Introduction
The multistep model of breast cancer progression suggests a transition from normal epithelium to invasive carcinoma via intraductal hyperplasia and
in situ carcinoma [
1]. These presumptive precursor lesions are currently defined by their histological features. Ductal carcinoma
in situ (DCIS) is a pre-invasive lesion with diverse histological morphologies and molecular alterations [
2]. The risk of DCIS progressing to invasive carcinoma is not well ascertained and robust biomarkers capable of stratifying the most aggressive from the more benign forms of the disease are currently lacking.
Cancer progression is due to the accumulation of genomic alterations leading to oncogene overexpression and tumour suppressor loss inducing growth advantage and clonal expansion. The transition of DCIS to invasive ductal cancer (IDC) is a poorly understood key event in breast tumour progression. Copy number alterations are already present in DCIS but their frequency tends to increase in IDCs [
3]. Such genomic aberrations lead to altered gene expression, and comprehensive gene expression studies comparing DCIS and IDCs have identified stage-specific markers ([
4‐
6] and Muggerud
et al., submitted) along with a gene expression classifier which differed between DCIS and invasive breast cancer [
7]. On the other hand, the frequency of
TP53 mutations in DCIS is similar to what is observed in invasive tumours and
in situ and invasive components from the same tumour exhibit the same mutations, indicating the same cellular origin of the two components [
8‐
10].
Epigenetic changes are considered to be an early event during tumour development and one of the hallmarks of cancer [
11]. Hypermethylation of CpG islands represents an alternative mechanism to inactivate tumour suppressor genes and is a prevalent early molecular marker for cancer. Specific patterns of CpG island methylation could result from clonal selection of cells having growth advantages due to silencing of associated tumour suppressor genes, DNA repair genes, cell-cycle regulators and transcription factors. Previous candidate gene studies investigated promoter hypermethylation of
in situ lesions and identified aberrant methylation at the promoters of
GSTP1,
CyclinD2,
RARB2,
Twist,
RASSF1A, HIN-1, CDKN2A, 14-3-3σ and
APC1 [
12‐
17]. However, only
GSTP1 promoter hypermethylation was reported to progress in frequency during breast carcinogenesis [
12].
Identification of early epigenetic changes in DCIS lesions might give valuable markers for early detection and may contribute to the understanding of how these changes affect the progression of the disease. The aim of this study was to identify informative progression markers by methylation analyses of eleven genes known to be methylated in breast tumours or breast cancer cell lines;
ABCB1 [
18],
CDKN2A/p16
INK4a
[
19],
ESR1 [
20],
GSTP1 [
21],
IGF2 [
22],
MGMT [
19],
MLH1 [
19],
PPP2R2B [
23],
PTEN [
24],
RASSF1A [
25] or displaying variation in breast cancer sub-type gene expression profiles;
FOXC1.
In a series of 27 DCIS, 28 IDCs, 34 mixed cases (invasive tumours with in situ components) and 28 normal tissues we show that methylation of CpG islands is already detectable in DCIS with the same frequency as within IDCs.
Discussion
In the present study we quantitatively determined the methylation levels in the promoter regions of 11 cancer-related genes in DCIS, small invasive breast cancers, mixed lesions and normal breast tissues. Aberrant DNA methylation was already present in DCIS for several of the genes studied. No DNA methylation changes specific for invasive breast cancer were identified.
Previous candidate gene studies have investigated promoter hypermethylation of
in situ lesions and have shown methylation for
GSTP1,
CyclinD2,
RARB2,
Twist,
RASSF1A, HIN-1, CDKN2A, 14-3-3σ [
12‐
14,
16]. We found frequencies of methylation for
RASSF1A in both DCIS and invasive tumours similar to previously published reports ranging from 60 to 88% in different populations [
14,
15]. Results seem thus to be very consistent over different technology platforms identifying the epigenetic inactivation of
RASSF1A as a very early step during breast carcinogenesis.
GSTP1 has been found to be frequently methylated in different stages of breast carcinomas. Again, similar frequencies of 20 to 30% were found here for
GSTP1 compared with recent reports [
21,
33‐
35]. In this study, methylation of
MGMT was rare in both DCIS and invasive tumours, also in concordance with previous studies [
36]. Infrequent methylation was also observed for the
MLH1 gene, in line with the absence of changes in
MLH1 expression during early breast carcinogenesis as assessed by IHC [
37,
38]. We also found minimal methylation of
CDKN2A within the CpGs studied here, which is in concordance with a previous study on DCIS and other proliferative lesions of the breast [
16]. No methylation of
ESR1 was observed in this study, although we have used the commonly studied region 400 bp downstream of
ESR1 transcription start site. A previous study observed methylation in DCIS samples [
15], however they could not show significant differences between normal and diseased tissue in North-American and Korean populations. A study by Feng
et al. (2007) shows the same result as we report with no increase in
ESR1 methylation in malignant compared to normal breast tissues [
39]. However, we can not exclude the possibility that the reason why we are not able to detect any
ESR1 methylation is that pyrosequencing is less sensitive compared to the Q-MSP technology used in [
15].
A novel finding of our study was the identification of aberrant DNA methylation of
ABCB1, FOXC1, PPP2R2B and
PTEN in DCIS. The methylation frequencies were similar in all diagnosis groups for
ABCB1, PPP2R2B and
PTEN. PTEN and
PPP2R2B are both candidate tumour suppressor genes [
23,
40] and our results suggest that epigenetic silencing might be involved in dysregulation of these genes in DCIS. We have observed the methylation of
PPP2R2B in locally advanced breast tumours (Dejeux
et al, submitted for publication), and
PTEN has been found to be frequently methylated in breast carcinomas [
24,
41].
ABCB1 is an ATP dependent p-glycoprotein involved in the efflux of various small molecules and xenobiotics in extra and intracellular membranes. Association of DNA methylation of
ABCB1 and drug resistance in breast cancer cell-lines has been reported [
42], and
ABCB1 expression has been associated with poor outcome in breast cancer patients [
43]. We have observed that
ABCB1 methylation is important for treatment response and overall survival in patients with advanced breast cancer treated with doxorubicin (Dejeux E,
et al. submitted). In this study,
ABCB1 methylation was associated with non-proliferative, Ki67 negative tumours supporting a positive role for
ABCB1 methylation in breast cancer progression and outcome.
FOXC1 is a transcription factor with an important role in the regulation of ocular development [
44].
FOXC1 is hypomethylated and highly expressed in CD44+ breast progenitor cells and might play an important role in the differentiation of mammary epithelial cell phenotypes [
45]. In our data set,
FOXC1 displayed a significantly increased methylation levels from normal breast tissue to invasive tumours with simultaneously lower
FOXC1 gene expression as measured by qRT-PCR. The tumours with less methylation and somewhat higher expression of
FOXC1 (Figure
3B) tended to be of the basal-like and normal-like breast cancer subtypes, as determined by gene expression profiling. Luminal B-like and ERBB2-like tumours had significantly lower
FOXC1 expression (
P = 0.026 and
P = 0.018, respectively) compared to the basal-like tumours whereas no statistical significance was found for basal vs. luminal A-like tumours (
P = 0.134) (Muggerud A
et al., unpublished results). This supports the view of heterogeneous
FOXC1 expression across molecular subtypes, which is in concordance with previously reported results [
45]. Bloushtain-Qimron
et al reported also increased methylation in matched distant metastases compared to the primary tumours, supporting a differential role for
FOXC1-methylated cells in the progression of the disease. The 28 normal breast tissue samples analysed in this study displayed on average significantly higher levels of
FOXC1 gene expression compared to the DCIS, small invasive and mixed lesions. Both the methylated and unmethylated DCISs and invasive tumours displayed significantly lower levels of
FOXC1 gene expression. This might indicate that histone modifications or other mechanisms in addition to promoter hypermethylation silence
FOXC1 in the unmethylated tumours. Further studies are needed to investigate the functional consequence of this increase of DNA methylation and the potential role of
FOXC1 in the progression from DCIS to invasive carcinoma.
Concomitant DNA methylation was observed between some of the genes studied suggesting connected epigenetic programs within tumours. In line with our result, significant correlation between
GSTP1 and
RASSF1A hypermethylation has previously been reported [
46]. The chromosomal region 6p25 harbouring the
FOXC1 gene is frequently gained in ER negative tumours [
47] in line with our observation of low DNA methylation and high expression in basal and normal-like tumours. The other chromosomal regions harbouring the genes with a high correlation to
FOXC1 methylation have all been reported to either have gain or losses in breast cancer [
48‐
50]. It is possible that concomitant DNA hypomethylation in these regions could induce chromosomal instability of the same regions as reported in colon cancer [
51].
By unsupervised hierarchical clustering analysis of a number of methylation markers and tumours from 148 breast cancer patients, Widschwendter
et al. (2004) [
52], showed that the tumours segregated naturally into groups with distinct methylation profiles that differed significantly in their hormone receptor status. Further, other studies have focused on the epigenetic differences between ER positive and ER negative breast cancers and their results imply that methylation profiles of ER-positive tumours are different from those of ER-negative tumours [
39,
46]. Moreover they found that promoter hypermethylation of
RASSF1A and
GSTP1 was more frequent in ER-positive than in ER-negative tumours in both early and advanced breast tumours. Our data are consistent with these previous reports suggesting that ER (or hormone receptor) expression may influence epigenetic changes.
Lower levels of DNA methylation were observed in
TP53 mutated tumours, especially at
FOXC1,
ABCB1,
PPP2R2B and
PTEN promoters. The association of
ABCB1 and
PPP2R2B with
TP53 status is also observed in more advanced tumours (Dejeux
et al. submitted). It has also been reported that breast tumours with
TP53 mutations lacked methylation in a number of regulatory genes [
39]. Further in concordance with our results, Toyota
et al. [
53] found a high number of
TP53 mutations in unmethylated colorectal tumours suggesting that
TP53 mutations and epigenetic alterations of other growth-suppressing genes can be two distinct mechanisms that inactivate tumour-suppressor genes in breast cancer. Similar to the results of [
52] we could not find associations between histopathological grade and DNA methylation patterns among the samples and genes investigated. This might be due to independency of grade for this limited gene panel investigated or the relatively small cohort size.
Almost all genes methylated in DCIS and IDC have been identified using candidate gene approaches. Genome-wide methylation studies might provide useful in identifying new DNA methylation events occurring during early breast tumourigenesis. Since this study was designed to find differences related to tumour progression from in situ to invasive breast cancer, follow-up studies are needed to investigate these biological markers and their potential in predicting prognosis.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AAM performed laboratory experiments, data analyses and wrote the manuscript. JAR performed laboratory experiments, data analyses and wrote the manuscript. FW was responsible for the patient cohorts, involved in the study design and wrote the manuscript. JB was responsible for the fresh frozen sample cohorts. FB and JJ were involved in laboratory experiments. HS was involved in the statistical analyses. IB provided normal breast tissue samples. ALBD, VNK, TS and JT initiated and designed the study and were involved in writing the manuscript. All authors have read and approved the final manuscript.