Distinct sets of gene alterations in endometrial carcinoma implicate alternate modes of tumorigenesis




Endometrial carcinomas seem to carry a different prognosis depending on the presence or absence of concomitant complex atypical hyperplasia (hyperplasia). The molecular genetic profile of these two pathogenetic types, based on the genes reportedly mutated in these cancers, remains to be defined. Although microsatellite inability is reported in approximately 25% of endometrial carcinomas, its relation with the 2 pathogenetic types is not investigated.


To elucidate their underlying genetic changes, we analyzed 53 sporadic endometrial tumors, including 19 with and 34 without hyperplasia, for microsatellite instability (MSI), DNA ploidy (by flow cytometry), and for mutations in different genes.


Microsatellite instability was present in 21%, DNA nondiploidy in 15%, and mutations in the PTEN, KRAS, CTNNB1/β-catenin, TP53, and CDKN2A genes were detected in 32, 11, 13, 17, and 0% of the tumors, respectively. Microsatellite instability and mutations in these genes were present in tumors both with and without complex atypical hyperplasia. All cases with complex atypical hyperplasia were early stage (I–II) endometrioid tumors and associated with long progression free disease (P = 0.0004). Furthermore, most tumors with hyperplasia had low World Health Organization or International Federation of Gynecology and Obstetrics grade, had less myometrial invasion, and showed expression of estrogen receptors. All MSI tumors were diploid and had a significantly higher rate of PTEN mutations, but similar rates of KRAS, β-catenin, and TP53 mutations compared with microsatellite stable tumors. TP53 mutations more often were found in nondiploid tumors but never in tumors with PTEN, KRAS, or β-catenin mutations, and all PTEN mutations occurred in diploid tumors.


Thus, PTEN, KRAS, β-catenin, and TP53 mutations occurred in tumors both with and without hyperplasia, but PTEN and TP53 mutations were more common in tumors without hyperplasia. However, none of these genes seems to clearly distinguish tumors with and without hyperplasia, suggesting that other factors may be involved. Conversely, alterations in the PTEN and TP53 genes seem to define distinct subgroups of endometrial carcinoma, the former associated with diploidy and MSI, the latter with macroscopic chromosomal instability. Cancer 2002;94:2369–79. © 2002 American Cancer Society.

DOI 10.1002/cncr.10498

Stimulation of the endometrium by estrogens without the differentiating effects of progestins is a primary etiologic factor associated with the development of endometrial hyperplasia and carcinoma.1 Evidence linking endometrial carcinoma with long-standing complex adenomatous hyperplasia (henceforth hyperplasia) is conclusive,2 but the cellular events in transformation are unknown. Two types of endometrial carcinomas thus are described: type 1 tumors are estrogen-related carcinomas, frequently preceded by hyperplasia, that show higher differentiation and less invasion; type 2 tumors are not estrogen-related and include poorly differentiated carcinomas, uterine papillary serous carcinoma (UPSC), and other anaplastic tumors. Biologic differences of the two pathogenetic types are suggested and relate actually to the presence of endometrioid and UPSC in the two groups.1 However, endometrioid carcinomas, forming the major group of endometrial carcinomas described in these two pathogenetic types, also seem to carry different prognoses.2–6 Molecular studies showed that endometrioid tumors without hyperplasia seem to display high levels of TP53 protein as compared with those with concomitant hyperplasia.6PTEN and KRAS mutations have been detected in hyperplasia, but their relation and frequency in endometrial carcinomas with and without hyperplasia are not determined.7–12 Hence, the molecular genetic profile of these two pathogenetic types remains to be defined.

Although approximately 25% of sporadic endometrial tumors are characterized by microsatellite instability (MSI),9, 13–18 most of these tumors are not causally associated with somatic frameshift mutations in cancer genes frequently found involved in hereditary nonpolyposis colorectal carcinoma or some other sporadic gastrointestinal carcinomas with MSI phenotype.16, 19 Instead, alterations in the PTEN gene or hMLH1 inactivation through promoter hypermethylation seem to be alternative pathways in these cancers.16, 20–28PTEN/MMAC1, the most frequently mutated tumor suppressor gene identified yet in endometrial carcinoma, has a substantially higher frequency of mutations in MSI tumors (78–88%) as compared with microsatellite stable (MSS) ones (30%). β-catenin mutations are reported in approximately 11–14% of endometrial carcinomas,29–32 bearing no association with the presence or absence of underlying MSI.33 The role of KRAS and TP53 in relation to the MSI phenomenon in endometrial carcinomas is controversial and needs to be investigated further.9, 14, 15, 34, 35 The CDKN2A gene, encoding the p16 protein, though infrequently mutated in endometrial carcinoma, has not been checked in MSI tumors.36–38CDKN2A mutations are, however, reported in endometrioid ovarian carcinomas,39 a histologically homologous variant of a major group of endometrial carcinomas. Mutations of PTEN, β-catenin, KRAS, TP53, and CDKN2A reported in endometrial carcinomas are detected more frequently in endometrioid histologic type; however, most reports concern results from the study of single genes. Based on the partly contradictory results from previous studies as well as the limitations of some of the studies, the genetic profile of endometrial carcinomas exhibiting MSI or MSS needs to be investigated further.

In the current study, sporadic tumors of endometrial carcinoma were histologically checked for presence/absence of hyperplasia. All the tumors were subsequently characterized for presence and absence of MSI, and results correlated to mutations in PTEN, KRAS, β-catenin, CDKN2A, and TP53 genes (TP53 mutation results have been published previously40). Furthermore, presence/absence of hyperplasia, MSI, and gene mutations detected were correlated to available clinicopathologic variables.


Tumor Samples

Fifty-three histopathologically verified endometrial tumors, stored frozen at −70 °C at the Department of Oncology in Lund, and selected from patients operated on between May 1993 and October 1994 were used for the current study. The median follow-up for these patients was 61 months (range, 8–81 months). The tumors were grouped as adenocarcinomas with endometrioid (n = 48) and nonendometrioid (n = 5; 3 mixed UPSCs + endometrioid and 2 UPSCs) features. Tissue imprints were taken adjacent to several frozen tumor sections used for DNA extraction, and those found rich in cancer cells were included in this study. DNA isolation and TP53 mutation analysis in exons 4–10 for all these tumors have been described previously.40

MSI Assay

Six microsatellite markers used to evaluate each tumor sample for presence of MSI phenotype were dinucleotide repeats D2S123, D18S61, TP53CA locus, D10S215, and mononucleotide repeats BAT26 and BAT40. Four of these markers have been recommended as useful markers to assess MSI in colorectal carcinoma.41 D10S215 was analyzed because the PTEN gene has been shown to be mutated in a high proportion of MSI endometrial carcinomas as well as to check for loss of heterozygosity at this locus in these tumors (results not shown). Radiolabeled polymerase chain reaction (PCR) products from blood and tumor DNA were loaded in adjacent wells and electrophoresed through 6% polyacrylamide 7 M urea gels. Size alterations in at least 3 of 6 markers tested (≥ 40%41) were used as criteria to define tumors as exhibiting MSI, whereas the remaining were considered MSS.

Estrogen and Progesterone Receptor Analysis

Tumor cytosol estrogen receptor (ER) and progesterone receptor (PgR) analyses were conducted by using enzyme immunoassays (Abbott Laboratories, North Chicago, IL), and a cutoff value of 25 fmol/mg−1 protein was taken to classify tumors as receptor positive or negative.42

DNA Flow Cytometry Analysis

Approximately 50 mg of each tumor tissue, stored at −70 °C, was analyzed by flow cytometry in an Ortho Cytofluorograph 50-H system (Ortho Instruments, Westwood, MA).43 Samples with one G0/G1 peak and a corresponding G2 peak were classified as diploid and those with two or more G0/G1 peaks as nondiploid.

PCR-Single Strand Conformation Polymorphism Analysis

Mutation analysis of PTEN exons 1, 3, 4, 6, and 7, and β-catenin exon 3 was conducted by using intron-based primers22, 29, 44 for amplification and radioactive labeling by PCR followed by single strand conformation polymorphism (SSCP) gel electrophoresis in MDEgels under two different conditions (with glycerol at room temperature, or without glycerol at 4 °C). Polymerase chain reaction-SSCP conditions and primer sequences for detecting the point mutations of codons 12, 13, and 61 of KRAS have been described previously.45 Radioactive PCR and SSCP electrophoresis of the amplified DNA fragments were performed in 6% nondenaturing polyacrylamide gels without glycerol at 4 °C. Variant bands were cut out of dried gels, DNA eluted, and reamplified (as above, excluding radioactivity step). Products were purified with PCR purification kit (Qiagen, Chatsworth, CA), and templates were sequenced.

DNA Sequencing

All samples with variant bands and PTEN exons 2, 5, 8, and 9 (> 300 base pair [bp] and too large for efficient SSCP analysis) were directly sequenced. We subjected PCR products to dye terminator sequencing, using performance-optimized polymer 6 (POP6) on the ABI PRISM 310 Genetic Analyzer (Applied Biosystems, Foster City, CA), according to the manufacturer's instructions. All tumor samples with mutations were reamplified and resequenced to ensure reproducibility.



Nineteen (36%) of the 53 tumors had hyperplasia adjacent to the carcinoma component. The mean age for patients with hyperplasia (61 years; range, 47–73) was lower than for those without hyperplasia (68 years; range, 51–93). Clinicopathologically, all cases in the hyperplasia group had early stage disease (Stage I–II) and endometrioid histology type, whereas most of these were of low grade (15 of 19) and were less invasive locally (15 of 19). Tumors with hyperplasia were essentially ER positive (17 of 19; 90%) and PgR positive (14/19; 74%). All the progressions/deaths from disease were recorded in patients without hyperplasia (P = 0.0013). We detected MSI equally frequently in tumors with and without hyperplasia. All but one tumor in hyperplasia group were diploid, whereas 20% (7 of 34) of tumors without hyperplasia showed nondiploidy (Table 1). In contrast, endometrioid tumors without hyperplasia (29 of 48) associated often with advanced stage (9 of 29; 31%), Grade 3 (13 of 29; 45%), locally invasive (greater than half myometrial thickness, 19 of 29; 66%), and ER negative (15 of 29; 52%), PgR negative (14 of 29; 48%), and nondiploid (7 of 29; 24%) tumors; all patients but one with progressions/deaths from disease belonged to this group (the remaining patient with death from disease had a seropapillary tumor).

Table 1. Clinicopathologic and Molecular Genetic Features of Endometrial Carcinoma Patients
VariableNo. of samplesNo. with hyperplasiaa(%)No. with MSIb(%)
  • MSI: microsatellite instability; WHO: World Health Organization; FIGO: International Federation of Obstetrics and Gynecology.

  • a

    Complex atypical hyperplasia.

  • b

    Microsatellite instability present.

  • c

    P < 0.05 (Fishers exact text).

  • d

    Three mixed (uterine papillary serous carcinoma [UPSC] + endometrioid) and two UPSC.

  • e

    Only endometrioid carcinomas.

  • f

    P < 0.01.

  • g

    Cutoff value as 25 fmol mg−1 protein.

  • h

    Progression of disease/death from disease.

  • i

    P < 0.0013.

All5319 (36)11 (21)
 I + II4419 (43)8 (18)
 III + IV90c3 (33)
 Endometrioid4819 (40)10 (21)
 Nonendometrioidd501 (20)
Grading (WHO)e
 Well + moderate3117 (55)5 (16)
 Poor172 (12)f5 (29)
Grading (FIGO)
 13215 (47)f5 (16)
 284 (50)f2 (25)
 3130f4 (31)
Myometrial invasion
 Less than half2815 (54)8 (29)
 More than half254 (16)f3 (12)
Tumor ploidy
 Diploid4518 (40)11 (24)
 Nondiploid81 (12)0
Complex atypical hyperplasia
 Present194 (21)
 Absent347 (21)
Microsatellite instability
 Present114 (36)
 Absent4215 (36)
Estrogen receptorg
 Present3617 (47)5 (14)
 Absent172 (12)c6 (35)
Progesterone receptorg
 Present3214 (44)6 (19)
 Absent215 (24)5 (24)
PTEN mutation
 Present174 (24)8 (47)
 Absent3615 (42)3 (8.3)f
KRAS mutation
 Present62 (33)1 (17)
 Absent4717 (36)10 (21)
CTNBB1/β-catenin mutation
 Present73 (43)1 (14)
 Absent4616 (35)10 (22)
TP53 mutation
 Present92 (22)2 (22)
 Absent4417 (39)9 (20)
 Progression free3619 (53)9 (25)
 Progression170i2 (12)


Eleven of the 53 (21%) endometrial carcinomas displayed MSI, defined as altered allele sizes for at least 3 markers. One additional tumor (classified as MSS) had altered bands for only the two mononucleotide repeat markers BAT26 and BAT40. Most tumors exhibiting MSI belonged to endometrioid histologic subtype (10 of 11), whereas the single nonendometrioid tumor exhibiting MSI had a mixed (UPSC + endometrioid) type. Presence of MSI did not show any statistical correlation with stage, World Health Organization/International Federation of Gynecology and Obstetrics grades, but these tumors tended to have less myometrial invasion than MSS tumors. The mean age of MSI cases (67 years; range, 51–84) was similar to that of MSS cases (65 years; range, 47–93). Microsatellite instability did not correlate to ER and PgR status (Table 1). Two of the 17 patients with progressions/death from disease were recorded in patients with MSI tumors, the 5-year survival rates for patients with MSI and MSS tumors being 81% and 64%, respectively (P = 0.26).


All 9 exons of the PTEN gene were analyzed, and 18 different mutations were detected in 17 (32%) of the 53 endometrial carcinomas (Table 2). These mutations included 10 frameshift deletions or insertions, 2 nonsense mutations and 2 splice site mutations, which were spread throughout the gene and are predicted to result in protein truncation. The series also included an in-frame 6-bp insertion, an in-frame 3-bp deletion, and 2 missense mutations. These were all located in exons 5 or 6, and both missense mutations affected Arg-130 in the phosphatase core motif. Two of the single nucleotide insertions were located in the poly(A)6 stretch encompassing residues 321–323 in exon 8, and one in the poly(T)5 stretch at residues 56–57 in exon 3. Matched blood cell DNA from all cases with mutations was sequenced and contained a normal sequence, indicating that the mutations identified are somatic in nature. PTEN mutations were detected in both endometrioid and nonendometrioid tumors, and one tumor containing a mixed papillary serous and endometrioid histology type had two different frameshift mutations (Table 2). PTEN mutations were common in tumors both with and without hyperplasia (21% vs. 38%; P = 0.24; Table 1) and detected in 8 of 11 tumors (73%) with MSI and in 9 of 42 (21%) MSS tumors (P = 0.002). All PTEN mutated tumors were DNA diploid (P = 0.04). Of the PTEN-mutated tumors, three had KRAS and two had β-catenin mutation, but none had TP53 mutation (Table 3).

Table 2. Molecular Genetic and Clinicopathologic Features of Endometrial Tumors with Gene Mutations
TumorExonNucleotideAmino acid changeStageHistologyFIGO gradePFS
  1. FIGO: International Federation of Obstetrics and Gynecology; PFS: progression free survival; E: endometrioid carcinoma; PD: progression/died of disease; A: alive without disease; M: mixed uterine papillary serous + endometrioid carcinoma; PA: progression of disease, alive at present; DW: died without disease; UPSC: uterine papillary serous carcinoma.

 E5Int 81027(−2)a→gUndeterminedIIIE128 PD
 E29Ex 5389G→AR130QIE163 A
 E30Ex 3279del26ter 97IE373 A
 E31Ex 3171insTter 62IE166 A
 E34Ex 117delAAter 9IM161 A
Ex 5406insAter 179
 E37Ex 5389delGter 133IE168 PA
 E73Ex 8969insAter 324IE257 A
 E82Int 81026(+1)g→aUndeterminedIE269 A
 E85Ex 5391delAter 133IE161 A
 E98Ex 5389G→CR130PIM161 A
 E108Ex 5340G→TE114XIE148 PA
 E113Ex 8969insAter 324IE372 A
 E137Ex 5408ins6137insGMIIE264 A
 E153Ex 4249C→AC83XIIIE324 PD
 E163Ex 8923delGter 316IIM115 PA
 K540Ex 6526delTATdelY176IIE264A
 K544Ex 8955insAter 324IE161A
 E30Ex 138G→AG13DIE373 A
 E37Ex 135G→CG12AIE168 PA
 E90Ex 135G→TG12VIIE115 PA
 E96Ex 135G→AG12AIIE211 PD
 E106Ex 135G→AG12AIIE170 A
 K544Ex 135G→TG12VIE161 A
 E5Ex 3121A→GT41AIIIE128 PD
 E18Ex 3133del3delS45IIIE320 PD
 E33Ex 397T→GS33AIE272 A
 E59Ex 3110C→AS37FIE172 A
 E98Ex 3110C→TS37FIM161 A
 E124Ex 394G→AD32NIE165 A
 E136Ex 393del3−delD32IE128 PD
 E19Ex 5455C→TP153LIE159 DW
 E57Ex 7733G→AG245SIE339 PD
 E74Ex 7694A→GI232VIE174 A
 E76Ex 8817C→TR273CIIE113 PD
 E92Ex 4215insCter148IVE311 PD
 E103Ex 7743G→AR248QIUPSC165 A
 E121Ex 5503A→GH168RIVE359 A
 E131Ex 6637C→TR213stopIIIE320 PD
 K538Ex 7722C→GS241CIIIE317 PD
Table 3. Mutation Analysis of Endometrial Carcinomas with and without Complex Atypical Hyperplasia (Hyperplasia)
With hyperplasiaWithout hyperplasia
TmMSIDNA ploidyPTENKRASβ-cateninTP53TmMSIDNA ploidyPTENKRASβ-cateninTP53
  1. MSI: microsatellite instability; no: mutation absent. yes: mutation present;



KRAS exons 1 and 2, including the frequently mutated codon 12, 13, and 61, were analyzed.

Six (11%) of the 53 endometrial tumors had KRAS point mutations (Table 2). Five were present at codon 12 and one at codon 13 in exon 1. No mutations were detected at codon 61 in exon 2. The mutations included 3 G → A transitions and 3 G → C:T transversions. Three of these six tumors (one MSI, two MSS) had a concomitant PTEN mutation, but no other gene alterations were detected in KRAS-mutated tumors (Table 3). All KRAS mutations occurred in the predominant Stage I–II endometrioid tumors but were approximately equally common in tumors with and without hyperplasia or MSI (Table 1).


Analysis of the β-catenin gene was restricted to exon 3, where stabilizing alterations are known to occur. Mutations were detected in 7 (13%) of the 53 tumors and included 5 missense (3 transitions and 2 transversions) and 2 in-frame deletions of single residues (Table 2). All mutations occurred at known Ser or Thr phosphorylation sites or at adjacent residues. No interstitial mutations, as detected previously in colon carcinoma, were found in any of these tumors.44 Two of the mutated tumors showed a second mutation in PTEN, but none had mutation in KRAS or TP53 (Table 3). Presence of β-catenin mutations did not correlate with tumors with and without hyperplasia/MSI.

TP53 and CDKN2A

Results from screening of TP53 exons 4–10 have been described previously.40 Mutations were found in 9 (17%) of these tumors and included 7 missense, 1 nonsense, and 1 frameshift insertion (Table 2). None of the TP53 mutations had a concomitant mutation in PTEN, KRAS, or β-catenin.TP53 mutation was found twice as frequently in tumors without hyperplasia as compared with those with hyperplasia (20% vs. 10%). Although these mutations were detected with equal frequency among tumors with MSI and MSS (18% vs. 17%), there was a correlation to DNA nondiploid status (P < 0.05). No mutation was detected in the three exons of the CDKN2A encoding the p16 product. However, detection of previously known polymorphic variants G → A transversion at codon 140 (6% of tumors) and G → C transversion at position 494 of the 3′ untranslated region (26% of tumors) illustrates the sensitivity of the method used in analysis of these tumors.

Gene Alterations and Clinicopathologic Correlations

Gene alteration and clinicopathologic correlation data are summarized in Table 2.


Altogether two-thirds of endometrial tumors (34 of 53; 64%) had mutations in the PTEN, KRAS, β-catenin, or TP53 genes, indicating a significant role of these genes in the pathogenesis of endometrial carcinoma. However, there was no major overlap between the different mutated genes, because 29 of the 34 tumors had a single mutant gene, thereby suggesting that these genes may operate in different genetic pathways involved in endometrial carcinogenesis.

PTEN was the most frequently (32%) mutated gene, which is in agreement with previous studies.16, 20–27 Although most of the PTEN mutations occurred in the predominant endometrioid tumor type, three of the five nonendometrioid tumors (two mixed UPSC/endometrioid, one UPSC) also contained PTEN mutation, which may seem contradictory to previous reports associating PTEN mutations with better prognostic factors (including endometrioid type) in endometrial carcinoma.22, 24, 26 However, in agreement with others, we found a significant correlation between PTEN mutations and MSI phenotype. Because one of the three PTEN-mutated nonendometrioid tumors was highly MSI (four microsatellite loci), and a second (although stated as MSS) did show instability for mononucleotide markers, this indicates the role of PTEN alterations in the MSI pathway of tumorigenesis irrespective of histology status. The current data provide no evidence as to whether PTEN alteration is a consequence of MSI, or if it precedes and even have a causal role for the MSI phenotype. However, only 3 of the 17 PTEN mutations, and 2 of the 8 PTEN mutations in MSI tumors, were located at polynucleotide stretches so typically mutated in MSI cells. This suggests that strong selective forces are responsible for the establishment of PTEN mutations in endometrial carcinoma, and that these may even precede the MSI state.

PTEN mutations have been reported in atypical hyperplasias,10–12 though at a frequency less than carcinomas, which suggests that they either play a role in the progression of this disease or represent these arising in tumors without a premalignant stage. Mutter et al. suggested that altered PTEN expression is an early event in most endometrial carcinomas with a premalignant phase and participates in their progression to carcinoma.12 One of the noteworthy observations in our study was presence of PTEN mutations in endometrioid tumors without hyperplasia. Microsatellite instability was detected with equal frequency in tumors both with and without hyperplasia and did not explain the presence of PTEN mutations in the nonhyperplasia group. As expected, hyperplasia was significantly more common in ER positive and PgR positive tumors and, accordingly, PTEN mutations were more common in ER negative tumors. However, PTEN mutations also were found in receptor positive tumors, in keeping with the suggested role of the PTEN-PI3K-AKT pathway in ER regulation.46 PTEN alterations have been linked to advanced disease in certain cancers47 and through promoter region hypermethylation also linked to metastatic potential in endometrial carcinomas.48 This suggests that PTEN inactivation is achieved through different mechanisms and that the biologic role of PTEN in endometrial carcinoma needs to be investigated further, preferably with tumors stratified for presence/absence of hyperplasia.

In the current study, we identified β-catenin mutations in 13% of endometrial carcinomas, all mutations being located at critical residues in exon 3 and predicted to give rise to protein stabilization via decreased phosphorylation by GSK-3β and increased signaling through the Tcf/Lef transcription factors. Nei et al. found that strong nuclear staining of β-catenin was more frequent in endometrial hyperplasia than in endometrial carcinoma samples, suggesting a role in the early development of this tumor type.49 Their study did not show any correlation between the presence of hyperplasia and expression of ER, suggesting that the CTNNB1/wnt-1 signaling pathway is activated irrespective of ER stimulation. In another study, two endometrial tumors with β-catenin mutations had adjacent hyperplasia with wild0type β-catenin, suggesting a role of β-catenin activation in the transition from benign to malignant phase, though not necessarily a gatekeeper-type event, in endometrial tumorigenesis.33 In our study, the frequency of β-catenin mutations was similar in tumors with and without hyperplasia (16% vs. 12%). However, all β-catenin mutated tumors were ER positive and most were PgR positive (4 of 7), suggesting a dependence of estrogen stimulation during endometrial carcinogenesis.

KRAS mutations were detected in 11% of the endometrial tumors in our series and were present among tumors both with and without concomitant hyperplasia. KRAS mutations have been identified in endometrial hyperplasia, though at a lower frequency than carcinomas.8, 50, 51 Conversely, KRAS mutations seem to provide a growth advantage to endometrial cells in low estrogen/progesterone milieu, thereby explaining the aggressive behavior of the mutant tumors in postmenopausal women.52 Japanese women, epidemiologically less exposed to estrogenic steroids, show a lower incidence but, paradoxically, higher mortality from endometrial carcinoma as compared with North American women. Of note, the frequency of KRAS mutations is twice as high in the Japanese group as compared with the North American group.8 Although KRAS mutations occurred with equal frequency in tumors with and without hyperplasia in our study, the epidemiologic results seem to suggest that KRAS activation is associated with malignant progression of endometrial tumors without the need for transition via hyperplasia. The prognostic significance of KRAS mutation in this cancer is controversial. Most studies report no association with clinical outcome,53–56 whereas others implicate KRAS activation as an unfavorable prognostic factor57 or, conversely, as associated with improved prognosis.18 These inconsistencies may depend on inclusion of cases from pathologically different groups. In two separate European studies, the mutations were detected in aggressive histologic types (type 2 tumors) of endometrial carcinoma.54, 55 Thus, the role of KRAS in the two pathogenetic types of tumors needs to be further clarified. In this study, three tumors with KRAS mutation also showed a concomitant PTEN mutation, a finding in contrast with one previous study in which these mutations were found to be mutually exclusive.58 RAS binds and activates the 110-kilodalton alpha subunit of P13-K, and PTEN has been reported to potentially inhibit RAS transformation through its ability to suppress the P13-K-signaling cascade.59 That study raised the possibility that endometrial precursor lesions harboring RAS mutations may acquire a growth advantage by PTEN inactivating. In our study, one of the three tumors with both KRAS and PTEN mutations had adjacent hyperplasia, and it would be of interest to further investigate whether either or both mutations are present also in the hyperplasia component. Advanced laser capture microdissection techniques have made it possible to exploit targeted cells without contamination from neighboring cells and may be useful to conduct this type of analysis.

We found that TP53 mutations are twice as frequent in tumors without hyperplasia (21%) than in those with hyperplasia. This finding is in agreement with a previous study in which TP53 overexpression associated significantly with tumors without concomitant hyperplasia.6 Of note, despite the finding that PTEN mutations were detected in a large proportion of endometrial carcinomas without hyperplasia (38%), these did not coexist with TP53 mutations. Thus, alterations of these genes seem to arise independently and represent alternative routes in (type 2) endometrial carcinoma development. This is in accordance with a previous observation that TP53 overexpression was less frequent in endometrial tumors with PTEN mutations.24

Alternative pathways of carcinogenesis in MSI and MSS sporadic colorectal carcinomas are suggested based on the observations of prevalence of mutations in APC, KRAS, TP53, and TGF-BII genes, i.e., genes commonly found involved in sporadic colorectal carcinomas.60 This was evident in our study too, because MSI seemed to confer mutations in several of the genes reported to be involved in endometrial carcinoma. Although 10 of the 11 tumors exhibiting MSI had mutation in either of the genes, PTEN accounted for the majority, an observation in agreement with previous studies.16, 20–27 Mutations in KRAS and TP53 are reported in endometrial carcinoma, although their relation to the MSI phenotype is not clear. KRAS activation was implicated as an early event50 and TP53 inactivation as a late event61 in endometrial tumorigenesis and occurred independently of each other in these studies, which lacked data on MSI. Others report that mutant KRAS alleles were more frequent in MSI than MSS endometrial carcinomas,9, 35 whereas some found no significant correlation of MSI with KRAS14, 15, 34 or TP53 mutations.14, 34 As in sporadic colorectal carcinomas,62 we did not observe any significant difference in the frequency of KRAS and TP53 mutations among MSI or MSS tumors, and these events seemed to arise independently of each other. In colorectal carcinomas, β-catenin mutations occur most frequently in the less common MSI tumors lacking APC mutations than in tumors with an APC alteration.63 In contrast, in endometrial carcinoma, activation of the Wnt-signaling pathway via a mutation in β-catenin, and not APC, seems to dominate.64 In agreement with a previous study, we observed that β-catenin mutations were present in both MSI and MSS tumors, suggesting that β-catenin mutations may arise irrespective of this phenotype.33 Insertion/deletion mutations occurring frequently at iterated bases and in palindromic or repeat sequences are postulated to occur as a result of a slippage mechanism consequent to a mismatch repair.65 The substantially higher frequency of frameshift mutations of PTEN in endometrial carcinoma as compared with glioblastomas thus may be partly attributed to its association with MSI.25 However, we found no difference in the mutation profile of PTEN, nor of other genes in MSI and MSS tumors.

No mutation was detected in the CDKN2A/p16 gene in endometrial tumors. Although the alternative exon 1β, encoding parts of the ARF product, was not included in the analysis, our study does not support a role of CDKN2A/p16 or ARF inactivation in development of endometrial carcinomas, which is in agreement with the majority of previous studies.36, 37 However, the alternative mechanism of promoter hypermethylation or homologous deletion of the gene may be involved in these tumors.66

Few tumors included in this study make it impossible to suggest a model for endometrial tumorigenesis. However, we observed that whereas mutations were present in endometrial tumors both with and without hyperplasia, all progressions/deaths from disease occurred in patients having tumors without concomitant hyperplasia. Thus, this may simply reflect the surrogate role that estrogen-related/nonrelated pathways play in characterizing these genes as high- or low-risk markers in this tumor type. Furthermore, the two types of genetic instabilities seem to occur early in the premalignant stage of adenomas in colorectal carcinomas. Thus, it may be proposed that the presence/absence of a hormone-driven pathway may underlie the various types of genetic instability, which predispose tumors to distinct genetic alterations. These genetic alterations eventually may determine the biologic behavior of these tumors. Therefore, the simultaneous assessment of a panel of histopathologic features and molecular genetic markers could help in defining different molecular profiles of endometrial patients characterized by a different outcome.