A potential mechanism of chemotherapy resistance in acute myeloid leukemia (AML) is the multidrug resistance (MDR-1) gene product P-glycoprotein (P-gp), which is often overexpressed in myeloblasts from refractory or relapsed AML. In a multicenter phase II clinical trial, 37 patients with these poor risk forms of AML were treated with PSC 833 (Valspodar; Novartis Pharmaceutical Corporation, East Hanover, NJ), a potent inhibitor of the MDR-1 efflux pump, plus mitoxantrone, etoposide, and cytarabine (PSC-MEC). Pharmacokinetic (PK) interactions of etoposide and mitoxantrone with PSC were anticipated, measured in comparison with historical controls without PSC, and showed a 57% decrease in etoposide clearance (P =.001) and a 1.8-fold longer beta half-life for mitoxantrone in plasma (P <.05). The doses of mitoxantrone and etoposide were substantially reduced to compensate for these interactions and clinical toxicity and in Cohort II were well tolerated at dose levels of 4 mg/m2 mitoxantrone, 40 mg/m2 etoposide, and 1 g/m2 C daily for 5 days. Overall, postchemotherapy marrow hypoplasia was achieved in 33 patients. Twelve patients (32%) achieved complete remission, four achieved partial remission, and 21 failed therapy. The PK observations correlated with enhanced toxicity. The probability of an infectious early death was 36% (4 of 11) in patients with high PK parameters for either drug versus 5% (1 of 20) in those with lower PK parameters (P =.04). P-gp function was assessed in 19 patients using rhodamine-123 efflux and its inhibition by PSC. The median percentage of blasts expressing P-gp was increased (49%) for leukemic cells with PSC-inhibitable rhodamine efflux compared with 17% in cases lacking PSC-inhibitable efflux (P =.004). PSC-MEC was relatively well tolerated in these patients with poor-risk AML, and had encouraging antileukemic effects. The Eastern Cooperative Oncology Group is currently testing this regimen versus standard MEC chemotherapy in a phase III trial, E2995, in a similar patient population.