A sense of tumour for the immune system

The field of tumour immunology encompasses the study and description of immunological properties and processes associated with tumours, which affect or control the way in which tumours are sensed by the immune system. Recently, a great deal of interest in the relationship between the immune system and tumours has been reflected in a large number of publications in the field. It has emerged that the immune system has not only a role in preventing the development of many tumour types but also a potential role in eliminating or containing the growth of established tumours. Based on laboratory research, an array of clinical trials of cancer immunotherapy are being undertaken and, although in their early phases, some have shown immunological and/or clinical effects. As more mechanisms are deciphered in immunological processes generally, and a greater knowledge of the interaction between the immune system and tumour emerges, it may become possible to design rationally immune therapies that are complementary, or even preferable to certain therapies in current use. Some of the important issues surrounding tumour immunotherapy include the heterogeneic nature of cancers in terms of their origin, location and other phenotypes, and their ability to evade immune responses. In particular, the ways in which the immune system prevents autoimmunity – mechanisms that need to be circumvented during certain immunotherapies, are only recently clearly emerging. The complicated nature of the immune manipulations required may pose obstacles to providing widely available therapies, especially as high technology and a very personalized regimen may be suggested by preclinical research.

This issue of Immunology aims to highlight certain recent findings in tumour immunology and immunotherapy, but without duplicating the content of other excellent extensive reviews.^1,2 This introductory article sets out some of the main issues in the field and is followed by articles which focus more specifically on areas of brain tumour immunology, leukaemia, microbial-based immunotherapy, and regulatory T cells.

As in every area of immunology, tumour immunology dogma is constantly evolving to accommodate new findings. Although there are numerous basic immunological principles revealed by tumour immunology, most research has some bearing on potential immunotherapeutic approaches. To this end, much is owed to technological advances in the identification of tumour-associated antigens (and measuring immune responses to these, e.g. at the single-cell level with tetramers and by ELISPOT), knockout and transgenic mouse development, gene therapy vector design, and the ability to propagate patient antigen-presenting cells (APC), in particular dendritic cells (DC), for adoptive immunization. Moreover, a need to provide new treatments is driving extensive experimentation.

Although early observations in nude mice suggested that these T-cell-deficient mice are not more susceptible to tumour development, more stringent knockout mice lacking interferon-γ (IFN-γ)-signalling, Rag, or perforin genes show a clear role for these T-lymphocyte molecules.³ Furthermore, lymphocytes controlling tumour emergence have been shown to be either T cells of the innate γδ T-cell lineage, or natural killer (NK) cells.^4,5 For both of these interactions a number of novel ligands and receptors have been identified. Moreover, elimination of tumour cells by NK cells appears to generate specific T-cell-mediated immunity.^5,6 This may question an aspect of the ‘danger theory’ which proposes that cytotoxic killing of target cells does not potentiate immunity.⁷

The role of tumour-specific T cells in controlling spontaneous tumour emergence may be less important than their therapeutic effect on established tumours. An overall consensus on established tumours (and transplantable model tumours) is that they are not being contained by the adaptive immune system for at least one of a number of reasons. They may be poor targets for immune recognition because of a lack of antigen-presenting molecules or a lack of actual potent antigens, and are thus unlikely to initiate directly the immune responses. A lack of associated immune stimulus (danger) may be an obstacle, together with possible tolerization to their ‘self’ tumour antigens. Tumour location may limit detection by the immune system. Additionally, tumours may cause local or generalized immune suppression by expressing various molecules [such as FasL, transforming growth factor-β, and interleukin-10, and cytotoxic T lymphocyte (CTL)-inhibiting serpins⁸], and they may actively switch off or kill T cells that recognize them.⁷

Enhancing immune responses specifically against tumours, particularly by increasing numbers of tumour-specific activated CTL and T helper type 1 (Th1) (IFN-γ-secreting) cells, is considered to be a way of circumventing the associated immune deficits. The question posed then is how to achieve this. The most basic cancer vaccines, and ones in greatest use at present in clinical trials, are whole cell vaccines based on tumour cells of autologous (self) or allogeneic [major histocompatibility complex (MHC)-disparate] origin,⁹ or derived from patient DC. Whole tumour cell vaccines, which usually contain some form of immunological adjuvant either admixed (e.g. with mycobacteria) or transfected [e.g. with granulocyte–macrophage colony-stimulating factor (GM-CSF)], are irradiated and administered in a number of doses. Allogeneic vaccines may additionally be able to enhance immunity through an anti-graft reaction. Whole cell vaccines appear to work best when certain intrinsic mechanisms designed to control autoimmunity are removed, such as the regulatory CD4⁺ CD25⁺ T cells¹⁰ and T-cell-inhibitory CTL-antigen-4 (CTLA-4) engagement.¹¹ Tumour-specific regulatory T cells have yet to be identified, but there are likely to be regulatory T cells specific for normal self-tissue antigens expressed by tumour cells, and these may inhibit the generation or function of T cells specific for tumour antigen in the context of whole tumour cells. Thus, the type of T-cell response may be critical in addition to its presence.

There remains some debate as to whether whole tumour cells are more efficient at priming T cells via host APC (cross-priming)¹² or directly.¹³ This may depend on the model being used. If haematological malignancies are investigated it is feasible that they can prime T cells directly because they have greater access to lymphoid tissue and possess more co-stimulatory molecules than epithelial-derived tumour cells. The influx into lymphoid tissue of large numbers of such cells expressing foreign antigens (viral antigens in the case of ref. 13) at a high cell surface density may well resemble APC influx during an infection and so will be highly immunostimulatory. Nevertheless, cross-priming of cell-associated tumour antigens by host APC is considered most important and efforts to enhance this process use GM-CSF-gene transduction.¹⁴ More recently, whole tumour cells have been fused to DC to produce vaccine cells capable of directly priming T cells, and these have demonstrated some clinical and immunological effects.¹⁵ A number of studies have suggested that generation of effective anti-tumour immunity, particularly CTL, either requires additional help from Th cells, or reproduction of their effect by ligating receptors on DC.¹⁶

Immune modulation can also be directed to an in situ tumour, e.g. mycobacteria for bladder cancer or adenovirus encoding interleukin-12 for non-resectable tumours. Transduction of tumours with genes encoding enzymes that impart sensitivity to administered prodrugs (‘suicide gene therapy’) has been shown to generate local and systemic immune responses which greatly amplify their effect. Thus, although the direct killing of tumour cells per se is highly desirable, adaptive immunity generated in this process is highly beneficial against disseminated disease. In the HSVtk/GCV suicide gene therapy system necrotic cell death appeared more immune-stimulatory than apoptotic, with the expression of stress-induced danger signals such as heat-shock protein 70.¹⁷ This makes sense because an immune response is not beneficial when associated with apoptotic death that occurs during normal tissue remodelling. In contrast, immunity is beneficial following necrotic cell death associated with infection and tissue damage. It is tempting to speculate that stress and necrosis associated with NK cell-mediated tumour destruction may enhance adaptive immunity via heat-shock proteins.¹⁸

While apparently a crude approach, the use of whole tumour cells for cancer immunization continues to be investigated because tumour cells comprise an array of tumour antigens that can match tumours in vivo, and which do not necessarily need to be identified. Moreover, tumour cells are potentially efficient at contributing to the process of T-cell priming.

The next generation of cancer vaccines under investigation include subunit or component vaccines based on tumour-associated antigens, particularly those antigens that promote tumour rejection. Mirroring many vaccines against pathogens, the evolution of cancer vaccines has moved from whole cells, to extracted proteins (e.g. antigen-carrying heat-shock proteins¹⁹) and recombinant proteins, to synthetic peptides and DNA/RNA-based vaccines. The mode of antigen delivery is critical to the successful generation of immunity. Protein or peptide antigens are delivered in vehicles that enhance antigen uptake and also stimulate APC, such as oils and liposomes. DNA vaccines are administered as naked DNA or carried by microbially derived vectors that are not only efficient at delivering their genes but also have stimulatory effects on the immune system. Examples of these vectors are those derived from adenovirus, herpes virus and retrovirus, and Mycobacteria, Listeria and Salmonella. Such vectors can be engineered to comprise a number of tumour antigens in an attempt to avoid the limitations of host MHC restriction. However, specific immune responses against vectors may reduce their ability to deliver tumour genes repeatedly, and so vectors with minimal immunogenicity are being developed, or the vector may be varied for priming and boosting.

Short-circuiting the antigen-delivery issue, studies have used DC generated from patients' blood, pulsed with antigens (tumour cell lysates or peptides), and re-administered; and have demonstrated some clinical and immunological effects.²⁰ However, such studies generally need to standardize their methodologies and resulting cell populations for meaningful comparisons between studies to be made. Most recently, T cells are being modified to combat tumours either by their transfection with tumour-specific T-cell receptors²¹ or by using specific T cells to home to tumours and deliver immunotherapeutic viruses.²²

The concept that tumours are ignored by the immune system is suggested to be a result either of their location, such as in ‘immune-privileged’ sites of the brain, testes, or even subcutaneous sites, or of the lack of immune stimuli associated with tumours. Thus, for immune therapy to be successful it appears that a specific immune response needs to be generated of appropriate type, sufficient magnitude, and correct location. Clearly, immune responses do occur in such ‘immune-privileged’ sites, such as those that must eliminate viruses and bacteria. Safeguards are also required to avoid collateral damage to healthy tissue, especially in delicate areas such as the brain. The resistance of tumours to immune-mediated killing will reflect not only their histological origin and current location but also features acquired during their transformation from normal to malignant cells. Although it may appear daunting, overall, the integration of information on properties of tumours, immune-regulatory processes and vaccination approaches is bringing us closer to providing immunotherapeutic options for treating cancer.

This article highlights certain important points concerning current thinking on tumour immunology and immunotherapy, particularly issues on tumour location, histological origin, methods of immunotherapy, and avoidance of autoimmunity, all subjects to be expanded upon in the following articles.

Table 1. Issues Surrounding anti-tumour immunity.

Immunologically related features of tumours promoting/affecting their persistence	Immunological approaches to combat tumours
Histological origin giving certain phenotype	Whole tumour cell vaccines of autologous or allogeneic origin ± djuvants; transfection with cytokine-, or co-stimulatory genes
Location of primary tumour or metastases
Lack of associated immune stimulus
Generation of tumour antigen-specific immune tolerance or inappropriate type of immune response (Th1/Th2/Tr…)	Fusion of tumour cells with ex-vivo DC for vaccines
	In situ transfection of tumours with cytokine or ‘suicide’ genes
Anti-inflammatory modes of cell death	Tumour antigen; in lysate, protein or peptide form, with adjuvant or presented by ex-vivo-pulsed DC
Limited or lack of target antigen
Reduction or lack of antigen presentation capability	Viral, bacterial, or naked DNA vectors for tumour antigen immunization
Expression of immune-suppressive molecules
T-cell homeostatic mechanisms (CTLA-4)	Provision of actual T-cell help, or mimicking this
Regulatory T cells (CD25+; IL-10-secreting)	Inhibition of CTLA-4 engagement
	Depletion of CD25+ regulatory T cells
	Generate an appropriate immune response or re-direct an inappropriate one

An inexhaustive selection of immunological features of tumours, and approaches to combat tumours.