Annals of Surgical Oncology

, Volume 20, Issue 6, pp 1865–1871

Prognostic Value of Disseminated Tumor Cells in the Bone Marrow of Patients with Operable Primary Breast Cancer: A Long-term Follow-up Study

Authors

  • Christoph Domschke
    • Breast UnitUniversity of Heidelberg
  • Ingo J. Diel
    • CGG-Clinic Mannheim
  • Stefan Englert
    • Institute of Medical Biometry and Informatics, University of Heidelberg
  • Silvia Kalteisen
    • Breast UnitUniversity of Heidelberg
  • Luisa Mayer
    • Breast UnitUniversity of Heidelberg
  • Joachim Rom
    • Breast UnitUniversity of Heidelberg
  • Joerg Heil
    • Breast UnitUniversity of Heidelberg
  • Christof Sohn
    • Breast UnitUniversity of Heidelberg
    • Breast UnitUniversity of Heidelberg
Breast Oncology

DOI: 10.1245/s10434-012-2814-4

Cite this article as:
Domschke, C., Diel, I.J., Englert, S. et al. Ann Surg Oncol (2013) 20: 1865. doi:10.1245/s10434-012-2814-4

Abstract

Background

Detection of disseminated tumor cells (DTC) in primary breast cancer (BC) patients’ bone marrow (BM) seems to be a surrogate marker of tumor spread and an independent prognostic factor for disease-free and overall survival.

Methods

Here we present the largest single-center cohort of patients (n = 1378) with the longest observation time (median 82.0 months). Immunocytochemical staining was performed using murine monoclonal antibody 2E11 with the avidin–biotin complex technique.

Results

At primary surgery, 49 % of patients showed MUC-1 positive cells inside their BM. Patients without BM DTC had significantly more often T1-tumors (P = 0.007) with less often affected axillary lymph nodes (P < 0.001). We observed a significantly higher incidence of distant metastases in DTC positive patients (P < 0.001). This leads to a reduced disease-free survival (P < 0.0001). Furthermore, in DTC positive patients there was a higher mortality rate and, accordingly, a reduced overall survival (P < 0.0001).

Conclusions

Due to the presence of BM DTC, patients with a clinically poorer outcome can be identified at primary surgery. We therefore suggest that DTC analysis can be used as a prognostic factor and monitoring tool in clinical trials. Future study concepts relating to DTC should aim at identification of BC patients who may profit from adjuvant systemic therapy.

In recent years, mostly due to earlier diagnoses primary breast cancer (BC) survival rates have steadily increased, although still approximately one-third of patients will develop distant metastasis.1 Having in mind that many BC patients can be cured by a surgical procedure alone and the fact that systemic therapies can only prevent a minority of treated patients from cancer recurrence, the question of who is to be treated with which kind of therapy should critically be considered. The first step has to be an identification of those patients with a higher risk of developing distant metastases. In this context, many methods and parameters have been analyzed. Next to the classic parameters as tumor size, axillary lymph node involvement and grading, tumor-biological characteristics like estrogen-/progesterone-receptor and HER2-receptor status have become main criteria for decision making in adjuvant systemic therapy. However, these parameters do not reflect the capability of breast cancer cells to survive as single cells or minimal residual disease for several years and to lead to metastatic spread itself. Because some tumors with a favorable prognosis are able to relapse even after years and many tumors with unfavorable characteristics will never relapse, additional factors are desirable to identify those patients with a high risk of metastatic spread.

The detection of disseminated tumor cells (DTC) in bone marrow (BM) of primary breast cancer patients has been described by many authors as a specific way of defining a subgroup of patients with a high risk of distant relapse especially in the bone system.28 The presence of DTC was reported as an independent risk factor in addition to the classic ones.9,10 A meta-analysis with more than 4700 patients from nine different institutions was done by the “Collaborative Group Bone Marrow Metastasis.”11 Although different techniques were used in the individual centers, the analysis showed shorter progression-free and overall survival for BC patients with DTC in BM.

Despite those observations a clinical routine use of this method has never been recommended by any guideline or expert panel due to a lack of consensus regarding methodological and institutional problems. Next to the question how to analyze the samples there is no standard which antibody against which tumor associated antigen should be used although a German guideline has been published.12 Furthermore, there are no long-term follow-up data of a large group of patients available that really prove the prognostic value of DTC.

In 1996, our group published data from 727 primary BC patients who underwent BM aspiration during surgical procedure.2 For the first time we were able to show in a large cohort of patients the independent prognostic value of DTC in regard to distant disease-free and overall survival. A second follow-up of this cohort was published in 2001 demonstrating that there was still a high impact on these important parameters.13 Here we present the largest group of patients who underwent BM aspiration for tumor cell detection with the longest follow-up ever published.

Methods

Patient Population

BM samples of 1378 primary BC patients were collected in a monocenter observation trial at the Breast Unit of Heidelberg University Hospital, Germany. Study protocol was approved by the ethical committee of the University of Heidelberg, Heidelberg, Germany. All patients gave their informed consent to participate. From 1985 to 1996, 1378 primary BC patients underwent a surgical procedure of the breast with BM aspiration for analysis of disseminated tumor cells. Exclusion criteria were regional or distant metastasis within 3 months after surgery, breast biopsy and/or lumpectomy before definitive surgery, and a history of other malignant disease or simultaneous second primary tumor as well as incomplete follow-up data.

Follow-up was done from March 2005 until September 2005 by evaluating patients′ records or, if there was a time-period of longer than 6 months after the last visit, patients or their physicians were contacted by mail and/or phone.

BM Aspiration and Immunocytology

BM samples were taken under standardized conditions during the primary surgical procedure before or after tumor resection. Both anterior iliac crests were punctured by using a BM aspiration set with heparinized syringes. Samples of 5–7 ml of each crest were taken in order to avoid contamination with peripheral blood.14 The immunocytochemical staining method has been described in our previous study.2 Briefly, the aspirate was separated by density centrifugation (Ficoll), and the cell suspension (4–5 × 106 cells) was smeared onto slides. Immunocytochemical staining was performed using murine monoclonal antibody 2E11 with the avidin-biotin complex technique, which is directed against MUC-1 molecule that is very common on breast cancer cells while it is absent or expressed on low level in normal mammary gland.15 Consequently, the MUC-1-specific antibody 2E11 does positively react with more than 96 % of primary breast carcinomas. The sensitivity of this method allows an identification of 1 positive cell in 106 normal BM cells. Negative as well as positive controls were also analyzed against 4 smears per patient. The membrane and cytoplasm of tumor cells stained bright red. Positive smears were defined as those containing one or more than one stained cell. Additionally, according to morphological criteria proposed by a European Working Group (with coauthor I.J.D. serving as member) immunopositive cells were only classified as tumor cells if nuclei were clearly enlarged or atypical and clusters of two or more positive cells were present (in case of multiple cells).16 All slides were assessed by two independent investigators with an inter-observer agreement of over 99 %. In discordant smears patients were considered to be tumor cell negative. All smears were analyzed without knowing patients′ clinical as well as histopathological results.

Statistical Methods

The association of DTC with established prognostic markers was analyzed by χ2 test or t-test according to the distribution of the data. Overall survival and distant disease-free survival rates were analyzed. Distant disease-free survival was defined as survival without the development of distant metastases. Survival curves were calculated by the Kaplan-Meier method, and the comparison of two survival curves was based on the log rank test according to Peto and Peto. A multivariate Cox regression analysis was performed to assess the independent prognostic value of DTC adjusted for other prognostic factors. The impact of each variable in the Cox regression model was tested by the Wald χ2 test and described by the risk ratio (i.e., the hazard ratio). All reported probabilities are two-sided, and P values of < 0.05 were considered statistically significant.

Results

All Patients

We were able to include 1255 patients of the initially DTC-analyzed 1378 patients into our follow-up analysis. 8.9 % of our study patients were lost to follow-up (equally distributed to both groups). At primary surgery, 621 (49 %) of those patients showed MUC-1 positive cells inside their BM while 634 (51 %) did not. The medium observation time was 82.0 months (lower quartile: 61.3 months; upper quartile 104.2 months).

Relationships between patients and tumor characteristics and DTC are shown in Table 1. Patients without disseminated tumor cells in bone marrow had significantly more often T1-tumors (P = 0.007) with less often affected axillary lymph nodes (P < 0.001). Patients with disseminated tumor cells had significantly more often ER positive tumors (P = 0.041). Grading, PR status and histological type of tumor were not significantly different between both groups. Concordantly to those histopathological results we found a significant difference in the number of adjuvant chemotherapies (P < 0.001), while the different number of adjuvant endocrine therapies with tamoxifen did not reach the level of statistical significance (P = 0.089). Patients with disseminated tumor cells in bone marrow received significantly less breast conserving therapy (P < 0.001). In the number of primary chemotherapies and radiotherapies, however, there was no difference (Table 2).
Table 1

Patients and tumor characteristics of 1255 primary BC patients with (n = 621) and without (n = 634) disseminated tumor cells in BM

Characteristic

BM negative

BM positive

All

P

(n = 634)

(n = 621)

(n = 1255)

Age (years)

 Mean ± SD

50.9 ± 6.9

51.1 ± 7.2

51.0 ± 7.0

0.554

Histology

 Ductal invasion

459 (72.6 %)

447 (72.0 %)

906 (72.3 %)

0.808

 Lobular invasion

100 (15.8 %)

95 (15.3 %)

195 (15.6 %)

 

 Other

73 (11.6 %)

79 (12.7 %)

152 (12.1 %)

 

Tumor size (mm)

 Mean ± SD

23.6 ± 15.6

26.0 ± 18.7

24.7 ± 17.2

0.020

T stage

 1

305 (49.4 %)

255 (41.7 %)

560 (45.5 %)

0.007

 2–4

313 (50.6 %)

357 (58.3 %)

670 (54.5 %)

 

Nodal status

 0

375 (63.6 %)

256 (47.4 %)

631 (55.8 %)

<.001

 1

215 (36.4 %)

284 (52.6 %)

499 (44.2 %)

 

Grading

 1

56 (10.9 %)

51 (9.9 %)

107 (10.4 %)

0.531

 2

294 (57.4 %)

286 (55.3 %)

580 (56.4 %)

 

 3

162 (31.6 %)

180 (34.8 %)

342 (33.2 %)

 

ER

 Negative

238 (39.7 %)

196 (33.9 %)

434 (36.8 %)

0.041

 Positive

362 (60.3 %)

382 (66.1 %)

744 (63.2 %)

 

PR

 Negative

248 (41.3 %)

245 (42.4 %)

493 (41.9 %)

0.714

 Positive

352 (58.7 %)

333 (57.6 %)

685 (58.1 %)

 

Data are presented as mean ± SD or n (%)

ER estrogen receptor, PR progesterone receptor

Table 2

Antitumor therapies of 1255 primary BC patients with (n = 621) and without (n = 634) disseminated tumor cells in BM

Characteristic

BM negative

BM positive

All

P

(n = 634)

(n = 621)

(n = 1255)

Surgery

 Breast-conserving therapy

455 (72.3 %)

378 (61.5 %)

833 (67.0 %)

<0.001

 Mastectomy

174 (27.7 %)

237 (38.5 %)

411 (33.0 %)

 

Primary chemotherapy

 No

546 (86.9 %)

518 (84.5 %)

1064 (85.7 %)

0.219

 Yes

82 (13.1 %)

95 (15.5 %)

177 (14.3 %)

 

Adjuvant chemotherapy

 No

345 (61.0 %)

265 (48.8 %)

610 (55.0 %)

<0.001

 Yes

221 (39.0 %)

278 (51.2 %)

499 (45.0 %)

 

Adjuvant endocrine therapy

 No

302 (53.4 %)

262 (48.3 %)

564 (50.9 %)

0.089

 Yes

264 (46.6 %)

281 (51.7 %)

545 (49.1 %)

 

Radiotherapy

 No

178 (28.7 %)

180 (29.8 %)

358 (29.2 %)

0.688

 Yes

442 (71.3 %)

425 (70.2 %)

867 (70.8 %)

 

BC breast cancer

Table 3 describes the incidence of local and distant recurrences as well as mortality rates. Local recurrence occurred in 287 patients (128 DTC negative and 159 DTC positive patients; P = 0.022). We observed, furthermore, a significantly higher incidence of distant metastases in DTC positive patients (233 vs. 136 events, P < 0.001). Bone metastases (P < 0.001) as well as pulmonary (P = 0.002) and hepatic (P = 0.015) lesions occurred more often in DTC positive patients. This higher incidence of distant metastases leads to a reduced disease-free survival (Fig. 1a; P < 0.0001). Furthermore, there was a higher mortality rate in DTC positive patients (173 vs. 97 deaths; Table 3; P < 0.001) and a reduced overall survival (Fig. 1b; P < 0.0001). Figure 1c shows Kaplan-Meier analysis for disease-free survival in regard to DTC as well as nodal status. Patients without involvement of bone marrow and lymph nodes do have the best prognosis while patients with tumor cells in both compartments show a highly reduced disease-free survival. Solitary analysis of patients without lymph node involvement revealed a significantly reduced disease-free survival in patients with DTC in BM (P = 0.0418). Accordingly, in patients with lymph node involvement absence of DTC in BM was correlated to a significantly increased disease-free survival (P = 0.0002). As patients with DTCs have significantly fewer T1-tumors and more often positive lymph nodes, we performed a multivariate Cox survival analysis with T and N status as adjusted covariates, showing DTC detection as an independent, additional prognostic factor of both disease-free and overall survival (P = 0.0003; P = 0.0005).
Table 3

Incidence of local recurrence and distant metastases as well as mortality rates of 1255 primary BC patients with (n = 621) and without (n = 634) disseminated tumor cells in BM

Characteristic

BM negative

BM positive

All

P

(n = 634)

(n = 621)

(n = 1255)

Local recurrence

 No

506 (79.8 %)

462 (74.4 %)

968 (77.1 %)

0.022

 Yes

128 (20.2 %)

159 (25.6 %)

287 (22.9 %)

 

Distant recurrence

 No

498 (78.5 %)

388 (62.5 %)

886 (70.6 %)

<0.001

 Yes

136 (21.5 %)

233 (37.5 %)

369 (29.4 %)

 

Site of distant recurrence

 Bone

  No

548 (86.4 %)

481 (77.6 %)

1029 (82.1 %)

<0.001

  Yes

86 (13.6 %)

139 (22.4 %)

225 (17.9 %)

 

 Liver

  No

586 (92.4 %)

548 (88.4 %)

1134 (90.4 %)

0.015

  Yes

48 (7.6 %)

72 (11.6 %)

120 (9.6 %)

 

 Lung

  No

587 (92.6 %)

541 (87.3 %)

1128 (90.0 %)

0.002

  Yes

47 (7.4 %)

79 (12.7 %)

126 (10.0 %)

 

 Other sitesa

  No

588 (92.7 %)

527 (85.1 %)

1115 (89.0 %)

<0.001

  Yes

46 (7.3 %)

92 (14.9 %)

138 (11.0 %)

 

 Death

  No

537 (84.7 %)

448 (72.1 %)

985 (78.5 %)

<0.001

  Yes

97 (15.3 %)

173 (27.9 %)

270 (21.5 %)

 

BC breast cancer, BM bone marrow

aOther sites include brain, skin, adrenal, thyroid, and ovary

https://static-content.springer.com/image/art%3A10.1245%2Fs10434-012-2814-4/MediaObjects/10434_2012_2814_Fig1_HTML.gif
Fig. 1

a Kaplan-Meier analysis of disease-free survival of patients with (BMM+) and without (BMM−) DTC in BM. b Kaplan-Meier analysis of overall survival of patients with (BMM+) and without (BMM−) DTC in BM. c Kaplan-Meier analysis of disease-free survival in regard to DTC in bone marrow (BMM) and tumor spread to axillary lymph nodes (N)

Subgroup Analysis of Tumors with a Histopathologically Good Prognosis (pT1 pN0)

A total of 348 BC patients with tumors less than 20 mm in diameter (pT1) and negative lymph node involvement (pN0) were included in a subgroup. In this cohort with an excellent prognosis the impact of DTC on disease-free (DFS) and overall survival (OS) was analyzed. During follow-up 27 patients of this subgroup died and 35 patients developed distant metastases. Due to this small number of events there are high confidence intervals for hazard ratios. Therefore, in univariate and multivariate analyses we were not able to show that in this small subgroup DTC are an independent prognostic factor for DFS and OS (Tables 4 and 5). Whether those results are underpowered or were different in a larger patients’ cohort can not be examined any further in our study population.
Table 4

Univariate and multivariate analyses in regard to disease-free survival in BC patients (n = 348) with favorable primary disease (pT1pN0)

Variable

HR

95 % CI

P

Univariate analysis regarding disease-free survival

 DTC

1.59

0.96–2.63

0.069

 Poorly differentiated (G3)

2.04

1.10–3.80

0.025

 ER negative

1.56

0.89–2.72

0.120

 No chemotherapy

0.49

0.28–0.84

0.009

 No endocrine therapy

2.24

1.22–4.10

0.009

Multivariate analysis regarding disease-free survival

 DTC

1.36

0.73–2.52

0.336

 Poorly differentiated (G3)

1.76

0.89–3.48

0.103

 ER negative

1.23

0.61–2.44

0.564

 No chemotherapy

0.78

0.37–1.62

0.506

 No endocrine therapy

1.54

0.70–3.37

0.282

BC breast cancer, HR hazard ratio, CI confidence interval, DTC disseminated tumor cells, ER estrogen receptor

Table 5

Univariate and multivariate analyses in regard to overall survival in BC patients (n = 348) with favorable primary disease (pT1pN0)

Variable

HR

95 % CI

P

Univariate analysis regarding overall survival

 DTC

1.32

0.62–2.85

0.471

 Poorly differentiated (G3)

4.95

1.97–12.42

0.001

 ER negative

1.78

0.76–4.15

0.182

 No chemotherapy

0.37

0.16–0.85

0.019

 No endocrine therapy

1.37

0.59–3.19

0.466

Multivariate analysis regarding overall survival

 DTC

1.70

0.68–4.21

0.255

 Poorly differentiated (G3)

4.44

1.67–11.81

0.003

 ER negative

1.05

0.38–2.90

0.920

 No chemotherapy

0.59

0.19–1.86

0.370

 No endocrine therapy

1.12

0.33–3.79

0.851

BC breast cancer, HR hazard ratio, CI confidence interval, DTC disseminated tumor cells, ER estrogen receptor

Discussion

Prognostic factors are used to differentiate between patients with favorable and unfavorable prognosis and to assign appropriate therapies. In BC patients most prognostic factors refer to the biology of the primary tumor itself (e.g., grade, ER, PgR, HER2). Only a few, including axillary lymph node status and tumor cell detection in distant organs, are morphological correlates of tumor spread and thus true markers of dissemination. Immunocytochemical detection of DTC in BM of early BC patients is an independent prognostic factor. All investigators have linked the detection of epithelial cells inside bone marrow with a poor prognosis.210 However, almost all publications are flawed because of small numbers of patients investigated or inadequate follow-up times. Here we describe the largest group of patients in one study with the longest follow-up time who were treated at the Breast Unit of the University of Heidelberg. With a median observation time of 82.0 months, detection of DTC proved to be an excellent prognostic factor for both overall as well as disease-free survival. Ten-year-survival-rate was 66.9 % for patients with DTC in BM in comparison to 81.8 % for patients without. DTC in BM are a strong prognostic factor for the development of metastasis to bone as well as visceral organs. Furthermore, the local recurrence rate was higher in patients with disseminated tumor cells. These results have been confirmed by other authors.35

A subgroup-analysis was not able to show that in patients with excellent prognosis (T1N0) identification of DTC can discriminate between high or low risk for the development of distant metastasis. Although in an earlier publication we have discussed studying tumor cell detection in bone marrow as a prognostic factor instead of nodal status in patients with small breast tumors, the data presented here do not support this hypothesis.2 However, in all patients detection of DTC represented a worse prognosis and a shorter distant disease-free survival. These observations make us believe that in a majority of BC patients DTC in BM are the missing link of metastatic spread several years after primary disease.

The longest follow-up period for DTC in bone marrow has been published by Mansi and colleagues.4 In a first analysis after 6 years they found tumor cell detection in bone marrow to be an independent prognostic factor for both disease-free and overall survival.4 In a subsequent investigation at a median of 12.5 years after the start of the study, tumor cell detection in BM was still prognostically significant, but was no longer an independent factor.17 The detection of epithelial cells therefore appears only to be predictive for early progression. Braun et al. confirmed those results although the number of investigated patients was smaller (n = 552) and the BM preparation technique and antibodies used were different.18 Patients with cytokeratin-positive cells in BM had a significant reduction in disease-free and overall survival. A combination with the prognostic factor nodal status could clearly define patients with a very poor prognosis (node-positive with micrometastases) and patients with an excellent prognosis (node-negative without micrometastases). Patients with either involved nodes or micrometastases in bone marrow had a prognosis in between.

A limitation of our study is the high detection rate of MUC-1 positive cells inside bone marrow which varied between 43 and 55 % over the time, although based on a European Working Group initiative we checked immunostained cells for additional morphologic criteria of malignancy (e.g., nuclear to cytoplasmatic ratio and cell size) to reduce false-positive results.16 Nevertheless, our DTC detection rates are higher than those in other studies, showing detection rates of epithelial cells inside bone marrow of 16 up to 42 %.5,11,17,19 In a very recent report, DTC were detected in 39 % of the patients.20 Those analyses, however, were mostly done with cytokeratin (CK), epithelial membrane antigen (EMA) or epithelial cell adhesion molecule (EpCam) antibodies, making a comparison to our findings more difficult. On the other hand, the observed impact on disease-free as well as overall survival was comparable to our data regardless of the technique of detection used. Clearly, as to date no general consensus has been established to standardize the laboratory protocols for micrometastasis detection, a comparative evaluation of the various DTC immunostaining techniques is highly desirable with respect to sensitivity and specificity rates of the different antibodies available presently.12 Such studies will be mandatory to arrive at a definitive recommendation regarding which antibody to use.

However, even if we were able to identify patients with a higher risk of distant metastases by DTC detection, it is difficult to draw an appropriate clinical conclusion from this information. Longitudinal observational studies showed that DTC persist in 15–30 % of breast cancer patients despite intensive adjuvant therapies like taxane-based chemotherapy.1,18,19,2123 The prognosis for those patients in regard to the development of distant metastasis is even worse.24,25 Whether DTC are cells of an early or a late metastatic event is not well known but there seems to be evidence that the metastatic capability of breast cancer cells can be determined through the identification of a “metastatic phenotype” in primary BC cells by genome and transcriptome analysis.26,27 Furthermore, Balic et al. 28 found out that DTC in bone marrow of primary BC patients have a stem cell-like signature (CD44+CD24/low) which might be a possible explanation for the resistance of those cells against adjuvant therapies.

In conclusion, detection of DTC in primary breast cancer patients’ bone marrow seems to be a surrogate marker of tumor spread and an independent prognostic factor for disease-free and overall survival. Due to the presence of BM DTC, patients with a clinically poorer outcome can be identified at primary surgery. If we do, however, accept that the detection of single tumor cells in bone marrow—regardless of the laboratory technique used—is a prognostic factor in primary BC patients and the link to distant metastases, we have to think about the therapeutical consequences. We therefore suggest that DTC can not only be used as a prognostic factor but also as a monitoring tool in clinical trials. Preferably, the aim of any study concept regarding DTC should be to investigate, whether breast cancer patients with favorable prognostic parameters (node negative, low grading, hormone receptor positive) are to be treated with adjuvant systemic therapy or not. Nevertheless, however, further studies are needed to develop standardized techniques for DTC identification/quantification and to analyze the clinical outcome in respect to relevant prognostic variables as tumor subtype and Her-2/neu status.

CONFLICT OF INTEREST

None.

Copyright information

© Society of Surgical Oncology 2012