This dose-escalation study was performed to evaluate the hematologic activity, biological effects, immunogenicity, and toxicity of PIXY321 (an interleukin-3/granulocyte-macrophage colony-stimulating factor fusion protein) administered after high-dose carboplatin (CBDCA) treatment. Patients with advanced cancers received CBDCA at 800 mg/m2 intravenously on day 0 of repeated 28-day cycles. In part A of the study, patients were treated with CBDCA alone during cycle 1 and then received PIXY321 on days 1 through 18 of cycle 2 and later cycles. In part B, patients received 18 days of PIXY321 beginning on day 1 of all CBDCA cycles, including cycle 1. PIXY321 was administered subcutaneously in 2 divided doses. Total doses of 135, 250, 500, 750, and 1,000 micrograms/m2/d were administered to successive cohorts of 3 to 6 patients in part A. In part B, patient groups received PIXY321 doses of 750, 1,000, and 1,250 micrograms/m2/d. The hematologic effects of PIXY321 were assessed in the first 2 cycles of therapy. Anti-PIXY321 antibody formation was assessed by enzyme-linked immunosorbent assay (ELISA) and neutralization assay. Of the 49 patients enrolled, 31 were fully evaluable for hematologic efficacy. When comparing the first B cycle (cycle B-1; with PIXY321) with the first A cycle (cycle A-1; without PIXY321), the fusion protein had no significant effect on platelet nadirs or duration of platelets less than 20,000/microL but was able to speed the time of recovery of platelet counts to 100,000/microL (15 v 20 days; P =.01). Significant improvements in neutrophil nadir and duration of ANC less than 500 were observed in cycles A-2 and B-1 (with PIXY321) as compared with cycle A-1 (without PIXY321). Initial PIXY321 prophylaxis (cycle A-2 and cycle B-1), enhanced the recovery of ANC to greater than 1,500/microL by an average of at least 8 days as compared with cycle A-1 (without PIXY321; P </=.004). However, positive PIXY321 hematologic effects were lost in the second course of PIXY321 among patients treated in part B. ELISA analysis showed that 92% of patients had developed neutralizing anti-PIXY321 antibodies by the completion of 2 PIXY321-containing cycles. The incidental action of PIXY321 to depress serum cholesterol levels was also abrogated during cycle B-2. We conclude that PIXY321 was active in speeding hematologic recovery but that neutralizing anti-PIXY321 antibody formation suppressed the hematologic and biochemical effects by the second cycle of PIXY321 administration. The immunogenicity of this fusion protein provides a cautionary warning that clinical development of bioengineered human molecules requires thorough testing for immune neutralization.