[Skip to Navigation]
Sign In
Figure.  Study Flowchart
Study Flowchart

IOTA5 indicates International Ovarian Tumor Analysis phase 5 study.

aFrom the 8494 patients in 36 centers, the tumor outcome was benign in 5720 (67%), malignant in 1342 (16%), and uncertain in 1432 (17%).

bPredefined criteria for including centers: at least 50 patients recruited, consecutive recruitment, and good follow-up information for at least 70% of recruited patients. Good quality of follow-up data was defined as a recorded study outcome (spontaneous resolution, surgery with histology, or death) or last follow-up scan 10 months or more after the inclusion scan.

Table 1.  Definition of Tumor Outcome Based on Histology or Clinical Information
Definition of Tumor Outcome Based on Histology or Clinical Information
Table 2.  IOTA Terms and Variables Used to Translate the IOTA Terminology Into the O-RADS Lexicona
IOTA Terms and Variables Used to Translate the IOTA Terminology Into the O-RADS Lexicona
Table 3.  Observed Prevalence of Malignancy per O-RADS Groupa
Observed Prevalence of Malignancy per O-RADS Groupa
Table 4.  Sensitivity, Specificity, and Positive and Negative Predictive Value Regarding Malignancya
Sensitivity, Specificity, and Positive and Negative Predictive Value Regarding Malignancya
Supplement.

eTable 1. Patient and Tumor Characteristics

eTable 2. Outcome Determination by O-RADS Category

eTable 3. Observed Number and Percentage of Different Tumor Types in Each Ovarian-Adnexal Reporting and Data System (O-RADS) Risk Group When Using the O-RADS Lexicon and When Using the International Ovarian Tumor Analysis (IOTA) 2-Step Strategy to Estimate the Malignancy Risk (Pooled Analysis)

eTable 4. Observed Number and Percentage of Malignant Tumors in Each Ovarian-Adnexal Reporting and Data System (O-RADS) Subgroup Category (Pooled Analysis)

eTable 5. Observed Prevalence of Malignancy per O-RADS Group Depending on Menopausal Status, Type of Center, or the Actual Management When Using the Ovarian-Adnexal Reporting and Data System (O-RADS) Lexicon and the International Ovarian Tumor Analysis (IOTA) 2-Step Strategy to Estimate the Risk of Malignancy (Pooled Analysis)

eTable 6. Sensitivity and Specificity With Regard to Malignancy of the Ovarian-Adnexal Reporting and Data System (O-RADS) Lexicon and International Ovarian Tumor Analysis (IOTA) 2-Step Strategy Depending on Menopausal Status, Type of Center, and Actual Management (Meta-analysis)

eTable 7. Observed Prevalence of Malignancy in Ovarian-Adnexal Reporting and Data System (O-RADS) Groups 3a, 3d, 3e, 4d, and 5c When Number of Cyst Locules (O-RADS 3d), Echogenicity of Cyst Fluid (O-RADS 3a, 3d), and Shadowing (O-RADS 3e, 4d, and 5c) Are Taken Into Account (Pooled Analysis)

eAppendix. Search Strategy to Find Publications on Validation of Ovarian Adnexal Reporting and Data System (O-RADS)

eTable 8. Summary of Studies Validating the Ovarian Adnexal Reporting and Data System (O-RADS)

eReferences.

1.
Sung  H, Ferlay  J, Siegel  RL,  et al.  Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries.   CA Cancer J Clin. 2021;71(3):209-249. doi:10.3322/caac.21660 PubMedGoogle ScholarCrossref
2.
Querleu  D, Planchamp  F, Chiva  L,  et al.  European Society of Gynaecologic Oncology quality indicators for advanced ovarian cancer surgery.   Int J Gynecol Cancer. 2016;26(7):1354-1363. doi:10.1097/IGC.0000000000000767 PubMedGoogle ScholarCrossref
3.
Woo  YL, Kyrgiou  M, Bryant  A, Everett  T, Dickinson  HO. Centralisation of services for gynaecological cancer.  Cochrane Database Syst Rev. 2012;2012(3):CD007945. doi:10.1002/14651858.CD007945.pub2PubMedCrossref
4.
Verleye  L, Vergote  I, van der Zee  AG.  Patterns of care in surgery for ovarian cancer in Europe.   Eur J Surg Oncol. 2010;36(suppl 1):S108-S114. doi:10.1016/j.ejso.2010.06.006 PubMedGoogle ScholarCrossref
5.
Engelen  MJ, Kos  HE, Willemse  PH,  et al.  Surgery by consultant gynecologic oncologists improves survival in patients with ovarian carcinoma.   Cancer. 2006;106(3):589-598. doi:10.1002/cncr.21616 PubMedGoogle ScholarCrossref
6.
Bristow  RE, Chang  J, Ziogas  A, Anton-Culver  H.  Adherence to treatment guidelines for ovarian cancer as a measure of quality care.   Obstet Gynecol. 2013;121(6):1226-1234. doi:10.1097/AOG.0b013e3182922a17 PubMedGoogle ScholarCrossref
7.
Andreotti  RF, Timmerman  D, Strachowski  LM,  et al.  O-RADS US Risk Stratification and Management System: a consensus guideline from the ACR Ovarian-Adnexal Reporting and Data System Committee.   Radiology. 2020;294(1):168-185. doi:10.1148/radiol.2019191150 PubMedGoogle ScholarCrossref
8.
Van Calster  B, Van Hoorde  K, Valentin  L,  et al; International Ovarian Tumour Analysis Group. Evaluating the risk of ovarian cancer before surgery using the ADNEX model to differentiate between benign, borderline, early and advanced stage invasive, and secondary metastatic tumours: prospective multicentre diagnostic study.  BMJ. 2014;349:g5920. doi:10.1136/bmj.g5920PubMedCrossref
9.
Van Calster  B, Valentin  L, Froyman  W,  et al. Validation of models to diagnose ovarian cancer in patients managed surgically or conservatively: multicentre cohort study.  BMJ. 2020;370:m2614. doi:10.1136/bmj.m2614PubMedCrossref
10.
Landolfo C, Bourne T, Froyman W, et al. Benign descriptors and ADNEX in two-step strategy to estimate risk of malignancy in ovarian tumors: retrospective validation on IOTA 5 multicenter cohort.  Ultrasound Obstet Gynecol. Published online September 30, 2022. doi:10.1002/uog.26080PubMedCrossref
11.
Hack  K, Gandhi  N, Bouchard-Fortier  G,  et al.  External validation of O-RADS US risk stratification and management system.   Radiology. 2022;304(1):114-120. doi:10.1148/radiol.211868 PubMedGoogle ScholarCrossref
12.
Lai  HW, Lyu  GR, Kang  Z, Li  LY, Zhang  Y, Huang  YJ.  Comparison of O-RADS, GI-RADS, and ADNEX for diagnosis of adnexal masses: an external validation study conducted by junior sonologists.   J Ultrasound Med. 2022;41(6):1497-1507. doi:10.1002/jum.15834 PubMedGoogle ScholarCrossref
13.
Hiett  AK, Sonek  J, Guy  M, Reid  TJ.  Performance of IOTA Simple Rules, Simple Rules risk assessment, ADNEX model and O-RADS in discriminating between benign and malignant adnexal lesions in North American population.   Ultrasound Obstet Gynecol. 2022;59(5):668-676. doi:10.1002/uog.24777 PubMedGoogle ScholarCrossref
14.
Cao  L, Wei  M, Liu  Y,  et al.  Validation of American College of Radiology Ovarian-Adnexal Reporting and Data System Ultrasound (O-RADS US): analysis on 1054 adnexal masses.   Gynecol Oncol. 2021;162(1):107-112. doi:10.1016/j.ygyno.2021.04.031 PubMedGoogle ScholarCrossref
15.
Xie  WT, Wang  YQ, Xiang  ZS,  et al. Efficacy of IOTA Simple Rules, O-RADS, and CA125 to distinguish benign and malignant adnexal masses.  J Ovarian Res. 2022;15(1):15. doi:10.1186/s13048-022-00947-9PubMedCrossref
16.
Basha  MAA, Metwally  MI, Gamil  SA,  et al.  Comparison of O-RADS, GI-RADS, and IOTA Simple Rules regarding malignancy rate, validity, and reliability for diagnosis of adnexal masses.   Eur Radiol. 2021;31(2):674-684. doi:10.1007/s00330-020-07143-7 PubMedGoogle ScholarCrossref
17.
Chen  GY, Hsu  TF, Chan  IS,  et al.  Comparison of the O-RADS and ADNEX models regarding malignancy rate and validity in evaluating adnexal lesions.   Eur Radiol. Published online May 18, 2022. doi:10.1007/s00330-022-08803-6 PubMedGoogle ScholarCrossref
18.
Wang  R, Li  X, Li  S,  et al.  Clinical value of O-RADS combined with serum CA125 and HE4 for the diagnosis of ovarian tumours.   Acta Radiol. Published online March 15, 2022. doi:10.1177/02841851221087376 PubMedGoogle ScholarCrossref
19.
Chen  H, Yang  BW, Qian  L,  et al.  Deep learning prediction of ovarian malignancy at US compared with O-RADS and expert assessment.   Radiology. 2022;304(1):106-113. doi:10.1148/radiol.211367 PubMedGoogle ScholarCrossref
20.
Solis Cano  DG, Cervantes Flores  HA, De Los Santos Farrera  O, Guzman Martinez  NB, Soria Céspedes  D. Sensitivity and specificity of ultrasonography using Ovarian-Adnexal Reporting and Data System classification versus pathology findings for ovarian cancer.  Cureus. 2021;13(9):e17646. doi:10.7759/cureus.17646PubMedCrossref
21.
Pi  Y, Wilson  MP, Katlariwala  P,  et al.  Diagnostic accuracy and inter-observer reliability of the O-RADS scoring system among staff radiologists in a North American academic clinical setting.   Abdom Radiol (NY). 2021;46(10):4967-4973. doi:10.1007/s00261-021-03193-7 PubMedGoogle ScholarCrossref
22.
Guo  Y, Zhao  B, Zhou  S,  et al.  A comparison of the diagnostic performance of the O-RADS, RMI4, IOTA LR2, and IOTA SR systems by senior and junior doctors.   Ultrasonography. 2022;41(3):511-518. doi:10.14366/usg.21237 PubMedGoogle ScholarCrossref
23.
Ahmed  HEK. The usefulness of the ultrasound diagnosis of suspicious ovarian masses based on the O-RADS classification system.  Al-Azhar Int Med J. 2021;2(10):1-6. doi:10.21608/aimj.2021.90943.1551Crossref
24.
Jha  P, Gupta  A, Baran  TM,  et al. Diagnostic performance of the Ovarian-Adnexal Reporting and Data System (O-RADS) ultrasound risk score in women in the United States.  JAMA Netw Open. 2022;5(6):e2216370. doi:10.1001/jamanetworkopen.2022.16370PubMedCrossref
25.
Froyman  W, Landolfo  C, De Cock  B,  et al.  Risk of complications in patients with conservatively managed ovarian tumours (IOTA5): a 2-year interim analysis of a multicentre, prospective, cohort study.   Lancet Oncol. 2019;20(3):448-458. doi:10.1016/S1470-2045(18)30837-4 PubMedGoogle ScholarCrossref
26.
Bossuyt  PM, Reitsma  JB, Bruns  DE,  et al; STARD Group. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies.  BMJ. 2015;351:h5527. doi:10.1136/bmj.h5527PubMedCrossref
27.
Education and Practical Standards Committee, European Federation of Societies for Ultrasound in Medicine and Biology.  Minimum training recommendations for the practice of medical ultrasound.   Ultraschall Med. 2006;27(1):79-105.PubMedGoogle ScholarCrossref
28.
Timmerman  D, Valentin  L, Bourne  TH, Collins  WP, Verrelst  H, Vergote  I; International Ovarian Tumor Analysis (IOTA) Group.  Terms, definitions and measurements to describe the sonographic features of adnexal tumors: a consensus opinion from the International Ovarian Tumor Analysis (IOTA) Group.   Ultrasound Obstet Gynecol. 2000;16(5):500-505. doi:10.1046/j.1469-0705.2000.00287.x PubMedGoogle ScholarCrossref
29.
Valentin  L.  Use of morphology to characterize and manage common adnexal masses.   Best Pract Res Clin Obstet Gynaecol. 2004;18(1):71-89. doi:10.1016/j.bpobgyn.2003.10.002 PubMedGoogle ScholarCrossref
30.
Timmerman  D, Testa  AC, Bourne  T,  et al; International Ovarian Tumor Analysis Group.  Logistic regression model to distinguish between the benign and malignant adnexal mass before surgery: a multicenter study by the International Ovarian Tumor Analysis Group.   J Clin Oncol. 2005;23(34):8794-8801. doi:10.1200/JCO.2005.01.7632 PubMedGoogle ScholarCrossref
31.
Prat  J; FIGO Committee on Gynecologic Oncology.  Staging classification for cancer of the ovary, fallopian tube, and peritoneum.   Int J Gynaecol Obstet. 2014;124(1):1-5. doi:10.1016/j.ijgo.2013.10.001 PubMedGoogle ScholarCrossref
32.
Ameye  L, Timmerman  D, Valentin  L,  et al.  Clinically oriented three-step strategy for assessment of adnexal pathology.   Ultrasound Obstet Gynecol. 2012;40(5):582-591. doi:10.1002/uog.11177 PubMedGoogle ScholarCrossref
33.
Testa  A, Kaijser  J, Wynants  L,  et al.  Strategies to diagnose ovarian cancer: new evidence from phase 3 of the multicentre international IOTA study.   Br J Cancer. 2014;111(4):680-688. doi:10.1038/bjc.2014.333 PubMedGoogle ScholarCrossref
34.
Reitsma  JB, Glas  AS, Rutjes  AW, Scholten  RJ, Bossuyt  PM, Zwinderman  AH.  Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews.   J Clin Epidemiol. 2005;58(10):982-990. doi:10.1016/j.jclinepi.2005.02.022 PubMedGoogle ScholarCrossref
35.
Leeflang  MM, Deeks  JJ, Rutjes  AW, Reitsma  JB, Bossuyt  PM.  Bivariate meta-analysis of predictive values of diagnostic tests can be an alternative to bivariate meta-analysis of sensitivity and specificity.   J Clin Epidemiol. 2012;65(10):1088-1097. doi:10.1016/j.jclinepi.2012.03.006 PubMedGoogle ScholarCrossref
38.
de Groot  JAH, Bossuyt  PMM, Reitsma  JB,  et al. Verification problems in diagnostic accuracy studies: consequences and solutions.  BMJ. 2011;343:d4770. doi:10.1136/bmj.d4770PubMedCrossref
39.
Moons  KGM, Wolff  RF, Riley  RD,  et al.  PROBAST: a tool to assess risk of bias and applicability of prediction model studies: explanation and elaboration.   Ann Intern Med. 2019;170(1):W1-W33. doi:10.7326/M18-1377 PubMedGoogle ScholarCrossref
Original Investigation
December 15, 2022

External Validation of the Ovarian-Adnexal Reporting and Data System (O-RADS) Lexicon and the International Ovarian Tumor Analysis 2-Step Strategy to Stratify Ovarian Tumors Into O-RADS Risk Groups

Author Affiliations
  • 1Department of Development and Regeneration, KU Leuven, Leuven, Belgium
  • 2Department of Obstetrics and Gynecology, University Hospital Leuven, Leuven, Belgium
  • 3Department of Obstetrics and Gynecology, Skåne University Hospital, Malmö, Sweden
  • 4Department of Clinical Sciences Malmö, Lund University, Malmö, Sweden
  • 5Laboratory of Tumor Immunology and Immunotherapy, Department of Oncology, Leuven Cancer Institute, KU Leuven, Leuven, Belgium
  • 6Dipartimento Scienze della Salute della Donna, del Bambino e di Sanità Pubblica, Fondazione Policlinico Universitario A. Gemelli, IRCSS, Rome, Italy
  • 7Queen Charlotte’s and Chelsea Hospital, Imperial College, London, United Kingdom
  • 8Department of Obstetrics and Gynecology, Ziekenhuis Oost-Limburg, Genk, Belgium
  • 9First Department of Obstetrics and Gynecology, Alexandra Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece
  • 10Clinic of Obstetrics and Gynecology, University of Milan–Bicocca, San Gerardo Hospital, Monza, Italy
  • 11Department of Clinical Science and Education, Karolinska Institutet and Department of Obstetrics and Gynecology, Södersjukhuset, Stockholm, Sweden
  • 12Preventive Gynecology Unit, Division of Gynecology, European Institute of Oncology IRCCS, Milan, Italy
  • 13Department of Perinatology and Oncological Gynecology, Faculty of Medical Sciences, Medical University of Silesia, Katowice, Poland
  • 14Department of Gynecologic Oncology, National Cancer Institute of Milan, Milan, Italy
  • 15Department of Obstetrics and Gynecology, Clinica Universidad de Navarra, School of Medicine, Pamplona, Spain
  • 16Department of Obstetrics and Gynecology, Biomedical and Clinical Sciences Institute L. Sacco, University of Milan, Milan, Italy
  • 17Institute for Maternal and Child Health–IRCCS “Burlo Garofolo,” Trieste, Italy
  • 18Department of Biomedical, Experimental and Clinical Sciences, University of Florence, Florence, Italy
  • 19Department of Obstetrics and Gynecology, University of Cagliari, Policlinico Universitario Duilio Casula, Monserrato, Cagliari, Italy
  • 20Department of Obstetrics and Gynaecology, Whipps Cross Hospital, London, United Kingdom
  • 21Department of Obstetrics and Gynecology, Ikazia Hospital, Rotterdam, the Netherlands
  • 22Department of Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, Tennessee
  • 23Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, Nashville, Tennessee
  • 24Department of Biomedical Data Sciences, Leiden University Medical Centre, Leiden, the Netherlands
JAMA Oncol. 2023;9(2):225-233. doi:10.1001/jamaoncol.2022.5969
Key Points

Question  Can the Ovarian-Adnexal Reporting and Data System (O-RADS) lexicon proposed by the American College of Radiology and the International Ovarian Tumor Analysis (IOTA) 2-step strategy be used to determine the likelihood of malignancy?

Findings  In this diagnostic study of 4905 patients with ovarian tumors, both the O-RADS lexicon and the IOTA 2-step strategy could be used to stratify patients into risk groups. However, the observed malignancy rate in O-RADS 2 was not clearly below 1%.

Meaning  Findings suggest that risk stratification using the O-RADS lexicon or the IOTA 2-step strategy can be used to guide management.

Abstract

Importance  Correct diagnosis of ovarian cancer results in better prognosis. Adnexal lesions can be stratified into the Ovarian-Adnexal Reporting and Data System (O-RADS) risk of malignancy categories with either the O-RADS lexicon, proposed by the American College of Radiology, or the International Ovarian Tumor Analysis (IOTA) 2-step strategy.

Objective  To investigate the diagnostic performance of the O-RADS lexicon and the IOTA 2-step strategy.

Design, Setting, and Participants  Retrospective external diagnostic validation study based on interim data of IOTA5, a prospective international multicenter cohort study, in 36 oncology referral centers or other types of centers. A total of 8519 consecutive adult patients presenting with an adnexal mass between January 1, 2012, and March 1, 2015, and treated either with surgery or conservatively were included in this diagnostic study. Twenty-five patients were excluded for withdrawal of consent, 2777 were excluded from 19 centers that did not meet predefined data quality criteria, and 812 were excluded because they were already in follow-up at recruitment. The analysis included 4905 patients with a newly detected adnexal mass in 17 centers that met predefined data quality criteria. Data were analyzed from January 31 to March 1, 2022.

Exposures  Stratification into O-RADS categories (malignancy risk <1%, 1% to <10%, 10% to <50%, and ≥50%). For the IOTA 2-step strategy, the stratification is based on the individual risk of malignancy calculated with the IOTA 2-step strategy.

Main Outcomes and Measures  Observed prevalence of malignancy in each O-RADS risk category, as well as sensitivity and specificity. The reference standard was the status of the tumor at inclusion, determined by histology or clinical and ultrasonographic follow-up for 1 year. Multiple imputation was used for uncertain outcomes owing to inconclusive follow-up information.

Results  Median age of the 4905 patients was 48 years (IQR, 36-62 years). Data on race and ethnicity were not collected. A total of 3441 tumors (70%) were benign, 978 (20%) were malignant, and 486 (10%) had uncertain classification. Using the O-RADS lexicon resulted in 1.1% (24 of 2196) observed prevalence of malignancy in O-RADS 2, 4% (34 of 857) in O-RADS 3, 27% (246 of 904) in O-RADS 4, and 78% (732 of 939) in O-RADS 5; the corresponding results for the IOTA 2-step strategy were 0.9% (18 of 1984), 4% (58 of 1304), 30% (206 of 690), and 82% (756 of 927). At the 10% risk threshold (O-RADS 4-5), the O-RADS lexicon had 92% sensitivity (95% CI, 87%-96%) and 80% specificity (95% CI, 74%-85%), and the IOTA 2-step strategy had 91% sensitivity (95% CI, 84%-95%) and 85% specificity (95% CI, 80%-88%).

Conclusions and Relevance  The findings of this external diagnostic validation study suggest that both the O-RADS lexicon and the IOTA 2-step strategy can be used to stratify patients into risk groups. However, the observed malignancy rate in O-RADS 2 was not clearly below 1%.

Introduction

Ovarian cancer is the eighth most common cancer worldwide in women and is the most lethal gynecologic malignancy.1 Because patients with ovarian cancer have a better prognosis when treated in tertiary oncology centers than in other centers, correct preoperative diagnosis of adnexal masses should result in optimal management.2-6

In 2020, an international multidisciplinary committee, sponsored by the American College of Radiology, published the Ovarian-Adnexal Reporting and Data System (O-RADS).7 The system suggests a uniform lexicon to describe ultrasonographic images of adnexal masses, uses this lexicon to stratify masses into different risk groups of malignancy, and suggests management for each risk group.

Another way to classify adnexal masses into O-RADS risk groups than using the O-RADS lexicon is to use a mathematical model to calculate the risk of malignancy (eg, the Assessment of Different Neoplasias in the Adnexa [ADNEX] model developed by the International Ovarian Tumor Analysis [IOTA] group). The ADNEX model is a polynomial logistic regression model that estimates the risk of 5 tumor types: benign, borderline, stage I primary invasive, stage II to IV primary invasive, and secondary metastasis. It is based on ultrasonographic and clinical information and can be used with or without information on serum cancer antigen 125 (CA-125).8,9 A third method of risk stratification is to apply the IOTA 2-step strategy (ie, the modified IOTA benign simple descriptors [BDs], which do not require access to a computer, are used first, and if these do not apply, ADNEX is used).10 When externally validated with patients treated with surgery or expectantly, this 2-step strategy had good performance.10

The ability of the O-RADS lexicon to place patients in the correct O-RADS risk group and the sensitivity and specificity of the O-RADS lexicon regarding malignancy have been validated in studies using retrospective review of images and in 1 small prospective study. In most of these studies, histology was the reference standard.11-24 According to the original O-RADS publication, masses with risk of malignancy less than 1% could be managed with follow-up.7 Therefore, validation studies should be performed for all patients with an adnexal mass, irrespective of whether they are treated conservatively or with surgery. The aim of this study was to estimate the diagnostic performance of the O-RADS lexicon and the IOTA 2-step strategy when used for both surgically and conservatively managed adnexal masses.

Methods
Study Design and Population

This diagnostic study was a retrospective external validation study using the 2-year interim data from IOTA phase 5 (IOTA5), an ongoing international multicenter prospective observational cohort study (ClinicalTrials.gov Identifier: NCT01698632).9,25 The IOTA5 study protocol was approved by the ethics committee of the University Hospitals Leuven as the coordinating center and the local ethics committee of each contributing center. The study design has been described elsewhere and is briefly outlined here.9,25 The study followed the Standards for Reporting of Diagnostic Accuracy (STARD) reporting guideline.26

Patients were eligible for IOTA5 if they were aged 18 years or older at recruitment and had at least 1 adnexal mass (ovarian, paraovarian, or tubal) detected on ultrasonographic examination. Premenopausal patients with a clearly physiologic cyst with largest diameter less than 3 cm were not eligible. The IOTA5 2-year interim analysis data set contains data from patients recruited at 36 centers (oncology or nononcology centers) in 14 countries between January 1, 2012, and March 1, 2015, with follow-up data until June 30, 2017. Patients with an adnexal mass who were already in follow-up in the recruitment center before the start of the study were not used in this analysis.

Written or oral informed consent was obtained from every participant at inclusion. All patients underwent a standardized transvaginal ultrasonographic examination by an IOTA-certified examiner (these examiners had completed a standardized IOTA course and passed an IOTA test on assessment of adnexal masses). Most ultrasonographic examiners were level 2 (experienced) or 3 (advanced) examiners.27 By design, the examiners were blinded to the outcome. Results of ultrasonography and clinical information were registered in accordance with research protocol.9,24 IOTA terminology was used to describe the ultrasonographic findings,28 and a set of predefined ultrasonographic variables was collected for each patient.9,25 In addition, the ultrasonographic examiner’s diagnosis (benign, borderline, or malignant; specific diagnosis from a drop-down list), based on pattern recognition (ie, based on knowledge of the typical ultrasonographic appearance of different types of adnexal pathology29), was recorded together with the confidence with which the diagnosis was made (certainly benign, probably benign, uncertain, probably malignant, or certainly malignant). The ultrasonographic examiner suggested management based on clinical findings and the ultrasonographic diagnosis. The treating clinician decided on the final management together with the patient. Conservative management included ultrasonography and clinical follow-up at intervals of 3 months, 6 months, and then every 12 months. At follow-up visits, clinical data (including symptoms) were collected, and an ultrasonographic examination was performed in the same manner as at the inclusion scan. Some patients initially treated conservatively underwent surgery after 1 or more follow-up visits (eg, because of suspicion of malignancy, symptoms, patient anxiety). In some patients, the mass resolved spontaneously during follow-up. In case of multiple masses, the mass with the ultrasonographic features most suggestive of malignancy was defined as dominant and was used in our statistical analysis.

Patients who underwent surgery were treated according to local protocols, with histologic examination of surgically removed masses. Central pathology review was not performed because, in a previous study, we did not observe important differences in diagnoses between local and central pathology reports.30 Malignant tumors were classified according to International Federation of Gynecology and Obstetrics recommendations.31

Data Collection and Cleaning

Patient data were collected on a secure electronic platform developed for the study (IOTA5 Study Screen; astraia software, version 2.0_58). Patients were pseudonymized on inclusion with a unique identifier, ensuring encryption of all data communication. A team of biostatisticians and ultrasonographic examiners cleaned the data, which included correcting inconsistencies and retrieving missing information. Before analyzing the data, we defined the criteria for a study center to be included in our analysis: recruitment of at least 50 patients, consecutive recruitment, and good-quality follow-up data (ie, a recorded study outcome or last follow-up visit ≥10 months after inclusion) for at least 70% of the recruited patients.9

Reference Standard

The reference standard describes the nature of the adnexal mass as benign or malignant at inclusion (borderline tumors were classified as malignant). It is based on histology for masses removed by surgery. Pathologists were blinded to ultrasonographic variables, risk predictions, and O-RADS groups but might have received information on the diagnosis suspected by the ultrasonographic examiner. If surgery was not performed, the reference standard was based on clinical history and status and the ultrasonographic examiner’s diagnosis and diagnostic confidence at inclusion and during follow-up until 12 months (±2 months). Table 1 shows how we determined tumor outcome.9

O-RADS Risk Stratification

O-RADS 1 indicates a normal ovary, which is not applicable here. O-RADS 2 indicates an almost certainly benign tumor (malignancy risk <1%), O-RADS 3 indicates low risk (malignancy risk 1% to <10%), O-RADS 4 indicates intermediate risk (malignancy risk 10% to <50%), and O-RADS 5 indicates high risk (malignancy risk ≥50%).7 Table 2 shows how we derived the O-RADS lexicon from the 2-year interim IOTA5 database.

Two-Step Strategy

The modified IOTA BDs almost always indicate a benign tumor according to studies of patients who underwent surgery.32,33 This corresponds to O-RADS 2 (risk of malignancy <1%). The BDs are BD1 (unilocular cyst, ground-glass echogenicity, largest diameter <10 cm, and premenopausal patient), BD2 (unilocular cyst, mixed echogenicity, acoustic shadows, largest diameter <10 cm, and premenopausal patient), BD3 (unilocular cyst, anechoic cyst fluid, smooth internal walls, and largest diameter <10 cm), and BD4 (remaining unilocular cysts with smooth internal walls and largest diameter <10 cm).

If the BDs do not apply, ADNEX is used, which calculates the risk of 5 tumor types: benign, borderline, stage I invasive cancer, stage II to IV invasive cancer, and secondary metastasis.9 The risk of malignancy is obtained by adding the risks of the 4 malignant tumor types. In this study, we used ADNEX without serum CA-125 because CA-125 results are usually not known when a patient presents with an adnexal mass. The risks of malignancy generated by ADNEX were divided into 4 categories corresponding to the malignancy probabilities of O-RADS 2 to 5.

Statistical Analysis

We used multiple imputation to deal with uncertain outcomes (U1-U4 in Table 1). We generated 100 imputations and refer to previous work for details.9 Our results are based on the multiply imputed values for the outcome.

The percentage of patients was calculated, as well as the histologic outcome (benign, borderline, stage I invasive, stage II-IV invasive, and secondary metastatic) in each O-RADS risk group (pooled analysis). Sensitivity, specificity, positive predictive value, and negative predictive value were calculated for the O-RADS lexicon and the IOTA 2-step strategy at 3 O-RADS risk thresholds: 1% (O-RADS ≥3 vs O-RADS 2), 10% (O-RADS 4-5 vs O-RADS 2-3), and 50% (O-RADS 5 vs O-RADS 2-4) (meta-analysis). To deal with multiply imputed data, logit-transformed values for sensitivity and specificity were combined using Rubin rules to obtain center-specific results. These center-specific results (logit transformed) were combined with bivariate random-effects meta-analysis to calculate the overall sensitivity and specificity.34 Center-specific positive and negative predictive values (logit transformed) and their variance were computed and combined using Rubin rules to obtain a final center-specific estimate. These estimates were used in a bivariate random-effects model to calculate the final estimate.35

We performed predetermined subgroup analyses depending on menopausal status and type of center (oncology center vs other). We also performed 2 post hoc subgroup analyses defined by actual management: patients who underwent surgery within 120 days of the inclusion scan without any follow-up scan, and patients with at least 1 follow-up scan.

Some ultrasonographic features are not included in the original O-RADS classification. At the request of the members of the American College of Radiology O-RADS committee, we calculated the observed prevalence of malignancy when taking into account the echogenicity of cyst fluid (anechoic vs other), number of cyst locules in multilocular cysts (2 cyst locules vs >2), and presence of shadowing in smooth solid tumors.

The statistical analysis was performed with R, version 4.1.0 (R Foundation for Statistical Computing). We used the mice package for multiple imputation and the metafor package (rma.mv function) for the meta-analysis of diagnostic performance. Data were analyzed from January 31, 2022, to March 1, 2022.

Results

The Figure shows patient flow, with an initial consecutive cohort of 8519 patients from 36 centers. Our statistical analysis included the data of all 4905 patients in the 17 centers that fulfilled our predetermined data quality criteria9: 3441 (70%) tumors were benign, 978 (20%) were malignant, and for 486 tumors (10%), the outcome was uncertain and imputed. Patient and tumor characteristics are presented in eTable 1 in the Supplement. Median age was 48 years (IQR, 36-62 years). Data on race and ethnicity were not collected. eTable 2 in the Supplement shows how final tumor outcome was determined for each O-RADS risk group.

Using the O-RADS lexicon for risk stratification resulted in 1.1% (24 of 2196) observed prevalence of malignancy in O-RADS 2, 4% (34 of 857) in O-RADS 3, 27% (246 of 904) in O-RADS 4, and 78% (732 of 939) in O-RADS 5; the corresponding results for the IOTA 2-step strategy were 0.9% (18 of 1984), 4% (58 of 1304), 30% (206 of 690), and 82% (756 of 927) (Table 3; eTable 3 in the Supplement). With the exception of O-RADS 2 when the O-RADS lexicon was used, the observed proportion of malignant tumors per O-RADS group fell within the targeted risk range. eTable 4 in the Supplement shows the observed number and percentage of malignant tumors in each of the 33 O-RADS subgroup categories. The observed malignancy rate was less than 1% for simple cysts, classic hemorrhagic cysts, dermoid cysts, and endometriomas less than 10 cm, and for nonsimple unilocular cysts with smooth inner margin less than or equal to 3 cm.

Sensitivity, specificity, positive predictive value, and negative predictive value regarding malignancy when the O-RADS lexicon and the IOTA 2-step strategy were used are shown in Table 4. At the 10% risk threshold (O-RADS 4-5 vs O-RADS 2-3), the O-RADS lexicon had 92% sensitivity (95% CI, 87%-96%) and 80% specificity (95% CI, 74%-85%), and the 2-step strategy had 91% sensitivity (95% CI, 84%-95%) and 85% specificity (95% CI, 80%-88%).

The results of our subgroup analyses are shown in eTable 5 and eTable 6 in the Supplement. For both the O-RADS lexicon and the IOTA 2-step strategy, the observed proportion of malignant tumors in O-RADS 2 was greater than 1% in postmenopausal patients (1.8%) and in oncology centers (1.4% and 1.2%), whereas it was less than 1% in premenopausal patients (0.8% and 0.6%) and in nononcology centers (0.7% and 0.6%). The observed proportion of malignant tumors in the other O-RADS groups fell within the targeted risk range irrespective of menopausal status and type of center (eTable 2 in the Supplement). Sensitivity was higher and specificity lower in postmenopausal than premenopausal patients and in oncology centers than in other centers. In patients who underwent immediate surgery, the observed malignancy rate was higher in all O-RADS risk groups than in those with at least 1 follow-up scan, and sensitivity was higher and specificity lower.

eTable 7 in the Supplement shows the observed prevalence of malignancy in O-RADS groups 3a, 3d, 3e, 4d, and 5c when number of cyst locules (O-RADS 3d), echogenicity of cyst fluid (O-RADS 3a and 3d), and shadowing (O-RADS 3e, 4d, and 5c) were taken into account. For the 183 bilocular cysts in O-RADS subcategory 3d, the observed prevalence of malignancy was 0.7%. In O-RADS subcategories 3e, 4d, and 5c (smooth solid tumors), the observed prevalence of malignancy was substantially lower if acoustic shadows were present.

Discussion

Our results showed that both the O-RADS lexicon and the IOTA 2-step strategy performed well to stratify patients into O-RADS groups 3 to 5. For O-RADS 2, the target proportion of malignant tumors is less than 1%, but the observed proportions were not clearly below 1%. The large amount of multicentric data from patients treated conservatively or surgically suggests generalizability of these results.

We reported the observed rate of malignancy in the O-RADS groups separately for premenopausal and postmenopausal patients and, to our knowledge, for the first time, validated O-RADS separately in oncology centers and other types of centers. The higher sensitivity and lower specificity and the higher observed prevalence of malignancy in lesions classified as O-RADS 2 in postmenopausal patients and in patients examined in oncology centers are likely to be explained by differences in histologic diagnoses (“case mix”) between premenopausal and postmenopausal patients and between oncology and other centers. The difference in performance between patients who underwent surgery and those cared for with follow-up also illustrates the association between study population characteristics and results. Patients who undergo surgery constitute a highly select population in which the proportion of malignant tumors is higher than in a total population of patients with an adnexal mass. Patients treated expectantly with follow-up constitute another select population, in which the malignancy rate is very low. We believe that the best estimate of performance of the O-RADS lexicon and the IOTA 2-step strategy is obtained by including all patients with an adnexal mass irrespective of how they were treated. Our results also show that clinicians were good at selecting patients for conservative management by using clinical information and pattern recognition to interpret ultrasonographic images.

Use of the O-RADS lexicon to stratify patients into O-RADS risk groups has been validated retrospectively in 13 other studies11-22,24 and prospectively in 1 study23 (search strategy and study details shown in the eAppendix and eTable 8 in the Supplement). The prospective study was small (50 patients) and included only tumors in O-RADS 3 to 5.23 All retrospective studies used review of saved ultrasonographic images with or without supplementary review of ultrasonographic reports. In 9 retrospective studies, the reference standard was histology,12-15,17-20,22 in 3 it was histology or results of follow-up,11,16,24 and in 1 it was the agreed diagnosis of 2 radiologists according to the ultrasonographic images.21 The reported observed malignancy rates per O-RADS group and the sensitivities and specificities when O-RADS 4 to 5 was used (≥10% malignancy risk) to indicate malignancy are variable. The point estimates of observed prevalence of malignancy in O-RADS 2 ranges from 0% to 5%, with 4 centers reporting it to be greater than 1%12,15,18,22; that in O-RADS 3 ranges from 0% to 19%, with 4 centers reporting it to be greater than 10%15,18,22,23; that in O-RADS 4 ranges from 21% to 79%, with 4 centers reporting it to be greater than 50%12,15,18,23; and that in O-RADS 5 ranges from 66% to 95%. The divergent results are likely to be explained by differences in study population characteristics (eg, types of tumors included sample size, study design, interpretation of the O-RADS lexicon) and by a varying proportion of images that were not fully representative of the tumor. Three studies compared sensitivity and specificity at the 10% risk cutoff (O-RADS 4-5) between use of the O-RADS lexicon and use of ADNEX to estimate the risk of malignancy, again with divergent results.12,13,17 To our knowledge, no study reported the observed prevalence of malignancy when risk stratification was done with ADNEX or the IOTA 2-step strategy.

Risk categorization may help when patients are selected for treatment. To facilitate the use of O-RADS and ADNEX, web applications and mobile applications are available,36,37 and ADNEX is incorporated in some ultrasound machines.10 The advantage of the IOTA 2-step strategy over the O-RADS lexicon is that it provides an individual risk estimate of malignancy and estimates the likelihood of different tumor types (benign, borderline, stage I invasive malignancy, stage II to IV invasive malignancy, and secondary metastasis).8,10 However, neither the O-RADS lexicon nor the 2-step strategy confidently identified tumors with malignancy risk less than 1%. When the 2-step strategy was used to place tumors in O-RADS 2, the upper 95% confidence limit for the observed prevalence of malignancy exceeded 1%, and when the O-RADS lexicon was used, the point estimate exceeded 1%. The performance of the O-RADS lexicon might be improved by taking into account the number of cyst locules, echogenicity of cyst fluid, and shadowing, subdividing the O-RADS subcategories further. However, increasing the number of subgroups will make use of the O-RADS lexicon more difficult.

Limitations

Our study has limitations. First, although ultrasonographic information was collected prospectively, the O-RADS lexicon was applied retrospectively. A prospective study would yield results more similar to expected performance in clinical practice. Second, several centers were excluded because of few recruited patients, nonconsecutive recruitment, or insufficient data quality. Third, our reference standard is based on different methods: histology or results of clinical and ultrasonographic follow-up (differential verification).38 In some cases, the outcome was unclear because of insufficient or inconsistent information (partial verification). We dealt with this by using multiple imputation.39

Conclusions

The findings of this external diagnostic validation study suggest that both the O-RADS lexicon and the IOTA 2-step strategy can be used to stratify patients into risk groups. However, the observed malignancy rate in O-RADS 2 was not clearly less than 1%. The advantage of the 2-step strategy is that it provides an individual risk estimate, as well as risk estimates of 4 types of malignancy. Prospective validation of the 2 approaches to risk stratification is needed.

Back to top
Article Information

Accepted for Publication: September 15, 2022.

Published Online: December 15, 2022. doi:10.1001/jamaoncol.2022.5969

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Timmerman S et al. JAMA Oncology.

Corresponding Author: Wouter Froyman, MD, PhD, Department of Obstetrics and Gynecology, University Hospital Leuven, Herestraat 49, 3000 Leuven, Belgium (wouter.froyman@uzleuven.be).

Author Contributions: Profs Van Calster and Froyman had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: S. Timmerman, Valentin, Testa, Landolfo, Andreotti, D. Timmerman, Bourne, Froyman.

Acquisition, analysis, or interpretation of data: S. Timmerman, Valentin, Ceusters, Testa, Landolfo, Sladkevicius, Van Holsbeke, Domali, Fruscio, Epstein, Franchi, Kudla, Chiappa, Alcazar, Leone, Buonomo, Coccia, Guerriero, Deo, Jokubkiene, Kaijser, Scambia, D. Timmerman, Bourne, Van Calster, Froyman.

Drafting of the manuscript: S. Timmerman, Landolfo, D. Timmerman, Bourne, Froyman.

Critical revision of the manuscript for important intellectual content: Valentin, Ceusters, Testa, Sladkevicius, Van Holsbeke, Domali, Fruscio, Epstein, Franchi, Kudla, Chiappa, Alcazar, Leone, Buonomo, Coccia, Guerriero, Deo, Jokubkiene, Kaijser, Scambia, Andreotti, D. Timmerman, Bourne, Van Calster, Froyman.

Statistical analysis: Ceusters, Van Calster.

Obtained funding: D. Timmerman, Froyman.

Administrative, technical, or material support: S. Timmerman, Testa, Landolfo, Sladkevicius, Van Holsbeke, Domali, Epstein, Kudla, Buonomo, Andreotti, Bourne.

Supervision: Valentin, Testa, Coccia, D. Timmerman, Bourne, Van Calster, Froyman.

Conflict of Interest Disclosures: Dr Valentin reported receiving grants from the Swedish Research Council, Malmö University Hospital and Skåne University Hospital, Allmänna Sjukhusets i Malmö Stiftelse för bekämpande av cancer (the Malmö General Hospital Foundation for Fighting Against Cancer), Avtal om läkarutbildning och forskning (ALF)–medel, and Landstingsfinansierad Regional Forskning during the conduct of the study; and teaching fees from Samsung outside the submitted work. Dr Sladkevicius reported receiving grants from Skåne (skåne regionalt forskningsstöd) during the conduct of the study. Dr Andreotti reported receiving honoraria from Philips Healthcare outside the submitted work. Dr D. Timmerman reported receiving grants from the Research Foundation–Flanders (FWO) during the conduct of the study. Dr Bourne reported receiving speaking honoraria and departmental funding from Samsung Healthcare and grants from Roche Diagnostics, Illumina, and Abbott. No other disclosures were reported.

Funding/Support: The IOTA5 study is supported by the FWO (projects G049312N, G0B4716N, and G097322N) and Internal Funds KU Leuven (project C24/15/037). Dr Valentin is supported by the Swedish Research Council (grant K2014-99X-22475-01-3, Dnr 2013-02282), funds administered by Malmö University Hospital and Skåne University Hospital, Allmänna Sjukhusets i Malmö Stiftelse för bekämpande av cancer (the Malmö General Hospital Foundation for Fighting Against Cancer), and 2 Swedish governmental grants (Avtal om läkarutbildning och forskning [ALF]–medel and Landstingsfinansierad Regional Forskning). Dr Landolfo was supported by Linbury Trust grant LIN 260. Dr D. Timmerman is a senior clinical investigator of FWO and is supported by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Imperial College Healthcare National Health Service (NHS) Trust and Imperial College London. Prof Froyman was a clinical fellow of the FWO.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed are those of the authors and not necessarily those of the NHS, NIHR, or Department of Health.

Additional Contributions: We thank everyone who provided assistance for monitoring the study and completing the database. In particular, we thank Bavo De Cock, MS, and Laure Wynants, MSc, PhD (Department of Development and Regeneration, KU Leuven, Leuven, Belgium), for statistical analysis; Willem Mestdagh, MSc (KU Leuven, Department of Electrical Engineering [ESAT], STADIUS Center for Dynamical Systems, Signal Processing and Data Analytics, Leuven, Belgium), for data management; Mona Aboulghar, MD, PhD (Department of Obstetrics and Gynecology, Faculty of Medicine, Cairo University, Cairo, Egypt; The Egyptian IVF Center, Cairo, Egypt), Maria José dos Santos Bernardo, MD (Department of Obstetrics and Gynecology, Centro Hospitalar Lisboa Central, Lisbon, Portugal), Doriana Scardigno, MD (Department of Obstetrics, Gynecology and Neonatology, Ospedale Consorziale Policlinico, University of Bari, Bari, Italy), Paola Pollastri, MD (Istituti Ospitalieri di Cremona, Cremona, Lombardy, Italy), Ulrike Metzger, MD (Centre Européen de Diagnostic et d’Exploration de la Femme [CEDEF], Le Chesnay, France), Anna Knafel, MD (Department of Gynecology and Obstetrics, Jagiellonian University, Krakow, Poland), Chiara Lanzani, MD (Department of Obstetrics and Gynecology, Children’s Hospital Buzzi, Milan, Italy), Fatima Alves, MD (Department of Gynecology, Obstetrics and Reproductive Medicine, University Hospital de Santa Maria, Lisbon, Portugal), Thierry Van den Bosch, MD, PhD (Department of Development and Regeneration, KU Leuven, and Department of Obstetrics and Gynecology, University Hospitals KU Leuven, Leuven, Belgium), Lauri Hochberg, MD (Image-Based Gynecology Service, Department of Obstetrics and Gynecology, University of South Florida Morsani College, Tampa, Florida), Jing Zhang, MD (Ultrasound Department, Chinese PLA General Hospital, Beijing, China), Alberto Rossi, MD, PhD (Department of Obstetrics and Gynecology, University of Udine, Udine, Italy), Luca Savelli, MD, PhD (Gynecology and Physiopathology of Human Reproduction Unit, S. Orsola-Malpighi Hospital of Bologna, Bologna, Italy), Daniela Fischerová, MD, PhD (Gynecologic Oncology Centre, Department of Obstetrics and Gynecology, First Faculty of Medicine, Charles University and General University Hospital, Prague, Czech Republic), Artur Czekierdowski, MD, PhD (First Department of Gynecological Oncology and Gynecology, Medical University of Lublin, Lublin, Poland), Perrine Capmas, MD (Department of Gynecology Obstetrics, Hospital of Kremlin-Bicêtre, University Paris-Saclay, Le Kremlin Bicêtre, France), Maria Concetta Blanco, MD (Department of Obstetrics and Gynecology, Garibaldi Hospital, Catania, Italy), Samir Helmy, MD (Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria), and Maria Angela Pascual, MD (Department of Obstetrics, Gynecology and Reproduction, Hospital Universitario Dexeus, Barcelona, Spain) for data collection; and Maria Björklund, MLIS (Faculty of Medicine, Library & ICT, Lund University, Malmö, Sweden), for helping with the literature search. No one received financial compensation for his or her contributions.

References
1.
Sung  H, Ferlay  J, Siegel  RL,  et al.  Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries.   CA Cancer J Clin. 2021;71(3):209-249. doi:10.3322/caac.21660 PubMedGoogle ScholarCrossref
2.
Querleu  D, Planchamp  F, Chiva  L,  et al.  European Society of Gynaecologic Oncology quality indicators for advanced ovarian cancer surgery.   Int J Gynecol Cancer. 2016;26(7):1354-1363. doi:10.1097/IGC.0000000000000767 PubMedGoogle ScholarCrossref
3.
Woo  YL, Kyrgiou  M, Bryant  A, Everett  T, Dickinson  HO. Centralisation of services for gynaecological cancer.  Cochrane Database Syst Rev. 2012;2012(3):CD007945. doi:10.1002/14651858.CD007945.pub2PubMedCrossref
4.
Verleye  L, Vergote  I, van der Zee  AG.  Patterns of care in surgery for ovarian cancer in Europe.   Eur J Surg Oncol. 2010;36(suppl 1):S108-S114. doi:10.1016/j.ejso.2010.06.006 PubMedGoogle ScholarCrossref
5.
Engelen  MJ, Kos  HE, Willemse  PH,  et al.  Surgery by consultant gynecologic oncologists improves survival in patients with ovarian carcinoma.   Cancer. 2006;106(3):589-598. doi:10.1002/cncr.21616 PubMedGoogle ScholarCrossref
6.
Bristow  RE, Chang  J, Ziogas  A, Anton-Culver  H.  Adherence to treatment guidelines for ovarian cancer as a measure of quality care.   Obstet Gynecol. 2013;121(6):1226-1234. doi:10.1097/AOG.0b013e3182922a17 PubMedGoogle ScholarCrossref
7.
Andreotti  RF, Timmerman  D, Strachowski  LM,  et al.  O-RADS US Risk Stratification and Management System: a consensus guideline from the ACR Ovarian-Adnexal Reporting and Data System Committee.   Radiology. 2020;294(1):168-185. doi:10.1148/radiol.2019191150 PubMedGoogle ScholarCrossref
8.
Van Calster  B, Van Hoorde  K, Valentin  L,  et al; International Ovarian Tumour Analysis Group. Evaluating the risk of ovarian cancer before surgery using the ADNEX model to differentiate between benign, borderline, early and advanced stage invasive, and secondary metastatic tumours: prospective multicentre diagnostic study.  BMJ. 2014;349:g5920. doi:10.1136/bmj.g5920PubMedCrossref
9.
Van Calster  B, Valentin  L, Froyman  W,  et al. Validation of models to diagnose ovarian cancer in patients managed surgically or conservatively: multicentre cohort study.  BMJ. 2020;370:m2614. doi:10.1136/bmj.m2614PubMedCrossref
10.
Landolfo C, Bourne T, Froyman W, et al. Benign descriptors and ADNEX in two-step strategy to estimate risk of malignancy in ovarian tumors: retrospective validation on IOTA 5 multicenter cohort.  Ultrasound Obstet Gynecol. Published online September 30, 2022. doi:10.1002/uog.26080PubMedCrossref
11.
Hack  K, Gandhi  N, Bouchard-Fortier  G,  et al.  External validation of O-RADS US risk stratification and management system.   Radiology. 2022;304(1):114-120. doi:10.1148/radiol.211868 PubMedGoogle ScholarCrossref
12.
Lai  HW, Lyu  GR, Kang  Z, Li  LY, Zhang  Y, Huang  YJ.  Comparison of O-RADS, GI-RADS, and ADNEX for diagnosis of adnexal masses: an external validation study conducted by junior sonologists.   J Ultrasound Med. 2022;41(6):1497-1507. doi:10.1002/jum.15834 PubMedGoogle ScholarCrossref
13.
Hiett  AK, Sonek  J, Guy  M, Reid  TJ.  Performance of IOTA Simple Rules, Simple Rules risk assessment, ADNEX model and O-RADS in discriminating between benign and malignant adnexal lesions in North American population.   Ultrasound Obstet Gynecol. 2022;59(5):668-676. doi:10.1002/uog.24777 PubMedGoogle ScholarCrossref
14.
Cao  L, Wei  M, Liu  Y,  et al.  Validation of American College of Radiology Ovarian-Adnexal Reporting and Data System Ultrasound (O-RADS US): analysis on 1054 adnexal masses.   Gynecol Oncol. 2021;162(1):107-112. doi:10.1016/j.ygyno.2021.04.031 PubMedGoogle ScholarCrossref
15.
Xie  WT, Wang  YQ, Xiang  ZS,  et al. Efficacy of IOTA Simple Rules, O-RADS, and CA125 to distinguish benign and malignant adnexal masses.  J Ovarian Res. 2022;15(1):15. doi:10.1186/s13048-022-00947-9PubMedCrossref
16.
Basha  MAA, Metwally  MI, Gamil  SA,  et al.  Comparison of O-RADS, GI-RADS, and IOTA Simple Rules regarding malignancy rate, validity, and reliability for diagnosis of adnexal masses.   Eur Radiol. 2021;31(2):674-684. doi:10.1007/s00330-020-07143-7 PubMedGoogle ScholarCrossref
17.
Chen  GY, Hsu  TF, Chan  IS,  et al.  Comparison of the O-RADS and ADNEX models regarding malignancy rate and validity in evaluating adnexal lesions.   Eur Radiol. Published online May 18, 2022. doi:10.1007/s00330-022-08803-6 PubMedGoogle ScholarCrossref
18.
Wang  R, Li  X, Li  S,  et al.  Clinical value of O-RADS combined with serum CA125 and HE4 for the diagnosis of ovarian tumours.   Acta Radiol. Published online March 15, 2022. doi:10.1177/02841851221087376 PubMedGoogle ScholarCrossref
19.
Chen  H, Yang  BW, Qian  L,  et al.  Deep learning prediction of ovarian malignancy at US compared with O-RADS and expert assessment.   Radiology. 2022;304(1):106-113. doi:10.1148/radiol.211367 PubMedGoogle ScholarCrossref
20.
Solis Cano  DG, Cervantes Flores  HA, De Los Santos Farrera  O, Guzman Martinez  NB, Soria Céspedes  D. Sensitivity and specificity of ultrasonography using Ovarian-Adnexal Reporting and Data System classification versus pathology findings for ovarian cancer.  Cureus. 2021;13(9):e17646. doi:10.7759/cureus.17646PubMedCrossref
21.
Pi  Y, Wilson  MP, Katlariwala  P,  et al.  Diagnostic accuracy and inter-observer reliability of the O-RADS scoring system among staff radiologists in a North American academic clinical setting.   Abdom Radiol (NY). 2021;46(10):4967-4973. doi:10.1007/s00261-021-03193-7 PubMedGoogle ScholarCrossref
22.
Guo  Y, Zhao  B, Zhou  S,  et al.  A comparison of the diagnostic performance of the O-RADS, RMI4, IOTA LR2, and IOTA SR systems by senior and junior doctors.   Ultrasonography. 2022;41(3):511-518. doi:10.14366/usg.21237 PubMedGoogle ScholarCrossref
23.
Ahmed  HEK. The usefulness of the ultrasound diagnosis of suspicious ovarian masses based on the O-RADS classification system.  Al-Azhar Int Med J. 2021;2(10):1-6. doi:10.21608/aimj.2021.90943.1551Crossref
24.
Jha  P, Gupta  A, Baran  TM,  et al. Diagnostic performance of the Ovarian-Adnexal Reporting and Data System (O-RADS) ultrasound risk score in women in the United States.  JAMA Netw Open. 2022;5(6):e2216370. doi:10.1001/jamanetworkopen.2022.16370PubMedCrossref
25.
Froyman  W, Landolfo  C, De Cock  B,  et al.  Risk of complications in patients with conservatively managed ovarian tumours (IOTA5): a 2-year interim analysis of a multicentre, prospective, cohort study.   Lancet Oncol. 2019;20(3):448-458. doi:10.1016/S1470-2045(18)30837-4 PubMedGoogle ScholarCrossref
26.
Bossuyt  PM, Reitsma  JB, Bruns  DE,  et al; STARD Group. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies.  BMJ. 2015;351:h5527. doi:10.1136/bmj.h5527PubMedCrossref
27.
Education and Practical Standards Committee, European Federation of Societies for Ultrasound in Medicine and Biology.  Minimum training recommendations for the practice of medical ultrasound.   Ultraschall Med. 2006;27(1):79-105.PubMedGoogle ScholarCrossref
28.
Timmerman  D, Valentin  L, Bourne  TH, Collins  WP, Verrelst  H, Vergote  I; International Ovarian Tumor Analysis (IOTA) Group.  Terms, definitions and measurements to describe the sonographic features of adnexal tumors: a consensus opinion from the International Ovarian Tumor Analysis (IOTA) Group.   Ultrasound Obstet Gynecol. 2000;16(5):500-505. doi:10.1046/j.1469-0705.2000.00287.x PubMedGoogle ScholarCrossref
29.
Valentin  L.  Use of morphology to characterize and manage common adnexal masses.   Best Pract Res Clin Obstet Gynaecol. 2004;18(1):71-89. doi:10.1016/j.bpobgyn.2003.10.002 PubMedGoogle ScholarCrossref
30.
Timmerman  D, Testa  AC, Bourne  T,  et al; International Ovarian Tumor Analysis Group.  Logistic regression model to distinguish between the benign and malignant adnexal mass before surgery: a multicenter study by the International Ovarian Tumor Analysis Group.   J Clin Oncol. 2005;23(34):8794-8801. doi:10.1200/JCO.2005.01.7632 PubMedGoogle ScholarCrossref
31.
Prat  J; FIGO Committee on Gynecologic Oncology.  Staging classification for cancer of the ovary, fallopian tube, and peritoneum.   Int J Gynaecol Obstet. 2014;124(1):1-5. doi:10.1016/j.ijgo.2013.10.001 PubMedGoogle ScholarCrossref
32.
Ameye  L, Timmerman  D, Valentin  L,  et al.  Clinically oriented three-step strategy for assessment of adnexal pathology.   Ultrasound Obstet Gynecol. 2012;40(5):582-591. doi:10.1002/uog.11177 PubMedGoogle ScholarCrossref
33.
Testa  A, Kaijser  J, Wynants  L,  et al.  Strategies to diagnose ovarian cancer: new evidence from phase 3 of the multicentre international IOTA study.   Br J Cancer. 2014;111(4):680-688. doi:10.1038/bjc.2014.333 PubMedGoogle ScholarCrossref
34.
Reitsma  JB, Glas  AS, Rutjes  AW, Scholten  RJ, Bossuyt  PM, Zwinderman  AH.  Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews.   J Clin Epidemiol. 2005;58(10):982-990. doi:10.1016/j.jclinepi.2005.02.022 PubMedGoogle ScholarCrossref
35.
Leeflang  MM, Deeks  JJ, Rutjes  AW, Reitsma  JB, Bossuyt  PM.  Bivariate meta-analysis of predictive values of diagnostic tests can be an alternative to bivariate meta-analysis of sensitivity and specificity.   J Clin Epidemiol. 2012;65(10):1088-1097. doi:10.1016/j.jclinepi.2012.03.006 PubMedGoogle ScholarCrossref
38.
de Groot  JAH, Bossuyt  PMM, Reitsma  JB,  et al. Verification problems in diagnostic accuracy studies: consequences and solutions.  BMJ. 2011;343:d4770. doi:10.1136/bmj.d4770PubMedCrossref
39.
Moons  KGM, Wolff  RF, Riley  RD,  et al.  PROBAST: a tool to assess risk of bias and applicability of prediction model studies: explanation and elaboration.   Ann Intern Med. 2019;170(1):W1-W33. doi:10.7326/M18-1377 PubMedGoogle ScholarCrossref
×