PURPOSE: To determine if variability in toxicity of amonafide during phase II trials could be correlated with pharmacokinetic variability.

PATIENTS AND METHODS: Seventy-three patients enrolled onto three Cancer and Leukemia Group B (CALGB) phase II trials of amonafide (300 mg/m2 daily for 5 days) were studied, using a limited sampling strategy (45 minutes and 24 hours) to estimate the amonafide area under the plasma concentration-time curve (AUC). Concentrations of N-acetyl-amonafide, an active metabolite, were also determined.

RESULTS: The primary determinant of toxicity at a fixed dose of amonafide was the extent of N-acetylation. Fast acetylators (36% of patients) had significantly greater toxicity than slow acetylators (64% of patients), with median WBC nadirs of 500/microL and 3,400/microL, respectively (P < or = .001). In a multivariate analysis, lower pretreatment WBC count, lower albumin level, and nonwhite race were also independently associated with toxicity. Further analysis of interracial differences demonstrated that minority women had slower clearance of amonafide (P = .026) and a higher incidence of grade 4 leukopenia (P = .042).

CONCLUSION: The highly variable toxicity of amonafide is primarily due to genetic differences in N-acetylation. Other genetic (race) and acquired factors (baseline WBC count and albumin level) also appear to influence the extent of toxicity observed following administration of this agent.