Ovarian Cyst Management: 4 Practical Approaches Compared
How the IOTA Simple Rules, the ADNEX model and O-RADS ultrasound stratify adnexal masses to safely reduce unnecessary surgery.
One-Minute Takeaway
- Ovarian cyst management increasingly relies on structured ultrasound risk models instead of reflex surgery, and each major system is backed by meta-analytic evidence.
- Subjective assessment by an expert pooled 93% sensitivity and 89% specificity — the informal benchmark for every other tool.
- The IOTA Simple Rules pooled 92% sensitivity and 92% specificity and classified roughly 86% of masses; the IOTA ADNEX model reached an AUC near 0.93.
- O-RADS ultrasound pooled 95–97% sensitivity, and O-RADS 2 lesions carried only about 0.6% malignancy risk — a rate that favours surveillance.
- In a prospective cohort, 37% of masses were managed conservatively; 2-year invasive malignancy was 0.4% and torsion 0.4%.
Why Ovarian Cyst Management Matters
Adnexal masses are common and are frequently found incidentally on imaging performed for unrelated reasons. For decades many were removed surgically, largely because clinicians could not be confident a given mass was benign. That caution had a cost.
Across pooled surgical series, roughly three-quarters of removed adnexal masses prove benign. In one meta-analysis of nearly 11,818 lesions, 74.6% were benign at histology. A large share of operations therefore treats lesions that never threatened the patient.
Surgery is not free of harm. It carries anaesthetic and operative risk, adhesion formation, possible loss of ovarian tissue and fertility, and financial cost. For a benign cyst, those harms come with no offsetting oncological benefit.
Yet missing an ovarian malignancy is a serious error, because outcome depends heavily on stage at diagnosis and on specialist surgical care. The task of modern ovarian cyst management is to separate the many benign masses that can be watched from the few that need a gynaecological oncologist — without operating on everyone as a precaution.
The Clinical Problem in Ovarian Cyst Management
The core difficulty is overlap. Benign and malignant masses share sonographic features, so appearance alone is often ambiguous. A haemorrhagic cyst, an endometrioma or a mature teratoma can look complex, while some early cancers look deceptively simple.
No single ultrasound feature reliably separates benign from malignant. Serum markers do not close the gap either: CA-125 is raised in many benign conditions in premenopausal women, which limits its specificity as a stand-alone test.
Expert examiners resolve much of this ambiguity through pattern recognition, but that expertise is unevenly distributed. A generalist sonographer working without a subspecialist nearby needs a structured, reproducible route to a similar answer.
This is the gap that risk-stratification systems address. Each converts a defined set of ultrasound descriptors into a malignancy probability or category, so that management — discharge, surveillance or referral for surgery — follows the estimated risk rather than an examiner’s unaided impression.
Available Approaches
Four approaches dominate current practice, and this review compares them: unstructured subjective assessment by an experienced examiner, the IOTA Simple Rules, the IOTA ADNEX model, and O-RADS ultrasound. The table summarises the strongest pooled or multicentre evidence for each; study names link to the source.
| Study / source | Design | N | Approach assessed | Key result (95% CI) | Tier |
|---|---|---|---|---|---|
| Meys 2016 | Meta-analysis of diagnostic studies | 47 studies / 19,674 masses | Subjective assessment (expert) | Sensitivity 93% (92–95); specificity 89% (86–92) | V |
| Gareeballah 2025 | Meta-analysis of diagnostic studies | 27 studies / 7,841 masses | IOTA Simple Rules (expert operators) | Sensitivity 92% (89–94); specificity 92% (89–94); applicable in 85.7% | V |
| Van Calster 2014 | Prospective multicentre development/validation | 5,909 women | IOTA ADNEX model | Temporal-validation AUC 0.94 (0.93–0.95); at 10% cut-off ~96% sensitive, ~71% specific | V |
| Barreñada 2024 | Meta-analysis of external-validation studies | 47 studies / 17,007 tumours | IOTA ADNEX model (with CA-125) | Summary AUC 0.93 (0.92–0.94); 91% of validations at high risk of bias | V |
| Vara 2022 | Meta-analysis of diagnostic studies | 11 studies / 4,634 masses | O-RADS ultrasound | Sensitivity 97% (94–98); specificity 77% (68–84) | V |
| Lee 2023 | Meta-analysis of category-specific rates | 18 studies / 11,818 lesions | O-RADS ultrasound categories | Malignancy: O-RADS 2 = 0.6% (0.3–1.0); 3 = 3.9% (2.5–5.4); 4 = 43.5% (33.8–53.2); 5 = 87.3% (83.0–91.7) | V |
| Froyman 2019 (IOTA5) | Prospective cohort (interim) | 3,144 conservatively managed (of 8,519) | Conservative management of benign-appearing masses | 2-yr spontaneous resolution 20.2% (18.4–22.1); invasive malignancy 0.4% (0.1–0.6); torsion 0.4% (0.1–0.7) | V |
Subjective Assessment by an Experienced Examiner
Subjective assessment — expert pattern recognition — remains the informal benchmark against which the structured systems are judged. In a meta-analysis of 47 studies and 19,674 masses, expert subjective assessment pooled a sensitivity of 93% (95% CI 92–95%) and a specificity of 89% (95% CI 86–92%), and it outperformed the older Risk of Malignancy Index.
Its limitation is reproducibility. Performance depends on training and case exposure, so it does not transfer reliably to non-expert settings. That dependence is exactly what the rule-based and model-based systems were built to reduce.
The IOTA Simple Rules
The IOTA Simple Rules apply five ultrasound features that suggest malignancy and five that suggest a benign lesion. A mass with only benign features is called benign, one with only malignant features is called malignant, and any other combination is inconclusive.
In a meta-analysis of 27 studies and 7,841 masses, the rules pooled a sensitivity of 92% (95% CI 89–94%) and a specificity of 92% (95% CI 89–94%) when applied by experienced operators, and they returned a conclusive result in about 86% of masses. Inconclusive masses are then passed to expert assessment, a two-step strategy that has matched expert accuracy.
The IOTA ADNEX Model
The ADNEX model combines three clinical variables — age, serum CA-125 and centre type — with six ultrasound variables to output a malignancy probability. It also splits that risk across four malignant subtypes, which supports more granular counselling than a single benign-or-malignant label.
In its development and temporal validation across 5,909 women, ADNEX with CA-125 reached an AUC of 0.94 (95% CI 0.93–0.95) for benign versus malignant tumours; at a 10% risk cut-off it was roughly 96% sensitive and 71% specific. A later meta-analysis of 47 external-validation studies (17,007 tumours) reported a summary AUC of 0.93 (95% CI 0.92–0.94), though most validations carried a high risk of bias.
O-RADS Ultrasound
O-RADS, from the American College of Radiology, pairs a standardised lexicon with risk categories running from 1 (normal) to 5 (high risk). It was designed so that each category maps to a defined malignancy band and a corresponding management step.
In a meta-analysis of 11 studies and 4,634 masses, O-RADS ultrasound pooled a sensitivity of 97% (95% CI 94–98%) and a specificity of 77% (95% CI 68–84%). Category-specific pooled malignancy rates from 18 studies (11,818 lesions) tracked the intended bands: 0.6% for O-RADS 2, 3.9% for O-RADS 3, 43.5% for O-RADS 4 and 87.3% for O-RADS 5. The 2022 update refined the lexicon and now recommends that lower-risk O-RADS 3 lesions be followed with ultrasound rather than excised.
Comparative Analysis
Across the four approaches a consistent pattern emerges. Sensitivity is uniformly high — roughly 92–97% — so all four rarely miss malignancy when correctly applied. The systems differ mainly in specificity, which governs how many benign masses are flagged for surgery.
O-RADS ultrasound tends to maximise sensitivity at the expense of specificity, so it errs toward flagging lesions. Expert subjective assessment offers the highest specificity, while the IOTA Simple Rules sit between them with balanced figures. The ADNEX model adds subtype probabilities and can resolve masses that the Simple Rules leave inconclusive.
For the specific goal of avoiding surgery, the decisive numbers are the low-risk categories rather than the headline sensitivities. When a system calls a mass benign — a Simple Rules “benign” result, a low ADNEX probability, or an O-RADS 2 score — the residual malignancy risk is small enough to justify surveillance. Pooled O-RADS 2 malignancy was 0.6% (95% CI 0.3–1.0%).
The clearest surgery-avoidance evidence comes from actually following such masses. In the prospective IOTA5 cohort, 37% of masses were selected for conservative management. Among newly diagnosed masses, the 2-year cumulative incidence of spontaneous resolution was 20.2% (95% CI 18.4–22.1%).
Complications over the same period were uncommon. Invasive malignancy found at surgery was 0.4% (95% CI 0.1–0.6%), torsion 0.4% (95% CI 0.1–0.7%) and cyst rupture 0.2% (95% CI <0.1–0.4%). These low rates support watching, rather than removing, masses that a validated system classifies as benign.
A practical two-step workflow endorsed by the ESGO/ISUOG/IOTA/ESGE consensus applies the Simple Rules or benign descriptors first, then reserves expert subjective assessment or the ADNEX model for inconclusive masses. Masses that remain suspicious for malignancy are referred to a gynaecological oncology centre rather than characterised further in a general clinic.
Quality and Consistency of the Evidence
- High — discrimination of benign versus malignant by IOTA and O-RADS. Several independent meta-analyses converge on high sensitivity and similar AUCs. Strong evidence supports the diagnostic accuracy of these systems in study settings.
- Moderate — specificity and false-positive burden. Specificity varied widely between systems and settings, and between-study heterogeneity for specificity was high. Evidence suggests meaningful over-flagging of benign masses, particularly with O-RADS.
- Moderate — safety of conservative management. Supported by one large prospective cohort with 2-year interim follow-up. Evidence suggests conservative management is safe for benign-appearing masses, but longer-term and randomised data are lacking.
- Low-to-moderate — performance in non-expert hands. Accuracy is reported mostly from expert or referral centres, and most ADNEX validations carried a high risk of bias. Preliminary evidence suggests some fall-off outside expert settings.
What the Evidence Does Not Show
- It does not show that any system removes the need for clinical judgement; each was validated as an aid to an examiner, not a replacement for one.
- It does not confirm the long-term safety of conservative management beyond two years, and it provides no randomised comparison of surveillance versus surgery.
- It does not confirm equivalent accuracy in low-resource or non-expert settings, since most validations were performed at referral centres.
- It does not identify a single best system. The approaches perform comparably, and superiority for hard outcomes — operations avoided or cancers missed — is not established head-to-head.
- It does not make a high-risk category diagnostic. Because specificity is limited, a proportion of benign masses are still flagged, so a high O-RADS or ADNEX result signals the need for referral rather than confirmed cancer.
Practical Implications
The evidence supports a structured, risk-based pathway rather than reflex surgery for adnexal masses. A reproducible system — the Simple Rules, ADNEX or O-RADS — lets a generalist reach a defensible malignancy estimate and reserve surgery for masses that genuinely warrant it.
Because sensitivity is high but specificity is limited, these systems are best used to rule out malignancy and justify surveillance for low-risk masses, while a high-risk category prompts referral. A two-step approach — a rule-based first pass, then expert assessment or ADNEX for inconclusive cases — reflects how the pooled data perform best.
For a mass classified as benign, prospective data support conservative follow-up with serial ultrasound, given the low 2-year rates of malignancy and complications. Masses that are suspicious for malignancy should be referred to a gynaecological oncology centre.
These are evidence-informed options rather than directives. Individual decisions should weigh symptoms, menopausal status, fertility goals and patient preference, and drug or contrast choices should follow current local protocols.
Evidence Grade and Bottom Line
Overall grade: Moderate-to-high for diagnostic accuracy; moderate for the safety of avoiding surgery.
- Structured ultrasound risk models discriminate benign from malignant adnexal masses with high sensitivity, supported by multiple independent meta-analyses.
- The systems differ mainly in specificity, so their principal value is confidently identifying low-risk masses that can be watched.
- Prospective cohort data indicate that conservatively managing benign-appearing masses carries low 2-year malignancy and complication rates.
- No system replaces clinical judgement or an expert examiner, and long-term and non-expert-setting data remain limited.
Article Information and References
- Gareeballah A, Gameraddin M, Alshoabi SA, et al. The diagnostic performance of International Ovarian Tumor Analysis: Simple Rules for diagnosing ovarian tumors—a systematic review and meta-analysis. Front Oncol. 2025;14:1474930. DOI: https://doi.org/10.3389/fonc.2024.1474930
- Barreñada L, Ledger A, Dhiman P, et al. ADNEX risk prediction model for diagnosis of ovarian cancer: systematic review and meta-analysis of external validation studies. BMJ Med. 2024;3(1):e000817. DOI: https://doi.org/10.1136/bmjmed-2023-000817
- Vara J, Manzour N, Chacón E, et al. Ovarian-Adnexal Reporting Data System (O-RADS) for classifying adnexal masses: a systematic review and meta-analysis. Cancers (Basel). 2022;14(13):3151. DOI: https://doi.org/10.3390/cancers14133151
- Lee S, Lee JE, Hwang JA, Shin H. O-RADS US: a systematic review and meta-analysis of category-specific malignancy rates. Radiology. 2023;308(2):e223269. DOI: https://doi.org/10.1148/radiol.223269
- Meys EMJ, Kaijser J, Kruitwagen RFPM, et al. Subjective assessment versus ultrasound models to diagnose ovarian cancer: a systematic review and meta-analysis. Eur J Cancer. 2016;58:17–29. DOI: https://doi.org/10.1016/j.ejca.2016.01.007
- Van Calster B, Van Hoorde K, Valentin L, et al. 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: https://doi.org/10.1136/bmj.g5920
- 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: https://doi.org/10.1016/S1470-2045(18)30837-4
- 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: https://doi.org/10.1148/radiol.2019191150
- Strachowski LM, Jha P, Phillips CH, et al. O-RADS US v2022: an update from the American College of Radiology’s Ovarian-Adnexal Reporting and Data System US Committee. Radiology. 2023;308(3):e230685. DOI: https://doi.org/10.1148/radiol.230685
- Timmerman D, Planchamp F, Bourne T, et al. ESGO/ISUOG/IOTA/ESGE Consensus Statement on preoperative diagnosis of ovarian tumours. Facts Views Vis Obgyn. 2021;13(2):107–130. DOI: https://doi.org/10.52054/FVVO.13.2.016