Endorsing consensus interpretation for diagnosing placenta accreta spectrum disorders on MRI

Area of Science:

• Radiology
• Obstetrics & Gynecology
• Medical Imaging

Background:

• Placenta accreta spectrum (PAS) disorders pose significant maternal risks, necessitating accurate diagnostic methods.
• Inter-observer variability and lack of validated criteria can limit the reliability of MRI for PAS diagnosis.
• Developing standardized MRI criteria is crucial for improving noninvasive diagnosis of PAS.

Purpose of the Study:

• To evaluate the diagnostic performance of individual and combined MRI features for placenta accreta spectrum (PAS) disorders.
• To assess the utility of a multivariable model incorporating specific MRI findings for PAS diagnosis.
• To address the need for reliable, noninvasive diagnostic tools for PAS in pregnant women.

Main Methods:

• Retrospective analysis of 131 pregnant women undergoing MRI for placental or fetal indications.
• Assessment of 11 predefined MRI features by three independent observers, with consensus adjudication for discrepancies.
• Application of univariable/multivariable logistic regression and ROC analysis to evaluate feature performance and model accuracy.

Main Results:

• Consensus review indicated that most MRI features and prior delivery history were associated with PAS (p ≤ 0.0003).
• "Loss of T2-retroplacental line" and "myometrial thinning" demonstrated high sensitivity (0.87) at consensus.
• A multivariable model combining "uterine/placental bulge," "focal exophytic placental mass," and "bladder vessel sign" achieved an AUC of 0.83 for PAS diagnosis.

Conclusions:

• A diagnostic model based on consensus interpretation of MRI findings demonstrates good performance for PAS diagnosis.
• Standardized interpretation of MRI features can enhance diagnostic accuracy for PAS.
• This model aids in the noninvasive diagnosis of PAS, crucial for mitigating maternal morbidity and mortality.