Predictive value of MRI image-based scoring model for diagnosis and adverse clinical outcomes of invasive placenta accrete / 中华围产医学杂志

ABSTRACT

Objective:To explore the predictive value of a scoring model based on MRI images for diagnosing invasive placenta accreta and associated adverse clinical outcomes.Methods:This retrospective cohort study involved 260 patients delivered at Peking University Third Hospital from January 2015 to December 2018, who were suspected to be placenta accreta with two or more ultrasound image findings and underwent MRI examination. Placenta accreta was finally diagnosed and classified based on the intraoperative clinical findings or pathological examination. Adverse clinical outcomes were defined as intraoperative bleeding ≥1 500 ml and/or having hysterectomy. Quantitative and qualitative interpretation of five MRI signs were performed, including intraplacental low-intensity band on T2 weighted imaging, abnormal intraplacental vascularization, vascularization of uterovesical interface, uterine bulging and cervical involvement. Chi-square and t test were used for univariate analysis of the five MRI signs and the receiver operating characteristics (ROC) curve of each MRI sign for predicting invasive placenta accreta and adverse clinical outcomes were drawn. The predictive value was assigned as 1 when ≥ the cutoffs that matched to the maximum Yoden index values, and was assigned as 0 when below the cutoffs. A scoring model based on the five MRI signs was established, ROC curves of the model for predicting invasive placenta accreta and adverse clinical outcomes were drawn and the area under the curve (AUC), sensitivity, specificity and Youden index were calculated. Results:(1) Univariate analysis showed that all five MRI signs were significantly associated with invasive placenta accreta and adverse clinical outcomes. Except for cervical involvement, the other four signs had an AUC value of greater than 0.5 in predicting invasive placenta accreta and adverse clinical outcomes. (2) The predictive cut-off values of abnormal intraplacental vascularization image and intraplacental dark band area on T2 weighted imaging were 2.0 cm 2 and 0.6 cm 2, respectively, and were all 1.0 for the other three signs. The AUC value of MRI signs-based scoring model for predicting invasive placenta accreta was 0.863. When the score was ≥ 2 points, the diagnostic sensitivity was 0.836 and the specificity was 0.726. The scoring model predicted adverse clinical outcomes with an AUC of 0.841. When the score was ≥3 points, the predictive sensitivity was 0.707 and the specificity was 0.818. Conclusions:The scoring model based on MRI signs is of good value for the diagnosis of invasive placenta accreta and the prediction of adverse clinical outcomes.