Should We Stop Assessing Intraoperative Frozen Section Marrow Margins During Ewing Sarcoma Resection?

Abstract

Background

Ewing sarcoma is a rare and highly aggressive pediatric bone cancer that is histologically composed of small, round blue cells. These histologic findings can make it difficult to assess intraoperative frozen section bone marrow margins because the bone marrow that regenerates after preoperative chemotherapy has a similar appearance, especially on frozen section analysis. Prior studies have more broadly questioned the utility of intraoperative assessment of bone marrow margins using frozen sections during the resection of bone sarcomas; however, to our knowledge, no prior study has specifically characterized the accuracy or clinical utility of evaluating bone marrow margins on frozen sections during long bone Ewing sarcoma resection.

Questions/purposes

(1) How accurate is the assessment of intraoperative bone marrow margins using frozen sections during the resection of long bone Ewing sarcoma? (2) What changes to the prespecified surgical plan were made in response to positive intraoperative assessments of bone marrow margins? (3) Is intraoperative assessment of bone marrow margins on frozen sections associated with improved survival free from local recurrence, development of metastatic disease, or Ewing sarcoma–specific death?

Methods

Sixty-four patients who underwent primary resection of a conventional Ewing sarcoma of a long bone at our institution were analyzed. In this cohort, 81% (52 of 64) of patients had frozen bone marrow margins assessed intraoperatively. There were no identifiable reasons for why some patients had or did not have a frozen section performed, and we could not detect differences in demographic or surgical features between patients who did versus those who did not have intraoperative margins assessed. Intraoperative margins were assessed as negative on frozen sections in 88% (46 of 52) of patients and positive in the remaining 12% (6 of 52) of patients. To determine the rates of false-positive and false-negative intraoperative assessments, the results of intraoperative frozen sections were compared with the assessments of those same initial intraoperative margins as reviewed on final pathology reports. In patients with positive intraoperative assessment of bone marrow margins on frozen sections, we reviewed the surgical records and operative notes to determine whether additional bony resection was performed or if any other changes were made to the prespecified operative plan as a result of the concern for a positive intraoperative margin. Data were available on all study endpoints in 86% (55 of 64) of patients at a minimum follow-up time of 2 years. Kaplan-Meier curves and log-rank tests were used to compare survival free from local recurrence, development of metastatic disease, and Ewing sarcoma–specific death between patients with intraoperative margin assessment and those without. We also compared these same oncologic outcomes between patients whose margins were called positive versus negative intraoperatively.

Results

All bone marrow margins that were assessed as negative on intraoperative frozen sections were confirmed to be negative when examined on final pathology reports (100% [46 of 46]). All bone marrow margins that were assessed as positive on intraoperative frozen sections were actually negative when the same tissue margins were examined on final pathology results (6 of 6) and confirmed by re-review by an experienced bone pathologist for this study. Five of those six patients had an additional, unnecessary bone resection; in the sixth patient, the orthopaedic surgeon documented a high suspicion for false-positive intraoperative assessment and did not perform additional resection. When comparing patients who had an intraoperative margin assessed by frozen section versus those who did not, there were no differences in local recurrence-free survival at 2 years (93% [95% confidence interval (CI) 81% to 99%] versus 100% [95% CI 72% to 100%]; p = 0.99), development of metastatic disease-free survival at 2 years (85% [95% CI 71% to 94%] versus 78% [95% CI 40% to 97%]; p = 0.62), or Ewing sarcoma–specific death-free survival at 2 years (91% [95% CI 78% to 97%] versus 100% [66% to 100%]; p = 0.99). Similarly, when comparing patients whose margins were true negatives versus false positives intraoperatively, there were no differences in local recurrence-free survival at 2 years (92% [95% CI 79% to 98%] versus 100% [95% CI 54% to 100%]; p = 0.99), development of metastatic disease-free survival at 2 years (86% [95% CI 71% to 95%] versus 80% [95% CI 28% to 99%]; p = 0.56), or Ewing sarcoma–specific death-free survival at 2 years (90% [95% CI 77% to 97%] versus 100% [95% CI 54% to 100%]; p = 0.99).

Conclusion

During long bone Ewing sarcoma resection, in a study of our patients, routine assessment of intraoperative bone marrow margins on frozen sections appears to provide no demonstrable clinical benefit and may lead to excessive resection of normal bone. If an orthopaedic surgeon has a specific concern for a positive bone marrow margin, then an intraoperative frozen section may certainly still be warranted. However, in the era of modern MRI imaging, routine intraoperative assessment of bone marrow margins using frozen sections is likely unnecessary in this setting and may be omitted to save time and cost.

Level of Evidence

Level III, diagnostic study.

Introduction

The goal of Ewing sarcoma surgical resection is to completely excise the primary tumor and achieve negative surgical margins, as this is associated with decreased rates of local recurrence and improved overall survival [2,6,7]. To help ensure that negative margins are obtained during surgical resection of bone sarcomas such as Ewing sarcoma, many orthopaedic oncologists routinely send a frozen bone marrow margin from the residual bone after resection for intraoperative pathologic assessment. Although the assessment of intraoperative margins for Ewing sarcoma is routinely performed at many institutions, the clinical benefit of this practice has not been fully characterized in previous reports. In fact, some studies suggest that intraoperative assessment of frozen section margins is of little utility in osteosarcoma [5] and bone sarcomas more broadly [1,4,8]. Furthermore, intraoperative margins are particularly difficult to interpret on frozen sections in the context of Ewing sarcoma, as the histologic appearance of this tumor type (classically described as small, round blue cells) is visually similar to the bone marrow that regenerates after preoperative chemotherapy [3]. This visual similarity can cause surgical pathologists to incorrectly call margins positive on intraoperative frozen sections (Fig. 1). However, to our knowledge, no prior study has specifically investigated the occurrence of false-positive intraoperative bone marrow margins at the time of resection in patients with Ewing sarcoma, and it is possible that misinterpretation of frozen section marrow margins can affect surgical decision-making and oncologic outcomes. Additionally, no study has characterized the impact of these false-positive intraoperative margins on surgical or long-term patient outcomes specifically for patients with Ewing sarcoma of long bones.

Fig. 1.

Example images of a single patient’s bone marrow margins as assessed on intraoperative false-positive frozen sections (left), compared with when those same margins were sectioned and stained on true-negative final pathology (right). 200× magnification (mag) images: the frozen section pictures show small round blue cells (SRBCs) that because of the artifacts of the frozen section (such as cellular swelling and tissue rolling up/not sticking to the slide, which is especially problematic if the sample has any element of fat from the bone marrow) appear much larger than they do on the permanent sections. On the frozen slides, some neutrophils in the background seem much smaller than the clusters of SRBCs. In contrast, on the permanent sections, the size of the cells is much more homogeneous. The large, pink, and blobby cells that are most notable are megakaryocytes. What was interpreted as suspicious on frozen sections is just regenerating B and red blood cells. 400× magnification (mag) images: the frozen sections show cannonballs of SRBC, and this clustering together is very suspicious because normal hematopoietic marrow should not display this feature on frozen sections. In contrast, the permanent section shows the marrow nicely spread out and dotted with erythroid islands of immature red blood cells. The cells are not as large as they were in the frozen section images.

Therefore, the purpose of this study was to address the following questions: (1) How accurate is the assessment of intraoperative bone marrow margins using frozen sections during the resection of long bone Ewing sarcoma? (2) What changes to the prespecified surgical plan were made in response to positive intraoperative assessments of bone marrow margins? (3) Is intraoperative assessment of bone marrow margins on frozen sections associated with improved survival free from local recurrence, development of metastatic disease, or Ewing sarcoma–specific death?

Patients and Methods

Study Design and Setting

This was a retrospective, comparative study of patients with Ewing sarcoma of a long bone who underwent primary tumor resection at a single tertiary sarcoma center. All operative procedures were performed between 1994 and 2023 by one of four orthopaedic surgeons (JLH, HSS, GEH, JML) with fellowship training in musculoskeletal oncology.

Patients

After receiving institutional review board approval, we queried our institution’s orthopaedic oncology database, which includes all patients treated for a musculoskeletal bone or soft tissue tumor at our institution since 1987. Of the initial 266 patients identified between 1987 and 2023, we excluded 150 patients with Ewing sarcoma of a non–long bone (including the pelvis, vertebrae, scapula, soft tissue) or who had preoperatively known skip metastases that might confuse margin assessment, 39 patients who underwent initial resection at an outside hospital, and 13 patients with a final pathology report that was not consistent with conventional Ewing sarcoma. Based on the exclusion criteria described, no eligible patients were identified in the database between the years of 1987 and 1993. Therefore, our final cohort included 64 patients who underwent primary resection of a conventional Ewing sarcoma of a long bone at our institution between 1994 and 2023 (Fig. 2).

Fig. 2.

Flow chart of inclusion and exclusion criteria used to identify a cohort of patients with conventional Ewing sarcoma of a long bone undergoing primary surgical resection at a single institution. ^aOf note, we initially screened all patients in our institutional database from its inception in 1987 to the end of 2023. However, after screening, there were no eligible patients in the database between the years of 1987 and 1993, so our cohort of included patients only spans from 1994 to 2023.

The median (IQR) duration of clinical follow-up in this study was 5 years (3 to 8). Duration of clinical follow-up was defined as the time between the date of a patient’s index operation and the last date they were seen by an orthopaedic oncology provider at our institution. For our first and second research questions regarding the accuracy of intraoperative margins and their impact on predefined surgical plans, we had complete data for all study variables in 100% of patients (64 of 64). For our third research question regarding the impact of assessing intraoperative margins on local recurrence rates, development of metastatic disease, and Ewing sarcoma–specific death, we had complete data on all study endpoints in 86% (55 of 64) of patients at a minimum follow-up of 2 years, meaning that 14% of patients (9 of 64) were lost to follow-up at 2 years. There was no difference in the proportion of patients with a minimum of 2 years of follow-up between the groups who did versus did not have a margin assessed (87% [45 of 52] versus 83% [10 of 12]; p = 0.67). One-half of the patients (32 of 64) in this cohort had followed up at our institution within the last 5 calendar years.

Data Variables and Collection

Two independent reviewers (SWC, AR) extracted data from the electronic medical record; in instances of discordance between the two reviewers, the senior author (JML) was consulted, and consensus was reached. Data collected included demographic, surgical, and pathologic data and information on local recurrence, development of metastases, and disease-specific death. After the completion of preoperative chemotherapy, all patients had a preoperative MRI scan for surgical planning purposes. At our institution, the goal of long bone Ewing sarcoma resection is an en bloc resection with a 2-cm margin as measured on the postchemotherapy T1-weighted image sequence, using the joint line as a reference point for intraoperative measurement.

Descriptive Data

The median (IQR) age of patients in this study was 14 years (11 to 19). The cohort was 63% (40 of 64) self-identified male, and there were no differences in demographic or surgical characteristics between male and female patients (data not shown). Given the limited sample size of this study, a further by-sex analysis was not conducted when answering our research questions, as any multivariate statistical model with sex as a factor would be underpowered to detect differences between outcomes of interest.

At our institution, patients are asked to self-identify their race as: (1) American Indian or Alaska Native, (2) Asian, (3) Black or African American, (4) Hispanic/Latino/a, (5) Middle Eastern or North African, (6) Native Hawaiian or other Pacific Islander, (7) White, (8) none of these, (9) prefer not to answer, or (10) unable to provide. In this cohort, 88% (56 of 64) of patients self-identified their race as White, 2% (1 of 64) as Black or African American, 6% (4 of 64) as Hispanic/Latino/a, and 5% (3 of 64) as none of these, prefer not to answer, or unable to provide. No patients in our cohort self-identified as any of the other races listed above. Similarly, at our institution, patients are asked to self-identify their ethnic group as (1) Cuban; (2) Mexican, Mexican American, or Chicano/a; (3) Puerto Rican; (4) other Hispanic, Latino/a, or Spanish origin; (5) none of these; (6) prefer not to answer; or (7) unable to provide. In this cohort, 94% (60 of 64) of patients self-identified as none of these, prefer not to answer, or unable to provide. Three patients self-identified as other Hispanic, Latino/a, or Spanish origin. One patient self-identified as Mexican, Mexican American, or Chicano/a. No patients in this cohort self-identified as any of the other ethnicities listed above.

As for tumor location, 41% (26 of 64) of this cohort had Ewing sarcoma of the femur, 28% (18 of 64) of the tibia, 14% (9 of 64) of the fibula, 14% (9 of 64) of the humerus, 2% (1 of 64) of the radius, and 2% (1 of 64) of the ulna. All 64 patients received preoperative and postoperative chemotherapy, and no patients received preoperative radiation to their primary tumor.

Assessment of Frozen Sections

It has not historically been our institutional practice to routinely split resected bone specimens for gross inspection of marrow margins. Rather, when intraoperative bone marrow margins were assessed, they were principally evaluated by standard hematoxylin and eosin staining of frozen sections. In this study, 81% (52 of 64) of patients had intraoperative bone marrow margins assessed on frozen sections during their index surgical resection. There were no differences in demographic features or surgical characteristics between patients who had a margin assessed intraoperatively compared with patients who did not have a margin assessed intraoperatively (Table 1). During the time period that this study investigated, our institution did not have a set protocol describing when intraoperative margins needed to be evaluated or not. Thus, the decision to send an intraoperative margin was at the sole discretion of the attending orthopaedic surgeon and was likely based on patient-specific features and surgical characteristics. However, this decision process was not well documented in the reviewed operative notes, and it was not possible to retrospectively identify a rationale for why some patients had intraoperative frozen sections assessed while others did not. The year of surgery, specific orthopaedic surgeon, tumor location, reconstruction type, preoperative presence of metastasis, and other demographic features were investigated and could not be retrospectively associated with the decision to intraoperatively assess a bone marrow margin (see Limitations section).

Table 1.

Patient characteristics

Parameter	All patients (n = 64)	Intraoperative margins assessed (n = 52)	No intraoperative margins assessed (n = 12)	p valuea
Age in years	14 (11-19)	14 (10-18)	17 (12-27)	0.14
Male sex	63 (40)	65 (34)	6 of 12	0.34
Race/ethnicityb				0.51
White	88 (56)	88 (46)	10 of 12
Black	2 (1)	2 (1)	0 of 12
Hispanic	6 (4)	6 (3)	1 of 12
Other/multiple/unknown	5 (3)	4 (2)	1 of 12
Tumor location				0.08
Femur	41 (26)	44 (23)	3 of 12
Tibia	28 (18)	31 (16)	2 of 12
Fibula	14 (9)	10 (5)	4 of 12
Humerus	14 (9)	13 (7)	2 of 12
Radius	2 (1)	0 (0)	1 of 12
Ulna	2 (1)	2 (1)	0 of 12
Pre- and postoperative chemotherapy	100 (64)	100 (52)	12 of 12
Preoperative metastatic disease	13 (8)	12 (6)	2 of 12	0.64
Pathologic fracture present	22 (14)	23 (12)	2 of 12	0.99
Soft tissue mass present	83 (53)	79 (41)	12 of 12	0.11
Type of reconstruction				0.38
Allograft	59 (38)	63 (33)	5 of 12
Noncustom megaprosthesis	23 (15)	21 (11)	4 of 12
Custom megaprosthesis	2 (1)	2 (1)	0 of 12
Amputation	3 (2)	2 (1)	1 of 12
None	9 (6)	8 (4)	2 of 12
Other	3 (2)	4 (2)	0 of 12
Follow-up duration in years	5 (3-8)	5 (3-9)	4 (3-6)	0.23
Patients with minimum 2-year follow-up	86 (55)	87 (45)	10 of 12	0.67
Tumor size in cm	6.9 (3.9-12.0)	7.0 (3.6-12.0)	6.4 (4.6-13.1)	0.91
Treatment-induced necrosis	99 (90-100)	100 (98-100)	99 (90-100)	0.16

Data presented as median (IQR), % (n), or proportion of cases when the denominator is < 25.

^aThe p values were derived from statistical tests comparing values for patients who did versus did not have an intraoperative margin assessed. Mann-Whitney U test was used to compare continuous variables, and Fisher exact test was used to compare categorical variables.

^bInformation on race and ethnicity were manually extracted from the electronic health record and are indicative of self-reported identity.

In patients where margins were assessed intraoperatively, frozen section samples were obtained by scooping out the marrow at the cut residual long bone margin with a curette. These samples were immediately transferred to the pathology department, where they were further processed for frozen section analysis. Of the 52 patients who had an intraoperative bone marrow margin assessed, 88% (46 of 52) had their initial bone marrow margin called negative on intraoperative analysis of frozen sections, and the remaining 12% (6 of 52) had their initial bone marrow margin called positive. The 12% (6 of 52) of patients who had a positive intraoperative assessment of bone marrow margins on frozen section had their initial intraoperative pathology report re-reviewed by a blinded, fellowship-trained musculoskeletal anatomic pathologist (RS) at the time of the current study. We did not re-review the 46 histologic slides from patients who were called negative on frozen sections intraoperatively, as all 46 were confirmed to be negative on final pathology at the time of the initial resection. There was concordance between the initial intraoperative assessment and the reviewing pathologist in 100% (6 of 6) of these patients’ histologic slides.

Accuracy of Intraoperative Margins

To answer our first study question, we reviewed patients’ historical pathology reports and systematically compared the intraoperative frozen section results with the final pathology results for the same intraoperative margin sample. Specifically, we calculated the true-positive, true-negative, false-positive, and false-negative rates of intraoperative frozen bone marrow margins. A true-positive margin was defined as one that was called positive intraoperatively and confirmed to be positive on final pathology. Conversely, a true-negative was a margin that was called negative intraoperatively and confirmed to be negative on final pathology. False-positive margins were defined as those that were called positive intraoperatively but found to be negative on final pathology, whereas false-negative margins were called negative intraoperatively but found to be positive on final pathology.

Impact of Margin Assessment on Surgical Management

To answer our second study question, we evaluated the surgical implications of positive intraoperative assessment of bone marrow margins. After identifying all patients with positive intraoperative margins on frozen section, we reviewed the surgical records and operative notes to determine whether additional resection was performed as a result of the concern for a positive intraoperative margin. As the goal of sarcoma surgery at our institution is to remove the tumor as a single en bloc resection with negative margins, we assumed that all subsequent bony resections after the initial resection were additional, unplanned resections, unless otherwise stated in the operative note. For patients who underwent additional bony resections, we also documented whether any additional margins were assessed by intraoperative frozen sections, and if so, what their intraoperative pathology results were. We then compared the results of all subsequent intraoperative margins with the results of those margins on final pathology to identify any discrepancies and their impact on surgical management. In patients with positive intraoperative margins on frozen sections where additional bony resection was not performed or additional margins were not assessed by intraoperative frozen sections, we recorded the surgeon’s rationale.

Impact of Margin Assessment on Local Recurrence, Metastasis, and Survival

To answer our third study question, we compared the rates of 2-year survival free from local recurrence, development of metastasis, and Ewing sarcoma–specific death between patients who had intraoperative margins assessed and those who did not. Additionally, we compared these same oncologic outcomes between patients with false-positive intraoperative margins and those with true-negative margins to assess whether a false-positive assessment may suggest a worse prognosis. All patients were assessed for local recurrence and Ewing sarcoma–specific death; patients with preoperatively known metastasis (n = 8) were not included in the analysis of postoperative development of metastasis.

Ethical Approval

We obtained ethical approval for this study from the institutional review board at Vanderbilt University Medical Center (IRB# 231707).

Statistical Analysis

Results of descriptive statistics are reported as median and IQR, unless otherwise indicated. We used the Mann-Whitney U test to compare differences between groups of continuous variables. We used the Fisher exact test to compare proportions between groups of categorical variables. We generated Kaplan-Meier curves and performed log-rank tests to compare survival free from local recurrence, development of metastasis, and Ewing sarcoma–specific death between groups. Statistical analyses were performed using GraphPad Prism, version 10.4.1, for Macintosh (GraphPad Software). A threshold of p value < 0.05 was considered statistically significant.

Results

Accuracy of Intraoperative Margins

For the six patients who had an initial intraoperative frozen section assessed as positive during their operation, none were confirmed to be positive on the final pathology review. Thus, the rate of true-positive intraoperative frozen bone marrow margins was 0% (0 of 6) and the rate of false-positive intraoperative margins was 100% (6 of 6). For the 46 patients who had an initial intraoperative frozen section assessed as negative during their operation, all 46 were determined to be truly negative on the final pathology review. Thus, the rate of true-negative intraoperative frozen bone marrow margins was 100% (46 of 46) whereas the rate of false-negative intraoperative margins was 0% (0 of 46). On final pathologic assessment, all patients had negative surgical margins after their initial preplanned surgical resection, regardless of whether a bone marrow margin was assessed intraoperatively, and in patients in whom one was assessed, regardless of what the intraoperative frozen bone marrow margin assessment demonstrated (Fig. 3).

Fig. 3.

Flow chart describing the number of patients with Ewing sarcoma who had frozen bone marrow margins assessed intraoperatively.

Impact of Margin Assessment on Surgical Management

For the patients who had a false-positive intraoperative bone marrow margin, we recorded the pathologic analysis of the intraoperative frozen section and compared it with the pathologic analysis reported on the final pathology report (Table 2). Five of the six patients with a false-positive intraoperative bone marrow margin underwent additional resection of bone; in the remaining patient, the orthopaedic surgeon documented a high suspicion for false-positive assessment and did not perform additional resection. For the five patients who underwent additional bone resection, a second bone marrow margin from the new resection level was assessed by a subsequent intraoperative frozen section. In all five of these patients, the second bone marrow margin was intraoperatively assessed as negative for residual disease, and the operating surgeon proceeded with reconstruction. In all five of these patients, the additionally resected bone was confirmed to be negative for residual tumor on final pathologic assessment.

Table 2.

Cohort of six false-positive assessments of intraoperative bone marrow margins as evaluated by frozen section taken during the resection of long bone Ewing sarcomas

Age in years	Sex	Location	Reconstruction	Assessment of initial intraoperative bone marrow margin on frozen section	Assessment of the same bone marrow margin when reviewed on final pathology report	Additional resection performed
11	Male	Distal tibia	Allograft	“Suspicious for malignancy; discussed with surgeon and additional margin requested”	“Marrow with trilineage hematopoiesis, negative for malignancy”	Yes
13	Female	Distal femur	Allograft	“Bone and marrow elements with scattered atypical cells”	“Negative for malignancy”	Yes
12	Female	Proximal femur	Megaprosthesis	“Positive for malignancy”	“Bone marrow with trilineage maturation; negative for malignancy”	Yes
14	Female	Proximal humerus	Allograft	“Hematopoietic elements with focal atypical cells, cannot exclude involvement by tumor”	“Bone marrow with normal trilineage hematopoiesis and scattered hemosiderin-laden macrophages; negative for malignancy”	No
17	Male	Proximal femur	Megaprosthesis	“Rare clusters of atypical cells suspicious for malignancy”	“Hematopoietic bone marrow with erythroid hyperplasia; negative for malignancy”	Yes
13	Male	Distal femur	Allograft	“Tumor focally present”	“Hematopoietic marrow with focal clusters of immature differentiated myeloid cells; negative for malignancy”	Yes

Impact of Margin Assessment on Oncologic Outcomes

To assess the impact of intraoperative margin assessment on oncologic outcomes, we collected patient data on local recurrence, development of metastatic disease, and Ewing sarcoma–specific death. When comparing patients who had an intraoperative margin assessed versus those who did not, there was no difference in local recurrence-free survival (HR not calculated because of no local recurrences in the “margins not assessed” group; p = 0.20), development of metastatic disease-free survival (HR 1.0 [95% confidence interval (CI) 0.2 to 4.5]; p = 0.98), or Ewing sarcoma–specific death-free survival (HR 2.3 [95% CI 0.6 to 9.8]; p = 0.40) as assessed by Kaplan-Meier curves and log-rank testing at a median (IQR) follow-up time of 5 years (3 to 8) (Fig. 4). When comparing outcomes at 2-year minimum follow-up, when we had complete follow-up data for 86% of patients, there were no differences in local recurrence-free survival (93% [95% CI 81% to 99%] versus 100% [95% CI 72% to 100%]; p = 0.99), development of metastatic disease-free survival (85% [95% CI 71% to 94%] versus 78% [95% CI 40% to 97%]; p = 0.62), or Ewing sarcoma–specific death-free survival (91% [95% CI 78% to 97%] versus 100% [66% to 100%]; p = 0.99) between patients who had an intraoperative margin assessed by frozen section versus those who did not (Table 3).

Fig. 4.

Kaplan-Meier curves demonstrating differences between patients who had intraoperative margins assessed via frozen sections (blue lines) versus those who did not have intraoperative margins assessed via frozen sections (red lines) in terms of (A) survival free from local disease recurrence, (B) survival free from the development of metastatic disease, and (C) survival free from Ewing sarcoma–specific death. The p values are derived from log-rank tests comparing Kaplan-Meier curves at a median (IQR) follow-up time of 5 years (3 to 8). A color image accompanies the online version of this article.

Table 3.

Two-year oncologic outcome-free survival for patients who did versus did not have intraoperative bone marrow margins assessed by frozen section

	Margins assessed (n = 52)	No margins assessed (n = 12)	p value
Two-year local recurrence-free survival	93 (81-99)	100 (72-100)	0.99
Two-year metastatic disease-free survival	85 (71-94)	78 (40-97)	0.62
Two-year Ewing sarcoma–specific death-free survival	91 (78-97)	100 (66-100)	0.99

Data presented as % (95% CI).

For patients who had an intraoperative margin assessed, we compared those who had a true-negative intraoperative margin versus a false-positive intraoperative margin and found that there was no difference in local recurrence-free survival (HR not calculated because of no local recurrences in the “margins not assessed” group; p = 0.30), development of metastatic disease-free survival (HR 1.1 [95% CI 0.2 to 8.0]; p = 0.91), or Ewing sarcoma–specific death-free survival (HR 1.6 [95% CI 0.3 to 8.4]; p = 0.65) as assessed by Kaplan-Meier curves and log-rank testing at a median (IQR) follow-up time of 5 years (3 to 9) (Fig. 5). When comparing outcomes at 2 -year minimum follow-up, there were no differences in local recurrence-free survival (92% [95% CI 79% to 98%] versus 100% [54% to 100%]; p = 0.99), development of metastatic disease-free survival (86% [95% CI 71% to 95%] versus 80% [28% to 99%]; p = 0.56), or Ewing sarcoma–specific death-free survival (90% [95% CI 77% to 97%] versus 100% [54% to 100%]; p = 0.99) between patients whose margins were true negatives versus false positives intraoperatively (Table 4).

Fig. 5.

Kaplan-Meier curves demonstrating differences between patients who had true-negative intraoperative margins assessed via frozen sections (blue lines) versus those who had false-positive intraoperative margins assessed via frozen sections (red lines) in terms of (A) survival free from local disease recurrence, (B) survival free from the development of metastatic disease, and (C) survival free from Ewing sarcoma–specific death. The p values are derived from log-rank tests comparing Kaplan-Meier curves at a median (IQR) follow-up time of 5 years (3 to 9). A color image accompanies the online version of this article.

Table 4.

Two-year oncologic outcome-free survival for patients with true-negative versus false-positive bone marrow margins as assessed by intraoperative frozen section

	True negatives (n = 46)	False positives (n = 6)	p value
Two-year local recurrence-free survival	92 (79-98)	100 (54-100)	0.99
Two-year metastatic disease-free survival	86 (71-95)	80 (28-99)	0.56
Two-year Ewing sarcoma–specific death-free survival	90 (77-97)	100 (54-100)	0.99

Data presented as % (95% CI).

Discussion

As obtaining negative surgical margins is of paramount importance during the resection of musculoskeletal sarcomas, sending tissue for intraoperative assessment of margins on frozen sections has been routine clinical practice at some institutions. Prior studies have called the utility of this practice into question, suggesting that appropriate preoperative imaging and gross inspection of intraoperative margins may be adequate to achieve negative final surgical margins. However, these studies have primarily focused on patients with osteosarcoma or have included few patients with Ewing sarcoma, which may be more difficult to distinguish from normal bone marrow on gross inspection. Furthermore, the histologic appearance of Ewing sarcoma as small, round blue cells can be difficult to distinguish from regenerating bone marrow on assessment of frozen sections. Therefore, we sought to determine the accuracy and utility of assessing intraoperative bone marrow margins specifically during the resection of long bone Ewing sarcoma. In our cohort, we observed a 12% (6 of 52) false-positive rate of intraoperative bone marrow margin assessments, which led to unnecessary additional bone resections in 5 of 6 patients. Taking a frozen section margin or having a false-positive assessment was not associated with a detectable difference in long-term oncologic outcomes. These data have led our musculoskeletal oncology group to stop routinely assessing bone marrow margins intraoperatively in the resection of long bone Ewing sarcoma and rather evaluate intraoperative margins only when there is a high clinical suspicion for a positive margin.

Limitations

The results of our study must be interpreted in the context of certain limitations. First and foremost, we did not find any differences in the rates of survival free from local recurrence, development of metastatic disease, or Ewing sarcoma–specific death between patients who did versus did not have a margin assessed intraoperatively. While we did not detect any differences in these outcomes, it must be noted that we did not perform an a priori power analysis, and given the small sample size of this study, our investigation may be underpowered to detect true differences, if such differences were to exist. Furthermore, approximately 14% (9 of 64) of our cohort was lost to follow-up at a minimum time point of 2 years. While the proportion of missing patients was not different between the compared groups, because missing patients are logically not doing as well as patients who are accounted for, our estimations of event-free survival could certainly overestimate the health of our patient sample. Given these limitations, no sweeping claims should be made about the impact of assessing an intraoperative margin on long-term oncologic outcomes in the setting of long bone Ewing sarcoma resection.

Furthermore, based on our limited sample size, we did not perform a sex-based analysis for any of our research questions. One cannot assume that findings drawn from a population consisting of males and females would apply equally to both males and females separately. While there is no biological rationale that we are aware of as for why male or female patients may be at a higher risk for false-positive intraoperative margins, this is an important factor to consider in future investigations with increased sample sizes.

This study is also at risk for selection bias, as we were unable to ascertain why some patients had intraoperative margins assessed while others did not. During the time period this study investigates, our institution did not have a set protocol describing when intraoperative margins needed to be evaluated or not. Thus, the decision to send an intraoperative margin was at the sole discretion of the attending orthopaedic surgeon and was likely based on patient-specific features and surgical characteristics. However, this decision process was not well documented in the reviewed operative notes, and it was not possible to retrospectively identify a rationale for why some patients had intraoperative frozen sections assessed while others did not. The year of surgery, specific orthopaedic surgeon, tumor location, reconstruction type, preoperative presence of metastasis, and other demographic features were investigated and could not be retrospectively associated with the decision to intraoperatively assess a bone marrow margin. Future studies could address this limitation by randomizing patients to have margins assessed or not.

Another limitation of this study is that we only re-reviewed the frozen section specimens for the patients in whom intraoperative margins were retrospectively identified to be false positives. At the time of the current study, a blinded, senior musculoskeletal anatomic pathologist (RS) re-reviewed the six intraoperative frozen section bone marrow margins from the historical Ewing sarcoma resections. At the time of this re-review, the pathologist was not informed of the previous assessment (positive versus negative) and did not know the purpose of the current study. The goal of this exercise was simply to ensure that the histologic assessments that were called positive intraoperatively were not fluke errors or misdiagnoses by a previous on-call pathologist. In all six patients, it was agreed that the intraoperative section appeared to be positive for malignancy or was strongly suspicious for malignancy. Furthermore, it was agreed that the final marrow margins were negative for malignancy after the initial resection. We did not re-review the 46 patients who were called negative on frozen sections intraoperatively, as all 46 were confirmed to be negative on final pathology at the time of the initial resection. Rather, only the six intraoperative margins that were historically discrepant when they were assessed on frozen sections versus final pathology were re-reviewed at the time of the current study.

Furthermore, we did not include flat bones such as the scapula, pelvis, or ribs in this study. This was because it can be difficult to ascertain whether and how bone marrow margins should be obtained in these patients, including where they should be taken from and how many locations should be included. Therefore, it is not routine clinical practice at our institution to send intraoperative bone marrow margins for frozen sections in patients with any type of flat bone tumor unless there is clinical suspicion for a positive margin in a specific location.

Finally, this was a retrospective study performed at a single institution with a large orthopaedic oncology referral population, which may have decreased the generalizability of our findings to centers that do not specialize in sarcoma.

Accuracy of Intraoperative Margins

In this study, we found that intraoperative assessment of frozen sections was of limited accuracy during the resection of long bone Ewing sarcomas. This finding further emphasizes the questionable utility of routinely assessing intraoperative margins via frozen sections during sarcoma resections, as has previously been suggested by prior studies. Our findings are in concordance with a previous multicenter analysis that described the limited utility of intraoperative frozen sections in 475 patients with primary malignant long bone tumors [8]. However, in that study, only 1.5% (7 of 475) of patients had a diagnosis of Ewing sarcoma. In their cohort, all bone marrow margins were classified as negative intraoperatively, except for 2 of 475 patients. Similar to our findings, when these two intraoperative margins were reviewed on final pathology, both were deemed to have actually been negative after initial resection. Importantly, both false-positive findings were in patients with Ewing sarcoma, meaning that for the subset of their cohort with Ewing sarcoma, the false-positive rate for intraoperative margins was 29% (2 of 7). Another previous study of 142 patients with primary sarcomas compared frozen section versus gross examination for bone marrow margin assessment and found that frozen sections are commonly redundant or disregarded for intraoperative surgical decisions [1]. While this study included 28 patients with Ewing sarcoma (20% of the cohort), the study did not report the specific rates of false-positives by tumor histology and included both long and flat bones, making it difficult to assess whether frozen sections may be more or less useful for certain tumor locations or types of sarcomas. A similar, more recent study investigated the utility of intraoperative frozen sections and gross assessment of margins in 166 patients with extremity bone sarcomas [4]. This study included 15 patients with Ewing sarcoma (9% of this cohort), none of whom had a misdiagnosed margin intraoperatively by either gross or frozen section assessment. Given the strategic homogeneity of our cohort, our study of 64 patients is the largest investigation of intraoperative margins for patients with Ewing sarcoma reported to date and includes more patients with Ewing sarcoma than the three studies described above combined. Thus, our study contributes to the field by specifically answering the question of whether intraoperative assessment of bone marrow margins using frozen sections is helpful in this setting, and we do not extrapolate our findings to other histologic sarcoma subtypes. Future studies should strive to replicate this homogeneous type of cohort in other rare forms of sarcoma rather than lumping many tumor types and locations together.

Impact of Margin Assessment on Surgical Management

We found that false-positive margin assessments led to additional resection of bone in 5 of 6 patients, resulting in increased operative time and alterations in preplanned reconstructions. Although not directly applicable to this cohort, the need for additional resection because of a false-positive margin could be complicated for a patient receiving a custom oncologic endoprosthesis. Furthermore, eliminating routine assessment of intraoperative margins would reduce the average operative time, as sending intraoperative margins for frozen section can add between 15 and 45 minutes to an operative procedure [6]. While not a focus of the current study, assessing frozen sections is also associated with financial costs, including both the cost of pathologic assessment and the cost of increased time in the operating room. In 2018, Thévenin-Lemoine et al. [9] showed in a series of 20 patients who underwent Ewing sarcoma resection that a 2-cm bone margin measured from the postchemotherapy T1-weighted sequence led to negative final pathologic margins in all patients. The results of our study, which show no true-positive intraoperative margins or final resection margins, underscore the ability of appropriate preoperative imaging and planning alone to adequately guide resection boundaries in long bone Ewing sarcoma.

Impact of Margin Assessment on Oncologic Outcomes

This was the first study to our knowledge that sought to evaluate the relationship between assessing intraoperative margins during Ewing sarcoma resection and long-term oncologic outcomes, including local recurrence, development of metastatic disease, and disease-specific mortality. We were unable to detect a difference in these critical outcomes between patients who had intraoperative margins assessed and those who did not. While intriguing, these preliminary results must be interpreted with caution as these findings are likely underpowered, as extensively described in the Limitations section. Future studies with larger sample sizes should be conducted to definitively investigate these findings. With these limitations in mind, the findings from this study preliminarily suggest that the elimination of routine intraoperative margin assessments would not compromise long-term oncologic outcomes.

Conclusion

During the resection of long bone Ewing sarcomas at our institution over the last three decades, the routine assessment of intraoperative bone marrow margins on frozen sections provided no demonstrable clinical benefit. Routine intraoperative assessment of frozen sections was associated with a 12% (6 of 52) false-positive rate, leading to unnecessary additional resections in 5 of 6 patients. Based on these findings, our updated institutional practice pattern is to avoid perfunctory assessment of intraoperative bone marrow margins during the resection of long bone Ewing sarcomas. Rather, if an orthopaedic surgeon has a specific concern for a positive intraoperative margin, then an intraoperative frozen section of the area of concern may certainly still be conducted. We conclude that in the era of modern MRI imaging, routine intraoperative assessment of bone marrow margins using frozen sections is likely unnecessary. If confirmed in larger studies, we believe that frozen sections can be omitted, and this may avoid resecting additional bone and reduce operative time and cost in this specific setting.