Heterogeneous responses of an in vitro model of human stomach cancer to anticancer drugs.

Four permanent clones of a human adenocarcinoma of the stomach and the parent line from which they were isolated were used as an in vitro model system to evaluate the effects of 8 anticancer agents on cell survival. The drugs tested were actinomycin D (Act-D), Bleomycin (Bleo), adriamycin (adria), melphalan, chlorambucil, 5 Fluorouracil (5FU), 1,2:5,6-Dianhydrogalactitol (DAG), and 1-(2-chloroethyl)-3-(4-methyl cyclohexyl-1-nitrosourea) (MeCCNU). Although the cell lines had similar growth properties, morphologies and modal chromosome numbers, the clones expressed heterogeneous survival responses to each of six drugs tested. A comparison of the doses lethal to 90% of a clonal population (LD90) for each drug indicated large differences between the most sensitive and least sensitive clones. For chlorambucil there was a 160% difference between the LD90 values of the most and least sensitive clones. For MeCCNU the difference was 200%; for adria, 230%; Bleo, 280%; 5FU, 360%; and melphalan, 600%. Despite the heterogeneity in response among the clones to these agents, no particular clone was always the most sensitive or resistant. Of particular interest was the finding that these stomach cancer clones demonstrated uniform responses to both Act-D and DAG. Since the differential drug sensitivities expressed by heterogeneous tumor populations could be a cause of treatment failure in the patient, the demonstration of uniform sensitivities to Act-D and DAG are encouraging and suggest that other anticancer drugs which produce uniform cell killing may be identified and tested. Act-D and adria were the most effective of the drugs tested when compared on a dose for dose basis. Both agents killed more than 99.9% of the parent cell line with doses below 3 micrograms/ml (1-h treatments). The cells were least sensitive to 5FU, with only 30% of the cells killed at 100 micrograms/ml. The studies reported here indicate that this human stomach cancer model can provide valuable insight into the design of clinical protocols for treatment of gastric carcinoma in man.