Ultrasonic risk model for predicting invasiveness of placenta accreta spectrum disorders / 中国医学影像技术

ABSTRACT

Objective To explore the value of a risk model established based on ultrasonic features for predicting invasiveness of placenta accreta spectrum disorders(PAS).Methods Data of 133 PAS patients were retrospectively analyzed.According to being invasive PAS or not,the patients were divided into invasive group(n=63)and non-invasive group(n=70).PAS-related ultrasonic features and distance between the lower margin of placenta and internal os of cervix(D value)were compared between groups.Univariate logistic regression and the receiver operating characteristic(ROC)curve were used to define the optimal cut-off value of figure of ultrasonic features for identifying invasiveness of PAS,then a dichotomous variable of the above figure was created.Multivariate logistic regression was performed to detect whether the dichotomous variable of the above figure and D were the independent impact factors for identifying invasiveness of PAS,and the risk prediction model was constructed.Results Among 12 PAS-related ultrasonic features,the detection rates of 10 features,including interruption or disappearance of retroplacental clear zone,thinner myometrium,lacunae,thickened placenta,cervix involvement,interrupted or disappeared bladder wall,feeding vessels of lacunae,bridge vessels,as well as hypervascularity of uteroplacental interface and between uterus and bladder in invasive group were higher than those in non-invasive group(all P＜0.05),while those of lumpy contour and placental bulge were not significantly different between groups(both P＞0.05).In invasive group,anterior placenta mainly located on the anterior wall and multiple PAS-related ultrasonic features were more common.Multiple ultrasonic features and smaller D value were both independent risk factors for identifying invasiveness of PAS(both P＜0.05).The prediction model for identifying invasiveness of PAS was logit(P)=-0.717+1.551 × Positivemultiple ultrasonic features-0.216 × D value,with the area under curve(AUC)of 0.905.Conclusion Multiple PAS-related ultrasonic features and shorter distance of the lower margin of placenta to the internal os of cervix were independent risk factors for identifying invasiveness of PAS.The constructed risk model was effective for predicting invasiveness of PAS.