T2*-weighted placental MRI: basic research tool or emerging clinical test for placental dysfunction?
Introduction
Placental dysfunction is a common obstetric problem that complicates 5–10% of all pregnancies1. It is a progressive condition, in which the supply of oxygen and nutrients to the fetus is insufficient to maintain normal growth and organ development. The association between low birth weight and adverse neonatal outcome is well described2. Moreover, there is evidence to support the hypothesis, known as Barker's hypothesis3, that common adult diseases, such as metabolic syndrome and cardiovascular disease, originate from abnormal fetal programming due to placental dysfunction. There is no treatment available currently to improve placental function. However, correct identification of placental dysfunction antenatally leads to a four-fold improvement in neonatal outcome, as it allows for timely delivery, thereby reducing the risk of irreversible fetal organ damage4.
Antenatal screening for placental dysfunction focuses on fetal-weight estimates and fetal and umbilical Doppler flow measurements. However, fetal size does not reflect precisely placental function. In addition, in late-onset placental dysfunction, Doppler flow usually remains normal5. Therefore, additional markers that reflect directly placental function have the potential to improve considerably antenatal screening for placental dysfunction.
There is increasing interest in placental magnetic resonance imaging (MRI), due to its potential to detect placental dysfunction in vivo. In particular, T2*-weighted MRI has proved to be a simple and useful method with which to assess placental health, either by measuring directly the value of the transverse relaxation time, T2*, or by assessing the relative change in the raw T2*-weighted signal in response to a given challenge (i.e. the blood-oxygen-level-dependent effect (ΔBOLD)). T2* depends primarily on oxygenation, but also on other tissue characteristics, including villus density, inhomogeneities in the distribution of oxygenated blood, magnetic field inhomogeneities and the presence of other paramagnetic molecules6.
In this Opinion, we explain the physiological basis for changes in T2*-weighted signal intensity in the placenta. We review previous literature on T2*-weighted placental MRI and thereby provide reference values for scanning at 1.5- and 3-Tesla (T). We include practical guidance on how to optimize T2*-weighted scans, and outline the clinical potential of T2*-weighted MRI as a test of placental dysfunction.
Basic concepts of T2*-weighted MRI
T2* is the time taken for the observed decay of transverse magnetization to reach 37%. It is a multifactorial, empirical measure that depends on many biological and physical features of the tissue. Specifically, in the placenta, T2* depends on the underlying fundamental nuclear magnetic resonance transverse relaxation time (T2), on the volume fractions of the maternal and fetal blood and the villous tissue, and on spatial variations in the magnetic field across each voxel. The T2 of blood depends on the oxygen saturation of hemoglobin, as well as factors such as water diffusion and exchange between water and other molecules, and water binding. Spatial variations in the magnetic field can be caused by a number of factors, but arise primarily from the heterogeneous distribution of deoxygenated blood. Oxyhemoglobin and tissue both have similar, slightly diamagnetic (negative), magnetic susceptibility, whereas deoxyhemoglobin has paramagnetic (positive) magnetic susceptibility. Variations in magnetic susceptibility will lead to variations in the induced magnetic field within a tissue. Therefore, any spatial variations in the concentration of deoxyhemoglobin across the placenta, for instance due to deoxygenated blood confined within fetal arterioles or deoxygenated maternal blood draining from the placental intervillous space, could lead to a variation in magnetic field across the voxel.
In addition, most T2*-weighted images retain some residual dependence on the longitudinal relaxation time, T1, and dissolved oxygen may influence the T1 of blood, which could cause a further increase in signal in T2*-weighted imaging7, 8. Biologically, infarcts can contain intracellular methemoglobin, which is particularly paramagnetic, and fibrosis, which can also reduce T29. In some cases, incoherent movement of the blood during the echo time may reduce the signal intensity of T2*-weighted scans. Finally, large-scale magnetic field variations will affect T2*. While static effects can be minimized by shimming, maternal respiration can cause fluctuation in T2*, which may not be addressed by shimming.
T2*-weighted images form the basis of the BOLD effect used in functional MRI (fMRI) of the brain, in which changes in blood oxygenation and blood volume in response to neuronal activation lead to a change in the observed MRI signal. In contrast to measuring the absolute value of T2*, the BOLD effect is a relative measure, usually obtained by monitoring the T2*-weighted signal intensity over time, to determine the percentage change in signal, relative to the baseline, in response to intervention. For instance, a number of studies have involved increasing the oxygen saturation of the maternal blood within the placenta by having the mother breathe 100% oxygen. Assuming that there is no other physiological effect of the hyperoxic challenge, then any change in the placental T2*-weighted signal (the BOLD effect) reflects the change in placental oxygenation.
How to obtain T2*-weighted placental images
If T2* is measured quantitatively, then normal reference values can be established to differentiate between normal and pathological placentae. There are several ways in which to obtain T2*-weighted MR images of the placenta, including: fast gradient echo (GE) sequence or GE echo planar imaging (EPI) sequences, run in either single or multi-echo mode. Compared with EPI, GE sequences can achieve higher spatial resolution, but take longer to acquire, which makes them more sensitive to motion artifacts. For multi-echo EPI data, there is effectively no motion during the time (< 200 ms) required to acquire each slice, rendering it very robust to intraslice motion, although the spatial resolution achievable is generally lower than that possible on GE scans. At term, the whole placenta can be scanned with EPI during a single maternal breath-hold using multiband/simultaneous multislice approaches. In principle, GE-EPI will be affected by susceptibility artifacts in the abdomen, but the spherical nature of the pregnant uterus and the fluid filling the uterine and fetal cavities make GE-EPI a very robust sequence for use in pregnancy, even at 3 T. It is preferable, however, to use image-based B0 shimming for GE-EPI at 3 T when available.
We recommend using a transaxial imaging plane, with the field of view aligned with the scanner coordinates to simplify planning. While coronal and sagittal imaging planes may also be used, the transaxial plane is commonly the most useful, as it assures coverage of the main functional placental direction (maternal basal plate to fetal chorionic plate) within each image, without having to consider movement between slices. It is important that the field of view covers the whole of the abdomen, in order to avoid artifacts from parallel imaging reconstruction techniques (e.g. SENSE artifacts). In addition, good fat suppression is required, for which the B0 shim box must cover all of the maternal subcutaneous fat.
In addition, we suggest that the subject lies in the left lateral position (10°–20° tilt), mainly to avoid caval vein compression, as even minor changes in the maternal circulation may have a detectable impact on the placental T2* value. A recent study by Humphries et al.10 found that a maternal supine position in late pregnancy reduced the flow in the inferior caval vein by 85% compared with the left lateral position. While the venous return in the azygos vein was increased, probably to compensate, cardiac output was reduced by 16%. We also find that the left lateral position makes it easier to image posterior placentae, as often it allows placement of the receiver coils closer to the placenta. If the woman is scanned in a supine position, we recommend ensuring frequent monitoring of maternal blood pressure, heart rate and oxygen saturation, using MR-compatible monitoring equipment. In addition, splitting the scanning session into shorter sessions, each lasting no longer than 30–45 min, will improve maternal comfort. Accurate recording of the position and tilt angle is important for analysis purposes, given the possibility that the maternal position may influence her physiology. Using this approach, a baseline T2*-weighted dataset covering the entire placenta can be acquired in under 5 min, from the time the woman enters to the time she leaves the scanning room, which suggests that this simple measurement could be used for rapid monitoring of placental health.
Regarding the optimal field strength for placental imaging, 1.5 T is the field strength used in most placental MRI studies reported so far. However, systems with a field strength of 3 T are replacing the 1.5-T systems in many centers and it is likely that 3 T will be used increasingly in future placental MRI studies. As the field strength increases, the sensitivity of MRI increases, and the sensitivity to the BOLD contrast increases superlinearly. 3-T images can provide exquisite detail on oxygenation in the human placenta. Yet, for a standard clinical test of placental dysfunction, 1.5 T can give sufficient information, while the intrinsically shorter T2* values at higher field strength can make it difficult to acquire data at sufficiently short echo times as to be sensitive enough to pick up changes in T2*. In addition, at 3 T, the radiofrequency (RF) fields can be highly inhomogeneous, leading to signal dropout in the center of the field of view, particularly when scanning the larger abdomen at late gestation, and this can make it harder to image posterior placentae. Higher field strength also increases the risk of geometric distortion due to air–tissue interfaces, requiring special attention to B0 shimming, especially if using high-resolution EPI. While RF heating and acoustic noise are general safety concerns for fetal MRI, 3 T magnets tend to have stronger gradient coils and hence higher acoustic output. Based on current knowledge, we recommend using Level-0 normal operating mode of RF, i.e. 2 W/kg, labeled by some vendors as ‘low SAR’ or ‘fetal MRI’, since the fetus has reduced heat-loss mechanisms compared with the adult. Similarly, we recommend not using parallel transmit approaches unless the system vendor explicitly advises their use in pregnancy. We also recommend using the scanner's options for reduced acoustic noise.
Data analysis
Quantitative placental T2* relaxation time
For more detailed placental assessment, it may be preferable to measure the absolute T2* relaxation time rather than just acquiring a T2*-weighted image, since quantitative measurements make the results more comparable over time for longitudinal studies, between patients in cross-sectional studies and between scanners and sites for a given field strength in multicenter studies.
We generally assume that T2* describes a monoexponential decay. As such, in principle, T2* can be estimated by a measurement at two echo times, the first being the shortest possible echo time and the second being an echo time equal to T2* of the particular tissue of interest. The two echoes can be collected either from a single RF pulse (double-echo sequence) or by repeating a single echo acquisition at two echo times. T2* can then be estimated by calculating –dTe ln (S1/S2), where S1 and S2 are the signals from the first and second echo images, respectively, and dTe is the difference in the echo times of these two images.
Alternatively, multiple echoes can be fitted for T2*. Comparing the signal decay to a monoexponential function will highlight any systematic errors in the data, for example from shimming errors or high noise floors due to Riccian noise. We therefore recommend sampling the signal decay of at least three echo times (with the longest echo time providing a signal that is above the noise floor). If the data are fitted for a monoexponential decay, then a weighted linear fit should be used and efforts should be made to eliminate data points influenced by the noise floor. Alternatively, the data could be fitted to a more complex function, for instance, to take into account the effects of through-slice dephasing11. Multi-echo sequences are automatically recorded simultaneously, but will be more sensitive to motion during the acquisition, while repeated single-echo acquisitions allow a wider range of echo times to be explored, which can provide better discrimination of the signal decay curve (Figure 1).
Because of the heterogeneous nature of placental tissue and placental oxygenation, we recommend including as much placental tissue as possible in the region of interest (ROI) to improve the reproducibility of the T2* measurements. If possible, a 3D volume should be obtained, or the T2* estimate should be based on an average of at least three slices spaced across the placenta. In each slice, the ROI should cover the entire placenta in a cross-section perpendicular to the placental surface, and amniotic fluid and large maternal vessels should be avoided by careful segmentation (Figure 2). In apparently inhomogeneous placentae, the segmentation should include all placental tissues, even regions with very low T2* values that appear almost homogeneously black. If necessary, segmentation can be assisted by checking the placental outline in T2-weighted (fast spin echo) anatomical images, taking into account any movement between the acquisitions. The T2* measurement should be repeated at least three times, each separated by a minimum of 3 min, in order to identify artifacts related to uterine contractions12. Extreme T2* values should be checked for susceptibility artifacts, movement artifacts (both maternal and fetal) and contractions, and any affected values should be excluded from the analysis. If both magnitude and phase maps are stored for T2* mapping datasets, then susceptibility-weighted imaging and susceptibility mapping can be performed, providing information that is more specific to oxygenation than is T2* alone.
Placental T2* mapping
In addition to reporting the mean T2* relaxation time for the entire placental slice (2D) or placental organ (3D), a map of the T2* value of each pixel can be created, to illustrate the T2* heterogeneity (Figure 3a,b). A histogram showing the distribution of T2* values normalized by placental volume may provide a quantitative evaluation of the degree of tissue heterogeneity, which may be related to placental pathology13, 14 (Figure 3c,d). It also provides an opportunity to relate the T2* values to specific anatomic features.
) and two of PE (
). (d) Histogram of T2* values normalized by placental volume, color-coded according to gestational age at MRI, in normal pregnancies at: 22.1 weeks (
), 34 weeks (
) and 37 weeks (
). Dotted lines indicate cases of PE. (Reproduced with permission from Hutter et al.13.)Placental BOLD effect
If a time series of T2*-weighted MRI scans is acquired across the placenta, then the dynamic change in signal intensity can be estimated; this is generally known as the BOLD effect (ΔBOLD) (Figure 4). For instance, experimental studies have investigated the placental response to maternal hyperoxia using the following equation to estimate the hyperoxic BOLD effect15: ΔBOLD (%) = (SHyperox − SNormox)/SNormox × 100, where SHyperox is the signal intensity during maternal hyperoxia and SNormox is the signal intensity during maternal normoxia. In order to achieve reliable estimates of the hyperoxic BOLD effect, the dynamic scan should be continued until the placental response reaches a steady state. In the normal placenta, the steady-state level is reached within 5 min of maternal hyperoxia. However, when the placenta is dysfunctional, this time can be prolonged, sometimes even beyond 10 min. Stopping the scan before a steady-state level is reached may lead to underestimation of the BOLD effect. Also, T2*-weighted scans are highly sensitive to uterine contractions, and contractions may lead to underestimation of the hyperoxic BOLD effect. Uterine contractions may be identified by uterine movements in the dynamic scan or by the pregnant woman herself. However, a recent publication suggests that most contractions are subclinical; therefore, one cannot rely on maternal reports alone12. Alternatively, if T2* is measured, then the BOLD signal can be characterized by the absolute change in T2* (ΔT2*), as follows: ΔT2* = T2*Hyperox − T2*Normox.
Literature search
A systematic search was performed of PubMed/MEDLINE (last updated 8 November 2018) by combining the following terms: ‘placenta’,‘MRI’,‘T2*’ and ‘BOLD’. The precise search string is given in Appendix S1. This search yielded 142 results. These papers were reviewed by title and abstract using prespecified inclusion criteria, which required that: (1) papers were original studies and (2) they included T2*-weighted MRI of the placenta. There were no language restrictions, and we considered both human and animal studies. We assessed the resulting 37 studies by full-text review, identifying 22 studies which fulfilled the inclusion criteria. An additional literature search using the reference lists of the included studies revealed no further studies for this review. The included studies on the placental BOLD effect (n = 11) and those on placental T2* (n = 13) are listed in Tables S1 and S2, respectively.
Dynamic T2*-weighted placental MRI: placental BOLD MRI
Most human studies investigated the changes in placental signal intensity during maternal hyperoxia (3–10 min of breathing 100% oxygen)15-21, and one study investigated the effect of uterine contractions12. In addition, three animal studies described changes in placental signal intensity during maternal hypoxia22, normoxic hypercapnia23 and injection of a vasoconstrictor24. Human studies were performed at both 1.5 T12, 15-18, 21 and 3 T19, 20 using single-shot GE-EPI sequences. However, the specific echo-time (TE) and repetition time (TR) varied between studies (Table S1).
Placental BOLD effects in experimental animal studies. The experimental animal studies investigated the correlation between acute changes in placental oxygenation and changes in placental BOLD signal intensity. In a sheep model, maternal hypoxia led to a mean reduction of the BOLD signal of a placental cotyledon of 29% (range, 9–43%)22. In a rat model, injection of prostaglandin F2α, which is known to cause placental vasoconstriction, led to a 10% reduction in placental BOLD signal intensity24. A recent study conducted in mice investigated the effect of normoxic hypercapnia on the placental BOLD signal and demonstrated a 44% reduction in signal, probably caused by placental hypoperfusion due to the vasoconstrictive effect of hypercapnia23.
Hyperoxic BOLD effect (ΔBOLD) in normal human pregnancy. In normal human pregnancy, the placental hyperoxic BOLD effect has been addressed in three studies15, 16, 21. A significant increase in the placental signal intensity (ΔBOLD) was demonstrated during hyperoxia15, 16, 21, with ΔBOLD being 20% in normal pregnancies at term21.The findings are presented in Figure 5.
) and 13 pregnancies complicated by fetal growth restriction (
). Lines indicate linear correlation between ΔBOLD and gestational age in normal pregnancy (mean and 95% prediction interval). (Reproduced with permission from Sinding et al.21.)Hyperoxic BOLD effect (ΔBOLD) in human pregnancies complicated by fetal growth restriction (FGR). Three studies included FGR pregnancies, with FGR defined by low birth weight16, 19 or low birth weight in combination with abnormal postnatal placental examination21 (Table S1). Initially, a small study including four FGR cases demonstrated inconsistent results regarding the hyperoxic BOLD effect16. However, a larger case–control study, including 13 FGR cases and 49 healthy controls, demonstrated that placental ΔBOLD was significantly increased among FGR cases21 (Figure 5). In addition, the study by Lou et al.19 demonstrated that, in monochorionic twin pairs complicated by selective FGR, the time to reach a hyperoxic steady state is prolonged in the smaller twin19.
Quantitative placental T2* relaxation time (baseline T2*)
The placental T2* relaxation time has been estimated in seven human studies7, 8, 13, 21, 25-27 and six animal studies14, 28-32 (Table S2). The main purpose of these studies was to estimate the baseline T2* value in normal pregnancy and in pregnancy complicated by FGR. In addition, the hyperoxic placental T2* response (ΔT2*) has been addressed in five of these studies7, 8, 21, 28, 32.
Five human studies were performed at 1.5 T7, 8, 21, 25, 27 and two human studies were performed at 3 T13, 26 using multi-echo gradient-recalled-echo sequences. However, the MRI protocols were different with regards to the number and values of the included echo times (Table S2). Such protocol differences result in different accuracies in short and long T2* measurements. In all human studies, the T2* value was obtained in large ROIs covering the entire placental cross-section. Moreover, in the studies of Sinding et al.21, 25, 27, the T2* value was calculated as an average of three slices, the studies by Hutter et al.13 and Armstrong et al.26 were based on 3D placental models and, in the remaining studies, the T2* value was based on a single slice only7, 8.
Quantitative placental T2* relaxation time in normal pregnancy (baseline T2*). In normal pregnancy, the association between baseline T2* and gestational age has been investigated in six studies, one using a 3-T MRI system13 and five using a 1.5-T MRI system7,8,21,25,27. The largest studies found a strong negative correlation between baseline T2* and gestational age13,21,25,27, while Ingram et al. found a non-significant negative correlation7 and a small study on 14 pregnant women failed to demonstrate any correlation with gestational age8.
In normal pregnancies at term, the placental baseline T2* was estimated to be 47–58 ms at 1.5 T7, 8, 25, 27 and 25 ms at 3 T13 (Figures 6 and 3, respectively). The reproducibility of baseline T2* measurements at 1.5 T has been addressed by Sinding et al.25, who found the 95% limits of agreement for within- and between-session variation for a single-slice placental T2* measurement to be –2.1 ± 10.4 ms and –0.6 ± 22.6 ms, respectively. In addition, their study demonstrated that including large ROIs and averaging values for several slices improves the reproducibility of the T2* measurement. A recent publication by Hutter et al.13 demonstrated that the within- and between-session variation of T2* at 3 T was 1.83 ± 2.42 ms and 1.91 ± 21.60 ms, respectively.
T2*-weighted 1.5-T placental MRI in normal pregnancy and pregnancy complicated by placental dysfunction. (a) T2*-weighted MRI of placentae (arrows) in four normal pregnancies (upper row) and four cases of placental dysfunction (lower row). Dysfunctional placentae appear markedly darker than normal placentae. (Reproduced from Sinding et al.25.) (b–d) Association between placental T2* measurements and gestational age in 49 normal pregnancies (
) and 13 pregnancies complicated by placental dysfunction (
). Lines indicate least square fit of normal pregnancies (mean and 95% prediction interval). (b) Baseline placental T2*. (c) Hyperoxic placental T2*. (d) Hyperoxic change in T2* value (ΔT2*). (Reproduced with permission from Sinding et al.21.)
Using placental T2* mapping, it has been demonstrated that the within-placental heterogeneity of T2* is in the range of 10–200 ms, and the hyperintense areas correspond with the center of the cotyledons. In addition, the degree of heterogeneity is increased as pregnancy advances13. The placental histogram is presented in Figure 3.
Baseline T2* in pregnancy complicated by FGR. Four human studies included FGR pregnancy7, 21, 25, 27. In these studies, FGR was defined by either low birth weight25, 27, a combination of low birth weight and abnormal umbilical artery Doppler7, or low birth weight and abnormal postnatal placental examination21. Three of these studies were designed as case–control studies7, 21, 25. In each study, the baseline placental T2* value was significantly reduced in FGR cases when compared with normal controls (Figure 6). In a prospective cohort study, a low placental T2* value was a strong predictor of low birth weight, and it performed significantly better than did uterine artery Doppler flow measurements27.
Hyperoxic T2* response (ΔT2*). ΔT2* in response to maternal hyperoxia has been investigated in three human studies7, 8, 21, all of which found a significant increase in placental T2*. The largest study was conducted in 49 normal pregnancies, and demonstrated a strong positive correlation between ΔT2* and gestational age21 (Figure 6). The remaining two smaller studies, Ingram et al.7 (n = 28) and Huen et al.8 (n = 14), failed to demonstrate any correlation with gestational age. The increase in placental T2* during hyperoxia is supported by two experimental animal studies in rats32 and mice28.
Two human studies included ΔT2* in FGR pregnancies7, 21. FGR was defined by either low birth weight and abnormal postpartum placental examination21 or low birth weight and abnormal Doppler ultrasound findings7. These studies found no significant difference between ΔT2* in FGR pregnancies and normal controls. In animal studies, there have been conflicting results concerning ΔT2* in FGR. One experimental rat study by Chalouhi et al.32 reported that placental ΔT2* in response to hyperoxia was reduced among FGR cases, whereas a recent experimental study in mice, by Collinot et al.28, reported that placental ΔT2* in response to hyperoxia was increased in FGR.
Discussion
Based on our literature review, we conclude that T2*-weighted placental MRI is a promising marker of placental dysfunction. Approximate reference values for the placenta, scanned at 1.5 T and 3 T, and tips and guidance on how to obtain placental baseline T2* values, are presented in Table 1.
| Placental MRI T2* | |
| Approximate reference values | |
| At 20 weeks | 150 ms (1.5-T system); 90 ms (3-T system) |
| At 40 weeks | 47–58 ms (1.5-T system); 25 ms (3-T system) |
| Correlation with gestational age | Negative |
| Placental T2* in FGR | Decreased |
| Placental hyperoxic ΔBOLD | |
| Approximate reference values | |
| At 20 weeks | 5% (1.5-T system) |
| At 40 weeks | 20% (1.5-T system) |
| Correlation with gestational age | Positive |
| Hyperoxic ΔBOLD in FGR | Increased |
| Tips for placental MRI protocol |
|
| Practical guidance |
|
| Tips for image analysis |
|
- ΔBOLD, blood-oxygen-level-dependent response; EPI, echo planar imaging; FGR, fetal growth restriction; GA, gestational age; ROI, region of interest.
It is a consistent finding that placental dysfunction is associated with a lower baseline T2* value7, 14, 21, 25, 28. The underlying physiological explanation for the low T2* value in placental dysfunction is not fully understood. It is well known that placental dysfunction is associated with impaired maternal placental perfusion due to defective transformation of the spiral arteries, which may lead to placental hypoxia33. As T2*-weighted MRI is particularly sensitive to the amount of deoxyhemoglobin present in tissue, placental hypoxia may reduce T2*. However, in addition, changes in placental morphology, such as altered villous density, deposits of fibrin and infarcts, may contribute to a lower T2* value. The latter is supported by a study by Wright et al.34, demonstrating a significant positive correlation between placental T2 and morphological findings such as the relative density of fibrin deposition and villous volume.
The majority of studies support a negative correlation between baseline T2* and gestational age7, 13, 21, 25. Direct measurements of oxygenation in the intervillous space have demonstrated that oxygenation decreases as gestation advances35, probably as a result of increased extraction of oxygen from the intervillous space due to increased fetoplacental metabolic demand as pregnancy advances. In addition, normal physiological placental maturation is associated with morphological changes that may alter the placental MRI signal intensity. Any comparison between groups should always include adjustment for gestational age at MRI.
It has been demonstrated that the placenta has a heterogeneous appearance in T2*-weighted images15, and maps of T2* demonstrate that the heterogeneity increases with gestational age13. This heterogeneity may reflect the oxygenation of the placental cotyledons. The cotyledon has a highly oxygenated center, where the maternal spiral arteries enter the intervillous space, and the oxygenation decreases towards the margins of the cotyledon, where maternal blood drains back into the maternal venous circulation. Combining contrast-enhanced MRI and placental T2* maps in non-human primates supports this hypothesis, as areas with a higher T2* value correspond with the entry point of the spiral artery in the center of the cotyledon31.
Several studies have demonstrated that maternal hyperoxia increases the placental signal intensity in the T2*-weighted image, also known as the hyperoxic BOLD effect (ΔBOLD)15, 16, 21. The direct correlation between changes in placental signal intensity and changes in placental oxygenation has been confirmed by an invasive experiment in a sheep model22. In these experiments, deoxyhemoglobin can be regarded as an intrinsic contrast agent31. At room air (21% O2), maternal arterial blood is already fully saturated; hyperoxia leads to increased saturation of the maternal venous blood in the intervillous space, which is generally accepted to be the main contributor to the hyperoxic BOLD effect. Fetal blood may also contribute to the hyperoxic BOLD effect, but this contribution is expected to be marginal, as the proportion of fetal blood in the placenta is markedly lower than that of maternal blood.
In the dysfunctional placenta, ΔBOLD is increased when compared to that of the normal placenta21. To understand this finding, one should remember that ΔBOLD is a relative measurement. Any increase in ΔBOLD may be due to an increase in absolute signal intensity and/or a reduced baseline absolute signal intensity. In the dysfunctional placenta, it has been demonstrated that the baseline T2* is markedly reduced when compared to normal controls21. However, the change in absolute placental oxygenation, estimated by ΔT2*, did not differ between normal and dysfunctional placentae21. This finding suggests that the higher hyperoxic BOLD effect in dysfunctional placentae may simply reflect a lower T2* baseline value. Similarly, the positive correlation between ΔBOLD and gestational age in normal pregnancy may also be explained by a decreasing baseline T2*. These findings suggest that, as an initial clinical biomarker, the baseline T2* signal, which is fast and simple to obtain, may provide sufficient information with which to discriminate between a normal and a dysfunctional placenta.
Placental dysfunction is one of the main challenges in modern obstetrics and the leading cause of stillbirth36. Baseline T2* may provide a valuable tool with which to differentiate between the constitutionally small fetus and the fetus suffering from placental dysfunction. A reliable marker of placental dysfunction in vivo may increase the identification of placental dysfunction, which would allow for timely delivery and thereby reduce the risk of stillbirth. Baseline T2* may provide early subclinical identification of placental dysfunction, before abnormal fetal growth occurs. In addition, longitudinal studies based on baseline T2* offer an opportunity to evaluate the effect of treatment to improve placental function. An area of particular interest will be late-onset placental dysfunction, which can present with appropriately grown fetuses and normal umbilical artery Doppler, so that dysfunction often remains undetected. Other MRI markers of placental function, such as baseline T1-, T2- and diffusion-weighted MRI, still need to be evaluated properly, since a combination of different MRI markers may further improve the detection of placental dysfunction.
Conclusion
This Opinion provides approximate reference values for 1.5-T and 3-T MRI and useful guidance on how to obtain and analyze T2* placental values. Based on previous publications, placental T2* is a promising marker with which to discriminate between a normal and a dysfunctional placenta in vivo. Initial studies suggest that placental T2* has great clinical potential. As the value of T2* obtained depends on the particular pulse sequence used, pooling of results to provide clear guidelines for a specific T2* cut-off value is difficult at this stage. Future multicenter trials, with a consistent MRI protocol, performed at specific gestational ages, in clinically well-defined populations, should be the next step to determine the clinical potential of T2*-weighted placental MRI.
