Cancer Biomarkers 11 (2011/2012) 89–98 89DOI 10.3233/CBM-2012-0263IOS Press

Circulating cell-free DNA: A promisingmarker of regional lymphonode metastasis inbreast cancer patients

M. Agostinia,b,c,∗, M.V. Enzoa,c, C. Bedina,c, V. Belardinellia, E. Goldina, P. Del Biancod,E. Maschiettoa,c, E. D’Angeloa,c, Leo Izzie, A. Saccanie, G. Zavagnoa and D. NittiaaDepartment of Surgical, Oncological and Gastroenterological Sciences, 2nd Surgical Clinic, Padova, ItalybThe Department of Nanomedicine, The Methodist Hospital Research Institute, Houston, TX, USAcIstituto di Ricerca Pediatrica- Citta della Speranza, Viale della Ricerca, Padova, ItalydClinical Trials and Biostatistics Unit, Veneto Oncological Institute – IRCSS, Padova, ItalyeEuroclone S.p.a, Pero (MI), Italy

Abstract. Purpose: We undertook the current study with untreated breast cancer to (1) role the variations in the plasma levels ofcfDNA and the size distribution in early stage, (2) determine the frequency in plasma of methylation of three candidate genes,RASSF1A, MAL, and SFRP1, and (3) to determine whether detection of cfDNA variations and methylation changes in plasmamight have specific clinical utility.Methods and materials: Thirty-nine patients woman patients (median age 64 years; range, 36–90 years) who underwent surgeryfor primary BR and 49 healthy females’ subjects (control group without any breast lesion) were evaluated. The cfDNA levelswere analyzed using quantitative real-time polymerase chain reaction of β-globin. Based on the ALU repeats, the cfDNA wasconsidered as either total (fragments of 115 bp, ALU115) or tumoral (fragments of 247 bp, ALU247). The association betweenthe levels of the ALU247, ALU115 repeat, and ALU 247/115and the pathologic tumor characteristics was analyzed.Used methylight qPCR method, cfDNA from plasma samples of healthy donors and patients with breast cancer were evaluatedfor the diagnotic value of the methylation status of three genes (RASSF1A, MAL, SFRP1) frequently methylated in breast cancer.Results: The baseline levels of cfDNA were significantly higher in the patients with cancer, and the level of ALU247 was themost accurate circulating cfDNA marker in discriminating the cancer from non-cancer subjects.A high statistical significance was found by considering the T stage and patients with regional LN metastasis positive cancersshowed significantly higher cfDNA level of ALU247. Moreover, patients with methylation of at least one of the gene underinvestigate showed a higher quantity of cfDNA ALU115 (p < 0.0001) and ALU247 level (p < 0.0001).Conclusions: We observed that necrosis could be a potential source of circulating tumour-specific cfDNA ALU247; and thatcfDNA ALU247 and methylated cfDNA (RASSF1A, MAL and SFRP1) are both a phenotypic feature of tumour biology.

Keywords: Breast cancer, regional LN metastasis, cfDNA

1. Introduction

In most developed and many developing countries,breast cancer (BR) is the most frequent cancer andthe leading cause of cancer death in women. Current

∗Corresponding author: Marco Agostini, PhD, Department ofOncological and Surgical Gastroenterological Sciences, SurgicalClinic II, University of Padova, Via Giustiniani 2, 35128 Padova,Italy. Tel.: +39 049 8214374, Fax: +39 049 651891, E-mail:m.agostini@unipd.it.

screening methods fail to detect many cancers thatpresent at a later date as a result of symptoms [1,2].Early detection and a reliable follow-up of breast can-cer are crucial for successful treatment. For a definitivediagnosis, a tumour biopsy is required. However, anon-invasive test for early detection of the disease andfor monitoring disease progression has been a goal formany researchers.

Although the potential of biomarkers to aid in theearly detection, diagnosis, prevention, and treatmentof breast cancer is broadly recognized and numerous

ISSN 1574-0153/11/12/$27.50 2011/2012 – IOS Press and the authors. All rights reserved

AUTH

OR

COPY

90 M. Agostini et al. / Breast cancer and circulating cell-free DNA

biomarker candidates have been reported in the liter-ature, few molecular markers have been adopted intoclinical use to date [3]. Particular attention is reservedto study in plasma and serum of epigenetic events. Epi-genetic changes, such as DNA methylation are one ofthe most common molecular alterations in human neo-plasia [4–7] including breast cancer [8]. DNA hyper-methylation in the promoter region of many genes isfrequently associated with “gene silencing” (i.e., thegene is expressed in the absence of methylation) [9]and gene-specific hypermethylationare associated withmalignancy [10–12]. Studies in animals and in hu-mans have demonstrated that these epigenetic changesare an early event in carcinogenesis and are presentin the precursor lesions of a variety of cancers includ-ing breast [13–15]. Recently, also the cell-free DNA(cfDNA) circulating in plasma or serum shows charac-teristics of a potential candidate biomarker of tumourprogression and chemo-treatment monitoring [16]. Al-though circulating cfDNA is present in healthy individ-uals, it has been indicated as a strong diagnostic andprognostic marker of malignancy at different type oftumours [17–19].

Its limitations are related to the contradictory find-ings reported concerning the proportion of tumour- andnon-tumour-derived cfDNA. Interestingly, it has beenhypothesised that cfDNA released into circulation fromtumour necrosis varies in size, whereas cfDNA re-leased from non-tumoural apoptotic death is uniformlytruncated into fragments shorter than 200 bp [20–22].In healthy individuals, the main source of cfDNA isthought to be apoptotic cells. In contrast, necrotic celldeath is a frequent event in solid tumours, and DNAfragments released from the tumour cells are variablein length [20]. As a consequence, the amount of longerDNA fragments and the ratio between the longer andshorter fragments, named the integrity index, may re-flect the presence of cancer [23]. Recently, integrity ofthe free circulating DNA in serum using quantitativereal-time PCR for ALU repeats seem to be a reliablebiomarker for the detection of cancer [24]. Accord-ingly, recent studies have reported an increased ratio(i.e., the cfDNA integrity) between the long and shortfragments in patients with cancer [23–26].

Thus far, however, there are only limited data avail-able on the detection of epigenetic alterations in serumand/or plasma and role of cfDNA in primary breastcancer. Therefore we undertook the current study withuntreated BR to (1) role the variations in the plasmalevels of cfDNA and the size distribution in early stageBR, (2) determine the frequency in plasma of methy-

lation of three candidate genes, RASSF1A, MAL, andSFRP1 which preliminary studies suggest are impor-tant in breast cancer, and (3) to determine whether de-tection of cfDNA variations and methylation changesin plasma might have specific clinical utility.

2. Material and methods

2.1. Plasma samples and clinicalpathologicalinformation

This studywas undertaken at the 2nd Surgical Clinic,University of Padova, Italy.

Blood samples were obtained from 39 woman pa-tients with primary BR and from 49 healthy females’subjects (control group without any breast lesion). Fol-lowing UICC classification the BR group consisted outof n = 31 patients with cancer stage I and stage II, n =4 pts with stage III and n = 4 pts no classified. Bloodwas drawn before any therapeutic treatment and within15 days before the time of surgery. No women reporteda history of previous cancer treatment or a history ofbreast biopsy due to benign or malignant breast lesions.

The median age was 59 (32–75) years for 49 healthyfemales and 64 (range 36–90) years for 39 patients withprimary breast cancer.

All cancers were ductal carcinomas and the meantumour size was 1.5 ± 0.9 (SD) cm.

All patients enrolled underwent breast surgery withsentinel node lymphonodectomy or axillary lymphon-odectomy according to the stage of disease, betweenJune 2006 and October 2006 and the clinicopatholog-ical data are listed in Table 1. Plasma were sampledpreoperatively.

To confirm the ability of the cfDNA to discriminatethe patients with breast cancer and the healthy subjects,blood samples of healthy individuals without a person-al history of cancer were obtained following mammo-grafty that showed neither cancer nor other type of neo-plasia. A ratio of patients/healthy subjects of 1/1 wasconsidered adequate.

The study protocol was reviewed and approved bythe local ethics committee (protocol number 448), andeach patient provided written informed consent.

2.2. Immunohistochemical analysis of EgR, PgR,HER2, Ki-67

A breast cancer pathologist interpreted the slideswithout knowledge of the clinical outcome of each case.

AUTH

OR

COPY

M. Agostini et al. / Breast cancer and circulating cell-free DNA 91

Table 1Patient and tumour characteristics of the 39 patients included in the study

Clinical information Number %

Age Median (range) yrs Cancer 64 (36–90)Control 59 (32–75)

Tumour size (cm) < 2 26 67>= 2 12 31nv 1 2

Grading 1 6 152 26 673 6 15nv 1 3

Stage (pTNM) I/II 31 80III/IV 4 10nv 4 10

Histopathology Infiltrating 33 85not infiltrating 6 15

EgR < 10% positive tumor cells 8 18> 10% positive tumor cells 32 82

PgR < 10% positive tumor cells 16 41> 10% positive tumor cells 23 591–19% 24 62

Ki-67 20–30% 11 2831–50 4 10

HER2 negative 17 44positive 18 46nv 4 10

Vascular invasion absent 32 82present 7 18

EgR = Estrogen receptor, PgR = Progesterone receptor.

Breast cancer tissue was obtained from prechemother-apy core needle biopsy, fixed in 10% neutral bufferedformalin for 48 hours, and embedded in paraffin. Se-rial sections of 4-micron thickness were obtained fromtissue blocks and processed for immunohistochemicalstudies after confirmation of histopathology and tumourgrade.

Slides were then deparaffinized and rehydrated withdescending grades of ethyl alcohol. Immunohistochem-istry staining was performed using the following anti-bodies: EgR (BioGenex ER88 prediluted, Clone 1D5,Mouse IgG), PgR (BioGenex PR88 prediluted, dilu-tion 1:20), HER2 (BioGenex prediluted, Clone CB11,Mouse IgG), Ki-67 (Biogenex Recombinant humanKi67 protein, Clone Ki88, Mouse Ig G1 kappa).

Immunostainingwas performedmanually while em-ploying a biotin-streptavidin complex. 3,3-Diamino-benzidine tetrahydrochloride was used as chromogen,and sections were counterstained with haematoxylin.Microwave pretreatment (slides were immersed in a10-mM citrated buffer, pH 6.0, at 95◦C, 5 min × 3)for antigen retrieval was carried out prior to incuba-tion with primary antibody. Positive tissue controls ofthe breast carcinoma as well as negative control slides

run simultaneously were used to assess the quality ofimmunostaining. Immunoreactivity was quantified byevaluating a minimum of 1,000 carcinoma cells in ran-domly selected fields on histological sections using ahigh-powered objective lens.

The results for EgR, and PgR protein were consid-ered positive when more than 10% of the carcinomacells showed nuclear positivity. For the evaluation ofHER2 expression, semiquantitative analysis was per-formed according to scoring guidelines from the Her-cepTest instruction guide. In brief, a score of 0 was giv-en to sections showing no staining or membrane stain-ing in less than 10% of tumour cells; 1+ was given tosections showing faint or barely perceptible membranestaining in more than 10% of tumour cells; 2+ wasgiven to sections showing weak to moderate completemembrane staining in more than 10% of tumour cells;and 3+ was given to specimens showing strong com-plete membrane staining in more than 10% of tumourcells. Scores of 0, 1+, and 2+ were considered nega-tive for HER2 overexpression, whereas a score of 3+was considered as HER2 overexpression. Only mem-branous staining intensity and patterns were evaluatedas defined in the guidelines.

AUTH

OR

COPY

92 M. Agostini et al. / Breast cancer and circulating cell-free DNA

The Ki-67 index was calculated as the percentage ofcells demonstrating nuclear Ki-67-positive carcinomacells. A Ki-67 index equal to or more than 20% wasconsidered to be a high level. For Bcl-2 and BAX,positive status was considered when more than 25%of the carcinoma cells were positive or when strongto moderate immunostaining was observed. Faint tonegative immunostaining was considered negative. Astrong correlation was seen between the percentages ofcells positive for stain vs. the intensity of staining.

2.3. Extraction of DNA from plasma samples

Seven ml of peripheral blood was drawn into ablood collection tube (containing EDTA additive) be-fore physical examination or biopsy. Plasma sampleswere left at room temperature for molecular assaysand all were transferred to the study laboratory within4 hours of collection for processing. The plasma sam-ples were obtained by centrifugation of 7 ml of periph-eral blood at 3,000 g for 10 min. The plasma sampleswere carefully collected from the upper portion of thesupernatant and stored in aliquots at −80◦C. The DNAwas purified from 500 μl of plasma with a QIAampDNA Mini Kit (Qiagen, Hilden, Germany) accordingto the manufacturer’s instructions; the washes with thebuffer AW2 were increased to two to remove inhibitorsof the polymerase chain reaction (PCR) [27]. The DNApreparations were eluted in 80 μl of elution buffer. Theeluted DNA was stored at −20◦C until further use.

2.4. Quantitative PCR of plasma DNA fragments

The quantification of the DNA fragments was per-formed by quantitative real-time PCR (qPCR), whichamplified and quantified the shorter and longer frag-ments.

To maximize the sensitivity of the DNA quantifica-tion, the ALU repeats, which are the most abundantrepeat sequence in the human genome, were used as atarget of the qPCR.

Two primer pairs were used as previously report-ed [25]. One set of primers (ALU115) amplified boththe shorter (115 bp) and longer fragments, whereas thesecond primer set (ALU247) amplified only the longer(247 bp) DNA fragments. The results obtained usingthe ALU115 primers represent the total free plasmaDNA, while the results of the ALU247 primers reflectthe amount of DNA released from non-apoptotic cells.The quantitative values from the 115-bp primers rep-resent the total level of cfDNA in ng/ml, while the ra-

tio of longer to shorter fragments show the integrity ofcfDNA in each sample.

The sequences of the ALU115 primers were as fol-lows: forward, 5’-CCTGAGGTCAGGAGTTCGAG-3’ and reverse,5’-CCCGAGTAGCTGGGATTACA-3’;the ALU247 primers were as follows: forward, 5’-GTGGCTCACGCCTGTTAATC-3’ and reverse, 5’-CAGGCTGGAGTGCAGTGG-3’ (PRIMM, Milan).

The reaction was performed using 7300 Real-TimePCR System (AppliedBiosystem, Milan) as previouslyreported [25]. The absolute amount of cfDNA in eachsample was determined by a standard curve using 3,3-fold dilutions (from 10 ng to 1 pg) of genomic DNAfrom the peripheral lymphocytes of healthy subjects.

A negative control (without the template) was per-formed in each plate. All qPCR assays were performedin a blinded fashionwithout knowledgeof the specimenidentity. Following amplification, melting curves wereobtained to confirm the accurate amplification signal.In this setting, each amplicon displays a specific melt-ing behavior. The 115-bp amplicon produced a peak atapproximately 79◦C, whereas the 247-bp amplicon hada higher melting temperature of approximately 83◦C.The quantitative ratio of ALU247 to ALU115 reflectedthe integrity of the cfDNA (cfDNA integrity index =ratio of ALU 247/115).

The total plasma cfDNA was also measured using areal-time quantitative PCR assay of the β-globin gene.β-globin forward 5’-TGAGTCCAAGCTAGGCCCTTT-3’; β-globin reverse 5’-CCAGGAGCTGTGGGAGGAA-3’ (Applied Biosystem, Milan); and a la-beled fluorescent TaqMan β-globin probe, (FAM) 5’-CTAATCATGTTCATACCTCTTAT-3’ (MGB) (Ap-plied Biosystem, Milan).

2.5. Analysis in plasma samples of cfDNAMethylation

cfDNA from plasma samples of healthy donors andpatients with breast cancer were used to evaluate the di-agnostic value of the methylation status of three genes(RASSF1A, MAL, SFRP1). These genes had been re-ported to be frequentlymethylated in breast cancer [28–30]. It was used Quantitative Real-Time PCR methy-lation (methylight qPCR) method. MethyLight is atechnology that combines the specificity of MSP with(the sensitivity of real time PCR) the ability to de-tect small amount of DNA. Patients were classified asmethylation-positive if at least one of the genes includ-ed in the combination showed aberrant methylation.

AUTH

OR

COPY

M. Agostini et al. / Breast cancer and circulating cell-free DNA 93

2.6. Sodium bisulfite conversion of unmethylatedcytosine in DNA

Plasma cfDNA was modified with EpiTect BisulfiteKit (Qiagen, Hilden, Germany) following the manu-facturer’s protocol. For bisulfite conversion, we addedbisulfite mix solutions and DNA protect buffer to 40 ulof cfDNA in a 0.2 mL elution tube and performed theconversion using a GeneAmp PCR System 9700 (Ap-plied Biosystems, Milan) for 5 h at 60◦C with 3 thermalspikes at 99◦C. We purified the converted single strand-ed DNA using a EpiTect spin column based purifica-tion kit for bisulfite-converted DNA (Qiagen, Hilden,Germany). Purified DNA was eluted in 25 uL elutionbuffer (10 mmol/L Tris, pH 7.2) and used directly inreal-time PCR analysis.

2.7. Quantitative real-time PCR methylation(MethyLight)

Sodium bisulfite-treated plasma DNA was analyzedby the MethyLight, a fluorescence-based, realtime PCRassay, as previously described [31].

Briefly, four sets of primers and probes designedspecifically for bisulfite-converted DNA were used:three sets for genes of interest (RASSF1A, MAL,SFRP1) and one set for MYOD to control for bisul-fite conversion and to normalize for the amount ofinput DNA. Primers and probes for RASSF1A, MAL,SFRP1, and MYOD were described following. TheRASSF1A forward primer is 5’-CGTATTCGGTTGGAGCGTGT-3’, the RASSF1A reverse primer is 5’-GAACTAAAAACGATAACCACGACCA-3’, and theRASSF1A probe is 6FAM-5’-TATCGCGTGTAATTTTATAC-3’-MGB; the MAL forward primer is 5’-CGTTCGTTTCGTTTTAAGGTCG-3’, the MAL reverseprimer is 5’-CCCGTCGCCGCTACG-3’, and the MALprobe is 6FAM-5’-CGTTAGTACGTCGTTATGGT-3’-MGB; the SFRP1 forward primer is 5’-TTTTCGCGTCGGTGACG-3’, the SFRP1 reverse primer is 5’-AATCAACTCCCGACGAAACG-3’, and the SFRP1probe is 6FAM-5’-CGTGGTAACGAGTGCG-3’-MGB; MYOD forward primer is 5’-TTTTAGTTAGAGTGTTGAGAGGATTGTGT-3’, the MYOD reverseprimer is 5’-CATACCGACCACCCCCATAA-3’, andthe MYOD probe is 6FAM-5’-ATTGTAGATTTAGGAAGAGGTT-3’-MGB.

We used a 7500 Real-Time PCR System (AppliedBiosystems, Milan). The reaction was performed in20 μl reaction volume containing 1X TaqMan Univer-sal PCR Master Mix No AmpErase UNG, 0.9 μM of

each primer, 0.25 μM of probe and 4 μL of modifiedDNA. The real time PCR conditions consisted of aninitial denaturation for 10 minutes at 95◦C followedby 50 cycles of denaturation for 15 seconds at 95◦Cand annealing/extension for 1 minute at 60◦C. The ap-propriate positive and negative reference samples wereincluded.

Absolute quantity was derived from standard curvegenerated by serial dilution of methylated and sodi-um bisulfite modified DNA. The serial diluted standardDNA was subjected to bisulfite modification togetherwith the sample DNA at the same time. The concen-tration of plasma DNA was expressed as ng/mL. Mul-tiple negative controls, no template control, unmethy-lated control DNA (Qiagen, Hilden, Germany), and nobisulfite conversion control were included for each setof reaction.

The specificity of reaction for methylated DNA wasconfirmed separately using fully methylated controlDNA (Qiagen, Hilden, Germany).

The amount of methylated cfDNA at a specific locuswas calculated by dividing the GENE: MYOD ratio of asample by the GENE: MYOD ratio of fully methylatedhuman genomic DNA (Qiagen, Hilden, Germany) andmultiplying by 100 (PMR, percentage of methylated.Reactions using fully methylated DNA were used tonormalize for any difference in amplification efficien-cies between target gene and MYOD.

A gene was deemed methylated if the percentage offully methylated reference value was > 0.

2.8. Statistical methods

The distributions of cfDNA between cases and con-trols and among categories of clinical and biologicalvariables, such as Progesterone receptor, Estrogen re-ceptor, MIB, C-ERB2, methylation status, were com-pared by the Kruskal-Wallis test.

The receiver operating characteristics curves weredrawn for patients as compared with controls and fornodal positive patients as compared with node negativepatients. The role of cfDNA in predicting case/controlstatus and positive/negative nodal status was evaluatedwith the areas under the curve with the 95% CI. Thesensitivity and specificity of the optimumcut-off valueswere subsequently estimated.

For each clinical variable, a logistic regression analy-sis was performed to explore the association with nodalstatus.

All tests were two-sided, and a P < 0.05was consid-ered statistically significant. Statistical analyses wereperformed by using SAS version 9.1 (SAS Institute,Cary, NC).

AUTH

OR

COPY

94 M. Agostini et al. / Breast cancer and circulating cell-free DNA

Table 2Baseline plasma levels of cfDNA in patients with breast cancer and in the healthy controls

Markers Patients n = 39 Controls n = 49 p-valuemedian (Q1–Q3) median (Q1–Q3)

ALU115 ng/ml 52,47 (21,40–147,22) 0,38 (0,09–1,72) < 0.0001ALU247 ng/ml 48,85 (17,21–100,6) 0,3 (0,08–0,64) < 0.0001β−globin ng/ml 300,00 (88,00–580,00) 5,78 (1,69–12,05) < 0.0001Ratio ALU247/115 0,62 (0,44–1,23) 0,33 (0,00–0,82) 0.0002

3. Results

3.1. The cfDNA distribution in healthy individualsand primary breast cancer patients

The ability of cfDNA to discriminate the patientswith primary breast cancer and the healthy subjectswas confirmed using a control group of plasma sam-ples from healthy subjects. These plasma samples wereobtained immediately after a mammography yieldednegative findings of cancer and benign lesion.

In the patients, the baseline levels of ALU115,ALU247 and β-globin were significantly higher (p <0.0001) than in the control group (Table 2), and the cfD-NA integrity index was 2 times higher in the patientsthan in the healthy controls.

Using the ROC curve analysis, each method of cfD-NA quantification showed a high ability to discrimi-nate healthy individuals from patients (AUCALU115 =0.99, 95% CI: 0.95–1.0; AUCβ−globin = 0.98, 95% CI:0.92–1.0; and AUC ALU247 = 1.00, 95% CI: 0.96–1.0).It should be noted that, with a cut-off of 2.0 ng/ml, thediagnostic power of ALU247 showed 100% specifici-ty (95% CI: 92.7–100) and 100% sensitivity (95% CI:91–100). Instead the same sensitivity with ALU115is obtained with a cut-off more than four times: witha cut-off > 9.3394 ng/ml 100% specificity (95% CI:92.7–100) and 94.8% sensitivity (95% CI: 82.7–99.4).

Interestingly, a high statistical significance wasfound by considering the T stage. For tumours less than2 cm (T1) the baseline levels of ALU115, ALU247 andβ-globin were significantly higher (p < 0.0001, p <0.0001 and 0.0027) than in the control group.

No statistical difference was found considering infil-trating condition. In six cases with in situ lesions cfD-NA levels were comparable to those of other T stages:in situ medial level cfDNA ALU247 = 51.35 ng/ml,IQR = 14.07–100.03; infiltrating T1 medial level cfD-NA ALU247= 48.85 ng/ml, IQR = 22.21–108.54;T2-T3-T4 medial level cfDNA ALU247 = 46.05 ng/ml,IQR = 16.61–86.28; in situ medial level cfDNAALU115 = 36.75 ng/ml, IQR = 15.43–181.61; infil-trating T1medial level cfDNAALU115= 50.62ng/ml,

IQR = 28.73–111.85; T2-T3-T4 medial level cfDNAALU115 = 130.65 ng/ml, IQR = 22.21–108.54; in situmedial level cfDNA β-globin = 186 ng/ml, IQR = 48–460; infiltrating T1 medial level cfDNA β-globin =540 ng/ml, IQR = 500–600; T2-T3-T4 medial levelcfDNA β-globin = 271 ng/ml, IQR = 60–600).

3.2. Plasma DNA integrity and pathologycharacteristics of primary breast cancers

Mean cfDNA integrity, ALU115, and ALU247 wereindependent of age in breast cancer patients (data notshowed) then age was not a confounding factor.

In the 39 patientswith invasive primary breast cancer,mean plasma DNA integrity, cfDNA ALU 115 andprimary tumour size were not significantly correlated.Conversely a trend association with plasma level ofcfDNA ALU247 and tumour size seemed to be present(p = 0.0056): plasma level of ALU247 increased withthe tumour expansion.

Regional lymphnode (LN) metastasis determined byhistopathology and results of 33 primary tumours wasavailable. Eleven tumours were classified as regionalLN metastasis positive tumours and 22 were classifiedas regional LN metastasis negative tumours It should benoted that total cfDNA, measured by ALU115 and β-globin, did not show any differences between regionalLN metastasis positive and regional LN metastasis neg-ative tumours (p = 0.89 and p = 0.47, respectively).On the other hand, patients with regional LN metasta-sis positive cancers showed significantly higher cfDNAlevel of ALU247 (median 100.56 ng/ml, IQR = 28.54–380.31) than patients with regional LN metastasis neg-ative cancers (median 39.36, IQR = 16.35–88.58) (p =0.041) (Fig. 1). This lead the cfDNA integrity index toshow a statistically significant difference between thegroups (p = 0.025).

Using the ROC curve analysis, cfDNA integrityquantification showed a moderate ability to discrim-inate patients with regional LN metastasis from pa-tients without LN metastasis (AUC = 0.74, 95% CI0.54–0.88). With a cut-off of 0.6, the diagnostic powerof cfDNA integrity showed 78% specificity (95% CI:54.6–92.2) and 73% sensitivity (95% CI: 39–94).

AUTH

OR

COPY

M. Agostini et al. / Breast cancer and circulating cell-free DNA 95

Table 3Baseline plasma levels of cfDNA ALU247 and relationship with clinical data

Median (Q1–Q3)N ng/ml p-value K-W

Grading 1 6 (14.07–88.72) 0.91272 26 (22.21–108.54)3 6 (16.61–100.56)

Tumor stage vs. healthy pT1 32 48.85 (19.71–104.55) < 0.0001Healthy 49 0.29 (0.08–0.64)

Tumor stage T1 vs. other T stage pT1 32 48.85 (19.71–104.55) 0.6889pT2-pT3-pT4 6 46.05 (16.61–86.28)

Lymph node metastasis N0 22 39.36 (16.35–88.58) 0.0410N+ 11 100.56 (28.54–380.31)

MIB < 20% 22 32.75 (16.35–88.72) 0.5584� 20% 8 48.85 (30.70–98.56)

Estrogen receptor Absent 5 71.12 (57.69–100.03) 0.6580Low 2 73.99 (25.46–122.52)High 32 39.36 (16.78–94.64)

C-ERB2 Negative 17 48.85 (25.46–88.58) 0.9211Positive 18 44.63 (16.61–112.62)

Progesterone receptor Absent 9 57.69 (16.61–100.03) 0.6758Low 7 28.54 (17.21–380.31)Medium 2 98.63 (88.72–108.54)High 21 44.79 (22.21-88.58)

Histology Infiltrating 33 48.85 (22.21–100.56) 0.6685not infiltrating 6 51.35 (14.07–100.03)

Vascular Invasion negative 32 46.82 (16.78–104.29) 0.6738positive 7 48.85 (23.57–100.56

No relationshipwas observed between all markers ofcfDNA levels and pathologic stage, histology, gradingstage, estrogen receptor (EgR), progesterone receptor(PgR) and C-ERB2 status and vascular invasion (Ta-ble 3).

3.3. Quantitative real-time PCR methylation

In healthy subject no methylation of three genes wasfound (100% of specificity). In breast cancer patientsRASSF1A was hypermethylated in 9 of 39 (23%), MALin 5 of 39 (13%) and SFRP1 in 4 of 39 (10%). Intotal 16 of 39 (41%) patients cfDNA were positive forhypermethylation of at least one of the genes studied.

Interesting, patients with methylation of at least oneof the gene under investigate showed a higher quantityof cfDNA ALU115 (p < 0.0001) and ALU247 level(p < 0.0001).

cfDNA integrity was also higher in subjects withmethylation than subjects without it but there was notstatistical significance.

Using the ROC curve analysis, cfDNA ALU247quantification and methylation status showed a moder-ate correlation (AUC = 0.80, 95% CI 0.7–0.9). Witha cut-off of 6.4514 ng/ml, the correlation power of

Fig. 1. Baseline level of ALU 247 in plasma samples of regional LNmetastasis positive and negative patients.

cfDNA ALU247 quantification and methylation statusshowed 69% specificity (95%CI: 57.1–79.2) and 100%sensitivity (95% CI: 79.4–100).

4. Discussion

The main finding of the study was that the plasmalevel of cfDNA is able to differentiate the patients with

AUTH

OR

COPY

96 M. Agostini et al. / Breast cancer and circulating cell-free DNA

early stage breast cancer from the healthy subjects andthat the variations in the plasma levels of cfDNA longfragments and the DNA integrity are associated withthe regional LN metastasis. However, ours results indi-cating that plasma levels of circulating cell-free DNAcould have diagnostic value to discriminate betweenhealthy individuals and patients with breast lesions, butthey have no relevance to discriminate between patientswith late stage malignant and in situ breast lesions. Asdescribe also in previous evidences, nucleic acids maybe released into the blood circulation by damaged cellswhich are present in both, benign and malignant dis-eases. A factor for these high serum plasma concen-trations in patients with benign breast tumors could bethe delayed clearance of nucleic acids [32–34].

However, identifying a lesion since its first manifes-tation (benign, in situ, malignant) is of considerableimportance in the screening of this disease and howstressed could be useful as a noninvasive test and as asupport to traditional screening methods.

Furthermore, circulating cfDNA and the cfDNA in-tegrity index have been shown to be promising diag-nostic biomarkers of many type of tumors [24,25,35].

To date, there is no agreement on the mechanismsthat are responsible for the presence of tumour DNAshed into the bloodstream. Among the most commonhypothesis is also the shedding of lysed circulating tu-mour cells (CTC). However, this seems to be unlikelybecause the number of CTC is inadequate to explain theobserved amount of DNA in the plasma or serum [36].Therefore the most likely hypothesis remains apoptosisand necrosis of tumour cells that increase the levels ofcirculating DNA [21] and that DNA is actively releasedinto the bloodstream by the tumour [36].

Apoptosis and necrosis [20,37] can be distinguishedby the dimension of the DNA fragments: the apop-totic death of cells releases DNA fragments shorterthan 200 bp into circulation, whereas tumour necro-sis is characterized by the presence of fragments thatvary in size and are generally greater than 200 bp. Insome patients there are fragments that indicate the eventof apoptosis, while in other patients, both fragmentsof longer and shorter dimension are present [21]. Thefindings of our study seem to confirm this hypothesis.Compared with healthy controls, significantly higherlevels of cfDNA were present in the patients with can-cer, and the level of long DNA fragments (ALU247)was the most accurate circulating cfDNA marker in dis-criminating the cancer from non-cancer subjects [38].It should be noted that, with a cut-off of 2.0 ng/ml,the levels of ALU247 showed a sensitivity and speci-

ficity of 100%, with an overall diagnostic accuracy of97%. On the other hand significant evidence was madeby the cfDNA integrity and ALU247 quantificationthat showed a moderate ability to discriminate patientswith regional LN metastasis from patients without LNmetastasis.

However, the presence of altered DNA is not infor-mative of the tumour site. It is unclear whether all can-cer types can release altered cfDNA at the same rate.Moreover it is likely that differences in cfDNA releaseexist according to organ site, tumour location, histolo-gy, vascularization, grade and stage.

Nevertheless an increasing number of studies havereported the presence of methylated DNA in serum/plasma of patients with various types of malignancies,including breast cancer, and the absence of methylatedDNA in normal control patients [39–41].

In this contest methylation markers offer an oppor-tunity for the clinical application of sensitive cancerdetection. Although the detection of DNA methylationmarkers in plasma cfDNA has fairly low sensitivity,they excel in specificity.

Therefore, methylation markers could be very usefulas an ancillary tool in risk assessment or disease detec-tion, by enhancing the specificity of existing screeningmethods with lower specificity, such as cfDNA ALUand integrity index quantification and it is merely in-dicative of the release of sufficient amounts of tumourDNA into the bloodstream, which is likely to be corre-lated with invasiveness.

In any healthy subject of our study no methylation ofRASSF1A, MAL, and SFRP1 (100% of specificity) wasfound and an interesting correlation (AUCALU247 =0.8, 95% CI: 0.7–0.9) exist with cfDNA level ALU247.

RASSF1A, MAL, and SFRP1 are involved in path-ways counteracting metastasis: mediation of intercel-lular adhesion, stabilization of the cytoskeleton, regu-lation of the cell cycle, and apoptosis [30,42–46]. Ir-respective of the mechanistic role of methylated DNAwith regards to metastasis in breast cancer patients,these epigenetic changes in serum or plasma have sev-eral advantages as indicators of tumour site [29,30,47,48].

Therefore, this could play an important role in non-invasive diagnosis of some cancers such as breast can-cer: a simple blood draw (which can be repeated anytime throughout the follow-up period) is sufficient.This plays an important role when considering thatthe more screeningmammographies are performed, themore small cancers are treated, and after histopatho-logical examination no tumour material will remain

AUTH

OR

COPY

M. Agostini et al. / Breast cancer and circulating cell-free DNA 97

to perform RNA and/or protein-based assays for riskevaluation.

In our study we observed that necrosis could be apotential source of circulating tumour-specific cfDNAALU247 and that cfDNAALU247 andmethylated cfD-NA (RASSF1A, MAL and SFRP1) are both a pheno-typic feature of tumour biology. In conclusion, thisstudy provides evidence that the detection of amountsof cfDNA and of methylated genes is associated withpresence of breast cancer. Furthermore, the combinedassessment of these two molecular markers was predic-tive for tumour early diagnosis. However, others stud-ies are necessary to verify the clinical utility of cfDNAALU247 quantification and serum/plasma methylatedcfDNA as potential prognostic factors in breast cancerpatients.

Acknowledgements

This article is dedicated to Leo Izzi.A person who looked beyond the simple rational

things, a person who believed in us young researchers,and he knew how to convey the passion for this worldof research. His words went away with him. We stilltrust in what we are doing. Thank you dear friend.

This study was supported in part by grants fromthe Euroclone, Banca AntonVeneta, CARIPARO andAIRC Foundation, and Euroclone S.p.a.

Biological samples were provided by 2nd SurgicalClinic, Tumour Tissue Biobank.

No conflicts of interest to declare. All patients signedinformed consent form.

References

[1] Kattlove, H., Liberati, A., Keeler, E., andBrook, R.H.Benefitsand costs of screening and treatment for early breast cancer.Development of a basic benefit package. Jama, 273: 142-148,1995.

[2] Chen, X., Bonnefoi, H., Diebold-Berger, S., Lyautey, J., Led-errey, C., Faltin-Traub, E., Stroun, M., and Anker, P. Detectingtumor-related alterations in plasma or serum DNA of patientsdiagnosed with breast cancer. Clin Cancer Res, 5: 2297-2303,1999.

[3] Wang, W. and Srivastava, S. Strategic approach to validatingmethylated genes as biomarkers for breast cancer. Cancer PrevRes (Phila), 3: 16-24.

[4] Baylin, S. B., Herman, J. G., Graff, J. R., Vertino, P. M., andIssa, J. P. Alterations in DNA methylation: a fundamentalaspect of neoplasia. Adv Cancer Res, 72: 141-196, 1998.

[5] Bird, A. The essentials of DNA methylation. Cell, 70: 5-8,1992.

[6] Merlo, A., Herman, J. G., Mao, L., Lee, D. J., Gabrielson, E.,Burger, P. C., Baylin, S. B., and Sidransky, D. 5’ CpG islandmethylation is associated with transcriptional silencing of thetumour suppressor p16/CDKN2/MTS1 in human cancers. NatMed, 1: 686-692, 1995.

[7] Herman, J. G., Umar, A., Polyak, K., Graff, J. R., Ahuja,N., Issa, J. P., Markowitz, S., Willson, J. K., Hamilton, S.R., Kinzler, K. W., Kane, M. F., Kolodner, R. D., Vogelstein,B., Kunkel, T. A., and Baylin, S. B. Incidence and functionalconsequences of hMLH1 promoter hypermethylation in col-orectal carcinoma. Proc Natl Acad Sci USA, 95: 6870-6875,1998.

[8] Muller, H. M., Widschwendter, A., Fiegl, H., Ivarsson, L.,Goebel, G., Perkmann, E., Marth, C., and Widschwendter,M. DNA methylation in serum of breast cancer patients: anindependent prognostic marker. Cancer Res, 63: 7641-7645,2003.

[9] Leonhardt, H. and Cardoso, M. C. DNA methylation, nuclearstructure, gene expression and cancer. J Cell Biochem Suppl,Suppl 35: 78-83, 2000.

[10] Robertson, K. D. and Jones, P. A. DNA methylation: past,present and future directions. Carcinogenesis, 21: 461-467,2000.

[11] Momparler, R. L. Cancer epigenetics. Oncogene, 22: 6479-6483, 2003.

[12] Ehrlich, M. DNA hypomethylation, cancer, the immunodefi-ciency, centromeric region instability, facial anomalies syn-drome and chromosomal rearrangements. J Nutr, 132: 2424S-2429S, 2002.

[13] Umbricht, C. B., Evron, E., Gabrielson, E., Ferguson, A.,Marks, J., and Sukumar, S. Hypermethylation of 14-3-3 sigma(stratifin) is an early event in breast cancer. Oncogene, 20:3348-3353, 2001.

[14] Evron, E., Dooley, W. C., Umbricht, C. B., Rosenthal, D.,Sacchi, N., Gabrielson, E., Soito, A. B., Hung, D. T., Ljung,B., Davidson, N. E., and Sukumar, S. Detection of breast can-cer cells in ductal lavage fluid by methylation-specific PCR.Lancet, 357: 1335-1336, 2001.

[15] Evron, E., Umbricht, C. B., Korz, D., Raman, V., Loeb, D. M.,Niranjan, B., Buluwela, L., Weitzman, S. A., Marks, J., andSukumar, S. Loss of cyclin D2 expression in the majority ofbreast cancers is associated with promoter hypermethylation.Cancer Res, 61: 2782-2787, 2001.

[16] Agostini, M., Pucciarelli, S., Enzo, M. V., Del Bianco, P.,Briarava, M., Bedin, C., Maretto, I., Friso, M. L., Lonardi, S.,Mescoli, C., Toppan, P., Urso, E., and Nitti, D. CirculatingCell-Free DNA: A Promising Marker of Pathologic TumorResponse in Rectal Cancer Patients Receiving PreoperativeChemoradiotherapy. Ann Surg Oncol.

[17] Taback, B. and Hoon, D. S. Circulating nucleic acids andproteomics of plasma/serum: clinical utility. Ann N Y AcadSci, 1022: 1-8, 2004.

[18] Taback, B. and Hoon, D. S. Circulating nucleic acids in plasmaand serum: past, present and future. Curr Opin Mol Ther, 6:273-278, 2004.

[19] Fujimoto, A., O’Day, S. J., Taback, B., Elashoff, D., andHoon,D. S. Allelic imbalance on 12q22-23 in serum circulatingDNAof melanoma patients predicts disease outcome. Cancer Res,64: 4085-4088, 2004.

[20] Gormally, E., Caboux, E., Vineis, P., and Hainaut, P. Circulat-ing free DNA in plasma or serum as biomarker of carcinogen-esis: practical aspects and biological significance. Mutat Res,635: 105-117, 2007.

AUTH

OR

COPY

98 M. Agostini et al. / Breast cancer and circulating cell-free DNA

[21] Jahr, S., Hentze, H., Englisch, S., Hardt, D., Fackelmayer, F.O., Hesch, R. D., and Knippers, R. DNA fragments in theblood plasma of cancer patients: quantitations and evidencefor their origin from apoptotic and necrotic cells. Cancer Res,61: 1659-1665, 2001.

[22] Giacona, M. B., Ruben, G. C., Iczkowski, K. A., Roos, T.B., Porter, D. M., and Sorenson, G. D. Cell-free DNA inhuman blood plasma: length measurements in patients withpancreatic cancer and healthy controls. Pancreas, 17: 89-97,1998.

[23] Wang, B. G., Huang, H. Y., Chen, Y. C., Bristow, R. E.,Kassauei, K., Cheng, C. C., Roden, R., Sokoll, L. J., Chan, D.W., and Shih Ie, M. Increased plasma DNA integrity in cancerpatients. Cancer Res, 63: 3966-3968, 2003.

[24] Umetani, N., Kim, J., Hiramatsu, S., Reber, H. A., Hines, O.J., Bilchik, A. J., and Hoon, D. S. Increased integrity of freecirculating DNA in sera of patients with colorectal or peri-ampullary cancer: direct quantitative PCR for ALU repeats.Clin Chem, 52: 1062-1069, 2006.

[25] Umetani, N., Giuliano, A. E., Hiramatsu, S. H., Amersi, F.,Nakagawa, T., Martino, S., and Hoon, D. S. Prediction ofbreast tumor progression by integrity of free circulating DNAin serum. J Clin Oncol, 24: 4270-4276, 2006.

[26] Jiang, W. W., Zahurak, M., Goldenberg, D., Milman, Y., Park,H. L., Westra, W. H., Koch, W., Sidransky, D., and Califano,J. Increased plasma DNA integrity index in head and neckcancer patients. Int J Cancer, 119: 2673-2676, 2006.

[27] Huang, Z. H., Li, L. H., and Hua, D. Quantitative analysis ofplasma circulating DNA at diagnosis and during follow-up ofbreast cancer patients. Cancer Lett, 243: 64-70, 2006.

[28] Hoque, M. O., Feng, Q., Toure, P., Dem, A., Critchlow, C. W.,Hawes, S. E., Wood, T., Jeronimo, C., Rosenbaum, E., Stern,J., Yu, M., Trink, B., Kiviat, N. B., and Sidransky, D. Detectionof aberrant methylation of four genes in plasma DNA for thedetection of breast cancer. J Clin Oncol, 24: 4262-4269, 2006.

[29] Dulaimi, E., Hillinck, J., Ibanez de Caceres, I., Al-Saleem, T.,and Cairns, P. Tumor suppressor gene promoter hypermethy-lation in serum of breast cancer patients. Clin Cancer Res, 10:6189-6193, 2004.

[30] Veeck, J., Niederacher, D., An, H., Klopocki, E., Wiesmann,F., Betz, B., Galm, O., Camara, O., Durst, M., Kristiansen, G.,Huszka, C., Knuchel, R., and Dahl, E. Aberrant methylation ofthe Wnt antagonist SFRP1 in breast cancer is associated withunfavourable prognosis. Oncogene, 25: 3479-3488, 2006.

[31] Eads, C. A., Danenberg, K. D., Kawakami, K., Saltz, L. B.,Blake, C., Shibata, D., Danenberg, P. V., and Laird, P. W.MethyLight: a high-throughput assay tomeasure DNA methy-lation. Nucleic Acids Res, 28: E32, 2000.

[32] Zanetti-Dallenbach, R. A., Schmid, S., Wight, E., Holzgreve,W., Ladewing, A., Hahn, S., and Zhong, X. Y. Levels of cir-culating cell-free serum DNA in benign and malignant breastlesions. Int J Biol Markers, 22: 95-99, 2007.

[33] Holdenrieder, S., Nagel, D., Schalhorn, A., Heinemann, V.,Wilkowski, R., von Pawel, J., Raith, H., Feldmann, K., Kre-mer, A. E., Muller, S., Geiger, S., Hamann, G. F., Seidel, D.,and Stieber, P. Clinical relevance of circulating nucleosomesin cancer. Ann N Y Acad Sci, 1137: 180-189, 2008.

[34] Holdenrieder, S. and Stieber, P. Clinical use of circulatingnucleosomes. Crit Rev Clin Lab Sci, 46: 1-24, 2009.

[35] Kamat, A. A., Sood, A. K., Dang, D., Gershenson, D. M.,Simpson, J. L., and Bischoff, F. Z. Quantification of totalplasma cell-free DNA in ovarian cancer using real-time PCR.Ann N Y Acad Sci, 1075: 230-234, 2006.

[36] Stroun, M., Maurice, P., Vasioukhin, V., Lyautey, J., Lederrey,C., Lefort, F., Rossier, A., Chen, X. Q., and Anker, P. Theorigin and mechanism of circulating DNA. Ann N Y AcadSci, 906: 161-168, 2000.

[37] Stroun, M., Lyautey, J., Lederrey, C., Olson-Sand, A., andAnker, P. About the possible origin and mechanism of cir-culating DNA apoptosis and active DNA release. Clin ChimActa, 313: 139-142, 2001.

[38] Stroun, M., Lyautey, J., Lederrey, C., Mulcahy, H. E., andAnker, P. Alu repeat sequences are present in increased pro-portions compared to a unique gene in plasma/serum DNA:evidence for a preferential release from viable cells? Ann NY Acad Sci, 945: 258-264, 2001.

[39] Goessl, C., Krause, H., Muller, M., Heicappell, R., Schrader,M., Sachsinger, J., and Miller, K. Fluorescent methylation-specific polymerase chain reaction for DNA-based detectionof prostate cancer in bodilyfluids. Cancer Res, 60: 5941-5945,2000.

[40] Wong, I. H., Lo, Y. M., Zhang, J., Liew, C. T., Ng, M. H.,Wong, N., Lai, P. B., Lau, W. Y., Hjelm, N. M., and Johnson,P. J. Detection of aberrant p16 methylation in the plasma andserum of liver cancer patients. Cancer Res, 59: 71-73, 1999.

[41] Silva, J. M., Dominguez, G., Villanueva, M. J., Gonzalez, R.,Garcia, J. M., Corbacho, C., Provencio, M., Espana, P., andBonilla, F. Aberrant DNA methylation of the p16INK4a genein plasma DNA of breast cancer patients. Br J Cancer, 80:1262-1264, 1999.

[42] Yazici, H., Terry, M. B., Cho, Y. H., Senie, R. T., Liao,Y., Andrulis, I., and Santella, R. M. Aberrant methylationof RASSF1A in plasma DNA before breast cancer diagno-sis in the Breast Cancer Family Registry. Cancer EpidemiolBiomarkers Prev, 18: 2723-2725, 2009.

[43] Dammann, R., Schagdarsurengin, U., Strunnikova, M.,Rastetter, M., Seidel, C., Liu, L., Tommasi, S., and Pfeifer,G. P. Epigenetic inactivation of the Ras-association domainfamily 1 (RASSF1A) gene and its function in human carcino-genesis. Histol Histopathol, 18: 665-677, 2003.

[44] Agathanggelou, A., Cooper, W. N., and Latif, F. Role of theRas-association domain family 1 tumor suppressor gene inhuman cancers. Cancer Res, 65: 3497-3508, 2005.

[45] Mimori, K., Shiraishi, T., Mashino, K., Sonoda, H., Ya-mashita, K., Yoshinaga, K., Masuda, T., Utsunomiya, T.,Alonso, M. A., Inoue, H., and Mori, M. MAL gene expres-sion in esophageal cancer suppresses motility, invasion and tu-morigenicity and enhances apoptosis through the Fas pathway.Oncogene, 22: 3463-3471, 2003.

[46] Lo, P. K., Mehrotra, J., D’Costa, A., Fackler, M. J., Garrett-Mayer, E., Argani, P., and Sukumar, S. Epigenetic suppressionof secreted frizzled related protein 1 (SFRP1) expression inhuman breast cancer. Cancer Biol Ther, 5: 281-286, 2006.

[47] Martens, J. W., Margossian, A. L., Schmitt, M., Foekens, J.,and Harbeck, N. DNA methylation as a biomarker in breastcancer. Future Oncol, 5: 1245-1256, 2009.

[48] Burbee, D. G., Forgacs, E., Zochbauer-Muller, S., Shivaku-mar, L., Fong, K., Gao, B., Randle, D., Kondo, M., Virmani,A., Bader, S., Sekido, Y., Latif, F., Milchgrub, S., Toyooka, S.,Gazdar, A. F., Lerman, M. I., Zabarovsky, E., White, M., andMinna, J. D. Epigenetic inactivation of RASSF1A in lung andbreast cancers and malignant phenotype suppression. J NatlCancer Inst, 93: 691-699, 2001.

AUTH

OR

COPY