We have completed a phase I study to test feasibility and toxicity of immunotherapy of hepatic metastases from colorectal carcinoma by direct gene transfer of HLA-B7, a MHC class I gene. Eligible patients were HLA-B7 negative, immunocompetent by PHA lymphocyte stimulation and had at least two measurable hepatic lesions on CT scan for measurement of response of the injected lesion, as well as evaluation of possible distant response. Under ultrasonographic guidance the hepatic lesions were injected with Allovectin-7, a liposomal vector containing the combination of the HLA-B7 gene with beta 2-microglobulin formulated with the lipid DMRIE-DOPE. Eligible patients were injected on two schedules. On the first schedule patients received an injection on day 1 and the injected lesion was biopsied to determine transfection every 2 weeks for 8 weeks. Doses were escalated from 10 micrograms to 50 micrograms to 250 micrograms with three patients treated at each level. The second schedule included multiple injections of 10 micrograms. Three patients received injections on days 1 and 15. Three patients received injections on days 1, 15 and 29. A total of 15 patients have completed treatment. The plasmid DNA was detected in 14 of 15 patients (93%) by PCR. In five of 15 patients (33%) mRNA was also detected. The HLA-B7 protein was detected in five of eight patients (63%) by immunohistochemistry and in seven of 14 patients (50%) tested by fluorescence activated cell sorting (FACS) analysis. There has been no serious toxicity directly attributable to allovectin-7. Our results suggest that liposomal gene transfer by direct injection is feasible and non-toxic. Further studies will be necessary in order to establish the therapeutic efficacy.