Detecting Accuracy of Three Dimensional Power Doppler (3DPD) Vascular Indices for Prenatal Diagnosis of Morbidly Adherent Placenta in Patients with Placenta Previa

Objective: The study’s objective was to assess the accuracy of using prenatal 3-dimensional power Doppler analysis of vascular placental indices to accurately diagnose morbidly adherent placenta objectively. Background: Traditionally, 2D ultrasound was used for the diagnosis of a suspected morbidly adherent placenta (MAP) previa. More objective techniques like 3D power Doppler haven’t been well studied. Study Design: A prospective cohort study is designed for women with gestational age between 28 and 32 weeks with suspected placenta previa. Patients were examined by 2D ultrasound which was used in management decisions. 3D Power Doppler’s VI, FI and VFI were measured during the same examination after manual tracing of placenta; data were blinded to obstetricians. Histopathology was performed to confirm MAP. Results: Our results showed that the 3D power Doppler VI ≥ 16 predicted the diagnosis of MAP with 100% sensitivity, 100% specificity which is better than those of 2D ultrasound. While VI > 33.1 measured by 3D Doppler predicted severe MAP with a sensitivity of 73.9% and specificity of 86.4%, which was superior to 2D ultrasound. Conclusion: In patients with placenta previa, the 3D Doppler’s vascular index accurately predicts MAP. Furthermore, vascular and vascular flow indices of 3D Doppler were more predictive of severe cases of MAP compared to 2D ultrasound.


Introduction
Normally, the placenta attaches to the wall of the uterus, however, sometimes the placenta invades deeply into the uterine wall that part or all of the placenta remains attached [1]. Recently, the term "morbidly adherent placenta" (MAP) has been more frequently used to describe abnormal implantation of the placental villi into the uterine wall [2]. It is histopathological term-occurs when the placenta fails to detach from the wall of the uterus due to abnormal implantation at the basal plate [3].
Approximately 1 out of every 533 pregnancies, according to the American College of Obstetricians and Gynecologists (ACOG) experience placenta accrete, percreta or increta [3]. The difference between percreta, increta or accreta which are the pathological types used to describe placental invasion, is determined by how deep the placental villi invade into the uterine wall [4] [5] [6].
There is a systematic review showed a wide heterogeneity in terminology used to describe the grades of accreta placentation using the 2D parameters. The authors concluded that the evaluation of the accuracy of ultrasound imaging in screening and diagnosing placenta accreta is limited [5]. In the same review, when combining all cases of placenta accreta, increta and percreta, the loss of sonolucency was the most common finding [5].
Several studies used the 3D color Doppler techniques to diagnose MAP and reported a sensitivity ranging between 39% and 100% based on abnormal vascular patterns [7] [8]. In these studies, the placenta was assessed subjectively for abnormal vascularity using the 3D power Doppler technique without any quantification. Shih et al. made the diagnosis of placenta accreta when intraplacental hyper vascularity, inseparable cotyledonal (fetal) and intervillous (maternal) circulations, or tortuous vascularity was noted [7]. Hyper vascularity of the uterine serosa-bladder wall interface and tortuous confluent vessels across the placental width were used as diagnostic criteria for morbidly adherent placenta by Collins et al. [8].
A recent study quantified the amount of vessels involved in the diagnosis of morbidly adherent placenta [9]. However, this was applied to the utero-placental interface only and not the entire placenta. They measured the largest area of confluent 3D power Doppler signal [9]. Although abnormal vascularity of the utero-placental interface is highly suggestive of abnormal placentation, abnormal vessels running throughout the placenta could be missed using this particular technique. In addition, choosing the most confluent area is operator dependent and other areas might have been missed due to subjectivity of the operator's assessment [10].

Aim of the Work
The study's objective was to assess the accuracy of prenatal 3-dimensional power Doppler to predict the morbidly adherent placenta objectively.  Exclusion criteria: Fetal anomalies and multiple gestations.
Combination of trans abdominal and transvaginal ultrasound performed to confirm the location of placenta using routine 2D imaging using Voluson E8 GE system equipped with 4 -8 MHz transducers. Uniform diagnostic criteria were applied to diagnose the suspected MAP using at least one of the following findings: 1) The presence of hypo echoic areas in the body of the placenta (placental lacunae).
2) The loss of the normal hypo echoic myometrium adjacent to the base of the placenta (loss of sonolucency).
3) Absent or disrupted hyper echoic line separating the uterus from the urinary bladder (abnormal uterine serosa-bladder line).
They have all been identified as markers of MAP on 2D ultrasound [2].
Pregnant women with at least one identified 2D findings, underwent an additional imaging using 3D power Dopplertrans abdominal ultrasound of the placenta. After full bladder confirmation, obtained using a GE Voluson E8 RAB4-8 transducer probe this technique was previously described for evaluation of placental vascularity [11] [12].
The study was approved by the Ethical Committee of Faculty of Medicine, Ain-Shams University. Verbal consent is taken from all the enrolled patients after explaining the aim of the study and was voluntarily participated when it was clear that it was an observational study that has no harm to them or to their babies.
The placental images were optimized for every patient to visualize the maximum thickness of the placenta with the suspected MAP at the center of the imaging area. Histogram analysis was applied to the taken images. The 3D placental volumes were assessed by manual tracing at 30˚ angle increments to include the maximum viewed placenta. The vascularization index (VI; which is calculated by dividing colored/total voxels, voxels are the cubes that occupy the volume of interest, which is in this case the placenta), the flow index (FI; the average of the color value of all blood flow or the mean intensity of the colored voxels), and the vascular flow index (VFI; is obtained by multiplying VI and FI and dividing the result by 100) which is calculated using the same software. Data was blinded to the maternal fetal medicine specialist interpreting the 2D ultrasound findings and making recommendations for the decision making of the patient's management of care. 2D results were blinded too to the sonographer performing Open Journal of Obstetrics and Gynecology the 3D examination and calculating VI, FI and VFI.
All patients were delivered by caesarean section followed by hysterectomy (cesarean hysterectomy) as planned by the operative physician-based on combination of suspected diagnosis from 2D ultrasound and clinical assessment which were used as tools to make the decision before surgery.
If 2D findings were not suggestive of MAP, hysterectomy was performed when the placental invasion was obvious with increased vascular markings and visualization of placenta through the uterine window, when the placenta failed to separate spontaneously or when there is resistance to manual placental separation with or without hemorrhage was encountered.
The final diagnosis of MAP for the analysis in this study was made based on the pathological examination of hysterectomy specimen by a histo-pathologist.
The patients were divided according to the collected data into 2 groups (group A and group B). Severe MAP (sMAP) which was identified for this study by the presence of the following components: Increta or percreta placenta on histopathology.
Transfusion of ≥ 2 units of PRBCs and estimated blood loss > 2000 ml.

Statistic Analysis Method
Descriptive statistics were done for quantitative data as minimum & maximum of the range as well as mean ± SD (standard deviation) for quantitative normally distributed data, median and 1st & 3rd inter-quartile range for quantitative non-normally distributed data, while it was done for qualitative data as number and percentage.
Inferential analyses were done for quantitative variables using Shapiro-Wilk test for normality testing, independent t-test in cases of two independent groups with normally distributed data and, ANOVA test with for more than two independent groups with normally distributed data and Kruskal Wallis test with non-normally distributed data. In qualitative data, inferential analyses for independent variables were done using Chi square test for differences between proportions and Fisher's Exact test for variables with small expected numbers as well as Kappa and test for agreement between paired categorical data.
K values > 80% were considered very good agreement [15]. Post hoc Bonferroni test was done to find homogenous groups in significant tests among more than two independent groups. ROC curve used to assess the performance of different tests and differentiate between certain groups. The level of significance at

Results
100 pregnant patients were included in our study ( Figure 1). All cases were delivered by C.S 97% of them underwent LUS C.S. Among MAP group 6 cases suffered from visceral injuries and 2 cases were admitted to the ICU with mean estimated blood loss of 2.4 ± 1.3 L (Table 1).      the diagnosis of MAP with a 100% sensitivity and 100.0% specificity (Table 4).

Prediction of Severe MAP (sMAP) Using 3D Doppler Indices
sMAP was found in 23/45 (51.2%) patients. The mean of calculated blood loss volume was on average 3.5 ± 0.9 L. The placenta pathology in the sMAP was 7 increta and 11 percreta. Similarly, using VI, FI, and VFI measured by 3D power Doppler for the prediction of severe MAP. VI of more than 33.1 predicted sMAP with a sensitivity of 73.9% and 86.4% specificity (Table 3).
ROC curve was used to evaluate the ability VI, FI, VFI to predict (sMAP) and an (AUC) was calculated for each parameter (Table 2, Figure 5). Among the 3D Doppler indices, VI had the highest AUC and VFI had the lowest (Table 2). Table 2(b) and Figure 5: Only VI had significant moderate diagnostic performance in differentiating sMAP from non-sMAP. Table 3 shows that: VI ≥ 33.1 had highest diagnostic characteristics in differentiating sMAP from non-sMAP.

Comparing 2D Findings and 3d Findings
Comparison between obtained data of all the studied cases to assess the best diagnostic method for prediction of MAP and sMAP weather 2D or 3D power Doppler. This comparison is better seen in Table 3 and Table 4. Our study assessed the entire viewed placenta. We found a significant increase of VI in MAP cases in the current study. This strongly suggests that the use of the 3D Doppler vascular index (VI) can accurately predict the diagnosis of morbidly adherent placenta by quantifying the colored voxels within the placental volume indicating an increased vascularity. Furthermore, the values are higher when the clinical severity is increased. The sensitivity and the negative predictive value in predicting sMAP can reach 100% when a threshold value for VI ≥ 33.1 is used. Similar increases were seen with VFI. However, we found that FI was not increased in association with MAP. Table 4: VI ≥ 16.0 and VFI ≥ 11.0 had perfect diagnostic characteristics in differentiating MAP from non-MAP.
2D and FI ≥ 42.8 had significant low diagnostic characteristics in differentiating MAP from non-MAP.

Discussion
The incidence of MAP has continued to rise primarily due to the increase in    cesarean delivery rates with an overall incidence of approximately 1 in 533 pregnancies [3]. MAP can result in major maternal morbidities such as life-threatening hemorrhage and intra-operative organ injuries [16]. Therefore, it is important to be able to detect this entity antenatally with high reliability to allow for a proper preparation for delivery in an appropriate unit equipped to handle potentially complicated surgery [17] [18].

Findings and Interpretation
Our study assessed the entire viewed placenta. We found a significant increase of indicating an increased vascularity with a 100% sensitivity and 100.0% specificity. Furthermore, the indices are higher when the clinical severity is increased.
The sensitivity of 73.9%, 86.4% specificity and the negative predictive value in predicting sMAP was 90.7% when a threshold value for VI of greater than 33.1 is used. Similar increases were seen with VFI. However, we found that FI was not increased in association with MAP.
In the current study the sensitivity and the specificity of 2D ultrasound for the diagnosis of MAP was 60.0% and 89.1% respectively when at least one abnormal parameter was present. The sensitivity and specificity dropped to 69.6% and 50.0% respectively when severe MAP was the outcome.
There is a significant difference in the ability of prenatal prediction of MAP and sMAP between 2D ultrasound and 3D power Doppler which state that the 3D power Doppler is a more accurate diagnostic tool of MAP than traditional 2D ultrasound.

Comparison with Previous Studies
Previously reported studies to diagnose MAP by 2D ultrasound have concluded a range between 76% -94% and 74% -99% for sensitivity and specificity respec- [20]. This wide range is probably attributed to the subjectivity of the 2D parameters and the variation in sample size.
Another factor that affected this range was the review by a single examiner.
When multiple examiners blindly reviewed every ultrasound examination, the sensitivity and the specificity were found to be even lower (63% and 87% respectively) [21]. There was a systematic review showed a wide heterogeneity in terminology used to describe the grades of accreta placentation using the 2D parameters [22]. The authors concluded that the evaluation of the accuracy of ultrasound imaging in screening and diagnosing placenta accreta is limited [22].
All these factors were avoided in our study by using the latest and the most accurate description of 2D findings, considerable sample size and the examination was reviewed by single examiner for 2D ultrasound and another for 3D power Doppler.
Several studies used the 3D color Doppler technique to diagnose MAP and reported a sensitivity ranging between 39% and 100% based on abnormal vascular patterns [8] [23] [24]. In these studies, the placenta was assessed subjectively for abnormal vascularity using the 3D power Doppler technique without any quantification.
Another study quantified the amount of vessels involved in the diagnosis of morbidly adherent placenta [9]. However, this was applied to the utero-placental interface only and not the entire placenta. They measured the largest area of confluent 3D power Doppler signal [9]. Although abnormal vascularity of the Open Journal of Obstetrics and Gynecology utero-placenta interface is highly suggestive of abnormal placentation, abnormal vessels running throughout the placenta could be missed using this particular technique. In addition, choosing the most confluent area is operator dependent and other areas might have been missed due to subjectivity of the operator's assessment. While in our study the placenta was traced manually with examination of vascularity of placenta and utero-vesicle space and at the maximum placental thickness using offline analysis of an optimized 3d Doppler images and the vascular indices were measured to decrease the percentage of subjective operator's error.

Clinical Implications
Clinically, the use of VI can detect a number of cases with invasive placentation that the conventional 2D ultrasound cannot, which allows for a better preparation for surgery.
Despite the limitations, the major strength of our study is the objective nature of the 3D Doppler technique for the diagnosis of MAP. The technique has a short learning curve and is reproducible. It has the potential to reduce the variation between examiners and increase the rate of antenatal detection of MAP leading to a decrease in maternal and fetal morbidities.

Weaknesses and Strengths
The small sample size is one of the limitations of our study. In addition, only women with placenta previa were included as we considered this a risk factor for MAP that would allow us to have a better yield for this disease in our study population. MAP may also occur in women without placenta previa; however our study did not address this population. Our study period was limited to an interval of gestational age of 28 -32 weeks and diagnosing the condition prior to 28 weeks gestation needs to be studied further. We also could not adjust for other factors such as the number of previous CD due to the sample size. Additionally, 3D color Doppler volume studies are currently available only in the GE Voluson machines.
This may not apply to other technologies and with changing technology the indices may change for the diagnosis, which will need to be revalidated. Our study did not assess MRI ability to predict MAP in comparison to 3D Doppler indices. However, MRI is more expensive and less readily available.
Despite the limitations, the major strength of our study is the objective nature of the 3D Doppler technique for the diagnosis of MAP. The technique has a short learning curve and is reproducible. It has the potential to reduce the variation between examiners and increase the rate of antenatal detection of MAP leading to a decrease in maternal and fetal morbidities.

Conclusions
In this study, our results showed that the 3D color Doppler VI ≥ 16 predicted the diagnosis of MAP with a 100% sensitivity, 100% specificity which are better than those of 2D ultrasound (60.0% and 89.1% respectively).
Severe MAP occurred in 51.2% of MAP and 3D color Doppler of VI > 33.1 predicted severe MAP with a sensitivity of 73.9% and specificity of 86.4%, which was superior to 2D ultrasound.
In the current study, the sensitivity and the specificity of 2D ultrasound for Open Journal of Obstetrics and Gynecology the diagnosis of MAP were 60.0% and 89.1% respectively when at least one abnormal parameter was present. The sensitivity and specificity dropped to 69.6% and 50.0% respectively when severe MAP was the outcome.

Future Research Implications
3D power Doppler is introduced as an objective technique that may limit the variations in diagnosing MAP due to the subjectivity of 2D ultrasound interpretations. So, a multi-center prospective study is needed to validate our findings, and confirm the usefulness of this methodology and its ability to improve the prenatal diagnosis of this condition compared to 2D ultrasound before an appropriate clinical application is determined.
Other available imaging technologies such as placental elastography and 3D MSV Doppler should be assessed to measure these indices with a larger sample size to design definitive prenatal diagnostic measures of MAP to decrease maternal and fetal morbidity and mortality.