Ann Surg Oncol (2014) 21:3185–3191 DOI 10.1245/s10434-014-3905-1
ORIGINAL ARTICLE – BREAST ONCOLOGY
Reasons for Re-Excision After Lumpectomy for Breast Cancer: Insight from the American Society of Breast Surgeons MasterySM Database Jeffrey Landercasper, MD1,2, Eric Whitacre, MD3, Amy C. Degnim, MD4, and Mohammed Al-Hamadani, MBChB, MPH5 Department of Surgery, Gundersen Health System, La Crosse, WI; 2Norma J. Vinger Center for Breast Care, Gundersen Health System, La Crosse, WI; 3Breast Center of Southern Arizona, Tucson, AZ; 4Department of Surgery, Mayo Clinic, Rochester, MN; 5Department of Medical Research, Gundersen Health System, La Crosse, WI
1
ABSTRACT Background. There is marked variability of re-excision rates after initial lumpectomy for breast cancer. Reasons for re-excision and variability across surgeons have not been well documented. We hypothesized the American Society of Breast Surgeons (ASBrS) MasterySM Program can identify reasons for re-excision. Methods. Data from January 1 to 7 November 2013 were evaluated in the ASBrS MasterySM Program to determine re-excision lumpectomy rate (RELR). On 1 June 2013, a tool to track reasons for re-excision was developed. Variation in re-excision rate by surgeon and patient characteristics was performed by Chi square test and Fisher’s test for univariate analysis, then logistic regression with backwards elimination method for multivariate analysis. Results. For 6,725 patients undergoing initial lumpectomy for cancer, 328 surgeons reported 1,451 (21.6 %) patients had one or more re-excisions. The most common reasons for re-excision were ink positive margins in 783 (49.7 %), margin \1 mm (34.3 %), and margin 1–2 mm (7.2 %). By multivariate analysis, re-excision rates were lower in patients aged less than 35 years, with White (non-) Hispanic ethnicity, and, among surgeons in solo practice, more years in practice and higher-volume practice. Conclusion. Half of re-excisions after initial lumpectomy were performed for margins that are positive. Most of the
Ó Society of Surgical Oncology 2014 First Received: 14 April 2014; Published Online: 22 July 2014 J. Landercasper, MD e-mail: [email protected]
remainder were for negative close (\1–2 mm) margins. This information corroborates surgeon survey data regarding reasons for re-excision and provides proof of concept the MasterySM Program can capture surgical outcome data in real time, providing opportunity and a method for future performance improvement.
Re-excision of positive margins after initial lumpectomy for breast cancer is guideline-compliant and represents quality care.1–3 Breast cancer recurrence is reduced by reexcision when cancer is present at the lumpectomy surface;4–6 however, there is marked variation of national and international re-excision rates, which range from 5 to 70 %.7–10 The National Quality Forum (NQF) and other stakeholders in quality improvement would consider this degree of performance variation to be a potential marker for a gap in care quality because variation is part of their ‘importance criteria’ to justify development of a new quality measure (QM).11,12 Much discussion has occurred regarding whether re-excision lumpectomy rate (RELR) is a measure of quality.3,13,14 Briefly, the argument against naming RELR as a QM is that negative unintended consequences may occur, such as performing larger-volume lumpectomies (leading to worse cosmetic outcomes), omitting re-excision when needed in order to maintain a benchmark, or possibly increased local recurrence rates. ‘Risk aversion’ to lumpectomy, a surgeon fear of being penalized for initial positive margins, could also occur, resulting in a higher mastectomy rate if individual surgeon performance was tied to some measure of accountability, such as payment or patient steerage. Lastly, agreement on the definition of a ‘positive’ margin has been elusive, but would be necessary prior to using RELR as a QM.
3186
METHODS The ASBrS Mastery of Breast SurgerySM Program was developed in 2006 to help surgeons document their breast procedures, patient care, and to allow quality measurement.20,21 Surgeons can enter information on patient demographics, operations, and QM with immediate peerperformance comparison. After Institutional Review Board approval, de-identified MasterySM data from 1 January 2013 to 7 November 2013 were evaluated to determine RELR. On 1 June 2013, a dropdown pop-up menu was added to MasterySM to track reasons for re-excision. Overall, RELR was defined as the number of patients undergoing any type of re-excision (lumpectomy or any type of mastectomy) after initial lumpectomy divided by the number of patients having initial lumpectomy for cancer. We excluded patients undergoing bilateral operations and patients/procedures with incomplete or missing data. Patient age was determined by year of birth, not exact date of birth, to prevent patient identification and to comply with security requirements for de-identified data review. Statistical analysis was used to compare RELR stratified by patient and physician factors using univariate analysis (Chi square test, Fisher’s test). The RELR for comparisons and for Fig. 1 was defined as the number of patients undergoing re-excision lumpectomy, excluding mastectomy
100
80
Re-excision lumpectomy rate (%)
In contrast to the robust literature detailing variability of re-excision, the causes of re-excision after initial lumpectomy have not been well documented. Survey research suggests some re-excisions may be performed unnecessarily due to differences in surgeon and radiation oncology opinion regarding adequacy of margin width.15–17 The motivation to identify reasons for re-excision is to aid a performance improvement plan to reduce secondary operations. The NQF endorses efforts to reduce reoperations, specifying them as QM for many operation types.18 Furthermore, reoperations and readmissions are prioritized for improvement in the National Quality Strategy, a report commissioned under the Affordable Care Act.19 Reducing secondary operations has safety, cost, and patient-centered benefits. The recipe for a performance improvement plan and the design of clinical trials to reduce re-excisions will require information on causes of re-excision. The primary aim of this study was to identify contemporary causes of reexcision. We hypothesized the American Society of Breast Surgeons (ASBrS) MasterySM Program could identify reasons for re-excision. Secondary aims were to identify causes of variation of re-excision in ASBrS members and to search for patient and provider demographic factors associated with re-excision.
J. Landercasper et al.
60
40
20
0 Surgeon #1
Surgeon #183
Individual breast surgeons N= 183
FIG. 1 Variability of breast re-excision lumpectomy rate for 183 surgeons
as a re-excision type, divided by the number of patients having initial lumpectomy, because the database was deidentified for security for exact date of surgery preventing differentiation of a patient with initial mastectomy from a mastectomy for re-excision. Statistical significance was reported at an alpha level of 0.05 using p value with 95 % confidence interval (CI). For analysis regarding surgeons’ volume of procedures and re-excision rate, surgeons with ten or less lumpectomies were excluded, then Wilcoxon ranksum and Kruskal–Wallis tests were used. Multivariate analysis was performed with logistic regression with backwards elimination method (at an alpha level of 0.05) to control for patient and physician factors. Significant main predictors for outcome of re-excision were reported with adjusted odds ratio and 95 % CI. RESULTS Overall, 328 surgeons reported on 6,725 patients undergoing initial lumpectomy for cancer, with 1,451 (21.6 %) patients undergoing one or more re-excisions of any type (lumpectomy or mastectomy). Of these 6,725 patients, the re-excision type was lumpectomy in 1,131 (16.8 %). Re-excision rate ranged from 2 to 91 % in surgeons reporting more than ten cases (Fig. 1). In patients undergoing re-excision, the mean, median, and range of the number of re-excisions was 1.1 (±0.27), 1, and 1–4, respectively. Mean patient age was 61.9 years (±13.12) and ranged from 13 to 90. The causes of 1,575 re-excisions are detailed in Table 1. The most common causes of reexcision were ink positive margins in 783 (49.7 %), margin \1 mm (34.3 %), and margin 1–2 mm (7.2 %). In the univariate analysis for patient factors, we found that the re-excision rate was significantly highest in patients who were 35–44 years of age (17.5 %), and
Causes of Re-Excision Lumpectomy in Breast Cancer
3187
TABLE 1 Reasons for re-excision lumpectomy procedures Reason Ink positive margin
N
% 783 49.7
Margin \1 mm
540 34.3
Margin 1–2 mm
114 7.2
Post-lumpectomy imaging demonstrated evidence of residual disease
38 2.4
Prior surgery elsewhere, margin status uncertain
25 1.6
Margin C2 mm, but desire wider margin
16 1.0
Tumor board recommended wider margin Fragmented specimen, margin status uncertain
16 1.0 3 0.2
Radiation oncologist recommended wider margins Other Total procedures
2 0.1 48 3.1 1,575 100
Hispanics (21.2 %) [p-values were 0.005 and 0.018, respectively]. For physician factors, re-excision rate was lowest in solo practice (13.5 %), community population of less than 50 K (7.7 %), physician with more than 30 years of experience (7.8 %), and for those who were enrolled in any quality program (15.2 %). p-Values were \0.01 for all (Table 2). In multivariate analysis, we found similar trends to the univariate analysis (Table 3). Adjusted analysis for the mean rate of re-excision for surgeons was compared with the volume of their surgeries. Mean re-excision rate was highest for those surgeons who performed less than ten initial lumpectomies during the 11month study period. There was no statistical difference between those who performed 25–49, 50–99 and C100 initial lumpectomies (Fig. 2). DISCUSSION Reoperation for lumpectomy margin status averages 20– 40 % and ranges from 5 to 70 %.7–10 The European Union of Breast Cancer Specialists (EUSOMA) recommends measurement of secondary breast operations after either lumpectomy or mastectomy.22 Certified EUSOMA breast units audit ‘one-step surgical success rate’ for invasive cancer, endorse it as a measure of quality, then specify a minimum standard (80 %) and a target goal (90 %).22 Multiple operations compared with a single operation decrease value. Value has been defined as the ratio of performance and cost.23,24 Costs include both the financial and non-financial burdens of care endured by the patient.23 Higher-value care is more affordable for patients and affordability is one of the three aims of the US National Quality Strategy for healthcare.19,25 The goals of this study were to identify reasons for reexcision after initial lumpectomy for breast cancer and to identify the factors correlating with variability of re-
excision as a first step towards reducing secondary operations. There is no doubt that RELR varies widely by provider and institution.7,10,26 Moreover, Eck et al.27 reported that reoperations in breast-conserving therapy patients were an outlier compared with other general surgical operations. Using the National Surgical Quality Improvement Program (NSQIP) database, they demonstrated an observed/expected morbidity ratio of 7.75, with 89 % of morbidity attributed to reoperation.27 These reoperations with occasional readmissions add psychological, emotional, and financial stress to the patient encounter, as well as cost to the public healthcare burden. Prior reports of re-excision lumpectomy have interrogated the provider, patient, tumor, histologic, surgical technique, and pathologic processing factors associated with margin status and RELR.10,26–33 For example, the Mayo Clinic reports a low RELR of less than 15 % with a methodology of radioisotope localization techniques for non-palpable cancers, along with routine use of intraoperative frozen section margin assessment.29,30 A recent metaanalysis also supports use of immediate intraoperative margin assessment to lower RELR.26 However, other institutions report low rates of re-excision without using either of these adjuncts.32,34 In the absence of randomized controlled trials (RCT) it is unclear which factors have the most influence on margin status and which factors and resources are transferable and available for national performance improvement. Consequently, there is lack of contemporary meaningful information to incorporate into any plan to reduce secondary breast operations. Aggregate RELR at the patient level was 21.6 % in this study. There was wide variability of performance. Rates varied by patient and surgeon characteristics. Re-excision rates were higher in Black and Hispanic patients and those greater than 35 years of age. Rates were lowest among surgeons with solo practice type, with more years in practice, and higher practice volume. Re-excision rates did not differ by patient insurance type, US geographic region, surgeon proportion of practice in breast surgery, surgeon use of ultrasound, or Physician Quality Reporting System (PQRS) enrollment. A critical analysis of the cause of the differences described above is beyond the scope of this study but these results add to the existing literature documenting variability of performance, and emphasize the importance of including these characteristics into the data fields for future margin research. The most common cause of re-excision was a positive margin, accounting for nearly 50 % of all re-excisions (see Table 1). The next most common reasons for re-excision were a negative close margin, \1 mm in 34 % and 1–2 mm in 7 %. This contemporary data corroborates prior survey research data that concluded that many re-excisions are performed for close but negative margins.15–17
3188
J. Landercasper et al.
TABLE 2 Univariate analysis of patient and provider factors associated with re-excision lumpectomy rates Total
N
Re-excision rate (%) (95 % CI)
\35
147
11
7.48 (3.23–11.73)
35–44
548
96
17.52 (14.34–20.7)
45–54
1,449
239
16.49 (14.58–18.4)
55–64
2,000
327
16.35 (14.73–17.97)
65–74
2,080
353
16.96 (15.35–18.57)
C75
1,292
175
13.53 (11.66–15.4)
p value
Patient factors Age group (years) 0.0053
Race White (non-Hispanics)
6,083
939
15.44 (14.53–16.35)
Black Asian/Pacific Islander
715 278
127 41
17.76 (14.96–20.56) 14.74 (10.58–18.92)
Hispanics
330
70
21.21 (16.8–25.62)
0.018
Insurance Commercial
3,671
613
16.70 (15.49–17.91)
Medicare
2,595
380
14.64 (13.28–16)
Medicaid
327
45
13.76 (10.03–17.49)
Tricare
60
10
16.67 (7.24–26.1)
Uninsured
70
14
Unknown/Missing
793
139
17.53 (14.88–20.18)
Academic
558
117
20.97 (17.59–24.35)
Group (private)
3,351
570
17.01 (15.74–18.28)
Hospital-employed
1,760
264
15 (13.33–16.67)
Solo (private)
1,847
250
13.54 (11.98–15.1)
Years of practice B10
0.1383
20 (10.63–29.37)
Physician factors Type of practice facility
2,220
392
17.66 (16.07–19.25)
11–20
2,818
478
16.96 (18.48–21.44)
21–30
2,081
300
14.42 (12.91–15.93)
C30
397
31
7.81 (5.17–10.45)
Midwest
1,720
293
17.03 (15.25–18.81)
Northeast
2,393
387
16.17 (14.69–17.65)
Northwest
516
79
15.31 (12.2–18.42)
Southeast
1,400
220
15.71 (13.8–17.62)
Southwest
1,419
214
15.08 (13.22–16.94)
68
8
11.76 (4.95–18.57)
[150,000
5,310
856
16.12 (15.13–17.11)
50,000–150,000
1,816
315
17.35 (15.61–19.09)
390
30
7.69 (5.05–10.33)
\25
54
14
25.93 (14.24–37.62)
25–50
434
62
14.29 (11–17.58)
[ 50
1,518
236
15.55 (13.73–17.37)
100
5,510
889
16.13 (15.16–17.1)
\0.001
\0.001
Practice location
International
0.6246
Surgeon’s community population
\50,000 Practice proportion breast (%)
\0.001
0.1566
Causes of Re-Excision Lumpectomy in Breast Cancer
3189
TABLE 2 continued Total
N
Re-excision rate (%) (95 % CI)
Yes
6,488
1,051
16.20 (15.3–17.1)
No
1,028
150
14.59 (12.43–16.75)
Yes
353
2,348
15.03 (13.58–16.48)
No
848
5,168
16.41 (15.4–17.42)
p value
If surgeon performed ultrasound 0.1911
If surgeon performed stereotactic biopsy 0.1317
If surgeon was enrolled in a PQRS quality program Yes
3,340
531
15.90 (14.66–17.14)
No
4,176
670
16.04 (14.93–17.15)
0.8640
If surgeon has participated in any quality program Yes
5,543
842
15.19 (14.25–16.13)
No
1,973
359
18.20 (16.5–19.9)
0.0017
CI confidence interval, PQRS physician quality reporting system
The findings that many re-excisions occur in patients with negative but close margins is timely given the recent creation of position statements by the ASBrS, the Society for Surgical Oncology (SSO) and the American Society of Therapeutic Radiology (ASTRO).35,36 An updated metaanalysis describing the relationship between margin width and breast cancer recurrence served as the primary evidence to support the SSO–ASTRO position statement.4 Randomized trial data does not exist to guide the decision of re-excision for close final margins. The existing metaanalysis is the strongest evidence available and therefore has been endorsed by the ASBrS and multiple other organizations. Both position statements provide opportunity for lowering national RELR by recommending that routine re-excision not be performed for negative but close margins. Reasons for re-excision after initial lumpectomy still exist that are not addressed by the ASBrS and SSO– ASTRO consensus statements, but recognition that most patients with negative but close margins do not benefit from re-excision should reduce variability of RELR and lower national RELR . The patient and tumor factors influencing margin status cannot usually be changed to reduce re-excision. Therefore, reduction will require better optimization of preoperative patient evaluation for extent of disease, advancements in technical aspects of the operation, and, armed with the most common causes of re-excision identified in this study, education to cease routine re-excision of negative but close margins. The possibility of prospective RCT also remains open to determine optimal management of close margins in biologic subgroups of concern (young patients, triple-negative tumor subtypes). Additional steps to address the variability of RELR are to advance beyond observational studies of re-excision variability to
comparative effectiveness studies and RCT that investigate different methods for achieving negative margins and to investigate novel methods of improving surgical technical performance, such as real-time optical margin assessment, as well as ‘video review’ and ‘coaching’.28,37,38 Linking operative technique, as judged by surgeon peers, to outcomes has recently been reported.37 Weaknesses of this study include the following. Tumor characteristics known to contribute to margin statues were not audited. Omission of these factors for risk adjustment reduces our discriminatory performance; however, it is unlikely that variation based on these would have been large enough to account for the patient and surgeon factors associated with re-excision rates. Another weakness of the study is that consecutive patient entry was not required for surgeon participants to enter patients into MasterySM, with the exception of surgeons enrolled in PQRS. Surgeons enrolled in PQRS did attest to submit consecutive cases and were subject to auditing. There was no statistical difference in RELR between PQRS and non-PQRS participants (Table 2). The major strength of this study is that it links prior survey data on reasons for re-excision with actual care, corroborating the findings of survey research that postulated that many re-excisions are being performed for negative margins. This information will aid improvement projects to reduce secondary breast operations. The information collected on causes of re-excisions was contemporary and robust, with 1,575 re-excision encounters collected in 5 months, confirming the feasibility of using the ASBrS MasterySM program for vetting performance measures and assisting in performance improvement activities. Another strength is the level of information on causes of re-excision to include margin
3190
J. Landercasper et al. 100%
TABLE 3 Multivariate analysis of patient and provider factors associated with re-excision lumpectomy rates p value
Patient factorsa Age group (years) \35 35–44
Reference 2.64 (1.37–5.07)
45–54
2.44 (1.30–4.59)
55–64
2.43 (1.30–4.55)
65–74
2.58 (1.38–4.83)
C75
1.95 (1.03–3.69)
0.0059
Reference
Black
1.18 (0.96–1.45)
Asian/Pacific Islander
0.94 (0.67–1.32)
Hispanics
1.48 (1.13–1.95)
Insurance Physician factors
Not significant
60% 40% 20% 0%
ORIGINAL ARTICLE – BREAST ONCOLOGY
Reasons for Re-Excision After Lumpectomy for Breast Cancer: Insight from the American Society of Breast Surgeons MasterySM Database Jeffrey Landercasper, MD1,2, Eric Whitacre, MD3, Amy C. Degnim, MD4, and Mohammed Al-Hamadani, MBChB, MPH5 Department of Surgery, Gundersen Health System, La Crosse, WI; 2Norma J. Vinger Center for Breast Care, Gundersen Health System, La Crosse, WI; 3Breast Center of Southern Arizona, Tucson, AZ; 4Department of Surgery, Mayo Clinic, Rochester, MN; 5Department of Medical Research, Gundersen Health System, La Crosse, WI
1
ABSTRACT Background. There is marked variability of re-excision rates after initial lumpectomy for breast cancer. Reasons for re-excision and variability across surgeons have not been well documented. We hypothesized the American Society of Breast Surgeons (ASBrS) MasterySM Program can identify reasons for re-excision. Methods. Data from January 1 to 7 November 2013 were evaluated in the ASBrS MasterySM Program to determine re-excision lumpectomy rate (RELR). On 1 June 2013, a tool to track reasons for re-excision was developed. Variation in re-excision rate by surgeon and patient characteristics was performed by Chi square test and Fisher’s test for univariate analysis, then logistic regression with backwards elimination method for multivariate analysis. Results. For 6,725 patients undergoing initial lumpectomy for cancer, 328 surgeons reported 1,451 (21.6 %) patients had one or more re-excisions. The most common reasons for re-excision were ink positive margins in 783 (49.7 %), margin \1 mm (34.3 %), and margin 1–2 mm (7.2 %). By multivariate analysis, re-excision rates were lower in patients aged less than 35 years, with White (non-) Hispanic ethnicity, and, among surgeons in solo practice, more years in practice and higher-volume practice. Conclusion. Half of re-excisions after initial lumpectomy were performed for margins that are positive. Most of the
Ó Society of Surgical Oncology 2014 First Received: 14 April 2014; Published Online: 22 July 2014 J. Landercasper, MD e-mail: [email protected]
remainder were for negative close (\1–2 mm) margins. This information corroborates surgeon survey data regarding reasons for re-excision and provides proof of concept the MasterySM Program can capture surgical outcome data in real time, providing opportunity and a method for future performance improvement.
Re-excision of positive margins after initial lumpectomy for breast cancer is guideline-compliant and represents quality care.1–3 Breast cancer recurrence is reduced by reexcision when cancer is present at the lumpectomy surface;4–6 however, there is marked variation of national and international re-excision rates, which range from 5 to 70 %.7–10 The National Quality Forum (NQF) and other stakeholders in quality improvement would consider this degree of performance variation to be a potential marker for a gap in care quality because variation is part of their ‘importance criteria’ to justify development of a new quality measure (QM).11,12 Much discussion has occurred regarding whether re-excision lumpectomy rate (RELR) is a measure of quality.3,13,14 Briefly, the argument against naming RELR as a QM is that negative unintended consequences may occur, such as performing larger-volume lumpectomies (leading to worse cosmetic outcomes), omitting re-excision when needed in order to maintain a benchmark, or possibly increased local recurrence rates. ‘Risk aversion’ to lumpectomy, a surgeon fear of being penalized for initial positive margins, could also occur, resulting in a higher mastectomy rate if individual surgeon performance was tied to some measure of accountability, such as payment or patient steerage. Lastly, agreement on the definition of a ‘positive’ margin has been elusive, but would be necessary prior to using RELR as a QM.
3186
METHODS The ASBrS Mastery of Breast SurgerySM Program was developed in 2006 to help surgeons document their breast procedures, patient care, and to allow quality measurement.20,21 Surgeons can enter information on patient demographics, operations, and QM with immediate peerperformance comparison. After Institutional Review Board approval, de-identified MasterySM data from 1 January 2013 to 7 November 2013 were evaluated to determine RELR. On 1 June 2013, a dropdown pop-up menu was added to MasterySM to track reasons for re-excision. Overall, RELR was defined as the number of patients undergoing any type of re-excision (lumpectomy or any type of mastectomy) after initial lumpectomy divided by the number of patients having initial lumpectomy for cancer. We excluded patients undergoing bilateral operations and patients/procedures with incomplete or missing data. Patient age was determined by year of birth, not exact date of birth, to prevent patient identification and to comply with security requirements for de-identified data review. Statistical analysis was used to compare RELR stratified by patient and physician factors using univariate analysis (Chi square test, Fisher’s test). The RELR for comparisons and for Fig. 1 was defined as the number of patients undergoing re-excision lumpectomy, excluding mastectomy
100
80
Re-excision lumpectomy rate (%)
In contrast to the robust literature detailing variability of re-excision, the causes of re-excision after initial lumpectomy have not been well documented. Survey research suggests some re-excisions may be performed unnecessarily due to differences in surgeon and radiation oncology opinion regarding adequacy of margin width.15–17 The motivation to identify reasons for re-excision is to aid a performance improvement plan to reduce secondary operations. The NQF endorses efforts to reduce reoperations, specifying them as QM for many operation types.18 Furthermore, reoperations and readmissions are prioritized for improvement in the National Quality Strategy, a report commissioned under the Affordable Care Act.19 Reducing secondary operations has safety, cost, and patient-centered benefits. The recipe for a performance improvement plan and the design of clinical trials to reduce re-excisions will require information on causes of re-excision. The primary aim of this study was to identify contemporary causes of reexcision. We hypothesized the American Society of Breast Surgeons (ASBrS) MasterySM Program could identify reasons for re-excision. Secondary aims were to identify causes of variation of re-excision in ASBrS members and to search for patient and provider demographic factors associated with re-excision.
J. Landercasper et al.
60
40
20
0 Surgeon #1
Surgeon #183
Individual breast surgeons N= 183
FIG. 1 Variability of breast re-excision lumpectomy rate for 183 surgeons
as a re-excision type, divided by the number of patients having initial lumpectomy, because the database was deidentified for security for exact date of surgery preventing differentiation of a patient with initial mastectomy from a mastectomy for re-excision. Statistical significance was reported at an alpha level of 0.05 using p value with 95 % confidence interval (CI). For analysis regarding surgeons’ volume of procedures and re-excision rate, surgeons with ten or less lumpectomies were excluded, then Wilcoxon ranksum and Kruskal–Wallis tests were used. Multivariate analysis was performed with logistic regression with backwards elimination method (at an alpha level of 0.05) to control for patient and physician factors. Significant main predictors for outcome of re-excision were reported with adjusted odds ratio and 95 % CI. RESULTS Overall, 328 surgeons reported on 6,725 patients undergoing initial lumpectomy for cancer, with 1,451 (21.6 %) patients undergoing one or more re-excisions of any type (lumpectomy or mastectomy). Of these 6,725 patients, the re-excision type was lumpectomy in 1,131 (16.8 %). Re-excision rate ranged from 2 to 91 % in surgeons reporting more than ten cases (Fig. 1). In patients undergoing re-excision, the mean, median, and range of the number of re-excisions was 1.1 (±0.27), 1, and 1–4, respectively. Mean patient age was 61.9 years (±13.12) and ranged from 13 to 90. The causes of 1,575 re-excisions are detailed in Table 1. The most common causes of reexcision were ink positive margins in 783 (49.7 %), margin \1 mm (34.3 %), and margin 1–2 mm (7.2 %). In the univariate analysis for patient factors, we found that the re-excision rate was significantly highest in patients who were 35–44 years of age (17.5 %), and
Causes of Re-Excision Lumpectomy in Breast Cancer
3187
TABLE 1 Reasons for re-excision lumpectomy procedures Reason Ink positive margin
N
% 783 49.7
Margin \1 mm
540 34.3
Margin 1–2 mm
114 7.2
Post-lumpectomy imaging demonstrated evidence of residual disease
38 2.4
Prior surgery elsewhere, margin status uncertain
25 1.6
Margin C2 mm, but desire wider margin
16 1.0
Tumor board recommended wider margin Fragmented specimen, margin status uncertain
16 1.0 3 0.2
Radiation oncologist recommended wider margins Other Total procedures
2 0.1 48 3.1 1,575 100
Hispanics (21.2 %) [p-values were 0.005 and 0.018, respectively]. For physician factors, re-excision rate was lowest in solo practice (13.5 %), community population of less than 50 K (7.7 %), physician with more than 30 years of experience (7.8 %), and for those who were enrolled in any quality program (15.2 %). p-Values were \0.01 for all (Table 2). In multivariate analysis, we found similar trends to the univariate analysis (Table 3). Adjusted analysis for the mean rate of re-excision for surgeons was compared with the volume of their surgeries. Mean re-excision rate was highest for those surgeons who performed less than ten initial lumpectomies during the 11month study period. There was no statistical difference between those who performed 25–49, 50–99 and C100 initial lumpectomies (Fig. 2). DISCUSSION Reoperation for lumpectomy margin status averages 20– 40 % and ranges from 5 to 70 %.7–10 The European Union of Breast Cancer Specialists (EUSOMA) recommends measurement of secondary breast operations after either lumpectomy or mastectomy.22 Certified EUSOMA breast units audit ‘one-step surgical success rate’ for invasive cancer, endorse it as a measure of quality, then specify a minimum standard (80 %) and a target goal (90 %).22 Multiple operations compared with a single operation decrease value. Value has been defined as the ratio of performance and cost.23,24 Costs include both the financial and non-financial burdens of care endured by the patient.23 Higher-value care is more affordable for patients and affordability is one of the three aims of the US National Quality Strategy for healthcare.19,25 The goals of this study were to identify reasons for reexcision after initial lumpectomy for breast cancer and to identify the factors correlating with variability of re-
excision as a first step towards reducing secondary operations. There is no doubt that RELR varies widely by provider and institution.7,10,26 Moreover, Eck et al.27 reported that reoperations in breast-conserving therapy patients were an outlier compared with other general surgical operations. Using the National Surgical Quality Improvement Program (NSQIP) database, they demonstrated an observed/expected morbidity ratio of 7.75, with 89 % of morbidity attributed to reoperation.27 These reoperations with occasional readmissions add psychological, emotional, and financial stress to the patient encounter, as well as cost to the public healthcare burden. Prior reports of re-excision lumpectomy have interrogated the provider, patient, tumor, histologic, surgical technique, and pathologic processing factors associated with margin status and RELR.10,26–33 For example, the Mayo Clinic reports a low RELR of less than 15 % with a methodology of radioisotope localization techniques for non-palpable cancers, along with routine use of intraoperative frozen section margin assessment.29,30 A recent metaanalysis also supports use of immediate intraoperative margin assessment to lower RELR.26 However, other institutions report low rates of re-excision without using either of these adjuncts.32,34 In the absence of randomized controlled trials (RCT) it is unclear which factors have the most influence on margin status and which factors and resources are transferable and available for national performance improvement. Consequently, there is lack of contemporary meaningful information to incorporate into any plan to reduce secondary breast operations. Aggregate RELR at the patient level was 21.6 % in this study. There was wide variability of performance. Rates varied by patient and surgeon characteristics. Re-excision rates were higher in Black and Hispanic patients and those greater than 35 years of age. Rates were lowest among surgeons with solo practice type, with more years in practice, and higher practice volume. Re-excision rates did not differ by patient insurance type, US geographic region, surgeon proportion of practice in breast surgery, surgeon use of ultrasound, or Physician Quality Reporting System (PQRS) enrollment. A critical analysis of the cause of the differences described above is beyond the scope of this study but these results add to the existing literature documenting variability of performance, and emphasize the importance of including these characteristics into the data fields for future margin research. The most common cause of re-excision was a positive margin, accounting for nearly 50 % of all re-excisions (see Table 1). The next most common reasons for re-excision were a negative close margin, \1 mm in 34 % and 1–2 mm in 7 %. This contemporary data corroborates prior survey research data that concluded that many re-excisions are performed for close but negative margins.15–17
3188
J. Landercasper et al.
TABLE 2 Univariate analysis of patient and provider factors associated with re-excision lumpectomy rates Total
N
Re-excision rate (%) (95 % CI)
\35
147
11
7.48 (3.23–11.73)
35–44
548
96
17.52 (14.34–20.7)
45–54
1,449
239
16.49 (14.58–18.4)
55–64
2,000
327
16.35 (14.73–17.97)
65–74
2,080
353
16.96 (15.35–18.57)
C75
1,292
175
13.53 (11.66–15.4)
p value
Patient factors Age group (years) 0.0053
Race White (non-Hispanics)
6,083
939
15.44 (14.53–16.35)
Black Asian/Pacific Islander
715 278
127 41
17.76 (14.96–20.56) 14.74 (10.58–18.92)
Hispanics
330
70
21.21 (16.8–25.62)
0.018
Insurance Commercial
3,671
613
16.70 (15.49–17.91)
Medicare
2,595
380
14.64 (13.28–16)
Medicaid
327
45
13.76 (10.03–17.49)
Tricare
60
10
16.67 (7.24–26.1)
Uninsured
70
14
Unknown/Missing
793
139
17.53 (14.88–20.18)
Academic
558
117
20.97 (17.59–24.35)
Group (private)
3,351
570
17.01 (15.74–18.28)
Hospital-employed
1,760
264
15 (13.33–16.67)
Solo (private)
1,847
250
13.54 (11.98–15.1)
Years of practice B10
0.1383
20 (10.63–29.37)
Physician factors Type of practice facility
2,220
392
17.66 (16.07–19.25)
11–20
2,818
478
16.96 (18.48–21.44)
21–30
2,081
300
14.42 (12.91–15.93)
C30
397
31
7.81 (5.17–10.45)
Midwest
1,720
293
17.03 (15.25–18.81)
Northeast
2,393
387
16.17 (14.69–17.65)
Northwest
516
79
15.31 (12.2–18.42)
Southeast
1,400
220
15.71 (13.8–17.62)
Southwest
1,419
214
15.08 (13.22–16.94)
68
8
11.76 (4.95–18.57)
[150,000
5,310
856
16.12 (15.13–17.11)
50,000–150,000
1,816
315
17.35 (15.61–19.09)
390
30
7.69 (5.05–10.33)
\25
54
14
25.93 (14.24–37.62)
25–50
434
62
14.29 (11–17.58)
[ 50
1,518
236
15.55 (13.73–17.37)
100
5,510
889
16.13 (15.16–17.1)
\0.001
\0.001
Practice location
International
0.6246
Surgeon’s community population
\50,000 Practice proportion breast (%)
\0.001
0.1566
Causes of Re-Excision Lumpectomy in Breast Cancer
3189
TABLE 2 continued Total
N
Re-excision rate (%) (95 % CI)
Yes
6,488
1,051
16.20 (15.3–17.1)
No
1,028
150
14.59 (12.43–16.75)
Yes
353
2,348
15.03 (13.58–16.48)
No
848
5,168
16.41 (15.4–17.42)
p value
If surgeon performed ultrasound 0.1911
If surgeon performed stereotactic biopsy 0.1317
If surgeon was enrolled in a PQRS quality program Yes
3,340
531
15.90 (14.66–17.14)
No
4,176
670
16.04 (14.93–17.15)
0.8640
If surgeon has participated in any quality program Yes
5,543
842
15.19 (14.25–16.13)
No
1,973
359
18.20 (16.5–19.9)
0.0017
CI confidence interval, PQRS physician quality reporting system
The findings that many re-excisions occur in patients with negative but close margins is timely given the recent creation of position statements by the ASBrS, the Society for Surgical Oncology (SSO) and the American Society of Therapeutic Radiology (ASTRO).35,36 An updated metaanalysis describing the relationship between margin width and breast cancer recurrence served as the primary evidence to support the SSO–ASTRO position statement.4 Randomized trial data does not exist to guide the decision of re-excision for close final margins. The existing metaanalysis is the strongest evidence available and therefore has been endorsed by the ASBrS and multiple other organizations. Both position statements provide opportunity for lowering national RELR by recommending that routine re-excision not be performed for negative but close margins. Reasons for re-excision after initial lumpectomy still exist that are not addressed by the ASBrS and SSO– ASTRO consensus statements, but recognition that most patients with negative but close margins do not benefit from re-excision should reduce variability of RELR and lower national RELR . The patient and tumor factors influencing margin status cannot usually be changed to reduce re-excision. Therefore, reduction will require better optimization of preoperative patient evaluation for extent of disease, advancements in technical aspects of the operation, and, armed with the most common causes of re-excision identified in this study, education to cease routine re-excision of negative but close margins. The possibility of prospective RCT also remains open to determine optimal management of close margins in biologic subgroups of concern (young patients, triple-negative tumor subtypes). Additional steps to address the variability of RELR are to advance beyond observational studies of re-excision variability to
comparative effectiveness studies and RCT that investigate different methods for achieving negative margins and to investigate novel methods of improving surgical technical performance, such as real-time optical margin assessment, as well as ‘video review’ and ‘coaching’.28,37,38 Linking operative technique, as judged by surgeon peers, to outcomes has recently been reported.37 Weaknesses of this study include the following. Tumor characteristics known to contribute to margin statues were not audited. Omission of these factors for risk adjustment reduces our discriminatory performance; however, it is unlikely that variation based on these would have been large enough to account for the patient and surgeon factors associated with re-excision rates. Another weakness of the study is that consecutive patient entry was not required for surgeon participants to enter patients into MasterySM, with the exception of surgeons enrolled in PQRS. Surgeons enrolled in PQRS did attest to submit consecutive cases and were subject to auditing. There was no statistical difference in RELR between PQRS and non-PQRS participants (Table 2). The major strength of this study is that it links prior survey data on reasons for re-excision with actual care, corroborating the findings of survey research that postulated that many re-excisions are being performed for negative margins. This information will aid improvement projects to reduce secondary breast operations. The information collected on causes of re-excisions was contemporary and robust, with 1,575 re-excision encounters collected in 5 months, confirming the feasibility of using the ASBrS MasterySM program for vetting performance measures and assisting in performance improvement activities. Another strength is the level of information on causes of re-excision to include margin
3190
J. Landercasper et al. 100%
TABLE 3 Multivariate analysis of patient and provider factors associated with re-excision lumpectomy rates p value
Patient factorsa Age group (years) \35 35–44
Reference 2.64 (1.37–5.07)
45–54
2.44 (1.30–4.59)
55–64
2.43 (1.30–4.55)
65–74
2.58 (1.38–4.83)
C75
1.95 (1.03–3.69)
0.0059
Reference
Black
1.18 (0.96–1.45)
Asian/Pacific Islander
0.94 (0.67–1.32)
Hispanics
1.48 (1.13–1.95)
Insurance Physician factors
Not significant
60% 40% 20% 0%
