Adjunctive MRI in the diagnosis of placenta accreta spectrum in major placenta previa: incidence, risk factors, and maternal morbidity

Abstract

BACKGROUND:

Antenatal assessment of maternal risk factors and imaging evaluation can help in diagnosis and treatment of placenta accreta spectrum (PAS) in major placenta previa (PP). Recent evidence suggests that magnetic resonance imaging (MRI) could complement ultrasonography (US) in the PAS diagnosis.

OBJECTIVES:

Evaluate the incidence, risk factors, and maternal morbidity related to the MRI diagnosis of PAS in major PP.

DESIGN:

A 10-year retrospective cohort study.

SETTING:

Tertiary care hospital.

PATIENTS AND METHODS:

We report on patients with major PP who had cesarean delivery in Abha Maternity and Children's Hospital (AMCH) over a 10-year period (2012-2021). They were evaluated with ultrasonography (US) and color Doppler for evidence of PAS. Antenatal MRI was ordered either to confirm the diagnosis (if equivocal US) or to assess the depth of invasion/extra-uterine extension (if definitive US).

MAIN OUTCOME MEASURES:

Risk factors for PAS in major PP and maternal complications.

SAMPLE SIZE:

299 patients

RESULTS:

Among 299 patients, MRI confirmed the PAS diagnosis in 91/299 (30.5%) patients. The independent risk factors for MRI diagnosis of PAS in major PP included only repeated cesarean sections and advanced maternal age. The commonest maternal morbidity in major PP with PAS was significantly excessive intraoperative bleeding.

CONCLUSION:

MRI may be a valuable adjunct in the evaluation of PAS in major PP, but not as a substitute for US. MRI may be suitable in major PP/PAS patients who are older and have repeated cesarean deliveries with equivocal results or suspected deep/extra-uterine extension on US.

LIMITATION:

Single center, small sample size, lack of complete histopathological diagnosis.

CONFLICT OF INTEREST:

None.

INTRODUCTION

Historically, a morbidly adherent placenta or abnormally invasive placenta or the placenta accreta spectrum (PAS) is a broad-spectrum disorder related to the pathologically adherent and deeply invading placental tissues through the uterine myometrium.¹ By March 2018, the International Federation of Gynecology and Obstetrics (FIGO) had proposed the terminology (PAS), as a standardized terminology, to describe all placental implantation disorders.² Several theories have been offered to explain the etiology of PAS disorder; however, the most acceptable explanation is uterine decidual destruction, mostly by previous cesarean surgeries, hysterectomies¹ and many surgical gynecologic procedures.³ In fact, the presence of placenta previa (PP) and a history of previous cesarean deliveries are the two major risk factors for PAS, and the incidence increases linearly with the number of repeated cesarean deliveries or hysterotomies.^4,5 PP at one stage was considered necessary in PAS diagnosis.⁶ Such antenatal situations (PP and PAS) could be associated with high maternal and neonatal morbidity and even mortality.⁷

The latest publications of the American College of Obstetricians and Gynecologists (ACOG) “Obstetric Care Consensus” No. 9, 2019⁸ and ACOG “Committee Opinion” No 7, 2018⁹ confirm that both antenatal assessment of maternal risk factors and imaging evaluation can help in guiding the antenatal diagnosis of PAS. In practice, the outcomes are maximally optimized when delivery occurs at highly equipped maternal care tertiary centers, before the onset of labor or bleeding with availability of multidisciplinary staff for managing complex and complicated maternal morbidities.

Ultrasound (US) is the first-line imaging technique for PAS diagnosis because it is effective, low cost, readily available, and less time is required for exploration, with an overall sensitivity of 100% and specificity of 100% for “multiple” ultrasound signs; however, this modality has lower specificity for the analysis of the depth of myome-trial invasion (specificity 64.8%-sensitivity 100%).¹⁰ On the other hand, magnetic resonance imaging (MRI) has a high sensitivity and specificity for PAS diagnosis (sensitivity 100%, specificity 100%) and higher accuracy for detecting the depth of placental invasion/extension in comparison to US (specificity 96.8%-98.8%, specificity 86.5%–100%).¹¹

An updated review by Fratelli et al 2022,¹² on the diagnostic accuracy of US and MRI in PAS disorders, showed that the advantages of each imaging modality counteract the limitations of the other. Thus, the disadvantages of US, such as operator-dependence and limited penetration and narrow field of view, are compensated by MRI's reproducible large field of view. Obviously, the most evident advantage of MRI is its high quality contrast resolution and tissue specific description, allowing a detailed evaluation of the whole placental-myometrial junction with accurate assessment of deep invasion or extra-uterine extension. Notably, its limited availability and high cost are recognized disadvantages of MRI, which is also dependent on the technician experience.¹³

The current study is a retrospective analysis of a cohort of consecutive patients with major PP with “equivocal” (suspicious) or “definitive” evidence of PAS on US and color Doppler examination. MRI was ordered to confirm the diagnosis (if equivocal) or depth of invasion and/or extra-uterine extension (if definitive). We aimed to determine the incidence, risk factors and maternal morbidity related to the MRI diagnosis of PAS in major PP patients.

PATIENTS AND METHODS

This retrospective cohort study initially included consecutive patients diagnosed with major PP admitted to Abha Maternity and Children's Hospital (AMCH) over a 10 year-period (2012–2021), Aseer region, Saudi Arabia. US and color Doppler examinations were used to diagnose PAS. Antenatal MRI was ordered either to confirm the diagnosis or to assess invasion and/or extra-uterine extension. The Research Ethics Committee at King Khalid University approved the study (ECM# 2023-607).

We diagnosed major PP when the placenta reached the internal os, and partially or completely covers it after 24 weeks gestation. Major PP was confirmed after 32 weeks; based on ultrasonographic types III and IV. The antenatal identification of PAS was primarily performed by trans-abdmonial US with color Doppler examination. This was done beyond 24 weeks’ gestation and at 32–36 weeks to re-evaluate placental location and to diagnose PAS. If two or more of the characteristic PAS ultrasound signs were present, a diagnosis of PAS was made. With only one (isolated) sign, the case was diagnosed as equivocal/suspicious; if no ultrasound signs were present, the case was diagnosed as negative for PAS.¹⁴

We consider MRI to be adjunctive to US because the accuracy of US/color Doppler in PAS diagnosis is greatly dependent on the expertise of the ultrasonographer (up to 2021 we did not have a specialized PP/PAS US unit in AMCH).¹⁵ The US diagnostic accuracy of PAS screening in non-specialist referring US units is only 50%.¹⁶ Furthermore, none of the US signs (or even combinations of signs) associated PAS reliably predicts depth of myometrial invasion or type of PAS.¹⁷ Thus, the MRI appears to be complementary to US, when there is an equivocal result or to assess depth of placental invasion/extension.¹² According to the MRI findings, the PAS was classically subdivided into three grades: accreta, increta and percreta, depending on the depth of invasion into the myometrium of the uterus.^10,11 All patients were delivered by cesarean surgery (emergency or planned); because if the PP reached the internal os or covered it after 35 weeks, this is an indication for a cesarean delivery.⁷ A postoperative histopathological examination was performed to confirm the PAS diagnosis in hysterectomy specimens when possible, as this is neither always achievable nor accurate.¹⁸ PAS diagnosis cannot be established among patients who have heavy intraoperative bleeding with difficult placental separation but are non-hysterectomized. Intraoperative blood loss was assessed by visual estimation. Up to 1000 mL was considered normal, 1000-2000 mL as mild, 2000-3000 mL as moderate and more than 3000 mL was considered heavy bleeding.

The records of all PP patients with US/Doppler evidence of PAS (equivocal or definitive diagnosis) and confirmed with MRI were reviewed to collect the following data: age, obstetric characteristics, placental site (anterior or posterior PP), presence of PAS on ante-natal imaging (US/color Doppler and MRI), and maternal morbidity. Maternal morbidity included: admission and discharge hemoglobin levels, RBCs/fresh frozen plasma transfusions, admission and termination gestational ages, emergency/planned cesarean deliveries, emergency cesarean hysterectomy, intraoperative bleeding, visceral injuries, and length of postoperative hospital stay. The extracted data were recorded and coded in a predesigned excel sheet.

Data analysis was performed using IBM SPSS version 19.0 for Windows (IBM SPSS Inc., Armonk, NY: IBM Corp, USA). The medians, means and standard deviations were used to describe continuous data, which was not normally distributed. We applied the Mann-Whitney U test to compare numerical data and chi-square test (Χ²) for categorical data. Multivariable logistic regression analysis was performed and presented with crude and adjusted odds ratio (OR) and 95% confidence intervals. A P value of less than .05 was considered statistically significant. The Hosmer-Lemeshow test was used to test goodness of fit of the logistic regression analysis.

RESULTS

At admission, the median age was 33 weeks (Table 1). Women with PP/PAS were older (P=.003), had higher parity (P=.002) and more repeated cesarean sections (P=.001) (Table 2a) (P=.003). Placental location is shown in Table 2b. Other admission obstetrical characteristics did not show significant differences. Among 299 patients with major PP patients identified over the ten-year period, nearly 33% (77/236 patients) had definitive results (with ≥2 US signs), while the remainder (n=159) had equivocal PAS diagnosis (with 1 US sign). PAS disgnosis was confirmed by MRI in 91 (30.5%) of the major PP patients with negative results in 145. (Figure 1). Compared to the entire cohort, 30.5% had confirmed PAS diagnosis (91/299). The independent risk factors for MRI diagnosis of PAS in major PP included repeated cesarean sections (adjusted OR 3.923, 95%CI 2.082–7.390) and advanced maternal age (adjusted OR 2.147, 95%CI 1.183–4.259) (Table 3). Other variables analyzed in the multivariable regression model were not independent risk factors (gravidity, abortions, previous uterine surgery, and placental location). The commonest maternal morbidity in major PP with PAS was excessive intraoperative bleeding (61.5% vs. 24.1%, P<.001). All differences in maternal morbidity are shown in Tables 4, 5 and 6. Variables that did not differ included emergency versus elective cesarean section, preoperative hemoglobin level, postoperative hemoglobin level, and termination gestational age.

Table 1.

Characteristics of the study group at admission (n=236).

Variable	Values
Maternal age (years)	33 (9) [17–47]
Gravidity	4.93 (2.68) [1–16]
Parity	3.14 (2.12) [0–12]
0–4	184 (77.9)
5+	52 (22.1)
Abortions	0.8 (1.1) [0–6]
Number of cesarean deliveries	1.86 (1.63) [0–6]
0–2	166 (70.3)
3+	70 (29.7)
Admission gestational age (weeks)	32.7 (3.33) [25–37]
Termination gestational age (weeks)	34.56 (6.13) [27–39]
Previous Uterine Surgery (D&C-D&E)
No	182 (77.1)
Yes	54 (22.9)
Antepartum hemorrhage
No bleeding	18 (7.6)
Bleeding	218 (92.4)
Placental site
Previa anterior	75 (31.8)
Previa posterior	161 (68.2)
Placenta Accreta Spectrum (PAS) by USS+ Doppler (n=236)
Equivocal for PAS	159 (67)
Definitive PAS	77 (33)
Placenta Accreta Spectrum (PAS) by MRI (n=236)
No PAS	145 (61.5)
Definitive PAS	91 (38.5)
MRI Grading of PAS disorder (n=91)
Accreta	56 (61.5)
Increta	15 (16.6)
Percreta	20 (21.9)

Data are n (%) for categorical variables and median (minimum-maximum) for continuous variables. D&C: dilatation and curettage; D&E: dilatation and evacuation

Table 2a.

Obstetrical characteristics in major placenta previa patients with and without placenta accreta spectrum by MRI.

Variable	Value	Mann-Whitney U value	P value
Maternal age (years)
PP without PAS (n=145)	34 (10, 18–47)	2.558	.003
PP with PAS (n=91)	33 (7.5, 16–47)
Parity
PP without PAS (n=145)	3 (2, 0–12)	3.127	.002
PP with PAS (n=91)	3 (2.5, 0–12)
Number of cesarean deliveries
PP without PAS (n=145)	1 (3, 0–6)	6.032	.001
PP with PAS (n=91)	2 (3, 0–6)

Data are median (interquartile range and minimum-maximum). Comparison by Mann-Whitney U test.

Table 2b.

Placental location (n=236).

Placental location	Anterior 75 (31.8)	Posterior 161 (68.2)	Pearson chi-Square	P value
PP without PAS (n=145)	37 (25.5)	108 (74.5)	5.802	.009
PP with PAS (n=91)	38 (41.8)	53 (58.2)

Data are n (%).

Figure 1.

Study flow diagram.

Table 3.

Multiple logistic regression analysis of the independent risk factors for MRI diagnosis of PAS in major PP.

Risk Factor	PP with PAS (n=91)	PP with out PAS (n=145)	P value	Crude OR (95% CI)	P value	Adjusted OR (95% CI)
Maternal age (years)
≤30 (reference)	19 (20.9)	56 (38.6)	.004	2.38 (1.30–4.37)	.029	2.147
31+	72 (79.1)	89 (61.4)				1.183–4.259
Number of cesarean deliveries
1–2 (reference)	47 (51.6)	119 (82.1)	<.001	4.285 (2.37–7.73)	<.001	3.923
3+	44 (48.45)	26 (17.9)				2.082–7.390

Data are n (%) for variables. Model fit summary: Deviance 282.386, Overall model chi-square 32.314, P<.001, Cox&Snell R square 0.128, Nagelkerke R square 0.174

Table 4.

A comparison of maternal morbidities in major PP patients with and without PAS by MRI.

Variable	Value	Mann-Whitney U value	P value
Packed RBCS Transfusion (units)
PP without PAS (n=145)	1 (1, 0–5)	6092.5	<.001
PP with PAS (n=91)	2 (1, 0–13)
Fresh frozen plasma transfusion (units)
PP without PAS (n=145)	0 (1, 0–4)	23728	<.001
PP with PAS (n=91)	2 (3, 0–24)
Post operative stay (days)
PP without PAS (n=145)	3 (0, 2 5)	45.5	<.001
PP with PAS (n=91)	3 (1.8, 1–15)

Data are median (interquartile range and minimum-maximum); Commparison by Mann-Whitney U test.

Table 5.

Intra-operative bleeding.

	NO to mild bleeding	Moderate to heavy	Pearson chi-square	P value
PP without PAS (n=145)	110 (75.9)	35 (24.1)	37.727	<.001
PP with PAS (n=91)	35 (38.5)	56 (61.5)

Table 6.

Additional maternal morbidity.

	DONE	NOT DONE	Pearson chi-square	P value
Emergency hysterectomy (n=28)
PP without PAS (n=145)	1 (0.7)	144 (99.3)	44.907	<.001
PP with PAS (n=91)	27 (29.7)	64 (70.3)
Urologic injuries (n=14)
PP without PAS (n=145)	1 (0.7)	144 (99.3)	18.521	<.001
PP with PAS (n=91)	13 (14.3)	78 (85.7)

Data are n (%).

DISCUSSION

In this is 10-year retrospective analysis, 299 consecutive patients were diagnosed with major PP. MRI confirmed the PAS diagnosis in 91 patients. Generally, previous cesarean delivery and PP are the two most important risk factors for PAS with further substantial increased risk if both are present together.⁶ Even alone, PP is one of the most predictive risk factors in the development of PAS.⁵ In the current study the incidence of PAS was 30.5% (91/299) among major PP patients; a previous systematic review showed a wide range in incidence (2.9%–71.6%) of PAS in women presenting with PP in different studies.¹⁹ Nevertheless, epidemiological studies have revealed a notable rise in the prevalence of caesarean deliveries and, consequently, an increase in the occur-rence of PAS over time.^4,16,19–21 Unfortunately, a recent review confirmed that if PAS is left undiagnosed in 50% to 75% of patients, increasing maternal morbidity and mortality (i.e. determined only at the time of delivery).²²

Notably, our comparison of the obstetrical characteristics of PP patients (with and without PAS by MRI) demonstrated that patients with major PP/PAS were older and had a higher mean number of deliveries than patients. Such obstetric features of the PP/PAS patients are typically reported by previous studies from AMCH by Bahar et al²³ Zaki et al²⁴ and by studies from other regional hospitals in Saudi Arabia.^25–27 Earlier multivariate analyses showed that women aged 34-35 years or older²⁸ and multiparity (para >3) had higher risks of PP,²⁹ which in turn increased the risk of PAS. A clear dose-response pattern was observed with increasing age or parity.

As expected, our major PP/PAS group had a higher mean number of cesarean deliveries indicating that the majority had more than one previous cesarean delivery. Nearly 44% and 30% of them had 2+ and 3+ previous cesarean deliveries; respectively. Such a high mean number of repeated cesarean deliveries is higher than in similar Saudi studies on PP/PAS (2+ previous cesarean deliveries was 33.5% by Mansour and Mousa³⁰ 24.5% by Kassem and Alzahrani²⁷ and 20.2% by Bahar et al 2009²³). The number of repeated cesarean deliveries dramatically increases the risk of PP.^20,31 On the other hand, reports showed that women with repeated cesarean deliveries are particularly at risk for PAS reaching 6.74% in women undergoing their sixth or more cesarean deliveries. However, the presence of both PP in a scarred uterus substantially increased the risk of PAS (up to 67% for fifth or more repeat cesarean deliveries).⁴

The commonest placental location among our patients was the posterior position, in 74.5% of PP patients without PAS, but only in 58% of PP with PAS (P=.009). This is consistent with similar studies showing that 60% to 80% of PP women exhibited a posterior placental position.^32,33 These studies suggested that the placenta prefers to develop on the posterior uterine wall in PP. However, the incidence of PAS was significantly higher in the anterior group, irrespective of PP degree. Nevertheless, some studies contradict our findings; Morgan et al³⁴ and Koai et al³⁵ showed that anterior PP is more common than a posterior position with significantly higher maternal hemorrhagic morbidity and preterm delivery.

The multivariable logistic regression analysis of the independent risk factors for PAS in PP patients showed that only previous cesarean deliveries (adjusted OR of 3.923) and older maternal age (adjusted OR of 2.147) significantly increased the odds ratios of PAS on MRI examination. Two previous 10-year analyses were published from AMCH on the risk factors and outcomes of PP/PAS, by Bahar et al²³ and Zaki et al²⁴ together, these studies reported high odds ratios of PP/PAS due to increased number of repeated cesarean deliveries (OR of 3.2 and OR of 7.9; respectively). This is in agreement with larger international multicenter studies which revealed comparable or higher odds ratios of PAS due to repeated cesarean deliveries as Jenabi et al³⁶ (OR, 1.60) Farquhar et al³⁷ (OR, 13.8) and OR of 34.9 by Bowman et al.²¹

Maternal morbidity among our PP patients with PAS by MRI showed significantly increased mean numbers of packed RBCS and fresh frozen plasma transfusions (P<.001), and increased percentage of intraoperative moderate/heavy bleeding (P<.001). This pattern of serious hemorrhagic morbidity is a fixed main feature of the studies on PP/PAS from AMCH^23,24 and similar studies from different regions in Saudi Arabia.^25–27,30 An updated review article by Fonseca and Ayres de Campos³⁸ indicated that PAS is associated with 18-fold increase in maternal morbidity; and heavy bleeding is the commonest morbidity. Moreover, PP/PAS patients had significantly increased rates of emergency hysterectomy (29.7% vs. 0.7%, P<.001), and consequently longer mean number hospital stay (P<.001). Similarly, many Saudi studies recorded emergency hysterectomy as a major morbidity related to the life threatening bleeding due to PP/PAS; the percentage of hysterectomy due to PP alone ranged from 4.5% to 22% and was as high as 50% to 84% among PP with PAS.^23–27,30 We reported 14 (5.9%) intraoperative urological injuries (almost bladder injuries) secondary to the extra-uterine placental extension (percreta) with bladder invasion. The recorded incidences in some Saudi studies ranged from 4% to 9.8%.^25,27,30 A recent study by Chen et al³⁹ MRI confirmed a 100% specificity for predicting the extra-uterine placental extension; where all patients with evidence of bladder invasion received partial bladder resection and all those with parametrial or cervical invasion underwent cesarean hysterectomy.

In fact, tentative trials of separating an invasive placenta indicate a risk of hysterectomy in up to 100% of cases.⁴⁰ Accordingly, the current recommendations of the ACOG, Society of Gynecologic Oncology (SGO); the Society for Maternal–Fetal Medicine (US-SMFM)^8,9 and the consensus of the FIGO^2,41 expert panel: “among patients with high suspicion for PAS during a cesarean delivery, most of surgeons proceed to hysterectomy after delivery of the fetus, with the placenta left in-place”.

We suggest that MRI is a valuable adjunct in the evaluation of PAS in major PP; as a compliment but not as a substitute to US. Both repeated cesarean sections and older age appeared as independent risk factors for the existence of PAS on MRI in major PP. These high risk PP/PAS patients had substantially higher rates of maternal morbidity. MRI may be recommended in major PP/PAS patients who are older with repeated cesarean deliveries and having equivocal results or suspected deep/extra-uterine extension on US. Limitations of the study were that it was single-center with a relatively small sample size, and there was lack of complete histopathological diagnosis and MRI is costly.

Funding Statement

None