​

​

          Endometrial cancer (EC) or carcinoma of corpus uteri is a neoplastic change, mostly adenocarcinoma,  arising from uterine columnar epithelial lining [1, 2]. Abnormal uterine bleeding is the main presenting symptom especially in postmenopausal women [3, 4]. EC is estimated to be the seventh most commonly diagnosed cancer in women [5]. It is considered the 16th leading cause of death in women with cancer worldwide, with 382 000 estimated new cases and 89 900 deaths in 2018 [5]. Factors as obesity, parity, diabetes mellitus, unopposed estrogen exposure, genetics and hormonal therapies are recognizable risks for EC [6, 7]. Other factors as chronic comorbidities, tumor size and organ metastasis influence staging, prognosis, and management protocols [8, 9].  

           FIGO staging has been adopted as the standard classification system in the management of endometrial cancer. However, this staging system does not consider all factors that affect treatment decision and prognosis including but not limited to patient demographics, tumor grade, and lymphovascular space invasion. In addition, some interventions are still debatable particularly in the presence of intermediate disease e.g, grade II early EC. Available evidence supports combined pelvic and para-aortic lymphadenectomy in management of patients with EC [10]. However, combined pelvic and para-aortic lymphadenectomy carry the risk of long term morbidities as lymphedema. Thus, hysterectomy alone as management plan is suggested in patients with low risk EC [11]. More comprehensive studies of the multiple confounders that determine patient’s risk are essential to reach an individually adjusted management plans and to predict prognosis of each individual case [12]. Therefore, availability of large multicenter studies will provide robust evidence regarding optimal management of EC and hence, improve treatment outcome and prognosis particularly in the era of machine learning and artificial intelligence.  

​

​

​

​

​

            This project aims to determining prognostic factors and individualizing management decision per patient characteristics and EC features. The ultimate objective is to create an individualized alternative to the current staging system and identify as many of the factors that affect prognosis and the ideal management of based on these factors.

​

​

​

​

​

           This retrospective study will include at least 10 centers from different countries that present at least Europe, South America, Asia, and Africa. Data will be retrospectively collected from January 2008 to December 2015 with a total follow-up of at least 5 years (December 2020). Each center is anticipated to recruit at least 100 patients to participate in the study. All women who were diagnosed with endometrial cancer at any stage, of all histological types and grades, during this period will be eligible for the study. This study will include all the clinical and demographic information of the eligible patients.

 

Inclusion criteria:

 

• Women diagnosed with endometrial cancer, aged between 2008 and 2015.

• Women should be diagnosed and managed by the corresponding center.

• Patients with adequate clinical and pathological data.

​

Exclusion criteria:

​

• Inadequate information and follow-up for at least 5 years.

• Authorization to use anonymous patient data for research purposes.

​

Data will be collected using an excel spreadsheet designed for this study or an online form (as per center preference). Data include patient age, parity, body mass index, ethnicity, smoking index, contraception method, medical comorbidities [coronary artery disease, diabetes on insulin, hypertension, chronic renal disease, chronic lung disease, thyroid dysfunction], preoperative imaging [tumor size, myometrial involvement (< 50% or > 50%), parametrial invasion, cervical involvement (radiological, clinical), vaginal involvement (upper third, middle third or lower third), iliac lymph nodes (number, largest diameter), paraaortic lymph nodes (number, largest diameter), omental/peritoneal involvement (isolated, disseminated), lung metastasis (single, multiple, largest nodule), pleural effusion, liver metastasis (single, multiple, largest nodule), ascites], staging, histopathology [histological type, grading, lymph vascular space invasion “LVSI”], genetics [positive family history, lynch syndrome], treatment approach [Surgical approach, Hysterectomy BSO, Omentectomy, LNs (sentinel lymph nodes, pelvic lymphadenectomy, para-aortic lymphadenectomy, lymph node sampling), Brachytherapy, External pelvic radiation, Adjuvant chemotherapy], and related outcomes (overall survival, disease free survival, and rate of distant metastases at 1 year, 3 years, and 5 years, pelvic organ functioning, postoperative complications, incidence of lymphedema, tumor-related deaths, frequency of residual disease on definitive surgical specimens ). Data will not include any identifiable information. With the data gathered from this database, the investigators could further demonstrate different aspects of endometrial cancer management and prognosis.

Data will be collected and analyzed using patient demographics, disease features and interventions as inputs. Primary outcome

​

Statistical analysis

 

           Data will be described using (mean, median, standard deviation, range) in the final sample. Machine learning method is superior to traditional statistical methods as it provides robust and automatic estimation of complex relationships between different variables and clinical outcomes. Data will be utilized as Xi and Yi where Xi presents input (features) and Yi presents dependent variables (outcomes). XGboost algorithm will be used to create prediction model and data will be split at 0.8:0.2 to allow model testing. Model Validation will be performed via K-fold cross-validation. Machine learning models will be created using python 3.8®.

 

Institutional review board (IRB) approval

​

The study meets the criteria of accelerated institutional review board (IRB) review as it conveys minimal risk to research participants.

​

​

​

​

​

1. Bell, D.W. and L.H. Ellenson, Molecular Genetics of Endometrial Carcinoma. Annu Rev Pathol, 2019. 14: p. 339-367.

2. Lax, S.F., [New features in the 2014 WHO classification of uterine neoplasms]. Pathologe, 2016. 37(6): p. 500-511.

3. Hamilton, C.A., et al., Endometrial cancer: A society of gynecologic oncology evidence-based review and recommendations. Gynecol Oncol, 2021.

4. Soja, M., et al., Analysis of the results of invasive diagnostic procedures in patients referred to gynecologic department due to abnormal uterine bleeding. Prz Menopauzalny, 2020. 19(4): p. 155-159.

5. Bray, F., et al., Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin, 2018. 68(6): p. 394-424.

6. Ignatov, A. and O. Ortmann, Endocrine Risk Factors of Endometrial Cancer: Polycystic Ovary Syndrome, Oral Contraceptives, Infertility, Tamoxifen. Cancers (Basel), 2020. 12(7).

7.  Raglan, O., et al., Risk factors for endometrial cancer: An umbrella review of the literature. Int J Cancer, 2019. 145(7): p. 1719-1730.

8. Kruse, A.J., et al., Vaginal hysterectomy with or without bilateral salpingo-oophorectomy may be an alternative treatment for endometrial cancer patients with medical co-morbidities precluding standard surgical procedures: a systematic review. Int J Gynecol Cancer, 2019.

9. Murali, R., et al., Evolving Roles of Histologic Evaluation and Molecular/Genomic Profiling in the Management of Endometrial Cancer. J Natl Compr Canc Netw, 2018. 16(2): p. 201-209.

10. Petousis, S., et al., Combined pelvic and para-aortic is superior to only pelvic lymphadenectomy in intermediate and high-risk endometrial cancer: a systematic review and meta-analysis. Arch Gynecol Obstet, 2020. 302(1): p. 249-263.

11. Mariani, A., et al., Low-risk corpus cancer: is lymphadenectomy or radiotherapy necessary? Am J Obstet Gynecol, 2000. 182(6): p. 1506-19.

12. Reijnen, C., et al., Preoperative risk stratification in endometrial cancer (ENDORISK) by a Bayesian network model: A development and validation study. PLoS Med, 2020. 17(5): p. e1003111.