Presence Of Bone Marrow Micrometastases Research Articles

Bone marrow micrometastases in esophageal carcinoma: a 10-year follow-up study

Detection of bone marrow micrometastases (BMMs) in patients with esophageal carcinoma may indicate a metastatic phenotype. We assessed if the presence of BMMs had adverse prognostic significance in a 10-year follow-up study. Patients undergoing surgery for esophageal cancer were prospectively recruited between February 1999 and August 2000. Bone marrow aspirates were obtained from the iliac crest of patients under general anesthesia at the time of surgery. Immunocytochemical analysis using anticytokeratin antibodies CAM 5.2 and AE1/AE3 was undertaken to determine the presence of BMMs. Union International Contre le Cancer staging was recorded for all patients. Patient follow-up was completed over a 10-year period through analysis of the Northern Ireland Cancer Registry. Forty-two patients (male = 35) were included, with a mean age of 67.2 years (range 39-83). BMMs were detected in 19 patients (45.2%). International Contre le Cancer tumor staging was stage I = 6, stage II = 10, stage III = 24, and stage IV = 2. BMMs were associated with lymphovascular invasion (P= 0.02) and advanced T stage (P= 0.02). Overall, 10-year survival was 21.4% (n= 9), with a median follow-up of 877.5 days (interquartile range 391.5-2546.3). There was no statistically significant difference between the survival of patients with or without BMMs (1451.4 vs. 1431.6 days, P= 0.99). Univariate analysis demonstrated a trend toward decreased survival for patients with positive lymph nodes (P= 0.07), an increased T stage (P= 0.06), and lymphovascular invasion (P= 0.07). Multivariate analysis demonstrated that none of the variables were significant predictors of mortality. Although the presence of BMMs correlates with recognized adverse tumor characteristics in patients with esophageal cancer, micrometastases detected in the bone marrow at time of surgery does not influence long-term survival.

Use of bone marrow micrometastasis to predict disease-free and overall survival in esophageal cancer: Evaluation of 304 cases

4519 Background: The presence of bone marrow micrometastases (BMM) yields prognostic information for patients at higher risk for recurrence and death. Aim of the study was to evaluate the role of BMM in esophageal cancer (EC). Methods: 304 consecutive patients operated for EC and with BMM status were included into the prospective study. BMM were assessed by a pan-anti- cytokeratin monoclonal antibody, A45-B/B3. Associations between BMM status and clinicopathological parameters as well as disease-free (DFS) and overall survival (OS) were calculated with chi-square-, log-rank test and Cox multivariate analysis. Results: Out of 304 patients 109 (35.9%) were BMM positive (BMM+) and 195 (64.1%) negative (BMM-). Significant correlation was found between BMM and tumor size, nodal stage and recurrence (P=0.02, 0.002 and <0.0001). DFS and OS were significantly better in BMM- compared to BMM+ patients (median 44.2 and 31.6 vs. 10.4 and 12.4 months, P<0.0001). Sub-analyses revealed that nodal negative (N-) patients with BMM- had a significant better DFS and OS compared to N- and BMM+ patients (50.8 and 41.5 vs. 18.0 and 16.3 months, P<0.0001). Nodal positive patients (N+) with BMM- had a better DFS and OS compared to N+ and BMM+ patients (22.1 and 24.8 vs. 9.0 and 11.3 months, P<0.0001). In the Cox multivariate analysis BMM was the strongest predictor of DFS and OS with a hazard ratio of 3.1 and 2.9 (P<0.0001). Also tumor size and nodal stage were found to be significant predictors, whereas age and sex were not. All analyses were repeated after stratification of the study group to underlying tumor type (adenocarcinoma (AC) and squamous cell carcinoma (SCC)). Significant differences were found at all levels as reported for the entire study group above within the AC and SCC group as well. Conclusions: To our knowledge this is the largest study evaluating the role of BMM in EC. Survival sub-analyses stratified to nodal and bone marrow status revealed a significant role of BMM in EC. Presence of BMM proved to be the strongest predictor of DFS and OS in AC and SCC of the esophageus. No significant financial relationships to disclose.

Bone marrow micrometastases in advanced stage non-small cell lung carcinoma patients

The clinical relevance of bone marrow micrometastases in non-small cell lung cancer (NSCLC) is undetermined, and the value of such analyses in advanced stage patients has not been assessed previously. Immunomagnetic selection with the MOC31 (anti-EpCam) antibody was performed to isolate and detect tumor cells in bone marrow aspirates obtained from 196 patients with NSCLC, and the patients were subjected to follow-up for the assessment of survival. Repeated bone marrow samples, 2-7 samples per patient, were obtained from 13 long-term survivors. MOC31 positive tumor cells were detected in 107 of 196 (55%) samples, a frequency similar to results reported in low-stage patients prior to curative surgical resection for NSCLC. No association was found between the presence of bone marrow micrometastases and disease stage or histological subgroup, and survival was similar in patients with and without detectable tumor cells in bone marrow. Repeated bone marrow analysis revealed, for the first time in this group of patients, the continued presence of tumor cells regardless of the given therapy and treatment response. The immunomagnetic method utilized is a feasible strategy for the detection of bone marrow micrometastases in NSCLC. In advanced stage patients, the presence of MOC31 positive cells in bone marrow does not predict survival. Repeated analyses of bone marrow samples from long-term survivors revealed tumor cells in bone marrow which may represent dormant cells of particular interest for further characterization and follow-up.

Correlation of bone marrow micrometastases with nodal status in gastrointestinal tumors

14577 Background: Presence of bone marrow micrometastases (BMM) is an important prognostic factor for patients (pts) with solid tumors. Sentinel lymph node (SLN) mapping (M) has been found to upstage pts with gastrointestinal (GI) malignancies. However, a direct correlation between the presence BMM and nodal metastases (mets) is lacking. Hence, a retrospective study was undertaken to determine the relationship between BMM and nodal status in various GI malignancies. Methods: A total of 159 consecutive pts with GI malignancies were analyzed. Of these, 6 pts with non-epithelial tumors were excluded. All pts underwent bilateral posterior-superior-iliac spine bone marrow (BM) aspiration. BM samples were examined for micrometastases by Automated Cellular Imaging System using ChromaVision Cytokeratin Detection Kit to detect cells expressing CK8 as defined by CAM 5.2 monoclonal antibody. When feasible, SLNM was performed followed by standard oncologic resection (n=105). The SLN were examined by H&E and immunohistochemical (IHC) staining. Results: A total of 153 pts were analyzed in the study. Of these 13.72% (21/153) were found to have BMM (CI 8.70–20.21, p=0.05). When analyzed by sites, the incidence of BMM was 33.3% (4/12) for stomach cancer, 9.1% (1/11) for pancreatic cancer, 12.36% (11/89) for colon cancer and 16.13% (5/31) for rectal cancer. No BMM were detected in pts with cancer of esophagus (n=7), small bowel (n=2) and liver (n=1). The BMM was positive bilaterally in 61.9% (13/21), and unilaterally in 38.1% (8/21). In pts who underwent SLNM (n=105), BMM were detected in 11.11% of SLN positive pts vs. 10.14% for SLN negative pts (p=0.87). For a subgroup of pts with colorectal cancer who underwent SLNM (n=97), BMM were detected in 9.67% of SLN positive pts vs. 9.09% for SLN negative pts (p=0.92) Conclusions: BMM did not correlate with nodal status for GI malignancies suggesting a possible different mechanism for metastases. Detection of BMM may have a significant clinical value in SLN negative pts who may benefit from adjuvant therapy. [Table: see text] No significant financial relationships to disclose.

Association of the Presence of Bone Marrow Micrometastases with the Sentinel Lymph Node Status in 410 Early Stage Breast Cancer Patients: Results of the Swiss Multicenter Study

The sentinel lymph node (SLN) status has proven to accurately reflect the remaining axillary lymph nodes and represents the most important prognostic factor. It is unknown whether an association exists between the SLN status and the presence of bone marrow (BM) micrometastases. The objective of the present investigation was to evaluate whether or not such an association exists. In the present investigation 410 patients with early stage breast cancer (pT1 and pT2 <or=3 cm, cN0) were prospectively enrolled between 1/2000 and 12/2003. All patients underwent SLN biopsy and bone marrow aspiration. The histological examination of the SLN consisted of step sectioning, H&E, and immunohistochemistry (Lu-5, CK 22) staining. Cancer cells in the BM were stained with monoclonal antibodies A45-B/B3 against cytokeratin and counted by an automated computerized digital microscope. BM micrometastases were detected in 28.8% (118/410) of all patients. The SLN contained metastases in 32.4% (133/410). Overall 51.2% of the patients (210/410) were SLN negative/BM negative and 12.4% (51/410) SLN positive/BM positive. Of all patients, 16.4% (67/410) were SLN negative/BM positive and 20.0% (82/410) SLN positive/BM negative. There was a statistically significant association between the SLN and BM status, both in unadjusted (Fisher's exact test: P = .004) and multiple logistic regression analysis (P = .007). In the present investigation a significant association was found between a positive SLN status and the presence of BM micrometastases. Nonetheless, the percentage of non-concordance (SLN negative/BM positive and SLN positive/BM negative) was considerable. The prognostic impact of BM micrometastases in our patient sample remains to be evaluated.

The impact of bone marrow micrometastases on metastatic disease‐free survival in patients with colorectal carcinoma

The biological relevance of bone marrow micrometastases (BMM) in colorectal cancer remains unknown. Here, we investigate their nature by examining the impact of the presence of BMM on metastatic disease-free survival in a cohort of patients with this disease. Sixty-three consecutive patients undergoing surgery for colorectal cancer of any stage were studied after approval of the study protocol by the local ethics committee and with full individual informed consent. All had bilateral iliac crest bone marrow aspirates prior to operation. Aspirates were then examined for the presence of aberrant cytokeratin-18-positive cells by a blinded observer using both flow cytometric and APAAP immunohistochemical techniques. Mean follow-up after surgery was 4.6 years (range 1.9-6.9) for those without hepatic metastases at diagnosis. Seven of 34 patients with Dukes' stage A or B developed metastatic disease after a mean interval of 4.7 years (range 3.8-6.8). However, only 2 of these patients demonstrated BMM at the time of surgery. Nine of 15 patients with Dukes' C carcinoma at the time of surgery subsequently developed metastases after a mean interval of 4.4 years (range 1.9-6.9). Again, only two of these patients had BMM detectable initially. In only three of the 14 patients known to have metastases at the time of operation (i.e. Dukes''D' disease) were BMM found. The presence of BMM as detected by this methodology was not predictive of tumour recurrence or metastasis. This study does not support the consideration of adjuvant therapy based on the presence of BMM at a single pre-operative time point in patients with colorectal cancer.

Relative microvessel area of the primary tumour, and not lymph node status, predicts the presence of bone marrow micrometastases detected by reverse transcriptase polymerase chain reaction in patients with clinically non-metastatic breast cancer

About 50% of patients with breast cancer have no involvement of axillary lymph nodes at diagnosis and can be considered cured after primary locoregional treatment. However, about 20–30% will experience distant relapse. The group of patients at risk is not well characterised: recurrence is probably due to the establishment of micrometastases before treatment. Given the early steps of metastasis in which tumour cells interact with endothelial cells of blood vessels, and, given the independent prognostic value in breast cancer of both the quantification of tumour vascularisation and the detection of micrometastases in the bone marrow, the aim of this study was to determine the relationship between vascularisation, measured by Chalkley morphometry, and the bone marrow content of cytokeratin-19 (CK-19) mRNA, quantified by real-time reverse transcriptase polymerase chain reaction, in a series of 68 patients with localised untreated breast cancer. The blood concentration of factors involved in angiogenesis (interleukin-6 and vascular endothelial growth factor) and of factors involved in coagulation (D-dimer, fibrinogen, platelets) was also measured. When bone marrow CK-19 relative gene expression (RGE) was categorised according to the cut-off value of 0.77 (95th centile of control patients), 53% of the patients had an elevated CK-19 RGE. Patients with bone marrow micrometastases, on the basis of an elevated CK-19 RGE, had a mean Chalkley count of 7.5 ± 1.7 (median 7, standard error [SE] 0.30) compared with a mean Chalkley count of 6.5 ± 1.7 in other patients (median 6, SE 0.3) (Mann–Whitney U-test; P = 0.04). Multiple regression analysis revealed that Chalkley count, not lymph node status, independently predicted CK-19 RGE status (P = 0.04; odds ratio 1.38; 95% confidence interval 1.009–1.882). Blood parameters reflecting angiogenesis and coagulation were positively correlated with Chalkley count and/or CK-19 RGE. Our data are in support of an association between elevated relative microvessel area of the primary tumour and the presence of bone marrow micrometastases in breast cancer patients with operable disease, and corroborate the paracrine and endocrine role of interleukin-6 and the involvement of coagulation in breast cancer growth and metastasis.

Bone marrow micrometastases and adjuvant treatment of breast cancer.

The immunohistochemical detection of epithelially derived cells in the bone marrow of patients with primary breast cancer has been shown to be associated with increased risk of distant relapse as well as higher rates of cancer-related death. Despite the correlation between bone marrow micrometastases and poor outcome in breast cancer patients, bone marrow status does not yet have an established role in patient management. In this prospective study, adjuvant therapy recommendations for 43 patients with stage I, II, or III breast cancer treated with lumpectomy or mastectomy, sentinel lymph node biopsy and/or axillary dissection, and intraoperative bone marrow aspiration were recorded. Recommendations were made by a multidisciplinary tumor board both blinded and unblinded to the results of the bone marrow aspiration. In our study, 10 of the 43 breast cancer patients were found to have bone marrow micrometastases. Four of these patients (40%) had axillary lymph node metastases. When blinded to the results of the bone marrow aspiration, the tumor board recommended adjuvant chemotherapy for these four node-positive patients, as well as two node-negative patients. When unblinded to the results of the bone marrow aspiration, the tumor board did not change its recommendations for any of these six patients. The remaining four node-negative, bone marrow-positive patients were not advised to have adjuvant chemotherapy by the tumor board when blinded to bone marrow status. However, once the tumor board was informed of the presence of bone marrow micrometastases, adjuvant chemotherapy was recommended for all of these patients. The results of this pilot study indicate that the presence of bone marrow micrometastases in breast cancer patients with stage I, II, or III disease does influence recommendations for adjuvant chemotherapy, particularly in patients with node-negative disease.

Cyclooxygenase-2 expression in primary human colorectal cancers and bone marrow micrometastases

Background. Both the expressions of the inducible form of cyclooxygenase-2 and the presence of bone marrow micrometastases are poor prognostic markers in patients with colorectal carcinoma. Aims. As cyclooxygenase-2 expression in these tumours is associated with increased metastatic potential in vitro, our objectives were to determine the relationship between cyclooxygenase-2 and haematogenous spread to bone marrow. Patients and methods. Thirty-two patients with resection of colorectal carcinoma were evaluated (median age: 69.5 years). Bone marrow was obtained from all patients from both iliac crests before manipulation of the primary tumour. The tumours were of varying stages at diagnosis (5 Dukes’ A, 14 Dukes’ B, 11 Dukes’ C and 2 Dukes’ D). Tumour sections were stained for cyclooxygenase-2 using the avidin–biotin immunohistochemical technique. Extent of staining was graded depending on the percentage of epithelial cells staining positive for cyclooxygenase-2. Micrometastases were detected by staining contaminant cytokeratin-18 positive cells in the bone marrow aspirates by either immunohistochemical (ARAAP) or immunological (flow cytometry) methods. Fisher’s exact probability test was used to calculate statistical significance. Results. Cyclooxygenase-2 expression in the primary tumour was detected in 72% of the patients. Twelve (38%) patients had bone marrow micrometastases detected by either immunohistochemistry or flow cytometry. Of the 12 patients who had bone marrow micrometastases, 8 tumours demonstrated increased expression of cyclooxygenase-2 protein (66.6%). In contrast, 9 out of the 20 (45%) patients in whom micrometastases were not detected expressed increased levels of cyclooxygenase-2 ( P=0.29). When dividing the patients into subgroups of localised (Dukes’ A and B) versus disseminated (Dukes’ C and D) disease, there was no further association between cyclooxygenase-2 expression and bone marrow micrometastases ( P=0.179 and 1.0). Conclusion. In this pilot study, there was no association between cyclooxygenase-2 expression and bone marrow micrometastases in patients with otherwise localised or disseminated disease.

Micrométastases et cancers bronchiques non à petites cellules

Non-small cell lung cancer has a poor prognosis, including those with operable, apparently localised, disease. Preoperative staging investigations and histo-pathological analysis are poor at detecting small clusters of tumour cells, particularly in lymph nodes. New methods based on immunohistochemistry, or molecular biology, have been developed to detect these so-called micro-metastases. We present a ten-year review of the literature published on this topic. These publications primarily reported on the detection of micro-metastases within mediastinal lymph nodes removed at operation in order to identify patients at risk of recurrence, for whom adjuvant therapy might be offered. Lymph node micro-metastases have been demonstrated to be an independent prognostic factor for survival, especially in stage I patients. On the other hand, the presence of bone marrow micro-metastases did not appear to be of significant prognostic value in non-small-cell lung cancer. Finally, the clinical relevance of circulating tumour cells is still debatable, although recent published studies show interesting results.

Bone marrow micrometastasis detected by RT-PCR in esophageal squamous cell carcinoma.

The clinical implications of bone marrow micrometastases (BMM) detected by RT-PCR in esophageal squamous cell carcinoma (ESCC) have not been elucidated. We evaluated the relation between the presence of BMM, both before and after surgery, and clinicopathologic findings in patients with ESCC. Bone marrow samples from 48 patients with ESCC were obtained from the iliac crest before and after surgery. After total RNA was extracted from each bone marrow sample, carcinoembryonic antigen (CEA)-specific RT-PCR was performed. BMM was detected by RT-PCR in 10 of the 48 patients. Four patients each had positive signals only before or only after surgery and 2 patients had positive signals both before and after surgery. There were no significant differences in clinicopathologic factors, including neoadjuvant therapy, between patients with BMM and without BMM. To date, the rates of recurrent disease in patients with BMM and without BMM are 80% (8/10) and 50% (19/38), respectively, a difference which is not significant. The 4-year survival rates of patients with BMM and without BMM are 10.0% and 47.3%, respectively. Recurrence and survival rates were poorer in patients with RT-PCR positivity, although the differences were not significant. A larger study is required to clarify the clinical impact of BMM.

The presence of bone marrow cytokeratin-immunoreactive cells does not predict outcome in gastric cancer patients.

The independent prognostic significance of isolated tumour cells in bone marrow is still a matter of debate. This study evaluated the possible association of bone marrow micrometastases with tumour progression and prognosis in patients affected by gastric cancer. Bone marrow aspirates from both iliac crests were obtained from 114 consecutive patients operated on for gastric cancer. The specimens were stained with monoclonal antibody CAM 5.2 which reacts predominantly with cytokeratin filaments 8 and 19. Among 114 cases analysed, 33 cases (29%) had cytokeratine-positive cells in the bone marrow. There was no significant relationship between the presence of bone marrow micrometastases and site, depth of tumour invasion, lymph node metastases, presence of metastases. Patients with cytokeratine-positive cells had a trend towards a diffuse type histology (P=0.06). Among the 88 curatively resected patients, median survivals were 40 months and 36 months for cytokeratine-negative and cytokeratine-positive subsets respectively (P=0.9). Recurrence of the disease was observed in 39 cases (44.3%); 11 of 24 (45.8%) in the cytokeratine-positive subset and 28 of 64 (43.7%) in the cytokeratine-negative subset. In conclusion in our experience the presence of cytokeratine-positive cells in the bone marrow of curatively resected gastric cancer patients did not affect outcome and its independent prognostic significance remains to be proven before its official acceptance in the TNM classification.British Journal of Cancer (2002) 86, 1047–1051. DOI: 10.1038/sj/bjc/6600211 www.bjcancer.com© 2002 Cancer Research UK

Bone marrow micrometastases in breast cancer patients

Background: The status of the axillary nodes has historically been the most important indicator of prognosis in breast cancer patients. However, approximately one third of node-negative patients recur with systemic disease. The detection of bone marrow micrometastases (BMM) may represent additional information in predicting distant recurrence and survival. Methods: Bone marrow aspiration is obtained from the patient’s anterior iliac crest at the time of breast cancer surgery. Cytospins are prepared from this aspirate and stained for polymerase chain reaction (PCR) analysis. Results: Multiple studies have evaluated the clinical implications of BMM. The majority of studies have found a significant correlation between the presence of BMM and decreased survival. The information of bone marrow status may serve as a complement to axillary lymph node status in assessing the prognosis of breast cancer patients. Conclusions: There is a strong correlation between the presence of bone marrow micrometastases and poorer survival. These results may have an impact upon therapeutic recommendations in breast cancer patients.

Outcome of primary-breast-cancer patients with micrometastases: a long-term follow-up study

Bone-marrow micrometastases have been found in patients with primary breast cancer. We report long-term follow-up of women with primary breast cancer, diagnosed between 1981 and 1986, who had multiple aspirates taken at the time of initial surgery. 350 women with primary breast cancer were examined immunocytochemically with antibody to epithelial membrane antigen. We investigated associations with various prognostic factors as well as the effect of micrometastases on relapse-free survival and overall survival. At median follow-up of 12.5 years, 151 patients had metastatic disease and 136 patients had died from breast cancer. 10-year relapse-free and overall survival were 43.9% (95% CI 33.4-54.7) and 44.9% (34.2-55.9) in patients with micrometastases, and 62.7% (56.5-68.6) and 65.7% (59.4-71.5) in patients without micrometastases at presentation (p<0.001). For relapse-free survival and overall survival, allowing for tumour size, lymph-node status, and vascular invasion, the effect of micrometastases decreased and was no longer significant, with a hazard ratio of 1.09 (0.74-1.61) for relapse-free survival and 1.21 (0.84-1.75) for overall survival. The presence of bone-marrow micrometastases in patients with primary breast cancer is associated with a shorter relapse-free survival and overall survival, but is not an independent prognostic factor. This immunocytochemical technique may be of value in patients for whom pathological tumour size and lymph-node status are unavailable (ie, patients receiving primary medical treatment).

Association of tumor angiogenesis with bone marrow micrometastases in breast cancer patients.

The microscopic detection of tumor cells (micrometastases) in bone marrow and the extent of blood vessel formation (angiogenesis) in primary tumor specimens are recognized as independent prognostic markers in patients with breast cancer. Since micrometastases occur as a consequence of interaction between the neoplastic cells and the tumor neovasculature, we have examined the relationship between these markers to determine whether the degree of angiogenesis is related to the presence of micrometastases. Micrometastases were identified in bone marrow aspirates collected from multiple sites in 214 breast cancer patients prior to surgery (mastectomy or lumpectomy). Tumor cells were detected through an examination of epithelial membrane antigen expression and an analysis of cell morphology. Tumor vascularity was graded semiquantitatively or quantitatively (Chalkley point count) after immunohistochemical staining of the CD31 antigen expressed by the endothelial cells. The reproducibility and accuracy of the vascular grading were validated by use of kappa statistics. Associations between micrometastases and clinicopathologic characteristics, including angiogenesis, were examined using chi-squared and logistic regression techniques. All tests of statistical significance were two-sided. Of the 214 patients, 42 (20%) were positive for bone marrow micrometastases and 75 (35%) had tumors of high vascular grade. There was 86% agreement between vascular grades assessed twice for 35 tumors (kappa statistic = 0.66); for 22 evaluated tumors, there was absolute concordance between vascular grade and Chalkley point count. There were significant positive associations between tumor angiogenesis and micrometastasis (P = .01), tumor grade (P = .003), and estrogen receptor expression (P = .007); however, no significant associations were observed with tumor size (P = .9), lymph node status (P = .33), vascular invasion (peritumoral blood or lymph vessels) (P = .9), menopausal status (P = .17), or age (P = .12). Adjusting for confounding factors, multivariate analysis showed that only tumor angiogenesis (odds ratio = 2.7; P = .016) and vascular invasion (odds ratio = 2.7; P = .012) were significant determinants for the presence of micrometastases. This study suggests that an assessment of tumor angiogenesis and vascular invasion gives a reliable indication of the likelihood of the presence of bone marrow micrometastases in patients with breast cancer and that both processes contribute to metastases.

Prediction of early relapse in patients with operable breast cancer by detection of occult bone marrow micrometastases.

We used monoclonal antibodies to identify occult micrometastases in the bone marrow of 49 patients with operable (stage I and II) breast carcinoma. Follow-up (mean, 29 months; median, 30 months) revealed that 12 patients recurred. The presence of bone marrow micrometastases (BMM) was significantly associated with early recurrence (P less than .04). The estimated 2-year recurrence rate for patients with no BMM detected (BMM-) was 3%; in patients with BMM, the 2-year recurrence rate was 33%. When BMM and axillary lymph node (LN) status were combined, groups of patients at low risk (LN-, BMM-; 2-year recurrence rate, 0%) and high risk (LN+, BMM+; 2-year recurrence rate, 42%) for early recurrence were identified. Bone marrow tumor burden was related to early recurrence. Among patients with BMM, those who did not recur had on average fewer extrinsic cells in their marrow than those who recurred (15 v 43 cells, respectively). Multivariate analysis comparing BMM, LN+ versus LN-, and tumor size (less than or equal to 2 cm v greater than 2 cm) revealed no factor independently associated with early recurrence. Peripheral tumor burden of BMM (0 or less than 10 extrinsic cells v greater than or equal to 10 extrinsic cells) was the only independent predictor of early recurrence (P less than .003). In conjunction with conventional prognostic factors, particularly axillary LN status, evaluation for BMM might be used to stratify patients for adjuvant treatment programs. Because this pilot study involved few patients with short-term follow-up, the results should be interpreted with caution. The examination of bone marrow for micrometastases remains an experimental procedure; the clinical usefulness of the test will be established through larger studies with long-term follow-up.