IOTA Logistic回归模型2对良性与非良性附件肿物的鉴别诊断价值

Diagnostic performance of the IOTA Logistic regression model 2 for differentiating benign from non-benign adnexal masses

摘要

目的 评价IOTA Logistic回归模型2 （IOTA Logistic regression model 2,IOTA·LR2）在附件肿物良性与非良性（恶性+交界性）鉴别诊断中的效能,并比较其在绝经前与绝经后人群中的表现。方法 回顾性纳入河北医科大学第二医院2019年7月—2020年6月经手术及病理确诊的卵巢/附件肿物患者198例（其中非良性66例）,收集术前IOTA·LR2输出的风险值（LR2%）,以术后病理诊断为金标准。以LR2%≥10%判定为非良性,计算敏感度、特异度、阳性预测值（positive predictive value,PPV）、阴性预测值（negative predictive value,NPV）及似然比,并绘制受试者工作特征（receiver operating characteristic,ROC）曲线评价判别能力;分别在绝经前与绝经后亚组进行分析。结果 IOTA·LR2的敏感度为90.91%,特异度为85.61%,曲线下面积（area under the curve,AUC）为0.917（95%CI：0.875~0.960）。Kappa值为0.729（P<0.001）,提示与病理诊断具有较强一致性。亚组分析显示,绝经前患者敏感度85.00%、特异度89.42%（Kappa=0.641;AUC=0.893,95%CI：0.799~0.986）;绝经后患者敏感度93.48%、特异度71.43%（Kappa=0.673;AUC=0.847,95%CI：0.745~0.948）。总体PPV为75.95%,NPV为94.96%。结论 IOTA·LR2诊断模型对附件肿物良性与非良性的鉴别诊断具有较高诊断价值,敏感度较高、漏诊风险较低;但在绝经后人群中特异度相对下降,临床上对“阳性（≥10%）”结果应结合肿瘤标志物及其他影像学征象综合判断,以减少假阳性导致的过度干预风险。

Abstract

Objective To evaluate the diagnostic performance of the IOTA Logistic regression model 2 （IOTA-LR2） for differentiating benign from non-benign （malignant + borderline） adnexal masses, and to compare its performance between premenopausal and postmenopausal women. Methods This retrospective study included 198 patients with ovarian/adnexal masses confirmed by surgery and pathology at the Second Hospital of Hebei Medical University from July 2019 to June 2020 （including 66 cases of non-benign masses）. Preoperative IOTA-LR2 risk estimates （LR2%） were collected, with postoperative pathology as the reference standard. Using a predefined cutoff of LR2% ≥10% to classify a mass as non-benign, we calculated sensitivity, specificity, positive predictive value （PPV）, negative predictive value （NPV）, and likelihood ratios. Receiver operating characteristic （ROC） analysis was performed to assess discrimination, and subgroup analyses were conducted in premenopausal and postmenopausal women. Results IOTA-LR2 achieved a sensitivity of 90.91% and a specificity of 85.61%, with an area under the curve （AUC） of 0.917 （95%CI： 0.875―0.960）. The kappa statistic was 0.729 （P<0.001）, indicating strong agreement with the pathological diagnosis. In subgroup analyses, the premenopausal group showed a sensitivity of 85.00% and a specificity of 89.42% （Kappa=0.641; AUC=0.893, 95%CI： 0.799―0.986）, whereas the postmenopausal group showed a sensitivity of 93.48% and a specificity of 71.43% （Kappa=0.673; AUC=0.847, 95%CI： 0.745―0.948）. The overall PPV was 75.95% and the NPV was 94.96%. Conclusion The IOTA-LR2 diagnostic model demonstrates good performance for differentiating benign from non-benign adnexal masses, with high sensitivity and a low risk of missed diagnosis; however, specificity is relatively lower in postmenopausal women. Therefore, in dinical practice, positive （≥10%） results should be interpreted in conjunction with serum biomarkers and other imaging findings to reduce unnecessary interventions due to false positives.

关键词

子宫附件 / IOTA·LR2 / 诊断效能

Key words

adnexa uteri / IOTA-LR2 / diagnostic performance