A ten markers panel provides a more accurate and complete microsatellite instability analysis in...

Cancer Biomarkers 6 (2009/2010) 49–61 49DOI 10.3233/CBM-2009-0118IOS Press

A ten markers panel provides a more accurateand complete microsatellite instabilityanalysis in mismatch repair-deficientcolorectal tumors

Marco Agostinia,1,∗, Maria Vittoria Enzoa,1, Luca Morandib,1, Chiara Bedina, Silvia Pizzinia,Silvia Masonc, Roberta Bertorelled, Emanuele Ursoa, Claudia Mescolie, Mario Lisef ,Salvatore Pucciarellia and Donato NittiaaII Section of Surgery Clinic, Department Oncologic and Surgical Sciences, University of Padova, ItalybDepartment of Ematology and Oncological Science “L. & A. Seragnoli” Section of Pathology, Bellaria Hospital,University of Bologna, ItalycAB ANALITICA s.r.l. Padova, ItalydIstituto Oncologico Veneto, IRCCS, Padova, ItalyeSection of Pathology,University of Padova ItalyfSurgical Oncology, Regional Oncologic Center (CRO), Aviano, Italy

Abstract. Tumour microsatellite instability (MSI) is useful in identifying patients with hereditary non-polyposis colorectal cancer(HNPCC) with defective DNA mismatch repair (MMR) genes. A reference Bethesda panel has limitations resulting from theinclusion of dinucleotide markers, which are less sensitive and specific for detection of tumours with MMR deficiencies. Wedeveloped a multiplex PCR assay with additional four mononucleotide markers and one dinucleotide marker (NR-21, NR-24,BAT-40, TGF-BetaR and D18S58) for a rapid and proper classification of MSI-H, MSI-L and MSS colorectal cancers. Twotetranucleotide markers were added to identify sample mix-ups and/or contamination. Results: all the 44 cases test cases were inagreement with previous classification except for three cases: one case MSI-H-Bethesda unstable only for dinucleotides markersshifted to MSI-L category and two cases MSI-L-Bethesda unstable for mononucleotide markers shifted to MSI-H category.Immunohistochemistry analysis revealed that these two MSI-H cases did not expressed hMLH1 and they were found to bemethylated at the MLH1 promoter, while the first one that shifted to MSI-L showed MMR protein expression. Conclusion: acomplete panel of ten markers including four dinucleotide and six mononucleotide microsatellites allows accurate evaluation oftumor MSI status.

Keywords: Microsatellite instability, colorectal cancer, mononuclotide repeats

1. Introduction

Colorectal carcinoma (CRC) is the second leadingcause of cancer-related deaths in the Western world.

1The first three authors contributed equally to this article.∗Corresponding author: Marco Agostini PhD, Clinica Chirurgica

II, Dipartimento di Scienze Oncologiche e Chirurgiche, Universitadi Padova, Via Giustiniani 2, 35128 Padova, Italy. Tel.: +39 0498214374; Fax: +39 049 651891; E-mail: m.agostini@unipd.it.

There are ∼110,000 new cases and 50,000 deaths dueto the disease per year in the U.S. A small fraction ofpatients (∼5%) have a hereditary colorectal cancer syn-drome [1]. One of these syndromes, hereditary non-polyposis colorectal cancer (HNPCC), accounts for 1%to 2% of all colorectal carcinomas [2]. In most HN-PCC colorectal tumors, MSI has been shown to resultfrom defects in DNA mismatch repair. Mutations in thehMLH1 or hMSH2 genes are the most common defects

ISSN 1574-0153/09,10/$27.50 2009/2010 – IOS Press and the authors. All rights reserved

50 M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis

in these families; in equal proportions, these make upabout 94% of the germline mutations detected. About10% to 15% of sporadic colorectal cancers also exhibitMSI, and loss of expression of one or more of the MMRproteins has been found in these tumors [3]. Most spo-radic MSI-positive tumors lack expression of hMLH1,as the result of promoter methylation [4]. MSI analy-sis is a useful tool in identifying patients with HNPCCand sporadic CRC with defective DNA mismatch repair(MMR) genes.

In 1997, a reference panel of 5 markers was sug-gested for MSI testing by the National Cancer Insti-tute aiming to standardize the test [5]. Based on thenumber of microsatellites found to be unstable, tumorsare classified as stable, with low instability (MSI-L)and with high instability (MSI-H). However this pan-el has limitations resulting from the inclusion of dinu-cleotide markers, which are less sensitive and specif-ic for the detection of tumors with MMR deficienciescompared to other markers that are currently available.Dinucleotide repeats in the Bethesda panel (D2S123,D17S250 and D5S546) generally show instability inonly 60%–80% of MSI-H tumors [6]. Dinucleotide re-peat are also highly polymorphic and their use in MSIscreening of tumor DNA requires the analysis of cor-responding germline DNA. The interpretation of sizealterations in dinucleotide repeats is difficult and canlead to misclassification [7]. Two dinucleotide markers(using the Bethesda reference panel) are observed tobe unstable in the absence of BAT-26 deletions [8,9].Hoang JM et al. [9] used only the analysis of mononu-cleotide repeats BAT-25 and BAT-26 to establish MSIstatus without reference to the germline DNA, becausethese markers are quasi-monomorphic. However sta-bility in BAT-26 locus is strongly evocative for the pres-ence of wide intragenic deletion in the MSH2 gene [10]and polymorphic BAT-25 and BAT-26 alleles have beenidentified, respectively, in 18.4% and 12.6% of Afro-Americans and in a small percentage of Caucasian in-dividuals [11]. Additionally a panel of five quasi-monomorphic mononucleotide repeats that make un-necessary the analysis of corresponding germline DNAwas proposed by Suraweera et al. [12] for the detec-tion of high-frequency microsatellite instability (MSI)in colorectal cancer within a Caucasian study popula-tion for which matching normal DNA would not beavailable without microdissection.

In this work we developed a ten markers basedmethod to evaluate accurately MSI status following theBethesda guidelines [13].

2. Patients and methods

2.1. DNA sample

Germline and tumor DNA was obtained from aprospective group of 44 CRC patients, who underwentsurgery at the University Hospital of Padua (Italy) be-tween 2003–2004. Lymphocytes from peripheral bloodand tumor biopsy histologically confirmed were usedfor DNA extraction. Patients gave their formal in-formed consent for the study according to the Helsin-ki declaration. MSI status was previously determinedfollowing the Bethesda panel for MSI testing.

DNA was extracted from fresh-frozen tumor tissuesand from the EDTA-preserved blood samples with theuse of standard methods (QiAamp DNA Mini Kit, Qi-agen, Hilden, Germany). The histologic features of thetumor were reevaluated by analysing of the paraffin-embedded tissue block. The proportion of tumor cellsin the tumor tissues used for the extraction of DNAexceeded 50 percent in all cases.

2.2. Microsatellite instability testing

To determine the microsatellite instability of the tu-mor, we ascertained the genotypes using twelve poly-morphic markers (BAT25, BAT26, BAT40, D2S123,D5S346, D18S69, D17S250, NR21, NR24, TGF-Beta,TPOX, TH01) in tumor tissue and unaffected tissue(Table 1) [5,12,13]. The two tetranucleotide markers(TPOX, TH01) have been selected for their high levelof polymorphism and low degree of MSI to uniquelyidentify samples, confirming that tumor and matchingnormal samples are from the same individual. Fourmultiplex amplifications were performed in a total vol-ume of 25 µl using 1–5 ng of genomic DNA in AB Su-perTaq buffer (AB Analitica, Padova, Italy), containing100 mM of Tris HCl (pH 8.3), 500 mM of KCl, using1.5 mM of MgCl2, 200 µM of each four dNTPs and1 U of AB SuperTaq DNA Polymerase (AB Analiti-ca, Padua, Italy). 32 cycles of touch-down PCR (TD-PCR) were repeated according to the following proto-col: initial denaturation and AB SuperTaq activation at95◦C for 8 minutes, ten cycles at 94 ◦C for 30 seconds,60◦C decreasing to a final temperature of 55◦C, at arate of 0.5◦C per cycle, for 45 seconds (annealing), and72◦C for 30 seconds (extension). 22 cycles of con-ventional PCR were then performed using an identicalcycling profile and a constant annealing temperature of55◦C. Incubation at 72◦C for 10 min (final extension)followed the cycling protocol.

M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis 51

Table 1List of primers used in this study and markers location in four multiplex reactions

Name Gene Repeat type and GenBank Primer sequence Average PCR Multiplex groupchromosomal number 5’ to 3’ fragment size and primers

location (bp) concentration

TGF-βRII TGF-ß-RII (A)n 3p22 BC040499 Fw: FAM –ATGCTGCTTCTCCAAAGTGCATTARw:GCACTCATCAGAGCTACAGGAACA

80–95 A: 0.25 µM

NR21 SLC7A8 (T)n 14q11.2 XM 033393 Fw: NED –TAAATGTATGTCTCCCCTGGRw: ATTCCTACTCCGCATTCACA

102–108 A: 0.5 µM

NR24 ZNF-2 (T)n 2q11.2 X60152 Fw: FAM –CCATTGCTGAATTTTACCTCRw: ATTGTGCCATTGCATTCCAA

130–145 A: 0.5 µM

BAT40 HUMBHSD (T)n 1p12–13.3 M38180 Fw: HEX –ATTAACTTCCTACACCACAACRw: GTAGAGCAAGACCACCTTG

120–130 B: 0.5 µM

D18S58 CYB5A (CA)n 18q23 Z16735 Fw: HEX –ATCCCTTAGGAGGCAGGAAARw: CTCCCGGCTGGTTTTATTTA

144–160 B: 0.5 µM

D2S123 hMSH2 (CA)n 2p15 Z16551 Fw: FAM –AAACAGGATGCCTGCCTTTARw: GGACTTTCCACCTATGGGAC

210–230 B: 0.5 µM

BAT25 c-kit (T)n 4q11–q12 L04143 Fw: HEX –TCGCCTCCAAGAATGTAAGTRw: TCTGCATTTTAACTATGGCTC

120–128 C: 0.5 µM

D17S250 LASP1 (CA)n 17q12 X54562 Fw: FAM –AAAAGGAAGAATCAAATAGACARw:GCTGGCCATATATATATTTAAACC

140–154 C: 0.5 µM

TH01 TH01 (TCAT)n 11p15–15.5 D00269 Fw: HEX –GTGGGCTGAAAAGCTCCCGATTATRw:GTGATTCCCATTGGCCTGTTCCTC

154–178 C: 0.5 µM

D5S346 APC (CA)n 5q22–23 G15921.1 Fw: FAM –ACTCACTCTAGTGATAAATCGRw:AGCAGATAAGACAGTATTACTAGTT

110–130 D: 0.5 µM

BAT26 hMSH2 (A)n 2p22–p21 U41210 Fw: FAM –ATGAAATTGGATATTGCAGCAGTCRw:AGCTCCTTTCTAAGCCTTCTTCACT

205–225 D: 0.5 µM

TPOX TPOX (TGAA)n 2p23–2pter M68651 Fw: HEX –ACTGGCACAGAACAGGCACTTAGGRw:GGAGGAACTGGGAACCACACAGGT

232–248 D: 0.5 µM

2.2.1. Fragment Analysis and statistical evaluation1 µl of PCR products were mixed with 12 µl of for-

mamide and 0.2 µl of LIZ500 standard length (Ap-plied Biosystems, Milan). DNA fragments were sep-arated on ABI PRISM 3130 Genetic Analyzer (Ap-plied Biosystems) using Filter Set D. Size, height, andprofiles of microsatellite peaks were analyzed usingthe GeneMapper 3.7 software (Applied Biosystems).When the analysis was performed using the NCI panelof microsatellites, presence of peaks in the fluorescenceprofile of the amplified tumor DNA that were absentin a corresponding profile derived from germline DNA

was interpreted as microsatellite instability. For the in-stability analysis using the pentaplex panel of mononu-cleotide markers, microsatellites were considered un-stable if PCR fragments showed deletions of at least 3bp at a given locus.

2.2.2. ImmunohistochemistrySlides 7 through 9 were selected for immunohisto-

chemical staining with antibodies to hMLH1 (Pharmin-gen), hMSH2 (Oncogene Research Products), hMSH6(Transduction Laboratories) and hPMS2 (Pharmingen)as previously described [14].

52 M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis

Fig. 1. MSI fragment Analysis between normal and tumor DNA: NR21, BAT40, D2S123, D18S58 and BAT26 showed microsatellite instabilityfor this case.

2.2.3. Mutation analysisMLH1, MSH2 andMSH6 mutations analyses were

carried out in all MSI-H cases and in cases with negativeIHC for MMR proteins, by bi-directional sequencingon an automatic ABI PRISM 3130 DNA analyser (Ap-plera, Milan). Point mutations of a gene were searchedby polymerase chain reactions of genomic DNA withexon-specific primer pairs and bidirectional sequenc-ing.

The purified PCR product was used as template incycle sequencing with the BigDye Terminator v1.1Kit (Applied Biosystems, Foster City, CA). The reac-tion mix consisted of 1X Big Dye Terminator Premix,1X Sequencing Buffer, 3.2 nM of primer and 2 µl ofcleaned template in a 20 µl total volume. The reactionswere run on a termocycler 9700 GeneAmp PCR Sys-tem (Applied Biosystems) according to the following

protocol: one cycle of 95◦C for 1 minute; 25 cyclesof 95◦C for 10 seconds, 50◦C for 5 seconds, 60◦C for4 minutes. The sequencing reactions were purified inEDTA/ethanol precipitated and run on an ABI3130 Ge-netic Analyser (Applied Biosystems). Sequencing datawere analysed using SeqScape v2.5.

Mutations without clear indication of pathogenicitywere classified as variant of uncertain significance.

2.2.4. Methylation analysis of MLH1 promoterThe methylation status of the CpG islands within the

MLH1 gene promoter was analyzed by methylation-specific PCR (MSP) with DNA modified by sodiumbisulfite. The DNA of individuals with high microsatel-lite instability and negative IHC was extracted fromperipheral blood lymphocytes (PBLs) and from tumor

M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis 53

fresh tissue using QIAmp DNA mini kit (QIAGEN,Germany).

Genomic DNA was modified with sodium bisul-fite following the manufacturer’s protocol (BIRD s.r.l.Varese, Italy). To summarise, genomic or tumor DNAfrom the sample (1 µg) was diluted in 10 µl of distilledwater and then denatured for 15 min at 37◦C. 100 µlof sodium bisulfite at pH 5.0 were then added, and thesamples were incubated at 50◦C for 18–20 h. Aftertreatment, the DNA was desulfonated and purified us-ing the DNA Clean-Up System (BIRD s.r.l., Varese,Italy), following the manufacturer’s protocol. The sam-ples were resuspended in 15 µl of distilled buffer.

The modified DNA was then subject to MSP usingprimer pairs engineered to amplify either methylatedor unmethylated DNA.

MSP was performed with primers and conditionsdescribed in the AMPLI-Carcinoma Colon-retto kit(BIRD s.r.l., Varese, Italy).

The PCR was performed in 25 µl reaction volumescontaining 1X MIX-PCR Methylated or Unmethylat-ed, 100–150 ng of modified DNA, and 1.25 units ofTaq Polymerase. For both methylated and unmethylat-ed PCR the thermocycler conditions were as follows:95◦C for 10 min, 40 cycles of 95◦C for 40 seconds,60◦C for 40 seconds, 72◦C for 40 seconds, and finalextension of 72◦C for 10 min. The appropriate positiveand negative reference samples were included.

The PCR products were subject to gel electrophore-sis through a 3.5% agarose gel, stained with ethidiumbromide, and then visualized with UV illumination us-ing a digital imaging system (Minibis biolmaging sys-tem) (Fig. 3).

2.3. High-Resolution Melting Analysis (HRMA) andDNA Sequencing of BRAF V600E (1799 T > A)

HRMA, was performed to identify BRAF V600Emutation. Primer sequences for BRAF in HRMA were:FWD (5’ ATG AAG ACC TCA CAG TAA AAA TAG3’) REV (5’ GAC AAC TGT TCA AAC TGA TGG 3’)for the shorter 88 bp amplicon.

30 ng of tumoral DNA was amplified in a final vol-ume of 20 µL by using the following: 1X AmpliTaqGold PCR Master Mix (Applied Biosystems) which in-cluded 2.5 mM of MgCl2 and 0.2 mM of each dNTPs,1.5 µM of SYTO9 and 0.3 µM of each primer. RealTime PCR was performed on the 7500 Fast Real TimePCR System (Applied Biosystems) to an initial denat-uration at 95◦C for 10 minutes followed by 40 cyclesof 95◦C for 15 seconds, 58◦C for 15 seconds, 60◦C for

15 seconds and then a melt from 60 to 95◦C rising at0.3◦C per second. The data were analized with HRMv2.0 software (Applied Biosystem).

To confirm HRMA results, sequencing analysis wasalso performed in all samples. Primer sequences forBRAF were as follows: BRAF-F (5’ TGC TTG CTCTGA TAG GAA AAT GAG A 3’), BRAF-R (5’ GGGCCA AAA ATT TAA TCA GTG GA 3’), that generat-ed fragment lengths of 219 base pairs. Samples wereamplified according to the following conditions; 30 ngof tumoral DNA, 0.625 U of GoTaq DNA Polymerase(Promega), 1X Colorless Reaction Buffer (Promega)which included.5 mM of MgCl2, 0.2 mM of eachdNTPs, 0.2 µM of primers and water to a final volumeof 25 µL. PCR cycling was performed on the 9700 Ge-neAmp PCR System (Applied Biosystems). One stepof 95◦C for 5 minutes; 35 cycles of 95◦C for 1 minute,55◦C for 1 minute, 72◦C for 1 minute and a final ex-tension of 72◦C for 5 minutes. Then the PCR productswere purified with ExoSAP-IT (GE Healthcare, Buck-inghamshire, England) according to the manufacturer’sinstructions. The purified PCR product was used astemplate in the cycle sequencing as mentioned previ-ously mentioned.

3. Data analysis

CRC were classified using the new panel and com-pared with previous results obtained using the Bethes-da panel [13], combined with immunohistochemical(IHC) analysis of MLH1, MSH2, MSH6.

4. Results

Six mononucleotide markers, BAT-25, BAT-26,BAT-40, NR-21, NR-24, TGF-βRII, four dinucleotidemarkers D5S346, D2S123, D18S58 and D17S250, andtwo tetranucleotide markers (TPOX and TH01) werecoamplified in four multiplex PCR mixes (see Table 1for details). The PCR products was analyzed for sizein an automated DNA sequencer (ABI PRISM 3130Genetic Analyzer, Applied Biosystems). Nonspecificbands within the 80–300 bp range were not observed,thus allowing accurate identification of the 12 markers.Figure 1 shows an example of the fluorescent peaksobserved for each marker in germline DNA or tumorDNA. The size of PCR products and the correspondingfluorescent labels were chosen to allow the simultane-ous analysis of normal-sized alleles, with the smaller-

54 M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis

Fig. 2. MLH1 immunohistochemistry. A) normal expression of hMLH1. B) not expression of hMLH1.

sized alleles containing deletions typically seen in MSI-H tumors. Along with the smaller alleles, visible asleft-shifted peaks, most MSI-H primary tumors alsoshowed normal sized alleles presumably due to contam-inating non-cancer cells or tumor heterogeneity. Clini-cal data of the 44 cases enrolled in this study are sum-marized in Table 2. All patients found to have tumorswith microsatellite instability and/or lack of MSH2 orMLH1 protein expression underwent germline genetictesting for MSH2 and MLH1 by.sequencing analysis.

The 44 cases test cases were in agreement with pre-vious classification except for three cases: one caseMSI-H-Bethesda unstable only for dinucleotides mark-ers shifted to MSI-L category and two cases MSI-L-Bethesda unstable for mononucleotide markers shiftedto MSI-H category.

MSI data for MSI-H cases and three discordant casesare summarized in detail in Table 3. No mutations werefound in MLH1 and MSH2 genes for these discordantcases but the two which shifted to MSI-H did not ex-

M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis 55

Table 2Clinical data of the 44 cases enrolled in this study

Patient features NCI MSS NCI MSI-L NCI MSI-H new profile for MSS new profile MSI-L new profile MSI-H

Number of patients 20 13 11 20 12 12mean age 50 72 65 50 73 64gendermale 10 8 4 10 7 5female 10 5 7 10 5 7localizationright 8 5 8 8 3 10left 6 4 2 6 4 2rectal 5 3 1 5 4 0colona 1 1 0 1 1 0grading1 5 2 1 5 3 02 10 8 6 10 7 73 4 1 3 4 1 34 0 0 0 0 0 0n.e.b 1 2 1 1 1 2stage (pTNM)I 5 5 4 5 5 4II 7 4 3 7 3 4III 3 2 4 3 2 4IV 4 1 0 4 1 0n.e.b 1 1 0 1 1 0histopathologymucinous 2 1 1 2 0 2mucinous aspects 2 2 3 2 2 3not mucinous 16 10 7 16 10 7familiar historyc

HNPCC 0 0 3 0 0 3s-HNPCC 6 0 3 6 0 3familiarity 0 2 4 0 2 4sporadic 14 11 1 14 10 2

alocalization not specified; bn.e.: not estimable; cHNPCC: patients with Hereditary Nonpolyposis Colorectal Cancer who followAmsterdam Clinical Criteria II; s-HNPCC (suspect HNPCC) patients with family history for CRC but not satisfy all AmsterdamClinical Criteria II; familiarity: patients with CRC family history; sporadic: patients with no family history for CRC.

press MLH1 protein. All cases MSI-H showed normalexpression of hPMS2.

5. Germline epimutation of MLH1 is associatedwith monoallelic expression

DNA from tumor and peripheral blood lymphocyteswas analyzed using methylation-specific PCR (MSP),to determine hypermethylation status in individualsMSI-H without germline mutation in MLH1 and lack-ing IHC protein expression.

Out of a total 12 cases of MSI-H, we analyzed 7 cas-es who matched with the selection including the twocases that were MSI-L with bethesda panel and thenwere shifted to MSI-H with the new panel. The methy-lation results are summarized in Table 4. All the pa-tients presented somatic hypermethylation of promoterof MLH1 in tumor cells but only two patients showedgerminal hypermethylation. One of these carriers was

a suspected subject of HNPCC, the other patient hadno family history and showed high instability only withthe new panel. (Fig. 3)

5.1. BRAF V600E point mutation

BRAF is mutually exclusive in MSI tumors. Knownmutations in the BRAF V600E oncogene were ana-lyzed in this study. Three out of 12 analyzed MSI tu-mors displayed the BRAF V600E point mutation. Casen◦ 4756 had somatic hypermethylation of MLH1 pro-moter, whereas case n◦ 4676 had germline and somatichypermethylation. (Table 4)

6. Discussion

The importance of nucleotide and microsatellite in-stability in carcinogenesis is widely accepted. Further-more, these aspects today provide additional informa-

56 M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis

Tabl

e3

MSI

-Hpa

tient

sch

arac

teri

stic

san

dm

olec

ular

data

.E

xp=

prot

ein

expr

esse

d;no

exp

=pr

otei

nno

tex

pres

sed;

low

exp

=lo

wpr

otei

nex

pres

sion

;pa

t=

path

olog

ic;

u.v.

=un

clas

sifie

dva

rian

t;n.

e.=

not

estim

able

;H

NPC

C:

patie

nts

with

Her

edita

ryN

onpo

lypo

sis

Col

orec

tal

Can

cer

who

follo

wA

mst

erda

mC

linic

alC

rite

ria

II;

s-H

NPC

C(s

uspe

ctH

NPC

C)

patie

nts

with

fam

ilyhi

stor

yfo

rC

RC

butn

otsa

tisfy

allA

mst

erda

mC

linic

alC

rite

ria

II;f

amili

arity

:pa

tient

sw

ithC

RC

fam

ilyhi

stor

y;sp

orad

ic:

patie

nts

with

nofa

mily

hist

ory

for

CR

C

IDA

geFa

mili

arC

olon

Muc

inou

sM

SIM

SIB

ethe

sda

pane

lN

ewpa

nel

IHC

Mut

atio

nB

ethe

sda

new

hist

ory

site

pane

lpa

nel

D5S

346

D2S

123

D17

S250

BA

T26

BA

T25

NR

21N

R24

BA

T40

TG

FB-r

D18

S58

hMSH

2hM

SH6

hML

H1

hML

H1

hMSH

2

4594

72fa

mili

arity

left

noH

HI

II

IS

SS

SS

SE

xpn.

e.E

xpno

no45

7051

HN

PCC

righ

tye

sH

HI

IS

II

II

II

IN

oex

pN

oex

pE

xpno

u.v.

7185

85fa

mili

arity

righ

tye

sH

HI

SI

II

II

IS

IN

oex

pN

oex

pL

owex

pno

no46

7674

S-H

NPC

Cri

ght

noH

HS

SI

II

II

II

IE

xpE

xpN

oex

pno

no47

3660

fam

iliar

ityri

ght

noH

HS

II

IS

II

SS

Ilo

wE

xpN

oex

pno

no47

4192

S-H

NPC

Cri

ght

noH

HI

II

II

II

II

IE

xpE

xpN

oex

pno

no47

5653

fam

iliar

ityri

ght

yes

HH

SI

II

IS

SS

SS

Exp

Exp

No

exp

nono

7449

38H

NPC

Cle

ftno

HH

II

SS

II

SI

SI

No

exp

No

exp

Exp

nopa

t47

6639

HN

PCC

righ

tno

HH

SS

II

II

II

SS

Low

exp

Exp

Exp

pat.

no47

7482

S-H

NPC

Cri

ght

yes

HH

II

SI

II

II

SS

No

exp

No

exp

low

exp

u.v.

no72

8864

spor

adic

rect

umno

HL

SI

IS

SS

SS

SS

Exp

Exp

Exp

nono

4571

70sp

orad

icri

ght

yes

LH

SS

SI

SS

SI

II

Exp

Exp

No

exp

n.e.

n.e.

4722

55sp

orad

icri

ght

noL

HS

SS

IS

II

SS

IE

xpE

xpN

oex

pn.

e.n.

e.

M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis 57

Table 4Molecular epigenetic analysis in MSI-H patients

ID MSI-Beth MSI new panel IHC hMLH1 somatic MLH1 germinal MLH1 Mut.BRAF V600Emethylation methylation

7185 H H low yes no wt4676 H H Not expressed yes yes 1799T>A het4736 H H Not expressed yes no wt4741 H H Not expressed yes no wt4756 H H Not expressed yes no 1799T>A het4571 L H Not expressed yes no wt4722 L H Not expressed yes yes wt

M UM 4571

M UM 4722

M UM 7185

M UM4756

M UM Negative

Fig. 3. hMLH1 germ line promoter methylation. M = methylated, UM = unmethylated.

tion, beyond clinical facts, to perform an earlier di-agnosis of hereditary nonpolyposis colorectal cancer(HNPCC).

Thus it is of great importance to develop efficientscreening protocols to make specific and early diagno-sis possible [15].

The test concerning microsatellite instability is per-formed by comparing DNA extracted from normal andtumoral tissue respectively through the amplification ofthe analogous sequences. The instability of those lociis assessed if the sequences differ in length.

MSI testing of tumors is a valuable tool in select-ing cases with an underlying mismatch repair defect.HNPCC-associated tumors almost invariably displaymicrosatellite instability (MSI) as a direct consequenceof impaired mismatch repair activity.

However, pathogenic sequence mutations of the mis-match repair genes fail to be identified in about onethird of cases that meet the clinical criteria for HNPCC.

Another confounding factor is the identification ofsporadic colorectal cancers that also demonstrate MSIand loss of MLH1 expression [16].

There is a large quantity of literature that describesa wide range of different results concerning the studyand analysis of microsatellites and genomic mutations.

These contradictory findings pinpoints the limits of thetumor classification according to the Bethesda paneland its limited diagnostic credibility [17,18].

In tumors classified as MSI-H the presence of mu-tations in genes MSH2 and MLH1 is very likely, whilemutations in the gene MSH6 are found in both tumorsclassified as MSI-L and MSS. This is due to the fact thatMSH6 is particularly involved in the process of repa-ration of insertions/deletions of single bases. Defectsin this gene are easily highlighted by mononucleotiderepetitions but not easily by the panel with dinucleotidemarkers [13,19–22].

In tumors classified according to MSI-H, analysesdemonstrated that when only the dinucleotidic markersare instable the evidence for a deficit in the reparationsystem could not often be confirmed.

All these observations lead to the conclusion thatdinucleotidic markers cannot be interpreted unambigu-ously. Their polymorphic character can lead to under-estimating cases with high levels of instability by mis-interpreting the instability as a loss of heterozygositythat is not connected to HNPCC. In contrast, an over-estimation can be the result of high rates of somaticmutations [19,20].

58 M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis

Nowadays, it is possible to identify false positive testresults in order to reduce costs and the stress caused byfalse diagnosis on patients and their families.

This evidence emphasize the need to extend the tra-ditional tests introducing four mononucleotidic mark-ers (NR21, NR24, BAT40, TGF-ßRII), one dinucleo-tidic marker (D18S58) and two tetranucleotidic mark-ers (TPOX, TH01). TPOX and TH01 are not useful foranalysis of microsatellite instability but help to verifycontaminations.

The aim of this study was to clarify the diagnos-tic potential of new markers panel (BAT-25, BAT-26,BAT-40, NR-21, NR-24, TGF-βRII, D5S346, D2S123,D18S58 and D17S250) and their ability to identify pa-tients with MMR defects.

A total of 44 patients were enrolled into this study.The same patients had already been tested for mi-crosatellite instability with the Bethesda panel, im-munohistochemistry and mutational analysis.

According to the Bethesda panel, 20 patients wereclassified as stable (MSS), 13 as low grade of instability(MSI-L) and 11 as high grade of instability (MSI-H).

The new panel provided the same results concerningthe group of 20 cases and thus they were classified asMSS. In contrast, the other two groups’ results werediscordant. Accordingly, 12 patients were classified asMSI-L (1 of whom was previously classified as MSI-H,for the other 11 patients accordance was found) and12 as MSH-H (2 of whom were previously classifiedas MSI-L) for the other patients accordance was foundrespectively.

The two cases previously classified as MSI-L wereupgraded to MSI-H as a result of the analysis of thenew panel. If only the markers data of Suraweera pan-el [12] were considered, only one of the two patientspreviously classified as MSI-L would shift to MSI-Hcategory (Table 3).

In contrast, the newly added markers demonstratedextensive and elevated instabilities. Accordingly, thepercentage of instability was estimated to be approxi-mately 50% respect to the previous 20 %. These twocases were regarded as sporadic on the basis of the fam-ily anamnesis and the presence of defining patterns ofHNPCC (affection of the proximal colon, and relativelyearly onset of the disease).

Although these two cases did not show any mutationin MLH1 and MSH2 genes, somatic hypermethylationof MLH1 was found in both.

More interestingly, in one of the early onset cases agermline hypermethylation was found as well.

In recent years, an aetiological role for the epige-netic inactivation of the mismatch repair genes MLH1

and MSH2 has been revealed in cases with an HNPCCphenotype but normal sequence of the mismatch repairgenes [16]. Studies have shown that hypermethylationof MLH1 is not limited to neoplastic cells. Rather, insome subjects, hypermethylation of a single allele ofMLH1 originates in the germline and is thus widespreadin normal somatic cells.

Constitutional epimutations are characterized bymethylation of a single allele of the promoter accompa-nied by transcriptional silencing of the affected allelein the normal somatic tissues, in an otherwise intactgene [23,24]. Subjects who have a germline epimuta-tion of one allele of MLH1 have a predisposition for thedevelopment of cancer in a pattern typical of hereditarynonpolyposis colorectal cancer.

They appear to confer a similar phenotype as se-quence mutations of the same gene, thus can serve asan alternative aetiological mechanism for HNPCC.

However, constitutional epimutations of MLH1are reversible during meiosis and so display non-Mendelian inheritance, in contrast to the strict autoso-mal dominant inheritance pattern associated with ge-netic mutations of the mismatch repair genes [25]. Themechanism that underlies this defect and its inheritancepattern remain to be elucidated.

Inheritance of epimutations is weak, so family his-tory is not a useful guide for screening.

Regarding the other case with tumor MLH1 expres-sion loss and without identified germline mutations orepimutations, the cause of this protein loss has not beenexplained yet. This might be due to either the pres-ence of an MLH1 genetic or an epigenetic change inthe germline which cannot be detected with currenttechniques.

This information corroborates the credibility of theclassification corrected according to the new panel.

In all the cases that presented an elevated insta-bility concerning the new markers, it is observablethat at least one of the two Bethesda-mononucleotidicmarkers shows instability. Additionally, the mark-ers have a noteworthy capacity to discriminate tu-mors with defects in MSH2 e MLH1 (BAT26) orMSH6(BAT25) [21].

As a result of the analysis with the new markers, onesample within the MSI-H group was classified as MSI-L. Here, the percentage of instability was estimated tobe lower than 30–40%. The findings were supportedand in total accordance with the fact that neither theclinical-pathological characteristics of non-polyposiscolorectal cancer nor the family history for CRC wereevident. Furthermore, the analysis of mutation of the

M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis 59

Table 5Predictive values (%) of the microsatellite panels in detecting tumors withabsence of mismatch repair protein expression

Microsatellite markers Sensitivity+ Specificity ‡ PPV § NPV ||Bethesda panel 80 91 73 94New panel 100 94 83 100

PPV = positive predictive value; NPV = negative predictive value;+Sensitivity = (high microsatellite instability [MSI-H] tumors with lackof mismatch repair protein expression/total number of tumors with lack ofmismatch repair protein expression) × 100.‡ Specificity = (non – MSI-H tumors with normal expression of mismatchrepair proteins/total tumors with normal mismatch repair protein expression)× 100.§ PPV = (MSI-H tumors with lack of mismatch repair protein expression/allMSI-H tumors) × 100.|| NPV = (non – MSI-H tumors with normal mismatch repair protein/allnon-MSI-H tumors) × 100.

two most frequently involved genes (MSH2 and MLH1)that are connected with HNPCC, did not show any ab-normalities and therefore supported the reclassificationof the case. Interestingly, the previously MSI-H classi-fied sample showed high instability only for the dinu-cleotidic markers. This result is in accordance with thelimitations mentioned above. The problems of inter-preting the genotypic profile of those markers and theirhigh level of spontaneous somatic mutations result inthe high rate of false diagnosis [19].

Furthermore, results for the sample that was clas-sified as MSI-L and revealed instability only for onesingle dinucleotidic marker could not be confirmed byanalyzing the new markers which showed contrastingstability. These results strengthen the published disser-tation that dinucleotidic markers are less specific thanthe mononucleotidic ones [26].

In particular a study conducted by Bacher describesthe markers D2S123 and D17S250 that are showed to bemost commonly modified in tumors, although normallevels of the MMR protein were found.

The 12 samples classified as MSI-L tumors accord-ing to the new set of markers, were consequently an-alyzed immunohistochemically. The results revealedthat, in all of the samples, the proteins connected to thegenes MSH2, MLH1 and MSH6 were expressed.

This confirms the relatively poor diagnostic potentialof the markers. From the 12 samples that were clas-sified as MSI-L, 7 (58%) showed instability concern-ing D2S123 and 4 (33%) showed instability concerningD17S250. Finally, one of the probes showed instabilityfor both, D2S123 and D17S250.

In the group of MSI-H the reliable mononucleo-tidic markers showed high frequency of instability onceagain with positive test results in the largest part ofthe cohort (BAT26 n = 11/12; BAT 25 n = 8/12).

In contrast, the dinucleotidic ones presented instabilityonly in a smaller part of the cases (D2S123 n = 7/12;D5S346 n = 6/12). Furthermore, the limited role ofD5S346 was confirmed.

The incidence of instability of the newly added mark-ers was much higher (NR21 and BAT40 n = 9/12,NR24 and D18S58 n = 8/12, TGFß-RII n = 4/12).

The new panel sensitivity (proportion of tumorswith instability in a specific marker among the tumorswith mismatch repair protein deficiency) and specifici-ty (proportion of tumors without instability in a specif-ic marker among the tumors without mismatch repairprotein deficiency) were 100% and 94% respectively.(Table 5)

This data impressively corroborates numerous stud-ies that report the possibility of eliminating the dinu-cleotidic markers from the prototypic panel.

Interestingly, the elimination of the class of dinucleo-tidic markers would mean, in other words, to eliminateto whole class MSI-L.

Scientific debate is fervent about MSI-L because,currently, it is not entirely clear whether the low gradeinstability is the result of any kind of genetic defector if it is generally the result of the carcinogenesisprocess [19].

The reason for the controversy in this debate de-rives from the fact that MSI-H tumors present clinical-pathologically different characteristics than tumorswith MSI-L and MSS. In contrast, both MSI-L andMSS show very similar clinical characteristics. Oftenthey are only distinguishable by the total number ofanalyzed markers. It was shown that the largest part ofcolorectal cancers presents a particular grade of insta-bility when being analyzed by an elevated number ofmicrosatellites [8].

60 M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis

Without any further information, the aim of this testis to identify the group of MSI-H tumors which be-comes clinical evident as HNPCC. Furthermore, thetest provides the possibility to discriminate the groupsof MSI-H and non-MSI-H from each other. In this con-text it seems reasonable to discuss an elimination of theclass MSI-L in order to reduce costs that are connectedto this diagnosis.

Moreover, the assessment of MSI status may be use-ful in establishing the oncological outcome of CRCpatients and also in predicting tumor response tochemotherapy. High levels of microsatellite instabil-ity was associated with a favorable outcome, but themechanism of this effect is uncertain.

High levels of microsatellite instability improve theprognosis and may also increase the likelihood of sur-vival after chemotherapy because cancers with highlevels of microsatellite instability usually retain 18qalleles, loss of heterozygosity in such tumors is un-likely to be a determinant of outcome after adjuvantchemotherapy [27–30].

As a conclusion, this study on 44 patients with CRCrevealed noteworthy results concerning the accuracy ofthe new markers.

Further evidence was given that these new markerscan provide additional information concerning the sta-tus of microsatellites instability. In particular, it wasproved that, with these new markers, it is possible toclearly discriminate two conditions: MSS and MSI-Hwith a much higher credibility than with the prototypicmarkers according to Bethesda or with other panels.

As a matter of fact, it was shown that low and highgrades of instability could not be reliably distinguishedby the use of dinucleotidic markers in total accordancewith current literature. Thus, it seems appropriate toargue whether the use of these markers should be rec-ommended for the future.

The aforementioned considerations can only beproved analyzing a larger cohort of patients. In thisway, it would be possible to give concrete proposalsabout how to replace markers according to Bethesdathrough markers that are more powerful and usable forthe clinical practice.

Acknowledgments

We are grateful to Dr. Galdi F, Dr. Cesaro L for sup-porting the laboratory analysis. This study, was par-tially supported by Clinica Chirurgica II Tissue Tu-mor Biobank, grants from the Regione Veneto (Azione

Biotech 2), from Fondazione Antonveneta, the ItalianMinistry of Health (Programma speciale ex. art.12),AIRC (Associazione Italiana per la Ricerca sul Cancro)and the Italian Ministry of Education, University andResearch (MIUR: n◦ 3933).

Competing Interests: The Authors declare no con-flict of interest.

References

[1] K.W. Kinzler and B. Vogelstein, Lessons from hereditary col-orectal cancer, Cell 87(2) (1996), 159–170.

[2] R. Salovaara, A. Loukola, P. Kristo, H. Kaariainen, H. Ahtola,M. Eskelinen et al., Population-based molecular detection ofhereditary nonpolyposis colorectal cancer, J Clin Oncol 18(11)(2000), 2193–2200.

[3] P. Chaves, C. Cruz, P. Lage, I. Claro, M. Cravo, C.N. Leitao etal., Immunohistochemical detection of mismatch repair geneproteins as a useful tool for the identification of colorectalcarcinoma with the mutator phenotype, J Pathol 191(4) (2000),355–360.

[4] J.G. Herman, A. Umar, K. Polyak, J.R. Graff, N. Ahuja, J.P. Is-sa JP et al., Incidence and functional consequences of hMLH1promoter hypermethylation in colorectal carcinoma, Proc NatlAcad Sci USA 95(12) (1998), 6870–6875.

[5] C.R. Boland, S.N. Thibodeau, S.R. Hamilton, D. Sidransky,J.R. Eshleman, R.W. Burt et al., A National Cancer InstituteWorkshop on Microsatellite Instability for cancer detectionand familial predisposition: development of international cri-teria for the determination of microsatellite instability in col-orectal cancer, Cancer Res 58(22) (1998), 5248–5257.

[6] C. Sutter, J. Gebert, P. Bischoff, C. Herfarth and M. von KnebelDoeberitz, Molecular screening of potential HNPCC patientsusing a multiplex microsatellite PCR system, Mol Cell Probes13(2) (1999), 157–165.

[7] A. Loukola, S. Vilkki, J. Singh, V. Launonen and L.A. Aal-tonen, Germline and somatic mutation analysis of MLH3 inMSI-positive colorectal cancer, Am J Pathol 157(2) (2000),347–352.

[8] A. Loukola, K. Eklin, P. Laiho, R. Salovaara, P. Kristo, H.Jarvinen et al., Microsatellite marker analysis in screening forhereditary nonpolyposis colorectal cancer (HNPCC), CancerRes 61(11) (2001), 4545–4549.

[9] J.M. Hoang, P.H. Cottu, B. Thuille, R.J. Salmon, G. Thomasand R. Hamelin, BAT-26, an indicator of the replication errorphenotype in colorectal cancers and cell lines, Cancer Res57(2) (1997), 300–303.

[10] C. Pastrello, S. Baglioni, M.G. Tibiletti, L. Papi, M. For-nasarig, A. Morabito et al., Stability of BAT26 in tumours ofhereditary nonpolyposis colorectal cancer patients with MSH2intragenic deletion, Eur J Hum Genet 14(1) (2006), 63–68.

[11] A.K. Sood, R. Holmes, M.J. Hendrix and R.E. Buller, Appli-cation of the National Cancer Institute international criteria fordetermination of microsatellite instability in ovarian cancer,Cancer Res 61(11) (2001), 4371–4374.

[12] N. Suraweera, A. Duval, M. Reperant, C. Vaury, D. Furlan,K. Leroy et al., Evaluation of tumor microsatellite instabili-ty using five quasimonomorphic mononucleotide repeats andpentaplex PCR, Gastroenterology 123(6) (2002), 1804–1811.

[13] A. Umar, C.R. Boland, J.P. Terdiman, S. Syngal, A. de laChapelle, J. Ruschoff et al., Revised Bethesda Guidelines for

M. Agostini et al. / A ten markers panel provides a more accurate and complete microsatellite instability analysis 61

hereditary nonpolyposis colorectal cancer (Lynch syndrome)and microsatellite instability, J Natl Cancer Inst 96(4) (2004),261–268.

[14] N.M. Lindor, L.J. Burgart, O. Leontovich, R.M. Goldberg,J.M. Cunningham, D.J. Sargent et al., Immunohistochemistryversus microsatellite instability testing in phenotyping col-orectal tumors, J Clin Oncol 20(4) (2002), 1043–1048.

[15] A. Loukola, A. de la Chapelle and L.A. Aaltonen, Strategiesfor screening for hereditary non-polyposis colorectal cancer,J Med Genet 36(11) (1999), 819–822.

[16] M.P. Hitchins and R.L. Ward, Constitutional (germline) MLH1epimutation as an aetiological mechanism for hereditary non-polyposis colorectal cancer, J Med Genet, 2009.

[17] L.M. Baudhuin, L.J. Burgart, O. Leontovich and S.N. Thi-bodeau, Use of microsatellite instability and immunohisto-chemistry testing for the identification of individuals at riskfor Lynch syndrome, Fam Cancer 4(3) (2005), 255–265.

[18] T.M. Pawlik, C.P. Raut and M.A. Rodriguez-Bigas, Colorectalcarcinogenesis: MSI-H versus MSI-L, Dis Markers 20(4–5)(2004), 199–206.

[19] J.W. Bacher, L.A. Flanagan, R.L. Smalley, N.A. Nassif, L.J.Burgart, R.B. Halberg et al., Development of a fluorescentmultiplex assay for detection of MSI-High tumors, Dis Mark-ers 20(4–5) (2004), 237–250.

[20] O. Buhard, N. Suraweera, A. Lectard, A. Duval and R.Hamelin, Quasimonomorphic mononucleotide repeats forhigh-level microsatellite instability analysis, Dis Markers20(4–5) (2004), 251–257.

[21] A. Muller, T.B. Edmonston, W. Dietmaier, R. Buttner,R. Fishel and J. Ruschoff, MSI-testing in hereditary non-polyposis colorectal carcinoma (HNPCC), Dis Markers 20(4–5) (2004), 225–236.

[22] W.S. Samowitz, M.L. Slattery, J.D. Potter and M.F. Leppert,BAT-26 and BAT-40 instability in colorectal adenomas and

carcinomas and germline polymorphisms, Am J Pathol 154(6)(1999), 1637–1641.

[23] C.M. Suter, D.I. Martin and R.L. Ward, Germline epimutationof MLH1 in individuals with multiple cancers, Nat Genet 36(5)(2004), 497–501.

[24] M. Hitchins, R. Williams, K. Cheong, N. Halani, V.A. Lin, D.Packham et al., MLH1 germline epimutations as a factor inhereditary nonpolyposis colorectal cancer, Gastroenterology129(5) (2005), 1392–1399.

[25] M.P. Hitchins, J.J. Wong, G. Suthers, C.M. Suter, D.I. Martin,N.J. Hawkins et al., Inheritance of a cancer-associated MLH1germ-line epimutation, N Engl J Med 356(7) (2007), 697–705.

[26] Y. Abe and H. Masuda, Genetic alterations of sporadic col-orectal cancer with microsatellite instability, especially char-acteristics of primary multiple colorectal cancers, J Surg On-col 74(4) (2000), 249–256.

[27] R. Gryfe, H. Kim, E.T. Hsieh, M.D. Aronson, E.J. Holowaty,S.B. Bull et al., Tumor microsatellite instability and clinicaloutcome in young patients with colorectal cancer, N Engl JMed 342(2) (2000), 69–77.

[28] S. Popat, R. Hubner and R.S. Houlston, Systematic reviewof microsatellite instability and colorectal cancer prognosis, JClin Oncol 23(3) (2005), 609–618.

[29] C.M. Ribic, D.J. Sargent, M.J. Moore, S.N. Thibodeau, A.J.French, R.M. Goldberg et al., Tumor microsatellite-instabilitystatus as a predictor of benefit from fluorouracil-based adju-vant chemotherapy for colon cancer, N Engl J Med 349(3)(2003), 247–257.

[30] E. Vilar, M. Scaltriti, J. Balmana, C. Saura, M. Guzman, J. Ar-ribas, et al., Microsatellite instability due to hMLH1 deficien-cy is associated with increased cytotoxicity to irinotecan inhuman colorectal cancer cell lines, Br J Cancer 99(10) (2008),1607–1612.