A multimarker qPCR platform for the characterisation of endometrial cancer

  • Authors:
    • Anna Supernat
    • Sylwia Łapińska-Szumczyk
    • Hanna Majewska
    • Jacek Gulczyński
    • Wojciech Biernat
    • Dariusz Wydra
    • Anna J. Żaczek
  • View Affiliations

  • Published online on: December 13, 2013     https://doi.org/10.3892/or.2013.2924
  • Pages: 1003-1013
Metrics: Total Views: 0 (Spandidos Publications: | PMC Statistics: )
Total PDF Downloads: 0 (Spandidos Publications: | PMC Statistics: )


Abstract

The molecular background of endometrial cancer (EC) has not been fully elucidated. In the present study, we developed a quantitative PCR (qPCR) platform to examine the gene dosages of the potential molecular markers MGB1, TOP2A, ERBB1-4, MYC, CCND1, ESR1 and PI3K. The platform was applied in samples collected from 157 EC patients (stage I-IV) to verify its clinical utility and to examine the diagnostic and prognostic significance of the analysed biomarkers. The gene dosage pattern of the ERBB family and its downstream effectors PI3K and MYC showed particularly strong correlations with clinicopathological data. The ERBB PI3K/Akt pathway was upregulated in 31 (20%) of 156 cases. Activation of the ERBB PI3K/Akt pathway was positively correlated with a higher stage (p=0.001), higher grade (p=0.001), histological type II disease (p=0.0003) and metastases (p=0.02). The implemented hierarchical clustering revealed that cluster 2 was characterised by high copy numbers of the studied genes. Cluster 2 was associated with shorter overall survival (p=0.05). The platform was found to be a fast and simple method for direct analysis of the genes involved in uterine carcinogenesis, making it feasible for EC biology characterisation.

Introduction

Endometrial cancer (EC) is the most frequent malignancy of the female genital tract in the Western world, with an estimated incidence of 10–20 per 100,000 women (1). There were 5,594 new cases and 1,028 deaths estimated in 2012 in Poland (2). The prevalence of EC is increasing annually (3). Despite the high prevalence, our understanding of the molecular background of EC with regard to its pathogenesis, tumour growth and disease progression remains insufficient. Data concerning copy number variations within EC are scarce.

EC encompasses a group of histologically and biologically diverse tumours, which are characterised by a distinct pathogenesis. Two pathogenic types have been distinguished based on the differential characteristics, as proposed by Bokhman (4). This classification has been challenged in daily practice, as some tumours possess overlapping or combined morphologic and molecular characteristics. Nevertheless, the dualistic model is still followed in clinical practice (5), with endometrial carcinomas being classified as either type I or II.

The more frequent endometrioid type I carcinomas (80%) (6) are rather well-differentiated, with high oestrogen receptor α (ERα) and progesterone receptor (PR) expression. They are generally confined to the uterus and carry a favourable prognosis. Type II carcinomas, which have non-endometrioid serous or clear cell histology, follow an oestrogen-unrelated pathway. Such poorly-differentiated tumours behave in a more aggressive manner, resulting in a poor prognosis (4,7,8). Type II carcinomas appear spontaneously and they are not clearly related to the transition from atypical hyperplasia; rather, they arise in the background of an atrophic or inert endometrium (9).

Type I endometrioid carcinomas have a near-diploid karyotype and are associated with microsatellite instability, which is the consequence of alterations in the mismatch repair genes, MLH1, MSH6 and MSH3. Aberrations of K-RAS oncogene together with tumour-suppressor gene mutations, especially PTEN, have also been reported. Additionally, type I ECs contain PI3K and β-catenin mutations. On the other hand, type II serous carcinomas are known for their chromosome aneuploidy, ERBB2 overexpression and mutant p53 protein accumulation within the nucleus. Other mutations, encountered in both types of EC, include aberrations of MYC, INT2 and CTNNB1 (6,1015).

The prognostic value of the EC dualistic model is limited. Up to 20 and 50% of type I and II carcinomas, respectively, recur (16). The aim of the present study was to further explore the molecular basis of EC by developing a quantitative PCR (qPCR) platform that would allow for more precise tumour characterisation. We searched for potential molecular markers that may be of diagnostic and prognostic significance.

We analysed the RNA expression level of the MGB1 gene (mammaglobin), as this secretory protein is recognised as a sensitive marker in breast cancer and was also found to be expressed in the female genital tract and in EC (17). We also measured the copy number variations of the following genes: ESR1 (oestrogen receptor 1), ERBB1 (epidermal growth factor receptor), ERBB2 (v-erb-b2 erythroblastic leukaemia viral oncogene homolog 2, also known as HER2), ERBB3 (receptor tyrosine-protein kinase erbB-3), ERBB4 (receptor tyrosine-protein kinase erbB-4), PI3K (phosphatidylinositol 3-kinases), MYC (cellular homolog of the retroviral v-myc oncogene), CCND1 (cyclin D1) and TOP2A (DNA topoisomerase 2-α).

ERBB2 is a central oncogene of prognostic and predictive value. Its amplification and increased expression have been reported in EC, indicating a higher stage and poor differentiation (18). Previous reports indicated the importance of the interaction of ERBB2 with other ERBB receptors (ERBB1 ERBB3, ERBB4). This interaction may be associated with a more aggressive phenotype (19,20). PI3K and MYC are downstream effectors of ERBB receptors. Excessive activation of PI3K, as a result of amplification, is associated with a poorer prognosis in EC patients (21). Expression of the MYC oncogene is often impaired in several types of cancer and gene amplification is one of the mechanisms of its activation (22). Another gene implicated in endometrial carcinogenesis is CCND1; its amplification may result in cyclin D1 overexpression, which is found in 40–56% of EC cases (23). TOP2A is an enzyme that plays a key role in maintaining the integrity of the genome. In addition, TOP2A is a target of many anticancer drugs. Amplification and overexpression of TOP2A has been observed in 7.5 and 45% of EC cases, respectively (24).

The gene dosage of the aforementioned genes seems to be of clinical importance in EC. However, these genes have not been examined together in EC. Therefore, we developed a potentially highly sensitive and reproducible qPCR-based platform for the reliable assessment of specific genes associated with EC and in order to analyse its utility in a representative group of EC samples.

Materials and methods

Patients and tissues

The present study included 157 fresh frozen tumour samples retrospectively collected from a cohort of EC patients who were operated on in the Department of Gynaecology, Gynaecological Oncology and Gynaecological Endocrinology (Medical University of Gdańsk) between 2005 and 2011. Each patient was primarily treated by surgery, with the possible option of radiotherapy and/or chemotherapy administration. The inclusion criteria were operable EC (IVB stage patients underwent cytoreductive surgery) confirmed by histological examination and a signed consent form. The present study was approved by the Ethics Committee of the Medical University of Gdańsk.

The samples were collected from the core of the tumour by surgical excision prior to any systemic treatment and were immediately frozen and stored at −80°C. The samples were kept on ice during transport. The tissue samples included all pathologic stages of endometrial carcinoma, from non-invasive IA to metastatic IVB cancer as distinguished by the International Federation of Gynecology and Obstetrics (FIGO) in 2009 (25). The patient characteristics are summarized in Table I. The mean age was 63.6 years (range 30–87 years). Patients with a body mass index >30 were classified as obese (26). The PgR, ER and ERBB2 status was determined immunohistochemically (IHC) using Allred score (positivity cut-off, ≥3) for PgR and ER and the HercepTest score (positivity cut-off, 3) for ERBB2. A survival analysis was performed for 82 patients. After a median follow-up of 4.4 years (range 0.04–7.58 years), 15 patients (18%) had died. The last follow-up data were collected in January 2013.

Table I

Clinicopathological data (N=157).

Table I

Clinicopathological data (N=157).

VariableNo. of cases (%)
Menopausal status
 Premenopausal8 (5.1)
 Perimenopausal11 (7)
 Postmenopausal124 (79)
 Missing data14 (8.9)
Histology
 Endometroid123 (78.3)
 Non-endometroid34 (21.7)
Stage (FIGOa)
 IA-IB112 (71.3)
 II22 (14)
 IIIA-IIIC20 (12.7)
 IVA-IVB3 (1.9)
Grade
 I60 (38.2)
 II69 (43.9)
 III24 (15.3)
 Missing data4 (2.5)
Obesity
 Absent68 (43.3)
 Present87 (55.4)
 Missing data2 (1.3)
Cervical invasion
 Absent113 (72)
 Present42 (26.8)
 Missing data2 (1.3)
Myometrial infiltration
 ≤1/287 (55.4)
 >1/269 (43.9)
 Missing data1 (0.6)
Metastases
 Absent105 (66.9)
 Present49 (31.2)
 Missing data3 (1.9)
PgR status
 Negative12 (7.6)
 Positive94 (59.9)
 Missing data51 (32.3)
ER status
 Negative8 (5.1)
 Positive98 (62.4)
 Missing data51 (32.3)
ERBB2 status
 Negative89 (56.7)
 Positive17 (10.8)
 Missing data51 (32.3)

a FIGO, International Federation of Gynecology and Obstetrics.

Controls

Healthy endometrium from patients treated for ailments other than EC was collected as control samples. Control samples were accessed histopathologically; no hyperplasia was observed. Pooled DNA and RNA was isolated from frozen tissue samples of five healthy donors and used for qPCR assay optimisation, standard curve generation and as a calibrator for the gene dosage and expression evaluation test in 157 EC tumour samples.

The utility of the assay was verified using the human cancer cell line SK-BR-3 with confirmed amplification of the genes ERBB2, MYC, TOP2A and human cancer cell line MDA-MB-361 with confirmed amplification of ERBB2 (27). Additionally, we used reference DNA (Roche, Switzerland) and DNA isolated from healthy breast tissue to verify the performance of the EC DNA.

DNA and RNA isolation

Prior to nucleic acid isolation, tissue specimens (25 mg/sample) were homogenised (1 min, 6,000 rpm speed) using a MagNA Lyser (Roche). DNA and RNA were isolated with AllPrep DNA/RNA Mini kit (Qiagen, Germany) according to the manufacturer’s instructions. After the isolation, the DNA/RNA concentration and purity were determined using a Spectrophotometer ND-1000 (NanoDrop, USA). Good quality of the DNA was defined as an A260/280 nm ratio between 1.70 and 1.90. Good quality of the RNA was defined as an A260/280 nm ratio of ~2. The DNA and RNA samples were stored at −20 and −80°C, respectively.

RNA was subsequently reverse transcribed to cDNA with the Transcriptor First Strand cDNA synthesis kit (Roche) according to the manufacturer’s instructions using random hexamer primers. The total amount of 1,000 ng RNA was used per each reverse transcription reaction.

Genes and primers

The gene dosages of ERBB1, ERBB2, ERBB3, ERBB4, MYC, CCND1, ESR1, PI3K and TOP2A were determined by qPCR with Power SYBR-Green Mastermix (Applied Biosystems, USA), using the amyloid precursor protein (APP) gene and 3P (RNA, U4 small nuclear pseudogene) gene as a reference. The potential reference genes were chosen based on a search performed in the Atlas of Genetics and Cytogenetics in Oncology and Haematology (http://www.atlasgeneticsoncology.org/). APP and 3P stability was verified against superoxide dismutase 2 (SOD2) using geNorm software (28). The primers were designed using Beacon Designer Software version 7.6 (Bio-Rad, USA). Their specificity was examined by PCR, followed by electrophoretic separation on 1.5% agarose gel stained with ethidium bromide (final concentration of 0.5 μg/ml). Electrophoresis was performed in 0.5X TBE buffer at 100 V (5 V/cm) for 30 min. For visualisation of the bands, the gel was placed in a Molecular Imager® GelDoc™ XR photoimaging system and analysed using Quantity-One software (both from Bio-Rad) after irradiation with UV light (302 nm). The primer parameters are summarised in Table II. The quantity of DNA/well was 40 ng.

Table II

Primer parameters.

Table II

Primer parameters.

NameSequence 5′>3′No. of base pairsProduct length (bp)Reference sequenceFinal primer concentration (nM)
3PF: CTC ATA GGC GAA GGC ACC AG20112NT_022517300
3PR: GGT CAA GTT CCG CAC ACA CC20300
APPF: AGC CCA GAA GGT GTC AAA CA2060NG_007376300
APPR: CAT CTT CAT GTC CGT TGC AT20300
CCND1F: AGG AGG TGT GAG GAG GAG18102NG_007375300
CCND1R: CTG GAA GTC AAC GGT AGC18300
MYCF: CGT GAC CAG ATC CCG GAG TT20134NG_007161300
MYCR: CGT TTC CGC AAC AAG TCC TCT21200
ERBB1F: GCA GAC AGG ATG ACC AAG AG2091NG_007726300
ERBB1R: GTA GGC AGA TGA ACA GGA ACC21300
ERBB2F: CTG CTG GTC GTG GTC TTG G1993NG_007503300
ERBB2R: CTG CAG CAG TCT CCG CAT C19300
ERBB3F: GAG AGG TGT GAG GTG GTG ATG G22114NG_011529300
ERBB3R:AAG AGG AGC AGG TTG AGG AAG G22300
ERBB4F: CGC TTC GTC TCT TCT CGT TTC C2290NG_011805300
ERBB4R: GAA CAA CAA TGG CAC GCT AAT CC23200
ESR1F: ACA TGG ACA CCT CCC AGT C1960NG_008493.1300
ESR1R: ACA GAC TAA CAC AGC CCA TC20300
PI3KF: GCT TCC AAC AAT CCT CTT CCG TAG24118NG_012113.1200
PI3KR: GCA CTG AAT CTG TAG CGA ACT TCC24300
TOP2AF: GCC AAC TCA GCC GTT CAT AGG21101NG_027678200
TOP2AR: CGA AGC AGA CCA GCC AAT CC20200

The MGB1 RNA expression level was determined by RT-qPCR with TaqMan® Universal PCR Master Mix (Applied Biosystems), using hypoxanthine phosphoribosyltransferase 1 (HPRT1) as a reference. The stability of the HPRT1 gene expression was verified relative to the expression of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and β-actin (ACTB). We used TaqMan® Expression Assays (Applied Biosystems) as follows: HPRT1 Endogenous Control Hs99999909_m1; MGB1 Gene Expression Assay Hs00935948_m1. The quantity of cDNA/well was 75 ng.

Calculating the optimal primer concentration

The optimal forward/reverse primer concentration was based on determination of the minimum quantification cycle (Cq) and the maximum ΔRn while minimizing nonspecific amplification. The reaction volumes were 20 μl. We used 40 ng of genomic DNA as a template. Each of the nine forward/reverse primer concentration combinations (200/200, 300/200, 400/200, 200/300, 300/300, 400/300, 200/400, 300/400 and 400 nM/400 nM) was run in triplicate. Quantities of gene expression assays and TaqMan® Universal PCR Master Mix were used according to the manufacturer’s instructions.

Assessment of gene dosage and mRNA expression

After the optimisation, the components of the prepared mastermix in the final protocol used for qPCR standard curve generation as well as ERBB1, ERBB2, ERBB3, ERBB4, MYC, CCND1, ESR1, PI3K and TOP2A gene dosage determination in 157 EC tumour samples were: 10 μl of 2X Power SYBR-Green Mastermix, 0.4–0.6 μl of 10 μM forward/reverse primer (Sigma-Aldrich, Germany), 40 ng of DNA in the volume of 4 μl and water added into the reaction mixture to obtain a final volume of 20 μl.

Reactions amplifying ERBB1, ERBB2, ERBB3, ERBB4, MYC, CCND1, ESR1, PI3K, TOP2A and two reference genes (APP and 3P) were performed in separate wells. The reactions were always prepared in duplicate on 96-well plates and sealed with optical tape (both from Applied Biosystems). For each tested gene, a negative control (no template DNA) was always included. A melting curve analysis was also included in all gene dosage determinations.

Reactions amplifying MGB1 and HPRT1 cDNA were performed in separate wells. The reactions were always prepared in triplicate on 96-well plates and sealed with optical tape (both from Applied Biosystems). For each tested gene, a negative control (RNA and reverse transcription mix without reverse transcriptase) was always included.

The thermal profiles used were the default settings of the manufacturer for either SYBR-Green or TaqMan® probe assays. Reports containing the Cq values were generated using StepOne Software v2.2.2 (Applied Biosystems).

Assay evaluation and quality control

To overcome the shortcomings of the in-built StepOnePlus™ software normally used to analyse data, we utilized qbasePLUS software, version 2.3 (Biogazelle, Belgium) (29) for gene dosage and expression quantification. The software uses a modified ΔΔCt method based on multiple reference genes and gene specific amplification efficiencies. It also possesses an algorithm that allows for inter-run calibration, thus taking into account run-to-run variation.

Average Cq values from the duplicates/triplicates were used for gene dosage/mRNA expression calculation. Inter-run calibration was implemented. The reference target stability was verified, the values of M and CV were <0.5 and 0.2, respectively. The targets were scaled to the average. User-defined amplification efficiencies separate for each gene were assumed. The reaction efficiencies of ERBB1, ERBB2, ERBB3, ERBB4, MYC, CCND1, ESR1, PI3K, TOP2A, APP, 3P, MGB1 and HPRT1 under the optimised conditions were calculated from the five-point standard curves prepared by amplifying the pooled control genomic DNA/cDNA, diluted over four orders of magnitude. Standard curves were automatically generated by the StepOne Software v2.2.2. Baseline cycles and thresholds were set manually and the Cqs were calculated automatically by the software using an in-built algorithm.

The integrity of the nucleic acids in randomly selected samples was checked by visualisation of the electrophoretically separated post-PCR products (6,4 kb fragment of CYP2D6 gene) and when compared with the molecular marker GeneRuler™ 1 kb DNA Ladder (Thermo Scientific, USA).

The presence of the PCR inhibitors was evaluated by performing qPCR with pooled control DNA and genomic DNA from healthy breast tissue and reference DNA from Roche, followed by comparison of the Cq values. The reactions were conducted in triplicate on a test plate, with a negative control (no template DNA) included.

The reaction specificity was verified through a melt curve analysis (included in every run) and by visualisation of the electrophoretically separated random post-PCR products for all studied genes.

Statistical analysis

STATISTICA software version 10 (StatSoft Co., USA) was used for all calculations, with the exception of unsupervised hierarchical clustering. Data visualisation, dendrograms and complete linkage computations were performed with the use of the Java Development kit 7u21 script running in Java Runtime Environment 7. The tests that were used and their applications were: testing normality of the data set, Shapiro-Wilk test, correlations between continuous relative gene quantities, Spearman’s correlation, correlations between continuous or categorical relative gene quantities and clinicopathological data of the patients, Mann-Whitney U test or crosstabs statistics with Pearson’s Chi-square test, respectively. The Kaplan-Meier estimator was employed for survival analysis and the results were verified with the Cox F-test. The endpoint for the present study was overall survival (OS). OS was defined as the time from sample collection to mortality or censoring. Censoring was defined as loss of follow-up or alive at the end of follow-up. Statistical significance in all the aforementioned tests was assumed when p≤0.05. Cox proportional hazards regression analysis was used to identify the independent predictors of OS. Univariate predictors significant with a value of p≤0.10 were entered into a step-wise multivariate model to identify those with independent prognostic information. The study was performed in accordance with the REMARK criteria (30).

Results

Flow of samples

Of the 157 RNA samples, 156 (99.4%) had amplifiable cDNA. Amplifiability of the DNA samples differed depending on the target, ranging from 152 (97.5%) to 156 (99.4%).

Assay evaluation and quality control results

The parameters of the standard curves obtained for ERBB1, ERBB2, ERBB3, ERBB4, MYC, CCND1, ESR1, PI3K, TOP2A, APP, 3P, MGB1 and HPRT1 indicated good reaction performance and the calculated efficiencies of the reactions were within the range of 90.4–109.9% (Table III).

Table III

Reaction efficiencies.

Table III

Reaction efficiencies.

GeneEfficiencySlopeR2
APP109.6−3.1120.997
3P109.7−3.1060.999
MYC91.4−3.5460.993
CCND197.3−3.3890.992
ERBB196.6−3.4060.994
ERBB290.4−3.5750.987
ERBB391.7−3.5360.987
ERBB496.7−3.4040.98
ESR1109.9−3.1040.992
PI3K90.5−3.5720.98
TOP2A96.8−3.40.992
HPRT196.8−3.40.999
MGB1108.2−3.140.964

The isolated genomic DNA was confirmed to have high integrity and was seen as a single band on the electrophoretic gel. To confirm the high quality of the DNA, we performed a long PCR reaction on randomly chosen isolated material; the amplified CYP2D6 gene (6,4 kb) was seen as a single clear band. Each of the analysed targets was examined through a melt curve analysis and electrophoresis for the presence of primer dimers or unspecific bands; none were observed. No inhibition of the real-time PCR reaction was observed as the recorded Cq values for the APP and 3P reference genes in the calibrator DNA, the DNA from healthy breast tissue, and the DNA supplied by Roche, the latter being free of qPCR inhibitors, had differences in Cqs <0.5.

The repeatability (intra-assay variation) and reproducibility (inter-assay variation) of the assay were measured as the average mRNA expression/gene dosage in pooled controlled mRNA/DNA on one plate (three reactions) or on four randomly selected separate plates, respectively. The assay was characterised by low intra-assay and inter-assay variation (data not shown).

The utility of the assay was verified with the use of the cell lines SK-BR-3 and MDA-MB-361, which had confirmed amplification of the genes ERBB2, MYC, TOP2A and ERBB2, respectively (27). Additionally, one copy of the TOP2A gene is known to be deleted in MDA-MB-361 (31). All recorded values for ERBB2, MYC and TOP2A in the cell line SK-BR-3 were higher than the assumed cut-offs. The ERBB2 gene dosage in MDA-MB-361 was also higher than the cut-off and, although the assay is not designed to measure deletions, the relative quantity of TOP2A was indeed decreased in this cell line (data not shown).

mRNA expression and gene dosage in clinical samples, cut-point determination and unsupervised data analysis

The relative quantities of gene expression and gene dosage were scaled to average. MGB1 overexpression was recognised when there was at least 2-fold change of MGB1 expression in comparison to the calibrator sample. The experimentally determined cut-off values were used in the evaluation of gene aberrations; as cut-offs, we used the most extreme values within the control samples.

Unsupervised hierarchical clustering was implemented to distinguish the gene aberration/expression profiles within the group of 157 EC patients. Out of the studied genes, ERBB4, CCND1 and ESR1 were excluded from the analysis as factors carrying little information. Instead, we included the RUNX1 and RAD21 mRNA expression levels, which were previously reported (32). Unsupervised analysis of gene aberrations and mRNA expression distinguished two major groups of tumours (clusters 1 and 2; Fig. 1). Patients within cluster 2 generally had increased gene dosages of ERBB3 and elevated expression of MGB1, RUNX1 and RAD21.

Correlation of gene dosages and mRNA expression with clinical and pathological data

The gene dosages of the individual genes often correlated with each other. Particularly strong correlations occurred in the case of ERBB1 and CCND1 (p=0.000002), ERBB1 and ERBB3 (p=0.000003), PI3K and MGB1 (p=0.00003), ERBB3 and TOP2A (p=0.00005) and ERBB3 and PI3K (p=0.00005). A higher stage of the disease was correlated with increased levels of ERBB2, ERBB3, ERBB4, PI3K, MYC and TOP2A (p=0.05, p=0.003, p=0.02, p=0.02, p=0.007, p=0.04, respectively). Histology type II was associated with elevated levels of ERBB1 (p=0.002), ERBB3 (p=0.01), PI3K (p=0.05) and MYC (p=0.02), and decreased expression of MGB1 (p=0.002). The presence of metastases was associated with higher PI3K, MYC and ESR1 gene dosage (p=0.04, p=0.002, p=0.02, respectively), while a high grade was associated with increased levels of ERBB1 (p=0.02), PI3K (p=0.04) and MYC (p=0.006) and decreased level of MGB1 (p=0.0008). Cervical invasion was correlated with increased gene dosage of MYC (p=0.05) and ESR1 (p=0.02). Myometrial infiltration was associated with elevated TOP2A levels (p=0.02; Table IV).

Table IV

Median gene dosage and expression in the context of clinicopathological data.

Table IV

Median gene dosage and expression in the context of clinicopathological data.

CCND1ERBB1ERBB2ERBB3ERBB4
No. of casesMedian gene dosageP-valueNo. of casesMedian gene dosageP-valueNo. of casesMedian gene dosageP-valueNo. of casesMedian gene dosageP-valueNo. of casesMedian gene dosageP-value
Histology0.150.0020.280.010.33
 Type I1211.001210.981230.971210.971230.97
 Type II321.01321.08330.99321.00331.00
Stage0.090.220.050.0030.02
 I, II1331.001330.981340.971330.971340.97
 III, IV201.04201.03220.99201.10221.03
Grade0.070.020.380.060.20
 1, 21271.001270.981290.971270.971290.97
 3221.03221.11230.99221.00231.00
Cervical invasion0.550.750.340.640.35
 Absent1111.001110.981120.971110.981120.97
 Present401.01400.99421.00400.95421.01
Myometrial infiltration0.640.400.250.380.72
 ≤1/2871.01870.97870.99870.98871.00
 >1/2650.99651.00680.96650.96680.96
Distant metastases0.260.230.430.560.18
 Absent1050.991050.981050.971050.971050.97
 Present461.01461.00480.99460.97481.01

PI3KMYCESR1TOP2AMGB1

Histology0.050.020.270.380.002
 Type I1230.961230.941211.001211.001221.49
 Type II331.00331.01311.05320.99340.19
Stage0.020.0070.060.040.50
 I, II1340.961340.931321.001331.001341.08
 III, IV221.02221.00201.22201.05220.89
Grade0.040.0060.200.560.0008
 1, 21290.961290.931271.011271.001281.49
 3231.02231.04211.18220.99240.10
Cervical invasion0.080.050.020.880.53
 Absent1120.961120.931100.991111.001130.95
 Present420.99420.98401.17401.00411.21
Myometrial infiltration0.820.330.200.020.82
 ≤1/2870.96870.93871.00871.02870.95
 >1/2680.98680.95641.06650.98681.67
Distant metastases0.040.0020.020.420.84
 Absent1050.951050.931030.981051.001050.95
 Present480.99480.99461.17461.00481.15

[i] P-values in bold indicate statistical significance.

When we compared the groups of positive vs. negative samples within all the studied genes, particularly strong positive correlations were observed between the ERBB1 status and histology type II (p=0.00001), the ERBB1 status and a higher grade (p=0.00005), the ERBB3 status and a higher stage (p=0.00005) and the ESR1 status and cervical invasion (p=0.0003). Based on the analysis of separate genes, the gene dosage pattern of the ERBB family and its downstream effectors, PI3K and MYC, was analysed in more detail. Upregulation of the ERBB PI3K/Akt pathway was assumed whenever any of the ERRB genes together with PI3K and/or MYC were found to be amplified. The ERBB PI3K/Akt pathway was upregulated in 31 (20%) of 156 cases. Activation of the pathway was positively correlated with a higher stage: p=0.001, 21/134 (15.7%) positive cases vs. 10/22 (45.5%); grade: p=0.001, 19/129 (14.7%) vs. 10/23 (43.5%); histological type II of the disease: p=0.0003, 17/123 (13.8%) vs. 14/33 (42.4%), as well as with the presence of metastases: p=0.02, 16/105 (15.2%) vs. 15/48 (31.3%).

Although the clusters were distinctive with generally higher gene dosages reported for cluster 2, there was no correlation between the results of the clustering and the clinicopathological characteristics.

Survival analysis

Of all the studied genes, with additional analysis of the ERBB PI3K/Akt pathway and hierarchical clustering, statistical significance with respect to the OS was reached only in the case of ERBB3 and the generated clusters (both p=0.05; Fig. 2A and B). The pattern of the remaining graphs, with p-values >0.05, either had the expected tendency (shorter OS within the groups with gene amplification) or was inconclusive.

A univariate analysis was performed for all the clinicopathological data, the studied genes and the distinguished clusters. Stage, grade, PR status, ERBB3 status and the clusters correlated with shorter OS (all p≤0.10), and these parameters were included in a step-wise multivariate analysis, eventually yielding p-values of 0.01 and 0.13 for the stage, and ERBB3, respectively. The hazard ratios for the stage and ERBB3 were: 4.94 (95% CI, 1.40–17.40), 3.27 (95%CI, 0.68–15.65; Table V).

Table V

Univariate and multivariate analysis of clinicopathological and molecular parameters as prognostic factors in endometrial cancer patients.

Table V

Univariate and multivariate analysis of clinicopathological and molecular parameters as prognostic factors in endometrial cancer patients.

Univariate analysisMultivariate analysis
VariableHR95% CIP-valueHR95% CIP-value
Histology (type II vs. I)2.080.71–6.160.18
Stage (III, IV vs. I, II)6.021.84–19.660.0034.941.4–17.400.01
Grade (3 vs. 1, 2)1.871.07–3.250.03NS
Obesity (present vs. absent)1.700.54–5.360.36
Cervical invasion (present vs. absent)1.450.46–4.590.52
Myometrial infiltration (present vs. absent)2.230.80–6.250.13
Metastases (present vs. absent)2.340.83–6.610.11
PgR status (positive vs. negative)0.330.10–1.050.06NS
ER status (positive vs. negative)0.300.07–1.350.12
ERBB2 status (positive vs. negative)1.430.40–5.120.58
Clusters (2 vs. 1)2.570.87–7.540.09NS
CCND1 status (positive vs. negative)1.530.23–10.190.66
ERBB1 status (positive vs. negative)1.050.27–4.130.94
ERBB2 status (positive vs. negative)2.720.78–9.550.12
ERBB3 status (positive vs. negative)5.311.24–22.900.023.270.68–15.650.13
ERBB4 status (positive vs. negative)1.080.25–4.610.91
PI3K status (positive vs. negative)0.820.11–5.900.84
MYC status (positive vs. negative)1.730.56–5.370.34
ESR1 status (positive vs. negative)0.120.00–495.960.62
TOP2A status (positive vs. negative)2.120.51–8.800.30
MGB1 status (positive vs. negative)0.460.13–1.640.23

Discussion

Copy number variations belong to frequently found genetic alterations within cancerous tissues (33) and we have established a multimarker quantitative PCR (qPCR) platform to examine genes that may carry such aberrations and thus be of significance in endometrial cancer (EC) biology. The gene dosages of TOP2A, ERBB1, ERBB2, ERBB3, ERBB4, MYC, CCND1, ESR1, PI3K and RAD21 were determined in fresh-frozen tumour samples using a SYBR-Green-based qPCR assay. Additionally, the RNA expression level of the MGB1 gene was examined by RT-qPCR.

The designed platform, exploited on a large sample size, proved to be a fast and simple way for direct analysis of the genes involved in uterine carcinogenesis. The results obtained from versatile quality control tests showed good reaction performance, specificity of the designed primers, low intra-assay and inter-assay variation, high DNA/RNA integrity and quality, and thus low sample loss. We performed a utility verification using cancer cell lines, and the results, revealed the same aberrations that have been reported in the literature. The developed protocol allows for highly coherent gene dosage and transcriptomic measurements.

Mammaglobin 1 was first detected as a protein that is specifically overexpressed in breast cancer (34). Studies by Classen-Linke et al (35) demonstrated that this expression is not restricted to breast tissue; it may also be detected in the endometrium where it changes in a hormone-dependent manner. To the best of our knowledge, this is the first study to measure the MGB1 RNA level in EC to understand its clinical significance. Our findings show that in EC, the loss of MGB1 expression, not overexpression, strongly correlates with a higher tumour grade and is characteristic of histological type II.

Our results demonstrate that elevated ERBB1, ERBB3 and MYC gene dosages are correlated with more aggressive tumour behavior. Yeramian et al also found MYC to be deregulated in EC (10). However, these findings only involve gene expression. Data on MYC, ERBB1 and ERBB3 copy number variations in EC are scarce or non-existent. Raeder et al analysed the amplifications of the 8q24 region in EC samples but did not observe any link between MYC copy number alterations and disease progression (36). Esteller et al studied the role of ERBB1 in EC, but they did not observe the ERBB1 gene amplification in any of the cases studied (37). This may be due to the limitations of the method of differential PCR, which is much less sensitive than qPCR. Liang et al demonstrated that the overexpression of wild-type ERBB3 significantly increased the survival of Ba/F3 cells (38), which is consistent with the proposed role of ERBB3 as an oncogene. Our studies are also in agreement with the study of Saghir et al where ERBB3 overexpression was indicated as a characteristic typical for EC (39). It is well established in breast cancer that the amplification of genes belonging to the ERBB family results in the overexpression of tyrosine kinase receptors and our findings suggest that the same mechanism occurs in EC.

The borderline statistical significance of ERBB2 gene dosage in the context of the stage of the disease may be explained by the fact that ERBB2 amplification was found to represent a rather rare event in EC (40). Nevertheless, Rolitsky et al found ERBB2 amplification to be correlated with type II EC, which supports our findings (41).

We demonstrated that the activation of the PI3K pathway is crucial in EC. These results are in agreement with studies by Salvesen et al (21) where amplification of a region including PIK3CA was significantly associated with unfavourable recurrence-free survival. Our data also indicated that increased ERBB/PI3K pathway gene dosage correlated with histological type II, higher stage and higher grade as well as the presence of metastases.

The results described in the present study may facilitate the development of targeted therapies, developed to inhibit cellular signalling that leads to cell growth and proliferation in EC. Several tyrosine kinase receptor inhibitors are available, including ERBB1 inhibitors (gefitinib, erlotinib, lapatinib, cetuximab) and the ERBB2 inhibitor, trastuzumab. A phase II study on the use of cetuximab in recurrent EC is currently being carried out (42). There have also been trials on the efficacy of trastuzumab in ERBB2-positive EC patients. Although these studies did not report any activity (43,44), further clinical evaluation of the use of trastuzumab or combination regiments may still be worth verifying as it is possible that this drug may be effective in a properly selected group of EC patients (45).

Although the clusters generated in the present study did not possess strikingly different clinical and histopathologic characteristics, cluster 2 in general did contain higher gene dosages with the exception of MYC and RAD21. Additionally, the survival analysis revealed a significantly shorter OS in cluster 2, suggesting that this clustering distinction may in fact serve as a prognostic factor.

The key limitation of this study was the short follow-up period and lack of survival data for a considerable number of patients. As a result, the survival analyses rarely reached statistical significance, not only in the case of the studied genes but also in the analysis of the clinicopathological data. Nevertheless, even without statistical significance, almost all the studied potential markers retained the expected tendency (shorter OS for increased gene dosage/decreased expression), with few Kaplan-Meier curves being inconclusive due to a small sample size. The survival studies by Konecny et al (40), which were performed on a larger cohort of EC patients, did reach statistical significance, but these analyses focused mainly on type II EC, with a median follow-up of 83 months. The survival data for the cohort examined in the present study are still being collected. The prognostic value of the studied biomarkers is to be updated.

Another limitation of the present study is associated with tumour heterogeneity. The samples used for DNA/RNA isolation may not reflect the heterogeneity across a tumour, as they were excised manually from a single tumour region and only small tissue fragments were used. Perhaps a microdissection technique of formalin-fixed paraffin-embedded EC samples may be applied in future studies.

Our results demonstrate the utility of cancer genome characterisation in clinical specimens. The gene dosage pattern of the ERBB signalling network and the PI3K/Akt pathway differs in EC depending on the stage of the disease and the presence of metastases. The PI3K pathway is of particular importance in patients with aggressive disease. However, the prognostic value of the genes studied here remains to be determined in further survival studies. The results facilitate the understanding of the pathogenesis of EC and may also contribute to the development of more effective diagnostic tools. The potential for the application of this direct and rapid assay for EC patients is demonstrated in the case of the ERBB PI3K/Akt signalling pathway, whose status may provide additional diagnostic information. The developed highly sensitive and reproducible platform is feasible for EC characterisation.

Acknowledgements

The present study was supported by a grant from the National Science Centre (5715/B/P01/2010/38) and a grant from the Foundation for Polish Science Parent-Bridge Programme co-financed by the European Union within the European Regional Development Fund (DPS-424-5053/11). The present study was co-financed by the European Commission in the frame of the European Social Fund, by the European Social Fund, the State Budget, and the Pomorskie Voivodeship Budget according to the Operational Programme Human Capital, Priority VIII, Action 8.2, Under-action 8.2.2: ‘Regional Innovative Strategy’ within the system project of the Pomorskie Voivodeship ‘InnoDoktorant-Scholarships for PhD students, Vth edition’. The publication was also financed by the European Social Fund as a part of the project ‘Educators for the elite-integrated training program for PhD students, post-docs and professors as academic teachers at University of Gdansk’ within the framework of the Human Capital Operational Programme, Action 4.1.1, improving the quality of educational offer of tertiary education institutions. The sponsors had no involvement in the present study design, collection, analysis and interpretation of the data, writing of the manuscript, or the decision to submit the manuscript for publication. The authors acknowledge Aleksandra Markiewicz for her support during the development of the multimarker platform and Bartosz Supernat for data visualisation, dendrogram generation, complete linkage computations and artwork adjustments. The manuscript was edited by Elsevier Language Editing Services.

References

1 

Mirantes C, Espinosa I, Ferrer I, Dolcet X, Prat J and Matias-Guiu X: Epithelial-to-mesenchymal transition and stem cells in endometrial cancer. Hum Pathol. 44:1973–1981. 2013. View Article : Google Scholar : PubMed/NCBI

2 

Wojciechowska U and Didkowska J: Cancer in Poland in 2010. Nowotwory. 63:197–216. 2013.

3 

Bray F, Dos Santos Silva I, Moller H and Weiderpass E: Endometrial cancer incidence trends in Europe: underlying determinants and prospects for prevention. Cancer Epidemiol Biomarkers Prev. 14:1132–1142. 2005. View Article : Google Scholar : PubMed/NCBI

4 

Bokhman JV: Two pathogenetic types of endometrial carcinoma. Gynecol Oncol. 15:10–17. 1983. View Article : Google Scholar : PubMed/NCBI

5 

Horn LC, Meinel A, Handzel R and Einenkel J: Histopathology of endometrial hyperplasia and endometrial carcinoma: an update. Ann Diagn Pathol. 11:297–311. 2007. View Article : Google Scholar

6 

Amant F, Moerman P, Neven P, Timmerman D, Van Limbergen E and Vergote I: Endometrial cancer. Lancet. 366:491–505. 2005. View Article : Google Scholar : PubMed/NCBI

7 

Randall TC and Kurman RJ: Progestin treatment of atypical hyperplasia and well-differentiated carcinoma of the endometrium in women under age 40. Obstet Gynecol. 90:434–440. 1997. View Article : Google Scholar : PubMed/NCBI

8 

Sherman ME, Sturgeon S, Brinton LA, et al: Risk factors and hormone levels in patients with serous and endometrioid uterine carcinomas. Mod Pathol. 10:963–968. 1997.PubMed/NCBI

9 

Ayhan A, Gultekin M and Dursun P: Textbook of Gynaecological Oncology. Günes Publishing; Ankara: 2010

10 

Yeramian A, Moreno-Bueno G, Dolcet X, et al: Endometrial carcinoma: molecular alterations involved in tumor development and progression. Oncogene. 32:403–413. 2012. View Article : Google Scholar : PubMed/NCBI

11 

Busmanis I, Ho TH, Tan SB and Khoo KS: p53 and bcl-2 expression in invasive and pre-invasive uterine papillary serous carcinoma and atrophic endometrium. Ann Acad Med Singapore. 34:421–425. 2005.PubMed/NCBI

12 

Caduff RF, Johnston CM and Frank TS: Mutations of the Ki-ras oncogene in carcinoma of the endometrium. Am J Pathol. 146:182–188. 1995.PubMed/NCBI

13 

Mutter GL, Lin MC, Fitzgerald JT, et al: Altered PTEN expression as a diagnostic marker for the earliest endometrial precancers. J Natl Cancer Inst. 92:924–930. 2000. View Article : Google Scholar : PubMed/NCBI

14 

Mutter GL, Boynton KA, Faquin WC, Ruiz RE and Jovanovic AS: Allelotype mapping of unstable microsatellites establishes direct lineage continuity between endometrial precancers and cancer. Cancer Res. 56:4483–4486. 1996.

15 

Strissel PL, Ellmann S, Loprich E, et al: Early aberrant insulin-like growth factor signaling in the progression to endometrial carcinoma is augmented by tamoxifen. Int J Cancer. 123:2871–2879. 2008. View Article : Google Scholar : PubMed/NCBI

16 

Rose PG: Endometrial carcinoma. N Engl J Med. 335:640–649. 1996. View Article : Google Scholar : PubMed/NCBI

17 

Onuma K, Dabbs DJ and Bhargava R: Mammaglobin expression in the female genital tract: immunohistochemical analysis in benign and neoplastic endocervix and endometrium. Int J Gynecol Pathol. 27:418–425. 2008. View Article : Google Scholar : PubMed/NCBI

18 

Grushko TA, Filiaci VL, Mundt AJ, et al: An exploratory analysis of HER-2 amplification and overexpression in advanced endometrial carcinoma: a gynecologic oncology group study. Gynecol Oncol. 108:3–9. 2008. View Article : Google Scholar : PubMed/NCBI

19 

Zaczek A, Wełnicka-Jaśkiewicz M, Bielawski KP, et al: Gene copy numbers of HER family in breast cancer. J Cancer Res Clin Oncol. 134:271–279. 2008. View Article : Google Scholar

20 

Zaczek A, Brandt B and Bielawski KP: The diverse signaling network of EGFR, HER2, HER3 and HER4 tyrosine kinase receptors and the consequences for therapeutic approaches. Histol Histopathol. 20:1005–1015. 2005.PubMed/NCBI

21 

Salvesen HB, Carter SL, Mannelqvist M, et al: Integrated genomic profiling of endometrial carcinoma associates aggressive tumors with indicators of PI3 kinase activation. Proc Natl Acad Sci USA. 106:4834–4839. 2009. View Article : Google Scholar : PubMed/NCBI

22 

Ponzielli R, Katz S, Barsyte-Lovejoy D and Penn LZ: Cancer therapeutics: targeting the dark side of Myc. Eur J Cancer. 41:2485–2501. 2005. View Article : Google Scholar : PubMed/NCBI

23 

Semczuk A and Jakowicki JA: Alterations of pRb1-cyclin D1-cdk4/6-p16INK4A pathway in endometrial carcinogenesis. Cancer Lett. 203:1–12. 2004. View Article : Google Scholar : PubMed/NCBI

24 

Tsiambas E, Alexopoulou D, Lambropoulou S, Gerontopoulos K, Karakitsos P and Karameris A: Targeting topoisomerase IIa in endometrial adenocarcinoma: a combined chromogenic in situ hybridization and immunohistochemistry study based on tissue microarrays. Int J Gynecol Cancer. 16:1424–1431. 2006. View Article : Google Scholar

25 

Pecorelli S: Revised FIGO staging for carcinoma of the vulva, cervix, and endometrium. Int J Gynaecol Obstet. 105:103–104. 2009. View Article : Google Scholar : PubMed/NCBI

26 

Consultation WE: Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 363:157–163. 2004. View Article : Google Scholar : PubMed/NCBI

27 

Jönsson G, Staaf J, Olsson E, et al: High-resolution genomic profiles of breast cancer cell lines assessed by tiling BAC array comparative genomic hybridization. Genes Chromosomes Cancer. 46:543–558. 2007.PubMed/NCBI

28 

Vandesompele J, De Preter K, Pattyn F, et al: Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 3:research00342002. View Article : Google Scholar : PubMed/NCBI

29 

Hellemans J, Mortier G, De Paepe A, Speleman F and Vandesompele J: qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biol. 8:R192007. View Article : Google Scholar : PubMed/NCBI

30 

McShane LM, Altman DG, Sauerbrei W, et al: REporting recommendations for tumor MARKer prognostic studies (REMARK). Nat Clin Pract Oncol. 2:416–422. 2005. View Article : Google Scholar

31 

Järvinen TA, Tanner M, Rantanen V, et al: Amplification and deletion of topoisomerase IIα associate with ErbB-2 amplification and affect sensitivity to topoisomerase II inhibitor doxorubicin in breast cancer. Am J Pathol. 156:839–847. 2000.

32 

Supernat A, Lapińska-Szumczyk S, Sawicki S, Wydra D, Biernat W and Zaczek AJ: Deregulation of RAD21 and RUNX1 expression in endometrial cancer. Oncol Lett. 4:727–732. 2012.

33 

Inaki K and Liu ET: Structural mutations in cancer: mechanistic and functional insights. Trends Genet. 28:550–559. 2012. View Article : Google Scholar : PubMed/NCBI

34 

Watson MA and Fleming TP: Mammaglobin, a mammary-specific member of the uteroglobin gene family, is overexpressed in human breast cancer. Cancer Res. 56:860–865. 1996.PubMed/NCBI

35 

Classen-Linke I, Moss S, Gröting K, Beier HM, Alfer J and Krusche CA: Mammaglobin 1: not only a breast-specific and tumour-specific marker, but also a hormone-responsive endometrial protein. Histopathology. 61:955–965. 2012. View Article : Google Scholar : PubMed/NCBI

36 

Raeder MB, Birkeland E, Trovik J, et al: Integrated genomic analysis of the 8q24 amplification in endometrial cancers identifies ATAD2 as essential to MYC-dependent cancers. PLoS One. 8:e548732013. View Article : Google Scholar : PubMed/NCBI

37 

Esteller M, García A, Martínez i Palones JM, Cabero A and Reventós J: Detection of c-erbB-2/neu and fibroblast growth factor-3/INT-2 but not epidermal growth factor receptor gene amplification in endometrial cancer by differential polymerase chain reaction. Cancer. 75:2139–2146. 1995. View Article : Google Scholar

38 

Liang H, Cheung LW, Li J, et al: Whole-exome sequencing combined with functional genomics reveals novel candidate driver cancer genes in endometrial cancer. Genome Res. 22:2120–2129. 2012. View Article : Google Scholar

39 

Saghir FS, Rose IM, Dali AZ, Shamsuddin Z, Jamal AR and Mokhtar NM: Gene expression profiling and cancer-related pathways in type I endometrial carcinoma. Int J Gynecol Cancer. 20:724–731. 2010. View Article : Google Scholar : PubMed/NCBI

40 

Konecny GE, Santos L, Winterhoff B, et al: HER2 gene amplification and EGFR expression in a large cohort of surgically staged patients with nonendometrioid (type II) endometrial cancer. Br J Cancer. 100:89–95. 2009. View Article : Google Scholar : PubMed/NCBI

41 

Rolitsky CD, Theil KS, McGaughy VR, Copeland LJ and Niemann TH: HER-2/neu amplification and overexpression in endometrial carcinoma. Int J Gynecol Pathol. 18:138–143. 1999. View Article : Google Scholar : PubMed/NCBI

42 

Tsikouras P, Bouchlariotou S, Vrachnis N, et al: Endometrial cancer: molecular and therapeutic aspects. Eur J Obstet Gynecol Reprod Biol. 169:1–9. 2013. View Article : Google Scholar : PubMed/NCBI

43 

Fleming GF, Sill MW, Darcy KM, et al: Phase II trial of trastuzumab in women with advanced or recurrent, HER2-positive endometrial carcinoma: a Gynecologic Oncology Group study. Gynecol Oncol. 116:15–20. 2010. View Article : Google Scholar

44 

Santin AD, Bellone S, Roman JJ, McKenney JK and Pecorelli S: Trastuzumab treatment in patients with advanced or recurrent endometrial carcinoma overexpressing HER2/neu. Int J Gynaecol Obstet. 102:128–131. 2008. View Article : Google Scholar : PubMed/NCBI

45 

Santin AD: Letter to the Editor referring to the manuscript entitled: ‘Phase II trial of trastuzumab in women with advanced or recurrent HER-positive endometrial carcinoma: a Gynecologic Oncology Group study’ recently reported by Fleming et al, (Gynecol Oncol., 116;15–20;2010). Gynecol Oncol. 118:95–96; author reply 96–97. 2010.

Related Articles

Journal Cover

2014-February
Volume 31 Issue 2

Print ISSN: 1021-335X
Online ISSN:1791-2431

Sign up for eToc alerts

Recommend to Library

Copy and paste a formatted citation
x
Spandidos Publications style
Supernat A, Łapińska-Szumczyk S, Majewska H, Gulczyński J, Biernat W, Wydra D and Żaczek AJ: A multimarker qPCR platform for the characterisation of endometrial cancer. Oncol Rep 31: 1003-1013, 2014.
APA
Supernat, A., Łapińska-Szumczyk, S., Majewska, H., Gulczyński, J., Biernat, W., Wydra, D., & Żaczek, A.J. (2014). A multimarker qPCR platform for the characterisation of endometrial cancer. Oncology Reports, 31, 1003-1013. https://doi.org/10.3892/or.2013.2924
MLA
Supernat, A., Łapińska-Szumczyk, S., Majewska, H., Gulczyński, J., Biernat, W., Wydra, D., Żaczek, A. J."A multimarker qPCR platform for the characterisation of endometrial cancer". Oncology Reports 31.2 (2014): 1003-1013.
Chicago
Supernat, A., Łapińska-Szumczyk, S., Majewska, H., Gulczyński, J., Biernat, W., Wydra, D., Żaczek, A. J."A multimarker qPCR platform for the characterisation of endometrial cancer". Oncology Reports 31, no. 2 (2014): 1003-1013. https://doi.org/10.3892/or.2013.2924