Immune selection after antigen-specific immunotherapy of melanoma

Background. Melanoma antigen (MA)-specific vaccination strongly enhances antitumor reactivity in vivo and is capable of producing strong cytotoxic T lymphocyte responses in vitro. Furthermore, specific human leukocyte antigen- restricted T cell activation is hypothesized to occur in response to peptide- based immunotherapy, which may lead to the preferential killing of tumor cells bearing the relevant MA. The development of melanoma antigen-loss variants may subsequently occur in vivo. Methods. Analysis of 532 melanoma lesions from 204 patients was performed on fine-needle aspiration biopsy specimens. Lesions were graded for the expression of the MAs gp100 and MART- 1 with use of immunocytochemistry. A total of 351 melanoma lesions were divided into cohorts on the basis of the treatment received. The pretreatment group (n = 175) consisted of lesions obtained before any form of gp 100 immunotherapy, with the posttreatment group (n =176) consisting of lesions obtained after vaccination with a modified gp100 epitope, gp209-2M +/- interleukin 2 (IL-2). Results. The percentage of lesions not expressing the gp100 antigen is greater than the percentage not expressing MART-1 (26% vs 14%). The frequency of lesions with high expression (> 75%) of gp100 significantly decreased with therapy (47% vs 34%) and conversely negative lesions increased (18% vs 29%). Treatment of lesions with peptide alone (no IL-2) revealed a significant decrease in gp100 expression (47% vs 32%), enhanced with the addition of IL-2 to therapy (47% vs 35%). The expression of MART-1 remained essentially unchanged unless IL-2 was added (54% vs 54%, MART-1 peptide alone, 54% vs 43%, MART-1 peptide + IL-2). Of 94 patients (181 lesions)assessed for gp100 expression before treatment, 10 patients responded to therapy. Pretreatment lesions in responding patients expressed some level of gp100 in all cases compared with 27% of nonresponding lesions, which were negative for gp100 expression. Conclusions. This study indirectly demonstrates that vaccination with an MA-derived peptide can result in immune selection in vivo. Furthermore, it provides strong immunologic evidence for the specificity of MA vaccines and to the relevance of MA expression in predicting the response to vaccination.