Ultrasound as the Primary Predictor of Perioperative Hemorrhage in Low-to-Moderate Risk Placenta Accreta Spectrum: A Prospective Comparison with MRI in Women with Placenta Previa

Background/Objectives: Placenta accreta spectrum (PAS) is an increasingly prevalent and potentially life-threatening complication in women with placenta previa. Despite widespread clinical use, the inter-modality agreement between prenatal ultrasound and MRI and their comparative value for predicting perioperative hemorrhage remain poorly characterized, particularly in low-to-moderate risk populations where placenta accreta predominates. We aimed to compare inter-modality agreement between standardized ultrasound and MRI impressions and to evaluate each modality’s predictive value for perioperative hemorrhage. Methods: This prospective cohort study enrolled 47 women with placenta previa who underwent both standardized ultrasound and MRI prospectively between 28 + 0 and 32 + 6 weeks of gestation, with perioperative outcomes collected at the time of cesarean delivery. Both modalities were classified using a three-tier impression system (None/Suspected/Likely) based on standardized structural, vascular, and invasive marker composites. The primary outcome was inter-modality agreement (linearly weighted Cohen’s κ); secondary outcomes were the association of each modality’s impression with postpartum hemorrhage (PPH; estimated blood loss ≥ 1000 mL) and estimated blood loss (EBL). Results: PAS was confirmed in 18 of 47 women (38.3%), predominantly placenta accreta (83.3%). Inter-modality agreement was fair (weighted κ = 0.263), structural concordance was moderate (κ = 0.539), while vascular agreement was near-absent (κ = 0.085). Ultrasound impression demonstrated a dose-dependent association with PPH rates (38.5%, 52.9%, and 82.4% across None, Suspected, and Likely tiers; p = 0.048) and EBL (800, 1000, and 1800 mL; p = 0.003), with logistic regression confirming a 2.70-fold increase in PPH odds per tier (p = 0.018; AUC 0.657). MRI impression was not associated with PPH (p = 1.000), EBL (p = 0.743), or PAS status (p = 0.741; AUC 0.543). Serum AFP was significantly elevated in women with PPH (p = 0.005). Conclusions: In this accreta-predominant, low-to-moderate risk cohort, ultrasound—but not MRI—demonstrated a significant dose-dependent association with perioperative hemorrhage. These findings should not be interpreted as evidence of general MRI inadequacy but rather as reflecting the specific imaging context in which MRI’s strengths in deep invasion characterization are less clinically determinative. These results support ultrasound as the primary tool for hemorrhage risk stratification in this population.