Neo approaches to cancer vaccines

See allHide authors and affiliations

Science  15 May 2015:
Vol. 348, Issue 6236, pp. 760-761
DOI: 10.1126/science.aab3465

The recent success of cancer immunotherapies is rapidly changing the face of both cancer care and cancer biology. The excitement has been driven by various antibodies that block so-called “immune checkpoints” to enhance antitumor immune responses (1). Although this approach has produced durable responses for patients across a variety of tumor types, it is also the case that only a minority of patients benefit from these agents. It seems likely that among patients who do not respond or respond poorly to immunotherapies, there will be individuals who lack preexisting antitumor T cell responses. In principle, this situation can be addressed with antitumor vaccines, a strategy that has yet to yield much success despite decades of effort. The recent finding that tumor-specific mutations (neoantigens) may drive potent antitumor responses has provided hope and prompted renewed interest in the field (2). On page 803 of this issue, Carreno et al. (3) report, in a first proof of concept study, that CD8 T cell responses to tumor neoantigens can be enhanced through vaccination in melanoma patients.

Current cancer immunotherapeutic approaches act largely by reinvigorating or expanding preexisting antitumor T cell responses, overcoming a natural homeostatic mechanism designed to prevent T cell overstimulation during infection. This includes inhibiting the immune checkpoint receptors cytotoxic T lymphocyte–associated protein 4 (CTLA-4) and programmed cell death protein 1 (PD-1) on T cells. Over the past year, another conceptual advance suggests that the panoply of mutations that accompany many cancers generate potent neoantigens that provide protective immunity (2). The potency of these neoantigens reflects the fact that the immune system could not have been tolerized against them during fetal development, enhancing the chance that they will be recognized as “foreign.”

Recent studies have shown that CD8 T cells that recognize neoantigens can attack tumors, which suggests that enhancing neoantigen-specific T cell responses could improve antitumor responses (46). Because the bulk of mutations are patient-specific, any vaccine would require the development of a personalized approach. This presents some obvious challenges to implementation, such as how to determine what fraction of tumor mutations is immunogenic, which neoantigens should be included in a vaccine, what vaccine platform would be best to deliver neoantigens (see the figure), and whether this approach will work in human tumors as it apparently does in mice (7, 8).

Personalized cancer vaccines.

Routine identification of tumor-specific mutations is enabling the exploitation of their immunogenicity through vaccination. Challenges to this approach are described in the text.


One big challenge is the identification of immunogenic neoantigens. Only a very small fraction of mutations leads to the formation of a neoantigen presented on major histocompatibility complex (MHC) class I molecules (2). Carreno et al. identified several hundred such mutations in each melanoma patient. MHC class I binding affinity, determined from prediction algorithms and biochemical assays or from the affinity differential between the mutant peptide and corresponding wild-type peptide, was used to reduce the number of mutant peptides considered. Potential candidates were further prioritized according to mRNA expression level. Only seven of the possible candidates were included in the vaccine for each patient, and one-third of the selected neoepitopes proved to be immunogenic. The other neoepitopes either failed to stimulate T cells or were not processed or presented by the tumors (“cryptic antigens”). Increasing the efficiency of neoantigen identification and selection will therefore be important to extend the utility of this approach. For example, a more detailed characterization of antigen processing pathways and improved understanding of the biochemical properties of immunogenic peptides (e.g., T cell receptor affinity) should improve prediction algorithms.

The vaccine platform used to elicit immunity is another important variable. Vaccination strategies using dendritic cells generated in vitro, such as that used by Carreno et al., are based on the superior capacity of these cells to induce effective T cell responses (3, 9). Such an approach may be particularly useful when the target antigen is poorly immunogenic or to bypass the risk of developing immune tolerance, as in the case of tumor-associated self-antigens. However, implementing a cell-based vaccine approach may not be needed for strongly immunogenic antigens. A simpler immunization with long synthetic peptides combined with adjuvant or with RNA-based vaccines would be more suited for broader application of a neoantigen-based vaccine and more feasible to implement in the clinic. In mouse tumor models, mutant peptide vaccines elicited potent therapeutic immunity that resulted in tumor regression even in the absence of checkpoint blockade therapy (8). In addition, an RNA-based approach elicits effective antitumor T cell response to mutant antigens (10). Nevertheless, as Carreno et al. demonstrate, neoantigen vaccination not only can amplify existing CD8 T cell clones but also can produce responses that might have been silent prior to vaccination, thus yielding a highly diverse antitumor T cell repertoire more likely to be therapeutically effective.

Because patients underwent resection surgery before vaccination, Carreno et al. could not monitor tumor regression. Important questions remain as to whether such vaccinations will be sufficient or must be accompanied by immune adjuvants, T cell costimulatory agonists, or inhibitors of immune-suppressive mechanisms (including checkpoint inhibitors). With improved T cell monitoring and readouts of clinical efficacy, future studies should soon begin to address these issues. There could, however, be a few possible hurdles to seeing a clinical response despite induction of CD8 T cell responses upon vaccination, such as T cell exclusion from tumor, loss of tumor mutation, and immune suppression. Furthermore, although limited, there is increasing evidence that recognition of MHC class II mutant neoantigens by CD4 T cells occurs in cancer, and it is possible that their potentiation is required to further boost the activity of CD8 T cells targeting neoantigen (10-12). The optimal individualized vaccine may therefore have to include both CD8 and CD4 T cell-specific neoantigens.

Despite decades of study, cancer vaccines have not had a smooth ride and have earned a less than stellar reputation in many quarters. The findings of Carreno et al. and others (1012) reinvigorate the field and point to new strategies for achieving successful antitumor vaccines. Although there are many problems to be addressed, efficacy to be confirmed, and mechanisms to be understood, neoantigen-specific vaccines merit serious consideration.

References and Notes

  1. Acknowledgments: L.D., I.M., and M.Y. are employees of Genentech Inc., which develops cancer immunotherapies, incuding checkpoint inhibitors, and cancer vaccines.
View Abstract

Stay Connected to Science

Navigate This Article