Article Text

Download PDFPDF

2022-LBA-1282-ESGO Identifying women with early-stage cervical cancer at low risk of lymph node metastases, in a large international cohort – A logistic regression analysis, without sharing privacy-sensitive patient data
Free
  1. Hans Wenzel1,2,
  2. Anna Norberg Hardie3,4,
  3. Ruud Bekkers5,6,
  4. Henrik Falconer3,4,
  5. Claus Høgdall7,
  6. Pernille Jensen8,9,
  7. Valery Lemmens1,10,
  8. Frank Martin1,
  9. Arturo Moncada-Torres1,
  10. Hans Nijman2,
  11. Maaike van der Aa1,
  12. Tine Schnack7,11,
  13. y (DANDELION) research group Dutch,
  14. dANish and sweDish gynaEcoLogIcal ONcologog
  1. 1Department of Research and Development, Netherlands Comprehensive Cancer Organisation (IKNL), Utrecht, Netherlands
  2. 2Department of Obstetrics and Gynaecology, University Medical Centre Groningen, University of Groningen, Groningen, Netherlands
  3. 3Department of Pelvic Cancer, Karolinska University Hospital, Stockholm, Sweden
  4. 4Department of Women’s and Children’s Health, Karolinska Institutet, Stockholm, Sweden
  5. 5Department of Obstetrics and Gynaecology, GROW school for oncology and developmental biology, Maastricht University Medical Centre+, Maastricht, Netherlands
  6. 6Department of Obstetrics and Gynaecology, Catharina Hospital, Eindhoven, Netherlands
  7. 7Department of Gynaecology, Juliane Marie Centre, Rigshospitalet, Copenhagen, Denmark
  8. 8Department of Gynaecology, Aarhus University Hospital, Aarhus, Denmark
  9. 9Department of Clinical Medicine, Aarhus University, Aarhus, Denmark
  10. 10Department of Public Health, Erasmus MC University Medical Centre, Rotterdam, Netherlands
  11. 11Department of Gynaecology, Odense University Hospital, Odense, Denmark

Abstract

Introduction Standard treatment of early-stage cervical cancer is radical hysterectomy with pelvic lymphadenectomy. However, a generic benefit of systematic lymphadenectomy is questionable, as most women will not have metastatic nodes. This study aimed to classify the most important risk factors of lymph node metastases (pN+) and to identify a group of women at low risk of pN+, in a large cohort of Danish, Swedish and Dutch women, using federated learning.

Methods Women diagnosed with cervical cancer between 2005–2020 were identified from nationwide population-based registries: the Danish Gynaecological Cancer Database, Swedish Quality Registry of Gynaecologic Cancer and Netherlands Cancer Registry. Inclusion criteria were: squamous cell carcinoma, adenocarcinoma or adenosquamous carcinoma; FIGO 2009 stage IA2, IB1 and IIA1; treatment with radical hysterectomy and pelvic lymph node assessment. We applied privacy-preserving federated logistic regression to identify risk factors of pN+ (figure 1). Significant factors were used to identify a low-risk group (pN+ &le5%).

Results A total of 3,606 women were included. The most important risk factors of pN+ were lymphovascular space invasion (LVSI) (odds ratio [OR] 5.16, 95% confidence interval [CI] 4.59–5.79), tumour size 21–40 mm (OR 2.14, 95% CI 1.89–2.43) and depth of invasion &gt10 mm (OR 1.81, 95% CI 1.59–2.08]). Tumours without LVSI, a size ≤20 mm and a depth of invasion ≤10 mm were associated with a low risk of pN+ (2%, 95% CI 2–3%) (table 1).

Abstract 2022-LBA-1282-ESGO Figure 1

Approaches to analysing data from different sources; a) Centralisation. This is the traditional approach, but has several disadvantages such as loss of data control logistics data governance and (most importantly) putting at risk sensitive patient data. b) Federated warring in this docentralised approach, privacy-sensitive patient data are not shared, but kept undisclosed and safe at their original location. Communication within the infrastructure is end-to-end encrypted

Abstract 2022-LBA-1282-ESGO Table 1

Risk of lymph node metastases, stratified by the most important risk factors

Conclusions LVSI, tumours size and depth of invasion were the most important risk factors of pN+. Based on that, we identified a group at very low risk of pN+, in whom sentinel lymph node mapping should be considered to replace radical pelvic lymphadenectomy.

Statistics from Altmetric.com

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.