Ultrasound is the screening modality of choice for fetal imaging. However, there are cases where sonographic diagnosis is limited. Fast magnetic resonance imaging (MRI) is being increasingly used as a correlative imaging modality in pregnancy because it uses no ionizing radiation, provides excellent soft-tissue contrast, and has multiple planes for reconstruction and a large field of view, allowing for better depiction of anatomy in fetuses with large or complex anomalies. Although fast magnetic resonance (MR) techniques are widely available, knowledge of fetal anatomy and pathology is important in performing fast fetal MR examinations. It is important to tailor the examination to answer specific questions raised by either patient history or prior sonographic examination. Because fetal diagnosis differs from that of even the pediatric population, the appropriate specialists in obstetric and pediatric imaging need to be available to interpret the examination and to correlate MR findings with those of ultrasound. This article compares and contrasts the use of ultrasound and MR in fetal evaluation.
WHY ULTRASOUND IS IMPORTANT PRIOR TO FETAL MR
Ultrasound is the screening modality of choice in evaluating the pregnant patient. This is because it is relatively inexpensive, has real-time capability, and can identify the vast majority of clinically significant fetal anomalies. When a patient is referred for a fetal MR examination, it is typically because of an anomaly seen on ultrasound. At our laboratory in Beth Israel Deaconess Medical Center, we perform a confirmatory sonogram on all of our referred patients. This protocol was established for research purposes, in order to have a standard of comparison for the sonographic and MR diagnosis. Because the quality of ultrasound is variable in the community at large, this allows for a high-quality standardized examination to be performed.
In addition to sonographic image quality, interpretation of the examination may differ even with good-quality images. In a 1999 study by Hubbard et al. (1), nine of 18 fetuses with chest masses studied by ultrasound and fast MRI were diagnosed incorrectly by ultrasound. Hubbard et al. (1) stated that the examinations were performed “at several institutions,” and that the image quality overall was good but the incorrect diagnoses were due to “misinterpretation of the sonographic findings.” Another example of the importance of standard-quality sonography is from an ongoing study from our laboratory (D. Levine, unpublished data) in which 37 (35%) of 107 fetuses referred for sonographic central nervous system (CNS) anomalies had confirmatory sonograms that changed the fetal diagnosis. In 17 cases the confirmatory sonogram was normal, and in 20 cases the confirmatory sonogram either added additional information or completely changed the diagnosis (Table 1). In cases where the referral and confirmatory results differed, the MR results most closely agreed with those of the confirmatory ultrasound. From these examples it is evident that the quality of the ultrasound and the quality of the interpretation will affect any perceived or real benefit from performing a fetal MR examination.
One final consideration in comparing MR to sonographic findings in the fetus is that fetal anomalies can change in appearance over time. Waiting a few days or weeks between a referral sonogram and MR can lead to both improved visualization of certain structures due to the larger size of the fetus and its organs as well as a different appearance of the anomaly overall due to progression (or regression) of a developmental or disease process. For example, in Fig. 1, a chest mass was visualized sonographically and, at the time of the diagnosis, the stomach was in the fetal abdomen. Therefore, the chest mass was incorrectly thought to represent a congenital cystic adenomatoid malformation (CCAM). The individual who interpreted the ultrasound had not previously seen a small-bowel containing hernia without stomach and, therefore, did not recognize the appearance of the chest mass. MR was performed 41 days later, at which time the stomach was in the fetal chest. Due to a lapse in protocol, the chest mass was not sonographically evaluated at the time of the MR. Although operator experience limited the sonographic diagnosis on the initial scan, it is difficult to compare the correct diagnosis of congenital diaphragmatic hernia (CDH) by MR to the incorrect diagnosis of CCAM made 6 weeks previously at the time of the original sonogram.
Having a high-quality ultrasound is important for comparison for research purposes, but it also is important for clinical care. There are anomalies for which MR performs less well than ultrasound. For example, cardiac anomalies are not well evaluated. Because MR scans are not gated for fetal cardiac motion, the chambers of the heart are not adequately assessed (2). In addition, small encephaloceles, the sacs of neural tube defects, and small arachnoid cysts may be difficult to visualize with MR due to partial volume averaging of the thin-walled structure surrounded by cerebrospinal or amniotic fluid (Fig. 2) (3).
There are a number of other reasons to perform a confirmatory sonogram prior to MR. The first is the unlikely event of a recent fetal demise. Many fetuses referred for MR have significant congenital anomalies that could result in in utero demise. It is always possible that a fetal death could occur in the magnet, so it is prudent to perform a sonogram of the fetus prior to MR to ensure it is alive prior to beginning the examination. In our laboratory to date we have performed MR examinations of over 200 fetuses. One scan was cancelled at the time of confirmatory sonography due to fetal demise that had occurred in the 3-day interval since the most recent sonogram.
Another reason a confirmatory sonogram is helpful is for proper placement of the surface coil. Given the relatively large size of the fetus in the third trimester and the shape of the maternal abdomen, it frequently is not possible to optimally image the entire fetus with a surface coil. If the intracranial contents are of clinical concern and the fetal head is located low in the maternal pelvis, the surface coil should be centered over the region of interest. This decision can be made prior to placing the patient in the magnet. This decreases the need for repositioning of the patient. Similarly with multiple gestations, elucidation of the position of the abnormal gestation will aid in placement of the surface coil and interpretation of the images.
BENEFITS OF MR IN THE EVALUATION OF THE SONOGRAPHICALLY ABNORMAL FETUS: CNS ANOMALIES
One area that holds particular promise for fetal MRI is in evaluation of the abnormal CNS. CNS abnormalities affect approximately 6,000 neonates each year in the United States (4). Sonography is the standard approach for evaluating these anomalies prenatally. However, the sonographic evaluation of the fetal CNS is limited by (1) a nonspecific appearance of some anomalies;(2) technical factors that make visualization of the brain near the transducer difficult and visualization of the posterior fossa difficult late in gestation;(3) subtle parenchymal abnormalities that frequently cannot be visualized sonographically; and (4) at times, the low position of the fetal head in the maternal pelvis that limits visualization of intracranial anatomy. Because of these limitations, MRI has been suggested as a useful adjunct in cases where sonography is nonspecific (3,5,6). In a study from our laboratory, we found that MRI findings led to changed diagnoses in 26 (40%) of 66 fetuses with abnormal confirmatory sonograms. MR findings not visualized by ultrasound included partial or complete agenesis of the corpus callosum (N = 9), porencephaly (N = 6), hemorrhage (N = 5), tethered cord (N = 3), cortical gyral abnormality (N = 2), cortical cleft (N = 2), midbrain abnormality (N = 2), partial or complete agenesis of the septi pellucidi (N = 3), and one case each of holoprosencephaly, cerebellar hypoplasia, subependymal and cortical tubers, vascular malformation, and vermian cysts. Abnormalities better defined by MRI than ultrasound included three cephaloeles, a dural arteriovenous malformation, one distal sacral neural tube defect, and the mass effect of three arachnoid cysts (7).
BENEFITS OF MR IN THE EVALUATION OF THE SONOGRAPHICALLY ABNORMAL FETUS: NON-CNS ANOMALIES
MR can be helpful outside of the CNS. An example is when sonologists have differring opinions of the type of anomaly present. For example, Fig. 3 shows a fetus with a multiseptated cystic mass in the fetal abdomen and pelvis. A portion of the right kidney was visualized, but this was the only normal-appearing renal tissue. Some bowel loops in the right abdomen and pelvis appeared prominent. One sonologist believed the mass was of bowel origin (i.e., a distal small-bowel obstruction), and another sonologist believed that the mass was of renal origin (a duplex kidney with multicystic dysplastic kidney being the inferior moiety). The MR clearly showed the renal origin of this mass in this patient with a crossed fused ectopia, with the lower moiety being multicystic dysplastic.
Another example is a fetus with a chest mass on the right with unexplained mediastinal shift to the right, i.e., toward the side of the mass (Fig. 4) (8). MR showed a cystic mass at the carina with obstructed hyperexpanded lung on the left. The right lung was compressed and was of lower signal intensity than expected, which was believed to be due to compression. The left lung appeared hyperinflated with stretched vessels. The lung parenchyma protruded between the ribs. The findings were best explained by a complex cystic mediastinal mass, likely a bronchogenic cyst, causing obstruction of the left main stem bronchus leading to hyperinflation of the left lung. The compressed right lung and hyperinflated obstructed left lung raised concerns about the ability of the neonate to breathe after birth. Because a well-controlled, planned delivery was necessary for this fetus, an ex utero intrapartum treatment procedure was planned for delivery at 34 weeks. This procedure entailed partial delivery of the fetus with placement of the fetus one extracorporeal membrane oxygenation prior to clamping the umbilical cord.
Another diagnosis in which MR shows promise for improved prenatal counseling is in the assessment of CDH. Benefits of MR in the assessment of CDH include documentation of the position of the fetal liver (9,10), which is important for prognosis because isolated “liver-up” and “liver-down” CDH have respective mortalities of 57% and 7%(9,11,12), and possibly assessment of the amount of normal-appearing lung remaining (9), because the lung is poorly visualized with sonography but is well depicted on MRI (Fig. 1).
COMPARING ULTRASOUND TO MR: ASSESSING THE IMPACT OF ADDITIONAL FINDINGS
That MR can demonstrate anatomy in a manner similar or different from ultrasound is unimportant, unless MR can document anomalies not seen with ultrasound. There is no doubt that the depiction of fetal anatomy with fast T2-weighted MR is excellent. But performing an MRI to evaluate a fetal anomaly is unlikely to change outcome unless additional findings beyond that available with ultrasound can be seen and are of a type that make a difference in the counseling and/or management of the fetus (Fig. 5). An example of this is the evaluation of fetuses with neural tube defects (NTDs). If a fetus has an NTD that is well documented by ultrasound and if the fetus is not being evaluated for fetal surgery, it is unlikely that a fetal MR will change the diagnosis, counseling, or management of the fetus. In our series of 10 fetuses with spinal NTDS, MR contributed no additional information in seven obvious defects (7). However, in three cases there was a Chiari malformation but the distal NTD was difficult to visualize sonographically. On confirmatory sonogram the defect was seen in two and suspected in the third. MR tailored to the distal spine documented the defect in all three fetuses. This information was clearly helpful to the patient bearing the fetus with the poorly visualized defect on ultrasound, because it confirmed the diagnosis and allowed visualization of the abnormality (Fig. 2).
There is no doubt that MR is better than ultrasound in depicting the posterior fossa and is more precise than ultrasound in describing the Chiari malformation, but the incremental benefit of performing an MR for the purpose of evaluating a Chiari malformation has not been established. For this reason, in our laboratory, we no longer routinely perform MR examinations on fetuses with NTDs that have been well evaluated with ultrasound.
A region in which MR frequently provides additional information beyond that available with ultrasound is in the assessment of the normal and abnormal corpus callosum. The corpus callosum frequently is difficult to visualize sonographically. Callosal agenesis is common in cases of fetal ventriculomegaly, but it is missed in up to 50% of cases even when a targeted sonogram is performed (13). In a report by Bennett et al. (14), 15 cases of callosal agenesis remained undiagnosed at less than 22 weeks, with diagnosis being made at late second or third trimester scanning (Fig. 6). Because MRI allows for direct imaging of the corpus callosum, it is very helpful in making the diagnosis of callosal agenesis. This is especially important in cases with partial agenesis of the corpus callosum in which the contour of the lateral ventricles may not have the typical appearance seen in complete agenesis of the corpus callosum. In our study, we found that in six cases of mild-to-moderate ventriculomegaly, agenesis of the corpus callosum or destruction of the corpus callosum was visualized with MRI, which was not clearly visualized sonographically (7). In cases with complete or partial agenesis of the corpus callosum, MRI also was helpful in identifying additional abnormalities, which were present in six of seven cases. This type of information is important in counseling patients, because isolated agenesis of the corpus callosum has a good outcome in 85% of cases, whereas agenesis of the corpus callosum with other associated abnormalities has a poor prognosis (15).
COMPARING ULTRASOUND TO MR: ALLOWING COUNSELING BY INDIVIDUALS UNFAMILIAR WITH ULTRASOUND
Prenatal counseling frequently is performed by pediatric specialists such as pediatric neurosurgeons, who have experience reading MR examinations but have limited ability to interpret sonograms. At times, the benefit of performing a fetal MR is that a specialist can feel more confident about a specific diagnosis and therefore can better counsel the patient.
One question that remains to be evaluated with respect to fetal counseling is how the MR findings in utero correlate with MR findings postnatally. For example, we know that the degree of hydrocephalus is a less important factor than the intrinsic parenchymal damage (16). However, it is possible that the consequences of parenchymal damage in utero may not be the same as parenchymal damage postnatally. Just as it is difficult to assess the amount of residual normal cortex in fetuses with severe ventriculomegaly prior to shunting, the same may be true in our assessment of porencephaly. There is no difficult to assess the amount of residual normal cortex in fetuses with severe ventriculomegaly prior to shunting, the same may be true in our assessment of porencephaly. There is no doubt that the visualization of parenchymal damage portends worse for the outcome than non visualization of cortical destruction, but we must temper our counseling of patients because our knowledge of the natural history of prenatally diagnosed porencephaly with postnatal correlation still is limited (Fig. 7). Similarly, we must temper our counseling of patients with large arachnoid cysts, because our knowledge of the natural history of these lesions also is limited (Fig. 8).
DO MR FINDINGS CHANGE THE LEVEL OF CONFIDENCE IN THE DIAGNOSIS?
One of the major benefits of MR in assessment of congenital anomalies is that it increases the confidence in the diagnosis. This increased confidence can affect patient management and counseling, but it is difficult to quantify this effect. For example, how should the benefit of performing the MR in a fetus with sonographic mild ventriculomegaly in whom the MR provides no additional findings be evaluated? This reassurance to the patient and her referring clinician are valuable but are not easily quantified.
DO THE FINDINGS CHANGE THE MANAGEMENT OF THE PATIENT?
When MR shows additional information beyond that available with ultrasound, the subsequent question is whether the additional information changes management and/or counseling. The quantitative way to address this question is to evaluate change in management and change in outcome. However, these issues become blurred when dealing with the reality of counseling patients. In a study by Coakley et al. (17) MR directly influenced fetal care in four (17%) of 24 fetuses evaluated for complex fetal disorders. In an additional eight cases (33%), MR provided supplementary findings but did not affect fetal care (17). Similar findings were shown in our study of fetal CNS anomalies in which the additional information provided by MRI was of the type that changed patient counseling in 33 (55%) of 66 sonographically abnormal cases (Fig. 9) (7).
This question of change in management or outcome, although seemingly straightforward, frequently is not easy to assess in real-life situations. Patient decision making is based on a number of factors, including imaging findings, bias of the persons performing patient counseling, and personal religious and moral beliefs. Action plans can be requested prior to imaging and compared to those plans made after imaging, but the discussed plans are often ill defined and subject to change even without additional imaging. Three examples of fetuses with encephaloceles are given that demonstrate the types of influence MR has on the decision-making process.
A patient with a fetus at 20 weeks' gestational age was trying to decide whether to terminate the pregnancy. The MR finding of a small encephalocele (Fig. 10) with otherwise normal-appearing brain directly influenced the decision to continue the pregnancy.
In the second case of a fetus with a small encephalocele and no normal-appearing ventricular system on ultrasound, the MR findings of totally disorganized cerebral tissue with no normal-appearing supratentorial structures was used to counsel the patient regarding the dismal outcome for the pregnancy (Fig. 11). The family decided to continue the pregnancy. In this case, the benefit of the MR was to give reasonable expectations to the patient as to the outcome of the pregnancy and prepare the family for the appearance of the neonate at birth.
In a third case of encephalocele, the benefit of MR was more clear-cut. In a fetus with sonographic diagnosis of orbital encephalocele, the distortion of the brain was evident on ultrasound (Fig. 12). However, the decision needed was whether to perform a cesarean if fetal distress occurred, and whether to give life support to the fetus once it was delivered. In this case, the MR findings of no normal supratentorial structures enabled the family, hospital ethicists, clergy, and obstetric and neonatal intensive care team to devise a plan for delivery and perinatal care. The obstetrician was grateful for the diagnosis being made ahead of time of delivery, because at the time of delivery when the neonate had his eyes closed, he looked fairly normal.
SCREENING FOR ANOMALIES WITH MR
There are a number of anomalies for which prenatal screening is not available or is suboptimal. MRI has the potential to aid in prenatal diagnosis of hemachromatosis (17,18), in which T2* imaging can be used to show low signal intensity in the iron-overloaded liver; tuberous sclerosis (Fig. 13), in which subependymal tubers have been demonstrated as early as 21 weeks (19); and lissencephaly, in which MR has the potential to demonstrate migrational abnormalities that are not visible with ultrasound.
HIDDEN RISKS OF MR COMPARED WITH ULTRASOUND
When addressing the issue of using MR in prenatal diagnosis, a hidden risk of MR arises. Visualization of anatomy is possible in a manner not possible with ultrasound. Because of this ability, we will inevitably encounter incidental findings in the fetus that will not change management and area unrelated to the referral diagnosis, but will increase patient anxiety. Additionally, we will discover findings of uncertain significance, which again will increase patient anxiety. Examples of this are shown from our CNS study [(7) and Levine, unpublished data], where normal sonograms were followed by MR incidental findings of an enlarged subtemporal vein, a subependymal bleed (Fig. 14), and focal thoracic diastematomyelia.
An incidental finding is relatively less important when evaluating fetuses with previously diagnosed sonographic abnormalities, because patients already realize that there is a congenital anomaly. However, these incidental findings are of crucial importance when using MR as a screening modality in otherwise normal fetuses. Patients enter into a screening program with the hope that the fetus will be normal. Even a minor abnormality, with no apparent consequence to the outcome of the pregnancy, can increase patient anxiety.
Ultrasound is of crucial importance in evaluating the pregnant patient and the fetus. The information provided by ultrasound is useful, not only for screening but also as a guide for tailoring the MR examination. Proper use and interpretation of ultrasound will limit the cases in which MR is necessitated. The individuals interpreting the MR examination also should use the information provided by the ultrasound. Differential diagnosis in the fetus is not the same as that of the neonate. Persons familiar with prenatal diagnosis and those familiar with neonatal diagnosis should be involved in the interpretation of MR examinations, as both types of subspecialized information will complement each other. Good communication between individuals interpreting the examinations is vital, because the information provided by both ultrasound and MR is integral in allowing for appropriate counseling of patients. MR has a valuable contribution to make in the counseling of pregnant patients, even when findings are similar to those of ultrasound. Although MR commonly provides additional information beyond that available with ultrasound, it less commonly results in change in care of pregnant patients or alterations in pregnancy outcome.
This work was supported by NIH grant NS37945.
1. Hubbard AM, Adzick NS, Crombleholme TM, et al. Congenital chest lesions: Diagnosis and characterization with prenatal MR imaging. Radiology 1999; 212:43–48.
2. Levine D, Barnes P, Sher S, et al. Reproducibility, technical quality, and conspicuity of organs in evaluating fetuses with fast MRI. Radiology 1998; 206:549–54.
3. Levine D, Barnes PD, Madsen JR, et al. Fetal central nervous system anomalies
: MR imaging augments sonographic diagnosis. Radiology 1997; 204:635–42.
4. Carrasco CR, Stierman ED, Harnsberger HR, et al. An algorithm for prenatal ultrasound
diagnosis of congenital CNS abnormalities. J Ultrasound
Med 1985; 4:163–8.
5. Thickman D, Mintz M, Mennuti M, et al. MR imaging of cerebral abnormalities in utero. J Comput Assist Tomogr 1984; 8:1058–61.
6. Dinh DH, Wright RM, Hanigan WC. The use of magnetic resonance imaging
for the diagnosis of fetal intracranial anomalies. Childs Nerv Syst 1990; 6:212–5.
7. Levine D, Barnes PD, Madsen JR, et al. Central nervous system abnormalities assessed with prenatal magnetic resonance imaging
. Obstet Gynecol 1999; 28:169–74.
8. Levine D, Jennings R, Barnewolt C, et al. Progressive fetal bronchial obstruction caused by a bronchogenic cyst diagnosed by prenatal MR imaging. AJR Am J Roentgenol
9. Leung JWT, Coakley FV, Hricak H, et al. Prenatal MR Imaging of congenital diaphragmatic hernia. AJR Am J Roentgenol 2000; 174:1607–12.
10. Hubbard AM, Adzick NS, Crombleholme TM, et al. Left-sided congenital diaphragmatic hernia: Value of prenatal MR imaging in preparation for fetal surgery. Radiology 1997; 203:636–40.
11. Metkus AP, Filly RA, Stringer MD, et al. Sonographic predictors of survival in fetal diaphragmatic hernia. J Pediatr Surg 1996; 31:148–52.
12. Adzick SN, Harrison MR, Glick PL, et al. Diaphragmatic hernia in the fetus: Prenatal diagnosis
and outcome in 94 cases. J Pediatr Surg 1985; 20:357–61.
13. Pugash D, Langlois MN, McFadden DE, et al. Prenatal ultrasound
diagnosis of ventriculomegaly associated with dysgenesis of the corpus callosum. J Ultrasound
Med 1997; 16:S12.
14. Bennett GL, Bromley B, Benacerraf BR. Agenesis of the corpus callosum: Prenatal detection usually is not possible before 22 weeks of gestation. Radiology 1996; 199:447–50.
15. Gupta JK, Lilford RJ. Assessment and management of fetal agenesis of the corpus callosum. Prenat Diagn 1995; 15:301–12.
16. Giudetti B, Occhipinti E, Riccio A. Ventriculo-atrial shunt in 200 cases of non-tumoroural hydrocephalus in children: Remarks on the diagnostic criteria, postoperative complications and long-term results. Acta Neurochir 1969; 21:295–308.
17. Coakley FV, Hricak H, Filly RA, et al. Complex fetal disorders: Effect of MR imaging on management—Preliminary clinical experience. Radiology 1999; 213:691–6.
18. Marti-Bonmati L, Baamonde A, Poyatos CR, et al. Prenatal diagnosis
of idiopathic neonatal hemochromatosis with MRI. Abdom Imaging 1994; 19:55–6.
19. Levine D, Barnes P, Korf B, et al. Tuberous sclerosis: Second trimester diagnosis of subependymal tubers with fast MRI. AJR Am J Roentgenol 2000; 175:1067–9.
20. Levine D, Barnes PD, Madsen J, et al. Fetal CNS anomalies depicted with ultrafast MR imaging. Am J Roentgenol 1999; 172:813–8.
21. Levine D, Barnes PD, Edelman RR. State of the art: obstetric MR imaging. Radiology 1999; 211:609–617.
Issue Editor: Deborah Levine