Cancer and the Cellular Immune Response

Dendritic cells are specialized antigen-presenting cells, which display an extraordinary capacity to stimulate naïve T cells and initiate primary immune responses. ^11,12 This established function of DCs has now offered the hope to apply DC-based immunotherapy for cancers. Recent studies suggest that DCs also play critical roles in the induction of peripheral immunological tolerance, regulate the types of T-cell immune responses, and function as effector cells in innate immunity against microbes. In both humans and mice, DCs can be grouped into two major subsets: the myeloid conventional DCs (mDC) and the plasmacytoid DCs (pDCs). Both display functional plasticity depending on the types of activation signals and also their resident microenvironments. However, mDCs and pDCs display different sets of TLRs and appear to regulate innate and adaptive immunity in different ways (Figure 51-2 ).

In humans, mDCs can be identified by CD4+CD11c+lineage and pDCs can be identified by CD4+CD11c-lineage-BDCA-2+ILT7+. Strikingly, whereas mDCs express TLR2, 3, 4, 5, 6 and 8, pDCs express only TLR7 and TLR9. In response to the microbial ligands for TLR2, 3, 4, 5, 6 and 8, mDCs produce the TH1 polarizing cytokine IL-12 and cytokines IL-1, IL-6, IL-10, and TNF-α and undergo maturation by upregulation of MHC class I/class II and co-stimulatory cytokines CD80, CD83, and CD86. ^10–12 In response to microbial ligands for the TLR7 and TLR9 and viral infection, pDCs rapidly produce massive amounts of type 1 IFNs, including IFN-α, IFN-β, and IFN-ω. Therefore, pDCs are also known as specialized type 1 IFN-producing cells (IPC), which represent the key cell type in antiviral innate immunity. pDCs also have the ability to produce IL-6 and TNF-α on activation through TLR7 and TLR9 and undergo differentiation into mature DCs that express high MHC class I/class II and co-stimulatory molecules CD80/CD86 and acquire the ability to prime naïve T-cell activation. ^10–12

In humans, DCs were found to display different effector functions in directing T-cell responses that are regulated by the maturation stage of DCs and the maturation signals. ^11,12 Whereas mDCs at the mature stage induce T_H1 differentiation and strong cytotoxic T lymphocyte (CTL) responses, mDCs at the immature stage induce IL-10–producing CD4⁺ and CD8⁺ regulatory T cells. Two groups of signals were shown to stimulate immature mDCs to induce T_H1 differentiation: (1) LPS derived from gram-negative bacteria (TLR4-L), gram-positive bacteria Staphylococcus aureus (SAC) (may trigger multiple TLRs), and double-stranded viral RNA (TLR3-L); and (2) T-cell signals such as CD40L and IFN-γ. Several signals were shown to stimulate immature mDC to induce T_H2 differentiation, including epithelial cell–derived cytokine TSLP and helminth Schistosoma mansoni egg antigen. ^17,18 pDC-derived mature DCs also display different effector functions depending on the types of differentiation factors. Whereas pDCs activated by IL-3 and CD40L preferentially promote T_H2 differentiation, pDCs activated through TLR7 or TLR9 prime naïve T cells to produce IFN-γ and IL-10. ¹²

A major class of adjuvants for vaccines used in humans or in experimental animal models are killed microbials or microbial-derived products that trigger different TLRs. ¹⁹ The current understanding of TLR biology and DC biology reveals that the immune system has been evolved to fight against microbial pathogens, but does not have an optimal system for sensing and effectively responding to cancer. Therefore, a basic principle of developing a cancer vaccine is to instruct DCs to recognize tumor antigens as foreign by introducing microbial-derived adjuvants together with tumor antigens (Figure 51-3 ). In animal models, stimulation of TLR9 mainly expressed on plasmacytoid DCs with CpGs has been shown to increase the immunogenicity of different forms of cancer vaccines, including peptide vaccines, DNA vaccines, tumor cell–based vaccines, and DC-based vaccines. ¹⁹ Stimulation of TLR4 (mainly expressed by mDCs) by MPL, BCG, or murine β-defensin can also promote the immunogenicity of cancer vaccines. Several studies suggest that the ability of mDCs to present antigens and activate antigen-specific T cells can be greatly enhanced by activated pDCs through a type-1 IFN-dependent mechanism in both antiviral immune responses and autoimmune responses. ²⁰ We have shown that pDCs activated by CpG promote the ability of mDCs to present melanoma antigens to T cells and induce strong tumor-specific CTL responses in vitro and in vivo. In addition, pDCs activated by CpG also strongly activate NK cells, which kill tumor cells and further enhance the ability of mDCs to take up dead tumor cells and cross-present tumor antigens to CD8⁺ T cells. ²¹

Over the past 10 years, numerous tumor antigens have been described that can be recognized by T cells. ²² These have been identified by two major methods: (1) molecular cloning using tumor antigen–specific T cells derived from cancer patients, and (2) analysis of candidate antigens based on gene expression and molecular profiling of tumors. Many of these antigens are expressed on normal tissues and are therefore considered to be “self” antigens. Examples of this class of antigen include melanocyte differentiation antigens that are expressed on melanoma cells as well as normal melanocytes. These include tyrosinase, MART-1, gp100, and TRP-1. ²² Differentiation antigens, expressed on tumor as well as the normal tissue of origin, can be targeted in immunotherapeutic strategies, as long as the normal tissue is nonessential. Mutated antigens, endogenous retroviral antigens, and antigens expressed in tumor and testis have also been described to be expressed in the context of MHC class I and class II molecules, capable of being recognized by CD8⁺ and CD4⁺ T cells, respectively. Although mutated antigens may be more easily recognized than “self” antigens, it is also possible that peripheral tolerance develops against the mutations, because co-stimulation may be absent in tumors, which are often present for many years before clinical diagnosis.

Because many tumor antigens are also expressed by normal host tissues, such as normal melanocytes in the case of melanoma antigens, obtaining an effective antitumor immune response requires overcoming self-tolerance. Indeed, in some settings, effective immunotherapy has been correlated with autoimmune responses against host tissues. IL-2, a cytokine that can stimulate the proliferation of T cells, can result in significant long-lasting regression of disease in some patients with metastatic melanoma and renal cell cancer (see later discussion). Interestingly, Rosenberg found that whereas 0 of 104 renal cancer patients treated with high-dose IL-2 developed vitiligo, the immune destruction of normal melanocytes, 11 of 74 melanoma patients treated with high-dose IL-2 developed vitiligo. ²³ Furthermore, vitiligo was seen in 26% of melanoma patients who demonstrated objective tumor response to IL-2, whereas no vitiligo was observed in patients who did not respond to the IL-2. This suggests that tolerance to self-antigens can be overcome in the induction of an effective antitumor immune response.

Perhaps the strongest evidence that immune responses can result in a significant antitumor effect in patients is the fact that a subset of patients with metastatic melanoma can have complete tumor regressions and long-term survival following the administration of the T-cell growth factor IL-2. Of 270 patients with metastatic melanoma treated with high-dose IL-2, 43 (16%) had an objective response (complete or partial response). ²⁴ More importantly, 60% of those patients who achieved a complete response had prolonged disease-free intervals and long-term survival (more than 10 years). This demonstrates that it is possible, by activation of the immune system, to induce clinically meaningful responses in patients. Future studies are needed to focus on understanding more fully the mechanism of response in patients so that improved strategies can be designed that will result in higher response rates and improved survival.

With clear evidence that the immune system can play an important role in mediating clinical responses in cancer patients, current efforts are focused on developing cancer vaccines in order to enhance efficacy as well as specificity, because nonspecific immune stimulation can result in autoimmunity, as discussed. Two major approaches have been pursued in cancer vaccine strategies: the use of whole tumor cells or the use of specific tumor antigens (Table 51-1 ).

Cancer vaccine strategies have been performed using derivatives of both autologous and allogeneic tumor cells. Although the use of autologous tumor cells is more labor intensive, in that vaccines need to be prepared individually for each patient, autologous tumor has the advantage of containing specific mutations for that patient, which may be seen as more foreign compared to shared self antigens. Cell lysates fed to autologous dendritic cells, isolation of heat shock proteins bound to autologous antigens, and gene modification of autologous tumor with immune-enhancing cytokines have been evaluated in clinical trials. In murine models, transduction of tumor cells to express GM-CSF results in enhanced antitumor immune responses against parental non-transduced tumor cells. ²⁷ Antitumor activity of GM-CSF–expressing tumors was found to be dependent on host bone-marrow–derived antigen-presenting cells, CD1d-restricted NK T cells, CD4⁺ and CD8⁺ T cells, and antibodies. ^28,29

As discussed earlier, numerous tumor antigens have now been identified that can be recognized by T cells in the context of MHC class I and class II molecules. Many of these antigens are shared among specific types of tumors, and represent a feasible target for vaccine development. Antigen-specific vaccine approaches have included the use of specific peptide epitopes, whole proteins, and recombinant DNA and viral vaccines. The advantage of whole protein and recombinant approaches using the entire antigen gene is that multiple class I and class II epitopes may be presented. ³⁰ However, whole proteins have been expensive and challenging to produce clinically, and viral vaccines can induce neutralizing antibodies that prevent the efficacy of serial doses of vaccine. Peptide vaccines, in combination with specific adjuvants, have demonstrated potential clinical efficacy in the case of chronic myelogenous leukemia and have been the most consistent method of inducing high levels of circulating antigen-specific T cells. ^31,32 However, in the case of patients with metastatic melanoma, high levels of tumor-specific T cells in the blood do not always result in tumor regression. ³³ Therefore, future efforts are focused on stimulating stronger and more effective T-cell priming through the use of specific adjuvants such as TLR agonists, using concepts learned from basic studies of antiviral immunity as described earlier. This may result in T cells with increased affinity and specificity as well as enhanced memory and effector function. One method to more carefully manipulate the specific phenotype of tumor-reactive immune cells is to generate and select specific T cells in the laboratory followed by reinfusion into patients. This is termed adoptive immunotherapy.

In the setting of metastatic melanoma, T cells can be found at the tumor site (tumor-infiltrating lymphocytes, or TILs) that are specific for melanoma antigens, such as MART-1 and gp100 (Figure 51-4 ). Because the tumor is growing, either these T cells are nonfunctional or the tumor is resistant to recognition or lysis. However, when TILs are expanded ex vivo and reinfused, clinical regressions are seen in patients with metastatic disease. A number of reasons may explain the effectiveness of ex vivo expanded lymphocytes compared to endogenous T cells. First, large numbers can be generated in the laboratory that may be difficult to achieve in vivo. Second, expanding the lymphocytes ex vivo takes them out of the suppressive tumor microenvironment. Finally, growth ex vivo may allow reactivation of lymphocytes rendered anergic or nonfunctional by in vivo toleragenic mechanisms. Initially, transfer of tumor-reactive T cells alone resulted in response rates of greater than 20%, but clinical regressions were often transient, and it was clear from gene marking studies that the T cells did not survive long in vivo. ^36,37

Because tumor-reactive lymphocytes are naturally found in melanomas, much of the work in adoptive therapy has focused on this disease. Although tumor-reactive lymphocytes are rarely found in other common cancers, antibodies that recognize these tumors in a relatively specific fashion have been described. Therefore, chimeric receptors have been designed using antibody variable regions extracellularly fused to T-cell signaling chains intracellularly (Figure 51-5 ). The initial studies demonstrated the ability to redirect T-cell specificity in vitro against ovarian cancer. ⁴² Lysis of ovarian cancer cells by human lymphocytes redirected with a chimeric gene composed of an antibody variable region and the Fc receptor gamma chain, but subsequent receptors have been designed that recognize human immunodeficiency virus (HIV) as well as a number of other tumor types. ^43–45

CD4⁺D25⁺, naturally occurring regulatory T cells (Treg), constitute 5% to 10% of peripheral CD4⁺ T cells, which play an essential role in the active suppression of autoimmunity in both humans and rodents. Treg appear to differentiate as a unique T-cell lineage in the thymus from immature T cells expressing T-cell receptor (TCR) with medium to high affinity for self-antigens, which depends on IL-2 and co-stimulatory molecules provided by activated APCs. ^48–52 Foxp3, a member of the forkhead transcriptional factor family, has been demonstrated to be the master regulator of Treg development in the thymus, as well as Treg suppressive function. Increasing evidence suggests that tumor-specific Treg exist and play an essential role in immune tolerance to tumors and thus represent a major hurdle for antitumor immunotherapy. In addition, the tumor microenvironment appears to be the site where tumor-specific infiltrating T cells are actively converted into tumor-specific Treg_. ⁵³

There is accumulating evidence that progressive tumor growth is correlated with an increased frequency of immature myeloid cells (iMC) and immature DCs (iDC) in the tumor microenvironment, which can inhibit the function of tumor-specific T lymphocytes. ^64–66 These immature myeloid cells can be induced by tumor-derived factors such as vascular endothelial growth factor (VEGF), macrophage colony-stimulating factor, IL-6, granulocyte-macrophage colony-stimulating factor, IL-10, and gangliosides. The proangiogenic cytokine VEGF is a major contributor to immune suppression, and tumors often produce large quantities of VEGF that can be detected in the serum of cancer patients. ^67,68 In vitro studies have shown that VEGF can block DC maturation, leading to the production of iMCs; conversely, blocking VEGF promotes normal DC differentiation and enhances antitumor function. ⁶⁹

CTLA-4 is expressed by activated CD4 and CD8 T cells. It is a homologue of T-cell co-stimulator CD28 but has a higher binding affinity for its ligands. On T-cell activation, signaling pathways lead to the expression of CTLA-4, which is then mobilized from intracellular vesicles to the cell surface, where it outcompetes co-stimulator CD28 for binding to its ligands. Binding of CTLA-4 to CD86 proteins interrupts CD28 co-stimulatory signals and, as a result, limits T-cell responses (see Figure 51-6). In addition to the intrinsic restriction of effector T-cell responses due to expression of CTLA-4 on effector cells, there can also be extrinsic restriction of effector T-cell responses due to the expression of CTLA-4 on regulatory T cells as well. ^87–89

Another important T-cell checkpoint that can be targeted clinically is the interaction of PD-1 and its ligands (see Figure 51-6). PD-1 is mainly expressed by activated CD4 and CD8 T cells, and it has two ligands, PD-L1 and PD-L2, with distinct expression profiles. ⁹⁵ PD-L1 is expressed not only on APCs, but also on T cells, B cells, and nonhematopoietic cells, including tumor cells. Expression of PD-L2 is largely restricted to APCs, including macrophages and myeloid DCs. The role of PD-1 as a negative regulator of T and B cells was best demonstrated by the findings that PD-1–deficient mice developed significant autoimmunity. ^96,97 Subsequently, blocking antibodies against PD-1 were shown to induce a reduction of tumor growth and metastasis in a number of experimental mouse models. ^98,99 Consistent with the immune inhibitory role of PD-1/PD-L1/2 signaling, enforced expression of PD-L1 on tumor cells caused enhanced tumor growth in vivo, which could otherwise be kept in check by T cells. Again, this augmentation of tumor growth could be reversed with the use of blocking antibodies against PD-L1. ¹⁰⁰

As with most experimental cancer therapies, treatment with immune checkpoint blockade agents is sometimes associated with a distinct set of on-target toxicities due to immune-related adverse events. For example, among melanoma patients treated with ipilimumab, up to 60% of patients manifested autoimmune reactions including dermatitis, colitis, hepatitis, and hypopituitarism. ⁹¹ Early recognition, diagnosis, and treatment of such drug-induced inflammatory conditions with steroids are critical for minimizing the adverse effects while maximizing the therapeutic benefits of this promising anticancer agent. By contrast, the early Phase I results targeting the PD-1 pathway have shown less overall toxicity in cancer patients, although future studies are required to confirm this. Regardless, investigations into biomarkers/pathways that are associated with or predict clinical benefit or toxicities will be important as the field of cancer immunotherapy moves forward, and recent studies in these areas have been promising. ^103,104 Perhaps future strategies combining checkpoint blockade strategies with an effective cancer vaccine may skew the immune response toward tumor-specific antigens and away from normal tissue-associated self-antigens.