Intra-operative Peripheral Frozen Margin Assessment in Soft Tissue Sarcoma

Abstract

Background / Objectives

Intraoperative peripheral margin sampling in soft tissue sarcoma (STS) is a routine practice amongst musculoskeletal oncologists. Practice patterns are variable and evidence to support it is lacking. Rates of peripheral margin sampling at our institution were analyzed in addition to its clinical utility and cost-effectiveness.

Methods

Peripheral margin sampling patterns at a tertiary sarcoma center were retrospectively evaluated. Concordance between peripheral margins and final pathology was assessed using McNemar’s Test and Kappa Coefficient. Clinical outcomes were compared and a cost-utility analysis was performed.

Results

179 patients were included. 66% had peripheral margins sampled of which 23% had frozen margins analyzed. 10 patients had positive margins (5.5% of all patients; 8.4% in those with margins sampled) and R1 margins on the final tumor specimen were identified in 15 patients (8.4%). There were no R2 resections. 3 patients underwent repeat surgical resection (20%). 3 patients with R1 resections had negative peripheral margins sampled, suggesting falsely reassuring peripheral margins. Peripheral margin sampling averaged $5000/patient.

Conclusions

Routine peripheral margin sampling in STS resection is of questionable utility with added cost. Prospective studies are warranted to determine the optimal approach to surgical margin assessment.

Introduction

Surgical margin assessment in sarcoma resection is debated, with several classification schemes and variables factoring into the adequacy of margins [1]. Because of this, and in efforts to obtain negative margins, several intraoperative frozen margin assessments are often performed to assess adequacy of resection. In the setting of bone sarcomas, the utility of the marrow margin frozen assessment has been scrutinized--Anderson et al. reported the practice as either redundant or disregarded for intraoperative decision making, and thus accounted for increased cost and time without clinical utility [2]. Contrary to this finding in bone sarcomas, in breast conserving surgery, intra-operative margin assessment was helpful in over a quarter of the studied patients to yield a margin negative surgery and avoiding a subsequent re-excision [3]. The utility of frozen section margin analysis in STS specifically has yet to be studied to guide evidence-based decision making.

In efforts to minimize positive margin resection and ideally minimize local recurrences, recommendations regarding intra operative frozen assessments have been made although there lacks an evidence-based consensus. Byerly et al. suggested “6–8 perpendicular sections from all margins < 2cm” [4]. Additionally, it was recommended by the Sarcoma Sustainable Development Goals (Sarcoma SDG) to obtain margin sampling with two samples taken from the closest margin and 1–2 sections from all other margins [5]. Lastly, Cates et al. recommended margin assessment from a gross appearance after resection and six or more specimens taken from margins <2cm [6]. The challenge with these recommendations is that a planned close surgical margin relative to adjacent critical structures has not been shown to increase risk for local recurrence, especially when treated at sarcoma centers. This is with the implication that other adjuvant therapies were also utilized [7,8]. According to Gundle et al., a microscopically positive (R1) margin on critical structures after preoperative RT was not significantly different from R0 at 10 years follow up [7]. Further, they reasoned that a negative but < 1mm margin may be adequate with multidisciplinary treatment [8]. Focusing on STS, Endo et al. analyzed the use of frozen margins and comments on the limitations of this practice. They suggest that a negative frozen margin does not guarantee a negative final resection margin, but the rate at which this occurs has yet to be elucidated [9].

Considering this, we sought to assess the use of intra-operative margin assessment at a tertiary sarcoma center. We identified the frequency of sampling, the cost of use, and lastly whether the results from the intraoperative frozen assessments correlated with final margin assessment of the resected specimen. Lastly, we assessed if this subsequently affected future treatment and oncologic outcomes. We hypothesized that intra-operative frozen margin assessment in STS offers little to the intraoperative surgical decision algorithm, especially as it has been reported that close or microscopically positive margins adjacent to critical structures do not affect local recurrence rates when treated in a multidisciplinary fashion. We correlated the results of the frozen margin assessments to determine if they are concordant with final pathology margin assessments from the resected specimens. Additionally, we further postulated that this practice may result in increased cost to the patient and health care system that in select cases could be avoided.  

Materials and Methods

Study design and Setting

We performed a retrospective chart review of patients treated with STS of the trunk/extremities at a tertiary sarcoma center by fellowship trained Orthopaedic/Surgical Oncologists from 2005 to 2019.

Participants/Study Subjects

Inclusion criteria included all patients with extremity/truncal soft tissue sarcoma treated primarily at a tertiary referral center by fellowship trained orthopedic/surgical oncologists with a practice focused on sarcoma from 2005 to 2019. Exclusion criteria included retroperitoneal/thoracic disease as well as osseous disease. Additionally, those who had previous inadvertent resections were excluded, as were those treated non-operatively with definitive radiation and/or chemotherapy.

Description of experiment, treatment or surgery

Data on sarcoma subtype, tumor size/stage of disease, approximation of critical structures, intra-operative frozen sampling, and final resection specimen pathology margins were abstracted. Peripheral margins represent samples taken outside of the tumor proper, vs final margins representing the status of the resected specimen. Pathology results were reviewed for the final diagnosis, and were classified according to the R0, R+1, R1, and R2 UICC residual tumor resection scheme [7].

Variables, Outcome measures

We assessed the concordance of negative margins with the final R classification for the resected tumor specimen. We then evaluated the frequency of intra-operative frozen margin sampling; how many specimens were taken and whether this impacted intra-operative decision making.

We determined rates of local recurrence and incidence of pulmonary metastatic disease at least 2 years post-operatively through medical record review. Lastly, the cost of frozen margin sampling was determined based on the billing practices of the institution to in an effort to quantify financial implications of this practice and ideally comment on its cost utility.

Statistical analysis

McNemar’s test and the Kappa coefficient were used to assess the degree of agreement between positive peripheral margins and the final for frozen and non-frozen margins, both separately and combined. Chi square tests were used to assess associations between two-year recurrence and positive peripheral margins by subtype; and between two-year recurrence and subtype, AJCC 8th ed. classification, and final grade. All analyses were conducted using SAS^® software version 9.4 for Windows^® (SAS Institute Inc., Cary, NC). A p-value of <0.05 was considered statistically significant. 

Results

179 patients were identified for inclusion in the study. 119 patients (66%) had peripheral margins sampled, 27 of which (23%) had intraoperative frozen margins. If peripheral margins were obtained, five or six samples were most common with number of samples ranging from 1 to 9 in our dataset. (56%). Positive peripheral margins were identified in 10 patients, one of which was on intraoperative frozen with the remainder on permanent (5.5% of all patients; 8.4% in those with margins sampled). R1 margins on the final tumor specimen were identified in 15 patients (8.4%). There were no R2 final tumor resections (Table 1). The most common tumor subtype was undifferentiated pleomorphic sarcoma (UPS) representing 31.3% (Table 2). Only one case was identified to have an intra-operative decision to resect more tissue because of the results of the frozen sample. Of the 15 patients with R1 margins, 1 had frozen margins sampled and 7 had margins sampled but sent for permanent. Three of those patients underwent repeat surgical resection at a later date (20%), 2 of which had peripheral margins sent but only as a permanent (Table 3). Additionally, there were 3 patients with R1 final resections with negative frozen peripheral margins sampled, suggesting these 3 patients had falsely reassuring peripheral margins, although they were all sampled as permanent margins with results unknown at the time of index operation (Table 3).

Table 1:

Final margin resection classification with sampling patterns of peripheral margins.

Final Resection Margin by Peripheral Margins obtained
	R0	R+1	R1	R2	Total
# Patients: (% )	126 (70.4%)	38 (21.2%)	15 (8.4%)	0 (0%)	179 (100%)
Frozen Margins	22	4	1	0	27
Permanent Margins	61	24	7	0	92
No Margins	43	10	7	0	60

Table 2:

Subtype of Soft Tissue Sarcomas identified for study analysis.

Subtype	N	Percent
Leiomyosarcoma	13	7.26
Myxofibrosarcoma	34	18.99
Myxoid Liposarcoma	24	13.41
Others	43	24.02
Synovial Sarcoma	9	5.03
UPS	56	31.28

Table 3:

Margin sampling patterns among resections with R1 final tumor margins.

Table of Patients with R1 final Margins
	Obtained	Positive	Negative	Intra-Op Decision	Subsequent Re-Excision
Frozen Margins	1	1	0	1 Excised more	0
Permanent Margins	7	4	3	0	3 (2 with positive peripheral margins)
No Margins	7	NA	NA	NA	0

If a positive peripheral margin was identified, there was a 50% chance that final resection margins would be positive (R1), vs if margins were negative only 4.6% had an R1 final margin which was statistically significant (P= 0.003). There was weak to moderate agreement between positive peripheral margins and final specimen margins (kappa = 0.42). When subdivided between frozen and permanent only, there was moderate agreement between positive final specimen margin and positive frozen margins (kappa=0.74, McNemar’s p-value=.31) and weak agreement between positive permanent only peripheral margins and margins of the resected tumor specimen (kappa= 0.42, McNemar’s Test p= 0.32). There was no statistically significant difference in final margins and AJCC tumor size or stage (p-value = 0.7043, 0.6800 respectively). The overall 2-year local recurrence rate was 8.86%, which was not statistically significantly different by final margin status (P= 0.9765) or by sarcoma subtype (P= 0.8149), although likely underpowered. 10 of the 14 patients who developed a local recurrence had R0 resections, 3 were R+1, and 1 was R1. Two-year local recurrence was identified in 14 patients (7.8%). 15 patients had metastatic disease at diagnosis, while 57 patients (32%) had metastatic disease by 2 years after diagnosis. 26 patients (14.5%) were deceased by 2 years after diagnosis.

At this institution, the inpatient charge for each frozen margin is $503.69 whereas permanent billed for $500.80. Every frozen sample is reflexively processed for a permanent. This results in an average total charge of $5022.45 per patient based on an average of 5 margin samples. This does not account for multiple sectioning of a frozen assessment, or any increases in intra-operative time that may further contribute to overall cost of care.

Discussion

Intra-operative frozen margin assessment is a long-standing practice that is utilized by many sarcoma surgeons. In the setting of musculoskeletal oncology, its utility has become a topic of debate. Anderson et al. reported that the examination of split gross specimens alone appeared to be an adequate adjunct when compared to frozen section in osseous sarcoma resections [2]. These conclusions along with others in the literature have focused mostly on osseous sarcomas. In this study we aimed to determine the practice patterns in soft tissue sarcoma and assess the utility of intraoperative frozen margin assessment. This is among the first studies to retrospectively review the practice patterns of peripheral margin sampling amongst fellowship trained musculoskeletal and surgical oncologists at a single sarcoma institution.

Despite our series demonstrating moderate agreement between frozen peripheral margin sampling and final R margin classification, the utility of routine margin sampling at the time of resection remains questionable. When considering a scenario such as a planned close or R1 margin along a critical structure after neoadjuvant radiotherapy, it can be assumed that the outcome of intraoperative sampling of that margin would not impact surgical decision making. In this case, it would be prudent to not send a peripheral margin as it adds little benefit but does have the potential to increase cost and operative time. Alternatively, however, if further tumor resection could be undertaken, it would be logical to send a frozen sample if there is concern for adequacy of the resection at a particular area of the tumor bed as this can be completed at the time of index operation. This is of particular importance in single-stage procedures involving resection and often complex soft tissue reconstruction. In this setting, the added time and cost may be useful if it prevents additional surgical intervention due to inadequate margins or eventual compromise to the soft tissue reconstruction due to revision surgery. With the aforementioned scenarios in mind, certain conclusions can be made. The practice of routine intraoperative margin sampling for permanent analysis appears unjustified. When used in a selective manner, however, frozen sampling may be a useful tool that provides real-time feedback.

Interest has continued to grow in the development of novel intraoperative adjuncts to guide oncologic resection [10]. Computer assisted tumor surgery has shown promise in the osseous sarcoma literature with reports suggesting higher accuracy of resection when used by experienced surgeons [11, 12]. Similar adjuncts have been investigated in the setting of soft tissue sarcoma. Brookes et al. reported a case series using indocyanine green fluorescence-guided surgery (FGS) and showed reduced unexpected positive margin rates [13]. Whitley et al. describe a phase I trial using a novel protease-activated fluorophore in humans with soft tissue sarcoma and showed significantly increased tumor fluorescence which can be used to guide resection [14]. Advances are also being made in the use of augmented reality with multiple case reports of the technology being used to guide intraoperative management [15, 16]. Much of this technology, while promising, remains in its infancy and is not without its limitations. Prior to widespread adoption, there is a need for more investigation on the utility of these adjuncts, comparative analysis with respect to established clinical practices such as frozen sampling, and further research regarding cost-effectiveness and feasibility of application [17].

Regarding associations between margin sampling and subsequent disease specific outcomes such as local recurrence, this retrospective dataset is likely underpowered. Further, it is reasonable to suggest a practice of selective margin sampling when there is concern for focal inadequate margins or in select cases of particular histologies. For example, Miwa et al. have reported on histological grades of frozen section for the diagnosis of soft tissue tumors and found that frozen section yielded low diagnostic accuracy in adipocytic and fibroblastic tumors, and that diagnosis should be based on permanent section [18]. Since the diagnostic utility was low on these frozen samples, the utility of frozen sampling is scrutinized to a greater extent given it may not identify positive peripheral margins, particularly in these tumor types [19, 20]. As previously stated, exceptions may exist and there may be utility to frozen margin sampling prior to definitive complex soft tissue reconstruction given the moderate agreement between positive frozen margins and final pathology. There does not appear to be utility for intra-operative margins for permanent analysis without frozen analysis due to weak agreement that does not guide surgical decision making.

Lastly, there is an inherent cost to sampling with intra-operative frozen assessment which does not account for operative time. This cost impacts both patient and hospital infrastructure, case time and turnover, and likely contributes an overall unnecessary cost to both the patient and health care system. In light of current cost-conscious practices and improving access to care, timeliness remains relevant. If a practice does not guide surgical decisions at the time of intervention, it is unlikely that the cost of this practice is of benefit to either the patient or the hospital system.

The present study is not without limitations. For one, this collection of cases is based on documented CPT procedure codes for radical resections. This presents a possible selection bias or incomplete capture of previously treated truncal and extremity sarcomas. Further, given our study’s retrospective nature, it is subject to sampling biases and sampling patterns of individual surgeons at one institution. Additionally, the relatively small size of our dataset limits the power of some of our conclusions, namely in determining the existence of associations between the practice of margin sampling and disease specific oncologic outcomes. Given this, our study alone cannot provide formal recommendations regarding how or when margins should be sampled. A large prospective and randomized study is needed to allow sufficient statistical power to fully assess the utility of this practice in the clinical setting.

Conclusions

Our paper is novel in the sense that it is amongst the first to seek to establish evidence for or against routine peripheral margin sampling amongst fellowship trained surgical and musculoskeletal oncologists at a single sarcoma institution. We have drawn attention to the practice of routine frozen sampling of peripheral margins in soft tissue sarcoma resections and have demonstrated, at least in this single institution series, its low clinical utility and contributes a non-negligible expense to patients and the health system. There is also additional need for studies to assess alternatives to frozen margin sampling in the context of soft tissue sarcomas specifically. Ultimately, we propose clinicians adopt a selective margin sampling practice, with use of frozen margins for real time data when it would influence surgical decision making.

Table 4:

Contingency Table showing rate of positive peripheral margins by final margin status with corresponding McNemar’s test, indicating no evidence of disagreement, and the kappa coefficient, showing evidence of weak agreement.

Table of positive peripheral by Final Status
Positive peripheral	Final Status
Key
Row %
Col %	Negative	Positive	Row Total
Negative	104 95.41 95.41	5 4.59 50.00	109 91.60
Positive	5 50.00 4.59	5 50.00 50.00	10 8.40
Column Total	109 91.60	10 8.40	119 100.00
Frequency Missing = 60, McNemar’s p-value > 0.99

Simple Kappa Coefficient
Estimate	Standard Error	95% Confidence Limits
0.4541	0.1454	0.1691	0.7392

Synopsis for Table of Contents:

Routine intraoperative peripheral margin sampling is commonly employed by musculoskeletal surgical oncologists in the operative treatment of soft tissue sarcoma. There are several recommendations that encourage peripheral margin sampling but there remains no evidence-based consensus. We attempt to determine the rate of frozen peripheral margin sampling at a single tertiary sarcoma center, analyze its clinical utility, and assess its cost to patients and the health system.