a systematic review and meta-analysis Diagnostic accuracy of ultrasound in detecting the severity of abnormally invasive placentation: a systematic review and meta-analysis GIORGIO PAGANI1, GIUSEPPE CALI2 , GANESH ACHARYA3,4 , ILAN-TIMOR TRISCH5, JOSE PALACIOS-JARAQUEMADA6 , ALESSANDRA FAMILIARI7, DANILO BUCA8 , LAMBERTO MANZOLI9, MARIA E. FLACCO10 , FRANCESCO FANFANI8, MARCO LIBERATI8, GIOVANNI SCAMBIA7 & FRANCESCO D’ANTONIO4,11 1Department of Obstetrics and Gynecology, Fondazione Poliambulanza, Brescia, 2Department of Obstetrics and Gynecology, Arnas Civico Hospital, Palermo, Italy, 3Department of Clinical Science, Intervention and Technology, Karolinska Institute, Stockholm, Sweden, 4Women 0 s Health and Perinatology Research Group, Department of Clinical Medicine, Faculty of Health Sciences, UiT-The Arctic University of Norway, Tromsø, Norway, 5Department of Obstetrics and Gynecology, Division of Maternal-Fetal Medicine, New York University SOM, New York, NY, USA, 6Center for Medical Education and Clinical Research (CEMIC), University Hospital, Buenos Aires, Argentina, 7Department of Obstetrics and Gynecology, Catholic University of The Sacred Heart, Rome, 8Department of Obstetrics and Gynecology, University of Chieti, Chieti, 9Department of Medical Sciences, University of Ferrara, Ferrara, 10Local Health Unit of Pescara, Pescara, Italy, and 11 Department of Obstetrics and Gynecology, University Hospital of Northern Norway, Tromsø, Norway ACTA Obstetricia et Gynecologica Scandinavica Journal Club January 2018 Edited by Francesco D’Antonio

Introduction Accurate prenatal diagnosis of AIP is fundamental because it has been shown to reduce the burden of maternal and fetal morbidity associated with this condition, by allowing implementation of pre-planned management strategies. Prenatal diagnosis of AIP is commonly accomplished by ultrasound during the second and third trimesters of pregnancy and has been shown to have an overall good diagnostic accuracy in women at risk, such as those with placenta previa and previous cesarean section (CS), especially when a combination of maternal characteristics and imaging signs are integrated into an individualized diagnostic algorithm. Intra- and post-surgical outcomes of women affected by AIP are directly related to the depth and topography of placental invasion with cases affected by placenta percreta and/or showing parametrial invasion being at the highest risk of morbidity. Despite this, the actual performance of ultrasound in detecting the severity of placental uterine invasion remains elusive.

Aim of the study To elucidate the overall diagnostic accuracy of prenatal ultrasound in detecting the severity of placental invasion in women at risk And To explore the strength of association and the predictive accuracy of each ultrasound sign suggestive of AIP in identifying the type of placental invasion.

Material and Methods Study design: Systematic review and meta-analysis Literature search: Medline, Embase, CINAHL, and The Cochrane databases were searched. Outcomes’ measures (1): Diagnostic accuracy of ultrasound in recognizing the severity of AIP, defined by the depth and topography of placental invasion. For the depth of placental invasion, the reference standard was histopathological examination of the removed uterus. For the assessment of the topography of placental invasion, the anatomical classification of AIP proposed by Palacios-Jaraquemada et al. was adopted. According to such classification, anterior placental invasion is divided into two sectors delimited by a plane perpendicular to the upper bladder axis, and the uterine sector bordering; the upper posterior bladder wall is called S1, and the uterine sector adjacent to the lower posterior wall is called S2 (Figure). From an anatomical perspective, S1 invasion refers to an invasion situated in the uterine body while S2 described invasion that is mainly located in the lower uterine segment or below it. Reference standard was the topography of invasion observed at surgery

Material and Methods Outcomes’ measures (2): The strength of association between each ultrasound sign of AIP and the depth of placental uterine invasion and their individual predictive accuracy in detecting such invasion. Ultrasound signs explored: Ultrasound signs were differentiated into those identified on gray scale and those on color Doppler ultrasound. Gray-scale ultrasound signs of AIP were: Loss of clear zone, defined as a loss, or irregularity, of hypoechoic plane in myometrium underneath placental bed

Material and Methods Placental lacunae, defined as the presence of numerous lacunae, often containing turbulent flow visible on gray-scale imaging Bladder Bladder wall interruption, defined as loss or interruption of bright bladder wall (hyperechoic band or ‘line’ between uterine serosa and bladder lumen)

Material and Methods Myometrium Myometrial thinning, defined as thinning of myometrium overlying placenta to <1 mm or undetectable Bulging Bladder Focal exophytic mass, defined as placental tissue seen breaking through uterine serosa and extending beyond it; most often seen inside filled urinary bladder.

Material and Methods Color Doppler signs of AIP were: Placental lacunar flow, defined as the presence of color Doppler signal within placental lacunae. Subplacental vascularity, defined as striking amount of color Doppler signal seen behind the placental bed.

Material and Methods Uterovesical hypervascularity, defined as striking amount of color Doppler signal seen between myometrium and posterior wall of bladder, including vessels appearing to extend from placenta, across myometrium and beyond serosa into bladder or other organs; often running perpendicular to myometrium.

Material and Methods Statistical analysis: Summary estimates of sensitivity, specificity, positive and negative likelihood ratios (LR+ and LRÀ), and diagnostic odds ratio (DOR) were computed using the hierarchical summary receiver operating characteristics (HSROC) model. Rutter and Gatsonis HSROC parameterization was used because it models functions of sensitivity and specificity to define a summary ROC curve, and its hierarchical modeling strategy can be used for comparisons of test accuracy when there is variability in threshold between studies. However, when the number of studies is small, the uncertainty associated with the estimation of the shape parameter could be very high, and models may fail to converge. Hence, for all meta-analyses in which less than four study estimates could be pooled, the DerSimonian–Laird random-effect model was used.

Results 20 studies (3209 pregnancies at risk for AIP, mainly because of the presence of placenta previa and previous CS or uterine surgery) were included in the systematic review. 407 (12.7%, 95% CI 11.6–13.9) women had AIP. The occurrences of placenta accreta, increta and percreta were 37.8% (95% CI 33.1–42.7), 32.2% (95% CI 27.7–37.0), and 30.0% (95% CI25.6–34.7), respectively.

Results Ultrasound had an overall good diagnostic accuracy in identifying the depth of placental invasion with sensitivities of 90.6% (95% CI 80.7–96.5), 93.0% (95% CI 80.9–98.5), 89.5% (95% CI 73.2–96.3), and81.2% (95% CI 51.8–94.6) for placenta accreta, increta, accreta/increta, and percreta, respectively; the corresponding figures for specificity were 97.1% (95% CI 95.4–98.3),98.4 (95% CI 97.0–99.2), 94.7 (95% CI 91.0–96.9), and 98.9 (95% CI 95.0–100). Only two studies (30,39) explored the role of ultrasound in identifying the topography of the invasion. Overall ultrasound correctly identified 93.4% (95% CI 64.7–100) of women with S1 and 90.3% (95% CI 80.7–97.4) of those with S2 invasion confirmed at surgery.

Results The presence of lacunae was independently associated with placenta accreta, increta, and percreta with OR of 7.8, 16.1, and 8.2 respectively. When translating these findings into figures of diagnostic accuracy, placental lacunae had sensitivities of 74.8% (95% CI55.4–87.6), 88.6% (95% CI 55.3–98.0), and 76.3% (95% CI 42.2–93.4) for the detection of placenta accreta, increta, and percreta respectively, whereas the corresponding figures for specificity were 87.9% (95% CI 52.6–97.9),77.4% (95% CI 46.8–93.0), and 74.0% (95% CI 45.0–90.9). Loss of the clear zone was associated with a higher risk of placenta accreta, increta, and percreta, with OR of 23.8, 20.8, and13.0 respectively. Sensitivity and specificity of loss of the clear zone in identifying placenta accreta were 74.9% (95% CI 33.5–94.6) and 92.0% (95% CI 68.8–98.3), whereas the corresponding figures for placenta increta and percreta were 91.6% (95% CI 59.9–98.8) and 76.9% (95% CI 45.4–93.0), and 88.1% (95% CI 64.7–96.8) and71.1% (95% CI 42.2–89.2)

Results The presence of an abnormal myometrial thickness, defined as <1 mm, was associated with a higher risk of every type of AIP. Sensitivity and specificity were 100% (95% CI 31.0–100) and 85.0% (95% CI 72.9–92.5) for placenta accreta, 100% (95% CI 47.8–100) and 74.3% (95% CI 62.4–84.0) for placenta increta and 85.7% (95% CI 57.2–98.2) and 76.0% (95% CI 66.4–84.0) for placenta percreta. Focal exophitic mass had sensitivity of 16.7% (95% CI 0.42–64.2), a specificity of 100% (95% CI 88.6–100), and a DOR of 21.0 (95% CI 0.76–583) in detecting placenta percreta. The presence of any abnormality at the bladder–uterine interface, such as loss or interruption of bright bladder wall, was significantly associated with placenta accreta, increta, and percreta, with OR of5.3, 27.6, and 77.6, respectively. Abnormalities of the bladder wall had a sensitivity, a specificity, and a DOR of 17.0% (95% CI 0.06–85.8), 96.8% (95% CI 86.0–99.3), and 6.13 (0.13–294.1) in identifying cases with placenta accreta.

Results The presence of lacunar flow in women at risk of AIP was significantly associated with the occurrence of placenta accreta (OR 21.6), increta (OR 20.4), and percreta (OR 2.51). Lacunar flow had sensitivities of 81.2% (95% CI 57.2–93.3), 84.3% (95% CI50.8–96.5), and 45.2% (95% CI 27.3–64.0) for the detection of placenta accreta, increta, and percreta respectively; the corresponding figures for specificity were 84.0% (95% CI 65.4–93.6), 79.7% (95% CI 57.4–91.9), and 75.3% (95% CI 69.8–80.2). Subplacental hypervascularity had a low sensitivity in detecting placenta accreta (40.7%, 95% CI 22.4–61.2), increta (17.4%, 95% CI5.0–38.8), and percreta (40.0%, 95% CI 12.2–73.8), while specificity was 95.5% (95% CI 91.3–98.0), 93.8% (95% CI 88.8–97.0), and 92.5% (95% CI 85.1–96.9), respectively. Uterovesical hypervascularity was significantly associated with placenta accreta (OR 4.6), increta (OR32.4) and percreta (OR 48.9). Sensitivity was low for the detection of placenta accreta (12.3%, 95% CI 2.59–100) but high for placenta increta (94.4%, 95% CI 29.2–100) and percreta (86.2%, 95% CI 60.0–96.3); the corresponding figures for specificity were 90.8% (95% CI 75.2–97.0),88.0% (95% CI 72.8–95.3), and 88.2% (95% CI 71.9–95.6), respectively.

Limitations Small number of included studies. Retrospective design. Heterogeneity in ultrasound signs explored. Heterogeneity in gestational ages at assessment. The large majority of studies exploring the predictive accuracy of ultrasound in detecting AIP did not report the diagnostic performance of ultrasound in detecting the topography of placental invasion according to the classification system provided by Palacios-Jaraquemada et al.

Conclusion Ultrasound has an overall good diagnostic accuracy in recognizing the depth and the topography of placental invasion.