Complete resection, as defined by macroscopically negative margins (R0-R1), was achieved in 95% of all operations performed (Table ​(Table1).1). In three patients, complete resection was not possible because of extensive peritoneal sarcomatosis at laparotomy (n = 2) and significant intraoperative bleeding (n = 1, discussed below). Complete resection was feasible, but achieved slightly less often in resections performed for recurrent tumors (93%) versus primary tumors (97%, Table ​Table1).1). Tumor involvement of adjacent organs necessitated concomitant en bloc resection to achieve macroscopically negative margins in 58% of all cases (Table ​(Table1).1). The kidney, colon, and pancreas were the most frequently involved and resected organs. In 8% of cases, major vascular (e.g. inferior vena cava, iliac vessel) resection was required. In each case, anatomic vascular reconstruction was done immediately after resection by using synthetic tube grafts (for inferior vena cava) or cryopreserved human vein (for iliac vessels) (Figure ​(Figure1).1). Among patients who had complete resection, the rates of multi-organ or major vascular resection were higher in those with primary disease (71%; n = 24/34) than in those with recurrent disease (62%; n = 16/26).

Several pre-operative and intraoperative adjunct procedures were used to facilitate safe and complete resection in 54% of cases (Table ​(Table1).1). The decision to use any of these procedures was based on careful review of preoperative imaging studies that raised concern for tumor involvement of critical structures. For example, in three patients, a posterior laminotomy was done by neurosurgeons because preoperative imaging suggested spinal nerve root involvement. In two patients whose CT scans suggested high tumor vascularity, preoperative angioembolization was done by interventional radiologists to limit intraoperative blood loss. To help identify and avoid iatrogenic injury to the ureter, ureteral stents were placed by urologists before laparotomy in 35% of resections. Intraoperative femoral nerve monitoring, which we described previously [10], was done in 11% of cases to help identify the femoral nerve and prevent postoperative disability when involvement was suspected based on the location of the tumor and/or symptoms (e.g. paresthesias, pain, weakness).

With the exception of one patient with significant intraoperative bleeding (discussed below), all patients tolerated resection of their retroperitoneal tumor without intraoperative complications. A total of 24 complications were observed in the postoperative period. When scored according to the revised Accordion Classification scheme [11], the majority of complications were mild to moderate (Grade 1-3, Table ​Table2).2). Major postoperative complications, as defined by organ failure, need for reoperation, or death (Grade 4-6), occurred in 16% of cases. The most common complication overall was intra-abdominal abscess, which was managed by percutaneous drainage and parenteral antibiotics in all cases. The second most common complication was postoperative atrial fibrillation, which occurred in 4 patients. One patient had pre-existing atrial fibrillation, and none experienced a myocardial infarction. Atrial fibrillation was thought to be related to perioperative intravascular volume shifts and atrial stretch.

One patient had significant intraoperative bleeding that precluded complete resection and ultimately died (2% 30-day series mortality). This patient had a primary epithelioid tumor with invasion into the vagina and rectum that required partial vaginectomy and abdominoperineal resection. Tumor dissection close to the sacrum led to profuse venous bleeding and massive blood transfusion requirements that necessitated damage control measures and cessation of surgery. The patient developed multi-organ failure and died on postoperative day four.

In total, median length of hospital stay for all patients was 7 days (range 4-72). 38% (24/63) required postoperative recovery in the intensive care unit and among those patients, median length of stay was 4 days (range 1-23).

The final pathology results of resected retroperitoneal tumors revealed that 69% of cases had malignant disease. Among the malignant tumors, retroperitoneal sarcoma was the most common diagnosis (72%, n = 32/44). Liposarcoma was the most common sarcoma subtype (55%; n = 24/44), followed by leiomyosarcoma (18%; n = 8/44) (Figure ​(Figure2a).2a). Rare diagnoses such as undifferentiated pleomorphic sarcomas and spindle cell sarcomas were also seen. Among the benign tumors, schwannomas and paragangliomas were the most frequent (Figure ​(Figure2b).2b). Tumor sizes ranged from 2 to 48 cm, with a median of 14 cm. Only four tumors (7%) were less than 5 cm.

For retroperitoneal tumors, large case series indicate that multi-organ resection is to be anticipated in up to 80% of patients, with the kidney being the most common organ resected en bloc with the primary tumor [3,4,6]. Although preoperative imaging studies may suggest adjacent organ involvement, often, definitive assessment can only be made intraoperatively. This underscores the importance of preoperative planning, especially if the visceral organs to be resected are outside of the surgeon's area of technical expertise. Assistance from consulting services may be needed.

Recently, Bonvalot et al. have advocated "complete compartmental surgery", which involves multi-organ resection, even without obvious tumor involvement of adjacent organs at the time of surgical exploration [14]. In a retrospective review of 382 patients undergoing retroperitoneal sarcoma resection, the authors found that this technique was associated with a 3-fold decrease in local recurrence in comparison to standard multi-organ resection only for tumor involvement. However, no difference in survival was noted. Santos et al. found that the technique of compartmental surgery did not appear to impact either recurrence rates or survival and was in fact, associated with higher intraoperative blood transfusion requirements and postoperative morbidity [15]. We did not utilize complete compartmental surgery in our series and believe that more studies are needed to validate its routine use.

Retroperitoneal tumors can also involve major abdominal vascular structures in up to 18% of patients [16], necessitating concomitant en bloc resection, as was done in five (8%) of our cases. Preoperative planning is critical and should include consultation with a vascular surgeon for resection and reconstruction. For right-sided tumors, the surgeon must be prepared to resect the inferior vena cava (IVC) when a solid mass abuts it. Most published reports in patients with non-hepatic, non-renal primary retroperitoneal tumors advocate use of synthetic graft (e.g. polytetrafluoroethylene, PTFE) to reconstruct the IVC [16-19]. Graft patency rates are very good (90-94%), as shown by studies with 19 to 36 months of follow-up review [16-19]. Alternatively, proximal ligation of the IVC after resection may be appropriate in select patients whose infrarenal tumors have extensive venous collateralization, or when concern for bowel anastomotic leakage would make synthetic graft placement risky. Minimal morbidity was reported in 11 patients who had proximal IVC ligation [20], but a subsequent report found that postoperative leg edema was twice as common after IVC ligation than after graft reconstruction [21]. Finally, although more rare and arguably more challenging, aortic resection for retroperitoneal tumors has also been described [16,22,23]. A variety of graft materials have been used for aortic resection, including Dacron [16], PTFE [16,22], polyethylene terephthalate [22], and even autologous superficial femoral vein [23]. Five patients with aortic resection and prosthetic reconstruction were reported to have a patency rate of 89% at 19 months [16]. For both venous and arterial reconstruction, the option of using cryopreserved human vein or extra-anatomic bypass (i.e. axillary-femoral) also exist, particularly when there is concern for enteric contamination within the resection site. In select cases, extracorporeal circulatory bypass may be helpful to permit complete and safe resection of retroperitoneal tumors with major vascular involvement [24].

In our series, complete resection was feasible for recurrent retroperitoneal tumors, although it was achieved slightly less often than for primary tumors (93% versus 97%, Table ​Table1).1). Even more dramatic differences were reported by Lewis et al. more than a decade ago - 80% of patients with primary disease had complete resection versus 57% for those with local recurrence, and not surprisingly, the rate of complete resection continued to decrease with each subsequent recurrence [3]. Similar to our findings, more recent studies indicate a difference in the rate of complete resection, but suggest that the difference may not be as dramatic (99% versus 90%) [9]. The recent higher rates of complete resection for recurrent disease may reflect improved surgical technique, a more aggressive approach to resection of all retroperitoneal tumors, or improved patient selection. Complete resection should always be considered for recurrent disease as it is critical for improved survival [25] and in fact, may even result in comparable survival to patients after complete resection of primary disease [26].

Careful review of preoperative imaging is essential to anticipate potential operative scenarios and determine whether to utilize adjunct procedures to minimize complications. For example, ureteral stent placement and femoral nerve monitoring should be considered to identify these important structures when masses are near to their expected location in the retroperitoneum. Posterior laminotomy should be considered to permit complete resection of retroperitoneal tumors involving spinal nerve roots and when masses abut the vertebral bodies. Preoperative angioembolization of highly vascular tumors should be considered to minimize intraoperative blood loss.

Despite use of adjunct procedures, perioperative complications are common given the magnitude of the operations often required in retroperitoneal tumor resection. In terms of perioperative mortality, our 2% rate compares favorably to the 1-3% rates reported for large single institution patient cohorts [3,6] and a recent ACS-NSQIP national database review [27]. Our complication rate of 16% is also comparable to the recent 13-26% rates reported in the literature [6,27]. The caveat of reporting complications, of course, is that they are heavily dependent on the experience and expertise of the surgeon(s) and the resources of the institution, and are to some extent, driven by extent of the operation, which may differ from one patient to another.