Systematic review and meta-analysis on the role of prenatal magnetic resonance imaging in the era of fetal neurosonography: mild and moderate ventriculomegaly

Objectives : To report the rate of additional anomalies detected exclusively on prenatal MRI in fetuses affected by isolated mild or moderate ventriculomegaly (VM) according to the type of ultrasound protocol adopted (dedicated neurosonography vs standard assessment of fetal brain) and to explore whether the diagnostic performance of fetal MRI in detecting such anomalies is affected by gestational age at scan and laterality of ventricular dilatation. Methods : The primary aim was to report the rate of additional anomalies detected exclusively on prenatal MRI in fetuses affected by isolated mild and moderate VM (ventricular dilatation between 10-15 mm) undergoing compared to those not undergoing dedicated neurosonography, defined as a detail assessment of fetal brain according to the International Society of Ultrasound in Obstetrics and Gynecology guidelines. Medline, Embase, Cinahl and Clinicaltrials.gov databases were searched. Additional anomalies were classified in: callosal, septal, posterior fossa, white matter, intra-ventricular hemorrhage, cortical, peri-ventricular heterotopia, peri-ventricular cysts and complex malformations. Furthermore, we calculated the rate of additional CNS anomalies missed at prenatal MRI and detected only at birth in fetuses who had early (  24 weeks) compared to late (>24 weeks of gestation) MRI. A sub-analysis was performed according to the laterality (uni- vs bilateral VM) and the degree (mild vs moderate, defined as ventricular dilatation between 10-12 and 13-15 mm respectively) of ventricular dilatation. Finally, we explore whether MRI assessment led to a significant change in the prenatal management. Random-effect meta-analyses of proportions were used to analyze the data. Results: 16 studies (1159 fetuses) were included in the systematic review. MRI (95% CI 6.2-14.5; 92/1159) of fetal anomalies not detected on ultrasound. However, Both proportion meta-analyses and single-group meta-analyses of continuous data were performed using a random-effect model to account for inter-study heterogeneity, and all analyses were carried out using Stata, version 13.1 (Stata Corp., College Station, TX, 2013).


INTRODUCTION
Ventriculomegaly (VM) encompasses a spectrum of conditions characterized by a dilatation of the lateral ventricles of the brain, typically defined as greater than 10 mm at the level of the atria, with or without dilatation of the third or fourth ventricles [1][2][3][4] . VM is the most common brain anomaly diagnosed during fetal life, with a reported prevalence ranging from 1:50 to 1:1600 newborns. [1][2][3][4] VM is commonly categorized according to the degree of ventricular dilatation as mild (10-12 mm), moderate (13-15 mm), or severe (>15 mm); the rationale for such classification relies on the fact that the risk of associated anomalies and abnormal neurodevelopmental outcome is higher with increasing degree of dilation. 5 VM is not a unique condition, but rather a sonographic sign that represents a common endpoint of various pathologic processes with different outcomes and prognosis; it may also be a physiologic finding representing an extreme normal variation. [6][7][8] The main determinants of post-natal outcome in fetuses presenting with VM are the etiology, presence of associated anomalies and severity of ventricular dilation. Thus, the main issues to be resolved during the diagnostic work-up of a fetus with VM are to rule out infection, chromosomal, central nervous system (CNS) and extra-CNS anomalies. 7 Fetal magnetic resonance imaging (MRI) has been shown to add additional information compared to ultrasound in fetuses affected by CNS anomalies. 10 The overall prevalence of MRI-detected additional brain abnormalities prenatally has been reported to be as high as 19% in fetuses with VM, irrespective of the degree of ventricular dilatation. 11 However, the majority of these studies do not report the type of ultrasound assessment performed and the MRI is performed at varying gestational ages. The International Society of Ultrasound in Obstetrics and Gynecology recommends that fetuses presenting with CNS anomalies should undergo a multiplanar assessment of the brain through axial, coronal and sagittal views of the fetal head. 9 Therefore, it is plausible that the rate of associated anomalies detected exclusively on MRI reported in the published literature may be affected by the type of ultrasound assessment performed.
The primary aim of this systematic review was to report the rate of additional anomalies detected exclusively on prenatal MRI in fetuses affected by isolated VM according to the type of ultrasound protocol adopted. The secondary aim was to explore whether the diagnostic performance of fetal MRI in detecting such anomalies is affected by gestational age at scan.

Protocol, eligibility criteria, information sources and search
This review was performed according to a priori designed protocol recommended for systematic reviews and meta-analysis. [12][13][14] Medline, Embase, Cinahl and Clinicaltrials.gov databases were searched electronically in June 2018, utilizing combinations of the relevant medical subject heading (MeSH) terms, key words, and word variants for "magnetic resonance imaging" and "ventriculomegaly" (Supplementary Table 1). The search and selection criteria were restricted to English language. Reference lists of relevant articles and reviews were hand searched for additional reports. PRISMA and MOOSE guidelines were followed. [15][16][17] The study was registered with the PROSPERO database (Registration number: CRD42018104173).

Study selection, data collection and data items
The primary aim was to report the rate of additional anomalies detected exclusively on prenatal MRI in fetuses affected by isolated mild and moderate VM, defined as ventricular dilatation between 10-15 mm, undergoing compared to those not undergoing dedicated neurosonography, defined as a detail assessment of fetal brain through axial, sagittal and coronal views according to ISUOG guidelines.
Additional anomalies were classified into:  Callosal anomalies, including complete and partial agenesis (ACC), hypoplasia and dysgenesis of the corpus callosum  Septal anomalies, including all the anomalies characterized by a primary involving the septum pellucidum with a normally present corpus callosum  Posterior fossa anomalies, including all defects involving the cerebellar vermis and/or hemispheres  Intra-ventricular hemorrhage  Cortical anomalies, including all abnormalities associated with a primary defect in neuronal migration towards the cortical surface of the brain  Peri-ventricular heterotopia  Other white matter anomalies  Peri-ventricular cysts  Complex brain anomalies including all defects characterized by the presence of multiple intra-cranial anomalies  Other cerebral anomalies

Accepted Article
For the purpose of the analysis, we did not consider as associated anomalies biometric variation in brain structures, such as mega cisterna magna, increased or reduced degree of ventricular dilation or of cranial size. Furthermore, we aimed to perform a sub-group analysis according to the laterality (unilateral-vs bilateral VM) and degree (10-12 mm vs 13-15 mm) of ventricular dilatation.
The secondary aim was to elucidate whether the diagnostic performance of fetal MRI for detected additional CNS anomalies was affected by gestational age at scan. For the purpose of the analysis, we calculated the rate of additional CNS anomalies missed at prenatal MRI and detected only at birth in fetuses who had early (24 weeks) compared to late (>24 weeks of gestation) MRI. Finally, we explored whether MRI detection of associated anomalies led to a change in the prenatal management of the pregnancy, mainly defined as termination of pregnancy (TOP) for parental request due to the higher risk of abnormal neurodevelopmental outcome followed by the MRI detection of associated anomalies.
Only studies reporting the prevalence of brain anomalies diagnosed on MRI in fetuses affected by isolated VM and confirmed at post-natal imaging or post-mortem examination, in case of termination of pregnancy or fetal demise, were considered eligible for the inclusion in the present systematic review. VM was defined as isolated when no other CNS and extra-CNS anomalies, were detected on ultrasound.
Studies including cases with fetal anomalies, those including exclusively cases of severe VM (defined as ventricular dilation >15 mm) and those not reporting the degree of ventricular dilation were excluded in view of the higher risk of associated brain anomalies in cases presenting with additional anomalies and/or severe VM. Case reports, conference abstracts and case series with fewer than 3 cases were excluded to avoid publication bias. Furthermore, studies published before ISUOG statement for a detailed assessment of fetal brain was released (2007) were also excluded as advances in prenatal imaging and difficulties in extrapolating the type of ultrasound protocol make them less relevant.
Two authors (DDM, FGS) reviewed all abstracts independently. Agreement regarding potential relevance was reached by consensus or resolved by discussion with a third reviewer (FDA). Full text copies of those papers were obtained, and the same reviewers independently extracted relevant data regarding study characteristics and pregnancy outcome. Inconsistencies were discussed by the reviewers and consensus reached or by discussion with a third author. If more than one study was published on the same cohort with identical endpoints, the report containing the most Accepted Article comprehensive information on the population was included to avoid overlapping populations. For those articles in which information was not reported but the methodology was such that this information would have been recorded initially, the authors were contacted.
Quality assessment of the included studies was performed using the Newcastle-Ottawa Scale (NOS) for cohort studies. According to NOS, each study is judged on three broad perspectives: the selection of the study groups, the comparability of the groups, and the ascertainment outcome of interest. 18 Assessment of the selection of a study includes the evaluation of the representativeness of the exposed cohort, selection of the non-exposed cohort, ascertainment of exposure and the demonstration that outcome of interest was not present at start of study. Assessment of the comparability of the study includes the evaluation of the comparability of cohorts based on the design or analysis. Finally, the ascertainment of the outcome of interest includes the evaluation of the type of the assessment of the outcome of interest, length and adequacy of follow-up. According to NOS a study can be awarded a maximum of one star for each numbered item within the Selection and Outcome categories. A maximum of two stars can be given for Comparability. 18

Statistical analysis
First, we performed random-effect meta-analyses of proportions to estimate the pooled rates of each brain anomaly in fetuses affected by isolated VM undergoing MRI assessment. Second, we used random-effect head-to-head meta-analyses to directly compare the risk for detecting an associated anomaly in fetuses undergoing early (24 weeks) compared to late (>24 weeks) fetal MRI, expressing the results as summary odds ratio (OR) and relative 95% Confidence Interval (CI), and evaluating the statistical heterogeneity through I 2 metric. Data from individual studies were also combined to estimate the weighted mean gestational age at diagnosis (of brain anomalies) among fetuses with VM detected through multiplanar approach. Only sparse data on gestational age were available for the standard US group, as well as on the time interval between US and MRI exams for both groups. In these cases, no meta-analyses were performed, and the results were reported only narratively for descriptive purposes. All included studies reported single-group analyses, and no outcome comparison between groups was available, thus no head-to-head meta-analysis could be performed.
Publication bias was assessed graphically, through funnel plots, and formally, through Egger's regression asymmetry test; formal tests for funnel plot asymmetry were not performed when the total number of publications included for each outcome is <10 because the power of the test is too low to distinguish chance from real asymmetry.

Accepted Article
Both proportion meta-analyses and single-group meta-analyses of continuous data were performed using a random-effect model to account for inter-study heterogeneity, and all analyses were carried out using Stata, version 13.1 (Stata Corp., College Station, TX, 2013).

General characteristics
Six-hundred and twenty articles were identified, 66 were assessed with respect to their eligibility for inclusion (Supplementary Table 2) and 16 studies were included in the systematic review (Table 1, Associated anomalies confirmed at birth or autopsy were found in 10.56% (95% CI 6.7-15.2, I 2 : 78.7% 100/1159) of cases with a prenatal diagnosis of isolated VM on ultrasound.
The results of the quality assessment of the included studies using Newcastle-Ottawa Scale (NOS) are presented in Table 2. Most of the included studies showed an overall good score regarding the selection and comparability of the study groups, and for ascertainment of the outcome of interest.
The main weaknesses of these studies were their retrospective design, small sample size, different gestational ages at scan and lack of stratification of the results according to laterality and degree of ventricular dilatation for most of the included studies. Furthermore, a comprehensive stratification of the analysis according to the gestational age at scan was possible only by adopting 24 weeks of gestation as cut-off, while we could not perform any meaningful sub-analysis considering different time intervals between US and MRI assessments.
When stratifying the analysis according to the type of anomaly, the incidence of undetected callosal anomalies on ultrasound was 0.7% (95% CI 0.2-1.5) in the group fetuses undergoing neurosonography and 2.9% (95% CI 1.1-5.7) for those having standard assessment of fetal brain,
Hemorrhagic lesions were missed at ultrasound and detected exclusively on MRI in 1.1% (95% CI 0.3-2.4) of cases undergoing detailed neurosonography and 2.0% (95% CI 0.6-4.1) of those receiving standard assessment (Table 4). Finally, white matter anomalies were missed at ultrasound and detected on MRI in 1.4% (95% CI 0.6-2.5) of cases with an ultrasound diagnosis of isolated VM. When stratifying the analysis according to the type of ultrasound protocol adopted, the detection rate of fetal MRI in identifying white matter anomalies missed at the scan was 1.6% (95% CI 0.7-2.7) of cases undergoing neurosonography and 1.8% (95% CI 0.3-4.4) of cases receiving a standard ultrasound assessment.
Sub-analyses according to laterality and degree of ventricular dilatation were carried out including exclusively fetuses undergoing multiplanar assessment and are shown in Supplementary Table 4 and 5.
Finally, we explore the rate of the change in perinatal management after prenatal MRI. Overall 4.6% (95% CI 2.1-8.0; I 2 : 61%) of fetuses with a prenatal diagnosis of isolated VM on ultrasound had a significant change in perinatal management, mainly TOP for parental request, following MRI detection of associated anomalies. When stratifying the analysis according to the type of ultrasound assessment performed, a significant change in perinatal management involved 2.9% (95% CI 0.01-

Accepted Article
This article is protected by copyright. All rights reserved.

Main findings
The findings from this systematic review show that, in fetuses affected by VM undergoing dedicated neurosonography, the rate of CNS anomalies detected exclusively on MRI is lower compared to what has been previously reported. Fetal MRI has a high diagnostic accuracy in identifying associated anomalies regardless of gestational age at MRI. Finally, the risk of detecting associated anomalies is higher in fetuses with moderate compared to mild while there was no difference between cases with unilateral compared to bilateral VM.

Limitations
Retrospective design, small sample size, different gestational ages at scan and lack of stratification of the results according to laterality and degree of ventricular dilatation for most of the included studies represents the main limitations of the represent systematic review. Furthermore, not all cases presenting additional anomalies on MRI were screened for aneuploidies or infection. In this scenario, it might be entirely possible that some cases affected by aneuploidies of infection diagnosed only later on after birth were included in the analysis thus affecting the robustness of the results. Finally, it was not possible to completely rule-out inclusion of cases complicated by maternal medical conditions potentially leading to an increased risk of associated CNS in the setting of VM, which may explain the relatively high rate of acquired brain anomalies such as hemorrhage reported in this review.

Implications for clinical practice and research
Ruling out associated anomalies in fetuses with a prenatal diagnosis of VM is fundamental as neurodevelopmental outcome of these children is largely affected by the presence of associated malformations.
The International Society of Ultrasound in Obstetrics and Gynecology recommends that fetuses presenting with CNS anomalies should undergo a multiplanar assessment of fetal head, in order to rule associated anomalies which can be potentially missed on a standard assessment. Likewise, the Society for Maternal-Fetal Medicine suggests that MRI should be considered in cases of mild or moderate fetal VM although it may be of less value if the women has had a detailed ultrasound performed by an individual with specific experience and expertise in sonographic imaging of the fetal brain. These statements from the two major bodies society of maternal-fetal medicine highlight the need for a detailed assessment of fetal brain when VM is detected on the scan. Despite this, it is common clinical practice to refer fetuses presenting with VM to MRI without a detailed assessment Accepted Article of the brain. This has led to a high reported incidence of associated anomalies detected only on MRI in fetuses presenting with isolated VM on ultrasound.
In the present review, the large majority of associated malformations undetected on ultrasound in cases undergoing standard assessment were callosal anomalies. This underlines the need for detailed neurosonography in fetuses presenting with ventricular dilatation in axial views in order to reduce the number of cases incorrectly labelled as isolated VM. This is also fundamental as fetal MRI may not be immediately available and a late diagnosis of associated anomalies may lead to increased parental stress and lack of legal options for the prenatal management of the pregnancy. This is also highlighted by the different rates in the change of prenatal management after MRI in fetuses undergoing compared to those not undergoing neurosonography.
Fetal MRI has been suggested to have a high diagnostic accuracy in detecting cortical malformations, such as polymicrogyria and schizencephaly. 10,35 The true prevalence of congenital cortical anomalies in the general population is largely unknown and some of these cases are suspected on fetal ultrasound during the third trimester of pregnancy based on the subjective findings of immature sulcation, thin and smooth cortex or wide and thick gyri. In the present systematic review, the rate of cortical anomalies detected exclusively on MRI was about 1.6%, highlighting the need for a detailed MRI assessment of fetal brain in order to rule out such anomalies.
The severity of ventricular dilatation is associated with an increasing risk of adverse neurodevelopmental outcome in fetuses presenting with isolated VM. In the present systematic review, the rate of associated anomalies detected only on MRI was 3.5% and 22.6% with mild and moderate VM, suggesting that increasing ventricular dilatation is associated with a higher risk of associated anomalies. Conversely, there was no difference in the rate of associated anomalies detected only on MRI in fetuses with unilateral compared to bilateral VM, although the very small number of cases included in this sub-analysis may have led to a lack of statistical power.
Gestational age at MRI is another important clinical information when assessing fetuses with an ultrasound diagnosis of isolated VM. Early (second trimester) MRI assessment is useful to confirm the diagnosis and rule out associated anomalies especially in those countries where termination of pregnancy beyond the second trimester is not allowed but may potentially overlook anomalies such as cortical malformations or hemorrhage which can become evident only later on in gestation. In the present systematic review, the overall rate of additional anomalies detected only at birth and Accepted Article This article is protected by copyright. All rights reserved. missed at prenatal MRI was 1.3% (95% CI 0.5-2.4) when MRI was performed before and 0.8% (95% CI 0.3-1.7) when carried out after 24 weeks of gestation. Although there was no statistical difference in the rate of undetected CNS anomalies between early and late MRI, it would be reasonable to perform an MRI scan in the third trimester of pregnancy as most of the anomalies coexisting with VM, such as cortical, white matter or hemorrhage can become more evident after 24 weeks" gestation. However, parents should be reassured that, in cases with VM apparently isolated on fetal MRI, the probability of undetected anomalies at birth is low.

Conclusion
Fetuses affected by isolated VM on ultrasound screening examination should be referred to dedicated neurosonography in order to rule out associated anomalies which can potentially affect the short and long-term neurodevelopmental outcome of these children. Fetal MRI assessment of these fetuses is recommended to detect associated anomalies that can be missed on neurosonography in about 5% of the cases. Although early MRI is reliable, a third trimester scan is technically more feasible and may detect conditions potentially missed at second trimester scan, such as hemorrhage. Future, large, prospective studies sharing objective protocols for ultrasound imagining of fetal brain are needed to better elucidate the actual role of fetal MRI when fetal neurosonography is performed.

Accepted Article
This article is protected by copyright. All rights reserved.   Accepted Article Accepted Article Table 2. Quality assessment of the included studies according to Newcastle-Ottawa Scale (NOS) for cohort studies; a study can be awarded a maximum of one star for each numbered item within the Selection and Outcome categories. A maximum of two stars can be given for Comparability. Accepted Article