Five-year outcome of patients classified using the American Society for Radiation Oncology consensus statement guidelines for the application of accelerated partial breast irradiation

An analysis of patients treated on the American Society of Breast Surgeons MammoSite Registry Trial

Authors


Abstract

BACKGROUND:

The American Society for Radiation Oncology (ASTRO) consensus statement (CS) for the application of accelerated partial breast irradiation (APBI) was applied to patients who were treated with this technique on the American Society of Breast Surgeons MammoSite Registry Trial to determine potential differences in clinical outcome based on classification group.

METHODS:

Patients were classified based on the CS groups of “suitable,” “cautionary,” and “unsuitable.” Rates of ipsilateral breast tumor recurrence (IBTR), regional lymph node failure, distant metastases, disease-free survival, cause-specific survival, and overall survival were assessed.

RESULTS:

Of the 1449 cases who were treated, 1025 patients (71%) could be classified according to the CS groupings, including 419 patients (41%) who fit the “suitable” criteria, 430 patients (42%) who fit the “cautionary” criteria, and 176 patients (17%) who fit the “unsuitable” criteria. At a median follow-up of 53.5 months, the 5-year actuarial rates of IBTR for the “suitable,” “cautionary,” and “unsuitable” groups were 2.59%, 5.43%, and 5.28%, respectively (P = .1884). Univariate analysis of factors potentially associated with IBTR indicated that negative estrogen receptor status was the only variable associated with IBTR among patients with invasive breast cancer (odds ratio [OR], 4.01; P = .0003). Larger tumor size was associated with a greater risk of distant metastasis (OR, 3.05; P = .0001). Among patients with ductal carcinoma in situ, only age <50 years and close-positive margins were associated with IBTR (OR, 1.12 [P = .0079] and OR, 7.81 [P = .0131], respectively).

CONCLUSIONS:

The ASTRO CS groupings did not differentiate a subset of patients with a significantly worse rate of IBTR when they were treated with the MammoSite breast brachytherapy catheter to deliver APBI. Cancer 2010. © 2010 American Cancer Society.

Whole breast irradiation (WBI) has become the standard of care after breast-conserving surgery in patients with early stage breast cancer. However, patterns of failure in these patients suggest that limiting radiation therapy to the lumpectomy cavity (eg, accelerated partial breast irradiation [APBI]) may produce acceptable clinical outcomes in properly selected patients.1-4 Consequently, numerous multicenter, randomized clinical trials (including the National Surgical Adjuvant Breast and Bowel Project [NSABP] B-39/Radiation Therapy Oncology Group [RTOG]-0413 phase 3 trial) have been initiated or completed to compare the effectiveness and safety of APBI versus WBI.5-10 It has been widely acknowledged that APBI provides several potential advantages over WBI, including both decreased overall treatment times and reduced radiation dose delivery to normal breast tissue and adjacent organs. However, among several theoretical disadvantages to APBI, the most salient is the potential that occult foci of cancer elsewhere in the breast may remain untreated.

Despite this concern, the use of APBI outside the framework of a clinical trial has increased markedly.11 This trend has occurred simultaneously with the emergence of a growing body of evidence that WBI significantly improves long-term overall survival after breast-conserving surgery.12, 13 In light of these potentially conflicting observations, most breast cancer experts agree that conservative patient selection criteria for APBI should be used until long-term follow-up data from large, randomized phase 3 trials become available.

Historically, few studies have been published to help guide physicians in assessing which patients could safely be treated with APBI off-protocol and which were better suited to receive WBI. The American Society for Radiation Oncology (ASTRO) Health Services Research Committee responded to these concerns by convening a task force of experts in the field of breast cancer (the Task Force). The Task Force recently published a consensus statement (CS) addressing patient selection criteria and best practices for the application of APBI outside of a clinical trial.14, 15 These recommendations were based on the results from a systematic review of the APBI literature supplemented by the expert opinions of members of the Task Force.

The CS divided patients into 3 groups, those deemed “suitable,” “cautionary,” and “unsuitable” to receive APBI outside of a clinical trial based on particular patient, tumor, and treatment criteria (Table 1). The groupings (and their subsequent descriptor terms) were based primarily on an analysis of the characteristics of patients most frequently included in trials of APBI. It must be emphasized that the CS groupings were not based on data that identified subsets of patients with higher rates of ipsilateral breast tumor recurrence (IBTR) when treated with APBI. Cognizant of the ongoing assessment of the efficacy of APBI, the Task Force acknowledged that the CS would require frequent modifications to account for new research findings. The Task Force indicated that the CS was meant to serve not only as a guide to the use of APBI outside clinical trials but also as a framework for research into the optimal role of APBI in treating breast cancer.

Table 1. Criteria Defining “Suitable,” “Cautionary,” and “Unsuitable” Patients for Accelerated Partial Breast Irradiation According to the American Society for Radiation Oncology Consensus Statement
FactorSuitable: All Criteria RequiredCautionary: Any CriteriaUnsuitable: Any Criteria
CS DefinitionASBS Registry TrialCS DefinitionASBS Registry TrialCS DefinitionASBS Registry Trial
  1. CS indicates consensus statement; ASBS, American Society of Breast Surgeons; BRCA 1/2, breast cancer susceptibility genes 1 and 2; LVSI, lymphovascular space invasion; ER, estrogen receptor; SN Bx, sentinel lymph node biopsy; ALND, axillary lymph node dissection; DCIS, ductal carcinoma in situ; EIC, extensive intraductal component; NA, not applicable.

Age, y≥60≥6050-5950-59<50<50
BRCA1/2 mutationNot presentNot availableNANAPresentNot available
Size, cm≤2≤22.1-3.02.1-3.0>3.0>3.0
Tumor classificationT1T1T0 and T2T0 and T2T3-T4T3-T4
MulticentricityUnicentricNot availableNANAPresentNot available
Multifocality≤2 cmNot available≤3 cmNot available>3 cmNot available
GradeAnyAnyNANANANA
LVSINoNot availableLimitedNot availableExtensiveNot available
ER statusPositivePositiveNegativeNegativeNANA
Lymph node statusNegativeNegativeNANAPositivePositive
Lymph node surgerySN Bx or ALNDNot availableNANANone performedNot available
MarginsNegative (≥2 mm)Negative (≥2 mm)Close (<2 mm)Close (<2 mm)PositivePositive
HistologyInvasive ductal onlyInvasiveDCIS (≤3 cm) or invasive lobular (if 2.1-3.0 cm)DCIS (≤3 cm) or invasive (if 2.1-3.0 cm)DCIS >3 cmDCIS >3 cm
EICNoNoYes and tumor size ≤3 cmYes and tumor size ≤3 cmYes and tumor size >3 cmYes and tumor size >3 cm
Neoadjuvant therapyNot allowedNot allowedNANAUsedNA

In light of the above, we applied the ASTRO CS guidelines for the application of APBI to outcome data from patients who were treated on the American Society of Breast Surgeons (ASBS) MammoSite Registry Trial (the Registry Trial). This APBI trial was initiated by the manufacturer (Hologic, Inc., Bedford, Mass) after clearance of the MammoSite breast brachytherapy device by the United States Food and Drug Administration (FDA) for clinical use in May 2002. The goals and objectives of the Registry Trial were to prospectively, objectively, and systematically collect data on the clinical use of the MammoSite brachytherapy applicator, focusing in particular on efficacy, acute toxicity, cosmesis, and adherence to placement in appropriate patients. On November 17, 2003, the ASBS assumed primary management of the trial. In total, 1449 breasts in 1440 patients have been treated on the Registry Trial. To our knowledge, this is the largest prospectively followed cohort of patients treated with APBI, with published 5-year results. Consequently, an analysis of these data was undertaken to apply the CS groupings and to assess their significance and utility in guiding the use of APBI.

MATERIALS AND METHODS

Study Participants

After FDA approval of the MammoSite device for clinical use, 97 institutions participated in a registry trial that was designed to collect data on the use of the device to deliver APBI between May 4, 2002 and July 30, 2004. The details of patient enrollment criteria and data collection and management have been described in prior reports.16-18 In brief, recommended criteria for patient enrollment in the Registry Trial were based on guidelines published by the American Society of Brachytherapy on the use of APBI.19, 20 Additional technical guidelines were followed in the protocol to exclude patients with inadequate balloon-to-skin distance, excessive cavity size, poor balloon-cavity conformance, or asymmetry of the central cavity shaft. All centers that were trained in and clinically using the MammoSite device were offered and encouraged to participate in the Registry Trial at its inception. Eighteen percent of sites enrolled ≥30 patients, 27% enrolled between 10 and 29 patients, and 55% enrolled ≤10 patients. Patients could be enrolled in the trial at any time (before, during, or after treatment), although enrollment before treatment was strongly encouraged. Two full-service, independent contract research organizations, Synergos, Inc. (The Woodlands, Tex) and BioStat International, Inc. (Tampa, Fla) have provided support in data management and statistical analysis of the Registry Trial for the ASBS. Data entry and processing for the Registry Trial are ongoing; a data cutoff date of January 29, 2010 was used for the purposes of the current analysis. At that time, a total of 1449 breasts in 1440 patients had been treated with the MammoSite breast brachytherapy device.

ASTRO CS Groups

All patients with sufficient data were categorized into 1 of 3 ASTRO CS groups. The criteria for each of the 3 groupings are set forth in Table 1. It is noteworthy that the Registry Trial did not collect data prospectively on the following parameters (“missing characteristics”): multifocality, multicentricity, presence of lymph-vascular space invasion (LVSI), histology of invasive cancer, breast cancer susceptibility genes 1 and 2 (BRCA1/2) mutation, and type of lymph node surgery performed. Patients who were classified as “suitable” fulfilled all of the criteria of this ASTRO CS group, excluding an assessment for the “missing characteristics” outlined above. Patients who were classified as “cautionary” met at least 1 of the criteria outlined in the ASTRO CS, again, excluding an assessment of the “missing characteristics” detailed above. Patients were classified as “unsuitable” if they met at least 1 criterion of this ASTRO CS group; for this group alone, any omission of data used to define the grouping was allowed, because it would not potentially alter the patient's group classification.

Outcome Measures

An IBTR was defined as the recurrence of cancer in the treated breast before or at the time of regional lymph node failure (RNF) or distant metastasis (DM) and was confirmed pathologically. RNF included recurrence in an axillary, supraclavicular, or internal mammary lymph node. DM was defined as recurrence of disease outside of an IBTR or an RNF. Predictors of IBTR, RNF, and DM were assessed separately for patients with invasive breast cancer and those with ductal carcinoma in situ (DCIS). Disease-free survival (DFS) was defined as survival free of any event attributed to breast cancer (ie, local, regional, or distant recurrences). Cause-specific survival (CSS) was defined as survival free of death attributed to breast cancer. Overall survival (OS) was defined as survival free of death from any cause. All time intervals were calculated from the date of completion of radiation therapy.

Statistical Analysis

The estimated likelihood of events for IBTR, RNF, DM, DFS, CSS, and OS were calculated by using the Kaplan-Meier method, and the statistical significance of differences among CS groups were calculated using the log-rank test. The association of individual categorical variables (eg, margin status) and continuous variables (eg, patient age in years) contained in the CS groupings with clinical outcome were analyzed, and odds ratios (ORs) and 95% confidence intervals (CIs) were generated using a Cox proportional hazards model. All P-values ≤.05 were considered statistically significant, and all statistical tests were 2-tailed. Statistical analyses were performed using SAS version 8.0 (SAS Institute, Cary, NC).

RESULTS

Table 1 presents a comparison of the CS “suitable,” “cautionary,” and “unsuitable” groups and any areas of omission of these data in the Registry Trial. Again, it should be noted that the Registry Trial did not collect data prospectively on “missing characteristics.” No patients in the Registry Trial received neoadjuvant chemotherapy. Table 2 shows the distribution of variables that were included in the CS groups among patients in the Registry Trial. In total, 1025 of 1449 cases (70.7%) in the Registry Trial were able to be categorized into the ASTRO CS groupings (Table 3).

Table 2. Patients Categorized by the American Society for Radiation Oncology Consensus Statement Group and Divided by Invasive Breast Cancer and Ductal Carcinoma In Situ (n = 1025 Patients)
CharacteristicNo. of Patients (%)
Invasive Breast CancerDCIS
  1. DCIS indicates ductal carcinoma in situ; NA, not applicable; ER, estrogen receptor; EIC, extensive intraductal component.

Tumor classification
 T0NA34 (100)
 T164 (6.5)NA
 T2926 (93.4)NA
 Missing1 (0.1)0 (0)
Tumor size, cm
 ≤2929 (93.7)23 (67.7)
 >2 to ≤360 (6.1)0 (0)
 >32 (0.2)4 (11.8)
 Missing0 (0)7 (20.6)
ER status
 Positive830 (83.8)12 (35.3)
 Negative117 (11.8)3 (8.8)
 Missing44 (4.4)19 (55.9)
EIC
 Present54 (5.5)NA
 Absent931 (94.0)NA
 Missing6 (0.6)NA
Surgical margins
 Positive11 (1.1)2 (5.9)
 Close66 (6.7)6 (17.7)
 Negative914 (92.2)26 (76.5)
 Missing0 (0)0 (0)
Lymph node status
 Positive905 (91.3)12 (35.3)
 Negative38 (3.8)0 (0)
 Missing48 (4.8)22 (64.7)
All parameters above included895 (90.3)5 (14.7)
Any parameter above excluded96 (9.7)29 (85.3)
Table 3. Study Population Versus the American Society for Radiation Oncology Consensus Statement Group
ASTRO CS GroupAll Patients: No. (%)
  1. ASTRO CS indicates American Society for Radiation Oncology consensus statement.

Suitable419 (40.9)
Cautionary430 (42.0)
Unsuitable176 (17.2)
Total categorized1025
Unknown424 (29.3)
Total study population1449

Suitable

Of the 1025 patients who were classified by the CS groupings, 419 patients (40.9%) had all of the “suitable” characteristics that were collected by the Registry Trial (Table 4). This excluded a determination regarding the “missing characteristics” noted above (see ASTRO CS Groups, above). Because of this limitation, it is possible that patients in the “suitable” group may have been misclassified and, indeed, would have been categorized as “cautionary” or “unsuitable” if all ASTRO CS factors had been available for review.

Table 4. Breakdown of Patient Characteristics Among Patients Classified by the American Society for Radiation Oncology Consensus Statement Group
CharacteristicASTRO CS Group: No. of Patients (%)
Suitable, n = 419Cautionary, n = 430Unsuitable, n = 176
  1. ASTRO CS indicates American Society for Radiation Oncology consensus statement; DCIS, ductal carcinoma in situ; ER, estrogen receptor; EIC, extensive intraductal component.

Age, y
 ≥60419 (100)188 (43.7)32 (18.2)
 50-590 (0)242 (56.3)14 (8.0)
 <500 (0)0 (0)130 (73.9)
Tumor classification
 T1419 (100)372 (86.5)135 (76.7)
 T20 (0)49 (11.4)15 (8.5)
 DCIS ≤3 cm0 (0)8 (1.9)15 (8.5)
 DCIS >3 cm0 (0)0 (0)4 (2.3)
 DCIS of unknown size0 (0)0 (0)7 (4.0)
 Missing0 (0)1 (0.2)0 (0)
Tumor size, cm
 ≤2419 (100)382 (88.8)151 (85.8)
 >2 to ≤30 (0)48 (11.2)12 (6.8)
 >30 (0)0 (0)6 (3.4)
 Missing0 (0)0 (0)7 (4.0)
ER status
 Positive419 (100)326 (75.8)97 (55.1)
 Negative0 (0)104 (24.2)16 (9.1)
 Missing0 (0)0 (0)63 (35.8)
EIC
 Present0 (0)47 (10.9)11 (6.3)
 Absent419 (100)383 (89.1)148 (84.1)
 Missing0 (0)0 (0)17 (9.7)
Surgical margins
 Positive0 (0)0 (0)13 (7.4)
 Close0 (0)53 (12.3)19 (10.8)
 Negative419 (100)377 (87.7)144 (81.8)
 Missing0 (0)0 (0)0 (0)
Lymph node status
 Positive0 (0)0 (0)38 (21.6)
 Negative419 (100)378 (87.9)120 (68.2)
 Missing0 (0)52 (12.1)18 (10.2)

Cautionary

Four hundred thirty patients (42%) had at least 1 of the “cautionary” characteristics (Table 4). The distribution of characteristics that placed them into this CS group was as follows: 1) ages 50 to 59 years (n = 242; 56.3%), 2) tumor size from 2.1 cm to 3.0 cm (n = 48; 11.2%), 3) negative estrogen receptor (ER) status (n = 104; 24.2%), 4) margins negative but <2 mm (close; n = 53; 12.3%), 5) the presence of an extensive intraductal component with tumor size <3 cm (n = 47; 10.9%), and 6) pure DCIS ≤3 cm (n = 8; 1.9%). Because the Registry Trial did not collect data on the “missing characteristics,” it is possible that some of these patients would have been categorized as “unsuitable” according to the full CS criteria.

Unsuitable

One hundred seventy-six patients (17.2%) had at least 1 of the “unsuitable” characteristics (Table 4). The distribution of characteristics that placed them into this CS group was as follows: 1) age <50 years (n = 130; 73.9%), 2) tumor size >3 cm (n = 6; 3.4%), 3) positive margins (tumor touching ink: n = 13; 7.4%), 4) positive lymph nodes (n = 38; 21.6%), and 5) DCIS >3 cm (n = 4; 2.3%).

Table 5 presents the 5-year actuarial rates of IBTR, RNF, DM, DFS, CSS, and OS by CS groups. At a median follow-up of 53.5 months, applying the CS groupings resulted in statistically significant differences only in the rate of DM (P = .0392).

Table 5. Five-Year Clinical Outcomes Versus the American Society for Radiation Oncology Consensus Statement (CS) Group for Patients Categorized by CS Group (n = 1025)
OutcomeASTRO CS Group, % 
SuitableCautionaryUnsuitableP
  1. ASTRO indicates American Society for Radiation Oncology.

Ipsilateral breast tumor recurrence2.595.435.28.1884
Regional lymph node failure0.581.390.63.5075
Distant metastases0.874.443.01.0392
Disease-free survival86.483.686.0.8859
Cause-specific survival98.898.698.1.8566
Overall survival91.291.693.0.4300

An analysis was conducted on the 991 patients who had invasive breast cancer in the Registry Trial categorized by the CS groupings to assess for predictors of IBTR, RNF, and DM. Univariate analysis indicated that the only significant predictor of IBTR among these patients was negative ER status (OR, 4.01; 95% CI, 1.87-8.57; P = .0003) (Table 6). It is noteworthy that analyses by CS grouping of “suitable” versus “cautionary”/“unsuitable,” “suitable”/“cautionary” versus “unsuitable,” and “suitable”/“unsuitable” versus “cautionary” were not associated with IBTR. A 1-mm increase in tumor size resulted in an increase >3-fold in the risk of DM. These results were similar to those produced when a univariate analysis was performed in a subgroup of 870 patients who had invasive breast cancer for whom all of the variables that were included in the CS groupings were available among the variables that were collected consistently in the Registry Trial, again excluding the “missing characteristics” (data not shown).

Table 6. Univariate Analysis of Clinical Outcomes for Patients With Invasive Breast Cancer Categorized by the American Society for Radiation Oncology Consensus Statement Group (n = 991)a
VariableIBTRRNFDM
OR [95% CI]POR [95% CI]POR [95% CI]P
  • IBTR indicates ipsilateral breast tumor recurrence; RNF, regional lymph node failure; DM, distant metastases; OR, odds ratio; CI, confidence interval; EIC, extensive intraductal component; ER, estrogen receptor; ASTRO CS; American Society for Radiation Oncology consensus statement.

  • a

    Estimates that were unstable because of small sample size are not displayed.

Age (<50 y vs ≥50 y)0.90 [0.27-2.97].86770.90 [0.21-3.87].8855
Age (continuous)1.00 [0.97-1.03].87811.01 [0.95-1.08].73500.98 [0.94-1.02].2872
Tumor stage (T1 vs T2)2.56 [0.31-21.3].38322.59 [0.76-8.84].1288
Tumor size (≤2 cm vs >2 cm)0.37 [0.05-3.11].36280.37 [0.11-1.27].1150
Tumor size (continuous)0.81 [0.41-1.59].53431.80 [0.55-5.85].33153.05 [1.75-5.29].0001
EIC (present vs absent)2.61 [0.91-7.45].0740
ER status (negative vs positive)4.01 [1.87-8.57].00032.89 [0.56-14.9].20490.15 [0.06-0.37]<.0001
Surgical margins (≤2 mm vs >2 mm)1.21 [0.37-3.99].75041.98 [0.24-16.5].52600.62 [0.08-4.65].6440
Lymph node status (positive vs negative)1.10 [0.15-8.10].92531.33 [0.18-9.94].7823
ASTRO CS group
 “Suitable” vs “cautionary”/“unsuitable”0.64 [0.30-1.35]0.23790.53 [0.10-2.73].44770.34 [0.11-1.01].0525
 “Suitable”/“cautionary” vs “unsuitable”1.10 [0.38-3.14].86101.01 [0.30-3.45].9864
 “Suitable”/“unsuitable” vs “cautionary”0.63 [0.31-1.27].19360.30 [0.06-1.56].15380.40 [0.16-0.99].0480

A separate analysis was performed on patients in the Registry Trial who had DCIS. Because only 34 patients with DCIS could be categorized by CS grouping, a univariate analysis of the predictors of clinical outcomes that were included in the CS criteria was preformed on the larger group of all 194 patients with DCIS. For these patients, the only variables that were associated with IBTR on univariate analysis were age <50 years (OR, 1.115; 95% CI, 1.029-1.208; P = .0079) and close-positive margins, which were defined as ≤2 mm (OR, 7.813; 95% CI, 1.539-39.659; P = .0131) (Table 7). Limitations in the data available on this smaller group precluded making a conclusion regarding a possible association between CS grouping and IBTR.

Table 7. Univariate Analysis of Variables Associated With 5-Year Ipsilateral Breast Tumor Recurrence, Regional Lymph Node Failure, and Distant Metastases for Patients With Ductal Carcinoma In Situ (n = 194)a
VariableIBTRRNFDM
OR [95% CI]POR [95% CI]POR [95% CI]P
  • IBTR indicates ipsilateral breast tumor recurrence; RNF, regional lymph node failure; DM, distant metastases; OR, odds ratio; CI, confidence interval; ER, estrogen receptor.

  • a

    Estimates that were unstable because of small sample size are not displayed.

Age (continuous variable)0.953 [0.879-0.034].25070.748 [0.481-1.162].19570.750 [0.492-1.144].1819
Age at diagnosis (<50 y vs ≥50 y)1.115 [1.029-1.208].0079
Tumor size (continuous variable)0.728 [0.194-2.724].63690.967 [0.081-11.496].97850.945 [0.084-10.619].9634
ER status (negative vs positive)
Surgical margin status (≤2 mm vs >2 mm)7.813 [1.539-39.659].0131

DISCUSSION

In the current study, we observed that ASTRO CS groupings for the use of APBI off-protocol failed to differentiate between patients who may or may not be treated appropriately using the MammoSite breast brachytherapy catheter as the form of APBI. No statistically significant differences were demonstrated in the 5-year actuarial rates of IBTR between patients categorized in the “suitable,” “cautionary,” or “unsuitable” groups (2.59%, 5.43%, and 5.28%, respectively; P = .1884). A univariate analysis of factors potentially associated with IBTR indicated that negative ER status was the only variable associated with IBTR among patients with invasive breast cancer (OR, 4.01; P = .0003). In patients with DCIS, only age <50 years and close-positive margins were associated with IBTR (OR, 1.115 [P = .0079] and OR, 7.813 [P = .0131], respectively). Despite the significant limitations of the current study, we believe these data suggest that further refinement of the ASTRO CS groupings may be needed.

Development of the ASTRO CS on APBI

It is estimated that >50,000 women in the United States have been treated with the MammoSite breast brachytherapy catheter, a device that was designed for the delivery of APBI.21 Despite limited long-term data on the use of APBI, the implementation of this treatment to deliver adjuvant breast radiation therapy outside of a clinical trial has increased markedly.11 Other techniques of delivering APBI, including external-beam radiation, single-fraction radiation, interstitial brachytherapy, and newer intracavitary brachytherapy devices and intraoperative radiation units, also have been used to treat thousands of additional women.11 Most breast cancer experts agree that conservative patient selection criteria for APBI should be used until the results of phase 3 trials comparing APBI and WBI become available, particularly given the growing body of evidence that postlumpectomy WBI provides a benefit in long-term survival.12, 13

In response to these trends, the ASTRO Health Services Research Committee convened a task force of experts in the field of breast cancer to develop a consensus statement regarding patient selection criteria and best practices for the use of APBI outside the context of a clinical trial. The recommendations from this committee were based on the results from a systematic literature review supplemented by the expert opinions of the Task Force members. The objective of the CS is to guide clinicians in identifying patients who may be treated safely with APBI outside of a clinical trial. It is noteworthy that the Task Force acknowledged that there was a paucity of clinical and pathologic data used to determine which patients were deemed inappropriate to receive APBI. Consequently, the Task Force relied heavily on the expertise of its members in the knowledge of risk factors associated with IBTR after WBI to identify patients who were deemed “cautionary” and “unsuitable” to receive APBI outside the confines of a clinical trial.

It should be noted that the groupings selected by the ASTRO Task Force were formulated largely without the use of long-term clinical data on the efficacy of APBI and were not based specifically on comparisons of IBTR rates or other outcomes. Patients who are characterized as “cautionary” and “unsuitable,” in fact, may be appropriate candidates for APBI, as suggested by our current results. However, by necessity, the ASTRO Task Force included patients in the “suitable” category only if there were sufficient data to support the use of APBI off-protocol. Therefore, patients in the “cautionary” and “unsuitable” classification were considered acceptable candidates only for protocols that specifically addressed the efficacy of APBI. The ASTRO Task Force explicitly acknowledged the need for continued data collection, particularly among the “cautionary” and “unsuitable” groups, and they anticipated a need to periodically update and modify the current guidelines.

The ASTRO Task Force faced the burden of defining patients as “suitable” to receive APBI despite limited data from pathology and prospective clinical studies of patients with breast cancer to precisely define this subset.22 The identification of optimal patients to receive APBI necessitates the identification of subsets of patients at low risk of having clinically occult disease remote from the lumpectomy cavity. The Task Force, as a proxy for identifying these patients, relied on the common inclusion criteria and characteristics of patients enrolled in prospective, single-arm studies of APBI with a minimum follow-up of 4 years and a low reported risk of IBTR.14, 15 The members of the Task Force acknowledged that there may be additional groups of patients for whom APBI is suitable. However, at the time of publication, insufficient evidence was available to identify these groups.

It should be emphasized again that there are no contemporary data demonstrating that patients who receive APBI in either the “cautionary” or “unsuitable” group have significantly worse rates of IBTR, RNF, DM, DFS, CSS, or OS than patients in the “suitable” group. In fact, several small, recently published studies that included these “higher risk” patients who received APBI demonstrated acceptable and comparable rates of IBTR.23, 24 Thus, it remains both ethical and appropriate to randomize these eligible patients on the NSABP-B39/RTOG-0413 phase 3 trial. Indeed, accrual of a wide breadth of patients onto this trial is critical to allow for accurate conclusions on the noninferiority of APBI to WBI. The ASTRO Task Force strongly recommended that those patients who remain eligible for enrollment on the NSABP B-39/RTOG-0413 trial, patients aged <50 years or patients of any age with ER-negative or pathologic N1 tumors should be strongly encouraged to participate on this trial.5

Current Analysis/Study Limitations

In this article, we present results from the application of the ASTRO CS guidelines for APBI to the largest published cohort of patients who have received APBI to date. The ASBS MammoSite Registry Trial, as discussed above, contains prospectively collected data on 1449 breast cancers treated with APBI. Among this large cohort of patients, the ASTRO CS groupings failed to differentiate patients who may or may not be treated more appropriately with APBI.

The current study has the benefit of containing a large number of patients treated at multiple institutions, thereby allowing a broad application of the results. However, caution must be used in interpreting these findings. Only 71% of the entire group could be included in this analysis because of the lack of sufficient data to categorize all patients. Because of the retrospective nature of this analysis, the “suitable,” “cautionary,” and “unsuitable” groups used here encompassed the majority but not all of the characteristics outlined in the ASTRO CS (Table 1). Again, data on multifocality, multicentricity, LVSI, histology of invasive cancer, BRCA1/2 mutation, and type of lymph node surgery performed were not collected. Consequently, only the unsuitable patients clearly met all criteria of the CS groupings. To control for the potential misclassification of patients because of a lack of complete information on each patient, univariate analysis on the predictive ability of the CS groupings was performed comparing the “suitable” versus “cautionary”/“unsuitable,” “suitable”/“cautionary” versus “unsuitable,” and “suitable”/“unsuitable” versus “cautionary” groups, none of which were associated with IBTR. Despite these significant limitations, we believe the results presented here serve as a useful reference with which to judge the appropriateness of the CS groups and form the foundation for the development of more useful categories as more APBI data become available.

Risk Factors of Cancer Recurrence After APBI

We observed that ER status and tumor size were the only factors associated with clinical outcome among patients with invasive breast cancer. Other variables also may need to be explored as possible risk factors for the development of IBTR in the setting of APBI. It is possible that these risk factors may be different from those observed with traditional WBI. In addition, MammoSite delivery of APBI is 1 of multiple modalities for delivering APBI, and it will be useful to verify these findings among cohorts that received interstitial brachytherapy, external-beam irradiation, and intraoperative radiation.

Limited data have been published on potential risk factors associated with the development of IBTR in the setting of APBI. Vicini et al. recently reported long-term data (median follow-up, 11.1 years) demonstrating equivalent rates of IBTR among a matched pair of 199 patients who received APBI as interstitial brachytherapy versus 199 patients who received WBI.25 Those authors then applied the CS groupings to these same patients and provided an analysis similar to that in the current study. Among the patients who received APBI and WBI (taken together as 1 group), the CS groupings were associated with progressively worse rates of RNF, and a trend toward worse DM rates emerged. However, the CS groupings were not associated with IBTR in either the APBI cohort or the WBI cohort when they were analyzed both separately and together.

We hope that additional data like those described in this analysis and detailed above will emerge to help clarify important associations and guide any future changes to the ASTRO CS on APBI. Until then, conservative patient selection should continue to be used in treating patients with APBI off-protocol. Because many unsettled issues remain regarding APBI, we continue to encourage any patient interested in APBI to enroll in prospective clinical trials, and we particularly encourage those patients who meet current eligibility criteria for NSAPB-B39/RTOG-0413 to enroll in this high-priority national study.

In conclusion, the ASTRO CS groupings for the use of APBI off-protocol defined a subset of patients that can appropriately receive APBI because of a very low rate of IBTR at 5 years (the “suitable” CS group). However, the groupings failed to differentiate other patients who may or may not be treated more appropriately with WBI. These results await validation from other large APBI databases and suggest that further refinement of the ASTRO CS groupings may be needed. These findings also support the continued enrollment of “high-risk” patients in phase 3 trials, from which definitive data on the appropriateness of delivering APBI should emerge.

CONFLICT OF INTEREST DISCLOSURES

Martin Keisch is a consultant for Hologic, Inc. Maureen Lyden receives financial support from Hologic, Inc. for her statistical analysis.

Ancillary