Metastasis to the breast is rare, but it must be considered in the differential diagnosis of a breast mass. The purpose of this study was to identify clinical characteristics and outcomes associated with this entity to identify determinants of survival.
Between 1983 and 1998, 169 patients were confirmed by pathology to have metastasis to the breast from nonbreast solid organ primary tumors at University of Texas M. D. Anderson Cancer Center. Medical records were retrospectively reviewed for clinicopathological characteristics. Survival was determined by Kaplan-Meier analysis.
The median age was 51 years (range, 13–85). One hundred forty-nine (88.2%) patients had a prior history of cancer. Ninety-one (53.9%) patients presented with additional systemic metastases. The most common histology identified was melanoma (65 patients, 38.5%). In most patients (77%), the diagnosis was initially made through physical examination. The median survival from the time the breast metastasis was diagnosed was 10 months (range, 0.4–192.7). On univariate analysis, a significantly better survival was observed in patients who had no evidence of other disease at the time of diagnosis (P = .0036), patients with neuroendocrine tumors (P = .023), and patients who underwent surgical resection for breast metastases (P = .0001). On multivariate analysis, patients who did not have surgery were 88% more likely to die than those who did (P < 0.001).
The breast is rarely the site of metastatic disease. Breast metastases develop most commonly from a primary cancer of the contralateral breast.1 Breast metastases from extramammary malignancies to the breast are even more rare and, therefore, difficult to study in a meaningful manner. The first reported case of extramammary breast metastasis occurred in 1903.2 In the century that has passed, case reports and reviews published in the English-language literature have reported fewer than 500 cases of metastases to the breast.3, 4 Despite its rarity, metastatic disease to the breast is an important diagnostic concern, because its treatment differs greatly from that of primary breast cancer.
We conducted a single-institution retrospective review of patients with pathologically confirmed extramammary metastases to the breast to identify their clinical characteristics and to assess their clinical outcomes in order to determine what factors affect survival. To our knowledge, this is the largest series to date to address this topic.
MATERIALS AND METHODS
The patient population was identified through a search of the database maintained by the Department of Medical Informatics at the University of Texas M. D. Anderson Cancer Center. Between January 1983 and September 1998, 674 patients were registered at M. D. Anderson Cancer Center with a diagnosis of metastasis to the breast. Of these, 373 patients were excluded because of metastasis from a contralateral primary breast neoplasm or because of a diagnosis of hematologic malignancy. Thirty-four patients were omitted because of incomplete information. An additional 98 patients were excluded for the following reasons: 49 were clinically diagnosed with metastases but lacked pathologic confirmation, 19 had a biopsy performed at an outside facility but their pathology was not reviewed at our institution, 16 were diagnosed with an unknown primary, 6 had benign breast pathology, 4 were diagnosed with primary breast cancer, 2 had pathology that was nondiagnostic, 1 was negative for metastasis on pathologic assessment, and 1 was diagnosed with a breast sarcoma. The final study group comprised 169 patients with biopsy-proven metastasis to the breast from another solid primary tumor site, with pathology reviewed and confirmed at M. D. Anderson Cancer Center.
The medical records of the 169 patients were retrospectively reviewed, and the following data were assessed: age, race, sex, breast involvement, cancer history, cancer status at the time of breast metastasis, initial mode of diagnosis, mammographic findings, clinical findings, surgical treatment, surgical specimen pathology, first subsequent distant metastasis, and clinical follow-up. Local and systemic therapies were rendered at the discretion of the treating physicians.
Descriptive statistics were performed to assess frequency distributions in the study population. Overall survival interval probabilities were calculated by the product-limit method of Kaplan and Meier,5 and differences in survival were assessed by the log-rank test. P values ≤.05 were considered to be statistically significant. The Stata statistical software program (StataCorp LP, College Station, Tex) was used for statistical analyses.
Demographic characteristics for the overall study population are summarized in Table 1. The median age was 51 years (range, 13 to 85 years), and the group was predominately women (91.7%). The median follow-up duration was 10 months after diagnosis of metastatic disease to the breast.
Table 1. Demographic Characteristics of Patients Presenting With Metastasis to the Breast
No. of patients (%) N = 169
Prior cancer history
No prior cancer history
Cancer status at presentation
No evidence of disease (NED)
History of metastases, now NED
Intact primary/active disease
Given the rarity of breast metastases, we postulated that breast metastases would usually occur in the setting of advanced disease and widespread systemic metastasis. A total of 149 (88.2%) patients had a documented history of cancer, with metastatic disease appearing in the breast a median of 12 months after the initial diagnosis of cancer (range, 0–269 months). Fifty (29.6%) patients had no evidence of disease at the time of presentation and no history of metastatic disease elsewhere. Eleven (6.5%) patients presented with intact primaries. Seventeen (10.1%) patients had previously experienced metastasis from their primary cancer to another site that had already been treated and were otherwise disease-free at the time they presented with metastatic disease to the breast. Overall, 91 (53.8%) patients presented with other systemic metastases. The median time to diagnosis of additional metastases was 2 months (range, 0 to 58.9 months).
Table 2 summarizes the patients' clinical presentations. The initial mode of detection of metastasis was physical examination in 130 (76.9%) patients. In 144 (85.2%) patients, breast involvement was unilateral (46% right and 39% left). Clinically, 89 (52.7%) patients presented with a solitary mass, 40 (23.7%) had multiple nodules, 26 (15.4%) had skin and/or subcutaneous nodules, 6 (3.6%) had a normal breast examination, 3 (1.8%) had a distinct solitary mass with inflammatory skin changes, 3 (1.8%) had skin nodules with inflammatory skin changes, and 3 (1.2%) had diffuse parenchymal involvement. Mammograms were obtained in 86 patients. The findings are summarized in Table 3.
Table 2. Clinical Presentation of Patients With Metastasis to the Breast
No. of patients (%) N = 169
Initial mode of diagnosis
Unknown mode of diagnosis
Distant solitary mass
Distinct solitary mass with inflammatory skin changes
Skin nodules with inflammatory skin changes
Diffuse parenchymal involvement
Table 3. Mammographic Findings in Patients With Metastasis to the Breast
Only 84 patients had mammography results available.
Distinct solitary mass
The sites of primary tumors that metastasized to the breast are shown in Table 4. The most common primary tumor site was the skin (66 patients, 39.1%), followed by the respiratory system (41 patients, 24.3%). Other tumor primaries included gynecologic, gastrointestinal, head and neck, genitourinary, soft tissue, and other cancers.
Table 4. Primary Tumor Sites for Patients With Metastasis to the Breast
Primary tumor sites
No. of patients (%) N = 169
Head & Neck
The distribution of primary tumor histology is listed in Table 5. The most common primary histology was melanoma, which was found in 65 (38.5%) patients, followed by adenocarcinoma in 42 (24.9%) patients. Other tumor histologies included small cell carcinoma, sarcoma, neuroendocrine tumors, squamous cell carcinoma, and others.
Table 5. Primary Histology of Breast Metastases
No. of patients (%) N = 169
Small cell carcinoma
Squamous cell carcinoma
Only 76 (45.0%) patients had surgery for their breast lesions. Of the 76 patients who underwent surgery, 74 had breast-conserving surgery. A total mastectomy and a modified radical mastectomy were performed in 1 patient each. The total mastectomy was performed for a large metastatic thyroid cancer. The modified radical mastectomy was performed for a breast tumor that was presumed to be poorly differentiated on preoperative core-needle biopsy. Postoperatively, final pathology revealed clear-cell histology, and the patient was found to have a lung mass and mediastinal adenopathy. A retrospective comparison performed by the pathologist determined that the breast lesion was a metastasis from a lung primary.
Not surprisingly, this patient cohort had an extremely poor outcome (Fig. 1). The median duration of survival from diagnosis of breast metastasis was 10 months (range, 0.4 to 192.7 months). At the date of last contact, only 6 (4%) patients were still alive. Two of the 6 continue to be followed at M. D. Anderson.
Determinants of Survival
Univariate analysis was performed to measure the effects of tumor histology, cancer history, disease status, and surgical intervention on overall survival. Disease status at the time the patient presented with metastasis to the breast, whether or not the patient had surgery, and neuroendocrine histology were found to be of prognostic significance (Figs. 2, 3, 4). Patients with neuroendocrine tumors had a median overall survival duration of 18.5 months compared with 9 months in patients whose tumors were of non-neuroendocrine histologies (P = .023). Patients who had surgery had a median survival duration of 15.5 months, whereas their cohorts who did not have surgery survived only a median 8.1 months (P = .0001). Patients with no history of cancer did not have a survival advantage compared with patients with a known history of cancer (P = .22). On multivariate analysis, patients who did not have surgery were 88% more likely to die than those who did (P < .001).
We found that metastatic disease to the breast, although rare, can occur with a myriad of primary cancers. It occurs in an age group at high risk for primary breast cancer and can be misdiagnosed if not considered. Treatment differs greatly from primary breast cancer and, given the poor prognosis of metastatic disease, the treatment should be individualized. In patients with metastatic disease limited to the breast, or with minimal disease burden elsewhere, wide local excision of breast metastases may be considered.
The majority of reports in the literature about metastasis to the breast are isolated case reports, small series involving a dozen or fewer cases from a single institution, larger collective reviews, or meta-analyses of existing cases in the literature.3, 4, 6–12 The largest such collective review, published by Alva and colleagues, identified a total of 431 cases reported in the English-language literature.4 Previously, Hadju and colleagues from Memorial Sloan-Kettering Cancer Center (New York, NY) had reported the largest single-institution experience. They described 51 cases occurring over a 10-year time period.8 Both M. D. Anderson and Memorial Sloan-Kettering Cancer Center benefit from referral bias, which results in these institutions seeing a higher number of unusual manifestations of cancer than other institutions. Furthermore, both institutions aggressively follow patients who have been diagnosed with other primary tumors, resulting in a higher likelihood of detecting these metastases. These 2 factors are likely somewhat responsible for our vast experience. Nonetheless, we believe our series actually underestimates our overall experience, because we excluded patients with nonsolid organ primaries as well as those with clinically evident, but not histologically confirmed, breast metastases.
In prior studies, patients were typically diagnosed by clinical examination after presenting with a palpable breast lump. Although the majority (76.9%) of our patients presented in a similar fashion, 21 (12.4%) patients presented for a mammogram, and metastasis was detected as an incidental finding. With these patients, it may be difficult to distinguish a breast metastasis from a primary breast cancer. Several authors have retrospectively examined mammographic findings in patients subsequently diagnosed with breast metastases.13–15 Most agree that calcifications are exceedingly rare and are seen only in patients with metastatic papillary carcinoma with psammoma bodies.13, 16 Our data reinforce this notion, as we found calcifications on only 2 mammograms. In patients with a known history of cancer, multiple masses or a solitary mass in the absence of calcifications on a mammogram should raise suspicion of metastatic disease to the breast and warrants further evaluation.
Given the rarity of metastases to the breast coupled with the prevalence of primary breast cancer, even in the presence of a past history of an extramammary cancer, a breast mass is likely to represent a new primary breast tumor. Thus, it is recommended that a diagnostic bilateral mammogram and an ultrasound be performed and followed by percutaneous biopsy to establish a tissue diagnosis. Metastases typically do not have an invasive ductal or in situ component when examined microscopically.17, 18 In patients with a history of cancer, it is useful for the pathologist to review slides from the primary tumor and compare the primary histology with that of the suspicious breast lesion. This is particularly true in cases diagnosed by fine-needle aspiration or those in which there is unusual histology.19 Armed with histology from a previous cancer, the pathologist can often differentiate a primary breast tumor from metastasis to the breast.
Immunohistochemistry can be very valuable when trying to differentiate between a primary cancer originating in the breast and a metastasis to the breast.20 The most common invasive breast cancer is invasive ductal carcinoma followed by invasive lobular carcinoma. Lobular carcinomas lack E-cadherin protein expression, whereas ductal carcinomas are generally E-cadherin positive.21–23 Additional immunohistochemical (IHC) analyses often used in primary breast cancer include the Ki-67 cell proliferation marker, estrogen and progesterone hormone receptors, and the HER-2/neu oncoprotein.
There are a multitude of IHC stains available to aide in tissue diagnosis of tumors originating in other sites. For the primary histologies identified in our study, the following IHC stains are commonly used: CA-125 (breast carcinoma, ovarian cancer, lung cancer, and gastrointestinal cancer), CA-19-9 (pancreatic cancer, but it can also be elevated in breast cancer, uterine cancer, cancers of the liver and lung, uterine cancer, and mucinous ovarian cancers), calcitonin (thyroid carcinoma), carcinoembryonic antigen (colorectal cancer, but it may also be elevated in breast, lung, pancreatic, and gastric cancers), chromogranin-A (neuroendocrine tumors), cytokeratins (carcinomas), desmin (soft tissue tumors), epithelial membrane antigen (carcinomas), HMB-45 (melanoma), MART-1 (melanoma), myoglobin (rhabdomyosarcoma), renal cell carcinoma antigen (renal cell carcinoma), S-100 antigen (melanoma, neuroendocrine tumors), synaptophysin (neuroendocrine tumors), thyroglobulin (thyroid cancer), thyroid transcription factor-1 (lung and thyroid cancer), and vimentin (soft tissue tumors).24–28 Although no marker should be considered 100% specific, these markers, along with comparison of pathology from the primary tumor and the suspicious breast lesion, can help to differentiate between a primary cancer and metastasis to the breast. In the near future, newer technologies, such as gene-expression profiling, may help further elucidate the tissue of origin in patients with an unknown primary.
Other series have shown metastatic melanoma to be the most common solid-organ malignancy to metastasize to the breast.4, 17, 29–31 More than 20% of cases of breast metastases reported to date were from melanoma, the largest single histologic variant.5 Similarly, we found melanoma to be the most common histology, although at nearly double the proportion previously reported (38.5% of all cases). This disproportionate experience may be attributable to 2 factors: 1) referral bias, as M. D. Anderson is the regional center of excellence in melanoma care, and 2) geographic bias, because of our location in the sunny South. Currently, the incidence of melanoma is rising faster than that of any other malignancy, and this is especially true in the South and in young women.32 In that context, our data highlight the importance of assessing a patient's cancer history when making a diagnosis on presentation of a breast lump. Although, our patients typically presented with breast metastases within 16 months of diagnosis of primary cancer, previous authors have found delayed presentation of metastases in excess of 10 years after diagnosis.17, 29, 33 We identified 11 patients presenting more than 10 years after diagnosis of their primary. This knowledge may guide diagnostic and therapeutic decisions for older women and those in high-risk categories who have had a previous cancer diagnosis.
Breast metastasis is associated with an extremely poor prognosis. We found that patients had median survival duration of 10 months from the time of diagnosis of their breast metastasis. Despite treatment advances over the years, our outcomes are comparable to those reported previously, with median survivals in the range of 8–10 months, reflecting the biology of disease.7, 14, 33, 34 Most patients, in our series and those reported historically, presented with a breast lump as their chief complaint but also had concomitant widespread systemic metastases. At the date of last follow-up, only 6 of our patients were still alive.
Most prior reviews are too small and the overall outcomes are too poor to provide a meaningfully analysis of survival in this patient cohort. We are the first to analyze a large population of patients with breast metastases with the goal of identifying determinants of survival. We found that neuroendocrine histology, disease status, and surgical intervention were all significant predictors of survival on univariate analysis. Among breast metastases, neuroendocrine tumors are rare and are often misdiagnosed as primary breast cancer.35 Metastasis of neuroendocrine tumors to other sites has been associated with a relatively more favorable prognosis. Patients with neuroendocrine tumors, even those with liver metastases, can have good long-term survival if they are treated with aggressive surgery to remove all sites of tumor.36, 37 In fact, outcomes in these patients are so favorable that some centers will perform an orthotopic liver transplant in those with isolated neuroendocrine metastases to the liver.38 It may be especially worth considering more aggressive local therapy in patients with breast metastases from neuroendocrine tumors.
In our study, patients who underwent surgery did have significantly better overall survival than those who did not. Although this also held true on multivariate analysis, we still cannot rule out selection bias. It is likely that patients with clearly advanced disease or those in poor health who had had a diagnostic study were spared surgery. In addition, disease status likely reflects the inherent survival advantage of patients with favorable histology who have no evidence of disease at presentation. Further data are needed to determine whether surgical resection of breast metastases in the setting of limited metastatic disease may result in a survival advantage.
Metastasis to the breast is rare. It can mimic primary breast cancer and, thus, must be considered in the differential diagnosis of a suspicious breast mass, especially if the patient has a history of cancer. This entity carries a poor prognosis, and local therapy should be individualized. Further work is needed to confirm whether, in selected patients, surgical resection of metastasis to the breast may confer a survival advantage.
We thank Marlen Banda and Laura Gardner for assistance in manuscript preparation, Dr. DeLinda D. Wills for assistance in data collection, and Dr. Nadeem Mirza for data entry.