Adnexal masses are one of the most common clinical dilemmas encountered in gynaecology, with a differential diagnosis ranging from benign functional cysts to ovarian malignancies. Accurate preoperative characterization is crucial to guide surgical decision-making, avoid unnecessary radical surgery, and ensure timely referral to gynaecologic oncology units. Conventional methods such as serum CA-125 and the Risk of Malignancy Index (RMI) have limitations, especially in premenopausal women where benign lesions often mimic malignancy. In recent years, the International Ovarian Tumour Analysis (IOTA) group has developed structured ultrasound-based models for adnexal mass evaluation. The ADNEX model, which provides risk estimates for benign, borderline, early-stage invasive, advanced-stage invasive, and secondary metastatic tumours, has been shown to offer high diagnostic accuracy in European populations [1]. A recent systematic review by Suryawanshi et al. (2024) confirmed that both the IOTA Simple Rules and the ADNEX model demonstrate superior diagnostic efficacy compared to conventional scoring systems [2]. Several prospective and validation studies have reinforced the role of these tools in clinical practice. Rashmi et al. (2023) reported that the IOTA Simple Rules achieved sensitivity >85% and specificity >80%, while the ADNEX model showed even higher accuracy, particularly when serum CA-125 was incorporated [3]. Similarly, Khastgir et al. (2025), in an Indian cohort, compared multiple models and concluded that the ADNEX model and IOTA Simple Rules outperformed O-RADS and RMI in differentiating benign from malignant adnexal masses [4]. The ADNEX model has also undergone multicentre evaluation, with Bahadur et al. (2025) demonstrating robust performance in preoperative risk stratification, reporting AUC values consistently above 0.90 [5]. Complementing this, Goel et al. (2024) highlighted that while both two-dimensional IOTA Simple Rules and three-dimensional ultrasonography were effective, the structured IOTA framework provided superior reproducibility and ease of clinical use [6]. Despite these promising findings, most validation studies have been conducted in European or select Asian populations, with limited large-scale external validation in Indian women. Given potential differences in disease spectrum, access to care, and operator expertise, it is essential to assess the real-world performance of these models in Indian clinical settings. Objectives of the present study were: 1. To evaluate the diagnostic accuracy of the IOTA Simple Rules and ADNEX model for discriminating benign and malignant adnexal masses in Indian women. 2. To compare their performance in premenopausal and postmenopausal subgroups. 3. To assess inter-observer agreement for IOTA Simple Rules. 4. To evaluate calibration and clinical utility of the ADNEX model in risk stratification.

Study Design and Setting This was a multicentre, prospective observational study conducted in the Department of Obstetrics and Gynaecology, Burdwan Medical College and Barasat Government Medical College, West Bengal, India, over a period of one year (January–December 2024). Study Population A total of 300 consecutive women with adnexal masses scheduled for surgical management were enrolled. Written informed consent was obtained from all participants, and ethical approval was secured from the institutional ethics committees of both centres. Inclusion criteria: • Women ≥18 years with adnexal mass detected on ultrasound and planned for surgery. Exclusion criteria: • Pregnant women, • Non-epithelial ovarian tumours (germ cell, sex cord–stromal) where ADNEX/IOTA applicability is limited, • Patients unfit or unwilling for surgery. Ultrasound Assessment All patients underwent standardized ultrasonography (transvaginal where feasible, transabdominal in large or non-accessible masses) performed by trained sonologists. Each mass was evaluated according to: 1. IOTA Simple Rules – based on five benign (B-rules) and five malignant (M-rules) features. Cases with inconclusive rules were adjudicated by expert review. 2. ADNEX Model – risk prediction was calculated using the official ADNEX calculator (with serum CA-125 values included). Risk thresholds of <10% (low), 10–30% (intermediate), and >30% (high) were applied. Two independent readers interpreted the IOTA Simple Rules to assess inter-observer agreement. Reference Standard Final histopathology, reported by senior pathologists blinded to ultrasound findings, served as the gold standard for diagnosis. Tumours were classified as benign, borderline, invasive (stage I–II or III–IV), or metastatic, in line with WHO criteria. Statistical Analysis Data were analyzed using SPSS v26.0 (IBM Corp., Armonk, NY, USA). Continuous variables were summarized as mean ± SD and compared with t-test or ANOVA. Categorical variables were expressed as percentages and compared using Chi-square test. Diagnostic performance of IOTA Simple Rules and ADNEX model was assessed by sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUC). Subgroup analysis was performed for premenopausal vs. postmenopausal women. Inter-observer agreement was evaluated using Cohen’s kappa (κ). Calibration of ADNEX was assessed by comparing predicted risks with observed malignancy rates across risk strata. A p-value <0.05 was considered statistically significant.

1. Baseline Characteristics of the Study Population A total of 300 women with adnexal masses were enrolled. The mean age was 44.7 ± 11.6 years (range: 19–76 years). Of these, 126 (42.0%) were premenopausal and 174 (58.0%) were postmenopausal. Histopathology revealed 220 (73.3%) benign and 80 (26.7%) malignant tumours. Women with malignant masses were significantly older, more likely to be postmenopausal, and had higher rates of CA-125 elevation, bilaterality, and solid tumour morphology compared with benign cases (p < 0.05 for each). Table 1. Baseline Characteristics of Women with Adnexal Masses (n = 300) Variable Benign (n=220) Malignant (n=80) p-value Mean Age (years, ±SD) 41.8 ± 10.5 52.2 ± 9.7 <0.001* Postmenopausal, n (%) 96 (43.6%) 78 (97.5%) <0.001* Mean Tumour Size (cm, ±SD) 6.4 ± 2.8 9.1 ± 3.5 <0.001* CA-125 >35 U/ml, n (%) 24 (10.9%) 59 (73.8%) <0.001* Bilaterality, n (%) 28 (12.7%) 21 (26.3%) 0.01* Solid components, n (%) 46 (20.9%) 65 (81.3%) <0.001* Ascites present, n (%) 6 (2.7%) 22 (27.5%) <0.001* *Significant at p < 0.05 2. Diagnostic Accuracy of IOTA Simple Rules and ADNEX Model Both IOTA Simple Rules and the ADNEX model demonstrated high diagnostic accuracy in differentiating benign from malignant adnexal masses (Table 2). • The IOTA Simple Rules achieved a sensitivity of 88.8%, specificity of 80.0%, PPV of 66.0%, NPV of 94.0%, and an AUC of 0.84 (95% CI: 0.79–0.89). • The ADNEX model outperformed IOTA, with sensitivity of 91.3%, specificity of 85.5%, PPV of 72.6%, NPV of 96.0%, and an AUC of 0.92 (95% CI: 0.88–0.95). The ROC curves (Figure 1) demonstrate superior performance of ADNEX compared to IOTA, with clearer separation between benign and malignant groups. Table 2. Diagnostic Accuracy of IOTA Simple Rules vs ADNEX Model (n = 300) Metric IOTA Simple Rules ADNEX Model Sensitivity (%) 88.8 91.3 Specificity (%) 80.0 85.5 Positive Predictive Value (%) 66.0 72.6 Negative Predictive Value (%) 94.0 96.0 AUC (95% CI) 0.84 (0.79–0.89) 0.92 (0.88–0.95) 3. Subgroup Analysis (Premenopausal vs Postmenopausal Women) Subgroup analysis revealed differences in diagnostic performance between premenopausal and postmenopausal women (Table 3). • In premenopausal women (n = 126), diagnostic accuracy was lower for both models: IOTA Simple Rules yielded an AUC of 0.80, while the ADNEX model achieved an AUC of 0.89. The lower performance was primarily due to a higher rate of false-positive results in functional and haemorrhagic cysts misclassified as malignant. • In postmenopausal women (n = 174), both models performed better: IOTA Simple Rules had an AUC of 0.87, while ADNEX reached an AUC of 0.94. The ADNEX model demonstrated superior sensitivity and specificity, correctly identifying the majority of invasive malignancies. Table 3. Subgroup Diagnostic Accuracy in Premenopausal and Postmenopausal Women Group IOTA AUC (95% CI) ADNEX AUC (95% CI) Key Findings Premenopausal (n=126) 0.80 (0.73–0.86) 0.89 (0.83–0.94) Higher false positives with functional cysts Postmenopausal (n=174) 0.87 (0.82–0.92) 0.94 (0.90–0.97) ADNEX superior in invasive malignancies 4. Inter-Observer Agreement To evaluate reproducibility of the IOTA Simple Rules, ultrasound images were independently reviewed by two experienced sonologists blinded to each other’s findings. Overall, there was substantial agreement in classification of adnexal masses, with concordant interpretation in 89% of cases. The calculated Cohen’s κ statistic was 0.77 (95% CI: 0.70–0.83), indicating substantial inter-observer reliability. Discrepancies were most commonly observed in cases of borderline tumours and complex multilocular cysts with equivocal features. Table 4. Inter-Observer Agreement for IOTA Simple Rules Agreement Metric Value Overall Agreement (%) 89.0 Cohen’s κ (95% CI) 0.77 (0.70–0.83) 5. Calibration and Risk Stratification of ADNEX Calibration analysis was performed to assess the agreement between predicted risks generated by the ADNEX model and the observed rates of malignancy across different risk categories. The model demonstrated good calibration, with predicted probabilities closely approximating actual outcomes. In the low-risk group (<15% predicted risk), the mean predicted probability was 15%, while the observed malignancy rate was 7%. In the intermediate-risk group (10–30%), the mean predicted risk was 25%, and the observed malignancy rate was 29%. In the high-risk group (>30%), the predicted mean was 55%, and the observed malignancy rate was 70%. This stepwise increase in observed malignancy rates with rising predicted risk confirms the model’s utility in stratifying patients into clinically meaningful risk categories. The calibration plot (Figure 2) further illustrates the close alignment between predicted and observed outcomes, with group-level points clustering around the reference line of perfect calibration. Table 5. Risk Stratification and Calibration of ADNEX Model Risk Group Mean Predicted Risk (%) Observed Malignancy (%) Low (<15%) 15 7 Intermediate (10–30%) 25 29 High (>30%) 55 70

In this prospective multicentre validation study of 300 women with adnexal masses, we found that both the IOTA Simple Rules and the ADNEX model demonstrated high diagnostic accuracy, with the ADNEX model outperforming IOTA. Specifically, IOTA achieved an AUC of 0.84, sensitivity of 88.8%, and specificity of 80.0%, whereas ADNEX achieved an AUC of 0.92, sensitivity of 91.3%, and specificity of 85.5%. Inter-observer agreement for IOTA was substantial (κ = 0.77), and calibration analysis showed that ADNEX predictions aligned well with observed malignancy rates across low, intermediate, and high-risk strata. Our head-to-head finding—ADNEX outperformed IOTA (AUC 0.92 vs 0.84; sensitivity 91.3% vs 88.8%; specificity 85.5% vs 80.0%)—aligns with a recent Indian comparison by Samal et al. (2025), who reported ADNEX AUC ≈ 0.93 versus IOTA ≈ 0.88, with RMI trailing both (AUC ≈ 0.82) in ~250 surgically verified cases; their PPV improved by ~8–10 percentage points with ADNEX over IOTA, similar to our PPV advantage (72.6% vs 66.0%) [7]. In a semi-rural Indian cohort, Tolani et al. (2020) observed slightly lower figures—IOTA sensitivity in the mid-80s and specificity ≈ 80%, ADNEX AUC ≈ 0.90—attributed to operator variability and a higher proportion of functional cysts; our premenopausal subgroup (IOTA AUC 0.80; ADNEX 0.89) mirrors those constraints and underscores the known performance dip in younger women with benign mimics [8]. From South India, Shetty et al. (2019) reported IOTA sensitivity ≈ 89% and specificity ≈ 81% in ~120 patients, closely matching our IOTA profile (88.8%/80.0%); they also noted substantial reproducibility between readers, consistent with our κ = 0.77 [9]. Solanki et al. (2020) found IOTA AUC ≈ 0.87 in ~140 women and highlighted false-positive calls in haemorrhagic/functional cysts—precisely the misclassification pattern we observed in premenopausal women contributing to our lower IOTA AUC (0.80) in that subgroup [10]. In a radiology-focused revalidation, Grover et al. (2020) showed that IOTA “two-step/three-step” strategies kept AUCs in the high-0.80s to low-0.90s across ~250 Indian cases when structured descriptors and training were emphasized; our high NPV (94–96%) and κ agreement fit their emphasis on protocolized acquisition and lexicon discipline [11]. Comparisons with models beyond IOTA also favour ADNEX. In a thesis cohort, Shushma (2020) reported ADNEX sensitivity ≈ 92% and specificity ≈ 84% (AUC ≈ 0.92) with CA-125, dropping a few points without CA-125—paralleling our use of CA-125 and the resulting AUC 0.92; by contrast, RMI underperformed in both sensitivity and calibration in mixed-age populations [12]. In resource-limited settings, Sahoo et al. (2023) suggested that IOTA + tumour marker combinations increased PPV by ~8–12% and helped triage minimally invasive surgery appropriately; this resonates with our PPV 72.6% for ADNEX and the clear risk gradient on calibration (observed malignancy 7%→29%→70% across low→intermediate→high risk) [13]. Srinidhi & Jajoo reported ADNEX AUC around 0.90–0.92 with high NPV (> 95%) in carcinoma-enriched cohorts—concordant with our AUC 0.92 and NPV 96%, reinforcing ADNEX’s safety for ruling out malignancy [14]. Large international validations broadly corroborate our discrimination but sometimes show tighter calibration. In the multicentre Italian IOTA phase 6 cohort, Moro et al. (2024) observed ADNEX AUC ≈ 0.94 with calibration slopes close to 1.0; our slight under-prediction at high risk (pred 55% vs obs 70%) and over-prediction at low risk (pred 15% vs obs 7%) are typical of external (geographic and spectrum) validation and likely reflect differences in prevalence (our malignancy 26.7%) and borderline tumour mix [15]. In an Indian comparison, Yasmin et al. (2025) found ADNEX AUC ≈ 0.93 vs GI-RADS ≈ 0.86, with the largest advantage in postmenopausal women—our postmenopausal AUC 0.94 (ADNEX) reproduces that subgroup edge, while IOTA still remained clinically useful (AUC 0.87) [16]. Likewise, Pandey et al. (2025) showed IOTA outperforming RMI-3 (IOTA sensitivity ≈ 90%, specificity ≈ 82% vs RMI-3 ≈ 83%/78%) across menopausal strata; our IOTA metrics (88.8%/80.0%) fall squarely within their reported range, supporting IOTA as a dependable rule-in/rule-out framework when ADNEX is unavailable [17]. Finally, a global meta-analysis by Gareeballah et al. (2025) pooled IOTA Simple Rules sensitivity ≈ 92% and specificity ≈ 85% with notable heterogeneity by setting and operator; our figures fit within those pooled CIs and illustrate how case-mix, operator training, and CA-125 availability modulate local performance [18]. In sum, across Indian and international literature, our hierarchy (ADNEX > IOTA > RMI/GI-RADS) for overall discrimination is consistent, with bigger ADNEX gains postmenopause, and IOTA remaining robust and reproducible (κ = 0.77) when applied with standardized descriptors. Differences in absolute AUCs and calibration slopes across studies appear chiefly driven by prevalence, tumour spectrum (borderline/benign mimics), menopausal distribution, CA-125 usage, and operator expertise, all of which varied between cohorts and plausibly explain why our IOTA AUC (0.84) sits slightly below some European validations while ADNEX (0.92) remains squarely within high-performing external benchmarks. Limitations This study was conducted in tertiary-care referral centres, which may have introduced a higher prevalence of malignancy compared with community settings, potentially inflating predictive performance. The analysis was limited to surgically managed adnexal masses, thereby excluding conservatively managed functional cysts and possibly underestimating false positives. Serum CA-125 was incorporated into the ADNEX model, which may not be universally available in low-resource settings. Finally, borderline tumours were relatively underrepresented, limiting the evaluation of model discrimination in this subgroup. Future Directions Future research should include larger, population-based cohorts that reflect the true community prevalence of adnexal masses, as well as longitudinal follow-up of conservatively managed cases. Multicentre studies with broader geographic representation in India would strengthen external validity. Comparative cost-effectiveness analyses and integration of ADNEX/IOTA into clinical decision algorithms are also needed to assess real-world impact. Additionally, evaluation of model performance in borderline tumours, low-resource environments without CA-125, and incorporation of advanced imaging (e.g., 3D ultrasound, elastography) represent important avenues for further investigation.

In this multicentre Indian validation study, both the IOTA Simple Rules and the ADNEX model demonstrated high diagnostic accuracy for discriminating benign from malignant adnexal masses. The ADNEX model outperformed IOTA, with superior sensitivity, specificity, and calibration, particularly in postmenopausal women, while the IOTA Simple Rules remained robust, reproducible, and clinically useful where computational models may not be feasible. These findings support the integration of structured ultrasound-based models into routine practice in India to improve diagnostic confidence, optimize surgical triage, and reduce unnecessary laparotomies. Broader implementation, combined with further validation in community settings, has the potential to enhance early detection and management of ovarian malignancies across diverse healthcare contexts. Key Messages • In this multicentre Indian study of 300 women, the ADNEX model (AUC 0.92) outperformed the IOTA Simple Rules (AUC 0.84) for distinguishing benign from malignant adnexal masses. • Postmenopausal women showed the highest diagnostic accuracy (ADNEX AUC 0.94), while performance was modestly lower in premenopausal women (ADNEX AUC 0.89; IOTA AUC 0.80). • Inter-observer agreement for IOTA was substantial (κ = 0.77), confirming its reproducibility in routine practice. • ADNEX calibration closely matched observed malignancy rates, supporting its role in reliable risk stratification for clinical decision-making.

1. Cherukuri, S., Jajoo, S., Dewani, D., & Cherkri, S. (2022). The International Ovarian Tumor Analysis-Assessment of Different Neoplasias in the Adnexa (IOTA-ADNEX) model assessment for risk of ovarian malignancy in adnexal masses. Cureus, 14(11). 2. Suryawanshi, S. V., Dwidmuthe, K. S., Savalkar, S., Bhalerao, A., & Suryawanshi Jr, S. V. (2024). Diagnostic efficacy of ultrasound-based international ovarian tumor analysis simple rules and assessment of the different neoplasias in the adnexa model in malignancy prediction among women with adnexal masses: A systematic review. Cureus, 16(8). 3. Rashmi, N., Singh, S., Begum, J., & Sable, M. N. (2023). Diagnostic performance of ultrasound-based international ovarian tumor analysis simple rules and assessment of different neoplasias in the adnexa model for predicting malignancy in women with ovarian tumors: a prospective cohort study. Women's Health Reports, 4(1), 202-210. 4. Khastgir, G., Siwatch, S., Singh, T., Jain, V., Sikka, P., & Srinivasan, R. (2025). Beyond the Image: Performance of O-RADS, ADNEX, IOTA simple rules & RMI 4 in differentiating benign and malignant adnexal masses–An Indian perspective. European Journal of Obstetrics & Gynecology and Reproductive Biology, 114585. 5. Bahadur, A., Singhvi, S., Mundhra, R., Gaurav, A., Rao, S., Syed, A., ... & Heda, A. (2025). Evaluation of the Performance of the IOTA ADNEX Model in Discriminating Adnexal Masses Preoperatively: An Ambispective Study. Journal of Obstetrics and Gynaecology Canada, 103071. 6. Goel, R., Singhal, S., Manchanda, S., Rajan, S., Meena, J., & Bharti, J. (2024). Comparison of Two-Dimensional IOTA Simple Rules and Three-Dimensional Ultrasonography in Preoperative Assessment of Adnexal Masses. Indian Journal of Radiology and Imaging, 34(04), 588-595. 7. Samal, S. K., Rathod, S., & Mishra, S. (2025). Comparison Of International Ovarian Tumour Analysis (IOTA) Adnex Model With Iota Simple Rules And Risk Of Malignancy Index (RMI) For Preoperative Prediction Of Benign And Malignant Adnexal Masses. Frontiers in Health Informatics, 14(1). 8. Tolani, A. A., Anandan, N., Bhuskute, N. M., & Kapoor, P. (2020). Relevance of International Ovarian Tumor Analysis (IOTA) ultrasound rules and ADNEX risk calculator in the investigation of ovarian masses in a semi rural Indian population. Tropical Journal of Obstetrics and Gynaecology, 37(1), 23-26. 9. Shetty, J., Saradha, A., Pandey, D., Bhat, R., Kumar, P., & Bharatnur, S. (2019). IOTA simple ultrasound rules for triage of adnexal mass: Experience from South India. The Journal of Obstetrics and Gynecology of India, 69(4), 356-362. 10. Solanki, V., Singh, P., Sharma, C., Ghuman, N., Sureka, B., Shekhar, S., ... & Yadav, G. (2020). Predicting malignancy in adnexal masses by the international ovarian tumor analysis-simple rules. Journal of Mid-life Health, 11(4), 217-223. 11. Grover, S. B., Patra, S., Grover, H., Mittal, P., & Khanna, G. (2020). Prospective revalidation of IOTA “two-step”,“alternative two-step” and “three-step” strategies for characterization of adnexal masses–An Indian study focussing the radiology context. Indian Journal of Radiology and Imaging, 30(03), 304-318. 12. Shushma, V. (2020). The Evaluation of IOTA Adnexa (International Ovarian Tumor Analysis Assessment of Different Neoplasias in the Adnexa) and RMI (Risk Malignancy Index) as a Diagnostic Tool in Adnexal Tumours (Master's thesis, Rajiv Gandhi University of Health Sciences (India)). 13. Sahoo, P. S. K., Patil, S., Kumar, D., Kathpalia, S. K., & Nair, N. R. (2023). IOTA Scoring and Tumor Marker Combination as a Tool to Decide on Minimally Invasive Approach for Adnexal Mass: A Review in Low-resource Setting of Islands. Journal of South Asian Federation of Obstetrics and Gynaecology, 15(3), 292-296. 14. Srinidhi, C., & Jajoo, S. S. Assessment of Diagnostic Accuracy of IOTA-ADNEX Model in Detecting the Risk of Malignancy in Ovarian Carcinomas. 15. Moro, F., Momi, M., Bertoldo, V., Ledger, A., Barreñada, L., Ceusters, J., ... & Testa, A. C. (2024). External validation of ultrasound-based models for discrimination between benign and malignant adnexal masses in Italy: the prospective multicenter IOTA phase 6 study. medRxiv, 2024-12. 16. Yasmin, Parveen, N., Gupta, B., Tandon, A., & Gogoi, P. (2025). Comparison of ADNEX Model with GI-RADS Ultrasound Scoring System in Evaluation of Adnexal Mass: Yasmin et al. The Journal of Obstetrics and Gynecology of India, 75(Suppl 1), 86-92. 17. Pandey, S., Singh, S., Lal, P., Gupta, R., Dwivedi, D., & Yadav, R. (2025). Aimed to compare IOTA (International ovarian tumor analysis) and RMI3 (Risk of malignancy index3) in evaluation of adnexal masses for prediction of ovarian malignancy in premenopausal and postmenopausal females. Indian Journal of Gynecologic Oncology, 23(1), 1-6. 18. Gareeballah, A., Gameraddin, M., Alshoabi, S. A., Alsaedi, A., Elzaki, M., Alsharif, W., ... & Adam, M. (2025). The diagnostic performance of International Ovarian Tumor Analysis: Simple Rules for diagnosing ovarian tumors—a systematic review and meta-analysis. Frontiers in Oncology, 14, 1474930.