Sentinel lymph node dissection (SLND) is the standard method of nodal staging in patients with clinically node-negative breast cancer. Until the publication of the American College of Surgeons Oncology Group (ACOSOG) Z0011 trial results, completion axillary lymph node dissection (ALND) was recommended when the sentinel lymph node (SLN) demonstrated metastatic carcinoma. The ACOSOG Z0011 randomized trial was designed to determine whether SLND alone was not inferior to completion ALND in patients with clinical T1–T2, N0 breast cancer found to have one or two positive SLNs. All patients in the Z0011 trial underwent breast conserving therapy (BCT), including lumpectomy and whole breast irradiation. The primary endpoint was overall survival (OS): at a median follow-up of 6.3 years, there was no difference in OS between the ALND and SLND arms (91.8% vs. 92.5%, respectively). Local-regional recurrence (LRR) was a secondary endpoint, and again, no differences were seen between the arms. Local recurrence rates were 3.6% in the ALND arm and 1.8% in the SLND arm, whereas ipsilateral axillary recurrences occurred in 0.5% of patients in the ALND arm and 0.9% of patients in the SLND arm [1], [2].

Invasive lobular carcinoma (ILC) is the second most common histologic type of invasive mammary carcinoma, comprising 5%–15% of all invasive breast carcinomas [3]. Practically, ILC is not treated differently from invasive ductal carcinoma (IDC); however, ILC has several unique features. Most ILCs are well differentiated and estrogen receptor (ER) positive; they tend to have a multifocal, multicentric and bilateral distribution; and they often have a dispersed growth pattern both in the breast and at metastatic sites including the axillary lymph nodes. Detection of nodal disease sometimes requires immunohistochemical staining for cytokeratin for identification. ³ In addition, ILC patients tend to be at risk for distant recurrence for more than 5–10 years [4].

Only 63 (7%) of the 856 patients in ACOSOG Z0011 had ILC; thus, it is unclear whether the trial's results are applicable to such patients. The objective of the current study was to determine whether the Z0011 trial's results could be safely applied to ILC patients. We used the Surveillance Epidemiology and End Results (SEER) database to identify ILC patients who met the ACOSOG Z0011 eligibility criteria. Patients were evaluated on the basis of the extent of axillary surgery (SLND alone or ALND), and the clinical outcomes of these 2 groups were compared.

Breast conserving therapy including lumpectomy and whole breast irradiation has become widely used in the treatment of early-stage breast cancers. Although there was initial concern about treating ILC with BCT because of its tendency to be multifocal and multicentric, available data confirm that BCT is as effective for ILC as for IDC. There is no difference in the reported LRR or OS rates between ILC and IDC after BCT [6], [7]. The results of the ACOSOG Z0011 trial will likely lead to further reductions in the extent of surgery for early-stage invasive breast cancers; in particular, ALND can be omitted in patients undergoing BCT for clinical T1–T2, N0 breast cancer found to have 1 or 2 positive SLNs. Because only 7% of participants in the ACOSOG Z0011 trial had ILC, we used SEER data to confirm that the Z0011 trial results are applicable to ILC patients.

The primary endpoint of the ACOSOG Z0011 study was OS; therefore, it is important that the current study revealed no differences in OS between patients with ILC who underwent SLND and those who underwent ALND. This is in part due to the fact that these patients' tumors had very favorable biological characteristics: most were ER+ and HER2 negative, and thus likely of the molecular luminal A subtype. These patients are generally treated with endocrine therapy for five years. Although data regarding systemic therapy is not available in the SEER database, given the years included in this analysis (1998–2009), most patients were likely treated with adjuvant endocrine therapy.

A secondary endpoint of the ACOSOG Z0011 study was LRR. In that trial, after a median follow-up of 6.3 years, the local recurrence rates were1.8% and 3.6% in the SLND and ALND arms, respective [2]. The regional recurrence rates were 0.9% and 0.5% in the SLND and ALND arms. In the current study, we found that after a median follow-up duration of 73 months, no LRR was reported in the SLND group, and only 3 IBTRs (0.34%) and 1 (0.11%) regional recurrence was reported in the ALND group. Of note, the SEER database does not specifically report LRR data; therefore, we used surrogates. Specifically, we identified patients with 2 or more record entries after the primary surgery. If the same breast was affected it was counted as an IBTR and if lymph nodes were affected, it was counted as an ipsilateral regional recurrence. This likely underestimates the LRR risk. Despite this, the LRR rates in this population may be low regardless of the extent of axillary surgery performed, partly because of the overall favorable biological characteristics of ILC, which is predominantly ER+, luminal A type. In the current study, 90.6% of patients had ER+ tumors versus 74.8% in the ACOSOG Z0011 study (with ER status unknown in 9.5%) [1]. Our results are consistent with those of a recent study by Arvold et al that found a LRR rate of 0.8% in patients with luminal A breast cancer approximated as hormone receptor (HR)-positive, HER2-negative, grade 1–2 after BCT compared to 2.3% in luminal B (HR+, HER2−, grade 3), 1.1% in HER2+ luminal B (HR+, HER2+), 10.8% in HER2-enriched (HR−, HER2+) and 6.7% in basal (HR−, HER2−) cancers [8].

This study has several limitations. First, in contrast to the ACOSOG Z0011 study which was a randomized trial, our study was a retrospective review of a large, population-based database. There may have been bias with respect to which patients underwent SLND alone in that surgeons selected patients for this limited axillary surgery on the basis of perceived favorable biologic characteristics. This is consistent with previously published data from the SEER database showing a trend towards omitting ALND in selected patients, specifically, older women with low-grade, ER-positive tumors [9]. Similarly, a review of National Cancer Data Base (NCDB) data revealed a trend towards omitting ALND in patients with micrometastases in the SLN [10]. A patient selection bias was also present in the ACOSOG Z0011 trial. In a study evaluating factors influencing participation in the trial, Leitch et al. reported that 69% of SLN positive patients who were eligible to participate but did not enroll and underwent an ALND instead. This bias contributed to the overall very favorable characteristics of patients enrolled in the ACOSOG Z0011 trial [11]. Although there were some differences between the patients reported in the current study and those in the ACOSOG Z0011 trial—slightly fewer T1 tumors (64.7% vs. 68.6%) and more ER-positive tumors (90.6% vs. 74.8%)—both studies included patients with very favorable biologic characteristics.

A second limitation is that the SEER database does not specify whether a patient underwent SLND alone or ALND. We therefore used surrogates; patients with 5 or fewer lymph nodes removed were categorized as having undergone SLND alone, whereas patients with >5 lymph nodes removed were categorized as having undergone ALND. There may be a concern that using 5 or fewer nodes removed as a surrogate for SLND may potentially lead to poorer prognosis due to insufficient axillary node dissection. In contrast, using 6 and more nodes removed as a surrogate for ALND may potentially lead to better prognosis. Nevertheless, the analyses from the SEER database showed no difference in OS or DSS in the SLND and ALND groups using the above surrogates, despite the potential survival difference caused by arbitrary subgrouping, indicating that it may be safe to perform SLND in early stage (clinical T1 and T2, N0) ILC. These surrogates are consistent with the AJCC definition of an ALND and were used in the above referenced study of the National Cancer Data Base dataset.

A third limitation is that the SEER database does not provide specific data on the use of radiation therapy. An important aspect of the ACOSOG Z0011 trial is that all patients underwent opposing tangential field whole breast irradiation. With these tangents, 51% of level I and 26% of level II axillary lymph nodes receive 95% of the prescribed dose [12], radiation may have contributed to the favorable local-regional control demonstrated in the ACOSOG Z0011 trial. Thus, the ACOSOG Z0011 data are only applicable to patients undergoing BCT with whole breast irradiation and should not be applied to those underwent accelerated partial-breast irradiation (APBI) or radiation administered in the prone position [13]. With respect to APBI, the American Society of Radiation Oncology (ASTRO) consensus statement recommends that patients with ILC be included in a “cautionary” group with respect to considering APBI based on randomized clinical trial data showing a higher IBTR risk in patients with ILC compared to IDC when treated with APBI [14]–[16]. The SEER database does not provide specific information regarding the modality by which radiation was administered; however, given the years of the study, the relatively newness of accelerated partial-breast irradiation, and the identification of ILC patients as a cautionary group by by the American Society of Radiation Oncology (ASTRO), we believe that most patients in the current study would have undergone standard whole breast irradiation administered in the supine position.

Finally, our study population was restricted to patients with pathologic T1 and T2 tumors, whereas the ACOSOG Z0011 study enrolled patients with clinical T1 and T2 tumors. Although all patients in the ACOSOG Z0011 trial had clinical T1 or T2 tumors, the median tumor size and range indicated a percentage of patients had pathologic T3 disease (the median tumor size in the ALND arm was 1.7 cm; range 0.4–7.0) [1]. ILC has a diffuse growth pattern and rarely forms a mass lesion; therefore, clinical measurement of the tumor size by palpation or radiographic evaluation may underestimate the disease extent. Caution is advised regarding application of the ACOSOG Z0011 data to patients with clinical T1 or T2 ILC that has larger extent identified on pathologic evaluation.

In conclusion, the results of current study demonstrate that among patients with T1–T2 ILC and low volume nodal metastasis, there is no difference in LRR, DSS or OS for patients undergoing SLND alone compared to those undergoing ALND. These findings suggest that, consistent with the findings from the ACOSOG Z0011 trial, SLND alone, without completion ALND may result in comparable outcomes for patients with early-stage ILC. However, given the limitations of this retrospective cohort study, further investigation is warranted to validate these findings.