Breast Cancer Research and Treatment

, Volume 130, Issue 1, pp 109–117

Prognostic significance of methylated RASSF1A and PITX2 genes in blood- and bone marrow plasma of breast cancer patients

Authors

  • Georg Göbel
    • Department of Medical Statistics, Informatics and Health EconomicsInnsbruck Medical University
  • Doris Auer
    • Department of Obstetrics and GynecologyInnsbruck Medical University
  • Inge Gaugg
    • Department of Obstetrics and GynecologyInnsbruck Medical University
  • Alois Schneitter
    • Department of Obstetrics and GynecologyInnsbruck Medical University
  • Ralf Lesche
    • Department of Biomedical Research and Development and Technology DevelopmentEpigenomics AG
  • Elisabeth Müller-Holzner
    • Department of Obstetrics and GynecologyInnsbruck Medical University
  • Christian Marth
    • Department of Obstetrics and GynecologyInnsbruck Medical University
    • Department of Obstetrics and GynecologyInnsbruck Medical University
Preclinical study

DOI: 10.1007/s10549-010-1335-8

Cite this article as:
Göbel, G., Auer, D., Gaugg, I. et al. Breast Cancer Res Treat (2011) 130: 109. doi:10.1007/s10549-010-1335-8

Abstract

Free circulating DNA is increased in the serum/plasma of cancer patients, and methylation of certain genes has been found to be characteristic for malignancy. Therefore, we investigated the prognostic value of two promising genes, PITX2 and RASSF1A, in peripheral blood-plasma (PB-P) and bone marrow plasma (BM-P) of breast cancer patients. Peripheral blood and bone marrow samples from patients with primary breast cancer were prospectively collected during primary surgery at the Department of Obstetrics and Gynecology in Innsbruck (n = 428) from June 2000 to December 2006. The study has been approved by the ethical committee of the Medical University of Innsbruck. Methylation analysis was performed using MethyLight, a methylation-specific quantitative PCR-method. In univariate survival analysis, methylated PITX2 in PB-P was found to be a significant indicator for poor overall survival (OAS) and distant disease-free survival (DDFS) (P = 0.001 and P = 0.023). Methylated RASSF1A in PB-P was also an indicator for poor OAS and DDFS (P = 0.001 and P = 0.004). RASSF1A had also significant prognostic potential when determined in BM-P (P = 0.016). In multivariate survival analysis methylated PITX2 and RASSF1A in PB-P remained as therapy-independent prognostic factors for OAS (P = 0.021, P < 0.001). For DDFS only RASSF1A in PB-P showed prognostic significance (P = 0.002). Methylated RASSF1A and PITX2 in PB-P appear to have promising potential as prognostic markers in clinical use.

Keywords

RASSF1APITX2DNA-methylationBreast cancerPrognosis

Introduction

The phenomenon of elevated free circulating DNA levels in the serum/plasma of cancer patients was first shown 30 years ago [1]. A variety of DNA alterations have been reported within the circulating free DNA of cancer patients, including point mutations, microsatellite instability, and losses of heterozygosity [1].

In addition to the molecular genetic alterations described above, the silencing of genes by promoter hypermethylation is a common feature in human cancer [2]. The inhibition of gene expression by DNA methylation plays an important role in the regulation of DNA repair, cell cycle, apoptosis, cell–cell adhesion, metastasis, tissue and organ architecture, and various signaling pathways.

The tumor suppressor gene RASSF1A is one of the most frequently inactivated proteins—mostly by inappropriate promoter methylation—ever identified in human cancer. It lacks apparent enzymatic activity but contains a Ras association domain and is potentially an effector of the Ras oncoprotein. RASSF1A modulates multiple apoptotic and cell cycle checkpoint pathways. Current evidence supports the hypothesis that it serves as a scaffold for the assembly of multiple tumor suppressor complexes and may relay pro-apoptotic signaling [3]. Furthermore, we and others have shown that RASSF1A is frequently hypermethylated in breast cancer [46], and methylated RASSF1A has been demonstrated in patients with ductal carcinomas of the breast, both in cancer tissue and in paired serum DNA samples of these patients [7].

Moreover, RASSF1A has been reported to be present in nearly all breast cancer cell fractions, but is rare in the serum of patients with non-neoplastic breast conditions [8].

DNA-methylation of another gene, PITX2, a bicoid-related homebox transcription factor, has been shown to be an independent prognostic marker in patients with hormone receptor-positive lymph node-negative breast cancer, implying that tumors with a methylated PITX2 promoter are more aggressive [9]. Furthermore, DNA-methylation of PITX2 in tumor tissue has been reported to reliably predict the risk of distant disease recurrence in tamoxifen-treated, node-negative breast cancer patients [10]. However, the role of PITX2 in breast carcinogenesis and progression is unclear. PITX2 is regulated by WNT/DVL/beta-catenin and hedgehog TGFβ pathways [11]. It is essential for normal development of the pituitary gland, craniofacial region, eyes, heart, abdominal viscera, and limbs. It has also been suggested that it plays a role in regulating regionally specific terminal neuronal differentiation of the ventrolateral thalamus and midbrain [12]. Moreover, it is differentially expressed in pituitary adenomas [13]. Germ line mutations of PITX2 cause Axenfeld-Rieger Syndrome (ARS) [14], but no association with neoplastic disorders has been reported for ARS. Collectively, these data imply a growth or differentiation control function for PITX2, which could contribute to malignancy when out of equilibrium [10].

The mechanisms of free circulating DNA release in the blood are poorly understood. Generally, two models are used to explain this phenomenon. First, cells (tumor cells and surrounding normal cells) are thought to undergo apoptosis and/or necrosis in situ, and their DNA is released in the blood stream. Second, cells are thought to detach and extravasate into the blood stream, where they undergo lysis and release their DNA content. In these two models, macrophages may play an intermediary role in the release of free circulating DNA in the blood stream [1]. In line with this second model, the death of circulating tumor cells (CTCs) has been considered one of the possible sources for free circulating DNA in cancer patients [15, 16]. In metastatic breast cancer the presence of CTCs is a prognostic indicator, and a decrease in CTC levels during therapy may indicate clinical response [17, 18]. Koyanagi et al. [19] have shown a correlation between CTCs, detected by three mRNA markers, and serum tumor-related methylated DNA (RASSF1A) in peripheral blood of melanoma patients. Methylated RASSF1A DNA in the serum of breast cancer patients has been reported to be a surrogate marker for circulating breast cancer cells [8].

This study was designed to evaluate the prognostic significance disseminated tumor cells (DTCs) in bone marrow which are believed to be associated with poor prognosis in primary breast cancer patients [2, 2024] (results of this part of the study will be published elsewhere) and to investigate whether RASSF1A and PITX2 DNA-methylation in peripheral blood-plasma (PB-P) and in bone marrow plasma (BM-P) are surrogate markers of DTCs or have prognostic or predictive potential independent of DTCs.

Materials and methods

General study design

Peripheral blood (PB-P) and bone marrow (BM-P) samples from patients with primary breast cancer were prospectively collected during primary surgery at the Department of Obstetrics and Gynecology at our Medical University Hospital in Innsbruck. After DNA isolation and quantitative PCR analysis of the RASSF1-A and PITX2 DNA-methylation by using Methylight, we applied univariate and multivariate survival models to explore the potential of RASSF1A and PITX2 DNA-methylation in predicting distant disease-free (DDFS) and overall survival (OAS) in patients with primary breast cancer. All included patients gave written consent to the use of their PB-P and BM-P for research purposes. The study was approved by the local Institutional Ethics Review Board and it was performed in concordance with the Reporting Recommendations for Tumor Marker Prognostic Studies of the National Cancer Institute [25]. Clinical, pathological, and follow-up data were stored in a database in accordance with our hospital privacy rules.

Patients

All patients for this study were treated at the Department of Obstetrics and Gynaecology of the Innsbruck Medical University, Austria between June 2000 and December 2006 and staged according to the International Federation of Gynaecology and Obstetrics (FIGO) system. We included 428 female patients with primary breast cancer in our study. None of the patients was diagnosed to have a distant metastatic disease at primary surgery. All patients were monitored within the outpatient follow-up program of the Department of Obstetrics and Gynecology, Innsbruck Medical University, and the median observation period of the patients was 51 months (IQR 35–68). Median age of the patients at surgery was 57.4 years (IQR 50.0–66.3). Follow-up information was available for all patients. The clinicopathological characteristics of the patients included are summarized in Table 1.
Table 1

Associations of clinicopathologic characteristics with negative versus positive DNA-methylation in peripheral blood-plasma and bone marrow plasma (χ2-Statistic)

 

n

Peripheral blood-plasma (PB-P)

Bone marrow plasma (BM-P)

PITX2 (P-value)

RASSF1A (P-value)

PITX2 (P-value)

RASSF1A (P-value)

Age

 <57.4

214

0.763

<0.001

0.135

<0.001

 ≥57.4

214

Menopausal status

 Pre

131

0.870

0.009

0.012

<0.001

 Post

292

 Unknown

5

pT

 pT-0

9

0.511

0.237

0.784

0.921

 pT-1

318

 pT-2/pT-3

77

 Unknown

25

LN

 Negative

275

0.318

0.679

0.333

0.290

 Positive

141

 Unknown

12

Grade of malignancy

 1

76

0.098

0.108

0.054

0.899

 2

292

 3

55

 Unknown

5

Estrogen receptor

 Negative

66

0.662

0.586

0.025

0.612

 Positive

358

 Unknown

4

Progesterone receptor

 Negative

76

1.0

0.405

0.427

0.148

 Positive

348

 Unknown

4

Her-2

 Negative

277

0.726

0.773

0.504

0.055

 Positive

124

 Unknown

27

Total

428

 

neg negative, pos positive, pT pathologic tumor staging, LN lymph nodes

Types of primary surgical treatment were: mastectomy (23%), skin sparing mastectomy (13%), and breast-conserving therapy (64%). Neoadjuvant chemotherapy was applied in 13% of patients. Adjuvant therapies were chemotherapy (32%), endocrine therapy (85%), radiotherapy (79%), and anti-HER2 therapy (5%). 12% of the patients received only chemotherapy (34% anthracycline, 56% combined anthracycline/taxane, 10% others), 64% only endocrine therapy. 20% of the study population received chemotherapy and endocrine therapy (33% anthracycline, 53% combined anthracycline/taxane, 14% others). 55% of the patients, who were treated with an endocrine therapy, received tamoxifen (13% in combination with GnRH-Analoga), whereas for 23% of this groups aromatase inhibitors were used.

Specimen characteristics and laboratory work

For bone marrow sampling a bilateral BM aspiration was performed, as described in [20, 26]. 10 ml BM was taken out from each side of the iliac crest/pelvis crest, immediately transferred to a 50-ml Falcon tube (BD Biosciences, USA), mixed with 1 ml EDTA to avoid coagulation and pooled afterwards. Blood samples were collected in 10 ml EDTA containing collection tubes (S-Monovette, Sarstedt, Germany). Both, PB and BM samples were either worked up immediately or kept at 4°C (no longer than 4 h) before processing. Estrogen and progesterone receptor levels were determined in patient paraffin sections by immunohistochemistry with a DAKO Autostainer. The following primary antibodies were used: monoclonal Mouse Anti-Human Estrogen-Receptor α (DAKO, clone 1D5, dilution 1:100); and Monoclonal Mouse Anti-Human Progesterone-Receptor (DAKO, clone PgR 636, dilution 1:200). Chemmate DAKO EnvisionTM K5007 was used as the secondary antibody (DAKO, Glostrup, Denmark). Receptor status was recorded as a percentage of stained cells (0–100%) and staining intensity (1–3) for both ER and PR. Samples with a percentage >10% staining were considered receptor-positive. HER2/neu status was analyzed immunohistochemically using the HercepTest (DAKO, Glostrup, Denmark). Membrane immunoreactivity and membrane staining patterns were evaluated and scored using the 0 to 3+ scoring system according to the manufacturer’s protocol. Tumors with a HER-2/neu score of 2+ in immunohistochemical analysis were also tested for HER-2/neu gene amplification by fluorescence in situ hybridization (FISH) using the PathVysion kit (Vysis, Downers Grove, IL, USA). According to the HER-2/neu testing recommendations of the American Society of Clinical Oncology and the College of American Pathologists, tumors were classified as positive when the immunohistochemical score was 3+ or gene amplification was detected by means of FISH [27].

DNA isolation

Blood samples were centrifuged at 1000×g for 5 min at room temperature. Bone marrow samples were centrifuged at 500×g for 5 min; supernatant was taken and again centrifuged at 1000×g for 5 min at room temperature. 1 ml aliquots of PB-P and BM-P were stored at −50°C.

Genomic DNA from plasma and bone marrow supernatant was isolated using the High Pure Viral Nucleic Acid kit (Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s protocol with some modifications, as already published [8].

Analysis of DNA methylation

Sodium bisulfite conversion of genomic DNA was performed using the EZ DNA Methylation-Gold KitTM (Zymo Research, Orange, CA, USA) according to the manufacturer’s protocol. Sodium bisulfite-treated genomic DNA was analyzed by MethyLight, a fluorescence-based, quantitative real-time PCR assay, as described previously [6]. In brief, a set of primers and probe, designed specifically for bisulfite-converted DNA, was used for the genes of interest, RASSF1A and PITX2. Primer and probe sequences were as follows: RASSF1A forward primer, 5′-ATTGAGTTGCGGGAGTTGGT-3′; RASSF1A reverse primer, 5′-ACACGCTCCAAC CGAATACG-3′; RASSF1A probe, 5′-FAM-CCCTTCCCAACGCGCCCA-BHQ1-3′, according to Fiegl et al. [8]; PITX2 forward primer, 5′-AGTTCGGTTGCGCGGTT-3′; PITX2 reverse primer, 5′-TACTTCCCTCCCCTACCTCGTT-3′; and PITX2 probe, 5′-FAM-CGACGCTCGCCCGAACGCTA-BHQ1-3′, sequences for PITX2 primers and probe were provided by Epigenomics AG, Berlin, Germany. For each MethyLight reaction 10 μl of bisulfite-treated genomic DNA was used.

Interassay variance of the MethyLight reactions and the whole procedure (including DNA isolation and sodium bisulfite conversion) was determined using SssI treated (completely methylated) human white blood cell DNA (New England Biolabs, Ipswich, MA, USA) and control-plasma (CP) from a healthy volunteer that was supplemented with DNA from diverse breast cancer cell lines (MCF-7, T-47D, MDAMB231 and ZR75-1). Variance was defined as a 90% range of Ct-values for each gene at the median Ct-value. For comparison we calculated the corresponding variation coefficient. All of these values are shown in Table 2. Specificity and accuracy of MethyLight reactions was controlled using SssI completely methylated human white blood cell DNA as a positive control and distilled water as a negative control.
Table 2

Interassay variance of MethyLight reaction only (using treated SssI) and the entire method including DNA isolation, sodium bisulfite conversion and MethyLight reaction (using control plasma)

Control

Gene

Median Ct-value

90% Range of variation

Variation coefficient

SssI treated DNA

Methylated PITX2

35.51

0.85

18.5%

SssI treated DNA

Methylated RASSF1A

35.04

1.31

26.4%

Control plasma

Methylated PITX2

29.91

1.71

47.2%

Control plasma

Methylated RASSF1A

27.57

1.37

30.2%

SssI treated (completely methylated) human white blood cell DNA standard

Real time PCR for RASSF-1 was performed within 36 h after bisulfite modification. qPCR for PITX-2 was performed one half to one year after bisulfite modification, the samples being stored at −80°C. It could be proved with 10 different samples that the storage under above conditions had no influence at all on the result of qPCR (data not shown). The thermal cycling conditions for RASSF-1 and PITX-2 comprised an initial denaturing step at 95°C for 10 min and 50 cycles of 95°C for 15 s and 60°C for 1 min.

Patients’ samples were analyzed in triplicate. A patient sample was defined as positive when the minimum Ct-value of the triplicates was ≤40. As we were interested in the absolute amounts of methylated DNA, we did not normalize results to a reference gene. Results were only related to the deployed serum volume as this was kept constant.

Bone marrow preparation and ICC

The procedure for bone marrow preparation and ICC has been described previously [9]. In short, bone marrow aspirates were separated by Ficoll density-gradient centrifugation to enrich mononuclear cells. Mononuclear cells were removed from the interface, washed, and cytospin slides were prepared at 1 × 106 MNC/slide. They were air-dried overnight at room temperature, fixed in 3.7% formalin and either immunostained directly or stored at −80°C until immunostaining was performed. For immunohistochemical staining we used monoclonal antibody A45-B/B3 (Micromet, Munich, Germany), which is directed against a common epitope on cytokeratin polypeptides, including the cytokeratin heterodimers 8–18 and 8–19. The reaction of the primary antibody was developed with the alkaline phosphatase anti-alkaline phosphatase technique combined with the new fuchsine stain to indicate antibody binding, as previously described [9]. Evaluation of ICC was carried out by the Automated Cell Imaging System ACIS (ChromaVision GmbH, Karlsruhe, Germany) and verified by our pathologist. For each patient a total of 4 × 106 mononuclear cells were screened for the presence of DTCs. To identify cells we not only used immunocytochemical staining but also morphologic features according to Borgen et al. [28].

Statistical analysis

Associations between RASSF1A and PITX2 methylation status in PB-P and BM-P and clinicopathologic characteristics of the patients were assessed with Pearson’s chi-square test. DDFS was defined as the time from surgery to histopathological confirmation of distant metastases or death. OAS was defined as the time from surgery to death from any cause or to the last clinical inspection. Univariate survival analyses were conducted using the Kaplan–Meier method, and differences between groups were determined with the log-rank test. A Cox proportional hazards regression model with stepwise backwards variable-selection was applied for multivariate survival analyses. Variables with P < 0.05 in the univariate survival analysis were included into the multivariate model.

In sensitivity analysis we further investigated the effect of RASSF1A and PITX2 methylation status on the above endpoints, modeling the methylation status of RASSF1A and PITX2 as a continuous variable, using penalized splines (P-splines) in extended, restricted maximum-likelihood (REML-) optimal Cox-type additive hazard regression [29].

A P-value of <0.05 was considered statistically significant. Statistical analysis was performed using SPSS software (Version 15.0 for Windows).

Results

DNA methylation status of RASSF1A and PITX2 was assessed in PB-P and BM-P of primary breast cancer patients. The prevalence of methylated RASSF1A was 21.8% in PB-P and 20.6% in BM-P. For methylated PITX2 it was 13.9% in PB-P and 44.2% in BM-P. RASSF1A or PITX2 methylation was found in 30.7% of PB-P samples and 47.6% of BM-P samples. Unexpectedly, methylated PITX2 and RASSF1A were also observed in healthy controls (data not shown).

DNA methylation levels of RASSF1A and PITX2 were significantly associated in PB-P (P = 0.005) and BM-P (P < 0.001). Significant association was also observed between PB-P and BM-P for both RASSF1A (P < 0.001) and PITX2 (P = 0.009).

We compared the analyzed markers with clinical indicators for prognosis: stage, age, grade of malignancy, menopausal status, HER-2 expression, ER, PR, lymph node status, and the presence of DTCs. A significant association was observed between methylation status of RASSF1A (in PB-P and BM-P) and age (<median vs. >median) as well as menopausal status (P < 0.002 for all). PITX2 methylation was associated with ER and menopausal status in BM-P. Neither RASSF1A nor PITX2 in both PB-P and BM-P showed any association with the presence of DTCs.

In univariate Kaplan–Meier survival analyses, methylation of RASSF1A as well as PITX2 in PB-P was shown to be a prognostic factor for OAS and DDFS (P = 0.001 and 0.023 resp.). Combining the methylation status of both genes in PB-P lead to a higher proportion of positive results and also to a higher prognostic significance for OAS and DDSF (P < 0.001 for both). Kaplan–Meier curves for PITX2 and RASSF1A are shown in Fig. 1. All other information about univariate survival analysis is provided in Tables 3 and 4. Neither PITX2 nor RASSF1A showed prognostic relevance in patients who received only chemotherapy. In patients treated with adjuvant hormone therapy methylated RASSF1A in PB-P was shown to be a prognosticator for poor OAS and DDFS (P = 0.035 and 0.008). Furthermore methylated PITX2 and RASSF1A in PB-P were associated with significantly lower OAS and DDFS in patients who received combined chemo-hormone-therapy (DDFS: P = 0.036, P = 0.001; OAS: P = 0.002 for both PITX2 and RASSF1A). In this group also patients with methylated RASSF1A in BM-P showed lower OAS and DDFS (P = 0.017 and P = 0.002).
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-010-1335-8/MediaObjects/10549_2010_1335_Fig1_HTML.gif
Fig. 1

Univariate Kaplan–Meier survival analysis of patients with low/high RASSF1A and PITX methylation in peripheral blood-plasma (PB-P). P-values indicate results of the log-rank test

Table 3

Kaplan–Meier survival analysis-1 (log-rank test)

 

OAS

DDFS

P-value

5-year survival

5-year survival

P-value

5-year survival

5-year survival

Positive

Negative

Positive

Negative

Peripheral blood-plasma

 Methylated PITX2 (n = 359)

0.001

80.1%

(n = 50)

(ne = 10)

95.1%

(n = 309)

(ne = 15)

0.023

79.1%

(n = 50)

(ne = 10)

91.1%

(n = 309)

(ne = 25)

 Methylated RASSF1A (n = 357)

0.001

85.8%

(n = 78)

(ne = 13)

95.4%

(n = 279)

(ne = 12)

0.004

80.6%

(n = 78)

(ne = 15)

91.3%

(n = 279)

(ne = 21)

 Methylated PITX2 or RASSF1A (n = 355)

<0.001

85.6%

(n = 109)

(ne = 18)

96.6%

(n = 246)

(ne = 7)

0.001

82.3%

(n = 109)

(ne = 20)

92.4%

(n = 246)

(ne = 15)

Bone marrow plasma

 Methylated PITX2 (n = 406)

0.481

91.4%

(n = 179)

(ne = 16)

93.9%

(n = 227)

(ne = 12)

0.718

89.4%

(n = 179)

(ne = 19)

88.8%

(n = 227)

(ne = 21)

 Methylated RASSF1A (n = 403)

0.016

87.8%

(n = 83)

(ne = 11)

94.1%

(n = 320)

(ne = 17)

0.087

87.3%

(n = 83)

(ne = 13)

89.8%

(n = 320)

(ne = 27)

 Methylated PITX2 or RASSF1A (n = 403)

0.239

91.5%

(n = 211)

(ne = 19)

94.1%

(n = 192)

(ne = 9)

0.767

89.8%

(n = 211)

(ne = 22)

88.0%

(n = 192)

(ne = 18)

OAS overall survival, DDFS distant disease-free survival, n number of cases, ne number of events

Table 4

Univariate Kaplan–Meier survival analysis stratified by the type of adjuvant therapy (log-rank test)

 

Chemotherapy

n = 51

Hormone therapy

n = 275

Chemotherapy + hormone therapy

n = 83

DDFS

OS

DDFS

OS

DDFS

OS

Methylated PITX2 in PB-P

0.454

0.543

0.516

0.187

0.036

0.002

Methylated RASSF1A in PB-P

0.925

0.801

0.035

0.008

0.001

0.002

Methylated PITX2 in BM

0.872

0.633

0.458

0.446

0.778

0.770

Methylated RASSF1A in BM

0.490

0.532

0.939

0.257

0.017

0.002

PB-P peripheral blood-plasma, BM bone-marrow, OAS overall survival, DDFS distant disease-free survival

In a multivariate time-independent Cox regression adjusted for stage, age, grade of malignancy, menopausal status, Her-2 expression, ER, PR, and lymph node status, methylated PITX2 in PB-P and methylated RASSF1 in PB-P remained as independent prognostic factors for OAS whereas for DDFS only RASSF1A methylation in PB-P together with LN- and ER-status had significant impact. Further details are shown in Table 5.
Table 5

Therapy-independent significant prognostic factors for OAS and DDFS

Factor

Hazard ratio

95%CI

P-value

OAS

 Methylated PITX2 in PB-P (low vs. high)

3.4

1.2–9.8

0.021

 Methylated RASSF1A in PB-P (low vs. high)

5.6

2.1–14.5

<0.001

DDFS

 Methylated RASSF1A in PB-P (low vs. high)

3.4

1.6–7.3

0.002

 LN (neg. vs. pos)

2.5

1.1–5.8

0.036

 ER (neg. vs. pos)

0.3

0.1–0.9

0.039

Time-independent multivariate Cox Regression Model adjusted for clinicopathological factors and stratified by the type of therapy (backward selection of variables)

LN lymph-nodes, PB-P peripheral blood-plasma, OAS overall survival, DDFS distant disease-free survival, ER estrogen receptor, CI confidence interval

Analyzing the associations of PITX2 and RASSF1A methylation status with OAS and DDFS, using PITX2 and RASSF1A methylation status as a continuous covariate in a flexible, univariate, and multivariate non-parametric P-spline regression model widely confirmed the results of the main multivariate Cox model with PITX2 and RASSF1A as dichotomized variables.

Discussion

Numerous prognostic factors have been described for breast cancer. However, many of them lack practicability, as they are determined in fresh/frozen tissue, which is not available in low stage tumors. Prognostic parameters analyzed in a patient’s blood would therefore be much more convenient. Disseminated tumor cells were analyzed as a prognostic factor in our breast cancer center. Because this analysis was performed in bone marrow, we were able to investigate the two markers in both peripheral blood and in the corresponding bone marrow. Free circulating DNA in serum and plasma has been shown to be increased in cancer patients [1, 30]. Therefore, the analysis of this DNA fraction might be a useful tool to characterize a tumor.

It was interesting to note that the prevalence of positive results was higher in BM-P than in PB-P for methylated PITX2, but not for methylated RASSF1A, although these two markers correlate significantly with each other in PB-P as well as in BM-P. This indicates that PITX2 and RASSF1A are not co-expressed in tumor cells—or are methylated in a different manner—or that they derive from different cell types. Assuming that methylated PITX2 and RASSF1A occur exclusively in tumor cells, methylated PITX2 would be the more sensitive marker for breast cancer cells. However, the fact that patients with benign lesions of the breast also show positive results for methylated PITX2 and RASSF1A speaks against considering them classic tumor markers, as both sensitivity and specificity are low. Occurrence of methylated PITX2 and RASSF1A in benign conditions speaks against the concept that they are malignancy-specific, at least concerning the serum fraction. It could be speculated that in all cases where increased cell turn over occurs, methylated PITX2 and RASSF1A appear in the serum. Associations of analyzed markers with clinical indicators of prognosis were observed for RASSF1A (in PB-P and BM-P) and for PITX2 in BM-P, but only with age and menopausal status. PITX2 in PB-P did not correlate with any of the clinical parameters. This speaks for an essential influence of the host on the methylation of these two genes. Kaplan–Meier survival analysis indicated a variety of significant differences for OAS and DDFS for both markers. Results in BM-P were only significant for RASSF1A methylation and OAS. In the different treatment groups the strongest differences occurred in the group of combined chemo-hormone-therapy. This could mean that the significant differences obtained when analyzing the whole population of patients may derive mainly from this group of patients. In a multivariate analysis PITX2- and RASSF1A-methylation in PB-P remained independent prognosticators. This implies, together with the results of univariate analysis, that PB-P is superior to BM-P as a sample source for the determination of these two biomarkers.

When comparing the prognostic potential of methylated PITX2 and RASSF1A, it is not clear which one of them allows better discrimination between good and bad prognosis. A combination of both markers seems to improve the diagnostic validity. The great advantage of these markers is that they can be determined in blood, which would represent a step forward in the evaluation of the course of disease in breast cancer patients.

Acknowledgment

This work was supported by the COMET Center ONCOTYROL.

Copyright information

© Springer Science+Business Media, LLC. 2011