基于超声的预测模型预测腹腔镜子宫内膜异位症手术中输尿管松解术的需求
本研究为回顾性观察研究,纳入177例接受经阴道超声(TVS)及后续腹腔镜±子宫内膜异位症切除的患者,构建决策树模型预测输尿管松解术需求。最终模型纳入排便困难、卵巢活动度缺失、左右宫骶韧带深部子宫内膜异位症及肠道深部子宫内膜异位症,模型曲线下面积0.82,基线风险20%。
要点
- 左卵巢固定者87.5%需输尿管松解术(p<.001),右卵巢固定者82.5%(p<.001)。
- 右宫骶韧带深部内异症者96.2%需输尿管松解术(p<.001),左宫骶韧带者64.6%(p=.043)。
- 肠道深部内异症者95.5%需输尿管松解术(p<.001)。
核心结论
- 临床与超声数据可预测腹腔镜子宫内膜异位症手术中输尿管松解术的需求。
- 具备模型变量组合(排便困难、卵巢活动度缺失、宫骶韧带及肠道深部内异症病灶)的患者输尿管松解术风险高。
- 无这些特征的患者需输尿管松解术的风险低,约20%。
研究设计与样本
- 回顾性观察研究,纳入2017年1月至2021年2月期间接受经阴道超声(按国际深部子宫内膜异位症分析共识)及后续腹腔镜±子宫内膜异位症切除的177例连续患者。
- 研究地点为澳大利亚新南威尔士州悉尼大学Nepean医学院、Nepean医院及Blue Mountains医院。
- 通过电子临床记录提取人口学、临床、TVS及术中数据,构建多分类决策树和基线模型。
主要发现
- 177例患者中,输尿管松解术组占51.4%,非输尿管松解术组占48.6%。
- 左卵巢固定者中87.5%行输尿管松解术(p<.001),右卵巢固定者中82.5%(p<.001)。
- 右宫骶韧带深部内异症者中96.2%行输尿管松解术(p<.001),左宫骶韧带深部内异症者中64.6%(p=.043)。
- 肠道深部内异症者中95.5%行输尿管松解术(p<.001)。
- 最终模型纳入排便困难、卵巢活动度缺失、左右宫骶韧带深部内异症及肠道深部内异症,基线风险20%,AUC 0.82。
关键数值与推荐
- 模型预测变量:排便困难、卵巢活动度缺失、右或左宫骶韧带深部内异症、肠道深部内异症(TVS所见)。
- 模型整体性能:受试者工作特征曲线下面积0.82。
- 样本基线输尿管松解术风险为20%。
临床怎么用
- 作者认为,临床与超声数据可用于预测输尿管松解术需求。
- 作者指出,具备模型变量组合的患者输尿管松解术风险高,无这些特征者风险低(约20%)。
局限性
- 原文未报告局限性。
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 a model, including clinical features and ultrasound findings, to predict the need for ureterolysis (i.e., dissection of the ureter) during laparoscopy for endometriosis. DESIGN: A retrospective observational study of patients who had undergone transvaginal ultrasound (TVS) according to the International Deep Endometriosis Analysis consensus and subsequent laparoscopy ± excision of endometriosis between January 2017 and February 2021 was conducted. SETTING: Sydney Medical School Nepean, University of Sydney, Nepean Hospital, and Blue Mountains Hospital, New South Wales, Australia. PATIENTS/PARTICIPANT: 177 patients. INTERVENTION: The demographic, clinical, TVS, and intraoperative data were extracted through electronic clinical records. MEASUREMENTS AND MAIN RESULTS: Multicategorical decision-tree and baseline models were built to choose the variables most correlated to the outcome under study. Receiver operating characteristic analysis was performed on the binary classification. Based on our results, we selected the variables performing with significant statistical differences (p <.05). During the study period, 177 consecutive patients were recruited and divided into 2 subgroups, ureterolysis (51.4%) and nonureterolysis (48.6%). Ureterolysis was noted in 87.5% of patients in which the left ovary was immobile (p <.001) and in 82.5% in which the right ovary was fixed (p <.001). For patients with right uterosacral ligament (USL) deep endometriosis (DE), ureterolysis was performed in 96.2% patients (p <.001) and 64.6% (p = .043) for left USL DE. Among patients with bowel DE, the proportion of patients undergoing ureterolysis was 95.5% (p <.001). The prognostic variables used in the final model to predict ureterolysis included dyschezia, absence of ovarian mobility, presence of right or left USL DE, and presence of bowel DE on TVS. According to the developed model, the baseline risk for performing ureterolysis is 20% in our sample. The overall model performance demonstrated an area under the receiver operating characteristic curve 0.82. CONCLUSION: Our study demonstrates that it is possible to predict the need for ureterolysis with clinical and sonographic data. Furthermore, patients presenting with a combination of the variables of our model (dyschezia, ovarian immobility, USL, and bowel DE lesions) have a high risk of ureterolysis. In contrast, patients without these features have a low risk (approximately 20%) of needing ureterolysis.
引用
- Zanardi JVC, Rocha RM, Leonardi M, et al. An Ultrasound-based Prediction Model to Predict Ureterolysis during Laparoscopic Endometriosis Surgery. J Minim Invasive Gynecol. 2022. PMID: 35817365. doi:10.1016/j.jmig.2022.07.002
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。