www.elsevier.com/locate/humpath

Human Pathology (2013) 44, 218–225

Original contribution

Association of mammalian target of rapamycin withaggressive type II endometrial carcinomas and pooroutcome: a potential target treatment☆,☆☆

Gloria Peiró MD, PhDa,⁎, Francisca M. Peiró MDb, Fernando Ortiz-Martínez BS a,María Planelles MD c, Laura Sánchez-Tejada BS a, Cristina Alenda MD, PhDd,Segundo Ceballos MDd, José Sánchez-Payá MD, PhDe, Juan B. Laforga MD, PhD f

aResearch Unit, Hospital General Universitari d'Alacant, Pintor Baeza 12, 03010 Alacant, SpainbDepartment of Pathology, Hospital Marina Baixa, Av. Jaume Botella Mayor no. 7, 03570 La Vila Joiosa, Alacant, SpaincDepartment of Pathology, Hospital Virgen de los Lírios, Políg. Caramanchel s/n, 03800 Alcoi, Alacant, SpaindDepartment of Pathology, Hospital General Universitari d'Alacant, Pintor Baeza 12, 03010 Alacant, SpaineDepartment of Epidemiology, Hospital General Universitari d'Alacant, Pintor Baeza 12, 03010 Alacant, SpainfDepartment of Pathology, Hospital de Dénia, Marina Salud, Ptda. Beniadlà s/n, 03700 Dénia, Alacant, Spain

Received 9 March 2012; revised 22 April 2012; accepted 9 May 2012

(F

P

0h

Keywords:Endometrial carcinoma;mTOR;PTEN;β-Catenin;MMR genes;MSI;BRAF-V600E;Outcome

Summary The classification of endometrial carcinoma divided into types I and II has shown clinicalusefulness. Molecular alterations of PTEN and Wnt/β-catenin have been identified in this neoplasia.However, the role of mammalian target of rapamycin according to subcellular localization in thepathogenesis of this neoplasia and its prognostic significance are not well defined. We studied theexpression of phosphorylated mammalian target of rapamycin, PTEN, and β-catenin and theirrelationship with clinicopathologic features, molecular factors (microsatellite instability, mismatchrepair, and BRAF genes) and patients' survival in a series of 260 nonconsecutive endometrialcarcinomas. Tissue microarrays were manually constructed, and genomic DNA was extracted fromparaffin-embedded cylinders (1 mm thick) from preselected tumor areas. The mammalian target ofrapamycin in the nuclei (mTORC2; 47%) or cytoplasm (mTORC1; 48%) were seen in type IIendometrial carcinoma, the latter also in advanced stages (P ≤ .046). PTEN loss (58%) was detected intype I endometrial carcinoma of grade 1, at early stage, with mismatch repair gene loss (24.4%) andmicrosatellite instability–positive status (22%; P ≤ .05). Nuclear β-catenin (16%) was found in type Itumors of younger patients (P ≤ .003). In contrast, BRAF-V600E mutations were not detected (0%).Mammalian target of rapamycin cytoplasmic high expression implied poorer prognosis (P = .02;Kaplan-Meier, log-rank test), but grade 3 tumors, vascular invasion, advanced stage, or PTEN presencecorrelated independently with a negative impact on survival (all P ≤ .036; Cox analysis). Our resultsshow that mammalian target of rapamycin, PTEN, and β-catenin are independently involved in different

☆ This study was supported by a grant from Fundación de la Comunidad Valenciana para la Investigación en el Hospital General Universitario de AlicanteCVI-HGUA/2006).☆☆ Presented in part at the 97th Annual Meeting USCAP (Denver, CO; March 2008) and at the 25th Annual Meeting of the Spanish Society of Anatomic

athology (Zaragoza, May 2011).⁎ Corresponding author.E-mail address: peiro_glo@gva.es (G. Peiró).

046-8177/$ – see front matter © 2013 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.humpath.2012.05.008

219mTOR in endometrial carcinoma

molecular subtypes of endometrial carcinoma with diverse patients' prognosis and support theirdistinctive treatment based on targeted drugs.© 2013 Elsevier Inc. All rights reserved.

1. Introduction BRAF mutations; (4) to study the correlation between the

Table 1 Panel of antibodies for the immunohistochemicalanalysis

Antibody Vendor Clone Dilution

PTEN Dako 6H2.1 1:100β-Catenin Dako β-Catenin-1 1:1mTOR Cell Signaling 49F9 1:100p53 Dako DO7 1:50MLH1 BD-Pharmingen G168-15 1:30MSH2 Calbiochem Ab-2 1:100MSH6 Cell Marque 44 1:1PMS2 Biocare Medical A16-4 1:1

Dako, Copenhagen, Denmark; Cell Signaling, Beverly, MA, USA; BDPharmingen, San Jose, CA, USA; Calbiochem, La Jolla, CA, USA; CellMarque, Rocklin, CA, USA; Biocare Medical, Concord, CA, USA.

Endometrial carcinoma (EC) is the most prevalentmalignant neoplasia of the female genital tract. Womenpresenting tumors in the early stage are mostly alive at a 5-year survival (96%). In contrast, those diagnosed as havingan aggressive histology or cancer in the regional or distantstage have an increased risk of recurrence, and the survivalrate dramatically decreases to 67% and 17%, respectively.Current treatment protocols have only slightly improvedthe overall survival (OS), and they are not exempt fromtoxicity [1]. Therefore, there is a need for better knowledgeof the pathogenic molecular pathways involved in thisdisease that might be the rationale for the development ofnew targeted drugs.

The first pathogenic classification by Bokhman [2] intotypes I (endometrioid carcinomas) and II (nonendometrioidcarcinomas) EC based on their biology, clinical course, andpatient's outcome has been supported by numerous studies.Type I includes molecular changes in PTEN, β-catenin,PIK3CA, K-RAS, IGF1R, or mismatch repair (MMR) genes,along with microsatellite instability (MSI) and, rarely, BRAFmutations. In contrast, type II shows alterations, mainly, inp53, HER2, p16, and E-cadherin [3-5]. Despite the diversemolecular alterations, occasionally, the distinction of bothtypes is not accurate because a minority of tumors may showoverlapping features [6]. Moreover, the prognostic relevanceof most of these biomarkers is still controversial.

The mammalian target of rapamycin (mTOR) is a highlyconserved serine/threonine kinase activated in response toseveral growth factors and nutrients. It is part of 2 distinctmultiprotein complexes, mTORC1 (predominantly cytoplas-mic and highly responsive to rapamycin) and mTORC2(both cytoplasmic and nuclear, and relatively rapamycinresistant). They regulate their respective downstreameffectors p70S6K/4EBP1 and Akt, leading to proteinsynthesis, cell growth, and proliferation [7]. mTOR highexpression in the cytoplasm has been reported in 7% to 70%of ECs of both subtypes, but the results regarding itsassociation with clinicopathologic data are still controversial[8-11]. Recent studies suggest the usefulness of mTORinhibitors in cancer treatment [12-14].

In EC, although evidence has shown the involvement ofPTEN and β-catenin in this neoplasia, little is known aboutthe role of mTOR, as well as their relationship in thepathogenesis and prognosis of these patients. Therefore, theaims of our study were (1) to perform an immunohisto-chemical analysis of mTOR, PTEN, β-catenin, and MMRgene proteins (MLH1, MSH2, MSH6, and PMS2); (2) toanalyze the MSI status; (3) to determine the frequency of

previously mentioned factors; and (5) to evaluate theirclinical relevance in a large series of ECs.

2. Patients and methods

2.1. Study group

We selected 260 nonconsecutive uterine ECs from thesurgical pathology files of 4 departments of pathology(University General Hospital of Alicante, Hospital MarinaBaixa, Hospital Virgen de los Lírios, and Hospital MarinaSalud,Alicante, Spain). The Institutional EthicsCommittee ofeach participating hospital approved the study. Patients' agesranged from 31 to 90 years (median age, 63 years). Medianfollow-upwas 53months (range, 3-156months). In each case,all hematoxylin and eosin slides available were reviewed, and1 to 2 representative blocks containing at least 30% neoplasticcells were selected for immunohistochemical and DNAanalysis. All cases were classified using the 2003 WorldHealth Organization criteria [15]. Additional recorded histo-logic features were the presence of lymph-vascular invasion(LVI), myometrial invasion, and International Federation ofObstetrics and Gynaecology (FIGO) grade and stage.

2.2. Immunohistochemistry

Immunohistochemical staining was performed on paraf-fin-embedded tissue from tissue microarrays (Beechermanual tissue arrayer; Beecher Instruments, Silver Spring,MD, USA) and two 1-mm cores. We used heat-inducedantigen retrieval before exposure to the primary antibody,and the staining was performed on TechMate 500 automatedslide processing equipment (Dako, Copenhagen, Denmark)

220 G. Peiró et al.

by the standard avidin-biotin method. Antibodies andconditions are detailed in Table 1.

mTOR staining in the membrane/cytoplasm versus nucleiwere separately and semiquantitatively scored according tothe percentage of positive cells and intensity (from 0 to 3+)(score 0-300).

Staining of PTEN (cytoplasm) and MMR proteins(MLH1, MSH2, MSH6, or PMS2) (nuclei) were evaluatedas absent (no expression) or present (≥1% positive cells),independently of the intensity. Normal epithelial cells,lymphocytes, and stromal cells served as an internal positivecontrol. For β-catenin, only nuclear staining was consideredpositive, irrespective of the percentage.

2.3. MSI analysis

We extracted genomic DNA (QIAmp DNA miniKit;Qiagen, Hilden, Germany) from 2 to 3 paraffin-embeddedcylinders (1 mm thick) from preselected tumor areas with atleast 30% of tumor cell content, excluding areas of necrosis orhemorrhage. A total of 138 primary ECs had sufficientformalin-fixed, paraffin-embedded tissue available, withadequate quantity and quality. We used a pentaplex polymer-ase chain reaction assaywith 5mononucleotide repeats (NR21,NR24, NR27, BAT25, and BAT26), following the protocol of

Fig. 1 Typical allelic profile of NR27, NR21, NR24, BAT25, and

Suraweera et al [16]. Tumors with microsatellite sequencealterations at 2 or more (≥40%) loci were defined as high-frequency MSI (MSI-H) or positive, low-frequency MSI ifonly 1 of 5 markers was positive, and microsatellite stable(MSS) if none was unstable [17] (Fig. 1).

2.4. Real-time chemistry TaqMan minor groovebinder (MGB) probes

BRAF-V600E mutation detection was performed in 138ECs in an ABI PRISM 7500 (allelic discrimination software;Applied Biosystems, Singapore), based on the design of 2TaqMan probes labeled with a different fluorescent tag (VICand FAM), specific for the wild-type and the mutant alleles,respectively. Two sets of primers-probes and amplificationconditions were applied, according to Benlloch et al [18]. Acolorectal carcinoma cell line (HT29) with a known BRAF-V600E mutation was included as a positive control.

2.5. Statistical analysis

The χ2 and Fisher tests were used to determine thedistribution of the clinical-pathological, immunohistochem-ical, and molecular characteristics between both tumor types.

BAT26 in DNA from MSS (upper) and MSI (lower) tumors.

221mTOR in endometrial carcinoma

A receiver operating characteristic and area under thecurve (AUC) were generated to determine a cutoff value ofthe mTOR expression and the potential clinical utility topredict prognosis (death). Therefore, the cutoff with thelargest AUC was chosen. Kaplan-Meier survival plots andlog-rank tests were performed for the comparison of thesurvival curves, and for multivariate analysis, the Coxproportional hazards model was applied. P values less than.05 were considered statistically significant.

3. Results

To achieve the purpose of our study, 221 adenocarci-nomas were classified into type I (198 endometrioidadenocarcinomas, 16 mixed adenocarcinomas [composedof an admixture of endometrioid carcinoma, including itsvariants, or mucinous adenocarcinomas in which the minortype comprised at least 10% of the tumor], 2 pure

Table 2 Clinicopathologic features, MMR genes, MSI, and BRAF da

PTEN (negative) β-Catenin (nu

Age (y)b50 12 (54) 8 (40)≥50 128 (59) 25 (12)

P = .86 P = .001Histologic grade1 42 (67) 11 (18)2 65 (66) 12 (14)3 33 (43) 10 (13)

P = .002 P = .65Histologic typeI 130 (64) 30 (18)II 10 (28) 0 (0)

P b .0001 P = .006FIGO stageI-II 126 (66) 26 (15)III-IV 12 (25) 6 (15)

P b .0001 P = .94LVIAbsent/ND 86 (65) 21 (17)Present 26 (54) 8 (19)

P = .19 P = .83MMR expressionAll present 95 (53) 24 (14)Any loss 45 (76) 9 (17)

P = .001 P = .59MSI statusMSS/MSI-low 57 (57) 17 (18)MSI-H 22 (76) 4 (14)

P = .05 P = .64BRAF-V600EWild-type 86 (62) 19 (15)Mutated 0 (0) 0 (0)

P = NE P = NE

NOTE: Values are n (%). All statistical comparisons by χ2 test.Abbreviations: ND, nondefinitive; NE, not evaluable.

mucinous adenocarcinomas, and 5 poorly differentiatedadenocarcinomas with nondefinitive histologic features ofpapillary serous carcinomas [PSCs]); 39 were includedinto type II category (33 PSCs, 2 clear cell carcinomas,and 4 mixed tumors with N20% of PSC or clear cellcarcinoma component).

Tumors were predominantly of type I (85%; 221/260) andof FIGO grade 2 (40%; 105/260), had negative LVI (73%;141/192) and superficial myometrial invasion (68%; 142/209), and presented at early stages (I-II) (82%; 204/248).Only 9% (23/260) of the patients were younger than 50 years.

The cutoff point with the highest sensitivity andspecificity for defining high and low mTOR cytoplasmicexpression was set at 55, derived from receiver operatingcharacteristic curve to predict death, which had an AUC (thelargest) of 0.69 (0.58-0.80) (P = .002). mTOR highexpression (cytoplasm) (48%; 114/236), more consistentwith mTORC1, was seen in type II tumors (P = .026)presented at advance stages (P =.046). In contrast, mTOR

ta in relation to PTEN, β-catenin, and mTOR

clear) mTOR (cytoplasm) mTOR (nuclear)

11 (50) 11 (50)103 (48) 99 (46)P = .86 P = .73

30 (45) 35 (53)29 (31) 37 (40)35 (46) 38 (50)P = .07 P = .18

89 (45) 85 (43)25 (68) 25 (68)P = .01 P = .005

74 (40) 82 (45)25 (61) 22 (54)P = .046 P = .29

57 (44) 53 (41)15 (32) 24 (51)P = .13 P = .25

88 (49) 85 (47)26 (46) 25 (44)P = .64 P = .63

35 (34) 15 (15)12 (41) 7 (24)P = .46 P = .22

49 (37) 23 (17)0 (0) 0 (0)

P = NE P = NE

Table 3 Correlation of immunohistochemical results betweenPTEN, β-catenin, and mTOR according to subcellularlocalization

mTOR cytoplasm mTOR nuclear

222 G. Peiró et al.

nuclear expression (any positive) (47%; 110/236), mostconsistent with mTORC2, correlated only with type IIhistology (P = .005) (Table 2 and Fig. 2A). Lower levels ofmTOR were found in PTEN-negative tumors (P = .027), aswell as in those with nuclear β-catenin (P = .017) (Table 3).

Fig. 2 Representative photographs of protein expression formTOR (cytoplasm and focally nuclear) in type II EC (A), PTEN loss(B), and β-catenin in type I EC (C) (original magnification ×200).

Low High P Absent Present P

PTENNegative 80 (58) 57 (42) .027 71 (52) 66 (48) N.05Positive 41 (44) 53 (56) 52 (55) 42 (45)β-CateninNuclear 23 (70) 10 (30) .017 17 (52) 16 (48) N.05Cytoplasm 89 (47) 100 (53) 96 (51) 93 (49)

NOTE: Values are n (%). All statistical comparisons by χ2 test.

However, β-catenin nuclear expression was seen in 18% oftumors with PTEN loss (24/133) and in 10% of those withPTEN present (9/88) (P N .05).

We found PTEN loss (58%; 140/240) (Fig. 2B) in tumorsof type I (P b .0001), FIGO grade 1 (P = .002), FIGO stages Ito II (P b .0001), MMR-defective genes (P = .001), and MSI-H (P = .05). Nuclear β-catenin (range, 1%-95% positivenuclei) (Fig. 2C) was detected in 15% (33/224) of tumors,often found in areas with squamous differentiation, in tumorsof younger patients (P = .001) and of type I (P = .003).

MMR loss of expression (any protein absence) was seenin 24.4% (63/258), and the distribution was as follows:MLH1 in 20% (50/255), MSH2 in 3% (7/255), MSH6 in5% (11/236), and PMS2 in 20% (50/248). Of note,combined MLH1/PMS2 loss was seen in 19% (47/247)of tumors and MSH2/MSH6 loss in 2% (5/234) (P b.0001). Among those with MLH1 loss, 74% presented alsoPTEN loss (P = .021) and similarly occurred for MSH6(90%; P = .047) and PMS2 (75%; P = .015). Nevertheless,the results were not significant for MSH2 probably becauseof the small number of cases (6/7; 86%) (P = .24).Defective MMR genes (any one of them) were found morefrequently among tumors of type I (27%; 60/220) comparedwith type II (8%; 3/38) (P = .008), of early FIGO stages(27%; 55/202 versus 14% in stage III + IV) (P = .059), andwith PTEN loss (32%; 35/140) (P = .001); whereas noassociations were found with age, FIGO grade, LVI,nuclear β-catenin, or mTOR expression either in thecytoplasm or nuclear (all P N .05).

MSI-H was seen in 22% (30/138) of tumors, predom-inantly in NR27 locus (22.4%). There was a positivecorrelation between MMR gene expression loss (P b .0001),type I tumors (22% [29/131] versus 0% [0/7] for type II) (P= .16), and PTEN loss (28%; 22/79) (P = .05), whereas noassociations were found with age, LVI, FIGO grade andstages, β-catenin, or mTOR high expression (all P N .05).No BRAF-V600E mutations were identified (0%; 0/138).

Longer OS (Kaplan-Meier, log-rank test) was observed inpatients with type I tumors (89%; P = .003), FIGO grade 1(97%; P b .0001) and stages I to II (94%; P b .0001),superficial myometrial infiltration (94%; P = .001), no LVI

Table 4 Multivariate analysis (OS) of histologic andbiomarkers (Cox model)

Variables β Hazard ratio 95% CI P

Histologic grade 2.19 8.9 1.1-69.9 .036FIGO stage 1.32 3.7 1.4-10.0 .008LVI 2.55 12.8 3.5-4.65 b.0001PTEN 1.38 3.9 1.4-10.9 .008mTORC1 0.41 1.5 0.5-4.2 N.05

Abbreviation: CI, confidence interval.

223mTOR in endometrial carcinoma

(97%; P b .0001), or younger age (99%; P = .053). Inaddition, patients with PTEN-loss tumors also lived longerthan those with PTEN-present tumors (94% versus 79%,respectively; P = .001). In contrast, mTOR cytoplasmic highexpression implied poorer OS (83% versus 93%; P = .02).Further stratification of mTOR results according to tumorhistology showed that patients with type I mTOR-over-expressing tumors had shorter OS, with similar results inpatients with type II (P = .021) (Fig. 3A and B).Nevertheless, no differences were found for nuclear mTOR(88% versus 86%; P N .05) or β-catenin (92% versus 89%,respectively; P N .05). Differences in OS were neitherobserved for patients whose tumors had MMR loss (89%versus 87%; P N .05), nor when the analysis was individuallyperformed (all P N .05) (data not shown). Similarly, OS didnot differ between those with or without MSI-H (91% versus90%; P N .05). By multivariate analyses, only histologicgrade, stage, LVI, and PTEN showed an independent value(Cox regression) (Table 4).

Fig. 3 A and B, Kaplan-Meier curves (OS) for mTORC1 resultsaccording to EC histology: type I (A) and type II (B).

4. Discussion

Patients with EC who are diagnosed at an early stage havean excellent prognosis that contrast with the high mortalityrate in those with advanced disease. Therefore, currentresearch is mainly focused on elucidating new moleculartargets for more rational treatment. Our study of a large seriesof ECs showed that type I tumors characteristically had PTENloss, defective MMR genes, and MSI-H or β-catenin nuclearaccumulation. In contrast, type II tumors showed increasedlevels of mTOR and lower rates of PTEN loss and defectiveMMR genes, MSI-stable, and no β-catenin accumulation.Our results sustain previous observations that PTEN and β-catenin have a main role in the pathogenesis of type I EC andextend the knowledge that mTOR is mainly involved inaggressive mainly type II tumors. Hence, these might havecrucial implications for the use of targeted therapies.

mTOR localization has been reported critical for tumorprogression and outcomes in a few studies. In our series,mTOR cytoplasmic high expression (mTORC1) was seen in48%, predominantly type II tumors presented at advancestages, and mTOR nuclear expression (47%) (mTORC2)correlated only with type II histology. Previous investigatorshave reported mTOR cytoplasmic high expression in 7% to70% of ECs, either associated with high-grade and high-stagetumors [8] or unrelated with clinicopathologic features [9].Shen et al [19] in a comparative study including 32 ECs with24 proliferative and 23 secretory endometrium analyzed thesubcellular distribution of mTOR (plasmalemmal/cytoplasmicversus nuclear) expression. They showed moderate to strong(2-3+) nuclearmTOR staining in 9 (30%) of 32EC cases but innone of the 47 noncancerous cases. Therefore, the resultssuggested a major role of the rapamycin-insensitive mTORC2pathway in endometrial tumorigenesis. Nevertheless, thisstudy was performed in a small series of patients, and thehistologic type of ECs was not specified.

In our series, only the presence of cytoplasmic mTOR(mTORC1) implied poorer survival, selecting a subset ofpatients at high risk for death in both histologic types. Incontrast, Yoshida et al [20] in a series of 82 patients with typeI EC reported that nuclear mTOR was significantly elevatedin poorly differentiated tumors with lymph node involve-ment and shorter survival, whereas cytoplasmic expressioncorrelated only as a trend, the latter more in line with our

224 G. Peiró et al.

observations. However, longer disease-free survival [11] orno correlations with outcome [9] have also been reported. Itshould be noted that most of the published clinical serieshave the limitations of being small [8,10,19] or the inclusionof only type I tumors [11,20].

Multiple mechanisms have been proposed for mTORdysregulation/activation, including PTEN/PI3K/Akt alter-ations [7]. PTEN is a negative regulator of mTOR.Accordingly, it would be expected that in type I tumorswith PTEN loss, mTOR levels would be predominantlyincreased. Conversely, we found mTOR high expression inaggressive type II tumors, suggesting the role of othergenetic alterations. In agreement, Catasus et al [21] examinedthe expression profiling of 22 involved genes in 38 ECs. Ahierarchical clustering analysis identified 2 subgroups ofhigh-grade ECs with different molecular alterations. Inter-estingly, both clusters included a high proportion of pure ormixed type II tumors with PTEN up-regulation, includingmTOR up-regulation in cluster 1.

Liver kinase B1 (LKB1)/adenosine monophosphate–activated protein kinase (AMPK) pathway regulatesmTORC1 through the phosphorylation of tuberous sclerosiscomplex-2 [14]. LKB1-inactivating mutations and/or pro-moter hypermethylation has been related to endometrialcarcinogenesis. Apparently, LKB1 deregulation results in adifferent phenotype from EC-associated PTEN loss [22].Interestingly, Contreras et al [23] showed in mouse modelsthat heterozygosis for a null LKB1 allele induced, in morethan 50% of mice, an aggressive EC. Therefore, it isplausible that involvement of the LKB1/AMPK pathwaywould partially explain our findings.

Currently, mTOR is considered a rational target forinhibition of the PI3K/Akt or LKB1/AMPK pathways eitheralone or in combination with cytotoxic agents [14]. In fact,phase II trials have demonstrated a beneficial clinicalresponse in advanced and in pretreated patients with recurrenttype I EC. Although no clear advantages of one mTORinhibitor (rapamycin analogues that inhibit the proteincomplex mTORC1) over the others were identified, theresults suggested some antitumor activity, mainly in terms oflong-lasting stable disease [12]. Nevertheless, the dataconcerning mTOR and other related biomarkers by immu-nohistochemistry on tumor biopsies were available in veryfew patients, or the expression status of these biomarkers wasnot even a criterion for the inclusion of the patients in someclinical trials, but only histologic type. More recently,inhibitors targeting both mTORC1 and mTORC2 are indevelopment. Preclinical studies have shown these com-pounds to be more effective inhibitors of proliferation thanrapamycin analogues [13].

Overall, PTEN loss as a result of mutation, deletion, orpromoter hypermethylation was seen in 58% tumors withfavorable features, MMR-defective genes, and MSI pheno-type [5,24]. Furthermore, we also confirmed a longersurvival in patients with PTEN loss tumors [3,25]. Incontrast, association with a worse prognosis in early [26] or

advanced stages [27] has been described, but in relativelysmall series.

MMR-defective genes are responsible for MSI. We found24.4% MMR deficiency associated with MSI-H status,predominantly in type I tumors with PTEN loss, but unrelatedwith patient's age, FIGO grade, LVI, mTOR, and β-catenin.Awide variation (11%–45%) has been reported, which mightbe related with the histologic types included and differenttechniques or markers [17,25,28]. Combined MLH1/PMS2loss occurred in 19% andMSH2/MSH6 loss occurred only in2%, the latter commonly related with germline mutation inhereditary nonpolyposis colorectal cancer–associated tu-mors. Of note, promoter hypermethylation or direct DNAsequencing to confirm gene status was outside the scope ofour study. The MSI prognostic significance is still contra-dictory. Whereas we observed no differences [17,29], othershave reported an association with a worse survival at earlystages [25,28] or favorable outcome [30].

BRAF mutations have been described predominantly inMMR deficiency neoplasias [18]. Our data support thatBRAF is not the target gene for abnormal MMR or MSI insporadic EC [31,32]. However, a 21% incidence associatedwith abnormal MMR function has been reported [33].

β-Catenin was found in 15% of the tumors in youngerpatients, exclusively of type I with focal squamousmetaplasia, but it was unrelated with mTOR, PTEN, MMRgene, MSI, or BRAF status, supporting the independent roleof the Wnt/β-catenin pathway in EC [34,35]. Furthermore,nuclear expression denoted no prognostic significance in ourpatients, in contrast to favorable or poor outcome reported byothers [5].

In conclusion, our data support the independent roles ofmTOR, PTEN, and β-catenin in the carcinogenesis of ECtypes. On the one hand, type I tumors would develop mainlybecause of molecular alterations in PTEN, MMR genes, andMSI or through β-catenin, whereas mTOR is involved in ahigh proportion of type II ECs. Moreover, BRAF mutationsare not relevant. Regarding outcome, mTORC1 highexpression implied worse patients' prognosis, and β-cateninaccumulation was unrelated, but only PTEN loss wasrevealed as an independent good prognostic factor. Furtherstudies are needed to confirm our data to proportion noveltreatment strategies that are based on distinctive molecularfeatures of individual tumors.

Acknowledgments

We thank Patricia Picó and Cristina Pomares for theirtechnical assistance and Doreen A. Denecker for themanuscript preparation.

References

[1] Fleming GF. Systemic chemotherapy for uterine carcinoma: metastaticand adjuvant. J Clin Oncol 2007;25:2983-90.

225mTOR in endometrial carcinoma

[2] Bokhman JV. Two pathogenetic types of endometrial carcinoma.Gynecol Oncol 1983;15:10-7.

[3] Peiro G, Lohse P, Mayr D, Diebold J. Insulin-like growth factor-Ireceptor and PTEN protein expression in endometrial carcinoma.Correlation with bax and bcl-2 expression, microsatellite instabilitystatus, and outcome. Am J Clin Pathol 2003;120:78-85.

[4] Peiro G, Mayr D, Hillemanns P, Lohrs U, Diebold J. Analysis of HER-2/neu amplification in endometrial carcinoma by chromogenic in situhybridization. Correlation with fluorescence in situ hybridization,HER-2/neu, p53 and Ki-67 protein expression, and outcome. ModPathol 2004;17:227-87.

[5] Yeramian A, Moreno-Bueno G, Dolcet X, et al. Endometrialcarcinoma: molecular alterations involved in tumor development andprogression. Oncogene 2012, http://dx.doi.org/10.1038/onc.2012.76[Epub ahead of print].

[6] Voss MA, Ganesan R, Ludeman L, et al. Should grade 3 endometrioidendometrial carcinoma be considered a type 2 cancer—a clinical andpathological evaluation. Gynecol Oncol 2012;124:15-20.

[7] Bhaskar PT, Hay N. The two TORCs and Akt. Dev Cell 2007;12:487-502.

[8] Darb-Esfahani S, Faggad A, Noske A, et al. Phospho-mTOR andphospho-4EBP1 in endometrial adenocarcinoma: association withstage and grade in vivo and link with response to rapamycin treatmentin vitro. J Cancer Res Clin Oncol 2009;135:933-41.

[9] No JH, Jeon YT, Park IA, et al. Expression of mTOR protein and itsclinical significance in endometrial cancer. Med Sci Monit 2009;15:BR301-5.

[10] Wahl H, Daudi S, Kshirsagar M, et al. Expression of metabolicallytargeted biomarkers in endometrial carcinoma. Gynecol Oncol2010;116:21-7.

[11] Choi CH, Lee JS, Kim SR, et al. Clinical significance of pmTORexpression in endometrioid endometrial carcinoma. Eur J ObstetGynecol Reprod Biol 2010;153:207-10.

[12] Delmonte A, Sessa C. Molecule-targeted agents in endometrial cancer.Curr Opin Oncol 2008;20:554-9.

[13] Tsoref D, Oza AM. Recent advances in systemic therapy for advancedendometrial cancer. Curr Opin Oncol 2011;23:494-500.

[14] van Veelen W, Korsse SE, van de Laar L, Peppelenbosch MP. The longand winding road to rational treatment of cancer associated withLKB1/AMPK/TSC/mTORC1 signaling. Oncogene 2011;30:2289-303.

[15] Silverberg SG, Kurman RJ, Nogales F. Tumours of the uterinecorpus. In: Tavassoli FA, Devilee P (eds). Pathology and genetics oftumours of the breast and female genital organs. IARC Press: Lyon2003. p. 217-28.

[16] Suraweera N, Duval A, Reperant M, et al. Evaluation of tumormicrosatellite instability using five quasimonomorphic mononucleo-tide repeats and pentaplex PCR. Gastroenterology 2002;123:1804-11.

[17] Peiro G, Diebold J, Lohse P, et al. Microsatellite instability, loss ofheterozygosity, and loss of hMLH1 and hMSH2 protein expression inendometrial carcinoma. HUM PATHOL 2002;33:347-54.

[18] Benlloch S, Paya A, Alenda C, et al. Detection of BRAF V600Emutation in colorectal cancer: comparison of automatic sequencingand real-time chemistry methodology. J Mol Diagn 2006;8:540-3.

[19] Shen Q, Stanton ML, FengW, et al. Morphoproteomic analysis revealsan overexpressed and constitutively activated phospholipase D1-mTORC2 pathway in endometrial carcinoma. Int J Clin Exp Pathol2010;4:13-21.

[20] Yoshida Y, Kurokawa T, Horiuchi Y, Sawamura Y, Shinagawa A,Kotsuji F. Localisation of phosphorylated mTOR expression is criticalto tumour progression and outcomes in patients with endometrialcancer. Eur J Cancer 2010;46:3445-52.

[21] Catasus L, D'Angelo E, Pons C, Espinosa I, Prat J. Expressionprofiling of 22 genes involved in the PI3K-AKT pathway identifiestwo subgroups of high-grade endometrial carcinomas with differentmolecular alterations. Mod Pathol 2010;23:694-702.

[22] Alessi DR, Sakamoto K, Bayascas JR. LKB1-dependent signalingpathways. Annu Rev Biochem 2006;75:137-63.

[23] Contreras CM, Gurumurthy S, Haynie JM, et al. Loss of Lkb1provokes highly invasive endometrial adenocarcinomas. Cancer Res2008;68:759-66.

[24] Risinger JI, Hayes K, Maxwell GL, et al. PTEN mutation inendometrial cancers is associated with favorable clinical andpathologic characteristics. Clin Cancer Res 1998;4:3005-10.

[25] Mackay HJ, Gallinger S, Tsao MS, et al. Prognostic value ofmicrosatellite instability (MSI) and PTEN expression in women withendometrial cancer: results from studies of the NCIC Clinical TrialsGroup (NCIC CTG). Eur J Cancer 2010;46:1365-73.

[26] Athanassiadou P, Athanassiades P, Grapsa D, et al. The prognosticvalue of PTEN, p53, and beta-catenin in endometrial carcinoma: aprospective immunocytochemical study. Int J Gynecol Cancer2007;17:697-704.

[27] Uegaki K, Kanamori Y, Kigawa J, et al. PTEN-positive andphosphorylated-Akt–negative expression is a predictor of survivalfor patients with advanced endometrial carcinoma. Oncol Rep2005;14:389-92.

[28] Steinbakk A, Malpica A, Slewa A, et al. Biomarkers and microsatelliteinstability analysis of curettings can predict the behavior of FIGO stageI endometrial endometrioid adenocarcinoma. Mod Pathol 2011;24:1262-71.

[29] Zighelboim I, Goodfellow PJ, Gao F, et al. Microsatellite instabilityand epigenetic inactivation of MLH1 and outcome of patients withendometrial carcinomas of the endometrioid type. J Clin Oncol2007;25:2042-8.

[30] Black D, Soslow RA, Levine DA, et al. Clinicopathologic significanceof defective DNA mismatch repair in endometrial carcinoma. J ClinOncol 2006;24:1745-53.

[31] Moreno-Bueno G, Sanchez-Estevez C, Palacios J, Hardisson D,Shiozawa T. Low frequency of BRAF mutations in endometrial and incervical carcinomas. Clin Cancer Res 2006;12:3865 [author reply3865-6].

[32] Kawaguchi M, Yanokura M, Banno K, et al. Analysis of a correlationbetween the BRAF V600E mutation and abnormal DNA mismatchrepair in patients with sporadic endometrial cancer. Int J Oncol2009;34:1541-7.

[33] Feng YZ, Shiozawa T, Miyamoto T, et al. BRAF mutation inendometrial carcinoma and hyperplasia: correlation with KRAS andp53 mutations and mismatch repair protein expression. Clin CancerRes 2005;11:6133-8.

[34] Moreno-Bueno G, Hardisson D, Sanchez C, et al. Abnormalities of theAPC/beta-catenin pathway in endometrial cancer. Oncogene 2002;21:7981-90.

[35] Bansal N, Yendluri V, Wenham RM. The molecular biology ofendometrial cancers and the implications for pathogenesis, classifica-tion, and targeted therapies. Cancer Control 2009;16:8-13.