The objectives of this study were to determine the locoregional recurrence (LRR) rate and to evaluate the correlation between surgical resection volume (RV) and LRR in patients with breast cancer who underwent segmental mastectomy after achieving a pathologic complete response (pCR) on neoadjuvant chemotherapy.
The authors reviewed the outcomes of all 109 patients who underwent segmental mastectomy after the complete eradication of invasive disease by neoadjuvant chemotherapy at their institution between 1987 and 2002. LRRs were recorded, and RVs after segmental mastectomy were calculated and categorized as small, medium, or large.
At a median follow-up of 6.6 years, 3 patients (2.7%) developed LRR. In 2 of those patients, the recurrence was located in the ipsilateral breast; in the other patient, the recurrence was located in the supraclavicular lymph nodes with synchronous distant metastases. The median RV was 73.12 cm3 (range, 2.82–451.51 cm3). Large RVs (>125 cm3) were less common than small RVs (up to 70 cm3) or medium RVs (between 70 cm3 and 125 cm3; P = .009 and P<.0001, respectively). One patient with a small RV had an LRR at 4 years, and 2 patients with medium RVs had LLRs at 2.3 years and 6 years, respectively. The 5-year and 10-year LRR-free survival rates were 98.1% and 96.5%, respectively, and the corresponding overall survival rates were 96% and 92%, respectively.
Neoadjuvant chemotherapy is a widely accepted treatment that is in use currently not only for locally advanced breast cancer but also for earlier stage disease.1, 2 There are 3 main advantages of neoadjuvant chemotherapy: It permits in vivo determination of an individual tumor's chemosensitivity; it can minimize micrometastatic disease; and it can downstage tumors, making previously ineligible patients eligible for breast-conserving surgery (BCS).3–5
In the largest prospective, randomized trial of neoadjuvant chemotherapy versus adjuvant chemotherapy for breast cancer (National Surgical Adjuvant Breast and Bowel Project [NSABP] B-18), the safety of BCS was demonstrated after tumor shrinkage, and the rates of BCS were higher among women who received neoadjuvant chemotherapy compared with the rates among women who received adjuvant chemotherapy (67.8% vs. 59.8%, respectively).6 Overall, there was no difference in outcome between patients who received neoadjuvant chemotherapy and those who received adjuvant chemotherapy.6 However, in the NSABP study and in other studies, a pathologic complete response (pCR), defined as the absence of invasive tumor in the breast only or in the breast and in the axilla, was associated with better overall survival.6–8 Thus, efforts have been made to increase pCR rates by using more effective drugs and regimens. With the addition of taxanes to traditional anthracycline-based regimens and with the addition of trastuzumab to the treatment for Her2/neu-positive patients, pCR rates up to 34% and 65%, respectively, have been achieved.1, 9, 10
The need for BCS is expected to increase as the incidence of pCR increases. However, there is some concern regarding discrepancies in locoregional recurrence (LRR) rates between previous studies that investigated LRR after BCS in patients who received with neoadjuvant chemotherapy. Some of those studies showed high LRR rates, up to 22.6% at 10 years,11, 12 whereas others showed lower LRR rates, up to 10.7% at 9 years.7, 13, 14 These discrepancies have led to uncertainty and to a questioning of the benefits of BCS. The majority of those previous studies included small numbers of patients who achieved a pCR. In addition, the previous studies used different selection criteria and treatments. Thus, currently, to our knowledge, little is known regarding LRR in patients who undergo BCS and achieve a pCR in the breast and axilla.
Selection criteria for BCS include the ability to remove residual disease completely and to maintain optimal cosmetic results; however, to our knowledge, there have been no reports concerning the extent of BCS in patients with no residual tumor. The objective of this study was to determine the total resection volume (RV), the LRR rate, and overall outcome in patients from prospective institutional trials who underwent segmental mastectomy after they achieved a pCR to neoadjuvant chemotherapy.
MATERIALS AND METHODS
A retrospective review of medical records identified 243 women with breast cancer who experienced a pCR (defined as the absence of invasive tumor in the breast and the axilla) after receiving neoadjuvant chemotherapy on prospective institutional trials at The University of Texas M. D. Anderson Cancer Center between 1987 and 2002. Of these, 109 consecutive patients underwent BCS. The initial diagnosis of invasive breast cancer was made on the basis of core-needle biopsy. Disease stage was determined on the basis of the 2003 staging system of the American Joint Committee on Cancer.15 The full details of the neoadjuvant chemotherapy regimens have been reported previously elsewhere.16–18 Tumor response to neoadjuvant chemotherapy was monitored by physical examination, mammography, and sonography performed during chemotherapy and before surgery (at completion of chemotherapy). Metallic coils were placed after 2 cycles of chemotherapy in the tumor bed under ultrasound guidance to enable identification of the primary tumor site at the time of surgery. After completion of neoadjuvant chemotherapy, a multidisciplinary team determined eligibility for BCS. Patients with inflammatory breast cancer, extensive calcifications, multicentric disease, or extensive invasive lobular disease were considered ineligible for BCS. Selection criteria for BCS included the ability to maintain optimal cosmesis after resection of the primary tumor site. In general, surgeons attempted not to resect the tumor volume observed at initial diagnosis. BCS consisted of segmental mastectomy together with standard Level I and II axillary lymph node dissection or sentinel lymph node biopsy. On final histopathologic examination, the macroscopic size of the resection specimen and the greatest width, length, and height axis dimensions in centimeters were recorded. Other items recorded included the presence or absence of invasive cancer, chemotherapy-associated changes in the tumor bed, the presence or absence of residual intraductal carcinoma, and results of lymph node examination. Pathologic examination of sentinel lymph node(s) included hematoxylin and eosin staining and immunohistochemical staining.
To calculate RVs for this study, we used the formula for an ellipsoid volume: 4/3 × width axis radius × length axis radius × height axis radius. The ellipsoid formula estimates the volume of breast specimens more accurately than the formula for rectangular solids, which overestimates the true RV.19 The total RV was defined as the sum of all RVs for each surgical candidate. RVs were categorized as small (<70 cm3), medium (70–125 cm3), or large (>125 cm3). These cut-off sizes were applied previously in a study evaluating the patients who would be eligible for balloon catheter-based, accelerated, partial breast irradiation20 after BCS based on the RV.
The treatment protocols called for adjuvant hormone therapy to be given to patients with hormone receptor-positive tumors and for adjuvant external-beam radiotherapy to be given to all patients. Follow-up examinations were performed at least every 6 months during the first 5 years and annually thereafter.
The outcome endpoints examined were LRR and overall survival. LRR was defined as recurrent disease in the ipsilateral breast or in the axillary, supraclavicular, infraclavicular, or internal mammary lymph nodes. The time to LRR was defined as the time from initial tumor diagnosis to time of last follow-up or development of LRR. Kaplan–Meier estimates were used to calculate LRR-free and overall survival rates at 5 years and 10 years. Frequency distributions were tested by using the chi-square test. The analysis-of-variance F test was applied to analyze the means between groups. Significance tests were 2-tailed, and differences were considered to be statistically significant at P<.05.
Before treatment, all patients underwent a physical examination and staging work-up. Patient and treatment characteristics are shown in Table 1. The median clinical tumor size at presentation was 3.5 cm (range, 1.2–9 cm). Sixty-nine patients had a clinically positive axilla, and metastatic disease was confirmed cytologically in 53 of those patients (76.8%) at diagnosis. The majority of patients had Stage IIA, IIB, or IIIA disease. Seventy-two patients (66.1%) received both neoadjuvant and adjuvant chemotherapy as part of their treatment research protocol, and 37 patients (33.9%) received only neoadjuvant chemotherapy. Neoadjuvant chemotherapy was anthracycline-based in 100 patients (91.7%), and 45% of patients also received taxane-based chemotherapy either preoperatively or postoperatively. Of the 32 patients who presented with hormone receptor-positive tumors, 31 patients (96.9%) received adjuvant hormone therapy. One hundred seven patients (98.2%) received postoperative external-beam radiotherapy to the affected breast with tangential fields. The median dose to the breast was 50 grays (Gy) delivered in 25 fractions over 5 weeks. These patients also received an electron boost to the tumor bed (median dose, 10 Gy). One patient declined radiotherapy, and 1 patient did not receive radiotherapy because she developed distant metastasis shortly after completing adjuvant chemotherapy. A clinical complete response (no residual tumor on palpation and on radiologic imaging studies) was estimated in 64% of patients, and a clinical partial response (≥50% decrease in tumor size) was estimated in 36% of the patients. Eighty patients (73.4%) underwent segmental mastectomy and Level I and II axillary lymph node dissection, and 25 patients (22.9%) underwent segmental mastectomy and sentinel lymph node biopsy. Four patients (3.7%) declined axillary surgery. The mean number of lymph nodes removed after Level I and II axillary dissection was 16.4 (range, 5–36 lymph nodes removed). The mean number of lymph nodes removed after sentinel lymph node biopsy was 3.4 (range, 1–9 lymph nodes removed). The median RV at segmental mastectomy was 73.12 cm3 (range, 2.82–451.51 cm3). There was no significant difference in median RVs between patients with initial clinical tumor sizes of T1, T2, T3, and T4 (P = .31) (Table 2). Forty-nine patients (44.9%) underwent small-volume resection, 39 patients (35.8%) underwent medium-volume resection, and 21 patients (19.3%) underwent large-volume resection. Large RVs were less common than small RVs (P = .009) or medium RVs (P<.0001). There was no significant difference between the proportion of patients who had small RVs and medium RVs (P = .21). Univariate analysis showed that patients with more advanced disease (≥Stage IIIA) were more likely to have large RVs (odds ratio;, 3.3; 95% confidence interval, 1.2–8.9; P = .017). For patients who had a clinical complete response and patients who had a clinical partial response, there were no significant differences observed in RVs (75.20 cm3 vs. 66.78 cm3, respectively; P = .53). Histopathologic examination of the surgical specimens revealed residual intraductal carcinoma in 21patients (19.3%).
Table 1. Patient and Treatment Characteristics
No. of patients (%)
Clinical tumor category
Initial clinical stage
Grade 1 or 2 (well or moderately differentiated)
Grade 3 (poorly differentiated)
Hormone receptor status
Adjuvant hormone therapy
Table 2. Resection Volumes and Clinical Tumor Category
Median resectionvolume (Range), cm3
At a median follow-up of 80 months (range, 12–239 months), 3 patients (2.7%) had developed an LRR. The median time to LRR was 48 months. Two patients had LRR in the ipsilateral breast—near the primary tumor site in 1 patient who had a small RV and in a different quadrant from that of the primary tumor site in the other patient, who had a medium RV. Those patients underwent mastectomy followed by breast reconstruction; and, at the time of this report, both patients were alive and well 185.2 months and 72.1 months, respectively, after their initial diagnosis. One patient who had a medium RV had an LRR in the supraclavicular lymph nodes with synchronous distant metastases in the liver and bone. This patient subsequently was treated with systemic therapy but died 2 months later. This patient had a shorter time to recurrence (27.6 months) and a shorter overall survival (30 months) than the patients who had LRR without distant metastases. Characteristics of the patients with LRR are shown in Table 3. Characteristics that they had in common were young age at diagnosis (all age 40 years or younger), clinical T2 tumors, clinically positive lymph nodes (confirmed by cytology in 2 patients), and high nuclear grade. Of 21 patients who had ductal carcinoma in situ in the surgical specimen, only 1 patient developed an in-breast recurrence (4.8%). All 3 patients with recurrences underwent an initial segmental mastectomy and Level I and II axillary dissection. The mean number of lymph nodes dissected was 14.3 (range, 13–16 lymph nodes dissected).
Table 3. Characteristics of Patients with Local Recurrence
SM indicates segmental mastectomy; ALND, axillary lymph node dissection; A, anthracycline; A/T, anthracycline plus taxane; DCIS, ductal carcinoma in situ; LN, lymph node; LRR, locoregional recurrence.
Age at diagnosis, y
Clinical tumor size, cm
Adjuvant hormone treatment
Resection volume, cm3
Macroscopic size of surgical specimen
6 × 5 × 4 cm
8 × 7 × 4 cm
8 × 7 × 3.5
DCIS in surgical specimen
Site of LRR
Time to LRR, mo
Status at last follow-up
Overall survival, y
For the entire series of 109 patients, the 5-year and 10-year LRR-free survival rates were 98.1% and 96.5%, respectively (Fig. 1), and the 5-year and 10-year overall survival rates were 96% and 92% (Fig. 2), respectively. At the time of this analysis, 11 patients (10%) had developed distant metastases, and 4 patients had died with disease.
To our knowledge, the current study is the first to analyze locoregional control in a large number of patients who achieved a pCR who underwent BCS after neoadjuvant chemotherapy. Our observations suggest that segmental mastectomy is associated with excellent local control and overall survival in patients who achieve a pCR after neoadjuvant chemotherapy. Only 3 patients (2.7%) experienced an LRR at a median follow-up of 80 months, and 1 of those patients had synchronous distant metastases. Overall survival rates for the whole population were 96% at 5 years and 92% at 10 years. In this study, 80.7% of patients had small RVs (up to 70 cm3) or medium RVs (70–125 cm3). An RV of approximately 70 cm3 would correspond to a breast specimen that measures 8 × 4 × 4 cm, and an RV of 125 cm3 would correspond to a breast specimen that measures 10 × 6 × 4 cm. These sizes demonstrate the extent of segmental mastectomy at our institution, although other institutions may perform more extensive or less extensive BCS. Our findings indicate that segmental mastectomy is associated with excellent locoregional control and survival when a median volume of 73.12 cm3 (range, 2.8–451.5 cm3) is resected. Extending the RVs in this population is unlikely to lead to significant improvements of local control.
In a previous study from The University of Texas M. D. Anderson Cancer Center that evaluated RVs in 445 patients with ductal carcinoma in situ or invasive breast cancer who underwent BCS prior to adjuvant therapy, the median RV was 67.71 cm3 (range, 2.0–588.4 cm3), which is slightly lower than the RV in the current study population.20 In our study, the patient who had an LRR at 4 years had a small RV, and the 2 patients who had LRRs at 2.3 years and 6 years, respectively, had medium RVs. Although information regarding the minimum RV that correlates with the best local control would be useful, no conclusions can be made concerning the optimal RV on the basis of the current study given the low LRR rate. The main clinical objective of tumor downstaging in patients who receive neoadjuvant chemotherapy is to avoid mastectomy. However, there has been some concern regarding the appropriateness of BCS after neoadjuvant chemotherapy in patients who achieve a pCR. In an institutional series of 226 patients who achieved a pCR after neoadjuvant chemotherapy, the mastectomy rate was 59%, and the BCS rate was 41%.17 Buzdar et al.10 showed that, in a patient population with Her2/neu overexpression, adding the most active regimens in breast cancer treatment resulted in a pCR rate of 65%. In that series, only 52.6% of patients in the control group and 56.5% of patients in the experimental group underwent BCS despite the high pCR rate. In addition to patient convenience and preference, there are several potential reasons for choosing mastectomy over BCS in patients who achieve the complete eradication of tumor after neoadjuvant chemotherapy. These include the uncertainty caused by inconsistent reports concerning locoregional control after BCS, the discrepancy between clinical and pathologic response because of the moderate accuracy of imaging, and primary tumor characteristics that represent contraindications for BCS. Our results show that segmental mastectomy is associated with a minimal risk of LRR in patients who have no residual tumor after neoadjuvant chemotherapy. Thus, mastectomy can be avoided in such patients.
Our findings regarding patient outcome after pCR and BCS compare favorably with findings from previous studies. In an institutional analysis of 340 patients who underwent BCS after neoadjuvant chemotherapy, the LRR rate was 9%, and the ipsilateral breast tumor recurrence rate was 5% at a median follow-up of 60 months.21 For the subset of 67 patients who achieved a pCR, the LRR rate was 4%, and the 5-year LRR-free rate was 93.21 In the NSABP B-18 study,14 the ipsilateral breast tumor recurrence rate for all patients who underwent BCS was 10.7%, and the recurrence rate for the subset of 88 patients who achieved a pCR was 6.7%. In that series, overall survival at 9 years of follow-up for patients who achieved a pCR was 85%. In our study, the ipsilateral breast tumor recurrence rate was only 1.8%. Possible reasons for the discrepancy between our findings and those from NSABP B-18 include the use of different chemotherapy drugs and regimens and differences in criteria used to determine BCS eligibility. Newman et al.22 showed that lobular histology, multicentricity, and calcifications may influence BCS eligibility. In addition, a pCR is defined by the NSABP as the absence of invasive tumor in the breast only, regardless of residual disease in the axillary lymph nodes.
The current findings indicate that better methods are needed for the preoperative identification of patients who achieve a pCR. At our institution, tumor response to neoadjuvant chemotherapy is assessed routinely by physical and radiologic examination, and a multidisciplinary team plans definitive treatment. Sixty-four percent of the patients in our study had a clinical complete response to neoadjuvant chemotherapy; however, 36% of patients were categorized with only a partial response. These observations show that conventional assessment of response may fail to predict a pCR in some patients. Furthermore, conventional radiologic methods have only moderate accuracy in predicting tumor response preoperatively.23 More precise imaging methods are needed to enable more accurate identification of pCR and to simplify planning of the surgical RV.
Several questions regarding BCS after pCR to neoadjuvant chemotherapy remain to be answered. For example, can excellent local control be achieved even with small RVs? Can we standardize the appropriate RV? What is the role of adjuvant radiotherapy in patients who achieve a pCR after neoadjuvant chemotherapy and undergo BCS? Ring et al.24 compared outcomes after surgery plus radiotherapy and radiotherapy alone in a nonrandomized series of 136 patients who had a clinical complete response. In that study, BCS was offered to 85% of patients in the surgery group, and all patients in the radiotherapy group received radiotherapy to the axilla. The LRR rate was 10% in the surgery group and 21% in the no-surgery group (P = .09). Although those authors did not report the total number of patients who achieved a pCR, their results showed that a high LRR rate occurred when surgery was omitted. Considering that a proportion of patients who achieve a pCR will have a noninvasive tumor component in the resection specimen, we believe that surgery is necessary for appropriate local control. The impact of adjuvant radiotherapy on LRR and survival was addressed extensively in a recent meta-analysis by the Early Breast Cancer Trialists' Collaborative Group,25 which reported that, for every 4 local recurrences avoided as a result of adjuvant radiotherapy, 15-year overall mortality would be reduced by 1 death. In our study, 98% of patients, including the patients with LRR, received adjuvant radiotherapy. Future studies should examine the extent to which conventional radiotherapy can prevent LRR in patients who achieve a pCR and whether the use of alternative approaches, like partial breast irradiation, which is being investigated currently in NSABP protocol B-39, would be of interest in selected patients who receive neoadjuvant chemotherapy. Such dramatic increases in pCR rates emphasize the need to increase BCS and indicate that information regarding the appropriate extent of the surgical resection is essential to encourage surgeons to perform BCS rather than mastectomy.
In summary, our observations suggest that segmental mastectomy with adjuvant radiotherapy is associated with excellent local control and overall survival in carefully selected patients who achieve a pCR after neoadjuvant chemotherapy. Improvements in imaging modalities may be helpful in the more appropriate selection of patients to avoid mastectomy as pCR rates increase because of novel chemotherapy regimens. The optimal RV and the role of innovative radiotherapy methods are issues that remain to be investigated.
We thank Keith Anderson, MS (Department of Biostatistics and Applied Mathematics, The University of Texas M. D. Anderson Cancer Center) for expert statistical support and Sean Eric McGuire, MD (Department of Radiation Oncology, The University of Texas M.D. Anderson Cancer Center) for his assistance in the preparation of this study