Published online before print May 30, 2008
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Cancer Immunology Program, Trescowthick Laboratories, Peter MacCallum Cancer Centre, East Melbourne, Victoria, Australia; and
Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, Missouri, USA
2 Correspondence: Cancer Immunology Program, Peter MacCallum Cancer Centre, Locked Bag 1, A’Beckett St., 8006, Victoria, Australia. E-mail: mark.smyth{at}petermac.org
ABSTRACT
This brief review discusses the role of the immune system in tumor development, covering a history of cancer immunity and a summary of the concept of cancer immunoediting, including its three phases: elimination, equilibrium, and escape. The latter half of this review then focuses specifically on the equilibrium phase, making note of previous work, suggesting that immunity might maintain cancer in a dormant state, and concluding with a description of a tractable mouse model unequivocally demonstrating that immunity can indeed hold preformed cancer in check. These findings form a framework for future studies aimed at validating immune-mediated cancer dormancy in humans with the hopes of devising new, immunotherapeutic strategies to treat established cancer.
Key Words: immunoediting • elimination • escape
CANCER IMMUNITY—A BRIEF HISTORY
The history behind the study of the immune reaction to cancer is a long and often controversial one, most extensively reviewed by Dunn and colleagues [1
]. Briefly, in 1909, Paul Ehrlich first hypothesized that a key function of the immune system was to protect the host from cancer. However, the lack of immunological tools and knowledge about the immune system meant that this hypothesis was not tested experimentally. As the field of immunology developed, this concept was reintroduced by Burnet and Thomas in the 1950s as the cancer immunosurveillance theory. Burnet and Thomas proposed that lymphocyte populations of the immune system continuously recognized and eliminated cancerous and/or precancerous cells arising in the host before they could cause harm. However, when this hypothesis was formally tested in CBA/H nude mice, the most congenitally immunodeficient mice available in the 1970s, the mice were found to develop spontaneous tumors and methylcholantherene (MCA)-induced sarcomas in an equivalent manner to wild-type control mice [2
, 3
]. Only now in hindsight do we appreciate that there were caveats associated with these initial experiments. Nude mice that lack a thymus are not entirely immunodeficient, containing NK cells and even low numbers of some functional populations of 
β+T cells. Furthermore, the CBA/H strain of mice used in the MCA carcinogenesis experiments expressed a highly active isoform of the enzyme that metabolized MCA to its carcinogenic form. Thus, it was possible that any protective effect of the immune system was masked by the overwhelming efficiency of MCA-induced cellular transformation in the CBA/H strain.
Nonetheless, at the time, these experiments were so convincing that they essentially caused the whole-scale rejection of the cancer immunosurveillance hypothesis. Interest in the field of cancer immunosurveillance was however rekindled as a result of four key studies published approximately 10 years ago. The first demonstrated that use of neutralizing antibody to IFN-
in tumor-bearing mice caused accelerated tumor growth compared with control, antibody-treated, wild-type mice [4
]. Subsequently, mice lacking an intact IFN- receptor were found to have an increased incidence of spontaneous and carcinogen-induced tumors compared with age-matched, wild-type mice [5
, 6
]. Although IFN- has been shown to play multiple roles in promoting protective host responses against carcinogen-induced tumors [7
], the finding that it enhanced tumor cell immunogenicity by up-regulating tumor cell MHC class I antigen processing and presentation was particularly important in demonstrating that rejection of these types of tumors was dependent on the development of anti-tumor immunity [6
]. Lastly, mice lacking perforin (pfp–/–) had an increased incidence of spontaneous B cell lymphomas compared with wild-type mice [8
]. Since then, an extensive amount of experimental data from various mouse models of cancer, together with convincing, correlative clinical data from human patients has provided unequivocal evidence that cells of the immune system, innate and adaptive, are required for the prevention of cancer (reviewed in refs. [9
10
11
]).
Nonetheless, tumors can and do arise in the presence of a functional immune system. We now appreciate that the immune system, through its interaction with tumors, can, in part, sculpt the cancer phenotype, developing less immunogenic variants that ultimately facilitate tumor outgrowth. Indeed, the dual nature of the immune system to impede and aid tumor growth and progression has led to the refinement of the cancer immunosurveillance theory into one now termed cancer immunoediting [1 ].
CANCER IMMUNOEDITING—AN INTRODUCTION TO THE THREE Es
The cancer immunoediting process is envisaged to consist of three phases: elimination, equilibrium, and escape that have been termed the "three Es of cancer immunoediting" [1 , 10 ]. The elimination phase corresponds to the original concept of cancer immunosurveillance, whereby cancer cells are successfully recognized and eliminated by the immune system, thus returning the tissues to their normal state of function. Numerous studies have now demonstrated the requirement for cells of the innate and adaptive immune system for this phase (reviewed in refs. [9 10 11 ]). However, it has become apparent that each model of immune rejection of cancer not only shares many common effector cell subsets and molecules but also often involves some unique aspects, and there is still much to be learned about the specifics of immune tumor elimination. Tumor cells not completely eliminated by the immune system were predicted to proceed into a phase of equilibrium, a poorly understood process where the immune system was envisaged to control tumor cell growth but not completely eliminate the transformed cells. It was hypothesized that in the equilibrium phase, two outcomes may eventuate. In the first, the immune system may eventually eliminate all tumor cells, in essence, leading to an outcome that was not physiologically distinct from elimination. In the second scenario, the constant interaction of the immune system with tumors over a long period of time may ultimately "edit" or sculpt the phenotype of developing tumors, resulting in the selection for tumors that have been shaped into a less-immunogenic state. Tumors that are no longer susceptible to immune attack thus progress into the third phase of the immunoediting process, termed "escape." It is thought that the emergence of clinical symptoms of cancer generally correlates with this stage. Much work on tumor immunology has focused on the escape phase with the complex dynamic interactions that occur between tumors and the immune system now being dissected. Numerous experimental data have demonstrated that tumors in the escape phase subvert the immune system through direct and/or indirect mechanisms to aid in their growth (reviewed in refs. [9 10 11 ]). Intensive work is now being carried out to devise strategies that can target these mechanisms of escape, as this represents new means of cancer immunotherapy [12 ]. Of the three phases of cancer immunoediting, evidence for the equilibrium phase, during which the tumor and anti-tumor immune response co-evolve, is less advanced. We will now discuss what is known about this phase from mouse models and humans and highlight some recent progress that we have made in this area.
EQUILIBRIUM—WHAT WE HAVE LEARNED FROM MOUSE MODELS
The existence of an equilibrium phase of immunoediting was difficult to deduce in the past because of an inability to identify and monitor preneoplastic or small/early neoplastic lesions that may be subject to immunological control. In the past, evidence for an equilibrium or tumor-dormancy phase was indirectly provided by transplant data, demonstrating that tumors arising in immunodeficient backgrounds can be more immunogenic than those arising in wild-type mice. MCA-induced sarcomas from IFN-, type I IFN, and T cell-deficient mice and lymphomas from pfp–/– mice could be eliminated when transplanted into immunocompetent recipients (i.e., wild-type mice) but grew progressively when transplanted into mice of the same immunodeficient genetic background as those in which they arose. These data suggested that the immune system sculpts the immunogenic profile of evolving tumors, the exact process of which is thought to occur during the equilibrium phase.
Other studies suggested that some residual tumor cells may persist in an immunocompetent host and that the clinical dormancy exhibited by these cells may have been a result of the actions of the immune system [13
, 14
]. Thus, the concept emerged that the immune system can prevent expansion of proliferating tumor cells [15
]. In other studies, interruption of tumor dormancy of mouse lymphoma and leukemia was proposed to result from reduced expression of tumor-associated antigens and evasion of the immune system [14
, 15
]. An interesting, early demonstration of tumor dormancy was documented using the BALB/c B cell leukemia/lymphoma 1 (BCL1) model [16
]. In this model, mice were immunized with BCL1-derived Ig (i.e., treated with an idiotype vaccine) and then challenged with BCL1 cells. After 25–30 days, all control mice developed splenomegaly and were killed, and 70% of immunized mice remained protected from splenomegaly for >60 days. Importantly, low numbers of BCL1 cells could still be detected in these mice by flow cytometry, and mice relapsed at a steady rate over a period of more than 610 days [17
]. A similar finding was also made when mice were vaccinated with irradiated BCL1 cells and then challenged with live tumor cells. In this case, tumor cells could be detected in the spleen of 40% of long-term survivors 250 days after tumor challenge, and throughout this period, host mice were completely asymptomatic [18
]. A humoral response controls a regulatory network of anti-idiotypic antibodies that activates BCR signaling, which inhibits proliferation and is followed by apoptosis or dormancy (for a comprehensive review, see ref. [19
]). Interestingly, in the case of BCL1 Ig-vaccinated mice, depletion of CD8+ T cells or the neutralization of IFN- reduced (but did not completely ablate) the period and incidence of dormancy, implicating these factors in the maintenance of the dormancy state [16
]. These findings are consistent with a role for T cells and IFN- in the equilibrium phase of immunoediting proposed in the MCA-induced sarcoma model.
Adoptive immunotherapy has also been demonstrated to protect mice long-term from prostate carcinoma in a transgenic model, and histological analysis of long-term survivors has demonstrated that tumors are not entirely eliminated but are restricted to small foci [20 ]. Adoptively transferred, tumor-specific T cells were also present in protected mice, and these findings are again consistent with a state of equilibrium between the tumor and anti-tumor immune response. Additional studies have shown that proliferating mouse lymphoma cells are kept at a low number in the bone marrow, owing to persistent antigen and memory T cells that are able to coordinate a CD4+ and CD8+ T cell-mediated response [21 22 23 ]. The idea that bone marrow is a special compartment for immunological memory and tumor dormancy is supported by clinical studies showing a higher proportion of memory T cells among the CD4+ and CD8+ cells in the bone marrow of patients with breast cancer who carry cytokeratin-positive cells (breast cancer cells) compared with normal, healthy donors [22 ].
CD8+ T cells may also be the critical cell type involved in a model of surveillance of UVB-induced skin, where there is potential equilibrium between the immune system and a developing cancer [24 ]. The E3 ligase Casitas B cell lymphoma-b (Cbl-b) is a key signaling molecule that controls spontaneous, anti-tumor activity of CTL. CD8+ cells were depleted in UVB-treated cbl-b–/– mice that had received UVB irradiation 130 days earlier but never developed a tumor. Remarkably, only 10 days after starting the depletion by injection of the anti-CD8, 50% of UVB-treated cbl-b–/– mice developed rapidly growing tumors, whereas all control-treated, UVB-treated cbl-b–/– mice remained tumor-free. Whether CD8+T cells also control skin tumors for long periods of time in similar, UVB-irradiated, tumor-free, wild-type mice was not tested in this experiment. Regardless, these results suggested that CTL might potentially enter equilibrium with UVB-induced skin cancers. Overall, in some situations, the immune system might still be operating to suppress residual tumor cell expansion.
EQUILIBRIUM IN HUMAN CANCER PATIENTS
Clinical evidence suggests that tumors can remain dormant in patients for many years, and cases of relapse after long periods (at least 10 years, although sometimes exceeding 20 years) of tumor remission have been noted [25 26 27 28 29 30 ]. The case of tumor remission may provide one possible setting in which potential immunological control of tumors may be investigated. Cancer recurrence after therapy and long periods of remission is frequent. For example, 20–45% of patients with breast or prostate cancer will relapse years or decades later [31 32 33 ]. Recent intriguing studies have shown that patients who are free of clinically detectable disease >20 years after treatment still have circulating, disseminated tumor cells [34 ]. In fact, most cancer types are associated with disseminated disease that after treatment might persist as minimal residual disease. However, the lack of mechanistic insight into this stage has been a major shortcoming in our understanding of the full complexities of metastatic growth. In one report, six cases of recurrence of nonsmall cell lung carcinoma between 7 and 14 years after achieving remission were reported, and all were associated with immunosuppressive treatments [25 ] indicative of a role for the immune system in growth suppression; however, not all cases of recurrence have been associated with immunosuppression.
In a similar vein, cases of unintentional transplantation of cancer from organ donor to immunosuppressed recipient have also been noted. In these cases, it has been documented that tumors of donor origin can come from donors who were in clinical remission [35 ] or from donors with no clinical history of malignancy [36 ]. In such cases, it is possible that the tumor was being held under control by an immunological mechanism in the donor, and transplant of the organ into an immunosuppressed, naïve host allowed subsequent tumor outgrowth. Once again, however, although consistent with the concept of an equilibrium phase in a cancer immunoediting process, other explanations for the growth of donor-derived malignancies are possible. For example, fluctuation in oncogene expression has been implicated in tumor cell dormancy [37 ] and certainly could have occurred during organ transplantation. One interesting point to make here comes from donor cell leukemia. Arising from donor cells, following allogeneic bone marrow transplants, leukemias occur in rare cases [38 ]. From an immunoediting perspective, these cases are of interest, as living donors could potentially be observed for evidence of preneoplastic changes or the presence of an anti-tumor immune response. Although not reported for the majority of cases, no instances of leukemia were noted in the donors from whom donor-derived malignancies arose in graft recipients, suggesting that malignancies are controlled in some manner in the donor or arise in the host environment following transplantation.
Clinical evidence supportive of the equilibrium phase of immunoediting is provided by a number of other findings. First, the existence of an immune response to preneoplastic monoclonal gammopathy of unknown significance (MGUS) cells that eventually can progress to multiple myeloma is consistent with the equilibrium phase, and the immune system controls but does not eliminate MGUS cells, which eventually evolve and progress to malignancy. Further clinical evidence for an equilibrium phase in tumor immunoediting is provided by two other hematological malignancies: low-grade B cell lymphoma and acute myeloid leukemia. Passive immunization with anti-idiotype antibody, in conjunction with cytokine or chemotherapy, can induce remission in some low-grade, B cell lymphoma patients. Interestingly, however, tumor cells are not eliminated completely and can be detected in the blood or bone marrow for up to 8 years following clinical remission [39 ]. It is unclear if an immunological mechanism is responsible for keeping residual lymphoma cells in check in this disease; however, this finding is reminiscent of results from an idiotype vaccine against the BCL1 murine tumor (see above). A role for the immune system in establishing long-term remissions has also been suggested by studies of pediatric acute myeloid leukemia patients treated with chemotherapy or chemotherapy with autologous bone marrow transplant [40 ]. In this study, eight of 18 patients in remission following induction chemotherapy had detectable anti-tumor CTL precursors in their peripheral blood, and no instances of relapse were noted amongst these patients. In contrast, seven of eight patients, in which no anti-tumor CTL precursors could be detected, suffered from subsequent relapse. These results further suggest that cancer immunoediting can synergize with chemotherapy to achieve or maintain clinical remissions, a potential state of equilibrium. Transcriptional profiles from dormant, disseminated tumor cells or experimental models of dormancy might help determine whether primary tumors carry a cancer dormancy "signature", which might have prognostic value [41 ].
IMMUNE-MEDIATED VERSUS OTHER FORMS OF DORMANCY
In general, a long delay in tumor development has been explained by nonimmune types of cancer dormancy. Cancer dormancy is poorly understood; however, clinical evidence supports the existence of various mechanisms of cancer dormancy including cellular ("intrinsic") dormancy (G0–G1 arrest) and angiogenic ("extrinsic") dormancy (reviewed in ref. [42 ]; Fig. 1 ). Cellular dormancy is thought to occur when tumor cells enter a state of quiescence or senescence, although whether these programs drive such dormancy is not understood completely. Alternatively, angiogenic dormancy is when cancer cell proliferation is counterbalanced by apoptosis, owing to poor vascularization—the tumor does not develop beyond a certain size because of limitations in its blood supply. Tumor dormancy is observed in local recurrences or metastases. In the case of a primary tumor, the term commonly used is latency, the time between carcinogenic insult and clinical detection of the tumor. These mechanisms of dormancy may actually favor immune-system evasion; however, additional immune control with subsequent escape remains an intriguing possibility in these cases. How the immunological response intersects with angiogenic dormancy or cellular dormancy is unknown. Genes that control escape from dormancy (for example, quiescence- or angiogenic-switch-dependent) might also coordinate the ability of tumor cells to evade an immune response. However, we question whether all forms of tumor dormancy are mutually exclusive, how long they last, or whether they occur at different times.
 View larger version (26K): [in a new window] |
Figure 1. Various forms of cancer dormancy. (A) Subclinical disease may occur when dormant cells enter a G0–G1 arrest (Cellular dormancy), and these cells may additionally develop mechanisms to evade immune system recognition and eradication. (B) Angiogenic dormancy results from the balance between pro- and antiangiogenic factors (e.g., vascular endothelial growth factor). (C) Immune dormancy. Proliferating tumor cells are kept at low numbers (subclinical) by an active immune system that may include the actions of CD8+ T cells, IFN-, and antibodies. It is unclear whether these forms of dormancy are mutually exclusive.
|
As discussed above, the idea that proliferating tumor cells may sometimes be kept at low numbers (subclinical) by an active immune system has been proposed by others; however, until recently, it has not been possible to visualize or validate this latent state in any experimental or clinical setting. Now, for the first time, our work has identified small, inert lesions containing tumor cells in an equilibrium state, a first step toward better characterization of tumor latency [43 ]. An attenuated dose of the carcinogen MCA causes an initial wave of tumors affecting a small proportion of mice. The apparently healthy, surviving mice show no evidence of growing tumors; however, dormant tumors still exist, and they are kept in check by the immune system. This state of equilibrium can be disrupted by specific immunosuppression, allowing the dormant tumors to escape from immune control and grow out and kill their host.
Some important principles were established in this study (Fig. 2
). First, dormant tumors in immunocompetent mice only developed into progressive tumors after treatment with mAb, depleting T lymphocytes (T cells) or neutralizing the action of the cytokines IL-12 or IFN-. Depletion of innate NK cells or neutralization of the NKG2D and TRAIL pathways had no effect. This clearly suggests that highly specific, adaptive T cell immunity is the component of the immune system that maintains equilibrium. Second, by morphology, many cells in dormant lesions showed features of cancer cells, including large cells with variable degrees of atypia, enlarged nuclei, and prominent nuclear bodies, all reminiscent of the histology of progressively growing, primary, MCA-induced cancers. Stable, dormant masses showed infiltration by immune cells, including T cells, B220+ cells, and macrophages and a much lower percentage of proliferating, atypical cells but increased cell death compared with growing sarcomas. Transient culture of cells from dormant lesions yielded atypical, fibroblast-like cells that grew out as tumors when injected into immunodeficient but not wild-type mice. By contrast, the few tumors escaping from dormancy grew out in a high percentage of cases in wild-type mice. The dormancy observed in this MCA induction of a sarcoma model is similar to the behavior of chronic infections such as caused by the pathogen Mycobacterium tuberculosis, which typically establishes latent infection that can last for decades before the pathogen reactivates, and clinical disease presents, often coincidental with a period of immune suppression.
 View larger version (44K): [in a new window] |
Figure 2. Equilibrium. In a proportion of MCA-treated mice, small, stable masses (2–8 mm diameter), exclusively at the site of carcinogen injection, harbor transformed sarcoma cells, the outgrowth of which was immunologically restrained. Half of the stable masses contains clusters of large cells showing variable degrees of atypia mirroring the immunocytological spectrum observed in progressively growing, primary MCA sarcomas. Immunohistochemical staining of stable masses containing atypical cells revealed the presence of CD3+ T cells, B220+ cells, and F4/80+ mononuclear phagocytes, infiltrating into regions containing atypical cells. Stable masses are also characterized by a combination of increased apoptosis and decreased tumor cell proliferation. An interruption to this equilibrium might be caused by specific immune depletion of T cells or neutralization of IFN- or IL-12. Alternatively, many distinct mechanisms of tumor cell escape from the immune system might allow outgrowth of the tumor mass, which will then display an edited tumor phenotype.
|
ACKNOWLEDGEMENTS
M. W. L. T. and M. J. S. thank the National Health and Medical Research of Australia for fellowship and program support. We acknowledge the contribution of J. B. S. to sections of this review drafted from his Ph.D. literature review. We thank all of the members of the Smyth and Schreiber Laboratories who have contributed to the work described in this review.
FOOTNOTES
1 These authors contributed equally to this work. 
Received November 18, 2007; revised March 10, 2008; accepted March 11, 2008.
REFERENCES
receptors Immunity 1,447-456[CrossRef][Medline]-dependent tumor surveillance system in immunocompetent mice Proc. Natl. Acad. Sci. USA 95,7556-7561 and lymphocytes prevent primary tumor development and shape tumor immunogenicity Nature 410,1107-1111[CrossRef][Medline] in establishing and maintaining the tumor-dormant state J. Immunol. 162,2842-2849
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
