Comparative evaluation of cell-free tumor DNA in blood and disseminated tumor cells in bone marrow of patients with primary breast cancer

The present study was conducted at the Department of Obstetrics and Gynecology at the University Hospital in Essen. In total, 81 patients with primary breast cancer were studied from April 1998 until January 2003. Additionally, 10 healthy female controls aged between 30 and 50 years and with no history of cancer were recruited. Overall survival data of these patients were obtained from the local municipal registry; the median follow-up time was 6.2 years (range 0.2 to 9.8 years). Informed written consent was obtained from all patients, and the study was approved by the Local Essen Research Ethics Committee (05/2856). The clinical data of the patients are summarized in Table 1.

The median age of the patients was 56 years (range 33 to 81 years). All four initial tumor stages were included, with a predominance of stage I and II. Most patients had ductal breast cancer and 49 women were node negative. High and moderately differentiated tumors were predominant. Two-thirds of the tumors were estrogen receptor (ER) negative and progesterone receptor (PR) positive, respectively. All patients undergoing breast-conserving therapy received an adjuvant radiation. Patients with hormone receptor-positive tumors received an adjuvant hormonal treatment with tamoxifen or an aromatase inhibitor. The chemotherapeutic adjuvant treatment mostly contained anthracyclines and taxanes.

For each of the 81 patients, the tumor type, TNM-staging and grading were assessed according to the World Health Organization-classification of tumors of the breast [16] and the sixth edition of the TNM Classification System [17]. The ER and PR receptor status were determined by immunohistochemistry.

For the quantitative determination of carcino embryonal antigen (CEA)/CA15-3 in human serum and plasma, 10 ml serum were collected and assessed using the Elecsys CEA/CA15-3 immunoassays (Roche, Mannheim, Germany). The serial measurement of CEA/CA15-3 was intended to aid in the management of cancer patients. These assays were performed on a Cobas^® immunoassay analyzer in the central laboratory of University Hospital in Essen according to the manufacturer's instruction. The central laboratory has a valid certification for the performance of these assays following international guidelines.

BM cells were isolated from heparinized BM (5000 U/ml BM) by Ficoll-Hypaque density gradient centrifugation (density 1.077 g/mol; Pharmacia, Freiburg, Germany) at 400 g for 30 minutes. Interface cells were washed (400 g for 15 minutes) and resuspended in PBS. For the detection of cytokeratin-positive (CK+) cells, 3 × 10⁶ cells (1 × 10⁶ per slide and area of 240 mm²) from each aspiration side were directly spun (400 g for 5 minutes) onto glass slides coated with poly-L-lysine (Sigma, Deisenhofen, Germany) using a Hettich cytocentrifuge (Tuttlingen, Germany).

After overnight air drying, staining of CK+ cells was performed using the Epimet^® kit (Micromet, Munich, Germany). The identification of epithelial cells using this kit is based on the reactivity of the murine monoclonal antibody (Mab) A45-B/B3, directed against a common epitope of CK polypeptides. The kit uses Fab fragments of the pan-Mab conjugated with alkaline phosphatase molecules. The method includes: permeabilization of the cells by a detergent (5 minutes); fixation by a formaldehyde-based solution (10 minutes); binding of the conjugate Mab A45-B/B3-alkaline phosphatase to cytoskeletal CKs (45 minutes); and formation of an insoluble red reaction product at the binding site of the specific conjugate (15 minutes). Subsequently, the cells were counterstained with Mayer's hematoxylin for one minute and finally mounted with Kaiser's glyzerine/gelatine (Merck, Darmstadt, Germany) in Tris-EDTA buffer (Sigma, Deisenhofen, Germany). A conjugate of Fab-fragment served as a negative control. For each test a positive control slide with the breast carcinoma cell line MCF-7 (ATTC, Rockville, MD, USA) was treated under the same conditions. The microscopic evaluation was carried out independently by two investigators. Patients were evaluated as tumor cell-positive if at least one CK-positive cell was detected as analyzed by immunocytochemistry.

From each patient, 10 ml whole blood was collected in routine S-Monovette^® tubes (Sarstedt AG&Co, Nümbrecht, Germany) and immediately stored at 4°C. Serum and leukocyte preparations were performed within four hours. The blood samples were centrifuged at 2500 g for 10 minutes. The upper phase contained the blood serum, from which 3 to 4 ml was removed for the extraction and analysis of the circulating DNA. The remaining 16 to 17 ml blood was supplemented up to 50 ml with lysis buffer containing 0.3 M sucrose, 10 mM Tris-HCl pH 7.5, 5 mM MgCl₂ and 1% Triton X100 (Sigma, Taufkirchen, Germany). Following incubation for 15 minutes on ice, the isolation and purification of the leukocytes were carried out by two centrifugation steps at 2500 g, 4°C for 20 minutes.

Tumor tissue of 22 patients was available. Specimens were retrieved from the Institute of Pathology and Neuropathology of the University Hospital of Essen. Tumor pieces of 3 mm in size were processed from paraffin-embedded tumor blocks and embedded in paraffin again to perform six sections of 10 to 20 μm thickness. The sections were dewaxed in 1 ml xylene on a shaker incubator at 45°C for five minutes. After centrifugation at full speed and room temperature for five minutes, the supernatant was removed. Pellets were washed in 1 ml of ethanol and centrifuged at full speed for five minutes. The supernatant was removed, and the pellet was dried at 45°C for two to five minutes until the ethanol had evaporated.

For the PCR-based fluorescence microsatellite analyses we used our former microsatellite method without extended fractionation step [15], because the fractionation technique which separates blood DNA in short and long DNA fragments was not yet established [18] when blood samples were collected between 1998 and 2003.

Genomic DNA was extracted from tumor tissues, leukocytes and serum of peripheral blood using the QIAamp Blood DNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Quantification and quality of the extracted DNA were determined spectrophotometrically using the BioPhotometer (Eppendorf, Hamburg, Germany) or the NanoDrop Spectrometer ND-1000 (Peqlab Biotechnologie, Erlangen, Germany). To determine the lowest portion of tumor-specific DNA which can be flawlessly detected, dilution experiments were performed. For this study we mixed and amplified known quantities and proportions of normal leukocyte and serum DNA, as described before [18].

Serum, tumor and leukocyte (reference) DNA were amplified with a PCR using primer pairs binding to microsatellite markers as summarized in Table 2. PCR conditions were described before [15]. To confirm the microsatellite alterations, each PCR was repeated at least twice.

The fluorescence-labeled PCR products were separated by capillary gel electrophoresis and detected on an automated Genetic Analyzer 310 (Applied Biosystems, Freiburg, Germany). Fragment length and fluorescence intensity were evaluated by the GeneScan software. The 500-ROX size marker (Applied Biosystems, Freiburg, Germany) served as an internal standard. The LOH incidence was determined by calculating the ratio of the intensities of the two alleles from a serum or tumor sample corrected by the ratio of the intensities of the two alleles from the corresponding leukocyte sample which served as reference DNA. LOH was interpreted if the final quotient was less than 0.6 or more than 1.67. Homozygous and non-analyzable peaks were designated as non-informative cases.

DNA was extracted from blood serum of 81 breast cancer patients. The range of DNA concentrations was between 58 and 5317 ng/ml of serum with a mean value of 886 ng/ml and a median value of 519 ng/ml. As shown in the box plot of Figure 1a, the patients with lobular breast cancer had significantly higher DNA levels in their blood than patients with ductal breast cancer (P < 0.03, Table 3). Patients with DTC-positive BM had higher DNA yields than patients with DTC-negative BM (P < 0.05, Figure 1b).

Furthermore, the comparison of the DNA levels in the blood of the tumor patients with those of 10 healthy controls showed that minor DNA yields circulate in blood of healthy individuals. This background level of normal DNA probably results from apoptotic lymphocytes present in the blood sample [13, 15]. The range of DNA concentrations of the control group was between 36 and 156 ng/ml of serum with a mean value of 60 ng/ml and a median value of 45 ng/ml (data not shown).

A PCR-based fluorescence microsatellite analysis was performed using a panel of six different polymorphic microsatellite markers (Table 2). Tumor tissue specimens of 22 patients were available and served as positive controls. The matched leukocyte DNA (normal DNA) of each patient served as a reference sample. LOH was found in every second tumor tissue (50%). Of the 11 patients, 4 patients showed 1 LOH in their tumor sample, 4 patients showed 2, 1 patient showed 3 and 2 patients showed 4 LOH in their tumor sample (data not shown). The overall LOH incidence in the specimens of all analyzed microsatellite markers was 27.5%.

Following PCR-based microsatellite analyses on blood serum samples from 81 patients with a panel of six different polymorphic microsatellite markers, at least one LOH was found on cell-free DNA in serum of 27 (33.5%) patients (Table 1). Three patients displayed LOH at two markers in their blood (data not shown). The overall LOH incidence of all analyzed microsatellite markers in blood was 9.5%. Figure 2a depicts the LOH distribution at the different microsatellite markers in the serum samples of the 81 patients. Moreover, blood of the healthy controls did not display LOH at any microsatellite marker used (data not shown).

From 22 patients, both serum and tumor tissues were available. Comparative analyses showed a lower LOH frequency in serum than in tumor tissues (Figure 2b). The LOH distribution of the corresponding serum and tumor samples was heterogeneous and only in three cases was LOH simultaneously detected in both clinical materials, which were all derived from patients with lobular breast cancer (data not shown).

DTC were detected in BM of 32 of 81 patients (39.5%). Four of the five patients with metastatic disease at primary diagnosis were postmenopausal, hormone receptor positive and DTC positive (Table 1). Recurrent disease was diagnosed in 15 of 81 patients (18.5%) after a median of 21 months (range 4 to 54 months), and 22 of 81 patients (27.0%) died after a median of 4.3 years (range 2 months to 9 years). The occurrence of DTC in BM significantly correlated with distant metastases (P < 0.05) but with no other clinicopathological factors (Table 3). Both ER-positive and PR-positive patients more often tended to have DTC in their BM than receptor-negative patients (Table 1).

Statistical evaluations of the LOH frequency in blood from breast cancer patients were performed with the following clinical data: age, family history, tumor size, nodal status, histology, grading, ER and PR status, CEA and CA15-3 (Table 1). Table 3 summarizes the significant associations between these parameters.

The occurrence of LOH at the entire marker set correlated positively with the histologic type of carcinoma (P = 0.05) and negatively with the histopathologic grading (P = 0.006, Tables 1 and 3). Most LOH were detected in blood of patients with ductal breast cancer (38%) and a histopathologic grading of I to II (45%, Table 1). Statistical evaluations of the LOH frequency at the single markers showed that two markers (D3S1255 and D9S171) were associated with clinicopathological parameters of the patients. The marker D9S171 was only affected by LOH in patients with higher tumor stages of pT2-4 (P < 0.05) and patients older than 55 years (P = 0.05, Table 3).

In respect to the 76 patients with primary breast cancer without overt metastases, LOH at both markers (D3S1255, P < 0.05 and D9S171, P < 0.02) significantly correlated with the prevalence of recurrence (Table 3). As estimated by Kaplan-Meier analysis, the median recurrence periods were 109 and 110 months (95% confidence interval (CI) 101 to 117 and 103 to 118) in patients who had no LOH at D3S1255 and D9S171, respectively, whereas the periods were 69 and 43 months (95% CI 39 to 100 and 16 to 71) in patients who displayed LOH at D3S1255 (P = 0.009, Figure 3a) and D9S171 (P = 0.001, Figure 3b), respectively.

None of the LOH data correlated with a reduced overall survival rate. Whereas nodal status significantly correlated with recurrence (P < 0.05) and overall survival (P < 0.001), histological grading was associated with a significant shorter overall survival (P < 0.03; data not shown).

Finally, we compared the occurrence of LOH at the particular microsatellite markers in blood with the presence of DTC in BM. However, our statistical assessments between LOH and the BM status did not reach any statistical significance for the marker analyzed (Table 3).