MRI-Derived Uterine Sector 2 Placental Area and Cervical Area Are Associated with Massive Hemorrhage in Complete Placenta Previa with Placenta Accreta Spectrum

Introduction

Placenta accreta spectrum (PAS) disorders is defined as the invasion of the chorion into the myometrium, including placenta accreta, increta and percreta.¹ PAS is an obstetric complication related to severe maternal morbidity and mortality, such as postpartum hemorrhage, peripartum hysterectomy, mechanical ventilation, infection, intensive care unit admission, prolonged hospitalization, psychological disturbances, and even maternal death.^2,3 Prior cesarean section and placenta previa are high-risk factors for PAS.⁴ Imaging examination for PAS is essential for evaluating haemorrhage risk and determining treatment options before surgery. Ultrasound is widely used for the initial diagnosis of PAS because it is non-invasive and widely available. Magnetic resonance imaging (MRI) studies on PAS have recently focused more on the evaluation of relevant MRI features to predict the bleeding risk and the adverse outcomes of the maternal caesarean section.^5,6 MRI provides better soft-tissue contrast and multiplanar imaging than ultrasound, which is useful in cases of equivocal ultrasound findings or posterior placenta.⁷

Depending on the anatomic origin of the blood vessels, uterine vascular areas were divided into S1 (above the peritoneal reflection) and S2 (below the peritoneal reflection). The blood supply to the S2 area is provided by 5 branches of the retroperitoneal vessels, namely the superior, middle and inferior vaginal arteries, the bladder artery, and the internal pudendal artery. Bilateral uterine artery embolization or ligature was effective for hemostasis in S1 area, but ineffective in S2 area.⁸ Moreover, the operation space in S2 area was narrow and hemostasis was difficult, which might lead to massive hemorrhage (MH) (estimated blood loss >2000 mL) during cesarean section.⁹ When the amount of bleeding exceeds 2000 mL, the patients usually have severe coagulopathy.¹⁰ It has been reported that large placental area and short cervical length might be associated with a high risk of MH and adverse maternal–fetal perinatal outcomes in patients with complete placenta previa.^6,11 However, in the presence of PAS, the cervix and lower segment are often distorted. Measuring the placental area in sector 2 and the cervical area may provide a more reliable and reproducible indicator of the extent of placental invasion and the resulting distortion of the cervix. Reducing blood loss during cesarean section and maternal mortality is an important responsibility of obstetricians. A comprehensive preoperative MRI assessment facilitates the tailoring of management strategies (referral to PAS center, planned hysterectomy vs conservative intent, blood product preparation, interventional radiology planning, et al) to the individual patient’s risk profile in cases of PAS. Therefore, the risk assessment of MH before delivery and appropriate treatment are important ways to improve the clinical outcomes in PAS patients with complete placenta previa and perinatal infants. We speculated that further evaluation of placenta area in S2 and cervical area in sagittal plane by MRI would be helpful to assess the risk of MH before surgery and choose appropriate treatment.

Therefore, the purpose of our study was to explore the role of placenta area in S2 and cervical area in identifying MH in complete placenta previa patients with PAS.

Materials and Methods

Study Population and MRI Protocol

The retrospective study was approved by the Institutional Review Board (IRB) of the affiliated Suzhou Hospital of Nanjing Medical University (K-2022-015-K01), and an informed verbal consent was obtained from all the participants. Patients were excluded due to not delivering in our hospital (n=117), not undergoing MRI examination (n=220), poor image quality (n=11), or insufficient urine in the bladder on MRI (n=32). At last, 138 complete placenta previa women who were diagnosed to have PAS after cesarean section, between January 2016 and December 2023, were reviewed carefully (Figure 1). Diagnosis of complete placenta previa was made by transabdominal ultrasound in the third trimester of pregnancy when the lower placental edge covered completely the internal os of the uterine cervix. The amount of bleeding during all the cesarean sections was measured using the weighing method. PAS women were grouped in terms of the amount of blood loss during cesarean section, MH (estimated blood loss >2000 mL) and non-MH groups (estimated blood loss ≤2000 mL). The diagnosis of PAS was based on the clinical criteria of conservative treatment or pathological diagnosis after hysterectomy according to FIGO criteria.^12,13

Figure 1 Flow chart of PAS patients with complete placenta previa included in the study.

All placental MRIs were performed on a 3-T MRI (Siemens Medical Solutions, Erlangen, Germany). One hundred and thirty-eight patients with bladder moderately full (about 400–500 mL urine) who underwent placenta MRI by T₂-weighted half-fourier acquisition single-shot turbo spin echo (HASTE) sequence (TR 700 ms, TE 87 ms, 432×432 matrix over a field of view of 380×380 mm, 5 mm slice thickness). According to the anatomical classification of PAS proposed by Palacios Jaraquemada et al, a line perpendicular to and across the midpoint of the posterior wall of the bladder was drawn, the area above the line was S1 and the area below the line was S2 (Figure 2)¹⁴ From an anatomical perspective, S1 refers to the body of the uterus, S2 refers to the lower segment of the uterus and the cervix. The region of interest was manually traced along the outer contour of the cervix, extending from the internal os to the external os. The tracing followed the cervical stroma, excluding the endocervical canal. The lateral boundaries were defined by the anterior and posterior cervical lips. All tracings were performed independently by two experienced radiologists (>20 years in obstetric MRI) who were blinded to the patients’ hemorrhage status and all other clinical outcomes. The average of the two measurements was used for statistical analysis, and any disagreements were resolved by consensus. The placental area in S2 and the cervical area were measured on two-dimensional (2D) MRI. For each patient, the mid-sagittal slice that best visualized the entire cervical canal was selected for analysis. On this single representative slice, the placenta area in S2 and cervical area were traced by Image J software version 1.50 (National Institute of Health, Bethesda, USA), and then their areas were calculated.

Figure 2 (a) Sagittal placenta image of a patient with PAS by MRI. (b) The pelvis of PAS patient was divided into S1 and S2 sectors by line S that passed vertically through the midpoint of the posterior wall of bladder in sagittal MRI image. (c) Placenta area in S2 was marked with green paint and measured in sagittal MRI image (62.35 cm2). (d) Cervical area was marked with blue paint and measured in sagittal MRI image (4.68 cm2).

Statistical Analysis

The data was analyzed by SPSS Version 23.0 (SPSS Inc., Chicago, IL, USA) statistical package. Continuous variables were reported as mean ± standard deviation (SD) and compared using Student’s t test. Categorical data were presented as number and percentage and compared by chi-squared test. Non-normally distributed numerical data were presented as median and inter-quartile, and their differences were compared by Mann–Whitney test. Multivariable logistic regression analysis was used to identify MRI features that may lead to MH. Spearman analysis was performed to analyze the correlation between placental area in S2, cervical area and amount of blood loss. Receiver operating characteristics (ROC) curves were constructed to determine the accuracy of placenta area in S2 and cervical area to identify PAS patients at high risk for MH. Intraclass correlation coefficients (ICCs) were calculated to determine the agreement of placenta area in S2 and cervical area between the radiologists according to the following definitions: excellent agreement (≥0.90), good agreement (range, 0.75 to 0.90), moderate agreement (range, 0.50 to 0.75), poor agreement (<0.50). A two-tailed value of P<0.05 was considered statistically significant.

Results

A total of 138 PAS patients were finally included in the present study. The general characteristics of participants were listed in Table 1, including maternal age, BMI, gravidity, parity, number of abortions, number of previous cesarean delivery, previous history of placental previa, gestational age during MRI, and prenatal vaginal bleeding showed no differences between the two groups. There was a significant difference between patients with or without MH with regard to gestational age at delivery (P=0.003), neonatal birth weight (P=0.001), operation time (P=0.003), intraoperative blood loss (P<0.001), and blood transfusion (P<0.001). ICU admission and hysterectomy were also significantly greater in patients with MH (P = 0.012 and P = 0.014, respectively).

Table 1 Patients’ Clinical Features of the Study Groups

The ICC for the measurement of the placenta area in sector 2 was 0.913 (95% confidence interval [CI]: 0.882–0.937), and for cervical area it was 0.905 (95% CI: 0.764–0.931) (Table 2). The placenta area in S2 of the MH group was significantly higher than that of the group without MH (P<0.001), on the contrary, cervical area was significantly lower in PAS patients with MH than that of the group without MH (P<0.001) (Figure 3). After adjusting for the type of PAS, we used placenta area and cervical area for identifying MH, which were shown to have good performance that placenta area in S2 ≥45 cm² (OR, 4.805; 95% CI, 2.776–8.794; P<0.05) and cervical area <4.0 cm² (OR, 4.015; 95% CI, 1.905–7.862; P<0.05) (Figure 4). ROC analysis revealed that the area under the curve (AUC) of placenta area in S2 in identifying MH was 0.853 and that the optimal cut-off was 45.0 cm², with a sensitivity of 86.431% and a specificity of 85.335%. For predicting MH, the AUC of cervical area was 0.831, and the optimal cut-off was 4.0 cm², the sensitivity was 85.548%, and the specificity was 82.504% (Table 3 and Figure 5). When we combined placental area in S2 and cervical area, the sensitivity, specificity, and AUC for the predictive MH were significantly improved, and their values were 89.237%, 91.548%, and 0.910 (95% CI 0.858 ~ 0.963), respectively.

Table 2 Interobserver Reliability of MRI in the Measurement of MRI Features

Table 3 Receiver Operating Characteristic Analyses for Prediction of Massive Hemorrhage Based on Cervical Length and Cervical Area

Figure 3 (a) Placenta area in S2 in PAS patients with and without MH. (b) Cervical area in PAS patients with and without MH.

Figure 4 Multivariate logistic regression analysis of risk factors for patients with MH.

Figure 5 Receiver operating characteristic (ROC) curve of different magnetic resonance imaging (MRI) features in patients with placenta accreta spectrum disorders (PAS).

In PAS patients, a positive association between placental area in S2 and amount of blood loss (r=0.748) was observed, and negative association was observed between cervical area and amount of blood loss (r=−0.682), between cervical area and placenta area (r=−0.575). As the placental area in S2 became larger and cervical area became smaller, the bubbles representing the amount of blood loss tended to increase gradually (Figure 6).

Figure 6 (a) Association between placental area in S2 and amount of blood loss in PAS patients (r= 0.752). (b) Association between cervical area and amount of blood loss in PAS patients (r=−0.671). (c) Association between placental area in S2 and cervical area in PAS patients (r=−0.559). (d) Association between placental area in S2, cervical area and amount of blood loss in PAS patients (bubble sizes denote the amount of blood loss).

Discussion

Complete placenta previa women with prior cesarean section are at increased risk of MH requiring an emergency cesarean delivery that is difficult to be dealed with for obstetricians. Therefore, it is necessary to assess perioperatively the risk of MH in those patients with either ultrasound or MRI investigations. Ultrasound is still an important means of diagnosing PAS, but MRI can provide more information about the depth of placenta invasion.¹⁵ To our knowledge, most of the studies focused on the influence of cervical length on maternal and infant perinatal outcomes in PAS patients,^16,17 and the studies on cervical area in PAS were rare. In this study, we focused on addressing placenta area in S2 and cervical area as indicators that might help to anticipate MH during caesarean section.

An increased thickness of the lower placental edge had been reported by some to increase the risk of vaginal bleeding in patients with placenta previa.¹⁸ To stop MH in the uterus, not only the irrigation arterial pedicles of uterus, but also their anastomotic components at the bleeding sector should be managed. Knowledge of the blood supply of uterine sectors (S1 and S2) is a prerequisite to stop bleeding accurately and rapidly during surgery. The blood supply in S2 is more complex than that in S1, and the operating space in S2 is narrower, so the bleeding in the S2 is difficult to control, resulting in an increased risk of MH in PAS patients. The S1 (above the peritoneal reflection) and S2 (below the peritoneal reflection) regions of the uterus are supplied by different blood vessels in the pelvis. The S1 is supplied by ascendent branches of the uterine artery and the descendent branches of the ovarian artery, and the S2 is supplied by uterine, cervical, upper vesical, vaginal, iliac internal, and pudendal artery branches.⁸

The findings from this retrospective study showed that prenatal MRI has promising discriminatory ability in identifying MH during cesarean section. In this study, using ROC curve, if placenta area in S2 >45 cm² was taken as a cut-off, the risk of MH was significantly higher as compared to cases with placenta area in S2 below this threshold. PAS patients with placental accreta that occur in the S2 were at a higher risk of complications during surgery.¹⁹ The study of Palacios Jaraquemada et al found that up to 87.6% of placenta implantation occurred in the S2 region of the lower uterine segment.¹⁴ Uterine ligature, occlusion, or embolization was very effective for uterine bleeding in the S1 region, and the efficiency was even more than 90%.²⁰ Placental accreta in the S2 region was an important cause of postpartum hemorrhage in PAS patients, and even internal iliac vascular ligation was ineffective.²¹

Shorter cervical length was strongly associated with MH and adverse maternal–fetal outcomes in patients with PAS.¹¹ Many studies had focused on the association between cervical length and adverse perinatal outcomes in PAS patients. In this study, cervical area was strongly associated with the development of MH during surgery. PAS patients with placenta previa with a short cervix have a higher risk of MH, which may be caused by the low position of the placenta leading to dilated lower uterine segment and cervix, and the lack of muscle tissue that can constrict the torn vessels after placental separation.^22,23 In late pregnancy, the placenta previa tends to pull the cervix, which caused the cervical length to shorten and the cervical area to shrink.²⁴ Some researchers had pointed out that shortened cervical length means placenta previa cannot attach smoothly to the lower segment of the uterus and the cervix, which increased the risk of MH.²⁵ Cervical area includes the length and thickness of the cervix, which may be able to reflect the cervical condition of PAS patients more comprehensively to a certain extent.

In this study, we showed that the placenta area in S2 of the MH group was significantly higher than that of the group without MH, on the contrary, cervical area was significantly lower in PAS patients with MH than that of the group without MH. Our findings regarding a possible association between placenta area in S2, cervical area and the risk of MH in complete placenta previa patients with PAS may improve our ability to predict adverse clinical outcomes and to refine obstetric management in these cases.

One strength of our study was its clinically relevant population – PAS women were grouped in terms of the amount of blood loss during cesarean section, MH (estimated blood loss >2000 mL) and non-MH groups (estimated blood loss ≤2000 mL). These preliminary findings suggest that the placenta area in sector 2 and cervical area may contribute to identifying patients at potentially elevated risk of MH in patients with PAS. Another strength was that our study allowed us to assess the risk of MH in PAS patients before cesarean section and take well prepared for surgery.

Several limitations of this study should be acknowledged. First, the limitations of this study were its retrospective nature, the small number of women enrolled, and the fact that it involved only a single institution. The study may have selection bias due to MRI eligibility and image quality requirements. Second, the study period spanned from 2016 to 2023, during which there may have been subtle changes in perioperative management, surgical techniques, or transfusion thresholds that could influence outcomes but were not fully captured in the analysis. Third, different obstetric surgeons may have deviations in their estimations of the intraoperative blood loss for patients with PAS.

Conclusion

In summary, the placental area in S2 and the cervical area are associated with the risk of MH during cesarean section in complete placenta previa patients with PAS disorders. These MRI findings may aid in preoperative risk recognition. Due to the lack of validation, the adjusted analyses and validation are needed before clinical deployment.