Essay: The accuracy of Ultrasound versus Magnetic resonance imaging in the diagnosis of placenta previa-accreta: review of the literature

KEYWORDS: diagnosis, accuracy, placenta previa, ultrasound, MRI, accreta.
Abstract
Purpose: To evaluate the accuracy of ultrasound (US) versus magnetic resonance imaging (MRI) in the diagnosis of placenta previa (accreta) to determine the most useful imaging modality in suspected cases.
Methods: a search was performed on the following databases: Medline, PubMed and Science direct for appropriately reviewed publications on the selected topic. The search used a combination of terms in each database such as *accuracy *ultrasound *Magnetic resonance imaging *placenta previa *accreta and *diagnosis. All published in English and varied publications in the United States and Egypt. Most databases provided search results of 55 papers, with science direct yielding the most relevant results. From 19 relevant articles, a selection of 10 papers were made, chosen studies fulfilled predefined criteria.
Results: The highest noted sensitivity for ultrasound was 95.1% by a study from (Warshak et al. 2006) and the lowest sensitivity was presented by Riteau et al. (2014) and was 35.5%.
Conclusion: MRI can help with signs of signs PPA not visible on ultrasound e.g. dark intraplacental bands, a distinct sign that can predict the presence of accreta. Most authors have come to the conclusion that wherever ultrasound falls short in diagnosing PPA, MRI should be as a tool for further investigation.
Introduction
Placenta previa accreta (PPA) can described as the atypical deep attachment of the placenta to the uterine wall. Classification of placenta previa accreta is classified on the basis of the depth of myometrium invasion (Marrinan, 2013). In the incidents of PPA increases with a known risk factor such as increased maternal age and can lead to emergency hysterectomy with life threatening chance of haemorrhaging, renal distress or failure, and possibly death (Committee Opinion No. 529, 2012).
Suspicion of PPA can be accessed through previous patient’s clinical history and imaging. Cases have risen in the United States due to the rising number of caesarean deliveries. According to RCR (2011) Obstetric haemorrhage is a major causes of maternal death in less developed countries, up to 50% of the estimated 500,000 maternal deaths that occur globally each year. Deaths from obstetric haemorrhage are uncommon in the United Kingdom. Previous caesarean sections mean a greater chance of PPA occurrence again. As the number of caesarean sections increases, so does the risk. Accurate diagnosis will allows for optimum obstetric management, preparations can be made to recruit a skilled surgical and anaesthetic team, time and place for delivery and transfusion arrangements (Perez-Delboy and Wright, 2013).
Ultrasound and magnetic resonance imaging have been the modalities of choice in the diagnosis of PPA, but the there are questions of bias from previous review on the high accuracy of ultrasound due to its dependency on the skills of the sonographer. Both by ultrasound and MRI have features for the efficient visualisation of PPA signs e.g. intraplacental lacunae for U/S and uterine bulging for MRI (Teo et al., 2009). The aim of this review is to assess and determine the accuracy of ultrasound and MRI in the diagnosis of placenta previa accreta and also to look into the specific features of each modality that aid in the diagnosis of PPA (Mar et al., 2015).
Methods
A search of the following databases was carried out: Medline, PubMed (National Library of Medicine) and Science direct appropriately reviewed publications on the accuracy of ultrasound versus MRI in the diagnosis of PPA. The search used a combination of terms in each database such as *accuracy *ultrasound *MRI *placenta previa *accreta and *diagnosis.
All published in English and varied publications in the United States, France and Egypt. Most databases provided search results of 55 papers, with science direct yielding the most relevant results. From 19 relevant articles, a selection of 10 papers were made, chosen studies fulfilled predefined criteria. The chosen papers were then read and 10 full text papers were finally reviewed.
Results
Important features which make each modality suitable in the detection of PPA are addressed by the literature. Parikh et al., (2008) Sensitivity is the ability of a test to correctly identify an individual who has disease present and specificity as the tests to correctly identify an individual as free from the disease. The percentage of patients with a negative test who do not have the disease (negative predictive value) must be low while (positive predictive value) for a diagnostic tool should be high (Naeger et al, 2013). The above qualities are needed for a test to be seen as accurate.
The highest noted sensitivity from all 10 reviewed articles for ultrasound was 95.1% by a study from (Warshak et al. 2006) and the lowest sensitivity was presented by Riteau et al. (2014) and was 35.5% this study was looking at accuracy for solely ultrasound. In a study by Lim et al. (2011) MRI correctly predicted PPA in 23 of 26 cases and reflected most of the literature as MRI had a relatively high sensitivity all through ranging from the 70-89% but when it came to specificity 50% was the lowest (Riteau et al. 2014) and a range as high as 65-100% from a study from Berkley and Abuhamad. (2013).
Studies such as Hamisa et al. (2014) Uses features of U/S and MRI such as intraplacental lacunae or uterine bulging to ascertain accuracy, does not mention percentages for sensitivity, specificity or PPV or NPV. Agrees with (Comstock and Bronsteen, 2013) for e.g. interruption of bladder uterine interface showed 93% sensitivity and 84% specificity.
Discussion
Accuracy: sensitivity, specificity, PPV & NPV.
In the study done by Mansour and Elkhyat (2011), transvaginal ultrasound was not performed in cases where pregnancy was advanced to avoid risks such as premature rupture of membrane or in cases of already ruptured membrane to avoid infection. In this study they presented the data of a sensitivity of 86.6 % and 80% specificity for ultrasound with an accuracy of 82.8%. They stated that U/S had true positive diagnosis of 86.6% and MRI in 93.3%. U/S displayed false positive prediction in four cases, and in 16 cases true negatives and false negatives in 2 cases. MRI showed less with false positive predictions in only three cases with true negatives in 17 cases, one more than U/S and false negatives in only one case. It is ascertained that U/S had a positive predictive value (PPV) of 89% while MRI had 94.4%. Negative predictive values (NPV) were higher in MRI than U/S by 5.8%.
Bowman et al. (2014) examined 55 patients with PPA. All had ultrasound only to access the accuracy of just ultrasound. In the end they ascertained from their independent observations that U/S had a Sensitivity of 53.5%, specificity of 88.0%, positive predictive value of 82.1%, negative predictive value 64.8%, and accuracy were and 64.8%. 27% true positives, 5.9% false positives and for false negatives 16% and 12.3% had uncertain diagnosis. This study unlike Mansour and Elkhyat (2011) find ultrasound highly specific but disagree on how sensitive it can be as it is much lower in this study by bowman. It is stated that ultrasound may not be as sensitive as previously described by other studies. The accuracy for ultrasound for this study was lower as well. Hamisa et al., (2014) the mean age amongst women in this study was 32.25 years. 25 cases of previa were diagnosed and 4 complicated by accreta. This study had showed and agreed with the U/S and MRI features demonstrating similar findings of PPA as found in the Bowman study.
Written in three stages according to the first to third trimesters of pregnancy, the authors of this literature used specific findings of PPA in both U/S and MRI to ascertain accuracy of both modalities. Utility of clear space in the diagnosis of accreta was a sensitivity of 73% and a low sensitivity (20%) if using interrupted bladder line in diagnosis of PPA but high is PPV of 96%, sensitivity may be low as not all women had transvaginal ultrasound scans. The authors determined that the more lacunae the more likely the presence of PPA (Comstock and Bronsteen, 2013). Just like in the Comstock paper, Berkley and Abuhamad, (2013) also discusses specific signs seen in the three trimesters such as lacunae, myometrium thickness etc. grayscale sonography is the tool of choice in prenatal diagnosis of placenta accreta. With sensitivity has been reported in the range of 77% to 87%, specificity of 96% to 98%, PPV of 65% to 93%, and an NPV of 98%. The overall reported sensitivity of MRI is 80% to 85% with specificity of 65% to 100%.
In the Riteau et al. (2014) paper, 42 patients underwent both ultrasound and MRI to access suspected cases of PPA. Ultrasound was successful at diagnosing all 26 cases of PPA but also diagnosed adherent placenta. Sensitivity for U/S diagnosing accreta was 100% and 76.2% for MRI, specificity for U/S was 37.5% and for MRI 50%. Correct diagnosis was made using U/S in 76.2 of the cases and 66.7 with MRI. Specificity is very low in U/S in this study but higher for MRI. Dwyer et al. (2006) reviewed a cohort of 32 patients who were risk of PPA and underwent both U/S and MRI examinations were identified. U/S had sensitivity of 93% and specificity of 71% while MRI had a sensitivity of 80% and a specificity of 65%. Unlike the Riteau paper U/S is higher in specificity while they both roughly had similar sensitivity and specificity for MRI.
Maher et al. (2013) of 577 diagnosed with Placenta previa. 42 had placenta accreta, 49 of these were diagnosed prenatally. U/S accurately diagnosed placenta accreta in 33 of 39 of women and correctly ruled out placenta accreta in 512 of 514. Sensitivity was 95.1% and specificity 95.5%. MRI accurately PPA in six of 20 cases and correctly ruled out placenta accreta in 10 of 20 cases with sensitivity at 85.7% and specificity at 76.9%.
It is highly suggested in the study carried out by Lim et al. (2011) in three patients examined out of the thirteen accessed, MRI showed signs of PPA 4-7 weeks earlier than ultrasound did, however one of the three patients at 23 weeks had an ultrasound scan showing signs of lacunae. This patient had positive MRI findings of PPA at 24 weeks. In a report by Warshak et al. (2006) which presents the accuracy of both modalities measured on confidence intervals and believe that in general MRI performed better than U/S but suggest that with overlaps in confidence intervals the superiority of MRI with gadolinium over U/S is seen at statistically insignificant.
Specific U/S and MRI features
Ultrasound demonstrated some findings that most if not all reviewed literature agreed with finding such as presence of placental lacunae. Most of the literatures consider the presence of intraplacental lacunae to be the most indicative U/S feature in the diagnosis of PPA. This feature was a good indication for high sensitivity but meant PPV and specificity were relatively low. While in regards to MRI; findings of placenta accreta complicated by placenta previa presented the following signs, uterine bulging, heterogeneous signal intensity within the placenta, dark intraplacental bands on T2-weighted images, and focal interruptions in the myometrium wall, tenting of the bladder, direct visualization of the invasion of pelvic structures by placental tissue.
Ultrasound is a popular means of investigation when the diagnosis of PPA is needed. In many studies the author explore the use of U/S by delving into grey scale ultrasound, colour Doppler U/S or transvaginal and transabdominal techniques. Though U/S in more readily available, inexpensive, mobile and safe for patient to a certain degree. It is a modality just like MRI that depends heavily on the experience of the examiner, what they can see and their ability to differentiate PPA from other conditions. U/S faces limitation causes by posterior placenta and a large body habitus. MRI can be useful where U/S findings are limited and can be used in correlation with U/S. a study carried out by Mansour and Elkhyat (2011) seem to find the need for both modalities to work hand in hand whilst studies done by (Comstock and Bronsteen, 2013) think that MRI was not as useful as they hoped that it would be. A study carried out by Warshak et al. (2006) concluded that MRI used in addition to ultrasonography added little value to the diagnostic accuracy of PPA.
Bowman et al. (2014) agree with the featured U/S findings of PPA mention above they believed that a strong indication of the presence of accreta could be determined by placental lacunae. Hamisa et al., (2014) found that lacunae are seen in all cases of placenta accreta. This was statistically significant when predicting accreta. They carried out colour Doppler in order to be able to differentiate whether the cystic or vascular spaces are seen could be lakes. (Comstock and Bronsteen, 2013) made investigations using colour doppler and spectral ultrasound, and found that lacunae have high-velocity and low-resistance flow and are irregular in cross section on grey scale U/S, this view of lacunae being a strong U/S indication is shared by Berkley and Abuhamad. (2013). Lacunae is found deep in the placenta, irregularly shaped, and not rounded as is found with placental lakes. They also found that lacunae alone had a higher sensitivity of 93% and a PPV of 93%.
Riteau et al. (2014) study found the U/S features that presented a better sensitivity for the detection of placental invasion were intraplacental lacunae and loss of the normal retroplacental clear space. Lim et al. (2011) showed areas of high dark intraplacental bands on T2-weighted images were significantly different in the patients where PPA was not present. The study performed by Lim, performed MRI on at-risk women regardless if they had U/S scans showing indicative ultrasound findings of PPA, had pathological proof of PPA from caesarean hysterectomies and included Comstock showed that the MRI showed a moderate-size dark band as well most of the other papers that were reviewed.
Most authors agreed on myometrium thickness, most studies found that using U/S over the site of placental implementation where it was found that thickness was more than 1 mm was a strong indication of placenta accreta, this was statistically significant.
The accuracy of diagnosis appears to improve using more than one ultrasound technique combination. Using three grey scale signs such as retroplacental clear space, disruption of bladder line and intraplacental lacunae), some studies agree that a combination of these techniques can help in differentiating between accreta and percreta, as can be found in studies done by Hamisa et al. (2014) and Berkley and Abuhamad. (2013). Comstock et al. (2011) found that the grey scale signs mentioned above and found that when using two or more criteria the sensitivity was 80% and the PPV was 86% however, lacunae had a higher sensitivity and PPV value.
Warshak et al (2006) the oldest study compared U/S and MRI with gadolinium for diagnosis and evaluation of placenta accreta. Ultrasound was highly specific with 77% sensitivity while MRI with gadolinium had showed 100% specificity. The high statistical values in the study may be because they had performed routine both transvaginal ultrasound and transabdominal techniques when evaluating patients with accreta as well as using gadolinium in the MRI exams according the authors it had clearly defined the outer placental surface close to the myometrium clearly. Riteau et al. (2014) stated that they did not see an increased accuracy in MRI when using gadolinium or diffusion-weighted MR imaging.
In the Dwyer et al. (2008) study, they had lower specificity for both U/S and MRI than in the Warshak study. These differences could be due to ascertainment/referral bias e.g. in the patients they sampled and a retrospective study. No significant difference between the U/S and MRI could be shown in the Dwyer study. The difference in the specificity of U/S between studies could be that transvaginal U/S was used in Warshak study but was not routinely used in the Dwyer study. According to Webb and Thomsen, (2013) prior to 2006 most radiologists thought that gadolinium-based contrast media were safe for use in pregnant and lactating women till information of a link between nephrogenic systemic fibrosis (NSF) and exposure to gadolinium made it clear that they are absolutely contraindicated in pregnant women.
Only transabdominal sonography was used the study carried out by Lim et al. (2011), unlike who Berkley and Abuhamad. (2013) suggests that If the diagnosis of placenta accreta is needed, transvaginal sonography should be the go to imaging modality suggests and may for some patients the only modality needed for diagnosis. The Hamisa et al. (2014) suggests the combinations of both grey scale U/S and colour Doppler techniques for a higher prediction of PPA. Bowman et al. (2014) looks at ultrasound solely but came to a decision it may not be as sensitive a tool to detect PPA
In the study done by Lim accuracies of ultrasound and MRI were directly compared in the same group of patients, as in the study by Dwyer. A weakness was the small sample size, which was due to the fact that only patients who underwent both ultrasound and MRI studies were included. Bias was another inherent factor in their pilot cohort study Maher et al. (2013) is a study like Dwyer that didn’t use contrast but noted a limitation for not directly comparing the accuracy of U/S and MRI and the small amount of patients sampled.
Conclusion
The reviewed literature show some authors in agreement that ultrasound is the most commonly used diagnostic imaging modality and most sensitive for the diagnosis of PPA, due to the fact that it isn’t time consuming, non-invasive, accurate, inexpensive and more readily available than MRI. Grey-scale ultrasound is taught to be the tool of choice since it is the most widely available but some authors believe colour Doppler to be the best choice when sensitivity and PPV in ultrasound is concerned. Most authors also argue the best means of examination to be transvaginal while some other studies only looked at transabdominal scanning. Accessing the value of each feature of both modalities according to its PPV and NPV is important in accessing characteristics of PPA. MRI can help with signs of signs PPA not visible on ultrasound e.g. dark intraplacental bands, a distinct sign that can predict the presence of accreta. Most authors have come to the conclusion that wherever ultrasound falls short in diagnosing PPA, MRI should be as a tool for further investigation.
Acknowledgments
I would like to thank Dr. Martin Benwell for his contribution and critique and my family for all their love and support.
Abbreviations
NPV
negative predictive value
MRI
magnetic resonance imaging
PPV
positive predictive value
U/S
Ultrasound
PPA
Placenta Previa Accreta