Microsatellite instability and mutation analysis of candidate genes in unselected sardinian patients with endometrial carcinoma




Microsatellite instability (MSI) is due mostly to a defective DNA mismatch repair (MMR). Inactivation of the two principal MMR genes, hMLH1 and hMSH2, and the PTEN tumor suppressor gene seems to be involved in endometrial tumorigenesis. In this study, Sardinian patients with endometrial carcinoma (EC) were analyzed to assess the prevalence of both the mutator phenotype (as defined by the presence of MSI and abnormal MMR gene expression at the somatic level) and the hMLH1, hMSH2, and PTEN germline mutations among patients with MSI positive EC.


Paraffin embedded tissue samples from 116 consecutive patients with EC were screened for MSI by polymerase chain reaction-based microsatellite analysis. Immunohistochemistry (IHC) with anti-hMLH1 and anti-hMSH2 antibodies was performed on MSI positive tumor tissue sections. Germline DNA was used for mutational screening by denaturing high-performance liquid chromatography analysis and automated sequencing.


Thirty-nine patients with EC (34%) exhibited MSI; among them, 25 tumor samples (64%) showed negative immunostaining for hMLH1/hMSH2 proteins (referred to as IHC negative). No disease-causing mutation within the coding sequences of the hMLH1/hMSH2 and PTEN genes was found in patients with EC who had the mutator phenotype (MSI positive and IHC negative), except for a newly described hMLH1 missense mutation, Ile655Val, that was observed in 1 of 27 patients (4%). Although MSI was more common among patients with advanced-stage EC and increased as the tumor grade increased, no significant correlation with disease free survival or overall survival was observed among the two groups (MSI positive or MSI negative) of patients with EC.


In patients with MSI positive EC, epigenetic inactivations rather than genetic mutations of the MMR genes seem to be involved in endometrial tumorigenesis. No prognostic value was demonstrated for MSI in patients with EC. Cancer 2002;94:3157–68. © 2002 American Cancer Society.

DOI 10.1002/cncr.10606

Endometrial carcinoma (EC) represents the most common gynecologic malignancy and the fourth most common female tumor in Western countries, with approximately 150,000 new cases per year.1 Although its molecular mechanisms are not understood completely, it is believed that the pathogenesis of EC occurs by a sequential accumulation of genetic alterations. Several studies have implicated the inactivation of tumor suppressor genes (like TP532) as well as the activation of oncogenes (such as K-ras3 and erb-B24) in endometrial tumorigenesis. In recent years, it has been demonstrated that somatic mutations and/or epigenetic inactivations of the PTEN tumor suppressor gene5, 6 play a very major role in the pathogenesis of EC, accounting for up to 90% of sporadic EC tumors with endometrioid histology.7–9 However, disease-causing germline mutations of PTEN have been found in only 10–38% of EC tumors.10, 11 Loss of heterozygosity (LOH), which is suggestive for the presence of tumor suppressor genes, has been found at various rates within different genomic regions, strongly suggesting that new putative tumor suppressor genes may be involved in the pathogenesis of EC. In this regard, the greatest frequency of allelic deletions was reported for chromosome 10q; in addition to 10q23 (where PTEN is located5, 6), a new region, 10q25–q26, has been correlated strongly with endometrial tumorigenesis, as demonstrated previously by us and others.12–14

A molecular mechanism that also is involved in the genesis of some human malignancies, including EC, is represented by a defective replication fidelity. Tumors with a nonfunctional DNA mismatch repair (MMR) display genomic instability, as detected by ubiquitous somatic variations in the length of microsatellite sequences.15 Microsatellite instability (MSI) is characterized by small insertions or deletions within short tandem repeats in tumor DNA compared with the corresponding normal DNA. MSI has been demonstrated in patients with hereditary nonpolyposis colorectal carcinoma (HNPCC), an inherited syndrome that also predisposes to endometrial tumorigenesis (EC is the most common extracolonic malignancy in families with HNPCC).15, 16 The frequency of MSI has been reported to range from 15% to 43% in EC17–20 and from 75% to nearly all tumors from patients with HNPCC.21, 22

Genetic (gene mutation) or epigenetic (gene silencing) inactivation of both alleles of the two principal MMR genes, hMLH1 and hMSH2, leads to MSI at the somatic level. Although a high prevalence of germline mutations in MMR genes has been reported for colorectal carcinoma,21, 22 a low mutational rate of these genes and epigenetic silencing of hMLH1 have been described for sporadic endometrial neoplasia.20, 23, 24 Recent studies have shown a good correlation between the presence of MSI and abnormal MMR gene expression in both colorectal carcinoma25, 26 and EC,27, 28 suggesting that immunodetection of the MMR proteins may represent an additional approach for the identification of tumors with genetic instability. A combination of microsatellite analysis and immunohistochemical staining for hMLH1 and hMSH2 gene products may be helpful in better defining the so-called mutator phenotype as well as in selecting patients with MSI positive (MSI+) tumors for mutational analysis of these two MMR genes.

For patients with colorectal carcinoma, it has been demonstrated that patients with MSI+ tumors exhibit differences in clinicopathologic characteristics (preferential proximal tumor location and better clinical outcome) compared with patients who are without MSI (MSI−).16, 25, 26 Although the presence of the mutator phenotype in EC also has been associated with specific tumor behavior (prevalence of high tumor grade and endometrioid histologic type, indicating a poor prognosis17, 19, 29), the relatively low number of patients with EC who were analyzed for MSI in such studies did not contribute to an overall understanding of the role that genetic instability plays in endometrial tumorigenesis.

To contribute further in assessing both the role of the genetic alterations of candidate genes (hMLH1, hMSH2, and PTEN) in determining the mutator phenotype and the existence of some differences in the clinicopathologic characteristics that may be representative of distinctive tumor behavior, we performed microsatellite analysis on archival tissues from a subset of 116 unselected Sardinian patients with EC (the reported incidence of EC in Sardinia has been similar to that of other Western countries, with 18.7 new incidents per year per 100,000 inhabitants30). Thus, statistical correlations between these molecular features and histopathologic or clinical parameters were inferred.


Tissue Samples and DNA Extraction

Patients with a histologically proven diagnosis of EC were included in the study. Eighteen of 134 enrolled patients were excluded from the series because of DNA degradation. The remaining 116 patients with EC were classified according to the presence of tumor confined to the corpus uteri (Stages IA, IB, and IC) or extended beyond the corpus uteri (Stages IIA, IIB, IIIA, IIIB, and IIIC), as reported in the Fédération Internationale de Gynécologie et d'Obstétrique (FIGO) guidelines31 and corresponding to the following TNM categories of the International Union Against Cancer (UICC) classification: T1a, T1b, T1c, T2a, T2b, T3a, T3b, and N1, respectively.32

Tissue samples from patients with EC were collected consecutively at the Institute of Pathology, University of Sassari, Italy during the period 1989–1997. Pathologic review was performed on each sample to confirm the diagnosis of EC. Patients originated mainly from North Sardinia and were unselected for familial recurrence of the disease. However, family history for cancer was evaluated by questionnaire interviews in patients with EC who attended the Department of Obstetrics and Gynecology at the University of Sassari after initial surgical treatment. Among the patients interviewed, representing the majority of our cohort, the presence of the HNPCC syndrome21 had been excluded, and no significant evidence of tumors in first-degree or second-degree relatives was observed.

Genomic DNA was isolated from tumors samples and corresponding normal tissues for microsatellite analysis, and blood samples of selected patients (after obtaining written, informed consent) were taken for mutational analysis. The protocols for DNA isolation have been reported previously,33, 34 The percentage of neoplastic cells and normal cells in each tissue specimen was estimated by light microscopy. It was estimated that tumor samples contained at least 80% intact neoplastic cells.

Disease status at the time of diagnosis was defined depending on clinical staging, as assessed by medical history, physical examination, and instrumental tests; disease progression was determined by a worsening of the disease status. Clinical follow-up was performed over a median of 46 months (range, 18–132 months). The distribution of clinical and histopathologic characteristics for the patients analyzed in the current study is shown in Table 1.

Table 1. Patient Characteristics
CharacteristicNo. of patients%
  • FIGO: International Federation of Gynecology and Obstetrics.

  • a

    Staging was according to the FIGO criteria.

Total entered134
Total analyzed116100
Age (yrs)  
 ≤ 501210
 > 607565
FIGO Stagea  
 Serous carcinoma65

Microsatellite Analysis

The primers used to amplify simple sequence repeat markers were obtained from Life Technologies (Gaithersburg, MD). Genome-wide MSI was studied at five loci that contained single repeat or dinucleotide repeat sequences and mapping to different chromosomal locations: BAT-25 (at locus 4q12), BAT-26 (2p16), D2S123 (2p16–p21), D5S346 (5q21–q22), and D17S250 (17q11.2–q12). These loci were chosen on the basis of the suggestions of a panel of experts who indicated that this set of markers was the best tool to identify tumors with MSI (although its demonstrated sensitivity was even higher for colorectal carcinoma35). All primer sequences were as reported in the Genome DataBase (GDB at http://www.gdb.org). The presence of MSI (MSI+) was defined in each patient by detection of at least two unstable microsatellite markers (due to deletions or insertions) in tumor DNA compared with normal DNA, as indicated previously.22, 35

A set of microsatellite markers at chromosomes 3p22 and 2p16 (where hMLH1 and hMSH2 are mapped, respectively) was investigated for allelic deletion. In particular, distribution of markers at loci of the two MMR genes were as follows: D3S1619, 5′-hMLH1 (exons 1–12); D3S1611, hMLH1-3′ (exons 13–19); D3S3623 and D2S2240, 5′-hMSH2 (exons 1–5); BAT-26, hMSH2-3′ (exons 6–16) and D2S391 (both telomeric to centromeric). For each MMR gene locus, microsatellite markers spanned a chromosomal region measuring < 1 cm, as assessed by their genetic position reported in the Marshmed map (available at www.marshmed.org). LOH was defined by the absence or at least two-thirds reduction in the intensity of one allele in the tumor sample after comparison with the heterozygous normal tissue genotype (referred to as an informative case).

For each marker analysis, polymerase chain reaction (PCR) was carried out as described previously14 using fluorescence-labeled primers. Separation and analysis of the PCR products were performed by gel electrophoresis on an automated sequencer (model ABI377; Applied Biosystems, Foster City, CA).

Immunohistochemical Analysis of MMR Proteins

Immunohistochemistry (IHC) was performed on 2-μm sections of formalin fixed, paraffin embedded tissues that were obtained using standard procedures. IHC analysis was performed following previously described protocols.26, 28

Staining was evaluated semiquantitatively using normal epithelial cells or the centers of lymphoid follicles as internal controls. The intensity of nuclear staining was used to classify the tumor samples as positive (strong, moderate, or diffusely weak staining; referred to as IHC+) or negative (absent or focally weak staining; referred to as IHC−) for MMR protein expression. IHC scoring was performed by at least two investigators (only in a very few borderline cases, classification of immunostaining required additional investigators and was based on the consistency of the majority of them).

Mutational Analysis

Selected patients with EC who presented with the mutator phenotype (MSI+ and IHC−) were screened for germline mutations within the coding sequences of hMLH1 and hMSH2 genes and of the PTEN gene by using a two-step approach. The primers sequences used, corresponding to the 19 exons of hMLH1,36 the 16 exons of hMSH2,37 and the 9 exons of PTEN,7 were as reported previously. First, PCR products corresponding to exons and intron-exon boundaries were analyzed by denaturing high-performance liquid chromatography (DHPLC), using the WAVE Nucleic Acid Fragment Analysis System (Transgenomic, Omaha, NE), as described previously.38 Mutations and/or polymorphisms were visualized as a characteristic pattern of peaks corresponding to the mixture of homoduplex and heteroduplex formed when wild type and mutant DNA were hybridized. To analyze the nature of heteroduplexes detected with the WAVE system, DNA samples were then sequenced on an Automated Capillary Sequencer (model ABI3100; Applied Biosystems). Sequencing reactions were carried out using the ABI PRISM Big Dye Terminator Cycle Sequencing Ready Reaction kit (PE Biosystems, Foster City, CA).

Statistical Analysis

Using the Pearson chi-square test, the presence of MSI was assessed for its association with different clinical and pathologic parameters: disease stage, tumor grade, disease free survival (DFS), and overall survival (OS). The exact coefficient for sample proportion analysis was performed to determine all significant parameters (< the 0.05 level). All analyses were performed with the statistical package SPSS (version 7.5 for Windows; SPSS, Inc., Chicago, IL).


Analysis of Archival Tissues

Paired, paraffin embedded normal tissues and tumor tissues from 116 patients with EC were analyzed for genetic instability using polymorphic microsatellite markers. The clinicopathologic characteristics of the patients are listed in Table 1. In our series, patients with disease confined to the corpus uteri (81 patients with FIGO Stages IA–IC disease; 70%) were more prevalent than patients with disease that extended beyond the corpus uteri (35 patients with FIGO Stages II and III disease; 30%) (Table 1). The majority of patients with EC presented with endometrioid adenocarcinoma (97 patients; 84%) and moderately differentiated tumors (69 patients with Grade 2 tumors; 59%) as the most represented histologic variants and age > 60 years at the time of diagnosis (75 patients; 65%) (Table 1).

A comparison of amplified DNAs from tumor tissues and corresponding normal tissues revealed the existence of a genome-wide instability according to PCR-based screening with a panel of five polymorphic markers from different chromosomal regions (see Materials and Methods). Genetic instability was identified as an electrophoretic mobility shift due to contraction or expansion of microsatellite repeats, with consequent changes in microsatellite length. Examples of electrophoretic separation patterns, as detected on an automated sequencer, in patients with MSI are shown in Figure 2 (see below). Among the 116 patients analyzed, 39 patients (34%) exhibited an MSI+ phenotype and are listed in Table 2. Tumors were classified as MSI+ when at least two of five markers from the panel displayed evidence of mutant alleles in tumor DNA compared with corresponding normal tissue DNA. To avoid any PCR-based artifact, unclear or ambiguous results were confirmed in replicated experiments.

Figure 2.

Representative examples are shown of microsatellite instability (A) and loss of heterozygosity (B). Markers were from chromosomes 3p22 and 2p16 (see Materials and Methods). Peaks correspond to normal DNA (N) and tumor DNA (T). Sizes are indicated at the bottom of each peak in base pairs (along with the computer-calculated area under the peak).

Table 2. Endometrial Carcinomas with Microsatellite Instabilitya
Tumor sampleMicrosatellite markerTotal no. of markersIHC
  • IHC: immunohistochemistry; nc: not classified.

  • a

    The presence (+) or absence (−) of MMR gene proteins, as detected by IHC also is indicated.


Considering the expression of the hMLH1 and hMSH2 proteins in neoplastic tissue, negative nuclear staining was observed in 25 of 39 MSI+ tumors (64%). Representative examples of IHC staining for hMLH1 and hMSH2 proteins are shown in Figure 1. Table 2 shows that almost all IHC negative (IHC−) samples (23 of 25 samples; 92%) presented loss of hMLH1 expression, with only two samples showing loss of hMSH2 expression. In no MSI+ tumors was nuclear negativity for both MMR proteins was found. The remaining 14 MSI+ tumors with normal hMLH1 and hMSH2 expression (Table 2) also were screened with IHC for MSH6 protein. Although inactivation of MSH6, an additional MMR gene, also has been implicated in the pathogenesis of EC,39 no alteration in the expression of this gene product was detected in the subset of MSI+ tumors analyzed (data not shown). Finally, 20 additional samples that were classified as stable (MSI−) and that were used as negative controls for IHC had normal expression of both hMLH1 and hMSH2 gene products (IHC positive [IHC+]). Thus, the existence of a mutator phenotype was defined by the presence of MSI and abnormal MMR gene expression.

Figure 1.

Immunohistochemical staining is shown of hMLH1/hMSH2 proteins in samples of endometrial carcinoma. Some tumors were positive for hMLH1 (A) but negative for hMSH2 (B; arrow indicates a normal endometrial gland). Some tumors were negative for hMLH1 (C) but positive for hMSH2 (D). Positive nuclear staining for both hMLH1 (E) and hMSH2 (F) proteins was seen in some tumors. All tumors were positive for microsatellite instability.

DNA from paraffin embedded tissues from the 39 MSI+ tumors was amplified further by PCR analysis using microsatellite markers for the hMLH1 and hMSH2 loci (D3S1611 and BAT-26 were intragenic markers located within the hMLH1 intron 12 and hMSH2 intron 5, respectively; see Materials and Methods). Again, a comparison of the electrophoretic mobility of mononucleotide or dinucleotide repeats from normal and neoplastic DNAs revealed the following genetic events: MSI, LOH, retention of heterozygosity, or presence of homozygosity (noninformative samples for allelic deletion). Figure 2 shows typical examples of MSI and LOH using these six markers; these samples were separated electrophoretically on an automated sequencer. Figure 2A illustrates MSI+ samples that ranged from smears of abnormal peaks (in some samples with an evident shift of the electrophoretic pattern, like those reported for markers BAT-26, D3S1619, and D3S3623) to insertion/deletion of specific bands in tumor DNA. Most LOH+ results were represented by the quite complete deletion of one allele in tumor samples (like those shown in Fig. 2B; more evident for markers D2S391 and D3S1611). Unclear or ambiguous results, like LOH at the D3S1619 marker locus in Figure 2B, were confirmed in replicated experiments.

The results of the microsatellite analysis at chromosomes 2p16 and 3p22 for all 39 MSI+ EC samples are summarized in Figure 3. Overall, the prevalence of MSI also was observed at these chromosomal loci, confirming the presence of genome-wide instability. With the exception of samples 68, 107, and 145 at chromosome 3p22, LOH always was associated with MSI within the same MMR loci, suggesting that it may represent an additional expression of genetic instability (see Fig. 3). Considering data from both microsatellite analysis and IHC, 6 of 23 tumors (26%) and 1 of 2 tumors (50%) of the mutator phenotype (MSI+ and IHC−) presented LOH at the loci of the hMLH1 and hMSH2 genes, respectively (Fig. 3). However, allelic deletions within the MMR loci also were observed in tumors with MSI+ and normal expression of the MMR gene products (5 of 16 tumors [31%] and 6 of 37 tumors [16%] with the presence of LOH at the hMLH1 and hMSH2 loci, respectively) (Fig. 3), again supporting the hypothesis that LOH at this level may be due to genetic instability.

Figure 3.

Results from microsatellite analysis at mismatch repair (MMR) loci. Markers are ordered telomeric to centromeric (see Materials and Methods). Patients are indicated by their number in the series. Black boxes, loss of heterozygosity; gray boxes, microsatellite instability; white boxes, retained heterozygosity or constitutional homozygosity; X, not evaluated. Immunohistochemical (IHC) positivity or negativity to hMLH1/hMSH2 is reported. Asterisks indicate patients who underwent mutational analysis.

Mutational Analysis

Germline DNA from a subset of patients with EC who had tumors of the mutator phenotype (MSI+ and IHC−) and a prevalent association with LOH at the hMLH1/hMSH2 loci was screened for disease-causing mutations within the coding sequence of the hMLH1 and hMSH2 genes (see Fig. 3; selected tumors are indicated by asterisks). Mutation detection for all hMLH1/hMSH2 gene exons was performed by DHPLC analysis; all PCR products with an abnormal denaturing profile (due to the putative presence of heteroduplexes; also see Materials and Methods), compared with normal controls, were sequenced using an automated approach. Only two germline variants at exons 8 and 17 of hMLH1 were detected in the subset of 9 patients with EC who were analyzed. However, screening of blood DNA from an additional 18 unselected patients with EC (consecutively and recently collected) as well as from 53 unrelated normal individuals (corresponding to 106 chromosomes) who were used as controls revealed 1) that one variant (Ile655Val; hMLH1 exon 17) was a mutation that was detected in 1 of 27 EC samples (4%) and in no normal controls and that was not reported previously in the data base; and 2) that the other variant (Ile219Val; hMLH1 exon 8) was a high-frequency polymorphism that has been found at the same rate (about 50%) in both patients with EC and normal controls. It is interesting to note that the Ile219Val variant has been reported previously either as a low-penetrance mutation with a frequency of 1%40 or as a polymorphism with a frequency of 13%.41 Figure 4 presents both DHPLC and sequence profiles corresponding to the two germline variants identified within the coding sequence of hMLH1. Table 3 shows a list of all variants identified in the current study.

Figure 4.

Typical examples of results from mutational analysis. (A) Denaturing high-performance liquid chromatography profiles of two germline variants identified within the coding sequence of hMLH1; an arrow indicates the abnormal peak in the patient's DNA. (B) Electropherograms showing the corresponding nucleotide sequences; an arrow indicates the variation in the sequence.

Table 3. All Germline Variants Identified by Mutational Analysis of the Two Mismatch-Repair Genes
MutationhMLH1171963 A→GIle655Val
PolymorphismhMLH1865 A→GIle219Val
PolymorphismhMLH115IVS14 − 19 A→G
PolymorphismhMSH21IVS1 + 8 C→G
PolymorphismhMSH210IVS9 − 9 T→A
PolymorphismhMSH210IVS10 + 12 A→G

Because it has been demonstrated that mutations of the PTEN tumor suppressor gene are common in patients with MSI+ EC, a screening for germline mutations within the coding sequence of this gene was carried out in a series of 9 patients with MSI+ tumors using the same approach described previously. No mutations within the nine exons or the intron-exon boundaries of PTEN were found, indicating that PTEN was not involved in endometrial tumorigenesis in our series.

Clinical Correlations

All molecular alterations identified were correlated statistically with the histopathologic and clinical parameters. Table 4 shows that the presence of MSI was compared with the disease stage and the histologic tumor grade. Although the results showed that the MSI+ phenotype was associated slightly with a worse tumor grade (the incidence of MSI increased from well differentiated tumors to poorly differentiated tumors; P of linearity = 0.04), a more significant correlation between the presence of MSI and disease stage was observed when we compared patients who had tumors confined to the corpus uteri (FIGO Stage I) with patients who had tumors that extended beyond the uterus (FIGO Stage II–III; 20 of 81 patients [25%] vs. 19 of 35 patients [54%], respectively; P < 0.01) (see Table 4).

Table 4. Correlation between Microsatellite Instability and International Federation of Gynecology and Obstetrics Disease Stage or Tumor Grade
CriteriaNo. of patientsMicrosatellite instability (%)
  • MSI: microsatellite instability; FIGO: International Federation of Gynecology and Obstetrics.

  • a

    P = 0.01 (P value for linearity).

  • b

    P = 0.04 (P value for linearity).

FIGO Stage   
 I8120 (25)61 (75)
 IA133 (23)10 (77)
 IB4511 (24)34 (76)
 IC236 (26)17 (74)
 II and III3519 (54)16 (46)
  II103 (30)7 (9)
  III2516 (64)9 (36)
 Totala11639 (34)77 (66)
Tumor grade   
 1337 (21)26 (79)
 26926 (38)43 (62)
 3146 (43)8 (57)
 Totalb11639 (34)77 (66)

After a clinical follow-up over a median of 42 months (range, 18–132 months), no molecular alteration was correlated with DFS or OS. Univariate analysis showed a significant association of DFS and OS only for disease stage (median DFS, 42 months; median OS, 43 months) in 81 patients with Stage I EC compared with a median DFS of 29 months and an OS of 32 months in 35 patients with Stage II–III EC (P < 0.001). No significant difference was observed in DFS and OS among the groups of EC patients with MSI (median DFS, 38 months; median OS, 39 month) or without MSI (median DFS, 40 months; median OS, 42 months; P > 0.05).


In this study, we performed a molecular evaluation of EC to investigate further the role of genomic instability in the pathogenesis of the disease. A reference panel of five polymorphic markers (two mononucleotide repeats and three dinucleotide repeats35) was used to screen genomic DNA from paraffin embedded tissues from 116 unselected patients with EC (although most patients were ascertained as sporadic by questionnaire interviews) for genome-wide MSI. Alterations in two or more marker loci, as suggested by different studies,22, 25, 35 are able to identify tumors with MSI (also referred to as the MSI+ phenotype). This instability can be revealed through the shift in the electrophoretic mobility of the analyzed fragments that is due to a change in the number of repeat units. PCR-based screening with these markers revealed the presence of this genetic alteration in about one-third of patients with EC (39 of 116 patients; 34%).

Many studies have documented the presence of MSI instability at various rate in many tumor types,16, 42, 43 suggesting that MSI may be considered a marker of the tendency for replication errors in human malignancies.41–43 This phenomenon also has been associated with nonfunctional mechanisms of DNA repair (in particular, inactivation of the MMR repair genes hMLH1 and hMSH2).15, 16, 27, 43 In our series, data from IHC studies using both anti-hMLH1 and anti-hMSH2 antibodies revealed an absence of protein expression in 25 of 39 MSI+ tumors (64%). The rate of concordance between the down-regulation of hMLH1/hMSH2 gene expression and the presence of MSI was lower than the rate reported previously (the percentages of MSI+ tumors that showed the loss or reduction of MMR gene expression varied from 70% to 92%, with an average of 78%).25, 27, 29 Even taking into consideration the probability that the rate of immunostaining observed in the current study was underestimated slightly due to misclassification of a very few borderline tumors with weak nuclear staining, the percentage still remained lower than expected. Nevertheless, IHC-based observation of normal expression of the MSH6 protein in the remaining 14 MSI+ tumors clearly indicated that additional genes (such as the recently described MBD4/MED1 gene44) may be implicated in defects of replication fidelity underlying endometrial tumorigenesis. Conversely, it must be emphasized that no tumor classified as negative for MSI presented with abnormal hMLH1/hMSH2 protein expression, confirming the good sensitivity of the microsatellite analysis in detecting genetic instability.

Altogether, our findings may be considered helpful in confirming that inactivating mechanisms of hMLH1/hMSH2 have the major pathogenetic role in EC tumors of the mutator phenotype (MSI+ and IHC−). Although mutational analysis for the hMLH1 and hMSH2 genes was performed on a limited number of patients with MSI+ EC, the lack of disease-causing mutations within the coding sequences of these two MMR genes is consistent with previously reported data,23, 24, 27, 45 strongly suggesting that epigenetic mechanisms (i.e., promoter hypermethylation) may affect such genes and may explain the development of the MSI+ tumorigenic pathway.

Because it has been hypothesized that the PTEN tumor suppressor gene may represent a mutational target of genetic instability for endometrial tumorigenesis (MSI+ is considered to be associated closely with slippage-related mutations of this gene),20 we also assessed the PTEN germline status in a subset of patients with MSI+ EC. Although most of the PTEN alterations (mutations or epigenetic inactivations) have been observed at the somatic level,7–9 an unexpected absence of germline mutations within the coding sequence of this gene has been reported in our series compared with the previously reported data, in which the prevalence ranged from 10% to 38%).10, 11

A greater incidence of MSI was observed in patients with FIGO Stage III disease (also, see Table 4), suggesting that genetic instability may become particularly evident in patients with advanced-stage tumors due to the progressive accumulation of errors. In other words, a deficit in MMR with a subsequent increase of the replication error rate and a sequential accumulation of genetic mutations may be the causative mechanism in EC with MSI, as also suggested by other authors.18, 20, 27, 28

Considering the correlations with clinical parameters, predictive value as a prognostic factor was found only for disease stage. Although they presented with more advanced-stage disease and were more likely to have tumors associated with higher grades and more aggressive histologic subtypes, patients with MSI+ tumors had survival (in terms of both DFS and OS) that was similar to the survival of patients with MSI− tumors. Whereas data from different studies of patients with colorectal and gastric carcinoma are consistent in defining a better outcome for patients with tumors of the MSI+ phenotype,46, 47 some conflicting results on the clinical significance of MSI in patients with EC were obtained by different research groups.19, 29, 48 Considering the most significant studies and with the exception of a recent report by Fiumicino and colleagues (who observed a worse prognosis in patients with MSI+ tumors among 65 patients with Stage I–II sporadic endometrioid endometrial adenocarcinomas),48 the majority of authors found similar recurrence and survival rates for patients with MSI+ EC and MSI− EC.19, 20, 27, 29

Surprisingly, the presence of MSI+ tumors in a greater proportion of patients who had Stage III EC compared with patients who had Stage I EC was not associated with a poorer survival in our series. Taking into consideration data on patients with MSI+ colorectal carcinoma, which demonstrated a better survival only for patients who received adjuvant chemotherapy (the survival among untreated patients was similar in the two groups, with and without the MSI+ phenotype49, 50), it is possible that additional factors may interact with MSI in determining the clinical outcome among patients with EC. Further development of biotechnologies and the isolation of new MMR genes may be helpful to better characterize the mutator phenotype and to identify subsets of patients who have MSI+ EC, with different tumorigenic pathways predicting different clinical outcomes.