Evaluation of response rates to various antitumor agents of human gastric tumors implanted in nude mouse.

To ascertain the clinical predictability and, in the long run, the usefulness of the human tumor/nude mouse model as a screening tool for antitumor agents, it seems particularly important to use as many tumor lines as possible and to evaluate the therapeutic effectiveness of antitumor agents in terms of the overall response rate of a range of tumors. In this study, using 11 strains of established human gastric tumor xenografts with various histological characteristics and proliferation behavior, the experimental response rates to 8 typical antitumor drugs (mitomycin C, cyclophosphamide, ACNU, cisplatin, adriamycin, vincristine, vinblastine and 5-fluorouracil) were determined and compared with the clinical values. The experimental response rates to adriamycin, vincristine and 5-fluorouracil were in good accordance with the clinical results. However, with the other drugs, significantly higher response rates were observed with the nude mouse model as compared to clinical therapy, indicating that this model tends to overestimate the responsiveness of human tumor to a number of antitumor agents. These results strongly suggest the importance of using appropriate dose levels in the nude mouse to reproduce clinically equivalent effects in this model.