Comparative Analysis of Bone Marrow Micrometastases with Sentinel Lymph Node Status in Early-Stage Breast Cancer
- First Online:
- Cite this article as:
- Saha, S., Ali, S., Ghanem, M. et al. Ann Surg Oncol (2009) 16: 276. doi:10.1245/s10434-008-0244-0
- 105 Views
Bone marrow micrometastases (BMM) and sentinel lymph node (SLN) status are both prognostic factors in breast cancer (BRCa) patients (pts). A definitive relationship between the two has not yet been proven and the data available is controversial. Thus, a retrospective study was conducted to determine the relationship of BM status and SLN status in pts with early BRCa (T1/T2). All female pts with early BRCa (T1/T2) operated upon by a single surgeon were included in the study. Prior to surgery, all pts underwent bone marrow aspiration from the posterior superior iliac spine bilaterally. Subsequently, pts underwent SLN biopsy and definitive primary breast surgery. BM samples were examined by using a Cytokeratin Detection Kit using CAM 5.2 monoclonal antibody. All pts with BMM underwent repeat BM analysis 6 months after completing all treatments. Data was collected for SLN, BM, estrogen receptor/progesterone receptor (ER/PR), and human epidermal growth factor receptor 2 (Her-2/neu) status and analyzed using chi-square (χ2) analysis or Fischer’s exact test. A total of 270 consecutive pts with early BRCa were studied. SLN mapping was successful in all pts. SLN metastases (mets) were detected in 28.9% (78/270) pts. Of the 270 pts, 77.0% (208/270) had T1 disease. BMM were detected in 9.6% (26/270) pts, of whom 69.2% (18/26) were found to have BMM unilaterally. BMM were detected in 11.5% (9/78) pts with SLN mets versus 8.9% (17/192) in pts with node-negative disease (p = 0.65). Of the pts with T1 BRCa, BMM were observed in 9.1% (19/208) pts versus 11.3% (7/62) in pts with T2 BRCa (p = 0.6). In pts with ER/PR-negative (−ve) BRCa, BMM were found in 7.7% (2/26) pts versus 9.9% (24/242) in pts with ER/PR-positive (+ve) BRCa (p = 0.27). BMM were detected in 12.3% (9/73) pts with Her-2/neu +ve BRCa and in 8.6% (16/187) pts with Her-2/neu −ve BRCa (p = 0.11). After completion of adjuvant therapy all pts with BMM (n = 26) converted to BM negative status. We conclude that BM status did not correlate with SLN status and occurs independently of lymphatic metastasis possibly through a different mechanism. BMM occur in node-negative pts and may assist in identifying pts at high risk for disease recurrence. Obtaining bone marrow aspirate from two locations resulted in a significant increase in detection of micrometastases.
We report on early breast cancer and show that bone marrow micrometastases (BMM) often occur independently of sentinel node metastases. These findings suggest that BMM could potentially be used to further individualize treatment in selected node-negative patients.
Breast cancer (BRCa) remains the most common malignancy in females and is the second leading cause of death from all cancers in females in the USA, with an estimated 182,460 new cases to be diagnosed in the year 2008.1 Despite advances in treatment, approximately 30% of patients with node-negative breast cancer will die of the disease. Accurately identifying these high-risk patients in the past has been difficult, and subsequently the use of systemic therapy has become more widespread. More recently, two multigene assays (i.e., Oncotype DX®, Genomic Health, Redwood City, CA, USA and MammaPrint®, Agendia, Amsterdam, The Netherlands) have been used to assess the likelihood of distant recurrence and benefit from therapy in an effort to individualize breast cancer treatment.2–4 Although there have been advances in assessing an individual patient’s quantitative risk of distant recurrence, challenges still remain when attempting to predict the biology of an individual tumor. Current research has attempted to identify micrometastases in the lymphatics, circulating peripheral blood (i.e., circulating tumor cells, CTCs) and bone marrow (i.e., disseminated tumor cells, DTCs) to further understand the biology of this disease and individualize patient care. Micrometastases would be of particular importance if identified in node-negative patients with other favorable features as these patients may benefit from additional treatment options.
Randomized trials with 20-year follow-up have shown that long-term survival of breast cancer patients is determined by factors other than local-regional treatment.5,6 Braun et al. reported that presence of bone marrow micrometastases (BMM) at time of diagnosis is associated with poor prognosis and is an independent prognostic indicator for both disease-free and overall survival. Braun also reported that BMM is associated with lymph node metastasis.7,8 In a multicenter study of 410 early-stage breast cancer patients, Langer and colleagues report that presence of BMM is associated with sentinel lymph node metastasis.9 However, the relationship between sentinel lymph node status and BMM has remained controversial. In a retrospective analysis of 124 patients, Trocciola and colleagues suggest that BMM and axillary lymph node metastases are not concordant findings in most patients.10 Although the largest studies show an association, all studies have identified a significant number of BMM in node-negative patients, and some of these patients have had small tumors with favorable features.
We hypothesized that presence of BMM does not correlate with sentinel lymph node metastasis, negative ER/PR receptor status or Her-2/neu overexpression, all of which are poor prognostic factors. A multicenter retrospective study was undertaken in 270 consecutive early breast cancer patients (primary tumor = T1 and T2), to evaluate the concordance of BMM with sentinel lymph node metastasis and poor prognostic factors of the primary tumor.
Patients and Methods
A retrospective analysis approved by the Institutional Review Boards (IRB) was performed on early breast cancer patients between January 2000 and April 2007. We included female patients with T1 and T2 primary tumors without evidence of distant disease. Patients presenting with carcinoma in situ, larger invasive tumors (T3, T4), inflammatory carcinoma, patients undergoing neoadjuvant hormonal or chemotherapy treatment, and patients with a history of previous invasive epithelial cancer were excluded. All patients had a confirmed histological diagnosis of primary invasive breast cancer. All patients underwent preoperative physical exam, bilateral mammogram, ultrasound of breasts if indicated, chest X-ray, routine blood analysis, abdominal and pelvic computed tomography, and bone scan.
Informed consent was obtained to perform simultaneous bone marrow aspiration from bilateral posterior superior iliac spines followed by lumpectomy or mastectomy and SLN biopsy on 270 patients. Data was collected for SLN status, tumor size, tumor grade, age, estrogen/progesterone (ER/PR) receptor status, human epidermal growth factor receptor 2 (Her-2/neu) expression, and bone marrow status.
SLN mapping was performed using 1% isosulfan blue (Lymphazurin, US Surgical Corp., Norwalk, CT) or 1% methylene blue and Tc99m-labeled sulfur colloid (Bristol-Meyers Squibb, Princeton, NJ). All lymphatic draining patterns were identified preoperatively by lymphoscintigram. No internal mammary lymphatic draining patterns were identified, and no undue allergic reactions were observed. SLNs were sent to pathology fresh for intraoperative touch preparation (cytology) or frozen-section (histologic) analysis. If metastases >2 mm were confirmed during intraoperative evaluation or on permanent sections of the SLN, completion axillary LN dissection (level I and II) was recommended.
Primary breast tumors and SLNs were staged and graded by pathologists according to the American Joint Committee on Cancer (AJCC) 6th edition staging system. Bone marrow samples were evaluated by separate independent laboratories (IMPATH Los Angeles, CA and US Labs, Irvine, CA), which were blinded to the histopathologic characteristics of the primary breast tumor and nodal status. Cytokeratin-positive cells reflecting micrometastases within the bone marrow were detected by CAM 5.2 monoclonal anticytokeratin antibody. The bone marrow analysis involved direct visualization of positive cells, thereby minimizing false-positives events.
The SLNs were sectioned grossly at 2–3 mm intervals and then two touch preparations were air-dried and stained for intraoperative evaluation. After intraoperative analysis had been completed, the remaining SLN was fixed in 10% phosphate-buffered neutral formalin prior to embedment in paraffin. For each node, five representative paraffin sections were obtained at 20- to 40-micron intervals. The first four sections were stained by a standard hematoxylin and eosin (H&E) method and the fifth section was stained for immunohistochemical (IHC) anticytokeratin antibody cocktail (AE-1/AE-3; Ventana Medical Systems, Tucson, AZ). SLN staging was performed by H&E and IHC staining of cells with recognizable morphology within the multiple sections of the node. Micrometastasis (pN1mi) was defined as a foci of carcinoma within a lymph node measuring 0.2–2 mm and isolated tumor cells [pN0(i+)] were defined as foci measuring <0.2 mm in accordance with the AJCC 6th addition staging manual.
Bone marrow specimens were analyzed by ChromaVision Automated Cellular Imaging System (ACIS) (San Juan Capistrano, CA). Bone marrow samples were evaluated by independent laboratories that were blinded to the final pathology results. This protocol employs highly sensitive staining methods combined with automated microscopy to differentiate cytokeratin positive cells of epithelial origin from normal hematopoietic cells in bone marrow. The cytospin slides were stained with the monoclonal anti-CAM 5.2 antibody from Becton Dickinson at 1:25 dilution for 32 min. Slides were then counterstained with Ventana hematoxylin for 4 min. The ChromaVision Cytokeratin Detection Kit detects cells expressing cytokeratin protein 8 as defined by the monoclonal antibody CAM 5.2. One or more detected tumor cells were considered as bone marrow micrometastases positive.
All patients received radiation treatment after lumpectomy. Patient selection for adjuvant chemotherapy and hormonal therapy was independent of BM status. Details of the adjuvant therapy did not impact this study of concordance and were therefore not reported.
This study was designed to clarify the relationship between BMM and SLN status, as well as other features of the primary tumor, at the time of diagnosis of early breast cancer. A two-sample t-test between proportions was performed to determine whether there was a significant difference between SLN-positive and SLN-negative patients with regard to the rate of BMM. The t-statistic was calculated as significant at the 0.05 critical alpha level using a two-tailed test. The likelihood of observing BMM was compared among the subsets of tumor stage, SLN status, ER/PR receptor status, and Her-2/neu overexpression. Chi-square (χ2) analysis was the test used to determine significance using a two-tailed probability and a p-value cutoff of 0.05. Fischer’s exact test was employed for samples with small event rates.
Detection of Micrometastases in Bone Marrow
BM aspirates were successfully obtained in all 270 patients enrolled in the study. There were no adverse complications, and all patients tolerated the procedure well. Of the 270 patients in the study, 77% (208/270) were T1 and 23% (62/270) were T2. SLN metastases (pN1) were found in 28.9% (78/270) of pts. Of the 270 patients in the study, 26 (9.6%) were found to have micrometastatic disease involving the BM. Unilateral bone marrow metastasis was present in 69.2% (18/26) of patients. Bilateral bone marrow metastases was found in 30.8% (8/26) of patients. No significant correlation between tumor grade and BMM was identified in this series.
Correlation of bone marrow (BM) status with sentinel lymph node (SLN) status
BM +ve (26)
BM −ve (244)
SLN +ve (78)
SLN −ve (192)
p = 0.65
Correlation of sentinel lymph node (SLN) status with T-stage in early breast cancer
SLN +ve (78)
SLN −ve (192)
p < 0.0001
Correlation of bone marrow (BM) status with T-stage
BM +ve (26)
BM −ve (244)
p = 0.6
Correlation of sentinel lymph node (SLN) status with ER/PR status
SLN +ve (78)
SLN −ve (190)
ER/PR −ve (26)
ER/PR +ve (242)
p = 0.5
Correlation of sentinel lymph node (SLN) status with Her-2/neu status
SLN +ve (76)
SLN −ve (184)
Her-2/neu +ve (73)
Her-2/neu −ve (187)
p = 0.7
Characteristics of Breast Cancer and Micrometastases in SLNs
All 270 patients, stage I–II, were women, with mean age of 62 years, and age range of 31–90 years. All patients had new diagnosis of early breast cancer, with 77% (208/270) having T1 disease, and the remainder (23%, 62/270) having T2 disease. The rate of breast conservation therapy was 90% (243/270).
Two hundred and six patients (76.3%) had infiltrating ductal carcinoma, and 33 (12.2%) patients had invasive lobular carcinoma. Eleven (4%) patients had features of both ductal and lobular carcinoma. The remaining 7.5% were colloid, mucinous, adenoid cystic, medullary, and comedo type.
An average of two SLNs per patient were obtained (range 1–6). The average number of lymph nodes sampled per complete axillary dissection in this series was 16. Forty-six (46/208; 22.1%) of the T1 breast patients and 48.4% (30/62) of the T2 patients were staged histologically as having lymph node metastasis (pN1) or micrometastasis (pN1mi: >0.2–2 mm) in the SLNs. Metastasis in the SLN (pN1) was identified in 22.6% (61/270) of patients by careful H&E examination. Micrometastases (pN1mi) were identified within the SLN in 17/270 (6.3%) additional patients. The SLNs were the only site of nodal metastasis in 80.8% (63/78) of node-positive patients, whereas 19.2% (15/78) had positive SLNs and additional positive non-SLNs. SLN status as determined by IHC staining was positive in 29.8% (72/242) of the ER/PR receptor +ve patients. Her-2/neu status was determined in all 270 patients, and Her-2/neu overexpression was observed in 27% (73/270) of patients.
Many patients harbor occult metastatic disease that is undiagnosed due to clinical or pathologic understaging that contributes to decreased disease-free survival and overall survival. Since long-term survival of breast cancer patients may be determined by the presence or absence of occult systemic disease, detection of occult metastatic disease has become a primary focus in early breast cancer patients. Patients thought to be disease free after surgical resection may have occult metastatic cells identified by further focused pathologic examination of bone marrow.
Sentinel lymph node biopsy used to detect axillary metastases remains one of the most important prognostic indicators for breast cancer patients. However, even subgroups of node-negative patients pN0(i−)(sn) have occult metastasis leading to disease progression. Detecting occult metastatic cells in bone marrow may assist oncologists in identifying a high-risk subgroup of sentinel lymph node-negative patients [pN0(i−)(sn) with BMM] that may benefit from early adjuvant treatment and perhaps a low-risk subgroup [pN0(i−)(sn) without BMM] in whom adjuvant chemotherapy could be avoided, limiting overtreatment. To stratify node-negative patients, genetic analysis of the primary tumor has been used to assist in determining risk of recurrence and potential benefit from chemotherapy, yet several patients are of “intermediate risk” where additional risk factors (i.e., BMM) would be important to consider. Additionally, even patients determined to be of “low risk” by genetic assay can develop recurrence and die of disease, thus implying that some risk factors remain unrecognized in this patient population. The concept of undetected disseminated tumor cells in the form of BMM remains a factor to be looked into as a possible cause or marker of such systemic failures in these subgroups.
Since both BMM and SLN status are prognostic factors in breast cancer, it is important to clarify the frequency of BMM and its relationship to the SLN status to help determine the clinical utility of BMM. If bone marrow micrometastases occur independently of SLN metastases, then determining the presence of BMM in node-negative patients with favorable primary tumors is of particular interest. The studies comparing BMM and nodal status suggest a correlation. Although there may be a statistical correlation that could be identified between BMM and SLN status with a large study, it is the clinical impact of this correlation that is most important. We feel that a statistical correlation may be meaningless if by multivariate analysis both factors have been determined to be independently prognostic and if these factors can be mutually exclusive in a subset of patients that remain at risk of distant disease.
The value of assessing axillary metastasis by focused pathologic analysis of SLNs has been well described and it is well accepted that lymph node metastases in breast carcinoma have both prognostic and predictive value.11 The value of determining the presence of BMM in early breast cancer patients is less established. A pooled analysis published by Braun et al. with over 4,700 patients established the prognostic value of BMM in stage I–III breast cancer patients and suggested a correlation to nodal metastasis.7 Braun reported patients with BMM to have poor overall, breast-cancer-specific, disease-free, and distant disease-free survival during a 10-year observation period. A study by Mansi et al. that included 350 women with follow-up of 12.5 years showed that BMM was associated with decreased overall and disease-free survival, but was not an independent prognostic factor.12 Our study suggests no association between BMM and SLN metastases in stage I–II breast cancer and is consistent with a previous report by Trocciola et al.10 Additionally, we found that presence of BMM did not correlate with other features, such as ER/PR receptor status, or Her-2/neu overexpression, all of which are poor prognostic factors in breast cancer.
This study brings forth compelling evidence suggesting no clinically significant association between positive SLN status and presence of BMM in early-stage breast cancer. We feel that BMM provides additional patient-specific information that could be used in conjunction with other conventional methods including genomic analysis to individualize breast cancer treatment. Further assessment of molecular characteristics of these disseminated tumor cells may allow more targeted therapy, thereby further individualizing patient treatment planning to eradicate micrometastases well before becoming clinically evident.
In conclusion, it is reasonable to continue to research markers that may one day impact upon treatment in breast carcinoma, such as BMM; prospective trials are needed to definitively confirm the overall clinical utility and predictive value of BMM. Currently, since only a subset of patients could potentially benefit from this additional information, one must consider the risk–benefit ratio of obtaining a bone marrow biopsy in each individual patient. At this time, identifying BMM should not be used to guide treatment outside of a clinical trial. Research resulting in novel therapies designed to target BMM could make this prognostic marker critically important in the treatment of such disease.