Although immune cells infiltrate tumours, they are frequently unable to mount a full-scale attack, because tolerance mechanisms prevent them from destroying cells that express 'self' antigens. Immunization of patients with cancer antigens has only been shown to cause a slight increase in the number of tumour-specific circulating lymphocyte numbers, but has not been observed to promote tumour regression. In the 19 September issue of Science, Steven Rosenberg and colleagues describe a new immunotherapy approach that replaces up to 90% of a patients' normal lymphocytes with activated tumour-specific ones, resulting in regression of metastatic melanoma.

Thirteen patients with refractory metastatic melanoma received high levels of chemotherapy to deplete their immune systems, followed by administration of tumour-specific T cells and high doses of the cytokine interleukin (IL)-2. The T cells were initially isolated from the patient's own tumour samples and expanded in vitro, and shown to react against melanoma cells. Six of these patients experienced significant regression of metastatic melanoma, whereas four others had mixed responses with significant shrinkage of one or more metastases. The responses lasted for up to 24 months after therapy.

But what tumour antigen activated this immune response? Analysis of peripheral-blood cells isolated from the patients revealed that they contained a large percentage of T cells that specifically recognized the MART1 antigen — a non-mutated differentiation antigen that is expressed by both melanomas and normal melanocytes. The high frequency and extended persistence of individual T-cell clones that react to non-mutated self antigens has not been previously observed in humans.

The authors suggest that this approach succeeded when previous immunotherapy strategies failed because the expanded cultures that were used to treat the patients included both helper and cytolytic T cells. The non-myeloablative conditioning regimen that was used in this study might also have eliminated regulatory cells or other mechanisms that limit lymphocyte expansion.

Five of the patients who underwent tumour regression also experienced autoimmune melanocyte destruction, leading to vitiligo. This study shows, however, that 'self antigens' can be used as targets for human cancer immunotherapy, and, if autoimmunity can be controlled, similar approaches might be useful in treating other cancers and viral diseases.