Article Text

Download PDFPDF
Development and validation of an endometrial carcinoma preoperative bayesian network using molecular and clinical biomarkers (ENDORISK): an ENITEC collaboration study
  1. C Reijnen1,2,
  2. E Gogou3,
  3. L van der Putten1,
  4. N Visser4,
  5. K van de Vijver5,
  6. M Santacana6,
  7. J Bulten4,
  8. E Colas7,
  9. A Gil-Moreno8,
  10. A Reques9,
  11. G Mancebo10,
  12. C Krakstad11,
  13. J Trovik12,
  14. I Haldorsen13,
  15. H Engerud11,
  16. J Huvila14,
  17. M Koskas15,
  18. V Weinberger16,
  19. L Minar16,
  20. E Jandakova17,
  21. A van der Wurff18,
  22. X Matias-Guiu6,
  23. F Amant19,
  24. H Küsters-Vandevelde20,
  25. J Ramjith21,
  26. L Massuger1,
  27. M Snijders2,
  28. P Lucas3 and
  29. J Pijnenborg1
  1. 1Obstetrics and Gynaecology, Radboud UMC Nijmegen
  2. 2Obstetrics and Gynaecology, Canisius-Wilhelmina Hospital
  3. 3Department of Computing Sciences, Radboud University
  4. 4Pathology, Radboud UMC Nijmegen, Nijmegen, The Netherlands
  5. 5Pathology, Ghent University Hospital, Ghent, Belgium
  6. 6Department of Pathology and Molecular Genetics and Research Laboratory, Hospital Universitari Arnau de Vilanova, Lleida
  7. 7Biomedical Research Group in Gynecology, Vall Hebron Institute of Research, Universitat Autònoma de Barcelona
  8. 8Gynecological Department
  9. 9Pathology Department, Vall Hebron University Hospital
  10. 10Obstetrics and Gynecology, Hospital del Mar, Barcelona, Spain
  11. 11Centre for Cancer Biomarkers, Department of Clinical Science, University of Bergen
  12. 12Obstetrics and Gynecology
  13. 13Mohn Medical Imaging and Visualization Centre, Department of Radiology, Haukeland University Hospital, Bergen, Norway
  14. 14Pathology, University of Turku, Turku, Finland
  15. 15Obstetrics and Gynecology Department, Bichat-Claude Bernard Hospital, Paris, France
  16. 16Gynecology and Obstetrics, Faculty of Medicine
  17. 17Institute of Pathology, Faculty of Medicine, Masaryk University, Brno, Czech Republic
  18. 18Pathology, Elisabeth-TweeSteden Hospital, Nijmegen
  19. 19Gynaecologic Oncology, Center for Gynaecologic Oncology Amsterdam, Location Antoni van Leeuwenhoek, Amsterdam
  20. 20Pathology, Canisius-Wilhelmina Hospital
  21. 21Radboud University, Department for Health Evidence, Nijmegen, The Netherlands


Introduction/Background The presence of pelvic and/or para-aortic lymph node metastasis (LNM) is one of the most important prognostic factors for poor outcome in endometrial carcinoma (EC). Current risk stratification for lymphadenectomy is mainly based on preoperative tumor grade, results in over- and undertreated of approximately 25% and 15% of the patients. Use of preoperative prediction models allow a personalized risk estimation and contribute to shared decision making, balancing risks and clinical benefit in tailored treatment. The aim of this study is to develop a Bayesian network (BN), based on easily-accessible clinical, histopathological and molecular biomarkers, for the prediction of lymph node metastasis and outcome in endometrial carcinoma patients. Second, the calibration and discrimination this network will be tested by means of external validation.

Methodology This network was constructed within the European Network for Individualized Treatment of Endometrial Cancer (ENITEC), using a cohort including 809 patients treated for EC. The network was based both on expert knowledge of EC progression and learned from data of the construction cohort. Variables used to construct to BN included: preoperative tumor grade; immunohistochemical profile including estrogen receptor-, progesterone receptor-, p53- and L1CAM-expression; cancer antigen 125 serum levels, thrombocyte count, imaging results and cervical cytology. Internal cross-validation and external validation was performed using two independent validation cohorts comprising 431 and 400 patients.

Results A Bayesian network was constructed to predict the presence of lymph node metastasis and 1-, 3- and 5-year disease-specific survival (figure 1). Internal cross-validation showed good discrimination (area under the receiver operator characteristic curve 0.86) and was calibrated well with respect to the prediction of lymph node metastasis. External validation will be completed soon.

Abstract – Figure 1

Bayesian network

Conclusion We have developed and externally validated a Bayesian network predicting lymph node metastasis in endometrial carcinoma using preoperative markers with high diagnostic accuracy.

Disclosure Nothing to disclose.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.