A cancer diagnosis often results in any number of relatively nonspecific treatments, such as surgery, radiation, or chemotherapy, all of which can destroy healthy tissue along with the tumor. Seeking approaches that could successfully eradicate tumors while avoiding such collateral damage from aggressive therapy, researchers have developed a number of treatments targeted to specific types of tumors and, more recently, a handful of therapies aimed at modulating the body’s immune cells to more effectively fight its cancer. Mounting evidence suggests that such immunotherapies can effectively turn patients’ own immune systems against the very molecules that distinguish the tumor from normal cells, allowing the body’s T cells to serve as guided missiles that seek and destroy only the intended target.
Mounting evidence suggests that immunotherapies can effectively turn patients’ own immune systems against the very molecules that distinguish the tumor from normal cells.
This approach is based on the progressive mutational process that drives cancer evolution and generates antigens that are expressed exclusively in and on tumor cells. (See “Trunks and Branches.”) By training the immune system to target those tumor-specific antigens, called neoantigens, researchers hope to selectively eradicate the cancer cells while leaving healthy tissue unharmed.
Advances in genomic sequencing and bioinformatics over the past decade have synergized to produce a clearer picture of the immune response to cancer and to move this concept from the laboratory to clinical practice. Hailed as nothing short of a revolution in oncology, immunotherapies have the potential to upend the field’s standard of nonspecific, often damaging treatment regimens. Understanding the nature of cancer neoantigens is critical to continued development of these precision therapies.
Cancer as a disease of mutations
The accumulation of mutations that control critical cellular functions is believed to occur throughout a normal cell’s progression towards neoplasia—the stage at which a cell can be considered cancerous. In support of this notion, researchers have observed that early events in cancer development frequently involve loss-of-function mutations in DNA-repair proteins, thereby accelerating the rate of mutation accumulation in the tumor.1
Recent advances in both the cost and capacity of genomic sequencing and the development of powerful new computational methods for its analysis have enabled the mutational landscape of a number of histologically distinct tumors to be evaluated and cataloged. These efforts have revealed that a surprising range in the mutational burden exists among different tumor types, with those arising in mutagen-exposed tissues such as skin, lung, and bladder containing the greatest numbers, second only to those tumors lacking DNA mismatch repair (MMR) or proofreading functions, as occurs most commonly in certain subsets of colorectal and endometrial cancer.2 While some of these mutations are in known “driver” oncogenes, the majority occur in genes whose functions play no obvious role in either establishing or maintaining the transformed state, and are collectively referred to as “passenger” mutations.
Both driver and passenger mutations can lead to the cell’s production of tumor-specific neoantigens, which can be recognized by the T lymphocytes that are tasked with detecting foreign invaders in the body. T cells typically recognize short, linear peptides derived from proteins of intracellular and extracellular pathogens and presented on the major histocompatibility complex (MHC) molecules found at the surface of nearly all the cells in the body. While MHC-bound peptides that are derived from normal self proteins are largely ignored—a process known as tolerance—those that differ in sequence, even by a single amino acid, can be efficiently targeted for destruction by T cells.
Whereas the presentation of foreign peptides on the surface of cells infected with viral and bacterial pathogens is a well-studied phenomenon, only recently have researchers begun to consider the facts that tumor cells also display foreign molecules (in the form of mutated peptides) and that these antigens could be exploited for tumor control. Numerous preclinical animal-immunization studies have shown that both induced and spontaneous tumors possess varying degrees of intrinsic antigenicity, meaning that the immune system—specifically CD4+ and CD8+ T cells—can protect the animals from cancer development when they are rechallenged with the same tumor type. The earliest studies determined that the relevant antigens were those encoded by the viral oncogenes used to generate the experimental tumors, but subsequent work on spontaneous tumors showed that T cells can and do target mutated self proteins.3 And recent studies in both mice and humans have documented the appearance of T cells specific for neoantigens expressed by a tumor.4
One type of immunotherapy in which a tumor’s neoantigens are suspected to play a role is immune checkpoint inhibitors, which block inhibitory signals that would otherwise repress the body’s cancer-fighting T cells. In both preclinical models and human cancer patients, administration of antibodies to block immune checkpoint pathways, including PD-1/PD-L1 and CTLA-4, can elicit strong antitumor T-cell responses. In 2015, several groups discovered that responsiveness to immune checkpoint blockade correlates with neoantigen load;4,5 the more tumor-specific antigens the cancer cells have, the greater chance the body’s army of T cells will include some lymphocytes with matching receptors. Similarly, a growing number of clinical studies testing the use of T-cell transfusions, also known as adoptive cellular therapy, have demonstrated that mutant gene products are the immunological targets of the transferred lymphocytes.
These fundamental studies and single-patient results have provided a compelling case for targeting neoantigens as a class across a range of cancers. The next key developments must be to rapidly identify unique cancer markers and train the immune system to effectively target them.
The rocky path to the clinic
Neoantigens derive from somatic mutations that produce modified or novel peptide sequences within a tumor cell’s repertoire of expressed proteins. These include missense mutations, frameshifts, translocations, and mRNA splicing variants, as well as mutations that influence posttranslational processing, such as phosphorylation and glycosylation. All of these mutations can result in molecular changes that can be discriminated by an appropriate T-cell receptor.
Identification of tumor-expressed somatic mutations by sequencing is a relatively straightforward exercise that is increasingly within the grasp of most clinical research centers. The general strategy is to perform genomic or whole-exome sequencing of both a tumor and a reference genome (usually obtainable from peripheral lymphocytes or buccal swabs), as well as RNA sequencing to confirm that variants identified are indeed expressed in the tumor.
Predicting whether a patient will have an immune response to a particular mutation is challenging, however, as this depends not only on the presence of a suitable T cell within an individual’s immune repertoire, but also on myriad factors pertaining to the mutant protein’s ability to be processed and shuttled to the lymph nodes for interrogation by antigen-presenting cells. (See “Special Delivery.”) Nonetheless, researchers are now working to improve methods for identifying neoantigens in human cancer, in hopes of being able to develop personalized vaccine and cellular therapy approaches.
To this end, a number of computational tools have been developed to analyze a range of features thought to be relevant to a given peptide’s ability to be a T-cell target. These include the amino acid sequence of the mutated peptide, its similarity to the corresponding wild-type sequence, its predicted ability to undergo proteolysis, and its predicted binding affinity to relevant MHC molecules. The success rate of these analyses in forecasting which somatic mutations can be neoantigen targets is, to date, less than impressive, however. As an alternative approach to neoantigen identification, researchers have used sensitive mass spectrometry techniques to define the spectrum of peptides bound to a tumor’s surface MHC molecules. While this strategy has successfully identified neoantigen targets in murine tumors, its applicability to human cancers has yet to be established.6
A third approach is to marry the empiricism and sequence analysis themes inherent in the first two, but instead of working purely in computational space, researchers test a patient’s peripheral or tumor-infiltrating T cells for recognition of predicted neoantigens ex vivo. This has the advantage of confirming, rather than presuming, what the relevant targets are likely to be and allowing for the discovery of responses that would not have been evident from the computational models.
Predicting whether a patient will have an immune response to a particular mutation is challenging.
As our tools for the cellular- and molecular-level interrogation of tumors and for the identification of neoantigens continue to improve, other challenges to their development as therapeutics have become increasingly clear—namely, cancer’s ability to adapt. Tumor cells’ adaptations to maximize growth and therapeutic resistance likely represent the most significant impediment to neoantigen-guided precision immunotherapy. Tumors can counteract immune control via a number of extrinsic pathways of adaptive resistance, including those that exploit normal physiological pathways of immune suppression.7 By eliminating the presentation of antigens, for example, such pathways can render the tumor invisible to the immune system. (See “How Cancers Evolve Drug Resistance.”)
The genetic heterogeneity that results from tumor cells’ evolution can also present a significant obstacle to neoantigen-focused immunotherapeutic strategies, as not all cells will carry the targeted antigens. Retrospective studies on patients who underwent checkpoint blockade immunotherapies have found that positive responses were associated with targeting clonally expressed neoantigens, which are present on most or all tumor cells. Treatments targeting subclonal mutations, on the other hand, tended to result in little or no response in the patients.8
As sequencing costs continue to decrease, research efforts should be aimed at capturing the clonal diversity of somatic mutations present within an individual patient over the course of his or her disease. In this way, clinicians can have a chance of identifying the mutations present in the majority of tumor sites (the “trunk” mutations) versus those that arise either late in the development of the cancer or in a select subclonal population (the “branch” mutations). Although it is tempting to imagine that driver mutations would, by virtue of their potent effects on enhancing self-renewal, more likely be found in the trunk than the branches of a tumor’s mutational tree, it is just as likely that passenger mutations can provide the type of target coverage desired for an effective neoantigen-focused immunotherapy, in light of their greater number.
The wide variety in mutational burden among different cancers, however, may limit the number of instances in which this concept can be meaningfully tested. Cancers at extreme ends of the mutational burden spectrum may be less amenable, since those with a low mutational load will provide few neoantigen targets, while those with a high mutational load will have too many to test.
Given the personalized nature of a neoantigen-based vaccine, this strategy might be best employed when some cancer remains after prior treatment or in successfully treated cancers with a high rate of recurrence. More-aggressive therapies, such as the delivery of cancer-fighting T cells and checkpoint inhibitors to take the brakes off the immune system, will likely remain the better option for those with advanced disease and a high tumor burden. Nonetheless, it is unlikely that any monotherapy will be as effective as a combination. The pairing of two or more of these approaches could prove to be a synergistic intervention—one that provides a durable treatment benefit for the majority of cancer patients.
Stephen P. Schoenberger is a researcher at the La Jolla Institute for Allergy and Immunology (LIAI). Ezra Cohen is a medical oncologist at the University of California, San Diego (UCSD), where he is also associate director at the Moores Cancer Center. Both researchers are codirectors of the LIAI-UCSD Center for Precision Immunotherapy.
- M. Greaves, “Evolutionary determinants of cancer,” Cancer Discov, 5:806-20, 2015.
- M.S. Lawrence et al., “Mutational heterogeneity in cancer and the search for new cancer-associated genes,” Nature, 499:214-18, 2013.
- P.G. Coulie et al., “Tumour antigens recognized by T lymphocytes: At the core of cancer immunotherapy,” Nat Rev Cancer, 14:135-46, 2014.
- T.N. Schumacher, R.D. Schreiber, “Neoantigens in cancer immunotherapy,” Science, 348:69-74, 2015.
- N.A. Rizvi et al., “Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer,” Science, 348:124-28, 2015.
- M. Yadav et al., “Predicting immunogenic tumour mutations by combining mass spectrometry and exome sequencing,” Nature, 515:572-76, 2014.
- G.T. Motz, G. Coukos, “Deciphering and reversing tumor immune suppression,” Immunity, 39:61-73, 2013.
- N. McGranahan et al., “Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade,” Science, 351:1463-69, 2016.
- B.M. Carreno et al., “A dendritic cell vaccine increases the breadth and diversity of melanoma neoantigen-specific T cells,” Science, 348:803-08, 2015.
- L.M. Kranz et al., “Systemic RNA delivery to dendritic cells exploits antiviral defence for cancer immunotherapy,” Nature, 534:396-401, 2016.