1 2 3 1 4 2 5 6 7 3 6 8 9 10 11 13 14 15 PATIENTS AND METHODS Patient Characteristics 17 16 1 TABLE 1. a Characteristic n % Grade     I 10 14   II 23 32   III 40 54 Stage   I 10 14   II 1 1   III 50 69   IV 12 16 a 17 16 Perfusion Technique 18 19 20 Assessment of Tumor Response, Tumor Remnant, and Follow-up 21 Excision margins were also evaluated on pathological examination and classified as radical when the resection margins were free of tumor cells (complete resection, R0), as R1 when resection margins were microscopically involved, or as R2 when resection margins were macroscopically positive involved. Postoperative radiotherapy (60–70 Gy) was considered indicated in case of < 95% necrosis on pathological examination of the tumor or with marginal or microscopically positive resection margins. All patients were followed after perfusion treatment in a standardized protocol. Median follow-up was 27 (range 2–138) months. Statistical Analysis 22 P RESULTS Tumor Response 1 FIG. 1. Percentage of necrosis estimated at pathological examination of resected tumor remnant in relation to number of patients.  Amputations and Limb Salvage 2 2 TABLE 2. Amputations performed in 21 patients according to interval duration Patient no. Diagnosis Age (y) Interval (mo) Resection RT Rationale for amputation Current status 1 PUS 60 0 R0 No Postperfusion necrosis38 NED 120 months 2 Angiosarcoma 74 1 R2 No Local recurrence DOD 11 months 3 Fibrosarcoma 76 1 R0 No Postperfusion necrosis NED 2 months 4 PUS 67 2 R0 No Postperfusion necrosis DOD 9 months 5 Epithelioid Sarcoma 21 2 R0 No Postperfusion necrosis DOD 54 months 6 Leiomyosarcoma 17 2 R0 No Insufficient clinical response DOD 7 months 7 Liposarcoma 60 2 R1 No R1 resection, RT not possible AWD 10 months 8 PNET 62 3 R0 No Local recurrence DOD 17 months 9 Synovial sarcoma 39 3 R0 No Postperfusion necrosis DOD 50 months 10 PUS 63 3 R1 No Postperfusion necrosis NED 72 months 11 Angiosarcoma 80 4 R1 No Local recurrence DOD 10 months 12 Synovial sarcoma 65 4 R1 No R1 resection, RT not possible NED 6 months 13 Epithelioid Sarcoma 22 6 R2 No a DOD 39 months 14 Hemangioma pericytoma 50 8 R0 Yes a AWD 65 months 15 PUS 71 12 R0 Yes a NED 14 months 16 PUS 61 15 R0 Yes Arterial occlusion AWD 17 months 17 Synoviosarcoma 42 18 R1 Yes Local recurrence NED 20 months 18 Liposarcoma 53 37 R0 Yes Local recurrence DOD 110 months 19 Liposarcoma 39 58 R0 No Local recurrence DOD 120 months 20 PNET 56 110 R0 Yes Critical leg ischemia NED 118 months 21 Chondrosarcoma 18 125 R0 Yes Critical leg ischemia NED 134 months PUS, pleomorphic undifferentiated sarcoma; RT, radiotherapy; NED, no evidence of disease; AWD, alive with disease; DOD, death of disease; PNET, malignant peripheral nerve sheath tumor; R1 resection, microscopically involved resection margin; ILP, isolated limb perfusion. a FIG. 2. Limb salvage curve in patients with locally advanced soft tissue sarcoma treated with tumor necrosis factor alpha, melphalan, and isolated limb perfusion.  3 FIG. 3. Clinical appearance of lower leg of patient 21 (Table 3) before amputation for critical leg ischemia.  Systemic Metastases and Survival P 4 4 FIG. 4. Overall survival in patients with locally advanced soft tissue sarcoma treated with isolated limb perfusion (ILP), tumor necrosis factor alpha, and melphalan. Statistically significant difference was observed between patients with no distant metastases at time of ILP (mets −) compared with patients with metastases at time of ILP (mets +).  DISCUSSION 15 15 23 25 23 15 23 25 26 We used the Kaplan-Meier method to calculate limb salvage because this method adjusts for censored observations, i.e., patients who were alive and well at the time of last contact or patients who died of distant metastases but who retained preserved limb function. By use of this method, we calculated a 1-year limb salvage rate of 80%. Amputations were performed mostly because of postperfusion-related complications or early local recurrence in the first year after ILP. A second curve in limb salvage was observed within 5 years after TNF-α ILP in two patients with late local recurrences. A third bend in the limb salvage curve was observed approximately 10 years after ILP. This was a new observation in two patients who had critical leg ischemia with ulceration and continuous pain. Besides ILP with TNF-α and melphalan, both patients received adjuvant radiotherapy (66 and 70 Gy) after marginal tumor resection. 27 28 29 30 31 32 33 34 23 24 35 36 37 Since 1992, we have not changed the indication for TNF-α perfusion. Patients who were candidates for amputation of the involved limb, as assessed by preoperative magnetic resonance imaging, were offered an ILP with TNF-α and melphalan, with the goal of preserving the limb with a locally advanced STS. After ILP patients received a delayed surgical resection, adjuvant radiotherapy was provided to patients with marginal or microscopically positive resection margins. This treatment resulted in a high limb salvage rate in patients with locally advanced STS, although late morbidity can occur, especially when adjuvant postoperative radiotherapy is applied. Therefore, continuous follow-up of these patients is warranted.