The Variability in Surgical Margin Reporting in Limb Salvage Surgery for Sarcoma

Abstract

Background

Surgical margins are a standard reported measurement in tumor surgery that has implications for functional outcome, local control, and overall survival. There is no single accepted classification, and it is unclear what form or margin reporting predominates in the sarcoma literature.

Methods

We performed a PubMed literature search to identify articles that reported surgical margins and oncologic outcomes in limb salvage surgery for sarcoma from 1980 to 2013. We recorded the margin classification, specialty of the journal, specialty of the author, and location of the authors' institution.

Results

We found that 159/448 (35%) of articles included in the study did not report surgical margins. Of the 289 papers that did include data on margins, 160 (55%) of articles used Enneking's classification. There has been an increase over time in the proportion of articles reporting surgical margins by the residual tumor (R) classification and the proportion of articles reporting margins dichotomously as “positive” or “negative.”

Conclusions

We did not find a common method for reporting margins in the limb salvage sarcoma literature. Of most concern was over 1/3 of clinical reports of oncologic outcomes did not include margin status, which substantially compromises any conclusions that readers may infer about treatment success, local recurrence, or survival. We believe there should be renewed efforts to encourage use of a common surgical margin reporting system that is simple, reproducible, and prognostic.

Introduction

Soft-tissue and bone sarcomas classically arise from the mesodermal embryonic cell layer and account for less than 1% of all cancer-related deaths.¹ In 2014,15,000 new patients in the United States were estimated to be diagnosed with a sarcoma¹. Historically, the primary treatment for sarcoma was amputation of the affected limb to achieve acceptable rates of local control². Advances in chemotherapy in the 1980s dramatically increased the survival of adolescent bone sarcoma and coincided with a movement to minimize the number of amputations required for adequate management of sarcoma^3,4. In addition, increasing use of computed tomography scans, magnetic resonance imaging, and peri-operative radiation therapy allowed limb salvage surgery to become the preferred intervention to amputation^2,5.

Limb salvage surgery challenges practitioners to find a balance between removing tissue sufficient to prevent recurrence or metastatic spread while retaining structures integral to maintain acceptable functional outcomes⁶. The quantification of this measure of “closeness” to the tumor is reflected in the reported margin, the significance of which remains a consistent source of controversy and may be dependent on the specific type of sarcoma in question^7-9. The final surgical margin is a standard measurement for all operatively treated sarcomas and has been shown to be a key prognostic factor for local recurrence, metastatic spread, and overall survival for a variety of sarcomas^2,10-12. Therefore, it is imperative that the oncologic team (surgeons, medical oncologists, radiation oncologists, radiologists, and pathologists) utilize a common classification for margin reporting.¹³ The ideal classification allows the treating team to understand the risk of recurrent disease, and assists with treatment decisions, surveillance protocols, patient counseling, and comparative research.

All systems of margin reporting account for tumor at the edge of the resection (a “positive” margin), but may differ on whether it is further categorized as “grossly” or “microscopically” positive^13-15. In addition, some systems will equate all “negative” margins, while others will account for the extent of a negative margin by differentiating between a resection through the reactive zone of peri-tumorous tissue or reporting a distance from the closest margin^16-17. This results in surgical margin criteria that are not always interchangeable, making comparison of outcomes and risk factors for recurrence following limb salvage surgery across multiple investigations difficult or impossible. These inconsistencies suggest a clear need for a universal system of margin reporting, however, the scope of the issue and magnitude of the problem has not been previously defined. Therefore, we seek to identify 1) the common classification systems for reporting surgical margins utilized within the current literature, 2) the trends in the use of these classifications over time, and 3) characteristics of the article (journal, specialty of author, location of authors' institution) that may demonstrate a preference for a specific classification system.

Materials and Methods

A PubMed database search was conducted on July 4, 2014 using the keyword combination “sarcoma limb salvage.” PubMed filters for human species, English language, and publication dates between January 1,1980 and December 31, 2013 were applied to the search. The starting year of 1980 was selected because it was the year the Enneking surgical margin classification system was first reported¹³. Article titles, MeSH terms, abstracts, methods, and results were examined to exclude: 1) articles that did not report new surgical margin data, 2) case reports, 3) articles unrelated to sarcomas, 4) animal studies, 5) articles that could not be retrieved, and 6) articles that did not report any oncologic treatment outcomes (overall survival, disease-free survival, event-free survival or local recurrence). Articles that did not report new surgical margin data included review articles, meta-analyses, editorials, comments, letters, practice guidelines, and technical reports.

For articles included in the study, the first author's departmental affiliation, the first author's institutional location, and the journal's specialty focus were recorded. The first author's departmental affiliation was categorized as orthopaedic surgery, general surgery, medical oncology, radiation oncology, or other. The “other” category included articles where the first authors were from epidemiology, pathology, pediatrics, radiology, or where the department was not listed in the article. The first author's institutional location was categorized as: Africa, Asia, Europe, Oceania, or Americas. There were only six articles from South America; therefore these articles were grouped with articles from North America to form the “Americas” category. The journal's subject was categorized using the Science Citation Index (SCI) as: “oncology,” “orthopaedic surgery,” “surgery, not listed,” or “radiology, nuclear medicine, & medical imaging.”

Articles were placed into one of six unique categories based on the surgical margins criteria used to report data (Table 1). Articles that reported surgical margins using Enneking's criteria (radical, wide, marginal, or intralesional) were categorized as “Enneking¹³.” Trovik used a classification system based upon Enneking's criteria to report adequate (radical or wide) or inadequate (marginal or intralesional) margins¹⁴. Articles that used this classification system were also categorized as “Enneking.” Articles that used the residual tumor (R) classification criteria (R0, negative margins, R1, microscopically positive margins, and R2, grossly positive margins) were categorized as “R Classification¹⁸.” Articles that described surgical margins simply as positive or negative without further description were categorized as “Dichotomous¹⁶.” Articles that reported surgical margins using three categories of margins without further description, typically as negative, close, or positive, were categorized as “Trichotomous.” Articles that reported surgical margins as the distance from the resection edge to the closest site of tumor were categorized as “Measurements¹⁷.” Articles that reported surgical margins using the criteria proposed by Kawaguchi were categorized as “Japanese Orthopaedic Association (JOA)¹⁵.” Articles that did not report surgical margins were categorized as “Margins Not Reported.” Additionally, if an article was found to use multiple surgical margin criteria to report data, then each criteria type was included and recorded separately.

Table 1.

Explanation of the various categories of surgical margin reporting

Name of Criteria	Values Used to Report Surgical Margins
Measurements17	Closest distance to tumor
Dichotomous16	Positive or negative
Enneking13	Intralesional, marginal, wide, or radical
Trovik14	Adequate (wide or radical) or inadequate (intralesional and marginal)
Trichotomous19,20	Negative (or clear), close (or uncertain), or positive
Japanese Orthopaedic Association15	Curative wide margin, wide margin, marginal margin, and intralesional margin.
Residual Tumor (R classification)18	R0 (no tumor at margin), R1 (microscopic positive), or R2 (grossly positive)

Statistical Analysis

Data analysis was performed with XLSTAT (version 2014.4.08, Addinsoft USA, New York, NY, USA). The articles were ordered by year of publication and separated by the surgical margin criteria used to report data. The trends over time in the number of articles and proportion of articles using a specific surgical margin criterion were analyzed. The Mann-Kendall statistics (S) and Mann-Kendall trend test p-values were recorded. A positive S indicates an increasing trend, while a negative S indicates a decreasing trend. The null hypothesis (H₀) for the Mann-Kendall trend test is that no monotonic trend exists. The alternative hypothesis (H_A) is that a monotonic trend exists. Additionally t-test p-values were recorded in order to determine if a relationship exists between the surgical margin criteria used and first author's departmental affiliation, first author's institutional location, or SCI subject matter of the article. Articles classified as JOA or trichotomous were analyzed from their year of first appearance in the literature, 2004 and 1993, respectively. Statistical significance was defined as p<0.05.

Results

Between 1980 and 2013, 1,207 articles were identified through PubMed with the keyword combination: sarcoma limb salvage. After applying PubMed filters, 969 articles remained for review. 521 articles were subsequently excluded leaving a collection of 448 scientific reports available for analysis (Fig. 1).

Figure 1.

Margin vs. Year

Of the articles between 1980 and 2013 that examined at least one oncologic outcome, 160 used the Enneking criteria, 38 used the R classification criteria, 38 used dichotomous criteria, 26 used trichotomous criteria, 23 used measurements, and 5 used the JOA criteria. One article reported margins using two different criteria (Enneking and R Classification). 159 articles (35%) did not report surgical margins.

From 1980 to 2013, there was an increasing trend in the number of articles using Enneking's criteria (S = 402, p < 0.01), articles using R classification criteria (S = 272, p < 0.01), articles using dichotomous criteria (S = 238, p < 0.01), and articles that did not report surgical margins (S = 324, p < 0.01). There was no statistically significant trend in articles that reported surgical margins using the JOA criteria, measurements, or trichotomous criteria. (Table 2) (Figure 2)

Table 2.

Statistical tests for monotonic correlation and trend for the number of articles reporting surgical margins from 1980 to 2013 (unless otherwise noted).

Surgical margin criteria used to report	Mann Kendall statistic (S)	Mann-Kendall trend test p-value
Enneking	S = 402	P < 0.01 **
R Classification	S = 272	P < 0.01 **
Dichotomous	S = 238	P < 0.01 **
Trichotomous (From 1993)	S = 50	P = 0.20
Measurements	S = 41	P = 0.51
JOA (From 2004)	S = −9	P = 0.41
Margins not reported	S = 324	P < 0.01 **

^**Denotes statistical significance.

Figure 2. Percent of total articles using surgical margin criteria from 1980 to 2013.

From 1980 to 2013, there was an increasing trend in the proportion of articles using R classification criteria (S = 231, p < 0.01) and dichotomous criteria (S = 144, p = 0.025) to report surgical margins (Table 3). There were no other statistically significant trends in the proportion of articles using specific criteria to report surgical margins.

Table 3.

Statistical tests for monotonic correlation and trend for the proportion of articles reporting surgical margins from 1980 to 2013 (unless otherwise noted).

Surgical margin criteria used to report	Mann Kendall statistic (S)	Mann-Kendall trend test p-value
Enneking	S = 47	P = 0.49
R Classification	S = 231	P < 0.01 **
Dichotomous	S = 144	P = 0.025 **
Trichotomous (From 1993)	S = −7	P = 0.88
Measurements	S = −87	P = 0.18
JOA (From 2004)	S = −14	P = 0.19
Margins not reported	S = −79	P = 0.25

^**Denotes statistical significance.

Margin vs. Department

Examination of the first authors' departmental affiliation revealed that 198 were from orthopaedic surgery, 89 were from general surgery, 40 were from medical oncology, 17 were from radiation oncology, and 105 were from “other” departments. Articles that included surgical margin data were further analyzed. First authors from orthopaedic surgery (P = 0.035) and medical oncology (P < 0.01) were more likely to use Enneking's system than other criteria to report surgical margins. First authors from general surgery (P = 0.025) and radiation oncology (P < 0.01) were less likely to use Enneking's criteria than other criteria to report surgical margins. There was no statistically significant difference (P = 0.99) in the use of the Enneking criteria by first authors from “other” departments.

Margin vs. Continent

Upon examination of the physical location of the first author, we found that 172 were from Europe, 153 were from the Americas, 111 were from Asia, 7 were from Oceania, and 6 were from Africa. Articles that included surgical margin data were further analyzed. First authors from Europe were more likely (P < 0.01) to use Enneking's criteria than other criteria to report surgical margins. First authors from the Americas were less likely (P < 0.01) to use Enneking's criteria than other criteria to report surgical margins. There was no statistically significant difference in the use of Enneking's criteria by authors from Africa (P = 0.31), Asia (P = 0.52), and Oceania (P = 0.31).

Margin vs. SCI

Examination of the SCI subject category revealed: 156 articles were published in orthopaedic surgery journals, 109 articles were published in surgery journals, 92 articles were published in oncology journals, 22 articles were published in radiology, nuclear medicine, and medical imaging journals, and 70 were published in journals not listed in the SCI. Articles that included surgical margin data were further analyzed. Articles published in journals that were classified as “radiology, nuclear medicine, and medical imaging” by the SCI subject category were less likely (P = 0.012) to report margins using Enneking criteria. There was no statistically significant difference in the use of Enneking's criteria to report surgical margins in articles published in orthopaedic surgery journals (p = 0.083), surgery journals (p = 0.21), oncology journals (p = 0.096), and journals not listed in the SCI (p = 0.21).

Discussion

There have been multiple authors who have recognized that the methods to report sarcoma surgical margins are disjointed^7,13. In our study, we found that there exists six unique surgical margin criteria commonly used in the scientific literature: Enneking, R classification, dichotomous, trichotomous, measurements, and the JOA. While the number of articles reporting surgical margins has increased since 1980, there remain a notable proportion of articles not reporting surgical margins. When surgical margins are reported, Enneking's criteria are the most commonly used among authors, especially orthopaedic surgeons. To our knowledge, this is the first study to quantitatively analyze surgical margin reporting in the sarcoma and limb salvage literature.

Our data demonstrate that there are several distinct surgical margin classification systems, none of which were represented in the majority of publications. In our opinion, it was surprising that 35% of articles did not report surgical margins, as this limits the information that may be gleaned, specifically for any conclusions regarding local control of the tumor. If we examine the remaining reports (when authors chose to report surgical margins), the majority (55%) used Enneking's criteria to report their data. The next most common surgical margin reporting systems were R classification and dichotomous (both 13%). Trichotomous criteria, measurements, and the JOA criteria were the least used surgical margin criteria (9%, 8%, and 2% respectively). This variability is also a concerning finding, as the investigations that do report margins utilize several different systems that may not be easily compared to one another.

Our trend analysis shows that there has been a positive linear trend in the number of articles that use Enneking's criteria, articles that use dichotomous criteria, and articles that do not report margins. This finding is likely due to the overall increase in the number of articles reporting oncologic outcomes after limb salvage surgery over time. The number of articles reporting surgical margins using trichotomous criteria, measurements, and the JOA criteria did not show any definable changes over the study period. When examining trends in the proportion of margins reported in each system over time to account for the overall increase in publications, we found that there has only been an increase in the proportion of articles using R classification and dichotomous criteria. One possible explanation for the growing popularity of these systems is that physicians are becoming convinced that additional detail apart from the presence of tumor at the surgical margin is not important. Another possibility is that this reflects an increase in articles written by those outside orthopaedic surgery. As the Enneking classification was motivated by surgical decision-making and developed and presented primarily in the arena of orthopaedic oncology, it is plausible that those outside this specialty are unlikely to find it as commonplace or useful as an orthopaedic surgeon. Finally, it may be explained by the simplicity of dichotomous criteria as the easiest and most reproducible method to report margins. In reality, this trend is likely due to a combination of factors.

We also examined criteria usage preferences of first authors in other departments, by continent, and journal subject category. We found that first authors from orthopaedic surgery, hematology and oncology, or Europe were more likely to use Enneking's criteria than other systems to report surgical margins. Alternatively, first authors from surgery, radiation oncology, the Americas, or journals in medical imaging were more likely to use a system other than the Enneking classification. The lack of preference for the Enneking classification system in the Americas may be explained by the presence of more authors from radiation oncology (who use non-Enneking criteria to report surgical margins 92% of the time) within the Americas. The preference for the Enneking classification system in Europe may be explained by the presence of more first authors from hematology and oncology, which use Enneking criteria to report surgical margins 86% of the time within Europe.

This report has two major findings that warrant further study. The first is that the substantial number of papers not reporting any detailed margin status, but still commenting on the oncologic outcome after surgical treatment, is problematic. Submitting authors and accepting journals should be expected to include some mention of margins if commenting on local or systemic control. The second is that there is a substantial need for a universal system of margin reporting. The lack of such a system inhibits interdisciplinary communication and comparative literature reviews. Some reporting can be extrapolated to other systems, such as “positive” in the dichotomous classification and “intralesional” in Enneking's. However, some labels cannot. For instance, a “RO” margin could be either a “marginal” or “wide” margin in Enneking's system. This is critical as margins are the language of limb salvage surgery and local control of sarcoma, and a mechanism of documentation that is simple, meaningful, reproducible, and prognostic is an important and attainable goal.

This study had a number of limitations. We only identified articles for inclusion using a PubMed search. Although the PubMed database contains numerous articles, it does not contain the entire scientific literature. Our purpose was not to determine the most appropriate margin classification system, and more work must be performed prior to any definitive recommendations on which classification is best. Clearly this is an important and complex issue that is outside the scope of this report. Therefore, our findings are generally observational, and any conclusions drawn from these data are necessarily limited.

To conclude, it is essential that sarcoma specialists have an understanding of the effects of surgical margins on the prognosis of their patients. Disjointed surgical margin reporting in the scientific literature creates difficulty in comparing treatment outcomes between articles that use different margin criteria. A single universal surgical margin reporting system to facilitate communication between physicians would enhance the understanding of all the subspecialties involved in sarcoma treatment.

Source of Funding

No funding was required for this investigation.