基于MRI定量参数预测侵袭性胎盘植入及不良临床结局的可行性研究

Feasibility based on MRI quantitative parameters for predicting invasive placenta accreta and adverse clinical outcomes

目的:探讨基于MRI定量参数的二元logistic回归预测侵袭性胎盘植入及不良临床结局的可行性。方法:本回顾性研究选择武汉大学人民医院确诊的80例胎盘植入患者,记录各患者MRI图像上胎盘内异常增生血管直径、胎盘及子宫壁厚度、胎盘内T 2低信号带面积、子宫外突度、膀胱子宫间隙低信号带不连续长度、宫颈受侵长度。采用二元logistic回归分析分别预测胎盘植入类型和不良临床结局独立风险因素,并构建鉴别模型。采用受试者操作特征(ROC)曲线模型诊断效能评估曲线下面积(AUC)、敏感度、特异度、约登指数。回归模型稳定性采用Bootstrap进行内部验证,所得AUC值与原回归模型AUC值进行DeLong检验。结果:80例患者中位年龄为33岁(20~44岁),中位分娩孕周为36.1周(28.0~39.4周)。粘连型胎盘18例(22.5%),侵袭性胎盘植入62例(77.5%)。二元logistic回归分析结果显示,流产次数[OR:4.271(95%CI:1.226~14.876)]、胎盘内增生血管直径[OR:3.789(95%CI:1.367~10.501)]、子宫外突度[OR:1.432(95%CI:1.088~1.885)]为预测侵袭性胎盘植入的独立危险因素(P均<0.05);胎盘内增生血管直径[OR:0.053(95%CI:0.009~0.314)]、胎盘内T 2低信号带面积[OR:1.016(95%CI:1.004~1.029)]、子宫外突度[OR:0.839(95%CI:0.727~0.968)]为预测不良临床结局的独立危险因素,P均<0.05。模型预测粘连型胎盘与侵袭性胎盘植入、是否发生不良临床结局的准确率均为91.3%。回归模型粘连侵袭性胎盘植入的AUC值为0.962(95%CI:0.923~1.000,敏感度为94.4%,特异度为85.5%);预测不良临床结局的AUC值为0.952(95%CI:0.908~0.996,敏感度为98.0%,特异度为82.8%)。Bootstrap分析显示胎盘植入类型和不良临床结局模型稳定性均较好。结论:基于MRI定量参数的二元logistic回归模型对于预测侵袭性胎盘植入及不良临床结局具有较好的价值,提高了整体诊断效能。

Objectives:To investigate the feasibility of binary logistic regression based on MRI quantitative parameters in predicting invasive placenta accreta and adverse clinical outcomes.Methods:The retrospective study included 80 placenta accreta patients.MRI parameters were assessed including diameter of abnormal proliferative blood vessels in the placenta,placental and uterine wall thickness,the placenta with T 2-dark bands,uterus,bladder,uterus protruding degrees low signal with discontinuous gap length,length of cervical invasion.Binary logistic regression analysis was used to predict the type of placenta accreta and independent risk factors for adverse clinical outcomes,and differential model was further constructed.The area under the curve(AUC),sensitivity,specificity and Youden index were evaluated by the receiver operating characteristic(ROC)curve.The stability of the regression model was verified internally by Bootstrap,and the AUC value was compared with the AUC value of the original regression model by DeLong test.Results:Among the 80 patients,the median age was 33 years(20~44 years),the median gestational age at delivery was 36.1 weeks(28.0~39.4 weeks),18 cases(22.5%)were adhesive placent and 62 cases(77.5%)were invasive placenta accreta.Binary logistic regression analysis showed that the number of abortions[OR(95%CI):4.271(1.226~14.876)],the diameter of proliferative vessels in the placenta[OR(95%CI):3.789(1.367~10.501)],extrauterine protrusion[OR(95%CI):1.432(1.088~1.885)]were independent risk factors for predicting invasive placenta accreta(P<0.05).Diameter of proliferative vessels in placenta[OR(95%CI):0.053(0.009~0.314)],area of T 2-dark bands[OR(95%CI):1.016(1.004~1.029)],extrauterine protrusion[OR(95%CI):0.839(0.727~0.968)]were independent risk factors for predicting adverse clinical outcomes(P<0.05).The accuracy of the model in predicting adhesion placenta,invasive placenta accreta and adverse clinical outcome was 91.3%.The AUC,sensitivity,specificity of the regression model was 0.962(95%CI:0.923~1.000),94.4%,and 85.5%,respectively.The AUC,sensitivity,specificity for predicting adverse clinical outcome was 0.952(95%CI:0.908~0.996),98.0%,and 82.8%,respectively.The AUC value of combined diagnosis was higher than those of all single indexes.Conclusion:The binary logistic regression model based on MRI quantitative parameters has a good value for predicting invasive placenta accreta and adverse clinical outcomes,and improves the overall diagnostic performance.