摘要
目的:基于临床和MRI特征构建预测侵袭性胎盘植入和严重产后出血的列线图模型并验证.方法:对2019年1月1日—2022年12月31日在深圳市妇幼保健院因产前检查疑有胎盘前置或胎盘植入性疾病(PAS)而行MRI检查的236例孕晚期患者的临床和MRI资料进行回顾性分析.按8︰2的比例将患者随机分入训练集(189例)或验证集(47例).通过单因素和多因素logistics回归分析筛选临床和MRI特征,分别构建侵袭性胎盘植入和严重产后出血的列线图预测模型.采用受试者工作特征(ROC)曲线、校准曲线和决策曲线分析(DCA)评估预测模型的区分度、校准度和诊断效能.结果:剖宫产史(OR=2.835,P=0.022)、胎盘局部膨隆(OR=3.526,P=0.044)、胎盘-子宫界面不连续(OR=7.97,P=0.006)、胎盘床异常血管(OR=4.16,P=0.013)和胎盘陷窝流空信号(OR=6.229,P=0.006)是预测侵袭性胎盘植入的独立危险因素,联合上述征象构建预测模型并绘制其列线图,其在训练集和验证集中的AUC分别为0.955(95％CI:0.928～0.983)和0.866(95％CI:0.752～0.980).胎盘局部膨隆(OR=4.033,P=0.033)、胎盘床异常血管(OR=4.585,P=0.01)、胎盘内异常血管(OR=3.684,P=0.034)和胎盘陷窝流空信号(OR=8.154,P=0.002)是预测严重产后出血的独立危险因素,联合上述征象构建严重产后出血预的测模型并绘制其列线图,其在训练集和验证集中的AUC分别为0.945(95％CI:0.912～0.977)和0.899(95％CI:0.806～0.993).结论:基于临床和MRI特征构建的预测侵袭性胎盘植入和严重产后出血列线图预测模型对临床诊治能提供一定帮助.
Abstract
Objective:This study was aimed to develop and validate nomogram models for predic-ting invasive placenta accreta and massive postpartum hemorrhage based on clinical and MRI charac-teristics.Methods:From January 1,2019 to December 31,2022,the clinical and MRI data of 236 late-pregnant patients in our hospital,who underwent MRI for suspected placenta previa or placenta accreta spectrum (PAS),were retrospectively analyzed.All the cases were randomly allocated into a training set (189 cases)and a validation set (47 cases)in an ratio of 8︰2.Univariate and multivariate logistic regression analyses were used to identify clinical and MRI characteristics associated with invasive pla-centa accreta and massive postpartum hemorrhage (≥1000mL).Subsequently,nomogram prediction models were established.The models were validated using the validation set data,and their perfor-mance was assessed using receiver operating characteristic (ROC)curves,calibration curves,and deci-sion curve analyses (DCA).Results:The history of cesarean section (OR=2.835,P=0.022),placental bulge (OR=3.526,P=0.044),discontinuity of the uteroplacental interface (OR=7.97,P=0.006),abnormal vascularization of the placental bed (OR=4.16,P=0.013)and flow void signals in the pla-cental lacunae (OR=6.229,P=0.006)were identified as independent risk factors for invasive placen-ta accreta.A predictive model incorporating these factors was developed,and its corresponding nomo-gram was plotted.The area under the curve (AUC)of the model was 0.955 (95％CI:0.928～0.983) in the training set and 0.866 (95％CI:0.752～0.980)in the validation set.For predicting massive postpartum hemorrhage,independent risk factors included placental bulge (OR=4.033,P=0.033),abnormal vascularization of the placental bed (OR=4.585,P=0.01),abnormal intraplacental vascu-larity (OR=3.684,P=0.034),and flow void signals in the placental lacunae (OR=8.154,P=0.002) were independent risk factors for predicting massive postpartum hemorrhage.A predictive model for massive postpartum hemorrhage was constructed using these factors,and its nomogram was plotted,with AUCs of 0.945 (95％CI:0.912～0.977)in the training set and 0.899 (95％CI:0.806～0.993)in the validation set.Conclusion:The nomogram prediction models of invasive placenta accreta and mas-sive postpartum hemorrhage,based on clinical and MRI characteristics,are beneficial for clinical diag-nosis and treatment.
NETL
NSTL
万方数据