Occult breast cancer (OBC), which generally presents as axillary metastases without clinical or radiologic evidence of a primary breast tumor, accounts for only 0.1% of all breast cancer cases. Despite ongoing advances in diagnostic breast imaging, the management of OBC remains a clinical challenge. The majority of the literature on OBC consists of small, single-institution series and case reports. Most studies show the importance of local–regional (LR) treatment in OBC and, to date, have demonstrated similar outcomes for mastectomy or breast conservation with radiation therapy (RT).1,2

Few published guidelines address the treatment of OBC, and there is no clear consensus on the optimal management of these patients. The current National Comprehensive Cancer Network (NCCN) guidelines emphasize the diagnostic importance of magnetic resonance imaging (MRI) and suggest either mastectomy with axillary lymph node dissection (ALND) with or without RT, or ALND with RT.3 The American Society of Breast Surgeons lists OBC as an indication for ALND.4 In this study, we utilized a large patient registry database (National Cancer Database [NCDB]) to examine factors associated with LR treatments and their respective outcomes in patients with OBC, with a primary focus on comparing breast-directed treatment with RT versus mastectomy in patients who underwent ALND.

Methods

Study Design

The NCDB was utilized for this study. The NCDB is a clinical oncology database sponsored by the American College of Surgeons and the American Cancer Society. Information in this database includes patient demographics, tumor characteristics, treatment facilities, types of treatment, and outcomes, and is collected from the cancer registries of over 1500 Commission on Cancer-accredited facilities. The NCDB contains information on approximately 70% of all newly diagnosed malignancies in the US.5 Institutional Review Board approval was not required for this study since no protected health information was utilized.

Female patients with clinical T0 N1/2 M0 breast cancer were selected from the NCDB (2004–2013). Patients were categorized into four treatment groups: MAST = mastectomy + ALND ± RT; RT + ALND = no breast surgery + ALND + adjuvant RT; ALND = no breast surgery + no RT + ALND; OBS = no breast surgery or RT or ALND. Following Walker et al.,1 ALND was considered regional lymph node (LN) surgery with the removal of four or more nodes. Patients who received neoadjuvant chemotherapy were included, whereas patients were excluded from analysis if any partial breast or unspecified/unknown breast surgery was performed, the clinical N stage was 2B (internal mammary lymph node involvement without axillary node involvement), no breast surgery was performed and RT was not documented, regional LN surgery was not documented for ALND groups, an unknown number of LNs were removed, and mastectomy or RT was performed with removal of fewer than four nodes.

Variables examined included age, comorbid conditions (Charlson/Deyo comorbidity score [CDCC]), income, insurance status, race, patient location, clinical and pathologic N stage, number of LNs examined, histology, laterality of tumor, estrogen receptor (ER) and progesterone receptor (PR) status, use of chemotherapy and endocrine therapy, timing of chemotherapy, use of RT, facility type, facility location, distance from the treatment facility, and year of diagnosis. Multivariable analysis (MVA) included the following variables: age (continuous variable), CDCC score, income (<US$48 K vs. ≥US$48 K), insurance status (private vs. other), race (White vs. non-White), patient location (urban vs. other), clinical N stage (1 vs. 2), pathologic N stage (0, 1, 2, or 3), number of LNs examined (<10, 10–19, ≥20), ER status (positive vs. negative), use of chemotherapy (yes vs. no), facility type (academic vs. other), and distance from the treatment facility (<50 miles vs. ≥50 miles). Facility location was divided into four regions and analyzed as a categorical variable: Northeast (CT, MA, ME, NH, NJ, NY, PA, RI, VT), South (AL, AR, DC, DE, FL, GA, KY, LA, MD, MS, NC, OK, SC, TX, VA, WV), Midwest (IA, IL, IN, KS, MI, MN, MO, ND, NE, OH, SD, WI), and West (AK, AZ, CA, CO, HI, ID, MT, NM, NV, OR, UT, WA, WY).

Statistical Analysis

All statistical analyses were performed using SPSS statistical software version 22 (SPSS for Windows, IBM Corporation, Armonk, NY, USA). Patient, tumor, and treatment characteristics were compared between treatment groups using a χ 2 test. Binary logistic regression was used for MVA to identify factors associated with treatment with MAST or RT + ALND among age, CDCC score, income, insurance, race, patient location, clinical N stage, ER status, facility type, facility location, and distance from the treatment facility. Cox regression MVA considering age, CDCC score, race, clinical and pathologic N stage, number of LNs examined, use of chemotherapy, ER status, facility type, facility location, and treatment with either MAST or RT + ALND or MAST ± RT as covariates was performed to identify factors associated with overall survival (OS). Kaplan–Meier estimates and a log-rank test were used to examine the impact of treatment type on OS. A two-tailed p value of <0.05 was considered statistically significant.

Results

Patient, Tumor and Treatment Characteristics

All Treatment Groups

From 2004 to 2013, 2.03 million BC cases were reported to the NCDB, and 1853 female patients with clinical T0 N1/2 M0 disease (0.09%) were identified. There were 1231 patients who met our inclusion criteria. Overall, the majority of patients were White (76%), had private insurance (55%), and had no comorbid conditions (85%). ER-positive tumors (58%) and clinical N1 disease (75%) were more common. Chemotherapy was administered to 77% of patients and endocrine therapy was administered to 46% of patients. The total number of patients and median age by treatment group were as follows: MAST: n = 592, 58 years; RT + ALND: n = 342, 60 years; ALND: n = 106, 63.5 years; and OBS, n = 191, 65 years. Patient, tumor and treatment characteristics for all treatment groups are listed in electronic supplementary Table 1.

When comparing all four treatment groups, significant differences were observed in age, insurance, clinical and pathologic N stage, use of chemotherapy and endocrine therapy, and facility type (electronic supplementary Table 1). Younger patients were more likely to be treated with MAST, whereas older patients were more often treated with OBS. A higher percentage of patients with clinical N1 disease were treated with OBS. The use of chemotherapy and endocrine therapy were more common in patients treated with MAST or RT + ALND compared with ALND or OBS. Patients were less likely to be treated with MAST if they received care at an academic center.

Treatment with MAST Versus Radiation Therapy (RT) + Axillary Lymph Node Dissection (ALND)

Significant differences were also observed between treatment with MAST and RT + ALND (Tables 1, 2). Patients were more likely to be treated with RT + ALND than MAST if they received treatment at an academic center (39.5 vs. 25.2%, p < 0.001), lived in an urban location (76.0 vs. 68.9%, p = 0.014), or were treated in the Northeast (20.5 vs. 13.5%, p = 0.041). MAST was more common in patients treated in the South (34.6 vs. 29.5%, p = 0.041), in those who had PR-positive tumors (46.3 vs. 37.7%, p = 0.017), and in younger patients (age <40 years; 6.8 vs. 2.9%, p = 0.043). The vast majority of MAST (N = 451) and RT + ALND (N = 280) patients who received chemotherapy also received multiagent therapy (95.3 vs. 97.2%, p = 0.198). On MVA, the only factor associated with type of treatment was care at an academic center, with a higher likelihood of treatment with RT + ALND (odds ratio 2.03, 95% confidence interval [CI] 1.50–2.74, p < 0.001).

Table 1 Patient and tumor characteristics by treatment group
Full size table
Table 2 Treatment characteristics by group
Full size table

Use of RT in MAST Patients

RT was documented in 571 patients treated with MAST, with 323 (56.6%) receiving RT. Patients were more likely to receive RT if they had a higher clinical N stage (cN2, 33.1 vs. 14.1%, p < 0.0001) and pathologic N stage (pN2/3, 56.3 vs. 26.3%, p < 0.0001). Patients who received RT were also more likely to receive chemotherapy (62.9 vs. 37.1%, p < 0.0001) and endocrine therapy (60.7 vs. 39.3%, p = 0.029). The groups were similar in terms of CDCC score, income, insurance, race, patient location, ER and PR status, facility type, facility location, and distance from the treatment facility.

Survival Analysis

All Treatment Groups

Univariate survival analysis comparing all four treatment groups showed that patients treated with MAST or RT + ALND had better OS compared with patients treated with ALND or OBS. The 5- and 8-year estimates of OS for each of the treatment groups were MAST 80.0 ± 2.2 and 72.8 ± 3.2%; RT + ALND 90.8 ± 1.9 and 84.7 ± 3.6%; ALND 76.2 ± 5.0 and 65.1 ± 6.7%; OBS 56.5 ± 4.8 and 49.0 ± 5.9%, respectively.

Treatment with MAST versus RT + ALND

Univariate survival analysis showed that patients treated with RT + ALND had a significantly better OS compared with patients treated with MAST (Fig. 1; hazard ratio [HR] 0.475, 95% CI 0.306–0.736, p = 0.001). Multivariable survival analysis showed that RT + ALND was independently associated with OS (HR 0.509, 95% CI 0.321–0.808, p = 0.004), in addition to age, CDCC score, pathologic N stage, number of LNs examined, and ER status (Table 3).

Fig. 1
figure 1

Kaplan–Meier curve illustrating overall survival for patients with occult breast cancer treated with RT + ALND and MAST. RT radiation therapy, ALND axillary lymph node dissection, MAST mastectomy with ALND ± RT, HR hazard ratio

Full size image
Table 3 Multivariate Cox regression for overall survival in patients treated with MAST versus RT + ALND
Full size table

When we limited our ALND inclusion criteria to those patients who had ≥10 axillary LNs removed (total patients 738; MAST = 463, RT + ALND = 275), significantly better OS was still shown on univariate survival analysis in patients treated with RT + ALND compared with MAST (HR 0.475, 95% CI 0.288–0.786, p = 0.004), and treatment with RT + ALND was independently associated with OS on multivariable survival analysis (HR 0.591, 95% CI 0.349–0.998, p = 0.049).

Similarly, when patients who received neoadjuvant chemotherapy were excluded (MAST = 185, RT + ALND = 11), a significantly better OS was observed in patients treated with RT + ALND compared with MAST on univariate survival analysis (HR 0.466, 95% CI 0.293–0.743, p = 0.001), and, on multivariable survival analysis, treatment with RT + ALND was independently associated with OS (HR 0.512, 95% CI 0.310–0.847, p = 0.009).

Treatment with MAST (±RT)

No significant difference in OS was shown on univariate survival analysis for patients treated with MAST with and without RT (HR 0.995, 95% CI 0.648–1.528, p = 0.982). Multivariable survival analysis showed that the use of RT was not independently associated with OS in patients treated with MAST; however, age, CDCC score, pathologic N stage, number of LNs examined, and ER status were all independently associated with OS.

Pathologic Findings on Mastectomy

Of the 592 MAST patients, 397 (67%) had primary tumors that were classified as T0 or TX on final pathology. However, tumors were identified in the breast after mastectomy in 195 patients (33%). The majority of these were in situ or T1 tumors (n = 124, 63.6%), while 71 patients (36.4%) had tumors that were T2 or greater.

Discussion

In this NCDB analysis examining treatment and outcomes in patients with OBC, we show that patients treated with RT + ALND had a significant improvement in OS compared with patients treated with MAST after adjusting for other covariates. Other factors that typically influence outcomes in patients with breast cancer, including hormone receptor status and pathologic N stage, were also independently associated with OS in patients treated with MAST or RT + ALND. This analysis lends support to the use of RT + ALND as a treatment choice for OBC. This study also reinforces the importance of LR therapy in OBC, as definitive LR treatment with MAST or RT + ALND conferred a survival advantage over treatment with either ALND or OBS.

Our analysis is consistent with existing data in terms of the importance of LR therapy.1,2 Multiple single-institutional studies have shown a clear advantage of RT and/or mastectomy with ALND in the management of OBC compared with less aggressive LR treatment approaches.6–10 In a recent meta-analysis that evaluated therapeutic options for OBC, LR recurrence rates were decreased and there was a trend toward improvement in mortality with the addition of RT to ALND compared with ALND alone.2 A population-based analysis of OBC using the Surveillance, Epidemiology, and End Results (SEER) database showed that LR treatment with either mastectomy + ALND ± RT or breast-conserving therapy (BCT) + RT + ALND resulted in an improvement in survival compared with ALND alone or observation.1 Both of these studies showed similar survival when comparing patients treated with mastectomy or RT, which is in contrast to the results of our study.

The use of adjuvant RT in patients with breast cancer as part of multimodal treatment has increased as studies have shown improvements in local control and breast cancer survival in patients treated with breast conservation or mastectomy with positive lymph nodes.11,12 These studies highlight the importance of RT in node-positive patients, regardless of local therapy with mastectomy or BCT. In the present study, the addition of RT to ALND resulted in a better OS compared with ALND alone. In patients who underwent mastectomy, the use of adjuvant RT did not improve OS and was not independently associated with OS on MVA. These results are likely due to patient selection since patients treated with postmastectomy RT had higher clinical and pathologic N stage.

One major obstacle when optimizing therapeutic management of OBC is the low incidence of the disease, which makes the conduct of randomized controlled trials challenging. As a result, no consensus exists on the optimal LR therapy. A survey conducted by the American Society of Breast Surgeons on the management of patients with OBC highlights this lack of consensus, with 43% of surgeons preferring mastectomy and 37% recommending whole-breast irradiation.13

Historically, modified radical mastectomy was the recommended treatment for patients with OBC. This was likely driven by the high percentage of patients with positive pathology in the breast after mastectomy and high local recurrence rates when no treatment was administered to the breast.7,8,14 However, advanced imaging techniques, especially breast MRI, have significantly improved our ability to identify OBCs that are not detected on mammography.15–18 A study by Olson et al. from Memorial Sloan Kettering Cancer Center demonstrated that breast MRI improved breast cancer detection over mammography and physical examination and that a negative MRI predicted a low tumor yield at mastectomy.15 More recent studies have demonstrated positive pathology in the breast at mastectomy in approximately 30% of OBC patients.6,10,16 This is consistent with our findings of 33% and may reflect the increased use of MRI in these patients.

As with other large database analyses, our study was limited to the reported data, leaving several questions unanswered. For example, use of breast MRI imaging is not included in the NCDB, therefore we were unable to evaluate what percentage of patients had baseline assessment with breast MRI and whether this correlated inversely with positive pathology on mastectomy. Since human epidermal growth factor receptor 2 (HER2) has only been recently reported in the NCDB, we did not evaluate this as a prognostic or treatment factor. Likewise, the use of neoadjuvant chemotherapy was not well documented in patients treated with RT + ALND. Finally, since information on LR recurrences is also not available to researchers, we were unable to evaluate the impact of LR treatment on this outcome.

Conclusion

The results of our study support the use of RT + ALND as a treatment option for patients with OBC, and emphasize the importance of LR treatment and multimodality therapy in these patients.