|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to analyze the utility of a mammaglobin multigene RT-PCR assay and a mammaglobin sandwich ELISA to detect peripheral blood samples of breast cancer patients.
Peripheral blood samples of 147 untreated Senegalese women with biopsy confirmed breast cancer were collected. The samples were tested for mammaglobin and 3 breast cancer associated gene transcripts using a multigene real-time RT-PCR assay and for secreted mammaglobin protein using a sandwich ELISA format. Patient information regarding demographic and clinical staging of disease was also collected.
In 77 % of the breast cancer blood samples a positive expression signal was found using the multigene RT-PCR assay detecting mammaglobin and three complementary transcribed genes. 50 samples from healthy female donors tested negative. Circulating mammaglobin protein was found in 68 % of the breast cancer sera, whereas 38 % showed significantly elevated protein levels in comparison to a mixed control population. Statistical correlations were found between the detection of mammaglobin protein in serum, presence of mammaglobin mRNA expressing cells in blood, stage of disease and tumor size.
The multigene mammaglobin RT-PCR assay and mammaglobin sandwich ELISA could be valuable tools to detect metastatic disease and to monitor therapeutic efficiency. Both assays together provided a diagnostic sensitivity of 83 %. Use of the multigene RT-PCR increased detection sensitivity from 61 to 77 % in comparison to mammaglobin expression alone.
The identification of sensitive and specific biomarkers for the detection of circulating breast cancer cells and for staging of breast cancer may be of considerable importance for clinical management of breast cancer and could provide an important tool for researchers. Previous studies have demonstrated the utility of RT-PCR based detection of mammaglobin, a homologue of the rat prostatic binding protein component 3 (1) and a member of the uteroglobin/clara cell protein family (secretoglobins) (2), for identification of disseminated breast cancer cells in blood, lymph nodes and bone marrow (3–5). However, mammaglobin, which is expressed in normal breast epithelial cells, is only expressed in a subset (70–80%) of primary and metastatic breast cancer tissues (6). Therefore, we recently developed a panel of four complementary expressed genes, mammaglobin, B305D, γ-aminobutyrate type A receptor π subunit (GABAπ), and B726P to provide a panel with high sensitivity and specificity for detection of breast cancer cells (7,8). In our previous study we assayed for the transcripts of these four genes (using a multigene RT-PCR assay) in 27 primary breast cancer tissues, 50 lymph nodes containing metastatic breast cancer, and 27 non-breast cancer lymph node specimens (9). All primary breast tumors and metatstatic breast cancer lymph nodes, but none of the control samples had positive expression signals of either mammaglobin, B305D, GABAπ, and/or B726P. In the present study we focus on transcript detection of this gene panel in peripheral blood. Although little is known about detection of circulating mammaglobin protein so far, previous studies have suggested that detection of mammaglobin protein has potential as a biomarker for breast cancer (10,11). Thus, the present study was undertaken to examine the relationship between the presence and stage of breast cancer and peripheral blood based detection of these four genes transcripts and/or increased levels of mammaglobin protein.
Beginning in February 2001, Senegalese women presenting to the Oncology service at the Dantec Hospital of the University of Dakar with masses which were clinically diagnosed for breast cancer, and who had not undergone previous biopsy, surgery or therapy for this pathology, were invited to enroll into a study of breast cancer. Written informed consent was obtained in compliance with the Human Subjects Institutional Review Boards of the University of Washington and the University of Dakar. A ten milliliter sample of blood was collected into EDTA vaccutainers tubes and was immediately sent to the laboratory for processing. After collection of blood, a physical examination was performed and tissue samples were obtained from the main tumor mass by needle core biopsy. Two adjacent needle core biopsies were obtained, one was placed into formalin for routine histologic processing and the other in RNAlater (Ambion) for molecular studies. Chest x-rays and ultrasounds were performed to stage disease. In total, 197 patients were enrolled, of which 147 (75%) where found to have biopsy confirmed breast cancer. Serum samples from 142 of those with confirmed breast cancer study were available for ELISA studies and peripheral blood samples from 84 women were available for RT-PCR testing for circulating breast cancer cells. Control serum samples were collected from 53 Senegalese and 41 U.S. women without breast cancer. Control blood samples for the RT-PCR assay were collected from 50 healthy, female volunteers at Corixa, Seattle.
In the laboratory, RosetteSep™ CD45 depletion cocktail for enrichment of circulating epithelial tumor cells (StemCell Technologies Inc, Vancouver, Canada) was added (at a concentration of 50 μl/mL of whole blood) to the vaccutainer and incubated at room temperature for 20 minutes. The antibody labeled blood was then transferred into Sigma Accuspin System-Histopaque-1077 (Cat. No. A6929) tubes, and centrifuged for 10 minutes at 1000 x g to separate the human epithelial cells from the hematopoietic cells antibody cross-linked to red blood cells. The cell layer was collected, washed once with PBS, and cell pellets were resuspended with 1.5 ml mRNA isolation (Roche) lysis buffer and stored in liquid nitrogen for shipping to Seattle.
Needle core biopsies in RNAlater (Ambion) were shipped to Seattle on liquid nitrogen. Tissue samples (10 – 30 mg each) were transferred into 1 ml lysis buffer (Ambion Poly(A)Pure mRNA Purification Kit), disrupted using 2 grams of 1 mm zirconia beads (BioSpec Products Inc) for 3 minutes on highest setting using a Mini-BeadBeater (MidWest Scientific). RNA was isolated according to the manufacturer’s protocol and eluted with 60 μl elution buffer. Reverse transcription was performed for 1 h at 42 °C using oligo(dT) primers (Gibco) and 10 μl Superscript (Gibco) in a final volume of 150 μl.
For peripheral blood cell samples the tumor cell enriched blood cell lysates were shipped to Seattle on liquid nitrogen and processed according to the manufacturer’s manual (Roche mRNA isolation kit). mRNA was eluted with 25 μl nuclease-free H2O and reverse transcribed into cDNA using oligo(dT) primers (Gibco) and 8 μl Superscript Reverse Transcripase (Gibco) in a final volume of 120 μl.
The following specific primers and 6-carboxy-fluorescein (FAM)-labeled Taqman® probes were used to detect mRNA expression of mammaglobin, GABAπ, B305D and B726P simultaneously.
|Gene||Primers and probe||Concentration||Amplicon size|
|mammaglobin||F: tgccatagatgaattgaaggaatg||100 nM||89 bp|
|R: tgtcatatattaattgcataaacacctca||100 nM|
|P: tcttaaccaaacggatgaaactctgagcaatg||4 pmol|
|GABAπ||F: caattttggtggagaacccg||300 nM||137 bp|
|R: gctgtcggaggtatatggtg||50 nM|
|P: catttcagagagtaacatggactacaca||4 pmol|
|B305D||F: tctgataaaggccgtacaatg||300 nM||239 bp|
|R: tcacgacttgctgtttttgctc||50 nM|
|P: atcaaaaaacaagcatggcctcacaccact||4 pmol|
|B726P||F: gcaagtgccaatgatcagagg||100 nM||110 bp|
|R: atatagactcaggtatacacact||100 nM|
|P: tcccatcagaatccaaacaagaggaagatg||4 pmol|
|β-actin||F: actggaacggtgaaggtgaca||300 nM|
|R: cggccacattgtgaactttg||300 nM|
|P: cagtcggttggagcgagcatccc||4 pmol|
Primers were designed to cross intron-exon junctions in order to exclude genomic DNA from amplification. Expression levels were measured by quantitative real-time PCR using the ABI 7700 Prism™ sequence detection system (Applied Biosystems, Foster City, CA). Actin expression was measured in separate reactions as a quality control for blood cDNA samples. Specimens with Actin expression < 50 copies were excluded from analysis. 50 PCR cycles were performed with TaqMan® 1000 Rxn PCR Core Reagents (Part. No. 430 4439, Applied Biosystems, Foster City, CA) using 0.0375 U/μl TaqGold, 1x Buffer A, 5 mM MgCl, 0.2 mM dCTP, 0.2 mM dATP, 0.4 mM dUTP, 0.2 mM dGTP, 0.01 U/μl AmpErase UNG, 8 % (v/v) Glycerol, 0.05 % (v/v), Gelatin, 0.01 % (v/v) Tween20. PCR conditions were one cycle at 50° for 2 min, one cycle at 95° for 10 min, 95° for 15″ and 60° for 1′ and 68° for 1′ for 50 cycles. Multigene copy numbers were calculated by determining a standard curve using TaqMan® SDS analysis software from serial dilutions of four plasmids containing target gene cDNA sequences. Final copy numbers were determined as medians of triplicate reactions. Triplicate reactions were pooled and a 20 μl aliquot was separated by agarose electrophoresis using a 4% E-Gel (Invitrogen). Gene identities were determined according to amplicon size.
Antibodies were generated against purified native Mammaglobin protein complex (12). 96-well microtiter plates were coated overnight at 4°C with the monoclonal antibody RO48 at 200 ng/well. 5% non-fat milk/PBS was used to block for 2 h at room temperature. For standard curve analysis, different dilutions of purified native Mammaglobin protein complex were made in normal human serum. A 1 μg/ml concentration of biotinylated RO28 was added and incubated at room temperature for 1 h. Plates were washed and streptavidin-horseradish peroxidase (HRP) was added and incubated for 30 minutes. Tetramethylbenzidine substrate was added for 15 min before the reaction was stopped with 1 N sulfuric acid and read at 450 nm.
Two-sided Mantel-Haenszel chi-square or Fisher’s exact tests were performed to assess univariate associations between gene detection and the demographic and clinical characteristics of the breast cancer subjects or their disease. Subject demographic characteristics evaluated included age, gravidity, birth control use, menopausal status, history of hormonal therapy, alcohol and tobacco use, and history of obesity. Each breast cancer was characterized by tumor size, lymph node involvement, metastasis, and was subsequently staged using standard breast cancer staging methods. Associations with ordered categorical factors were tested using Mantel-Haenszel tests for trend, and Student’s t-test or analysis of variance was used to compare groups with respect to continuous factors. Data analyses were conducted using SAS 8.2 for Windows (SAS Institute, Cary, NC). A receiver-operator curve (ROC) was utilized to determine the optimal (most sensitive and specific) mammaglobin protein level predictive of breast cancer.
Histologic and cytologic review of biopsies confirmed the presence of breast cancer in 147 Senegalese patients. The mean age of subjects was 47.7 years, with a range of 13 to 77. Approximately half (47%) were menopausal, and only ten currently used any form of hormonal contraception. Mean gravidity was 5.8, with a range of zero to 19, and mean age of first pregnancy was 19.4 years. Few (3%) had any history of hormone replacement therapy, and less than 1% reported tobacco or alcohol use. Only eleven women (7%) had a history of obesity. Most of these women had advanced disease and large tumors, with only 33% of tumors being 5 cm or less in size, while 43% were 10 cm or greater (Table 1). Evidence of nodal involvement was present in 92% of cases, and a vast majority (80%) of the cancers were stage III and higher.
We first examined the expression frequency of the four genes in breast cancer biopsies. Using single gene real-time RT-PCR, mammaglobin expression was detected in 85% (39 of 46) cases (Table 2). GABAπ was detected in 52% of the tumors, but only when mammaglobin was also present. In two other cases, biopsy samples were positive for B726P, but not mammaglobin, resulting in 89% of tissues examined expressing at least one gene. Gene expression analysis of cells collected from blood of 84 women with breast cancer and 50 healthy female volunteers was then performed using the multigene RT-PCR assay. Mammaglobin expression was detected in 51 (61%), GABAπ was found in 18%, B726P in 5% and B305D in 8% of the blood samples. Overall, expression of one or more of these genes was present in 77% of samples from women with proven breast cancer, but none of the 50 samples from women without cancer.
We next examined the associations between detection of gene transcripts in peripheral blood and tumor characteristics. Overall, among women with biopsy confirmed breast cancer, detection in peripheral blood of any of the four genes included in the multigene assay was not significantly associated with increasing tumor size (p=0.4) Likewise, detection of mammaglobin transcripts, alone, were not associated with tumor size (p=0.4). However, peripheral blood based detection of GABAπ was found to be associated with tumor size (p<0.001, test for trend) with no expression in the 30 small (2–5 cm) tumors, expression in 14% of medium (6–9 cm) tumors and in 40% of largest tumors 10 cm or greater. Mammaglobin (p=0.07) and GABAπ (p=0.05) were each marginally associated with increased nodal involvement; only 20% of those with N0 disease were positive for mammaglobin, compared to 60–70% of those with N1 or N2 disease. Similarly, 10% of women with N0 or N1, as compared to 29% of N2 had GABAπ detected in blood samples. In contrast, B726P detection was inversely associated with nodal involvement (p=0.01) as 20% of those who were N0 were B726P positive, compared to only 2% of those who were N1 and none of those who were N2 or N3. Interestingly, in the ten women with N0 cancer, six were positive in the multigene assay, including two positive for mammaglobin, two for B726P, one for B305D and one for GABAπ.
Lastly, detection of mammaglobin transcripts in the blood did not vary consistently with increasing tumor stage (p=0.4). In contrast, GABAπ was not detected in stage I, II, or IIIA cancers, but was expressed in blood samples from 29% of those with stage IIIB or IV disease (p=0.004, test for trend). Overall, the detection of the four gene transcripts was associated with stage of disease (p=0.03), being present in 53% of those with stage II, 82% of those with stage III, and 87% of those with stage IV breast cancer.
Demographic and behavioral characteristics were also examined in association with detection of gene transcripts in peripheral blood. Detection of mammaglobin or transcripts did not differ significantly by the age or gravidity of the patient, however, detection of transcripts of GABAπ was inversely associated with menopausal status, as 28% of premenopausal compared to 9% of postmenopausal women with breast cancer had transcripts of GABAπ detected in peripheral blood (p=0.02).
Blood samples from 79 women with breast cancer could be analyzed for both the mammaglobin protein as well as detection of the transcripts from the four genes of interest. In these women, 66 (83.5%) had either elevated mammaglobin protein, or transcripts of one of the four genes detected in a single peripheral blood sample. Over half (59.5%) of the blood samples had mammaglobin transcripts detected, fourteen more had gene transcripts from one or more of B305D, GABAπ and B726P, and an additional five had elevated mammaglobin protein noted.
Among the 142 women with confirmed breast cancer mammaglobin serum protein >1.706 ng/ml was marginally associated with increasing tumor size (p=0.09, test for trend), but not lymph node involvement (p=0.5). In addition, mammaglobin protein >1.706 ng/ml was marginally associated with increasing clinical stage of disease (p=0.10, test for trend), with only 29% of samples from women with stage I or II cancer being positive for mammaglobin protein, compared to 31% of those with stage IIIA, 34% of those with stage IIIB, and 48% of those with stage IV disease having elevated serum levels. However, among those with mammaglobin serum protein above the level selected as the upper limit of normal (>1.706 ng/ml, n=54), mammaglobin ELISA dilution values were predictive of disease severity. Mean natural log transformed ELISA dilution values in clinical stages I to IIIB were all between 0.9 – 1.4, but increased to 2.3 ng/ml in stage IV disease (p=0.02 in ANOVA, Figure 1). Similarly, mammaglobin ELISA dilution values were strongly association with increasing tumor size (p=0.001 in ANOVA), as the mean natural log transformed dilution values in tumors <10 cm was 1.2 compared to 2.3 in tumors 10 cm or larger.
Mammaglobin serum protein concentration was also associated with the detection of mammaglobin transcript in the blood by RT-PCR. In mammaglobin PCR-negative blood samples the mean ELISA value for circulating protein was 1.8 ng/ml compared to the mean ELISA value of 18.5 ng/ml in mammaglobin PCR-positive samples (p=0.01 in t-test). Mammaglobin ELISA values were also correlated to the log(copy numbers) of the multigene RT-PCR assay (p=0.02). In the patient population (n=40) with mammaglobin expressing tumors, mammaglobin ELISA values were associated with mammaglobin transcript detection in blood (p=0.008 in t-test). Mean ELISA values are 2.2 ng/ml in those without mammaglobin expression in blood, compared to 26.6 ng/ml in those MG RT-PCR positive.
Development of biomarkers for detection and staging of breast cancer is of importance for clinical management of the disease. A number of molecules, including carcinoembryonic antigen (CEA) and mucin-type markers (e.g. the MUC-1 gene and its glycoprotein antigens CA15-3 and CA27.29) have been used as biomarkers for metastatic breast cancer. Before-treatment sensitivities of the commonly used circulating tumor markers for breast cancer CEA and CA15-3 have been reported at 12 % (13), with an increase to 40 % in breast cancer patients with recurrence. Several studies report the detection of circulating tumor cells using cytokeratins, CEA and MUC-1 RT-PCR, (3,14) however, application of these assays has been hampered by lack of specificity (15–17). In this study, we examined the expression of mammaglobin, B305D, Gabaπ and B726P and the elevation of mammaglobin protein in peripheral blood samples of breast cancer patients. The study population consisted of a large number of women with untreated breast cancer, almost all of whom had breast cancer which had already metastasized to regional lymph nodes. Given this, it was anticipated that these women had a high likelihood of having circulating tumor cells present. Using a single 10 ml sample of blood obtained prior to physical examination and biopsy, we detected at least one of the four breast cancer-associated transcripts and/or elevated mammaglobin protein in 84% of women with breast cancer, but in only 3.3% of women without such pathology.
Although mammaglobin tissue expression has been shown in approximately 80% of breast cancers, previous studies in patient blood reported detection of mammaglobin transcript in 25 – 54 % of those with, and in 10 – 25 % of patients without metastatic breast cancer (3,14,18–21). In this study we found mammaglobin transcripts in 62 % of single blood samples from breast cancer patients. This increased rate of detection may be related to the fact that the patients we examined were untreated while many of the women examined in previous studies had undergone chemotherapy. Treatment may lower the number of circulating tumor cells. As mentioned above, since mammaglobin is not expressed in all breast cancers, we developed and recently reported on three complementary expressed genes which when used in combination with mammaglobin provided increased sensitivity for identification of disseminated breast tumor cells in lymph node specimens (9). In the present study, the addition of these 3 transcripts increased blood based detection of circulating cells from 62 % to 77 % of women with breast cancer. It is possible that obtaining additional repeated samples will further increase the number of breast cancer patients who are found to have such transcripts in their peripheral blood.
The sensitivity of the multigene assay was increased with increasing cancer stage, but the detection of mammaglobin transcript alone was only marginally associated with increased nodal involvement and not to other tumor or patient characteristics. Our findings correspond with a recent study by Lin et al (22), which evaluated the correlation between mammaglobin expression in peripheral blood and known prognostic factors for breast cancer patients. Whereas mammaglobin mRNA expression frequency was shown to be increased in patients with unfavorable prognostic factors (tumor size, stage), no significant differences could be confirmed. The same group also reported that the combination of mammaglobin mRNA detection with CEA or CA15.3 increased the sensitivity from 54 % of 33 metastatic breast cancer patients to 81 % and 90 % respectively, suggesting mammaglobin mRNA as a potential adjunct to routinely used serum markers.
We found that the detection of GABAπ was more often present in breast cancer patients with larger tumors, increased node stage, and advanced overall tumor stage. Interestingly, GABAπ expression was higher in women who were premenopausal. These findings demonstrate a possible application of this marker to monitor disease progression and treatment efficacy in particular in pre-menopausal patients. Surprisingly, B726P and B305D expression was detected only in a small subset of breast cancer blood specimens (5% and 8% respectively). To confirm the expression of these target genes in the primary breast tumors, we analyzed biopsy tissue samples in a subset (n=47) of patients. The percentage of the tumors expressing mammaglobin (85%) and GABAπ (53%) were consistent with earlier findings (7,8). However, B726P and B305D were only expressed in 6% of the tumor biopsies tested. Previously, we reported B726P and B305D over-expression in 40–50% and 60–70% of primary and metastatic breast cancer specimens respectively (9), which has been confirmed by others for B726P (10,23). Specific characteristics of the breast tumors or patients (e.g. median age, later stage, ethnicity) could be the reason for a different gene expression profile. This would argue that B726P and B305D might exhibit different detection sensitivities in other patient populations. We are currently analyzing US breast cancer patient samples with different ethnicities, age groups and early stages of disease to confirm the application of this assay and its components.
In the sera, increasing elevations of mammaglobin protein were associated with increasing clinical stage of disease and tumor size. In addition, a significant correlation between mammaglobin serum protein and mammaglobin expression found in blood cells was detected. Whereas circulating mammaglobin protein was detected in 70% of the samples from breast cancer subjects we tested, only 38% were considered elevated in comparison to the control group. The cut-off value of 1.706 ng/ml was established using a Senegal and a US cohort of normal sera samples. The average of mammaglobin concentration in the Senegal normal sera (0.37 ng/ml) was significantly higher than the average concentration in US samples (0.10 ng/ml). This finding might reflect a population difference regarding pregnancies, nursing and undiagnosed benign or malignant breast diseases. All these conditions could affect the amount of mammaglobin protein released into the blood stream, since mammaglobin expression is associated with mammary gland proliferation (24). Detailed studies in different normal populations have to be performed in order to identify conditions which could influence the protein concentration of mammaglobin in sera. Circulating mammaglobin protein could provide a good marker to monitor treatment and detect disease relapse. Moreover, the detection of serum mammaglobin protein could be combined with other markers, e.g. CEA, CA15-3 and circulating antibodies against tumor specific epitopes, to increase detection sensitivity.
This study was supported in part by NIH grants CA-75794 and CA-86673.