Based on the likelihood of antitumor interactions between cytokines and cytotoxic drugs, we designed a pilot study to evaluate feasibility, clinical, pharmacologic, and immunologic effects of concomitantly administered subcutaneous (SQ) recombinant interleukin-2 (r-IL-2) and doxorubicin (ADR) in patients with advanced solid tumors (AST). Patients received one injection of ADR alone (70 mg/m2) and 3 weeks later a combination of r-IL-2 (18 MIU/m2 days 1-5 s.q.) and ADR at the same dose either 3-4 h after the first r-IL-2 injection (arm 1) or 2 days after the last r-IL-2 injection (arm 2). The same combination was repeated every 4 weeks according to the evolution of the disease. Pharmacokinetics were assessed over 48 h after injection of ADR alone and after the first ADR-IL-2 cycle and immunologic monitoring at days 1 and 8 of the first ADR-IL-2 cycle. Tumors were measured at baseline, after ADR alone, and after each ADR-IL-2 cycle until progression. Twenty-one adult patients with various AST including 14 soft-tissue sarcomas (STS) entered the study, 11 in arm 1 and 10 in arm 2. All patients were heavily pretreated; 16 had received an anthracycline-containing chemotherapy regimen. Eleven patients were ADR refractory and 1 ADR resistant. Grade 4 neutropenia occurred in 28, 82, and 40% of patients after ADR alone, ADR-IL-2 in arm 1 and ADR-IL-2 in arm 2, respectively. Mucitis was higher in arm 1 (7 of 11 patients) compared with arm 2 (0 of 10) and ADR alone (0 of 21). SQ injections of r-IL-2 did not affect ADR pharmacokinetics. ADR injection in arm 1 prevented IL-2-induced lymphocyte rebounds in all patients but did not alter qualitatively non-major histocompatibility complex-restricted cytotoxicity. There was no response after ADR alone. Two patients, one in each arm, experienced a prolonged (8 and 5 months) objective response after ADR-IL-2. Both had ADR-refractory STS with a local relapse and metastatic metastases. Interestingly, both patients had unusually elevated TNF-alpha levels before and after the first ADR cycle. Combination ADR-IL-2, although toxic, is feasible and manageable with routine clinical support. r-IL-2 enhanced ADR hematologic and extrahematologic toxicities. The two objective responses observed in these heavily pretreated patients refractory to ADR supports the hypothesis of a modulation of ADR resistance, possibly mediated by means of a mechanism involving TNF-alpha. Elevated baseline TNF-alpha levels could be predictive of response to ADR-IL-2 and deserves further investigation.