Tumor Suppressor Gene Promoter Hypermethylation in Serum of Breast Cancer Patients

Detection of breast cancer at an early stage is the key to successful treatment and outcome. In tumors confined to the breast at diagnosis, stage is the most important prognostic factor. Precursor lesions include at least two histologically defined types. Ductal carcinoma in situ (CIS) is a preinvasive lesion detected more often because the availability of mammography. Patients with localized ductal CIS are candidates for breast conserving surgery. A second entity, lobular CIS, is believed to represent a marker of increased risk for the development of invasive cancer. Within the category of invasive breast cancer, stage I (<2 cm) tumors have a better disease-free survival rate than higher stage tumors. However, ∼30% of women diagnosed with localized breast cancer develop metastatic disease. In the roughly 15% of invasive breast tumors not detected by mammography, lobular carcinomas are overrepresented. Despite the availability of mammography and prevalence of self-examination, there is still additional benefit to be gained from additional screening methodologies (1, 2). Novel approaches to early detection of breast neoplasia therefore merit examination.

The genetic and epigenetic alterations that initiate and drive tumorigenesis are promising targets for early detection because they may precede clinically obvious cancer, can be detected at sensitive levels, may be specific to tumor cells, and can potentially provide information about the prognosis and treatment of the disease (3, 4). CpG islands located in promoter regions of genes are normally unmethylated. In cancer cells, aberrant hypermethylation of these promoter regions is associated with transcriptional silencing. Hypermethylation is therefore an alternative mechanism for inactivation of tumor suppressor genes (5, 6). Because gene hypermethylation has been found to be a common and early alteration in many tumor types (7, 8, 9), including breast (10, 11), it has emerged as a promising target for detection strategies in clinical specimens (3, 4). Several tumor suppressor and other cancer genes have been found to be hypermethylated in breast cancer but to be unmethylated in normal cells, e.g., RAS association domain family protein 1A (RASSF1A), adenomatous polyposis coli (APC), and death-associated protein kinase (DAP-kinase) genes (9, 10, 12, 13).

Detection of tumor suppressor gene hypermethylation in nipple aspirate fluid has been reported by ourselves (14) and in ductal lavage by others (15). Serum or plasma is a more readily accessible bodily fluid, and provision of a specimen does not require the presence of a specialist. DNA is known to be released into serum/plasma, which is enriched for tumor DNA in cancer patients (16). Several recent studies have shown that it is possible to detect tumor-specific genetic or epigenetic alterations in serum DNA from head and neck, lung, and colon cancer patients (17, 18, 19). Importantly, tumor cell-specific DNA alterations in serum were not limited to patients with metastatic cancer but were also present in serum from patients with early or organ-confined tumors (17, 18, 19). Neoplastic DNA in the serum most likely arises from cells that have left the site of the primary lesion and have invaded the circulatory system but lack the capacity of metastasis to new organs or may be released from the primary tumor as free DNA from nonviable (apoptotic) neoplastic cells (3, 4).

To determine the feasibility of detection of hypermethylated tumor DNA, we examined the methylation status of the APC, RASSF1A, and DAP-kinase genes in paired tumor and serum DNA from 34 patients with breast neoplasia, including ductal CIS, lobular CIS, stage I ductal and invasive lobular carcinoma, as well as in normal and benign disease tissue and serum control DNA.

After approval from the Institutional Review Board, we obtained breast tumor specimens via the Fox Chase Cancer Center Tumor Bank Facility and matched preoperative serum or plasma via the Fox Chase Cancer Center Biospecimen Repository from 34 patients, ages 34 to 90 years, with histologically verified breast neoplasia. There were 7 ductal CIS, 3 lobular CIS, 20 invasive ductal, and 4 cases of invasive lobular cancer. Tumors were graded and staged according to American Joint Committee on Cancer guidelines (20). Serum specimens from 20 normal, healthy age-matched women and 8 women with benign breast disease were also obtained via the Fox Chase Cancer Center Biospecimen Repository. DNA previously extracted from five specimens of fibrocystic breast disease and five specimens of histologically normal (nonneoplastic) breast tissue were available for analysis (14). Tumor tissue was obtained immediately after surgical resection and subsequently microdissected with the assistance of a pathologist. DNA was extracted from tissue or from ∼1.5 mL of serum or plasma using a standard technique of digestion with proteinase K in the presence of SDS at 37°C overnight followed by phenol/chloroform extraction (21). Tumor specimen DNA was spooled out after precipitation with 100% ethanol. Serum DNA was precipitated with one tenth volume of 10 mol/L ammonium acetate, 2 μL of glycogen (Roche Diagnostics Corporation, Indianapolis, IN), and 2.5 volumes of 100% ethanol, followed by incubation at −20°C and centrifugation at top speed (16,000 relative centrifugal force). Approximately 50 ng of DNA were obtained from 1 mL of serum. For paraffin-embedded tissue, 10 7-μm sections were cut with a microtome and put on glass slides. The area of lobular CIS or ductal CIS, indicated by the pathologist, was removed with a razor blade, or needle depending on size, using an inverted microscope. The dissected tissue was placed directly into a microcentrifuge tube, washed with xylene, and DNA isolated as above.

Specimen DNA (0.05 to 1 μg) was modified with sodium bisulfite, converting all unmethylated, but not methylated, cytosine to uracil followed by amplification with primers specific for methylated versus unmethylated DNA. The genes used in the breast cancer detection panel were APC (8), RASSF1A (22), and DAP-kinase (23). The primer sequences used have all been previously reported and can be found in the report referenced after each gene. The primers for RASSF1A include CpG site positions 7 to 9 on the forward primer and 13 to 15 on the reverse primer as described previously (22). PCR amplification of template DNA was performed for 31 to 36 cycles at 95°C denaturing, 58°C to 66°C annealing and 72°C extension with a final extension step of 5 minutes. Cycle number and annealing temperature depended on the primer set to be use, each of which had been previously optimized for the PCR technology in our laboratory. For each set of DNA modification and PCR, a cell line or tumor with known hypermethylation as a positive control, normal lymphocyte, or normal tissue DNA as a negative control and water with no DNA template as a control for contamination were included. If no tumor or cell line with known hypermethylation of particular gene was available, normal human lymphocyte DNA in vitro methylated with SssI methylase according to the manufacturers instructions (New England Biolab, Beverly, MA) was used as a positive control. After PCR, samples were run on a 6% nondenaturing acrylamide gel with a 20-bp ladder and analyzed.

The sensitivity of MSP-based detection of hypermethylation in serum was calculated as number of positive tests divided by number of cancer cases. The specificity was calculated as number of negative tests divided by number of cases without cancer and in a second, distinct approach as number of negative tests divided by number of cases without hypermethylation in serum. The association of tumor stage with positive detection in serum was compared using Fisher’s exact test. Results were considered statistically significant if the two-sided P was ≤0.05.

We examined the methylation status of the normally unmethylated APC and RASSF1A tumor suppressor genes and the candidate metastasis inhibitor gene DAP-kinase, previously reported to be frequently hypermethylated in breast cancer (8, 9, 10, 12, 13, 14), in 34 breast tumor and matched serum or plasma DNA. The tumor set we studied included preinvasive and early-stage lesions (three lobular CIS, seven ductal CIS, and five stage I invasive ductal carcinomas) that have a better chance of cure with existing treatment options (2). To detect promoter hypermethylation, we used the sensitive MSP assay, which can detect 0.1% cancer cell DNA from a heterogeneous cell population (24). We found APC to be hypermethylated in 15 of 34 (47%), RASSF1A in 22 of 34 (65%), and DAP-kinase in 17 of 34 (50%) tumor DNA. Because no one gene was known to be hypermethylated in more than a proportion of breast tumors, it was necessary to use a panel of genes to provide a target for detection in the paired sera. Overall, 32 of 34 tumor DNA showed hypermethylation of at least one of the three genes from the panel, which provided a diagnostic coverage (whether a hypermethylated gene was available as a target in each case) of 94%. Hypermethylation was found in ductal CIS, lobular CIS, and invasive ductal and lobular tumors of all pathological grades and stages, which indicated that hypermethylation can be a relatively early event in breast tumorigenesis. Frequent hypermethylation of RASSF1A in ductal CIS has been reported previously (10). In regard to whether lobular CIS is a high-risk marker or a precursor lesion, it will be interesting to additionally examine tumor suppressor gene hypermethylation in lobular CIS. Hypermethylation was found in patients of all ages (Table 1). Three of 3 lobular CIS, 5 of 7 ductal CIS, 3 of 5 stage I, and 15 of 19 stage II to IV, and in total 26 of 34 (76%), patient serum DNA were positive for hypermethylation of a gene known to be methylated in the corresponding tumor DNA (Table 1 and Fig. 1). Positive detection of hypermethylation in serum was not associated with tumor stage (11 of 15 ductal CIS, lobular CIS, and stage I invasive ductal carcinoma versus 15 of 19 stage II to IV invasive ductal carcinoma and invasive lobular carcinoma; P = 1.0 Fisher’s exact test). In four cases (patients 2, 10, 13, and 29), where more than one gene was hypermethylated in the tumor DNA, not all of the methylated genes were detected in the paired serum DNA. This is likely attributable to differences in PCR amplification efficiencies between primer sets but will require further study.

We wished to investigate detection of hypermethylation in serum because it is a more readily accessible bodily fluid compared with nipple aspirate or ductal lavage fluid. Obtaining a peripheral blood sample does not require the presence of a specialist and is relatively painless. Several studies have detected tumor-derived genetic and epigenetic alterations in serum or plasma DNA from cancer patients (17, 18, 19). Hypermethylation was detected at only a slightly lower sensitivity (76%) in serum compared with 82% in our previous study of nipple aspirate fluid (14). A pioneering report of detection of hypermethylation in ductal lavage had a diagnostic coverage of 87.5% and found fluid DNA from 2 of 7 ductal CIS and 17 of 20 invasive breast carcinoma patients to contain hypermethylated DNA (15). Another previous study detected methylation of the p16^INK4a tumor suppressor gene by a methylation-sensitive restriction enzyme site-based assay in 6 of 10 matched plasma DNA from invasive breast cancer patients (25).

There are several potential approaches to improving the level of sensitivity observed in our study. Most immediately, diagnostic coverage can likely be increased to 100% by addition of other genes hypermethylated in breast cancer to the panel. The need for a larger panel of genes reflects the genetic and epigenetic heterogeneity of breast cancer. Significantly larger panels of genes will remain manageable given emerging array technology. We believe that the sensitivity level of hypermethylation-based detection can also be improved upon by the study of optimal serum collection techniques, enrichment of neoplastic cells or DNA from the serum by antibody or oligo-based magnetic bead technology, and improvements in PCR technology.

We examined the tumor cell specificity of the three genes in our study with several different controls. In all cases, a gene unmethylated in the tumor DNA was found to be unmethylated in the matched serum DNA (Table 1). This approach has been validated in previous MSP-based detection studies (17, 18, 26). For example, tumor 14 in Fig. 1 did not have APC hypermethylation, and the matched serum DNA was also negative. Moreover, we did not observe APC, RASSF1A, or DAP-kinase hypermethylation in serum DNA from 20 normal, healthy age-matched controls and 8 patients with the benign breast disease (Fig. 2 and Table 2). We had previously examined six cases of fibrocystic disease and histologically normal breast tissue containing nonneoplastic cells from six breast cancer patients and observed no RASSF1A or DAP-Kinase hypermethylation in these nonneoplastic cells (14). We also found these 12 nonneoplastic DNA to have unmethylated alleles of APC. Methylation of APC has been reported to be absent in previous studies of normal breast cells (13). These findings indicate that aberrant promoter hypermethylation of tumor suppressor genes in serum can be highly specific for cancer. A recent study reported DAP-kinase hypermethylation in normal human lymphocytes by quantitative MSP analysis (27). However, at our routine cycle numbers for conventional MSP, we did not observe DAP-kinase methylation in normal tissue or serum DNA. Future studies could use sufficient controls to address larger issues beyond our pilot study.

Methylation-based detection might also provide differential diagnosis of breast cancer. The E-cadherin and DAP-kinase genes have been reported to show different frequencies of methylation in ductal versus lobular carcinomas (28, 29). The methylation status of other genes such as the estrogen receptor gene, may aid in prognostic subgrouping (30). A recent study reported that DNA methylation in serum was an independent prognostic marker in breast cancer patients (31). It is also possible to screen for more than one tumor type, for example, breast and ovarian neoplasia, simultaneously in serum from high-risk individuals with inherited BRCA1 or BRCA2 mutation. Methylation profiling of breast and ovarian cancer will likely identify genes methylated exclusively in one but not the other of these cancer types (32, 33).

In this study, we have demonstrated sensitive and specific detection of hypermethylation in serum from breast cancer patients. The sensitivity observed in preinvasive or stage I lesions was 73% only slightly lower than the 79% in stage II to IV cancer. All three cases of lobular CIS examined demonstrated hypermethylation in the paired serum. One lobular CIS patient had a bilateral mastectomy and the other two patients had a lumpectomy. All three lobular CIS patients were free of disease 6 to 22 months after surgery. Our study also provides the first demonstration of methylation-based detection of ductal CIS and stage I ductal carcinoma in serum. Interestingly, we detected hypermethylation in the serum DNA from all four invasive lobular carcinoma patients. It is known that invasive lobular carcinoma is more often missed by mammography. Additional studies of the use of hypermethylation-based screening of serum should address improvements in sensitivity, the validation of specificity in larger numbers and examination of high-risk populations. If our results are confirmed in such studies, promoter hypermethylation may have useful clinical application in the detection of breast neoplasia.