Geburtshilfe Frauenheilkd 2018; 78(10): 188
DOI: 10.1055/s-0038-1671322
Poster
Freitag, 02.11.2018
Gynäkologische Onkologie IV
Georg Thieme Verlag KG Stuttgart · New York

Estimating risk of malignancy in adnexal masses with ultrasound: a retrospective diagnostic accuracy study

H Reina
1  Universitätsfrauenklinik Basel, Basel, Schweiz
,
J Büchel
2  Universitätsfrauenklinik Basel, Ultraschallabteilung, Basel, Schweiz
,
A Butenschön
2  Universitätsfrauenklinik Basel, Ultraschallabteilung, Basel, Schweiz
,
F Vigo
1  Universitätsfrauenklinik Basel, Basel, Schweiz
,
A Schoetzau
1  Universitätsfrauenklinik Basel, Basel, Schweiz
,
V Heinzelmann-Schwarz
3  Universitätsfrauenklinik Basel, Gynäkologische Onkologie, Basel, Schweiz
,
G Manegold-Brauer
2  Universitätsfrauenklinik Basel, Ultraschallabteilung, Basel, Schweiz
› Author Affiliations
Further Information

Publication History

Publication Date:
20 September 2018 (online)

 

Objectives:

To evaluate the performance of ultrasound with pattern recognition by experts, Risk of Malignancy Index (RMI), IOTA simple rules and IOTA ADNEX in the differentiation between benign and malignant adnexal masses.

Methods:

This is a retrospective diagnostic accuracy study, based on data prospectively collected from patients with adnexal masses, who underwent transvaginal and/or transabdominal examination by experienced examiners in our department between December 2016 and December 2017. The risk of malignancy, defined as invasive or borderline tumors, was determined by pattern recognition and the use of three prediction models: the ADNEX model, IOTA simple rules and the Risk of Malignancy Index (RMI, cut off 200). Histological findings were the clinical reference standard.

Results:

In the studied period, we recorded adnexal findings in 417 consecutive examinations. Of these, 45% had surgery and were included in the analysis. 60% of patients were premenopausal and 40% were postmenopausal. Ninety percent of the masses were benign, 4% borderline-tumors, 6% invasive cancers. Expert pattern recognition and RMI had the highest specificity (93%) in differentiating malignant from benign tumors. The ADNEX model showed a sensitivity of 92% and correctly differentiated borderline from invasive tumors, at a specificity of 86%. The use of a cut off ≤5% of risk of malignancy improved sensitivity and specificity of ADNEX.

Conclusions:

Although pattern recognition by experts resulted to have the best discriminating power, especially for borderline tumors, RMI and IOTA-models help in preoperative planning and are a valuable tool for triage for referral and in teaching settings.