Efficacy of immunotherapy depends primarily on the inherent immunogenicity of the corresponding tumors. This can be partly predicted by clinical observations related to the natural behavior of immune responses of the host, such as infilitration of T lymphocytes in tumor tissues and spontaneous tumor regression. Infiltration of lymphocytes has been seen in a group of inflammatory soft tissue sarcomas, including inflammatory myofibroblastic tumors 18 and inflammatory malignant fibrous histiocytomas 19. However, lymphocytes infiltrating these tumors are considered to be attracted non-specifically by cyotokines produced by tumor cells and not by tumor antigens 20. Besides inflammatory tumors, tumor-infiltrating lymphocytes (TIL) were sporadically observed in soft tissue sarcomas 212223. Notably, infiltration of dendritic cells was also reported in some soft tissue sarcomas with reactive lymphoid hyperplasia 23. This suggests that presentation of tumor antigens may occur in the regional lymph nodes. To date, prognostic significance of lymphocyte- or dendritic cell-infiltration in bone and soft tissue sarcomas remains to be proved. Also, it remains unknown whether there are differences in the immunogenicity of sarcomas between children and adults.

Spontaneous tumor regression is far more rarely observed. To our knowledge, there are only two reported cases of spontaneously regressing osteosarcoma 2425. These observations suggest the relatively low immunogenicity of bone and soft tissue sarcomas in natural conditions. Compared to malignant melanomas that are co-localized with Langerhans cells in the dermis and epidermis, there are few, if any, such immuno-surveillant cells in the deep tissues where sarcomas are developed. Such micro-environmental features may also be attributed to low immunogenicity of bone and soft tissue sarcomas.

The identification of TIL suggests the question of whether antigen-specific CTL precursors exist in the peripheral blood of patients with bone and soft tissue sarcomas. CTL precursors could home to the tumor site or originate from the presentation of tumor antigens by antigen presenting cells in tumor draining lymph nodes. Two pioneering experimental observations 2627 answered this question. Ichino et al. 26 induced CTLs from peripheral blood mononuclear cells of six patients with osteosarcoma, which specifically lysed fresh autologous tumor cells. Subsequently, Slovin et al. 27 developed CTL clones from one patient with osteosarcoma, and two patients with malignant fibrous histiocytoma (MFH). We have also recently established three autologous tumor cell-CTL pairs from patients with osteosarcoma 16, MFH of the bone 28, and MFH of the soft tissue, respectively. In our experience, the main difficulty in identifying sarcoma-specific antigens comes from the poor adaptability of sarcoma cells to in vitro culture, while the induction of tumor-specific CTLs from peripheral blood mononuclear cells is relatively more reliable thanks for the advancement of techniques for in vitro T cell culture. In addition, the utilization of tumor-draining lymph node as a source of antigen-specific lymphocytes and the manipulation of co-stimulatory pathways such as B7/CD28 and 4-1BB/4-1BBL have enhanced the success rate of CTL induction 2829. The efficacy in manipulation of costimulatory pathways to elicit immunological responses in vivo has been applied to animal models with osteosarcoma 3031323334 and used in clinical trials on patients with malignant melanoma 3536.

Another important advancement in the characterization of natural immune responses to sarcomas has come from the direct enumeration of tumor antigen-specific T cells in the circulation of patients with cancer. The introduction of HLA/peptide tetramer complexes technology 37 has made it possible to more precisely monitor the presence and frequency of circulating CTL precursors when the antigenic epitope and its associated HLA alleles is known. In this technology, soluble HLA/peptide complexes are assembled and biotinylated in vitro, and then are forced to form tetrameric arrays by addition of fluorescent-labeled avidin molecules. The HLA/peptide tetramer complexes bind to a T cell receptor on lymphocytes, which are visualized by FACS analysis. Using HLA-A24/SYT-SSX peptide tetramers, we observed an increased frequency of CTL precursors specific for SYT-SSX-derived antigenic peptides in patients with synovial sarcomas, compared to those with other sarcomas and healthy donors 383940. In addition, increased in vivo frequency of CTL precursors recognizing SYT-SSX-derived peptides was significantly associated with present or past history of distant metastases 38. Thus, aberrantly expressed SYT-SSX gene products are likely to have primed SYT-SSX-specific CTL precursors during the process of systemic tumor spreading. Furthermore, CTLs specific for HLA-A24-positive synovial sarcoma cells were inducible from PBLs of patients who showed increased frequency of CTL precursors in the HLA/peptide tetramer. Together, these findings suggest that SYT-SSX fusion gene products serve as the tumor associated antigen of synovial sarcoma.

Identification of tumor associated antigens and related antigenic peptides is the prerequisite for the development of the antigen-specific peptide-based immunotherapy. The discovery of tumor-specific chromosomal translocation and resultant fusion genes as a common event in various soft tissue sarcomas was a significant aid in antigen identification 131415. Since the fusion regions of translocation products are strictly limited to the corresponding tumors 41 (Table 1), they serve as attractive targets for tumor antigen-specific therapies. This concept has been proven previously by the study of leukemias 4243.

A strategic scheme for identification of therapeutic antigens targeting fusion genes is shown in Fig. 1. Soft tissue sarcomas with defined chromosomal translocation may be subjected to search for putative antigenic sequences in the region spanning the fusion gene. This search is based on the recognition of anchor motifs for HLA class I allele commonly expressed by the population affected by the disease 4445. This approach is commonly referred to as a "reverse immunology" 4647. In this fashion, several antigenic peptides have been defined from SYT-SSX 384048, EWS-FLI1 49, and PAX3-FKHR 49 fusion gene products (Table 2).

Bone and soft tissue sarcomas without known chromosomal translocation may be subjected to the "forward immunology approach" 46 (Fig. 1). This approach is based on the establishment of an autologous pair of tumor cells and tumor cell-reactive CTLs. When the first CTLs were expanded from patients with osteosarcoma 2627 and MFH 27, the expression cloning procedure for antigen identification 11 had not yet been identified. By using an autologous pair of osteosarcoma cell line and CTL clone 16, we identified an antigenic peptide derived from Papillomavirus binding factor (PBF) 17 (Table 2). PBF was originally defined as a transcriptional regulator of genomic DNA of the human papillomavirus type 8 50. The role of PBF in bone and soft tissue sarcomas still remains uncertain. Nevertheless, PBF was found to be expressed in 57/76 cases of bone and soft tissue sarcomas (11/14 osteosarcomas) and 20/34 of epithelial carcinomas by reverse transcription polymerase chain reaction. Immunostaining with anti-PBF antibody was positive in 45/51 bone and soft tissue sarcoma tissues (16/20 osteosarcoma), whereas it was weak and restricted in the cytoplasm of the normal organs tested including ovary, pancreas, and liver 17. Although PBF could represent a universal target for active specific immunization because of the frequency and relative selectivity of its expression, its utilization is limited by the rarity of the HLA allele in association with whom its discovery occurred. The recognition by the original CTL of the paired autologous tumor cells was associated to HLA-B55. Given the low prevalence of the HLA-B55 allele, the identified antigenic epitope is not clinically applicable to a broad range of patients. Currently the antigenicity of other PBF-derived peptides with high affinity for the HLA-A2 and HLA-A24 alleles, common in both the Japanese and Caucasian population, are being evaluated for future clinical trials.

In addition to the above described approaches aimed at the identification of novel sarcoma antigens and related epitopes, it is also reasonable to target known tumor-associated antigens that had been previously identified in cancers other than sarcomas as long as these proteins are also expressed by these tumors (Fig. 1). Accordingly, the expression of several cancer-testis antigens 51525354 and other antigenic proteins over-expressed by cancers compared to normal tissues 5556 has been investigated in bone and soft tissue sarcomas. Despite reasonable expression levels, they have not been proved to be naturally antigenic in patients with bone and soft tissue sarcoma and lead to spontaneous CTL responses identifiable in the blood of these patients. Therefore, peptides derived from those known antigens are now being examined using reverse immunology to test their antigenicity and binding properties to the common HLA alleles, with the aim of utilizing them as targets for antigen-specific immunization.

SEREX (serological analysis of recombinant cDNA expression library) was developed as a procedure to identify antigens without using autologous tumor-CTL pair 57. SEREX defines putative antigens using serum IgG antibodies from cancer patients. Since isotype switching from IgM to IgG implies the presence of specific help from CD4+ T cells, antigens defined by SEREX are likely to contain epitopes for CD4+ T cells. SEREX analysis has been performed on patients with bone and soft tissue sarcomas 165859, displaying an array of cancer-testis antigens as possible candidates. However, antigenic peptides have yet to be isolated.