Background: Rhizoxin is a new macrocyclic lactone isolated from the fungus Rhizopus chinensis which displays broadspectrum antitumor activity against murine and human tumor xenografts and has activity against a number of vincristine-resistant tumors in vitro and in vivo. Purpose: This study describes the preclinical and clinical pharmacology of rhizoxin to apply a pharmacokinetically guided doseescalation (PGDE) strategy during the phase I trial. Methods: Rhizoxin was administered by a single intravenous bolus injection to female BALB/c mice over the dose range 7.5–18 mg/m from which we derived the dose that was lethal to 10% and 50% of the mice (i.e., LD 10 and LD 50 , respectively). The LD 10 was 11.7 ± 0.7 mg/m (mean ± SD), and the LD 50 was 14.7 ± 0.6 mg/m 2 . Pharmacokinetic studies were integrated with the toxicity study in female BALB/c mice at one-tenth the LD 10 , one-half the LD 10 , and the LD 10 (i.e., 1.2, 6, and 12 mg/m, respectively). From these data, a target area under the plasma drug concentration versus time curve (AUC) (i.e., 40% of the LD 10 AUC) was calculated for clinical studies. Phase I studies were initiated at 0.8 mg/m (one-tenth the equivalent LD 10 in male CD1 mice), with the intent of escalating the dose by an extended factor-of-two method until the target AUC and/or maximum tolerated dose (MTD) was reached. Results: The major drug toxic effects in mice were body weight, loss, sluggishness, ataxia, transient changes in hematological parameters, and hematuria. Diarrhea was universal at doses greater than 9 mg/m, and hind limb paralysis was observed in one of 10 mice, but only at supralethal doses (18 mg/m). Rhizoxin pharmacokinetics were best described by a two-compartment open model (half-life [ t 1/2 ] α = 4.4 minutes ±0.9 minute [mean ±SD), and t 1/2 β =84 minutes ± 20 minutes at 12 mg/m) and found to be nonlinear with respect to dose. At doses of 1.2, 6, and 12 mg/m, the respective AUC values were 1.3, 22.4, and 70.6 μ M × minute. From these data, a target AUC value of 28 μ M × minute (40% of the LD 10 AUC) was derived. Rhizoxin was not detectable in patient plasma (M × minute, considerably lower than the target AUC of 28 μ M × minute. Conclusion and implications: Although PGDE schemes have been successfully employed for other antitumor agents, this methodology could not be applied during the phase I trial of rhizoxin. PGDE studies in the future may incorporate comparative murine versus human metabolism studies in vitro with phenotyped liver microsomes. It may also be useful to assess the comparative myelotoxicity of a new drug by performing in vitro cytotoxicity studies on mouse and human bone marrow stem cells. [J Natl Cancer Inst 84:494–500, 1992].