Bone marrow transplantation arose from the realization that dose intensification could improve the response to cytotoxic chemotherapy in some patients. Likewise, experiments evaluating cytotoxicity in the laboratory often demonstrate a steep dose-response curve, in which a threefold increase in drug concentration can result in a major decrease in cell survival. If all cancer cells fell within the same range of sensitivity, then the dose intensity achieved by the addition of bone marrow transplantation would provide effective treatment for all cancers.In truth, broad variations exist in chemosensitivity, both within a given type of cancer and among different tumor types. Figure 1 displays doxorubicin dose response curves from the National Cancer Institute (NCI) drug screen database. In vitro cytotoxicity assays are performed simultaneously in an automated fashion on 60 cell lines and are displayed in disease-oriented panels. The IC50s (the concentration at which 50% of cells are growth inhibited) for doxorubicin vary over a 1000-fold concentration range. These major variations in chemosensitivity in cancer cell lines, derived principally from patients who had received no prior therapy, are most likely caused by a diverse assortment of drug resistance mechanisms, which will ultimately limit the success of the dose intensification achieved by bone marrow transplantation.One mechanism of drug resistance that has been studied intensively is that mediated by P-glycoprotein (Pgp), a 170 kilodalton membrane glycoprotein known to transport certain chemotherapeutic agents out of the cell (1,2). Pgp, which is encoded by the mdr-1 gene, mediates resistance in vitro to important chemotherapeutic agents, including doxorubicin, paclitaxel, daunorubicin, vinblastine, vincristine, VP-16 (etoposide), and mitoxantrone, by active efflux of the chemotherapeutic agent (1). In vitro, multidrug resistance can be reversed by the addition of agents that are also recognized by Pgp but are less toxic and, in large concentrations, block the efflux of chemotherapy. These agents include verapamil, cyclosporine, phenothiazines, and dihydropyridines (3). Even cells with very high levels of Pgp and several-hundredfold resistance can have efflux blocked by the concurrent addition of a Pgp antagonist with the chemotherapy. An example of Pgp antagonism is shown in Fig. 2 in SW620 Ad5, a multidrug-resistant human colon carcinoma subline with relatively low levels of Pgp, derived from SW620 parental cells by selection in Adriamycin. Rhodamine fluorescence is measured by fluorescenceactivated cell sorting analysis as an assay of Pgp function (4-8). An initial incubation with rhodamine is followed by a 120 minute incubation in rhodamine-free media to allow the Pgp-mediated efflux of rhodamine. A decrease in rhodamine fluorescence is observed after the efflux period, which is consistent with Pgp-mediated efflux. Both verapamil and cyclosporine A prevent the decrease in fluorescence by impeding efflux. There is no doubt that in laboratory models Pgp is ab...