In the past, the detection of DTC is most important in pathological staging of lymph node (LN) specimens. In the last few years, the existence of DTC in bone marrow has also been shown to provide prognostic information. Promising detection strategies for DTC in peripheral blood (PB) are also being examined. Regarding LN in breast cancer, the risk of metastatic disease is classically estimated by factors such as tumor size, tumor grade, estrogen (ESR1) and progesterone (PGR) receptor status, ploidy, ERBB2 (HER2/neu), cytokines, MMPs, NF-KB overexpression, and the number of positive axillary lymph nodes (ALN). Several studies have shown that the presence of DTC in ALN is the most powerful prognostic factor, being associated with significantly poor disease-free (DFS) and overall survival [1]. During the past few years, the theory of sentinel lymph node (SLN) has emerged. SLN biopsy gears mapping of one or two LNs that primarily drain the tumor (the sentinel nodes) and therefore are most likely to harbor the metastatic disease. SLN examination is now widely performed in breast cancer, as it can provide prognostic value with minimal associated morbidity in contrast to complete ALN dissection.

The prescreening of SLN with highly sensitive detection methods for micrometastases thus represents a promising strategy. Considering that significant numbers of LN-negative patients develop metastatic disease, the dependability of current staging procedures to detect DTC in LN has been uncertain.

Peripheral blood (PB) is historically one of the most potent diagnostic specimens. For example, circulating tumor markers have been evaluated in serum for years to give indicative values about metastatic or budding primary breast cancer. Serum markers may be good indications for tumor load, yet in most cases, they fail to provide information about minimal residual disease thus not up to mark. Technically speaking, PB appears as an ideal source for the monitoring of DTC. In fact, PB sampling is relatively trouble-free and can be done at frequent intervals (for instance, to permit an assessment of the patient’s recovery or potential to develop metastases). Several reports have demonstrated the presence of DTC in PB of patients with early-stage cancer without overt metastases [1, 24].

Contrary to PB sampling, blood marrow (BM) aspiration during surgery appears time-consuming and uncomfortable for the patient. However, among the distant organs, BM is a normal homing site for DTCs derived from breast cancer and other primary carcinomas, even in the absence of LN metastases or clinical signs of overt distant metastases [1]. Indeed, the screening rate of DTC in BM from nonmetastatic breast cancer patients has been demonstrated to be in the range from 0 % [44] to 100 % [45], and this corresponds to the variability of results obtained by the use of different techniques or marker genes. In a recent, large (more than 3,500 cases) study of stages I through III breast cancer patients, the incidence of DTC in BM detected by immunocytochemistry (ICC) ranged from 13 to 43 % [46]. The presence of DTC in BM may be supportive not only in predicting the development of bone metastases but also in predicting the development of metastases in other remote organs, such as the lung and liver. At present, however, it remains unsolved whether BM is a reservoir that allows for DTC to adapt and disseminate later into other organs or whether the presence of DTC in BM might reflect the general tendency of these cells to disseminate and survive in organs, rather than just in the BM. Until methods are developed to detect the presence of DTC in organs, such as the lung or liver, it will not be possible to distinguish between these two possibilities. The BM could serve as a reservoir in breast cancer and is supported by the presence of epithelial (cytokeratin-positive) cells in the PB of patients with overt remote metastases years after the removal of the primary tumor. This suggests that tumor cells could break from bone metastases to recirculate and disseminate to secondary tissues [1]. This “two-step” metastasis model could explain why the DTC in patients with overt metastases closely resemble each other genetically [47].

According to Ring et al. [48], in studies using antibody-based (cytometric) assays, cells with the characteristics of tumor cells have been shown in the PB of between 0 and 100 % of patients with operable (stages I through IIIa) breast cancer and in the PB of between 3 and 100 % of patients with metastatic disease. Several reports with nucleic acid-based techniques have shown cells with the characteristics of tumor cells in the PB of 0–88 % of patients with operable (stages I through IIIa) breast cancer and in 0–100 % of patients with metastatic disease. Along the same line, in a survey on a total of more than 3,500 stages I through III breast cancer patients, the incidence of DTCs in BM detected by ICC ranged from 13 to 43 % [46]. In fact, the detection rate of DTCs in BM from nonmetastatic breast cancer patients has been reported to be in the range from 0 % [44] to 100 % [45]. The variability of results obtained in DTC detection results from dramatic variations in methodology. Factors that may influence the results as heterogeneity of the studied populations may be:

Other factors such as sample handling and preparation, delay between collection and analysis, conditions of sample storage, and contamination with normal epithelial cells may influence the results. The introduction of skin cells into a PB sample at the time of venipuncture could lead to false-positive results. Many researchers advocate that the first few milliliters of sampled PB are discarded to avoid such contamination. It has also been suggested recently that false-positivity of SLN could result from iatrogenic displacement and transport of benign epithelial cells in patients with breast carcinoma [51]. Clearly, such epithelial cells do not represent metastasis.

Approaches by fluorescence microscopy (FM), ICC, and flow cytometry (FC) analysis aim to isolate and enumerate individual tumor cells. ICC is still a gold standard for DTC detection, and most of the available clinical data have been gathered by ICC screening, especially in BM [23]. An advantage of this approach is that it may permit further characterization of the cells at a molecular level, in terms of expression of key biological markers, such as ERBB2 (ERBB2 gene amplification estimated by FISH analysis) and morphological cell investigation. However, identification of intracellular targets, such as cytokeratins, by antibodies needs cell permeabilization. As a consequence, cell viability is lost, making the important discrimination of dead and viable DTC impossible. Since only viable cells might lead to metastasis, this valuable information cannot be evaluated [23].

Like IHC, FM and ICC are labor intensive and time-consuming, making these techniques too expensive for routine implementation. When compared with conventional, essentially qualitative FM and ICC, FC offers the advantage of a fully automated technique permitting quantitative measurements with high sensitivity, good resolution, speed, reproducibility, and statistical reliability. For breast tumors, the most used targets for antibody-based techniques are the cytokeratins. ERBB2, MUC1, and TACSTD1, the latter two being known under a variety of names, have also been used as antibody targets to isolate and/or identify DTC.

Two–color ELISPOT, an immunological assay based on enzyme-linked immunosorbent assay, has been recently used to detect DTC-secreting cathepsin D (CTSD) and mucin-1 (MUC1). However, antibody-based techniques have limitations. Many of the antibodies directed at epithelial and breast cancer cells are known to also stain hematopoietic cells, including cytokeratins (KRT19), TACSTD1, and MUC1. Nonspecific staining of plasma cells can also occur due to alkaline phosphatase reaction against the k and l light chains on the cell surface [52]. According to the antibody used, a false-positive detection rate of 1–3 % can be estimated [23]. Since tumor- and epithelial-specific cell marker antigens are expressed differentially in DTCs, the use of a panel of monoclonal antibodies may help to enrich DTCs and facilitate their finding [53].

PCR, either qualitative or quantitative, has been used to identify and characterize DTCs through the detection of genetic (allele-specific expression, microsatellite instability, loss of heterozygosity) and epigenetic alterations (methylation status) that are exclusively linked with cancer cells [54]. This includes the search for tumor-associated point mutations in oncogenes or tumor suppressors. This latter PCR approach, however, is complex by the substantial degree of genetic variability between tumors. For instance, TP53, the gene coding for p53, is mutated in about 25 % of breast tumors; however, more than 1,400 different mutations of this gene have been observed [55]. Of note, PCR has been used to screen free DNA within plasma. For instance, the analysis of DNA methylation status of specific genes (ESR1, APC, HSD17B4, HIC1, and RASSF1A) in serum of breast cancer patients has been shown to be of prognostic value [56]. The PCR-based measurement of RASSF1A methylation has been used for examining efficacy of adjuvant tamoxifen therapy [57]. However, this use of PCR is imperfect by poor specificity. This is due in part to the high stability of DNA in plasma when compared with mRNA [58]. As a result, it is unclear whether the free DNA that is amplified from plasma is from DTCs present in plasma or if the DNA is being shed from primary tumors, metastatic tumors, or from normal tissue [48]. To identify DNA gains and losses in single DTC, the technique of comparative genomic hybridization (CGH) is increasingly used [59].

Reverse transcription (RT)-PCR has been used to identify DTC through their expression of epithelial or breast cancer-associated mRNA transcripts. RT-PCR is generally more sensitive than antibody-based techniques but has also been hampered by false-positive results in samples from normal volunteers and from patients with hematological malignancies [48]. These false-positives stem from multiple sources, including issues with laboratory technique, primer selection, illegitimate expression of the target genes in normal cells, the existence of pseudogenes, or contamination (KRT19/CK19). When using assays based on RT-PCR for detection of DTCs, the balance between sensitivity and specificity must be considered. Generally, specificity decreases with the increase in sensitivity, and vice versa. One way to resolve this problem is to examine multiple tumor markers in samples. As mentioned below, multiplex RT-PCR assays have revealed a higher efficacy (in both sensitivity and specificity) in comparison with the assessment of single markers. To recover the reliability, especially the specificity of RT-PCR assays, quantitative RT-PCR (qRT-PCR) may be used. In addition, qualitative marker information, qRT-PCR uses cutoff values of marker transcript numbers, below which transcripts can be considered as tumor cell-derived. Moreover, when compared with conventional RT-PCR, qRT-PCR relies not only on primers but also on internal probes that specifically hybridize to the amplified sequences. In addition, due to the continuous measurement of the amplified signal, false-positive results, which could produce an abnormally shaped, nonlinear amplification curve, could be easily identified and removed [23]. Variations of the RT-PCR technique, such as nested RT-PCR and competitive nested RT-PCR, have also been used [60].

Fluorescence in situ hybridization (FISH) allows the detection of gene amplifications, for instance, ERBB2 amplification in breast cancer. FISH has been used to analyze genetic aberrations in DTC in BM. Considering the importance of ERBB2 as a novel target for successful antibody-based therapy, the use of FISH to identify ERBB2 amplification in DTC appears promising [61]. Among the cytologic methods that allow isolation and enumeration of individual cells, immunocytochemistry is the most widely used approach. Because of the absence of tumor-specific target antigens—most commonly antibodies against various epithelium-specific antigens such as cytoskeleton-associated cytokeratins—surface adhesion molecules or growth factor receptors are used for the screening of carcinoma cells [62]. The main advantage of cytologic methods is the opportunity to combine immunostaining with the morphology of the cells so that cell size and shape as well as the nucleusplasma relation might be predictable and illicit expression of the protein of interest in BM cells can be excluded.

The detection of DTCs in BM is not yet a routine part of the tumor staging in the clinical practice, but rising data anticipate a future role of DTC screening for risk estimation and therapeutic monitoring of breast cancer patients [63]. However, the detection rates of DTCs in BM from nonmetastatic breast cancer patients vary significantly [45]. This might reflect the different sensitivity, but also specificity, of the numerous detection methods and marker genes/proteins used thus far. The newly defined consensus concept for the detection of DTCs in BM, signifying enrichment of mononuclear cells from BM by Ficoll density gradient centrifugation and immunocytochemical detection of cytokeratin expression as standard procedure, should help overcome these troubles and provide the basis for future multicentric clinical trials. The researchers recommend the pan-anti-cytokeratin antibodies A45-B/B3 or AE1/AE3 against a wide spectrum of cytokeratins as standard application, thereby ensuring detection of DTCs also in cells that have downregulated the expression of individual cytokeratins in the course of epithelial–mesenchymal transition [42]. Microscopic screening of large amounts of immunostained cytologic preparations is accomplished by automatic microscopes using sophisticated imaging approaches. Criteria to examine morphology and staining results have also been defined to avoid false-positive and false-negative results [42].

Although there are existing recommendations for standard operation procedures, there are still restrictions to the standardization of immunocytochemical methods with respect to reproducibility of the staining procedure itself as well as microscopic interpretations. Therefore, both intra- and interlaboratory evaluation of the methods is required to ensure reliability of the results [64].

Besides immunocytochemical methods, very sensitive nucleic acid-based techniques now allow the detection of DTCs at the single-cell level. The main advantage of these methods is the nearly unlimited availability of primers for almost every gene of interest. Although numerous genetic alterations have been described in breast cancer cells, heterogeneity is enormous, so that at present no universally applicable DNA marker exists for the primary screening of a wide range of DTCs [9]. Further efforts have been made to detect free circulating DNA or epigenetic alterations of circulating DNA such as methylation in BM and blood plasma, but the results are still preliminary [65]. Therefore, the measurement of epithelium-specific or more organ-specific mRNA species such as cytokeratin 19 or mammaglobin mRNA by RT-PCR has been proven to be a promising approach to detect DTCs in BM samples [66]. Because of the lack of tumor-specific markers, the main disadvantage of using surrogate tissue-specific markers is false-positive results due to illegitimate low-level transcription of epithelial or breast tissue-specific genes in normal cells [48]. Furthermore, heterogeneity in the expression of particular genes is not recognizable and the expression level of a gene of interest per cell cannot be estimated. At present, analyses are mainly performed by quantitative real-time RT-PCR, ensuring the discrimination between different levels of expression. Moreover, multimarker real-time RT-PCRs have the potential to improve the method even in the case of downregulation of the expression of a single gene [45]. However, storage and sample preparation have to be performed under conditions avoiding RNA degradation, one of the major problems of RT-PCR approaches [66].

The application of multimarker assays might also compensate for low mRNA amounts due to the low number of tumor cells. There are numerous excellent reviews listing the marker genes currently used in RT-PCR approaches to detect DTCs in BM or CTCs in blood from breast cancer patients [48]. The methods explained above are not able to discriminate between viable and apoptotic DTCs. A new technique, designated EPISPOT (epithelial immunospot), offers the advantage of detecting viable tumor cells by their ability to secrete individual proteins. In a newly published study, it was demonstrated that BM samples from metastatic and nonmetastatic breast cancer patients contain viable tumor cells which secret Muc-1 and/or cytokeratin 19 in about 90 and 50 % of cases, respectively, whereas in controls from healthy women, cells secreting these proteins could not be detected [9].

KRT20 is found in breast cancer cells [89]. However, its expression is less linked to breast tissue and more related to gastric and intestinal epithelium, urothelium, and Merkel cells [23]. Additionally, KRT20 expression has been established in granulocytes [90]. Due to its lower specificity when compared with KRT19, the use of KRT20 is not suggested in breast cancer patients. KRT8 and KRT18 have been hardly ever used for DTC detection. In fact, the expression patterns of these epithelial cytokines are very similar to that of KRT19 and they are not expected to provide more specificity than KRT19. Of note, KRT8, KRT18, and KRT19 are expressed in the breast epithelium but at higher levels in the luminal than in the basal component. In view of recent observations that breast tumors may be classified into subtypes, or classes, including luminal epithelial-like and basal epithelial-like, one can believe that these cytokeratins will be less easily distinguished in DTCs originating from basal-like tumors.

Commonly known as CEA, it functions in several biological roles, including cell–cell adhesion. It is one of the most commonly expressed markers in breast, as well as in various other, cancer cells [48, 60]. Therefore, it suffers from low specificity, as also seen with KRT19, and can likewise be induced in peripheral blood (PB) by cytokines and growth factors [48].

This epithelial cell–cell adhesion protein is known under a range of names, of which GA733-2 and EpCAM are the most commonly used. Ubiquitously expressed on the surface of epithelial cells, it has been normally used as a target for positive IMS to enrich DTC for RT-PCR analysis [23]. Monoclonal antibodies against this antigen have been widely developed for diagnostic, but also therapeutic, approaches. Although highly sensitive for epithelial malignancies, including breast cancer, its use is, however, hindered by the fact that it is expressed in low amounts in PB cells [91].

Mucin-1 is an extensive, polymorphic, and heavily glycosylated mucin. The role of mucins is mainly one of the hydrating and lubricating epithelial linings, but these proteins have also been concerned in modulating both growth factor signaling and cell adhesion. Further, it has been suggested that MUC1 expression at the surface of tumor cells could decrease cell adhesion and favor dissemination [92]. Conversely, MUC1 could play a role in the initial attachment of breast tumor cells to tissue at remote sites, facilitating establishment of metastatic sites [93]. Extensively expressed in normal epithelial tissues, MUC1 is remarkably present on the apical surfaces of breast, bronchial, pancreatic, uterine, salivary, intestinal, and other glandular tissue cells. Like TACSTD1, MUC1 has been commonly used as a target for positive IMS to enrich DTC for RT-PCR analysis [23]. Many studies have reported the expression of MUC1 in a significant proportion of healthy blood donors. Indeed, MUC1 expression has been considerably found in PB cells [23]. Although it has low specificity, the assessment of MUC1 expression in DTC is supported by the increasing interest for MUC1-based immunotherapy [94]. Although MUC1 is expressed in a majority of breast tumors, its overexpression has been associated with a lower grade and a higher ER-positive phenotype [95].

A series of RT-PCR-based mono- or multimarker studies have assessed the relevance of this growth factor receptor for DTC detection [96, 97]. EGFR emerges as more specific but less sensitive than KRT19. Unluckily, it has also been found infrequently in the PB of healthy donors [23]. Furthermore, Weigelt et al. [97] have shown that the median expression of EGFR was higher in normal ALN than in DTC-positive ALN! Notably, EGFRvIII, a cancer-specific EGFR variant, has been now used to detect DTC in breast cancer patients. The mutant was seen in the peripheral blood in 30 % of 33 low-risk, early-stage patients, 56 % of 18 patients chosen for neoadjuvant chemotherapy, 63.6 % of 11 patients with disseminated disease, and, remarkably, 0 of 40 control women [98].

Involved in signal transduction, ERBB2 participates in breast tumor biology. Yet, it is not breast specific [99], and weak ERBB2 expression has been found in the PB of healthy women in several studies [23]. However, it is overexpressed in 20–35 % of breast cancer patients, mostly as a result of gene amplification, and this forecasts for reduced survival. Furthermore, in patients with breast cancer, ERBB2 overexpression by DTC in the BM predicts poor clinical outcome [82]. This, as well as the increasing use of ERBB2 as target for immunotherapy (trastuzumab) [94], supports its assessment in DTC, at both the mRNA (RT-PCR) and the DNA (FISH) levels.

No breast cancer marker has been shown to be never expressed in healthy volunteers, but some markers are hardly ever found in controls. SCGB2A2 [105], widely known as mammaglobin, is one of these markers. It is a member of the secretoglobin superfamily [106], a group of small, secretory, rarely glycosylated, dimeric proteins generally expressed in mucosal tissues that could be involved in signaling, immune response, chemotaxis [107], and, probably, as a carrier for steroid hormones in humans. SCGB2A2 has become a quasi standard in breast DTC detection by RT-PCR-based methods, being the most extensively studied marker after KRT19. It has been used to identify DTC in LN, PB, BM, and even in malignant effusions. SCGB2A2 expression has been noticed, rarely and in low levels, in various normal tissues. This could restrict its prospective use as an immunotherapeutic target [108], due to concerns about autoimmune toxicity.

Zafrakas et al. have found an abundant SCGB2A2 expression in malignant and normal tissues of the breast and in the female genital tract, namely, the cervix, uterus, and ovary, while lower expression levels were hardly ever found in other tumors and normal tissues [109]. These remarks might extend the diagnostic potential of SCGB2A2 to the detection of DTC from gynecologic malignancies. While SCGB2A2 is significantly more breast cancer specific than KRT19, it is less “universal” among these tumors. Indeed, SCGB2A2 expression level is highly changeable in breast tumors, with some of them showing no expression at all. SCGB2A2 expression, estimated at mRNA or protein level, has been reported in 61–93 % of primary and/or metastatic breast cancer biopsies [110–112]. By examining SCGB2A2 gene expression levels in 11 BCC lines, BT-474, Evsa-T, Hs578T, IBEP-1, IBEP-2, IBEP-3 [113], KPL-1, MCF-7, MDA-MB-231, MDA-MB-453, and T-47D, by microarray and RT-PCR, researchers have shown elevated SCGB2A2 mRNA level only in Evsa-T BCC, while mild expression was seen in BT-474 BCC [114]. Notably, most of these BCC lines are of metastatic origin [113].