Breast Cancer Research and Treatment

, 128:137

Mitotic activity and bone marrow micrometastases have independent prognostic value in node positive breast cancer patients

Authors

  • Bjørnar Gilje
    • Department of Haematology and OncologyStavanger University Hospital
  • Oddmund Nordgård
    • Department of Haematology and OncologyStavanger University Hospital
    • Laboratory for Molecular BiologyStavanger University Hospital
  • Kjersti Tjensvoll
    • Department of Haematology and OncologyStavanger University Hospital
    • Laboratory for Molecular BiologyStavanger University Hospital
  • Emiel A. M. Janssen
    • Department of PathologyStavanger University Hospital
  • Håvard Søiland
    • Department of SurgeryStavanger University Hospital
  • Rune Smaaland
    • Department of Haematology and OncologyStavanger University Hospital
    • Laboratory for Molecular BiologyStavanger University Hospital
    • Institute of MedicineUniversity of Bergen
    • Department of PathologyStavanger University Hospital
    • The Gade InstituteUniversity of Bergen
Clinical Trial

DOI: 10.1007/s10549-011-1487-1

Cite this article as:
Gilje, B., Nordgård, O., Tjensvoll, K. et al. Breast Cancer Res Treat (2011) 128: 137. doi:10.1007/s10549-011-1487-1

Abstract

The purpose of this article is to investigate the prognostic value of the mitotic activity index (MAI) and the presence of disseminated tumor cells (DTCs) in bone marrow (BM), in clinically operable breast cancer patients. We compared routinely assessed MAI, classic prognosticators and BM DTCs, detected by a real-time RT-PCR multimarker assay including cytokeratin 19, mammaglobin A and TWIST1 mRNA, in 179 consecutive patients with operable breast cancer. Over a median follow-up of 96 months (range: 1–126 months), 31 (17.3%) patients experienced a systemic relapse and 26 (14.5%) died of breast cancer-related causes. MAI (≥ 10) was strongly associated with breast cancer-related death in lymph node (LN)-negative patients (hazard ratio (HR): 7.0, confidence interval (CI) 1.74–27.9), whereas both BM DTC-status (HR: 3.3, CI 1.25–8.52) and MAI (HR: 3.1, CI 1.08–8.8) were significant in LN-positive patients. With multivariate Cox regression, MAI was the only significant predictor of breast cancer-specific survival (HR 7.0, CI 1.7–27.9) in LN-negative patients. In LN-positive patients, both BM DTC-status and MAI were strong independent predictors of breast cancer-specific survival (HR 3.3, CI 1.25-8.49 and HR 3.1, CI 1.1–8.9), respectively. Where, however, MAI and BM DTC-status as single parameters were replaced by a combination of these, this showed to be the most significant prognostic marker in both LN-negative (HR 7.7, CI 1.2-50) and LN-positive (HR 6.0, CI 1.4 to 26.4) patients with regard to breast cancer-specific survival. A combination of MAI and BM DTC detection identified both LN-negative and LN-positive breast cancer patients with poor prognosis.

Keywords

Breast cancerMAIDTCProliferationBone marrow

Introduction

Over the past decades, a tremendous effort has been made to improve breast cancer treatment. Despite this, breast cancer nonetheless remains a leading cause of death. Thus, classical prognostic factors such as TNM stage, estrogen receptor (ER) and progesterone receptor (PgR) status, histologic grade, and HER-2 amplification still seem to be insufficient to achieve optimal adjuvant treatment decisions with regard to patient cure. According to the St Gallen guidelines, over- and undertreatment will occur in 85 and 20% of LN-negative patients, respectively [1]. This has led to an increased effort in identifying new prognostic and predictive factors, including both histological and molecular markers. The transcript profiling of breast cancer tumors described by Sørlie et al. [2] classifies breast cancers into five subtypes with very different prognoses, which also contributes to a better understanding of the biology of the disease. Subsequently, supervised transcript profiling experiments have been used to develop standardized molecular prognostic indicators, e.g., Mammaprint and Oncotype DX [3, 4].

There is considerable evidence that detection of disseminated tumor cells (DTCs) in bone marrow (BM) from breast cancer patients is a strong prognostic factor, independent of classical prognostic factors. This has been demonstrated by both immunhistochemical and reverse transcription PCR-based detection of DTCs [510]. However, a considerable subgroup of patients without detected DTCs still experience disease recurrence (6–13%) [5, 7, 9, 10], suggesting that DTC detection in combination with known or novel prognostic factors should be added for optimal prediction of clinical outcome.

An alternative approach to prognostic classification is to study the portion of proliferating and dividing cancer cells; i.e., those cells undergoing mitosis. This is partly done when assessing tumor grade according to the Nottingham grading system, which combines the number of mitoses, degree of tubular formations and nuclear pleomorphism. There are, however, strong indications that the degree of tubular formations and nuclear pleomorphism do not add significant prognostic information to the mitotic count [11]. Moreover, the mitotic component of tumor grade is usually assessed by an overall visual impression, which is highly observer-dependent and thus prone to low accuracy and reproducibility [12]. The mitotic acitivity index (MAI), however, which is determined by a standardized quantification protocol, represents a more objective and reproducible approach to measuring proliferation [13]. The MAI has also been demonstrated to be a strong, independent predictor of clinical outcome, and there is evidence indicating that a high MAI may predict success of chemotherapy in LN-negative breast cancer patients [11, 14]. Moreover, a recent prospective analysis showed that the prognostic value of the MAI exceeded and overshadowed that of Adjuvant! Online and the treatment guidelines of the Norwegian Breast Cancer Group [15]. In agreement with this, Ki-67 has recently been shown to predict response to docetaxel treatment in LN-positive/ER-positive breast cancer patients [16, 17]. These observations suggest a prognostic and predictive value of proliferation markers in general.

We have previously demonstrated by a multimarker real-time RT-PCR panel consisting of the cytokeratin 19 (CK19), mammaglobin A (hMAM) and TWIST1 transcripts, that detection of DTCs in pre-operative BM samples predicts clinical outcome in operable breast cancer patients [18]. In this study, we investigate the novel combination of a proliferation marker from the primary tumor (MAI) and BM DTC-status in terms of prognostic value.

Patients and methods

This study is reported according to the recommendations for tumor marker prognostic studies [19].

Patients and control samples

We recruited 234 consecutive clinically operable female patients (median age 56 years, range 25–86 years) from the Stavanger University Hospital for this prospective study from 1998 to 2000 as previously described [5, 20]. We have included a REMARK diagram illustrating patient selection (Fig. 1). Fifty-five patients were excluded from the study due to DCIS/LCIS (25 patients), suspected malignancy not confirmed (7 patients), primary metastatic disease diagnosed within 1 month after surgery (1 patient), missing BM samples (10 patients) and technically inadequate histologic material for MAI assessment (12 patients), leaving 179 patients for statistical analysis. BM samples from 26 healthy female volunteers constituted the control group for the BM-analyses. Written informed consent was obtained from all participants, and the project was approved by the regional ethical committee. In brief, 20 ml were drawn unilaterally from the posterior iliac crest immediately prior to surgery. The patients underwent either a modified radical mastectomy or lumpectomy with breast conserving surgery, and all had level I and II axillary lymph node dissection. The treatment and clinical follow-up of the patients were done systematically as defined by the official Norwegian guidelines as previously described [5, 20]. In brief, high risk patients (axillary lymphnode involvement and/or tumors larger than 2 cm with histology grade II–III) were treated with adjuvant systemic therapy. Patients between 55 and 65 years of age received six cycles of CMF (cyclophosphamide, metotrexate and 5-fluorouracil) and patients younger than 55 years received six cycles of FEC (5-fluorouracil, epirubicin and cyclophosphamide). All high risk patients received 20 mg tamoxifen daily for 5 years if positive or uncertain estrogen or progesterone receptor status was determined by routine immunohistochemistry. The end of the follow-up period was October 2008, and the median follow-up time was 96 months (range 1–126), i.e., >8 years. Sample size calculations in the patient inclusion of this study have been presented previously [20].
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-011-1487-1/MediaObjects/10549_2011_1487_Fig1_HTML.gif
Fig. 1

REMARK diagram illustrating patient selection

Assessment of the mitotic activity index (MAI)

Following the Multicenter Mammary Carcinoma Project (MMMCP) protocol, the total number of well-defined mitotic figures was counted by trained technicians at ×400 magnification (objective 40, field diameter 450 μm at specimen level) in 10 consecutive neighboring fields of vision in the most poorly differentiated peripheral area of the tumor (=measurement area, representing a total area of 1.59 mm2) [13, 21]. Fields with necrosis or inflammation were avoided, and doubtful structures were ignored. The resulting total number of mitoses in the 10 fields of vision was defined as the mitotic activity index (MAI). The MAI is a continuous variable, and according to previous studies, the most important prognostic threshold is 10 mitotic counts, with MAI < 10 indicating a favorable prognosis and MAI ≥ 10, a worse prognosis (denoted MAI) [11, 22].

Other cut-off values of MAI have been utilized in correlation with treatment response, including MAI3 (0–2 is negative, ≥3 is positive) which is discussed in this article and is thus specifically denoted MAI3.

RNA isolation and cDNA synthesis

BM samples were collected unilaterally from the posterior iliac crest, and nucleated cells isolated from buffy coat as described previously [23]. Isolation of total RNA, DNase I treatment and reverse transcription were also performed as previously described [20].

DTC detection by quantitative RT-PCR

CK19, hMAM and TWIST1 mRNA were quantified by real-time RT-PCR as previously described [5, 18, 20]. The relative levels of these DTC surrogate markers in patient BM samples were compared to thresholds for positivity based on the highest levels in the normal control group. A BM sample was defined as DTC positive if at least one of the surrogate markers was positive, this multimarker panel being introduced by Tjensvoll et al. [18], where also the TWIST assay is described in detail.

Statistical analyses

The statistical analyses were performed in SPSS version 15.0 (www.spss.com). P values less than 0.05 were considered significant. Pearson χ2 or Fisher’s exact test were used to compare categorical data. Survival curves were calculated by the Kaplan–Meier method, and tested for difference using the Mantel–Cox log-rank test. The survival probabilities were estimated as the time intervals from primary surgery to systemic recurrence of the disease (denoted systemic recurrence-free survival) or death related to progression of breast cancer (denoted breast cancer-specific survival). Univariate and multivariate Cox proportional-hazards regressions were used to investigate the effects of MAI, BM DTC-status, LN status, tumor size, tumor grade, adjuvant chemotherapy, adjuvant endocrine therapy, estrogen receptor (ER) status, and progesterone receptor (PgR) status on systemic recurrence-free survival and breast cancer-specific survival. All these features were categorized according to previously proven strongest prognostic thresholds, and used as categorical features in the Cox regression analyses. The multivariate Cox regression analyses were performed using both forward and backward stepwise selection, which yielded similar results. We describe the results of the forward Wald Cox regression. The enter limit was defined to 0.05 and the removal limit was 0.10. The effect of each variable in these models was assessed by the Wald test and described by the hazard ratio, with a 95% confidence interval. The MAI and BM DTC-status as single parameters were omitted from the multivariate Cox regressions involving the combination of the two parameters.

Results

Univariate survival analysis

The clinicopathological characteristics of the 179 included patients are presented in Table 1 (122 LN-negative patients) and Table 2 (57 LN-positive patients). Systemic relapses occurred in 31 (17%) patients, and 26 (14%) patients died of the disease during a median follow-up of 96 months (range 1–126). Stratification according to LN status revealed systemic relapses in 12/122 (10%) of the LN-negative patients and of these, 9 (7%) were breast cancer-related deaths. Univariate log-rank survival analysis demonstrated that MAI was a highly significant prognostic factor with regard to both systemic recurrence-free survival (P = 0.02) and breast cancer-specific survival (P = 0.001) in this patient group (Table 1). BM DTC-status, however, was not significant for the entire LN-negative group, but still significant in LN-negative patients with MAI ≥ 3 (P = 0.045 for systemic recurrence-free survival). Of the classical prognostic factors tested, ER status was the only parameter significantly associated with prognosis in the LN-negative patients. Among the 57 LN-positive patients, systemic relapse occurred in 19 (32%) patients, and of these, 17 (29%) died from the disease. Univariate log-rank survival analysis revealed that BM DTC-status was the most significant prognostic factor in this group, with regard to both systemic recurrence-free survival (P = 0.002) and breast cancer-specific survival (P = 0.011, Table 2). None of the classical prognosticators, nor MAI, were significant prognostic factors in the LN-positive group.
Table 1

Survival analysis of prognostic factors on systemic recurrence-free survival and breast cancer-specific survival in LN-negative patients

 

Systemic recurrence-free survival

Breast cancer-specific survival

No. of events/patients at risk

Log-rank P

HR

95% CI

No. of events/patients at risk

Log-rank P

HR

95% CI

Age

 <55 years

6/48 (87.5%)

   

5/48 (89.6%)

   

 55–70

5/57 (91.2%)

 

0.7

0.21–2.23

4/57 (93.0%)

 

0.7

0.18–2.51

 >70

1/17 (94.1%)

0.75

0.6

0.07–4.58

0/17 (100%)

0.43

nd

 

Tumor size

 pT1

8/96 (91.7%)

   

6/96 (93.8%)

   

 pT2

4/24 (83.3%)

 

2.3

0.69–7.65

3/24 (87.5%)

 

2.2

0.55–8.75

 pT3 and pT4

0/2 (100%)

0.33

nd

nd

0/2 (100%)

0.48

nd

 

Histologic grade

 Grade 1

2/54 (96.3%)

   

0/54 (100%)

   

 Grade 2

7/47 (85.1%)

 

4.5

0.93–21.68

6/47 (87.2%)

 

nd

 

 Grade 3

3/21 (86.7%)

0.11

4.3

0.72–25.91

3/21 (85.7%)

0.017

nd

 

ER status

 ER-positive

7/102 (93.1%)

   

4/102 (96.1%)

   

 ER-negative

5/20 (75.0%)

0.08

0.24

0.08–0.76

5/20 (75.0%)

0.001

0.14

0.04–0.52

PgR status

 PR-positive

6/58 (89.7%)

   

5/64 (92.2%)

   

 PR-negative

6/64 (90.6%)

0.88

1.1

0.35–3.39

4/58 (93.1%)

0.81

0.85

0.23–3.17

Chemotherapy

 Yes

3/15 (89.0%)

   

6/107 (94.4%)

   

 No

9/107 (91.6%)

0.15

2.5

0.69–9.36

3/15 (80.0%)

0.048

3.7

0.92–14.8

Endocrine therapy

 Yes

1/13 (92.3%)

   

8/109 (92.7%)

   

 No

11/109 (89.9%)

0.84

0.8

0.10–6.27

1/13 (92.3%)

0.90

1.1

0.14–9.18

MAI

 0–9 (MAIneg)

6/91 (93.4%)

   

3/91 (96.7%)

   

 >9 (MAIpos)

6/31 (80.6%)

0.02

3.4

1.11–10.66

6/31 (80.6%)

0.001

7.0

1.74–27.9

BM DTC-status

 Neg

9/107 (91.6%)

   

7/107 (93.5%)

   

 Pos

3/15 (80.0%)

0.18

2.4

0.65–8.81

2/15 (86.7%)

0.40

2.0

0.41–9.43

MAI-BM DTC-status

 Neg

4/79 (94.9%)

   

2/79 (97.5%)

   

 Pos

8/43 (81.4%)

0.012

4.1

1.25–13.76

7/43 (83.7%)

0.004

7.2

1.49–34.5

MAI3

 0–2

0/44 (100%)

   

0/44 (100%)

   

 >2

12/78 (84.6%)

0.06

nd

 

9/78 (88.5%)

0.02

nd

 
Table 2

Survival analysis of prognostic factors on systemic recurrence-free survival and breast cancer- specific survival in LN-positive patients

 

Systemic recurrence-free survival

Breast cancer-specific survival

No. of events/patients at risk

Log-rank P

HR

95%CI

No. of events/patients at risk

Log-rank P

HR

95%CI

Age

 <55 years

10/33 (69.7%)

   

8/33 (75.8%)

   

 55–70

4/11 (63.6%)

 

1.0

0.29–3.10

4/11 (63.6%)

 

1.2

0.36–4.02

 >70

5/13 (61.5%)

0.65

1.6

0.54–4.66

5/13 (61.5%)

0.47

2.0

0.65–6.05

Tumor size

 pT1

9/26 (65.4%)

   

7/26 (73.1%)

   

 pT2

7/24 (70.8%)

 

1.0

0.38–2.75

7/24 (70.8%)

 

1.3

0.47–3.85

 pT3 and pT4

3/7 (57.1%)

0.87

1.4

0.38–5.22

3/7 (57.1%)

0.69

1.8

0.45–6.80

ER status

 ER-positive

13/43 (57.1%)

   

11/43 (74.4%)

   

 ER-negative

6/14 (57.1%)

0.24

0.6

0.21–1.48

6/14 (57.1%)

0.12

0.5

0.17–1.25

PgR status

 PR-positive

8/30 (73.3%)

   

7/30 (76.7%)

   

 PR-negative

11/27 (59.3%)

0.23

0.6

0.23–1.44

10/27 (63.0%)

0.21

0.5

0.21–1.43

Chemotherapy

 Yes

10/37 (73.0%)

   

8/37 (78.4%)

   

 No

9/20 (55.0%)

0.16

0.5

0.21–1.30

9/20 (55.0%)

0.08

0.4

0.17–1.14

Endocrine therapy

 Yes

13/40 (67.5%)

   

11/40 (72.5%)

   

 No

6/17 (64.7%)

0.56

0.8

0.29–1.98

6/17 (64.7%)

0.33

0.6

0.22–1.65

MAI

 0–9 (MAIneg)

7/28 (75.0%)

   

5/28 (82.1%)

   

 >9 (MAIpos)

12/29 (58.6%)

0.09

2.2

0.87–5.62

12/19 (58.6%)

0.027

3.1

1.08–8.80

BM DTC-status

 Neg

10/42 (76.2%)

   

9/42 (78.6%)

   

 Pos

9/15 (40.0%)

0.002

3.7

1.50–9.22

8/15 (46.7%)

0.011

3.3

1.25–8.52

MAI-BM DTC-status

 Neg

3/21 (85.7%)

   

2/21 (90.5%)

   

 Pos

16/36 (55.6%)

0.01

4.4

1.29–15.3

15/36 (58.3%)

0.007

6.0

1.37–26.4

MAI3

 0–2

5/15 (66.7%)

   

4/15 (73.3%)

   

 >2

14/42 (66.7%)

0.74

1.2

0.43–3.32

13/42 (69.0%)

0.58

1.4

0.45–4.24

We also combined MAI and BM DTC-status to create a marker which was considered positive if either BM DTC-status and/or MAI were positive (positive MAI/BM DTC-status). This combination was highly significant with regard to both systemic recurrence-free survival (P = 0.012 in LN-negative patients and P = 0.001 in LN-positive patients) as well as breast cancer-specific survival (P = 0.004 in LN-negative patients and P = 0.007 in LN-positive patients) as shown in Tables 1 and 2.

Kaplan–Meier analyses confirmed that BM DTC-status was highly prognostic among the LN-positive patients, but was insignificant in LN-negative patients (Fig. 2). Moreover, MAI identified both LN-negative and LN-positive patients with reduced breast cancer-specific survival, but did not identify LN-positive patients with high risk of systemic recurrence, according to these analyses. Similarly, Kaplan–Meier survival plots confirmed the exceptional prognostic strength of the combined MAI/BM DTC parameter (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-011-1487-1/MediaObjects/10549_2011_1487_Fig2_HTML.gif
Fig. 2

Kaplan Meier estimates of breast cancer-specific survival according to BM DTC-status, MAI status, and MAI/BM DTC-status

When we stratified the patients according to treatment groups, similar prognostic effects of MAI were found with the exception of the chemotherapy only group. The groups were no adjuvant treatment (P = 0.0049/P = 0.003), only endocrine adjuvant treatment (P = 0.109/P = 0.091), chemotherapy only (P = 0.53/P= 0.53) and the combination of chemotherapy and endocrine treatment (P = 0.052/P = 0.33), where the P values were obtained by MAI-stratified log rank tests of systemic recurrence-free survival and breast cancer specific survival, respectively. In all treatment groups, patients with MAI < 10 performed worse than patients with MAI ≥ 10 with regard to both systemic recurrence-free survival and breast cancer specific survival. However, the number of patients within each group was low, leaving the difference statistically significant only in the untreated group where the number of patients was sufficiently high.

Mitotic activity index and bone marrow status are independent prognostic factors

Multivariate Cox regression analyses showed that BM DTC-status, MAI and LN status were independent prognostic factors with regard to both breast cancer-specific survival and systemic recurrence-free survival (Table 3). Although BM DTC-status was not significant in the LN-negative group, it was the strongest independent prognostic factor for the LN-positive patients. MAI, however, was not significant for the LN-positive patients with regard to systemic relapse (P = 0.095), but was significant for breast cancer-specific survival (P = 0.035)
Table 3

Multivariate Cox regression with regard to systemic recurrence-free survival and breast cancer-specific survival

 

Parameter

Hazard ratio

95% CI

P value

LN-positive patients

 Systemic recurrence-free survival

BM DTC-status (pos. vs. neg.)

3.7

1.50–9.16

0.005

 Breast cancer-specific survival

BM DTC-status (pos. vs. neg.)

3.3

1.25–8.49

0.016

MAI

3.1

1.08–8.87

0.035

LN-negative patients

 Systemic recurrence-free survival

ER (pos vs. neg)

0.24

0.077–0.76

0.015

 Breast cancer-specific survival

MAI

7.0

1.74–27.92

0.006

All patients

 Systemic recurrence-free survival

BM DTC-status (pos vs. neg)

3.3

1.57–6.84

0.002

Lymph node status (N > 0 vs. N0)

2.7

1.28–5.81

0.009

 

MAI

2.7

1.29–5.63

0.008

 Breast cancer-specific survival

MAI

4.3

1.84–10.26

0.001

 

BM DTC-status (pos vs. neg)

2.9

1.31–6.56

0.009

 

Lymph node status (N > 0 vs. N0)

2.8

1.22–6.59

0.016

BM DTC-status bone marrow disseminated tumor cells-status, MAI mitotic activity index with 10 or more mitoses defined as positive, ER estrogen receptor

Importantly, the combined MAI/BM DTC-status was found to be the most significant prognostic marker in LN-positive (HR 4.4, CI 1.3–5.3/HR 6.0, CI 1.4 to 26.4) patients for both systemic recurrence-free survival and breast cancer-specific survival, respectively. In the LN-negative patients, the HR of combined BM DTC-status/MAI is slightly higher than the MAI alone.

Discussion

Proliferation measurement by MAI assessment in the primary tumor and BM DTC detection represent two well established, yet completely different, approaches in predicting the clinical outcome of breast cancer. Thus, it is not surprising that these parameters performed differently in terms of prognostic stratification in the patients in our study. We have demonstrated that proliferation measurement according to the MAI definition is a significant independent prognostic factor in both LN-negative and LN-positive patients in our unselected patient cohort, although the prognostic impact appeared to be strongest in the LN-negative group. This finding is in accordance with previous studies, where MAI was the strongest prognostic factor of all variables analyzed by multivariate Cox regression in LN negative patients [11, 15]. In this study, however, MAI was not found to be significant with regard to systemic recurrence-free survival, but highly significant with regard to breast cancer-specific survival. The explanation might be that LN-negative high-risk patients less than 55 years were treated with adjuvant chemotherapy (ACT), following the Norwegian Breast Cancer Group guidelines. We and others have shown before that ACT is very effective in these patients, especially when MAI > 3 [14]. This confounding factor might explain why MAI in this study did not predict systemic recurrence-free survival.

BM DTC-status, on the other hand, only reached statistical significance in the survival analysis of LN-positive patients, despite a clear trend also for LN-negative patients (Tables 1; 2). Contrary to this finding, Braun et al. [6] in their pooled analysis of 4,703 patients demonstrated a significant prognostic value of BM DTCs also among LN-negative patients. The lack of statistical significance for BM DTC status in survival analysis of our LN-negative patients may primarily be due to the lower number of events in this group. MAI was clearly significant in the LN-negative patients, and BM DTC-status was still significant in the subgroup of LN-negative patients with MAI ≥ 3, suggesting that MAI and BM DTC detection supplement each other. This hypothesis was confirmed by survival analyses of MAI and BM DTC-status in combination, demonstrating an even stronger prognostic impact of the combination compared with the single factors alone.

We asked whether the supplementary behavior of MAI and BM DTC-status might reflect a different biology in the poor-prognosis cancers identified by MAI assessment when compared to those identified by BM DTC detection only. Interestingly, there also seemed to be a difference between MAI and BM DTC-status regarding the time period to the predicted relapses, which may be explained by a different biology. Patients with MAI > 10 tended to relapse earlier and no relapses were registered after 75 months. In contrast, the relapses predicted by BM DTC detection were more evenly distributed over time (both early and late). Regarding tumor biology, excessive proliferation is a hallmark of almost all aggressive cancer types. However, recent evidence suggests that epithelial tumor cells need to undergo an epithelial-to-mesenchymal transition (EMT) in order to migrate and invade other organs [24]. Interestingly, this EMT process seems to be associated with reduced proliferation [25, 26], although only a minor fraction of the tumor cells in the margin of the tumor undergo EMT [26]. Hence, aggressive tumors with high proliferation may also harbor minor cell populations undergoing EMT, enabling spread to distant organs. On the other hand, disease recurrences many years after surgical removal of the primary tumor support the presence of dormant tumor cells with low proliferation over prolonged periods [27]. This is consistent with our finding that some patients with BM DTCs experience late relapses. Patients with slowly proliferating tumors might have minimal residual disease in the BM that remain below the clinical and radiologic detection level for years, which, however, ultimately at a later stage might cause a clinically detectable relapse.

When we investigated the LN-negative subgroup further, we found that 44 (36%) of the 122 patients had a very slowly proliferating tumor with MAI count < 3. No recurrences were found in these 44 patients, although 8 (18%) had positive BM DTC-status. It can thus be hypothesized that BM DTC-status might be unsuitable for very low risk patients (i.e., LN-negative with low proliferation, MAI < 3). The possible explanation for this being that very slowly proliferating primary tumors may lack metastatic potential. Thus, DTCs from these tumors may not reflect elevated risk of metastatic disease, but rather the presence of apoptotic tumor cells in the bone marrow, or metastatic cells with a too low proliferation rate to ever become clinically manifest. Alternatively, low proliferation and the presence of BM DTCs could indicate a slight risk for very late relapses but not aggressive early relapses. Nevertheless, this finding suggests that BM DTC-status might not be useful in prognostication of low proliferating LN-negative breast cancer tumors.

An alternative explanation for the additive value of MAI and BM-status, is the following: It has been suggested that all patients with invasive breast cancer may have distant metastases at the time of diagnosis, which explains why MAI of the primary tumor has prognostic value: high primary tumor MAI translates into a high number and aggressiveness of the metastases, which makes them grow and become clinically apparent. Finding BM micrometastases in a small volume of bone marrow taken with a biopsy sugggests that there are many BM micrometastases throughout the body, thereby increasing the chance that one of them may survive locally and grow. This interesting hypothesis requires further studies to be confirmed.

This is the first study investigating the combination of BM DTC-status with MAI. The superior prognostic impact of this combination suggests its future application in routine diagnostics. MAI is easy to assess and more reproducible than conventional mitosis counting [11, 21], and BM DTC detection is also feasible, reproducible and easily automated. However, several different methods for DTC detection are in use; quantitative PCR and immunocytochemical methods being the most frequent [2830]. In order to establish a combined marker composed of both BM DTC-status and MAI that is applicable to routine diagnostics, standardization of methods for use in DTC detection must be established.

In conclusion, MAI was confirmed to be a strong, independent prognostic factor in both LN-positive and LN-negative patients. Detection of DTCs in BM provided supplementary information to MAI, and a combination of MAI and BM DTC-status was shown to be an even stronger independent prognostic factor than the two factors alone. This highly prognostic combination may also be relevant when deciding administration of adjuvant treatment, possibly reducing both over- and under-treatment. Future clinical trials of adjuvant therapy or intensified adjuvant therapy administered according to combined MAI and BM DTC-status are thus warranted.

Acknowledgments

This study was supported by the Norwegian Regional Health Collaboration of Health West, the Norwegian Cancer Society and Stavanger University Hospital.

Supplementary material

10549_2011_1487_MOESM1_ESM.ppt (192 kb)
Supplementary material 1 (PPT 190 kb)

Copyright information

© Springer Science+Business Media, LLC. 2011