This study describes the effect on surgical outcomes of local injection of rhIL2 or placebo into the wound bed after surgical excision of PNST-like tumours. Treatment with rhIL2 did not significantly influence the MFI, RFI, or OS. In contrast, it has been reported to have a beneficial effect in various types of tumours in humans [16, 17] and animals [25, 29–31, 35, 36, 39], providing a 6% complete response (CR) and 10% partial response (PR) in one study  or a 50% decrease in nasal angiosarcoma size in another . In other studies, mice with inoculated lymphoma or mastocytoma treated intraperitoneally with IL-2 achieved a 70–90% CR [29–31], cattle with ocular squamous cell carcinoma treated with peritumoural IL-2 achieved a 69% CR , and horses with sarcoid tumours achieved a 14% PR and 43% CR with intratumoural IL-2 . While systemic administration of IL-2 has also been used, locoregional administration strategies are more effective and successful [23, 26]. Interestingly, local treatment is often effective systemically, eradicating metastases . Systemic IL-2 therapy can lead to life-threatening toxicities and to hepatic dysfunction, which might result in hypotension and vascular fluid accumulation .
The use of IL-2 as sole adjuvant agent after sarcoma resection is not warranted at the moment. Additional therapies, such as chemotherapy or radiotherapy, may increase the efficacy of IL-2, but a large effect is unlikely, especially when a wide surgical excision is used, which is the preferred approach for PNSTs. In the above-mentioned study of horses with sarcoid tumours, local IL-2 treatment combined with chemotherapy (cisplatin) resulted in better outcomes than local IL-2 monotherapy (53% PR and 80% CR compared with 14% PR and 43% CR, respectively) .
Incomplete tumour resection increases patient morbidity, treatment costs, risk of further recurrence, and ultimately decreases survival time [40–43]. Therefore, the likelihood of recurrence is much lower if surgery is performed by a specialist-certified surgeon rather than a referring veterinarian . We found a much lower recurrence rate in dogs that had not previously undergone surgery than in dogs that had previously undergone surgery. This difference can be explained in two ways: the veterinarians who removed the primary tumour might have used even smaller tumour margins than the specialist-certified surgeon or the previously removed tumour might have been more aggressive and more likely to recur. Of the 40 dogs, 8 had a grade 1 PNST, 20 a grade 2, and 12 a grade 3. There was no significant difference in grade of the removed PNST between the dogs with or without previous surgery. Nor did we find survival or recurrence rates to differ by tumour grade, but this might be a reflection of the relatively low number of cases.
The rate of recurrence was similar in the two treatment groups, 45% (9/20; rhIL2 group) versus 35% (7/20; placebo group). The overall recurrence rate (40%) was higher than the 15% reported in a comparable study of 41 dogs, in which all dogs that had surgery for recurrence had been referred after inadequate primary tumour resection . This difference might be because we had a longer follow-up of minimally 5 years. Interestingly, tumour recurred at the site of excision many years after primary surgery, suggesting that dormant cells were present. This suggests that future studies should have longer follow-up times (>2 years) and that postoperative radiotherapy is warranted. Radiotherapy has been used in many studies after marginal resection of STS, with recurrence rates varying from 15% to 31%. Irradiated, incompletely resected STSs had a recurrence rate of 15%  and 17% . However, because these studies did not investigate PNSTs and did not include rhIL2 therapy, we cannot compare results. Interestingly, a recent review of human cancer concluded that the response rate was higher when rhIL2 was used as adjuvant to radiotherapy , an approach that should be evaluated in the future. Alternative treatment options for incomplete tumour resection are re-excision of the wound bed with wider margins [40, 42, 44] and downstaging local disease with preoperative radiotherapy .
Most PNSTs in dogs have low metastatic rates. In this study, 15% of the PNSTs (6/40) had metastasized to lungs or lymph nodes, and rhIL-2 therapy did not influence the rate of metastasis. One study of IL-2 used in an adjuvant setting as local inhalation therapy for carcinoma reported that 2 of 7 treated dogs achieved full remission for more than a year (29%) . Other canine STS studies have reported lower metastasis rates than the rate reported here. The rate of metastasis was 8% in a study in which radiotherapy was used as adjuvant therapy after incompletely resection . The higher metastatic rate in our study might have been due to the longer follow-up, a more vigilant surveying system, or the presence of dormant cells in the tumour margins. The results of this study indicate that aggressive primary surgery is advisable for these tumours.
The higher incidence of metastasis may warrant the use of postoperative chemotherapy, a protocol not used in this study. However, the exact role of chemotherapy in preventing distant metastasis of STS is not known. Doxorubicin, ifosfamide, and mitoxantrone have been used in dogs and in humans. The overall response of canine sarcomas to doxorubicin was 23% in one study , but doxorubicin had no effect as adjuvant therapy in a recent study of high-grade STS . Cisplatin applied in a biodegradable implant delivery system directly after marginal resection of STS in 19 dogs resulted improved survival in 9 dogs (47%) with a median follow-up of 874 days; 8 dogs died of tumour-unrelated causes, and 3 dogs had a recurrence (fatal in 1 dog) . Recently, cyclophosphamide combined with piroxicam significantly prolonged the DFI of 30 dogs (median DFI >410 days) compared with that of 55 control dogs (median DFI 211 days) after marginal resection of STSs . The median OS was not reached in our study, even though follow up times reached over 5 years. This compares favourable to other studies reported for surgery alone (1416 days)  or 2270 days after incomplete resection combined with adjuvant radiotherapy  and the 309 days reported in a study of intraoperative chemotherapy  or four-fraction palliative radiotherapy . Chemotherapy is typically used in dogs with high-grade and incompletely resected tumours with a high metastatic rate, which might explain the high recurrence rate .
PNSTs on extremities often cannot be excised with wide margins and amputation of the limb is a suggested treatment alternative. However, marginal resection either as sole therapy or combined with radiotherapy may result in a similar long-term survival with less immediate morbidity. Marginal excision of low-grade (G1) STSs from the extremities of 35 dogs resulted in only 4 recurrences (11%) . Similarly, a study showed a grade-dependent recurrence after marginal excision in 7% (3/41) of grade 1 (G1) tumours, 34% (14/41) of grade 2 (G2) tumours, and 75% (3/4) of grade 3 (G3) tumours . Survival time was not influenced by grade in dogs treated with marginal excision . Radical resection of STSs on extremities by limb amputation should therefore be considered as a last resort for recurrent and high-grade tumours, taking into account that the risk of metastases is higher in cases that had a recurrence [49, 50]. A more aggressive approach, with preoperative incisional biopsies, seems warranted. Surgical margins can be smaller with lower grade tumours and still result in positive outcomes. The study reported here, had longer a follow up and a grade variation between groups that was identical but the comparison with historic data prevents any major conclusions when survival outcome is compared to the above-mentioned studies.
The study had several limitations. Although the entry criteria were strict and the double-blinded study had a well-executed follow-up regimen, a major limitation remains the compliance of owners. Unfortunately, owner compliance is never 100% and some owners lived too far away to bring their dogs to follow-up evaluations. In these cases, we consulted the referring veterinarian for follow-up information, but we have no way of knowing whether he/she had recently examined the patient. The variation in tumour size, location, and grade can also affect outcomes, especially in small case cohorts. In total, 4.5 million IU of rhIL2 was administered, but the area of distribution varied with the wound size, which limits extrapolation of the exact local dose of rhIL2. The size of the wound should have been measured prior to injection of rhIL2 or placebo to allow a better calculation of the exact dose per square cm2. Power analysis indicated that 40 dogs would be needed to detect a statistically significant difference between treatments. However, this is still a small number, especially when multiple observations and variables are used, such as the grade and histologic diagnosis. The use of two instead of one referral institute may have influenced the data although there were no significant differences between the two institutes for the variables examined (data not shown).