Identification of Relevant Cancer Antigens for Vaccine Development

In This Article

Selection of Antigens to Be Used for Vaccine Construction

The first and most critical step in the design of a cancer vaccine is the selection of the antigens that will be used for its construction. At a minimum, these antigens must be able to stimulate clinically effective tumor-protective immune responses in humans and must be expressed in vivo by a patient's own tumor. Unfortunately, little is known about the identity of these antigens.

Nonetheless, several approaches are used to select tumor antigens for vaccine construction. All have some limitations. One approach is simply to use antigens that are selectively expressed on cancer cells, thus satisfying the criterion of being tumor-associated antigens. Such antigens are usually identified using monoclonal antibodies or T-cell clones. However, nothing is known about the ability of such antigens to stimulate anti-tumor immune responses in humans.

In a modification of this approach, the antibodies or T-cell clones used for antigen identification can be raised by immunizing animals with tumor cells or their extract. This procedure identifies antigens recognized as foreign in animals but provides no information about their immunogenicity in humans. In a further modification, the antibodies or T cells can be generated by stimulating human B or T cells in vitro. Again, this method provides no information about the actual immunogenicity of the antigen in vivo and hence about its value as a component of a vaccine. Lastly, antigens that are recognized in vivo by human B or T cells can be identified by a number of techniques, which include probing extract of tumor cells with antibodies or T cells obtained from cancer patients and then identifying the individual antigens within the extract that are recognized by these immune probes, as well as using molecular techniques such as SEREX analysis or phage libraries. Although these approaches identify molecules that are antigens (they are recognized by antibodies or T cells), they provide no direct evidence of immunogenicity (the ability to stimulate a B- or T-cell response when used to immunize an individual).

The only way to directly identify antigens that are immunogenic in humans is to immunize persons with the antigen(s) in question and determine whether a B- or T-cell response has been induced. This has been accomplished with a number of tumor-associated antigens, most of which are derived from melanoma. These include the MAGE series of antigens, [3] MART-1/melanA, gp100, tyrosinase, the gangliosides GD2, O-acetylated GD-3 and GM-2, [4] and a urinary tumor-associated antigen. [5] However, this approach requires that the antigen(s) of interest be isolated and purified before study of its immunogenicity in humans.

VIIR (vaccine-induced immune response) analysis provides a way to circumvent this problem. It is conducted by immunizing patients with polyvalent vaccines that contain multiple potential immunogens and determining which stimulate immune responses in vivo by comparing the profile of antibody and/or T-cell responses to antigens in the vaccine before and after vaccine treatment. The use of this procedure to identify a number of immunogenic melanoma-associated antigens is illustrated in Figure 2.

Identification of tumor antigens that are immunogenic in vivo in humans by VIIR analysis. Antibody response to a polyvalent melanoma vaccine was analyzed by western immunoblotting in serum samples collected from 3 patients before (lane A) and after immunization to the vaccine (lanes B). The antigen source for the assay was a detergent extract of freshly resected human melanoma tissue. Note that vaccine-induced antibodies identify multiple immunogenic antigens. Also note that even though all patients were immunized similarly to the same amount of vaccine, there is a striking heterogeneity in the pattern of antigens that stimulate antibody responses in the different patients.

The antigen source was an extract of fresh, surgically resected melanoma tissue, which ensured that the antigens being detected were naturally expressed in vivo by melanoma. Using this assay, we found that multiple antigens in the vaccine with molecular weights of 45, 59, 68, 79, 89, 95, and 110 kD were able to stimulate antibody responses in one or more individuals -- and thus are immunogenic in humans. [6] VIIR analysis can also be applied to directly identify antigens and peptides that can stimulate CD8+ T-cell responses in vivo in humans. This requires candidate peptides that have the potential to be immunogenic, T cells collected from HLA-matched patients before and after immunization with a vaccine containing these peptides, and an assay sensitive enough to detect the very small number of peptide-specific CD8+ T cells appearing in the circulation following vaccine immunization. Using this approach we have identified a number of peptides expressed by several melanoma-associated antigens [7] and presented by a variety of class I HLA common in melanoma patients (ie, HLA-A1, A2, B7), which can stimulate CD8+ T-cell responses in humans. In contrast to studies conducted by in vitro immunization, no peptide was clearly immunodominant, [7] indicating that the immunogenicity of antigens in vivo and in vitro may differ.

However, a major limitation of all these procedures is that the functional activity of the immune responses induced by the selected antigen(s) remains unknown. Simply because an antigen stimulates an antitumor immune response does not mean that the response is protective. Certain responses are irrelevant; others may even enhance tumor growth. Consequently, identifying antigens that are immunogenic in humans is a critical but not conclusive step in selecting antigens that may be important for vaccine construction. A correlation between the induction of a particular anti-tumor immune response by an individual antigen and improved clinical outcome provides additional but still indirect evidence of clinical effectiveness. Incidentally, such correlations must be subjected to multifactorial analysis to exclude the clinical effects of other risk factors and/or concurrent immune responses to other antigens in the vaccine.

In the end, the only way to objectively determine whether a tumor antigen is relevant for vaccine construction is to demonstrate that it can increase resistance against the cancer -- and the only way to do so is to conduct a randomized clinical trial large enough to demonstrate a statistically significant difference in clinical outcome between vaccine-treated and control patients. Such a trial usually involves hundreds of patients. Unfortunately, the time and expense involved in conducting such trials precludes the use of this rational approach to evaluate the immunogenicity of any but a handful of the large number of tumor antigens that are candidates for vaccine construction.

A more reasonable alternative is to develop vaccine design strategies that take into account the fact that the antigens required for the vaccine to be effective are not known. One way to do so is to construct polyvalent vaccines that contain numerous tumor-associated antigens, greatly increasing the chances that the vaccine will contain at least some antigens that can stimulate protective antitumor immunity.

For vaccine immunotherapy to be effective, the antitumor immune responses induced by the vaccine must be directed to antigens expressed by the tumor being treated. In addition, such antigens should be located at a site on the tumor where they can be seen by, and interact with, immune-effector mechanisms -- ie, they must be on the external surface of the tumor cells. Unfortunately, the pattern of tumor antigens expressed by cancers of the same histological type varies in different individuals, [8] as do the patterns expressed by different tumor nodules in the same individual [9,10,11] and even by different tumor cells in the same nodule. [12,13] In addition, as a result of antigenic modulation, the profile of tumor antigens expressed by a tumor during its progression may be altered by the immune response of the host. Thus, the actual pattern of tumor antigens expressed by a tumor in an individual patient is difficult to ascertain, even when tumor tissue is available for analysis, as the sample may differ antigenically from the tumor(s) that remains. For the same reason, it is unlikely that a vaccine prepared from a single tumor antigen will be effective against a broad range of tumors of the same type.

Fortunately, some tumor antigens are commonly expressed on many, although not all, tumors of the same histological type. The expression of different common antigens is often complementary, so that tumor cells that lack a particular antigen may express other tumor antigens. [11] Thus, a viable strategy to circumvent the antigenic heterogeneity of tumors is to construct polyvalent vaccines that contain a broad range of widely expressed immunogenic tumor antigens [14] whose pattern of expression on different tumors is complementary.

Another variable that complicates antigen selection for vaccine construction is that there is marked heterogeneity in the ability of different patients to develop an immune response to the same antigen. Expressed another way, some antigens stimulate immune responses in one patient but not in another, while the reverse is true for other antigens. This heterogeneity is illustrated for antibody responses in Figure 2. In part, it is due to the HLA type of the patient. Cellular CD8+ T-cell responses can be stimulated only by antigen that is bound to a class I HLA molecule, and different antigens bind best to different class I HLA molecules. Thus, whether or not a particular antigen stimulates a T-cell response will depend on whether the patient expresses the class I HLA molecule that binds that antigen. In part, the heterogeneity is also due to HLA-independent mechanisms, such that the same antigen or peptide can vary in its ability to stimulate immune responses in different individuals with the same class I HLA type. [15]

The implication of both phenomena for vaccine construction is that to maximize stimulation of immune responses to tumors, vaccines should be constructed from multiple immunogenic antigens.

It is clear from animal studies that under the right circumstances an immune response targeted to a single tumor antigen can be sufficiently potent to cause tumor regression. It is unclear whether the same is true in humans. In any event, it seems logical that if a response to a single antigen on a tumor cell can be clinically effective, then responses directed to multiple targets should be even more effective. This is another argument for the belief that polyvalent vaccines that can stimulate responses to multiple antigens expressed by tumors will be more effective than monovalent vaccines targeting a single or a limited number of antigens.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.