Effectiveness of contingent screening for placenta accreta spectrum disorders based on persistent low‐lying placenta and previous uterine surgery

Maternal mortality related to placenta accreta spectrum (PAS) disorders remains substantial when diagnosed unexpectedly at delivery. The aim of this study was to evaluate the effectiveness of a routine contingent ultrasound screening program for PAS.


INTRODUCTION
Placenta accreta spectrum (PAS) disorders are a recognized cause of major maternal morbidity and mortality, with a reported prevalence of between 0.01% and 1.1% of pregnancies 1 . The incidence of PAS is increasing worldwide, which is attributed to the rising rate of Cesarean section [2][3][4][5] . Previous Cesarean birth is associated with an almost three-fold increase in the risk of PAS in the next pregnancy, as compared with previous vaginal delivery 6 . The main risk factor for PAS in the current pregnancy is a diagnosis of placenta previa, with more than 90% of PAS cases being associated with placenta previa combined with a history of uterine surgery [7][8][9][10] . PAS is associated with a significant increase in maternal morbidity and mortality from massive peripartum hemorrhage 11,12 . However, hemorrhagic morbidity can be reduced significantly if a diagnosis of PAS is made prior to admission at the time of delivery 13 . Systematic screening and diagnosis of PAS would permit referral of these high-risk women to tertiary hospitals with specialized multidisciplinary teams experienced in the management of pregnancies complicated by PAS 14,15 .
The antenatal ultrasound diagnosis of PAS is possible, with close to 90% accuracy, in specialist referral centers with expertise in the diagnosis of PAS 16,17 . However, the diagnostic accuracy of PAS screening in non-specialist referring hospitals is only 50%, as clinical suspicion for PAS and/or knowledge of risk factors is low 18,19 . There is a clinical need for an effective and systematic screening program for PAS in referring hospitals, so that those cases with suspected PAS can be referred to a PAS diagnostic center and, if PAS is confirmed, women can be referred for specialist surgical management to prevent the maternal morbidity associated with undiagnosed PAS.
The aim of this study was to evaluate the effectiveness of a contingent screening program for the identification of pregnancies with PAS based on a persistent low-lying placenta in the third trimester in women with a history of previous uterine surgery or Cesarean section.

METHODS
This was a retrospective cohort study of data obtained between 2009 and 2019 in the Fetal Medicine Unit of St George's University Hospitals NHS Foundation Trust, London, UK. Two cohorts of patients were included in the study: a 'screening cohort' of unselected women attending for routine mid-trimester ultrasound assessment at 18-23 weeks' gestation and a 'diagnostic cohort' consisting of local and external referrals to the PAS diagnostic service for suspicion of PAS. In the screening cohort, women diagnosed with a low-lying placenta (leading edge of the placenta less than 2 cm from the internal cervical os) at the mid-trimester fetal anomaly assessment were scheduled for a 32-34-week placenta-location scan 20 . At this assessment, placenta previa was confirmed if the leading edge of the placenta was still within 2 cm of the internal os 20,21 . Patients with confirmed placenta previa and a history of previous uterine surgery or Cesarean section were referred to the PAS diagnostic service. Women with a low-lying placenta at the mid-trimester ultrasound examination who transferred to another center before assessment of placental location at 32-34 weeks were excluded from the analysis.
All women undergoing routine ultrasound screening in pregnancy were assessed by qualified sonographers, and any cases with fetal or maternal complications were managed by maternal-fetal medicine specialists. In particular, the PAS diagnostic service was run by two consultants (B.T., A.B.) with significant experience and expertise in the prenatal diagnosis of PAS. Ultrasound assessment by the PAS diagnostic service consisted of two-dimensional grayscale ultrasound and color Doppler ultrasonography, as described previously [22][23][24][25] . The following markers were assessed: (1) the presence of multiple irregular lacunar spaces within the placenta with turbulent blood flow on Doppler ultrasonography (peak systolic velocity often > 10 cm/s); (2) loss of the 'clear zone' (i.e. the normal hypoechoic line between the placenta and the myometrium); (3) myometrial thinning of the retroplacental area; (4) increased placental thickness; and (5) bladder wall interruption, defined as loss or irregularity of the hyperechoic line between the uterine serosa and bladder. Color Doppler was used at the discretion of the examiner, mainly to differentiate lacunae from placental lakes 22,23,26 . The presence of two or more of these ultrasound signs was considered diagnostic of PAS. Cases with one isolated sign were classified as equivocal, and the absence of any ultrasound signs was considered negative for a PAS diagnosis. Magnetic resonance imaging was performed only in cases in which ultrasound signs of extrauterine invasion of the placenta (focal exophytic mass, distortion of cervix or parametrial anatomy) were detected. Women with a diagnosis of PAS were referred to the PAS surgical team for further management, usually with a conservative surgical technique (Triple P procedure) that is recommended by the International Federation of Gynecology and Obstetrics as an alternative to peripartum hysterectomy 15,27,28 .
Screening for placenta accreta spectrum 93 Eligible pregnancies were identified by searching the electronic database (ViewPoint version 5.6.26.148; ViewPoint Bildverarbeitung GMBH, Wessling, Germany) and by review of surgical notes and histological records. Demographic, surgical and histological data and obstetric history were retrieved from the clinical records. The final diagnosis of PAS was based on intraoperative clinical findings and histopathological examination of the surgical specimen 29 . A histological diagnosis of placenta accreta, increta or percreta was possible only when hysterectomy or partial-myometrial-resection specimens were accessible for examination. If a histological specimen was not available, the final PAS diagnosis was based on the surgical record of adherent placenta. Ethical approval and signed patient consent were not required, in line with the UK Regional Health Authority decision tool. The findings of this study were reported in agreement with the STROBE Statement 30 .

Statistical analysis
Comparisons between continuous and categorical variables were performed using the Kruskal-Wallis test and the χ 2 square test or Fisher's exact test, respectively. Univariate binomial logistic regression analysis was carried out to determine which risk factors were associated with PAS in women with placenta previa in the screening cohort. Owing to the limited number of PAS cases, forward stepwise multivariable logistic regression analysis was performed to determine which factors identified on univariate analysis contributed to the prediction of PAS in women with placenta previa and previous Cesarean section. Analysis was carried out using RStudio statistical software version 1.0.136 (RStudio, Inc., Boston, MA, USA) and MedCalc for Windows, version 19.4.1 (Med-Calc Software, Ostend, Belgium). Statistical significance was defined as P < 0.05.

Screening cohort
A total of 57 179 women underwent routine mid-trimester fetal anatomy assessment at 18-23 weeks' gestation between 2009 and 2019. Among the 4486 (7.8%) women with a repeat scan scheduled because of a low-lying placenta, 415 (9.3%) had a diagnosis of placenta previa at 32-34 weeks ( Figure 1). Two hundred and twenty women had a final diagnosis of placenta previa at subsequent assessments, of whom 75 (34.1%) were referred to the PAS diagnostic service because of a history of uterine surgery. This final cohort contained 22 cases of PAS, of which 21 were identified correctly by the PAS diagnostic service and managed surgically (Table S1). Additionally, five women who subsequently underwent emergency delivery before 32 weeks were followed up serially from the first trimester for a diagnosis of scar implantation, and there was one case of focal PAS in the non-placenta previa cohort in a woman with a previous myomectomy. The overall performance of the contingent PAS screening program in which women with a diagnosis of placenta previa and previous uterine surgery were referred to a specialist PAS diagnostic service was: sensitivity, 95.45% (95% CI, 77.16-99.88%); specificity, 100% (95% CI, 99.07-100%); negative predictive value, 99.75% (95% CI, 98.30-99.96%); and positive predictive value, 100% (95% CI not calculable).
The demographic characteristics and risk factors for PAS in women with placenta previa at 32-34 weeks who had confirmed PAS and those without evidence of abnormal placental invasion are shown in Table 1. Univariate analysis demonstrated that parity ≥ 2 (odds ratio (OR), 35.50 (95% CI, 6.90-649.00)), two or more previous Cesarean sections (OR, 94.20 (95% CI, 22.00-656.00)) and placenta previa at the final scan (OR, 20.50 (95% CI, 4.22-369.00)) were the strongest risk factors for PAS. Smoking (OR, 1.12 (95% CI, 0.06-5.92)), body mass index > 24 kg/m 2 (OR, 1.04 (95% CI, 0.95-1.11)) and previous uterine surgery other than Cesarean section (OR, 3.49 (95% CI, 0.77-11.60)) were not associated with PAS in this cohort. Paired stepwise multivariable logistic regression analysis demonstrated that the strongest risk factors for PAS, when taking into account the number of previous Cesarean sections, were variables related to placental location and ethnicity ( Table 2). Maternal age, Asian ethnicity and multiparity were no longer associated with an increased risk for PAS after controlling for previous Cesarean section.

Diagnostic cohort
One hundred and seventy-three (99 local and 74 external) suspected cases of PAS were referred to the diagnostic service between 2009 and 2019 ( Figure 2). There were one false-positive and three false-negative diagnoses of PAS in this cohort.

DISCUSSION
Routine contingent screening for PAS based on the finding of placenta previa in the third trimester and a history of Cesarean section is both feasible and effective. The capability of such a screening program to result in improved clinical outcome is contingent on access to a multidisciplinary PAS diagnostic and surgical service.

Comparison with other studies
The prevalences of placenta previa (0.4%) and PAS (0.04%) in our screening cohort are in line with the prevalences of 0.56% and 0.07%, respectively, described previously in a meta-analysis of population-based studies 31 , consistent with the assertion that our study was performed in a routine pregnancy population. The finding that the number of previous Cesarean sections and diagnosis of placenta previa were the most important risk factors for PAS is consistent with the findings of previous systematic reviews of PAS 1,6,7 . Even one previous Cesarean section increased the odds for PAS by over 25-fold in a cohort of women with placenta previa, and the majority of PAS cases in our cohort had had only one previous Cesarean section. Other previous uterine surgeries were not associated with PAS in this cohort, which is in contrast to the findings of some previous studies 32,33 .
It is difficult to decide whether the lack of an association between previous uterine surgery other than Cesarean section and PAS is a consequence of the small number of cases included or a true finding of this study 33,34 . Afro-Caribbean ethnicity remained associated with PAS even after statistical correction for previous Cesarean section. This finding has not been reported previously and might be explained either by socioeconomic determinants of health in black and minority ethnic populations or, more likely, by residual confounding from the relatively low number of patients with PAS in the screening cohort and the increased likelihood of Cesarean birth in Afro-Caribbean women.
Despite general consensus among international guidelines that women with previous uterine surgery and a low-lying placenta are at increased risk for PAS, there has been no previous systematic evaluation of the effectiveness of such a contingent screening program 9,22 . A first-trimester screening strategy for PAS based on the identification of a low-lying placenta in women with previous uterine surgery between 11 and 13 weeks has been described 35 . Although the benefit of such early screening would be to provide women with the option to have a first-trimester termination of pregnancy, the drawbacks are the number of additional follow-up visits and low specificity, which would result in a large number of referrals to a PAS diagnostic service. The contingent screening described in the current study does not require additional routine scan visits, but still resulted in a high rate of detection of PAS. The St George's Hospital PAS diagnostic service correctly identified 86 out of 89 cases of PAS using ultrasound alone, resulting in a sensitivity and specificity greater than 95%, superior to those in retrospective studies (sensitivity, 88% (95% CI, 81.0-93.0%); specificity, 90% (95% CI, 88.0-93.0%)) and comparable with those in prospective diagnostic studies (sensitivity, 97% (95% CI, 93.0-99.0%); specificity, 97% (95% CI, 97.0-98.0%)) 9 .

Clinical implications
Effective prenatal screening that correctly identifies pregnancies complicated by PAS would significantly improve maternal outcomes, because the women could have a scheduled birth in a center with tertiary level facilities specialized in PAS care 13 . The need for such screening is compounded by the continuing global trend for an increase in both Cesarean section rate and PAS disorders 36 . We have shown that contingent screening for PAS in women with previous Cesarean section and a current diagnosis of placenta previa is both feasible and effective. This screening and triage strategy is feasible in lower-resource medical settings with basic obstetric ultrasound facilities and does not require additional visits beyond those that are indicated routinely. The success of such a screening program depends on access to both a specialized PAS diagnostic service with experienced operators and a management service at a tertiary level hospital at which, if the diagnosis of PAS is confirmed, safe delivery can be arranged 37 .

Strengths and limitations
The major strengths of the present study are related to the relevant sample size in both the screening and diagnostic cohorts. Additionally, significant expertise was combined with a robust surgical service with distinctive skills in the conservative management of PAS to allow linkage of antenatal ultrasound findings with surgical and histological records. The Triple P procedure used includes myometrial excision alongside preservation of the uterus and allows a full histological diagnosis in the majority of cases 29 .
An obvious limitation is the retrospective nature of the study, and there is a need to validate these findings in prospective studies set in populations with variable Cesarean section rates in order to ensure validity in a wide range of settings. The absolute number of cases affected by PAS in the screening cohort was small and, as a consequence, it was not possible to undertake a detailed analysis of risk factors in the prediction of PAS. Finally, our screening protocol failed to detect a case of focal PAS following myomectomy because the placenta was more than 2 cm from the internal os.

Conclusions
A screening program for PAS based on the identification of women with a persistent low-lying placenta in the third trimester and a history of previous Cesarean section was both feasible and highly effective. This contingent strategy has the potential to improve the antenatal PAS detection rate and decrease maternal morbidity and mortality related to undiagnosed PAS. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Eficacia de la detección contingente para los trastornos del espectro de la placenta acreta a partir de la placenta baja persistente y en la cirugía uterina previa RESUMEN Objetivo La mortalidad materna relacionada con los trastornos de la gama espectral de la placenta acreta (EPA) sigue siendo considerable cuando se diagnostica de forma inesperada en el momento del parto. El objetivo de este estudio fue evaluar la efectividad de un programa rutinario de detección contingente mediante ecografía para el EPA.
Métodos Este fue un estudio retrospectivo de datos obtenidos entre 2009 y 2019, en el que participaron dos grupos: una cohorte de detección de mujeres no seleccionadas que acudieron a la evaluación ecográfica rutinaria de mitad de trimestre y una cohorte de diagnóstico, integrada por mujeres remitidas al servicio de diagnóstico del EPA con un presunto diagnóstico del EPA. En la cohorte de detección, a las mujeres con una placenta baja en la evaluación de mitad de trimestre se les hizo un seguimiento en el tercer trimestre, y a aquellas con una placenta baja persistente (es decir, placenta previa) que habían tenido cirugía uterina previa se las remitió al servicio de diagnóstico del EPA. La evaluación ecográfica por el servicio de diagnóstico del EPA consistió en una ecografía Doppler bidimensional en escala de grises y en color, y a las mujeres con diagnóstico del EPA se las trató habitualmente con una resección conservadora del miometrio. El diagnóstico final del EPA se basó en una combinación de indicadores clínicos intraoperatorios y el examen histopatológico de la muestra quirúrgica.