2023 年 11 巻 4 号 p. 57-62
Aim: Among the most serious complications of placenta previa is bleeding during cesarean delivery. We developed a method of predicting bleeding risk using predelivery MRI or ultrasonography and discuss our findings.
Methods: We conducted a retrospective cohort study of cesarean deliveries involving placenta previa between January 2014 and March 2020. We assessed ultrasonography and MRI measurements of the distance from the short side of the placental edge to the internal cervical os (PI) and assessed MRI measurements of placental height (PH). We then analyzed the relation between blood loss and each of these measurements.
Results: Of 57 qualifying deliveries, total previa was present in 38 patients (66.7%), and marginal or partial previa in 19 (33.3%). Median blood loss was 1,510 ml (interquartile range, 1,170–2,000 ml). PI distance was positively correlated with blood loss; measurements at 30 to 32 gestational weeks were more correlated than at 35 to 37 gestational weeks by ultrasonography. Moreover, MRI measurements at 33 to 34 gestational weeks were much more correlated than ultrasonography. There was also a positive correlation for PH.
Conclusion: Measurements of PI and PH, particularly measured by MRI at 33 to 34 gestational weeks, are useful predictors of bleeding risk for patients with placenta previa.
Placenta previa (PP) can cause severe bleeding before or during cesarean section (CS), sometimes requiring massive transfusion.1,2) The position of the placenta affects the risk for postpartum hemorrhage. Traditionally, PP is classified into three types according to the distance from the short side of the placental edge to the internal cervical os. For purposes of this study, we are calling that distance PI. In complete PP, the placenta completely covers the internal os (PI≥20 mm); in partial PP, the placenta only partially covers the internal os (0<PI<20 mm); and in marginal PP, the placenta just reaches the internal os but does not cover it (PI=0 mm).3) Typically, PP is diagnosed using transvaginal ultrasound.4,5) Because the frequency of vaginal bleeding gradually increases after 28 gestational weeks, it is highly desirable to diagnose PP no later than 32 gestational weeks.6,7) At our hospital, we perform routine transvaginal ultrasonography at 16 to 18 gestational weeks. If the placenta lies within a few centimeters of the cervical os, we repeat ultrasonography at 30 to 32 gestational weeks for diagnosis. After a diagnosis of PP, we use MRI performed at 32 to 34 gestational weeks. Compared to ultrasonography, MRI is a highly objective examination because the entire uterus can be viewed on the same slice and there is little inter-examiner error.
Previous studies have examined the correlation between blood loss and PI estimating the degree of cervical coverage in PP.8) According to one study, short cervical length (CxL) in the setting of PP is associated with massive bleeding.9) Other studies have looked into whether anterior, posterior, or lateral placenta also affects blood loss.10,11,12) Although various studies have been conducted, it is still difficult to accurately predict bleeding risk.
The aim of this study is to identify ultrasonography or MRI risk factors for massive intraoperative bleeding. First, we analyzed the relationship between the amount of bleeding during CS and PI distance measured using ultrasonography and MRI. Second, the timing of measurement was also examined because the appropriate timing of ultrasonography has not been clarified. Third, we analyzed the relationship between blood loss and placental parameters except PI distance (cervical length and placental height) measured using MRI. To determine whether there is a relationship between the area of attachment of the entire placenta and the amount of hemorrhage, we examined the height from the cervical os to the placental edge as a simple indicator to estimate the size of the placenta.
We designed a retrospective cohort study and reviewed medical records for 14,192 deliveries performed at Juntendo University Shizuoka Hospital between January 2014 and March 2020.
Our practice during the study period was to make a final diagnosis of PP when the functional internal os was covered by the placenta, as assessed by transvaginal ultrasonography at around 30 to 32 gestational weeks. To evaluate bleeding risk for patients with PP, we used MRI performed at 32 to 34 gestational weeks and repeated ultrasonography before performing CS at 35 to 37 gestational weeks.
A total of 97 pregnancies involved PP and were delivered by CS during the study period. Figure 1 shows the process of selecting patients for study inclusion. We excluded patients who did not have MRI results available; we also excluded patients with placenta accreta (n=4), multiple gestations (n=2), uterine leiomyoma that makes it impossible to measure the placenta (n=2), and those who underwent preterm delivery at less than 35 gestational weeks (n=1). The remaining 57 patients were included after they provided informed consent for study inclusion.
This study was approved by the ethics committee of Juntendo University (approval number 821) and was conducted in accordance with the Declaration of Helsinki.
Methods of hemostasisAfter placental abruption, uterine contractions (oxytocin and methylergometrine maleate) are administered. In addition, if bleeding from the abruption surface cannot be controlled during CS, balloon tamponade is performed as an intraoperative hemostatic procedure. The balloon is inserted intraoperatively and the wound is closed while checking the effect. Immediately after surgery, the volume of the balloon is adjusted while checking the bleeding.
If balloon tamponade is ineffective and bleeding persists postoperatively, uterine artery embolization is performed with sufficient blood transfusion. In most cases, hemostasis can be achieved, but if bleeding persists, a total hysterectomy is performed.
Placental measurementsWe measure PI distance using ultrasonography at 30 to 32 gestational weeks for diagnosis and repeat ultrasonography at 35 to 37 gestational weeks prior to delivery. To accurately assess bleeding risk, we took three distinct measurements on MRI (Figure 2): PI distance, CxL, and placental height (PH).
PI was measured from the internal cervical os to the lower edge of the placenta, while CxL was measured from the internal os to the external os. To determine PH, we drew a straight line connecting the internal cervical os and the midsagittal line of the uterine cavity, then drew a perpendicular line from the far tip of the long side of the placenta. The distance from the internal cervical os to the intersection of the two lines was used to define PH. When the placenta was lateral side, perpendicular line from the highest point to the midsagittal view and from the point to the measuring line (Figure 3).
Patient background and outcomes were tabulated in terms of frequency and percentage. The association between blood loss and placental and cervical parameters was analyzed by Pearson’s product moment correlation coefficient, a simple linear regression analysis. Statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Inc., Cary, NC).
Data availabilityData supporting the findings of this study are available from the corresponding author on request.
Patient background factors related to risk factors for PP (patient age, parity, smoking status, history of prior CS, history of dilation and evacuation, and conception using assisted reproductive technology) are shown in Table 1.13,14,15) The median maternal age was 34 years. Total PP was present in 38 patients (66.7%), and marginal or partial PP was present in 19 (33.3%). Posterior PP was present in 42 patients (73.7%), anterior PP in 8 (14.0%), and lateral PP in 7 (12.3%). A total of 25 patients (43.9%) experienced sentinel bleeding.
| n=57 | % | |
|---|---|---|
| Age >35 years | 26 | 45.6 |
| Primiparous | 26 | 45.6 |
| Prior CS | 4 | 7.0 |
| Prior D&C | 7 | 12.3 |
| Conception by ART | 15 | 26.3 |
| Current smoker | 9 | 15.8 |
| Degree of PP | ||
| Total | 38 | 66.7 |
| Marginal/partial | 19 | 33.3 |
| Placental position | ||
| Anterior | 8 | 14.0 |
| Posterior | 42 | 73.7 |
| Lateral | 7 | 12.3 |
| Sentinel bleeding | 25 | 43.9 |
Abbreviations: ART, assisted reproductive technology; CS, cesarean section; D&C, dilation and curettage; PP, placenta previa.
Elective CS was planned at 36 to 37 gestational weeks, according to the patient’s level of risk. The details and outcomes of CS are shown in Table 2. Elective CS was performed in 52 patients (91.2%), and emergent CS in 5 (8.8%). In 7 patients (12.3%), the uterus was accessed by a transplacental approach. The median intraoperative blood loss was 1,510 ml (interquartile range, 1,170–2,000 ml), and the median operative time was 43 minutes (interquartile range, 39–57 min). The median birth weight was 2,726 g (interquartile range, 2,476–2,968 g). A total of 19 pregnancies (33.3%) were delivered at 36 gestational weeks, and 38 (66.7%) were delivered at 37 gestational weeks. Intrauterine balloon tamponade hemostasis was performed for 12 patients (21.1%), uterine artery embolization was performed for 1 patient (1.8%), and 1 patient (1.8%) required hysterectomy.
| n=57 | % | |
|---|---|---|
| Elective | 52 | 91.2 |
| Emergent | 5 | 8.8 |
| Transplacental | 7 | 12.3 |
| Massive hemorrhage (>2,400 ml) | 10 | 17.5 |
| GA at delivery (weeks+days) | ||
| 36+0–36+6 | 19 | 33.3 |
| 37+0–37+6 | 38 | 66.7 |
| Hemostasis | ||
| Intrauterine balloon | 12 | 21.1 |
| UAE | 1 | 1.8 |
| Hysterectomy | 1 | 1.8 |
Abbreviations: GA, gestational age; UAE, uterine artery embolization.
Blood loss was positively correlated with PI as measured by either MRI or ultrasonography, as is well known. Ultrasonographic PI measurement was more correlated with blood loss when measured at 30 to 32 gestational weeks than when measured at 35 to 37 gestational weeks (30–32 weeks: r=0.43377, P=0.0016, linear regression equation: y=16.2x+1,180.9; 35–37 weeks: r=0.40702, P=0.0101, linear regression equation: y=14.0x+1,376.1). However, MRI measurements were more correlated with blood loss than ultrasonography (MRI: r=0.48945, P=0.0001, linear regression equation: y=13.7x+1,202.6; Figures 4A–4C, Table 3). PH also was weakly correlated with blood loss (r=0.14402, P=0.2852, linear regression equation: y=2.9x+1,094.7). In addition, while short CxL in the setting of PP is reportedly associated with massive bleeding,9) dot plots and linear regression analyses revealed little correlation between CxL and blood loss in our patient cohort (r=−0.01687, P=0.9009, linear regression equation: y=−1.1x+1,709.3; Figures 4D, 4E).
| Length | n | Average length (mm) | Range (mm) | Correlation coefficient | P-value |
|---|---|---|---|---|---|
| PI (30–32 GW ultrasound) | 50 | 29.0 | 0.0–93.1 | 0.43377 | 0.0016 |
| PI (35–37 GW ultrasound) | 39 | 26.8 | 0.0–85.0 | 0.40702 | 0.0101 |
| PI (MRI) | 57 | 33.9 | 0.0–119.9 | 0.48945 | 0.0001 |
| CxL (MRI) | 57 | 37.5 | 12.4–70.0 | 0.01687 | 0.9009 |
| PH (MRI) | 57 | 196.5 | 111.8–278.8 | 0.14402 | 0.2852 |
To devise a method of accurate evaluation of bleeding risk for patients with PP, we analyzed the relationship between the amount of bleeding during CS and placental parameters measured using ultrasonography and MRI. We confirmed that a long PI confers a bleeding risk, and we determined that PI, as measured by MRI, reflects the bleeding risk more accurately than ultrasound measurements. The reason is that MRI has several advantages. First, MRI is a highly objective examination. Second, as the cervix is compressed by pressure from the transvaginal ultrasound probe, it is difficult to measure PI accurately; therefore, intra-examiner errors are smaller for MRI than for ultrasonography. Third, MRI images allow visualization of larger areas of the placenta than those of ultrasonography. Particularly with long PI distances, the cervical os and placental edge may not be detected clearly on ultrasonography.
Ultrasonography unexpectedly showed a larger correlation with blood loss when performed at 30 to 32 gestational weeks than at 35 to 37 gestational weeks. Although it seems logical that the placental position immediately before delivery should reflect the bleeding risk, placental ultrasonographic measurements at 35 to 37 gestational weeks could be affected by other factors. The amount of amniotic fluid—and the space it takes up—decreases in the third trimester, and the uterine wall around the cervical os is compressed and displaced by the fetal head. Additionally, it may be difficult to detect the boundary between the placenta and a hematoma in patients who experienced sentinel bleeding. Thus, apparent placental migration may not actually reduce the risk for massive bleeding.
We also found that PH was weakly correlated with blood loss. Longer placentas (longer PH) might have a greater attachment area, and main blood loss comes from the area of placental abruption. Therefore, longer PH might be one of the risks for massive bleeding in patients with PP. Although we devised a method of measuring PH so that we could easily include lateral placentas in our analysis, there is a limitation in the measurement. Since the measurement is done using sagittal sections, the anterior and posterior walls are sliced perpendicular to the plane of adhesion, but the lateral walls are sliced horizontal to the plane of adhesion. Therefore, coronal slices are used for measurement, as shown in Figure 3. However, lateral placentas are reportedly at lower risk for bleeding than placentas in other positions.12) We intend to consider the effects of wall position in a future study.
Although the area of placental separation is a more reasonable factor for predicting massive bleeding with PP, calculating this area is complex and not routinely performed at most MRI facilities. We intend to assess the area of placental separation directly in future research.
Our study has some limitations. Although some reports suggest an effect of placental position on the amount of blood loss, we measured all placentas (anterior, posterior, and lateral) the same way because of the small number of patients. Moreover, various parameters can also influence the amount of blood loss during CS for PP cases, e.g., previous CS, ART pregnancy, prior D&C, adenomyosis, and myoma (even if small). Although the present results do not take these into account, it will be important to collect this information and consider these factors in future studies.
Correlation coefficients were considered alone in this study, and we could not determine a cutoff value that indicates the risk for massive bleeding with statistical significance. Determining such a cutoff value may help decrease overtriage and overtreatment.
In conclusion, a long PI or PH could be an important factor for predicting massive bleeding, particularly as measured by MRI at 33 to 34 gestational weeks. Furthermore, placental migration detected immediately before CS might not contribute to risk reduction.
We are grateful to the members of the Clinical Research Promotion and Support Center for Medical Science, Juntendo University, for their technical support.
The authors declare no conflicts of interest for this article.
None.
