MRI prognosticators for adverse maternal and neonatal clinical outcome in patients at high risk for placenta accreta spectrum (PAS) disorders.

BACKGROUND: Placenta accreta spectrum (PAS) disorders may be associated with significant mortality and morbidity for both mother and fetus. PURPOSE/HYPOTHESIS: To identify MRI risk factors for poor peripartum outcome in gravid patients at risk for PAS. STUDY TYPE: Prospective. POPULATION: One hundred gravid women (mean age: 34.9 years) at third trimester, with placenta previa. FIELD STRENGTH/SEQUENCE: T2 -SSTSE (single-shot turbo spin echo), T2 -TSE, T1 -TSEFS (TSE images with fat-suppression) at 1.5T. ASSESSMENT: Fifteen MRI features considered indicative of PAS were recorded by three radiologists and were tested for any association with the following adverse peripartum maternal and neonatal events: increased operation time, profound blood loss, hysterectomy, bladder repair, ICU admission, prematurity, low birthweight, and 5-minute APGAR score <7. STATISTICAL TESTS: Kappa (K) coefficients were computed as a measure of agreement between intraoperative information/histology and MRI results as well as for interobserver agreement; chi-square and Fisher's exact tests were used to explore the association of the MRI signs with clinical complications. A score was calculated by adding all recorded MRI signs and its predictive ability was tested using receiver operating characteristic (ROC) analysis, against all complications, separately; odds ratios (ORs) for optimal cutoffs were determined with logistic regression analysis.
RESULTS: There was excellent agreement (K >0.75, P < 0.001) between MRI and intraoperative findings for invasive placenta, bladder and parametrial involvement. Intraplacental T2 dark bands, myometrial disruption, uterine bulge, and hypervascularity at the utero-placental interface or parametrium, showed significant association (P < 0.005) with poor clinical outcome for both mother and fetus. The MRI score showed significant predictive ability for each adverse maternal event (area under the curve [AUC]: 0.85-0.97, P < 0.001). The presence of ≥3 MRI signs was the cutoff point for a complicated delivery (OR: 19.08, 95% confidence interval [CI]: 6.05-60.13) and ≥6 MRI signs was the cutoff point for massive bleeding (OR: 90.93, 95% CI: 11.3-729.23), hysterectomy (OR: 72.5, 95% CI: 17.9-293.7), or extensive bladder repair (OR: 58.74, 95% CI: 7.35-469.32). The MRI score was not significant for predicting adverse neonatal events including preterm delivery (P = 0.558), low birthweight (P = 0.097), and 5-minute Apgar score (P = 0.078). DATA
CONCLUSION: Preoperative identification of specific MRI features may predict peripartum course in high-risk patients for PAS. LEVEL OF EVIDENCE: 1 Technical Efficacy: Stage 5 J. Magn. Reson. Imaging 2019;50:602-618.