microsatellite analysis at 10q25-q26 in sardinian patients with sporadic endometrial carcinoma :...

10
Microsatellite Analysis at 10q25-q26 in Sardinian Patients with Sporadic Endometrial Carcinoma Identification of Specific Patterns of Genetic Alteration Giuseppe Palmieri, M.D. 1 Antonella Manca, Ph.D. 1 Antonio Cossu, M.D. 2 Giovanni Ruiu, M.D. 3 Marina Pisano, Ph.D. 1 PierLuigi Cherchi, M.D. 3 Salvatore Dessole, M.D. 3 Adriana Pintus, M.D. 2 Giovannino Massarelli, M.D. 2 Francesco Tanda, M.D. 2 Mario Pirastu, M.D. 1 1 Institute of Molecular Genetics, C.N.R., Alghero, Italy. 2 Institute of Pathology, University of Sassari, Sas- sari, Italy. 3 Department of Obstetrics and Gynecology, Uni- versity of Sassari, Sassari, Italy. Supported by Assessorato Igiene e Sanita ` ed As- sistenza Sociale, Regione Autonoma della Sardegna. The authors are grateful to Drs. Egidio Celentano for statistical analysis and Milena Casula, for mo- lecular analysis. Giuseppe Palmieri, M.D., Antonella Manca, Ph.D., and Antonio Cossu, M.D., contributed equally. Address for reprints: Giuseppe Palmieri, M.D., In- stitute of Molecular Genetics, C.N.R., Casella Post- ale, 07040 Santa Maria La Palma (SS), Italy; Fax: 39-079-946714; E-mail: [email protected] Received March 17, 2000; revision received June 5, 2000; accepted June 28, 2000. BACKGROUND. Loss of heterozygosity (LOH) at chromosome 10q25-q26 has been reported previously in endometrial carcinoma (EC), suggesting the presence of tumor suppressor gene(s). Nevertheless, frequency of genome-wide microsatellite instability (MSI) has been demonstrated higher in EC than in other common malignancy, mostly due to defective DNA mismatch repair. The authors further evaluated the role of the chromosome 10q25-q26 in endometrial tumorigenesis as well as the clinical significance of any observed genetic alteration in sporadic EC. METHODS. Paired normal and tumor samples from 94 Sardinian patients with sporadic EC at various stages of disease were screened by polymerase chain reaction (PCR)– based microsatellite analysis. Genomic DNA was isolated from paraffin embedded tissues and amplified by PCR using microsatellite markers spanning approximately 14 cM at 10q25-q26. Microsatellite instability was studied at four loci mapping to different chromosomal locations. RESULTS. Thirty-two (34%) EC patients were found negative for genetic alterations within the 10q25-q26 region. Among the remaining 62 (66%) EC cases, the authors identified 1) a minimum consensus region of LOH of approximately 1 cM, between D10S610 and D10S542 markers; and 2) a subset of tumors with prevalence of instability at 10q25-q26 (10qMI1), as expression of the presence of a MSI1 phe- notype. CONCLUSIONS. The authors’ data establish the existence of significant correlations between disease stages and 10qMI1 (with or without MSI1). However, longer follow-up and additional studies are required to define the clinical significance of these findings as prognostic factors. Moreover, the minimum region of LOH at 10q25-q26 will be further analyzed for identifying the putative tumor suppressor gene involved in EC pathogenesis. Cancer 2000;89:1773– 82. © 2000 American Cancer Society. KEYWORDS: endometrial carcinoma (EC), polymerase chain reaction (PCR), chro- mosome 10q25-q26, loss of heterozygosity (LOH), and microsatellite instability (MSI). E ndometrial carcinoma (EC) represents the most common gyneco- logic malignancy and the fourth most common cancer among females in Western countries, with approximately 150,000 new cases per year 1 Genetic events leading to EC pathogenesis are largely unknown. Mutational inactivations of pentaerythritol tetranitrate (PTEN), re- cently isolated from the 10q23 region, 2,3 and TP53 tumor suppressor genes, often accompanied by allelic deletions, have been described in 10 –38% of ECs. 4 –10 Activating alterations of the K-ras oncogene, par- ticularly at codon 12, have been found in 10 –20% of premalignant 1773 © 2000 American Cancer Society

Upload: giuseppe-palmieri

Post on 06-Jun-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Microsatellite Analysis at 10q25-q26 in SardinianPatients with Sporadic Endometrial CarcinomaIdentification of Specific Patterns of Genetic Alteration

Giuseppe Palmieri, M.D.1

Antonella Manca, Ph.D.1

Antonio Cossu, M.D.2

Giovanni Ruiu, M.D.3

Marina Pisano, Ph.D.1

PierLuigi Cherchi, M.D.3

Salvatore Dessole, M.D.3

Adriana Pintus, M.D.2

Giovannino Massarelli, M.D.2

Francesco Tanda, M.D.2

Mario Pirastu, M.D.1

1 Institute of Molecular Genetics, C.N.R., Alghero,Italy.

2 Institute of Pathology, University of Sassari, Sas-sari, Italy.

3 Department of Obstetrics and Gynecology, Uni-versity of Sassari, Sassari, Italy.

Supported by Assessorato Igiene e Sanita ed As-sistenza Sociale, Regione Autonoma dellaSardegna.

The authors are grateful to Drs. Egidio Celentanofor statistical analysis and Milena Casula, for mo-lecular analysis.

Giuseppe Palmieri, M.D., Antonella Manca, Ph.D.,and Antonio Cossu, M.D., contributed equally.

Address for reprints: Giuseppe Palmieri, M.D., In-stitute of Molecular Genetics, C.N.R., Casella Post-ale, 07040 Santa Maria La Palma (SS), Italy; Fax:39-079-946714; E-mail: [email protected]

Received March 17, 2000; revision received June5, 2000; accepted June 28, 2000.

BACKGROUND. Loss of heterozygosity (LOH) at chromosome 10q25-q26 has been

reported previously in endometrial carcinoma (EC), suggesting the presence of

tumor suppressor gene(s). Nevertheless, frequency of genome-wide microsatellite

instability (MSI) has been demonstrated higher in EC than in other common

malignancy, mostly due to defective DNA mismatch repair. The authors further

evaluated the role of the chromosome 10q25-q26 in endometrial tumorigenesis as

well as the clinical significance of any observed genetic alteration in sporadic EC.

METHODS. Paired normal and tumor samples from 94 Sardinian patients with

sporadic EC at various stages of disease were screened by polymerase chain

reaction (PCR)– based microsatellite analysis. Genomic DNA was isolated from

paraffin embedded tissues and amplified by PCR using microsatellite markers

spanning approximately 14 cM at 10q25-q26. Microsatellite instability was studied

at four loci mapping to different chromosomal locations.

RESULTS. Thirty-two (34%) EC patients were found negative for genetic alterations

within the 10q25-q26 region. Among the remaining 62 (66%) EC cases, the authors

identified 1) a minimum consensus region of LOH of approximately 1 cM, between

D10S610 and D10S542 markers; and 2) a subset of tumors with prevalence of

instability at 10q25-q26 (10qMI1), as expression of the presence of a MSI1 phe-

notype.

CONCLUSIONS. The authors’ data establish the existence of significant correlations

between disease stages and 10qMI1 (with or without MSI1). However, longer

follow-up and additional studies are required to define the clinical significance of

these findings as prognostic factors. Moreover, the minimum region of LOH at

10q25-q26 will be further analyzed for identifying the putative tumor suppressor

gene involved in EC pathogenesis. Cancer 2000;89:1773– 82.

© 2000 American Cancer Society.

KEYWORDS: endometrial carcinoma (EC), polymerase chain reaction (PCR), chro-mosome 10q25-q26, loss of heterozygosity (LOH), and microsatellite instability(MSI).

Endometrial carcinoma (EC) represents the most common gyneco-logic malignancy and the fourth most common cancer among

females in Western countries, with approximately 150,000 new casesper year1

Genetic events leading to EC pathogenesis are largely unknown.Mutational inactivations of pentaerythritol tetranitrate (PTEN), re-cently isolated from the 10q23 region,2,3 and TP53 tumor suppressorgenes, often accompanied by allelic deletions, have been described in10 –38% of ECs.4 –10 Activating alterations of the K-ras oncogene, par-ticularly at codon 12, have been found in 10 –20% of premalignant

1773

© 2000 American Cancer Society

hyperplasias and invasive ECs,11,12 and overexpressionof erb-B2 oncogene (found in approximately 10% ofadvanced EC cases) seemed to correlate with a poorprognosis.13 Similar to various human malignancies,progressive accumulation of several genetic events ap-pears to also be associated with the development ofEC. However, these findings provide only a small frac-tion of the complete molecular genetic profile of ECs.

Variation in length of microsatellite sequences hasbeen found ranging from 15% to 43% in endometrialmalignancies.14 –17 This genome-wide microsatelliteinstability (MSI) is mainly due to defective DNA mis-match repair (MMR) and has been revealed in bothsporadic and EC cases associated with hereditary non-polyposis colorectal carcinoma (HNPCC) syndrome(EC is the most common noncolonic tumor in thesefamilies).17,18 Germ line mutations of the MMR genesMLH1, MSH2, or MSH6 have been implicated as acause of ECs associated with HNPCC syndrome,whereas MLH1 methylation and gene silencing seemto account for most sporadic EC cases with MSI.19 –21

Recently, a new MMR gene, MBD4 (MED1), has beendemonstrated to be involved in human carcinomas,including EC, with MSI.22 Altogether, these findingssuggest that defects in replication fidelity may be acommon underlying factor in the pathogenesis of ECs.However, data on frequency and pattern of instabilityare still limited.

Several groups have investigated EC for loss ofheterozygosity (LOH) to elucidate the chromosomalareas harboring putative tumor suppressor genes. Al-though many allelotyping studies have revealed lossesat variable frequencies within different genomic re-gions, one of the most frequent site of allelic deletionswas the chromosome 10q.23 In addition to the 10q23region (which PTEN tumor suppressor gene has beenmapped to),2,3 more distal regions of LOH (at 10q25-q26) have been identified and correlated to endome-trial tumorigenesis.24,25 Deletion mapping studies in-dicated that at least two minimal consensus regions ofallelic loss exist at this location: between D10S587 andD10S172326 and between D10S221 and D10S610 loci27

To further address questions on the role of the10q25-q26 region in endometrial carcinoma, wescreened 94 sporadic EC cases from Sardinia, wherethe population is genetically homogeneous and al-ready demonstrated to be helpful in defining the mo-lecular basis of complex diseases like cancer.28 Fre-quency of EC in Sardinia has been reported similar tothat of other Western countries, with 18.7 new casesper 100,000 inhabitants.29 Using microsatellite mark-ers from a 14-cM region within the 10q25-q26 chro-mosomal band as well as from four additional chro-mosomal locations (for assessing genome-wide

instability), we identified subsets of patients with dif-ferent patterns of genetic alterations. Statistical corre-lations between these genetic features and histopa-thology or clinical parameters also were concluded.

MATERIALS AND METHODSTissue Samples and DNA ExtractionPatients with histologically documented diagnosis ofEC, with either localized disease or local/regionalspread, were included in the study. Disease stage wasrecorded as Stage I (A, B, and C), Stage II (A and B),and Stage III (A, B, and C) according to the Interna-tional Federation of Gynecology and Obstetrics (FIGO)guidelines30 and corresponding to the following TNMcategories of the International Union Against Cancer(UICC) classification: T1a, T1b, T1c, T2a, T2b, T3a,T3b, and N1, respectively.31 Table 1 shows patientcharacteristics; clinical status was recorded as no ev-idence of disease or presence of progressive disease(PD) at the time of enrollment.

Ninety-four paired samples of tumors and corre-sponding normal tissues, obtained from patients withEC, were analyzed in this study. Patients were unre-lated and originated mainly from North Sardinia. En-dometrial carcinomas were classified as sporadic after

TABLE 1Characteristics of the EC Patients Included in the Study

Characteristic No. of patients %

Total entered 94 100Age at diagnosis (yrs)

Median (range) 65 (35–88)#60 33 35.60 61 65

FIGO stageIA 8 9IB 45 48IC 17 18II 4 4III 20 21

HistologyEndometrioid 83 87Serus carcinoma 5 6Other (clear cell, squamous, andmucinous carcinomas) 6 7

GradingG1 29 31G2 57 60G3 8 9

Disease statusNED 64 68PD 22 23Lost in follow-up 8 9

FIGO: International Federation of Gynecology and Obstetrics; NED: no evidence of disease; PD:

progressive disease.

1774 CANCER October 15, 2000 / Volume 89 / Number 8

evaluation of patients’ family history to exclude thepresence of additional tumors (in particular, colorec-tal and ovarian carcinomas in which association withEC defines HNPCC syndrome). No family fulfilled theAmsterdam criteria for HNPCC syndrome,32 nor wassignificant evidence of cancers in first- and second-degree relatives observed. Clinical charts, pathologyreports, hematoxylin and eosin–stained sections, andparaffin embedded blocks were available in all cases.Pathologic review was performed on each case to con-firm the EC diagnosis. Disease status was defined de-pending on clinical staging as assessed by medicalhistory, physical examination, and instrumental tests;disease progression was determined by a change indisease status. Clinical follow-up was performed overa median period of 28 months (range, 7– 69). All pa-tients were informed about the objectives of thisstudy.

Histopathologic classification and grading of eachtumor was based on the criteria of World Health Or-ganization (as reported by Burton and Wells, 199833).The percentage of neoplastic and normal cells in eachtissue specimen was estimated by light microscopy.All tumor samples included into the study were esti-mated to contain at least 80% intact neoplastic cells.

Genomic DNA was isolated from paraffin embed-ded specimens by dissolvement of paraffin with xylol,sodium dodecyl sulfate lysis, and proteinase K treat-ment followed by phenol extraction and ethanol pre-cipitation as previously described.28

Microsatellite AnalysisPrimers used to amplify simple sequence repeat mark-ers were obtained from Life Technologies (Gaithers-burg, MD). The following marker loci at 10q25-q26were investigated for both allelic loss and instability:D10S187, D10S221, D10S190, D10S1230, D10S1213,and D10S1656 (centromeric to telomeric). In Table 2,genetic position of each marker is indicated in centi-

morgans (cM), as reported in the Marshmed map (at:http://www.marshmed.org). Four additional markerswere used to increase density within the D10S221–D10S1230 interval: D10S1693, D10S610, D10S542, andD10S1722 (centromeric to telomeric; see Fig. 2b). Allprimer sequences were as reported in Genome Data-Base (at: http://www.gdb.org ). For each marker anal-ysis, polymerase chain reaction (PCR) was performedusing 25–50 ng of DNA, 0.5 mM of each specific primer,1.5 mM MgCl2, 0.2 mM dNTPs, 1 pmol of 1 primerend-labeled with -g-32P-ATP, and 1 U AmpliTaq Poly-merase (Perkin-Elmer, Foster City, CA) under the fol-lowing conditions: 1 cycle of enzyme activation at94 °C for 2 minutes, followed by 30 cycles of denatur-ation at 94 °C for 30 seconds, primer annealing at55– 60 °C (depending on primers) for 1 minute, andpolymerase extension at 72 °C for 1 minute. All PCRswere terminated with a 10-minute extension at 72 °C.All PCRs were performed in a 9600 Thermal Cycler(Perkin-Elmer). Final products were diluted 1:1 in de-naturing load buffer (95% formamide, 10 mM NaOH,0.05% xylene cyanol FF, and 0.05% bromphenol blue),denatured at 94 °C for 5 minutes, and 4 mL wereloaded for electrophoresis on 6% polyacrylamide gelscontaining 7.0 M urea at 80 W. The gels were dried andexposed to Hyperfilm MP autoradiography film (Am-ersham, Buckinghamshire, UK) for 16 hours at roomtemperature.

Loss of heterozygosity was defined by the absenceor at least two-thirds reduction in the intensity of oneallele in the tumor sample after comparison to theheterozygous normal tissue genotype (referred to asinformative case). Loss of heterozygosity scoring wasperformed by at least two investigators. Microsatelliteinstability was defined by the presence of additionalbands (due to deletions or insertions) in the PCR-amplified product derived from tumor DNA comparedwith normal DNA. Microsatellite instability of themarkers within the 10q25-q26 region is reported as

TABLE 2AResults from PCR-Based Microsatellite Analysis at 10q25-q26for Detection of LOH1 Tumors

Locus(cM) Marker

LOH1cases

Informative/analyzed cases

LOH1(%)

135.2 D10S187 13 72/85 18136.6 D10S221 19 80/94 24138.4 D10S190 25 75/94 33142.7 D10S1230 11 70/94 16148.1 D10S1213 12 63/73 19149.2 D10S1656 6 51/76 12

PCR: polymerase chain reaction; LOH: loss of heterozygosity.

TABLE 2BResults from PCR-Based Microsatellite Analysis at 10q25-q26for Detection of 10qMI1 Tumors

Locus(cM) Marker

10qMI1cases

Analyzedcases

10qMI1(%)

135.2 D10S187 20 85 24136.6 D10S221 16 94 17138.4 D10S190 23 94 24142.7 D10S1230 12 94 13148.1 D10S1213 11 73 15149.2 D10S1656 16 76 21

PCR: polymerase chain reaction; LOH: loss of heterozygosity.

Genetic Alterations in Endometrial Carcinoma/Palmieri et al. 1775

10qMI1. The genome-wide microsatellite instability(referred to as MSI1) was studied at four loci contain-ing single- or dinucleotide repeat sequences mappingto different chromosomal locations: BAT-26 and BAT-40(both amplifying regions containing polyA tracts onchromosomes 2p16 and 1p13, respectively), D2S123(2p21), and D18S58 (18q22). MSI1 was defined by thepresence of instability in at least one locus, as alsosuggested by studies recently undertaken to deter-mine the best strategy for diagnosing HNPCC-associ-ated cancers on a population-wide scale.34

Statistical AnalysisUnivariate analysis of different variables (presence ofLOH and 10qMI1 and/or MSI1, FIGO stage, histo-logic type and grade, and age) was performed by Pear-son chi-square test. The exact coefficient for sampleproportion analysis was performed to determine allsignificant parameters (,0.05 level). Statistical analy-sis was performed with the statistical package SPSS/7.5 per Windows.

RESULTSPCR-Based Microsatellite AnalysisPaired normal and tumor specimens from 94 patientswith sporadic EC were examined for genetic alter-ations at various microsatellite loci within the 10q25-q26 chromosomal band. Most EC patients presentedlocalized disease (74 of 94 at Stages I and II; 79%) andendometrioid adenocarcinoma as histologic variant(83 of 94; 87%) and were older than 60 years at thetime of diagnosis (61 of 94; 65%) (Table 1).

DNA from paraffin embedded tissues was amplifiedby PCR using 6 markers for microsatellite loci spanningapproximately 14 cM at 10q25-q26 chromosomal band:cen-D10S187- D10S221-D10S190-D10S1230-D10S1213-D10S1656-tel. Comparison of the electrophoretic mobil-ity of di- and trinucleotide repeats from normal andneoplastic DNAs permitted to observe the following ge-netic events: allelic loss, MSI, retention of heterozygosity,or presence of homozygosity (noninformative cases forallelic loss). Typical examples of LOH and MSI are shownin Figure 1. Most of the LOH cases was represented by aquite complete deletion of one allele in tumor samples,whereas the MSI cases ranged from smears to insertion/deletion of specific bands in tumor DNAs (Fig. 1). Un-clear or ambiguous results, such as LOHs of D10S1230(middle) and D10S1213 markers in Figure 1, were con-firmed in replicated experiments.

The observed frequencies of LOH for the 6 mark-ers at 10q25-q26 used in this study ranged from 12%for D10S1656 to 33% for D10S190, with an average of79.6% informativeness at each locus (Table 2A; LOH1frequency was calculated taking into account the

number of informative samples for each marker). Sim-ilar LOH frequencies (ranging from 9% to 27%) wereobserved at chromosome 9p21 in primary melanomasfrom the same population by using 9 different micro-satellite markers (G. Palmieri et al., submitted). All ECcases with at least one allelic deletion are reported inFigure 2a. Tumors can be grouped on the basis of thepresence of allelic deletions encompassing theD10S190 locus (Cases 72 to 6 from left to right in Fig.2a) or the prevalence of instability mixed with discon-tinuous allelic losses at loci different from D10S190(Cases 53 to 30 from left to right in Fig. 2a). In thislatter group, most tumors (11 of 16; 69%) presentedLOH at D10S221 locus (Fig. 2a, right)

Ten tumors with common allelic deletion re-stricted to the D10S190 locus and no instability in anyof the remaining microsatellite markers (Cases 59 to 6from right to left in Fig. 2a) were further analyzed byincreasing the marker density within the D10S221–D10S1230 interval. Four additional markers wereused: D10S1693 and D10S610, centromeric toD10S190, and D10S542 and D10S1722, telomeric toD10S190 (Fig. 2b). Most of these tumors (6 of 10)showed no LOH to these additional loci, strongly sug-gesting that region of common allelic loss is com-prised between D10S610 and D10S542 (approximately1-cM region, encompassing D10S190; Fig. 2b).

Polymerase chain reaction– based analysis ofpaired normal and neoplastic DNAs from our series ofsporadic EC patients revealed instability of microsat-ellite markers within the same region. As shown inFigure 1, genetic instability was assessed by identify-ing electrophoretic mobility shifts due to changes(contraction or expansion of di- and trinucleotide re-peats) in microsatellite length. Presence of at least 1unstable marker at 10q25-q26 classified tumors as10qMI1: 42 of 94 (45%) cases with this genetic alter-ation were identified. Frequencies of 10qMI1 for the 6markers at 10q25-q26 ranged from 13% for D10S1230 to24% for both D10S187 and D10S190 markers (Table 2B).

To assess whether a correlation between 10qMI1and genome-wide MSI (referred to as MSI1) may ex-ist, we screened paired normal and tumor samplesfrom all 42 10qMI1 cases as well as from 16 patientswith no instability at 10q25-q26 (10qMI2, as negativecontrols) with a panel of 2 nonpolymorphic (BAT-26and BAT-40) and 2 polymorphic (D2S123 and D18S58)markers. Primers for these markers amplify four dif-ferent chromosomal regions, and tumors were classi-fied as MSI1 if at least one marker again displayedevidence of mutant alleles in tumor DNA comparedwith corresponding normal tissue DNA (see “Materialsand Methods”). Sixteen (28%) of the 58 tumors ana-lyzed were found to be MSI1 (Table 3A), with a sig-

1776 CANCER October 15, 2000 / Volume 89 / Number 8

nificant correlation between the increasing number ofunstable markers at 10q25-q26 and the presence ofMSI1 (P , 0.0001; Table 3A). None of the 16 EC caseswith stable markers at 10q25-q26 showed MSI1,strongly suggesting that 10qMI1 or 10qMI2 may beconsidered as expression of the presence or absence ofMSI1, respectively. Frequencies of MSI1 cases for eachmarker used in this study are reported in Table 3B.

On the basis of all these analyses, 32 (34%) ECpatients were found negative for any genetic alteration

within the 10q25-q26 region (LOH2, 10qMI2). Theremaining 62 (66%) patients were further classified in3 different groups, depending on the presence of in-stability alone (LOH2, 10qMI1; 21 patients), LOHalone (LOH1, 10qMI2; 20 patients), or both geneticfeatures (LOH1, 10qMI1; 21 patients).

Clinical CorrelationsAll genetic alterations detected by PCR-based micro-satellite analysis and previously described were statis-

FIGURE 1. Representative examples of genotyping analysis using six microsatellite markers from the 10q25-q26 region. Lanes corresponding to normal and tumor

DNA are labeled N and T, respectively. L: loss of heterozygosity; M: microsatellite instability. Loss of heterozygosity example for D10S1213 presents only a reduction

of band intensity, probably due to a contamination with DNA from normal cells.

Genetic Alterations in Endometrial Carcinoma/Palmieri et al. 1777

tically correlated to histopathologic and clinical pa-rameters (including the stage of disease). Patientswere classified according to the presence of tumorsconfined to corpus uteri (Stages IA to IC) or extendedbeyond corpus uteri (Stages II and III). Presence of10qMI1 was demonstrated to be significantly associ-ated to the worsening of the disease stage (P 5 0.0063),whereas positivity to LOH showed no statistically sig-nificant linearity (P 5 0.731; Table 4A). Among the 58tumors analyzed for genome-wide MSI, the 16 (28%)cases with MSI1 presented a significant correlationwith the stage of the disease (P 5 0.0021; Table 4B).

No statistically significant correlation betweenany microsatellite alteration (LOH1, 10qMI1, or

MSI1) and histology, tumor grading, or age at onsetwas observed.

After a clinical follow-up over a median period of28 months (range, 7– 69), 22 (23%) progressions ofdisease (PDs) were observed (see Table 1). Presence ofrecurrence was registered in 8 (11%) among the 70Stage I patients, and 14 (58%) among the 24 StagesII–III patients. No significant difference in recurrencerates was observed among the groups of EC patientsnegative or positive for genetic alterations at 10q25-q26. In particular, recurrences were registered in 7

FIGURE 2. Allelic deletion map at 10q25-q26. Marker loci are ordered centromeric to telomeric. Patients are indicated by their number in the series. Black boxes

indicate LOH. Empty white boxes indicate retained heterozygosity. White boxes with O or X indicate constitutional homozygosity or no data, respectively. Gray boxes

indicate microsatellite instability. (a) Patterns of deletion for 6 markers at 10q25-q26 in all 41 patients with LOH1. (b) More detailed map among the 10 tumors

with deletion of D10S190. Genetic position of each marker locus is indicated in centimorgans (cM), as reported in the Marshmed map. The four additional markers

are evidenced.

TABLE 3AResults from PCR-Based Microsatellite Analysis for Detectionof MSI1 Tumors

Locus MarkerMSI1cases

Analyzedcases

MSI1(%)

2p16 BAT-26 10 58 171p13 BAT-40 7 58 122p21 D2S123 8 58 1418q22 D18S58 5 56 9

PCR: polymerase chain reaction; MSI: microsatellite instability.

TABLE 3BResults from PCR-Based Microsatellite Analysis for Detection ofMSI1 Tumors with Correlation between No. of 10qMI1 Markersand Presence of MSI

No. of 10qMI1markers (no.of patients)

MSI1 (16patients) (%)

MSI2 (42patients) (%)

$3 (17) 11 (65) 6 (35)2 (13) 4 (31) 9 (69)1 (12) 1 (8) 11 (92)0 (16) 0 16 (100)

P , 0.001

PCR: polymerase chain reaction; MSI: microsatellite instability.

1778 CANCER October 15, 2000 / Volume 89 / Number 8

(22%) of 32 LOH2/10qMI2 patients, and 15 (24%) of62 patients with at least 1 genetic alteration (LOH1,10qMI1, or MSI1). After stratification by diseasestage, again striking was the absence of differences inrecurrence rates between the 2 groups, without (3 of25 [12%] Stage I and 4 of 7 [57%] Stage II–III patients)and with (5 of 45 [11%] Stage I and 10 of 17 [59%] StageII–III patients) genetic alterations. However, most ofthe patients with PD in the latter group (13 of 15; 87%)presented genetic instability (10qMI1, alone or asso-ciated with MSI1).

DISCUSSIONNinety-four sporadic ECs (most of them histologicallyclassified as endometrioid adenocarcinoma) were col-lected and investigated for allelic deletion at 10q25-q26, identifying a large subset (41 of 94; 44%) of LOH1tumors. Screening with 6 markers revealed the pres-ence of 2 patterns of genetic alteration: 1, quite sim-ple, with deletions of various length progressively nar-rowing down a 14-cM distance to a region of commonallelic loss around the D10S190 locus; and the second,really complex, with most of tumors showing discon-tinuous losses at different loci associated to a high rateof MSI (although LOH at D10S221 locus was the mostfrequently observed in this group; Fig. 2a). Further-more, LOH analysis using 4 additional markers into

the D10S221–D10S1230 interval allowed to assess aminimal region of deletion of approximately 1 cMencompassing the D10S190 locus, between D10S610and D10S542 markers (see Fig. 2).

Previous LOH deletion mapping studies in ECshave been conducted on patients from Japan26 andNorth America.27 Two discreet regions of allelic loss at10q25-q26 have been thus described: one maps to theD10S221-D10S610 interval27 and the second is locatedbetween D10S587 and D10S1723 loci (distal toD10S1230)26 The region of LOH identified in our study(in the D10S610 –D10S542 interval) should be consid-ered as the third one, slightly more telomeric to thatshown by the American authors and markedly morecentromeric to that identified by the Japanese group.However, 2 issues have to be precised: 1) 13 (62%)among the 21 tumors with LOH1/10qMI1 (see Fig.2a) showed the smallest region of overlapping (SRO)around the marker D10S221 (it seems to be very likelythat this SRO overlaps with the region defined byPeiffer-Schneider et al.27); and 2) some LOH patternsdescribed by the same American authors suggestedthe possibility that other putative tumor suppressorgene(s) could be distal to D10S610.

All these mapping data in EC patients could beexplained by the presence of more than one tumorsuppressor gene within the 10q25-q26 region, all ofthem involved in endometrial tumorigenesis. If true,an intriguing hypothesis could be that such discrep-ancies in deletion patterns and thus in genes involvedin EC pathogenesis might be due to the considerabledifferences in patients’ origin (North America, Japan,Sardinia). Taking into account also other evidences(see below), one could suppose the existence of twomolecular types of EC (with or without MSI—indica-tive or not, respectively— of a mismatch repair defect),which may give rise to different LOH mapping results.

When the same series of EC patients has beenanalyzed for the presence of MSI at the 10q25-q26region (designated as 10qMI1), we found a similarhigh incidence (42 of 94; 45%) of this genetic alter-ation. Nevertheless, presence of at least 1 10qMI1marker associated to at least 1 marker with instabilityat other chromosomal locations identified a specificsubset (16 of 42; 38%) of sporadic ECs displaying anevident MSI1 phenotype (see Table 3A). Assumingthat most of the remaining 52 EC patients from ourseries with absence of instability at 10q25-q26 mightbe also MSI2 (as demonstrated in 16 randomly ana-lyzed 10qMI2 patients; see Table 3A), we could spec-ulate that MSI1 is present in a fraction of sporadicECs.

Many studies have documented MSI in many spo-radic tumor types, suggesting that presence of MSI

TABLE 4ACorrelation between Disease Stages and Presence of GeneticAlterations at 10q25-q26

Stage (no. ofpatients)

10qMI1 (42patients) (%)

10qMI2 (52patients) (%)

LOH1 (41patients) (%)

LOH2 (53patients) (%)

IA (8) 1 (13) 7 (87) 3 (38) 5 (62)IB (45) 15 (33) 30 (67) 22 (49) 23 (51)IC (17) 8 (47) 9 (53) 6 (35) 11 (65)II–III (24) 18 (75) 6 (25) 10 (42) 14 (58)

P 5 0.0063 P 5 0.731

LOH: loss of heterozygosity.

TABLE 4BCorrelation between Disease Stages and Presence of Genome-WideMSI

Stage (no. ofpatients)

MSI1 (16patient) (%)

MSI2 (42patients) (%)

IA (7) 0 7 (100)IB (27) 4 (15) 23 (85)IC (12) 4 (33) 8 (67)II–III (12) 8 (67) 4 (33)

P 5 0.0021

MSI: microsatellite instability.

Genetic Alterations in Endometrial Carcinoma/Palmieri et al. 1779

might be a marker of a mutator phenotype in humancancers.17,35,36 In our study, the significant correlationbetween increasing number of 10qMI1 markers andpresence of MSI1 is what should be expected for thelocus unspecific nature of the MSI1 phenotype. Both10qMI1 and MSI1 also correlated significantly withthe disease stages, another point for the commonMSI1 phenotype in EC.

Considering the clinical outcome, the recurrencerates among the Stage I (8 of 70; 11%) and the StageII–III (14 of 24; 58%) patients confirm the predictivevalue of clinical stage as prognostic factor. AlthoughMSI (10qMI1, with or without MSI1) was significantlycorrelated to the disease stage in our series as well asfrequently associated to PD among the 62 EC patientswith genetic alterations, a surprising absence of dif-ferences in recurrence rates between this group(LOH1, 10qMI1, or MSI1 patients) and the subset ofLOH2/10qMI2 patients (15 of 62 [24%] vs. 7 of 32[22%], respectively) was observed. On this regard,many studies of colon carcinoma have suggested thatindividuals with MSI1 tumors even enjoy better over-all survival rates than those with MSI2 tumors.37,38

Thus, further studies are required to assess the clinicalsignificance of the presence of a MSI1 phenotype in EC.

What is the meaning of the presence in our seriesof two subsets of patients, one carrying a specificallelic loss and the other with preponderance of MSI?According to a model recently proposed, normal en-dometrial cells would transform into malignant onesthrough replication errors, and subsequent accumu-lation of mutations in oncogenes and tumor suppres-sor genes.39,40 Replication errors could represent aconsequence of functional alterations of mismatchrepair genes.35,41,42

Somatic mutations of known mismatch repairgenes do not account for most sporadic ECs with theMSI phenotype.43,44 Conversely, several of these caseshas been found associated with hypermethylation ofMLH1 promoter, resulting in down-regulation of itsexpression.17,20 It seems likely that deficit in mismatchrepair with a subsequent increase of the replicationerror rates may be the causative mechanism in mostsporadic colorectal, gastric, and endometrial carcino-mas with MSI.45,46 In fact, DNA polymerases are proneto slip while copying long stretches of the repetitivemicrosatellite DNA sequences; in the absence of effec-tive DNA mismatch repair, the resulting insertion ordeletion errors go undetected and unrepaired.47 Inthis regard, we started screening additional EC casesfor the presence of MSI1 phenotype, to study theinactivating mechanisms of all known mismatch re-pair genes at both somatic and germinal levels (workin progress). Of note, preliminary data from immuno-

histochemistry using anti-MLH1 antibody among the16 MSI1 EC tumors from our series revealed a highconcordance between down-regulation of this MMRgene expression (13 staining negative cases; 81%) andpresence of MSI.

Frameshift mutations of microsatellite repeatswithin specific coding regions are believed to inacti-vate the corresponding genes, contributing to tumordevelopment. In MSI1 gastrointestinal carcinomas,several genes have resulted affected through thismechanism.17 Of interest in this context is the dem-onstration that mutations of the PTEN tumor suppres-sor gene are more common in MSI1 than in MSI2 ECcases, suggesting that PTEN may represent a muta-tional target of genetic instability for endometrial tu-morigenesis (MSI1 is considered to be closely associ-ated with slippage-related mutations of this gene).17

Again, identification of a larger group of ECs with theMSI1 phenotype (work in progress) will allow to alsoassess the PTEN status in such a subset of EC patients.

In light of the results from our EC series and onthe basis of the model previously reported, one couldspeculate that two pathogenetic mechanisms involv-ing the 10q25-q26 chromosomal band may exist. Infact, this region could either carry tumor suppressorgene(s) in which alterations are directly promoting theendometrial carcinogenesis (tumors with specific al-lelic deletions—including the D10S190 locus—andlow incidence of MSI) or represent the target for de-letion/insertion errors secondary to alterations of themismatch repairing machinery (tumors with preva-lence of instability). Again, additional studies areneeded to identify the putative tumor suppressorgene(s) contained into this defined region and con-firm the real existence of such two mechanisms as wellas their role on development and progression of EC.

REFERENCES1. Rose PG. Endometrial carcinoma. N Engl J Med 1996;335:

610 –9.2. Steck PA, Pershouse MA, Jasser SA, Yung WKA, Lin H, Ligon

AH, et al. Identification of a candidate tumor suppressorgene MMAC1, at chromosome 10q23.3 that is mutated inmultiple advanced cancers. Nat Genet 1997;15:356 – 62.

3. Li J, Yen C, Liaw D, Podsypanina K, Bose S, Wang SI, et al.PTEN, a putative protein tyrosinase phosphatase gene mu-tated in human brain, breast and prostate cancer. Science1997;275:1943–7.

4. Honda T, Kato H, Imamura T, Gima T, Nishida JI, Sasaki M,et al. Involvement of p53 gene mutations in human endo-metrial carcinomas. Int J Cancer 1993;53:963–7.

5. Kong D, Suzuki A, Zou TT, Sakurada A, Kemp LW, WakatsukiS, et al. PTEN1 is frequently mutated in primary endometrialcarcinomas. Nat Genet 1997;17:143– 4.

1780 CANCER October 15, 2000 / Volume 89 / Number 8

6. Tashiro H, Blazes MS, Wu R, Cho KR, Bose S, Wang SI, et al.Mutation in PTEN are frequent in endometrial carcinomabut rare in other common gynecological malignancies. Can-cer Res 1997;57:3935– 40.

7. Risinger JI, Hayes KA, Berchuck A, Barrett C. PTEN/NMAC1mutations in endometrial cancers. Cancer Res 1997;57:4736–8.

8. Risinger JI, Hayes KA, Maxwell GL, Carney ME, Dodge RK,Barrett JC, et al. PTEN mutation in endometrial cancers isassociated with favorable clinical and pathologic character-istics. Clin Cancer Res 1998;4:3005–10.

9. Niederacher D, An HX, Camrath S, Dominik SI, Gohring UJ,Oertel A, et al. Loss of heterozygosity of BRCA1, TP53 andTCRD markers analysed in sporadic endometrial cancer.Eur J Cancer 1998;34:1770 – 6.

10. Simpkins SB, Peiffer-Schneider S, Mutch DG, Gersell DJ,Goodfellow PJ. PTEN mutation in endometrial cancers with10q LOH additional evidence for the involvement of multi-ple tumor suppressors. Gynecol Oncol 1998;71:391–5.

11. Enomoto T, Fujita M, Inoue M, Rice JM, Nakajima R, Tani-zawa O, et al. Alteration of the p53 tumor suppressor geneand its association with activation of the c-k-ras-2 proton-cogenein premalignant and malignant lesions of the humanuterine endometrium. Cancer Res 1993;53:1883– 8.

12. Imamura T, Arima T, Kato H, Miyamoto S, Sasazuki T, WakeN. Chromosomal deletions and K-ras gene mutations inhuman endometrial carcinomas. Int J Cancer 1992;51:47–52.

13. Hetzel DJ, Wilson TO, Keeney GL, Roche PC, Cha SS, PodratzKC. HER-2/neu expression: a major prognostic factor inendometrial carcinoma. Gynecol Oncol 1992;47:179 – 85.

14. Caduff RF, Johnston CM, Svoboda-Newman SM, Poy EL,Merajver SD, Frank TS. Clinical and pathological signifi-cance of microsatellite instability in sporadic endometrialcarcinoma. Am J Pathol 1996;148:1671– 8.

15. Helland A, Borresen-Dale AL, Peltomaki P, Hektoen M, Kris-tensen GB, Nesland JM, et al. Microsatellite instability incervical and endometrial carcinomas. Int J Cancer 1997;70:499 –501.

16. Catasus L, Machin P, Guiu XM, Prat J. Microsatellite in-stability in endometrial carcinoma: clinicopathologic cor-relations in a series of 42 cases. Human Pathol 1998;29:1160 – 4.

17. Gurin CC, Federici MG, Kang L, Boyd J. Causes and conse-quences of microsatellite instability in endometrial carci-noma. Cancer Res 1999;59:462– 6.

18. Kowalski LD, Mutch DG, Herzog TJ, Rader JS, Goodfellow PJ.Mutational analysis of MLH1 amd MSH2 in 25 prospective-ly-acquired RER11 endometrial cancers. Genes Chromo-somes Cancer 1997;18:219 –27.

19. Esteller M, Levine R, Baylin SB, Ellenson LH, Hermann JG.MLH1 promoter hypermethylation is associated with themicrosatellite instability phenotype in sporadic endometrialcarcinomas. Oncogene 1998;17:2413–7.

20. Simpkins SB, Bocker T, Swisher E, Mutch DG, Gersell DJ,Kovatich AJ, et al. MLH1 promoter methylation and genesilencing is the primary cause of microsatellite instabilityin sporadic endometrial cancers. Hum Mol Genet 1999;8:661– 6.

21. Wijnen J, de Leeuw W, Vasen H, van der Klift H, Moller P,Stormorken A, et al. Familial endometrial cancer in femalecarriers of MSH6 germline mutations. Nat Genet 1999;23:142– 4.

22. Riccio A, Aaltonen LA, Godwin AK, Loukola A, Percesepe A,Salovaara R, et al. The DNA repair gene MB4 (MED1) is

mutated in human carcinomas with microsatellite instabil-ity. Nat Genet 1999;23:266 – 8.

23. Sonoda G, du Manoir S, Godwin AK, Bell DW, Liu Z, HoganM, et al. Detection of gain and losses in primary endometrialcarcinomas by comparative genomic hybridization. GenesChromosomes Cancer 1997;18:115–25.

24. Peiffer S, Herzog TJ, Tribune DJ, Mutch DG, Gersell DJ,Goodfellow PJ. Allelic loss of sequences from the long arm ofchromosome 10 and replication errors in endometrial can-cers. Cancer Res 1995;55:1922– 6.

25. Nagase S, Sato S, Tezuka F, Wada Y, Yajima A, Horii A.Deletion mapping on chromosome 10q25-q26 in humanendometrial cancer. Br J Cancer 1996;74:1979 – 83.

26. Nagase S, Yamakawa H, Sato S, Yajima A, Horii A. Identifi-cation of a 790-kilobase region of common allelic loss inchromosome 10q25-q26 in human endometrial cancer.Cancer Res 1997;57:1630 –3.

27. Peiffer-Schneider S, Noonanm FC, Mutch DG, Simpkins SB,Herzog T, Rader J, et al. Mapping an endometrial cancertumor suppressor gene at 10q25 and development of a bac-terial clone contig for the consensus interval. Genomics1998;52:9 –16.

28. Pisano M, Cossu A, Persico I, Palmieri G, Angius A, Casu G,et al. Identification of a founder BRCA2 mutation in Sardi-nia. Br J Cancer 2000;82:553–9.

29. Budroni M, Cesaraccio R, Desole MG, Pirino DR, Sechi O,Massarelli G, et al. Tumor Registry of Sassari. In: Incidenzae mortalita per tumori nella provincia di Sassari. Sassari,Italy: Tipografia Moderna, Inc., 1998:75– 8

30. FIGO. Changes in gynecological cancer staging by the Inter-national Federation of Gynecology and Obstetrics. Am JObstet Gynecol 1990;162:610 –1.

31. Sobin LH, Wittekind C. TNM classification of malignanttumours. 5th ed. New York: Wiley-Liss, 1997.

32. Vasen HFA, Mecklin J-P, Khan PM, Lynch HT. The interna-tional collaborative group on hereditary non-polyposis colo-rectal cancer (ICG-HNPCC). Dis Colon Rectum 1991;34:424–5.

33. Burton L, Wells M. Recent advances in the histopathologyand molecular pathology of carcinoma of the endometrium.Histopathology 1998;33:297–303.

34. Loukola AM, Salovaara R, Aaltonen LA, de la Chapelle A.Strategies to screen colorectal cancer patients for hereditarynon-polyposis colorectal cancer (HNPCC). Proceedings ofthe 91st annual meeting of the American Association forCancer Research, April 1–5 2000. San Francisco.

35. Modrich P. Mismatch repair, genetic stability, and cancer.Science 1994;266:1959 – 69.

36. Loeb LA. Microsatellite instability: marker of a mutator phe-notype in cancer. Cancer Res 1994;54:5059 – 63.

37. Lothe RA, Peltomaki P, Meling GI, Aaltonen LA, Nystrom-Lahti M, Pylkkanen L, et al. Genomic instability in colorectalcancer: relationship to clinicopathological variables andfamily history. Cancer Res 1993;53:5849 –52.

38. Sankila R, Aaltonen LA, Jarvinen HJ, Mecklin JP. Better sur-vival rates in patients with Mlh1-associated hereditary coloncancer. Gastroenterology 1996;110:682–7.

39. Lengauer C, Kinzier KW, Volgestein B. Genetic instabilitiesin human cancers. Nature 1998;396:643–9.

40. Sakamoto T, Murase T, Urushibata H, Kato K, Takada H,Imamura T, et al. Microsatellite instability and somatic mu-tations in endometrial carcinomas. Gynecol Oncol 1998;71:53– 8.

Genetic Alterations in Endometrial Carcinoma/Palmieri et al. 1781

41. Aaltonen LA, Peltomaki P, Mecklin JP, Jarvinen H, Jass JR,Green JS, et al. Replication errors in benign and malignanttumors from hereditary nonpolyposis colorectal cancer pa-tients. Cancer Res 1994;54:1645– 8.

42. Dietmaier W, Wallinger S, Bocker T, Kullman F, Fishel R,Rushoff J. Diagnostic microsatellite instability: definitionand correlation with mismatch repair protein expression.Cancer Res 1997;57:4749 –56.

43. Katabuchi H, Van Rees B, Lambers AR, Ronnet BM, Blazes MS,Leach FS, et al. Mutations in DNA mismatch repair genes arenot responsible for microsatellite instability in most sporadicendometrial carcinomas. Cancer Res 1995;55:5556–60.

44. Millar AL, Pal T, Madlensky L, Sherman C, Temple L, Mitri A,et al. Mismatch repair gene defects contribute to the genetic

basis of double primary cancers of the colorectum andendometrium. Hum Mol Genet 1999;8:823–39.

45. Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, etal. Incidence and functional consequences of hMLH1 pro-moter hypermethylation in colorectal carcinoma. Proc NatlAcad Sci USA 1998;95:6870 –5.

46. Leung SY, Yuen ST, Chung LP, Chu KM, Wong MP, BranickiFJ, et al. hMLH1 promoter methylation and lack of hMLH1expression in sporadic gastric carcinomas with high-fre-quency microsatellite instability. Cancer Res 1999;59:159 –64.

47. Peltomaki P, de la Chapelle A. Mutations predisposing tohereditary nonpoliposis colorectal cancer. Adv Cancer Res1997;71:93–119.

1782 CANCER October 15, 2000 / Volume 89 / Number 8