RETURN TO ABSTRACT LISTING


Objective:
Placenta accreta spectrum (PAS) disorder relies on prenatal evaluation and diagnosis to guide delivery planning and reduce morbidity risk. Ultrasound is the screening modality of choice, and MRI is often used for inconclusive cases and surgical planning. Current guidelines recommend an MRI at 28 - 34 weeks of gestation and early MRI is considered suboptimal. The purpose of this study is to evaluate the accuracy and interrater concordance of MRI performed before 28 weeks gestational age for diagnosing the presence and severity of PAS disorder. A secondary aim is to examine associations between MRI features and clinical outcomes.

Materials and Methods:
This HIPAA-compliant, IRB-approved, retrospective study included pregnant patients who underwent MRI for PAS prior to 28 weeks gestation and delivered at the study institution over 6 years (2016 - 2021). Two placental MRI experts blind reviewed the images and documented the presence of 12 imaging findings as well as overall suspicion for PAS and myoinvasion. Using intraoperative and pathologic diagnoses as a reference, the sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated. Cohen’s kappa values were used to calculate interrater concordance. A multivariable regression analyzed the association between imaging features and clinical outcomes: conservative vs nonconservative management, estimated blood loss during delivery, FIGO clinical stage, and surgical pathology.

Results:
Of 152 patients, 34 patients met the inclusion criteria. Early gestation MRI performed between 12-28 weeks gestational age exhibited an overall sensitivity of 93.6% (95% CI 78.6-99.2%) and specificity of 86% (95% CI 42.1-99.6%) for diagnosis of PAS. Myoinvasion was detected with a 100% specificity (95% CI 85.8-100.0%) and 64% specificity (95% CI 35.1%-87.2%). There was substantial interrater concordance for diagnosing both PAS disorder (Cohen’s kappa = 0.77) and myoinvasion (Cohen’s kappa = 0.68). The earliest gestational age at which MRI was performed was 12 weeks. A subset analysis of MRIs performed between 12-24 weeks gestational age demonstrated a 93.75% sensitivity (95% CI 69.8%-99.8%), 100% specificity (95% CI 15.8-100%), 100% PPV, and 66.7% NPV (95% CI 23.1%-93.0%) for diagnosing PAS disorder. The presence of T2 dark bands, placental uterine bulge, loss of T2 hypointense interface, myometrial thinning, asymmetric thickening of the placenta, and abnormal intraplacental vascularity were positively associated with nonconservative management, higher FIGO clinical stage, and higher pathological stage (p < 0.05).

Conclusion:
MRI for assessing PAS performed in early gestation (prior to 28 weeks) has similar accuracy to MRI performed in later gestation, being both highly sensitive and specific. It also has high sensitivity for detecting myometrial invasion, although with low specificity. A subset analysis of MRIs performed prior to 24 weeks gestation has similar high sensitivity and specificity, but our data are limited by a small sample size. Imaging features can predict nonconservative outcomes for patients.