Sarcoma Guidelines

Does multivisceral resection improve outcomes for patients with primary retroperitoneal sarcoma?

  • Population: Adult patients with primary localised retroperitoneal sarcoma
  • Intervention: Multivisceral resection (including adjacent organs uninvolved on preoperative imaging)
  • Comparator: Simple surgical resection
  • Outcomes: Overall survival, recurrence free survival, perioperative morbidity

Authors: David Coker, Anna Lawless, Deborah Zhou, Angela Hong, David Gyorki and the ANZSA Sarcoma Guidelines Working Party

There is a trend in retroperitoneal sarcoma (RPS) surgery over the past two decades, particularly for cases of liposarcoma, towards contiguous organ resection (multivisceral resection, MVR). A single en-bloc specimen, encompassing the tumour and contiguous, uninvolved organs is performed in order to achieve macroscopically complete resection and maximise the chance of microscopic clearance (1, 2).  Adapting the operative approach with respect to the extent of resection, in the context of tumour histology is now recommended by leading international groups, where risks of recurrence locally and distantly vary according to tumour biology (2-5).  Since 2013, the Transatlantic Australasian Retroperitoneal Sarcoma Working Group (TARPSWG) has sought to standardise surgical management of RPS, including MVR, through international collaboration and the establishment of an international database to enable collaborative research in this rare subset of tumours.

This clinical question aimed to determine whether MVR improve outcomes for patients with primary retroperitoneal sarcoma. The literature search identified twenty six studies that met the inclusion criteria (see technical report for the full list and summary). Two of these studies were from the same unit with overlapping time periods, and were therefore analysed together (6, 7). The size of the patient cohorts is generally small, with the largest study from the TARPSWG having just over 1000 patients recruited from North America and Europe. Given the nature of the question, no randomised study was identified. 

Overall Survival

Twenty four studies reported survival outcomes, and there is conflicting evidence as to the effects of MVR on overall survival (OS) and disease specific survival (DSS) across the studies (see technical report).

Seven studies demonstrated non-statistically significant results with respect to MVR and OS (5, 8-13). In four of these studies, resection margin and tumour grade were both significant predictors of worsened overall survival (8, 12-14). The study by Bonvalot et al. (n=382), is considered to be one of the most relevant to the PICO model, and is of high quality [8]. The authors compared simple resection, resection of contiguously involved organs, and compartmental resection (defined as systematic resection of uninvolved contiguous organs performed to obtain a rim of normal tissue surrounding the tumour). The histologic margins were directly linked to the type of surgery: after compartmental surgery, simple complete resection, and contiguously involved organ resection, margins were positive in, respectively, 19%, 40%, and 36% of cases (p<0.0001). Complete gross resection was a strong prognostic factor for OS (R1 resection vs R0 resection HR 1.7, 95% CI 1.07-2.72, p=0.03). In the multivariable analysis, independent predictive factors associated with decreased OS were high grade, tumour rupture, gross residual disease, and positive margins. Compartmental surgery was not significant factor for OS.  As compartmental surgery was linked to achieving resection margins, hence the authors argued a link between compartmental surgery and the improved survival seen with negative margins. In a multivariate analysis of 571 patients using the United States Sarcoma Collaborative data,  MVR (1-2 organs vs 0, 3-4 organs vs 0, 5 organs vs 0) was not associated with a better OS (12). Worse OS was seen in older patients, high grade tumour, node positive disease and multifocal disease.

There are two small studies showed some OS benefit of MVR as compared to simple resection. Morizawa et al. (n=23), demonstrated in multivariate analysis that simple resection, compared with extended resection encompassing the tumour and adjacent organs, were predictive of decreased OS (HR 3.80, 95% CI 1.25-16.59) (15). Gonzalez Lopez et al. (n=56), compared enucleation which was defined as removal of the tumour within its pseudocapsule, with en bloc resection of the tumour and contact (contiguous) structures. In the multivariate analysis, there was a trend to improved overall survival in the en bloc resection (p<0.08) (16).

Conversely to the above results, Singer et al. (n=177), demonstrated that contiguous organ resection (excluding nephrectomy) versus no contiguous organ resection was associated with a decreased DSS (HR 1.9, 95% CI 1.01-3.5, p=0.05) (17). The margin status, gross (R2) versus negative (R0) was far more significant in prediction of DSS (HR 3.8, 95% CI 2.0-7.4, p<0.0001). Dedifferentiated liposarcoma was associated with worse DSS than x well-differentiated liposarcoma (HR 6.0, 95% CI 3.3-10.9, p<0.0001). The proportion of patients undergoing contiguous organ resection, who had dedifferentiated liposarcoma, or the proportions of patients that had gross margins in each cohort of resection are not reported.

In a study of 119 dedifferentiated liposarcoma patients, Keung et al. demonstrated in univariate analysis that in patients with an R0/R1 resection, there was a significant difference in median overall survival where less than two organs were resected (74.3 months, vs 59.3 months, p=0.045) [36]. However, the number of organ resection was not an independently prognostic in multivariate analysis. Factors detrimental to OS were high grade tumours, R2 resection, and intraoperative tumour fragmentation.

Three papers analysed the survival by the number of organs resected, or the types of organs resected (13, 18, 19). Abdelfatah et al. (n=115), demonstrated a statistically significant increased risk of death, where 5 or more organs were resected, albeit in a total of only 8 cases (7% of the cohort), as compared to no organs resected (HR 6.25, p=0.005) (18) . However, where less than five organs were resected this variable was not significant. Smith et al. (n=362), showed an increased risk of death where three or more organs were resected on univariate analysis (HR 4.11, p<0.001), however this did not reach significance on multivariate analysis, where only the grade of the tumour showed significance (13). In a propensity matched analysis of 24 patients, vascular resection correlated with worsened survival (HR 5.17, 95% CI 1.41-18.99, p=0.013) (19).

Four studies reported the effect of specific organ resections and survival (9, 17, 19, 20). Cho et al. (n=114), found no difference in 5-year cancer-specific rates when comparing patients who underwent nephrectomy as part of their resection, versus those patients that did not undergo nephrectomy, 75% versus 71% respectively (9). In a subgroup analysis there was a trend towards better survival where nephrectomy was performed for FNCLCC grade 2 tumours (p=0.077). Those patients undergoing nephrectomy, had a statistically significant increase in contiguous organ resections (35 cases with nephrectomy (53%) vs 11 cases without nephrectomy (22%), p=0.002); and were significantly more likely to have a liposarcoma as opposed to another histology (p<0.001). No difference in disease-specific survival was seen when nephrectomy was analysed by Singer et al. (17). In a more recent study, Kim et al. (n=88), reported that distal pancreatectomy had no effect on overall survival (20). Patients who underwent distal pancreatectomy were more likely to have dedifferentiated liposarcoma, and to have a higher number of contiguous organs resected (3.80.9 vs 1.00.9, p<0.001). Spolverato et al, found a similar trend towards a higher number of contiguous organ resection in those patients undergoing vascular resection (p=0.089), though no trend towards any specific histopathology (19). 

Recurrence Free Survival

Sixteen studies reported recurrence free survival (5, 8-13, 16-24). In the study by Bonvalot et al. (n=382), the 3-year abdominal recurrence rates were 10% in compartmental complete resection, compared with 47% in simple complete resection and 52% in contiguously involved organ resection (8). In the multivariate analysis, there was a statistically significant 2-fold increase in abdominal recurrence with simple complete resection, as compared to MVR in the form of compartmental resection (HR 1.99, 95% CI 1.03-3.84, p=0.04).  In a large study of 1007 patients with well differentiated liposarcoma from multicentre in Europe and North America, there was a statistically significant difference in rates of local recurrence by the number of organs resected and the use of radiotherapy (RT) between different centres (p=0.048) (5). There was significant variance between the number of organs resected from 1 (IQR 0-2) and 5 (IQR 4-7), and the implementation of RT (0-72%). The relative weight of impact of MVR and RT is not elaborated on in this study, and as such RT is a confounder with respect to the comparators of MVR and simple resection which are focused on in this guideline.

Five studies demonstrated that MVR had no statistical significance with respect to local recurrence (10, 11, 13, 23, 24). In the largest of these studies, Smith et al. (n=362), demonstrated that significant predictors of local recurrence were the grade of the tumour (Grade 1 vs 3) (HR 0.09, 95% CI 0.04-0.20, p<0.001), and positive macroscopic resection margins (R2 vs R0-1) (HR 2.64, 95% CI 1.36-5.14, p=0.004) (13). Keung et al. (n =119) demonstrated that multifocal disease was the only independent predictor of worsened local recurrence free survival on multivariate analysis (HR 1.89, 95% CI 1.1-3.23, p=0.021) (23) [36]. Rossi et al. (n=43), found that histologic subtype was the only significant predictor for local control in primary RPS (11).

Singer et al. (n=177), found on multivariate analysis that dedifferentiated histopathology (HR 3.6, 95% CI 2.2-6., p<0.0001), and contiguous organ resection was significantly associated with local recurrence (HR 1.7, 95% CI 1.02-2.8, p=0.04) (17).

In terms of the effect of specific organ resections on recurrence, Kim et al. reported the outcomes of 86 patients with primary RPS abutting the pancreas (20). Microscopic pancreatic invasion occurred in 42.5% of patients and resection of the distal pancreas was not associated with local recurrence. R2 resection was significant for local recurrence on univariate analysis, but not in the multivariate analysis (HR 3.65, 95% CI 0.89-15.09, p=0.073). Abdelfatah et al. (n=115), did not find that pancreatic resection was significant with respect to local recurrence (18). However that diaphragmatic resection was associated with a significant increase in the risk of local recurrence (HR 18.5, 95% CI 2.75-123, p=0.003). Cho et al. (n=114), found that nephrectomy improved local recurrence rates significantly for patients with FNCLCC Grade 2 tumours (p=0.048), though significance was not reached for patients with Grade 1 and Grade 3 (n=11) tumours (9). Spolverato et al. (n=24), demonstrated that vascular resection was significantly associated with local recurrence as compared to a propensity matched cohort who did not undergo vascular resection (45%, 95% CI 22-68 vs 24%, 95% CI 3-43, p=0.05) (19).

Keung et al., found in their cohort of 119 patients that R0/R1 resection was a significant predictor of distant recurrence free survival, compared to R2 resection (HR 3.18, 95% CI 1.32-7.64, p=0.010) (23). Where only the 80% of patients in that cohort with R0/R1 resection were analysed, intact tumour specimen resection was a significant predictor of distant recurrence–free survival, compared with fragmented tumour resection (HR 3.93, 95% CI 1.92-8.05, p<0.001) [35]. Lopez et al. (n=56), reported en bloc resection to have significantly improved disease-free survival as compared to enucleation (p<0.01); though they do not quantify this improvement in disease-free survival within the paper (16).

Abdelfatah et al. (n=115), found that it was only once 5 or more organs were resected that the risk of overall recurrence was significantly elevated (HR 17.8, 95% CI 3.77-84.5, p<0.001)(18). Schwartz et al. (n=571), found on multivariate analysis that larger en bloc resections (3-4 organs compared with 0 organs) were predictive of worsened disease-free survival (HR 1.56, 95% CI 1.03-2.37, p=0.04) (12). Other variables predictive of worsened disease-free survival were high grade disease (HR 2.66, 95% CI 1.88-3.77, p<0.01) and node positive disease (HR 2.08, 95% CI 1.22-3.52, p<0.01).

Perioperative Morbidity

Fourteen studies (two with overlapping patient cohorts) compared MVR with simple resection on perioperative morbidity outcomes. In the largest study with 1007 patients by TARPSWG, a weighted organ score was devised to account for differences in surgical complexity (25).  Severe postoperative complications occurred in 16.4% of patients, with 10.5% requiring reoperation, and a 1.8% mortality within 30 days of surgery. In the multivariate analysis, the resected organ score was a significant predictor of severe adverse events when a score of 8 was compared with 0 (OR 3.00, 95% CI 1.24-7.29, p=0.007); though was not significant when a score of 4 was compared with 1 (OR 1.21, 95% CI 0.85-1.73).

Bonvalot et al (n=249), found overall rates of 18% severe morbidity (Clavien-Dindo grade 3 or higher) and 3% of patients died of surgical complication (21). There was a significant effect on the rate of severe complications when more than three organs were resected, as compared with three or less organs were resected (p=0.007). When considering the most common type of organs resected (colon, kidney and psoas muscle), none were associated with any meaningful increase in morbidity. The resection of major abdominal vessels (vein OR 2.63, artery OR 3.57 with wide 95% CI) was associated with increased risk of morbidity. Three other studies reported the morbidity with respect to specific organ resections. Cho et al. (n=114) demonstrated that whilst nephrectomy was associated with a significant decrease in postoperative renal function, there was no difference between the two groups in terms of decline of long term glomerular filtration rate (9). Distal pancreatectomy was demonstrated to have no significant difference in terms of complications and severe complications (20). In contrast, vascular resection was associated with an increase in severe complications (54% vs 25%, p=0.002) (19). In the TARPSWG study of 1007 patients, there was no difference in postoperative morbidity and mortality when comparing different extents of resection across six major European and two North American centres (5).

There are two large cohorts derived from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) (26, 27). In the first study of 156 patients from 2005 to 2007, there was no difference in overall morbidity (p=0.7), severe morbidity (p=0.79), and mortality (p=1) when contiguous organ resection was performed compared with no contiguous organ resection (26). Multivariate analysis similarly found no difference where contiguous organ resection was performed for neither overall morbidity (OR 1.38, 95% CI 0.49-3.89, p=0.54), nor severe morbidity (OR 0.78, 95% CI 0.05-13.18, p=0.86).  In the second study of 564 patients between 2012-2015 with a higher proportion of patients (41%) underwent MVR in this cohort, there was no difference in overall morbidity (OR 0.72, 95% CI 0.38-2.36, p=0.31), and severe morbidity (OR 1.18, 95% CI 0.55-2.52, p=0.67) when MVR was performed (27).

Evidence Summary

Level*  

References

There is a lack of conclusive evidence to support the notion that MVR improves overall survival in primary retroperitoneal sarcoma. It does appear to improve histologic margins.

III-3, IV

(6-20, 22, 24, 25, 28-30)

There is conflicting evidence as to the effects of MVR on abdominal (local) recurrence. Most evidence found a decrease in abdominal recurrence with MVR as compared with simple resection. The conflicting evidence would appear to be effected by the influence of bias.

III-3, IV

(31-42)

The evidence is consistent that MVR does not increase morbidity significantly.

III-3, IV

(6-9, 11, 13, 19, 20, 22, 25-27)

Evidence-based recommendations

Grade*

MVR may be considered for localised resectable retroperitoneal sarcoma with the aim to improve histological margins and may decrease abdominal recurrence.

C

MVR is safe, and has comparable perioperative morbidity and mortality outcomes with simple resection.

B

Practice Points

Preoperative radiological assessment by an experienced radiologist as part of the sarcoma multidisciplinary team is also essential in operative planning.

As surgery remains the mainstay of curative therapy for retroperitoneal sarcoma, the oncologic benefit of multivisceral resection should be assessed and balanced against the expected perioperative morbidity in individual patients. 

*National Health and Medical Research Council. NHMRC levels of evidence and grades for recommendations for developers of guidelines. Canberra: NHMRC; 2009.

Administration Report

Technical Report

References

► Topic 2: Retroperitoneal Sarcoma