{"paper_id":"c27716e1-8c44-4479-a343-fd46c3f86a38","body_text":"Genetic analysis of uterine aspirates improves\nthe diagnostic value and captures the\nintra-tumor heterogeneity of endometrial\ncancers\nAlba Mota 1,2, Eva Colás 3, Pablo García-Sanz 1,2, Irene Campoy 4, Alejandro Rojo-Sebastián 5,\nSonia Gatius 3, Ángel García 6, Luis Chiva 5, Sonsoles Alonso 5, Antonio Gil-Moreno 4,7,\nXavier González-Tallada3, Berta Díaz-Feijoo4,7,A u g u s tV i d a l8, Patrycja Ziober-Malinowska9,\nMarcin Bobiński9, Rafael López-López10, Miguel Abal 10, Jaume Reventós 8,11,\nXavier Matias-Guiu3,8 and Gema Moreno-Bueno1,2\n1Departamento de Bioquímica, Universidad Autónoma de Madrid (UAM), Instituto de Investigaciones\nBiomédicas ‘Alberto Sols ’ (CSIC-UAM), IdiPAZ, Madrid, Spain; 2MD Anderson International Foundation,\nMadrid, Spain; 3Department of Pathology and Molecular Genetics and Research Laboratory, Hospital\nUniversitari Arnau de Vilanova, University of Lleida, IRBLLEIDA, Lleida, Spain; 4Biomedical Research Group\nin Gynaecology, Vall d ’Hebron Research Institute and Hospital, and Universitat Autonoma de Barcelona,\nBarcelona, Spain; 5MD Anderson Cancer Center Madrid, Madrid, Spain; 6Department of Pathology, Hospital\nUniversitario Vall Hebron, Barcelona, Spain; 7Department of Obstetrics and Gynaecology, Hospital\nUniversitari Vall d'Hebron, Barcelona, Spain; 8Hospital Universati de Bellvitge, Institut d ’Investigacions\nBiomèdiques de Bellvitge (IDIBELL), Barcelona, Spain; 9First Chair and Department of Gynaecological\nOncology and Gynaecology, Medical University of Lublin, Lublin, Poland; 10Translational Medical Oncology,\nHealth Research Institute of Santiago (IDIS), SERGAS, Santiago, Spain and 11Departament de Ciències\nBàsiques, Universitat Internacional de Catalunya, Barcelona, Spain\nEndometrial cancer is the most common cancer of the female genital tract in developed countries. Although the\nmajority of endometrial cancers are diagnosed at early stages and the 5-year overall survival is around 80%, early\ndetection of these tumors is crucial to improve the survival of patients given that the advanced tumors are\nassociated with a poor outcome. Furthermore, correct assessment of the pre-clinical diagnosis is decisive to guide\nthe surgical treatment and management of the patient. In this sense, the potential of targeted genetic sequencing of\nuterine aspirates has been assessed as a pre-operative tool to obtain reliable information regarding the mutational\nprofile of a given tumor, even in samples that are not histologically classifiable. A total of 83 paired samples were\nsequenced (uterine aspirates and hysterectomy specimens), including 62 endometrioid and non-endometrioid\ntumors, 10 cases of atypical hyperplasia and 11 non-cancerous endometrial disorders. Even though diagnosing\nendometrial cancer based exclusively on genetic alterations is currently unfeasible, mutations were mainly found\nin uterine aspirates from malignant disorders, suggesting its potential in the near future for supporting the\nstandard histologic diagnosis. Moreover, this approach provides the first evidence of the high intra-tumor genetic\nheterogeneity associated with endometrial cancer, evident when multiple regions of tumors are analyzed from an\nindividual hysterectomy. Notably, the genetic analysis of uterine aspirates captures this heterogeneity, solving the\npotential problem of incomplete genetic characterization when a single tumor biopsy is analyzed.\nModern Pathology (2017) 30, 134–145; doi:10.1038/modpathol.2016.143; published online 2 September 2016\nEndometrial cancer is the fourth most common\ncancer among women in developed countries and\nthe most frequent cancer of the female genital tract. 1\nEndometrial cancer is mainly classified into two\ngroups with different clinical, pathological, and\nmolecular features. 2,3 Type I or endometrioid endo-\nmetrial carcinomas are normally low-grade,\nCorrespondence: Dr G Moreno-Bueno, PhD, Departamento de\nBioquímica, Facultad de Medicina (UAM), Instituto de Investigaciones\nBiomedicas‘Alberto Sols’ CSIC-UAM, C/Arzobispo Morcillo 2, Madrid\n28029, Spain or MD Anderson International Foundation,\nC/ Gomez Hemans 2, Madrid 28033, Spain.\nE-mail: gmoreno@iib.uam.es or gmoreno.fundacion@mdanderson.es\nReceived 20 December 2015; revised 29 June 2016; accepted 4 July\n2016; published online 2 September 2016\nModern Pathology (2017) 30, 134 –145\n134 © 2017 USCAP, Inc All rights reserved 0893-3952/17 $32.00\nwww.modernpathology.org\n\nestrogen-related tumors with a good prognosis.\nThese tumors are the most common endometrial\ncancers and they usually arise in perimenopausal\nwomen, preceded by or coexisting with endometrial\nhyperplasia. Type II tumors are high-grade non-\nendometrioid endometrial carcinomas, unrelated to\nestrogen, which occur in older women and have a\npoor prognosis.\n4,5 The majority of non-endometrioid\nendometrial carcinomas are serous endometrial\ncarcinomas, although there are also less frequent\nhistological subtypes, such as uterine carcinosarco-\nmas (also known as malignant mixed müllerian\ntumors), uncommon biphasic neoplasms with malig-\nnant epithelial elements, and a sarcomatoid compo-\nnent.\n5 At the molecular level, significant differences\nare evident between type I and II carcinomas. Muta-\ntions in PTEN, PIK3CA, PIK3R1, KRAS, CTNNB1,\nFGFR2, and ARID1A are associated with endome-\ntrioid endometrial carcinomas, whereas mutations in\nTP53, PIK3CA, and PPP2R1A are frequent in non-\nendometrioid endometrial carcinomas. 6,7 In addi-\ntion, microsatellite instability is evident in one-third\nof endometrioid endometrial carcinomas, a feature\nthat is unusual in non-endometrioid endometrial\ncarcinomas, which are more commonly character-\nized by chromosomal instability.\n8,9\nThe majority of endometrial cancers are diagnosed at\nearly stages and the associa ted 5-year overall survival\nis around 80%. Neverthe less, the survival rate\ndecreases to 57–46% for high-grade tumors.5 Further-\nmore, carcinosarcomas account for a high percentage\nof mortality despite constituting only 5 –6% of endo-\nmetrial cancers, principally because 60% of the\npatients presents extra uterine disease at the moment\nof diagnosis. In these cases more than 50% will suffer\nrecurrence after surgery and adjuvant treatment.\n10,11\nAs with many other tumors, early detection of\nendometrial cancer is crucial to increase patient\nsurvival, particularly as advanced tumors are asso-\nciated with a worse outcome. Moreover, the correct\nassessment of pre-clinical diagnosis is also decisive,\nas this will guide the pre-operative and surgical\nmanagement of the patient.\n12 In this sense, the use of\nuterine aspirates (Pipelle biopsies) as diagnostic pre-\noperative biopsies is widely recommended, repre-\nsenting a minimally invasive and highly sensitive\nprocedure. However, the failure rate in obtaining\nsuch samples is around 8%, whereas 13% of the\nsamples turn out to be histologically inadequate,\nfigures that are significantly higher in postmenopau-\nsal women.\n12,13 Moreover, discrepancies between\npre- and post-operative biopsies have been observed\nwith respect to histological grade, which could lead\nto a misclassification and the use of inappropriate\ntherapeutic strategies.\n14,15 As such, there has been\nsome interest in identifying molecular markers in\nuterine aspirates, enhancing their potential as a\ndiagnostic sample for both histological classification\nand molecular characterization of tumors.\n16–18\nDescribing the genetic profile of tumors can be\ndecisive for their accurate diagnosis and for\ntherapeutic decision-making. However, intra-tumor\ngenetic heterogeneity represents a challenge that\nhampers the correct characterization of tumor\nsamples.\n19 The current study reveals how uterine\naspirates are a potentially useful tool to circumvent\nthe problems derived from intra-tumor heterogeneity\nwhen genetically characterizing endometrial cancer.\nWe defined the mutational profile of endometrial\ncancers in paired pre-operative uterine aspirates and\nhysterectomy specimens from patients. The data\nobtained not only confirmed the utility of these\naspirates to detect the mutations in primary tumors,\neven when a pathological diagnosis could not be\nachieved by other means, but importantly, they also\nreflected the high intra-tumor genetic heterogeneity\nfound in endometrial cancers. These results show\nthat the genetic analysis of uterine aspirates provides\ninformation that the pathologist may find useful to\nreduce the rate of false-negative diagnoses. In\nsummary, we show the importance of uterine\naspirates in studying endometrial cancer at the\nmolecular level, supporting the potential of non-\ninvasive biopsies for the diagnosis and characteriza-\ntion of certain tumor types.\nMaterials and methods\nSample Description\nA total of 62 endometrial cancer cases (44 endome-\ntrioid endometrial carcinomas, 9 serous endometrial\ncarcinomas, 9 carcinosarcomas) were collected at Vall\nd’Hebron Hospital (Barcelona), Arnau de Vilanova\nUniversity Hospital (Lleida), MD Anderson Cancer\nCenter (Madrid) and Medical University (Lubin)\nbetween 2010 and 2015. The median age of the\npatients was 67 (±12, endometrioid endometrial carci-\nnomas), 75 (±8, serous endometrial carcinomas), and\n72 (±8, carcinosarcomas) and the histopathological\ndata of the tumors studied can be found in Supple-\nmentary Table 1. Endometrial tissue from endometrial\naspirates and hysterectomy specimens were analyzed\nfrom each subject. In addition, samples obtained from\n10 patients diagnosed with atypical hyperplasia, were\ncollected at Hospital Universitari de Bellvitge and\nused as an example of precursor malignant neoplasia.\nA total of 27 patients not diagnosed with cancer were\nalso analyzed as controls for the studies of the\nmutational profile (7 non-atypical hyperplasia endo-\nmetrium, the endometrium from 7 patients with\nleiomyoma, and 13 normal endometrium). In 11 of\nthese controls uterine aspirates and their respective\nhysterectomy specimen (endometrial tissue) were\nanalyzed, whereas in the remainder only a uterine\naspirate was available. Uterine aspirates were col-\nlected using a Pipelle de Cornier to obtain the sample\nthat was then centrifuged for 20 min, as described\npreviously.\n16 The pellet containing the cells from the\nuterine cavity was processed as formalin-fixed and\nparaffin-embedded tissue for further DNA extraction.\nCancer diagnosis of uterine aspirates\nA Mota et al 135\nModern Pathology (2017) 30, 134 –145\n\nA second uterine aspirate was obtained in the\noperating room just before surgery, being the tumor\nmaterial frozen at − 80 °C for hematoxylin and eosin\nstain examination and DNA extraction. Only in which\nthe formalin-fixed and paraffin-embedded uterine\naspirate material was used up in the histologic ana-\nlysis, frozen tissue was used for the study. The study\nwas approved by the local ethical committee from\neach institution, and a complete written informed\nconsent was obtained from all patients.\nDNA Extraction and Mutational Analysis\nDNA was obtained from formalin-fixed paraffin-\nembedded and frozen samples using phenol extrac-\ntion and ethanol precipitation, and 10 ng were used\nfor sequencing. Multiplex PCR to prepare amplicon\nlibraries was performed using the Ion AmpliSeq\nLibrary Kit 2.0 and Ion AmpliSeq Cancer Hotspot\nPanel v2 (Life Technologies). For PCR, a total of 17\nand 20 cycles were used for the frozen and formalin-\nfixed paraffin-embedded samples, respectively. The\nPCR template preparation and enrichment were\nperformed using Ion PGM Template OT2 200 Kit\nand the Ion OneTouch 2 System. Finally, the Ion\nPGM Sequencing 200 Kit v2 and Ion PGM System\n(Life Technologies) were used for DNA sequencing\naccording to the manufacturer ’s protocols. Dupli-\ncates were analyzed for 10 % of the samples,\nrendering equivalent results. For the bioinformatics\nanalysis, see Supplementary Methods.\nSanger Sequencing\nTo validate the mutations, a total of 88 of the 476\nvariants found in the samples analyzed were Sanger\nsequenced. The PCR conditions and amplicon lengths\nused are indicated in Supplementary Table 2. Only 6\nof these variants were not confirmed by Sanger\nsequencing, which was probably due to their poor\nquality and/or their frequencies below 10% in the Ion\nPGM sequencing analysis (Supplementary Table 3).\nStatistical Analysis\nA paired t-test was used to compare the data from the\nhysterectomy tumor samples and uterine aspirates.\nTwo-tailed tests were performed and 95% confi-\ndence intervals (CIs) were accepted. The mutation\ndiscovery rate was calculated in each sample\n(aspirate or tumor region) from the same patient\naccording to the following equation:\nPSample mutation\nX\nð\nX\nAspirate mutation þ\nX\nTumor region 1 mutation\nþ y þ\nX\nTumor region n mutationÞ\n´ 100\nThe Pearson coefficient was used to analyze the\ncorrelation between the percentage of tumor cells in\npatient samples and the MDR. P values o 0.05 were\nconsidered statistically significant and the statistical\nanalyses were performed using the SPSS Statistics\n17.0 software (SPSS, Chicago, IL, USA).\nResults\nIdentification of the Mutational Profile in Paired\nUterine Aspirate and Hysterectomy Specimen Samples\nUterine aspirates are thought to be highly sensitive\nand specific biopsies for the pre-operative diagnosis\nof endometrial cancer, especially when based on\nbiomarker expression. 13,16–18 To investigate the\nusefulness of mutation detection in uterine aspirates,\nthe molecular profile of paired samples (pre-opera-\ntive uterine aspirates and the corresponding resected\nsurgical specimen) from 54 patients with endome-\ntrial cancer (37 endometrioid endometrial carcino-\nmas, 9 serous endometrial carcinomas, and 8\ncarcinosarcomas) and 10 patients with atypical\nhyperplasia was analyzed using AmpliSeq Cancer\nHotspot Panel v2. This panel analyzes approximately\n2800 cancer mutations of 50 oncogenes and tumor\nsuppressor genes, some of which are frequently\naltered in endometrial cancer ( PTEN, KRAS, FGFR2,\nCTNNB1, PIK3CA, FBXW7, and TP53). In addition, a\ntotal of 27 patients not diagnosed with cancer were\nalso analyzed as control cases (7 cases of non-\natypical hyperplasia, 7 cases with leiomyomas and\n13 with a normal endometrium).\nSequencing analysis revealed the presence of\nmutations in 51 of the 54 aspirates from cancer\npatients (Table 1; Supplementary Table 3A) and in 5\nof the 10 aspirates from atypical hyperplasia cases\n(Supplementary Table 4). By contrast, mutations\nwere only identified in 1 of the 27 control patients\n(data not shown). Although it is currently unfeasible\nto reach a diagnosis of endometrial malignancies\nbased exclusively on genetic alterations, these\nresults indicate that genetic analysis of uterine\naspirates may offer reliable support to histological\ndiagnosis.\nMutations identified in the different subgroups of\npatients were consistent with previous studies.\n6,7 In\nsummary, endometrioid endometrial carcinomas\ncarried mutations in PTEN (71.1% of patients),\nPIK3CA (39.5%), CTNNB1 (28.9%), TP53 (28.9%),\nFGFR2 (23.7%), KRAS (21.1%), and CDKN2A\n(10.5%). In addition, we also detected mutations in\nless commonly affected genes, such as: ABL1, AKT1,\nAPC, ATM, BRAF, ERBB2, FBXW7, KIT, RB1, and\nVHL1 (5.3%); and GNA11, GNAS, HNF1A, MET,\nMLH1, NRAS, RET, STK11, SMAD4, SMARCB1, and\nSMO (2.6%). As expected, the most frequently\nmutated gene in serous carcinomas and carcin-\nosarcomas samples was TP53 (77.7 and 87.5%,\nrespectively). Frequencies found in our series were\ngenerally higher than those detected in the The\nCancer Genome Atlas dataset\n6 (Supplementary\nFigure 1A). This could be explained taking into\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\n136 A Mota et al\n\nTable 1 Summary of the histological grade and mutational profile in endometrial cancers and their paired uterine aspirates\nPatient\nAspirate\ngrade\nHysterectomy\ngrade\nCommon variants\n(aspirate and\nhysterectomy)\na\nHysterectomy\nvariants (not\ndetected in aspirate)\nb\nAspirate variants\n(not detected in\nhysterectomy)\n% Hysterectomy\nvariants detected\nin aspirate\n(a/(a+b))\nEEC-1 2 3 PTEN (2), TP53 (2), APC 100\nEEC-2 3 3 PTEN, CTNNB1,\nCDKN2A\n100\nEEC-3 3 3 PTEN, PIK3CA (2), ABL1 100\nEEC-4 1 1 PTEN, KRAS PIK3CA (3) 100\nEEC-5 1 1 PTEN, APC 100\nEEC-6 1 3 FGFR2, PIK3CA, KIT PIK3CA 100\nEEC-7 1 3 FGFR2, PTEN (2), TP53 PTEN, TP53,\nSMARCB1, CTNNB1,\nCKN2A\n44.4\nEEC-8 1 3 PIK3CA, CTNNB1 100\nEEC-9 2 3 FGFR2, PTEN, PIK3CA,\nCTNNB1\n100\nEEC-10 3 3 PIK3CA 100\nEEC-11 1 1 FGFR2, FBXW7 100\nEEC-12 2 2 PTEN (3), KRAS, RB1,\nERBB2, TP53, PIK3CA,\nCTNNB1, FBXW7\n100\nEEC-13 1 3 NRAS, PTEN (2), ATM,\nHNF1A, PIK3CA, SMO,\nABL1, CDKN2A\nKRAS, GNA11 (2) PTEN, ATM,\nTP53 (2),\nSMAD4, GNAS,\nCTNNB1\n75\nEEC-14 2 3 PTEN (2), ERBB2 100\nEEC-15 1 1 PTEN RET, STK11, PIK3CA VHL 25\nEEC-16 2 3 PTEN(3), TP53 100\nEEC-17 2 3 KRAS, AKT1 100\nEEC-18 1 1 PTEN, PIK3CA 100\nEEC-19 2 3 FGFR2, PTEN (3),\nKRAS\n0\nEEC-20 2 3 KRAS 100\nEEC-21 3 1 TP53, PIK3CA —\nEEC-22 2 3 PTEN (2) 100\nEEC-23 1 3 FGFR2 100\nEEC-24 1 3 PTEN, PIK3CA CTNNB1,\nCDKN2A\n100\nEEC-25 2 3 TP53 100\nEEC-26 2 3 KRAS, TP53 100\nEEC-27 1 3 FGFR2, MLH1 PTEN 66.7\nEEC-28 2 2 PTEN (2), CTNNB1 100\nEEC-29 3 2 PTEN 100\nEEC-30 3 3 PTEN, TP53, PIK3CA 100\nEEC-31 2 2 PTEN 100\nEEC-32 2 2 PTEN, CTNNB1 100\nEEC-33 1 1 FGFR2, AKT1, CTNNB1 100\nEEC-34 2 2 FGFR2, PTEN 100\nEEC-35 1 3 PTEN (3), KRAS,\nPIK3CA\nATM, RB1, TP53,\nMET\nRB1, KIT 55.6\nEEC-36 1 3 PTEN, BRAF 100\nEEC-37 1 3 PTEN, PIK3CA (2) 100\nSEC-1 3 3 TP53 100\nSEC-2 3 3 TP53 100\nSEC-3 3 3 KRAS 100\nSEC-4 3 3 ABL1 TP53 0\nSEC-5 3 3 TP53, PIK3CA, BRAF,\nATM\n100\nSEC-6 3 3 TP53 0\nSEC-7 3 3 PIK3CA 100\nSEC-8 3 3 TP53 (2), FBXW7 0\nSEC-9 3 3 TP53 100\nCS-1 3 3 PTEN, TP53, PIK3CA FBXW7 KRAS 75\nCS-2 3 3 TP53 100\nCS-3 3 3 TP53 100\nCS-4 3 3 TP53 100\nCS-5 3 3 TP53 100\nCS-6 3 3 PTEN, KRAS, PIK3CA 100\nCS-7 3 3 KRAS, TP53 IDH2, TP53 (3),\nEGFR\n28.6\nCS-8 3 3 TP53 100\naNumber of common mutations (aspirate and hysterectomy).\nbNumber of hysterectomy variants (not detected in aspirate).\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\nA Mota et al 137\n\naccount the sequencing method applied in each\nstudy. Whereas The Cancer Genome Atlas study 6\nperformed whole-exome sequencing (mean coverage\naround to 50 × ) our study has been developed with\ntargeted sequencing (mean coverage around to\n1000 × ), allowing to detect more accurately the\nmutations, specially those with low frequency.\n20\nTP53 mutation frequency was particularly high in\nour series, probably due to the presence of mutations\nin 8 of the 24 high-grade endometrioid carcinomas\n(Supplementary Figure 1B). To be sure that these\ncases were not misclassified a second pathology\nreview was performed, confirming the initial diag-\nnosis (Supplementary Table 5).\nTo gain further insight into the suitability of\nuterine aspirates to detect mutations and conse-\nquently, to estimate the potential of uterine aspirates\nto characterize endometrial caner from a genetic\npoint of view, we analyzed the percentage of\npathogenic variants present in hysterectomy speci-\nmens that were also detected in aspirates (Figure 1).\nAll the mutations detected in the surgical tumor\ntissue were also found in 30 out of the 36 aspirates\n(83.3%) from endometrioid endometrial carcinoma\npatients. In terms of the rest of the samples, 50 –75%\nof the mutations detected in the hysterectomy\nspecimen also appeared in the corresponding aspi-\nrate in three of them (8.3% of the total), although in\ntwo of them (5.6% of the total) the aspirate contained\n25–50% of the mutations present in the surgical\ntissue. Only in 1 patient did we fail to detect any of\nthe mutations identified in the hysterectomy sample\nin the corresponding aspirate, accounting for 2.8% of\nthe total cases. Conversely, in 1 other patient\nmutations were detected in the uterine aspirate,\nwhereas none were identified in the surgical sample\n(Table 1). Furthermore, in 6 of the 9 uterine aspirates\nfrom serous carcinomas patients 100% of the muta-\ntions identified in the hysterectomy specimens were\ndetected in the aspirate (66.7% of the total cases), as\nin 6 of the 8 carcinosarcoma cases (75% of the total).\nMisclassifying the histological grade of pre-oper-\native biopsies can have grave consequences\n14,15,18\nand indeed, in our samples 22 of the 37 endome-\ntrioid endometrial carcinomas uterine aspirates\n(59.5%) were misclassified with respect to their\ngrade during the pathological diagnosis, the majority\nof them being attributed with a lower grade than that\ndetected in the definitive hysterectomy specimen\n(Table 1). However, in 17 of these 22 (77.2%)\ndiscordant classifications, the uterine aspirates were\nconcordant in the mutational analysis, showing all\nthe mutations detected in their respective surgical\nspecimen. Nevertheless, no relationship between\nmutational status and histological type or grade has\nbeen previously described, and nor was one found in\nour series. Consequently, these results confirm that\nthe genetic analysis of uterine aspirates as a pre-\noperative biopsy can reliably reproduce the mole-\ncular status of the tumor in a pre-clinical setting.\nHowever, further studies into the mutational profile\nand histological grade will be necessary to take the\nmutational information from uterine aspirates into\naccount when assessing the tumor grade.\nGenetic Analysis Helps to Reduce the Rate of\nFalse-Negative Diagnoses in Uterine Aspirates\nThe histologic analyses of uterine aspirates fail to\ndistinguish the presence or absence of malignancy in\naround 13% of the cases, either due to the small\nproportion of representative tumor cells or to the\npoor quality of the specimen.\n21–23 To further inves-\ntigate the potential of genetic analysis of uterine\naspirates as an informative tool for endometrial\ncancer diagnosis, we assessed the tumor mutations\nthat could be detected in uterine aspirates that could\nnot be evaluated on a pathological basis (Figure 2a).\nMutational analysis was performed on eight paired\nsamples of non-diagnosable uterine aspirates from\npatients who turned out to have endometrial cancer\nand on the corresponding hysterectomy specimens\n(7 endometrioid endometrial carcinomas and 1\ncarcinosarcoma). Interestingly, seven of the eight\nnon-evaluable uterine aspirates had a similar muta-\ntion profile to that of their paired surgical sample\n(Figure 2b; Supplementary Table 3B). We did not\nfind mutations in the uterine aspirate from one\npatient, as was also the case in the paired hyster-\nectomy tumor tissue. We validated these results by\nSanger sequencing and immunohistochemistry\nwhen material was available. For example, the\nCTNNB1 mutation in case EEC-38 was validated by\nSanger sequencing in the aspirate and surgical tissue\n(Figure 2c). To validate this, we also analyzed\nβ-catenin expression by immunohistochemistry in\nsurgical tissue (Figure 2d). These results demon-\nstrated that genetic sequencing complements patho-\nlogical analysis and contributes significantly to a\nmore comprehensive characterization of the tumor at\nvery early stages of diagnosis, providing valuable\ninformation for its correct classification.\nGenetic Analysis of Uterine Aspirates Captures the\nIntra-Tumor Heterogeneity Found in Endometrial\nCarcinomas\nIt is well known that human cancers display\nsubstantial intra-tumor heterogeneity, not only in\ncellular morphology or gene expression but also in\nterms of genetic variation. 24,25 This phenomenon\nrepresents an important challenge to accurate diag-\nnosis and therapeutic decision-making. 19 Although\nrecent studies showed intra-tumor genetic hetero-\ngeneity in gynecological cancers like ovarian\ncancer,26,27 heterogeneity at the mutational level\nhas not been described in endometrial cancer to our\nknowledge. Interestingly, the comparison between\nthe mutational profile of uterine aspirates and\nhysterectomy specimens highlighted the presence\nof additional mutations in 11 out of the 54 uterine\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\n138 A Mota et al\n\naspirates, mutations that were not present in the\ncorresponding paired surgical tissue (Table 1). Thus,\nwe examined whether these differences might reflect\nthe intra-tumor genetic heterogeneity in this clinical\ncontext.\nTo explore this hypothesis, genetic sequencing\nanalysis was performed on additional tumor regions\nfrom 21 of the endometrial cancer hysterectomy\nspecimens previously studied (14 endometrioid\nendometrial carcinomas, 5 serous endometrial carci-\nnomas and 5 carcinosarcomas: Table 2). Comparative\nmutation analysis revealed differences in the muta-\ntional profiles of the distinct regions of the endome-\ntrioid endometrial carcinomas tumor tissue analyzed\nfrom 10 out of 14 patients (71.4%), confirming the\npresence of intra-tumor heterogeneity (Supple-\nmentary Figure 2; Supplementary Table 3C). For\nexample, in the three different tumor regions\nanalyzed from case EEC-1 (Figure 3a), a total of 5\nmutations in PTEN, TP53, and APC were detected,\nwith one of the regions (tumor region 1) carrying all\nfive, whereas the other two (tumor region 2 and 3)\ncarried 2 and 3 mutations, respectively. In the\nremaining cases (4/14, 28.6%), a similar mutational\nprofile was seen in all the samples analyzed\n(Supplementary Figure 2; Supplementary Table\n3C), suggesting that these cases did not harbor\nsignificant intra-tumor heterogeneity, at least with\nrespect to the genes and tumor regions studied. For\nexample, this was the case of patient EEC-11 from\nwhom all the samples analyzed carried mutations in\nFGFR2 and FBXW7 (Figure 3b). In contrast to the\nendometrioid endometrial carcinomas, intra-tumor\nheterogeneity was only detected in 1 of the 5 (20%)\nserous carcinomas and 1 of the 5 (20%) carcinosar-\ncomas when additional tumor regions were analyzed\n(Supplementary Figure 3). The low proportion of\nmutational heterogeneity in cases with serous and\ncarcinosarcoma histology could be due to the fact\nthat chromosomal instability is a more frequent\nmolecular alteration than punctual genetic changes\nin these tumor types, 9 a modification that cannot be\nproperly detected with the sequencing platform\nused here.\nThe sensitivity of mutation detection in each\nsample was scored as the mutation discovery rate,\nwhich indicates the proportion of mutations\ndetected in each sample with respect to the total\nmutations observed in all the samples studied from a\ngiven patient (see ‘Materials and methods ’ section).\nThe mutation discovery rate was significantly higher\nin the endometrioid uterine aspirates than in the\nmatched surgical tumor tissue, with a mean of 94.1%\nfor uterine aspirates and 77.2% for individual\nhysterectomy tissue samples. This difference\nincreased when low-quality mutations were not\nconsidered, decreasing the mutation discovery rate\nfor surgical tumor samples to 67.5%, whereas the\nmutation discovery rate of the aspirates remained\nunaltered (Figure 3c). However, no significant\ndifferences were found in the serous carcinoma or\ncarcinosarcoma samples. Differences in the mutation\nFigure 1 Percentage of mutations in hysterectomy specimens identified in paired uterine aspirate. Graphs represents the percentage\n(100%, 75–50%, 50–25%, or 0%) of the mutations found in surgical tumor samples and paired aspirates in endometrioid carcinoma ( a),\nserous carcinoma ( b), and carcinosarcoma ( c) samples.\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\nA Mota et al 139\n\ndiscovery rate are mainly found in heterogeneous\ntumors, due to the differences observed in the\nmutational profile between each tumor region\n(Figure 4). In 8 of the 10 (80%) heterogeneous endo-\nmetrioid tumors, uterine aspirates reflected a higher\nmutation discovery rate than the tumor region used\nfor the pathological diagnosis (tumor region 1). Only\nin one patient (EEC-7) the mutation discovery rate of\nthe uterine aspirate was lower than that for the\ndiagnostic tumor region, although it was equal or\nhigher than that derived from the two other regions\nfrom that patient. In fact, the mutation discovery rate\nvalue was higher in uterine aspirates than in at least\none tumor region in all cases where there was tumor\nheterogeneity. These differences seem not to be\nrelated to the proportion of the tumor tissue in each\nregion analyzed as there was no significant correla-\ntion in a Pearson test (data not shown). These data\nconfirm that genetic analysis of uterine aspirates\ndetects a more representative mutational landscape\nof the tumor, reproducing in a single sample the\nintra-tumor heterogeneity found in the different\ntumor regions.\nDiscussion\nAdvances in next-generation sequencing have\nrevealed that genetic heterogeneity must be taken\ninto account to fully understand tumor\nbiology.19,28,29 Indeed, over and above the inter-\npatient heterogeneity, 30 intra-tumor heterogeneity\nrepresents a real challenge for the precise character-\nization and adequate management of tumors. 19,31\nThe presence of different cell populations within a\ntumor with specific genomic, genetic and/or epige-\nnetic characteristics has been demonstrated in\nnumerous tumor types, including solid tumors and\nhematologic malignancies. 19 Indeed, intra-tumor\nheterogeneity has been observed among gynecologi-\ncal cancers, particularly in high-grade serous ovarian\ncarcinomas,27,32–34 although this issue has not been\nstudied in endometrial cancer so far. Therefore, a\nFigure 2 Genetic analysis of non-evaluable uterine aspirates. Paired samples of non-evaluable uterine aspirates and hysterectomy\nspecimens were analyzed genetically. ( a) Representative hematoxylin and eosin image of a uterine aspirate (upper image) and its paired\nsurgical sample (lower image). ( b) Summary of the mutations detected in the paired uterine aspirate and hysterectomy samples. Analysis\nof CTNNB1 (β-catenin) mutation (S37P) found in patient EEC-38 by ( c) the Sanger sequencing in hysterectomy specimen and uterine\naspirate samples and by ( d) immunohistochemistry analysis. The white arrow label the nuclear localization of β-catenin, which is\nsuggestive of mutations (panel magnification × 20).\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\n140 A Mota et al\n\nbetter understanding of the genetic heterogeneity\nunderlying the biological and phenotypic evolution\nof endometrial is crucial to understand the clinical\nbehavior of this disease. In this sense, the majority of\nthe endometrioid carcinomas analyzed here have\nvariable mutational profiles in the different tumor\nregions. By contrast, only 20% of serous carcinomas\nand 20% of carcinosarcomas showed mutational\nheterogeneity, which perhaps reflects the more\nfrequent genetic mutations in endometrioid than in\nserous and carcinosacomas,\n6 the latter more often\ndisplaying large genomic changes. 9 Therefore, a\ngenomic study should be carried out on these tumor\ntypes to define the implication of copy number\nvariation in intra-tumor heterogeneity, as previously\ndescribed in high-grade serous ovarian\ncarcinomas.\n27,33\nIntra-tumor clonal heterogeneity is thought to\ninflu;ence therapeutic resistance and tumor\nprogression,35 with some studies suggesting that\nsome clones are genetically predisposed to resist\ntherapy.36 In this context, characterizing intra-tumor\nheterogeneity would seem to be necessary to better\npredict the clinical outcome of a specific tumor at\nthe moment of diagnosis and to establish the most\nappropriate treatment. The standard treatment for\nendometrial cancer is well established, involving\nsurgery followed by adjuvant radiotherapy in tumors\nwith a high-risk of recurrence. Chemotherapy is\nusually restricted to metastatic/recurrent and high-\ngrade endometrial cancers, although traditional\nchemotherapy regimens are less effective than in\ncancers of other organs.\n5 In this sense, numerous\nclinical trials have been stratified according to\ngenetic features, based on PTEN, PIK3CA ,o r FGFR3\nmutational status. Consequently, tumor heterogene-\nity represents a therapeutic challenge and the use of\na single diagnostic biopsy of a tumor may be\ninsufficient, leading to the misclassification of a\nsignificant proportion of patients.\nSeveral studies have centered on the feasibility of\nusing liquid biopsies to analyze intra-tumor genetic\nheterogeneity.\n37–40 In endometrial cancer, uterine\naspirates are used as minimally invasive and highly\nsensitive biopsies for histological diagnosis or\nmolecular characterization.\n16–18 In this regard, we\nfound that the genetic analysis of uterine aspirates\ncoupled to their pathological classification could be\na very sensitive approach to detect endometrial\nmalignant neoplasia. This implies that detecting a\ncancer-related mutation (such as those detected by\nthe method we employed) is related to a possible\nmalignant disorder or tumor. Although this seems to\nbe true in our series it remains controversial, and a\nsignificantly larger number of samples (both normal\nand malignant) should be analyzed to address this\nissue. Paired sequencing of uterine aspirates and\nhysterectomy specimens confirms the efficacy in\nrevealing malignant disorders (endometrial tumors\nor atypical hyperplasia) in uterine aspirates. Only\nthree samples (5.5%) of uterine aspirates from tumor\ncases did not show any of the surgical tumor sample\nmutations, whereas a total of 42 (77.8%) of the\naspirates carried all the mutations found in the\ncorresponding hysterectomy specimen.\nFurthermore, we detected mutations in aspirates\nthat could not be evaluated pathologically. The\namount of tissue obtained from endometrial biopsies\nfrom postmenopausal patients is sometimes insuffi-\ncient to obtain an adequate diagnosis, which in the\nmajority of cases is due to the presence of endome-\ntrial atrophy. However, patients with endometrial\ncancer on occasions provided poor quality samples.\nIn a recent study of 1120 endometrial samples\nclassified as unsuitable for diagnosis, a second\nbiopsy was obtained from 38% of the patients that\nwas suitable for diagnosis in 75% of cases, with 10%\nhaving a malignant tumor.\n23 Our results show that\nmutation analysis could indicate the presence of\nendometrial cancer or at least some pre-malignant\nanomaly, emphasizing the need for resampling in\nsuch cases and providing valuable information to\naccelerate the diagnosis.\nGenetic analysis of uterine aspirates captures the\nintra-tumor heterogeneity identified in endometrioid\nendometrial carcinomas. The mutation discovery\nrate, defined as the percentage of mutations detected\nin each individual sample with respect to all the\nmutations found in a given patient, was used to\nmeasure the sensitivity of mutation detection in each\nsample. In heterogeneous tumors, the uterine\nTable 2 Endometrial cancer cases studied in the intra-tumor\nheterogeneity analysis\nPatient\nTumor\nregions\nanalyzed\nTotal\nvariants\ndetected\nin tumor\nregions\nTotal\nvariants\ndetected in\nuterine\naspirates\nIntra-tumor\nheterogeneity\nEEC-1 3 5 5 Yes\nEEC-2 2 3 3 Yes\nEEC-3 3 4 4 Yes\nEEC-4 4 2 5 Yes\nEEC-5 3 2 2 Yes\nEEC-6 3 3 4 Yes\nEEC-7 4 9 4 Yes\nEEC-8 4 2 2 No\nEEC-9 4 4 4 Yes\nEEC-10 4 1 1 No\nEEC-11 4 2 2 No\nEEC-12 4 10 9 Yes\nEEC-13 3 12 16 Yes\nEEC-14 4 3 3 No\nSEC-1 3 1 1 No\nSEC-2 3 1 1 No\nSEC-3 3 1 1 No\nSEC-4 3 1 1 Yes\nSEC-5 3 4 4 No\nCS-1 4 4 4 Yes\nCS-2 3 1 1 No\nCS-3 3 1 1 No\nCS-4 3 1 1 No\nCS-5 2 1 1 No\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\nA Mota et al 141\n\naspirate mutation discovery rate was higher than that\nin at least one of the tumor regions. In fact, the\nmutation discovery rate value was higher in the\nuterine aspirate than in the tumor regions used for\npathological diagnosis (tumor region 1) in 8 of the 10\nheterogeneous endometrioid carcinomas. These\nresults highlight the potential utility of this type of\nbiopsy and reveal that the use of a unique tumor\nsample in diagnosis could underestimate the muta-\ntional burden in heterogeneous tumors. However,\nthe study of multiple samples of a given tumor as a\nroutine practice is still a difficult issue, as it would\nincrease significantly the time and cost of diagnosis.\nMoreover, combining DNA from different tumor\nsamples previously to the targeted sequencing is\nnot a good option, because it would lead to a\ndecrease in the frequency of those mutations, which\nare not present in all the tumor regions, causing\nsome low-requency variants to be undetected. It is\nalso worth pointing that it is fairly difficult to\ncalculate how many tumor regions need to be\nanalyzed to cover the intra-tumor heterogeneity\nfound in each case. Taken together, these arguments\nincrease the value of uterine aspirates as a genetic\nFigure 3 Characterization of the intra-tumor genetic heterogeneity in endometrial tumors. Representative mutational profile of genetically\nheterogeneous endometrioid carcinoma ( a, EEC-1) and of a homogeneous endometrioid tumor ( b, EEC-11). The colors in the squares\nrepresent the mutant allelic frequencies (MAFs). The squares marked as LQ identify low quality variants in the ion PGM analysis. The\nmutation discovery rate is defined as the percentage of mutations detected in each sample with respect to the totality of the mutations\nobserved in all the samples analyzed from the same patient (see ‘Materials and methods’ section). The graph represents the mean mutation\ndiscovery rate ( c) in endometrioid carcinomas, serous carcinomas, and carcinosarcomas. (*0.005 oP o0.05; **0.001 oP o0.005).\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\n142 A Mota et al\n\ndiagnostic biopsy, solving, at least in part, some of\nthe problems found in the study of hysterectomy\nspecimens. The fact that intra-tumor heterogeneity\nmay be represented in uterine aspirates is probably\nrelated to the nature of such samples, consisting of\ncells from many different parts of the uterine cavity,\nwhich could provide a more representative picture\nof the entire tumor specimen than samples from a\nspecific tumor region. Similar results were observed\nin ovarian carcinomas where intra-tumor genetic\nheterogeneity was evident when solid tumor biop-\nsies were compared, 32–34 but not when different\nascites from the same patient were compared. 41 In\nthis case, ascites could represent the entire cavity in\na similar way that uterine aspirates do in uterine\ncancers, capturing all the genetic mutations and\nrepresenting the heterogeneity found in the solid\ntumor biopsies.\nThe use of non-invasive biopsies to diagnose and\ncharacterize tumors is currently a relevant clinical\nchallenge. The data presented here shed light on the\nmolecular characterization of minimally-invasive\nbiopsies in endometrial cancer, and they provide\npotential solutions to the problem of detecting\ngenetic heterogeneity, as well as valuable informa-\ntion in the case of biopsies with insufficient material.\nThese data pave the way for the use of such analyses\nfor other diseases.\nFigure 4 Mutation discovery rate in heterogeneous endometrioid endometrial carcinoma. The mutation discovery rate was calculated for\neach sample from the heterogeneous endometrioid carcinoma patients as indicated in the ‘Materials and methods ’ section. Each bar\nrepresents a sample, from the bottom to the top: uterine aspirate and the different tumor regions. 1–4 The dark gray color represents the\npercentage of high quality variants detected and the light gray reflects the LQ variants identified in the ion PGM analysis.\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\nA Mota et al 143\n\nAcknowledgments\nWe thank all those at the Translational Research\nLaboratory and Immunohistochemical Laboratory\nfrom MD Anderson Madrid for their invaluable help.\nTissue samples were obtained with the support of\nMD Anderson Foundation Biobank (Record Number\nB.0000745, ISCIII National Biobank Record), the\n‘Xarxa Catalana de Bancs de Tumors ’ and ‘Plata-\nforma de Biobancos ’ ISCIII (PT13/0010/0014,\nB.000609). This work was supported by grants from\nthe AECC (Grupos Estables de Investigacion 2011-\nAECC- GCB 110333 REVE), the ‘Fundació La Marató,\nTV3’ (2/C/2013) to AG-M, JR, XM-G, and GM-B;\nInstituto de Salud Carlos III (ISCIII) (PI13/00132 and\nRETIC-RD12/0036/0007 to GM-B; RETIC-RD12-\n/0036/0035 to JR; PI13/01701, and RD12/0036/0013\nto XM-G; PI14/02043 and PI14/01942 to AG and\nMA); the ‘CIRIT, Generalitat de Catalunya ’ (2014\nSGR 1330 to JR; and 2014 SGR 138 to XM-G), GEIS\naward 2013 to GM-B and PG-S, and the ‘Commu-\nnidad de Madrid’ (S2010/BMD-2303) to GM-B. AM is\nfunded by the Spanish Ministry of Education,\nCulture, and Sports (FPU2012-5338). IC and PG-S\nare funded by PhD and postdoctoral contracts,\nrespectively, from the AECC Scientific Foundation.\nEC is funded by the Spanish Ministry of Economy\nand Competitiveness (FPDI-2013-18322).\nAuthor contributions\nAM and PG-S performed the sequencing experiment\nand analysis. AM, EC, PG-S, and IC contributed to\nthe sample processing. AR-S, SG, BD-F, AV, and AG\nperformed the pathological analysis of the samples.\nLC, SA, AG-M, XG-T, PZ-M, and MB helped obtain\nthe samples. MA and RL-L read and corrected the\nmanuscript. EC, XM-G, JR, and GM-B conceived the\nstudy, participated in its design, and helped draft the\nmanuscript. GM-B discussed and directed the study.\nAll the authors read and approved the final\nmanuscript.\nDisclosure/conflict of interest\nThe authors declare no conflict of interest.\nReferences\n1 Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer\nincidence and mortality worldwide: sources, methods\nand major patterns in GLOBOCAN 2012. Int J Cancer\n2015;136:E359–E386.\n2 Bokhman JV. Two pathogenetic types of endometrial\ncarcinoma. Gynecol Oncol 1983;15:10 –17.\n3 Lax SF, Kurman RJ. A dualistic model for endometrial\ncarcinogenesis based on immunohistochemical and\nmolecular genetic analyses. Verh Dtsch Ges Pathol\n1997;81:228–232.\n4 Llaurado M, Ruiz A, Majem B, et al. Molecular bases of\nendometrial cancer: new roles for new actors in the\ndiagnosis and the therapy of the disease. Mol Cell\nEndocrinol 2012;358:244 –255.\n5 Yeramian A, Moreno-Bueno G, Dolcet X, et al.\nEndometrial carcinoma: molecular alterations involved\nin tumor development and progression. Oncogene\n2013;32:403–413.\n6 Cancer Genome Atlas Research Network, Kandoth C,\nSchultz N, et al. Integrated genomic characterization of\nendometrial carcinoma. Nature 2013;497:67 –73.\n7 Hong B, Le Gallo M, Bell DW. The mutational land-\nscape of endometrial cancer. Curr Opin Genet Dev\n2015;30C:25–31.\n8 Zighelboim I, Goodfellow PJ, Gao F, et al. Microsatellite\ninstability and epigenetic inactivation of MLH1 and\noutcome of patients with endometrial carcinomas of the\nendometrioid type. J Clin Oncol 2007;25:2042 –2048.\n9 Tritz D, Pieretti M, Turner S, et al. Loss of hetero-\nzygosity in usual and special variant carcinomas of the\nendometrium. Hum Pathol 1997;28:607 –612.\n10 Cantrell LA, Blank SV, Duska LR. Uterine carcinosar-\ncoma: a review of the literature. Gynecol Oncol\n2015;137:581–588.\n11 Artioli G, Wabersich J, Ludwig K, et al. Rare uterine\ncancer: carcinosarcomas. Review from histology to\ntreatment. Crit Rev Oncol Hematol 2015;94:98 –104.\n12 Colombo N, Preti E, Landoni F, et al. Endometrial\ncancer: ESMO Clinical Practice Guidelines for diag-\nnosis, treatment and follow-up. Ann Oncol 2013;24:\nvi33–vi38.\n13 Clark TJ, Mann CH, Shah N, et al. Accuracy of\noutpatient endometrial biopsy in the diagnosis of\nendometrial cancer: a systematic quantitative review.\nBJOG 2002;109:313 –321.\n14 Karateke A, Tug N, Cam C, et al. Discrepancy of pre-\nand postoperative grades of patients with endometrial\ncarcinoma. Eur J Gynaecol Oncol 2011;32:283 –285.\n15 Wang XY, Pan ZM, Chen XD, et al. Accuracy of tumor\ngrade by preoperative curettage and associated clin-\nicopathologic factors in clinical stage I endometriod\nadenocarcinoma. Chin Med J 2009;122:1843 –1846.\n16 Colas E, Perez C, Cabrera S, et al. Molecular markers of\nendometrial carcinoma detected in uterine aspirates.\nInt J Cancer 2011;129:2435 –2444.\n17 Perez-Sanchez C, Colas E, Cabrera S, et al. Molecular\ndiagnosis of endometrial cancer from uterine aspirates.\nInt J Cancer 2013;133:2383 –2391.\n18 Stelloo E, Nout RA, Naves LC, et al. High concordance\nof molecular tumor alterations between pre-operative\ncurettage and hysterectomy specimens in patients with\nendometrial carcinoma. Gynecol Oncol 2014;133:\n197–204.\n19 Jamal-Hanjani M, Quezada SA, Larkin J, et al. Transla-\ntional implications of tumor heterogeneity. Clin Cancer\nRes 2015;21:1258 –1266.\n20 Sims D, Sudbery I, Ilott NE, et al. Sequencing depth and\ncoverage: key considerations in genomic analyses. Nat\nRev Genet 2014;15:121 –132.\n21 McCluggage WG. My approach to the interpretation of\nendometrial biopsies and curettings. J Clin Pathol\n2006;59:801–812.\n22 Phillips V, McCluggage WG. Results of a questionnaire\nregarding criteria for adequacy of endometrial biopsies.\nJ Clin Pathol 2005;58:417 –419.\n23 Kandil D, Yang X, Stockl T, et al. Clinical outcomes of\npatients with insufficient sample from endometrial\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\n144 A Mota et al\n\nbiopsy or curettage. Int J Gynecol Pathol 2014;33:\n500–506.\n24 Durrett R, Foo J, Leder K, et al. Intratumor hetero-\ngeneity in evolutionary models of tumor progression.\nGenetics 2011;188:461 –477.\n25 Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor\nheterogeneity and branched evolution revealed by\nmultiregion sequencing. N Engl J Med 2012;366:883–892.\n26 Khalique L, Ayhan A, Whittaker JC, et al. The clonal\nevolution of metastases from primary serous epithelial\novarian cancers. Int J Cancer 2009;124:1579 –1586.\n27 Cooke SL, Ng CK, Melnyk N, et al. Genomic analysis of\ngenetic heterogeneity and evolution in high-grade\nserous ovarian carcinoma. Oncogene 2010;29:\n4905–4913.\n28 Lawrence MS, Stojanov P, Mermel CH, et al. Discovery\nand saturation analysis of cancer genes across 21\ntumour types. Nature 2014;505:495 –501.\n29 Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Sig-\nnatures of mutational processes in human cancer.\nNature 2013;500:415 –421.\n30 Vogelstein B, Papadopoulos N, Velculescu VE, et al.\nCancer genome landscapes. Science 2013;339:1546–1558.\n31 Kleppe M, Levine RL. Tumor heterogeneity confounds\nand illuminates: assessing the implications. Nat Med\n2014;20:342–344.\n32 Khalique L, Ayhan A, Weale ME, et al. Genetic intra-\ntumour heterogeneity in epithelial ovarian cancer and\nits implications for molecular diagnosis of tumours.\nJ Pathol 2007;211:286 –295.\n33 Bashashati A, Ha G, Tone A, et al. Distinct evolutionary\ntrajectories of primary high-grade serous ovarian\ncancers revealed through spatial mutational profiling.\nJ Pathol 2013;231:21 –34.\n34 Mota A, Trivino JC, Rojo-Sebastian A, et al. Intra-tumor\nheterogeneity in TP53 null high grade serous ovarian\ncarcinoma progression. BMC Cancer 2015;15:940.\n35 Pribluda A, de la Cruz CC, Jackson EL. Intratumoral\nheterogeneity: from diversity comes resistance. Clin\nCancer Res 2015;21:2916 –2923.\n36 Bhang HE, Ruddy DA, Krishnamurthy Radhakrishna V,\net al. Studying clonal dynamics in response to cancer\ntherapy using high-complexity barcoding. Nat Med\n2015;21:440–448.\n37 Patel KM, Tsui DW. The translational potential of\ncirculating tumour DNA in oncology. Clin Biochem\n2015;48:957–961.\n38 De Mattos-Arruda L, Weigelt B, Cortes J, et al. Captur-\ning intra-tumor genetic heterogeneity by de novo\nmutation profiling of circulating cell-free tumor DNA:\na proof-of-principle. Ann Oncol 2014;25:1729 –1735.\n39 Raimondi C, Nicolazzo C, Gradilone A, et al. Circulat-\ning tumor cells: exploring intratumor heterogeneity of\ncolorectal cancer. Cancer Biol Ther 2014;15:496\n–503.\n40 Broersen LH, van Pelt GW, Tollenaar RA, et al. Clinical\napplication of circulating tumor cells in breast cancer.\nCell Oncol 2014;37:9 –15.\n41 Castellarin M, Milne K, Zeng T, et al. Clonal evolution\nof high-grade serous ovarian carcinoma from primary\nto recurrent disease. J Pathol 2013;229:515 –524.\nSupplementary Information accompanies the paper on Modern Pathology website (http://www.nature.com/\nmodpathol)\nModern Pathology (2017) 30, 134 –145\nCancer diagnosis of uterine aspirates\nA Mota et al 145","source_license":"CC0","license_restricted":false}