Introduction
Cancer can be considered as cumulative phenotypic consequence of acquired genetic and epigenetic alterations in cells [
1]. Epigenetic alterations, in particularly changes in the DNA methylation pattern, are known to play a crucial role in carcinogenesis. Aberrant DNA methylation occurs early in carcinogenesis, suggesting that DNA methylation alterations may precede classical transforming events such as gene mutations. Changes in the DNA methylation status occur more frequently than mutations or cytogenic abnormalities [
2].
In humans and other mammals, DNA methylation takes place at carbon-5 of cytosine residues within cytosine-phosphatidyl-guanosine (CpG) dinucleotides. CpG dinucleotides are heterogeneously distributed in the human genome, often clustered in so-called CpG islands. CpG islands are particularly present in promoter regions and first exons of genes that regulate important cell functions [
3].
In normal cells, CpG islands are generally unmethylated, resulting in gene expression, if the corresponding transcription factors are available [
2]. In cancer cells, however, the promoter region of certain genes is frequently hypermethylated, leading to a tightly packed chromatin and transcriptional gene silencing. Promoter hypermethylation commonly affects regulator genes that are involved in a wide range of cellular pathways, such as cell cycle, DNA repair, toxic catabolism, cell adherence, apoptosis and angiogenesis [
4].
Several studies have already investigated the applicability of promoter hypermethylation as specific and sensitive biomarkers, e.g., for the detection and diagnosis of cancer at an early stage [
5,
6], the prognosis of cancer [
7,
8] or the prediction of the response to a certain treatment scheme [
9,
10]. In contrast to genetic alterations, changes in the DNA methylation status are potentially reversible. Reactivation of epigenetically silenced genes by using DNA demethylating drugs is therefore regarded as a promising strategy in cancer therapy [
11,
12].
Recent studies have shown that molecular abnormalities occur not only in the tumor tissue but also in tissue that is adjacent to the tumor and appears histologically normal. The presence of such abnormalities in tissues surrounding tumors is called field cancerization or field defect [
13]. In addition to genetic abnormalities, e.g., chromosomal anomalies and loss of heterozygosity, epigenetic alterations, in particular changes in the DNA methylation status, have been found in normal-appearing tissues close to tumors. Molecular signatures of field cancerization have been reported for various epithelial tumors including those of the colon [
14‐
16] and the prostate [
17,
18].
Field cancerization is of clinical relevance because it is assumed to be an important factor in local recurrence of cancer [
19]. As the field showing aberrant DNA methylation may not be removed completely by surgery, these changes in the DNA methylation status might lead to neoplasms and subsequent transformation to a tumor. So far, only a limited number of studies have investigated changes in the DNA methylation status in histologically normal tissue adjacent to breast tumor tissues. Yan et al
. [
20] detected methylation changes in the promoter of Ras association domain family member 1 (
RASSF1A) in mammary tissue as far as 4 cm from the primary tumor. In a study of Feng et al. [
21] the DNA methylation status of reversion-induced LIM protein (
RIL), high in normal-1 (
HIN-1),
RASSF1A and cadherin-13 (
CDH13) in normal-appearing tissue (located at least 3 cm away from the tumor) was found to correlate with that in the breast tumor.
The main aim of the present study was to extend research on the applicability of the methylation status of candidate genes as biomarkers for field cancerization. We selected a panel of six tumor suppressor genes that have previously been reported to be frequently methylated in breast tumors, comprising cyclin D2 (
CCND2), death-associated protein kinase 1 (
DAPK1), glutathione S-transferase P1 (
GSTP1),
HIN-1, O6-methylguanine-DNA methyltransferase (
MGMT) and
RASSF1A [
22]. From each of 17 breast cancer patients, three biopsy samples were taken: the first one from the tumor tissue, the second one from tumor-adjacent tissue and the third one from tumor-distant tissue. In addition, we tested if the DNA methylation status of the six tumor suppressor genes in tumor, tumor-adjacent and/or tumor-distant tissues is associated with any clinicopathological parameters. We were also interested to see if there is a correlation between the DNA methylation status of the genes in the tissues of one and the same breast cancer patient.
Discussion
We determined the methylation status of a panel of six tumor suppressor gene promoters in tumor, tumor-adjacent and tumor-distant tissues of 17 breast cancer patients. CCND2, DAPK1, GSTP1, HIN-1, MGMT and RASSF1A were selected because they have previously been reported to be frequently methylated in primary breast cancers. Tumor-adjacent and tumor-distant tissues appeared histologically normal and were located about 1 cm and about 3 cm from the tumor, respectively. Normal breast tissues of women that had been undergoing reduction mammoplasty served as control.
The DNA methylation status was determined by MS-HRM analysis, which consists of the following steps: treatment of the DNA with sodium bisulfite in order to convert unmethylated cytosines into uraciles (methylated cytosines remain unchanged); amplification of the bisulfite-treated DNA by PCR (uracils are replicated as thymines and methylated cytosines as cytosines); brief denaturation and rapid reannealing of the PCR products; and finally high-resolution melting by gradually increasing the temperature. Guanine–cytosine-rich sequences (resulting from methylated CpG dinucleotides) melt at higher temperature than adenine–thymine-rich sequences (resulting from unmethylated CpG dinucleotides). Melting curves are obtained by plotting the fluorescence intensity against the temperature. The methylation status of unknown samples can be determined with the help of a calibration curve established by analyzing mixtures of unmethylated and fully methylated control DNA.
In our study MS-HRM analysis was the method of choice because we aimed at determining very low methylation levels and discriminating between small differences in the methylation status of the tumor suppressor gene promoters. MS-HRM methods allow adjusting the accessible methylation range by generating a (helpful) bias toward methylated or unmethylated alleles in PCR amplification [
29]. High sensitivity can be achieved by using primers containing a low number of CpG dinucleotides, favoring amplification of guanine–cytosine-rich sequences (resulting from methylated CpG dinucleotides) [
23]. In addition, the T
a and the MgCl
2 concentration can be varied in order to tailor the methylation range accessible. LODs (S/N = 3) and LOQs (S/N = 10) of the MS-HRM methods applied were determined by repeatedly analyzing bisulfite-treated, unmethylated control DNA. The methods proved to be very sensitive, with LODs and LOQs ranging from 0.1 to 1.5 % and 0.3 to 5.3 %, respectively.
In contrast to several other methods used in DNA methylation analysis, e.g., bisulfite pyrosequencing, MS-HRM methods do not allow determining the methylation status of individual CpG dinucleotides. From the melting profiles and the corresponding derivative plots one can, however, assess if the original template was methylated homogeneously or heterogeneously. In the reannealing step, carried out after briefly denaturing the PCR products, DNA sequences originating from homogenously methylated templates form homoduplexes, whereas those originating from partially methylated templates form heteroduplexes. Due to base-pairing mismatches, heteroduplexes begin to melt at lower temperature and lead to more complex melting profiles than homoduplexes. In biopsy samples analyzed in the present study, promoters of
CCND2,
GSTP1,
HIN-1 and
RASSF1A were found to be methylated homogenously (in case they were methylated at all). Most of the tumors showed homogenous methylation of
DAPK1 and
MGMT, in some tumor tissues the promoter of these genes was methylated heterogeneously. We did not find an association between the occurrence of heterogeneous methylation in
DAPK1 and
MGMT promoters and any of the clinicopathological parameters of the patients. Heterogeneous methylation of
DAPK1 promoter has been shown previously in patients with chronic lymphocytic leukemia [
30] or diffuse large B-cell lymphoma [
31]. The
MGMT promoter has been found to be methylated heterogeneously in patients with breast cancer [
24], in diffuse large B-cell lymphoma [
31] and in human breast cancer cell line HS578T [
32].
In MS-HRM analysis, the DNA methylation status is determined with the help of calibration curves established by analyzing mixtures of unmethylated and methylated human control DNA. In case of heterogeneous methylation, melting curves differ in shape from melting curves obtained for standards. In order to obtain accurate results in spite of differences in the shape of the melting curves, we established interpolation calibration curves by taking into account the average of the normalized fluorescence signal for each standard over the entire temperature.
None of the normal breast tissues of the control group showed methylation of DAPK1, HIN-1 and RASSF1A. In one tissue, the methylation status of GSTP1 was < LOQ. In the breast tissue of the oldest woman (age: 60 years), the promoters of CCND2 and MGMT were found to be methylated (CCND2 < LOQ, MGMT: 5.2 %). However, for none of the tumor suppressor genes did we find an association between the age of the women and the DNA methylation status.
In breast cancer patients, 94 % and 82 % of the tumor tissues showed methylation of
RASSF1A and
HIN-1 promoters, respectively. In 82 % of the tumors, even both promoters were methylated. Our results are in accordance with previous studies reporting frequent methylation of
RASSF1A and/or
HIN-1 in breast carcinoma.
RASSF1A methylation was found in 85 % [
33‐
35], 68 % [
36], 65 % [
37], 59 % [
38], 58 % [
21] and 33 % [
24] of the breast tumors analyzed,
HIN-1 in 74 % [
39]), 73 % [
33] and 49 % [
21] of the cases. In the present study, promoters of
MGMT,
DAPK1 and
GSTP1 were methylated in 65 %, 63 % and 53 % of the tumors. However, in case of
GSTP1, all tumors showed methylation status < LOQ. Frequency of methylation of these promoters was higher than reported in the literature (
MGMT: 22 % [
24];
DAPK1: 50 % [
37] and 37.5 % [
24];
GSTP1: 16.6 % [
24] and 14 % [
40]), most probably due to the high sensitivity (LOD < 1 %) of the MS-HRM methods applied in the present study. In contrast, the
CCND2 promoter was found to be less frequently methylated (in 35 % of the tumors) than in the study of Lewis et al
. (57 % [
38]).
In each tumor tissue, the promoter of at least one of the six genes was found to be methylated. A total of 71 % of the tumors showed promoter methylation of ≥ 4 genes. This finding is consistent with previous studies reporting methylation of more than one tumor suppressor gene in breast tumors [
24,
37]. Simultaneous methylation of several tumor suppressor genes indicates the important role DNA methylation is playing in breast cancer development.
In tumors, the DNA methylation status of
DAPK1 (
p = 0.012
),
HIN-1 (
p = 0.005) and
RASSF1A (
p < 0.001) was statistically significantly different from that in normal breast tissues of the control group. In addition, tumor tissues showed a higher methylation status of
MGMT (Fig.
4) compared to the control group, however, the difference was not statistically significant since one normal breast tissue had a methylation status of 5.1 % (other normal breast tissues of the control group had a methylation status < LOD). In the present study,
MGMT promoter methylation was detected in the oldest woman of the control group. In a previous study,
MGMT methylation has been associated with the age of breast cancer patients [
24]. However, in our study we did not find a significant correlation between the DNA methylation status of
MGMT and the age of the women, neither for breast cancer patients nor for the control group.
With the exception of
CCND2 and
DAPK1, the genes were not only frequently methylated in the tumors but also in the corresponding tumor-adjacent and tumor-distant tissues of the breast cancer patients. The
RASSF1A promoter was as frequently and the
HIN-1,
MGMT and
GSTP1 promoters were almost as frequently methylated as in tumors. With the exception of
GSTP1, the methylation status was significantly lower in tumor-distant tissues than in tumors. In case of
RASSF1A,
HIN-1 and
MGMT promoters, tumor-adjacent tissues showed higher methylation status than tumor-distant tissues, the difference was, however, not statistically significant. These results demonstrate that the methylation status of
RASSF1A,
HIN-1 and
MGMT promoters indicates field cancerization in breast cancers. To the best of our knowledge, field cancerization due to
MGMT promoter methylation has not been reported so far.
MGMT field cancerization has, however, been detected in colorectal cancers [
14,
41] and oral squamous cell carcinomas [
42]. Aberrant methylation in tissue adjacent to breast tumor has already been reported previously for
RASSF1A [
20,
21] and
HIN-1 [
21]. Methylation of the
HIN-1 promoter has been detected with high frequency (95 %) in preinvasive lesions such as ductal and lobular carcinoma in situ [
39]
, indicating that it is an early event in breast tumorigenesis. In the present study, the methylation status of the
HIN-1 promoter in tumor-adjacent tissues was found to correlate strongly with that in the corresponding tumors, but not with that in the corresponding tumor-distant tissues. Among the promoters investigated in the present study, the methylation status of the
HIN-1 promoter can thus be considered the best suitable biomarker for detecting field cancerization. Further investigation is needed to test whether it can be used for defining surgical margins in order to prevent future recurrence of breast cancer.
Breast cancer is known to be a heterogeneous disease with regard to histopathological and molecular characteristics, outcome and response to treatment. Previous studies have reported association between the methylation status of a variety of gene promoters and clinicopathological parameters.
HIN-1 is a putative cytokine reported to be highly expressed in normal but not cancerous mammary epithelial cells [
39,
43]. The methylation status of
HIN-1 in breast tumors has been associated with the ER, PR and/or HER2 status [
21,
43]. In our study, in tumor-adjacent tissues of patients with HER2-positive tumors the
HIN-1 promoter was more frequently unmethylated (< LOD) than in those from patients with HER2-negative status. In addition, we found association between the methylation status of
HIN-1 in the tumor-distant tissue and the PR and ER status of the tumor. However, the number of PR-negative and ER-negative tumors investigated was too low to allow a statistical conclusion. The methylation status of the
HIN-1 promoter in tumors was found to correlate with the age of the patients, which is in accordance with the study of Feng et al
. [
21].
MGMT plays a role in DNA repair. Silencing of
MGMT by promoter methylation is a predictor of overall survival and response to alkylating agents [
44]. In our study, the methylation status of
MGMT in tumor-distant tissues was associated with tumor grading. In patients with tumor grade 3, the
MGMT promoter was found to be more frequently methylated than in patients with tumor grade 2. In the study of Tserga et al
. [
24], the methylation status of the
MGMT promoter in the tumor itself was found to be associated with advanced tumor grade.
Several studies report association between
RASSF1A promoter methylation and ER status of the tumor [
35,
45,
46]. In the present study only one of the tumors was ER-negative. We therefore could not evaluate potential association between the methylation status of the
RASSF1A promoter and the ER status of the tumors.
As already mentioned above, more than one tumor suppressor gene has been found to be methylated in a high percentage of the tumors. Correlation analyses revealed that the methylation status of
RASSF1A positively correlated with that of
HIN-1 and
MGMT. In addition, we found a statistically significant positive correlation between the promoter methylation of
DAPK1 und
MGMT. Correlation between methylation levels of
RASSF1A and
HIN-1 in tumor of breast cancer patients has been previously published by Feng et al
. [
21].
Authors’ contributions
MS performed experiments, analyzed data, prepared figures and tables and performed statistical analyses. EH performed experiments and analyzed data. GP and SH carried out tissue procurement and patient data collection. MCM supported conception and design and drafted the manuscript. All authors read and approved the final manuscript.