Carbohydrate antoantigens have proven to be suitable targets for immune recognition and attack against cancer cells, because of their abundance at the cell surface and their unexpected immunogenicity.
Carbohydrates play key roles in intracellular interactions as targets for selectins and adhesins, which may be crucial, not discretionary, to tumor cell survival and the metastatic process. Passively administered and vaccine-induced antibodies in preclinical models are capable of interfering with these processes directly, inducing complement-mediated inflammation and lysis and mediating opsonization, inflammation, and tumor cell death by other Fc-mediated mechanisms.
In clinical studies, conjugate vaccines have induced antibodies against a variety of carbohydrate antigens on glycolipids and mucins, and these antibodies are correlated with an improved prognosis. Phase III clinical trials with monovalent and phase II clinical trials with polyvalent carbohydrate antigen vaccines are ongoing.
I. ANTIBODIES CAN ELIMINATE EARLY CANCER SPREAD
The effect of all commonly used vaccines against infectious agents is thought to be primarily a consequence of antibody induction. Antibodies eliminate viral or bacterial infections by preventing their spread through the bloodstream and by eliminating early tissue invasion. Antibodies are also ideally suited for eliminating circulating tumor cells and micrometastasis.
A. Evidence in Experimental Animals
The basis for vaccines that induce only antibodies is well documented in experimental animals. Experiments by Zhang involving the administration of monoclonal antibody (mAb) 3F8 against the ganglioside GD2 have been particularly informative. Administration of 3F8 prior to or up to 4 days after intravenous tumor challenge with EL4 lymphoma (which expresses GD2) results in cure of most mice. Vaccination with GD2-keyhole limpet hemocyanin (KLH) conjugate vaccine achieves the same effect. This timing is comparable to treatment in the adjuvant setting after surgical resection of the primary malignancy or lymph node metastasis, as in both cases the target is circulating tumor cells and micrometastasis. The adjuvant setting in the clinic has been more closely modeled by injecting EL4 cells into the footpad on day 1 and amputating the visible tumor on day 25. Treatment at that point with 3F8 or GD2-KLH once again protects most mice from tumor recurrence, while untreated mice expire with nodal, hepatic, and other systemic metastases.
B. Evidence in Patients
Naturally acquired or passively administered antibodies are also associated with a more favorable prognosis in the clinic.
1. Patients with resected melanoma having natural antibodies against GM2 ganglioside had an 80-90% 5-year survival compared to the expected 40% rate in studies from the John Wayne Cancer Center and Memorial Sloan-Kettering Cancer Center (MSKCC).
2. Patients with small-cell lung cancer (SCLC) and naturally acquired antibodies against SCLC had a prolonged survival compared to antibody-negative patients.
3. Paraneoplastic syndromes in cancer patients have been associated with high titers of natural antibodies against onconeural antigens expressed on neurons and certain malignant cells. The antibodies were apparently induced by tumor growth and have been associated with autoimmune neurologic disorders but also with delayed tumor progression and prolonged survival.
4. Patients with resected Dukes C colon cancer who were treated with mAb 17-1A in the adjuvant setting had a significantly prolonged disease free and overall survival compared to randomized controls. This is the only randomized clinical trial of mAb therapy conducted in an adjuvant setting reported to date.
C. Mechanism of Action
Some antibodies may have direct effects, such as by inhibiting tumor cell attachment or inhibiting growth hormone receptors, but in general the interaction of antibody and antigen is without consequence unless Fc-mediated secondary effector mechanisms are activated. Binding of antibody to antigen results in a functional change in the Fc portion of the antibody and activation of several effector mechanisms. For cancer carbohydrate antigens, IgM bound to antigen is the most active complement activator in the intravascular space, and in humans, IgG1 and IgG3 are the most important complement activators extravascularly.
Complement activation mediates inflammatory reactions, opsonization for phagocytosis, clearance of antigen-antibody complexes from the circulation, and membrane attack complex-mediated lysis (CDC). Opsonization for ingestion and destruction by phagocytosis or cytotoxic mechanism can occur through complement activation, but can also occur directly as a consequence of Fc receptors on phagocytic cells [antibody-dependent cell-mediated cytotoxicity (ADCC)].
II. BASIS FOR FOCUSING ON CARBOHYDRATE ANTIGENS AS TARGETS FOR ANTIBODIES
A. Abundant Expression at the Cancer Cell Surface
Cancer cells may occasionally express as many as 106 copies of individual protein antigens at the cell surface, as has been described in some cell lines for epidermal growth factor and HER2/neu, but most protein antigens are expressed in far smaller numbers. The number of carbohydrate epitopes expressed at the cell surface, however, is often far greater. Thus the median number of glycolipid molecules, such as gangliosides GM2 and GD2, on melanoma cells is close to 107 and for sarcomas and neuroblastomas is over 5 × 107. The median number of GD3 molecules expressed on melanoma and sarcoma cells is also approximately 5 × 107. The number of copies of a given glycoprotein carbohydrate epitope on cancer cells is less well defined. However, each mucin molecule has 30-100 tandem repeats, each of which can express 5-15 carbohydrate epitopes. Because the surface of epithelial cancer cells is covered by a dense glycocalyx of various mucins, most epithelial cancer cells are probably covered by at least 100,000 mucin molecules. Consequently, the number of small carbohydrate epitopes, such as TF, Tn, or sTn, on mucins at or near the cell surface is probably well in excess of 107.
B. Effective Targets for Immune Attack
In our series of 110 patients immunized with melanoma cell or melanoma cell lysate vaccines mixed with various adjuvants and in the series of patients immunized with allogeneic melanoma cell vaccines described by Tai, gangliosides GM2 and GD2 were the only antigens recognized by multiple patients. Gangliosides are acidic glycosphingolipids overexpressed at the cell surface of melanomas, sarcomas, and other tumors of neuroectodermal origin. Gangliosides have also been shown to be effective targets for immunotherapy with mAbs with major responses seen in patients after treatment with mAbs against GM2, GD2, and GD3. The experience with mono- or disaccharide epitopes Thomsen-Friedenrich antigen (TF), Tn, and sialyl Tn (sTn), which are expressed on mucins in a great variety of epithelial cancers, has been similar. Fung and Singhal have demonstrated that immunization with TF and sTn protects mice from a subsequent challenge with cancer cells expressing these antigens. Springer and MacLean have increased antibodies against TF and sTn in cancer patients by vaccination. Patients with naturally increased or vaccine-induced antibodies against GM2 and sTn had a more favorable prognosis. Hence, active and passive immunotherapy trials have identified carbohydrate epitopes on glycolipids and glycoproteins as uniquely effective targets for cancer immunotherapy. The simplified chemical structure of these carbohydrate antigens is demonstrated as they appear at the cell surface in Fig. 1.
FIGURE 1 Carbohydrate epitopes on cell membrane glycoconjugates. Glc, glucose; Gal, galactose; GalNAc, N-acetyl galactosamine; NCAM, neural cell adhesion molecule; KSA, a panepithelial cancer antigen also referred to as 17-1A and GA733.
C. Biological Roles of Cell Surface Carbohydrates
The great majority of the molecules of the mammalian plasma membrane are glycosylated such that glycan structures form a dense forest covering the cell surface. These glycan chains are found on glycolipids and integral membrane glycoproteins, as well as on more specialized glycoproteins, such as mucins and proteoglycans. To some extent, these carbohydrates serve structural, protective, and stabilizing roles, but it is becoming increasingly recognized that they can have information-bearing functions as selectins and adhesins in cell-cell recognition and adhesion as well.
Carbohydrate structures on glycoproteins and glycolipids have been implicated in such normal cell functions as proliferation, interaction with endothelial cells, leukocytes, and platelets, embryogenesis, neural cell adhesion, and the biology and metastatic potential of tumor cells. All tumors studied have changes in the expression of carbohydrate structures, which are characteristic of the tissue of origin of the tumor. As a general rule, tumors of neural crest origin (e.g., melanoma, sarcoma and neuroblastoma) exhibit overexpression of gangliosides (sialylated glycolipids), whereas epithelial cancers (carcinomas) have altered fucosylated structures and mucin core structures (TF, Tn, sTn) as their characteristic antigens. Numerous studies have shown a correlation between high expression of certain carbohydrate specificities (including Ley, sTn, and Tn blood group antigens) and metastatic potential and decreased patient survival.
III. SELECTION OF CARBOHYDRATE ANTIGENS FOR VACCINE CONSTRUCTION
Using panels of mAb against carbohydrate antigens Zhang screened a variety of malignant and normal tissues by immunohistochemistry. In general, ganglioside antigens have a very different distribution on various malignancies than the other carbohydrate antigens. Melanomas, sarcomas, and neuroblastomas express GM2, GD2, and GD3 but none of the other antigens, whereas epithelial cancers express a broad range of carbohydrate antigens but not GD2 or GD3. Carbohydrate antigens expressed on 50% or more of tumor cells in 60% or more of biopsy specimens of some of the most common types of cancers are listed in Table I.
Each of these antigens is also expressed on some normal tissues. GM2, GD2, and GD3 are all expressed in the brain, especially GD2, which is also expressed on some peripheral nerves and, unexpectedly, a subpopulation of B lymphocytes in the spleen and lymph nodes. Also unexpectedly, GM2, as defined by mAb 696, is expressed at the secretory borders of most epithelial tissues. GD2 and GD3 are also expressed at low levels in connective tissues of multiple organs, and GD3 is known to be expressed on some human T lymphocytes. Fucosyl GM1 is expressed on occasional cells in the islets of Langerhans and in some sensory neurons in the dorsal root ganglia. Polysialic acid is expressed significantly in brain and some bronchial epithelial cells. Globo H, Ley, TF, Tn, and sTn are expressed at the secretory borders of a variety of epithelial tissues. Lex and sialyl Lex are expressed at the secretory border of many epithelial tissues and also on polymorphonuclear leukocytes.
The broad expression of most of these carbohydrate antigens on normal tissues raises concern over suitability as targets for immunotherapy. However, there is now sufficient experience from clinical trials with mAbs against GD2, GD3, Lex, and sTn and with vaccine-induced antibody responses against GM2, GD2, GD3, TF, Tn, sTn, globo H, and Ley to draw conclusions about the consequences of antigen distribution on various normal tissues. Expression of gangliosides in the brain and expression of many of these antigens at the secretory borders of epithelial tissues have induced neither immunological tolerance nor autoimmunity once antibodies were present, suggesting that they are sequestered from the immune system.
The known expression of Lex and sialyl Lex on polymorphonucleocytes, in addition to epithelial surfaces, and the granulocytopenia seen after the treatment of patients with mAb FC-2.15 (later found by to recognize Lex) exclude these two carbohydrates as candidates for vaccine construction and so they have been omitted from Table I. The expression of fucosyl GM1 on subpopulations of cells in the islets of Langerhans and the dorsal root ganglia was initially of concern, but no evidence of autoimmunity has been detected in nine patients vaccinated with fucosyl GM1 who produced high titers of IgM and IgG antibodies against fucosyl GM1. Administration of high doses of some, but not other, IgG mAbs against GD2 has been associated with peripheral neuropathy in melanoma patients. High, but not lower, doses of anti-Ley mAb BR96 conjugated to doxorubicin have resulted in vomiting, hematemesis, and amylase elevations in some patients, whereas BR55, a second mAb against Ley studied by the same investigators, resulted in no such toxicity. With regard to vaccines, trials with vaccines against GD2, fucosyl GM1, and Ley have addressed these questions more directly.
Moderate to high titers of IgM and IgG antibodies have been induced and have not been associated with any evidence of autoimmunity or other adverse clinical effects. Consequently, Table I, which lists carbohydrate antigens expressed at the cell surface of each of the common solid tumors, has served as a guide as single antigen or polyvalent antigen vaccine trials were planned.
TABLE I Carbohydrate Targets for Vaccine ConstructionTumor Antigensa Melanoma GM2, GD2, GD3 Neuroblastoma GM2, GD2, GD3, polysialic acid Sarcoma GM2, GD2, GD3 B-cell lymphoma GM2, GD2 Small-cell lung cancer GM2, fucosyl GM1, polysialic acid, globo H, sialyl Lea Breast GM2, globo H, Ley, TF Prostate GM2, Tn, sTn, TF, Ley Colon GM2, Tn, sTn, TF, sialyl Lea, Ley Ovary GM2, globo H, sTn, TF, Ley Stomach GM2, Ley, Lea, sialyl Lea
aPresent on at least 50% of cancer cells in at least 60% of biopsy specimens.
IV. APPROACHES FOR AUGMENTING THE IMMUNOGENICITY OF CARBOHYDRATE ANTIGENS: RESULTS OF PRECLINICAL STUDIES
Vaccination of mice with irradiated melanoma cells selected for GD3 expression plus adjuvants was able to induce low levels of IgM antibodies against GD3, but this could be accomplished more effectively and simply by immunizing with purified GD3 plus immunological adjuvants. While GD3 alone induced no response at all, GD3 adherent to Salmonella minnesota mutant R595 or liposomes containing monophosphoryl lipid A (MPL) induced moderate titers of IgM antibodies in most mice. Attempts by Ritter at augmenting the immunogenicity of GD3 by making minor structural modifications to the GD3 so it would be foreign and not recognized as self were unsuccessful.
Based on progress with conjugate vaccines against bacterial polysaccharide antigens, Helling systematically compared the immunogenicity of conjugate vaccines constructed with different carriers and adjuvants using GD3 as antigen. KLH was the best of the six immunogenic carrier molecules tested, the conjugation method was important, and a potent immunological adjuvant was required. GD3 conjugated to KLH by the ceramide double bond and mixed with immunological adjuvant QS-21 (a purified saponin fraction obtained from the bark of the Quillaja saponaria Molina tree) was optimal, inducing higher titers of antibody (1/1280) and, for the first time, consistent IgG antibodies (median ~ 1/160). A simple mixture of GD3 and QS-21 or GD3, KLH, and QS- 21 induced no antibodies. These results have been confirmed more recently using a series of other antigens, including the gangliosides GM2 and fucosyl GM1, the neutral glycolipids globo H and LeY, and the carbohydrate epitopes expressed on mucins Thompson Freidenreich antigen (TF), Tn, and sialyl Tn (sTn). The KLH conjugate vaccines induced antibodies that reacted with the synthetic antigens and with tumor cells expressing these antigens in all cases except in the case of Tn, TF, and sTn. Trimers of these antigens resulted in a more efficient induction of antibodies reactive with the cancer cell surface.
V. CLINICAL TRIALS WITH CARBOHYDRATE ANTIGEN VACCINES
A. Results of Serological Studies
The availability of reliable serological assays to serve as surrogate markers in clinical trials has greatly speeded progress in developing consistently immunogenic vaccines against carbohydrate antigens. A series of clinical trials with vaccines containing purified carbohydrate antigens has been conducted with TF as target by Springer, with sTn as target by MacLean, and with these and other carbohydrate antigens by Livingston, Helling, Ragupathi, and collaborators at MSKCC. Results demonstrating the progress made in trials at MSKCC are demonstrated in Table II. Trials with other antigens, including vaccines containing unconjugated ganglioside vaccines, GD3 congeners, and unclustered sTn or TF conjugates, are not shown because the antibodies induced failed to react with antigen-positive tumor cells. The conjugate vaccines shown in Table II are currently able to induce antibodies reactive with purified antigens and tumor cells expressing these antigens in essentially 100% of cases for GM2, fucosylated GM1, and sTn(c) and 50% or more of cases for vaccines containing GD2, GD3, globo H, LeY, Tn(c), and TF(c).
TABLE II Summary of Serological Results in Vaccinated Patients: The MSKCC ExperienceAntigen Median ELISA IgM Median ELISA IgG IgG subclass Median FACS IgM Median FACS IgG Median IAb Median CDC % patients positivea References GM2 640 320 IgG1 + 3 +++ ++ ++ ++ >90 Livingston et al. (1994), Livingston and Ragupathi (1997)
GD3 80 16 + + + + 60 Ragupathi, G. (2000), Int. J. Can. Fucosyl GM1 320 640 IgG1 +++ ++ ++ >90 Dickler,k M.H. (1999), Clin Ca. Res. Globo H 640 40 IgG1 + 3 ++ + ++ + 75 Ragupathi, G. (1999), Angewardte Chem.
Slovin, S. (1999), Proc. Natl. Acad. Sci. LewisY 80 0 ++ + + + 50 Sabbatini, P. (2000), Int. J. Can. Tn(c) 1280 1280 ++ - + - 60 STn(c) 1280 160 IgG3 +++ - + - >90 TF(c) 1280 1280 + - + - 50
aPercentage of KLH conjugate plus QS-21-vaccinated patients positive by ELISA and FACS or CDC.
bIA, Immune adherence.
B. Clinical Impact
Vaccine-induced antibody responses against GM2 and sTn have been associated with a more favorable clinical course. A randomized trial with GM2/BCG was conducted in 122 AJCC stage III melanoma patients who were free of disease after resection of metastatic disease in regional lymph nodes. This trial was based on the previous demonstration that immunization with GM2/BCG induced IgM antibodies in 85% of patients and that the production of these antibodies correlated with a more favorable prognosis. Patients were randomized to receive five immunizations over a 6-month period with BCG alone (64 patients) or BCG with GM2 adherent to the BCG surface (58 patients). Fiftyseven patients had GM2 antibody, which was present naturally or vaccine induced, and these patients had a significantly increased disease free (p = 0.004) and overall (p = 0.02) survival. Comparing the GM2/BCG and BCG groups, exclusion of all patients with preexisting GM2 antibodies (1 in the GM2/BCG group and 5 in the BCG group) resulted in differences of 23% (from 27% of patients remaining disease free to 50%) in disease-free survival and 14% in overall survival at 50 months with a minimal follow-up of 50 months. However, when all patients in the two treatment groups were compared as randomized, these increases were 18 and 11% for disease-free and overall survival in favor of vaccination with GM2/BCG, with neither difference achieving statistical significance. Antibody responses were predominately IgM, of moderate titer, and short lived (returning to baseline within 3 months of the final immunization).
A correlation between vaccine-induced antibody responses and clinical course has also been seen after immunization with sTn-KLH plus Detox. In a series of 113 patients with various types of epithelial cancers treated at the Cross Cancer Institute by MacLean, the 51 patients with high sTn antibody responses after vaccination survived significantly longer than the 62 patients with lower antibody responses. Antibody responses against KLH showed no such correlation. In a separate study, improved survival was also seen in 25 patients with advanced breast cancer who had a high antibody response to the sTn vaccine in combination with intravenous cyclophosphamide compared to 25 patients with a low antibody response and who did not receive the intravenous cyclophosphamide. The median survival increased from 13.3 months in patients with low anti-sTn antibody titers compared to 26.5 months in those with high titers.
VI. PHASE II/III RANDOMIZED TRIALS WITH CARBOHYDRATE VACCINES
Based on the progress described earlier, a series of randomized phase III trials with antibody-inducing single antigen vaccines against GM2 and sTn have been initiated (see Table III). Patient accrual in the stage III NED (no evidence of disease) GM2 trial (850 patients) has been completed. Follow-up data, accumulated at a median period of 16 months, show that the GMK vaccine does not improve relapse-free survival over IFN-α2b. And at this early time point, disease free and overall survival were significantly lower in the GMK group compared with the interferon group. Further follow-up is needed to show whether the overall survival rates are equivalent. These trials involve (1) GM2 or sTn attached covalently to KLH for optimal induction of T-cell help and enhanced presentation to the immune system; (2) the use of a potent immunological adjuvant to further augment immunogenicity; and (3) immunization in the adjuvant or limited disease setting based on the preclinical and clinical studies described previously.
Although there is every indication that immunization with these single antigen vaccines may prove beneficial when administered in the adjuvant setting, in the long run, polyvalent vaccines offer greater promise. As described earlier, we have now induced antibodies against GM2, fucosyl GM1, and sTn in close to 100% of patients and against GD3, GD2, Globo H, Ley, Tn, and TF in 50-80% of patients. In all cases, these antibodies have been demonstrated to react with the cell surface of antigen-positive cancer cells and to activate complement. Consequently, phase I/II or phase II trials with polyvalent vaccines in the adjuvant setting have been initiated at MSKCC in patients with prostate cancer, ovarian cancer, and breast cancer. The antigens included in these vaccines and other aspects of these trials are listed in Table III.
TABLE III Phase II or Randomized Phase III Trials with Carbohydrate-Based Cancer VaccinesVaccine Patients Sponsora (clinical centers) Status Trial design Monovalent vaccines GM2-KLH+QS-21 Melanoma, stage III NED PP, BMS ECOG, SWOG, CALGB, NCTP Completed Randomized phase III GM2-KLH+QS-21 Melanoma, stage II NED PP EORTC Ongoing Randomized phase III sTn-KLH+Detox Breast cancer, CR or PR to chemotherapy Biomira Multicenter Ongoing Randomized phase III Polyvalent vaccines (PVb) PV Ovarian cancer, first remission, NED MSKCC Ongoing Phase I/II PV-Tn(c)-KLH Prostate cancer, rising PSA, NED MSKCC Ongoing Phase II PV Breast cancer, high-risk stage, NED MSKCC Ongoing Phase I/II
aPP, Progenics Pharmaceuticals Inc. (Tarrytown, NY); BMS, Bristol-Myers Squibb Co. (Wallingfold, CT); Biomira, Biomira Inc. (Edmonton, Alberta); MSKCC, Memorial Sloan-Kettering Cancer Center (New York, NY); ECOG, Eastern Cooperative Oncology Group (Boston, MA); SWOG; Southwestern Oncology Group (San Antonio, TX); CALGB, Cancer and Leukemia Group B (Chicago, IL); NCTG, North Central Treatment Group (Rochester, MN); EORTC, European Organization for Research and Treatment of Cancer (Brussels, Belgium).
bGM2-KLH, globo H-KLH, LeY-KLH, TF(c)-KLH, Tn(c)-KLH, sTn(c)-KLH, MUC1-KLH plus QS-21.
Philip O. Livingston
Memorial Sloan-Kettering Cancer Center
ANTIBODIES IN THE GENE THERAPY OF CANCER ; ANTIIDIOTYPIC ANTIBODY VACCINES ; CANCER VACCINES: GENE THERAPY AND DENDRITIC CELL BASED VACCINES ; CANCER VACCINES: PEPTIDE- AND PROTEIN-BASED VACCINES ; DNABASED CANCER VACCINES ; RESISTANCE TO ANTIBODY THERAPY ; TUMOR ANTIGENS
adjuvant Assisting or aiding; after surgical resection of all known disease, a treatment administered to prevent recurrence; any substance mixed with an antigen to increase immunogenicity.
conjugate vaccine A vaccine containing the relevant antigen covalently linked to an immunogenic carrier molecule.
ganglioside A sialic acid containing glycosphingolipid.
keyhole limpet hemocyanin A large highly immunogenic respiratory chromprotein obtained from the blood of the keyhole limpet (a mollusk) and frequently used as a carrier protein.
mucin A glycoprotein that is extensively o-glycosylated.
polyvalent vaccine Vaccines containing multiple antigenic epitopes.
saponin A class of compounds extracted from plants with potent adjuvant activity containing a hydrophilic carbohydrate moiety and a hydrophobic triterpene moiety.
trimer or cluster Three antigenic epitopes (i.e., TF, Tn, or sTn) linked to three threonines with a terminal linker.
Livingston, P. O. (1998). The case for melanoma vaccines that induce antibodies. In "Molecular Diagnosis, Prevention and Treatment of Melanoma" (J. M. Kirkwood, ed.), pp. 139-157. Dekker, New York.
Livingston, P. O., and Ragupathi, G. (1997). Carbohydrate vaccines that induce antibodies against cancer. II. Previous experience and future plans. Cancer Immunol. Immunother. 45, 10-19.
Livingston, P. O., Wong, G. Y. C., Adluri, S., Tao, Y., Padavan, M., Parente, R., Hanlon, C., Helling, F., Ritter, G., Oettgen, H. F., and Old, L. J. (1994). Improved survival in AJCC stage III melanoma patients with GM2 antibodies: A randomized trial of adjuvant vaccination with GM2 ganglioside. J. Clin. Oncol. 13, 1036-1044.
Livingston, P. O., Zhang, S., and Lloyd, K. O. (1997). Carbohydrate vaccines that induce antibodies against cancer. I. Rationale. Cancer Immunol. Immunother. 45, 1-9.
MacLean, G. D., Reddish, M. A., Koganty, R. R., and Longenecker, B. M. (1996). Antibodies against mucin-associated sialyl-Tn epitopes correlate with survival of metastatic adenocarcinoma patients undergoing active specific immunotherapy with synthetic sTn vaccine. J. Immunol. 9, 59-68. Riethmuller, G., Schneider-Gadicke, E., Schlimok, G., Schmiegel, W., Raab, R., Hoffken, K., Gruber, R., Pichlmaier, H., Hirche, H., Pichlmayr, R., Buggisch, P., Witte, J., and the German Cancer Aid 17-1A
Study Group (1994). Randomised trial of monoclonal antibody for adjuvant therapy of resected Dukes' C colorectal carcinoma. Lancet 343, 1177-1183.
Springer, G. F. (1984). T and Tn, general carcinoma autoantigens. Science 224, 1198-1206.