Fetal magnetic resonance imaging: luxury or necessity?

During its latest annual meeting, the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) dedicated two sessions to the use of magnetic resonance imaging (MRI) in obstetrics and gynecology. In doing so, it acknowledged the increasingly important role of MRI in our practices. Today, with hardware becoming increasingly affordable, patients get more and easier access to this type of imaging, and insurance companies or other authorities do not question its use for a wider range of indications. At the same time MRI has become faster so that fetal movements are less of a problem, and fetal imaging is no longer beyond our reach. Modern equipment allows the taking of ‘snap-shot’ images with resolution and contrast detail never before witnessed. Nevertheless, the examination is certainly not used widely in pregnancy, despite a reassuring guideline on the safety of MRI from the American College of Radiology, which should dispel any doubt amongst skeptics1-3. We believe it is time for a change. In units where radiologists and fetal medicine specialists, or other players in the field of perinatal therapy, become more familiar with fetal images, enthusiasm increases. So, too, does that of our Society: the White Journal publishes an ever-increasing number of papers involved with MRI applications, as is evident in this issue. At one point, some faculty members at the last ISUOG World Congress in Vancouver even suggested changing the name of our Society to the International Society of Imaging in Obstetrics and Gynecology. Admittedly, this was somewhat overzealous, but it reflects well the idea that, whatever the nature of the technology, we use imaging techniques as tools to solve clinical questions. However, for most of us there remain many barriers that prevent us from fully embracing fetal MRI. First, we very rarely feel the clinical need. Sonography is undoubtedly satisfactory for everyday screening activities and even for imaging the majority of fetal malformations. Furthermore, MRI is a tool in ‘other’ hands: in all units, fetal MRI is conducted by radiologists and their technicians. This constitutes both a physical and a logistical barrier. We do not always know what to expect from each other: gynecologists are not trained in the physics behind the impressive MRI examination technology; radiologists cannot be expected to have the entire subspecialty background on which a maternal–fetal medicine specialist relies. However, the time has come for us to liaise with each other and explore the advantages and limitations of this procedure, to define in which situations fetal MRI can supplement our ultrasound findings, and to explore new applications, stretching beyond anatomical evaluation, and including functional studies. Thus, the editorial office judged it time for a brief round-up of what has been established (recently) and what is in the pipeline, with respect to MRI in the field of fetal medicine. Clinically, the most useful images are classified as T1- or T2-weighted images and diffusion-weighted images4. The MRI technique relies on detection of the magnetic moments of hydrogen nuclei, present in abundance, and with a variety of bonds in multiple molecules, throughout our cells and tissues. Differences in the structural components of tissues translate into contrast differences on the images. Water and fat constitute the two extremes of MRI contrast, the former having high signal intensity on a T2-weighted image but low signal intensity on a T1-weighted image. Structures containing less water are therefore darker on a T2- weighted image. In contrast, fat molecules have high signal intensity on a T1- weighted image. Diffusion-weighted images measure the thermal motion of water molecules in the human body; in biological tissue this is not truly random, because membranes, vessels, cell structures and interactions of water with macromolecules present barriers to diffusion. Most pathological conditions have an impact on tissue structure, with measurable facilitation or restriction of water-molecule movement, resulting in changes in the diffusion-weighted image. At the moment MRI cannot compete with the dynamic qualities offered by high-resolution ultrasound, but ultrafast sequences enable nearly real-time imaging, so that even this hurdle might one day be overcome. At most institutions, fetal MRI is performed on standard clinical 1.5-Tesla (T) units, and these constitute the bulk of machines now used worldwide. Although there is currently no instrumentation specific for fetal MRI, a few conditions need to be met to enable successful imaging to be performed. Preferentially, the unit should be equipped with an advanced spatial encoding gradient system (with gradient switching capabilities of 25 mT/m in 300 µs), allowing for fast data acquisition (< 400 ms/slice) with high-resolution capabilities (matrix size up to 512). This is needed to reduce fetal motion-induced artifacts and image degradation (blurring) during acquisition. Moreover, stronger gradients (e.g. 3T units) allow acquisition of images with a higher signal-to-noise ratio and higher spatial resolution. Although this theoretically increases energy exposure, the latter can be controlled by adjusting other variables of the examination, so that safety is less likely to be an issue; however, no formal studies on this are available. Similarly, the theoretically better image quality has not yet been translated into a proven and relevant clinical advantage of 3T over 1.5T systems. To date, there have been no reports of clinical fetal MRI performed on 3T units. Practically, patients are positioned in the supine or left-lateral position to prevent aorta-caval compression. To maximize image signal capture, a combination of a surface coil (placed over the pelvis or pregnant uterus) and a lower spinal coil are used. The average duration of a fetal MRI examination ranges between 10 and 35 min, depending on the clinical indication and fetal movements. These movements can, but need not necessarily, be reduced by maternal pharmacological sedation5, 6. The exact MRI protocol will depend largely on the clinical question being addressed; therefore, consultation with the radiologist in advance is a must. With the exception of universal restrictions of MRI per se, no clinical hazardous effects have been reported. Delayed yet undiscovered sequelae seem to be unlikely3, 7. Most centers use a cautious approach, not examining fetuses in the first trimester or prior to 18 weeks8, 9. Theoretically, there is the potential for interference with organogenesis and, in any case, current technology cannot achieve very satisfactory imaging so early in gestation. Intravenous injection of contrast agents is not recommended10, but this was questioned in a recent study by Webb et al.11. Because of ‘back’-diffusion across the placenta from the fetal to the maternal circulation, they claimed use of a contrast agent to be ‘probably’ safe for pregnant patients. So far we have not used contrast agents clinically for fetal MRI. The CNS is the fetal organ system that is most widely studied by MRI and suspected CNS anomalies represent the most common indication for fetal MRI7, 12, 13. Its potential lies in the fact that, by using a combination of T1, T2 and diffusion weighting, MRI can provide structural, maturational and even functional information14. It can be used easily to determine both brain structure and maturation by assessing brain biometry13, 15, gyral formation16 and parenchymal morphology17 on T2-weighted imaging, and the degree of myelinization can be determined on T1-weighted imaging18. For clinical interpretation, a good knowledge of the normal chronology and pattern of CNS maturation, including individual variation, is needed. Conventional fetal MRI, just as with ultrasound, certainly still has its limitations. Small parenchymal lesions, such as ischemic foci, can be overlooked. On indication, diffusion-weighted images should be obtained. These are extremely sensitive in detecting cytotoxic and/or vasogenic edema19, 20. Diffusion weighting also provides additional information on brain maturation, i.e. premyelinating white matter. Spectroscopy is another technique that is informative for tissue metabolism, but its diagnostic potential is still being explored21, 22. Fetal MRI of the CNS is no longer a gimmick. In a study on 214 fetuses with CNS problems detected initially by ultrasound, Levine et al.23 demonstrated that fetal MRI yielded additional diagnostic and therapeutic information in 23% and 14% of cases, respectively. Even better, as a result of MRI, the ultrasound-based diagnosis and management were modified in 32% and 19% of cases, respectively (n = 145). Other studies on larger cohorts of 250 fetuses24 and 100 fetuses25 confirmed these findings. Spinal malformations are documented easily by ultrasound, but sacral spinal defects may benefit from fetal MRI examination to differentiate them from sacrococcygeal teratoma. In the latter, MRI can demonstrate intra-abdominal extension, as described by the classification of Altman et al.26, as well as intralesional hemorrhage and urinary tract obstruction, information that is very useful to the perinatal surgeon27. Sometimes fetal MRI is used to overcome the limitations of ultrasound in very individual cases: in selected patients fetal skull ossification and pelvic position of the fetal head preclude appropriate ultrasound but not MRI. The fetal lung is a fluid-filled structure which thus has a high signal intensity on T2 weighting, and is easily discernible from the surrounding structures28, 29. This remains so even in cases of oligohydramnios and obesity, conditions that make appropriate ultrasound evaluation difficult or at least inaccurate. Fetal MRI may be helpful in the diagnostic work-up of mediastinal and thoracic masses as well as lung lesions30. Hubbard31 described the use of fetal MRI to distinguish congenital diaphragmatic hernia (CDH) from congenital cystic adenomatoid malformation or less common chest tumors. Levine et al.32 showed that fetal MRI yielded information additional to that obtained by advanced ultrasound in 28 of 74 (38%) cases presenting with thoracic anomalies. CDH is the most common thoracic pathology, and besides the differential diagnostic problems, the main challenge in this condition is prediction of lethal pulmonary hypoplasia. Preferentially, prognostic evaluation should be done at the latest at the end of the second trimester, so as to avoid painful choices beyond theoretical viability33-35, helping parents to decide on further management, such as termination rather than expectant management. This challenge in fact applies to any condition leading to pulmonary hypoplasia and therefore is clinically highly relevant. The need for prediction of CDH recently became even more important, as selected fetuses with severe CDH may now be offered prenatal intervention36-38. Today, case selection is made by ultrasound, using liver position and measurement of the so-called lung-to-head ratio34. Liver position can be determined more unequivocally and even quantitatively on T1-weighted MRI. To determine lung size, several centers are exploring the potential of three-dimensional lung volumetric techniques both using ultrasound39 and using fetal MRI. Several studies have shown that MRI lung volume assessment can predict outcome, but nearly all of these were conducted late in gestation40-42. Usually, the result is expressed as observed lung volume/lung volume that can be expected in a comparable healthy fetus. In some studies normal fetuses were selected based on their gestational age, but it is much better to select them based on biometric variables, such as abdominal circumference and liver volume. Intuitively, biometric indices should be more reliable because they avoid problems such as incorrect dating or altered fetal growth43, 44. In a normative study in 36 normal fetuses, we recently proposed using fetal body volume45. The algorithm was also applied in a series of 37 fetuses at high risk for lung hypoplasia, assessed prior to 29 weeks with documented perinatal outcome. For 19 CDH fetuses, LHR measurements were also available. The widely accepted threshold of LHR = 1.0 correlated to an observed/expected lung volume ratio of 35%. Of interest is that survival could not be predicted by volume alone; only after addition of liver position did the prediction become accurate. It is not only lung volume and structure that can be appraised using MRI, but also more functional characteristics such as parenchymal signal intensity, diffusion and/or spectroscopic distribution of metabolites. Signal intensity of the lung increases throughout pregnancy46, and this could be used in a non-invasive prenatal functional lung assessment. Osada et al.47 used the lung/spinal fluid intensity ratio in combination with lung volume in fetuses between 24 and 39 weeks' gestation, concluding that prediction of respiratory outcome was better than that achieved when using lung volume alone. We eagerly await confirmation from larger studies as well as studies earlier in gestation. Fetal lungs can also be assessed functionally with diffusion weighting. This has been done in 33 normal fetuses in order to establish the normal range and the authors concluded that fetal lung diffusion is a marker for the degree of pulmonary vascularization48. Functional studies in fetuses with pulmonary hypoplasia are not yet available. Spectroscopy has theoretical potential for evaluation of lung maturation: it can estimate the amount of lecithin, a major component of lung surfactant49, 50. Unfortunately, the application of this technique is severely compromised in practice by several limitations, especially motion artifacts, and it can therefore only be seen as a more distant resource. Both T1 and T2 weighting are used to study the GI tract and abdominal wall51. Meconium has a typical hyperintense signal on T1 weighting and a hypointense signal on T2 weighting. Saguintaah et al.52 proposed its use as a natural contrast agent. It enables depiction of intestinal anatomy from 20 weeks onwards. They described the normal distribution of meconium and amniotic fluid within the fetal GI tract at different developmental stages in 40 fetuses, resulting in a MRI topography of the small bowel and colon. The same group applied this knowledge to 32 fetuses with suspected bowel pathology53 and found that fetal MRI determined much more easily the level of obstruction than did ultrasound. Therefore, the use of fetal MRI to rule out megacystis-microcolon-intestinal-hypoperistalsis syndrome is undeniable. When a normal-sized colon can be visualized, this lethal pathology can be excluded. Fetal MRI can also be applied to abdominal wall defects51. The typical content of the protruding mass can be appraised easily, but, more importantly, intestinal ischemia, suggested by a thickened intestinal wall (> 3 mm) or less commonly by meconium cysts, can be a sign of bowel perforation54. To evaluate the fetal renal system, T2 weighting is most valuable because it allows visualization of the renal parenchyma and excretion system. T1 weighting is less crucial, but may be helpful in very rare cases of renal tumors and the more frequently associated (extrarenal) structural fetal anomalies. Structural problems are usually visualized equally well in the fetus by ultrasound and by MRI. This was confirmed recently in studies by Poutamo et al.55 and Cassart et al.56. However, when oligo- or anhydramnios renders ultrasound unreliable, other than diagnostic amnioinfusion, MRI is the only option. Function is another clinical question to which MRI can provide answers. Diffusion-weighting studies might yield renal functional information. Using MRI, Witzani et al.57 documented the length of the fetal kidney, the signal intensity of the renal parenchyma and the apparently normal diffusion coefficient of the entire kidney in 107 normal fetuses between 17 and 28 weeks. These variables can now be determined in pathological renal conditions. Another possible application of fetal MRI is renal spectroscopy, to measure renal metabolites. However, experience with fetal renal spectroscopy is lacking, and its use can only be considered as experimental at this stage. There has been increasing interest in therapeutic intervention in selected monochorionic twin pregnancies. In this situation also, MRI can be applied. Theoretically, in the diagnostic-work up of discordant anomalies, fetal MRI could be used whenever ultrasound proves unsatisfactory. Twin-to-twin transfusion syndrome (TTTS) is a condition that is diagnosed on ultrasound58. Relevant surgical landmarks, such as fetal position, cord location and the best place of entry towards the presumed vascular equator or membrane insertion, can all be determined by ultrasound. The use of fetal MRI for the above was suggested by Luks et al.59 some years ago. We would also use fetal MRI occasionally for determination of cord insertion for those cases in which the stuck twin hides the placental insertion completely, or to confirm close cord insertions, which are usually associated with very large anastomoses, rendering surgery difficult if not impossible. Fetal MRI has been shown to be helpful in the determination of the vascular equator; in at least 75% of cases, it was recognized based on a difference in intensity on T2 weighting between donor and recipient parts. The clinical consequences and relevance of this need to be demonstrated, but so far it has unfortunately failed to predict which fetus, if any, is likely to die after laser surgery60. Another application could be (preoperative) evaluation of intracranial structures, in TTTS cases or other complicated monochorionic twins. Monochorionic twins have a much higher risk than do fetuses of white and fetal MRI could certainly be of in this application is of interest to the fetal MRI This may be an in the of cases which cannot be because of to the of their The was by et and theoretically all conditions for a very high-resolution image are motion artifacts are and theoretically acquisition can be used. In this first study the of and fetal MRI in 20 fetuses were MRI could for the detection of CNS but not for malformations. These were confirmed by the of et and et et demonstrated that MRI could CNS anomalies even better than could a that is when one into the of the brain during the of such small Therefore, MRI should certainly be offered to parents are to this examination will never as no pathological can be but this can be overcome by a and much less of the or of In that MRI has a role in fetal medicine can no longer be MRI has been for select clinical structural evaluation of brain and We are that the advantage and of fetal MRI is in the functional evaluation of the fetus. no does fetal MRI compete with ultrasound, which will remain the screening as a and real-time MRI is there to other diagnostic with its tissue contrast, large field of and for several a still to be this should not prevent from to to or even up a and in order to a selected number of In this that we have to an on the MRI sessions on