Human neuroblastoma (NB) is a highly heterogeneous childhood cancer secreting a high level of vascular endothelial growth factor (VEGF). Its vascularization has been clearly correlated with metastatic progression and poor outcome. Thus, molecules that target the vascular endothelium are regarded as new therapeutics of clinical interest. Angiostatin, an internal fragment of plasminogen containing the first four kringle structures, has been described as a powerful angiogenic inhibitor. We used a recombinant adenovirus encoding the human angiostatin kringle 1-3 directly fused to human serum albumin HSA (AdK3-HSA). Coupling to HSA has been previously shown to increase the in vivo half-life of this angiostatic factor, and to lead to tumor growth inhibition in the MDA-MB-231 carcinoma model. For the assessment of antiangiogenic gene therapy in the human NB IGR-N835 tumor model, 5 x 10(9) PFU of AdK3-HSA were intravenously injected in tumor-bearing athymic mice presenting either of the following experimental settings: early stage, established, and minimal residual tumors. No delay in tumor growth was observed in animals treated with AdK3-HSA as compared to those treated with the empty virus AdCO1. In early-stage tumors, kinetics of tumor occurrence and tumor growth were similar in AdK3-HSA- and AdCO1-treated animals. K3-HSA was found to be expressed at high levels (the mean value for the three experiments being 19.4+/-15.9 microg/ml) in the circulation of all animals up to 21-35 days after virus injection. In addition, IGR-N835 tumors were found to be highly vascularized and to release high amounts of angiogenic factors, in particular VEGF (665+/-370 pg/mg total protein). Thus, in spite of high circulating levels, K3-HSA may be unable to displace the NB proangiogenic switch. In this regard, a more promising target to inhibit NB angiogenesis seems to be the VEGF/VEGFR system.