The “rail sign”: an ultrasound finding in placenta accreta spectrum indicating deep villous invasion and adverse outcomes

Abstract

Ultrasound has demonstrated a high accuracy in the prenatal diagnosis of placenta accreta spectrum. However, it is not known whether ultrasound findings can pinpoint the depths of villous invasion, recommend surgical strategies, and predict clinical outcomes. We described an ultrasound descriptor for the placenta accreta spectrum and investigated whether it can predict the severity of villous invasion and clinical outcomes. The patients with placenta accreta spectrum in this retrospective cross-sectional study were diagnosed and managed in our hospital from 2002 to 2017. The placenta, with overlying myometrium and bladder, was mapped with color Doppler sonography while the patient's bladder was full. A "rail sign" was defined as 2 parallel neovascularizations depicted by color Doppler sonography over the uterovesical junction and bladder mucosa, with interconnecting bridging vessels perpendicular to both. The patients received serial ultrasound examinations and surgery at our hospital. An unpaired t test and Pearson chi-square test compared the pathology subtypes, surgical strategies, and clinical outcomes in patients with or without a rail sign. We enrolled 133 consecutive cases of placenta accreta spectrum confirmed either by surgical inspection or pathology examination. Patients with a rail sign had a significantly higher risk of an abnormally invasive placenta (placenta increta or placenta percreta) than those patients without a rail sign (83.3% [60 of 72] vs 27.9% [17 of 61]; odds ratio, 12.94; P<.001). In addition, patients with a rail sign had a higher probability of perioperative approaches, including preoperative vascular control (58.3% [42 of 72] vs 21.3% [13 of 61]; odds ratio, 5.17; P<.001) and uterine artery embolization (34.7% [25 of 72] vs 11.5% [7 of 61]; odds ratio, 4.1; P=.0002]. Furthermore, patients with a rail sign carried a higher risk of adverse clinical outcomes than patients without a rail sign, such as blood transfusion (80.6% [58 of 72] vs 36.1% [22 of 61]; odds ratio, 7.34; P<.001], admission to the intensive care unit (33.3% [24 of 72] vs 16.4% [10 of 61]; odds ratio, 2.55; P=.026), hysterectomy (75% [54 of 72] vs 24.6% [15 of 61]; odds ratio, 9.2; P<.001), and bladder invasion (16.7% [12 of 72] vs 4.9% [3 of 61]; odds ratio, 3.86; P=.033). Notably, the negative predictive value of bladder invasion was 95.1%, indicating a high confidence to reject bladder invasion while the rail sign was negative. When the rail sign was used as a screening test, the positive likelihood ratio of predicting deep villous invasion was 3.64 and correlated with an increased probability of 20% to 25%. Patients with a rail sign also had a greater blood loss (2944±2748 mL vs 1530±1895 mL; P<.001) and a longer hospital stay (11.9±10.9 days vs 8.6±7.1 days; P=.036) than patients without a rail sign. A "rail sign" depicted by color Doppler sonography correlates with deeper villous invasion, additional perioperative approaches, and more adverse outcomes.