Gynecologic Oncology xxx (2015) xxx–xxx

YGYNO-975829; No. of pages: 7; 4C:

Contents lists available at ScienceDirect

Gynecologic Oncology

j ourna l homepage: www.e lsev ie r .com/ locate /ygyno

Immunohistochemical and genetic profiles of endometrioid endometrialcarcinoma arising from atrophic endometrium

Yvette P. Geels a,1, Louis J.M. van der Putten a,⁎,1, Angela A.G. van Tilborg a,b, Irene Lurkin c, Ellen C. Zwarthoff c,Johanna M.A. Pijnenborg d, Saskia H. van den Berg-van Erp e, Marc P.L.M. Snijders f, Johan Bulten b,Daniel W. Visscher g, Sean C. Dowdy h, Leon F.A.G. Massuger a

a Department of Obstetrics and Gynaecology, Radboud University Medical Center, Nijmegen, The Netherlandsb Department of Pathology, Radboud University Medical Center, Nijmegen, The Netherlandsc Department of Pathology, Erasmus Medical Center, Rotterdam, The Netherlandsd Department of Obstetrics and Gynecology, TweeSteden Hospital, Tilburg, The Netherlandse Department of Pathology, Canisius-Wilhelmina Hospital, Nijmegen, The Netherlandsf Department of Obstetrics and Gynecology, Canisius-Wilhelmina Hospital, Nijmegen, The Netherlandsg Department of Anatomic Pathology, Mayo Clinic, Rochester, MN, USAh Division of Gynecologic Surgery, Mayo Clinic, Rochester, MN, USA

H I G H L I G H T S

• Endometrium next to endometrioid endometrial carcinomas are mostly hyperplastic, but sometimes atrophic, which predicts a worse prognosis.• The immunohistochemical and genetic profiles of endometrioid carcinomas next to hyperplastic and atrophic endometrium was assessed and compared.• Carcinomas next to atrophic endometrium were associated with fewer KRAS mutations and loss of E-cadherin.

⁎ Corresponding author at: Radboud University Meof Obstetrics and Gynaecology, P.O. Box 9101, 6500HBTel.: +31 243617768; fax: +31 243668597.

E-mail address: Louis.vanderputten@radboudumc.nl (1 Both authors contributed equally to this work.

http://dx.doi.org/10.1016/j.ygyno.2015.03.0070090-8258/© 2015 Elsevier Inc. All rights reserved.

Please cite this article as: Geels YP, et al, Immendometrium, Gynecol Oncol (2015), http://

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 22 August 2014Accepted 5 March 2015Available online xxxx

Keywords:Endometrial carcinomaEndometrioidBackground endometriumImmunohistochemistryGenetical

Objective. Endometrial carcinomas are divided into type I endometrioid endometrial carcinomas (EECs),thought to arise from hyperplastic endometrium, and type II nonendometrioid endometrial carcinomas, thoughtto arise from atrophic endometrium. However, aminority (20%) of EECs have atrophic background endometrium,which was shown to be a marker of a worse prognosis. This study compares the immunohistochemical andgenetic profiles of this possible third type to that of the known two types.

Methods. 43 patients with grade 1 EEC and hyperplastic background endometrium (type I), 43 patients withgrade 1 EEC and atrophic background endometrium (type III) and 21 patients with serous carcinoma (type II)were included (n = 107). Tissue microarrays of tumor samples were immunohistochemically stained forPTEN, L1CAM, ER, PR, p53, MLH1, PMS2, β-catenin, E-cadherin and MIB1. The BRAF, KRAS, and PIK3CA geneswere analyzed for mutations.

Results. A significantly higher expression of ER and PR, and a lower expression of L1CAM, p53 andMLH1werefound in type I and III compared to type II carcinomas. Expression of E-cadherinwas significantly reduced in typeIII compared to type I carcinomas. Mutation analysis showed significantly less mutations of KRAS in type IIIcompared to type I and II carcinomas (p b 0.01).

Conclusion. There appear to be slight immunohistochemical and genetic differences between EECs withhyperplastic and atrophic background endometrium. Carcinogenesis of EEC in atrophic endometrium seems tobe characterized by loss of E-cadherin and a lack of KRAS mutations. As expected, endometrioid and serouscarcinomas were immunohistochemically different.

© 2015 Elsevier Inc. All rights reserved.

dical Center, 791 Department, Nijmegen, The Netherlands.

L.J.M. van der Putten).

unohistochemical and geneticdx.doi.org/10.1016/j.ygyno.2

1. Introduction

Cancer of the uterine corpus is the most common gynecologicmalignancy among women in the developed world. In 2012, it affected47,130 women and caused the death of 8010 women in the US [1].

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007

2 Y.P. Geels et al. / Gynecologic Oncology xxx (2015) xxx–xxx

It is generally accepted that endometrial carcinomas (EC) can bedivided into two subtypes [2]. Type I endometrial carcinoma is themost common subtype. It affects women at a median age of 60 yearsand has a good prognosis. These tumors are usually related to unop-posed estrogen stimulation and show endometrioid histology, arisingfrom hyperplastic endometrium. In contrast, the less common type IIcarcinomas affect older women and have a poor prognosis. Thesetumors are not related to unopposed estrogen stimulation and arecharacterized by clear cell or serous histology, arising from atrophicendometrium [3–5]. Distinct carcinogenic pathways have beendescribed in each subtype. Type I carcinomas are characterized bymicrosatellite instability and alterations of the PTEN, KRAS, PIK3CA andCTNNB1 genes, whereas type II carcinomas are often aneuploid andshow over expression of p53 and Her2/neu [6–9].

However, some tumors do not fit within this dualistic model. In arecent study we reviewed slides from 527 patients with grade 1endometrioid endometrial carcinomas and found that 17% of thesecarcinomas had atrophic background endometrium [10]. Furthermore,the presence of atrophic background endometrium adjacent to EECwas associatedwith several predictors of poor survival, and an indepen-dent predictor of reduced progression free survival in endometrioidendometrial carcinomas. Moreover, recent studies looking at themolecular basis of endometrial carcinomas also show that this dualisticmodel is too simplistic and propose to categorize them based on theirmolecular profile [11–13]. These studies show that within the twogroups as defined by the dualistic model, it is able to distinguish certaingroups solely based on their molecular pattern. It might be possible thatendometrioid endometrial carcinomas with atrophic backgroundendometrium have a different molecular and therefore immunohisto-chemical pattern when compared to endometrioid endometrial carci-nomas with hyperplastic background endometrium.

The aim of the present study was to analyze the hypothesis thatendometrioid endometrial carcinomas with a background of atrophicendometrium arise through different carcinogenic pathways than typeI and II endometrial carcinomas. Therefore, the expression of severalimmunohistochemical markers and the presence of distinct geneticmutations in endometrioid endometrial carcinoma with a backgroundof atrophic endometrium was compared to those of type I and IIcarcinomas.

2. Materials and methods

2.1. Patients

For this study, patients with endometrial carcinoma from twocohorts, who were at least treated with a hysterectomy and bilateralsalpingo-oophorectomy and who did not have a personal history ofmalignancy, were evaluated for inclusion. The first cohort is comprisedof patients treated for grade 1 endometrioid endometrial carcinomaat the Radboud University Medical Center (Radboudumc) and theCanisius-Wilhelmina Hospital in Nijmegen, The Netherlands, betweenJanuary 1999 and December 2009, and at the Mayo Clinic in Rochester,Minnesota, USA, between January 2002 and December 2008 [10]. Thesecond cohort is comprised of patients with uterine serous carcinomatreated at the Radboud University Medical Center and the Canisius-Wilhelmina Hospital, Nijmegen between January 1999 and December2009 [14,15].

Slides of the primary carcinoma and background endometriumfrom the cohort of patients with grade 1 endometrioid endometrialcarcinoma were reviewed with special attention to the nature of thebackground endometrium by experienced pathologists (JB, SB, DV)who were unaware of the original pathology results and clinicaloutcome. In the case of doubt or discrepancywith the original pathologyreport a second review was performed by another pathologist andconsensus was reached. Background endometrium was categorized assimple hyperplasia, simple atypical hyperplasia, complex hyperplasia,

Please cite this article as: Geels YP, et al, Immunohistochemical and geneticendometrium, Gynecol Oncol (2015), http://dx.doi.org/10.1016/j.ygyno.2

complex atypical hyperplasia, disordered proliferative, atrophic, andnormal proliferative as previously described [5,10,16]. Some cases hadto be excluded because the tumor covered the entire cavity of the uterusand there was no background endometrium to be evaluated.

All patients with grade I endometrioid endometrial carcinoma and abackground of pure (100%) atrophic endometrium (abbreviated to typeIII) as well as a similar amount of patients with grade I endometrioidendometrial carcinoma and a background of hyperplastic endometrium(type I) were included. Subsequently, all patients from the uterineserous carcinoma (type II) cohort of whom uterine tissue could beretrieved from the archive were included. This cohort consisted ofcarcinomas with both pure and mixed serous histology. It has beenpreviously described that only about half of the serous carcinomashave pure serous histology [17].

2.2. Tissue microarray and immunohistochemistry

Tissue microarrays were created from the primary carcinoma [18].Two representative areas of the carcinomawere selected on hematoxy-lin and eosin-stained slides. For the type II cases, areas with pure seroushistology were selected. Two cylinders with a diameter of 2 mm werepunched out of every donor block from the selected areas, andmountedinto a recipient paraffin block using the Tissue-Tek Quick-Ray (SakuraFinetek, Torrance, CA, U.S.) manual tissue microarrayer.

The tissue microarrays were cut in 4 μm slides and immuno-histochemically stained. Several markers were selected to be stained,based on the difference in their expression in type I and type II endome-trial carcinoma [6–9,19,20]. An overview of the antibodies and dilutionsused as well as the area of the cell whichwas evaluated when scoring isshown in Table 1.

Immunohistochemical analysis of PTEN, L1CAM, ER, PR, p53, MLH1,PMS2, β-catenin, E-cadherin and MIB1 expression was performedaccording to the local protocols. These markers were chosen becauseprevious literature has shown that their expression is different in typeI and II EC [8,9,20]. In short, formalinfixedparaffin sectionswere stainedwith the primary antibody following EDTA antigen retrieval, blocking ofendogenous backgroundwith Peroxidase Blocking Reagent and proteinblocking using horse serum. Subsequently, a secondary antibody wasadded and visualization was performed with Vectastain and 3,3′-Diaminobenzidine (Zymed Lab. California, USA) as a substrate. Stainingwas enhanced in CuSO4 and slides were counterstained with Mayer'shematoxylin. Finally, slides were dehydrated and mounted.

Tumor samples were given a score ranging from 0–9 by twoindependent evaluators (YG, AT) whichwas the product of the percent-age of cells stained (0% = 0; 1–10% = 1; 11–50% = 2; 51–100% = 3)and intensity of staining (none = 0; weak = 1; moderate = 2;strong = 3) [21]. The evaluators were unaware whether the tissuecylinders were from type I, type II or type III carcinomas. Samples withtoo little tissue to assess or samples not containing anymalignant tissuewere not included in the calculations. In the case of a large discrepancybetween the scores of the two evaluators (i.e., a difference in percentageor intensity score N2 or disagreement on the presence of malignanttissue) a third independent reviewer (JB), who was unaware of thescore given by the first evaluators, scored the sample as well.

The final score per case (range 0–9) was calculated by adding allscores given to the two tissue samples and dividing themby the numberof scores in the sum (which varied depending on the presence of tumortissue in the sample and the need for a third review). The final semi-quantitative score was used for analysis of PTEN, ER and PR accordingto the literature as well as for β-catenin and E-cadherin, because thereis no consensus as to which scoring system should be used [22,23].L1CAM was considered positive when there was staining seen in atleast 10% of the malignant cells [24]. Cases with a final score of fourand up concerning p53 were considered to be positive while thosewith a score below four were negative [25]. MLH1 and PMS2 wereconsidered lost when there was no scoring at all, according to the

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007

Table 1Antibodies and dilutions used in this study, as well as the areas of the cell used for scoring.

Antibody Dilution Company Area scored

PTEN inactivation 6H2.1 1:100 Dakoa Cytoplasm/nucleusL1CAM UJ127 1:100 Thermo Scientificb MembraneER expression SP1 1:80 Thermo Scientific NucleusPR expression PgR 636 1:250 Dako NucleusP53 mutations DO-7 1:400 Thermo Scientific NucleusLoss of MLH1 G168-15 1:50 BDc NucleusLoss of PMS2 A16-4 1:100 BD Nucleusβ-Catenin alteration 14/Beta-Catenin 1:100 BD MembraneE-cadherin alteration SPM471 1:300 Thermo Scientific MembraneHigh proliferation rate MIB1 1:100 Thermo Scientific Nucleus

a Dako, Glostrup, Denmark.b Thermo Fisher Scientific Inc., Waltham, Massachusetts, US.c Becton, Dickinson and Company, Franklin Lakes, New Jersey, US.

3Y.P. Geels et al. / Gynecologic Oncology xxx (2015) xxx–xxx

international guidelines. Finally, MIB1 staining was considered positiveat every intensity and categorized as 1–10%, 11–50% and 51–100% cellpositive.

2.3. Mutation analysis

Slides with at least 10% representative tumor tissue were selectedfor DNA isolation. For the cases from the Mayo Clinic, the two TMAtumor biopsies were used instead. DNA was isolated with TET-lysisbuffer (10 mmol/L Tris–HCl, pH 8.5; 1 mmol/L EDTA, pH 8; 0.1%Tween-20) containing 5% Chelex-100 (Bio-Rad, Hercules, CA). Proteindigestion was performed by adding proteinase K to each sample andincubation at 56 °C for 48 h. Next, Protein K was inactivated at 95 °Cfor 10 min. The samples were centrifuged for 10 min at 14.000 rpm(RT) and the DNA concentration of the supernatant was measuredusing the Quant-iT Picogreen dsDNA Assay Kit (Invitrogen, Carlsbad,CA, U.S.) before storage at 4 °C. For the detection of mutations, DNAwas amplified for exons of the KRAS, BRAF and PIK3CA genes usingearlier published PCR primers [26]. The amplified exons were assessedfor mutations at 22 nucleotide positions by single nucleotide probeextension assays using a SNaPshot Multiplex Kit (Applied Biosystems,Foster City, CA) as described previously [26,27].

2.4. Statistical analysis

The differences between the immunohistochemical marker scoresof the three endometrial carcinoma subtypes were calculated usingthe Mann–Whitney-U test, the chi square test and Fisher's exact test.Differences between the amounts of cases with genetic mutations persubtype were calculated using the chi square test and Fisher's exacttest. As a measure for tumor heterogeneity the correlation betweenthe scores given to the two biopsies per tumor was calculated usingPearson's correlation coefficient. This test was also used to calculatethe correlation between differentmarkers. Differences were consideredto be statistically significant at a p-value ≤ 0.05. SPSS version 20 (SPSSIBM, New York, NY, U.S.) statistical software was used for analysis ofthe data.

2.5. Ethical approval

No ethical approval was needed for this study,whichwas performedaccording to the Code for Proper Secondary Use of Human Tissue (DutchFederation of Biomedical Scientific Societies, www.federa.org).

3. Results

3.1. Patients

Of the 527 patients with grade 1 endometrioid endometrialcarcinoma evaluated, background endometrium was hyperplastic in

Please cite this article as: Geels YP, et al, Immunohistochemical and geneticendometrium, Gynecol Oncol (2015), http://dx.doi.org/10.1016/j.ygyno.2

387 (73%) and atrophic in 88 (17%). The background endometriumwas normal premenopausal, proliferative in 27 (5%) patients andcould not be assessed due to the size of the tumor in 25 (5%) patients.Pure atrophy was found in 43 (48.9%) of the 88 patients with atrophicbackground endometrium. Out of the 387 patients with hyperplasticbackground endometrium, 43 (11.1%) were randomly selected ascontrols. From the cohort of patients with uterine serous carcinoma,all patient data as well as tissue was present in 21 cases (out of 47meeting the inclusion criteria), 14 (66.7%) of which had pure seroushistology, which is slightly higher than that in the general populationof patients with uterine serous carcinoma [17]. In total, 107 caseswere included for immunohistochemical and mutation analyses.

3.2. Immunohistochemistry

After initial evaluation of 2140 tissue samples, additional reviewbecause of discrepancy between the evaluators' scores was necessaryfor 107 samples (44 type I, 19 type II and 44 type III). This was equallydivided among all markers. Three type III cases were excluded fromthe analysis because both tumor samples contained too little represen-tative tissue or because they did not contain malignant tissue (one ERcase, one MLH1 case and one MIB1 case). The final marker scores pertype of endometrial carcinoma and the differences between the scoresare shown in Fig. 1.

The correlation between the scores given to the two biopsiesfrom the same tumor as a measurement of tumor heterogeneity isshown per marker in Table 2, categorized as high (r between 0.7–0.9),moderate (r between 0.4 and 0.7) and weak (r between 0.1–0.4),according to Dancey and Reidy [28]. It was moderate to high formost markers, only the scores for β-catenin were not correlated(r = −0.01, p = 0.92).

There was a strong positive correlation between the presence ofPMS2 and MLH1 (r = 0.78, p b 0.01) as well as between the presenceof ER and PR (r = 0.70, p b 0.01). L1CAM positivity had a slightly nega-tive correlation with ER (r = −0.33, p b 0.01) and PR (r = −0.42,p b 0.01).

When looking at the differences between the subtypes it can be seenthat in both type I and III EC expression of ER and PR was significantlyhigher than that in type II EC, while expression of L1CAM, p53 andMLH1 was significantly lower in type I and III EC than in that in type IIEC. Expression of E-cadherin and MIB1 was significantly lower in typeIII EC compared to type I and II EC. Examples of E-cadherin and MIB1staining in type I and III carcinomas are shown in Fig. 2.

3.3. Mutation analysis

A representative part of the tumor could be retrieved for every caseand DNA yield was adequate for every sample when assessed by spec-tral photometry. Mutation analysis of the PIK3CA gene was successfulin all cases, of the BRAF gene in all but one type I case and the KRAS

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007

Fig. 1. The different marker scores per type of endometrial carcinoma. p-Values of the Mann–Whitney test (for the boxplots) and chi-square test (for the other graphs) comparing theindividual types are shown.

4 Y.P. Geels et al. / Gynecologic Oncology xxx (2015) xxx–xxx

gene in all but two type I, one type II and two type III cases. The distribu-tion of wild type and mutated genes per type as well as the differencesbetween types is shown in Table 3.

The only significant difference was foundwith the KRAS gene, whichwasmutated in 15/43 (37.2%) of the type I and 6/21 (23.8%) of the typeII carcinomas, compared to only 1/43 (2.3%) of the type III carcinomas.In the type I carcinomas eight of the mutations were located on codon

Please cite this article as: Geels YP, et al, Immunohistochemical and geneticendometrium, Gynecol Oncol (2015), http://dx.doi.org/10.1016/j.ygyno.2

12, five on codon 13, one on codon 19 and one on codon 61 and in thetype II carcinomas three mutations were located on codon 12 and twoon codon 13. The mutation in the type III carcinoma was located oncodon 19.

The BRAF genewas notmutated in any of the cases, while the PIK3CAwasmutatedmore often in type II EC (23.8%) than in type I (14%) and III(11.6%) EC, but these differences were not significant.

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007

Table 2Marker heterogeneity, depicted as the strength of the correlation (according to Dancey and Reidy) between the score of the two biopsies per marker.

Strong Moderate Weak

PTEN (r⁎ = 0.72, p† b 0.01) L1CAM (r = 0.44, p b 0.01) β-Catenin (r = −0.01, p = 0.92)ER (r = 0.80, p b 0.01) p53 (r = 0.55, p b 0.01)PR (r = 0.78, p b 0.01) PMS2 (r = 0.48, p b 0.01)MLH1 (r = 0.78, p b 0.01) E-cadherin (r = 0.56, p b 0.01)

MIB1 (r = 0.60, p b 0.01)

⁎ Pearson's correlation coefficient.† p-Value for Pearson's correlation coefficient.

5Y.P. Geels et al. / Gynecologic Oncology xxx (2015) xxx–xxx

4. Discussion

This study compared the immunohistochemical and genetic profilesof endometrioid endometrial carcinomaswith hyperplastic backgroundendometrium (type I) to thosewith atrophic background endometrium(hypothetical type III). While they were quite comparable,endometrioid carcinomas with atrophic background endometriumshowed reduced expression of E-cadherin and less KRAS mutationscompared to endometrioid carcinomas with hyperplastic backgroundendometrium. The endometrioid carcinomas were also compared tothosewith serous histology (type II), which had a different immunohis-tochemical profile as expected.

4.1. Immunohistochemical profile

The immunohistochemical differences between type I and IIendometrial carcinoma in this study are in linewith previous literature.As expected, ER and PR expression were significantly higher in type Icarcinomas, while L1CAM, p53 and MLH1 were significantly higherin type II carcinomas. Moreover, while the differences in PMS2,E-cadherin, β-catenin and MIB1 expression were not significant, theobserved trends were according to the findings in previous literatureon these markers. Interestingly, PTEN expression was found to belowest in serous carcinomas, while previous studies have shown lossof PTEN in endometrioid carcinomas [8,9,20].

It was hypothesized that type III carcinomas would have a distinctimmunohistochemical profile. For most markers, there was no differ-ence in expression between type I and III. However, expression ofE-cadherin and MIB1 were significantly reduced in type III comparedto type I carcinomas.

Loss of E-cadherin has been described in both endometrioid andnon-endometrioid carcinomas, while expression levels were found tobe normal in hyperplastic endometrium [29–32]. In epithelial cells,E-cadherin is the major molecule of the cadherin family, which isessential for tight cell–cell connections [33]. Reduced expression ofE-cadherinwas associatedwith lymphnodemetastasis, deepmyometrialinvasion, tumor dedifferentiation and advanced disease stage [29–32].

Fig. 2. Examples of E-cadherin (A) and MIB1 (B) expression in

Please cite this article as: Geels YP, et al, Immunohistochemical and geneticendometrium, Gynecol Oncol (2015), http://dx.doi.org/10.1016/j.ygyno.2

Loss of E-cadherin is likely to be responsible, at least partially, for thefact that type III carcinomas are associatedwith lymph nodemetastasis,deep myometrial invasion and advanced disease stage as well [10].Moreover, while increased expression of MIB1, a proliferation markerassociated with aggressive tumors, might seem more likely in type IIIcarcinomas, we found the opposite to be true. However, this findingis in line with the conclusions from a recent publication showing theprognostic value of E-cadherin over proliferation markers [34].

As expected, staining was slightly heterogeneous for most markers.Only β-catenin staining showed very strong heterogeneity, the reasonfor which is unknown. Finally, a strong correlation between thepresence of ER and PR as well as MLH1 and PMS2 was shown, whichis in line with the literature [22,35]. A negative correlation betweenthe presence of L1CAM and the loss of ER and PR was also shown,though weaker than previously described [20].

4.2. Genetic profile

In line with previously published data, BRAF mutations were foundin none of the cases [36]. PIK3CAmutations are associated with invasivegrowth and poor prognosis and were most frequently found in type IIcarcinomas, in accordance with previous findings [37–40].

Mutation analysis of KRAS, which have been described in 30% of thetype I, but only 10% of the type II carcinomas showed several interestingresults [9]. In type I carcinomaswe observed a slightly higher amount ofKRASmutations than previously reported, whichmight be explained bythe fact that most studies only look at codon 12 mutations, while welooked at mutations at codons 12, 13, 61 and 19. Most of the KRASmutations that we found were located on codons 12, 13 and 61, whichare the most likely locations of KRASmutations according to the litera-ture [41]. Only in one type II and in the only mutated type III case wasthe mutation on codon 19, an infrequent location of mutations insome tumors that has not been described in endometrial carcinomasbefore [42]. Codon 19 mutations of KRAS have been suggested to alterthe treatment response of colorectal carcinomas [42]. However, thisdoes not explain the high amount of KRASmutations that we observedin type II carcinomas. This finding might be explained by the fact that

type I (left) and type III (right) endometrial carcinomas.

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007

Table 3Results of the mutation analysis per type of endometrial carcinoma.

Type I Type II Type III p (I–II)a p (I–III)a p (II–III)a

BRAFWild type 42 (97.7%) 21 (100%) 43 (100%) – – –

Mutated – – –

Unknown 1 (2.3%) – –

KRASWild type 26 (60.5%) 14 (66.7%) 40 (93%) 0.58 b0.01 b0.01Mutated 15 (34.9%) 6 (28.6%) 1 (2.3%)Unknown 2 (4.7%) 1 (4.8%) 2 (4.7%)

PIK3CAWild type 37 (86%) 16 (76.2%) 38 (88.4%) 0.48 1.00 0.28Mutated 6 (14%) 5 (23.8%) 5 (11.6%)Unknown – – –

a p-Value for Fisher's exact test (bold values had a p b 0.05 and were consideredsignificant).

6 Y.P. Geels et al. / Gynecologic Oncology xxx (2015) xxx–xxx

we chose to include representative type II cases, some of which had aminor component of non-serous histology [17]. The KRASmutations inthree of the six type II cases might have been present in minor areaswith endometrioid histology instead of those with serous histology[43]. The discrepancy between previously published results and theamount of mutations that we observed in type II carcinomas wasmuch larger. Finally, the amount of mutations in the KRAS gene in thisstudy was significantly lower in type III compared to type I and IIcarcinomas.

KRASmutations have been found to be an early event in the carcino-genesis of endometrioid endometrial carcinomas, present not only inendometrioid carcinoma, but endometrial hyperplasia as well [43,44].These findings support the lack of KRAS mutations in endometrioidcarcinomas arising from atrophic endometrium.

Ourfindings in type III carcinomas of loss of E-cadherinwith a lack ofKRAS mutations are reported in several other studies. Up-regulation ofE-cadherin expression by KRAS mutations was demonstrated in onestudy, while another describes expression of Zeb1, a transcription factorrepressing E-cadherin expression, in tumor cell lines that growindependent of KRAS mutations [45,46]. These findings might suggestthat loss of E-cadherin is an early event of carcinogenesis of type IIIcarcinomas because there is a lack of KRASmutations.

4.3. Implications for the classifications of endometrial carcinomas

In 1983 endometrial carcinomas were subdivided into two distinctmorphological subgroups and this division has played a distinct role inthe management of endometrial carcinomas ever since [2]. Thismorphological subdivision has proven to be very useful and in morerecent yearswas also shown to be present on the immunohistochemicaland genetic levels [8,9]. However, recent immunohistochemical andgenetic studies have also shown that a subdivision into two groups istoo simplistic [11–13]. Moreover, even on a morphological level thereis reason to believe that this subdivision into type I carcinomas arisingfrom hyperplastic endometrium and type II carcinomas arising fromatrophic endometrium is incomplete. Studies from our group have notonly shown that 17% of the endometrioid carcinomas have in factatrophic background endometrium – and a worse prognosis – but thatthe background endometrium of 9.3% of the pure and 34.7% of themixed serous carcinomas was in fact hyperplastic [10,15].

The results of the current study, show additional value of immuno-histochemistry and genetics next to morphology, which is in line withother recent studies. While studies by McConechy et al. and The CancerGenome Atlas have shown that type I and II carcinomas both have adistinct genetic profile, they also show the existence of more subgroupsbased on the genetic profile [11,12].McConechy et al. describe the use ofgenetics to aid in distinguishing cases with aworse prognosis which arehard to distinguish morphologically, for example grade 3 endometrioidcarcinomas and serous carcinomas [12]. Interestingly, while the TCGA

Please cite this article as: Geels YP, et al, Immunohistochemical and geneticendometrium, Gynecol Oncol (2015), http://dx.doi.org/10.1016/j.ygyno.2

publication only looked at the overlap between their four genetic sub-types and the twomorphological subtypes, they do describe two groupswith a low number of KRAS mutations, which have a bad tointermediate prognosis [11].

Based on our findings as well as the growing amount of literaturedescribing alternative subdivisions of endometrial carcinomas, webelieve that it is important to take a closer look at the overlap betweennewly found immunohistochemical and genetic profiles. It is, forexample, unknown whether serous carcinomas with hyperplasticbackground endometrium have a distinct immunohistochemical andgenetic profiles. Future research should focus on combiningmorphology,immunohistochemistry and mutation analyses in a way which can beused to identify patients with a worse prognosis more accurately, asproposed by Chiang et al. [47].

4.4. Strengths and weaknesses of the study

This study is the first to analyze the immunohistochemical andgenetic profiles of endometrioid endometrial carcinomas with atrophicbackground endometrium. A large set of immunohistochemicalmarkers and genes was analyzed to give a clear view of the similaritiesand differences between the different endometrial tumor types. Whilemany recent publications have shown that there are more than twotypes of endometrial carcinoma based on their molecular profile, thisstudy shows that this is also true on an immunohistochemical level.It also shows that there is a significant heterogeneity within tumors,making a clear subdivision difficult.

A weakness of the study may be the fact that DNAwas extracted formutation analysis from whole slides instead of selected tumor tissue.However, while this might be responsible for the high mutation ratethatwe found in type II carcinomas, it does not interferewith answeringthe question whether there is a difference between type I and IIIcarcinomas. If anything, it highlights more clearly that very few KRASmutations are found in type III carcinomas as well as the surroundingtissue. Lastly, it might have been better to match the patient character-istics of the patients with type I and III carcinomas, because it is possiblethat the latter have a higher age and a lower BMI compared to theformer, possibly influencing the carcinogenesis.

5. Conclusion

In conclusion, on the immunohistochemical and genetic levels,endometrioid carcinomas arising from atrophic background endometri-um were shown to be quite comparable to endometrioid carcinomasarising from hyperplastic background endometrium. However, whileKRAS mutations are an early event in carcinogenesis of endometrioidendometrial carcinoma in hyperplastic endometrium, these mutationswere rarely seen in endometrioid carcinomaswith atrophic backgroundendometrium. Instead, carcinogenesis of these carcinomas seemsto be characterized by loss of E-cadherin, which is uncommon inendometrioid carcinomas arising from hyperplastic endometrium.Immunohistochemical and genetic profiling might aid in distinguishingthese cases which were previously shown to have a worse prognosis.

Conflict of interest

The authors have nothing to declare.

References

[1] Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62:10–29.

[2] Bokhman JV. Two pathogenetic types of endometrial carcinoma. Gynecol Oncol1983;15:10–7.

[3] Liu FS. Molecular carcinogenesis of endometrial cancer. Taiwan J Obstet Gynecol2007;46:26–32.

[4] EllensonHL, Ronnett BM, Kurman RJ. Precursor lesions of endometrial carcinoma. In:Kurman RJ, Ellenson HL, Ronnett BM, editors. Blaustein's Pathology of the FemaleGenital Tract. Sixth ed. New York: Springer; 2011. p. 359–91.

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007

7Y.P. Geels et al. / Gynecologic Oncology xxx (2015) xxx–xxx

[5] Ellenson HL, Ronnett BM, Soslow RA, Zaino RJ, Kurman RJ. Endometrial carcinoma.In: Kurman RJ, Ronnett BM, Ellenson HL, editors. Blaustein's Pathology of the FemaleGenital Tract. Sixth ed. New York: Springer; 2011. p. 393–452.

[6] Ellis PE, Ghaem-Maghami S.Molecular characteristics and risk factors in endometrialcancer: what are the treatment and preventative strategies? Int J Gynecol Cancer2010;20:1207–16.

[7] Engelsen IB, Akslen LA, Salvesen HB. Biologic markers in endometrial cancertreatment. APMIS 2009;117:693–707.

[8] Matias-GuiuX, Prat J.Molecular pathology of endometrial carcinoma. Histopathology2013;62:111–23.

[9] Lax SF. Molecular genetic pathways in various types of endometrial carcinoma: froma phenotypical to a molecular-based classification. Virchows Arch 2004;444:213–23.

[10] Geels YP, Pijnenborg JM, van den Berg-van Erp SH, Bulten J, Visscher DW, Dowdy SC,et al. Endometrioid endometrial carcinoma with atrophic endometrium and poorprognosis. Obstet Gynecol 2012;120:1124–31.

[11] Cancer Genome Atlas Research N, Kandoth C, Schultz N, Cherniack AD, Akbani R, LiuY, et al. Integrated genomic characterization of endometrial carcinoma. Nature 2013;497:67–73.

[12] McConechy MK, Ding J, Cheang MCU, Wiegand KC, Senz J, Tone AA, et al. Use ofmutation profiles to refine the classification of endometrial carcinomas. J Pathol2012;228:20–30.

[13] SalvesenHB, Carter SL,MannelqvistM, Dutt A, Getz G, Stefansson IM, et al. Integratedgenomic profiling of endometrial carcinoma associates aggressive tumors withindicators of PI3 kinase activation. Proc Natl Acad Sci U S A 2009;106:4834–9.

[14] Roelofsen T, Geels YP, Pijnenborg JM, van Ham MA, Zomer SF, van Tilburg JM, et al.Cervical cytology in serous and endometrioid endometrial cancer. Int J GynecolPathol 2013;32:390–8.

[15] Roelofsen T, van Ham MA, Wiersma van Tilburg JM, Zomer SF, Bol M, Massuger LF,et al. Pure compared with mixed serous endometrial carcinoma: two differententities? Obstet Gynecol 2012;120:1371–81.

[16] McCluggage WG. Benign diseases of the endometrium. In: Kurman RJ, Ellenson LH,Ronnett BM, editors. Blaustein's Pathology of the Female Genital Tract. 6th ed.Springer; 2011. p. 305–58.

[17] Fader AN, Starks D, Gehrig PA, Secord AA, Frasure HE, O'Malley DM, et al. An updatedclinicopathologic study of early-stage uterine papillary serous carcinoma (UPSC).Gynecol Oncol 2009;115:244–8.

[18] Arafa M, Boniver J, Delvenne P. Progression model tissue microarray (TMA) for thestudy of uterine carcinomas. Dis Markers 2010;28:267–72.

[19] Ioffe OB, Papadimitriou JC, Drachenberg CB. Correlation of proliferation indices,apoptosis, and related oncogene expression (bcl-2 and c-erbB-2) and p53 in prolif-erative, hyperplastic, and malignant endometrium. Hum Pathol 1998;29:1150–9.

[20] Huszar M, Pfeifer M, Schirmer U, Kiefel H, Konecny GE, Ben-Arie A, et al. Up-regulation of L1CAM is linked to loss of hormone receptors and E-cadherin inaggressive subtypes of endometrial carcinomas. J Pathol 2010;220:551–61.

[21] Leake R, Barnes D, Pinder S, Ellis I, Anderson L, Anderson T, et al. Immunohistochem-ical detection of steroid receptors in breast cancer: a working protocol. UK ReceptorGroup, UK NEQAS, The Scottish Breast Cancer Pathology Group, and The Receptorand Biomarker Study Group of the EORTC. J Clin Pathol 2000;53:634–5.

[22] Engelsen IB, Stefansson IM, Akslen LA, Salvesen HB. GATA3 expression in estrogenreceptor alpha-negative endometrial carcinomas identifies aggressive tumors withhigh proliferation and poor patient survival. Am J Obstet Gynecol 2008;199:543.e1–7.

[23] Maiques O, Santacana M, Valls J, Pallares J, Mirantes C, Gatius S, et al. Optimalprotocol for PTEN immunostaining; role of analytical and preanalytical variables inPTEN staining in normal and neoplastic endometrial, breast, and prostatic tissues.Hum Pathol 2014;45:522–32.

[24] Zeimet AG, Reimer D, Huszar M, Winterhoff B, Puistola U, Azim SA, et al. L1CAM inearly-stage type I endometrial cancer: results of a large multicenter evaluation. JNatl Cancer Inst 2013;105:1142–50.

[25] Salvesen HB, Iversen OE, Akslen LA. Prognostic significance of angiogenesis andKi-67, p53, and p21 expression: a population-based endometrial carcinoma study.J Clin Oncol 1999;17:1382–90.

[26] Lurkin I, Stoehr R, Hurst CD, van Tilborg AA, Knowles MA, Hartmann A, et al. Twomultiplex assays that simultaneously identify 22 possible mutation sites in theKRAS, BRAF, NRAS and PIK3CA genes. PLoS One 2010;5:e8802.

Please cite this article as: Geels YP, et al, Immunohistochemical and geneticendometrium, Gynecol Oncol (2015), http://dx.doi.org/10.1016/j.ygyno.2

[27] Kompier LC, Lurkin I, van der AaMN, van Rhijn BW, van der Kwast TH, Zwarthoff EC.FGFR3, HRAS, KRAS, NRAS and PIK3CA mutations in bladder cancer and theirpotential as biomarkers for surveillance and therapy. PLoS One 2010;5:e13821.

[28] Dancey C, Reidy J. StatisticsWithoutMaths for Psychology: Using SPSS forWindows.London: Prentice Hall; 2004.

[29] Leblanc M, Poncelet C, Soriano D, Walker-Combrouze F, Madelenat P, Scoazec JY,et al. Alteration of CD44 and cadherins expression: possible association withaugmented aggressiveness and invasiveness of endometrial carcinoma. VirchowsArch 2001;438:78–85.

[30] Moreno-Bueno G, Hardisson D, Sarrio D, Sanchez C, Cassia R, Prat J, et al.Abnormalities of E- and P-cadherin and catenin (beta-, gamma-catenin, andp120ctn) expression in endometrial cancer and endometrial atypical hyperplasia. JPathol 2003;199:471–8.

[31] Saito T, Nishimura M, Yamasaki H, Kudo R. Hypermethylation in promoter region ofE-cadherin gene is associatedwith tumor dedifferentiation andmyometrial invasionin endometrial carcinoma. Cancer 2003;97:1002–9.

[32] Sakuragi N, Nishiya M, Ikeda K, Ohkouch T, Furth EE, Hareyama H, et al. DecreasedE-cadherin expression in endometrial carcinoma is associated with tumor dediffer-entiation and deep myometrial invasion. Gynecol Oncol 1994;53:183–9.

[33] Takeichi M. Cadherin cell adhesion receptors as a morphogenetic regulator. Science1991;251:1451–5.

[34] Gonzalez-Rodilla I, Aller L, Llorca J, Munoz A-B, Verna V, Estevez J, et al. TheE-cadherin expression vs. tumor cell proliferation paradox in endometrial cancer.Anticancer Res 2013;33:5091–5.

[35] Shia J. Immunohistochemistry versus microsatellite instability testing for screeningcolorectal cancer patients at risk for hereditary nonpolyposis colorectal cancer syn-drome. Part I. The utility of immunohistochemistry. J Mol Diagn 2008;10:293–300.

[36] Moreno-Bueno G, Sanchez-Estevez C, Palacios J, Hardisson D, Shiozawa T. Lowfrequency of BRAFmutations in endometrial and in cervical carcinomas. Clin CancerRes 2006;12:3865 (author reply-6).

[37] Catasus L, D'Angelo E, Pons C, Espinosa I, Prat J. Expression profiling of 22 genesinvolved in the PI3K–AKT pathway identifies two subgroups of high-grade endome-trial carcinomas with different molecular alterations. Mod Pathol 2010;23:694–702.

[38] Catasus L, Gallardo A, Cuatrecasas M, Prat J. PIK3CA mutations in the kinase domain(exon 20) of uterine endometrial adenocarcinomas are associated with adverseprognostic parameters. Mod Pathol 2008;21:131–9.

[39] Hayes MP,Wang H, Espinal-Witter R, DouglasW, Solomon GJ, Baker SJ, et al. PIK3CAand PTEN mutations in uterine endometrioid carcinoma and complex atypicalhyperplasia. Clin Cancer Res 2006;12:5932–5.

[40] Kuhn E, Wu R-C, Guan B, Wu G, Zhang J, Wang Y, et al. Identification of molecularpathway aberrations in uterine serous carcinoma by genome-wide analyses. J NatlCancer Inst 2012;104:1503–13.

[41] Caduff RF, Johnston CM, Frank TS. Mutations of the Ki-ras oncogene in carcinoma ofthe endometrium. Am J Pathol 1995;146:182–8.

[42] Smith G, Bounds R,Wolf H, Steele RJC, Carey FA,Wolf CR. Activating K-Rasmutationsoutwith ‘hotspot’ codons in sporadic colorectal tumours — implications forpersonalised cancer medicine. Br J Cancer 2010;102:693–703.

[43] Sasaki H, Nishii H, Takahashi H, Tada A, Furusato M, Terashima Y, et al. Mutation ofthe Ki-ras protooncogene in human endometrial hyperplasia and carcinoma. CancerRes 1993;53:1906–10.

[44] Enomoto T, Inoue M, Perantoni AO, Buzard GS, Miki H, Tanizawa O, et al. K-rasactivation in premalignant and malignant epithelial lesions of the human uterus.Cancer Res 1991;51:5308–14.

[45] Singh A, Greninger P, Rhodes D, Koopman L, Violette S, Bardeesy N, et al. A geneexpression signature associated with “K-Ras addiction” reveals regulators of EMTand tumor cell survival. Cancer Cell 2009;15:489–500.

[46] Guerrero S, Casanova I, Farre L, Mazo A, Capella G, Mangues R. K-ras codon 12mutation induces higher level of resistance to apoptosis and predisposition toanchorage-independent growth than codon 13 mutation or proto-oncogeneoverexpression. Cancer Res 2000;60:6750–6.

[47] Chiang S, Soslow RA. Updates in diagnostic immunohistochemistry in endometrialcarcinoma. Semin Diagn Pathol 2014;31:205–15.

profiles of endometrioid endometrial carcinoma arising fromatrophic015.03.007