Prenatal diagnosis of congenital head, face, and neck malformations—Is complementary fetal MRI of value?

The aim of this study was to evaluate the role of fetal magnetic resonance imaging (MRI) as a complement to ultrasound (US) in the prenatal diagnosis of craniofacial anomalies.

habitus, complex fetal anomalies, fetal position and overlying limbs, advanced gestational age, or the skills of the performing operator. 3 The evaluation of the head, face, and neck can be limited by interposition of the tongue and the acoustic shadowing caused by ossification of facial structures. 5 In these cases, alternative imaging modalities may provide additional information that can improve diagnostic accuracy and facilitate treatment decisions.
MRI has progressively emerged as a valuable adjunct to US in the work-up of fetal craniofacial pathology because of the development of ultrafast sequences that have few motion artefacts and offer excellent spatial resolution. [5][6][7] Fetal craniofacial anomalies described on MRI include facial clefts, retrognathia, micrognathia, craniosynostosis, cephaloceles, vascular anomalies, tumors, dacryocystoceles, ocular, and orbital abnormalities. 5,8 Although the role of MRI as a complement to US in the diagnosis of cleft lip and palate has been established, 4,7,9 the data regarding its efficacy in evaluation of other craniofacial anomalies is limited.
The aim of the present study was to evaluate the role of fetal MRI as a complement to US in prenatal diagnosis of head, face, and neck anomalies, through comparing antenatal sonography and MRI findings with physical examination and postnatal imaging.

| Study population
This historical cohort included all pregnant women who were referred for fetal MRI because of suspected head, face, or neck anomalies diagnosed by US between January 2011 and July 2016 in a single university-affiliated tertiary referral hospital. Demographic and clinical data were collected from the electronic records of the mother and neonate. Any missing data were gathered through direct phone interviews with the women.
In order to explore the validity of MRI in the diagnosis of craniofacial anomalies through calculation of sensitivity, specificity, and diagnostic accuracy, an additional cohort of women who were referred for fetal MRI for indications other than suspected head, face, or neck malformations was established. The MR studies of this cohort were used as a data source, and the images were not reevaluated for the current study. The study was approved by the Institutional Review Board of our institution (no. 1725-14-SMC).

| Imaging technique and interpretation
All patients underwent expert diagnostic antenatal ultrasound, fetal MRI, and were advised to have an amniocentesis for karyotype analysis.

| Magnetic resonance imaging
MRI is routinely performed in our institution once a congenital craniofacial anomaly is suspected on US. All patients underwent hence an MRI study of the fetal face and central nervous system. Although fetal MRI of the head, face, and neck can be performed during the second or third trimester, the preferred timing for fetal MRI in our institute is the 32nd week of gestation because at that time one can also assess brain maturation, parenchyma, and make a comprehensive brain structural assessment. 11 In cases of later craniofacial anomalies diagnosis, MRI was conducted within a few days of diagnosis. All patients routinely signed an informed consent form prior to MRI performance.
Fetal MRI was performed using a 1.5-Tesla system (Optima scanner, GE Healthcare Technologies, Milwaukee, Wisconsin) as previously described. 12,13 The anatomic area was mapped, and sequence selection and planes of acquisition were chosen according to sonographic findings and clinical context, as defined by a neuro-radiologist MRI expert (C.H.). Images were acquired in three orthogonal planes with respect to the fetal face. The sagittal view provides a good evaluation of the fetal profile, including the frontal and nasal bones, hard palate, tongue, and mandible. A coronal view is useful in assessing the integrity of the fetal lips and palate, as well as providing delineation of the What is already known about this topic?
• Fetal head, face, and neck anomalies represent a challenging prenatal diagnosis, mainly because of the wide range of morphological features involved.
• Ultrasound (US) is the primary imaging modality for the evaluation of the fetus, including craniofacial malformations.
What does this study add?
• Magnetic resonance imaging (MRI) provides valuable information in the antenatal evaluation of fetal craniofacial anomalies, and may be a useful adjunct to US in the prenatal work-up of craniofacial anomalies. eyes, nose, and ears. Axial view helps assess the different compartments of the neck and is also valuable in evaluating the fetal facial structures and variations in cranial morphology. In addition to morphological analysis, detailed brain biometry was performed using standard reference data. 14 All MRI scans were assessed by a joint MRI and US team, which included a specialist in fetal US (E.K.) and a neuroradiologist MRI expert (C.H.).

| Outcome measurements
US and MR images interpretations were compared for consistencies and discrepancies. US and MR imaging were categorized as correct or incorrect with respect to the final postnatal diagnosis, which was considered the reference standard. Postnatal diagnosis was obtained through physical examination, plain radiograph, US, computed tomography, MRI, surgery, or fetal autopsy in cases of termination of pregnancy. In addition, we noted cases in which MRI or US provided more information.

| Statistical analysis
The normality of distribution of continuous variables was tested by histogram and Q-Q plots. Continuous variables with normal distribution are presented as mean and standard deviation (SD); non-normal distributed continuous variables are presented as median and interquartile range (IQR). Categorical variables are reported as number and percentage. Continuous variables normally distributed were compared using Student t test, and Mann-Whitney test was used to compare continuous variables not normally distributed. Categorical variables were compared using Chi-square test or Fisher exact test, as appropriate. Agreement between US and MRI findings was measured using kappa-statistics (κ). According to Landis and Koch, 15 the kappa values were interpreted as follows: 0.81 to 1.00 = almost perfect agreement, 0.61 to 0.80 = substantial agreement, 0.41 to 0.60 = moderate agreement, 0.21 to 0.40 = fair agreement, 0 to 0.20 = poor agreement, and 0 = no agreement. The sensitivity, specificity, and diagnostic accuracy of MRI and US were evaluated with postnatal diagnosis as a standard of reference. A two-tailed P < .05 was considered statistically significant. Analyses were performed with SPSS version 24.0 for windows (SPSS Inc., IBM Corp., Armonk, New York).

| Demographic and clinical characteristics of the study populations
Sixty-four pregnant women with fetuses suspected to have craniofacial anomalies based on screening US were referred for fetal MRI in our institution over a 6-year period (2011-2016). Following preliminary record review and complementary phone interviews, 19 women were excluded: five women declined to participate in the study, three women performed termination of pregnancy without autopsy, and 11 patients were found to have incomplete records. Thus, the study group comprised a total of 45 pregnant women who underwent fetal MRI scans: 44 singleton pregnant women and one dichorionic diamniotic twin pregnancy with a single affected fetus. The control cohort comprised 61 pregnant women who were referred for fetal MRI for other indications. Final diagnosis was obtained from all fetuses at postnatal physical examination (n = 99) or at autopsy (n = 7).
Mean gestational age at time of fetal MRI was 32.86 (range 31.43-35), which was significantly higher than that of the study group (P = .03). Table 1

| Cleft palate
Fourteen cases with cleft palate suspected on US were referred to fetal MRI. MRI confirmed the diagnoses in eight cases, and the diagnoses established by both imaging techniques were correct postnatally.
Discrepancies occurred in 10 cases; in these cases, the diagnosis on US was rejected by MRI, and the MRI interpretation was correct. MRI raised suspicion for cleft palate in four additional cases, and diagnosis was confirmed by pediatric examination in only three cases (Figure 1).

| Cysts/masses/tumors of the face or neck
There were six cases with cervicofacial cysts or tumors. MRI confirmed the diagnosis in four, and these diagnoses were correct when compared with postnatal diagnosis: pharyngeal teratoma, namely, epignathus ( Figure 2); lymphatic malformation in a cheek; macrocystic lymphatic malformation in the posterior triangle of the neck; and an astrocytoma behind one eye. In two cases, diagnosis at US was rejected by MRI, the latter interpretation being correct after birth.

| Eye structure anomalies
Six cases with structural eye anomalies on US were referred for MRI: microphthalmia, anophthalmia, irregular structure of the eyeball, retinal detachment, and opaque lenses. MRI confirmed the diagnosis in

| Prenasal and frontal edema
Five cases with prenasal and frontal edema suspected on US were referred to fetal MRI. MRI confirmed that in two cases; however, only in one this was on both imaging methods as well as in the postnatal period. In the other case, the prenatal US and MRI diagnosis was incorrect on postnatal examination. In another case, the initial diagnosis on US was rejected by MRI, yet this was incorrect as the neonate had frontal edema. In two cases, the diagnosis at US was rejected by MRI, the latter being correct.

| Hypotelorism/hypertelorism
Of the five cases with hypo/hypertelorism on US, MRI confirmed the diagnoses in only one, and a diagnosis of semilobar holoprosencephaly was established. This was the correct postnatal diagnosis. In another case, the US diagnosis was rejected by MRI, yet this was incorrect: following termination because of other malformations and severe intrauterine growth restriction, hypertelorism was confirmed. In three cases, the US diagnosis was rightfully rejected by MRI.

| Ear structure anomalies
There were three cases with ear anomalies. MRI confirmed the diagnoses in two (microtia) and this was correct postnatally. In one case, MRI rejected the diagnosis at US and this was correct. Nonetheless,  three neonates in the study group were diagnosed with low set ears and one neonate was diagnosed with auditory canal stenosis; none of the diagnoses were suspected on US or MRI.

| Forehead structure anomalies
Three cases with either frontal bossing or flattened forehead suspected on US were referred for fetal MRI. In all cases, MRI declined this diagnosis, an interpretation that was correct.

| Cysts of the skull
Three cases with skull lesions/cysts suspected on US were referred to fetal MRI. MRI confirmed the diagnoses in two. In only one, this diagnosis established by both methods was correct. In the third case, diagnosis on US was rejected by MRI, and the MRI interpretation was correct.

| Nuchal edema
In the three cases with nuchal edema, MRI confirmed the diagnoses in all three. Nonetheless, in only two, this was correct after birth. One neonate in the study group was born with nuchal edema which was not detected in both US and MRI.

| Tracheoesophageal malformations
One case with suspected tracheoesophageal fistula and one case with suspected cervical mass obstructing the esophagus and trachea on US were referred to fetal MRI. In both, MRI declined the diagnosis, an interpretation which was correct.

| Craniosynostosis
Two cases with craniosynostosison US were referred for fetal MRI. In one case MRI confirmed the diagnosis wrongfully. In one case MRI confirmed the diagnosis wrongfully. In the other case, MRI declined the diagnosis, which was correct. There were no neonates diagnosed with craniosynostosis.

| Meningoencephalocele
There was one case with meningoencephalocele, which was confirmed by MRI, and this diagnosis was correct after birth.
Considering all cases in this study, there was complete agreement Using postnatal diagnosis as the ground truth,

| DISCUSSION
Anomalies of the fetal head, face, and neck include a wide range of malformations, which produce various degrees of disfigurement and can result in severe functional impairment postnatally. 1 It is essential to describe the complexity of the malformation and associated anomalies because of their neonatal implications. Thanks to improvement in prenatal diagnosis through US and MRI, the approach to fetuses with craniofacial anomalies has shifted from a diagnosis at birth to prenatal diagnosis. 2 Prenatal diagnosis facilitates counselling by an oral and maxillofacial surgeon, and careful planning with a multidisciplinary team of the timing and mode of delivery, level of neonatal care required, and procedures that need to be considered. [16][17][18] This study includes one of the largest published cohorts of craniofacial anomalies evaluated by fetal MRI. Our main objective was to assess the role of MRI as a complement to US in the diagnosis of craniofacial anomalies. As reflected by confident diagnosis and misdiagnosis rates obtained from our cohort, MRI was found valuable for prenatal diagnosis of craniofacial anomalies, and also provided additional information when compared with US. When stratified by the type of anomaly, in the majority of malformations, US was in substantial to almost perfect agreement with MRI. They were in moderate agreement regarding diagnosis of cleft palate and prenasal and frontal edema, and in fair to no agreement in diagnosis of hypotelorism or hypertelorism, forehead structure anomalies, and tracheoesophageal malformations.
Although US remains the method of choice for fetal screening, 19 the utility of MRI for fetal evaluation has considerably increased in recent years for several reasons. First, MRI popularity has risen because of its safety and technological improvements. 20 These include ultrafast sequences that significantly reduce motion artifacts, multiplanar capability, excellent soft tissue contrast, and a larger field of view. 6,19,21,22 Second, MRI is less operator-dependent than US.
The images can be stored for subsequent analysis or transmitted to a specialist for a second opinion. 19 Third, MRI has an important role in situations in which US findings are impaired during pregnancy. 23,24 US and MRI are complementary imaging techniques in the evaluation of fetal craniofacial anomalies. 2 MRI has inherent soft tissue contrast and shows fetal anomalies, especially of the head and trunk, with  Several strengths and limitations of this study should be acknowledged. First, because of the scarcity of craniofacial anomalies and the limited years of MRI practice, our study is limited by the size of the study group and mainly the diversity of fetal anomalies. Although we have calculated the sensitivity, specificity, and accuracy of US and MRI in detecting specific craniofacial anomalies, caution should be exercised before determining the accuracy of MRI in the diagnosis of fetal craniofacial anomalies as some anomalies were diagnosed only in a few cases. Second, the study is retrospective in design, and the addition of a prospective validation cohort would be optimal. Third, selection bias is another limitation, as only fetuses with possible anomalies identified by skilled obstetric sonographers were referred for MRI. However, this is the population that probably benefits most from advanced imaging. Fourth, this is not a blinded comparison between two imaging techniques-the results of the US had been known to the radiologists when MRI was performed and interpreted.
Although this knowledge was necessary for sequence selection and choosing planes of acquisition, this may induce a bias in favor of MRI. Nevertheless, it conforms to reality.
Among the strengths of this study are the relatively large cohort of congenital craniofacial anomalies evaluated by MRI, postnatal evaluation considered as the gold-standard, the well-defined inclusion criteria for both the study and the comparison groups, and meticulous statistical methods. To overcome the limitations of our study and to generalize our conclusions, a multicenter prospective study should be conducted in order to further evaluate the contribution of fetal MRI to clinical management of congenital craniofacial anomalies.

| CONCLUSION
In conclusion, thanks to a distinguished illustration of fetal face anatomy, multiplanar capabilities, large field of view, and high-contrast resolution, MRI has proven its complementary value to US in the prenatal evaluation of head, face, and neck anomalies. MRI provides valuable information that can alter patient counseling and care. We recommend that fetal MRI be conducted whenever craniofacial anomaly is suspected on prenatal US, and its use should further evolve with the increasing availability of fetal MRI.

DISCLOSURE
The authors report no conflict of interest.

FUNDING STATEMENT
No funding to declare.

DATA STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.