Micrometastatic Breast Cancer Cells in Bone Marrow at Primary Surgery: Prognostic Value in Comparison With Nodal Status

Background: Approximately 30% of the patients with primary breast cancer who have no axillary lymph node involvement (i.e., lymph node negative) at the time of surgery will relapse within 10 years; 10% -20% of the patients with distant metastases will be lymph node negative at surgery. Axillary lymph node dissection, as a surgical procedure, is associated with frequent complications. A possible alternative to nodal dissection in terms of prognosis may be the immunocytochemical detection of tumor cells in bone marrow. Purpose: In a prospective study, the value of tumor cell detection (TCD) in bone marrow was compared with axillary lymph node dissection in the prognosis of primary breast cancer after surgery. Methods: Data from 727 patients with primary, operable breast cancer were included in the analysis. All patients had surgery, including axillary lymph node dissection, from May 1985 through July 1994 at the Women's Hospital of the University of Heidelberg (Federal Republic of Germany). Bone marrow aspiration at two sites on each anterior iliac crest was performed immediately after surgery while the patients were under general anesthesia. Most patients received some type of systemic adjuvant therapy. The monoclonal antibody 2E11, directed against the polymorphic epithelial mucin TAG12, was used to detect tumor cells in bone marrow samples. The association of TCD with recognized prognostic indicators was evaluated by means of chi-squared tests. Survival without the development of distant metastases (i.e., distant diseasefree survival) and overall survival were estimated by use of the Kaplan-Meier method; the logrank test was used to compare survival curves. A multivariate Cox regression analysis with stratification according to adjuvant treatment type was used to assess the independent prognostic value of TCD in bone marrow in relation to other variables. Reported P values are two-sided. Results: Tumor cells were detected in the bone marrow of 203 (55%) of 367 lymph node-positive patients and in 112 (31%) of 360 lymph nodenegative patients. TCD was associated with larger tumors (P<.001), lymph node involvement (P = .001), and higher tumor grade (i.e., more undifferentiated) {P = .002). After a median follow-up of 36 months, patients with tumor cells in their bone marrow experienced reduced distant disease-free survival and overall survival (both P values <.001). TCD was an independent prognostic indicator for both distant disease-free survival and overall survival that was superior to axillary lymph node status, tumor stage, and tumor grade. Among patients with tumors less than 2 cm in diameter, TCD was the most powerful predictor of outcome. Conclusions and Implications: TCD in the bone marrow of patients with breast cancer is a valuable prognostic tool associated with negligible morbidity. Prospective randomized studies should be performed to determine whether TCD might replace axillary lymph node dissection in a defined subgroup of patients with small tumors. [J Natl Cancer Inst 1996;88:1652-64]