Local-regional recurrence following neoadjuvant endocrine therapy: Data from ACOSOG Z1031 (Alliance), a randomized phase II neoadjuvant comparison between letrozole, anastrozole, and exemestane for postmenopausal women with estrogen receptor–positive clinical stage 2 to 3 breast cancer

Abstract

Background:

ACOSOG Z1031 addressed the ability of three neoadjuvant aromatase inhibitors (NAIs) to reduce residual disease (Cohort A) and to assess whether switching to neoadjuvant chemotherapy (NCT) after 4 weeks on NAI with Ki67 > 10% increases pathologic complete response (pCR) in postmenopausal women with estrogen receptor-enriched (Allred score 6–8) breast cancer (BC).

Methods:

622 women with clinical stage 2–3 estrogen receptor-positive (ER+) BC were enrolled, (Cohort A: 377, Cohort B: 245). The analysis cohort consisted of 509 patients after excluding patients who did not meet trial eligibility criteria, switched to NCT or surgery due to 4-week Ki67 > 10%, or withdrew prior to surgery. Distribution of time to local-regional recurrence (LRR) was estimated using competing risk approach where distant recurrence and second primaries were considered competing risk events. Patients who died without LRR, distant recurrence or second primary were censored at last evaluation.

Results:

342 (67.2%) patients had breast conserving surgery (BCS). Of 221 patients thought to require mastectomy at presentation, 50% were able to have BCS. Five (1.0%) patients had no residual disease in the breast or nodes at surgery. Among 382 women still alive, 90% have been followed >5 years. Five-year cumulative incidence rate for LRR is estimated to be 1.53% (95% CI: 0.7 to 3.0%).

Conclusions:

NAI rarely results in pCR in stage 2–3 ER+ BC, however a significant proportion will have downstaging to allow for BCS. LRR after surgery is uncommon (1.5% at 5 years) and support the use of BCS after NAI.

BACKGROUND:

Neoadjuvant systemic therapy is emerging as the preferred approach for patients with locally advanced breast cancer. In addition to de-escalation of surgical treatment, the neoadjuvant setting provides an opportunity to explore new treatment strategies for women with early-stage breast cancer. Early studies conducted by the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-18 showed no significant difference in survival among patients treated with preoperative or postoperative chemotherapy.¹ These early studies not only provided rationale for a change in conventional treatment strategies, but also raised many questions that have challenged the fields of breast medical oncology and breast surgical oncology. Further studies have identified many advantages to providing neoadjuvant chemotherapy (NCT) prior to surgery. The initial hypotheses that neoadjuvant treatments could downsize tumors, converting inoperable tumors to operable, and those that would have otherwise required mastectomies to breast conserving surgery, have been supported by many large studies.² Additionally, neoadjuvant chemotherapy has been shown to decrease the rate of positive nodes, reducing the extent of axillary surgery.^3–5 Besides the benefits provided to patients in terms of treatment options, the neoadjuvant treatment setting provides a window of opportunity for the physicians to observe the in vivo effects of the treatment on the tumor and adjust the treatment strategy. For example, pathologic complete response (pCR) to NCT is associated with improved outcomes and can therefore be used as a prognostic indicator allowing for de-escalation of therapy for patients with pCR.^{1, 6–9}

pCR is achieved in nearly 20% of patients receiving NCT, but the rate depends on the subtype and stage of breast cancer, with pCR rates of 18% in estrogen receptor-positive/HER2-positive tumors and 0.3% in luminal A tumors.^10–12 The majority (>70%) of breast cancers in the United States fall into the luminal A or luminal B subtype being estrogen receptor-positive (ER+) and HER2-negative. Neoadjuvant endocrine therapy (NET), including tamoxifen and aromatase inhibitors (AIs), has been investigated as an alternative to NCT for luminal breast cancers. Studies have shown that pCR rates are similar in luminal breast cancers treated with NET or NCT, with rates of breast conservation being slightly higher with NET compared with NCT (33% vs 24%, p=0.058).¹³ However, despite growing evidence of the efficacy of NET, it is still underutilized.¹⁴ To compare NCT to NET, it must first be determined if there are clinically meaningful differences between the types of NET, using appropriate endpoints to address this question.

American College of Surgeons Oncology Group (ACOSOG) Z1031 was a randomized phase II neoadjuvant clinical trial in postmenopausal women with ER enriched breast cancer designed to assess the clinical response rate (cRR), reduction in proliferation, surgical outcomes, and preoperative endocrine prognostic index (PEPI) score after 16 to 18 weeks of either exemestane (2.5 mg daily), anastrozole (1 mg daily), or letrozole (2.5 mg daily) and to recommend an AI for further testing against chemotherapy in the neoadjuvant setting (Cohort A). ACOSOG is now part of the Alliance for Clinical Trials in Oncology. The cRR was found to be 74.8% (95% CI, 66.3% to 82.1%) for letrozole, 69.1% (95% CI, 60.1% to 77.1%) for anastrozole and 62.9% (95% CI, 53.8% to 77.4%) for exemestane. It was recommended that letrozole and anastrozole be carried forward for further testing.¹⁵

Prior to the completion of Cohort A enrollment, the trial was amended to enroll a second cohort of patients (Cohort B) where a tumor biopsy was taken after 2–4 weeks after AI initiation to measure Ki67¹⁶. If the Ki67 was >10%, patients were encouraged to switch to NCT or go directly to surgery. The objectives for this cohort were to assess whether switching to NCT would improve the pCR rate (to at least 20%) and to assess whether recurrence-free survival (RFS) differed by respect to whether the PEPI score was 0 or not among those with week 4 Ki67 ≤ 10%. This phase II therapeutic trial was monitored at least twice annually by the Data and Safety Monitoring Board, a standing committee composed of individuals from within and outside of the Alliance.

When the results of Cohort A were released, patients were given the choice of letrozole or anastrozole treatment. The pCR rate among the 35 patients whose week 2–4 Ki67 was >10% who switched to NCT was 5.7% (95% CI, 0.7% to 19.1%) and the RFS among the women with week 2–4 Ki67 ≤10% was found to be significantly decreased for those with a PEPI = 0 score (recurrence hazard ratio [PEPI = 0 v PEPI > 0], 0.27; P = .014; 95% CI, 0.092 to 0.764).¹⁵

Patients in both cohorts were followed annually for a maximum of 10 years post-surgery for adjuvant treatment (at the discretion of the patient’s local physician), disease progression, and survival information. This manuscript reports the local-regional recurrence (LRR) events following NET of women from ACOSOG Z1031 Cohorts A and B who completed 16–18 weeks of AI treatment and then went directly to surgery.

METHODS:

Study Cohort

Postmenopausal women with invasive, clinical stage 2 or 3 (T2 to T4c, N0 to N3, M0), ER enriched (Allred score of 6–8) breast cancer enrolled on Z1031 who met trial eligibility criteria and began protocol treatment were potential candidates for inclusion.¹⁵ All Z1031 participants signed institutional review board approved informed consents specific to the Z1031 protocol. Patients were excluded from the study cohort if they discontinued NET due to refusal to continue AI, toxicity, disease progression, or new primary diagnosis and then either received non-protocol treatment prior to surgery or underwent surgery off protocol. In addition, Cohort B patients with a 4-week biopsy showing a Ki67 >10% who switched to NCT or went immediately to surgery were excluded (Figure 1 CONSORT Diagram). As of 12/31/2019, approximately 10 years after the last patient was enrolled, all data collection was terminated.

FIGURE 1:

CONSORT Diagram

Statistical Considerations

The distribution of the time to LRR following surgery was estimated using a competing risk approach where distant disease recurrence and new primary cancers (including contralateral breast cancer) were considered competing risk events. Patients who died without documentation of a local, regional or distant breast cancer recurrence or second primary diagnosis were censored at their last disease evaluation. Data collection and statistical analyses were conducted by the Alliance Statistics and Data Management Center. Data quality was ensured by review of data by the Alliance Statistics and Data Management Center and by the study chairperson following Alliance policies.

RESULTS

The patient and disease characteristics of the 509 analysis cohort participants are shown in Table 1. The median age at registration was 66 years (range 43 to 74). The clinical T category was T2 in 394 (77.4%), T3 in 93 (18.3%) and T4 in 22 (4.3%) patients. The majority were clinically node negative (370, 72.7%), 129 were N1 (25.3%), and 10 were N2–3 (2.0%).

Table 1:

Patient and Treatment Characteristics with Surgical Outcomes

	Cohort A (n=329)	Cohort B (n=180)	total (n=509)
AT STUDY ENTRY
Age at registration
less than 50 years old	3 (0.9%)	2 (1.1%)	5 (1.0%)
50–59	78 (23.7%)	52 (28.9%)	130 (25.5%)
60–69	119 (36.2%)	69 (38.3%)	188 (36.9%)
70 or older	129 (39.2%)	57 (31.7%)	186 (36.5%
Race
Caucasian	269 (81.8%)	156 (86.7%)	425 (83.5%)
Black	47 (14.3%)	17 (9.5%)	64 (12.6%)
Not reported	13 (3.9%)	7 (3.9%)	20 (3.9%)
BMI
≥ 25.0 (overweight/obese)	250 (76.0%)	141 (78.3%)	391 (76.8%)
< 25.0 (unweight/normal)	77 (23.4%)	33 (17.8%)	110 (21.6%)
Not reported	2 (0.6%)	6 (3.3%)	8 (1.6%)
Clinical T category
T2	249 (75.7%)	145 (80.6%)	394 (77.4%)
T3	63 (19.1%)	30 (16.7%)	93 (18.3%)
T4A-C	17 (5.2%)	5 (2.8%)	22 (4.3%)
Clinical N category
N0	238 (72.3%)	132 (73.3%)	370 (72.7%)
N1	84 (25.5%)	45 (25.0%)	129 (25.3%)
N2–3	7 (2.1%)	3 (1.7%)	10 (2.0%)
HER2 findings (by IHC and/or FISH) at Diagnosis
Negative	293 (89.1%)	173 (96.1%)	466 (91.6%)
Positive	25 (7.6%)	4 (2.2%)	29 (5.7%)
Not done	11 (3.3%)	3 (1.7%)	14 (2.7%)
Histologic Grade at Diagnosis
Grade I	76 (23.1%)	60 (33.3%)	136 (26.7%)
Grade II	202 (61.4%)	103 (57.2%)	305 (59.9%)
Grade III	49 (14.9%)	17 (9.4%)	66 (13.0%)
Not reported	2 (0.6%)	0	2 (0.4%)
SURGICAL PROCEDURES AND FINDINGS
Surgical Impression prior to NET and Surgery Performed Post NET*
Prior to NET Surgeon’s Impression Candidate for Lumpectomy
Breast Conserving Surgery performed post NET	148 (45.0%)	80 (44.4%)	228 (44.8%)
Mastectomy performed post NET	31 (9.4%)	24 (13.3%	55 (10.8%)
Prior to NET Surgeon’s Impression Candidate for Mastectomy
Breast Conserving Surgery performed post NET	75 (22.8%)	36 (20.0%)	111 (21.8%)
Mastectomy performed post NET	71 (21.6%)	39 (21.7%)	110 (21.6%)
Prior to NET Surgeon’s Impression Inoperable
Breast Conserving Surgery performed post NET	3 (0.9%)	0	3 (0.6%)
Mastectomy performed post NET	1 (0.3%)	1 (0.6%)	2 (0.4%)
Nodal Surgery
None	16 (4.9%)	2 (1.1%)	18 (3.5%)
SLN only	150 (45.6%)	114 (63.3%)	264 (51.9%)
ALND only	83 (25.2%)	38 (21.1%)	121 (23.8%)
SLN and ALND	80 (24.3%)	26 (14.4%)	106 (20.8%)
Maximum dimension of residual breast lesion
no disease present in breast	2 (0.6%)	4 (2.2%)	6 (1.2%)
0.1–2.0 cm	129 (39.2%)	72 (40.0%)	201 (39.5%)
2.1–5.0 cm	153 (46.5%)	87 (48.3%)	240 (47.2%)
more than 5.0 cm	45 (13.7%)	17 (9.4%)	62 (12.0%)
Number of positive lymph nodes
0	176 (53.5%)	94 (52.2%)	270 (53.0%)
1–3	84 (25.5%)	54 (30.0%)	138 (27.1%)
4–9	35 (10.6%)	19 (10.6%)	54 (10.6%)
10 or more	17 (5.2%)	11 (6.1%)	28 (5.5%)
not evaluated/reported	17 (5.2%)	2 (1.1%)	19 (3.7%)
PEPI Status
pCR	1 (0.3%)	4 (2.2%)	5 (1.0%)
zero	53 (16.1%)	60 (33.3%)	113 (22.2%)
non-zero	244 (74.1%)	111 (61.7%)	355 (69.7%)
unable to determine	31 (9.4%)	5 (2.8%)	36 (7.1%)
ADJUVANT THERAPY
Adjuvant radiation	212 (64.4%)	93 (51.7%)	305 (60.0%)
Adjuvant chemotherapy	131 (39.8%)	40 (22.2%)	171 (33.6%)
Adjuvant endocrine therapy
Anastrozole	144 (43.7%)	72 (40.0%)	216 (42.4%)
Exemestane	90 (27.4%)	22 (12.2%)	112 (22.0%)
Letrozole	119 (36.2%)	57 (31.7%)	176 (34.6%)
Tamoxifen	32 (9.7%)	6 (3.3%)	38 (7.5%)

^*Prior to the start of neoadjuvant treatment, the patient’s surgeon was asked to provide their impression as to whether the patient was a candidate for mastectomy or lumpectomy or had inoperable disease.

ALND – axillary lymph node dissection; BMI – body mass index; IHC – immunohistochemistry; FISH – fluorescence in situ hybridization; N – nodal; NET – neoadjuvant endocrine therapy; pCR – pathologic complete response; PEPI – preoperative endocrine prognostic index; SLN – sentinel lymph node; T – tumor

Surgical treatment following NET

Most patients underwent breast conserving surgery (BCS) (342, 67.2%). One hundred fourteen (50.4%) of the 226 patients who were thought to require a mastectomy (111/221) or to have inoperable breast cancer (3/5) prior to NET had BCS following NET. However, 55 (19.4%) of the 283 patients who were thought to be candidates for BCS prior to NET had a mastectomy. Among the 207 patients whose residual tumor size was 0–2.0 cm, 45 (21.7%) underwent a mastectomy, while 78 (32.5%) of the 240 patients whose residual tumor size was 2.1–5 cm had a mastectomy. Nodal surgery was not performed in 18 (3.5%) patients, 264 (51.9%) had sentinel lymph node (SLN) surgery alone, 121 (23.8%) had axillary lymph node dissection (ALND) alone, and 106 (20.8%) had SLN surgery and ALND. There were 6 (1.2%) patients with no residual disease in the breast at surgery. The number of positive nodes was 0 in 270 (53.0%), 1–3 in 138 (27.1%), 4–9 in 54 (10.6%) and 10 or more in 28 (5.5%). There were 118 (23.2%) patients who had a pCR (5) or PEPI score of 0 (113), (Table 1).

Local-regional recurrence rates after NET

Time to local-regional recurrence was defined as the time from surgery to documentation of a local or regional recurrence as the first event. When data collection ceased on12/31/2019, 382 women were still alive, and 90% had been followed at least 5 years. The first disease events are presented in Table 2. The 5-year cumulative incidence rate for a local-regional recurrence in this patient cohort is estimated to be 1.53% (95% CI: 0.7 to 3.0%).

TABLE 2:

First Disease Events

First Disease event	(n=509)
None	400
Local recurrence	7
Regional recurrence	4
Regional recurrence concurrent with second primary cancer	1
Distant recurrence	57
Second primary cancer	40

Of the 12 patients who had a local or regional recurrence after surgery, 8 had breast conserving surgery with either SLN surgery alone (5 patients) or ALND +/− SLN surgery (3 patients) and 4 had mastectomy with either SLN surgery alone (1 patient) or ALND +/− SLN surgery (3 patients). Pathologic TN categories were: ypT1N0 (2 patients); ypT2N0 (2 patients); ypT2N1–3 (7 patients); and ypT3N0 (1 patient). There were 7 (58.3%) patients with non-zero PEPI scores with a finding of Ki67 ≤10% in the residual tumor; 4 (36.4%) patients with non-zero PEPI scores with a finding of Ki67 >10% in the residual tumor; and 1 (8.3%) patient with a PEPI 0 score. After LRR, 2 of these patients had a distant recurrence, and one had a new primary (Table 3).

TABLE 3:

Characteristics of local or regional recurrences

Pre-NET impression and surgery performed post NET
Prior to NET Surgeon’s Impression Candidate for Lumpectomy
Breast Conserving Surgery performed post NET	6 (50.0%)
Mastectomy performed post NET	1 (8.3%)
Prior to NET Surgeon’s Impression Candidate for Mastectomy
Breast Conserving Surgery performed post NET	2 (16.7%)
Mastectomy performed post NET	3 (25.0%)
Nodal Surgery
SLN only	6 (50.0%)
ALND +/− SLN	6 (50.0%)
Pathologic TN category
ypT1N0	2 (16.7%)
ypT2N0	2 (16.7%)
ypT2N1–3	7 (59.7%)
ypT3N0	1 (8.3%)
PEPI score
zero	1 (8.3%)
Non-zero
Ki67 ≤ 10%	7 (58.3%)
Ki67 >10%	4 (36.4%)

ALND – axillary lymph node dissection; N – nodal; PEPI – preoperative endocrine prognostic index; SLN – sentinel lymph node; T – tumor; yp – pathologic stage following neoadjuvant therapy

DISCUSSION

The data from the ACOSOG Z1031 study shows that the 5-year cumulative incidence rate for a local-regional recurrence is estimated to be 1.53% after 16–18 weeks of an aromatase inhibitor in women with estrogen enriched clinical stage 2 to 3 breast cancer. Approximately, 35% of these women (Cohort B) were known to have week 4 Ki67 ≤ 10%. Although luminal subtypes are generally associated with a favorable prognosis, the rate of LRR in previously published retrospective studies have ranged from 0.8% to 8%. In a group of 55 patients with luminal A breast cancers, 4% of patients experienced LRR after chemotherapy.¹⁶ In another study of 335 patients with clinical stage 2–3 breast cancer who received neoadjuvant chemotherapy followed by BCS, patients were classified according to six molecular subtypes including: luminal A (hormone receptor [HR]+/HER2-/Ki-67 < 15%, n=113), luminal B1 (HR+/HER2-/Ki-67 ≥ 15%, n=33), and, luminal B2. The 5-year LRR rates were luminal A, 3.6%; B1, 6.1%; and B2, 9.7%.¹⁷ It is difficult to compare LRR after NET reported in this study with those observed after NCT due to differences in patient populations and adjuvant treatments administered. Large, prospective studies directly comparing NCT and NET to determine LRR rates remain to be reported.

The low LRR rate in ACOSOG Z1031 is especially noteworthy considering the conversion rate of 50% from mastectomy to BCS. These data should reassure physicians and patients during their decision making that conservative surgery after downstaging due to NET is a safe option. Our unpublished data shows that 24% of the patients that underwent a mastectomy had ypT1 tumors, suggesting that additional patients could have benefitted from BCS based on tumor size. Interestingly, if these patients had undergone BCS, the combined conversion rate would be close to those reported by Dixon, et al. who showed conversion rates after neoadjuvant exemestane, anastrozole, or letrozole to be 80%, 89% and 93%, respectively.¹⁸

Our data also highlights the need for further analysis of biomarkers to predict response to NET. Despite low recurrence rates, NET results in rare pathologic complete response rates in stage 2–3 ER+ breast cancer and thus pCR cannot be used to predict favorable outcome. Our study showed pCR was achieved in only 5 patients (1.0%). This likely reflects the short duration of treatment, however others have shown an equivalent pCR rate.¹⁹ With the relatively low LRR rate, it is necessary to determine which patients can de-escalate treatment and which patients have a risk of LRR justifying further treatment. Of the 12 patients who experienced LRR in our study, one patient had a PEPI score of zero, and of those patients that had a non-zero PEPI score, 7 (58%) had a low (<10%) Ki67 proliferative index following NET. It is therefore necessary to discern molecular characteristics of the tumors that relapse to identify prognostic markers and therapeutic targets for novel agents and combinations to treat patients with predicted lack of sustained response to therapy.

Cohort B in Z1031 was designed to tailor treatment strategy based on real-time Ki67 measurement on tumor biopsies taken at 2–4 weeks following initiation of NET, patients could switch to chemotherapy if the Ki67 remained at >10%. This high Ki67 group was expected to represent patients with resistance to endocrine therapy that would presumably respond better to NCT. However, the data showed that only two of the 35 patients (5.7%) who switched to neoadjuvant chemotherapy experienced a pCR.¹⁵ The ALTERNATE (Alternate Approaches for Clinical Stage II or III Estrogen Receptor Positive Breast Cancer Neoadjuvant Treatment in Postmenopausal Women: A Phase III study) trial built on findings from the Z1031 study to interrogate the use of Ki67 as a tool for decision making in the neoadjuvant setting using 4-week biopsies. The 5-year RFS data from ALTERNATE will provide a prospective validation of the PEPI 0 and modified PEPI 0, (without Allred scoring) as a biomarker.

Another potential strategy to stratify patients for personalized treatment is to use genomic tools. The Oncotype Dx 21 Gene Recurrence Score assay was validated to predict clinical response to neoadjuvant letrozole in postmenopausal patients with ER+, HER2-negative, clinically node-negative breast cancer in the TransNEOS trial, a translational study of New Primary Endocrine-therapy Origination Study. The recurrence score (RS) was significantly associated with rate of BCS after neoadjuvant treatment (RS2E< 18 vs. RS2E≥ 31, p2E= 0.010).²⁰ The MammaPrint / BluePrint assay has also been used to identify patients with a luminal-type tumor, who were then treated with NET. Of the 53 patients with BluePrint luminal tumors who received NET, 36 (68%) had a clinical response. This response rate was due to reclassification of tumors as non-luminal in patients that would have otherwise been in the NET group had they been stratified using conventional assessment methods.²¹ The novel design of the I-SPY2 trial classifies breast cancer into 10 subtypes based on ER, HER2, and MammaPrint scores, which are then used to adaptively randomize patients to control arm or experimental arms.²² Novel drug combinations are already showing success in specific subgroups of patients on this trial.²³ In Z1031, tumor RNA from surgical specimens was subjected to PAM50 analysis; this revealed that 3.3% of patients had post-treatment AI-unresponsive tumors of non-luminal subtypes (human epidermal growth factor receptor 2 enriched or basal-like). This emphasizes the importance of using methods with more sensitivity than is provided by immunohistochemistry to approximate subtype, personalize therapy, and understand whether longitudinal changes in markers are due to tumor evolution or treatment effects. The search for biomarkers to determine optimal treatment strategies based on molecular characteristics is further exacerbated by the emergence of immune checkpoint therapies.

The ultimate goal of any prognostic or predictive biomarker is to elucidate the optimal treatment strategy for each patient, which would include de-escalation of treatment, especially considering the 1.5% local-regional recurrence following NET. We were able to downsize 50% of patients from mastectomy to BCS, comparable to other studies, which should decrease surgical morbidity and lead to improved cosmesis.^{19, 24} Of note, not all patients who were eligible for breast conserving surgery chose that option. Patients were encouraged to undergo at least a SLN biopsy to enable the ascertainment of the patient’s PEPI score. Only 3.5% of patients did not undergo axillary surgery. No residual nodal disease was found in 31 (24.0%) of 29 women with cN1 disease, and 2 (20%) of 10 women with cN2–3 disease.

A weakness of the study is the heterogeneity of treatments following surgery and the combining of cohorts where one is culled to exclude those with a week 4 Ki67 > 10%. The ALTERNATE trial recommends switching to NCT if week 4 Ki67 > 10% and reduces the issue of heterogeneity of treatments following surgery by recommending adjuvant therapeutic strategies based on mPEPI scores. The data from ALTERNATE will also be strengthened by future analysis of survival data once the patients have had longer follow-up.

In summary, our data supports the use of NET to improve breast conservation rates for women who were initially classified as definitively requiring a mastectomy. Local-regional recurrence rates are low in patients receiving NET further supporting this approach. This study also clearly shows that large neoadjuvant endocrine therapy trials are feasible in the cooperative group setting, and that they are needed to address the questions that remain concerning therapeutic approaches for those with large tumors and advanced nodal disease. However, even in the absence of a randomized trial against chemotherapy, NET with an AI is a reasonable standard of care for postmenopausal women with clinical stage 2 or 3 ER enriched breast cancer (Allred score of 6–8) with week 2–4 Ki67 ≤10% who would like a low toxicity option.

Synopsis:

ACOSOG Z1031 (Alliance) was designed to determine which aromatase inhibitor is most effective for postmenopausal women with estrogen receptor–positive stage 2 to 3 breast cancer in the neoadjuvant setting. The local-regional recurrence rate of 509 women is reported here.