The numbers shown within both graphs are numbers of total evaluations for the respective years.
eFigure 1. Procedure frequency by surgeon volume in mastectomy patients (A) and BCS patients (B)
eFigure 2. Rates of node assessment with mastectomy over time by low-volume surgeons (A), mid-volume surgeons (B), and high-volume surgeons (C) and with BCS over time by low-volume surgeons (D) mid-volume surgeons (E), and high-volume surgeons (F)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Coromilas EJ, Wright JD, Huang Y, et al. The Influence of Hospital and Surgeon Factors on the Prevalence of Axillary Lymph Node Evaluation in Ductal Carcinoma In Situ. JAMA Oncol. 2015;1(3):323–332. doi:10.1001/jamaoncol.2015.0389
Although axillary lymph node evaluation is standard of care in the surgical management of invasive breast cancer, a benefit has not been demonstrated in ductal carcinoma in situ (DCIS). Despite uncertainty regarding the efficacy, axillary evaluation is often performed in women with DCIS.
To determine the incidence of axillary evaluation in women with DCIS and identify clinical, hospital, and surgeon-related factors associated with axillary evaluation.
Design, Setting, and Participants
Cross-sectional analysis conducted from January 2006 through December 2012 of medical records contained in the Perspective database for women with DCIS who underwent breast-conserving surgery (BCS) or mastectomy. A total of 35 591 women aged 18 to 90 years were included in the analysis.
Main Outcomes and Measures
Receipt or nonreceipt of surgical axillary evaluation, categorized as sentinel lymph node biopsy (SLNB), axillary lymph node dissection (ALND), or none. Analyses were stratified by surgery type, and multivariable regression analysis was used to identify factors associated with axillary evaluation.
Of women identified with DCIS, 26 580 (74.7%) underwent BCS while 9011 (25.3%) underwent mastectomy; 17.7% undergoing BCS and 63.0% undergoing mastectomy had an axillary evaluation. Rates of axillary evaluation increased over time with mastectomy (2006, 56.6%; 2012, 67.4%) and were relatively stable with BCS (2006, 18.5%; 2012, 16.2%). Rates of ALND decreased in women undergoing mastectomy (2006, 20.0%; 2012, 10.7%) and BCS (2006, 1.2%; 2012, 0.3%), with increasing use of SLNB. In a multivariable analysis, hospital factors including nonteaching hospital (risk ratio [RR], 1.17; 95% CI, 1.05-1.30) and urban location (RR, 1.15; 95% CI, 1.03-1.29) influenced axillary evaluation with mastectomy. Surgeon volume was the most significant predictor of axillary evaluation among women undergoing BCS (mid vs low volume: RR, 0.87; 95% CI, 0.70-0.94; high vs low volume: RR, 0.54; 95% CI, 0.44-0.65).
Conclusions and Relevance
Despite guidelines recommending against axillary lymph node evaluation in women with DCIS undergoing BCS and uncertainty regarding its use with mastectomy, SLNB or ALND is performed frequently. Given the additional morbidity and cost of these procedures, alternative surgical approaches or prospective evaluation of the clinical benefit of axillary evaluation in women with DCIS is needed.
The widespread adoption of screening mammography has led to an increase in the diagnosis of ductal carcinoma in situ (DCIS). As a result, DCIS now makes up about 20% of all newly diagnosed breast cancers.1 The surgical treatment options of mastectomy and breast-conserving surgery (BCS) are similar for women with invasive cancer and DCIS.2 However, while lymph node evaluation is standard for women with invasive cancer, there is no demonstrated benefit for lymph node assessment in women with DCIS.3,4
The sentinel nodes are the first few lymph nodes into which a tumor drains. For women with invasive breast cancer, sentinel lymph node biopsy (SLNB) replaced initial full axillary lymph node dissection (ALND).5-9 In some patients, there is no benefit of ALND even when the SLNB is positive.10 However, there are no data suggesting benefit of axillary evaluation in women with DCIS. Despite this, some have suggested it may have value because axillary involvement can occur if there is a coexisting invasive tumor.11,12 Guidelines published both by the American Society of Clinical Oncology (ASCO)3 and the National Comprehensive Cancer Network (NCCN)13 recommended against axillary evaluation in women undergoing BCS. If invasive cancer were discovered, SLNB could be performed at a later time.3 Because a total mastectomy precludes future SLNB, the guidelines suggested that SLNB may be appropriate in some high-risk patients because axillary evaluation would be indicated subsequently if an invasive component was identified.3,13
The purpose of this study was to determine the prevalence of surgical axillary evaluation in women with DCIS and to identify the clinical, hospital, and surgeon-related factors that are associated with axillary node evaluation in women undergoing BCS and in women undergoing mastectomy.
Although axillary lymph node evaluation is standard of care in the surgical management of invasive breast cancer, a benefit has not been demonstrated in ductal carcinoma in situ (DCIS).
Among women with biopsy-proven DCIS, 17.7% of women with breast-conserving surgery (BCS) and 63% of patients undergoing mastectomy underwent lymph node evaluation.
The rate of axillary evaluation has been relatively stable over time in women undergoing BCS, but rates have been increasing in women undergoing mastectomy.
Surgeon volume was the most significant predictor of axillary evaluation with BCS, with higher-volume surgeons less likely to perform lymph node dissections than low-volume surgeons.
Though use of axillary evaluation in DCIS may be appropriate in some cases, the high rates of axillary evaluation indicate a need to demonstrate benefit.
Institutional review board exemption was obtained for this retrospective review of medical records from Columbia University.
We utilized the Perspective database (Premier Inc), a voluntary, fee-supported database originally developed to measure resource utilization and quality of care. Perspective is an all-payer database including data from over 600 acute-care hospitals throughout the United States, capturing about 15% of outpatient and inpatient hospitalizations.14 Each institution submits electronic updates on a quarterly basis, and the data undergo a rigorous quality-control process prior to use. In addition to patient demographics, disease characteristics, and procedures, the database collects information on all billed services. Hospitals within Perspective are more frequently urban, teaching centers, and located in the southern United States.
Our cohort included women aged 18 to 90 years with DCIS (International Classification of Diseases, Ninth Revision [ICD-9] 233.0) who underwent an excisional breast procedure from January 2006 through December 2012. Procedures included BCS (ICD-9 85.21, 85.22, or 85.23 or Current Procedural Terminology [CPT] 19302, 19125, 19126, or 19301), simple mastectomy (ICD-9 85.4, 85.41, 85.42, 85.43, or 85.44 or CPT 19303), total mastectomy with ALND (ICD-9 85.45, 85.46, 85.47, or 85.48 or CPT 19305, 19306, or 19307), and skin-sparing mastectomy (ICD-9 85.33, 85.34, 85.35, or 85.36 or CPT 19304). We ensured that each patient was only included in the analysis once by including only the first admission for which there were codes for both DCIS and a procedure. We excluded patients with any code for invasive breast cancer, as well as 564 patients who were missing data on surgeon characteristics.
Patients were categorized by the type of axillary evaluation that was performed: SLNB, full ALND, or no axillary node evaluation. Any patient with a code for radiopharmaceutical mapping (CPT 38792, 38900, 78195, 78800, or 78801) was defined as having an SLNB. Patients with codes specific for a full axillary dissection (ICD-9 40.3, 40.29, 40.5, or 40.51 or CPT 38740 or 38745) or who had a total mastectomy with ALND (ICD-9 85.45, 85.46, 85.47, or 85.48 or CPT 19305, 19306, or 19307) were defined as having ALND. Patients coded as having a lymphadenectomy (ICD-9 40.23 or CPT 38500 or 38525) or BCS or mastectomy with axillary node assessment (ICD-9 85.43 or 85.44 or CPT 19302) without mapping were classified as having an ALND. Patients without any of these codes were classified as having no axillary evaluation. For patients with multiple codes, a hierarchy was created, where a code for SLNB superseded ALND, with the assumption that a full dissection may have been performed after a positive lymph node finding.
We validated the data by performing a similar analysis with patients who underwent BCS or mastectomy with a diagnosis of invasive cancer. Within this cohort, our definitions yielded estimated rates of SLNB and ALND that were consistent with previously reported rates in the literature, 30% to 40% for ALND and 50% to 70% for SLNB.9,15-17
For each patient, we analyzed demographic data including age, race (white, black, Hispanic, other), marital status (married, single, or unknown), and insurance status (commercial, Medicare, Medicaid, uninsured, or unknown). Clinical data collected included the type of procedure (BCS or mastectomy), lymph node evaluation (SLNB, ALND, or none), year of procedure, and the presence of comorbid disease. To assess the prevalence of comorbid disease, we used the Charlson comorbidity index,18 defined using ICD-9 coding as reported by Deyo et al.19
The hospitals at which patients were treated were characterized by location (urban or rural), region of the country (Midwest, Northeast, South, or West), size (<400 beds, 400-600 beds, or >600 beds), and teaching status (teaching or nonteaching).
We determined the specialty of the physician who performed the procedure and categorized the physicians by whether they had training in surgical oncology.
For each surgeon and hospital, we determined the total number of procedures performed for patients with DCIS during the study period. Because not all physicians and hospitals contributed data for the entire study period, we calculated annualized procedure volumes by dividing the total number of procedures of a given surgeon or hospital by the number of years a given surgeon or hospital contributed at least 1 procedure to the database. The distribution of volumes was then inspected visually and cut points were selected to create 3 approximately equal tertiles of surgeon volume and hospital volume.20,21
The primary outcome of the study was receipt of surgical axillary evaluation, either by SLNB or full ALND.
Frequency distributions between categorical variables were compared using χ2 tests. The association between these variables and the outcome of interest was assessed using multivariable regression models that included patient, surgeon, and hospital characteristics to determine factors that predict receipt of axillary evaluation. We selected previously published clinical and demographic variables associated with receipt of lymph node evaluation. Analyses were then stratified by type of procedure (BCS or mastectomy). To accommodate variable cluster sizes with annualized surgeon volume, the cluster-weighted Generalized Estimating Equations with Poisson distribution was used, which accounts for cluster size through inverse weighting.22 We also analyzed the data using hierarchical modeling (the GLIMMIX model) and with surgeon volume as a continuous variable as a sensitivity analyses,23 which revealed similar results.
Hospital volume was removed from the multivariate analysis because it was correlated with surgeon volume (r=0.40; P < .001): low-volume surgeons were significantly more likely to be at low-volume hospitals, and high-volume surgeons were significantly more likely to be at high-volume hospitals.
Results are reported with risk ratios (RRs) and 95% confidence intervals. All analyses were performed with SAS software, version 9.4 (SAS Institute Inc).
We identified 35 591 women with DCIS who underwent BCS or mastectomy, of whom 25 205 (70.8%) had no evaluation of the axillary lymph nodes, and 10 386 (29.2%) had some form of axillary evaluation. Of those with an axillary evaluation, 8742 (84.2%) had SLNB, and 1644 (15.8%) had a full ALND.
The clinical and demographic characteristics for the full cohort are listed in Table 1 and Table 2. The mean (SD) age of patients at the time of the procedure in our study was 59.8 (12.5) years. Most patients were white (71.1%) and had commercial insurance (55.8%). Of the patients identified, 26 580 (74.7%) had BCS, while 9011 (25.3%) had a total mastectomy, of which most procedures (64.8%) were unilateral. Most patients were treated in small (<400 bed) hospitals (53.7%), with 18.5% treated at large (>600 bed) hospitals. Annualized hospital and surgeon volume were separated into tertiles: hospital volume ranged from 1.0 to 102.0 procedures per year, with a median of 26.7, while surgeon volume ranged from 1.0 to 23.8 procedures per year, with a median of 1.8. Most procedures were performed by general surgeons, with only 4.7% performed by a surgical oncologist.
The rate of axillary evaluation was 63.0% among women undergoing mastectomy, with 15.2% of patients undergoing full ALND and 47.8% having SLNB. Patients undergoing BCS had lower rates of axillary evaluation (17.7%), with 16.7% of patients undergoing SLNB and only 1.0% of patients having ALND (Figure 1). The rate of axillary evaluation increased over time among women undergoing mastectomy from 56.6% in 2006 to 67.4% in 2012, which is attributable to increasing rates of SLNB (from 36.5% in 2006 to 56.7% in 2012), while the rates of ALND decreased (20.0% in 2006 to 10.7% in 2012) (Figure 2). The rate of axillary evaluation was relatively stable in women undergoing BCS over the same period (18.5% in 2006 to 16.2% in 2012) with the large majority of these procedures being SLNBs. While ALND was performed at a rate of 1.2% in 2006, by 2012 only 0.3% of women undergoing BCS received ALND.
Low-volume surgeons performed two-thirds of mastectomies and only one-quarter of BCS procedures; most BCS procedures (41.9%) were performed by high-volume surgeons, while only 4.9% of mastectomies were performed by high-volume surgeons. Axillary evaluation with mastectomy was relatively stable regardless of surgeon volume, performed in 61% to 64% of patients (eFigure 1 in the Supplement). As surgeon volume increased, a higher proportion of these procedures were SLNBs rather than ALNDs (low-volume, 47.1%; medium volume, 48.3%; high-volume, 53.7%). However, rates of axillary evaluation decreased with increasing surgeon volume in women undergoing BCS (low-volume, 26.4%; medium-volume, 20.4%; high-volume, 10.4%). Medium- and high-volume surgeons performed fewer ALNDs (0.7% vs 1.9%). Over time, rates of axillary evaluation with mastectomy increased by low- and medium-volume surgeons, but rates of axillary evaluation were stable over time in patients undergoing BCS, regardless of surgeon volume (eFigure 2 in the Supplement).
A multivariate analysis in patients undergoing mastectomy revealed that year of procedure, Medicaid coverage, treatment at a nonteaching hospital, and population type (urban vs rural) were independent predictors of receipt of axillary evaluation (Table 3 and Table 4). Patients undergoing mastectomy were significantly more likely to receive an axillary evaluation after 2006. Patients with Medicaid insurance had a higher rate of axillary evaluation than patients with commercial insurance (RR, 1.17; 95% CI, 1.05-1.30). Patients treated at nonteaching hospitals had higher rates of axillary evaluation (RR, 1.13; 95% CI, 1.06-1.21), as did patients treated in an urban location (RR, 1.30; 95% CI, 1.09-1.55).
Among patients undergoing BCS, hospital and surgeon factors had a larger influence on receipt of axillary evaluation than patient demographics. Hispanic patients were more likely than whites to undergo axillary evaluation (RR, 1.32; 95% CI, 1.00-1.74), although the difference did not reach statistical significance. At nonteaching hospitals, BCS was also associated with higher rates of axillary evaluation (RR, 1.17; 95% CI, 1.03-1.33). Decreased axillary evaluation was seen with increasing surgeon volume (medium vs low volume: RR, 0.87; 95% CI, 0.78-0.98; high vs low volume: RR, 0.54; 95% CI, 0.45-0.64).
Given the correlation between hospital volume and surgeon volume, we conducted a sensitivity analysis stratified by hospital volume (Table 5). In all hospital size categories, a higher surgeon volume was associated with decreased use of axillary evaluation. A sensitivity analysis using volume as a continuous variable resulted in similar results in each of the analyses.
Among women with biopsy-proven DCIS, we found that 17.7% of women with BCS and 63.0% of patients undergoing mastectomy underwent lymph node evaluation between 2006 and 2012. The rate of axillary evaluation was relatively stable over this period in women undergoing BCS, but rates increased in women undergoing mastectomy, with increased use of SLNB. Surgeon volume was the most significant predictor of axillary evaluation with BCS, with higher-volume surgeons less likely to perform lymph node dissections than low-volume surgeons.
The 29.2% overall rate of axillary evaluation in women with DCIS is consistent with previously reported rates. Ponti et al24 found a 23.1% rate of axillary evaluation between 2004 and 2008, with about three-quarters of patients undergoing BCS. Using data from the SEER database (Surveillance, Epidemiology, and End Results), Porembka et al25 reported a similar 28% rate of axillary evaluation. Rates of axillary evaluation in women with DCIS from the 1990s through 1996 decreased, but from 1996 through 2002, that rate remained unchanged, around 21%.25-28 Our analysis revealed that from 2006 through 2012, rates of axillary evaluation increased in patients undergoing mastectomy and were relatively stable with BCS.
Almost one-fifth of women undergoing BCS received axillary evaluation despite recommendations against this practice by both NCCN13 and ASCO.3 With mastectomy, the guidelines are less clear. Almost two-thirds of patients had an axillary evaluation, and rates increased over time. Justification for axillary evaluation in women with DCIS is based on the risk of discovering invasive cancer on final pathology; although by definition DCIS is local disease, studies suggest that 10% to 30% of patients with DCIS on minimally invasive biopsy will be upstaged within the breast.4,29-31 Patients may want to avoid returning to the operating room for a second procedure, especially because performance of a mastectomy precludes future SLNB. The patients undergoing axillary evaluation may also have clinical characteristics concerning for the coexistence of invasive disease, which was not captured in our data set.
However, the incidence of axillary node metastasis in DCIS is low, with most estimates ranging from 0.5% to 3%.24,32-39 The yield of identifying axillary metastasis can be increased with serial sectioning and immunohistochemical analysis.12,31,40-43 The clinical significance of these metastases is unknown, with multiple studies demonstrating no prognostic significance of micrometastasis at long-term follow-up in women with invasive cancer.31,42,44,45 Even in women with invasive breast cancer, completion ALND after a positive SLNB does not lead to improved survival compared with patients receiving no further axillary-specific intervention.10
The increasing rate of SLNB likely reflects increased experience with the procedure by surgeons over time. The SLNB is known to reduce both short- and long-term surgical complications compared to ALND.46-48 However, some risks remain, with a reported incidence of 6% to 7% for lymphedema and 3% to 9% for paresthesias, compared with rates of 11% to 75% for lymphedema and 19% to 68% for paresthesias in patients undergoing ALND. In addition, although rare, anaphylactic reactions to the blue dye used in SLNB have also been reported.49 Some suggest that the addition of SLNB to total mastectomy does not change the complication rate, but procedure costs need to be considered as well.50
Better identification of women with DCIS who may be at higher risk of axillary metastasis may reduce the number of patients requiring axillary evaluation. Large, high-grade, and palpable DCIS lesions have an increased risk of occult malignancy.41-43 However, though these features correlate with invasive disease, they may not be similarly predictive of axillary metastasis in women with DCIS, and efforts thus far have been unsuccessful at reliably identifying at-risk patients.43,51
Receipt of axillary evaluation was influenced by hospital factors in patients undergoing mastectomy, and by both hospital and surgeon factors in women having BCS. Surgeon volume was the most significant predictor of axillary evaluation with BCS. High-volume surgeons were also more likely to perform SLNB over an ALND, which is consistent with several other studies that have shown the association between practice volume and guideline-based care.20,52,53
Though other large studies have investigated rates of axillary evaluation in DCIS, many used older time frames, were limited to older patients, and have not evaluated the influence of insurance, hospital, surgeon, and volume-related factors.24,25,27,28 A study in Ontario found that women with DCIS treated at low-volume hospitals were also more likely to undergo lymph node evaluation.27 Surgeon volume, surgeon specialty, and hospital characteristics were not predictive of axillary evaluation in that analysis.
There are several limitations to our study. While the Perspective database contains a large sample of women of all ages treated throughout the United States, the data include a relatively higher proportion of patients treated at small to mid-sized, nonteaching, urban facilities. In addition, surgical decision making cannot be captured with administrative data, and factors such as size, grade, and location of DCIS in the breast were not available.
We identified diagnoses based on ICD-9 billing codes and were unable to confirm these results directly with pathology reports; however, most hospitals report ICD-9 diagnosis codes for billing once the operative pathology is known. The CPT codes are more specific for procedures and commonly reported by physicians, but ICD-9 codes are more commonly reported at the hospital administrative level. The CPT codes may be underreported, but we found a 95% agreement between ICD-9 and CPT codes and were reassured that the results were consistent with previously reported values. It is possible that the use of CPT codes for radiologic tracer infusion as a proxy for SLNB and lack of information regarding number of node evaluated may have led to misclassification.
Another limitation is the fact that Perspective does not allow linkage of surgeon data across hospitals. This may have led to some misclassification of surgeon volume. We do not believe that this limitation significantly affected the accuracy of our findings because this approach has been used in other studies.20,21
Despite uncertainty regarding the clinical benefit of axillary evaluation in women with DCIS, we found that 17.7% of women undergoing BCS and 63% of women undergoing mastectomy had either an SLNB or ALND. Though use of axillary evaluation in DCIS may be appropriate in some cases, the high rates of axillary evaluation indicate that additional research is needed in this area. In addition to better predictive tools for axillary involvement, other surgical approaches should be evaluated, such as placing a marker in the node rather than removing it, thus allowing for sentinel node removal at a second operation should invasive cancer be identified on final pathology. Perhaps most importantly, additional prospective evaluation is needed to determine if there is a clinical benefit to axillary evaluation in women with DCIS.
Accepted for Publication: February 17, 2015.
Corresponding Author: Dawn L. Hershman, MD, MS, Columbia University Medical Center, 161 Fort Washington Ave, Room 1068, New York, NY 10032 (firstname.lastname@example.org).
Published Online: April 9, 2015. doi:10.1001/jamaoncol.2015.0389.
Author Contributions: Dr Hershman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Coromilas, Wright, Chen, Hershman.
Acquisition, analysis, or interpretation of data: Coromilas, Wright, Huang, Feldman, Neugut, Chen, Hershman.
Drafting of the manuscript: Coromilas, Wright, Hershman.
Critical revision of the manuscript for important intellectual content: Coromilas, Huang, Feldman, Neugut, Chen, Hershman.
Statistical analysis: Coromilas, Wright, Huang, Chen.
Obtained funding: Neugut, Hershman.
Administrative, technical, or material support: Feldman, Neugut, Hershman.
Study supervision: Hershman.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by a grant from the Breast Cancer Research Foundation (Dr Hershman).
Role of the Funder/Sponsor: The Breast Cancer Research Foundation had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Create a personal account or sign in to: