BACKGROUND: Mitoxantrone is an anthraquinone derivative that has demonstrated encouraging preclinical and clinical activity against a variety of human carcinoma cell lines and malignancies. Three Phase II studies of systemically administered mitoxantrone in patients with colorectal carcinoma failed to demonstrate any therapeutic activity, as did four Phase II studies of intravenous mitoxantrone in hepatocellular carcinoma. Two additional trials demonstrated limited activity when administered intravenously to patients with hepatocellular carcinoma. However, because this drug exhibits a steep dose-response curve, a Phase I-II trial of mitoxantrone by hepatic arterial infusion was initiated. METHODS: Patients with hepatocellular carcinoma and metastatic colorectal carcinoma with liver only or liver-predominant disease were eligible for therapy. All patients underwent the placement of a percutaneous hepatic arterial catheter before each course of therapy, and the first cohort of patients was treated at 10 mg/m2/course on day 1 on a 28-day cycle. Dosages were escalated in increments of 2 mg/m2/course based on side effects and tolerance. RESULTS: Twenty-eight patients with bidimensionally measurable unresectable, liver-predominant disease were entered into this trial. The therapy was well tolerated, with only 5 courses of 55 being complicated by neutropenia and none associated with fever. Only one patient required a dosage reduction on the basis of toxicity (neutropenia). No complete or partial responses were observed. CONCLUSION: These data are consistent with a lack of therapeutic activity of mitoxantrone when administered by hepatic arterial infusion for the treatment of hepatocellular carcinoma or metastatic colorectal cancer.