Detecting Tumor-related Alterations in Plasma or Serum DNA of Patients Diagnosed with Breast Cancer¹ Free

Adenocarcinoma of the breast is the most common malignancy in the female Western population, affecting up to one in eight women in the United States. Breast cancer is responsible for up to one in five cancer-related deaths among women, and long-term survival following breast cancer has improved little over the past 20 years (1).

An important strategy to reduce mortality from breast cancer is the introduction of mammography screening in an attempt to detect cancers at an asymptomatic and pathologically early stage. Although several studies indicate that mass screening might be a useful strategy for reducing breast cancer mortality, there are a number of disadvantages associated with this form of cancer screening (2, 3, 4, 5, 6). These include a high rate of false positive tests (7), frequent false negative tests, and the enormous public health costs involved (8, 9). Thus, when the benefits of mammography screening are weighed against its costs and other disadvantages, it is perhaps not surprising that this form of screening has engendered an enthusiastic and contentious debate over the past 20 years (10, 11).

Our knowledge of the genetic changes associated with cancer has grown rapidly since the introduction of the PCR in the late 1980s. Chromosomal abnormalities, including mutations, insertions, deletions, allelic losses of oncogenes and tumor suppressor genes, and microsatellite alterations have been discovered in cancer cell DNA, and the application of molecular biological techniques now allows us to identify these alterations in tumors. In relation to breast cancer, several chromosomes appear to be frequently affected by somatic genetic abnormalities. Allelic losses or imbalances have also been reported in numerous chromosomal subregions, and knowledge of microsatellite alterations in breast cancer is constantly expanding (12).

Increased quantities of DNA have been found in the plasma of patients suffering from different malignancies (13, 14, 15, 16, 17). This circulating extracellular DNA exhibits tumor-related alterations such as decreased strand stability (17), ras or p 53 mutations (18, 19, 20, 21, 22, 23, 24, 25), microsatellite alterations (26, 27, 28, 29), or aberrant promoter hypermethylation of tumor suppressor genes (30, 31). In relation to breast cancer, p53 mutations have been found in the plasma DNA of 2 patients in a series of 15 patients (25).

Thus, the aim of our study was to isolate DNA from the plasma or serum of patients with a suspected diagnosis of breast cancer to analyze this DNA for the presence of microsatellite instability and LOH³ and to compare it with corresponding tumor DNA.

Tumor samples and, in some cases, corresponding adjacent tissue samples were collected during surgery from consenting and informed patients. Part of each tumor and adjacent tissue specimen was histopathologically characterized to confirm its diagnosis, and the other part was used for molecular analysis. Blood was collected before surgery. For the first 38 patients, 10 ml of blood with EDTA for plasma collection were taken. For the last 23 patients, a sample of 10 ml of blood was collected without additive for serum. Altogether, we analyzed 61 patients. The following clinical and biological parameters were studied: tumor size, lymph node status (pTNM), histological type, histoprognostic grade, and hormonal receptors. All molecular analyses were performed by personnel who were blinded to clinical and pathological details.

Blood samples were also taken from healthy donors as controls. The EDTA fraction yielded the lymphocytes and the plasma that were separated by Ficoll gradient. Serum was collected by centrifugation (1000 × g; 10 min) after clotting.

Breast tissue, lymphocyte, plasma, and serum specimens were stored at −20°C until further use.

Fresh frozen tissue and lymphocytes were treated with SDS and proteinase K, followed by phenol and chloroform extraction. Paraffin-embedded tissue scraped from the slides was washed in xylol to remove paraffin. After the addition of l volume of ethanol, the mixture was centrifuged, and the pellet was digested with proteinase K and SDS, followed by phenol and chloroform extraction.

Plasma or serum (200 μl) was purified on Boehringer columns (High Pure Viral Nucleic Acid Kit; Boehringer Mannheim, Mannheim, Germany) according to the manufacturer’s protocol, followed by treatment with phenol/chloroform (1:1 v/v) and chloroform/isoamylic alcohol (24:1 v/v) and ethanol precipitation, and finally dissolved in 10 mm Tris (pH 8)-1 mm EDTA diluted five times.

The amount of plasma/serum DNA was estimated by minigel colored with either SYBR green (Molecular Probes, Eugene, OR) or Gelstar nucleic acid gel stain (FMC Bioproducts, Rockland ME).

After PCR amplification, plasma or serum DNA was compared to tumor DNA. Lymphocytic DNA of the peripheral blood served as a control of normality.

The following oligonucleotide primers (Amersham-Pharmacia, Freiburg, Germany) were selected according to the Genome Database: D6S311; D7S522; D8S137; D8S321; D9S169; D10S197; D11S488; D13S260; D16S398; D16S402; D16S421; D17S579; D17S1325; Tp 53; Tp53.5; Tp53.6; Tp53.8; ACTBP2; DM1; EABMD; ER; THRA; and UT5320.

For the first 38 patients, we used radioactive autoradiography. Ten and 20 ng of plasma DNA and 20 ng of lymphocyte, tumor, and adjacent normal tissue DNA were used as a template in a hot-start PCR in a 25-μl reaction mixture [1.5 mm MgCl₂, 20 μm deoxynucleotide triphosphate, 0.6 unit of Taq polymerase, and 10–20 pmol of each primer (depending on the primer)]. One primer was labeled with T4 polynucleotide kinase and [γ-³²P]ATP. Lymphocyte, tumor, and plasma DNA was subjected to 30–35 cycles of PCR at 94°C for 1 min, 56°C to 60°C for l min (depending on the primer), and 1 min at 72°C, with a final extension step of 72°C for 10 min, and products were separated by denaturing gel electrophoresis (5.6 m urea, 7% polyacrylamide, and 32% formamide gel) followed by autoradiography (26).

After having checked that the results were comparable, an additional series of 23 patients was studied by laser fluorescence. One primer of each set was labeled with a fluorescent dye (5′Cy 5) at the 5′ end, and all primers were purified by fast protein liquid chromatography. Ten and 20 ng of serum DNA and 20 ng of lymphocyte and tumor DNA were used as a template in a hot-start PCR in a 25-μl reaction mixture [1.5 mm MgCl₂, 20 μm deoxynucleotide triphosphate, 0.6 unit of Taq polymerase, and 2–10 pmol of each primer (depending on the primer)]. Each PCR reaction was carried out at 94°C for 5 min, 94°C for 1 min, 55°C to 64°C (depending on the primer) for 1 min, and 72°C for 1 min for 33 cycles, with a final extension of 10 min at 72°C. PCR products were separated electrophoretically on 8% polyacrylamide gels (Reprogel high resolution; Amersham-Pharmacia) and detected by laser fluorescence using an automated gene sequencer (Alfexpress; Amersham-Pharmacia). Fluorescent gel data were analyzed with the Allele-link 2 program. PCR products from lymphocytes and corresponding serum and tumor tissue were analyzed on the same gel. The size (in bp) of the microsatellite alleles was calculated automatically by using internal size markers. Automatic analysis of peak areas allowed relative quantification of PCR products and determination of allelic ratios. LOH was scored if the allelic ratios of tumor or serum PCR products and corresponding lymphocyte PCR products were below a cutoff value of 70% (29). Results indicating LOH were repeated at least twice.

A total of 27 initial cases were studied with two microsatellite markers (DM-1 and D17S1325). Of these 27 patients, 7 had benign tumors, and 20 had malignant tumors (2 had in situ tumors and 18 had invasive tumors). Neither tumor nor plasma DNA extracted from patients with benign disease displayed microsatellite alterations (Table 1). In contrast, DNA extracted from 7 of 20 tumors (35%) displayed microsatellite alterations (either LOH or a bandshift; Table 1). Three of these cases also displayed an identical alteration in corresponding plasma DNA. In addition, two cases exhibited identical changes in normal tissue adjacent to the tumor.

Our second series of 11 patients (all of whom had invasive cancer) was studied with 15 microsatellite markers (D6S311, D7S522, D8S137, D9S169, D10S197, D13S260, D16S402, D16S421, D17S1325, D17S579, THRA, ER, EABMD, DM1, and TP53). No LOH was observed in any case with markers D6S311, D8S137, D13S260, D16S402, D16S421, D17S579, THRA, and EABMD. Tumor tissue taken from 5 of the 11 cases (45%) demonstrated an abnormality of at least one of the remaining seven markers (Fig. 1; Table 2). Three of 11 cases (27%) had identical alterations within plasma DNA, and tissue adjacent to one tumor displayed LOH with markers D9S169 and DM1.

The serum of the third group of 23 patients was studied using laser fluorescence on an automatic sequencer. Samples from these cases (19 cases of invasive carcinoma, 2 cases of in situ carcinoma, 2 cases of benign disease) were studied with 11 markers (D8S321, D11S488, D16S398, D17S1325, D17S579, ACTBP2, DM1, Tp53.5, Tp53.6, Tp53.8, and UT5320). No alterations were observed in any samples with markers D16S398, Tp53.5, Tp53.6, and Tp53.8. However, using the seven other markers, LOH was found in the tumor DNA of 17 of 21 patients with cancer (81%) in at least one locus, and 13 cases were positive for multiple loci (Fig. 2; Table 3). Ten of 21 cases (48%) had LOH in the corresponding serum samples, and 1 case was positive for multiple loci. DNA extracted from the tissue of one patient with benign fibroadenoma displayed LOH at two loci, but these alterations were absent in the corresponding serum sample. The plasma and serum of all healthy controls (at least 10 controls/microsatellite marker) did not demonstrate microsatellite alterations.

No significant associations were found between plasma/serum DNA alterations and the clinical and pathological features of cancer patients.

Breast cancer is a common malignancy among the female population, and current screening methods fail to detect many cancers that present at a later date as a result of symptoms. The present study was performed to determine whether the plasma or serum of patients with breast cancer displays microsatellite alterations that are similar to those found within the primary tumor.

This study, which commenced in 1997, initially used two microsatellite markers that we found to be altered in only 35% of tumors and 15% of plasma samples. Later, 15 markers with reportedly higher alteration rates (50–80%) in breast cancer (12, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44) were used in an attempt to increase the sensitivity of the assay for malignant disease. We found a LOH rate of between 0 and 18% for these markers, which is lower than the rate given in previous reports. Let us stress that similar examples of discrepancies between laboratories have been found (32, 33). For instance, LOH at 7q31 has been claimed to occur in over 80% of all breast cancers and to be of prognostic significance (34), whereas a joint European study made by 17 institutes found a LOH rate ranging from 0–40%, with a mean of 19% (35). The reason for this is unclear but may be related to a number of methodological differences between laboratories. In preliminary experiments, we found that the microsatellite pattern of lymphocyte DNA and plasma DNA of healthy controls did not correspond in some cases. This led us to conduct an additional series of preliminary experiments assessing PCR quality as a function of the amount of DNA template contained in the primary reaction. Using between 5 and 80 ng of tumor, lymphocyte, or plasma DNA as a PCR template, it became clear that small quantities of template resulted in a poor-quality PCR product (see Fig. 3), probably as a result of incorrect reading of bases by Taq polymerase. Thus, it is important that all studies reporting results of microsatellite alterations in cancer should contain a large enough number of repeated experiments on healthy controls to identify potential problems with individual microsatellite markers. The quality of the DNA is also a determinant. When the DNA is not pure enough, Taq polymerase might also make mistakes. Neglecting these facts may result in an overestimation of LOH.

In our third series of 23 patients, we used serum in preference to plasma, because greater quantities of higher-quality DNA were extractable from serum. In this series, a higher proportion of patients demonstrated alterations in both tumor and serum DNA. The primers D16S398, D17S579, D17S1325, Tp53.5, Tp53.6, and Tp53.8 were chosen because of their high incidence of LOH in breast cancer, as reported previously (12, 32, 33, 36, 38, 40, 41). In contrast, we also chose ACTBP2, D8S321, D11S488, and UT5320 because they are tetranucleotide repeats that have demonstrated a high rate of abnormality in other cancers (44). Although rarely studied in breast cancer (43), we also continued to use DM1 because this marker showed a relatively high and constant rate of abnormality in our series. As in our previous two groups, the markers that we expected to result in high rates of LOH in breast cancer tended to perform poorly, whereas other markers yielded a relatively high rate of LOH in tumor DNA. It might be worthwhile to extend this work to look at other tetranucleotide repeats that have previously resulted a high incidence of alteration in lung cancer (44), bladder cancer, or head and neck cancer (45).

The three patients with distant metastases at the time of operation (one of whom later died of disease) displayed LOH in plasma or serum DNA at more than one locus. In addition two large (T₄) tumors displayed LOH in plasma DNA. Otherwise, no clear relationship between LOH and clinical or pathological characteristics was evident. Ductal and lobular cancers of all grades (grades 1–3) and all sizes demonstrated LOH in plasma DNA, indicating that other factors may be important in determining the presence of mutant DNA in plasma.

The alterations found in the samples of tissue adjacent to the tumor were detected in conjunction with those found in the plasma. One of the patients had a grade 1 tumor with size T₂, and another patient had a grade 2 tumor with size T₁ and no metastases (Table 1), which confirms that the molecular heterogeneity that characterizes invasive cancers may occur at an early detectable stage (46).

One of nine patients with a benign tumor (a patient with a fibroadenoma accompanied by a fibrocystic mastopathy) showed LOH with two markers in the tumor DNA only. An increased risk of breast cancer among women previously diagnosed with fibroadenoma as well as the presence of LOH and microsatellite instability has been reported previously (47).

We found that two small (T₁) tumors of histoprognostic grade 1 or in situ carcinomas also displayed DNA alterations in serum/plasma DNA. This result is encouraging because it indicates that the presence of mutant DNA in plasma may occur at an early pathological stage. It is possible that further advances in detection techniques and the addition of reliable markers with a high incidence of mutation in breast cancer, such as p53 mutations (25) or hypermethylation of tumor suppressor genes (30, 31), might eventually lead to a noninvasive test for breast cancer. Such a test could also be valuable in the follow-up of patients after surgical or medical therapies. It will be interesting to compare it with the breast cancer screening technique using X-ray diffraction of hair that has just appeared (48).

We thank Prof. A. P. Sappino (Department of Oncology, University of Geneva, Geneva, Switzerland) for helpful suggestions and Dr. H. Mulcahy (Digestive Disease Research Center, St. Bartholomew’s, London, United Kingdom) for careful reading of the manuscript.