[Predictability of preclinical evaluation of anticancer drugs by human gastrointestinal cancer--nude mouse panel].

: We evaluated the predictability to clinical response of experimental effects of various anticancer agents on human cancer--nude mouse panel established in our department. The human cancer lines used were 12 gastric, 4 colorectal, 3 breast, 2 pancreatic cancers and 1 melanoma xenografted into BALB/c athymic nude mice under SPF conditions. Seven mice each with equivalent mean volume of sc inoculated tumor (about 100 mm3) were subjected to the treatment and control groups. Experimental treatment was conducted daily 25 times for antimetabolites, and intermittently 5 times once or twice a week for other drugs. Dosage of each drug adopted was maximal tolerated dose predetermined for the treatment schedule. Four weeks after the initiation of treatment, the therapeutic effect of each experiment was evaluated by the tumor growth inhibition rate (IR) based on the comparison of mean tumor weight between the 2 groups. When the IR was greater than 58%, the drug was evaluated as effective. The clinical response rate of each drug was referred from the result of the phase II study. Direct comparison of effects on 16 experimental chemotherapies in xenografts with responses to the corresponding clinical therapy of each donor patient revealed a fairly high accordance rate (94%). To elucidate the value of human cancer--nude mouse panel as the preclinical secondary screening, the response rates of 8 anticancer drugs treated to 15 cancer xenografts were compared with the cumulative clinical data available in each drug. Generally, the response rates of the human cancer xenografts to the drugs showed fairly good correlations with the cumulative clinical response rates of the corresponding drugs to the same organs. Using this panel, preclinical examinations of 6 new agents under development, including 254 S and other 2 CDDP derivatives, were performed in expectation the positive correlation with further clinical data.