BACKGROUND: Cytostatic isolated lung perfusion has been advocated for treating pulmonary metastasis of soft tissue sarcoma. Different techniques of isolated lung perfusion have been developed. METHODS: Isolated lung perfusion with and without doxorubicin was performed on white pigs during 15 minutes either by a single-pass system (n = 7) or by a recirculating-blood perfusion system (n = 7). Three animals with endovenous drug application served as controls. Leakage was assessed using isotopic tracers. Perfusion-induced lung tissue injury was determined by postperfusion chest radiographs, by angiotensin-converting enzyme-to-protein ratio in the plasma and in the bronchioalveolar lavage fluid, and by wet-to-dry weight ratio and histologic examination of lung biopsy specimens at 20 and 50 minutes. Doxorubicin concentration in lung tissue and plasma was compared between the three study groups. RESULTS: All isolated lung perfusion studies were successfully performed without significant systemic leakage (< 0.6%). Wet-to-dry weight ratio was significantly lower after single-pass as compared with recirculating-blood perfusion and endovenous drug application at both time points (5.0 +/- 1.1 and 5.3 +/- 0.8 for single-pass versus 6.6 +/- 1.1 and 6.9 +/- 0.5 for recirculating-blood versus 6.6 +/- 0.2 and 5.9 +/- 0.7 for the control group, respectively; p < 0.05). Angiotensin-converting enzyme-to-protein plasma ratio in the single-pass group was significantly lower only at 20 minutes (6.3 +/- 2.4 versus 9.3 +/- 1.0 versus 9.7 +/- 1.9, respectively; p < 0.05) but not at 50 minutes. Angiotensin-converting enzyme-to-protein ratio in bronchoalveolar lavage fluid, histology of lung biopsy specimens, and chest radiographs did not differ significantly between the three groups. Doxorubicin lung tissue concentration was not significantly different after single-pass (17.5 micrograms/g) and recirculating-blood perfusion (21.9 micrograms/g), but was significantly higher than after endovenous drug application (3.0 micrograms/g; p < 0.01). CONCLUSIONS: Both isolated lung perfusion techniques resulted in a sixfold to sevenfold higher doxorubicin lung tissue concentration than after endovenous application. Isolated lung perfusion-induced lung injury was similar for both techniques, but recirculating-blood perfusion appeared to result in more acute lung injury and was technically more demanding than single-pass perfusion.