|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 22, No 6 (March 15), 2004: pp. 1136-1151 © 2004 American Society of Clinical Oncology. DOI: 10.1200/JCO.2004.10.041
Tolerance and Cancer: Mechanisms of Tumor Evasion and Strategies for Breaking ToleranceFrom the Department of Hematology and Oncology, University Medical Center Charité, Campus Virchow Klinikum, Humboldt University Berlin, Berlin, Germany; and the Transplantation Biology Research Center, Bone Marrow Transplantation Section, Transplantation Biology Research Center Massachusetts General Hospital/Harvard Medical School, Boston, MA. Address reprint requests to Megan Sykes, MD, Transplantation Biology Research Center, Bone Marrow Transplantation Section, Transplantation Biology Research Center, Massachusetts General Hospital, MGH-E Bldg149-5102, Harvard Medical School, Boston, MA 02129; e-mail: Megan.Sykes{at}tbrc.mgh.harvard.edu
The development of malignant disease might be seen as a failure of immune surveillance. However, not all tumors are naturally immunogenic, and even among those that are immunogenic, the uncontrolled rapid growth of a tumor may sometimes out-run a robust immune response. Nevertheless, recent evidence suggests that mechanisms of tolerance that normally exist to prevent autoimmune disease may also preclude the development of an adequate antitumor response and that tumors themselves have the ability to thwart the development of effective immune responses against their antigens. A major challenge has been to develop approaches to breaking this tolerance in tumor-bearing hosts, and recent advances in our understanding of antigen presentation and tolerance have led to some promising strategies. An alternative approach is to use T cells from nontumor-bearing, allogeneic hosts in the form of lymphocyte infusions, with or without hematopoietic cell transplantation. Immunotherapy may occur in this setting via the response of nontolerant, tumor antigen-specific T cells from nontumor-bearing hosts or via the powerful destructive effect of an alloresponse directed against antigens shared by malignant cells in the recipient. Approaches to exploiting this beneficial effect without the deleterious consequence of graft-versus-host disease in allogeneic hematopoietic cell recipients are discussed.
Immunologic tolerance to a particular set of antigens is the absence of an immune response against those antigens, while normal responses to other antigens are preserved [1]. The mechanisms underlying the induction and maintenance of tolerance have been a major focus of research aimed at understanding how the immune system discriminates between self and nonself and avoids autoimmunity. Whereas induction of tolerance to a given set of donor antigens is the ultimate goal in organ transplantation, the opposite goal prevails in the field of oncology, in which an inadequate immune response towards a tumor permits tumor growth. This perspective in oncology has sparked intense research focusing on the following two major questions: (1) Which mechanisms underlie immune evasion by tumors, permitting their development and progression? and (2) Is it possible to break tolerance to tumor antigens and induce an antitumor response? In this article, we will try to address these questions and describe recent progress toward the development of strategies for improving antitumor responses.
One hypothesis to explain the relatively low frequency of tumor development in immunocompetent hosts was based on the assumption that the immune system might be able to control or eliminate the majority of tumors early in their development. Tumor development might be associated with the acquisition of gene mutations and expression of neoantigens or the overexpression of cellular proteins, which could be targets for recognition by the immune system. Clinical data seem to corroborate this hypothesis because lymphocyte infiltration of tumors has been shown to correlate with improved survival for a great variety of solid tumor types [2-6]. The high incidence of malignancies of several types, including those that are and are not known to be virally induced or associated, in patients receiving chronic immunosuppressive therapy after organ transplantation suggests a role for T-cell immune surveillance against tumors in humans [7,8]. Further corroboration is provided by the high incidence of lymphomas, including those that are not known to be virally associated as well as virally associated lymphomas and Kaposi's sarcomas, in patients with HIV-1 infection and AIDS [9]. This concept of immune surveillance was initially formulated by Burnet [10] and has evoked strong criticism, primarily because of experimental data showing that nude mice, which have T-cell defects, do not have an increased incidence of tumor development [11,12]. However, it was argued that the innate immune system could compensate for deficient T-cell immunity and explain the normal frequency of tumor development in these mice. Recently, new data have revived this discussion because severely immunocompromised, Rag-/- and STAT-1-/- mice, which display deficiencies in the innate and the adaptive immune systems, have been found to have a significantly increased incidence of tumors [13]. In these mice, the tumors develop late in life (> 1 year), arguing that immune surveillance controls the gradual, spontaneous development of tumors. It is possible that the innate immune system helps to prevent the more rapid development of tumors. Indeed, natural killer (NK) cells have been shown to be capable of killing certain tumors in vivo and in vitro, and they may play a particularly important role in controlling the development of tumors that downregulate class I major histocompatibility complex (MHC) expression [14], thereby evading destruction by class I–restricted cytotoxic T lymphocytes. Furthermore, it has been recently suggested that alloreactive NK cells might be critically involved in mediating graft-versus-leukemia (GVL) effects in the setting of MHC haploidentical hematopoietic stem-cell transplantation [15]. Cellular and humoral immune responses against tumors can be detected spontaneously in the tumor-bearing host, and additional responses can be induced [16]. On the basis of their expression pattern, the following two groups of tumor antigens can be distinguished: true tumor-specific antigens and tumor-associated antigens (TAA). Whereas true tumor-specific antigens are encoded by mutant cellular genes, TAAs are encoded by normal cellular genes. The group of potential tumor-specific antigens includes mutation-carrying tumor suppressor genes (p53), tumor-specific chimeric fusion proteins (bcr-abl), mutated oncogene-encoded proteins (ras), mutated cell cycle regulators (CDC27, CDK4), and tumor-specific rearrangements of the immunoglobulin heavy-chain locus (idiotype [Id] of B-cell neoplasias). TAAs include differentiation antigens (prostate-specific antigen, alpha-fetoprotein, breast mucin, the melanocyte differentiation antigen melanoma antigen recognized by T cells-1 [MART-1], and tyrosinase) and the so-called cancer testes antigens. Cancer testes antigens are normally expressed in spermatozoa and are silenced in somatic cells. In the process of cancer development, the expression of these genes re-emerges. Additional antigens in this group include members of the melanoma antigen-encodin [17]. Amplification antigens include a group of TAAs that are amplified or overexpressed in tumors and could, therefore, be targeted for immune responses (eg, murine-double-minute-2 [mdm-2] oncoprotein and Wilm's tumor gene). The majority of known tumor antigens recognized by T cells were defined using transfection of genomic tumor DNA into cells expressing the appropriate MHC. Transfection of the tumor antigen gene made the target cells susceptible to killing by human T cells with specific antitumor reactivity. These antigens are presented primarily by MHC class I molecules and recognized by CD8+ T cells. Despite the discovery of MHC class I–restricted CD8+ tumor-reactive T cells, vaccination studies using these class I–presented peptides led to rather disappointing results with respect to the achievement of major tumor responses [18]. One explanation for the weak immune responses elicited by these peptides may be a lack of tumor-specific CD4 help [19-21]. In fact, induction of tumor-specific CD4+ T cells by vaccination with a specific viral T-helper epitope presented by MHC class II led to protective immunity against MHC II–negative, virus-induced tumor cells in a mouse model. Tumor rejection was mediated by CD8+ cytotoxic T lymphocyte (CTL) cells recognizing a different viral antigen, demonstrating the significance of crosspriming and CD4 orchestration of the immune response [22]. Recently, tumor antigens have been described that are presented by MHC class II and recognized by CD4+ T cells (eg, mutated CDC27 [23]). Identification of additional MHC class II–presented tumor antigens could, therefore, be crucial to the generation of optimal antitumor responses. The role of CD4+ T cells may be to initiate the immune response by activating antigen-presenting cells (APC) so that they can optimally present antigen to costimulate and support the differentiation of CTL [24-26] and to maintain CD8 cell survival by providing cytokines and maintaining costimulatory molecule expression on APC [26-28]. Because of these considerations, identification of MHC class II–restricted tumor antigens or TAAs has recently attracted considerable interest. Another approach to detecting immunogenic tumor antigens is based on the detection of a humoral immune response by screening a phage expression library with serum from cancer patients [29]. Using this approach, a number of cancer testes antigens have been detected. So far, it is unclear whether or not these antigens have in vivo relevance with respect to T cell–dependent immunity. Hypothetically, humoral immune responses are dependent on T-cell help, and therefore, it is conceivable that a T-cell response against these serologically defined antigens should be present. Indeed, the NY-ESO-1 antigen, which was initially identified using the serologic analysis of cDNA expression libraries approach, has been demonstrated to be a target for CD8 and CD4 T-cell responses [30,31]. Thus, there is clear evidence for the presence of a cellular and humoral immune response in such tumor-bearing hosts. For example, a humoral immune response to NY-ESO-1 could be detected in 40% to 50% of patients with tumors expressing this antigen [32]. Furthermore, using tetramers and enzyme-linked immunosorbent spot assays, a recent study showed that 10 of 11 patients with spontaneous antibody responses also had a CD8+ T-cell response. These HLA-A2–restricted CD8+ T-cell responses were only detectable in those patients with an antibody response [33]. However, in four of seven patients with NY-ESO-1–expressing tumors, intradermal vaccination led to the induction of a CD8+ NY-ESO-1–specific T-cell response in the absence of a pre-existing response. This induction of T-cell immunity was associated with stabilization and even regression of the disease [30]. In contrast, patients with pre-existing antibody responses against NY-ESO-1 did not show an increase in antigen-specific CD8+ T cells after vaccination, despite some evidence of clinical benefit. Thus, although spontaneous T- and B-cell responses can be detected in tumor-bearing hosts and T-cell responses can be induced by vaccination strategies, these data suggest that this immune response may be overwhelmed by tumor growth (or an antigen loss mutation permits tumor growth) or may be merely an epiphenomenon associated with tumor progression. These possibilities are further underscored by the observation that the level of detectable antibody correlates with the tumor burden [34].
Tolerance to a given antigen can be achieved by either deletional mechanisms (ie, elimination of the antigen-reactive cells) or by nondeletional mechanisms. The term nondeletional refers to mechanisms in which antigen-specific cells are unable to respond to the antigens their T-cell receptors (TCR) recognize. This can be a result of either inadequate activation stimuli, leading to anergy, or active suppression of the antigen-reactive T cell through regulatory cells. Deletional tolerance to tumor antigens could not be broken via immunostimulation in the tumor-bearing host because T cells with reactivity to these antigens would simply be absent. Because most antigens expressed by tumors are, in fact, normal self antigens to which deletional tolerance is likely to exist, the number of T-cell clones in a given immune repertoire that can recognize tumors is likely to be relatively small to begin with. Thus, it is easy to imagine that rapidly growing tumors might out-run this immune response. Another way of categorizing tolerance is based on the anatomic site and time in development when T cells are tolerized. Thus, for thymus-dependent tolerance induced during the maturation of a T cell, the term central tolerance has been coined. Induction of tolerance extrathymically, after the mature T cell has exited the thymus, is referred to as peripheral tolerance. Central tolerance is achieved primarily through the process of negative selection. Developing thymocytes undergo two selective processes based on the recognition of MHC/peptide complexes by their uniquely rearranged TCRs. Positive selection refers to the process in which thymocytes with a low affinity for a self MHC/peptide complex expressed on epithelial cells of the thymic cortex are rescued from programmed apoptosis by the signal provided by TCR ligation. The consequence of this process is selection of a T-cell repertoire that preferentially recognizes host MHC molecules complexed to peptides. These complexes in the periphery provide signals to the T cells that optimize survival and readiness to respond to self MHC-foreign peptide complexes for which they have higher affinity [35-38]. Because it would be undesirable to develop a mature T-cell repertoire that recognizes self MHC/peptide complexes sufficiently well to be activated by them in the peripheral tissues, thymocytes with high affinity for self MHC/peptide complexes undergo a process of negative selection in the thymus, which leads to deletion via apoptotic cell death. This occurs mainly at the CD4+CD8+ (double positive) stage of thymocyte development, largely in the corticomedullary junction and medulla of the thymus. Although antigen presented on a variety of cell types can lead to deletion, it occurs most effectively as a consequence of the interaction of these immature thymocytes with bone marrow-derived APC in the thymus, especially dendritic cells (DC) [39-41]. In addition, other nondeletional mechanisms have been implicated in the intrathymic tolerance of developing T cells [42]. The overall consequence of these processes is the emergence in the periphery of a T-cell repertoire that recognizes self MHC/self peptide complexes preferentially but weakly and that includes cells with potentially strong reactivity for self MHC/foreign peptide complexes. For several reasons, the thymus may induce tolerance among developing thymocytes that recognize tissue-specific antigens expressed in certain peripheral tissues. Peripheral antigens might gain access to the thymus by mean of transportation through migratory DCs, and the thymic epithelium displays promiscuous expression of antigens previously thought of as specific to certain peripheral tissues and, thereby, induces tolerance to extrathymic antigens expressed in the thymus [43,44]. Nevertheless, T-cell tolerance is not entirely centrally acquired because the many tissue-restricted antigens expressed in the body are unlikely to all be expressed in the thymus. Indeed, autoreactive T cells can clearly escape negative selection in the thymus, and mechanisms exist that can tolerize these T cells once they mature and reach the periphery. These mechanisms include anergy, deletion, and suppression. The processes are incompletely understood and have been variously implicated in tolerance induced under different circumstances. Some of these will be discussed later in the context of tolerance to tumor antigens. It is worth pointing out that the thymus has been implicated in peripheral tolerance in addition to central tolerance because the egress of regulatory cells from the thymus seems to play a critical role in certain forms of tolerance involving active suppression of the responses of mature T cells in the periphery [45]. There is growing evidence that these regulatory cells have a crucial role in controlling the immune response. CD4+, CD25+ T cells have been shown to be important regulatory cells that contribute to the prevention of autoimmunity [46]. These cells have also been recently described in humans and are currently a focus of intensive study [47]. In the context of tumor immunology, consideration of the cell types presenting tumor antigen in the periphery is of considerable importance. T cells require signals from costimulatory molecules, such as CD28, ICOS, and others, to be fully activated following ligation of their antigen receptors. In the absence of such costimulation, T cells may be tolerized by antigen recognition. As is discussed in the ensuing section, tumor cells themselves may not express the necessary costimulatory molecules to make them effective APC to induce tumor antigen-specific responses. Immature DCs also may fail to provide adequate costimulation and may, thereby, tolerize T cells recognizing antigens they present [48,49]. However, recent data suggest that mature DCs might also be able to induce tolerance. In addition to MHC/peptide and costimulation, another signal, such as that provided by a cytokine, might also be crucial in determining whether the interaction of a T cell with a DC results in activation or tolerization [50].
There are a number of mechanisms by which tumors may actively evade or silence/suppress an immune response (Fig 1). However, it should first be considered that the magnitude of an antitumor response may be limited to begin with and may be insufficient to overwhelm a rapidly expanding and mutating tumor. A major factor limiting immune recognition of cancer cells is the fact that tumors arise from the organism's own tissue and, therefore, mainly express self antigens to which the individual's T cells have been tolerized, either centrally or peripherally. This situation could be expected to be manifested as tolerance of T cells that display a high avidity for these normal self antigens expressed by the tumor, leaving only T cells with low avidity. This problem is exemplified for p53. Because of its high level of expression in certain malignancies, wild-type p53 is a potential target antigen for immunotherapy in a broad spectrum of neoplastic diseases. However, because of low-level expression in normal tissues, T-cell tolerance by clonal deletion of high-avidity T cells in the thymus could be an obstacle. Using p53+/+ and p53-/- HLA-A2.1/Kb transgenic mice, Theobald et al [51] showed that, indeed, p53-specific HLA-A2.1–restricted CTL in p53+/+ mice had a 10-fold lower avidity than CTL in p53-/- mice.
Nevertheless, it has been possible to detect and clonally expand T cells specific for TAAs from tumor-bearing hosts. For example, using peptide/MHC tetramers, Lee et al [52] demonstrated TAA-specific circulating T cells in six of 11 melanoma patients. MART-1–specific T cells were detected in four patients, with frequencies ranging from 0.014% to 0.16%. Tyrosinase-specific T cells were detected in two patients, with frequencies of 0.19% and 2.2% [52]. Analysis of tumor-infiltrating lymphocytes (TIL) revealed that up to 30% of TIL are able to react against target cells expressing the antigen and the appropriate MHC molecule [53,54]. Furthermore, in vitro it is possible to generate high avidity melanoma-reactive cytotoxic T cells [55]. However, in vitro cytolytic activity or cytokine release on interaction between T cells and target cells may not always be a valid indicator of tumor reactivity and cytolytic activity in vivo. Furthermore, the presence of TAA-specific T cells may not correlate with tumor rejection. Thus, Lee et al [52] were able to detect circulating TAA-specific T cells in melanoma patients. However, the patient with the highest frequency of TAA-specific T cells was completely unresponsive to tumor antigen-expressing cells in vitro, although strong antiviral responses were still detectable. These results indicate that, even if TAA-specific cells are present at detectable levels in tumor-bearing hosts, they may be incompetent to reject the tumor. Thus, it is conceivable that the T-cell response to tumor antigens may be further reduced by active tolerance induction by the tumor. For example, adoptive transfer experiments have demonstrated that tumor antigen-specific naïve T cells are rapidly anergized when transferred to tumor-bearing mice [56]. The major mechanism inducing this form of T-cell tolerance involves cross-presentation of tumor antigens by bone-marrow–derived APC [57]. Intentional in vivo activation of APC by CD40 ligation resulted in conversion of this T-cell tolerance to T-cell priming [58]. These results underscore the relevance of the activation state of APC with regard to the induction of T-cell tolerance or activation (see preceding section). Mechanisms that have been implicated in the induction of anergy or deletion of tumor antigen-reactive T cells include secretion of the immunosuppressive cytokines interleukin (IL)-10 and transforming growth factor beta [59-62] and the expression of apoptosis-inducing Fas ligand, resulting in apoptosis of tumor-reactive T cells and immune evasion [63-67]. In addition, it has been recently demonstrated that expression by tumor cells of human B7-H1, a member of the B7 family of costimulatory molecules, leads to induction of T-cell apoptosis and might, thereby, contribute to immune escape of these tumors [68]. Another recently described receptor, RCAS1, which is expressed by tumors, has been shown to induce apoptosis in T and NK cells [66]. Conversely, it has been recently demonstrated that colon cancer cells can actively evade Fas-ligand–mediated cell death induced by immune effector cells by secretion of DC3 decoy receptors that bind and neutralize Fas ligand [69]. Although specific pathways for its induction have not been fully delineated in most instances, anergy of T cells in tumor-bearing hosts has been attributed to dysfunctions of the TCR signaling pathway, including globally decreased expression of the TCR zeta chain and loss of Syk tyrosine kinase in T cells infiltrating tumors [70,71]. Another mechanism responsible for the downregulation of T-cell responses against tumors might be the presence of regulatory T cells within the tumor. CD4+, CD25+ regulatory T cells have been shown in mice to be crucially involved in the prevention of autoimmunity (see preceding). Depletion of these regulatory cells leads to the development of autoimmunity, which is also observed in CD25-deficient mice [72]. Recent evidence suggests that such regulatory T cells might be involved in thwarting the T-cell response against the tumor in the tumor-bearing host [73-76]. It is conceivable that a combination of host T-cell depletion followed by adoptive administration of tumor-reactive T cells might circumvent this immune evasion mechanism. A similar concept may underlie recent successes described in melanoma patients who received nonmyeloablative conditioning consisting of cyclophosphamide and the T-cell–suppressing purine-analog fludarabine, followed by the infusion of highly enriched autologous tumor-reactive T cells derived from TIL. This treatment strategy led to regression of melanoma lesions and the development of autoimmunity [77]. This tumor response was associated with the engraftment and persistent clonal repopulation of the T-cell compartment, which was not observed in previous studies with the administration of highly active in vitro expanded antitumor T-cell clones [78-81]. Another factor that may contribute to tumor development and progression may result from localization of a tumor so that it is not accessible to circulating T cells, which are, therefore, ignorant of its presence. For example, a sarcoma cell line expressing a viral model antigen was shown to elicit a strong CTL response when transferred subcutaneously in a single-cell suspension but did not do so when the same cells were transferred as solid tumor fragments [82]. Tumor antigen-reactive T cells were not deleted or anergized, because a specific cytotoxic T-cell response could be readily induced by adequate immunization. Furthermore, the ability of tumor cells to migrate from the tumor itself to secondary lymphoid organs was shown to be critical to the development of a tumor antigen-specific CTL response [83]. Cross-priming of CTL by tumor antigen represented on class I molecules expressed on host APC was shown to be relatively inefficient and not highly protective against tumor growth [83].
Despite the many mechanisms by which tumors can evade or subvert immune responses, a number of strategies for enhancing antitumor immunity can be envisaged (Fig 2). These strategies can be categorized as those which attempt to enhance the host's own immune response to a tumor and those that use T cells from a nontumor-bearing syngeneic or, more often, allogeneic donor. We will discuss each of these separately.
Induction of Host Immunity Toward Tumor Antigens As suggested by the presence of spontaneous T- and/or B-cell immunity against tumor antigens and the inducibility of immune responses, one approach is to use these known targets of immune reactivity in efforts to enhance this immunologic response (see Immune Recognition of Tumors regarding MART-1, NY-ESO 1, etc). Several approaches are currently under investigation. We will review two major strategies, vaccination and cytotoxic T-cell targeting. The concept of inducing an immune response against cancer is quite old and dates back to the end of the 19th century, when attempts were made to achieve this using bacterial toxins. The following two major issues have to be considered in the design of vaccination studies in cancer: the target antigen and the vaccination vehicle. Because of the fear of eliciting autoimmune responses and the fact that high-avidity T cells recognizing antigens shared between tumor and normal cells might be lacking in the tumor-bearing host, initial research focused on true tumor-specific antigens resulting from tumor-specific genetic aberrations (see "Immune Recognition of Tumors"). Therefore, human B-cell malignancies (eg, multiple myeloma) have been especially attractive targets because of the presence of a tumor-specific rearrangement of the immunoglobulin gene locus resulting in a clonal marker and an easily accessible tumor antigen, which in the case of multiple myeloma can even be retrieved from the serum of patients. Thus, a number of preclinical and clinical studies have focused on using the Id of the malignant B-cell clone as antigen for vaccination [84-90]. A pilot study using Id-loaded DC was promising, with three detectable anti-Id and antitumor responses in three of four patients with follicular non-Hodgkin's lymphoma [90]. A recently published study from the same investigators showed similar promising results in a larger cohort of patients. Among 10 patients treated with Id-loaded DC, an immune response could be detected in eight patients, and four clinical responses occurred (two complete responses, one partial response, and one molecular response). Sixteen of 25 additional patients who had received chemotherapy before DC-based therapy developed an immune response, and four of 18 patients with residual tumor at the time of vaccination showed tumor regression [84]. A number of vaccination approaches are currently being evaluated in clinical trials in efforts to induce host immune responses against a variety of solid tumors (colon cancer, prostate cancer, melanoma, renal cell carcinoma [RCC], etc) and have even entered phase III levels [91-97]. These strategies are all based on the observation that tumors are often poor APCs. The lack of costimulatory molecules on their surface and the failure to produce stimulatory cytokines may make them poorly immunogenic and sometimes even tolerogenic. The approaches investigated include the use of gene-modified tumor cells [98], the use of professional APC (DCs) or DC fused to tumor cells [99,100], and DNA transfer using naked DNA or viral vectors. Vaccination with DCs has led to systemic T-cell responses in treated patients. However, clinical responses have been less striking. Nevertheless, significant antitumor responses, including some complete responses, have been reported in patients with melanoma, colon, and prostate cancer [95,101-103]. Among the approaches that have been investigated for enhancing the immune-stimulatory capacity of tumor cells, cytokine transfection of tumor cells has been studied in great detail and was generally not superior to nonspecific adjuvants in the ability to immunize recipients against tumor challenge [104]. However, tumor-cell vaccination using autologous tumor cells transfected with granulocyte-macrophage colony-stimulating factor (GM-CSF) led to promising results in rodent studies [105,106] and has been introduced into phase I studies [98,107,108]. The main mechanism for these antitumor effects might be the ability of GM-CSF to recruit granulocytes, macrophages, and, most importantly, APC, which could enhance cross-presentation of tumor antigens. Furthermore, it has been shown that GM-CSF leads to the upregulation of CD1 on APCs, suggesting a possible interaction with NK T cells. Another approach to enhancing the immunogenicity of tumor cells is to transfer genes encoding costimulatory molecules into tumor cells to improve their ability to activate T cells by providing signal II, thus overcoming the failure of the natural tumor to provide costimulation. Expression of costimulatory molecules, such as B7, has been shown to significantly enhance the immunogenicity of tumor-cell vaccines [109,110]. The combination of cytokine and B7 expression has further enhanced the immunogenicity of tumor cells [111]. Activation of APC via triggering of the CD40-CD40-ligand pathway can overcome tolerance to tumors [112,113]. In keeping with these results, the APC function of malignant B cells has been enhanced using CD40-mediated stimulation [114,115]. Triggering of the CD40 antigen leads to upregulation of costimulatory molecules, especially B7, and enhances the T-cell stimulatory capacity of these cells [116]. The use of CD40-activated B-cell lymphoma cells for vaccination has also been studied in clinical phase I trials [117]. Nonspecific immune adjuvant strategies for enhancing the immune response are also under investigation, for example, by incorporating bacterial CpG DNA or using Freund's-like adjuvants (Montanide ISA-51) in the vaccine formulation to enhance its immunostimulatory capacity [118,119]. Although still an area of much controversy, it has been suggested that the role of direct immune recognition of tumor cells is of less importance than the indirect presentation of tumor antigens by DCs in determining antitumor immunity [83,120]. Therefore, many efforts have been focused on using DC for in vivo priming of T cells to elicit immune reactivity. DCs are professional APCs that play a key role in the afferent arm of the immune response by picking up antigen in the periphery and presenting it to T cells in the lymph node. Enhancing the stimulatory function of APCs and selection of the optimal methods for achieving antigen presentation by the APC are the key steps in designing a DC vaccine. Antigen loading of APC may be achieved ex vivo by either exogenously pulsing DC with peptides and proteins, cellular debris, exosomes [121], or apoptotic tumor cells [122], or by transfecting DC with DNA encoding tumor-specific or -associated antigens. The advantage of the latter approach is that it is easier to target the endogenous MHC class I pathway than to achieve presentation of exogenous antigen by class I molecules. For this purpose, various viral vectors have been studied in the past [123-125]. The outcome of the interaction between T cells and DC is critically influenced by the maturational stage of the DC. Although mature DCs have a potent ability to activate T cells, immature DC can be tolerogenic, probably because of their inadequate expression of costimulatory molecules and cytokines [49]. Incorporation of APC stimulatory molecules, such as GM-CSF or CD40-ligand [126], into the vaccine formulation has led to significantly improved immune responses compared with those achieved with antigen-loaded DC alone. Furthermore, in this study [126], DC vaccines were shown to be superior to tumor cell vaccines. In animal models, cotransduction of tumors with GM-CSF and CD40-ligand led to a significant increase in infiltrating DC within the tumor [127]. These results suggest that in vivo cross-priming might be enhanced by using tumor cells with improved ability to activate and mature DC. During the last decade, methods have been developed for the ex vivo maturation and large-scale preparation of DC under Good Manufacturing Practice conditions. Therefore, a number of phase I and II studies have been launched investigating the toxicity and clinical efficacy of DC vaccination in tumor-bearing patients. These ongoing clinical studies will reveal the efficacy of such vaccination approaches. However, these trials are usually conducted in patients with advanced diseases, weighing the balance of tumor growth versus the immune response in favor of the tumor. It seems likely that this approach would have the greatest potential for success in the setting of minimal residual disease. All these previously mentioned approaches are limited by the potential problem that T cells with high-affinity TCR for self antigens expressed by tumors are deleted or otherwise tolerized [128,129] in the tumor-bearing host, so that these boosting approaches might not lead to an efficient T-cell response. However, rodent studies have shown that, using vaccination approaches, it is possible to induce low-avidity CTL that can reject tumors bearing TAA that are self antigens [130]. Therefore, efforts aimed at enhancing the TCR-MHC/peptide interaction might help to elicit a CTL response in a tumor-bearing host harboring only low-avidity CTL against TAA. Such a strategy has been developed in rodents using altered peptides that carry amino acid substitutions that enhance the affinity of the MHC/peptide complex for TCR. These so-called heteroclitic peptides elicit significantly improved tumor rejection responses compared with the native peptide [131]. A potential side effect of the above vaccination approaches against TAAs might be the induction of an autoimmune response against normal host cells sharing the antigen [132]. Melanoma is an especially relevant model for studying the induction of T-cell–dependent autoimmunity because the most relevant tumor antigen (Melan-a/MART-1) melanoma is also shared by normal melanocytes. In a clinical setting, proof of this principle has been provided by the fact that adoptive T-cell therapy using tumor-reactive [77] or MART-1–specific T cells can lead to vitiligo in melanoma patients [81]. Furthermore, autoimmunity and antitumor responses were observed in preclinical models using the combination of a GM-CSF–expressing tumor cell vaccine and a blocking anti-CTLA4 antibody. This treatment resulted in powerful tumor rejection that was associated with tissue-specific autoimmune responses, as indicated by vitiligo in the B16 melanoma model. Similar results were obtained in a transgenic mouse model based on the spontaneous development of prostate cancer as a result of the prostate-specific expression of the large T antigen of the SV40 promotor [133-135]. Importantly, a recent clinical trial involving anti–CTLA4 antibody and melanoma peptide administered in 14 patients was associated with multi-organ autoimmunity in 43% and tumor responses in 21% of patients [135a]. The difficulties in activating the afferent immune response have led to the development of strategies focusing on the efferent arm of the immune response (ie, the targeting of tumor cells by effector T cells). One such strategy has involved the generation and expansion of antigen-specific T cells in vitro followed by adoptive transfer of these tumor-reactive T-cell clones [78-80]. Another approach aims at using recombinant antibody technology (using recombinant bispecific antibodies or recombinant chimeric TCRs) [136,137] to specifically target and activate T cells of all specificities at the tumor site, thus circumventing the problem of the limited numbers of high-avidity T-cell clones with reactivity to tumor antigens. The second approach attempts to take advantage of the powerful T-cell response associated with alloreactivity.
Adoptive T-Cell Therapy Similar approaches have been evaluated in melanoma. Although it is possible to detect MART-1–specific T-cell melanoma TIL with apparent high avidity in in vitro assays, it is still unclear whether this is correlated with in vivo tumor reactivity. However, such T cells can be generated either from peripheral-blood lymphocytes or TIL populations. Historically, antigen-specific T cells were generated by in vitro cultivation with antigen-expressing stimulators in the presence of cytokines. These can now be specifically selected and enriched by peptide/MHC tetramers [141]. So far, however, the main concern with the adoptive transfer of antigen-specific T-cell clones or populations has been poor engraftment and survival of the cells, which may be circumvented by the additional administration of IL-2 [142]. In view of the ability of IL-2 to promote activation-induced cell death, it would be of interest to evaluate IL-15, which promotes CD8 T-cell survival but otherwise shares T-cell activating properties with IL-2, as an alternative cytokine in studies of this kind.
Effector T-Cell Targeting Direct recruitment of T cells of multiple specificities has been achieved using bispecific monoclonal antibodies. These antibodies are directed against an activating T-cell antigen, such as CD3, and a surface (tissue restricted) antigen expressed on tumor cells (eg, against CD19 on B cells or human epithelial antigen on epithelial cells). In fact, such bispecific antibodies have been shown to have potent in vitro [148] and in vivo [149,150] activity against tumor cells. Concomitant costimulation with an anti-CD28 monoclonal antibody has been reported to enhance these effects by preventing apoptosis [136]. Currently, phase I clinical trials are testing these new antibodies in patients. Another approach that takes advantage of recombinant antibody technology is the generation of chimeric TCRs in which the extracellular antigen-binding domain of the TCR is replaced by a single-chain antibody and is linked to a signal transducing element of the TCR/CD3 complex (eg, the zeta chain of the CD3 complex) to stimulate intracellular signaling after binding occurs [137]. Several technical problems are still not solved with this approach, including the signaling defects in tumor-bearing individuals discussed above. To circumvent this problem, chimeric constructs have been developed that consist of a single-chain antibody coupled to members of the Syk-tyrosine kinase family [151,152].
Using the Alloresponse to Achieve Antitumor Responses: Lessons Learned From Transplantation Immunology In contrast, T cells recognizing alloantigens, especially MHC antigens, are present at much higher frequencies in the naïve T-cell repertoire and, if obtained from allogeneic donors, would not have been previously tolerized by the tumor. Therefore, alloantigens expressed on a tumor can be targets of rejection by allogeneic cells obtained from MHC-matched or -mismatched donors. Allogeneic bone marrow transplantation (BMT), which was originally performed in leukemic patients to provide hematopoietic rescue after high-dose chemotherapy and radiotherapy that destroyed recipient hematopoiesis, has proved to provide immunotherapy of this kind. The development of graft-versus-host disease (GVHD) was shown to be associated with reduced relapse rates after HLA-identical BMT [153-155]. Although reduced relapse rates were predominantly coupled to the development of GVHD, relapse incidence was also lower in patients receiving HLA-identical sibling grafts in the absence of GVHD compared with that in patients receiving syngeneic BMT [155], indicating that such GVL effects can occur in the absence of GVHD. In the HLA-identical transplantation setting, these alloresponses are directed against minor histocompatibility antigens (HA). The observation of higher relapse rates when T cells were depleted from allogeneic marrow grafts to prevent GVHD [156] confirmed that allogeneic BMT is a form of immunotherapy and that alloreactive T cells in the bone marrow inoculum were major mediators of this immunotherapeutic effect. Thus, the depletion or tolerization of alloreactive T cells in the donor inoculum, which has been used as an approach to prevent GVHD in numerous studies, is conceptually unappealing in the context of certain hematologic malignancies because it removes this beneficial immunotherapeutic effect of alloreactivity. The clinical efficacy of this alloresponse has been further underscored by the results of administration of donor lymphocyte infusions (DLI) to patients in relapse after allogeneic BMT. Impressive and durable remissions have been achieved via this treatment, particularly in patients with chronic myelogenous leukemia (CML) [157-159]. Rodent studies and clinical studies have demonstrated that the DLI-mediated GVL effect correlates with a brisk increase in alloreactive T cells [160,161]. Although GVHD is the most frequent complication after DLI, T-cell dose-escalation studies clearly showed that clinical tumor responses can be achieved by DLI in the absence of GVHD [162]. HLA-identical siblings, who were the major source of donor marrow in the studies that showed the immunotherapeutic effect of T cells in allogeneic BMT, invariably differ at multiple minor HA loci, resulting in an immune response of greater magnitude than those directed against a single or small number of tumor antigens. Minor antigens include determinants recognized by CD4+ and CD8+ cells. Because many of these minor HAs are shared by leukemia cells, it is not surprising that these graft-versus-host alloresponses against minor HAs are associated with GVL. T cells recognizing allogeneic MHC molecules, as in the setting of non–HLA-identical BMT, are present at an even higher frequency (approximately 7%) in the unprimed T-cell repertoire of a normal individual [163]. Thus, the potential immunotherapeutic benefit of MHC-mismatched transplantation is enormous. Indeed, both murine studies [164,165] and clinical data [166] support the conclusion that alloresponses directed against MHC alloantigens lead to even stronger GVL effects than those detected in the setting of MHC identity and minor histocompatibility differences only. Unfortunately, the powerful alloresponse against extensive HLA disparities is associated, as might be predicted, with an unacceptably high incidence of severe GVHD [167], thus far greatly limiting the ability to exploit this alloresponse for the achievement of maximal graft-versus-malignancy effects. Therefore, the development of a means of exploiting the graft-versus-malignancy effect of anti-MHC alloresponses while avoiding the associated GVHD would be highly desirable. Approaches to achieving this goal may be developed through an understanding of the pathophysiologic factors contributing to the development of GVHD. The development of GVHD is not only dependent on the presence of alloreactive T cells within the donor graft. Its development is critically influenced by the toxicity and the inflammatory response induced by conditioning. The conditioning-induced tissue damage is especially apparent in the gastrointestinal tract, where it leads to the translocation of bacterial-derived lipopolysaccharides (LPS) into the circulation and is involved in triggering the inflammatory cascade [168]. This hypothesis is supported by experimental [169,170] and clinical results [171] showing that reducing the level of available LPS leads to a significant decrease in the incidence and severity of GVHD. Furthermore, when mice with defective LPS signaling pathways were used as bone marrow donors, significantly reduced pulmonary and intestinal pathologies were observed in association with reduced tumor necrosis factor alpha levels [172]. Although translating these results into clinical practice is a difficult challenge, a number of experimental strategies have been developed toward this goal. Many of these involve the antagonism of proinflammatory cytokines, the administration of exogenous cytokines, or the administration of cells and treatments that alter the character of the GVH alloresponse, and they have shown partial efficacy [173]. Another approach to achieving the same goal is to control the trafficking of alloreactive T cells so that they stay inside the lymphohematopoietic system, where leukemias and lymphomas largely reside, and do not migrate to the GVHD target tissues, which include skin, liver, and the intestinal system. This has been achieved with several approaches. One is to administer DLI to established bone marrow chimeras late after BMT, at a time point when the conditioning-associated proinflammatory response has disappeared. Indeed, rodent studies using mixed chimeras demonstrated that the delayed administration of nontolerant donor lymphocytes resulted in conversion from mixed hematopoietic chimerism to full donor chimerism without any associated GVHD [174]. This demonstrated that nontolerant donor T cells can mediate a GVH response limited to the lymphohematopoietic system without causing GVHD. One possible explanation for this observation is that inflammation in the GVHD target tissues plays a major role in directing the migration of GVH-alloreactive T cells into these tissues and that, in the absence of this inflammation (eg, after sufficient time has passed for conditioning-induced inflammation to subside), the alloresponse will remain within the lymphohematopoietic system, where leukemias and lymphomas reside. An alternative possibility, that host-type veto or suppressor cells mediate this protection from GVHD in established mixed chimeras, seems less probable because fully allogeneic chimeras are also resistant to the induction of GVHD by the delayed administration of DLI [174]. Although suppressor cells of donor [175] and host [176] origin have both been implicated in protection from GVHD in mice receiving delayed DLI in various models, studies in the mixed chimera model show that a potent antihost alloresponse occurs despite the absence of GVHD, because alloreactive donor T cells in DLI mediate the destruction of normal host hematopoietic cells and the eradication of host-type leukemias [160,174,177-179]. Studies in a nonmyeloablative model showed no evidence for a role for suppressive host or donor T cells in the resistance to GVHD of mixed chimeras receiving DLI 35 days after BMT (Mapara et al, manuscript in preparation). The DLI-mediated GVH alloresponse is associated with potent GVL effects. This GVL effect, however, is critically dependent on the presence of host APC, as is evident in the markedly superior DLI-mediated GVL effects observed in mixed hematopoietic chimeras compared with fully allogeneic chimeras [179,180]. These results are in accordance with the observation [181] that host APCs at the time of transplantation play a critical role in driving the development of GVHD. Thus, direct immune recognition of host APC by donor T cells is probably the driving force behind both GVHD and GVL. The major variable controlling the outcome of this alloactivation (ie, whether or not GVL is accompanied by GVHD) is the time point when alloreactive T cells are administered to the host. As mentioned previously in this section, administration of donor lymphocytes to patients in relapse after allogeneic hematopoietic cell transplantation has clearly demonstrated the clinically relevant immunotherapeutic effect of DLI [157,158]. In this setting, impressive clinical remissions, particularly against chronic-phase CML, can be achieved by the administration of allogeneic cells without further cytoreduction. However, in the clinical situation, GVHD does occur after DLI, although it is frequently less severe than what would be expected from a similar number of T cells administered to a freshly conditioned recipient. Recently, tumor regressions have been reported in a small group of patients who received allogeneic lymphocyte inocula without prior hematopoietic cell transplantation [182]. The strategy of reducing the intensity of conditioning, combined with induction of mixed chimerism followed by delayed DLI, is highly promising as a clinical approach to achieving GVL without GVHD. On the basis of a murine model [177] and a nonhuman primate model [183], a clinical protocol for the induction of mixed chimerism followed by DLI has been evaluated for the treatment of hematologic malignancies. Conditioning involved in vivo T-cell depletion with equine antithymocyte globulin given before and after transplantation, cyclophosphamide, thymic irradiation, and a short course of cyclosporine as the only posttransplantation immunosuppression given in addition to antithymocyte globulin. A remarkably high success rate was achieved with this approach in patients with advanced, refractory lymphoid malignancies [184-186] in whom other therapies had been uniformly unsuccessful. In some patients who received delayed DLI with this approach, striking antitumor effects against advanced hematologic malignancies were observed without or with only mild GVHD. Although the factors predicting the development of GVHD after DLI have not been identified, it seems probable that improved T-cell depletion of the initial donor stem-cell inoculum to avoid even subclinical GVHD and, therefore, permit the recovery of a noninflamed environment in the epithelial GVHD target tissues before DLI are given, should allow better control of this approach in separating GVHD and GVL. In support of this possibility, recent studies involving in vivo T-cell depletion of the recipient with a more powerful agent than the antithymocyte globulin used in previous studies, in combination with administration of an increased dose of ex vivo T-cell–depleted haploidentical allogeneic hemopoietic stem cells, have permitted demonstration in a small number of patients that DLI can convert mixed to full donor T-cell chimerism in this extensively HLA-mismatched setting without causing severe GVHD [187]. Another approach to achieving in vivo T-cell depletion used the humanized anti–CD52 antibody Campath-1H (Ilex Pharmaceuticals, San Antonio, TX), which also leads to an almost GVHD-free posttransplantation period [188,189]. The major reason for this highly efficient prevention of GVHD might be ascribed to the fact that Campath antibodies also deplete DCs [190]. This phenomenon might occur at the expense of GVL effects, as exemplified for multiple myeloma, which seems to be sensitive to allogeneic GVL effects [191]. However, Campath-containing conditioning regimens have been reported to result in only transient GVL effects against multiple myeloma [192]. These observations underscore the clinical relevance of our experimental data showing the more pronounced GVL effect obtained in mixed chimeras compared with full chimeras. The approach of DLI administration to establish mixed chimeras relies on a delay in the administration of GVH-reactive T cells and the use of less toxic initial conditioning as the main strategies for separating GVHD and GVL. Within the last 5 years, a number of reduced-intensity protocols have been developed for clinical BMT. Some of these regimens are clearly nonmyeloablative [185,193], whereas others may be considered to be to intermediate dose-reduced protocols [188,189,194-196]. Dose reduction clearly decreases transplantation-related morbidity and mortality [188,189], and the approach attempts to capture the immunotherapeutic benefit of allogeneic hematopoietic cell transplantation instead of relying on high-intensity chemotherapy and radiotherapy for tumor cytoreduction. Although antitumor effects have been clearly seen in the setting of less aggressive hematologic malignancies, this dose reduction might be associated with an increased risk of relapse in patients with highly proliferating malignancies (acute leukemias and accelerated-phase or blast crisis CML) that might out-run the immunologically mediated GVL effect induced by DLI. Most of the clinical conditioning protocols do not specifically aim to establish mixed chimerism, and spontaneous, rapid development of full donor chimerism is often seen. In recipients of these protocols, DLI are given if mixed chimerism persists. Administration of DLI in this context has frequently been associated with GVHD [197]. The reason for this probably relates to the fact that conditioning protocols do not involve complete T-cell depletion of the donor product given immediately after conditioning. Therefore, GVH reactions occur initially, potentiating conditioning-induced inflammation that likely predisposes the recipients to GVHD after DLI. Indeed, the incidence of GVHD after the initial hematopoietic cell transplant has not been significantly reduced by the use of reduced-intensity conditioning protocols. On its own, reducing the toxicity associated with the conditioning regimen needed for achieving engraftment of allogeneic hematopoietic stem cells seems to be insufficient to prevent GVHD when unmodified donor peripheral-blood stem-cell inocula are administered around the time of conditioning. One limitation of the approach involving induction of mixed chimerism with a regimen that includes donor T-cell depletion, then giving immunotherapy in the form of DLI several weeks or months later, is that many hematologic malignancies are rapidly progressive, making the delay (needed for conditioning-induced inflammation to subside) before DLI is administered as immunotherapy undesirable. Thus, it would be desirable to develop an approach to confine the GVH alloresponse to the lymphohematopoietic system even in the presence of such inflammation so that optimal GVL effects could be obtained immediately after hematopoietic cell transplantation. Recent studies using the new immunosuppressive agent FTY720, which acts primarily to trap lymphocytes within the secondary lymphoid tissues and prevents their migration into solid organs, indicate that GVHD can be prevented and GVL preserved by controlling lymphocyte trafficking in freshly conditioned recipients [198]. Other approaches to separating GVHD and GVL attempt more directly to focus the allogeneic donor T-cell response on lymphohematopoietic antigens or tumor antigens [199,200]. In patients with CML, clinical antitumor responses after allogeneic BMT or interferon alpha therapy were associated with the detection of PR-1 tumor antigen-specific T cells [201]. However, efforts to exploit tumor antigen-specific responses come with the risk of decreasing the potency of the (primarily alloantigen-driven) GVL effect. A recently described strategy involved the administration of primed T cells specific for a single minor HA shared by the tumor and normal host cells. As long as T cells recognizing other minor antigens of the host were not coadministered, GVL could be achieved without GVHD [202]. These results suggest that spreading of GVH alloresponses to multiple epitopes may be important for the development of GVHD and that confining the response to a single antigen may prevent the disease. However, responses to a single minor antigen may be insufficiently potent to mediate significant GVL effects without extensive ex vivo priming and expansion. In humans, minor antigens have been described that are relevant for GVHD and GVL reactions. Goulmy et al [203] and den Haan et al [204] have identified a number of minor antigens that serve as targets for GVH responses in recipients of allogeneic BMT. Using MHC tetramers, antigen-specific T cells were detectable in patients with GVHD [205]. In a female-to-male in vitro skin explant system, it was shown that the response was directed predominantly against ubiquitously expressed Y chromosome–associated antigens (HY) and that hematopoietic-tissue–restricted minor antigen-specific T cells (HA-1,2) were inert as long as the skin tissue was not preincubated with the peptide from minor antigen (HA-1,2 peptide) [206]. However, disparities in these hematopoietic antigen-restricted epitopes have been associated with an increased incidence of GVHD, consistent with the possibility that lymphohematopoietic GVH reactions induce GVH responses that may spread to other tissue specificities under the appropriate proinflammatory conditions and induce GVHD. Nevertheless, the exploitation of minor antigen-specific T-cell responses might provide a strategy for achieving GVHD without GVL. Another interesting approach is based on the antigen-specific recognition of TAAs by allo-MHC–restricted CTLs. Dahl et al [207] have demonstrated that tumor cells from H2b mice could be lysed by BALB/c lymphocytes (H2d) specific for mdm-2 presented by H2b, whereas normal cells expressing low levels of mdm-2 were not killed. Syngeneic T cells from H2b mice were unable to lyse the tumor cells because they only contained low-avidity CTL. The allogeneic T cells were also proven to be efficient in vivo in the BMT setting. In a fully MHC-mismatched strain combination (B10.A to B6; H2a to H2b), mdm-2–specific Kb-restricted CTL clones derived from BALB/c (H2d) mice, which share one class I allele with the B10.A bone marrow donor, were injected after lethal irradiation on day +2. These cells were engrafted in the recipient and were detectable up to 14 weeks after injection and did not induce GVHD [208]. Using this approach, it has also been demonstrated that allorestricted human CTLs specific for WT1, a transcription factor expressed in normal CD34 cells and upregulated in CML cells, were able to eliminate CD34+ progenitor cells from patients with CML but did not affect normal CD34+ progenitor cells [209]. Another approach to overcoming tolerance to a TAA [210] used HLA-A2.1 transgenic mice. Mdm-2–specific mouse T cells were generated. The TCR from these T cells were cloned, and a partially humanized TCR was constructed and transduced into HLA-A2–positive human peripheral-blood mononuclear cells. Transfer of the transgenic TCR into the peripheral-blood mononuclear cells led to specific lysis of mdm-2 expressing A2-positive target cells. These results open the possibility of transferring these custom-made, tumor-reactive, high-affinity TCRs into T cells for cellular therapies. On the basis of the assumption that an allogeneic antimalignancy response (graft-versus-tumor [GVT]) might also be functional against solid tumors, which have a documented or suspected sensitivity to immunotherapeutic strategies, clinical studies have been initiated to evaluate the role of allogeneic transplantation in patients with malignant melanoma, RCC, or even breast cancer. The most promising results were obtained in RCC patients. However, clinical remissions occurred rather late and were only achieved in patients who had developed GVHD [211]. Because RCC is generally resistant to chemotherapy, clinical responses can be attributed to the GVT effect of the allogeneic transplantation. As suggested by the results in RCC, it is difficult to see how this GVT effect could be achieved without accompanying GVHD in the context of solid tumors that exist in nonhematopoietic parenchymal tissues. Thus, the ability to exploit the alloresponse to achieve powerful antitumor effects without GVHD using allogeneic hematopoietic cell transplantation may be restricted to lymphohematopoietic tumors, such as leukemias and lymphomas, in which confinement of the alloresponse to the lymphohematopoietic system could eliminate the tumor without causing GVHD. However, in addition to the powerful antitumor effect of the alloresponse, there are reasons why allogeneic T cells might mediate superior tumor antigen-specific responses compared with autologous T cells. These relate to the global defects in T-cell signaling that have been observed in tumor-bearing hosts and to the possibility that exposure to tumor antigens on nonprofessional APC may have tolerized any tumor antigen-specific T cells initially present in the host. For these reasons alone, the transfer of T cells from a normal donor might lead to improved antitumor responses. Because most people do not have an identical twin syngeneic donor, these T cells must usually be obtained from allogeneic donors. A variety of approaches to tolerizing or depleting allogeneic T cells that recognize recipient alloantigens, with efforts to preserve antitumor responses, have thus been developed. These include stimulation of donor T cells with host antigens in the presence of CTLA4Ig [212] or depletion of alloactivated T cells using antibodies or immunotoxins directed against cell surface markers associated with recent activation, such as CD25 [213] and CD69 [214]. It is not yet clear whether strong antitumor effects can be achieved with this approach. In view of the rarity of tumor antigen-specific T cells, it seems likely that the magnitude of any such tumor-specific responses would be limited and that this approach might be most effective in the presence of highly immunogenic tumors and minimal residual disease or in combination with approaches to expanding the rare tumor-reactive T cells. It is of interest that the delayed antitumor responses seen in a fraction of patients with advanced RCC receiving nonmyeloablative allogeneic BMT occurred after GVHD had subsided [211], suggesting that tumor antigen-specific responses may have slowly emerged from the transferred allogeneic T-cell repertoire [215]. Alternatively, subclinical GVH responses might be ongoing in these patients after overt GVHD has subsided, leading to the observed GVT effect. Recently, it has been demonstrated in a mouse model that DLI can activate host T cells with tumor antigen-specific reactivity, leading to regression of solid tumors. On the basis of the clinical observation that spontaneous loss of chimerism after nonmyeloablative hematopoietic cell transplantation can, in some instances, be associated with striking tumor responses [216,217], we recently demonstrated in a mouse model that intentional rejection of donor marrow induced by recipient lymphocyte infusions can, paradoxically, induce antitumor effects against host-type tumors [218]. These results suggest that alloresponses in the GVH and host-versus-graft directions may have as yet unappreciated immunostimulatory effects that lead to destruction of tumors.
Increasing evidence has demonstrated that the immune system is able to mount responses against tumors and that this immune response can be enhanced using a number of strategies. Several of these strategies are currently being evaluated in clinical trials, where their efficacy and cost effectiveness will be ascertained. The immunotherapy that has been shown to most reliably avoid the problem of self tolerance and mediate clinically significant antitumor effects is allogeneic cell therapy. However, allogeneic cell transplantation is still limited and overshadowed by the potentially lethal complication of GVHD. New developments in this field provide hope that the power of the alloresponse might be used in the future with less danger of inducing GVHD.
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Acted as a consultant within the last 2 years: Megan Sykes, Bio Transplant, Inc. Performed contract work within the last 2 years: Megan Sykes, Bio Transplant, Inc. Served as an officer or member of the Board of a company: Megan Sykes, Bio Transplant, Inc. Received more than $2,000 a year from a company for either of the last 2 years: Megan Sykes, Bio Transplant, Inc.
Supported in part by National Institutes of Health/National Cancer Institute grants 1 R01 CA 79989-03 and 1 R01 CA 79986-03. Authors' disclosures of potential conflicts of interest are found at the end of this article.
1. Schwartz RH: Immunological tolerance, in Paul W (ed): Fundamental Immunology. Philadelphia, PA, Lippincott Williams & Wilkins, 1999, pp 701-739
2. Zhang L, Conejo-Garcia JR, Katsaros D, et al: Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer. N Engl J Med 348:203-213, 2003 3. Nakano O, Sato M, Naito Y, et al: Proliferative activity of intratumoral CD8+ T-lymphocytes as a prognostic factor in human renal cell carcinoma: Clinicopathologic demonstration of antitum |
TOP
ABSTRACT
INTRODUCTION
