基于血液的无创检测助力子宫内膜异位症早期诊断与临床管理
本研究为多中心病例对照研究,纳入298名育龄女性,通过检测3种微小RNA、3种蛋白标志物、1种类固醇激素及年龄和BMI,用随机森林机器学习模型分类子宫内膜异位症。独立验证队列中AUC达0.944,敏感度0.80,特异度0.975,且在各月经周期阶段表现一致,并能识别61.5%影像学漏诊的病例。结论认为该无创血液检测可跨月经周期准确检测子宫内膜异位症,具有补充诊断价值。
要点
- 独立验证队列中AUC 0.944,敏感度0.80,特异度0.975。
- 增殖期AUC 0.935,分泌期AUC 0.993,跨周期表现稳定。
- 可识别61.5%经组织学确诊但影像学漏诊的病例。
核心结论
- 基于多组学血液标志物与机器学习的无创检测可跨月经周期准确检测子宫内膜异位症,并提供补充诊断价值。
研究设计与样本
- 多中心病例对照研究,采用机器学习分类并设独立队列验证。
- 共纳入298名育龄女性,训练队列218例(137例阳性、81例对照),独立回顾性验证队列80例(40例阳性、40例对照)。
主要发现
- 独立验证队列中AUC为0.944,敏感度0.80,特异度0.975。
- 增殖期样本AUC 0.935,敏感度0.767,特异度0.962;分泌期样本AUC 0.993,敏感度0.90,特异度1.00。
- 与经阴道超声和/或MRI相比,血液检测识别出61.5%经组织学确诊但影像学漏诊的子宫内膜异位症病例。
临床怎么用
- 对于有症状的疑似子宫内膜异位症女性,该血液检测可作为影像学检查的补充,协助早期发现影像学漏诊的病例。
局限性
- 验证队列为回顾性且样本量较小(80例),需在前瞻性、地理和种族多样化人群中进一步验证。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
STUDY OBJECTIVE: To develop and validate a non-invasive, blood-based diagnostic assay for endometriosis that performs accurately across menstrual cycle phases and complements existing imaging modalities. DESIGN: Multicenter case-control study with machine learning classification and independent cohort validation. SETTING: Clinical and search settings involving symptomatic women evaluated for suspected endometriosis. PATIENTS: A total of 298 reproductive-age women were included. The training cohort comprised 218 participants (137 endometriosis-positive and 81 controls). A modest, independent, and retrospective validation cohort included 80 participants (40 endometriosis-positive and 40 controls). INTERVENTIONS: Peripheral blood sampling with quantification of 3 microRNAs via qPCR, 3 protein biomarkers, one steroid hormone using immunoassay, as well as the participant's age and body mass index. Biomarker data were integrated using a random forest machine learning model to classify disease status. MEASUREMENTS AND MAIN RESULTS: In the independent validation cohort, the assay achieved an area under the curve (AUC) of 0.944, sensitivity of 0.80, and specificity of 0.975. Subgroup analysis by menstrual cycle phase demonstrated consistent performance: proliferative-phase samples achieved an AUC of 0.935, sensitivity of 0.767, and specificity of 0.962, while secretory-phase samples achieved an AUC of 0.993, sensitivity of 0.90, and specificity of 1.00. Compared with transvaginal ultrasound and/or MRI, the blood-based assay identified 61.5% histologically confirmed endometriosis cases that were missed by the imaging modalities. CONCLUSION: A minimally invasive, multi-omic blood-based assay integrating molecular biomarkers with machine learning can accurately detect endometriosis across menstrual cycle phases and provides complementary diagnostic values. This approach has the potential to improve early detection and guide timely clinical intervention. A prospective validation is ongoing in geographically and ethnically diverse populations to further assess its broad clinical utility.
引用
- Wong WH, Yu Y, Hu Y, et al. Noninvasive Blood-Based Detection of Endometriosis Can Improve Standard-of-Care by Facilitating Early Diagnosis and Clinical Management Among Symptomatic Women. J Minim Invasive Gynecol. 2026. PMID: 41791702. doi:10.1016/j.jmig.2026.02.042
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。