The first cancer antigen to be characterized at a genetic and molecular level and recognized by T cells was MAGE-1 [1]. Since then, hundreds of peptides derived from tumors have been identified and shown to be expressed by solid tumors of various histologies, and restricted to presentation on different subclasses of MHC molecules [2]. Tumor-associated antigens fall into several major categories: (1) overexpressed normal proteins (eg, carcinoembryonic antigen [CEA] or nonmutated p53); (2) nonmutated differentiation antigens (eg, MART-1, overexpressed in melanoma and found in normal melanocytes); (3) cancer-testis antigens (CTA), consisting of nonmutated genes expressed during fetal development, then silent in normal adult tissues and reactivated in cancer cells across multiple malignancies (eg, MAGE and NY-ESO); and (4) mutated antigens, unique to a single tumor or shared by a group of tumors (eg, BRAF with the V600E mutation in melanoma and other solid tumors, or EGFRvIII in glioblastoma).

Despite the fact that tumor-associated antigens recognized by T cells have been described, solid cancers in humans grow and disseminate in immune competent hosts. Two main reasons, not mutually exclusive, explain this reality: (1) most cancers are weakly or not immunogenic, hence the frequency of tumor-reactive lymphocytes is low or null in most patients; and (2) immunologic mechanisms of T-cell anergy or tolerance, or immunosuppressive factors, either systemic or in the tumor microenvironment, thwart antitumor immune reactions. Box 1 summarizes general mechanisms described to result in cancer progression despite a competent immune system. Because cancer antigens are commonly nonmutated self-proteins, the naturally occurring pool of T cells able to recognize those self-antigens do so with low avidity, otherwise they would have been deleted during negative selection in the thymus during development. If a T cell is capable of reacting to a self-antigen in tissues and tumors, mechanisms involved in the prevention of autoimmunity are at play and referred to as peripheral tolerance. The fate of T cells able to strongly recognize altered self-antigens, such as mutated cancer antigens, is less clear. However, because the tumor can participate in immune suppression and tolerance at the tumor site in multiple ways, anticancer cytotoxic T cells are expected to lose cytotoxic functions and proliferative capacity, and may be driven to apoptosis. The lack of appropriate costimulation signals provided to naïve T cells by antigen-presenting cells found in an immature or inactivated state has also been proposed as a mechanism to explain the poor lytic and proliferative capacity of T cells on second encounters with tumor antigens.

Box 1 General mechanisms that can inhibit antitumor immune reaction in cancer patients

Immune factors

Low or null frequency of antitumor T cells in vivo

Low-affinity recognition of self-antigens expressed on tumor cells

Chronic exposure to tumor antigens with low immunogenicity

Absence of appropriate T-cell priming by immature or nonactivated antigen-presenting cells

Immunosuppression of antitumor T cells by regulatory cells (eg, regulatory T cells, cells derived from the myeloid lineage)

Tumor factors

Secretion of immunosuppressive molecules (eg, transforming growth factor β, interleukin-10, arginase-1, and nitric oxide synthase 2)

Cell surface expression of immunosuppressive ligand (eg, PD-L1)

Competition for molecules essential for T-cell metabolism and proliferative advantage (eg, glucose and tryptophan)

Downregulation of major histocompatibility complex molecules

Downmodulation of proteins involved in antigen processing and presentation

Heterogeneity of antigen expression in transformed cells constituting the tumor mass

Multiple tumor mechanisms responsible for T-cell inhibition have been described, and include the secretion of soluble molecules able to suppress T-cell proliferation and functions (eg, transforming growth factor-β [TGF-β], interleukin-10, arginase-1, nitric oxide synthase 2), the competition for molecules essential for T-cell metabolism (eg, glucose, tryptophan), and the expression of surface molecules that inhibit immune cell activation (eg, programmed death ligand 1 [PD-L1, also called B7H1]) [3]. In addition, cancer cells can go unrecognized by T cells simply by downregulation of MHC molecules and by downmodulation of proteins involved in tumor antigen processing and presentation at the cell surface (eg, transporter associated with antigen processing 1 [TAP1]). The tumor microenvironment is also enriched in immunosuppressive cells, such as regulatory T cells (Treg) [4]. Recent studies also suggest that inhibition of T cells can be mediated by innate immune cells, granulocytic, monocytic, or their precursors, and are now generally referred to as myeloid-derived suppressor cells (MDSC) [5,6]. The potency of these regulatory immune cell subsets to impair antitumor cytotoxicity by T cells has been established mainly in mouse models, and the specific mechanisms by which T-cell inhibition occurs remains to be elucidated. In humans, in vitro assays have suggested the existence of Treg and MDSC in cancer patients, but in vivo studies are in their infancy. The heterogeneity of the transformed cells that constitute a tumor mass also contributes to tumor progression, because many cancer cells may go unrecognized by the immune system while partial tumor destruction by T cells occurs. This “natural selection” of less immunogenic tumor cells over time has been coined “tumor immunoediting” and is mainly supported in animal models [7,8].

Despite all mechanisms able to inhibit antitumor immune reaction and the uncertainty of naturally occurring in vivo immune responses to solid cancers, T-cell–mediated immunity plays the predominant role in mediating the rejection of tumors in preclinical animal models. The general goal of cancer immunotherapy is thus to provide an adequate number and enhance the function of antitumor cytotoxic T cells while overcoming immune suppression and tolerance at the tumor site in cancer patients.

Human cancer immunotherapies can be categorized into 3 major approaches: (1) nonspecific immunomodulation, (2) active immunization (cancer vaccines), and (3) passive transfer of activated immune cells with antitumor activity, called adoptive immunotherapy or cell transfer. The first 2 strategies attempt to enhance previously existing in situ anticancer immunity. Adoptive cell transfer aims at providing the quality and quantity of anticancer T cells that are lacking in vivo by using ex vivo manipulations and immune preconditioning of patients prior to treatment.

Nonspecific immunotherapy

Nonspecific immune modulation aims at promoting tumor rejection through stimulation of effector T cells or blockade of regulatory factors that inhibit T-cell function.

Stimulation of effector T cells

The T-cell growth factor interleukin (IL)-2 can activate endogenous tumor-reactive cells, and reproducibly mediate the regression of advanced metastatic melanoma and kidney cancer. In a consecutive series of 409 patients treated at the Surgery Branch of the National Cancer Institute (NCI), between 1985 and 1996, high-dose bolus intravenous IL-2 administration produced complete and durable regressions of metastatic melanoma and renal cell cancer in 6.6% and 9.3% of patients, respectively [9,10]. Responses were seen at all sites of disease, and more than 80% of complete responses appeared durable and were ongoing after a median follow-up of 7 years at the time of publication. IL-2 alone does not appear sufficient to induce regression of other solid cancers. The US Food and Drug Administration (FDA) approved the use of IL-2 for the treatment of metastatic renal cancer in 1992 and metastatic melanoma in 1998, based on the ability of IL-2 to mediate durable complete responses. For the same reason, IL-2 therapy should be the first-line treatment for patients with metastatic renal cancer and metastatic melanoma.

Experience with the administration of IL-2 has resulted in treatment-related mortalities of less than 1%. Toxicities occur owing to a capillary leak syndrome, and can be safely treated with appropriate monitoring and judicious fluid resuscitation [11]. Organ-specific autoimmunity seen in melanoma patients treated with IL-2, such as delayed-onset vitiligo in approximately 20% of patients and the development of autoimmune thyroiditis in approximately 55%, has been associated with the likelihood of cancer regression [12]. Of interest, patients with renal cell carcinoma do not develop vitiligo with IL-2, and autoimmune thyroid is seen less commonly in these patients, arguing for activation of a different subset of T cells in melanoma patients, able to recognize antigens expressed on tumor and on normal melanocytes (mainly MART-1, gp100, and Tyrosinase) [10].

Interferon-α2b (IFN-α2b), an important mediator of antiviral immunity, has also been used for the treatment of patients with melanoma and renal cell cancer. The role of IFN-α2b for the adjuvant treatment of patients at high risk of recurrence after definitive surgery is controversial, since improvements in recurrence-free survival have translated into little impact on overall survival. Long-term administration of a pegylated formulation, expected to maintain maximum exposure to IFN-α2b with less frequent subcutaneous injections than with the unpegylated formulation, has recently led to similar results in a large randomized control trial of patients with node-positive melanoma (stage III) [13,14]. In this trial, the relapse-free survival was improved by 9.3 months (34.8 months vs 25.5 months), without benefit in overall survival. The FDA approved pegylated IFN-α2b (Sylatron) on March 29, 2011 for the treatment of melanoma patients with microscopic or gross nodal involvement.

Blockade of negative regulators of T-cell function

Activated T cells express surface inhibitory molecules that, when bound by their ligands, are capable of inhibiting T-cell activity. Monoclonal antibodies directed against these surface inhibitory molecules have been studied as immunotherapeutic regents. The cytotoxic T-lymphocyte–associated 4 molecule (CTLA-4, CD152), part of the immunoglobulin-like family of surface proteins, is one of the inhibitory molecules expressed at the surface of activated T cells. Two fully human IgG monoclonal antibodies recognizing CTLA-4, ipilimumab (MDX-010) and tremelimumab (CP-675,206), have been tested, alone or in combination, in phase 2/3 trials. These antibodies are designed to prevent the binding of CTLA-4 to its ligand B7, mainly expressed on immature antigen-presenting cells and tumor cells [15].

The first demonstration of the ability of ipilimumab to mediate tumor regression in 2003 reported objective regressions in 3 of 13 patients with metastatic melanoma [16]. An updated summary of consecutive cohorts of 179 patients with metastatic melanoma revealed an overall response rate ranging from 13% to 25%, including 6% to 17% durable complete responses [17]. The highest response rates were seen in patients given ipilimumab in combination with high-dose IL-2. In this cohort, 6 of 36 (17%) patients enjoy ongoing complete responses after 7 years of median follow-up. It is interesting that prior response to IL-2 was not correlated with the likelihood of response to ipilimumab, pointing to a different quality of interaction between melanoma and the host immune system using these two agents alone. Important but delayed responses to treatment were observed in some patients.

The effectiveness of ipilimumab as second-line treatment for patients with advanced melanoma has now been confirmed in a large double-blinded, randomized, multi-institutional phase 3 trial [19]. A total of 676 HLA-A*0201 patients were randomized 3:1:1 to receive ipilimumab plus a gp100 vaccine, ipilimumab alone, or the gp100 vaccine alone. The overall median survival was equivalent in both groups of patients who received ipilimumab, approximately 10 months versus 6.4 months for the group who received the gp100 vaccine alone. Objective responses were seen in 38 patients among the 540 who received ipilimumab (7%), with 3 complete responses (0.5%). Immune-related side effects were seen at comparable rates to what has been described above, and 4 patients died following bowel perforation, due to ipilimumab-induced colitis. The addition of the vaccine did not confer a survival benefit or lead to unexpected side effects. Based on the results of this trial, the FDA approved ipilimumab (Yervoy) for the treatment of unresectable or metastatic melanoma on March 25, 2011. Ongoing trials are now assessing the efficacy anti–CTLA-4 in combination with other biological and cytotoxic agents, notably dacarbazine [20].

The efficacy of ipilimumab has been tested for other solid malignancies. In a nonrandomized phase 2 study of 40 patients with metastatic renal cell cancer treated with ipilimumab alone, 5 patients had a partial response [21]. A single or 2 doses of ipilimumab has been associated with a decrease of 50% or more of the prostate-specific antigen (PSA) level in 2 of 12 patients with metastatic hormone-refractory prostate cancer patients [22]. Of 27 patients with unresectable or metastatic pancreatic adenocarcinoma treated with ipilimumab, only one experienced a mixed response [23].

Tremelimumab, the second fully humanized anti–CTLA-4 monoclonal antibody, has been less studied than ipilimumab. In the most recent and largest phase 2 trial in patients with advanced melanoma treated with tremelimumab, a response rate of 6.6% in 246 patients was reported, without complete responses, 2 treatment-related deaths, and a median survival of 10 months [24]. In heavily pretreated metastatic colorectal cancer patients, 1 of 45 partial but sustained response was reported [25].

Programmed death 1 (PD-1, CD279) is another inhibitory molecule expressed at the surface of activated T cells after repeated encounter with antigen. It belongs to the same family of surface molecules as CTLA-4, but binds different ligands and provides distinct intracellular signaling that leads to shutdown of T-cell effector function. Tumor cells and antigen-presenting cells found in tumors can express a high level of PD-L1, the main PD-1 ligand (also called B7-H1), and this has been associated with poor prognosis in renal and ovarian cancer [26,27]. Interrupting the interaction between PD-1 and its ligand using the fully human IgG4 monoclonal antibody MDX-1106 has been reported to mediate cancer regression in a phase 1, dose-escalation trial [28]. Objective response was documented in 3 of 39 patients, one each with melanoma, renal cancer, and colorectal cancer.