyanin in the presence of Freund’s adjuvant, anti-gp72 antibodies were detected. Using the human equivalent anti-idiotype antibody precipitated in aluminum hydroxide, 9 of 13 patients with advanced colorectal cancer produced blastogenic responses to gp72-expressing tumor cells or produced detectable levels of IL-2 in their plasma (149) . They suggested that survival correlated with immune responses. In another study, with the same human anti-idiotype antibody, six patients with rectal cancer were immunized preoperatively (150) . This study demonstrated significant killing of autologous tumor cells using cryopreserved lymphocytes or lymph node cells from patients 1–2 weeks after immunization.
We have had a major interest in an anti-idiotype antibody designated CeaVac (Ref. 152 ). We generated the CeaVac anti-idiotype murine monoclonal antibody to an antibody designated 8019, which identifies a specific epitope on CEA that is highly restricted to tumor cells and not found on normal tissues. We demonstrated that the CeaVac anti-idiotype antibody functioned as an internal image of CEA by generating anti-idiotypic (Ab3) responses that recognize CEA in mice, rabbits, and monkeys and had a major antitumor effect in a murine tumor model (153) . Among 23 patients with advanced colorectal cancer, 17 generated anti-anti-idiotypic Ab3 responses, and 13 of these responses were proven to be true anti-CEA responses (Ab1'; Refs. 154 and 155 ). The antibody response was polyclonal, and sera from 11 patients mediated ADCC. Ten patients had idiotypic T-cell responses, and five had specific T-cell responses to CEA. None of the patients had objective clinical responses, but overall, median survival for the 23 evaluable patients was 11.3 months, with 44% 1-year survival (95% confidence interval, 23–64%). Toxicity was limited to local swelling and minimal pain. The overall survival of 11.3 months was comparable with other Phase II data, in which advanced colorectal cancer patients were treated with a variety of chemotherapy agents, including irinotecan, and had considerably less toxicity.
Thirty-two patients with resected colorectal cancer were randomized to treatment with 2 mg of aluminum hydroxide-precipitated CeaVac intracutaneously or 2 mg of CeaVac mixed with 100 µg of the QS-21 adjuvant s.c. every other week times four, then monthly until recurrent disease (156)
. Four patients were Dukes’ B2, 11 were Dukes’ C, 8 were completely resected Dukes’ D, and 9 were incompletely resected Dukes’ D. The incompletely resected Dukes’ D were those with positive margins after surgery. Fourteen of the patients received 5-FU-based chemotherapy regimens (11 leucovorin and 3 levamisole) simultaneously with CeaVac. Ten patients have relapsed or demonstrated disease progression at 6–30 months (Table 2)
. Two patients have died at 14 and 20 months. All 32 patients had high titer polyclonal anti-CEA responses (50–300 µg/ml) that mediated ADCC. The predominant Ab3 immunoglobulin was IgG, and the major subclasses were IgG1 and IgG4. All 32 patients generated idiotypic-specific T-cell responses, and 75% were CEA specific. A linear peptide derived from the CDR2 light chain region stimulated a Th1 CD4 proliferative response in vitro (157)
. These data demonstrate that 5-FU-based chemotherapy regimens do not adversely affect the immune response to CeaVac. In addition, high-titer anti-CEA immunoglobulin and Th1 helper cell response can be maintained indefinitely with monthly boosts of 3H1. Injections were well tolerated with only minor local reactions and minimal systemic side effects. Although longer follow-up is required, there appears to be a biological effect on tumor progression, suggested by the 17 patients with resected and incompletely resected Dukes’ D disease who continue on study from 6–29 months. We did not identify a major immunological difference among patients injected with aluminum hydroxide-precipitated CeaVac or QS-21 mixed with CeaVac. The present plan is to randomize Dukes’ B and C colorectal cancer patients to 5-FU and leucovorin versus 5-FU and leucovorin plus CeaVac.
|
Perspectives
There exists several promising immunological approaches to colon cancer therapy. The recent reports of responses in breast cancer patients treated with Herceptin, a humanized anti-Her-2/Neu antibody (162) , as well as promising results with Panorex, a murine anti-171A antibody, (6 , 15) suggest that there may exist an important role for antibody therapy in colon cancer. Successful combinations of antibody with chemotherapy as reported for Herceptin (163) would be another approach for metastatic colon cancer as well as postsurgical adjuvant therapy. Antibody-generating vaccines such as our CeaVac anti-idiotype antibody, which generates a high-titer polyclonal human anti-CEA response and Th1 T-helper cell response, should be tested in combination with chemotherapy drugs in the metastatic and adjuvant setting. We have demonstrated excellent immune responses with CeaVac combined with 5-FU regimens as well as irinotecan. Recombinant vaccines such as the recombinant vaccinia CEA vaccine (141) , as well as CEA DNA plasmids (143) and CEA-pulsed DCs, require further testing but appear promising in generating active immune responses and will soon be ready for Phase II/III clinical trials.
FOOTNOTES
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 To whom requests for reprints should be addressed, at Barrett Cancer Center for Prevention, Treatment, and Research, 234 Goodman Street, ML0502, Room 1097, Cincinnati, OH 45219-2316. Fax: (513) 584-5680. 
2 The abbreviations used are: 5-FU, 5-fluorouracil; BCG, bacillus Calmette-Guérin; TAA, tumor-associated antigen; IL-2, interleukin 2; M-CSF, macrophage colony-stimulating factor; GM-CSF, granulocyte/macrophage-CSF; ADCC, antibody-dependent cellular cytotoxicity; CEA, carcinoembryonic antigen; TCR, T-cell receptor; APC, antigen-presenting cell; DC, dendritic cell; TF, Thomsen-Friedenreich; sTn, sialyl-Tn.
Received 7/29/98; revised 11/ 4/98; accepted 11/ 9/98.
REFERENCES
-interferon and murine monoclonal antibody CD17-1A in advanced colorectal carcinoma. Cancer Res., 48: 2568-2573, 1988.4) monoclonal antibody in humans. Cancer Res., 51: 5461-5466, 1991. (TNF-) or interleukin-2 (IL-2). J. Immunol. Methods, 161: 77-90, 1993.[Medline]
(1,3)Gal antibodies in humans switch immune responses from cellular to humoral. Nat. Med., 4: 315-320, 1998.[Medline]
