Portal de Periódicos da CAPES

Fetal MRI: what is the future?

2008; Wiley; Volume: 31; Issue: 2 Linguagem: Inglês

10.1002/uog.5249

1469-0705

Cathérine Garel,

MRI in cancer diagnosis

For the first 10 years of its clinical application, the use of fetal MRI was focused on the brain, and it is impossible to refer to all of the numerous articles dealing with this topic. It rapidly became obvious that MRI depicts the fetal brain exquisitely and provides details completely overlooked by ultrasound. During the second decade, other organs have been studied. The contribution of MRI in the evaluation of cervical masses, pulmonary malformations, intestinal obstruction and renal abnormalities has been emphasized, generating numerous other articles and case reports. However, it must be stressed that even though MRI undoubtedly has a place in this field, in common practice this type of indication does not exceed 10–15% of all imaging examinations. A few comments are essential. First, the diagnostic benefits of ultrasound examination in pregnancy are well documented. Mothers' expectations and the psychological consequences of such a procedure are less well known, but have been analyzed1-3. The psychological impact of sonography in high-risk cases (in comparison with routine scans) has also been studied4. The psychological condition of a woman undergoing fetal MRI can be compared to that of a woman undergoing such an ultrasound scan, because the decision to perform MRI is usually a result of the detection of an abnormality on ultrasound or a suggestive family history. The situation, therefore, is never neutral. To my knowledge, the psychological impact on the mother of undergoing fetal MRI has not previously been analyzed. We performed such a study (unpubl. data) in collaboration with our colleagues from Hôpital Saint-Vincent de Paul in Paris (Professor C. Adamsbaum). It was difficult to compare our two cohorts because the way in which both teams worked was not identical. For example, in our institution (Hôpital d'Enfants Armand-Trousseau), we always perform ultrasound just before MRI, explaining to the mother during sonography what MRI consists of, and we sedate all mothers. In the cohort studied in Hôpital Saint-Vincent de Paul, ultrasound was not performed just before MRI and the mothers were not sedated. In spite of these differences, we noted in both cohorts that patients felt a greater level of anxiety before MRI than before ultrasound. During the MRI examination, sedated patients were less anxious than were non-sedated patients. Some non-sedated patients reported how hard it was for them to remain alone in the MRI unit without being able to see the physician's facial expression, unlike during sonography. The necessity of conversing with the patient during the examination was emphasized. We can conclude that MRI has a potential negative effect on the mother from a psychological point of view, and this must be considered when MRI is requested. It must also be remembered that MRI is much more expensive than ultrasound. Second, normal fetuses are screened regularly with ultrasound. Therefore, numerous sonographic norms and measurements have been established in large cohorts of fetuses. Conversely, for ethical reasons, one cannot plan to perform fetal MRI in normal fetuses. Thus, only high-risk normal fetuses have contributed to the establishment of MRI norms (for example, those with a suggestive family history or those with cardiac rhabdomyomas)5-12. Therefore, normal variants are not as well known in fetal MRI as they are in ultrasound. Third, the way in which fetal MRI is used throughout the world is highly dependent on the local legislation regarding termination of pregnancy (TOP). It is obvious that in countries where TOP is allowed very late in pregnancy (e.g. France, the UK, Belgium and Israel), radiologists tend to perform cranial MRI late in order to increase spatial resolution and to wait for gyration to develop. Conversely, when TOP is allowed only until the mid second trimester (e.g. the USA and the south of Europe), MRI is of course performed much earlier, when artifacts related to fetal movements are more frequent and spatial resolution is decreased. Another factor influencing the use of MRI is the level of experience of the sonographers referring the patients for MRI. It is obvious when reading the prenatal literature that this level is variable. One example of a commonly debated topic is ventriculomegaly and, more precisely, the contribution of MRI to the evaluation of what is considered isolated ventriculomegaly on ultrasound. The rate of detection of associated abnormalities overlooked by ultrasound varies from 5–9%13-15 to 50%16! Regarding the contribution of MRI to the analysis of the fetal brain, some authors claim the supremacy of ultrasound17 and others emphasize the necessity of performing MRI16. The truth may lie in between. The modalities are undoubtedly complementary and it is not necessary to champion one over the other. The sonographers should try to exploit the capabilities of ultrasound and a serious effort should be made to increase the level of sonographic experience in referral centers. Also, the radiologists who, unfortunately, do not usually have access to fetal ultrasound should at least be aware of its capabilities so as not to 'discover' on MRI that which was obvious on ultrasound at a lower financial and psychological cost. Finally, in parallel with the development of MRI, the quality of ultrasound images has increased dramatically during the last two decades, therefore reducing the gap between these modalities. One example is the visibility in three-dimensional ultrasound of the optic chiasma18, a structure which, until recently, could be depicted by MRI only. So far, ultrasound and MRI have provided an anatomical approach to the fetal body. Increasing numbers of subtle details have been depicted. It is indisputable that progress will continue to be made in both modalities. However, it is undoubtedly necessary to refine the anatomical approach and to add metabolic and functional studies. This is the future challenge in MRI. Three 'new' techniques in fetal MRI are currently being developed in a number of centers worldwide. These are well-known techniques that are commonly used after birth, they are not yet used in routine clinical practice prenatally. The physical principle of diffusion-weighted imaging (DWI) is based on the changes in proton spin cohesion that result from the random (Brownian) motion of fluid molecules, such as water. Two magnetic field gradients, opposite in direction and equal in amplitude, are applied during DWI. Because of the constant motion of water molecules, the second gradient does not completely rephase the proton spins, resulting in a decrease in signal intensity. The more freely diffusible the water molecules are, the more random is the orientation of the water protons and the faster is the decay in signal intensity. For example, diffusion is physiologically increased in cerebrospinal fluid (ventricles and pericerebral space), whereas it is decreased in the cerebral parenchyma19, 20. The signal intensity created with DWI has T2 characteristics, so it may be difficult on a DW image to distinguish hyperintensity related to true reduction of diffusion from hyperintensity reflecting T2 prolongation ('T2 shine-through'). Thus, DWI is also evaluated by measuring the apparent diffusion coefficient (ADC). The higher is the ADC in a structure, the greater is the diffusion. For example, ADC is much higher in the ventricles than it is in the cerebral parenchyma19, 20. Diffusion encoding gradient can be applied on any gradient axis and is therefore sensitive to a diffusion process in different directions. Diffusion is considered 'isotropic' if there is no directional preference and 'anisotropic' if movement is impeded in some directions, resulting in a preferential direction of water molecule motion. Diffusion tensor imaging (DTI) is used to depict the direction and degree of anisotropy; fractional anisotropy can be evaluated in different structures. DTI makes it possible to reconstruct white matter fiber paths by using the pixel-by-pixel information on diffusion anisotropy. This process is called 'tractography' or 'fiber tracking'19, 20. DWI is performed using echo planar imaging. It is a sequence that suffers from a lot of background noise and which is very sensitive to motion artifacts. Moreover, spatial resolution is poor, which may lead to partial volume artifacts. The acquisition time is short (less than 1 min). The fetal brain has a high water content, so the diffusion phenomenon is amplified. During brain maturation, there is a progressive decrease in total water content within the brain and an increase in lipid concentration. The so-called 'premyelination' state is characterized by a progressive restriction of water motion, with a decrease in ADCs, a restriction of water diffusion directionality and an increase in anisotropy. Changes in DWI and DTI are well documented in premature and term newborns21-25. Normal evolution of diffusion in fetal brains has been studied throughout pregnancy26 and mean normal ADC brain values have been reported in small series27-29. The development of multiple forebrain commissures (including the corpus callosum, anterior commissure and optic chiasma) has been demonstrated in vitro using DTI30. The mean renal tissue ADC value has been evaluated in normal fetal kidneys in small series. It has been reported to be independent of gestational age in one series31 and to decrease with gestational age in other series32, 33. Such a decrease in ADC throughout pregnancy was attributed to changes in blood flow, tubular flow, water content or cell density (increase in the number of glomeruli). ADC values have also been determined in fetal lungs. In one study, the values were correlated with neither gestational age nor lung volume. Mean ADCs decreased significantly from the apex towards the base. It was not possible to identify changes correlated with lung maturation34. However, in another study, a trend towards lung ADCs increasing throughout pregnancy was observed35. For postnatal imaging of the brain, the advent of DWI has been a major turning point in the use of MRI. The applications of DWI in the evaluation of the central nervous system are numerous; thus, it has become an inevitable tool in the evaluation of many disorders, such as ischemic damage, tumors, trauma, infection and metabolic disorders. In fetuses, tumors, trauma and metabolic disorders are very scarce, but the detection of ischemic lesions (which may or may not be infectious in origin) is a real challenge. Focal ischemic parenchymal and cortical damage can be detected by 'conventional' fetal MRI36, 37. Conversely, the evaluation of diffuse white matter lesions, based on signal abnormalities, is always subjective and difficult. DWI could help in identifying the earliest changes of hypoxia–ischemia affecting the fetal brain38. Correlations between histology or clinical outcome and DWI changes have been analyzed in animal models of gestational hypoxia. In pregnant rats, white matter lesions induced by hypoxia have been studied with DWI. At birth, white matter lesions were associated with microglial activation and increased ADCs, and subsequently, glial scars were associated with relative decreases in ADCs39. In pregnant rabbits, uterine ischemia followed by reperfusion was induced, and the value of ADC was significantly lower (with an incomplete recovery during reperfusion) in the group with hypertonia after birth than it was in the non-hypertonic group40. So far, clinical applications are limited and very few articles have reported the practical use of DWI in fetal brain evaluation. In one study41, we analyzed the contribution of DWI in the evaluation of microstructural changes in the white matter and the correlation between DWI, MRI and neuropathological findings. There was a strong correlation between increased ADCs, white matter signal abnormalities and vasogenic edema with astrogliosis of the cerebral parenchyma. Therefore, DWI provides a more objective way of analyzing the fetal white matter. However, it must be stressed that the prognosis of diffuse white matter abnormalities with increased ADCs is unknown. In a recent article, DWI permitted the early detection of focal ischemic damage in the survivor of a monochorionic twin pregnancy42. Diffuse restricted diffusion has been demonstrated in the enlarged hemisphere of the brain of a fetus with hemimegalencephaly43. A number of small series have analyzed the contribution of DWI in fetal renal pathology31, 33. DWI may be of interest in cases of vascular pathology with renal involvement (e.g. twin–twin transfusion syndrome, renal vein thrombosis). In two reported cases, the renal ADC of the donor twin was higher than that of the recipient twin and this difference could be correlated with the severity of the syndrome31. In one case of renal vein thrombosis, the difference in signal of the two kidneys was more obvious on DWI images (decreased ADC in the abnormal kidney)31. The ADC value may be elevated in multicystic dysplastic kidneys, huge dilatations or polycystic kidneys31, 33. The contribution of DWI (compared with ultrasound) is not yet obvious in this field. The application of DWI in the evaluation of lung pathologies has not been analyzed. Proton magnetic resonance spectroscopy (H-MRS) is a non-invasive imaging technique for examining cerebral metabolism. It is based on measurement of chemical shift, i.e. the local change in resonant frequency due to different chemical environments. It can identify the presence of certain metabolite substrates and the H-MR spectrum is obtained within a specific anatomical region of interest. The recommended size of the region of interest varies according to the authors, and a compromise must be found between a small region, with its consequent decreased signal-to-noise ratio, and a large region that can include several anatomical structures but there is the possibility of contamination by the subcutaneous fat of the fetal scalp44-46. H-MRS has been studied in pediatric neuroradiology for a long time. Normal patterns throughout brain maturation after birth have been assessed47, 48, and H-MRS is commonly used to evaluate seizures, metabolic disorders, brain tumors, neurodegenerative disorders and ischemic insults49, 50. However, it is difficult to perform H-MRS in a fetus because the coils used are not dedicated to brain imaging; the coil is distant from the fetal brain and the time of acquisition is long. Most authors have studied H-MRS in fetuses after 30 gestational weeks44, 45, 51-54. In normal fetuses, various metabolites can be identified whose presence is a marker of neurons and axons (N acetyl aspartate (NAA)), cellular energy metabolism (creatine (Cr)), myelination (choline (Cho)) or a glial marker (MyoInositol (mI)). Therefore, H-MRS may reflect brain maturation. At 22 weeks, Cr is clearly visible, whereas there is minimal NAA detected, represented by a small NAA peak. The spectrum is characterized mainly by two prominent peaks, assigned to mI and Cho46, 55. Throughout pregnancy, brain development changes result in the development of dendrites and synapses, inducing a progressive increase of the NAA peak. The process of myelination generates a reduction in the Cho peak. At 34 weeks, the metabolic pattern is very similar to the neonatal spectrum. In normal conditions, lactates are not detectable44, 45, 53-55. The presence of lactates in the brain is abnormal. They are formed as the end-product of glycolysis under anaerobic conditions and are generated as a consequence of hypoxia. Their presence is said to indicate poor neurodevelopmental outcome. Thus, the identification of lactates in fetuses might be useful during the clinical management of pregnancies with possible fetal compromise. Lactates have been detected in utero using H-MRS in the basal ganglia of two out of six fetuses presenting with hydrocephalus44 and in the basal ganglia and surrounding parenchyma of a fetus with gastroschisis and mild pre-eclampsia56. They were also detected in the back muscle of a fetus with severe intrauterine growth restriction on the day before he died57. A significant decrease in NAA level and the presence of lactates was found in six fetuses with intrauterine growth restriction and/or maternal hypertension58. So far, except in one case mentioned above57, there has been no reported application of H-MRS in the analysis of an organ outside the central nervous system. To date, functional magnetic resonance spectroscopy (f-MRI) has been used mainly to investigate brain activity in response to various stimuli. Its applications in children are numerous, and many articles have reported the use of this technique in autism, attention deficit hyperactivity disorder, dyslexia and epilepsy. Using f-MRI, it is possible to establish language mapping, to evaluate hemispheric language dominance, to study sensorimotor cortical activation, to evaluate language organization following stroke or to predict the potential sensorimotor consequences of a surgical procedure. The technique is based on the BOLD (blood oxygen level-dependent) effect. Local neuronal activity increases in response to a stimulus, inducing an increase in local blood flow and, consequently, an increase in venous blood oxygenation and in blood volume, which is known as the hemodynamic response. This leads to an increase in the local MR signal, known as BOLD contrast. Therefore, f-MRI may also provide information about blood oxygenation and blood flow. f-MRI is performed using echoplanar imaging, which is highly sensitive to differences in magnetic susceptibility; the difference between oxy- and deoxyhemoglobin generates the f-MRI signal59, 60. To my knowledge, only one center in the world (Nottingham, UK) has evaluated f-MRI in fetuses and studies this method in the framework of research protocols. In this center, f-MRI studies have been carried out on a low field unit (0.5 T), with consequent reduction of susceptibility artifacts, acoustic noise and tissue heating. Conversely, the signal-to-noise ratio and the BOLD sensitivity are decreased, which is a major disadvantage. Moreover, there are problems related to motion artifacts because of the long acquisition time (about 20 min). In order to reduce motion, imaging was performed in fetuses in a cephalad position and late in pregnancy (after 36 weeks), when the fetal head was engaged in the maternal pelvis. The exposure of the fetus to acoustic noise during the acquisition of the images must also be considered, which is loud for echoplanar imaging but is, however, attenuated by the maternal abdomen and the fluid filling the fetus's ear60. Two types of stimuli have been used: vibroacoustic and visual stimuli. To investigate vibroacoustic stimulation, a sound level of 95–100 dB was generated using headphones strapped to the maternal abdomen in 15 pregnancies. Activation was observed in the temporal lobe of seven of the fetuses61. To investigate visual stimulation, a red light-emitting diode (LED) cluster lamp was applied to the maternal abdomen. Ultrasound was performed before f-MRI to ensure that the fetal eyes were directed towards the maternal abdomen and the light source was positioned in order to overlay the fetal eyes. In three of eight fetuses, activation was not observed. In four of the five cases with activation, the activated area was observed within the frontal region, with no significant activation detected within the visual areas62. The hemodynamic response to these stimuli differs between fetuses and adults. This may reflect differences in the oxygen affinity of fetal and adult hemoglobin, immaturity of vascular control mechanisms, differences in activation in the immature brain (immaturity of the synaptic connections precluding neuronal activity in the usual cortical area of interest) and low sensitivity of the fetal f-MRI technique60. To my knowledge, there are no published applications of f-MRI in human fetal pathology. Recently, the consequences of maternal hypoxia in ewes was reported in an experimental study. BOLD imaging was obtained in brain, heart, lungs and liver; BOLD signal intensity appeared to be correlated with changes of oxyhemoglobin saturation during hypoxia. The decrease in signal intensity was more pronounced in the liver than in it was in the brain, reflecting the well-known redistribution of fetal blood flow during hypoxia63. Therefore, f-MRI could be a useful tool in the evaluation of blood oxygenation in human fetuses and might play a role in monitoring at-risk fetuses. Using the information on brain activity provided by f-MRI, it might be theoretically possible to evaluate sensorial activity in fetuses. The applications could be numerous, including, for example, searching for visual impairment in fetuses with septal agenesis, or for hearing loss in fetuses with cytomegalovirus seroconversion. This is currently aspirational. Ultrasound and conventional MRI will remain mandatory for studying fetal brain morphology, and new developments are expected in both modalities. The new fetal MRI techniques mentioned above that are currently being developed in a research setting will certainly have applications in common practice. This is already the case for DWI. These techniques will make it possible to analyze the fetal brain from a completely new point of view. However, we must be careful not to act as sorcerers' apprentices. We should remain cognizant of the fact that the safety of these techniques has not yet been proven. We do not know if high gradients, intense light or high-level sounds are safe for the fetus, and follow-up of these patients after birth is essential64. Moreover, these techniques dramatically lengthen acquisition time. For ethical and practical reasons, it will be very difficult to determine normal values of the fetal brain.