Purpose: Genomic alterations (including gene hypermethylation) are likely to precede the phenotypic changes associated with breast tumorigenesis. From a prospective collection of ductal lavage (DL) samples from women with a known mutation in BRCA1 or BRCA2, we have assessed promoter methylation with a comparison of results with several variables, including breast cancer (BC) outcome.

Experimental Design: Hypermethylation of p16, RASSF1A, twist, and RARβ was assessed using a qualitative, real-time, nested PCR assay. Associations between methylation status and variables were tested using Fisher's exact test or logistic regression. Analyses were done at three levels: a single breast, a single duct (both over time), and each DL sample in isolation.

Results: A total of 168 samples from 93 ducts in 54 breasts have been analyzed in 34 women (16 BRCA1 and 18 BRCA2 mutation carriers). A median of 2 DL was done (range, 1–5), with 7 women developing BC on study, 1 bilateral. Methylation of p16 was associated with a known BRCA1 mutation (P = 0.001, P < 0.001, and P < 0.001 for breast, duct, and sample levels, respectively) and women with a history of contralateral BC (P = 0.001 and P < 0.001 for duct and sample levels, respectively). An association was seen for women who developed BC on study and RASSF1A methylation (P = 0.001 for sample level).

Conclusions: Genetic methylation patterns could potentially be used to predict future BC risk. In addition, p16 methylation may be a predictor of BRCA1 mutation status. Further research is required to corroborate these findings. Cancer Epidemiol Biomarkers Prev; 19(1); 265–74

Breast cancer (BC) is the most common cancer affecting women in the western world. For women with a known germ-line mutation in a BC predisposition gene, the BC risks are much higher (1) than the average lifetime risk of 13.2% (2). The estimates for BC risk to age 70 years are as high as 87% for women with a mutation in BRCA1 (1, 3) and 84% in women with a BRCA2 mutation (1, 4). The significantly younger median age of BC (1, 2) and the differing biological nature of the tumors in mutation carriers (5-7) are likely to have contributed to a reduced efficacy of screening, and although the addition of breast magnetic resonance imaging may improve screening efficacy, clinical management of these women remains difficult. Preventative strategies such as bilateral risk-reducing mastectomy and chemoprevention seem acceptable only to a minority (8-10) and not all will develop BC. The demand for improved methods of risk prediction to enable a more tailored approach to management of BC risk is therefore warranted.

Most BCs are thought to arise from ductal epithelium initiated by an accumulation of multiple molecular alterations that precede phenotypic and architectural changes (11, 12). Evaluating these molecular changes is a potential means of identifying early markers of malignant development, with the intraductal approach [nipple aspiration, random periareolar fine-needle aspiration, and ductal lavage (DL); ref. 13] being an attractive means of accessing ductal epithelial cells. Hypermethylation of the promoter region in tumor suppressor genes is not only common in BC cells (14) but has also been identified in premalignant, atypical epithelial cells (12) and whereby potentially representative of an early event in carcinogenesis. Even more appealing is the reversible nature of methylation with the potential to reduce the risk of carcinoma development with targeted therapies (15, 16).

From a prospective collection of DL samples from breasts unaffected by cancer in women with a known mutation in BRCA1 or BRCA2, we report our findings for the assessment of gene methylation in four genes—p16, RASSF1A, twist, and RARβ [chosen based on previous reports documenting a >10% prevalence of methylation in primary BCs (14, 17, 18) and/or that they were of biological relevance in breast tumorigenesis (12, 19, 20)]—with a comparison of results with cytologic findings and BC outcome.

Subjects

Eligible subjects were recruited between March 2003 and February 2005 from three Australian familial cancer centers (Peter MacCallum Cancer Centre; Royal Melbourne Hospital, Melbourne; and Westmead Hospital, Sydney) as previously described (21). In brief, women were eligible for the study if they were ages between 25 and 65 y, had a documented mutation in either BRCA1 or BRCA2, and had at least one breast unaffected by BC. Reasons for exclusion included pregnancy or lactation and previous breast surgery that could have potentially disrupted the breast ductal system. All participants who had enrolled in the study by March 2005, who had at least one lavage specimen successfully collected, were included in this study, analyzing for gene methylation.

The study was approved by each Hospital Research Ethics Committee, and all participating subjects provided written informed consent.

Specimen Collection

DL, nipple aspiration, and venesection were completed every 6 mo from the time of study for a maximum of 3 y with annual clinical follow-up planned for 10 y, again as previously described (21, 22).

Ductal Lavage

Lavage was done as previously described with minor modifications to the operator techniques (21). From May 2005, 1 mL of 1% lignocaine was instilled at the base of the nipple using an intradermal needle to produce complete sensory block of the nipple. Cannulation was attempted on both nipple aspiration fluid (NAF)–producing ducts and any additional “dry” ducts detected by gentle probing, with a maximum of three ducts per visit. If NAF was produced, attempts were made to cannulate the specific NAF-producing ducts.

All biospecimens collected during the study were processed, logged, and tracked by Tissue Bank staff from the Peter MacCallum Cancer Centre.

Sample Processing

Between 10 and 12 mL of DL, washings were recovered from each duct with processing of the sample within 2 h by tissue bank staff as previously described (21, 22). Briefly, DL fluid was spun before removal of the supernatant down to 2 mL. Cell counts were estimated using a hemocytometer and trypan blue. If the cell count was estimated to be >20,000 cells, the sample was then divided into ∼10,000 cells/pellet before snap freezing in liquid nitrogen and storage at −80°C. If the cell count was estimated to be <20,000 cells, the sample was divided into two samples before freezing and storage.

DNA Extraction and Quantification

DNA was extracted from DL cellular pellets using the DNeasy Tissue kit (Qiagen), and lymphocytic DNA was processed using the Mini Blood kit (Qiagen). The DNA yield was estimated for a subset of samples using quantitative PCR, with DNA quantity estimated by plotting against a standard curve of known dilutions of DNA derived from normal lymphocytes as described previously(22, 23).

Bisulfite Modification

Extracted DNA (2 μL in each reaction) was modified with sodium bisulfite to convert unmethylated (but not methylated) cytosine to uracil using the MethylEasy DNA Bisulphite Modification kit according to the manufacturer's instruction (Genetic Signatures). The bisulfite-modified DNA was resuspended in a volume of 20 μL and stored at 4°C until use.

Methylation-Specific PCR

The primer (GeneWorks) and probe (BioSearch Technologies) sequences used to amplify the bisulfite-treated DNA are described in Table 1. Methylation-specific PCR (MSP) was done in two separate amplification reactions: an initial methylation-independent PCR (MIP) followed by a MSP. For quality control, two sets of water, “no-template” controls (NTC) were used; the first was amplified in the initial MIP reaction. For the MSP reaction, 1 μL of the NTC MIP product was placed into the follow-on MSP reaction. In addition, a second “fresh” NTC sample was also used in the MSP amplification round. The following controls were used to create standards in each run: 100% Sss1-treated lymphocytic DNA, 1% Sss1-treated lymphocytic DNA, and 0% Sss1-treated (bisulfite-converted) lymphocytic DNA (Fig. 1). All controls were aliquots of the same stock frozen at −80°C. A sample was considered to be methylated for a given gene if the product amplified indicated >1% methylation.

Table 1.

PCR conditions and sequences for primers and probes used

Primer nameSequence 5′-3′OrientationStatus
p16 Ext-F ggaggggagtagtatggagt Forward MIP 
p16 Ext-R acaaaccctctacccacctaaatc Reverse MIP 
p16 RT-F tggagttttcggttgattggtt Forward MSP 
p16 RT-R aacaacgcccgcacctcct Reverse MSP 
p16 RT-probe HEX-acccgaccccgaaccgcg-BHQ1 Reverse MSP 
p16: Mg concentration: 3 mmol/L; primer concentration: 1 μmol/L; PCR conditions: 95°C for 10 min, 40 cycles of 95°C for 30 s, 57°C for 30 s, 72°C for 45 s 
RASSF1A Ext-F gttttatagtttttgtattta Forward MIP 
RASSF1A Ext-R caactcaataaactcaaactccc Reverse MIP 
RASSF1A RT-F cccgtacttcgctaactttaaacg Forward MSP 
RASSF1A RT-R gcgttgaagtcggggttc Reverse MSP 
RASSF1A RT-probe FAM-acaaacgcgaaccgaacgaaacca-BHQ1 Reverse MSP 
RASSF1A: Mg concentration: 4 mmol/L; primer concentration: 5 μmol/L; PCR conditions: 95°C for 10 min, 35 cycles of 95°C for 30 s, 55°C for 30 s, 72°C for 45 s 
RARβ Ext-F gtaggagggtttattttttgtt Forward MIP 
RARβ Ext-R aattacattttccaaacttactc Reverse MIP 
RARβ RT-F tgtcgagaacgcgagcgattc Forward MSP 
RARβ RT-R cgaccaatccaaccgaaacga Reverse MSP 
RARβ RT-probe FAM-atcctaccccgacgatacccaaac-BHQ1 Reverse MSP 
RARβ: Mg concentration: 2 mmol/L; primer concentration: 5 μmol/L; PCR conditions: 95°C for 10 min, 40 cycles of 95°C for 30 s, 55°C for 30 s, 72°C for 45 s 
twist Ext-F gagatgagatattatttattgtg Forward MIP 
twist Ext-R cctcccaaaccattcaaaaac Reverse MIP 
twist RT-F gttagggttcgggggcgttgtt Forward MSP 
twist RT-R ccgtcgccttcctccgacgaa Reverse MSP 
twist RT-probe FAM-aaacgatttccttccccggccgaaa-BHQ1 Reverse MSP 
twist: Mg concentration: 4 mmol/L; primer concentration: 1 μmol/L; PCR conditions: 95°C for 10 min, 35 cycles of 95°C for 30 s, 53°C for 30 s, 72°C for 45 s 
Primer nameSequence 5′-3′OrientationStatus
p16 Ext-F ggaggggagtagtatggagt Forward MIP 
p16 Ext-R acaaaccctctacccacctaaatc Reverse MIP 
p16 RT-F tggagttttcggttgattggtt Forward MSP 
p16 RT-R aacaacgcccgcacctcct Reverse MSP 
p16 RT-probe HEX-acccgaccccgaaccgcg-BHQ1 Reverse MSP 
p16: Mg concentration: 3 mmol/L; primer concentration: 1 μmol/L; PCR conditions: 95°C for 10 min, 40 cycles of 95°C for 30 s, 57°C for 30 s, 72°C for 45 s 
RASSF1A Ext-F gttttatagtttttgtattta Forward MIP 
RASSF1A Ext-R caactcaataaactcaaactccc Reverse MIP 
RASSF1A RT-F cccgtacttcgctaactttaaacg Forward MSP 
RASSF1A RT-R gcgttgaagtcggggttc Reverse MSP 
RASSF1A RT-probe FAM-acaaacgcgaaccgaacgaaacca-BHQ1 Reverse MSP 
RASSF1A: Mg concentration: 4 mmol/L; primer concentration: 5 μmol/L; PCR conditions: 95°C for 10 min, 35 cycles of 95°C for 30 s, 55°C for 30 s, 72°C for 45 s 
RARβ Ext-F gtaggagggtttattttttgtt Forward MIP 
RARβ Ext-R aattacattttccaaacttactc Reverse MIP 
RARβ RT-F tgtcgagaacgcgagcgattc Forward MSP 
RARβ RT-R cgaccaatccaaccgaaacga Reverse MSP 
RARβ RT-probe FAM-atcctaccccgacgatacccaaac-BHQ1 Reverse MSP 
RARβ: Mg concentration: 2 mmol/L; primer concentration: 5 μmol/L; PCR conditions: 95°C for 10 min, 40 cycles of 95°C for 30 s, 55°C for 30 s, 72°C for 45 s 
twist Ext-F gagatgagatattatttattgtg Forward MIP 
twist Ext-R cctcccaaaccattcaaaaac Reverse MIP 
twist RT-F gttagggttcgggggcgttgtt Forward MSP 
twist RT-R ccgtcgccttcctccgacgaa Reverse MSP 
twist RT-probe FAM-aaacgatttccttccccggccgaaa-BHQ1 Reverse MSP 
twist: Mg concentration: 4 mmol/L; primer concentration: 1 μmol/L; PCR conditions: 95°C for 10 min, 35 cycles of 95°C for 30 s, 53°C for 30 s, 72°C for 45 s 

Abbreviations: Ext-F, external primer forward; Ext-R, external primer reverse; RT-F, real-time forward primer; RT-R, real-time reverse primer.

Figure 1.

Quantified data display of controls and samples assayed using RASSF1A. An example of quantified data obtained using primers and probes for the gene RASSF1A with controls is shown. Control 1 contains 100% Sss1-methylated DNA without any added bisulfite-treated unmethylated product. Control 2 contains Sss1-methylated DNA (1% of the total mix) together with bisulfite-treated but unmethylated DNA (99% of the total mix; all with DNA inputs of 10,000 pg/μL). Control 3 contains 100% bisulfite-modified but unmethylated product. Control 4 represents a no-template sample, containing purified water only. Amplification for RASSF1A methylation was only seen for the controls containing methylated DNA. In addition, two samples are shown. Using a BARD1 probe, the DNA input for the two samples was estimated to be 2,935 pg/μL for sample 1 and 1,839 pg/μL for sample 2 (data not shown). Orange, 100% and 1% methylated controls; purple, DL samples.

Figure 1.

Quantified data display of controls and samples assayed using RASSF1A. An example of quantified data obtained using primers and probes for the gene RASSF1A with controls is shown. Control 1 contains 100% Sss1-methylated DNA without any added bisulfite-treated unmethylated product. Control 2 contains Sss1-methylated DNA (1% of the total mix) together with bisulfite-treated but unmethylated DNA (99% of the total mix; all with DNA inputs of 10,000 pg/μL). Control 3 contains 100% bisulfite-modified but unmethylated product. Control 4 represents a no-template sample, containing purified water only. Amplification for RASSF1A methylation was only seen for the controls containing methylated DNA. In addition, two samples are shown. Using a BARD1 probe, the DNA input for the two samples was estimated to be 2,935 pg/μL for sample 1 and 1,839 pg/μL for sample 2 (data not shown). Orange, 100% and 1% methylated controls; purple, DL samples.

Close modal

In the MIP amplification step, 1 μL of bisulfite-modified DNA was added to 19 μL of reaction buffer [2 μL of PCR buffer (Qiagen), 1 μL of 200 μmol/L deoxynucleotide triphosphates, 1 μL primer mix (Table 1), MgCl2, 0.2 μL of 0.5 unit/reaction HotStar Taq (Qiagen), with additional dH2O to bring the total volume to 20 μL], with conditions optimized for each primer set (Table 1). For the second nested PCR round, 1 μL of the PCR product from the MIP amplification was added to 19 μL of reaction buffer [10 μL of 2× Master Mix, 2.5 μL × 1 μmol/L primer mix, 4 μL of the Taqman probe (SA Biosciences; at a concentration of 333 nmol/L, except for twist, where a concentration of 1.5 μmol/L was used), 2.5 μL dH2O]. All reactions used Taqman Universal PCR Master Mix (Applied Biosystems), except p16, where ABsolute QPCR Mix (Advanced Biotechnologies) was used. The reaction was carried out using a Corbett Rotorgene (Corbett Life Science Research) in 0.2 μL tubes and a 72-well rotor rack. PCR conditions used for all genes were 95°C for 10 min followed by 40 cycles of 95°C for 10 s and 60°C for 45 s.

Validation of Methylation Assay

A control sample of blood from a single healthy donor (nonstudy participant) was acquired from the Peter MacCallum Cancer Centre Tissue Bank. This was used to obtain lymphocytic DNA to be used as a control for all experiments. This DNA was first quantified by real-time PCR using a probe within the BARD1 gene as described previously (23) and then diluted to a concentration of 10,000 pg/μL. A proportion of the control was methylated in vitro using Sss1 methylase according to the manufacturer's instructions (New England Biolabs), with the remaining proportion left untreated. Both Sss1-treated and Sss1-untreated DNA was then bisulfite treated. These were used as the control DNA for all PCRs. The controls were assessed using RASSF1A-methylated and RASSF1A-unmethylated primer sets to ensure that the Sss1 methylase–treated control was 100% methylated and that the lymphocytic sample was 100% unmethylated. Aliquots (10 μL) of 10,000 pg/μL of both methylated and unmethylated controls were stored at −80°C for use as controls in all subsequent assays.

Attempts were made to develop an assay to quantitate post–bisulfite-treated DNA using a nested protocol; however, results were inconsistent between repeated experiments such that it was not possible to reliably quantify the amount of DNA for each sample. In a subset of 46 samples from BRCA1 carriers and 39 samples from BRCA2 carriers, the DNA was quantitated before bisulfite treatment using a quantitative real-time PCR as described above. As the majority of samples had very low quantities of DNA, as previously described (22), we elected to carry out all reactions using 1 μL of bisulfite-treated sample into the first PCR amplification and then 1 μL of PCR product for the nested amplification. For several samples, as the total cell count was estimated to be zero, but because it is possible for the sample to contain free DNA that can potentially be amplified using a nested technique, all samples were used to assess for methylation, regardless of their original estimated cell count. Using the BARD1 quantitation assay, some samples were assessed as having no amplifiable DNA. However, as the methylation assay involved nested PCR and was highly sensitive, we included all samples to assess for hypermethylation.

To assess the accuracy of the assay for each gene, serial dilutions of the 100% Sss1-methylated control were used with an input from 10,000 to 50 pg/μL. The reliability of the assay was reduced with inputs of <100 pg/μL of Sss1-treated DNA. In addition to this, we assessed the methods for each gene using mix of control DNA containing 1% Sss1-methylated DNA added to 99% bisulfite-treated but unmethylated DNA, with dilutions ranging from 10,000 to 1,000 pg/μL. The lower limits of reliable amplification were seen at an input of 2,000 pg/μL of total DNA, representing an input of 20 pg/μL of methylated DNA. Controls used for all subsequent assays included duplicates of 100% Sss1-methylated and 1% Sss1-methylated, bisulfite-treated DNA (but 100% unmethylated) samples, all containing a DNA input of 10,000 pg/μL together with duplicates of NTC (Fig. 1).

Statistical Analysis

The association between methylation status and several categorical variables was tested using Fisher's exact test. Logistic regression was used to examine the relationship between methylation status and patient age and to examine the effects of multiple variables. All P values are two-sided. As this is a hypothesis-generating study, no adjustments have been made for multiple testing. Statistical analysis was done using StatXact version 6.0 (Cytel Software Corp.) and S-plus 2000 (MathSoft, Inc.).

Analyses were done at three levels for subjects: first, by using a single breast as the unit of analysis; second, by means of a single duct as the unit of interest; and third, by analyzing each sample in isolation. We elected to do the analyses this way for several reasons. By examining per breast, we attempted to overcome a potentially dominating effect an individual may have had if methylation represented a global change; that is, if a single woman has multiple ducts with hypermethylation, this may have had an overriding effect on results given the small number of participants. By examining per duct, we were able to get an indication of whether there was a local duct effect on cancer risk rather than the methylation representing a field effect or global cancer risk; pooling the results from a duct over time reduces the possibility of sampling error at a single time point. Examining by individual sample gave a crude relationship between methylation and BC risk but was vulnerable to the dominating effect of a repeated positive methylation result from a single duct over time. A breast was declared as having a particular methylation if it was present in any sample from any duct within the breast and as having atypia if atypia was present in any duct on any occasion. A duct was declared as having a methylation of a particular gene if it was present in any sample from a single duct and as having atypia if atypia was present in the duct on any occasion.

Variables assessed included age, BRCA1 or BRCA2 mutation status, menopausal status, previous bilateral oophorectomy, parity, previous history of BC on the contralateral side, BC development after enrollment onto the study, NAF production, and cytologic atypia (mild and severe based on local cytologic scoring system; ref. 21).

Subjects

Up to March 2005, a total of 63 women were recruited to participate in the DL study. Of these, the positive controls (preoperative DL samples taken from women with known BC but no known BRCA1 or BRCA2 mutation and no family history to suggest a high risk for hereditary BC syndrome) and negative controls (women from families with a known BRCA1 or BRCA2 mutation who had tested negative for that mutation) were excluded from this study. In addition, four women were excluded from this study as, while they had NAF and bloods collected and stored, DL collection was attempted but not possible and therefore no sample was available. Finally, three women were excluded from the analyses because of incorrect date or duct details recorded on the sample tubes. Therefore, 34 women from the total study cohort were eligible for this study and were included in the methylation analysis: 16 who carried a mutation in BRCA1 (47%) and 18 with a BRCA2 mutation (53%; Table 2).

Table 2.

Subject characteristics

CharacteristicCategoryNo.%
Age (y) Median 43  
Range 27-60  
No. visits for DL Median  
Range 1-5  
No. distinct ducts lavaged Median  
Range 1-7  
BRCA gene mutation 16 47% 
18 53% 
Menopausal status Pre 18 53% 
Post 16 47% 
Bilateral oophorectomy No 18 53% 
Yes 16 47% 
Tamoxifen use No 32 94% 
Past 3% 
Current 3% 
Previous BC No 23 68% 
Yes 11 32% 
Parous No 18% 
Yes 28 82% 
Cytologic atypia seen on any DL No 25 74% 
Yes 26% 
BC development during study No 27 79% 
Yes 21% 
Previous breast feeding No 24% 
Yes 26 76% 
Use of contraceptive pill Never 6% 
Yes, past 30 88% 
Yes, currently 6% 
CharacteristicCategoryNo.%
Age (y) Median 43  
Range 27-60  
No. visits for DL Median  
Range 1-5  
No. distinct ducts lavaged Median  
Range 1-7  
BRCA gene mutation 16 47% 
18 53% 
Menopausal status Pre 18 53% 
Post 16 47% 
Bilateral oophorectomy No 18 53% 
Yes 16 47% 
Tamoxifen use No 32 94% 
Past 3% 
Current 3% 
Previous BC No 23 68% 
Yes 11 32% 
Parous No 18% 
Yes 28 82% 
Cytologic atypia seen on any DL No 25 74% 
Yes 26% 
BC development during study No 27 79% 
Yes 21% 
Previous breast feeding No 24% 
Yes 26 76% 
Use of contraceptive pill Never 6% 
Yes, past 30 88% 
Yes, currently 6% 

The median age of the subjects was 43 years (range, 27-60 years). Nipple aspiration and DL were attempted in all women on at least one occasion (median, 2 visits; range, 1-5 visits). Menopausal status was assessed at the baseline entry into the study. Those women who were premenopausal and then became postmenopausal while ongoing samples were collected were considered premenopausal for the purpose of these analyses. All postmenopausal women had undergone a bilateral. Since commencement of the study, seven women have withdrawn: one due to increased anxiety following a cytologic diagnosis of severe atypia and six following bilateral mastectomy. Seven women (six with BRCA1 mutations) have been diagnosed with BC (one with bilateral BC) since enrolling on the study. All tumors from BRCA1 mutation-positive women were estrogen and progesterone receptor negative and c-erbB2 negative and classified as grade 3. The tumor occurring in the woman with a BRCA2 mutation was estrogen and progesterone positive and c-erbB2 negative and classified as grade 3.

A total of 168 samples from 93 ducts in 54 breasts have been analyzed for methylation in the promoter region of p16, RASSF1A, RARβ, and twist. The association between the presence of methylation and other variables has been done for all genes inclusively and separately for each specific gene.

Methylation Using a Single Breast as the Unit for Analysis

Of the 54 breasts eligible for analysis, 24 (44.4%) were from BRCA1 mutation carriers, whereas 30 (55.6%) were from BRCA2 mutation carriers. Methylation was detected in a total of 33 (61.1%) breasts: 17 in BRCA1 mutation carriers (70.8%) and 16 in BRCA2 mutation carriers (53.3%). There were no significant associations found between factors measured and methylation if all genes are combined.

Methylation of p16 was detected in 20 (37.0%) breasts. The presence of p16 methylation was associated with having a mutation in BRCA1 (P = 0.001), but no other factors measured were significantly associated with p16 at an individual breast level (Table 3).

Table 3.

p16 methylation: summary

Breast (n = 54)Duct (n = 93)Sample (n = 168)
MethylationMethylationMethylation
YesNoPYesNoPYesNoP
BRCA1 15 0.001 26 20 <0.001 39 44 <0.001 
BRCA2 25  41  78  
Premenopausal 13 16 0.263 22 34 0.269 33 59 0.009 
Postmenopausal 18  10 27  13 63  
No BSO 13 16 0.263 22 34 0.269 33 59 0.009 
Previous BSO 18  10 27  13 63  
No PHx CBC 14 29 0.294 18 54 0.001 20 107 <0.001 
PHx CBC  14  26 15  
Parous 14 29 0.294 25 49 0.793 38 100 1.000 
Nulliparous  12  22  
Cyt. atypia detected 0.744 0.167 10 0.098 
No atypia 15 27  23 52  38 112  
BC developed on study 0.450 0.359 10 13 0.079 
No BC 16 30  26 54  36 109  
NAF producer 16 21 0.229 25 47 1.000 37 82 0.127 
No NAF 13  14  40  
Age (cont)   0.219   0.150   0.004 
Breast (n = 54)Duct (n = 93)Sample (n = 168)
MethylationMethylationMethylation
YesNoPYesNoPYesNoP
BRCA1 15 0.001 26 20 <0.001 39 44 <0.001 
BRCA2 25  41  78  
Premenopausal 13 16 0.263 22 34 0.269 33 59 0.009 
Postmenopausal 18  10 27  13 63  
No BSO 13 16 0.263 22 34 0.269 33 59 0.009 
Previous BSO 18  10 27  13 63  
No PHx CBC 14 29 0.294 18 54 0.001 20 107 <0.001 
PHx CBC  14  26 15  
Parous 14 29 0.294 25 49 0.793 38 100 1.000 
Nulliparous  12  22  
Cyt. atypia detected 0.744 0.167 10 0.098 
No atypia 15 27  23 52  38 112  
BC developed on study 0.450 0.359 10 13 0.079 
No BC 16 30  26 54  36 109  
NAF producer 16 21 0.229 25 47 1.000 37 82 0.127 
No NAF 13  14  40  
Age (cont)   0.219   0.150   0.004 

Abbreviations: BSO, bilateral salpingo-oophorectomy; PHx, past history; CBC, contralateral BC; Cyt, cytologic; cont, continuous.

The presence of methylation in RASSF1A was detected in 13 (24.1%) breasts: 6 from BRCA1 mutation carriers (25%) and 7 from BRCA2 mutation carriers (23.3%). No significant associations were found for the factors measured. Hypermethylation of twist was detected in only three (5.6%) breasts: two BRCA1 mutation carriers and one BRCA2 mutation carrier (8.3% and 3.3% of all BRCA1 and BRCA2 mutation carriers, respectively). Methylation of twist was associated with the presence of cytologic atypia (P = 0.009; data not shown). A further 13 (24.1%) breasts, 4 from women with BRCA1 mutations and 9 from BRCA2 mutation carriers (16.7% and 30.0% of all BRCA1 and BRCA2 mutation carriers, respectively), were found to have methylation of RARβ, with a significant association between NAF production and the presence of methylation in this gene at the individual breast level (P = 0.043; data not shown).

Methylation Using a Single Duct as the Unit for Analysis

A total of 93 ducts were included in this analysis: 46 (49.5%) from BRCA1 mutation carriers and 47 (50.5%) from BRCA2 mutation carriers. Of the 93, 48 (51.6%) showed methylation in at least one of the genes analyzed. There was an association found between the presence of methylation and having a mutation in BRCA1 (P = 0.004) and a previous history of BC (on the contralateral side; P = 0.049). A linear logistic model incorporating BRCA status, indicator of previous BC, and their interaction was fitted:

where p̂ = fitted probability of methylation in any gene, BRCA = 0 (BRCA1) or 1 (BRCA2), and previous BC = 0 (no) or 1 (yes). For BRCA1 mutation carriers, the predicted probability for methylation in any gene is 56.7% (SE, 9.1%) for women who have not had previous BC and 87.5% (SE, 8.1%) for women who have had previous BC. For BRCA2 mutation carriers, the predicted probability of methylation in any gene is 38.1% (SE, 7.5%) for women with no history of BC and 20.0% (SE, 17.8%) for women with a history of BC.

Methylation of p16 was detected in 32 (34.4%) ducts: 26 from BRCA1 mutation carriers and 6 from BRCA2 mutation carriers (56.5% and 12.8% of all ducts from BRCA1 and BRCA2 mutation carriers, respectively). Methylation of p16 was associated with a mutation in BRCA1 (P < 0.001) and having had a previous BC (P = 0.001) but not for other factors measured (Table 3). A linear logistic model incorporating BRCA status, indicator of previous BC, and their interaction was fitted:

where p̂ = fitted probability of methylation of p16, BRCA = 0 (BRCA1) or 1 (BRCA2), and previous BC = 0 (no) or 1 (yes). Among BRCA1 mutation carriers, the predicted probability of methylation of p16 is 40% (SE, 8.9%) for women who have not had previous BC and 87.5% (SE, 8.3%) for women who have had a previous BC. For BRCA2 mutation carriers, the predicted probability of methylation of p16 is 14.3% (SE, 5.4%) for women with no history of BC and 0.0% (SE, 0.4%) for women with a history of BC.

A total of 14 (15.1%) ducts were found to be methylated for RASSF1A: 7 from BRCA1 mutation carriers (15.2%) and 7 from BRCA2 mutation carriers (14.9%). RARβ was found to be methylated in 14 (15.1%) ducts: 5 from BRCA1 mutation carriers (10.9%) and 9 from BRCA2 mutation carriers (19.1%). Methylation of twist was seen in three ducts: two from BRCA1 mutation carriers (4.3%) and one from BRCA2 mutation carrier (2.1%). No associations were found for factors measured and the methylation for RASSF1A, RARβ, or twist.

Methylation Using a Single Sample as the Unit for Analysis

In total, 168 samples were included for analysis: 83 (49.4%) from BRCA1 mutation carriers and 85 (50.6%) from BRCA2 mutation carriers. Overall, in 67 (39.9%) samples, methylation of at least one gene was detected: 45 in BRCA1 mutation carriers (54.2%) and 22 in BRCA2 mutation carriers (25.9%). Methylation of any gene was associated with a mutation in BRCA1 (P < 0.001), being premenopausal (P = 0.004), no history of bilateral oophorectomy (P = 0.004), a history of previous BC on the contralateral side (P < 0.001), NAF production (P = 0.01), and increasing age (P = 0.013). In a multivariate analysis, the following model was identified as the best fitting:

where p̂ = fitted probability of methylation in any gene, BRCA = 0 (BRCA1) or 1 (BRCA2), Menopause = 0 (post) or 1 (pre), and previous BC = 0 (no) or 1 (yes).

Methylation of p16 was seen in 45 (27.4%) samples: 39 from BRCA1 mutation carriers (47.0%) and 7 from BRCA2 mutation carriers (8.2%). At this level of analysis, methylation of p16 was associated with a mutation in BRCA1 (P < 0.001), having had a previous BC (P < 0.001), being premenopausal (P = 0.009), not having had a bilateral oophorectomy (P = 0.009), and age (continuous; P = 0.004) but not for other factors measured (Table 3). In a multivariate analysis, the following model was identified as the best fitting:

where p̂ = fitted probability of methylation of p16, BRCA = 0 (BRCA1) or 1 (BRCA2), and previous BC = 0 (no) or 1 (yes).

In 19 (11.3%) samples, methylation of RASSF1A was detected: 11 samples from BRCA1 mutation carriers (13.3%) and 8 from BRCA2 mutation carriers (9.4%). An association was found between methylation of RASSF1A in a sample and developing BC while on study (P = 0.001) but not for other factors measured. Methylation of twist was detected in three (1.8%) samples: two in BRCA1 mutation carriers (2.4%) and one in a BRCA2 mutation carrier (1.2%). RARβ was found to be methylated in 14 (8.3%) samples: 5 collected from BRCA1 mutation carriers (6.0%) and 9 from BRCA2 mutation carriers (10.6%). No significant associations were found for factors measured and the methylation of either twist or RARβ.

A summary of the individual methylation findings, together with tumor data, is presented in Supplementary Table S1 to facilitate a greater depth of appreciation of the contribution a specific individual had on the analysis of results together with the specifics of those individuals who developed a BC.

This study describes the findings of promoter hypermethylation in a prospectively collected series of DL samples from 34 women at high risk of BC development due to an inherited germ-line mutation in either BRCA1 or BRCA2. Analysis was done at three levels: first, by using a single breast as the unit of measurement (over time); second, by using a single duct (over time); and third, by analyzing each sample in isolation. Encouragingly, similar results are seen at all three levels, although the power to detect an association between factors measured was greater at the duct and sample level of analyses. As it is highly unlikely that methylation in a single gene will occur in all premalignant lesions or in 100% of BCs (of all histologic types), a small panel of genes was selected based on published data suggesting a role in breast tumorigenesis (7, 14, 17-19, 24, 25).

The cyclin-dependent kinase inhibitor 2A (p16INK4a) plays an integral role in cell cycle progression through the G1 phase of cell cycling via its inhibition of Rb phosphorylation (12, 26). The reported rate of methylation of p16 in breast tumors varies from 0 of 32 to 10 of 15 (0-67%); (7, 17, 24, 25, 27). However, its role in BC development (14, 20), and particularly in precursor lesions (12), made it an ideal candidate for this study.

Methylation of p16 was seen in 37.0% of breasts (34.4% of ducts and 27.4% of samples) included in this study. Methylation was more likely in women who had a previous history of BC (P = 0.001 and P < 0.001 for duct and sample as unit measured, respectively). Previous reports suggest lower rates of p16 methylation than seen in this study (7, 17). These disparate results may reflect several differences between the studies, the most notable being that we have not assessed methylation in the actual tumors but in DL samples. It is possible that the methylation signals may arise from cells present in the intraductal microenvironment, such as foam cells and lymphocytes, and therefore reflect early changes occurring in the microenvironment rather than the actual tumor. However, significant rates of p16 methylation have been reported in subpopulations of both morphologically normal breast epithelial cells and in early premalignant lesions in women at increased risk for BC development (12). It has been suggested that early loss of p16 function provides a pivotal trigger for cells to subsequently accumulate further CpG island methylation and somatic genetic mutations, which in turn promote malignant progression. Perhaps even more interesting is the consistently strong association seen between a known BRCA1 mutation and methylation of p16 (P = 0.001, P < 0.001, and P < 0.001 for breast, duct, and sample as unit measured, respectively). The biological explanation for this association is unclear and beyond the scope of this study, but it is recognized that both breast and ovarian cancers occurring in the setting of a germ-line mutation in BRCA1 or BRCA2 have a unique and consistent molecular expression pattern (28, 29) and patterns of gene methylation (7, 17, 30, 31).

The RAS association domain family 1 isoform A (RASSF1A) is a tumor suppressor gene likely to have a major role in the regulation of cell proliferation and apoptosis. Loss of function of RASSF1A due to methylation has been shown with high frequency in many tumors, including BCs (25, 32-35). Of particular relevance to this study population is the finding that loss of function of RASSF1A (due to the polymorphism A133S) coupled with the presence of a germ-line mutation in BRCA1 or BRCA2 mutations was associated with an increase in BC risk and BC at a younger age (36). In addition, the finding of a high frequency of RASSF1A methylation in preneoplastic lesions, such as hyperplasia and papillomatosis (25, 37), and in situ carcinomas (low to high grade; refs. 18, 25, 33, 38) is indicative that methylation of this gene may be an early event in BC development. The association of RASSF1A methylation in this study and BC development (P = 0.382, P = 0.103, and P = 0.001 for breast, duct, and sample as unit measured, respectively) is therefore an interesting finding. Although at a breast and duct level the association was not significant, this may reflect insufficient power due to the small number of samples included in these analyses. In addition, several samples included in this study contained very low levels of DNA, so our results may underestimate this association. For both subjects who developed BC while on study, and who were not found to have methylation in either RASSF1A or p16, all lavage samples that preceded the BC diagnoses were acellular and therefore had extremely low levels of amplifiable DNA, again potentially affecting the quality of assay result and therefore underestimating the potential association.

To our knowledge, this study represents the largest study assessing gene methylation in ductal fluid collected serially from BRCA1 and BRCA2 mutation carriers. Locke et al. (39) describe a smaller study in BRCA1 or BRCA2 mutation-positive women with a differing gene panel. Most notably, this group assessed methylation in free DNA taken from the supernatant rather than extracted from identified cell pellets; however, the group reported sufficient DNA in the majority of samples to carry out the analysis. No significant associations between methylation findings and variables measured, possibly due to the small numbers, were found. In other studies, although mutation-positive women were assessed for methylation, their results were not analyzed separately from other high-risk, nonmutation-positive women (40, 41).

In summary, this study forms the basis for hypothesizing that methylation profiling of tumor suppressor genes may identify potential markers of malignant transformation and, in particular, a potential explanation as to some of the expression profile variables unique to breast tumors occurring in BRCA1 mutation carriers. Further research is required to validate these findings, as the limitations of this study include a small study population and short duration of clinical follow-up. The rationale as to why inhibition of the p16-Rb regulatory controls should be specific to BRCA1 mutation-positive women also remains largely unexplored but an intriguing question for further research. An expanded repertoire of genes is also desirable, with only two of the four genes selected having a significant association with BC diagnosis. The development of technologies for whole-genome CpG methylation analysis is progressing rapidly and, with its application to BC arising in specific populations (such as BRCA1 or BRCA2 mutation carriers), is likely to assist in the identification of new candidate genes.

Although cell sampling using DL may represent an excellent method for examining the earliest epigenetic and genetic changes of breast tumorigenesis, particularly at-risk populations, accessing the same duct or acquiring a usable cellular sample from the same duct over time remains problematic. A potentially more desirable approach might be to apply methylation technologies to plasma or serum samples or even breast cells collected intermittently by random periareolar fine-needle aspiration, which overcomes the sampling issue (42). This approach has been used in women with early and advanced BC (43-45), among other tumor types (46-48); however, whether it will be sufficiently sensitive for detection of biomarkers released from small precursor lesions, considering the massive dilution that occurs once released into the systemic blood circulation, remains unknown.

No potential conflicts of interest were disclosed.

Grant Support: Y.C. Antill was the recipient of the Goodman Fielder National Breast Cancer Foundation Scholarship. G.J. Lindeman is the current National Health and Medical Research Council (NHMRC) Senior Research Fellow. K.A. Phillips is the Cancer Council Victoria, Dr. John Colebatch Clinical Research Fellow. I.G. Campbell is the current NHMRC Senior Research Fellow. This research project is funded by the NHMRC program grant “Beyond BRCA1/2.”

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Antoniou
A
,
Pharoah
PD
,
Narod
S
, et al
. 
Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case series unselected for family history: a combined analysis of 22 studies
.
Am J Hum Genet
2003
;
72
:
1117
30
.
2
Ries
L
,
Eisner
MP
,
Kosary
CL
, et al
. 
SEER cancer statistics review, 1975-2002
.
Bethesda
:
National Cancer Institute
; 
2004
.
3
Ford
D
,
Easton
DF
,
Bishop
DT
,
Narod
SA
,
Goldgar
DE
. 
Risks of cancer in BRCA1-mutation carriers. Breast Cancer Linkage Consortium
.
Lancet
1994
;
343
:
692
5
.
4
Ford
D
,
Easton
DF
,
Stratton
M
, et al
. 
Genetic heterogeneity and penetrance analysis of the BRCA1 and BRCA2 genes in breast cancer families. The Breast Cancer Linkage Consortium
.
Am J Hum Genet
1998
;
62
:
676
89
.
5
Porter
PL
,
El-Bastawissi
AY
,
Mandelson
MT
, et al
. 
Breast tumor characteristics as predictors of mammographic detection: comparison of interval- and screen-detected cancers
.
J Natl Cancer Inst
1999
;
91
:
2020
8
.
6
Phillips
K
. 
Re: Biologic characteristics of interval and screen-detected cancers [letter]
.
J Natl Cancer Inst
2001
;
93
:
151
2
.
7
Suijkerbuijk
KP
,
Fackler
MJ
,
Sukumar
S
, et al
. 
Methylation is less abundant in BRCA1-associated compared with sporadic breast cancer
.
Ann Oncol
2008
;
19
:
1870
4
.
8
Antill
Y
,
Reynolds
J
,
Young
MA
, et al
. 
Risk-reducing surgery in women with familial susceptibility for breast and/or ovarian cancer
.
Eur J Cancer
2006
;
42
:
621
8
.
9
Phillips
KA
,
Jenkins
MA
,
Lindeman
GJ
, et al
. 
Risk-reducing surgery, screening and chemoprevention practices of BRCA1 and BRCA2 mutation carriers: a prospective cohort study
.
Clin Genet
2006
;
70
:
198
206
.
10
Metcalfe
KA
,
Birenbaum-Carmeli
D
,
Lubinski
J
, et al
. 
International variation in rates of uptake of preventive options in BRCA1 and BRCA2 mutation carriers
.
Int J Cancer
2008
;
122
:
2017
22
.
11
Wellings
SR
. 
A hypothesis of the origin of human breast cancer from the terminal ductal lobular unit
.
Pathol Res Pract
1980
;
166
:
515
35
.
12
Bean
GR
,
Bryson
AD
,
Pilie
PG
, et al
. 
Morphologically normal-appearing mammary epithelial cells obtained from high-risk women exhibit methylation silencing of INK4a/ARF
.
Clin Cancer Res
2007
;
13
:
6834
41
.
13
Fabian
CJ
,
Kimler
BF
,
Mayo
MS
,
Khan
SA
. 
Breast-tissue sampling for risk assessment and prevention
.
Endocr Relat Cancer
2005
;
12
:
185
213
.
14
Widschwendter
M
,
Jones
PA
. 
DNA methylation and breast carcinogenesis
.
Oncogene
2002
;
21
:
5462
82
.
15
Fackler
MJ
,
Evron
E
,
Khan
SA
,
Sukumar
S
. 
Novel agents for chemoprevention, screening methods, and sampling issues
.
J Mammary Gland Biol Neoplasia
2003
;
8
:
75
89
.
16
Segura-Pacheco
B
,
Trejo-Becerril
C
,
Perez-Cardenas
E
, et al
. 
Reactivation of tumor suppressor genes by the cardiovascular drugs hydralazine and procainamide and their potential use in cancer therapy
.
Clin Cancer Res
2003
;
9
:
1596
603
.
17
Esteller
M
,
Fraga
MF
,
Guo
M
, et al
. 
DNA methylation patterns in hereditary human cancers mimic sporadic tumorigenesis
.
Hum Mol Genet
2001
;
10
:
3001
7
.
18
Fackler
MJ
,
McVeigh
M
,
Evron
E
, et al
. 
DNA methylation of RASSF1A, HIN-1, RAR-β, cyclin D2 and Twist in in situ and invasive lobular breast carcinoma
.
Int J Cancer
2003
;
107
:
970
5
.
19
Holst
CR
,
Nuovo
GJ
,
Esteller
M
, et al
. 
Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia
.
Cancer Res
2003
;
63
:
1596
601
.
20
Crawford
YG
,
Gauthier
ML
,
Joubel
A
, et al
. 
Histologically normal human mammary epithelia with silenced p16(INK4a) overexpress COX-2, promoting a premalignant program
.
Cancer Cell
2004
;
5
:
263
73
.
21
Mitchell
G
,
Antill
YC
,
Murray
W
, et al
. 
Nipple aspiration and ductal lavage in women with a germline BRCA1 or BRCA2 mutation
.
Breast Cancer Res
2005
;
7
:
R1122
31
.
22
Antill
YC
,
Mitchell
G
,
Johnson
SA
, et al
. 
Loss of heterozygosity analysis in ductal lavage samples from BRCA1 and BRCA2 carriers: a cautionary tale
.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
1396
8
.
23
Campbell
IG
,
Russell
SE
,
Choong
DY
, et al
. 
Mutation of the PIK3CA gene in ovarian and breast cancer
.
Cancer Res
2004
;
64
:
7678
81
.
24
Esteller
M
,
Corn
PG
,
Baylin
SB
,
Herman
JG
. 
A gene hypermethylation profile of human cancer
.
Cancer Res
2001
;
61
:
3225
9
.
25
Lehmann
U
,
Langer
F
,
Feist
H
,
Glockner
S
,
Hasemeier
B
,
Kreipe
H
. 
Quantitative assessment of promoter hypermethylation during breast cancer development
.
Am J Pathol
2002
;
160
:
605
12
.
26
Lukas
J
,
Parry
D
,
Aagaard
L
, et al
. 
Retinoblastoma-protein-dependent cell-cycle inhibition by the tumour suppressor p16
.
Nature
1995
;
375
:
503
6
.
27
Di Vinci
A
,
Perdelli
L
,
Banelli
B
, et al
. 
p16(INK4a) promoter methylation and protein expression in breast fibroadenoma and carcinoma
.
Int J Cancer
2005
;
114
:
414
21
.
28
Hedenfalk
I
,
Duggan
D
,
Chen
Y
, et al
. 
Gene-expression profiles in hereditary breast cancer
.
N Engl J Med
2001
;
344
:
539
48
.
29
Jazaeri
AA
,
Yee
CJ
,
Sotiriou
C
,
Brantley
KR
,
Boyd
J
,
Liu
ET
. 
Gene expression profiles of BRCA1-linked, BRCA2-linked, and sporadic ovarian cancers
.
J Natl Cancer Inst
2002
;
94
:
990
1000
.
30
Parrella
P
,
Poeta
ML
,
Gallo
AP
, et al
. 
Nonrandom distribution of aberrant promoter methylation of cancer-related genes in sporadic breast tumors
.
Clin Cancer Res
2004
;
10
:
5349
54
.
31
Krop
I
,
Maguire
P
,
Lahti-Domenici
J
, et al
. 
Lack of HIN-1 methylation in BRCA1-linked and “BRCA1-like” breast tumors
.
Cancer Res
2003
;
63
:
2024
7
.
32
Dammann
R
,
Li
C
,
Yoon
JH
,
Chin
PL
,
Bates
S
,
Pfeifer
GP
. 
Epigenetic inactivation of a RAS association domain family protein from the lung tumour suppressor locus 3p21.3
.
Nat Genet
2000
;
25
:
315
9
.
33
Yeo
W
,
Wong
WL
,
Wong
N
,
Law
BK
,
Tse
GM
,
Zhong
S
. 
High frequency of promoter hypermethylation of RASSF1A in tumorous and non-tumourous tissue of breast cancer
.
Pathology
2005
;
37
:
125
30
.
34
Fackler
MJ
,
McVeigh
M
,
Mehrotra
J
, et al
. 
Quantitative multiplex methylation-specific PCR assay for the detection of promoter hypermethylation in multiple genes in breast cancer
.
Cancer Res
2004
;
64
:
4442
52
.
35
Agathanggelou
A
,
Cooper
WN
,
Latif
F
. 
Role of the Ras-association domain family 1 tumor suppressor gene in human cancers
.
Cancer Res
2005
;
65
:
3497
508
.
36
Gao
B
,
Xie
XJ
,
Huang
C
, et al
. 
RASSF1A polymorphism A133S is associated with early onset breast cancer in BRCA1/2 mutation carriers
.
Cancer Res
2008
;
68
:
22
5
.
37
Pu
RT
,
Laitala
LE
,
Alli
PM
,
Fackler
MJ
,
Sukumar
S
,
Clark
DP
. 
Methylation profiling of benign and malignant breast lesions and its application to cytopathology
.
Mod Pathol
2003
;
16
:
1095
101
.
38
Krassenstein
R
,
Sauter
E
,
Dulaimi
E
, et al
. 
Detection of breast cancer in nipple aspirate fluid by CpG island hypermethylation
.
Clin Cancer Res
2004
;
10
:
28
32
.
39
Locke
I
,
Kote-Jarai
Z
,
Fackler
MJ
, et al
. 
Gene promoter hypermethylation in ductal lavage fluid from healthy BRCA gene mutation carriers and mutation-negative controls
.
Breast Cancer Res
2007
;
9
:
R20
.
40
Evron
E
,
Dooley
WC
,
Umbricht
CB
, et al
. 
Detection of breast cancer cells in ductal lavage fluid by methylation-specific PCR
.
Lancet
2001
;
357
:
1335
6
.
41
Fackler
MJ
,
Malone
K
,
Zhang
Z
, et al
. 
Quantitative multiplex methylation-specific PCR analysis doubles detection of tumor cells in breast ductal fluid
.
Clin Cancer Res
2006
;
12
:
3306
10
.
42
Fabian
CJ
,
Kimler
BF
,
Zalles
CM
, et al
. 
Short-term breast cancer prediction by random periareolar fine-needle aspiration cytology and the Gail risk model
.
J Natl Cancer Inst
2000
;
92
:
1217
27
.
43
Dulaimi
E
,
Hillinck
J
,
Ibanez de Caceres
I
,
Al-Saleem
T
,
Cairns
P
. 
Tumor suppressor gene promoter hypermethylation in serum of breast cancer patients
.
Clin Cancer Res
2004
;
10
:
6189
93
.
44
Muller
HM
,
Fiegl
H
,
Widschwendter
A
,
Widschwendter
M
. 
Prognostic DNA methylation marker in serum of cancer patients
.
Ann N Y Acad Sci
2004
;
1022
:
44
9
.
45
Fiegl
H
,
Millinger
S
,
Mueller-Holzner
E
, et al
. 
Circulating tumor-specific DNA: a marker for monitoring efficacy of adjuvant therapy in cancer patients
.
Cancer Res
2005
;
65
:
1141
5
.
46
Ulivi
P
,
Zoli
W
,
Calistri
D
, et al
. 
p16INK4A and CDH13 hypermethylation in tumor and serum of non-small cell lung cancer patients
.
J Cell Physiol
2006
;
206
:
611
5
.
47
Wang
J
,
Qin
Y
,
Li
B
,
Sun
Z
,
Yang
B
. 
Detection of aberrant promoter methylation of GSTP1 in the tumor and serum of Chinese human primary hepatocellular carcinoma patients
.
Clin Biochem
2006
;
39
:
344
8
.
48
Mori
T
,
O'Day
SJ
,
Umetani
N
, et al
. 
Predictive utility of circulating methylated DNA in serum of melanoma patients receiving biochemotherapy
.
J Clin Oncol
2005
;
23
:
9351
8
.

Supplementary data