Special Reviews

Cancer Immunotherapy: A Treatment for the Masses

See allHide authors and affiliations

Science  09 Jul 2004:
Vol. 305, Issue 5681, pp. 200-205
DOI: 10.1126/science.1100369


Cancer immunotherapy attempts to harness the exquisite power and specificity of the immune system for the treatment of malignancy. Although cancer cells are less immunogenic than pathogens, the immune system is clearly capable of recognizing and eliminating tumor cells. However, tumors frequently interfere with the development and function of immune responses. Thus, the challenge for immunotherapy is to use advances in cellular and molecular immunology to develop strategies that effectively and safely augment antitumor responses.

Ambrose Bierce's cynical description of the evolving field of medicine as “a stone flung down the Bowery to kill a dog on Broadway” often aptly describes conventional cancer therapies. By contrast, the immune system has evolved strategies, largely in response to infections, to efficiently search for and specifically destroy diseased targets. After nearly a century of debate as to whether the immune system can actually target tumors (13), compelling evidence now suggests that immune cells play an important role in the control of malignancy (4). This has first been implied by both occasional spontaneous regressions of cancers in immunocompetent hosts and increased cancer incidence in immunocompromised individuals. Second, tumor immunity can be demonstrated in experimental animal models. For example, mice with defined immunological defects exhibit greater susceptibility to spontaneous and induced tumors, with many of these tumors rejected if transplanted into normal hosts (4, 5). Third, the immune system often appears cognizant of tumors, as reflected by an accumulation of immune cells at tumor sites, which correlates with improved prognosis (6). Finally, with improved technologies, antitumor immune responses can now be detected directly from many patients. Augmenting these responses has started to yield therapeutic benefits not only in experimental models but also in cancer patients. Advances in cellular and molecular immunology in the past two decades have provided enormous insights into the nature and consequences of interactions between tumors and immune cells and continue to suggest strategies by which the immune system might be harnessed for therapy of established malignancies.

Cells of the innate immune system respond to “danger” signals, which can be provided by growing tumors as a consequence of the genotoxic stress of cell transformation and disruption of the surrounding microenvironment. Under ideal conditions, these signals will induce inflammation, activate innate effector cells with antitumor activity, and stimulate professional antigen-presenting cells (APCs), particularly dendritic cells (DCs), to engulf tumor-derived antigens and migrate to draining lymph nodes to trigger an adaptive response by T and B lymphocytes. Despite this well-orchestrated surveillance operation, the presence of a tumor indicates that the developing cancer was able to avoid detection or to escape or overwhelm the immune response. Progressing tumors often exhibit strategies that promote evasion from immune recognition (7), such as physical exclusion of immune cells from tumor sites, poor immunogenicity due to reduced expression of major histocompatibility complex (MHC) or costimulatory proteins, and disruption of natural killer (NK) and natural killer T (NKT) cell recognition (8). Additionally, some tumors prevent triggering of an inflammatory response by secreting proteins, such as interleukin (IL)–10 or vascular endothelial growth factor (VEGF) that interfere with DC activation and differentiation (9) or by blocking the production of pro-inflammatory molecules by increasing expression of the STAT3 protein (10) (Fig. 1A). Even if a response is induced, tumor cells may escape elimination by losing targeted antigens, rendering tumor-reactive T cells anergic, inducing regulatory T cells, or specifically deleting responding T cells (1114). Thus, there is often a cat and mouse game with the immune system exerting pressure to eliminate the tumor, and the tumor cells evading the immune response; the eventual tumor that develops reflects “immunoediting” with selection of poorly immunogenic and/or immune-resistant malignant cells (5). Despite these obstacles, modern immunebased therapies continue to show increased potential for treating malignant diseases.

Fig. 1.

Manipulating the innate immune response. (A) Tumor cells can avoid activating innate responses by producing inhibitory cytokines and down-regulating or secreting ligands for activating receptors. Mϕ, macrophage; TCR, T cell receptor. (B) Activation of innate responses can be enhanced by administering adjuvants, ligands for costimulatory proteins, cytokines, or drugs that directly trigger innate immune cells. αGalCer, α-galactosylceramide.

The Innate Immune System

Cells of the innate immune system have not only the responsibility of surveying and “informing” the host of a breach in integrity, but also intrinsic antitumor effector functions including lysis of tumor cells and the production of cytokines that inhibit tumor growth or block angiogenesis. NK, NKT, and γδ T cells express activating receptors such as NKG2D that recognize MHC class I chain–related (MIC) or ULI6-binding (ULBP) proteins that become up-regulated on tumor cells. Some NKT cells and γδ T cells also express T cell receptors with restricted diversity that recognize tumors through lipid or protein antigens presented in the context of MHC proteins (Fig. 1B).

Harnessing innate immune cells as effectors in therapy. NK, NKT, and γδ T cells produce interferon γ (IFNγ) after activation. IFNγ has tumoricidal activity and induces other cells of the innate immune system, including macrophages and DCs, to produce IL-12, which further activates cells mediating the innate response. The importance of this feedback activation pathway is highlighted by the increased susceptibility of IL-12–deficient mice to experimentally induced tumors and by evidence that NK and γδ T cells eliminate spontaneously appearing lymphomas (15). Although administering IL-12 systemically appeared promising in preclinical murine models, this approach proved unacceptable in humans because of liver toxicity. An alternative strategy to harness innate effector cells has been to expand and activate NK cells in vitro by culture with IL-2, followed by infusion of large numbers of these NK cells back into patients alone or with high doses of IL-2. This approach, or administration of high doses of IL-2 to expand and activate NK cells entirely in vivo, has yielded marked antitumor activity and complete remissions in a subset of patients (16). However, life-threatening toxicity often develops, largely due to the release of tumor necrosis factor (TNF) from activated NK cells. Separating antitumor activity from such toxicity has not proven feasible, relegating this approach to treatment of only the most responsive tumors, such as renal cell cancer. The selective activation of NKT cells by triggering in vivo with the glycolipid alpha-galactosylceramide or of γδ T cells by administration of bisphosphonate compounds is also being pursued, with encouraging results in preliminary trials (17, 18). Again, integrating these approaches into a coordinated treatment plan with acceptable toxicities will require further study.

An alternative to globally activating innate effector cells is to systemically provide effector molecules these cells produce, such as TNFα. Currently, systemic toxicity limits this approach to the treatment of tumors in which the blood supply can be isolated (19), and broader application awaits methods to link the activity of the cytokine with more specific targeting of the tumor.

Innate cells as initiators of the adaptive immune response. One of the first strategies to enhance immune responses to cancer was the administration of adjuvants directly into solid tumors to stimulate inflammation and recruit immune effector cells. This approach is still commonly used for treatment of superficial bladder carcinomas and has been used to treat melanoma and neurological tumors. It is now known that many of these adjuvants contain bacterial products such as lipopolysaccharide (LPS) or CpG-containing oligodeoxynucleotides recognized by toll-like receptors (TLRs) on innate immune cells, leading to the production of pro-inflammatory cytokines and facilitating productive interactions between the innate and adaptive immune responses (20). However, many tumors render this strategy ineffective by producing proteins such as transforming growth factor (TGF)–β to prevent activation of the immune response (21).

Insights into cellular and molecular events that lead to recruitment and activation of immune cells suggest that obstacles present at tumor sites might be bypassed and tumor immunity initiated by providing pro-inflammatory cytokines and/or chemokines at sites of solid tumors (22). Our knowledge of how to harness therapeutic chemoattractants and activators is still rudimentary, but expression of molecules such as secondary lymphoid-tissue chemokine or the TNF superfamily member LIGHT in tumor sites has been shown to convert these microenvironments into highly immunogenic structures. Although these approaches still await development and testing in humans, systemic administration of other activators of the innate response has already shown clinical promise. IFNα directly inhibits tumor cell growth, increases APC maturation, and despite some toxicity is now used to treat selected malignancies in the clinic, such as renal cell carcinoma and chronic myelogenous leukemia (CML). In responding CML patients, CD8 T cell responses to the leukemia have also been detected.

An alternative approach for increasing the number of APCs without inducing inflammation has been to provide a ligand for the receptor tyrosine-kinase Flt-3, expressed by hematopoietic precursor cells. Treatment of cancer patients with soluble Flt-3 ligand has induced large increases in circulating DCs (23), but these were largely unactivated and failed to augment immune responses. Thus, it will likely be necessary to also provide activation signals to DCs to effectively induce adaptive responses. Systemic administration of TLR agonists or antibodies to CD40 has resulted in potent activation of DCs and tumor responses in animal models and is currently being tested clinically. Further insights into how to effectively modulate APC number and function should translate into therapeutic benefits.

Humoral Immunotherapy

B cell activation results in the production of antibodies that can bind to immunogenic cell-surface proteins on tumor cells and initiate complement-mediated cell lysis, bridge NK cells or macrophages to the tumor for antibody-dependent cell-mediated cytotoxicity (ADCC), interfere with tumor cell growth by blocking survival or inducing apoptotic signals, or increase immunogenicity by facilitating the uptake and presentation of tumor antigens by APCs. Thus, enhancing B cell responses in vivo or providing large amounts of in vitro–generated antibodies has the potential to promote antitumor activity (Fig. 2A).

Fig. 2.

Manipulating humoral immunity. (A) B cell responses can be augmented by vaccination with tumor antigens to induce antibodies that kill tumors or promote antigen presentation. Passively transferred mAbs or engineered bispecific antibodies can bind to tumors and activate effector cells. BCR/lg, B cell receptor; Ag:Ab, antigen:antibody; MAC, membrane attack complex. (B) Modified and unmodified mAbs can kill tumor cells by many mechanisms independent of recruitment of effector cells.

Passive antibody transfer. One of the first mouse monoclonal antibodies (mAbs) tested therapeutically in humans targeted the IL-2 receptor expressed by many T cell leukemias and lymphomas, as well as by rapidly proliferating normal T cells, and provided a “proof of concept” for passive antibody transfer in cancer therapy. However, many disappointments followed, as numerous obstacles such as the immunogenicity of mouse proteins and modulation of target antigens had been underestimated. With elucidation of the reasons for failure and development of technologies to molecularly modify mAbs to remove immunogenic murine sequences and to enhance efficacy, an increasing number of mAbs are now becoming components of standard treatment regimens (24). The most widely used, rituximab, binds CD20 and, if given alone or with chemotherapy, can induce high rates of remission in patients with B cell lymphomas, predominantly by a mechanism involving ADCC (25). Some mAbs can mediate antitumor activity independent of effector cells, such as by blocking essential survival signals or inducing apoptotic signals (Fig. 2B). For example, two mAbs approved for clinical use, reactive with the Her-2/Neu receptor on breast cancer cells and the epidermal growth factor receptor on epithelial tumors, provide therapeutic benefits in part by blocking growth signals. The antitumor activity of mAbs can also be enhanced by attaching radioisotopes or drugs or by engineering recombinant bispecific antibodies that simultaneously bind tumor cells and activate receptors on immune effector cells such as CD3 or FcR (24, 26). A major remaining challenge with mAb therapy will be improving antitumor activity without inducing unmanageable toxicity to normal tissues, a problem now being addressed with pretargeting strategies to permit selective accumulation of the mAb at tumor sites. However, with the successes already enjoyed by mAb therapy, this modality is certain to become an increasingly important component of our therapeutic arsenal.

Enhancing B cell responses in vivo. The efficacy of stimulating a patient's own tumor-reactive B cells may be limited by the magnitude of the antibody response that can be achieved in vivo. Nevertheless, this approach remains appealing because of demonstrations with tumor cell expression libraries that sera from a large fraction of patients already contain tumor-reactive antibodies. The simplest means to stimulate such B cells in vivo is to provide tumor antigens in immunogenic vaccine formulations, such as mixed with adjuvants, or conjugated to antigens that can elicit helper T cell responses. Marked clinical results have been observed after priming patients with autologous DCs pulsed with the unique idiotypic immunoglobulin derived from the B cell receptor of a patient's own B cell lymphoma followed by boosting with the immunoglobulin conjugated to the helper protein keyhole limpet hemocyanin (KLH). The elicited anti-idiotypic antibodies induced signaling in lymphoma cells, and the generation of such antibodies correlated with antitumor responses and prolonged remissions (27). However, this vaccine also induced T cell responses, leaving the precise role of the antibody response undefined. Alternative approaches for activating and expanding existing B cell responses in vivo by ligation of costimulatory molecules such as CD40 or by administration of the B cell proliferative cytokine IL-4 have not met with much success in preclinical models and could potentially induce hazardous autoreactive antibodies. Thus, with the limitations of current techniques, humoral therapy will likely continue to be dominated by passive administration of mAbs specific for selected tumor antigens.

Cellular Immunotherapy

T cells express clonally distributed antigen receptors that in the context of MHC proteins can recognize either unique tumor antigens, such as those evolving from mutations or viral oncogenesis, or self-antigens, such as those derived from overexpression of proteins or aberrant expression of antigens that are normally developmentally or tissue-restricted. To mediate antitumor activity, T cells must first be activated by bone marrow–derived APCs that present tumor antigens and provide essential costimulatory signals (28), migrate and gain access to the tumor microenvironment, and overcome obstacles to effective triggering posed by the tumor. Activation results in the production of cytokines such as IFNγ and TNFα that can arrest proliferation of malignant cells and prevent the angiogenesis necessary for tumor growth, and also lysis of tumor cells mediated by perforin and/or Fas. Consequently, efforts have focused on identifying tumor antigens, providing the antigens in immunogenic formats to induce responses, manipulating T cell responses to increase the number of reactive cells, and augmenting effector functions (Fig. 3).

Fig. 3.

Manipulating T cell immunity. T cell responses can be augmented by (A) stimulation with immunogenic vaccines, pro-inflammatory cytokines, or antibodies that block negative signals, (B) adoptive transfer of large numbers of tumor-reactive T cells generated in vitro, or (C) administration of cytokines. (D) T cells can be genetically modified before adoptive transfer to acquire novel receptors for tumor recognition or regulated autocrine proliferative signals, or to block inhibitory signals that limit T cell responses.

Stimulation of responses in vivo. The poor immunogenicity of most tumor antigens largely reflects the nonconducive context in which these antigens are naturally presented, as well as tolerance resulting from most tumor antigens being normal proteins aberrantly expressed by the tumor. Therapeutic vaccines have attempted to circumvent these problems by presenting tumor antigens in a more enticing fashion, generally through activated DCs. This has been achieved either by isolating DCs and introducing the antigen ex vivo before returning the DCs to the host, by inoculating dead tumor cells modified to secrete factors such as granulocyte-macrophage–colony stimulating factor (GM-CSF) that promote local accumulation of DCs, by injecting activators of DCs such as TLR ligands or mAb to CD40 with the antigen, or by injecting recombinant vectors that provide both the antigen and a stimulus to the innate immune system (29). This last category includes plasmid DNA containing the antigen and immunostimulatory CpG sequences as well as recombinant attenuated pathogens, such as adenoviruses or Listeria monocytogenes, that express the antigen and provide TLR ligands to trigger innate responses. Preclinical murine studies with such approaches have demonstrated rejection of established tumor masses, and clinical trials exploring these strategies have engendered enthusiasm from occasional marked tumor regressions and complete responses (3032). However, most vaccinated patients exhibit only weak or undetectable T cell responses to the tumor antigen and no clinical benefit. Thus, methods to maintain APC activation and sustain immunogenic antigen presentation in vivo as typically occurs during encounter with a replicating foreign pathogen will likely be required before vaccines will become more predictably beneficial.

An alternative to improving antigen presentation has been to mitigate negative checkpoint signals that limit the T cell response. Cytotoxic T lymphocyte antigen–4 (CTLA-4) is a potent negative regulator of T cell activation, and administration of blocking antibodies to CTLA-4 has had marked effects in murine models and recent clinical trials, with lymphocytic infiltration into tumors and significant antitumor responses, including complete regressions of advanced disease in a fraction of patients (3335). However, global in vivo CTLA-4 blockade predictably had effects beyond the antitumor response, causing significant autoimmunity. These studies again demonstrate the potent antitumor activity of T cells and suggest that learning how to safely and effectively disrupt checkpoint signals should yield substantial therapeutic benefit.

T-regulatory (Treg) cells suppress T cell responses and provide another mechanism compromising the development of effective tumor immunity (36). These cells are usually CD4+ and are distinguishable phenotypically by expression of CD25 (the α chain of the IL-2 receptor required for high affinity binding), high levels of CTLA-4, the glucocorticoid-induced TNF-related receptor (GITR), and the forkhead transcription factor Foxp3. Treg cells can arise in response to persistent antigen stimulation in the absence of inflammatory signals, particularly in the presence of TGF-β, and have been detected in increased frequency in some cancer patients (12). Thus, depleting Treg cells in vivo may facilitate the elaboration of effective antitumor T-cell responses. Studies in murine tumor models targeting all CD25+ T cells for depletion have appeared promising (37). However, activated effector CD8 and CD4 T cells also express CD25, and depletion of these cells during the acute phase of the antitumor T cell response may severely limit the application of this approach. Thus, defining alternative molecules that permit selective targeting of Treg cells for depletion, such as GITR, should uncover greater anti-tumor activity. However, as with abrogation of CTLA-4 signaling, global ablation of Treg cells increases autoimmunity. General application of this approach to humans will likely await methods to preferentially deplete Treg subsets that regulate antitumor responses or to attenuate any induced autoimmunity.

Adoptive therapy. High-dose chemoradiotherapy followed by rescue from the resulting ablation of normal bone marrow with an allogeneic hematopoietic stem cell transplant (HSCT) has become standard therapy for many hematologic malignancies. One problem with this treatment is graft-versus-host disease (GVHD), due to allogeneic donor-derived T cells injuring the “foreign” normal tissues of the host. However, malignant cells that survive chemoradiotherapy are also of host origin, and patients who develop GVHD have lower relapse rates from an associated graft-versus-tumor (GVT) effect. T cells mediate this antitumor activity, as affirmed by the complete responses sometimes observed in patients who receive infusions of donor T cells to treat relapse after HSCT and in recipients of a newly developed non-myeloablative allogeneic HSCT regimen in whom, because of the absence of high-dose chemoradiotherapy, all antitumor effects must result from GVT effects (38). However, the GVT activity with these regimens is often associated with severe and life-threatening GVHD. Ongoing efforts to define antigenic targets with limited tissue distribution, permitting donor lymphocytes to preferentially target malignant cells and not critical normal tissues, coupled with methods to generate and/or select T cells with such specificities, should provide a much-needed refinement to this approach.

An alternative to using allogeneic T cells to mediate antitumor responses has been to isolate autologous tumor-reactive T cells, expand the cells in vitro, and then reinfuse the cells back into the patient. This approach circumvents many of the obstacles to generating an adequate response in vivo, as the nature of the APCs and components of the microenvironment can be more precisely controlled in vitro. However, this strategy has required the recent development of methods to extensively manipulate T cells in vitro with retention of specificity and function, such that after infusion the cells will survive and migrate to and eliminate tumor cells.

Initial therapies used tumor-infiltrating lymphocytes as an enriched source of tumor-reactive cells, but such cells can also usually be obtained from circulating blood lymphocytes. Although optimal methods for stimulating and expanding antigen-specific T cells in vitro are still being defined, in general, DCs presenting the antigen are used to initially trigger reactive T cells, which can then be selected and stimulated with antibodies to CD3. Supplemental cytokines are provided during cell culture to support lymphocyte proliferation, survival, and differentiation. With this approach, it has been possible to expand tumor-reactive T cells to enormous numbers in vitro, infuse billions of specific cells without overt toxicity to achieve in vivo frequencies beyond that attainable with current vaccine regimens, and mediate regression and occasionally complete elimination of large disseminated tumor masses. However, despite the high in vivo frequencies of tumor-reactive effector cells achieved, only a fraction of patients respond, pointing to the existence of additional hurdles. One clear requirement is that infused cells must persist to mediate an effective response. Analogous adoptive therapy trials for cytomegalovirus and Epstein-Barr virus infection in immunosuppressed hosts have demonstrated increased in vivo proliferation and persistence of CD8 effector T cells in the presence of specific CD4 helper T cells (39, 40). Such CD4 T cells likely provide many beneficial functions, including cytokine production and APC activation, which can improve the quality and quantity of the CD8 responses, as well as direct effector activities against infected or tumor targets. However, unlike viral responses that induce robust CD4 and CD8 responses, identifying and characterizing the specificity of tumor-reactive CD4 T cells has proven considerably more difficult than with CD8 responses. Additionally, obstacles to safely maintaining a CD4 response reactive with a potentially normal protein remain to be elucidated. Consequently, CD4 help is currently largely provided to transferred tumor-reactive CD8 cells in the form of surrogate exogenous cytokines. The largest experience is with IL-2, which clearly prolongs persistence and enhances the antitumor activity of transferred CD8 cells (41). Alternative cytokines such as IL-15, IL-7, and IL-21, as well as activation of APCs with antibodies to CD40, are currently being evaluated in preclinical studies.

The infusion of T cell clones, rather than polyclonal T cell lines, represents an appealing sophistication of adoptive therapy, because the specificity, avidity, and effector functions of infused cells can be precisely defined. This facilitates subsequent analysis of requirements for efficacy, basis for toxicity, and rational design of improved therapies. The transfer of antigen-specific CD8 T cell clones has been shown to be effective for prevention of viral infections and treatment of malignant disease (39, 41). Such studies have also formally demonstrated that low, nontoxic doses of IL-2 are sufficient to promote the in vivo persistence and antitumor activity of CD8 T cells.

Another means to enhance the activity and survival of transferred cells is to take advantage of endogenous homeostatic mechanisms that restore lymphocyte numbers after an episode of lymphopenia. The precise mechanisms, which likely include IL-7 and IL-15 production, are not entirely known, but intentional lympho-depletion of patients before T cell transfer can promote extensive proliferation of infused T cells, creating an in vivo repertoire dominated by the desired effector population (42). Additionally, this promising strategy, which is now being evaluated clinically, may create an environment more conducive to mediating an antitumor effect by eliminating Treg cells.

It is also possible to use integrating vectors to genetically modify T cells before infusion to enhance tumor recognition, cell survival, migration, or effector functions—in essence, engineering responses that might not be naturally achievable. For example, T cell recognition of tumors has been imparted by expressing high-affinity chimeric transmembrane receptors with the external recognition structure of an antibody and the signaling domain of a T cell receptor (43). Such T cells recognize tumor antigens in an MHC-independent fashion like mAbs but still employ effector mechanisms inherent to a T cell. Alternatively, T cell receptor genes can be rescued from T cell clones shown to be effective and safe in therapy and inserted into T cells from other patients with tumors that express the same tumor antigen and MHC-restricting allele, overcoming the need to isolate tumor-reactive effector cells in each individual patient. Survival of transferred cells can also be enhanced by introducing chimeric cytokine receptors that use cytokines such as GM-CSF that are produced by effector CD8 T cells after target recognition as regulated autocrine growth signals (44).

An alternative genetic strategy to introducing new structures or functions to T cells is to disrupt signaling pathways that normally serve to dampen responses such as those through Cbl-b, an adapter protein that negatively regulates T cell receptor signal strength (45). Preliminary studies in which dominant-negative Cbl-b proteins are expressed or Cbl-b expression is reduced through small interfering RNA approaches suggest that increasing signal strength can both reduce the threshold for T cell activation and restore regulated IL-2 production to effector CD8 T cells. Many other targets for these approaches have been identified by studies in knockout mice, suggesting the possibility of designing T cells capable of circumventing many of the obstacles posed by tumors.

Future Directions

It was not long ago that most scientists questioned whether the human immune system was capable of recognizing spontaneously arising tumors and whether immune therapy could ever become a meaningful treatment for human malignancy. The answers are a resounding and unequivocating “Yes!” The administration of cytokines and mAbs have already become components of some standard cancer treatment regimens. Many vaccines have advanced through preliminary testing to efficacy trials, and T cell therapy is being explored with new sophistication in many disease settings. As our understanding increases of the requirements for immune cell activation, homing, and accumulation at tumor sites, and for disrupting the regulatory mechanisms that limit responses, the ability to direct a coordinated and effective attack against tumors engaging multiple components of the immune system should evolve in parallel. Clinical trials of passive transfer of large numbers of tumor-reactive T cells or mAbs should not only help define the nature and magnitude of responses that will be necessary to achieve by stimulating endogenous responses, but will offer the possibility to genetically or chemically engineer immune responses with functional capacities beyond what can be elicited from the normal immune system. Despite Niels Bohr's caution that “prediction is very difficult, especially about the future,” we confidently predict that immunotherapy will become an increasingly essential component of future cancer therapies.

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article