cancers Article In Silico Approach for Validating and Unveiling New Applications for Prognostic Biomarkers of Endometrial Cancer Eva Coll-de la Rubia 1, * , Elena Martinez-Garcia 2 , Gunnar Dittmar 2 , Petr V. Nazarov 2 , Vicente Bebia 3 , Silvia Cabrera 1,3 , Antonio Gil-Moreno 1,3 and Eva Colás 1, * 1 Biomedical Research Group in Gynecology, Vall Hebron Institute of Research, Universitat Autònoma de Barcelona, CIBERONC, 08035 Barcelona, Spain; scabrera@vhebron.net (S.C.); agil@vhebron.net (A.G.-M.) 2 Luxembourg Institute of Health, L-1445 Strassen, Luxembourg; elena.martinezgarcia@lih.lu (E.M.-G.); gunnar.dittmar@lih.lu (G.D.); petr.nazarov@lih.lu (P.V.N.) 3 Gynaecological Department, Vall Hebron University Hospital, CIBERONC, 08035 Barcelona, Spain; vbebia@vhebron.net * Correspondence: eva.coll@vhir.org (E.C.-d.l.R.); eva.colas@vhir.org (E.C.); Tel.: +34-9-3489-4052 Simple Summary: Endometrial cancer (EC) mortality is directly associated with the presence of poor prognostic factors. Molecular prognostic factors have been identified, but none are used in clinical practice due to lack of validation studies. This study aims to validate a set of 255 prognostic biomarkers previously identified in an extensive literature review and explore new prognostic applications by analyzing them in The Cancer Genome Atlas (TCGA) and Clinical Proteomic Tumor  Analysis Consortium (CPTAC) databases. A total of 30 biomarkers were validated and associated  to a histological type (n = 15), histological grade (n = 6), FIGO stage (n = 1), molecular classification Citation: Coll-de la Rubia, E.; (n = 16), overall survival (n = 11), and recurrence-free survival (n = 5). Our results encourage further Martinez-Garcia, E.; Dittmar, G.; studies of understudied biomarkers such as TPX2, and validates already broadly studied biomarkers Nazarov, P.V.; Bebia, V.; Cabrera, S.; such as MSH6, MSH2, or L1CAM, among others. Finally, our results present a significant step to Gil-Moreno, A.; Colás, E. In silico advance the quest for biomarkers to accurately assess the risk of EC patients. Approach for Validating and Unveiling New Applications for Abstract: Endometrial cancer (EC) mortality is directly associated with the presence of prognostic Prognostic Biomarkers of Endometrial Cancer. Cancers 2021, 13, factors. Current stratification systems are not accurate enough to predict the outcome of patients. 5052. https://doi.org/10.3390/ Therefore, identifying more accurate prognostic EC biomarkers is crucial. We aimed to validate 255 cancers13205052 prognostic biomarkers identified in multiple studies and explore their prognostic application by analyzing them in TCGA and CPTAC datasets. We analyzed the mRNA and proteomic expression Academic Editor: Kazuaki Takabe data to assess the statistical prognostic performance of the 255 proteins. Significant biomarkers related to overall survival (OS) and recurrence-free survival (RFS) were combined and signatures generated. Received: 6 September 2021 A total of 30 biomarkers were associated either to one or more of the following prognostic factors: Accepted: 2 October 2021 histological type (n = 15), histological grade (n = 6), FIGO stage (n = 1), molecular classification Published: 9 October 2021 (n = 16), or they were associated to OS (n = 11), and RFS (n = 5). A prognostic signature composed of 11 proteins increased the accuracy to predict OS (AUC = 0.827). The study validates and identifies Publisher’s Note: MDPI stays neutral new potential applications of 30 proteins as prognostic biomarkers and suggests to further study with regard to jurisdictional claims in under-studied biomarkers such as TPX2, and confirms already used biomarkers such as MSH6, published maps and institutional affil- MSH2, or L1CAM. These results are expected to advance the quest for biomarkers to accurately iations. assess the risk of EC patients. Keywords: endometrial cancer; prognostic biomarker; uterine cancer; high-risk; bioinformatics; CPTAC; TCGA Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and 1. Introduction conditions of the Creative Commons Attribution (CC BY) license (https:// Endometrial cancer (EC) is the fourth most common cancer in women in developed creativecommons.org/licenses/by/ countries and the sixth in terms of mortality [1]. Unlike other cancers, in the last years EC 4.0/). has been rising in both incidence and associated mortality. By 2040, incidence is expected Cancers 2021, 13, 5052. https://doi.org/10.3390/cancers13205052 https://www.mdpi.com/journal/cancers Cancers 2021, 13, 5052 2 of 15 to increase 23% and mortality will rise by 33% worldwide [2]. Although most women diagnosed with EC are in early-stage disease and have a favorable outcome, the mortality increases dramatically for women with recurrent or advanced disease and for women diagnosed with a clinically aggressive tumor [3]. To manage the diagnosis, treatment, and follow-up of EC patients, multidisciplinary evidence-based guidelines on selected clinically relevant questions have been developed and updated over the years by the European Society for Medical Oncology (ESMO), the European Society of Gynaecological Oncology (ESGO), the European Society for Radiother- apy and Oncology (ESTRO) and the European Society of Pathology (ESP) consortiums [3,4]. These guidelines classify EC in different risk groups based on prognostic factors, including histological subtype (endometrioid or non-endometrioid), tumor grade (low, intermediate or high grade), depth of myometrial invasion, cervical involvement, tumor size, lympho- vascular space invasion (LVSI), lymph node status (LNS), tumor spread, and recently, based on molecular classification, which subdivides EC in four molecular groups: POLE ultramu- tated, microsatellite stability instable (MSI) hypermutated, copy-number low (CN-LOW) (microsatellite stable, MSS), and copy-number high (CN-HIGH) (serous-like) [4,5]. These prognostic factors should be determined at the moment of diagnosis since they will guide an optimal surgical treatment, which is the cornerstone treatment for EC; and after surgery, to provide the final diagnosis, stratification of the tumor, and guide the adjuvant treatment [3,4]. While prognostic factors are accurately determined by the histopathological examination of the tumor specimen after surgery, they are often inaccurately determined at early steps of the diagnostic process and need to benefit from multiple approaches. Specifically, histological subtype and tumor grade are assessed through the histopathological examina- tion of an endometrial biopsy, while depth of myometrial invasion, tumor size, cervical involvement, tumor spread, and LNS are determined though imaging techniques. The histological diagnosis is determined through the histopathological examination of an endometrial biopsy, preferably obtained by aspiration [4]. It should be reviewed by an expert pathologist to improve the accuracy of histological assessment and the reliability of tumor grading. Although conventional pathological analysis is critical for tumor stratification, it suffers from great interobserver variation [4]. A recent meta-analysis showed discrepancies in 33% of cases between preoperative and postoperative grading. Clinically relevant down- grading was reported in 26% and upgrading in 8% of the patient’s samples [6]. Additionally, interobserver variability also depends on the sampling method. Diagnosis performed by aspiration reached 73% agreement, while hysteroscopic biopsies had a significantly higher agreement (89%) in comparison to dilatation and curettage (70%) [6]. Importantly, the subjec- tive pathological analysis may result in either undertreatment or overtreatment of EC patients, which may lead to comorbidities or even life-threatening risks. Multiple imaging techniques are used to determine prognostic factors associated with the stage of the disease. Transvaginal ultrasonography (TVUS), or magnetic resonance imaging (MRI) will assess myometrial invasion, tumor size, and cervical invasion, while computed tomography (CT), MRI or positron emission tomography (PET-CT) will evaluate the lymph node status. However, these techniques are not sensitive enough, and they are not able to determine LVSI, which is a strong predictor of nodal metastasis, recurrence, and cancer-specific death [7,8]. Thus, LVSI is not studied at the time of diagnosis although being an important prognostic factor to predict the outcome of the patients. Lastly, the emerging molecular classification needs to be determined for the risk as- sessment of EC patients. To simplify the whole molecular classification, the TransPORTEC group proposed a surrogate classification which is highly effective for most of EC cases and is based on the assessment of POLE sequencing, immunohistochemistry of mismatch repair proteins (MMR-IHQ), and immunohistochemistry of the p53 (IHQ-p53) protein to classify EC patients [9–11]. Nevertheless, this approach is not broadly implemented in most hospitals due to methodological limitations and controversies for the evaluation of the test results [12–14]. Considering this inaccurate scenario for determining the risk classification of EC, an important part of EC research is directed to the hunt for biomarkers, particularly to Cancers 2021, 13, 5052 3 of 15 provide accurate information at the time of diagnosis. In Coll-de la Rubia E et al., 2020 [15], we reviewed 2507 publications ranging from 1991 to February 2020 and compiled a total number of 255 proteins described as prognostic EC biomarkers. Unfortunately, the vast majority of these biomarkers have not been introduced in clinical practice, probably due to a lack of validation in independent studies, reliability or existing evidence. In this publication we aim to validate and identify new prognostic applications for the 255 prognostic biomarkers described in Coll-de la Rubia E et al., 2020 [15] by performing a statistical analysis using the accessible datasets of TCGA and CPTAC studies [5,16]. Validated proteins were combined to achieve higher accuracy to predict patient’s outcome, and their biological significance was investigated. 2. Materials and Methods 2.1. Data Source Expression data profiles of EC patients were collected from the TCGA database though cBioPortal (https://www.cbioportal.org/, accessed on 27 June 2020). The RNA-Seq expression data of 333 EC patients from the Uterine Corpus endometrial Carcinoma (TCGA, Nature 2013) study (TCGA-RNAseq) was used. CPTAC—Uterine Corpus Endometrial Carcinoma data was obtained from LinkedOmics database (http://www.linkedomics. org/login.php, accessed on 10 June 2020). RNA-Seq (CPTAC-RNAseq) and proteomic (CPTAC-proteome) data corresponding to 95 EC patients was used for the analysis. Table 1 details the clinical information of the EC patients used for the analysis of this study. Table 1. Clinical, pathological, and molecular information of the patients. Detailed clinical, patholog- ical, and molecular information of the patients included in this study. CPTAC RNA-Seq + Proteome Characteristics TCGA RNA-Seq (n = 333) (n = 95) Age 1 Mean 63.23 ± 10.91 63.19 ± 9.78 Maximum 90 86 Minimum 33 38 Histological type Endometrioid 271 83 Serous 52 12 Mixed 10 0 Grade Grade 1 79 37 Grade 2 90 38 Grade 3 164 8 FIGO stage 2 I 226 71 II 19 8 III 70 13 IV 16 3 NA 2 0 Molecular Classification 3 POLE 31 7 MSI 92 25 CN-low 110 43 CN-high 78 20 Overall Survival 0: Living 282 36 1: Deceased 51 7 1 Age: means and standard deviations are shown. 2 FIGO stage: Federation of Gynecologists and Obstetricians for staging. 3 POLE: POLE ultramutated; MSI: microsatellite stable instable hypermutated; CN-low: copy number low, endometrioid-like; CN-high: copy number high, serous-like. Cancers 2021, 13, 5052 4 of 15 2.2. Data Processing and Identification of Differentially Expressed Genes (DEGs) and Proteins (DEPs) The 255 proteins identified in Coll-de la Rubia E et al., 2020 [15] were subtracted from the three datasets of expression data, which were separately analyzed using the limma and reportROC packages of R software. The criteria of false discovery rate (FDR) adjusted p-value < 0.25, | logFC | > 1, and Area Under the ROC Curve (AUC) > 0.75 were applied to screen the DEGs and DEPs (DEG/Ps). The DEG/Ps that were overlapping in the TCGA-RNAseq and [CPTAC-RNAseq OR CPTAC-proteome] EC datasets were named as validated biomarkers. 2.3. Survival Analysis We used the TCGA dataset to identify the potential genes with an impact on OS and RFS. DEGs with FDR < 0.05 and AUC > 0.6 at time points of 12, 24, 36 or 48 months were subsequently used to construct the Cox proportional hazards regression model to predict OS and RFS. 2.4. Statistical Analysis Comparisons between histological types (endometrioid EC vs. non-endometrioid EC) and histological grade (low-grade -G1 and G2- EC vs. high-grade -G3- EC) were performed using T Test. Tukey’s honest significance test was used to perform multiple comparisons for FIGO stage (I vs. II vs. III vs. IV) and for molecular classification (POLE vs. MSI vs. CN-LOW vs. CN-HIGH). An AUC value for each comparison was also calculated. The univariate Cox proportional hazards regression analyses were completed using the survival package of R software. AUC values were calculated using survivalROC package of R for OS and RFS at time points 12, 24, 36 and 48 months. Risk scores for each patient were calculated as follows: n Risc score = ∑ expi ∗ coe f i i=1 where “n” is the number of related prognostic genes, “expi ” is the expression value of the gene i, and “coefi ” is the log hazard ratio (LHR) in univariate Cox regression analysis [17]. Then, the median risk value was used to divide the patients into high and low-risk groups, while the Kaplan–Meier curve was applied to assess the survival difference between the two groups using the log-rank test. Receiver operating characteristic (ROC) curves for assessing the sensitivity and specificity of the prognostic signatures was generated using the survivalROC package implemented in R. 2.5. Functional Analysis of DEG/Ps, Interactions, and Tractability Information We used the Panther database to identify the biological processes and pathways asso- ciated with the DEGs. A FDR of < 0.05 was considered as statistically significant [18]. We reported the subcellular location of each protein using UNIPROT [19]. The potential rela- tionship between DEGs encoding proteins was analyzed using the STRING database [20]. Finally, to assess the current available drugs against our DEG/Ps we used the Open Targets Platform [21]. 3. Results 3.1. Study Workflow We recently used an exhaustive literature revision to compile a total number of 255 pro- teins as potential prognostic protein biomarkers [15]. These were defined as proteins which have been associated with one or more known clinical prognostic EC factors, including histological subtype, tumor grade, depth of myometrial invasion, cervical involvement, tumor size, LVSI, LNS, tumor spread; as well as the molecular classification, recurrence and/or survival [15]. Most of these biomarkers have not been validated in independent studies, jeopardizing their implementation in the clinical practice. To validate the poten- tial of those proteins as EC prognostic biomarkers and unveil novel potential prognostic Cancers 2021, 13, 5052 5 of 15 associations, we performed an in silico analysis of those proteins in 428 EC patients be- longing to the CPTAC and TCGA studies. The workflow of this study is depictured in Figure 1. Briefly, the 255 biomarkers were assessed in the RNA-Seq data of the TCGA and CPTAC datasets, in addition to the proteomic data of the CPTAC dataset. The most relevant prognostic factors, which are histological type, histological grade, FIGO stage, and molecular classification, were analyzed using a differential expression analysis and the calculation of the area under the ROC curve (AUC) values. Additionally, OS and RFS were assessed using a Cox analysis. Statistically significant biomarkers were identified for each parameter and dataset, and those that appeared significant in at least two datasets were considered as validated biomarkers. Among the 255 potential prognostic biomarkers, only 30 biomarkers were validated, and those were further studied using functional analysis and drug tractability studies. • Prognostic factors: TCGARNAseq • Histological type Validated in two different • Histological grade cohorts of patients: Literature. Prognostic CPTACRNAseq • FIGO stage TCGA- RNAseq proteomic biomarkers • Molecular classification AND CPTACProt [CPTAC-Seq OR CPTAC-Prot] Coll-de la Rubia E et al., J Clin Med, 2020 FDR < 0.25; | logFC | > 2; AUC > 0.75 Differential Expression Analysis DEG/Ps lists AUC calculation Functional Analysis 24 Drug tractability Set Up 255 biomarkers 30 Cox DEG/Ps lists Models and protein AUC calculation 14 panels of biomarkers FDR < 0.05; AUC > 0.6 (12, 24, 36 or 48 months) TCGA- RNAseq • Overall Survival • Recurrence Free Survival Figure 1. Workflow of the analysis. The 255 biomarkers compiled in Coll-de la Rubia E et al. 2020 [15] were assessed in two independent cohorts of patients from the TCGA and CPTAC studies. RNA-Seq data from both cohorts and proteomic data from the CPTAC cohort was used. Different prognostic factors were analyzed (histological type, histological grade, FIGO stage, Molecular classification), as well as overall survival and recurrence free survival. Finally, 30 biomarkers were identified as promising for the stratification of EC tumors. Among the 30 validated biomarkers, we encountered proteins that were significant for one or multiple prognostic parameters (Figure 2). TPX2 is protein associated with a major number of prognostic parameters, including histological type and grade, molecular classification, and OS and RFS. This protein was previously studied in Jiang T et al., 2018 where it was associated with worse prognosis [22], and recently its prognostic value in EC was further demonstrated [23,24]. Cancers 2021, 13, x FOR PEER REVIEW 6 of 15 Cancers 2021, 13, 5052 6 of 15 where it was associated with worse prognosis [22], and recently its prognostic value in EC was further demonstrated [23,24]. Figure Figure 2. Summary 2. Summary forfor thethe30 30 validated validated biomarkers biomarkers inin thethe analysis. analysis. Prognosticfactors Prognostic factorsororprognostic prognosticvalue valuedescribed describedininColl- Coll-de la Rubia E et al., 2020 for each of these biomarkers [15] are highlighted in grey. Additionally, the specific prognostic de la Rubia E et al., 2020 for each of these biomarkers [15] are highlighted in grey. Additionally, the specific prognostic factor factor or prognostic value validated in our analysis are indicated for each protein. The color green represents prognostic or factors prognostic foundvalue validated in both: in our literature analysis revision are indicated and statistical for each analysis, protein. whereas The blue the color colorrepresents green represents prognostic new prognostic factors features found in both: literature not described revision and before in literature. HT:statistical analysis, histological type; HG:whereas the color histological blue grade; represents FIGO: new MC: FIGO stage; prognostic features molecular classi-not described fication before definedinby literature. The Cancer HT:Genome histological Atlastype; HG: histological Network; grade; FIGO: OS: overall survival; FIGO stage; MC: RFS: recurrence-free molecular classification survival. defined by The Cancer Genome Atlas Network; OS: overall survival; RFS: recurrence-free survival. 3.2. Validated Prognostic Biomarkers in EC Regarding 3.2. Validated histological Prognostic subtype, Biomarkers inaEC total of 20, 36 and 18 genes/proteins were differen- tiallyRegarding expressed histological between endometrioid subtype, a (EEC) and total of 20,non-endometrioid (non-EEC)were 36 and 18 genes/proteins histolo- differ- gies in the TCGA-RNAseq, CPTAC-RNAseq, and CPTAC-prot datasets, entially expressed between endometrioid (EEC) and non-endometrioid (non-EEC) respectively (Fig-his- ure 3A).inHowever, tologies only 15 wereCPTAC-RNAseq, the TCGA-RNAseq, validated in at leastandtwo datasets (Figure CPTAC-prot 3B–D). datasets, From respectively those, eight biomarkers were previously described—CCNE1, CDKN2A, ERBB2, (Figure 3A). However, only 15 were validated in at least two datasets (Figure 3B–D). From ESR1, L1CAM, PAX8, PIGR, VIM—and seven are newly associated to the histological subtype: those, eight biomarkers were previously described—CCNE1, CDKN2A, ERBB2, ESR1, BUB1, CDC20, CDKN1A, HMGA1, S100A1, TPX2, UCHL1. The most confident biomarker L1CAM, PAX8, PIGR, VIM—and seven are newly associated to the histological subtype: proteins (present in the three datasets) were ERBB2, L1CAM, PIGR and TPX2. Among the BUB1, CDC20, CDKN1A, HMGA1, S100A1, TPX2, UCHL1. The most confident biomarker non-endometrioid subtypes analyzed in the TCGA dataset, 62 non-EEC cases were in- proteins (present in the three datasets) were ERBB2, L1CAM, PIGR and TPX2. Among the non-endometrioid subtypes analyzed in the TCGA dataset, 62 non-EEC cases were included, 10 cases of which were mixed subtype and 52 serous carcinomas. As seen in supplementary Figure S1, while some of the biomarkers behaved similarly between the non-endometrioid histologies, others such as CDC20, CDKN1A, ERBB2, HMGA1, L1CAM Cancers 2021, 13, x FOR PEER REVIEW 7 of 15 Cancers 2021, 13, 5052 7 of 15 cluded, 10 cases of which were mixed subtype and 52 serous carcinomas. As seen in sup- plementary Figure S1, while some of the biomarkers behaved similarly between the non- endometrioid histologies, others such as CDC20, CDKN1A, ERBB2, HMGA1, L1CAM and and PAX8 PAX8 werewere expressed expressed in mixed in mixed tumors tumors as a as a mixture mixture between between SECSEC andand EEC, EEC, similarly similarly to to the the nature nature of these of these tumors. tumors. Figure Figure3.3.Biomarkers Biomarkersrelated relatedtoto histological type histological and type andhistological grade. histological grade.(A,E) Table (A,E) of the Table of proteins that that the proteins werewerefound dif- found ferentially expressed differentially expressedbetween: (A) endometrioid between: (EEC)(EEC) (A) endometrioid and non-endometrioid and non-endometrioid (non-EEC), and (E) and (non-EEC), G1G2 (E)and G3, and G1G2 respec- G3, tively, in any of respectively, in the anytested of the cohorts. Highlighted tested cohorts. in yellow, Highlighted the specific in yellow, cohort in the specific which cohort inthat whichprotein was found that protein wasto be differ- found to be entially expressed between histologies and/or grades. Proteins highlighted in blue are those validated in more than one differentially expressed between histologies and/or grades. Proteins highlighted in blue are those validated in more than cohort, and therefore, the ones that we considered as validated biomarkers. (B–D) Boxplots showing the expression of the one cohort, and therefore, the ones that we considered as validated biomarkers. (B–D) Boxplots showing the expression 15 validated biomarkers for histological type in each cohort of patients: TCGA RNA-Seq data, CPTAC RNA-Seq data, and of the 15 CPTAC validateddata, proteomic biomarkers for histological respectively. type showing (F–H) Boxplots in each cohort of patients: the expression TCGA of the RNA-Seqbiomarkers six validated data, CPTAC for RNA-Seq histolog- data, and CPTAC proteomic data, respectively. (F–H) Boxplots showing the expression of the ical grade in each cohort of patients: TCGA RNA-Seq data, CPTAC RNA-Seq data, and CPTAC proteomic data, respec-six validated biomarkers for histological grade in each cohort of patients: TCGA RNA-Seq data, CPTAC RNA-Seq data, and CPTAC tively. Literature: literature revision from Coll-de la Rubia E et al., 2020 [15]; T_RNAseq: RNA-Seq data of the TCGA’s proteomic data, respectively. cohort; Literature: C_RNAseq: literature RNA-Seq datarevision from Coll-de of the CPTAC’s la Rubia cohort; E etproteomic C_prot: al., 2020 [15]; dataT_RNAseq: RNA-Seq of the CPTAC’s data of the TCGA’s cohort. cohort; C_RNAseq: RNA-Seq data of the CPTAC’s cohort; C_prot: proteomic data of the CPTAC’s cohort. Regarding histological grade, six proteins in the TCGA-RNAseq cohort, 21 proteins in theRegarding histological CPTAC-RNAseq grade, cohort, six proteins and seven in in proteins thethe TCGA-RNAseq cohort,showed CPTAC-prot cohort 21 proteins dif- in theprotein ferent CPTAC-RNAseq abundancescohort, and low-grade between seven proteins (gradein 1the CPTAC-prot and cohort grade 2) and showed high-grade different protein abundances between low-grade (grade 1 and grade 2) and high-grade (grade 3) EC. From those, six proteins were validated in two datasets, and were also previously described in other studies: ASRGL1, ATAD2, CDC20, and TPX2 (Figure 3E–H). cers 2021, 13, x FOR PEER REVIEW 8 of 15 Cancers 2021, 13, 5052 8 of 15 (grade 3) EC. From those, six proteins were validated in two datasets, and were also pre- viously described in other studies: ASRGL1, ATAD2, CDC20, and TPX2 (Figure 3E–H). Thus, highlighting Thus, highlighting their importance their importance and the need and of the needvalidation further of furtherof validation of those in further those in further prospectiveprospective clinical studies. clinical studies. Regarding FIGORegardingstage,FIGO threestage, and 17three and 17 proteins proteins showed showed differential differential abundancesabundances in the in the TCGA and CPTAC cohorts of patients, respectively, 10 TCGA and CPTAC cohorts of patients, respectively, 10 of which were previously de- of which were previously described in literature. scribed in the the literature. However, However, only was only PGR PGRvalidated was validated in two in two cohorts cohorts of patients of patients and and its performance its performance was limited wasto limited to the comparison the comparison betweenbetween stage stage I and I and II S2A–D). II (Figure (Figure S2A–D). Finally, thisFinally, this studyidentifying study permitted permittedaidentifying a significant significant number number of of biomarkers biomarkers that al- that allow low separating separating between between different groupsdifferent groups of classification. of the molecular the molecularAclassification. A total of 16 total of 16 pro- proteins teins (ATAD2, CAPG, (ATAD2, CCNE1,CAPG,CDKN2A, CCNE1, ESR1,CDKN2A, ESR1, HMGA1, HMGA1, L1CAM, MSH2, MSH6,L1CAM, MSH2, MSH6, PAX8, S100A1, SCGB2A1, TMEFF2, TPX2, TRA2B, UCHL1) were confirmed, i.e., were statisti- i.e., were PAX8, S100A1, SCGB2A1, TMEFF2, TPX2, TRA2B, UCHL1) were confirmed, statistically cally significant significant in at least in at least two datasets two datasets (Figure (Figure 4, Figure 4, Figure S2E-G). S2E–G). Specifically, Specifically, L1CAM, L1CAM, ATAD2, CAPG, CNNE1, CDKN2A, ESR1, HMGA1, ATAD2, CAPG, CNNE1, CDKN2A, ESR1, HMGA1, MSH2, MSH6, PAX8, S100A, TPX2, MSH2, MSH6, PAX8, S100A, TPX2, TRA2B,showed TRA2B, UCHL1, UCHL1, capacity showed capacity to distinguish to distinguish betweenbetween CN-LOW CN-LOW vs. vs. CN-HIGH; L1CAM, CN-HIGH; CDKN2A, HMGA1, MSH6, TMEFF2, UCHL1 L1CAM, CDKN2A, HMGA1, MSH6, TMEFF2, UCHL1 between MSI vs. CN-HIGH; and between MSI vs. CN-HIGH; and L1CAM and CDKN2A to differentiate between POLE vs. L1CAM and CDKN2A to differentiate between POLE vs. CN-HIGH subgroups. Remark- CN-HIGH subgroups. Remarkably, L1CAM seems to be the most informative biomarker to ably, L1CAM seems to be the most informative biomarker to differentiate the CN-HIGH differentiate the CN-HIGH from the other molecular from the other molecular groups. groups. Figure 4. Biomarkers related to the Figure 4. Biomarkers molecular related classification. to the molecular Table of Table classification. genes/proteins that were of genes/proteins found that were differentially ex- expressed found differentially pressed between any of the molecular subgroups in any of the tested cohorts (in grey). Highlighted in yellow, the specific between any of the molecular subgroups in any of the tested cohorts (in grey). Highlighted in yellow, the specific cohort in cohort in which each protein was found to be differentially expressed between subgroups and indicated the specific com- which each protein was found to be differentially expressed between subgroups and indicated the specific comparison. The parison. The numbers indicate the specific comparison where each biomarker was found differential. Proteins highlighted numbers indicate the specific comparison where each biomarker was found differential. Proteins highlighted in blue are those validated in more than one cohort. MC: Molecular classification defined by The Cancer Genome Atlas Network; 1: Cancers 2021, 13, x FOR PEER REVIEW 9 of 15 Cancers 2021, 13, 5052 9 of 15 in blue are those validated in more than one cohort. MC: Molecular classification defined by The Cancer Genome Atlas Network; 1: Comparison between POLE mutated vs. microsatellite instability(MSI); 2: Comparison between POLE mu- tated vs.between Comparison copy number low (CN-Low); POLE mutated 3: Comparison vs. microsatellite between POLE instability(MSI); mutated vs.between 2: Comparison copy number POLE high (CN-High); mutated vs. copy4:number Com- low parison between microsatellite instability (MSI) vs. copy number low (CN-Low); 5: Comparison between microsatellite (CN-Low); 3: Comparison between POLE mutated vs. copy number high (CN-High); 4: Comparison between microsatellite instability (MSI)instability (MSI) vs. vs. copy number lowcopy number5:high (CN-Low); (CN-High); Comparison 6: copy between number lowinstability microsatellite (CN-Low) vs. copy (MSI) number vs. copy high number (CN-High). high (CN-High); 6: T_RNAseq: RNA-Seq data of the cohort of the TCGA; C_RNAseq: RNA-Seq data of the cohort of the CPTAC; C_prot: copy number low (CN-Low) vs. copy number high (CN-High). T_RNAseq: RNA-Seq data of the cohort of the TCGA; C_RNAseq: proteomic data of the cohort of the CPTAC. RNA-Seq data of the cohort of the CPTAC; C_prot: proteomic data of the cohort of the CPTAC. 3.3. Validated Biomarkers Associated to OS and RFS in EC 3.3. Validated Biomarkers Associated to OS and RFS in EC Among the 255 biomarkers studied, a total of 11 and 5 genes presented significant Among the 255 biomarkers studied, a total of 11 and 5 genes presented significant correlation with survival rates and recurrence, respectively (FDR < 0.05 and AUC > 0.6 at correlation with survival rates and recurrence, respectively (FDR < 0.05 and AUC > 0.6 at time points of 12, 24, 36, and/or 48 months) in two datasets (Figure 5A,B). In particular, time points of 12, 24, 36, and/or 48 months) in two datasets (Figure 5A,B). In particular, the the genes genes with with significant significant association association to to OS OS were were ASRGL1, ASRGL1, ESR1, ESR1, FASN, FASN, HDGF, HDGF, MACC1, MACC1, MCM6, MCM7, MSH2, MSH6, PTK2, and TPX2, while MCM6, MCM7, MSH2, MSH6, PTK2, and TPX2, while the ones associated the ones associated with with RFS were RFS ATAD2, were ATAD2,BUB1,BUB1, MSH6,MSH6,TPX2, TPX2, and TRA2B. AmongAmong and TRA2B. them, ASRGL1 and ESR1 them, ASRGL1 andwere ESR1charac- were terized as low as characterized risk, lowwhile risk, the remaining while 14 were14categorized the remaining as high-risk were categorized genes. An as high-risk in- genes. creased prediction for OS and RFS was achieved by the development An increased prediction for OS and RFS was achieved by the development of biomarkerof biomarker panels. This panels.was performed This by using was performed byall the significant using genes associated all the significant to OS and genes associated to OSRFS andinRFS a Cox in regression analysis. We used the prognostic signature to calculate a risk score a Cox regression analysis. We used the prognostic signature to calculate a risk score (see (see equa- tion in materials equation and methods) in materials and methods)for each patient, for each while patient, the median while value the median waswas value used to di- used to vide the patients into a high-risk (n = 166), and low-risk groups (n = 167) (Figure divide the patients into a high-risk (n = 166), and low-risk groups (n = 167) (Figure 5A,B). 5A,5B). An An 11-protein 11-protein model model reached reached anan AUC AUC of of 0.827 0.827 to predict OS, to predict OS, while while aa 5-protein 5-protein model model ofof RFS RFS reached reached anan AUC AUC of 0.712, both of 0.712, both atat 48 48 months’ months’ time. time. Figure 5. Figure 5. Biomarkers Biomarkers related related to to overall overall survival survival and and recurrence recurrence free free survival survival and and model model performance. performance. Best Best performing performing individual biomarkers are shown for (A) overall survival (OS) and (B) recurrence free survival (RFS) for a period of 12, 24, individual biomarkers are shown for (A) overall survival (OS) and (B) recurrence free survival (RFS) for a period of 12, 24, 36, and 48 months, respectively. Additionally, models for both were performed. Regarding prediction of OS, a model of 36, and 48 months, respectively. Additionally, models for both were performed. Regarding prediction of OS, a model of 11 11 proteins was used, (A) while a model of 5 proteins was used to predict RFS (B). proteins was used, (A) while a model of 5 proteins was used to predict RFS (B). 3.4. Biological Significance of the Validated Biomarkers 3.4. Biological Significance of the Validated Biomarkers Gene ontology and KEGG enrichment analysis were used to explore the biological Geneof functions ontology and the initial KEGG 255 enrichment proteins, analysis as well as were used the subgroup to explore of the the biological 30 proteins that were functions of the initial 255 proteins, as well as the subgroup of the 30 proteins that were Cancers 2021,13, Cancers2021, 13,5052 x FOR PEER REVIEW 10 of of1515 validatedininthis validated thisstudy. study.All Allofofthem them were were significantly significantly associated associated to the to the following following biologi- biological cal processes: cellular processes, biological regulation, response to stimulus, processes: cellular processes, biological regulation, response to stimulus, signaling, and signaling, and metabolic metabolic processes. processes. The setTheof set of validated validated genes genes had anhad an increased increased association association with with repro- reproductive ductive processes processes (Figure(Figure 6A). Additionally, 6A). Additionally, we studied we studied the function the function of thoseof30 those 30 val- validated idated biomarkers. biomarkers. While BUB1,While BUB1, CDC20, CCNE1, CCNE1,CDKN1A, CDC20, CDKN1A, CDKN2A, CDKN2A, MCM6, MCM7 MCM6,and MCM7 TPX2, and TPX2, played a role in cell-cycle, proteins such as ERBB2, ESR1, played a role in cell-cycle, proteins such as ERBB2, ESR1, L1CAM, PGR, PIGR, PTK2 L1CAM, PGR, PIGR, are PTK2 are involved molecules molecules in involved in the the activation activation cascade cascade that that enhance enhance the tumor the tumor growth. growth. Interestingly, Interestingly, 14 out of the 3014 out of the proteins are 30 proteinsasare described described proteins as proteins related related to the epithelial- to the epithelial-mesenchymal mesenchymal transition (EMT),transition (EMT), crucial for crucial forprogression the malignant the malignant[25]progression (Figure S3).[25] (Figure S3). A B Biological processes Pathways Integrin signalling pathway (P00034) immune system process (GO:0002376) Cell cycle (P00013) growth (GO:0040007) p53 pathway by glucose deprivation (P04397) metabolic process (GO:0008152) Huntington disease (P00029) biological adhesion (GO:0022610) PDGF signaling pathway (P00047) EGF receptor signaling pathway (P00018) locomotion (GO:0040011) FGF signaling pathway (P00021) multicellular organismal process (GO:0032501) Interleukin signaling pathway (P00036) developmental process (GO:0032502) Parkinson disease (P00049) signaling (GO:0023052) FAS signaling pathway (P00020) TGF-beta signaling pathway (P00052) response to stimulus (GO:0050896) VEGF signaling pathway (P00056) biological regulation (GO:0065007) Cadherin signaling pathway (P00012) reproduction (GO:0000003) p53 pathway feedback loops 2 (P04398) Inflammation mediated by chemokine and cytokine signaling pathway (P00031) interspecies interaction between organisms (GO:0044419) Apoptosis signaling pathway (P00006) localization (GO:0051179) Wnt signaling pathway (P00057) multi-organism process (GO:0051704) p53 pathway (P00059) reproductive process (GO:0022414) Gonadotropin-releasing hormone receptor pathway (P06664) Alzheimer disease-presenilin pathway (P00004) biological phase (GO:0044848) Angiogenesis (P00005) cellular process (GO:0009987) CCKR signaling map (P06959) 0 10 20 30 40 0 10 20 30 40 Percent of gene hit against total Percent of gene hit against total Literature Validated upregulated Validated downregulated C D Nucleus Cytoplasm CM Extracellular A SR G L 1 A TA D 2 AURKB BU B1 C A PG CCN A 2 C C N B1 C C N E1 C D C 20 C D K N 1A C D K N 2A ER BB2 E SR 1 F A SN HDGF HDGF HM GA1 L 1C A M M ACC1 M CM 6 M CM 7 M SH 2 M SH 6 PA X 8 PG R P IG R PLK 1 PTK 2 S100A 1 SC G B 2A 1 T M E FF2 TPX 2 T R A 2B U CH L1 V IM Figure6.6.Functional Figure Functional analysis. analysis. Top Top 20 represented 20 represented biological biological processes processes (A) pathways (A) and and pathways (B)the (B) from from 255the 255 proteins proteins reviewedre- inviewed in the literature the literature (out of 56 (out of 56 pathways) pathways) and all theand all the related pathways pathways related to the set oftovalidated the set ofproteins. validated proteins. Highlighted Highlighted in green andin green are yellow andthe yellow are the upregulated upregulated and downregulated and downregulated pathways pathways in relation in relation to the to the literature, literature,and respectively, respectively, in grey, theand in grey, pathways the pathways represented by the proteins compiled in Coll-de la Rubia E et al., 2020 [15]; (C) Subcellular location of the 30 represented by the proteins compiled in Coll-de la Rubia E et al., 2020 [15]; (C) Subcellular location of the 30 validated validated prognostic biomarkers; (D) String analysis of the 30 biomarkers. prognostic biomarkers; (D) String analysis of the 30 biomarkers. Regardingthe Regarding thepathways pathwaysanalysis, analysis,the the255 255proteins proteinshad hadaabalanced balancedassociation associationwith with multiplepathways, multiple pathways,in incontrast contrastto tothe the30 30validated validatedbiomarkers biomarkershaving havingaastrong strongrelation relation with the gonadotropin-releasing hormone receptor pathway, p53 pathway, p53 pathway with the gonadotropin-releasing hormone receptor pathway, p53 pathway, p53 pathway Cancers 2021, 13, 5052 11 of 15 feedback loops 2, interleukin signaling pathway, cell cycle, and integrin signaling pathway, as shown in Figure 6B. Additionally, while downregulated validated biomarkers were associated to interleukin signaling pathway among others, upregulated biomarkers were associated to other pathways such as cell cycle, cadherin signaling pathway, angiogenesis, integrin signaling pathway, VEGF signaling pathway, Parkinson disease, CCLR signaling map, EGF receptor signaling pathway and FAS signaling pathway. The sub-cellular lo- cation of the validated biomarkers was mainly in the nuclear and cytoplasm component (Figure 6C). In Figure 6D, the STRING analysis of all validated biomarkers pointed to MCM6, MCM7, CDC20, CCNE1, MSH2, CDKN1A, CDKN2A, ESR1 and ERBB2 at the core of the interaction network. These are fundamental proteins involved in the most altered described pathways triggering EC: ERK, PI3K, WNT, and transcription signaling pathways. Moreover, these are key pathways widely described in cancer. Thus, to further support our findings, we explored each of our 30 validated biomarkers and their prognostic association between other types of cancer in The Human Protein Atlas. All but ESR1, PAX8, PGR, TMEFF2, were associated with prognosis of breast, cervical, colorectal, head and neck, liver, lung, melanoma, ovarian, pancreatic, renal, or urothelial cancers. Interestingly, proteins such as CCNE1, FASN, HDGF, MCM7, PIGR, PTK2, or TRA2B have been described as having a prognostic relation with some other type of gynecological cancer (breast, cervical or ovarian cancer) (see Figure S3). Finally, as part of our literature search, we identified chemical probes that have been developed and are currently in different phases of clinical trials targeting some of our validated biomarkers. Specifically, small molecules for different applications have been developed against AURKB, ERBB2, ESR1, PGR, PLK1, and PTK2, as well as therapeutic antibodies against ERBB2 and VIM. In addition, current treatment for EC includes medrox- yprogesterone acetate, hydroxyprogesterone caproate, and megestrol acetate targeting PGR, for inoperable patients or for advanced or recurrent tumors [26,27]. Their effectiveness has been reported to increase with the combined use of estrogenic compounds such as tamoxifen targeting ESR1 [28] (Figure S4). 4. Discussion EC is the most common malignancy of the female genital tract. Its early diagnosis is related to good prognosis and overall survival. During the diagnostic process, accurate identification of prognostic factors is crucial for assessing the preoperative risk of recurrence for each patient and guide the surgical treatment. Moreover, the assessment of prognostic factors after surgery is needed for final staging and to guide the adjuvant treatment. To increase the accuracy and objectivity on the diagnostic process, several studies have discovered and described prognostic biomarker candidates, but none of them have reached the clinical practice, mostly because of a lack of independent validation studies. This study focused on the validation of previously described 255 prognostic EC biomarkers. Here we performed an in silico validation of those biomarkers in two of the currently available molecular EC studies, specifically the RNA sequencing data generated by the TCGA and CPTAC and the proteomic data generated by the CPTAC, which compiled data from a total of 428 EC patients. The results obtained here are based on those datasets, which predominately include white and Non-Hispanic or Latino as the main representation for race and ethnicity, respectively. Additionally, in a significant number of cases race and ethnicity clinical data was not reported (60 and 154 cases, respectively). Knowing that EC impacts differently in the overall survival depending on race [1], the discovery and use of EC biomarkers, as well as other cancer biomarkers, could have a different behavior depending on race. Thus, we encourage the scientific community to further investigate and validate their biomarkers on this issue. Our results revealed 30 biomarkers that show strong evidence for being prognostic EC biomarkers. Remarkably, only ESR1, PGR, ERBB2, L1CAM, MSH2, and MSH6 have been broadly studied as EC biomarkers in the literature [15]. This study adds 30 proteins that may carry important prognostic information in EC and, therefore, should be prioritized Cancers 2021, 13, 5052 12 of 15 for external validation studies. Most of the validated biomarkers discriminate histological subtype and grade, molecular classification, OS and RFS. Regarding FIGO stage, we could only validate PGR as a good prognostic factor, as described by others [29,30]. Among the most outstanding biomarkers, the spindle assembly factor TPX2 merits further attention since its differential expression allowed for the discrimination of all the prognostic factors studied, except for the FIGO stage. TPX2 is a spindle assembly factor required for normal assembly of mitotic spindles, which mediates AURKA localization to spindle microtubules. It has been studied in a broad range of cancers as prognostic marker, including renal, liver, pancreatic and lung cancers [31]. Importantly, TPX2 was also described as prognostic biomarker in gynecological cancer such as breast cancer [32] or ovarian cancer [33]. Re- garding EC, it was identified in two bioinformatics studies [22,34], and it has been further investigated in vitro to demonstrate its prognostic ability [23,24,35,36]. Another highlight in this study is the identification of biomarkers to classify EC according to the molecular classification. This classification has acquired an increasing relevance in this last year since it has been incorporated in the most recent clinical guideline of EC to improve risk assessment [4]. Despite the existence of a simplified classification, the Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE), this molecular classification is not implemented in all centers due to the technical complexity of analyzing the POLE mutations. Our study identified 16 previously described biomarkers with high discrimination potential regarding the molecular groups. We could accurately separate the CN-HIGH group from the others by using the L1CAM biomarker. L1CAM was found frequently expressed in the CN-HIGH group by Kommoss FK et al., 2018 [37] and they also showed that the L1CAM status was predictive of worse outcome in tumors with no specific molecular profile. More studies are needed to facilitate the classification of EC patients within this molecular classification, particularly to identify the POLE group and the groups classified as multiple classifiers. Related to OS and RFS, we developed an 11-biomarker signature, including ASRGL1, ESR1, FASN, HDGF, MACC1, MCM6, MCM7, MSH2, MSH6, PTK2, and TPX2, to predict OS at 48 months with an AUC = 0.827. This model improved the accuracy for the best performing biomarker alone at time points 12, 24, and 36 months. Contrary to this, the model of 5 proteins, including ATAD2, BUB1, MSH6, TPX2, and TRA2B, that predicts RFS only reached AUC = 0.712 at time point 48 months, and its performance was worse than the best performing biomarker to predict RFS alone for time points 12 and 24 months. MSH6 and TPX2 were included in both models; thus, they are highly likely to have an impact on the outcome of the patients. MSH6 is already a well-described protein required to classify patients in the new molecular classification, specifically the MSI group. TPX2 was highlighted in our study as a protein related to multiple prognostic factors. These findings highlight the importance of designing new studies to assess the prognosis of EC patients. Particularly, the need of studies specifically designed to identify biomarkers that can help in the prediction of EC recurrence, as currently it is not clear which are the proteins highly influencing the recurrence of the patients. Moreover, the proteins validated in this study have been described to have an impor- tant function in the biology of the tumors, having a role in the key pathways triggering cancer such as ERK, PI3K, WNT, as well as transcription signaling pathways. To support our findings, we also explored the prognostic role of these 30 biomarkers in other types of cancer (such as of breast, cervical, colorectal, head and neck, liver, lung, melanoma, ovarian, pancreatic, renal, or urothelial cancers), finding association of those with favorable or unfavorable outcomes of the patients depending on the protein and the cancer type. In summary, the 30 validated highlighted in this study have shown prognostic power in at least three independent cohorts of patients and thus, they seem promising candidates that merit further validation. Notably, their differential subcellular location should be considered in the design of clinical studies. The cytoplasmic and cellular membrane proteins such as ASRLG1, ATAD2, L1CAM, PGR, TPX2 or UCHL1 should be explored in tissues, whereas secreted proteins such as HDGF, PIGR, SCGB2A1, or TMEFF2 might be Cancers 2021, 13, 5052 13 of 15 easily measured in biofluids, opening the way for a less invasive diagnostic process. In fact, some of the nuclear and cell membrane validated proteins (MSH2, MSH6 and L1CAM) are already used in clinics to diagnose EC patients [3,38], while secreted proteins, such as PIGR, have diagnostic value in uterine aspirates for discriminating between histological subtypes [39]. Considering all the above mentioned, our results confirm that literature revision and further in silico validation of the previously described biomarkers in currently available, broadly documented, and considerable in size cohorts of patients such as the TCGA and CPTAC datasets is a valid approach for prioritizing robust biomarkers in future studies. Furthermore, we encourage researchers to validate the 30 biomarkers described in this study in independent studies to achieve the main goal of improving the risk assessment in EC. 5. Conclusions In summary, this study validates and identifies new potential applications of 30 pro- teins as prognostic biomarkers to discriminate for histological subtype and grade, FIGO stage, molecular classification, overall and recurrence-free survival. A model combination of 11-protein and 5-protein yield a higher accuracy to predict overall and recurrence-free survival, respectively. Our results permit to advance the quest for biomarkers to accurately assess the risk of EC patients. Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cancers13205052/s1, Figure S1: Boxplots of the described validated biomarkers for histological type representing different non-endometrioid histologies as different entities, Figure S2: Biomarkers related to FIGO stage with its respective boxplots, and molecular classification of the validated biomarkers, Figure S3: Description of the functions of the proteins described as validated biomarkers in our study, their relation to the epithelial-mesenchymal transition, and their prognostic behavior in other types of cancer, Figure S4: Available drugs against validated proteins divided in small molecules and antibodies. Author Contributions: Conceptualization, E.C.-d.l.R., S.C., and E.C.; methodology, E.C.-d.l.R., and P.V.N.; software, E.C.-d.l.R.; validation, G.D., E.M.-G., V.B. and A.G.-M.; formal analysis, E.C.-d.l.R.; investigation, E.C.-d.l.R.; resources, E.C.-d.l.R.; data curation, E.C.-d.l.R.; writing—original draft preparation, E.C.-d.l.R.; writing—review and editing, E.C.-d.l.R., E.C., G.D., and P.V.N.; visualization, E.C.-d.l.R.; supervision, E.C., P.V.N., S.C., and A.G.-M.; project administration, E.C.; funding acquisi- tion, E.C. and A.G.-M. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by grants from the Instituto de Salud Carlos III (ISCIII) grant number PI17/02155, PI20/00644, and the IFI19/00029 to E.C.-d.l.R., the Ministerio de ciencia, Inno- vación y Universidades through a RETOS Colaboración (RTC-2017-6261-1), both co-financed by the European Regional Development Fund (FEDER); from Fundación Científica Asociación Española Contra el Cáncer (AECC) grant number GCTRA1804MATI and CIBERONC network grant number CB16/12/00328; and Grups Consolidats de la Generalitat de Catalunya (2017SGR1661). E.C. is supported by an Investigator Grant from AECC (INVES20051COLA). E.M.-G. was supported by Televie grant F5/20/5-TLV/DD. Institutional Review Board Statement: The study did not directly involve human or animals. How- ever, data used for the research here presented was obtained from studies conducted according to the guidelines of the Declaration of Helsinki, and approved by their Instructional Review board (see Getz et al., 2013 [5] and Dou et al., 2020 [16]). Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are available in this article (and supplementary material). Conflicts of Interest: The authors declare no conflict of interest. Cancers 2021, 13, 5052 14 of 15 References 1. Siegel, R.L.; Miller, K.D.; Fuchs, H.E.; Jemal, A. Cancer Statistics, 2021. CA Cancer J. Clin. 2021, 71, 7–33. [CrossRef] 2. Ferlay, J.; Colombet, M.; Soerjomataram, I.; Mathers, C.; Parkin, D.M.; Piñeros, M.; Znaor, A.; Bray, F. Estimating the global cancer incidence and mortality in 2018: GLOBOCAN sources and methods. Int. J. Cancer 2019, 144, 1941–1953. [CrossRef] [PubMed] 3. Colombo, N.; Creutzberg, C.; Amant, F.; Bosse, T.; González-Martín, A.; Ledermann, J.; Marth, C.; Nout, R.; Querleu, D.; Mirza, M.R.; et al. ESMO-ESGO-ESTRO Consensus Conference on Endometrial Cancer: Diagnosis, treatment and follow-up. Ann. Oncol. 2016, 27, 16–41. [CrossRef] [PubMed] 4. Concin, N.; Matias-Guiu, X.; Vergote, I.; Cibula, D.; Mirza, M.R.; Marnitz, S.; Ledermann, J.; Bosse, T.; Chargari, C.; Fagotti, A.; et al. ESGO/ESTRO/ESP guidelines for the management of patients with endometrial carcinoma. Int. J. Gynecol. Cancer 2021, 31, 12–39. [CrossRef] 5. Getz, G.; Gabriel, S.B.; Cibulskis, K.; Lander, E.; Sivachenko, A.; Sougnez, C.; Lawrence, M.; Kandoth, C.; Dooling, D.; Fulton, R.; et al. Integrated genomic characterization of endometrial carcinoma. Nature 2013, 497, 67–73. [CrossRef] 6. Visser, N.C.M.; Reijnen, C.; Massuger, L.F.A.G.; Nagtegaal, I.D.; Bulten, J.; Pijnenborg, J.M.A. Accuracy of endometrial sampling in endometrial carcinoma: A systematic review and meta-analysis. Obstet. Gynecol. 2017, 130, 803–813. [CrossRef] 7. Brennan, D.J.; Hackethal, A.; Metcalf, A.M.; Coward, J.; Ferguson, K.; Oehler, M.K.; Quinn, M.A.; Janda, M.; Leung, Y.; Freemantle, M.; et al. Serum HE4 as a prognostic marker in endometrial cancer—A population based study. Gynecol. Oncol. 2014, 132, 159–165. [CrossRef] 8. Mannelqvist, M.; Stefansson, I.; Salvesen, H.B.; Akslen, L.A. Importance of tumour cell invasion in blood and lymphatic vasculature among patients with endometrial carcinoma. Histopathology 2009, 54, 174–183. [CrossRef] 9. Talhouk, A.; McConechy, M.K.; Leung, S.; Li-Chang, H.H.; Kwon, J.S.; Melnyk, N.; Yang, W.; Senz, J.; Boyd, N.; Karnezis, A.N.; et al. A clinically applicable molecular-based classification for endometrial cancers. Br. J. Cancer 2015, 113, 299–310. [CrossRef] [PubMed] 10. Talhouk, A.; McConechy, M.K.; Leung, S.; Yang, W.; Lum, A.; Senz, J.; Boyd, N.; Pike, J.; Anglesio, M.; Kwon, J.S.; et al. Confirmation of ProMisE: A simple, genomics-based clinical classifier for endometrial cancer. Cancer 2017, 123, 802–813. [CrossRef] [PubMed] 11. Vermij, L.; Smit, V.; Nout, R.; Bosse, T. Incorporation of molecular characteristics into endometrial cancer management. Histopathology 2020, 76, 52–63. [CrossRef] [PubMed] 12. Kommoss, S.; McConechy, M.K.; Kommoss, F.; Leung, S.; Bunz, A.; Magrill, J.; Britton, H.; Kommoss, F.; Grevenkamp, F.; Karnezis, A.; et al. Final validation of the ProMisE molecular classifier for endometrial carcinoma in a large population-based case series. Ann. Oncol. 2018, 29, 1180–1188. [CrossRef] [PubMed] 13. Murali, R.; Davidson, B.; Fadare, O.; Carlson, J.A.; Crum, C.P.; Blake Gilks, C.; Irving, J.A.; Malpica, A.; Matias-Guiu, X.; Glenn McCluggage, W.; et al. High-grade Endometrial Carcinomas: Morphologic and Immunohistochemical Features, Diagnostic Challenges and Recommendations. Int. J. Gynecol. Pathol. 2019, 38, S40–S63. [CrossRef] 14. León-Castillo, A.; Gilvazquez, E.; Nout, R.; Smit, V.T.H.B.M.; McAlpine, J.N.; McConechy, M.; Kommoss, S.; Brucker, S.Y.; Carlson, J.W.; Epstein, E.; et al. Clinicopathological and molecular characterisation of ‘multiple-classifier’ endometrial carcinomas. J. Pathol. 2020, 250, 312–322. [CrossRef] [PubMed] 15. Coll-de la Rubia, E.; Martinez-Garcia, E.; Dittmar, G.; Gil-Moreno, A.; Cabrera, S.; Colas, E. Prognostic Biomarkers in Endometrial Cancer: A Systematic Review and Meta-Analysis. J. Clin. Med. 2020, 9, 1900. [CrossRef] 16. Dou, Y.; Kawaler, E.A.; Cui Zhou, D.; Gritsenko, M.A.; Huang, C.; Blumenberg, L.; Karpova, A.; Petyuk, V.A.; Savage, S.R.; Satpathy, S.; et al. Proteogenomic Characterization of Endometrial Carcinoma. Cell 2020, 180, 729–748.e26. [CrossRef] 17. Nazarov, P.V.; Wienecke-Baldacchino, A.K.; Zinovyev, A.; Czerwińska, U.; Muller, A.; Nashan, D.; Dittmar, G.; Azuaje, F.; Kreis, S. Deconvolution of transcriptomes and miRNomes by independent component analysis provides insights into biological processes and clinical outcomes of melanoma patients. BMC Med. Genom. 2019, 12. [CrossRef] [PubMed] 18. Thomas, P.D.; Campbell, M.J.; Kejariwal, A.; Mi, H.; Karlak, B.; Daverman, R.; Diemer, K.; Muruganujan, A.; Narechania, A. PANTHER: A library of protein families and subfamilies indexed by function. Genome Res. 2003, 13, 2129–2141. [CrossRef] 19. Bateman, A. UniProt: A worldwide hub of protein knowledge. Nucleic Acids Res. 2019, 47, D506–D515. [CrossRef] 20. Szklarczyk, D.; Morris, J.H.; Cook, H.; Kuhn, M.; Wyder, S.; Simonovic, M.; Santos, A.; Doncheva, N.T.; Roth, A.; Bork, P.; et al. The STRING database in 2017: Quality-controlled protein-protein association networks, made broadly accessible. Nucleic Acids Res. 2017, 45, D362–D368. [CrossRef] 21. Ochoa, D.; Hercules, A.; Carmona, M.; Suveges, D.; Gonzalez-Uriarte, A.; Malangone, C.; Miranda, A.; Fumis, L.; Carvalho-Silva, D.; Spitzer, M.; et al. Open Targets Platform: Supporting systematic drug-target identification and pri- oritisation. Nucleic Acids Res. 2021, 49, D1302–D1310. [CrossRef] 22. Jiang, T.; Sui, D.; You, D.; Yao, S.; Zhang, L.; Wang, Y.; Zhao, J.; Zhang, Y. MiR-29a-5p inhibits proliferation and invasion and induces apoptosis in endometrial carcinoma via targeting TPX2. Cell Cycle 2018, 17. [CrossRef] 23. Zhang, W.; Gao, L.; Wang, C.; Wang, S.; Sun, D.; Li, X.; Liu, M.; Qi, Y.; Liu, J.; Lin, B. Combining Bioinformatics and Experiments to Identify and Verify Key Genes with Prognostic Values in Endometrial Carcinoma. J. Cancer 2020, 11, 716–732. [CrossRef] 24. Besso, M.; Montivero, L.; Lacunza, E.; Argibay, M.; Abba, M.; Furlong, L.; Colas, E.; Gil-Moreno, A.; Reventos, J.; Bello, R.; et al. Identification of early stage recurrence endometrial cancer biomarkers using bioinformatics tools. Oncol. Rep. 2020, 44, 873–886. [CrossRef] [PubMed] Cancers 2021, 13, 5052 15 of 15 25. Dongre, A.; Weinberg, R.A. New insights into the mechanisms of epithelial–mesenchymal transition and implications for cancer. Nat. Rev. Mol. Cell Biol. 2019, 20, 69–84. [CrossRef] [PubMed] 26. Thigpen, J.T.; Brady, M.F.; Alvarez, R.D.; Adelson, M.D.; Homesley, H.D.; Manetta, A.; Soper, J.T.; Given, F.T. Oral Medroxyproges- terone Acetate in the Treatment of Advanced or Recurrent Endometrial Carcinoma: A Dose-Response Study by the Gynecologic Oncology Group. J. Clin. Oncol. 1999, 17, 1736. [CrossRef] 27. Reifenstein, E.C. The treatment of advanced endometrial cancer with hydroxyprogesterone caproate. Gynecol. Oncol. 1974, 2, 377–414. [CrossRef] 28. Whitney, C.W.; Brunetto, V.L.; Zaino, R.J.; Lentz, S.S.; Sorosky, J.; Armstrong, D.K.; Lee, R.B. Phase II study of medroxyprogesterone acetate plus tamoxifen in advanced endometrial carcinoma: A Gynecologic Oncology Group study. Gynecol. Oncol. 2004, 92, 4–9. [CrossRef] [PubMed] 29. Guan, J.; Xie, L.; Luo, X.; Yang, B.; Zhang, H.; Zhu, Q.; Chen, X. The prognostic significance of estrogen and progesterone receptors in grade I and II endometrioid endometrial adenocarcinoma: Hormone receptors in risk stratification. J. Gynecol. Oncol. 2019, 30. [CrossRef] 30. Smith, D.; Stewart, C.J.R.; Clarke, E.M.; Lose, F.; Davies, C.; Armes, J.; Obermair, A.; Brennan, D.; Webb, P.M.; Nagle, C.M.; et al. ER and PR expression and survival after endometrial cancer. Gynecol. Oncol. 2018, 148, 258–266. [CrossRef] 31. Expression of TPX2 in Cancer-Summary-The Human Protein Atlas. Available online: https://www.proteinatlas.org/ENSG00000 088325-TPX2/pathology (accessed on 11 August 2021). 32. Jiang, Y.; Liu, Y.; Tan, X.; Yu, S.; Luo, J. TPX2 as a Novel Prognostic Indicator and Promising Therapeutic Target in Triple-negative Breast Cancer. Clin. Breast Cancer 2019, 19, 450–455. [CrossRef] 33. Cáceres-Gorriti, K.Y.; Carmona, E.; Barrès, V.; Rahimi, K.; Létourneau, I.J.; Tonin, P.N.; Mes-Masson, A.M. RAN nucleo-cytoplasmic transport and mitotic spindle assembly partners XPO7 and TPX2 are new prognostic biomarkers in serous epithelial ovarian cancer. PLoS ONE 2014, 9. [CrossRef] 34. Shen, L.; Liu, M.; Liu, W.; Cui, J.; Li, C. Bioinformatics analysis of RNA sequencing data reveals multiple key genes in uterine corpus endometrial carcinoma. Oncol. Lett. 2017, 15, 205–212. [CrossRef] 35. Liu, Y.; Chen, P.; Li, M.; Fei, H.; Huang, J.; Zhao, T.; Li, T. Comprehensive Analysis of the Control of Cancer Stem Cell Characteristics in Endometrial Cancer by Network Analysis. Comput. Math. Methods Med. 2021, 2021, 1–11. [CrossRef] 36. Liu, J.; Wu, Z.; Sun, R.; Nie, S.; Meng, H.; Zhong, Y.; Nie, X.; Cheng, W. Using mRNAsi to identify prognostic-related genes in endometrial carcinoma based on WGCNA. Life Sci. 2020, 258, 118231. [CrossRef] [PubMed] 37. Kommoss, F.K.; Karnezis, A.N.; Kommoss, F.; Talhouk, A.; Taran, F.-A.; Staebler, A.; Gilks, C.B.; Huntsman, D.G.; Krämer, B.; Brucker, S.Y.; et al. L1CAM further stratifies endometrial carcinoma patients with no specific molecular risk profile. Br. J. Cancer 2018, 119, 480–486. [CrossRef] [PubMed] 38. León-Castillo, A.; de Boer, S.M.; Powell, M.E.; Mileshkin, L.R.; Mackay, H.J.; Leary, A.; Nijman, H.W.; Singh, N.; Pollock, P.M.; Bessette, P.; et al. Molecular Classification of the PORTEC-3 Trial for High-Risk Endometrial Cancer: Impact on Prognosis and Benefit From Adjuvant Therapy. J. Clin. Oncol. 2020, 38, 3388–3397. [CrossRef] [PubMed] 39. Martinez-Garcia, E.; Lesur, A.; Devis, L.; Cabrera, S.; Matias-Guiu, X.; Hirschfeld, M.; Asberger, J.; Van Oostrum, J.; Casares de Cal, M.D.L.A.; Gomez-Tato, A.; et al. Targeted proteomics identifies proteomic signatures in liquid biopsies of the endometrium to diagnose endometrial cancer and assist in the prediction of the optimal surgical treatment. Clin. Cancer Res. 2017, 23, 6458–6467. [CrossRef] [PubMed]