Breast cancer is likely to have systemic involvement. However, to the authors' knowledge there are few reports to date regarding clinically detected patterns of metastasis, meticulously studied in regard to the natural history of breast cancer, including unusual sites of metastasis.
Patients treated for invasive breast cancer from April 1983 to May 2007 were retrospectively analyzed. Patterns of clinically apparent tumor recurrence, focusing especially on unusual metastases, were studied as well as possible risk factors for unusual metastases and their influence on survival.
Overall, 3783 patients were eligible for the current analysis. The median duration of follow-up was 5.0 years (range, 0.6 years–20.4 years). Cumulative 5-year and 10-year survival rates were 89.7% and 81.5%, respectively. “Unusual metastasis” was defined as systemic failure with a frequency of ≤1%; in the current series; it was observed in 85 (2.2%) patients %. Of those, 70 (82%) had preceding metastasis in the usual sites. The median duration until the development of usual and unusual metastasis was 2.3 years and 3.6 years, respectively (P < .0001). Among764 patients with distant metastasis, the 5-year cumulative overall survival rate in those with or without unusual metastasis was 53.5% and 53.4 years, respectively (P = .33). No risk factors for unusual metastasis were identified.
Breast cancer has a long natural history and is likely to have systemic involvement. It is most likely to metastasize to bone, the lungs, or the liver, and metastasis can be observed in a variety of other organs and sites. The distribution of organ involvement has been studied in several autopsy series.1–3 However, to our knowledge, there are few studies published to date of clinically detected patterns of metastasis, meticulously studied in regard to the natural history of breast cancer, including unusual sites of metastasis.
This is a retrospective study of patterns of clinically apparent tumor recurrence, focusing especially on unusual metastases after curative treatment, in a large number of Japanese patients with breast cancer. We examined the frequency of metastases at each site as well as possible risk factors. The analysis was also extended to examine whether unusual metastasis itself influences survival outcome.
MATERIALS AND METHODS
From April 1983 to May 2007, a total of 4120 consecutive patients with pathologically confirmed invasive breast cancer were treated at our institutions. Of these, patients with distant metastasis at first treatment and those with a history of previous breast cancer, and men, were excluded from the analysis. Patients with simultaneous bilateral breast cancer were staged according to the more advanced site. All patients were followed for a minimum of 6 months.
The primary treatments consisted of mastectomy and breast-conserving surgery with or without preoperative systemic therapy. Although not particularly common, primary radiotherapy was indicated for selected patients with clinically complete responses after neoadjuvant systemic therapy.
For patients who underwent breast-conserving surgery, the whole breast was irradiated with a total dose of 50 grays (Gy) given in 25 fractions at a rate of 5 fractions per week, with or without an additional third port to the axillary and supraclavicular lymph nodes. A 10-Gy boost to the tumor bed was given if the excised tissue had a positive margin. Radiation to the chest wall and regional lymph nodes was administered to high-risk patients undergoing mastectomy.
Systemic chemotherapy was given depending on patients' risk factors. Until 2003, chemotherapy usually consisted of cyclophosphamide, methotrexate, and 5-fluorouracil (given at doses of 700 mg/m2, 40 mg/m2, and 600 mg/m2, respectively, each given intravenously on Days 1 and 8 every 4 weeks for 6 cycles); methotrexate was replaced by doxorubicin at a dose of 40 to 60 mg/m2 for high-risk patients. After 2003, doxorubicin and cyclophosphamide (at dosages of 60 mg/m2and 600 mg/m2, respectively, every 21 days for 4 cycles) were mainly used with or without weekly paclitaxel (60–80 mg/m2 weekly, for 6–12 weeks), depending on each patient's risk factors. For postmenopausal women with tumors overexpressing hormone receptors, tamoxifen (aromatase inhibitor in later years) was given, either alone or after chemotherapy. For premenopausal, hormone-responsive patients, tamoxifen and a lutenizing hormone-releasing hormone analogue were administered, if indicated.
Patients were followed monthly for 3 months after the completion of initial therapy, andthen every 3 to 6 months thereafter, depending on the type of adjuvant therapy. Routine checkups consisted of blood tests (complete blood count and serum biochemistry, as well as tumor markers), physical examination, annual mammography, chest x-rays, and ultrasound scanning of the breast and regional lymph nodes. Computed tomography, magnetic resonance imaging, bone scintigram, or positron emission tomography was added in the case of suspicious symptoms. All of the data had been prospectively collected by 2 of the authors (A.A. and A.T.) with database software for the corresponding years, and the data were analyzed retrospectively.
Time to any event was counted from the day of the initiation of any treatment for breast cancer. Distant metastasis associated with locoregional recurrence was counted as distant failure. Patients were defined as having a first recurrence when they exhibited the first evidence of recurrent disease (whether in the ipsilateral or the contralateral breast) after primary treatment. Local recurrence was defined as a recurrence in the ipsilateral breast or chest wall; regional recurrence was defined as metastasis to the ipsilateral supraclavicular, infraclavicular, parasternal, or axillary lymph nodes; and distant metastasis was defined as evidence of the tumor in all areas except those described above. Whenever possible, locoregional recurrences were confirmed cytologically or histologically, whereas many metastatic diseases were diagnosed by imaging. Cumulative recurrence rates and overall survival rates were calculated by the Kaplan-Meier method, and statistical significance was calculated with the log-rank test.
“Unusual metastasis” was defined as systemic treatment failure with a frequency of ≤1% in the current series. Chi-square and Fisher exact tests were used to determine statistically significant differences between sites of metastases (Dr. SPSS II, version11.0; SPSS Japan Inc., Tokyo, Japan).
Overall, 3783 patients were eligible for the current analysis. As of November 31, 2007, the median duration of follow-up was 5.0 years (range, 0.6 years–20.4 years) for all patients. The median age was 49 years (range, 22 years–94 years). Systemic chemotherapy was given to 2678 patients (70.8%), and 1408 (37.2%) received hormone therapy. Other characteristics of the study subjects are shown in Table 1. Cumulative 5-year and 10-year survival rates were 89.7% and 81.5%, respectively, with a median survival of 5.6 years (range, 0.5 years–20.7 years). Overall survival and disease-free survival by stage are shown in Figure 1.
Table 1. Patient Characteristics and Treatment Details
No. of Patients
ER indicates estrogen receptor; PgR, progesterone receptor; NOS, not otherwise specified.
Approximately 20% of patients with distant recurrence also demonstrated associated simultaneous local or locoregional treatment failures. Approximately 15% of patients with local recurrence presented with concurrent distant metastases.
The distribution of distant metastases with regard to frequency among all patients are listed in Table 2. With “unusual metastasis” defined as systemic failure with a frequency of ≤1% at each site, unusual metastases involve the central nervous system (meninges, choroidea, optic nerve, extraocular muscle, and pituitary gland), secretory/endocrine organs and glands (thyroid, adrenal gland, and lacrimal gland), internal organs and structures (endobronchus, pancreas, esophagus, stomach, kidney, spleen, peritoneum, and retroperitoneum), and gynecologic organs (uterus, ovary, and placenta).
Table 2. Frequency of Treatment Failure in Patients With Distant Metastasis
Sites of Metastasis
No. of Patients
Lymph nodes (nonregional)
Para-aortic plus retroperitoneal
Overall, unusual metastasis was observed in 85 patients (2.2% of all patients). Among those, 70 (82%) had preceding metastasis in the usual sites, whereas unusual metastasis was the first event in only 15 patients. Median durations until the development of usual and unusual metastases were 2.3 years (range, 0.3 years–11.2 years) and 3.6 years (range, 0.3 years–12.7 years), respectively (P < .0001). Among 764 patients who had distant metastasis, the 5-year cumulative overall survival in those with or without unusual metastasis was 53.5% and 53.4 years, respectively (P = .33) (Fig. 2).
Table 3 gives the frequency of treatment failure among patients with distant metastasis, as compared with 2 autopsy series.
Table 3. Distribution and Frequency of Distant Metastasis
Possible risk factors for developing unusual metastasis were calculated with regard to age, menopausal status, stage, use of chemotherapy, HER-2 status, hormone receptors, lymphovascular invasion, and histological/nuclear grade. However, none of these factors was found to demonstrate a statistically significant correlation with unusual metastasis. In terms of the correlation between histology (invasive lobular vs invasive ductal) and each site of metastasis, meningeal metastasis was more often associated with invasive lobular carcinoma (1.9% vs 0.4%; P = .03), whereas lung metastasis was observed more often with invasive ductal carcinoma (8.8% vs 2.8%; P = .03); no other significant difference was found to exist between the 2 pathologic types.
Survival rates in the current analysis are comparable to those in other reports.4-8 Although more recent antitumor agents such as taxanes or trastuzumab may have the potential to improve outcomes, survival did not appear to be influenced when all subjects were divided into 2 periods (ie, before vs after the new drugs became available) (data not shown). However, definitive comparison is not possible because of differences in follow-up duration.
This retrospective study demonstrated a high frequency of metastases at unusual sites after curative treatment in a large number of Japanese patients with initially nonmetastatic breast cancer (Table 2). Defining “unusual metastasis” as rare systemic failure with a frequency of <1% allowed various sites of unusual metastasis to be identified: the central nervous system (meninges, choroidea, optic nerve, extraocular muscle, and pituitary gland), secretory organs and glands (thyroid, adrenal gland, and lacrimal gland), internal organs and structures (endobronchus, pancreas, esophagus, stomach, kidney, spleen, peritoneum, and retroperitoneum), and gynecologic organs (uterus, ovary, and placenta). Although inguinal or abdominal lymph node metastasis is unusual, metastases at distant lymph nodes were all included in the category of nonregional lymph node metastasis, which accounted for 6.1% of all patients.
As the disease progresses, metastases usually involve >1 site, leading to widespread dissemination. The results of the current study reflect only the clinical behavior of breast cancer dissemination. The frequency of treatment failure among patients with distant metastasis was compared with that of 2 autopsy series, as shown in Table 3. Unusual metastases such as those in the thyroid, adrenal gland, pancreas, kidney, spleen, and ovary were not detected as often clinically as in autopsy series. In contrast, the percentages were similar among common metastases in bone, lung, pleura, liver, and brain.
Because the frequency of metastasis depends on the sensitivity of the diagnostic techniques used, the difference in the frequency of metastasis between clinical and autopsy series can be explained by the low sensitivity of clinical diagnostic methods as well as the lack of symptoms arising from the organs in which metastasis rarely occur.
There are reports that metastases at some specific sites correlate significantly with risk factors. According to Borst and Ingold, who reported the metastatic patterns of 2605 patients, lobular tumors were more likely to metastasize to the peritoneum, adrenal glands, uterus, and pleural surface.9 Other authors have also reported an association between lobular carcinoma and metastasis at unusual sites.10–13 In the current series, however, lobular carcinoma was associated only with an increased risk of having meningeal dissemination. The relatively less frequent population of lobular histology in the current series, 2.9%, may have diluted its impact on metastatic pattern.
Among patients with unusual metastasis, the majority (n = 70; 82%) had preceding metastasis at more usual sites, and the time until the development of unusual metastasis was 1.3 years longer than for usual metastasis. When the survival time was counted from the day of the first metastatic event for those with and without unusual metastasis, they were not found to be statistically different (data not shown). In addition, having unusual metastasis itself did not appear to influence survival (Fig. 2).
Table 4 lists a classification of first reported recurrences in comparison with the National Surgical Adjuvant Breast and Bowel Project trials.7, 8 The pattern of recurrence observed in the current series, which included a broad range of patients, was comparable to that of large prospective trials (Table 4). Further follow-up of the current series is expected to demonstrate that the frequency of distant metastasis or contralateral failure closely approximates the frequency documented in that of such large trials.
Table 4. Classification of Initial Recurrences
NSABP B-04 indicates National Surgical Adjuvant Breast and Bowel Project breast carcinoma trial B-04; NSABP B-06, National Surgical Adjuvant Breast and Bowel Project breast carcinoma trial B-06.
Tumor <4 cm, stage I/II
5.0 y (median)
Local failure only
Contralateral breast failure
This retrospective study examined the frequency of unusual metastases after curative treatment in a large number of Japanese patients with initially nonmetastatic breast cancer. The majority of unusual metastases were found to be associated with prior metastases at more usual sites, which often precede unusual metastasis by approximately 1 year. However, the prognosis of patients with unusual metastases was similar to that of patients with metastasis only at more usual sites. No risk factors for unusual metastasis were identified.