It has long been suspected that the immune system may be able to recognize tumours and destroy them, as it does an allogeneic transplant or a parasite. There is now good evidence for the old hypothesis that naturally occurring tumours are eliminated or contained by the immune system (‘immune surveillance’). This hypothesis predicts that the frequency or progression of tumours increases in immunosuppressed individuals, a prediction that was initially borne out by studies on virally induced tumours, but has recently been extended to other more common types. Immunologists have therefore hoped that by appropriate stimulation of stronger innate or specific immunity (vaccination) the immune system could contribute to the eradication of cancer. In the last few years the enormous effort devoted to this problem has begun to be translated into some clinical successes and the mood is one of cautious optimism.
Many mechanisms can contribute to tumour control, including those of both innate (e.g. NK cells, macrophages, cytokines) and adaptive immunity. Attention has been focused on the identification of tumourspecific B- and T-cell antigens, although it now seems likely that tumouassociated antigens (TAAs), normal proteins found more frequently or at higher levels on tumour cells than on normal tissues, are equally important. Older research concentrated on the study of experimentally induced tumours in animals, but probably these very fast-growing and aggressive tumours are much easier for the immune system to recognize than the more typical human tumours that usually develop gradually over years or even decades. Instead, attempts are being made to identify the naturally occurring immune responses to tumours in patients with cancer.
Nevertheless, tumours continue to pose formidable challenges to the immunologist. In its relationship to the host, a tumour cell (yellow, brown and black in figure) is rather like a successful parasite, but with special additional features. Parasite-like mechanisms that help prevent elimination include weak antigenicity and extensive cross-reaction with self; immunosuppression and tolerance induction; release of soluble antigens; antigen antibody complexes; and antigenic variation.
In mice, chemicals such as methylcholanthrene and benzopyrenes tend to induce tumours, each with unique ‘idiotypic’ antigens, whereas most of the common human cancers result from a slow and gradual accumulation of mutations in the genes of proteins that regulate the cell cycle. Such mutations can result in over-activation of a protein promoting cell growth (encoded by cellular oncogenes) or inactivation of a protein that normally slows down cell growth (encoded by tumoursuppressor genes). Some of these mutations are inherited, while others may result from exposure to chemicals in the environment. Normally, it requires several mutagenic events, which can occur over many decades, before a tumour develops. The mutated forms of these proteins may act as possible specific antigens for the adaptive immune system, especially the cytotoxic T cell.
BCG (an attenuated tubercle bacillus) has been tried against melanoma, sarcoma, etc., especially combined with other treatments. Its major immunological effect seems to be macrophage activation, but it may also affect NK cells. A tremendous range of bacterial and other immunostimulating agents has been tested for antitumour activ- ity (see Fig. 41), but so far with very limited success.
Cytokines The dramatic effects of ‘Coley’s toxin’ (a bacterial extract) 100 years ago may have been due to the vigorous induction of cytokines. Following the success of TNF in animals, numerous individual cytokines have also been tried on cancer patients. However, it is now becoming clear that inflammation, and excessive TNF-α production, can actually promote tumour growth, partly by increasing the blood supply to the tumour (angiogenesis). At present only IFNγ and IL-2 are in clinical use against some cancers, although a more targeted delivery to the site of the tumour (e.g. by gene therapy) may extend this approach.
MAC, NK Macrophages and natural killer cells (see Figs 8 and 15), especially when activated, can prevent growth of some tumours in vitro (‘cytostasis’) or actually kill them (‘cytolysis’). NK cells are also cytotoxic, and are activated by cells that have lost expression of MHC molecules, a common phenotype of many tumours. IFNγ is important in activating macrophages and NK cells. Some tumour cells can apparently activate complement via the alternative pathway. However, note that there are potential dangers of activating macro- phages and inflammation as discussed in the paragraph above.
Lymphocytes Tumours often contain large numbers of tumour- infiltrating lymphocytes (TILs), and the number and type of these cells can sometimes predict the rate of tumour progression. TILs are believed to be enriched for lymphocytes specifically recognizing the tumour cells, and such cells extracted from the tumour itself, expanded and then reinjected, have in some cases been successful in causing tumour rejection. Lymphocytes from the blood of tumour patients, activated non-specifically in vitro by IL-2 to kill (LAK cells) have also shown some promise.
Tumour antigens In the case of tumours induced by viruses, the viral antigens themselves are the target of the host immune response (see below). In non-viral tumours, the identification of TAAs has been much more difficult. In rare cases, embryonic antigens absent from normal adult cells may be re-expressed when they become malignant. Carcinoembryonic antigen (CEA) in the colon and α-fetoprotein in the liver are examples of diagnostic value. Other antigens found on the surface of some tumours are glycosylation variants of normal cell proteins (e.g. MUC-1 on epithelial tumours). However, it seems that the majority of antigens recognized by the host’s cellular immune response are normal self proteins, which are expressed at higher concentrations than normal in the tumour cells (sometimes because they are required for cell division). Unfortunately, it seems as though tumours are very heterogeneous and antigens common to a large number of tumours have been difficult to identify.
Viruses were once thought to be responsible for many human tumors, but most common cancers are now thought not to be virally induced. However, five important forms of cancer are firmly linked to viruses (all DNA): Burkitt’s lymphoma and nasopharyngeal carcinoma (EBV), Kaposi’s sarcoma (KSHV), hepatocarcinoma (hepatitis B virus [HBV]) and cervical cancer (papillomavirus). RNA retroviruses may be responsible for some other cases. Interestingly, all these tumours increase in frequency in immunosuppressed individuals (Kaposi’s sarcoma, for example, is commonly found in AIDS patients; see Fig. 28). Marek’s disease, a tumour of chickens, was the first example of the introduction of a successful tumour vaccine. HBV vaccination lowers the risk of hepatocellular carcinoma by preventing viral infection, and a vaccine against papillomavirus prevents most cases of cervical cancer
Antibody There is little evidence that antibody normally provides any host immunity to tumours. Nevertheless, passive immunization using antibodies against two TAAs, CD20 on B-cell lymphomas and Her2/ neu on epithelial cells, has been the first major success of tumour immunology, and these have entered the standard repertoire of drugs used by oncologists for treating these diseases. Much effort is under- way to extend these successes to other tumours, and several other antibodies are in advanced stages of clinical trial. Another approach is to enhance the effectiveness of antibodies by coupling them to potent cytotoxic drugs (‘magic bullet’). This aims to build up very high levels of anti-cancer drug in the immediate vicinity of the tumour, thus minimizing the general toxicity of the drug, which limits the concentrations that can normally be used for chemotherapy.
Cell-mediated immunity Cytotoxic CD8 T cells capable of lysing tumour cells in vitro have been isolated from both mice and humans (especially from individuals with melanoma). In mice such cytotoxic T cells can eliminate a tumour in vivo. Many tumours evade this by reducing their expression of MHC class I antigens. TH1 cells are also probably very important, because they can activate macrophages and NK cells via the release of IFNγ and are also essential for good CD8 T-cell memory. However, weak T-cell reactions may actually stimulate tumour growth and metastasis. Recent promising clinical vaccination trials using melanoma antigens have given a strong further impetus to this work, and there is also the possibility that T cells could be ‘redirected’ against a target tumor antigen by gene therapy of specific T-cell receptors (see Fig. 12).
Dendritic cells (see Fig. 4) are the most potent activators of cell- mediated immunity and it is therefore not surprising that many approaches have attempted to harness these cells for immunotherapy. One approach is to isolate dendritic cells from a patient, load them with tumour antigens and then reintroduce them into the body. Although these patient-specific adoptive immunotherapy procedures are difficult and expensive to implement, a dendritic cell-based immunotherapy for prostate cancer has recently been licenced in the USA, and further treatments of this type are likely to follow.
Breaking tolerance The immune response to most tumours is probably limited by the strong regulatory mechanisms that operate to prevent autoimmunity and maintain tolerance (see Fig. 22). Many strategies aimed at interfering with these mechanisms, and hence obtaining more effective immune responses are being explored. These include blocking molecules on the T-cell surface such as CTLA4 which transmit negative signals, depleting TREG cells and using gene therapy to produce large populations of T cells that carry specific receptors for tumour antigens. New biological drugs based on these strategies are now entering the clinic, and there is great excitement about their potential. Note that treatments may involve some unavoidable side effects in the form of autoimmunity (see Fig. 38).