MammoSite excision volume as a predictor for residual disease


  • Seth A. Kaufman M.D.,

    Corresponding author
    1. Department of Radiation Oncology, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, Massachusetts
    • Department of Radiation Oncology, Box 359, Tufts-New England Medical Center, Boston, MA 02111
    Search for more papers by this author
    • Fax: (617) 636-6131

  • Thomas A. DiPetrillo M.D.,

    1. Department of Radiation Oncology, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, Massachusetts
    2. Department of Radiation Oncology, Rhode Island Hospital, Brown University School of Medicine, Providence, Rhode Island
    Search for more papers by this author
  • Robin Ruthazer M.P.H.,

    1. Department of Clinical Care Research, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, Massachusetts
    Search for more papers by this author
  • David E. Wazer M.D.

    1. Department of Radiation Oncology, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, Massachusetts
    2. Department of Radiation Oncology, Rhode Island Hospital, Brown University School of Medicine, Providence, Rhode Island
    Search for more papers by this author

  • Presented in poster form at the 46th Annual Meeting of the American Society for Therapeutic Radiology and Oncology, Atlanta, Georgia, October 3–7, 2004.



The MammoSite catheter is a brachytherapy device used for accelerated partial breast irradiation. Currently, it is available as 2 spherically shaped balloons meant to fill 70-cc and 120-cc cavity volumes. This study was designed to define the relation of these excision volumes to the likelihood of microscopically detectable, residual disease based on tumor size, margin status, patient age, and histology.


The study data base was comprised of 531 patients with Stage 0, I, and II breast carcinoma (using American Joint Committee on Cancer staging criteria) who received breast-conserving therapy and underwent surgical reexcision. Patients in the data base were stratified based on the volume of their initial excision: ≤ 70 cc versus > 70 cc and ≤ 120 cc versus > 120 cc.


Surgical margin size was found to be a strong predictor of residual disease both for patients with smaller excision volumes (P = 0.0014) and patients with larger excision volumes (P = 0.0003); histology (extensive intraductal component [EIC] or pure ductal carcinoma in situ [DCIS]) also was a strong predictor. Tumor size was significant only for the larger volume group (P = 0.029). On multivariate analysis, only histology and initial margin status were significant correlates with residual disease. The adjusted odds ratio for residual disease with pure DCIS was 0.79, and the adjusted odds ratio for invasive ductal or lobular carcinoma (IDC/ILC) without EIC was 0.44 relative to IDC/ILC with EIC (P = 0.008). The adjusted odds ratio for residual disease with a positive initial margin versus a negative initial margin was 2.65 (P ≤ 0.0001).


For excisions amenable to use of the MammoSite catheter, a margin ≥ 1.0 mm appeared to afford at most a 35% risk of microscopically detectable residual tumor. Evidence of EIC on excision of IDC/ILC connoted a significantly higher risk. Age did not appear to be predictive for residual disease. Cancer 2005. © 2005 American Cancer Society.

Accelerated partial breast irradiation (APBI) has been investigated as an alternate means of delivering radiation therapy in the setting of breast conservation. Methods have included brachytherapy in the form of multiple rigid interstitial catheters, three-dimensional conformal external beam irradiation (including intensity-modulated radiotherapy), and intraoperative irradiation. Because only a portion of the breast tissue is treated, the toxicity to the surrounding tissue potentially may be reduced.

A balloon-tipped intracavitary brachytherapy device, the MammoSite RTS (Proxima Therapeutics, Inc., Alpharetta, GA) has been introduced recently for use in APBI. This applicator uses a saline-filled balloon tip to occupy the lumpectomy cavity. The catheter itself is placed transcutaneously, with the lumen aligned with the central axis of the balloon. A high-dose-rate brachytherapy source is placed at the center of the balloon through a remote afterloader to allow concentric dosing to the surrounding tissue. Hypofractionation permits a twice-daily fraction size of 340 centigrays prescribed to a 1 cm depth, up to a total dose of 34 grays. Therefore, the entire treatment course is completed within 1 week. A recent retrospective analysis showed that this prescription depth may be adequate to encompass residual disease in up to 90% of breast-conservation patients.1

The applicability of the MammoSite device is limited to a select group of patients. Both the American Brachytherapy Society (ABS)2, 3 and the American Society of Breast Surgeons (ASBS)4 have published criteria for APBI in terms of age, histology, tumor size, margin status, and lymph node involvement. In addition, use of the MammoSite catheter is feasible only if there is adequate separation of the cavity from the skin and chest wall and if the balloon fills the cavity completely and uniformly. The manufacturer has produced 2 spherical balloon sizes with 5 cm and 6 cm dimensions, which correspond to maximum fill volumes of 70 cc and 120 cc, respectively. Recently, the question has been raised regarding how many lumpectomy specimens fit this volume restriction.5 Another concern is that, for these resection volumes, what tumor margins are necessary to assure appropriately minimal risk of residual disease? Previous studies have established that surgical margin status is predictive for the probability of residual tumor for both invasive breast carcinoma6–11 and in situ breast carcinoma.12–15 It can be concluded from this that microscopic residual disease found adjacent to the resection cavity is a good surrogate for additional disease within the breast. We have performed an analysis of a data base consisting of women with early-stage breast carcinoma who underwent breast-conserving surgery and received whole breast irradiation over a 12-year period at our institution. Restricting our focus only to the patients who had excision volumes that meet the criteria for MammoSite placement, we have attempted to correlate clinical, histologic, and surgical parameters with the probability of finding residual tumor within the breast.


Patient Evaluation

Our invasive carcinoma data set consisted of all patients with Stage I–II breast carcinoma (according to the American Joint Committee on Cancer staging criteria) who underwent reexcision breast biopsies at our institution between 1983 and 1994 (n = 278 patients). Of those, 265 patients had data available for review. In addition, a total of 253 women underwent lumpectomy and reexcision for pure ductal carcinoma in situ (DCIS) at our institution between 1987 and 2000. Of these, 245 patients with available reexcision data were incorporated into the current analysis. Therefore, in total, 510 of 531 patients were included.

All excisions were performed with the objective of complete tumor removal with a minimum macroscopic normal tissue margin of 5 mm. Institutional guidelines dictated that the entire cavity be excised and that reexcision be performed for margins ≤ 2 mm, although the final treatment decision was at the discretion of the patient and physician. Specimen radiographs and postlumpectomy mammograms were performed as needed to ensure completion of the macroscopic total excision. Surgical specimens were oriented with marking sutures and immediately were delivered intact and unfixed to the pathologist. They were dried briefly, and the external surface was coated with India ink. Specimens were measured in 3 dimensions before multiple section levels (minimum, 10 section levels) were obtained with the intent of evaluating the deep, superficial, superior, inferior, medial, and lateral surfaces. Additional sections were obtained in areas where the tumor approached the surface of the specimen. No distinction was made between DCIS and invasive carcinoma in terms of defining the relative closeness of tumor to the specimen margin.

For each tumor, the histopathologic subtype was categorized broadly as invasive ductal carcinoma (IDC) with an extensive intraductal component (EIC), as defined by others9 (IDC with EIC), IDC without EIC, invasive lobular carcinoma (ILC), or DCIS. Samples of IDC and ILC were pooled for the purposes of the current analysis, resulting in a general invasive carcinoma (IDC/ILC) category stratified by EIC involvement. Tumor size was defined as the greatest tumor dimension measured pathologically. Tumor volume was calculated from the product of the greatest tumor dimension measured in three dimensions. The surgical margin status of the primary excision volume was defined as the closest surgical margin to the specimen edge, as measured with a micrometer. Surgical margin status was then scored as positive, 0.1–1.0 mm, or 1.1–2 mm.

Statistical Analysis

The unit of analysis was a surgically treated breast (n = 510). Clinical and therapeutic features were summarized using the mean, standard deviation, median, minimum, and maximum values for continuous variables and frequency distributions for nominal variables. Chi-square analysis was used to explore univariate correlations between histology (DCIS, IDC/ILC with EIC, IDC/ILC without EIC), tumor size (≤ 2 cm, > 2 cm), excision volume (dichotomized at 70 cc and at 120 cc), surgical margin status (positive, 0.1–1.0 mm, 1.1–2.0 mm, > 2.0 mm), and age (≤ 45 years, > 45 years) with the outcome of the presence of residual tumor. Factors with univariate associations of P < 0.10 were included in the building of a multivariable logistic regression model using a stepwise selection process. All analyses were performed using the SAS software system for Windows (version 8.2; SAS Institute Inc., Cary, NC).


The median age of the 510 patients included in the current study was 53 years (range, 25–90 years). The median initial volume (n = 385 patients) was 26.7 cc (range, 0.42–897.5 cc). The median reexcision volume (n = 352) was 79.5 cc (range, 3.0–4368 cc). The clinical and histopathologic features of all 510 patients are summarized in Table 1, which also shows the correlation of each feature with the incidence of residual tumor. Tumor size and surgical margin status showed strong correlations with the presence of residual tumor (P = 0.0063 and P < 0.0001, respectively; chi-square test). Histology was a statistically relevant factor only when invasive carcinomas were stratified by the presence or absence of EIC (P = 0.0156; chi-square test). Age (dichotomized at 45 years) and initial excision volume (stratified by the maximum cavity size acceptable to the balloon catheter: ≤ 70 cc vs. > 70 cc and ≤ 120 cc vs. > 120 cc) were not statistically significant at the P ≤ 0.05 level.

Table 1. Comparison of the Incidence of Residual Tumor (n = 510 patients)
ParameterPercentage (proportion) with residual tumorP value
  1. IDC/ILC: invasive ductal/lobular carcinoma; DCIS: ductal carcinoma in situ; EIC: extensive intraductal component.

Overall49.6 (253/510) 
Histology (IDC/ILC not stratified) 0.7957
 DCIS50.2 (123/245) 
 IDC/ILC49.1 (130/265) 
Histology (IDC/ILC stratified) 0.0156
 DCIS50.2 (123/245) 
 IDC/ILC without EIC41.4 (63/152) 
 IDC/ILC with EIC59.3 (67/113) 
Tumor size 0.0063
 Unknown44.4 (4/9) 
 ≤ 2cm42.1 (82/195) 
 > 2 cm54.6 (167/306) 
Initial volume 0.7830
 ≤ 70 cc49.5 (150/303) 
 > 70 cc51.2 (42/82) 
 Unknown48.8 (61/125) 
Initial volume 0.5050
 ≤ 120 cc50.4 (175/347) 
 > 120 cc44.7 (17/38) 
 Unknown48.8 (61/125) 
Initial surgical margin < 0.0001
 Positive58.9 (168/285) 
 0.1–1.0 mm41.3 (59/143) 
 1.1–2.0 mm24.0 (6/25) 
 > 2.0 mm0.0 (0/10) 
 Unknown42.6 (20/47) 
Age 0.0837
 ≤ 45 yrs56.2 (73/130) 
 > 45 yrs47.4 (180/380) 

Of the 510 patients with breast carcinoma, 385 patients had available data on the initial excision volume. The incidence of residual tumor stratified by histology, tumor size, and surgical margin status are presented for excision volumes ≤ 70 cc and ≤ 120 cc in Tables 2 and 3, respectively. These tables include the results of the analyses performed to evaluate the relations of these clinical and histopathologic factors with residual tumor in each of the excision volume subgroups. Univariate analysis in each of these subgroups was performed to look for residual tumor. Only surgical margin status and histology proved statistically significant with P ≤ 0.05 for the subset of patients who had excision volumes ≤ 70 cc. Surgical margin status, histology, and tumor size were significant factors in the larger subgroup of patients with excision volumes ≤ 120 cc. Figures 1 and 2 present the incidence of residual tumor as a function of surgical margin size for excision volumes ≤ 70 cc and ≤ 120 cc, respectively. These graphs show that, for both excision volume subgroups, the presence of extensive or pure DCIS was associated with a larger and more statistically relevant incidence of residual tumor than invasive histology alone.

Table 2. Incidence and Characteristics of Residual Tumor for Initial Excision Volume ≤ 70 cc (n = 303 patients)
ParameterPercentage (proportion) with residual tumorP value
  1. IDC/ILC: invasive ductal/lobular carcinoma; DCIS: ductal carcinoma in situ; EIC: extensive intraductal component.

Histology (IDC/ILC not stratified) 0.3858
 DCIS51.9 (81/156) 
 IDC/ILC46.9 (69/147) 
Histology (IDC/ILC stratified) 0.0251
 IDC/ILC without EIC36.1 (26/72) 
 IDC/ILC with EIC57.3 (43/75) 
 DCIS51.9 (81/156) 
Tumor size 0.0760
 ≤ 2 cm43.7 (52/119) 
 > 2 cm54.2 (97/179) 
 Unknown20.0 (1/5) 
Surgical margin status 0.0014
 Positive58.3 (98/168) 
 0.1–1.0 mm40.4 (38/94) 
 1.1–2.0 mm35.3 (6/17) 
 > 2.0 mm0.0 (0/6) 
 Unknown44.4 (8/18) 
Age 0.1645
 ≤ 45 yrs56.0 (47/84) 
 > 45 yrs47.0 (103/219) 
Table 3. Incidence and Characteristics of Residual Tumor for Initial Excision Volume ≤ 120 cc (n = 347 patients)
ParameterIncidence: Percentage (no.)P value
  1. IDC/ILC: invasive ductal/lobular carcinoma; DCIS: ductal carcinoma in situ; EIC: extensive intraductal component.

Histology (IDC/ILC not stratified) 0.4887
 DCIS52.2 (94/180) 
 IDC/ILC48.5 (81/167) 
Histology (IDC/ILC stratified) 0.0454
 IDC/ILC without EIC39.5 (34/86) 
 IDC/ILC with EIC58.0 (47/81) 
 DCIS52.2 (94/180) 
Tumor size: 0.0285
 ≤ 2 cm43.6 (58/133) 
 > 2 cm55.8 (116/208) 
 Unknown16.7 (1/6) 
Surgical margin status: 0.0003
 Positive59.1 (114/193) 
 0.1–1.0 mm41.7 (43/103) 
 1.1–2.0 mm33.3 (6/18) 
 > 2.0 mm0.0 (0/8) 
 Unknown48.0 (12/25) 
Age 0.5309
 ≤ 45 yrs53.2 (50/94) 
 > 45 yrs49.4 (125/253) 
Figure 1.

The percentage of residual tumor is illustrated as a function of surgical margin size for excision volumes ≤ 70 cc. DCIS: ductal carcinoma in situ; IDC/ILC: infiltrating ductal/lobular carcinoma; EIC: extensive intraductal component.

Figure 2.

The percentage of residual tumor is illustrated as a function of surgical margin size for excision volumes ≤ 120 cc. DCIS: ductal carcinoma in situ; IDC/ILC: infiltrating ductal/lobular carcinoma; EIC: extensive intraductal component.

On multivariate analysis, only surgical margin status (positive vs. > 0 mm) and histology displayed statistically significant correlations with residual tumor. The odds ratios for these results using IDC/ILC without EIC as a reference value are displayed in Table 4.

Table 4. Multivariate Model for Factors Correlated with Residual Tumor
VariableOR (95% CI)P value
  1. OR: odds ratio; 95% CI: 95% confidence interval; IDC/ILC: invasive ductal/lobular carcinoma; EIC: extensive intraductal component; DCIS: ductal carcinoma in situ; ROC: receiver-operator curve.

Histology: 0.0080
 IDC/ILC without EIC1.0 (reference) 
 IDC/ILC with EIC2.28 (1.31-.97) 
 DCIS1.80 (1.13–2.86) 
Surgical margin status (positive vs. > 0 mm)2.65 (1.78–3.96)≤ 0.0001
Model ROC area0.64 


APBI rapidly is becoming a popular means for breast-conserving therapy in the setting of early-stage, lymph node-negative breast carcinoma. Interest in APBI has grown among patients and physicians for several reasons: treatment is conducted over 5 days rather than over 6–7 weeks; less breast tissue is treated, which potentially may offer fewer side effects; and the option of reirradiation of breast tissue is opened. This is particularly relevant for patients who may have had a separate portion of the breast treated previously (e.g., in a mantle field for Hodgkin lymphoma).

The MammoSite catheter presents some unique benefits and obstacles for APBI. Brachytherapy is attractive particularly for APBI, because it eliminates the uncertainty in target volume localization that is inherent in external beam methods. Because the MammoSite catheter positions a single high-dose-rate source at the center of a saline-filled cavity, the challenges of both placement and dosimetric planning are less complicated relative to other interstitial methods. However, increased simplicity comes at the cost of versatility. Proper placement requires a volume limitation of the resected tissue. A recent publication from The University of Texas M. D. Anderson Cancer Center put forth a retrospective analysis of their institutional experience primarily in patients with lymph node-negative breast carcinoma who were treated with breast conservation and whole breast irradiation.5 By examining the volume of tissue excised in each patient, those investigators determined that 20–25% of patients would have been deemed acceptable for MammoSite catheter placement. The next logical question is, for those patients with an appropriate excision volume, which of them has an acceptably low risk of residual disease to recommend APBI with this device? Vicini et al. recently showed through a review of lumpectomy specimens and reexcisions at their institution that as many as 90% of breast conservation patients will have no residual disease outside of the 1-cm prescription depth of the MammoSite applicator.1 Furthermore, their report showed that patients who have residual tumor that extends beyond 1 cm may be predicted by relating the initial surgical margin status to the ratio of invasive tumor size and initial excision volume. Our current work has shown that a reasonable prediction of residual tumor status may be made from the width of the surgical margin alone. We also have observed an increased risk of residual disease in patients with EIC, as reported previously by our group16 and by others.17 This finding was corroborated by the observation from Vicini et al. that DCIS in the presence of invasive carcinoma extended farther, on average, from the initial excision margin than other histologies. It is our opinion that these two factors, surgical margin size and histology, are the primary elements of risk determination that must be used in an informed discussion with a patient concerning the use of this device.

The guidelines from the ABS and ASBS on eligibility for APBI differ with regard to the surgical margin requirement, recommending any negative surgical margin status versus a 2-mm minimum in all directions, respectively. In the current study, we qualified the risk of residual disease inherent within these parameters. Our data analysis revealed a 40.4% correlation with residual tumor for a negative surgical margin ≤ 1.0 mm in patients with the 70-cc maximum excision volume, and 41.7% for patients with the 120-cc maximum volume. These numbers dropped to 35.3% and 33.3%, respectively, for surgical margins of 1.1–2.0 mm. Given the previously stated benefits of APBI, many patients would find this acceptable in the absence of additional risk factors.

It should be noted that the cohort examined in the current study did not constitute a random sampling of patients undergoing breast conservation. The patients included were those at our institution who underwent reexcision after lumpectomy to obtain negative or adequate surgical margins during the study period. Additional criteria for reexcision and how these patients differed from other breast-conservation candidates at our institution were not specified. This is justified, because the data presented herein are meant as a useful guide in the determination of a given patient's eligibility for the MammoSite brachytherapy catheter. That assessment also must take into account the clinical attributes of the patient (such as age and medical comorbidities) and histologic features of the tumor. At our institution, we utilize this information within the guidelines set by the ABS.

The results of the current study indicate that breast tumor lumpectomies performed within the volume parameters of the MammoSite device offer at most a 35% risk of microscopically detectable residual disease for a surgical margin ≥ 1 mm. This risk may drop dramatically for surgical margins ≥ 2 mm. A large proportion of intraductal histology may connote a significantly greater risk of residual disease, with an adjusted odds ratio of 0.44 for purely invasive histology compared with tumors that have EIC. The relevance of patient age and tumor volume in predicting the probability of residual disease is not clear at this time and warrants further study.