Validation of a sonographic checklist for the detection of histologic placenta accreta spectrum

Abstract

To standardize research terminology and reduce unanticipated placenta accreta spectrum (PAS), the European Working Group for Abnormally Invasive Placenta (EW-AIP) developed a consensus checklist for reporting PAS suspected on antenatal ultrasound. The diagnostic accuracy of the EW-AIP checklist has not been assessed. To test the performance of the EW-AIP sonographic checklist in predicting histologic PAS. This is a multi-site, blinded, retrospective review of transabdominal ultrasound studies performed between 26-32 weeks gestation for subjects with histologic PAS between 2016-2020. We matched a 1:1 control cohort of subjects without histologic PAS. To reduce reader bias, we matched the control cohort for known risk factors including previa, number of prior cesarean deliveries, prior dilation and curettage (D&C), in vitro fertilization (IVF), and clinical factors affecting image quality including multiple gestation, body mass index (BMI) and gestational age at the ultrasound. Nine sonologists from 5 referral centers, blinded to the histologic outcomes, interpreted the randomized ultrasound studies using the EW-AIP checklist. The primary outcome was the sensitivity and specificity of the checklist to predict PAS. Two separate sensitivity analyses were performed: 1) we excluded subjects with mild disease (i.e. only assessed subjects with histologic increta and percreta); 2) we excluded interpretations from the 2 most junior sonologists. 78 subjects were included (39 PAS, 39 matched control). Clinical risk factors and image quality markers were statistically similar between cohorts. The checklist sensitivity (95% Confidence Interval, CI) was 76.6% (63.4%-90.6%) and specificity (95% CI) was 92.0% (63.4%-99.9%), with a positive and negative likelihood ratio of 9.6 and 0.3, respectively. When we excluded subjects with mild PAS disease, the sensitivity (95% CI) increased to 84.7% (73.6%-96.4%) and specificity was unchanged at 92.0% (83.2%-99.9%). Sensitivity and specificity were unchanged when the interpretations from the 2 most junior sonologists were excluded. The 2016 EW-AIP checklist for interpreting PAS has a reasonable performance in detecting and excluding histologic placenta accreta spectrum.