Fetal cerebral magnetic resonance imaging, neurosonography and the brave new world of fetal medicine

Stuart Campbell was the first to use the metaphor of parallel universes when discussing the utilization of ultrasound in pregnancy1. The controversy surrounding the use of magnetic resonance imaging (MRI) vs sonography to diagnose fetal cerebral malformations is another example in this regard. In one spatiotemporal continuum, fetal neurosonography allows an accurate diagnosis of central nervous system anatomy2-5 and malformation6-9, including complex lesions such as commissural anomalies10, 11, posterior fossa malformation12, 13, cerebral insults14-16 and cortical anomalies17; it performs similarly to MRI in the second and third trimesters18, 19 and it even allows a precise diagnosis of several malformations as early as in the first trimester20-22. In the other continuum, the laws of physics are quite different; sound waves propagate with more difficulty, and MRI has the upper hand, mostly because ultrasound fails to recognize even gross anomalies23-25. Fifteen years ago, this Journal published an Editorial in which the existing data regarding the alleged superiority of MRI in the diagnosis of fetal central nervous system (CNS) anomalies were analyzed26. It was postulated that the studies were biased for different reasons, the most important one being ambiguity in the definition of the type of ultrasound examination compared with MRI. It is well recognized that sonography of the fetal brain falls into two main categories27: a basic examination, usually involving axial views, and a dedicated examination, or ‘neurosonogram’, performed with a multiplanar approach and taking advantage, whenever possible, of the much greater resolution of transvaginal, rather than transabdominal, sonography. The former is essentially a screening examination performed by an operator with limited knowledge of brain malformations and the latter is a diagnostic examination performed by an experienced sonologist. Not surprisingly, the performance of the two exams is radically different. It seems that the recent MERIDIAN study, a prospective cohort investigation conducted in the UK, has fallen into this trap of not recognizing the difference between a screening sonogram and a neurosonogram28. The authors claim that the addition of prenatal MRI changed the prognostic information in 20% of cases of fetal cerebral malformation, and that it modified the management in one case out of three in comparison with ultrasound used alone. However, no allowance was made for the different types of ultrasound technique, which we would argue implies a flawed study design because it has long been known that different types of cerebral sonography exist. It would seem that most of the examinations were of the basic rather than the dedicated type, for the following reasons: the diagnostic accuracy of ultrasound was very poor; descent of the fetal head in the maternal pelvis was described as a limiting factor, suggesting that transvaginal scans were rarely, if ever, performed; and the 1109 eligible cases were evaluated over a 3-year period at 16 different fetal medicine units, which translates into an average of 22 cases of malformation per year per center, clearly a workload insufficient to develop adequate expertise. We are also puzzled by the statement that sonographic diagnoses were categorized using the nomenclature provided by the ViewPoint software. ViewPoint is a popular tool for documenting, reporting and archiving obstetric and gynecologic sonograms, but an inadequate reference standard for indexing fetal cerebral anomalies. The 2017 update of the 5.6 version that we use, and which is likely to have been the one utilized during the study, contains a list of only 11 items (six, if one excludes anencephaly and other neural tube defects), that does not include many of the most common abnormalities (e.g. agenesis of the corpus callosum, Dandy–Walker malformation, absence of the leaflets of the septum pellucidum, porencephaly, cortical malformations). This is a matter of serious concern because it implies that the authors either described incorrectly their Methods or they restricted the sonographic diagnoses to a handful of conditions that are not representative of the usual spectrum of anomalies. In a series of papers in this issue of the Journal, Griffiths and colleagues report secondary analyses from the MERIDIAN cohort, involving subgroups of fetuses with ventriculomegaly29, posterior fossa abnormalities30 and ‘failed commissuration’31. In these study groups, anomalies that should be straightforward to diagnose with contemporary sonography, such as complete agenesis of the corpus callosum, brain atrophy or rhomboencephalosynapsis, were consistently missed or overdiagnosed by ultrasound. In fact, we are astounded that ultrasound (performed before MRI) failed to recognize the brain anomalies underlying ventriculomegaly that are displayed in the MR images provided29, and this calls into question the statement that the sonographers were ‘appropriately trained’. The authors of the MERIDIAN studies conclude that MRI has greater accuracy than does ultrasound in diagnosing intracranial anomalies. We suggest a different interpretation of their data: in many tertiary care centers in the UK, obstetric ultrasound performs poorly. At the end of the day, this is a problem of allocation of resources. Developing and maintaining the skills to perform fetal neurosonography requires an investment and, in some obstetric facilities, it may be easier (albeit not necessarily more economical, or more convenient for the patients) to delegate to the neuroradiologists. We do not deny that fetal MRI is a powerful tool and that, in selected cases, it provides additional valuable information. What we find unreasonable is the non-selective approach suggested by the MERIDIAN collaboration and the concept that any fetus with a suspicion of a cerebral anomaly should undergo MRI. Fetal medicine is a crossroads where many specialties meet (and sometimes collide). As such, we believe that fetal medicine experts should take advantage whenever possible of the different skills and qualifications of their colleagues, while maintaining a pivotal role in the process of diagnosis, counseling and management. Fetal medicine centers should have an obligation to provide high-quality ultrasound and develop the best possible competence in diagnosing fetal anomalies, including brain malformations32, resorting to MRI when necessary, but not automatically18, 19, 26, 33. The MERIDIAN study may have far-reaching consequences for contemporary practice. We are concerned because, although this is a large and diligent survey, it has at least one major bias (the suboptimal performance of ultrasound) and possibly a second (inadequate indexing of sonographic diagnoses) and a third (only half of the cases with suspected CNS malformation underwent MRI, the other half were managed on the basis of ultrasound findings alone). We acknowledge the pragmatic nature of the study. However, it should be emphasized that the results may apply only to the geographic area in which the data were collected and/or to settings in which obstetric ultrasound performs suboptimally – they should not be generalized.