4874. Development and Validation of a Combined Clinical, US, and MRI Model for Predicting Placenta Accreta Spectrum in Placenta Previa Patients
Authors * Denotes Presenting Author
  1. Simone Maurea; University of Naples "Federico II"
  2. Francesco Verde; AORN Cardarelli
  3. Arnaldo Stanzione; University of Naples "Federico II"
  4. Valeria Romeo *; University of Naples "Federico II"
  5. Pier Paolo Mainenti; IBB CNR
  6. Luigia Romano; AORN Cardarelli
  7. Arturo Brunetti; University of Naples "Federico II"
This study aims to construct and validate a predictive model for identifying placenta accreta spectrum (PAS) in individuals with placenta previa (PP), utilizing a combination of clinical risk factors (CRFs) and indications from ultrasound (US) and MRI.

Materials and Methods:
In this retrospective investigation, we included patients diagnosed with PP from two different medical institutions. All participants underwent both US and MRI scans due to suspected PAS. We collected data on CRFs including maternal age, history of cesarean sections, smoking habits, and hypertension. Indicators from US and MRI suggestive of PAS were also assessed. Through logistic regression analysis, factors associated with PAS were identified, considering histology as the benchmark. Significant CRFs and imaging signs were integrated to create a nomogram. The diagnostic precision of the nomogram was gauged using the area under the ROCAUC, and the optimal threshold was determined using the Youden’s J statistic.

A total of 151 patients were included from the two participating institutions. The independent predictors for PAS incorporated into the nomogram encompassed: 1) smoking and the number of previous cesarean sections among the CRFs; 2) disappearance of the retroplacental clear space in US scans; 3) presence of intraplacental dark bands, focal disruptions in the myometrial border, and placental bulging in MRI images. A nomogram for predicting PAS was formulated incorporating these parameters, and an optimal threshold of 14.5 points was established. This threshold yielded the highest sensitivity (91%) and specificity (88%), with an AUC of 0.95 (0.80 AUC in the external validation cohort).

The integration of CRFs with US and MRI indicators in a nomogram-based model holds promise for predicting PAS in patients with PP. Notably, MRI appears to offer greater contributions to imaging assessment compared to US.