Objectives: Ultrasound detection of a placenta accreta spectrum (PAS) among women at risk is a key goal to reduce obstetric morbidity, but there is scarce information on its performance in real clinical settings. We report the effectiveness of a standardized ultrasound protocol to detect PAS in women with placenta previa (PP) in a secondary-level hospital. Methods: A retrospective analysis, including a cohort of 126 women with persistent PP among 27,975 pregnancies between 2008 and 2020, was performed. All 126 women underwent standardized transabdominal and transvaginal ultrasound scan that assessed 5 criteria: (1) loss of hypoechoic retroplacental zone and/or myometrial thinning <1 mm; (2) lacunar images with a flow of >15 cm/s; (3) thick and bulging placenta; (4) thinning or interruption of the uterine-bladder serous interface; and (5) placental or uterovesical hypervascularity. The presence of at least one criterion was considered a high risk for PAS. Diagnosis of PAS was confirmed during the caesarean section and by histopathological analysis. Results: Among 126 women with PP, 11 (8.7%) cases of PAS were diagnosed, of which 10 were detected prenatally by ultrasound scan. This resulted in a sensitivity of 90.9%, a specificity of 98.3%, a positive predictive value of 83.3%, and a negative predictive value of 99.1%. Histopathological assessment showed 6 placenta increta (54.5%), 4 percreta (36.4%), and 1 accreta (9.1%). All 10 cases of invasive placenta presented more than 3 ultrasound criteria. Conclusions: Standardized ultrasound screening protocol in women at risk due to PP in the third trimester was highly effective in detecting PAS in a secondary-level hospital setting.
What does this study add to current knowledge?
This study provides evidence that in a secondary-level hospital, screening for placenta accreta spectrum (PAS) in high-risk women and under real-practice conditions achieved a predictive performance similar to that described in larger prospective studies and under research conditions.
What are the main clinical implications?
Ultrasound evaluation achieved a high diagnostic accuracy before magnetic resonance imaging (MRI), and therefore the results support that under adequate staff training and using a standardized protocol, ultrasound detection of PAS can be higher in similar settings.
Placenta accreta spectrum (PAS) refers to an abnormal adherence of placental trophoblast to the uterine myometrium. PAS can lead to massive and potentially life-threatening bleeding during labour, especially when it is not identified prenatally [1‒4]. The main risk factor is the presence of a previous caesarean section (CS) associated with placenta previa (PP) [5, 6]. The risk of PAS increases with a greater number of CSs . The presence of PAS can also be associated with surgical procedures damaging the endometrium such as curettages, myomectomies, manual placenta detachment, or endometritis [8, 9]. PAS prevalence at delivery is still not well established due to the heterogeneity in series, with recent estimates ranging from 0.01% to 1.1% . The incidence has increased during the last 4 decades by 10-fold correlating with the increasing rate of CS worldwide [1, 11]. Prenatal detection of PAS is a key goal in obstetric care, since it can substantially reduce maternal mortality and serious morbidity [12, 13].
Ultrasonography in the second or third trimester of pregnancy is the backbone of the prenatal diagnosis of PAS [14‒16]. Magnetic resonance imaging (MRI) is also a useful diagnostic tool, especially for cases where visualization is arduous due to obesity, posterior placentas, or to assess the invasion of neighbouring organs [17, 18]. However, MRI is expensive and requires specialized expertise in PAS detection, which is not readily available in most healthcare settings. Detection rates of PAS using ultrasound can significantly increase if standardized screening criteria are used . Initial studies in expert centres reported detection rates over 90% [20, 21], but they were performed in university research settings. More recent series, including population studies, report that about 50% of cases of PAS were not detected prenatally [22, 23]. Therefore, there is a need of studies that can report the performance of ultrasound screening for PAS in women at risk under real-world practice settings. Here we report the effectiveness of a standardized ultrasound protocol for the detection of PAS in women with PP in a secondary-level hospital setting under real clinical conditions.
Materials and Methods
We conducted a retrospective cohort study on the basis of the data collected from our centre between 2008 and 2020 that included all pregnant women diagnosed with PP. Our institution, the Son Llàtzer Hospital, is a second-level public hospital located in the island of Mallorca that performs 2,100 deliveries per year on average. During this period, through scheduled sonographies of the second and third trimesters, we detected 126 cases of PP among a total population of 27,975 deliveries. We defined PP as occlusive or central when it covers partially or completely the internal cervical os (ICO) and as low-lying placenta when placental edge is less than 20 mm from ICO without covering it in the third trimester . All 126 pregnant women presenting PP underwent a transabdominal and transvaginal ultrasound test between 20 and 36 weeks of gestation using both two-dimensional greyscale and colour Doppler and/or power Doppler. In all cases, the evaluation was performed by 2 expert sonographers (M.J. and A.T.).
A standardized ultrasound protocol assessing the risk of PAS by transabdominal and transvaginal ultrasound was stablished in 2008 in our foetal medicine unit for all women with PP. This protocol was not modified during the study period between 2008 and 2020, and it was updated after a careful revision of the multiple prenatal sonographic signs of PAS reported in recent publications, especially those from Comstock’s review  and Wong et al. . We included 5 criteria that were adapted based on published and accepted standardized definitions: (1) loss of normal hypoechoic retroplacental zone and/or thinning of the myometrium less than 1 mm; (2) large and multiple irregular lacunar images that also present high-velocity flow of more than 15 cm/s on Doppler study; (3) presence of a thick and bulging placenta that deforms or distorts the uterine serosa; (4) thinning or interruption of the uterine-bladder serous interface, which includes the presence of focal exophytic masses; (5) hypervascularity in the placental bed with the presence of tortuous vessels and/or bridging vessels and/or hypervascularization of the uterovesical interface in colour Doppler and 3D power Doppler (shown in Fig. 1).
The 5 ultrasound criteria were always evaluated by 2 expert sonographers, and both reached a consensus in all cases. Our protocol defined the presence of 1 or more criteria as high risk for PAS.
In all suspected cases of PAS with one exception (a vaginal delivery at 21 weeks of gestation), a caesarean delivery was carried out, and the diagnosis of PAS was confirmed first clinically during surgery and later by a histopathological examination of the uterus following hysterectomy.
In those cases with a prenatal diagnosis of PAS, an elective CS was performed by a multidisciplinary team that included obstetricians, urologists, vascular surgeons, anaesthetists, and neonatologists. First, a cystoscopy was performed with a catheterization of both ureters, followed by a vascular balloon placed in both hypogastric arteries. Subsequently, we performed the CS through a midline infraumbilical incision with supraumbilical extension in most cases. The clinical diagnosis of PAS was made in all cases through the inspection of the surface of the uterus, finding in the invasive cases a bulging and bluish appearance in the placental bed area with hypervascularity, and an invasion of the uterine serosa, bladder, or other organs by placental tissue could also be observed. If there is no clinical evidence of an invasive form of PAS, a gentle cord traction can be attempted, and if this causes a dimple in the uterine wall, then PAS is clinically diagnosed. If these signs are not present, a very careful digital exploration may be attempted to assess the presence of a cleavage plane between the uterus and the placenta. After this macroscopical visual view of the uterine surface, the foetal extraction was conducted through an incision located in the uterine body, the most far away possible from the placenta, usually at the fundus. After umbilical cord clamping, if there are clinical signs of PAS, we proceed to close the hysterotomy leaving the placenta and the umbilical cord and the cord clamp in situ, followed by a hysterectomy preserving annexes.
For statistical analysis, we used the software Stata v.10.1. Descriptive data are presented with 95% confidence intervals (CIs). For comparisons between groups, a p value of 0.05 or below was considered for statistical significance, using the Fisher exact test for categorical variables and the t test for quantitative variables. The primary outcome of the study was the sensitivity, specificity, positive, and negative predictive values of the ultrasound protocol for the detection of PAS.
During the 13-year study period, we identified a total of 126 pregnant women presenting with ultrasound diagnosis of persistent PP. Of these, 43 (34.1%) had previous uterine surgery; 26 had at least 1 CS (20.6%), 16 had curettage (12.6%), and 1 had a myomectomy (0.79%). Occlusive central PP was found in 95 patients (75.4%) while the low-lying placenta was present in the remaining (24.6%). Regarding placental location at the inferior uterine segment, 81 (64.3%) were posterior and 45 (35.7%) anterior. Seventy-two patients (57.2%) had antepartum vaginal bleeding. The average age was 34.8 years (23–46) (shown in Table 1).
PAS was diagnosed postpartum in 11/126 (8.7%) patients, of which 10 cases following a CS and in 1 case after a vaginal birth. In all cases, a hysterectomy and a histopathological study were performed, showing 6 placenta increta (54.5%), 4 percreta (36.4%), and 1 accreta (9.1%). Placental implantation was anterior in 8/11 (72.7%) and posterior in 3/11 (27.3%). The proportion of PAS was higher in women with a previous CS (10/26, 38.4%), or previous uterine surgery (11/43, [25.6%]), as compared with women without any previous CS (1/100, 1%, p > 0.0001) and without any previous uterine surgery (0/83, p < 0.0001) (shown in Fig. 2). The rate of PAS was 33.3% (7/21) in women with 1 previous CS and 75% (3/4) in those with 2.
Among the 11 women with PAS, 10 were identified as high risk according to the standardized ultrasound protocol, for a sensitivity of 90.9% (95% CI: 58.7%–99.8%), a specificity of 98.3% (95% CI: 93.9–99.8), a positive predictive value of 83.3% (95% CI: 51.6–97.9), and a negative predictive value of 99.1% (95% CI: 95.2–100) (shown in Table 2). Among the 126 women with PP included in the study, 12 (9.5%) had at least 1 of the 5 criteria of ultrasound diagnosis of PAS: 2 with 1 criterion, 7 with 4, and 3 with 5. All the 10 women who presented more than 3 criteria had a confirmation of PAS during surgery and by histopathological analysis following hysterectomy (shown in Table 3). The remaining 2 cases presenting 1 ultrasound criterion were false positives (FPs).
Among the 10 true positives cases, all 6 cases of placenta increta presented 4 ultrasound criteria, while for placenta percreta 1 case presented 4 criteria, and 3 presented all 5 criteria. The 2 FPs presenting only 1 criterion, both showed loss of normal hypoechoic retroplacental zone. The false-negative case presented a posterior low-lying placenta at 20 mm from ICO with 2 previous CS and 3 curettages (shown in Fig. 3).
In all 12 cases suspicious of PAS, an MRI was performed to assess a possible invasion of neighbouring tissues. Signs suggesting PAS were found in all cases. Two cases with 5 ultrasound criteria had a suspicion of bladder invasion by MRI, although this was not confirmed by cystoscopy before and during surgery.
Therefore in our study, 10 out of 11 cases of confirmed PAS were considered high risk according to the screening protocol. Following the intraoperative confirmation of the clinical signs described above, a hysterectomy was performed with a subsequent histopathological study of the uterus that definitively confirmed the diagnosis of PAS: 6 placenta increta and 4 percreta. Among these 10 cases, 1 was a hysterectomy after a prompt vaginal delivery and massive bleeding in a 21-week-gestation woman who wanted a conservative management. In the only false-negative case, the clinical diagnosis was made when it was verified during a CS, a placental area could not be detached, and then massive haemorrhage occurred. After that, a hysterectomy was performed with the histopathological result of placenta accreta. In the 2 FP cases, no signs of possible PAS were observed upon the inspection of the uterus during the scheduled CS, and a uterine incision was made far from the placental implantation, and a manual detachment of the placenta was successfully carried out without any complications after a very careful verification of a cleavage plane between uterus and placenta.
Standardized ultrasound screening criteria are considered the best approach for the antenatal diagnosis of PAS in women with PP. However, there is some inconsistency in the previously reported evidence about the effectiveness of such screening, particularly, when applied outside research settings. Here we describe a cohort of 126 pregnant women with PP that underwent ultrasound evaluation for the detection of PAS in a second-level hospital setting.
In this clinical study, a structured ultrasound protocol achieved a high specificity and sensitivity for identifying women at a high risk of PAS among those with PP. The predictive performance is similar to that reported in prospective studies with larger cohorts [26‒28]. More recently, Countinho et al.  described a screening protocol similar to that used in the present study, with a similar performance to that here reported. The previous studies were prospective and mostly conducted in university tertiary hospitals under research protocols. In contrast, other retrospective series [16, 22, 23] have reported that the performance of ultrasound screening may be remarkably reduced in other settings. This has posed some concerns as to the performance of ultrasound detection of PAS under real-world conditions. This study provides evidence that in a secondary-level hospital, screening for PAS in high-risk women and under real-practice conditions achieved a predictive performance similar to that described in larger prospective studies and under research conditions. Ultrasound evaluation achieved a high diagnostic accuracy before MRI, and therefore the results support that under adequate staff training and using a standardized protocol ultrasound, ultrasound detection of PAS can be high in settings with limited access to MRI.
We believe that the study population was representative of a general obstetric population. The prevalence of PP (0.39%) and that of PP with PAS (8.7%) were comparable with those reported in a recent meta-analysis (0.56% and 11%, respectively) . Likewise, as widely reported , the previous CS was the most prevalent risk factor for PAS among women with PP. In-line with larger prospective series [20, 21, 26], lacunar images and placental and/or uterovesical hypervascularization were the strongest predictors of PAS, while the loss of normal hypoechoic retroplacental zone and thinning or interruption of the uterine-bladder serous interface were the weakest (shown in Table 2). Of note, as previously reported  the diagnostic accuracy was higher when the placenta was anterior and lower in posterior placentas. This might support the use of complementary imaging methods such as MRI when available in women at risk but with posterior placenta. Among the limitations, we acknowledge that this is a relatively small study. However, it represents one of the few series reporting the performance of screening for PAS in a secondary-level hospital and under real-practice conditions. Our ultrasound screening protocol considered the presence of 1 or more criteria for the diagnosis of PAS, in comparison with other studies where 2 or more criteria were considered [26, 31]. The reason behind this choice was to maximize sensitivity. Given the outcomes reported, changing the criteria to 2 would have virtually not influenced the outcomes reported. Surprisingly, we found a high proportion of increta and percreta placentas and just 1 case of placenta accreta in comparison with systematic reviews where accreta cases are higher . This could have biased towards a higher rate the diagnostic accuracy in our study.
In summary, this study reports that a structured ultrasound evaluation achieved a high predictive performance to detect high risk for PAS among pregnant women with PP in a secondary-level hospital and under real-practice conditions. The study supports the use and potential impact of such screenings in similar settings.
We would like to thank Drs. M. Mascaró, V. Bonet, R. Diaz, and P. Lozano for their involvement in the surgical management of the cases, and also the authors thank M. Torrent Quetglas and J. Juan-Mateu for their contribution in the English editing of the manuscript.
Statement of Ethics
This study protocol was reviewed and approved by the Ethics Committee of the hospital Son Llàtzer Audit Committee, Approval No. HUSLL 210421. Written informed consent was obtained from participants prior to the study.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
The authors received no financial support for the study or for publication of this article.
M.J.C. and A.T. contributed to study design analysis and ultrasound screening, M.J.C., M.T., and P.S. wrote the paper and contributed to data collection. E.A. and J.I. contributed to the diagnosis and management of the cases, M.J.C., A.T., M.T., P.S., E.A., and J.I. contributed to the discussion and conclusion sections, made the final revisions and corrections, and approved the final version of the paper.
Data Availability Statement
The data that support the findings of this study are openly available in figshare at http://doi.org/[doi], Reference No. 10.6084/m9.figshare.16866283.