Background
There is considerable evidence that detection of disseminated tumor cells (DTCs) in pre-operative bone marrow (BM) samples from non-metastatic breast cancer patients identifies a patient population at high risk for disease recurrence [
1‐
5]. DTCs have also been found in the BM after surgery, both before and after adjuvant treatment [
6‐
12]. The shedding of cells observed from primary tumors should be expected to end after removal of the tumor by surgery. Thus, DTCs detected after radical surgery in patients without evidence of distant metastases must originate from occult metastases or be able to persist in the BM after surgery. DTCs which persist post-operatively after surgery, and even after completion of adjuvant treatment, may be enriched for tumor cells with better capability to survive, and also grow, in the secondary site. Accordingly, it has been demonstrated that a large fraction of the breast cancer DTCs has a stem cell-like phenotype [
13,
14], which may cause resistance to conventional chemotherapeutic drugs [
15]. Thus, both a prognostic and a predictive value of BM DTC detection after surgery seem likely. Furthermore, one may hypothesize that repeated BM sampling in order to detect DTCs, in particular following administration of adjuvant treatment, may improve the prediction of disease recurrence and the selection of patients who might benefit from secondary or intensified adjuvant treatment.
There is limited evidence of the clinical usefulness of the suggested repeated BM sampling. Two studies, both using immunocytochemical detection methods, have reported the results of repeated BM sampling performed in women who were recurrence-free 2–3 years after primary diagnosis. The presence of DTCs in this group of patients significantly predicted shorter distant disease-free survival, but the prognostic impact seemed similar to that obtained from pre-operative analyses [
6,
7]. To select patients who would benefit by secondary adjuvant treatment, repeated BM sampling at an earlier time might be preferable. One such study has been reported by Daskalaki
et al. (2009), with samples collected before and after adjuvant chemotherapy [
10]. In contrast to the two prior studies, this group used real-time RT-PCR quantification of cytokeratin 19 (CK19) transcripts to detect DTCs. They observed a survival difference according to DTC status both before and after chemotherapy, however, the difference between the DTC positive and negative groups was statistically significant only for the BM samples obtained before chemotherapy [
10].
In the present study, we have used real-time RT-PCR quantification of a multimarker (MM) mRNA panel to detect DTCs in BM. The use of our MM mRNA panel is expected to result in high sensitivity since the markers may be differentially expressed in breast cancer cells. Previously we have demonstrated, using our MM mRNA panel, that detection of DTCs in pre-operative BM samples predicts clinical outcome in non-metastatic breast cancer patients [
16]. The present study is the first reporting repeated post-operative BM samples from non-metastatic breast cancer patients assessed by a MM quantitative RT-PCR assay for DTC detection. The BM samples were obtained three weeks and six months after surgery from 154 patients. Having 98 months (>8 years) median follow-up data for the patients, we have evaluated the prognostic significance of persistent DTCs in BM after surgery, and compared the prognostic and predictive information associated with the different sampling time points.
Discussion
We have previously demonstrated that the presence of DTCs, in terms of MM mRNA levels above the established cut-off values, in BM samples obtained prior to breast cancer surgery provide significant prognostic information (HR = 3.59,
p = 0.001 for systemic-recurrence-free survival) [
16]. In the present study, we have shown that DTCs in BM samples obtained three weeks and/or six months after surgery provide similar prognostic information as the DTC status in pre-operative BM, with hazard ratios in the range 5.8–6.8 (Table
4). These results were somewhat unexpected since the presumptive passive shedding of cells from primary tumors gives reason to believe that a higher proportion of clinically insignificant DTCs would be present in the BM before surgery. In this respect, BM sampling after surgery was expected to give more significant prognostic information. The results of the present study confirm the prognostic significance of DTCs after surgery as detected by our real-time RT-PCR assay. Most likely, the clinical importance of persistent DTCs in BM is even stronger than observed, as some of the included patients only agreed to one of the post-operative BM aspirations. Our results do not, however, suggest that patients with DTCs detected post-operatively have a prognosis that is inferior to patients with DTCs detected pre-operatively. Thus, a selection of DTCs with a higher capability of establishing clinical overt metastases after removal of the primary tumor, or even after six months of adjuvant treatment, is not supported. This finding is consistent with the report by Daskalaki
et al. (2009), in which the estimated prognostic effect of DTCs in BM samples collected shortly after adjuvant chemotherapy was similar to the effect in BM samples collected post-operatively, prior to chemotherapy [
10]. Similarly, the impact of DTCs in BM samples obtained 2–3 years after diagnosis seemed to have a similar magnitude as that observed pre-operatively [
2,
5‐
7].
The results in the present study do, however, suggest that a positive pre-operative DTC status, if confirmed in a second BM sample collected three weeks or six months post-operatively (double positive), predicts a particularly poor prognosis. The estimated 8-year systemic recurrence-free survival in this group of patients was in fact <20% (Figure
2A), as compared to >50% in the group defined as DTC positive based on post-operative BM samples alone (Figure
1A). A similar trend was reported by Wiedswang
et al. (2004), although it was not as striking as in our study. In their study, the estimated 5-year distant disease-free survival was <70% for the patients with DTC-positive BM both at diagnosis and after three years, as compared to ~80% for the whole group of patients who were DTC positive only in the second BM sample [
6]. However, the fact that they obtained the follow-up BM samples three years after surgery, and thus excluded the patients with recurrence during this 3-year interval, must be considered when comparing their estimated survival rates with ours. Our group of patients, with DTC positive BM both before and after surgery, included both LN-positive and LN-negative (N0) patients, the tumor size varying from T1-T4 with grade 1–3. Hence, these patients constituted a very heterogeneous group not necessarily destined to experience an unfavorable clinical outcome based on conventional prognostic factors.
It would be interesting to investigate whether DTC detection after adjuvant therapy could be a surrogate marker for evaluation of treatment efficiency. However, due to small patient numbers in the subgroups receiving adjuvant treatment it was not possible to conclude on the potential for monitoring or prediction of adjuvant treatment efficiency in this study. This important aspect should be addressed in new studies.
Discrepancies between the methods used to detect DTCs could make it difficult to compare studies using different methodologies. Immunocytochemistry has the advantages of being able to characterize cell size, cell shape and atypical enlargement of nucleus which may occur in malignant cells. However, due to the absence of tumor-specific targets, monoclonal antibodies against various epithelium-specific antigens, like the cytokeratins, are mostly used. In comparison, molecular methods are highly sensitive and may detect DTCs based on their expression of tumor-specific markers. One disadvantage is nevertheless that the cells cannot be morphologically characterized by the use of molecular methods. However, molecular profiling of breast cancer cells may be used to characterize and classify the tumor cells according to their protein, DNA or mRNA pattern.
In the present study we demonstrate that 15% of the patients with non-metastatic breast cancer have DTCs detected in BM after primary surgery (Table
4). Although the use of different methodologies complicates a direct comparison between the studies
, our results seem to correspond with Wiedswang
et al. (2004) also showing that 15% of the non-metastatic breast cancer patients had detectable DTCs in BM at a median 66 months from diagnosis [
6]. Janni
et al. (2005) detected DTCs in 13% of the patients after surgery [
7]. An European pooled analysis involving 676 breast cancer patients also corroborates our findings; 15.5% of the breast cancer patients having DTCs detected in BM after surgery, and this was found to be an independent predictor of subsequent reduced breast-cancer specific survival [
12]. The DTC frequency numbers published by Daskalaki
et al. (2009) using real-time RT-PCR for detection of CK19 mRNA-positive DTCs is, however, in contrast to these results. In their study, 58% of the non-metastatic breast cancer patients had BM DTCs detected after surgery, prior to chemotherapy, and 51% after chemotherapy [
10]. The low recurrence rate among the DTC-positive patients in their study suggests, however, that their assay is less specific with regard to clinical relevance. This may be partly due to a lower threshold for test positivity, as they have, as opposed to us, not used the highest determined normal BM level of the marker as a threshold [
10].
We observed that the contribution of CK19 to DTC detection, relative to hMAM and TWIST1, was lower after surgery than prior to surgery (Table
2). This observation supports the hypothesis that the DTC population after surgery may be enriched for tumor cells undergoing EMT (epithelial to mesenchymal transition), being able to persist in the BM after primary surgery, as decreased expression of cytokeratins is associated with the EMT process [
21]. A reduction in the number of CK19-positive patients after surgery has also been reported previously [
9]. These observations also contrast somewhat with Daskalaki
et al. (2009) regarding the high fraction of patients with CK19-positive DTCs in BM after surgery in their study [
10]. Thus, the change in the relative contribution of the three markers before and after surgery observed in our study is an interesting finding, and may suggest a differential marker expression in the DTCs detected at various sampling time points. Whether this reflects a general change in the expression profile of the DTC population present in the BM before and after surgery, may be a topic of future investigation.
Despite the establishment of the prognostic and predictive significance of BM DTCs [
2,
16,
19], detection of DTCs is yet to be adapted in clinical routine staging procedures. One of the reasons is the challenge of standardization of methods. This has been addressed for immunocytochemical methods [
22], but not in a similar way for RT-PCR methods. Additionally, patient discomfort and logistical challenges involved with BM sampling might be a hurdle to routine use. The rapid development of methods to detect circulating tumor cells (CTCs) in peripheral blood [
23‐
25], may offer a solution to this problem if CTCs are demonstrated to be of equivalent relevance to clinical outcome. However, in non-metastatic breast cancer patients only a limited number of studies have so far compared BM and peripheral blood examinations directly, by sampling both blood and BM at the same time point from the patients [
26‐
28]. Based on these studies the clinical significance of CTCs in peripheral blood seems less clear than for DTCs in BM in this patient group (reviewed in [
29]). Nevertheless, the clinical utility of the prognostic information from DTCs will also depend on the development of treatment options specifically targeting DTCs and CTCs. Detection and isolation techniques that allow a molecular characterization of DTCs may provide tools to guide novel, targeted therapies [
29]. However, presently ASCO guidelines state that the data from DTC detection in BM and CTC detection in blood, even in metastatic breast cancer patients, are insufficient to recommend assessment of minimal residual disease for the management of patients with breast cancer [
30]. Further validation in randomized trials is needed to confirm the clinical value of minimal residual disease detection.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
KT carried out all the qRT-PCR analyses, performed the statistical analyses and drafted the manuscript. SO did all the RNA purification, and participated in manuscript preparation. RH participated in sample collection, study design and manuscript preparation. Statistician JTK participated in the statistical evaluation, and preparation of the manuscript. BG participated in sample collection, and manuscript preparation. JMR gave scientific advice, contributed to data interpretation and manuscript preparation in addition to correcting the English grammar. RS is the group leader and contributed to result interpretation, supervision and manuscript preparation. ON participated in the study design, coordinated the study, gave statistical advice and contributed to the manuscript preparation. All authors read and approved the final manuscript.